NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.

Wilbur S, Williams M, Williams R, et al. Toxicological Profile for Carbon Monoxide. Atlanta (GA): Agency for Toxic Substances and Disease Registry (US); 2012 Jun.

Cover of Toxicological Profile for Carbon Monoxide

Toxicological Profile for Carbon Monoxide.

Show details

3HEALTH EFFECTS

3.1. INTRODUCTION

The primary purpose of this chapter is to provide public health officials, physicians, toxicologists, and other interested individuals and groups with an overall perspective on the toxicology of carbon monoxide. It contains descriptions and evaluations of toxicological studies and epidemiological investigations and provides conclusions, where possible, on the relevance of toxicity and toxicokinetic data to public health.

A glossary and list of acronyms, abbreviations, and symbols can be found at the end of this profile.

Focus of the Health Effects Review. The literature on the toxicology of carbon monoxide is immense. Relatively recent reviews of the literature have been prepared by various organizations, including the World Health Organization (WHO 1999) and EPA (2000, 2009g), and numerous reviews have been compiled on the clinical toxicology and management of acute carbon monoxide poisoning. The review of the literature provided in this section is not intended to be comprehensive; rather, it is focused on the more important, recent developments in the toxicological and epidemiological assessment of carbon monoxide. Studies that have potential relevance to understanding the lower end of the dose-response relationship for carbon monoxide are emphasized. Typically, dose-response information is presented in Toxicological Profiles in Levels of Significant Exposure (LSE) tables and figures. However, this information on carbon monoxide has been represented in various types of units (e.g., exposure concentration, blood COHb), and end points are not easily assigned into categories of NOAELs or LOAELs (e.g., odds ratio from epidemiological studies). Therefore, in place of an LSE table, Table 3-1 provides an overview of exposures (ppm) and blood COHb levels (percent) that have been associated with specific categories of health effects. Table 3-2 can be used to convert the reported ambient air carbon monoxide concentrations into equivalent predicted blood COHb levels. However, considerable uncertainty attends to these dose interconversions as well as their relevance to a particular outcome. Predicted COHb levels and corresponding exposure concentrations may have greater relevance to outcomes mediated through hypoxic mechanisms than to those mediated through non-hypoxic mechanisms. Furthermore, steady-state requires exposure durations of approximately 16–24 hours. Exposures to lower levels of carbon monoxide for longer durations and exposures to higher levels for shorter durations that achieve similar blood COHb levels may not yield equivalent responses.

Table 3-1. Carbon Monoxide Exposures and Carboxyhemoglobin Levels Associated with Health Effects from Selected Studies Representing the Lowest Adverse Effect Levels.

Table 3-1

Carbon Monoxide Exposures and Carboxyhemoglobin Levels Associated with Health Effects from Selected Studies Representing the Lowest Adverse Effect Levels.

Table 3-2. Predicted Steady-State Blood Carboxyhemoglobin (COHb) Levels.

Table 3-2

Predicted Steady-State Blood Carboxyhemoglobin (COHb) Levels.

The discussion of health effects has been limited to the inhalation exposure pathway. Carbon monoxide exists in the environment as a gas (Henry’s law constant >50,000 atm/mol fraction, 25 °C). As a result, humans can be exposed to carbon monoxide from breathing and/or contact with carbon monoxide in air. No information is available on the dermal absorption or toxicity of carbon monoxide resulting from exposures to gaseous carbon monoxide. However, as is the case for other gases that are avidly absorbed from the lung during inhalation (e.g., O2), dermal absorption of carbon monoxide through intact skin would be expected to make a substantially minor contribution to absorbed carbon monoxide, relative to inhalation. No information is available on the absorption or toxicity of carbon monoxide resulting from oral exposures to gaseous carbon monoxide. Although carbon monoxide can dissolve in water (23 mL carbon monoxide/L water, 20 °C), appreciable concentrations in drinking water would occur only at very high partial pressures of carbon monoxide in air, conditions in which inhalation would be the dominant absorption pathway. Therefore, the only relevant pathway of exposure to humans is the inhalation pathway, and oral and dermal exposures are not considered further.

3.2. DISCUSSION OF HEALTH EFFECTS

This section of the Toxicological Profile summarizes results obtained from clinical cases of carbon monoxide poisoning, studies of controlled exposures conducted in humans, epidemiological studies of health outcomes associated with ambient air carbon monoxide concentrations, and experimental studies conducted in various animal models. Reported clinical studies are limited to exposures of acute duration. Animal studies have examined longer-duration exposures. Epidemiological studies have examined outcomes in the context of chronic exposures or acute variations in exposure concentrations that occur during chronic exposures. Some studies (e.g., clinical studies) have reported carbon monoxide doses in terms of blood COHb levels, while most epidemiological studies and some animal studies have reported exposure concentrations (e.g., ppm, mg/m3). For comparability to human clinical studies, reported air carbon monoxide concentrations can be converted to corresponding equivalent steady-state COHb% values (i.e., the COHb% that would be achieved with continuous exposure to the reported air carbon monoxide concentration) through a model. This conversion has been made by application of the CFK model (see Section 3.4.5). Predicted steady-state blood COHb levels corresponding to a range of carbon monoxide exposure concentrations are presented in Table 3-2. This table can be used to convert the reported ambient air carbon monoxide concentrations into equivalent predicted blood COHb levels. Exposure concentrations are typically reported in units of ppm or mg/m3; in the Toxicological Profile, exposure units are presented in units of ppmv (indicated as ppm). At standard temperature and pressure (e.g., 25 °C, 760 Torr), the conversion factor is approximately 1 (i.e., 1 mg/m3≈0.87 ppm, 1 ppm≈1.15 mg/m3).

Overview of Acute Carbon Monoxide Toxicity and Modes of Action. Carbon monoxide exerts effects on cell metabolism through both hypoxic and non-hypoxic mechanisms. Both types of effects are thought to be largely (but not entirely) the result of the ability of carbon monoxide to bind to heme and alter function and/or metabolism of heme proteins. The binding affinity of carbon monoxide for hemoglobin is over 200 times greater than that of the affinity of oxygen for hemoglobin (Chakraborty et al. 2004). Formation of COHb decreases the O2 carrying capacity of blood and impairs release of O2 from Hb for its utilization in tissues. Through similar mechanisms, carbon monoxide decreases O2 storage in muscle cells by binding to, and displacing O2 from, myoglobin. Although all tissues are vulnerable to carbon monoxide-induced hypoxic injury, those having the highest O2 demand are particularly vulnerable, including the brain and heart. The developing fetus may also be a sensitive target of carbon monoxide, through hypoxic and/or non-hypoxic mechanisms (Carratu et al. 1993, 2000a, 2000b; De Salvia et al. 1995; Lopez et al. 2003, 2008; Stockard-Sullivan et al. 2003; Webber et al. 2003).

Acute carbon monoxide poisoning is largely the result of tissue hypoxia. Signs and symptoms of carbon monoxide toxicity, in order of increasing severity include: (1) headache, nausea, dilation of cutaneous vasculature, vomiting, dizziness, and blurred vision; (2) confusion, syncope, chest pain, dyspnea, weakness, tachycardia, and tachypnea rhabdomyolysis; and (3) palpitations, cardiac dysrhythmias, hypotension, myocardial ischemia, cardiac arrest, respiratory arrest, pulmonary edema, seizures, and coma (Kao and Nañagas 2006). Although binding of carbon monoxide to blood Hb is a primary component of the hypoxic mode of action of carbon monoxide, blood COHb levels have not been shown to be a reliable predictor of severity of acute toxicity, in part due to time elapsed from removal from carbon monoxide exposure to COHb measurement and to the effects of emergency medical intervention (i.e., treatment with oxygen) on COHb levels prior to COHb measurement (Hampson and Hauff 2008). In general, typical levels of COHb in nonsmokers are <2% (Adams et al. 1988; Allred et al. 1991; Anderson et al. 1973; Hinderliter et al. 1989; Kleinman et al. 1989, 1998; Sheps et al. 1987, 1990). Of this, approximately 0.2–1.0% is derived from endogenous production of carbon monoxide (Coburn et al. 1963; Delivoria-Papadopoulos et al. 1974; Longo 1977). Levels ranging from 2 to 6% have been shown to exacerbate underlying cardiovascular disease, including enhanced myocardial ischemia and increased cardiac arrhythmias (Adams et al. 1988; Allred et al. 1989, 1991; Anderson et al. 1973; Kleinman et al. 1989, 1998; Leaf and Kleinman 1996b). In general, signs and symptoms of acute carbon monoxide poisoning can present at COHb levels ranging from 3 to 24% (Hampson and Hauff 2008; Kao and Nañagas 2006). This level overlaps with levels found immediately following cigarette smoking, up to approximately 10% (Kao and Nañagas 2006). More severe signs of carbon monoxide poisoning are poorly correlated with blood COHb, with loss of consciousness occurring at a mean level of 24.3% (range: 2–70%; Hampson and Hauff 2008) and fatality at a mean level of 32.1% (range: 3.0–60%; Hampson and Hauff 2008). Exposures that result in COHb levels >50% are frequently fatal (Dolan 1985). Persistent neurologic sequelae, which can be delayed in onset, can also occur, including memory loss, impairments of concentration and language, affective disorders (e.g., depression), and Parkinsonism (Choi 2002; Klawans et al. 1982; Ringel and Klawans 1972), some of which may be related to pathologic changes in the brain (Gorman et al. 2003; Lo et al. 2007).

Most of the non-hypoxic mechanisms of action of carbon monoxide have been attributed to binding of carbon monoxide to heme in proteins other than Hb. Notable targets of carbon monoxide include components of several important physiological regulatory systems, such as brain and muscle oxygen storage and utilization (myoglobin, neuroglobin); nitric oxide cell signaling pathway (e.g., nitric oxide synthase, guanylyl cyclase); prostaglandin cell signaling pathway (cyclooxygenase, prostaglandin H synthase); energy metabolism and mitochondrial respiration (cytochrome c oxidase, cytochrome c, NADPH oxidase); steroid and drug metabolism (cytochrome P450), cellular redox balance, and ROS (catalase, peroxidases); and various transcription factors (e.g., neuronal PAS domain protein, NPAS2, implicated in regulation of circadian rhythm). Endogenously produced carbon monoxide may participate in the physiological regulation of some, if not all, of these systems. This has potentially important implications for the understanding of carbon monoxide toxicology and dose-response relationships for the following reasons: (1) carbon monoxide modulation of physiological processes may underlie some aspects of the toxicity of exogenous carbon monoxide; (2) exogenous carbon monoxide may disrupt physiological regulation of those systems that are responsive to endogenous carbon monoxide (e.g., vascular resistance); and (3) exposures to exogenous carbon monoxide may affect carbon monoxide-mediated physiological responses at levels that approach those resulting from endogenous production. One implication of this is that the dose threshold for effects of exogenous carbon monoxide on carbon monoxide-modulated physiological systems may lie near or below ambient air carbon monoxide concentrations.

Epidemiological Studies. Epidemiological studies of health outcomes associated with exposure to carbon monoxide fall into two major categories. Some studies have examined relationships between long-term average ambient air carbon monoxide concentrations and health outcomes, where the air carbon monoxide concentrations represent averages over years or decades. Other studies have examined associations between outcomes and air concentrations measured over relatively narrow time slices (e.g., 1–24-hour average or maxima). These short-term studies are particularly suited to outcomes that might be related to short-term elevations in carbon monoxide exposure in close temporal proximity to the outcome of interest, which might otherwise be lost in longer-term average carbon monoxide concentrations (e.g., change in heart rate, heart rate variability, arrhythmia). End points that have been assessed in epidemiological studies include mortality, morbidly, and rates of medical assistance. Examples of the latter category include rates of hospital admissions or emergency room visits recorded to have been prompted by a given morbidity outcome, rates of activation of recorded activation of implantable cardioverter defibrillator (ICD) devices, and rates of rescue medication use by asthmatics. These studies are typically conducted using time-series analysis and Poisson regression models for estimated associations between carbon monoxide concentrations and outcome rates. Case-crossover designs have also been widely applied in air pollution epidemiology. This design offers many of the advantages of the case-control design (e.g., matching of individual cases and controls for potential covariables and confounders), with the additional feature of each case serving as its own control, with each case assigned a case exposure period, usually within a few days of the outcome measurement, and a control exposure period, usually some time before or after the case exposure period. Another widely reported design in assessment of morbidity is the panel study in which members of the study cohort, usually a relatively small sample, are followed individually with respect to exposures and outcomes. Also reported are case-control studies, retrospective and cross-sectional cohort studies, and a few prospective studies of mortality.

Several important features of the epidemiological studies of ambient air carbon monoxide are of generic relevance to public health assessments. In particular, their utility for establishing dose-response relationships is limited by several factors:

(1)

Nearly all studies have relied on area monitoring for estimating exposure levels; this has the potential to produce errors in assigning exposure levels to individuals.

(2)

The temporal correspondence between monitored exposure levels and outcomes is highly uncertain. Outcomes that might be related to carbon monoxide exposure might occur in response to short-term elevations in carbon monoxide concentrations not captured in air monitoring data that are averaged over longer time periods (e.g., 24-hour average). Also, outcomes associated with a given exposure may have a latency period (e.g., myocardial infarction may occur days after the exposure that initiated the event). One approach to this problem has been to apply time-series techniques to quantify air concentration trends and to explore lag times between exposure concentration trends and the time of outcome.

(3)

Ambient air carbon monoxide concentrations tend to be strongly correlated with other air quality variables that can affect cardiovascular function, either directly or indirectly (e.g., through effects on the respiratory tract). Typical variables that correlate with air carbon monoxide concentrations include particulate matter (e.g., total suspended particulates, PM10, PM2.5), NO2, O3, and SO2. Ignoring co-variables that contribute to risk in regression models (e.g., single-pollutant models) can introduce upward bias in observed associations between air carbon monoxide concentrations and outcomes (e.g., inflate relative risk). On the other hand, assuming that strongly correlated co-variables are independent in multi-pollutant models can introduce downward bias in the carbon monoxide association. Although the strength of the carbon monoxide association may be underestimated in multi-pollutant models, for the reasons noted above, persistence of significant associations between the measured outcome and air carbon monoxide concentrations, when co-pollutants are accounted for in multivariate models, provides stronger support for a contribution of carbon monoxide to the outcome. Not all studies have explored multi-pollutant models; however, such models lend supporting evidence to the results of clinical studies that have shown carbon monoxide to exacerbate coronary vascular disease.

(4)

Outcomes investigated have tended to be serious, life-threatening events (e.g., cardiac arrhythmia, myocardial infarction, stroke, heart failure); these outcomes reflect the “late-stage” consequences of contributing pathophysiology that might be associated with exposures to lower levels of carbon monoxide for longer durations than indicated in the epidemiological studies.

(5)

Mean values of carbon monoxide exposure concentrations have ranged from 0.3 to 5 ppm, with the highest values ≤30 ppm. These values correspond to predicted steady-state COHb levels of <1% for the mean and <5% for the highest values (predicted from the CFK model). At these COHb levels, it is possible that only highly susceptible individuals exhibit the serious hypoxia-related outcomes. For these and other reasons, epidemiologic studies have typically focused on highly susceptible populations. A typical design has been to examine associations between temporal trends in ambient air carbon monoxide concentrations and hospital admissions or emergency department visits for which the reported diagnosis was some form of cardiovascular and/or respiratory disease of impairment.

3.2.1. Death

The Centers of Disease Control and Prevention (CDC) estimated that during the period 1999–2004, carbon monoxide was listed as a contributing cause of death on 16,447 death certificates in the United States, of which 2,631 deaths were categorized as unintentional and unrelated to fires (CDC 2007). Typically, fatal exposures to carbon monoxide produce coma, convulsions, and cardiorespiratory arrest (Raub et al. 2000; Wolf et al. 2008). Symptoms preceding death can include headache, dizziness, weakness, nausea, vomiting, mental confusion, visual disturbances, and loss of consciousness (Choi 2001; Raub et al. 2000). Clinical signs of life-threatening toxicity can include cardiac arrhythmia and myocardial ischemia, hypotension, pulmonary edema, and seizures (Kao and Nañagas 2006). Patients who have been resuscitated following carbon monoxide-induced cardiac arrest have a very low prognosis for survival (Hampson and Zmaeff 2001).

Although severe carbon monoxide toxicity is thought to primarily derive from hypoxia resulting from binding to, and displacement of, O2 from heme proteins, including blood Hb, the relationship between blood COHb levels and signs indicative of life-threatening toxicity is highly uncertain (e.g., loss of consciousness, convulsions, coma, cardiopulmonary depression). In a survey of 1,407 cases of acute carbon monoxide poisoning patients (i.e., record of history of carbon monoxide exposure, signs and symptoms, blood COHb >2%, treatment with hyperbaric O2), the mean blood COHb concentration was 22.3% (95% confidence interval [CI]: 22.7, 23.9, range: 2.1–72.3; Hampson and Hauff 2008). The group mean was only marginally lower than in patients who lost consciousness (24.3%, 95% CI: 23.3, 25.1, range: 2.1–73.3) or died (32.1%, 95% CI: 27.9, 36.4, range: 3.0–60.0). Although survivors had lower mean COHb levels (23.1, 95% CI: 22.5, 23.6) than that of fatalities, the upper end of the range exceeded that of fatalities (2.1–72.3%). The lower end of the range for COHb levels (2.1%) in this study reflects, at least in part, time delays between termination of exposure and measurement of blood COHb levels. However, that notwithstanding, this study suggests that blood COHb levels do not provide a reliable predictor of lethality. COHb levels can be converted to estimates of corresponding continuous air exposure levels by application of the CFK model. The mean for fatalities (32.1%) corresponds to a steady-state exposure (e.g., >500 minutes) to 300 ppm, or exposure to 1,000 ppm for approximately 80– 90 minutes.

Epidemiological Studies. Epidemiological studies have examined possible associations between ambient air carbon monoxide concentrations and mortality. These studies have examined relatively low carbon monoxide concentrations (from a toxicological perspective) that are typical of ambient conditions of the study period. Mean air carbon monoxide concentrations in most of the studies were <2 ppm, with highest values ≤10 ppm. Epidemiologic studies of ambient carbon monoxide and mortality can be grouped into two categories: (1) studies that have examined the relationship between exposures to carbon monoxide estimated from short-term (e.g., daily or monthly average) ambient air carbon monoxide concentrations; and (2) studies that have based exposure estimates on long-term average air carbon monoxide concentrations. Although both types of studies have yielded mixed results, the larger, multi-city studies of long-term exposure have yielded estimates of carbon monoxide-associated mortality risk that are zero or less than zero (Jerrett et al. 2003; Lipfert et al. 2006b; Miller et al. 2007; Pope et al. 2002). Results from short-term exposure studies have been more varied, with some studies showing increased mortality risk when carbon monoxide is considered in single-pollutant models (percent increase in mortality: 0.2–0.6), which did not persist when the models were adjusted for co-pollutants (Burnett et al. 2004; Dominici et al. 2003a, 2003b; Samoli et al. 2007). This conclusion is supported by at least one meta-analysis of time-series published between 1985 and 2001 (Stieb et al. 2002), which estimated excess mortality risk (pooled meta estimate) for a 1.1 ppm increment in air carbon monoxide concentration to be 1.6 (95% CI: 1.1, 2.1) based on a single-pollutant model and 0.7% (95% CI: −0.1, 1.5), based on a multi-pollutant model that included NO2, O3, PM10, and SO2 (both estimates from generalized additive model regression models). A study of survival among 1,073 carbon monoxide poisoning patients (followed from 1978 to 2005) found that survivors of carbon monoxide poisoning had an increased mortality (standardized mortality ratio [SMR]=1.9) (95% CI: 1.6, 2.2; NIOSH Life Table analysis as reference; Hampson et al. 2009; Leikin and Wills 2009).

While these studies provide some evidence in support of a mortality risk associated with inhalation exposures to carbon monoxide, as previously discussed in the introduction to Section 3.2, their utility for establishing dose-response relationships for these effects are limited by several factors: (1) reliance on area monitoring for estimating exposure levels; and (2) relatively strong correlations between ambient air carbon monoxide concentrations and other air quality variables that may contribute to mortality risk. The discussion presented below focuses on outcomes of the most recent follow-ups of larger (e.g., multi-city) studies. Numerous single-city studies have also been reported; however, results of these studies are, in general, consistent with findings of multi-city studies.

Three multi-city prospective mortality studies have examined possible associations between long-term average ambient air carbon monoxide concentrations and mortality. A prospective study examined mortality in a cohort of 552,138 adults in 151 U.S. metropolitan areas, enrolled in the study in 1982 and followed through 1998 (Pope et al. 1995, 2002). Based on estimated average ambient air carbon monoxide concentration during the period 1982–1998, relative risks for mortality per 1 ppm increase in carbon monoxide concentration (estimated from Figure 5 of Pope et al. 2002) were approximately 0.97 (95% CI: 0.93, 1.0) for all causes of death, 0.95 (95% CI: 0.88, 0.99) for cardiopulmonary death, and 0.90 (95% CI: 0.83, 0.96) for lung cancer death. A reanalysis of data from the same study included a more extensive treatment of covariables. The relative risk for death from all causes was estimated to be approximately 0.98 (95% CI: 0.92, 1.03) for an increase in ambient carbon monoxide concentration from 0.19 to 3.95 ppm (Jerrett et al. 2003). A smaller prospective study focused on cardiovascular deaths among postmenopausal women between the ages of 50 and 79 years (Miller et al. 2007). Women (n=65,893) without previous history of cardiovascular disease were enrolled in the study from 36 U.S. metropolitan areas during the period 1994–1998, and deaths were recorded up to mid-2003. Among the subset of the cohort that had complete data on ambient air pollutant levels from local monitoring stations (n=28,402), the hazard ratio for cardiovascular mortality was 0.92 (95% CI: 0.71, 1.21) per 1 ppm increase in ambient air carbon monoxide concentration based on a single-pollutant model and 0.93 (95% CI: 0.67, 1.30) based on a multi-pollutant model that adjusted for air concentrations of NO2, O3, PM2.5, PM10, and SO2. A prospective study of U.S. military veterans examined mortality outcomes in approximately 70,000 males who had been diagnosed with hypertension (Lipfert et al. 2000, 2006a, 2006b). Subjects were enrolled in the study in 1976 at the average age of 51 years and were followed through 2001 (Lipfert et al. 2006a). Mortality risk (all causes of death) was not significantly associated with ambient air carbon monoxide concentration. The relative risk was estimated to be 1.032 (95% CI: 0.954, 1.117) per 1 ppm increase in air carbon monoxide concentration based on a single-pollutant model and 1.023 (95% CI: 0.939, 1.115) after adjustment for NO2 and O3.

A multi-city, cross-sectional study examined mortality in all U.S. counties (excluding those of Alaska) for the period 1969–1997 (Lipfert and Morris 2002). Risk estimates were made for various exposure periods ranging from 2 to 9 years and for age categories. Estimates of mortality risk (all causes) attributable to average ambient air carbon monoxide concentrations for the exposure periods were mostly either not significant or significantly negative. Elevated risk was observed in some exposure periods for the age group 15–44 years; however, this outcome was not consistently observed across time periods.

Several multi-city studies have found significant associations between increasing short-term average ambient air carbon monoxide concentration and increasing mortality risk when carbon monoxide is considered in single-pollutant models, with the associations attenuated when models were adjusted for other co-pollutants. A time-series study examined mortality in 82 U.S. cities during the period 1987–1994 (Dominici et al. 2003b; HEI 2005; Samet et al. 2000). In the Dominici et al. (2003b) analysis of these data, a 1 ppm increase in ambient air carbon monoxide concentration (lag 1 day) was associated with increased mortality (from all causes) of 0.46% (95% CI: 0.18, 0.73). Mortality risk was not significant after adjustment for air concentration of PM10 alone or for PM10 and NO2. A time-series study examined morality in 12 Canadian cities during the period 1987–1994 (Burnett et al. 2004). Mortality risk (all non-accidental causes) was significantly associated with ambient air carbon monoxide concentration (lag 1 day) in single-pollutant models, with an increase of 0.68% (t=3.12) per 1.02 ppm increases in air carbon monoxide concentration. However, the association did not persist when adjusted for NO2 (0.07%, t=0.30). A time-series analysis examined mortality in 19 European cities during the period 1990–1997 (Samoli et al. 2007). This study found a significant association between carbon monoxide and total non-accidental and cardiovascular mortality in single-pollutant models. The estimated effect size varied depending on the model applied. The effect size for total mortality based on a single-pollutant model ranged from 0.59% increase (95% CI: 0.41, 1.79) to 1.20% (95% CI: 0.63, 1.77) per 1 mg/m3 increase in carbon monoxide concentration (0–1-day lag). The effect on cardiovascular mortality ranged from 0.8% increase (95% CI: 0.53, 1.07) to 1.25 (95% CI: 0.30, 2.21). Inclusion of black smoke or NO2 in the models substantially decreased the estimated risk attributable to carbon monoxide, with the effect sizes no longer significant in some models.

3.2.2. Systemic Effects

Respiratory Effects. Although cardiopulmonary arrest is an end point of fatal carbon monoxide poisoning, results of controlled clinical studies in healthy subjects indicate that the respiratory tract does not appear to be a primary target organ for carbon monoxide toxicity. Brief exposure to carbon monoxide at levels >1,000 ppm may decrease ventilatory performance, although conflicting results have been reported (Chevalier et al. 1966; Fisher et al. 1969; Koike et al. 1991; Ren et al. 2001; Vesely et al. 2004). Epidemiological studies have examined possible associations between ambient air carbon monoxide concentrations and mortality. These studies have examined relatively low carbon monoxide concentrations (from a toxicological perspective) that are typical of ambient conditions of the study period (mean concentrations ranging from 0.3 to 10 ppm with the highest values ≤30 ppm). Collectively, these studies have yielded mixed results, with some studies finding significant associations between increasing ambient air carbon monoxide concentrations and respiratory outcomes (e.g., exacerbation of asthma symptoms, hospitalizations and emergency room visits related to asthma) and few studies finding associations that persist after accounting for exposures to other air pollutants that also have been shown to contribute to respiratory disease risk (e.g., NO2, O3, particulate matter, and SO2). The lack of strong evidence for associations between ambient air carbon monoxide concentrations at <30 ppm and pulmonary function is also consistent with the results of human clinical and animal studies. Studies conducted in animals provide supporting evidence that the respiratory tract does not appear to be a primary target organ for carbon monoxide (EPA 1991, 2000; Sørhaug et al. 2006).

Clinical Studies. Few controlled clinical studies have evaluated adverse respiratory effects of carbon monoxide exposure. Available clinical studies have been conducted in small numbers (i.e., 4–12) of healthy subjects under acute exposure conditions (Chevalier et al. 1966; Fisher et al. 1969; Koike et al. 1991; Ren et al. 2001; Vesely et al. 2004). No controlled clinical studies evaluating respiratory effects of intermediate- or chronic-duration exposure in healthy subjects or of any duration exposure in patients with underlying respiratory diseases were identified.

Results of controlled clinical studies in healthy subjects indicate that brief exposure to carbon monoxide may decrease ventilatory performance, although conflicting results have been reported; however, based on results of available clinical studies, the respiratory tract does not appear to be a major target organ for carbon monoxide. Exposure of 10 healthy subjects to 5,000 ppm carbon monoxide for 2–4 minutes (COHb 4%) decreased inspiratory capacity by 7.5% (p<0.05), total lung capacity by 2.1% (p<0.02), and mean resting diffusing capacity of lungs by 7.6% (p<0.05) (Chevalier et al. 1966). However, other studies have found no effects of acute carbon monoxide exposure on ventilatory function. No changes in ventilatory mechanics (vital capacity, functional residual capacity, airway conductance, lung volume, pulmonary resistance, dynamic lung compliance, diffusing capacity) were observed in four healthy subjects exposed to carbon monoxide producing blood COHb levels of 11–20% (Fisher et al. 1969). Exposure to 1,000 ppm carbon monoxide for 10–30 minutes (COHb ~10%; 10 subjects) (Ren et al. 2001) or 1,200 ppm carbon monoxide for 30–45 minutes (COHb 10.2%; 10 subjects) (Vesely et al. 2004) did not affect resting ventilation. During submaximal exercise challenge, acute exposure of 10 subjects to carbon monoxide producing blood COHb levels of 11–20% did not alter ventilation mechanics (Koike et al. 1991). The reason for these conflicting results is not apparent, although the small number of subjects evaluated in each study may have been a contributing factor.

Epidemiological Studies. Epidemiological studies of exposure to carbon monoxide and respiratory outcomes fall into two major categories: (1) studies that have assessed respiratory morbidity in association with ambient air carbon monoxide concentrations; and (2) studies that have evaluated associations between ambient air carbon monoxide concentrations and the incidence of hospital admissions and/or emergency room visits related to respiratory disease (e.g., time-series studies, case-crossover). Collectively, these studies have yielded mixed results. Although some studies found significant associations between increasing ambient air carbon monoxide concentrations and respiratory outcomes (e.g., exacerbation of asthma symptoms, hospitalizations and emergency room visits related to asthma), few studies examined the robustness of the association in models that adjust for exposures to other air pollutants that also have been shown to contribute to respiratory disease risk (e.g., NO2, O3, particulate matter, and SO2). While these studies provide some evidence in support of adverse respiratory effects of inhalation exposures to carbon monoxide, as previously discussed in the introduction to Section 3.2, their utility for establishing dose-response relationships for these effects are limited by several factors: (1) reliance on area monitoring for estimating exposure levels; (2) uncertainty in knowledge of temporal correspondence between monitored exposure levels and outcomes; (3) relatively strong correlations between ambient air carbon monoxide concentrations and other air quality variables that can affect respiratory function; and (4) relatively low carbon monoxide exposures studied. Mean values of carbon monoxide exposure concentrations have ranged from 0.3 to 10 ppm, with the highest values ≤30 ppm. These concentrations are well below those explored in experimental studies that found minimal or no effects of carbon monoxide on pulmonary function in association with acute exposures to carbon monoxide (i.e., ≥1,000 ppm). The relatively low carbon monoxide concentrations studied in epidemiological studies, together with concurrent exposures to other air pollutants that may have more pronounced effects on respiratory function, may have contributed to the mixed results of these studies and relatively weak associations with carbon monoxide that have been observed (e.g., odds ratios and relative risks very close to 1). The lack of strong evidence for associations between ambient air carbon monoxide concentrations at <30 ppm and pulmonary function is also consistent with the results of human clinical and animal studies.

The presentation of epidemiology is organized by major categories of respiratory outcomes for which the studies were designed to evaluate, including pulmonary function, asthma and exacerbation of asthma symptoms, and hospital admissions and emergency room visits related to respiratory disease. Study conclusions are presented in the text, with selected supporting details presented in tabular form (Table 3-3). Although most studies explored various time lags between monitored air carbon monoxide concentrations and outcomes, as well as various sample strata, for the sake of brevity, only selected representative time lags (usually, those indicative of the strongest associations to carbon monoxide) are presented in the table. Where co-pollutant models have been explored, these results are also presented.

Table 3-3. Selected Epidemiological Studies of Associations Between Ambient Carbon Monoxide Concentrations and Respiratory Disease.

Table 3-3

Selected Epidemiological Studies of Associations Between Ambient Carbon Monoxide Concentrations and Respiratory Disease.

Studies of Pulmonary Function. Studies of possible associations between inhalation exposures to carbon monoxide and effects on pulmonary function have yielded mixed results (Chen et al. 1999; Fischer et al. 2002; Lagorio et al. 2006; Penttinen et al. 2001; Rabinovitch et al. 2004; Silkoff et al. 2005; Timonen et al. 2002). Several of these studies examined subjects who had ongoing lung disease (e.g., asthma, chronic obstructive lung disease) and who might be more sensitive to agents that affect pulmonary function (Canova et al. 2010; Rabinovitch et al. 2004; Silkoff et al. 2005; Timonen et al. 2002). Mean ambient air carbon monoxide concentrations in these studies ranged from 0.5 to 10 ppm, with the highest values ≤30 ppm (Lagorio et al. 2006). Studies that explored multi-pollutant models found that the association with carbon monoxide persisted (Canova et al. 2010) or was substantially weakened when adjustments were made for air particle matter (Penttinen et al. 2001). Therefore, is it difficult to sort out the effects of carbon monoxide from those of other urban air pollutants that have strong correlations with air carbon monoxide concentrations and that also could have affected pulmonary function.

Studies of children have yielded mixed results that are further complicated by lack of assessment of models in which correlated concentrations of other air pollutants were included (Chen et al. 1999; Fischer et al. 2002; Rabinovitch et al. 2004; Timonen et al. 2002). A study conducted in the San Joaquin Valley, California examined lung function in a cohort of 232 asthmatic children (age 6–11 years) and found that long-term average exposures to carbon monoxide in infancy and childhood (age 0–6 years) was associated with significant decreasing forced expiratory volume in 1 second (FEV1)/forced vital capacity (FVC) and forced midexpiratory flow rate (FEF25–75)/FVC (Mortimer et al. 2008). Mean air carbon monoxide concentrations during the study period ranged from approximately 1 to 1.5 ppm, with the highest values ≤6 ppm. The effect sizes (per 1 ppm increase in air carbon monoxide concentration) were approximately 2.5% decrease in FEV1/FVC and 4.8% decrease in FEF25–75/FVC. A panel study conducted in Denver, Colorado (1999–2002) examined lung function in asthmatic children (n=41–63, age 6–12 years) and did not find a significant association between air carbon monoxide concentration and pulmonary function end points, FEV1, and peak expiratory flow (PEF) (Rabinovitch et al. 2004). The mean air carbon monoxide concentration in the study was 1.0 ppm (range: 0.3–3.5 ppm). A similar panel study conducted in Kuopio, Finland examined lung function in children (n=33, age 7–12 years) who were diagnosed with chronic respiratory disease symptoms (Timonen et al. 2002). Increasing air carbon monoxide concentration was associated with decreasing FVC and FEV1, when measured at rest, but not during exercise. Other air pollutants were also associated with decrements in baseline pulmonary function (particulate matter, NO2); however, multi-pollutant models were not explored. The air carbon monoxide concentrations in this study were similar to that in the Rabinovitch et al. (2004) study (mean 0.6 mg/m3 [0.5 ppm], range: 0.1–2.8 mg/m3 [0.09–2.4 ppm]). A cross-sectional cohort study conducted in Taiwan (1995–1996) included asthmatic children (n=941, age 8–13 years; Chen et al. 1999). A significant association was found between increasing air carbon monoxide concentration and decreasing FVC; however, multi-pollutant models were not explored. Peak daily air carbon monoxide concentrations ranged from 0.4 to 3.6 ppm. Fischer et al. (2002) examined possible associations between carbon monoxide and pulmonary function in a panel of children (n=68, age 10–11 years) in The Netherlands. Increasing air carbon monoxide concentration was not associated with FVC, FEV1, PEF, or maximal midexpiratory flow (MMEF); however, it was associated with increasing levels of expired endogenous nitric oxide radical (NO•). Here again, multi-pollutant models were not explored. The mean air carbon monoxide concentration in this study was 0.9 mg/m3 (0.8 ppm) (range: 0.3–1.5 mg/m3 [0.3–1.3 ppm]).

Results of studies of pulmonary function in adults have also been mixed and conflicting (Canova et al. 2010; Lagorio et al. 2006; Penttinen et al. 2001; Silkoff et al. 2005). A panel study of adult asthmatics (n=40, mean age 39 years) conducted in Padua, Italy (2004–2005) followed 19 subjects for 2 years. Increasing air carbon monoxide concentration was associated with decreasing morning and evening PEF, and the association persisted when adjusted for co-exposure to SO2, NO2, and particulate matter (Canova et al. 2010). Another panel study of adult nonsmoking asthmatics (n=57, mean age 53 years) conducted in Helsinki, Finland (1996–1997) found that increasing air carbon monoxide concentration was associated with decreasing PEF; however, the association was not significant when adjusted for particulate matter (Penttinen et al. 2001). The mean air carbon monoxide concentration was 0.4 mg/m3 (0.3 ppm) (range: 0.1–1.1 mg/m3 [0.09–0.96 ppm]). A panel study of adult former smokers, diagnosed with chronic obstructive pulmonary disease (COPD), conducted in Denver, Colorado (1999–2001, n=16–18, age ≥40 years) found somewhat conflicting results. A significant association between increasing air carbon monoxide concentration and decreasing FEV1 was found in the second winter season of the study, but not in the first winter season of the study (Silkoff et al. 2005). Air carbon monoxide concentrations were not significantly different in the two seasons (mean: 1.1–1.2 ppm; ranges: 0.3–3.8 and 0.4–2.8 ppm), although concentrations of other air pollutants (e.g., PM10, PM2.5, and NO2) were higher in the second winter season, suggesting that other air pollutants may have contributed to the associations attributed to carbon monoxide in single-pollutant models. The study conducted by Lagorio et al. (2006) is notable because it included relatively high air carbon monoxide concentrations (mean range: 1.8 [spring]–10.6 [winter] ppm; range: 1.5–25.1 ppm). This panel study examined adults (n=29, age 18–80 years) in Rome, Italy (1999) who were diagnosed with asthma, COPD, or ischemic heart disease. Increasing air carbon monoxide concentration was not significantly associated with lung function parameters, and multi-pollutant models were not explored. A panel study conducted in Incheon, Korea that examined lung function in asthmatics (n=57, ages 16–75 years) found a significant association between air carbon monoxide concentrations and decreasing PEF; however, multi-pollutant models were not explored (Park et al. 2005a). The mean air carbon monoxide concentration was 0.6 ppm.

Studies of Asthma Incidence (or Prevalence) and Exacerbation of Asthma Symptoms. A birth cohort case-control study (n=3,248 cases, 16,240 controls) conducted in British Columbia, Canada, examined associations between exposures to carbon monoxide in utero or in the first year after birth and diagnoses of asthma up to ages 5 years (Clark et al. 2010). The odds ratio for asthma diagnosis was 1.07 (95% CI: 1.04, 1.11) for in utero exposure and 1.10 (95% CI: 1.06, 1.13) for exposure during the first year, both expressed per 0.1 mg/m3 (0.09 ppm) increase in carbon monoxide concentration. A cross-sectional study examined asthma diagnoses among school-age children (n=32,672, age 6–15 years) in Taiwan (Hwang et al. 2005). The odds ratio for 0.1 ppm increases in air carbon monoxide concentration (annual average) was estimated to be 1.045 (95% CI: 1.017, 1.074) and persisted after adjustment for O3, PM10, and/or SO2 in multi-pollutant models. The mean air carbon monoxide concentration was 0.6 ppm (range: 0.4–0.9 ppm). Several studies have examined possible associations between ambient air carbon monoxide concentrations and asthma symptoms (e.g., coughing, wheezing, chest tightness, shortness of breath, inhaler use). Results of these studies have been mixed; however, collectively, these studies provide evidence for associations between increasing air carbon monoxide concentration and increasing severity of asthma (Park et al. 2005a; Rabinovitch et al. 2004; Rodriguez et al. 2007; Schildcrout et al. 2006; Silkoff et al. 2005; Slaughter et al. 2003; von Klot et al. 2002; Yu et al. 2000). One of the larger studies examined asthma symptoms in a panel of 990 children (ages 5–12 years) in eight North American cities (1993–1995, Schildcrout et al. 2006). A 1 ppm increase in air carbon monoxide concentration (lag=0 days) was associated with an odds ratio of 1.08 (95% CI: 1.01, 1.14) for asthma symptoms. The association remained significant after adjusting for NO2, PM10, or SO2 in two-pollutant models. Air carbon monoxide concentrations ranged from approximately 0.4 to 2.4 ppm (10th–90th percentile range). Significant associations with increasing carbon monoxide concentration were also found when rescue medication use was evaluated as the asthma symptom end point.

The above results are consistent with those of several smaller panel studies. A study conducted in Seattle, Washington examined asthma symptoms (Yu et al. 2000) and asthma severity (Slaughter et al. 2003) in a panel of 133 asthmatic children (age 5–13 years). The mean air carbon monoxide concentration was 1.6 ppm (range: 0.6–4.8 ppm; Yu et al. 2000). Both studies found significant associations between ambient air carbon monoxide and asthma symptoms. In the Yu et al. (2000) study, a 1 ppm increase in air carbon monoxide (lag=2 days) was associated with an odds ratio of 1.26 (95% CI: 1.09, 1.46) for asthma symptoms. In the Slaughter et al. (2003) study, a 0.67 ppm increase in air carbon monoxide concentration was associated with an odds ratio of 1.21 (95% CI: 1.08, 1.35) for an increase in severity of asthma attack. A significant association with increasing carbon monoxide concentration was also found when rescue medication use was evaluated as the asthma severity end point (Slaughter et al. 2003). Rabinovitch et al. (2004) also found a significant association between increasing air carbon monoxide concentrations and increasing rescue medication use in a study of children in Denver, Colorado (1999–2002, n=41–63, age 6–12 years). A panel study of 263 children (age 0–5 years) in Perth, Australia, who were not diagnosed with asthma, also found a significant association between air carbon monoxide concentration and wheezing/chest rattle (Rodriguez et al. 2007).

Results of studies conducted in older asthmatic subjects have been mixed (Park et al. 2005a; Silkoff et al. 2005; von Klot et al. 2002). In a panel of 53 asthmatic adults (age 37–77 years) in Erfurt, Germany (1996–1997), a 0.6 mg/m3 (0.5 ppm) increase in air carbon monoxide concentration (5-day average) was associated with an odds ratio of 1.13 (95% CI: 1.05, 1.22) for prevalence of wheezing; however, the association was attenuated when adjusted for ultrafine particle number concentration in air. Two other panel studies of asthmatics, one conducted in Incheon, Korea (2002, n=54, age 16–76 years; Park et al. 2005a) and one in Denver, Colorado (n=16 or 18, age ≥40 years, Silkoff et al. 2005), did not find significant associations between air carbon monoxide concentrations and asthma symptoms.

Studies of Hospital Admissions and Emergency Room Visits Related to Respiratory Symptoms. A large number of studies have examined associations between ambient carbon monoxide concentrations and rates of hospital admissions (e.g., Burnett et al. 2001; Cakmak et al. 2006; Karr et al. 2006; Lin et al. 2003, 2005; Linn et al. 2000; Slaughter et al. 2005; Yang et al. 2003; Zanobetti and Schwartz 2006) or emergency room visits related to respiratory disease (e.g., Arbex et al. 2009; Peel et al. 2005; Slaughter et al. 2005; Tolbert et al. 2007). Collectively, these studies provide evidence for associations between increasing carbon monoxide concentrations and increasing risk of respiratory symptoms that trigger the need for medical assistance. However, only a few studies have examined associations in multi-pollution models; therefore, relative contributions of other air pollutants to the reported outcomes are highly uncertain.

In studies that did evaluate multi-pollutant models, the associations between carbon monoxide concentrations and outcomes persisted in some studies (Burnett et al. 2001; Lin et al. 2003; Yang et al. 2003), but not in others (Lin et al. 2005; Peel et al. 2005; Tolbert et al. 2007). In a time-series analysis conducted in Toronto, Canada (1980–1994), Burnett et al. (2001) found a significant association between increasing air carbon monoxide concentrations and hospital admissions for asthma, acute bronchitis/bronchiolitis, croup, and pneumonia in children (age <2 years). A 1.9 ppm increase in air carbon monoxide concentration (lag: 0–1 day) was associated with a 19.2% increase in hospital admissions, which remained elevated when adjusted for air O3 concentration. A case-crossover study of hospital admissions of children (ages 6–12 years) for asthma in Toronto, Canada (1981–1993) found that a 0.5 ppm increase in air carbon monoxide concentration was associated with an odds ratio (lag: 1 day) of 1.05 (95% CI: 1.00, 1.11) in males but not in females (odds ratio: 1.00, 95% CI: 0.93, 1.06) and persisted in males after adjustment for particulate matter (Lin et al. 2003). Another case-crossover study directed at respiratory infections in children (age <15 years, males), also conducted in Toronto, Canada (1988–2001), estimated the odds ratios for a 0.44 ppm increase in air carbon monoxide concentration (lag: 0–5 days) to be 1.13 (95% CI: 1.01, 1.25) and 1.08 (95% CI: 0.97, 1.20) after adjustment for PM10 and PM10–2.5 (Lin et al. 2005), respectively. Odds ratios for females were not significant. Yang et al. (2003) conducted a case-crossover study of hospital admissions for various respiratory disease categories in Vancouver, Canada (1986–1998) and found that an increase in carbon monoxide concentration of 0.5 ppm was associated with odds ratios of 1.04 (95% CI: 1.01, 1.07) for children (age <3 years) and 1.02 (95% CI: 1.00, 1.04) for elderly adults (age ≥65 years). The associations persisted when adjusted for air O3 concentration. In a time-series analysis of >400,000 emergency room visits for respiratory disease in Atlanta, Georgia (1993–2000), the estimated risk ratio for a 1 ppm increase in air carbon monoxide concentration (lag: 0–13 days) was 1.076 (95% CI: 1,047, 1.105) for all ages and 1.019 (95% CI: 1.004, 1.035) for asthma visits in children aged 2–18 years (Peel et al. 2005). The associations were not significant when adjusted for other air pollutants in two-pollutant models (e.g., PM10, NO2, or O3). This study also found a significant risk ratio for emergency room visits related to pneumonia (risk ratio: 1.045, 95% CI: 1.011, 1.080; lag: 0–13 days). A revisit to the Atlanta, Georgia study, with the study period expended to 1993–2004 (Tolbert et al. 2007), found essentially similar results as reported in Peel et al. (2005). The risk ratio for all emergency room visits for all respiratory disease categories was 1.016 (95% CI: 1.009, 1.022) per 1.22 ppm increase in 1-hour maximum carbon monoxide concentration (lag: 3 days); however, the association was attenuated in models that adjusted for O3, NO2, or PM10.

Studies that examined COPD admissions and emergency room visits yielded mixed results (Arbex et al. 2009; Peel et al. 2005; Slaughter et al. 2005; Yang et al. 2003). In the Yang et al. (2003) time-series study, conducted in Vancouver, Canada (1994–1998), the risk ratio for a 0.5 ppm increase in air carbon monoxide concentration (lag: 0–6 days) was 1.14 (95% CI: 1.03, 1.23) for COPD hospital admissions. A study conducted in São Paulo, Brazil found a 12.9% (95% CI: 1.3, 25.9) increase in COPD-related emergency room visits in association with a 1.9 ppm increase in air carbon monoxide concentration (lag: 0–6 days) (Arbex et al. 2009). On the other hand, significant associations between air carbon monoxide concentration and COPD hospital admissions or emergency room visits were not found in a time-series study conducted in Spokane, Washington (1995–2001; Slaughter et al. 2005). In the Peel et al. (2005) time-series study conducted in Atlanta, Georgia (1993–2000), relative risks for emergency room visits for COPD per 1 ppm increase in carbon monoxide concentration were 1.03 (95% CI: 1.00, 1.05) for lag 0–2 days and 1.03 (95% CI: 0.98, 1.09) for lag 0–13 days. The air carbon monoxide concentrations in these studies were similar (10th–90th percentile range: approximately 1.25–4.3 ppm)

Animal Studies. Most studies evaluating adverse respiratory effects of carbon monoxide in animals were conducted using carbon monoxide exposures that produced much higher COHb concentrations (i.e., COHb >50%) than those evaluated in controlled clinical studies in humans. Even at these high exposures, results of animal studies support those of controlled clinical studies indicating that the respiratory tract is not a major target organ for carbon monoxide. Animal studies published prior to 2000 have been reviewed in detail by the EPA (1991, 2000); therefore, these reviews largely form the basis of the following discussions for animals studies published prior to 2000.

Studies investigating adverse respiratory effects of carbon monoxide have evaluated the lung function and morphology, with most studies evaluating effects of brief exposure (i.e., a few minutes to <1 hour). In acute-duration studies, alterations in lung function (decreased compliance and increased airway resistance) were observed in rabbits and cats exposed to 8,000–28,400 ppm carbon monoxide for <1 hour (COHb >60%); however, no changes in lung function were observed in dogs exposed to 8,000–14,000 ppm carbon monoxide for 10–20 minutes (COHb 18–59%) (EPA 1991). In an intermediate-duration study, continuous exposure of rats to 50 ppm carbon monoxide for 21 days (COHb not reported) did not alter the pulmonary pressure-flow relationship (Carraway et al. 2002). Blood COHb concentrations were not reported for the Carraway et al. (2002) study; however, the CFK rat model predicts that COHb concentrations at the end of the 21-day exposure period would have been approximately 7%. Results of these studies suggest that carbon monoxide-induced changes in pulmonary function occur only following exposure to high carbon monoxide concentrations.

Studies evaluating effects of acute-duration carbon monoxide exposure on lung morphology have reported conflicting results (EPA 1991). Minimal changes in type 2 epithelial cells (fragmentation of lamellar bodies) and nonciliated bronchiolar cells (dilation of smooth endoplasmic reticulum and increased mitochondrial cristae) were reported in mice exposed to 50–90 ppm carbon monoxide for 1–5 days (COHb <10%) (Niden 1971). The study results were not presented in detail and the toxicological significance of these finding is not known; however, changes to the smooth endoplasmic reticulum and mitochondrial cristae are consistent with a compensatory response to altered electron transport due to possible carbon monoxide-induced interference with cytochromes involved in electron transport. In rabbits exposed to 8,000 ppm for 30 minutes (COHb 63%), examination of lung tissue by electron microscopy showed alveolar epithelial edema with detachment of the endothelium from the basement membrane (Fein et al. 1980). However, no morphological changes in lung tissue were observed following acute exposure (30 minute to 5 hours) of dogs and rabbits at carbon monoxide exposure producing blood COHb levels of 18–39% or following intermediate-duration exposure (6 weeks) of rabbits to 200 ppm carbon monoxide (COHb 11.9–19%) (Fisher et al. 1969; Hugod 1980). Results of a recent study in rats exposed chronically to 200 ppm carbon monoxide (COHb 14.7%) for 20 hours/day for 72 weeks showed no histopathological changes to lung tissue (Sørhaug et al. 2006).

Cardiovascular Effects. Cardiovascular effects of inhalation exposures to carbon monoxide have been evaluated in controlled human clinical studies, epidemiology studies, and various animal models (monkeys, dogs, rats, and rabbits). In general, these studies provide convincing evidence for adverse cardiovascular effects in association with carbon monoxide exposures that result in blood COHb levels ≥2.4%, with effects occurring at the lowest levels in subjects with compromised cardiovascular function (e.g., coronary artery disease). Results of controlled clinical studies in patients with coronary artery disease show that acute-duration exposure to carbon monoxide at levels producing blood COHb levels between 2.4 and 5.9% exacerbates underlying cardiovascular disease, including enhancing myocardial ischemia and increasing cardiac arrhythmias.

In healthy subjects, acute exposure to carbon monoxide at concentrations producing blood COHb levels of 3.35 and 20.5% resulted in compensatory cardiovascular responses and decreased exercise performance consistent with COHb-induced decreased O2 carrying capacity of the blood; however, subjects did not exhibit exercise-induced adverse cardiovascular effects (e.g., evidence of myocardial ischemia or cardiac arrhythmias). Continuous exposure of healthy subjects, which included smokers, to carbon monoxide resulting in blood COHb levels of 2.4 and 5.1% produced several P-wave deviations (P-wave inversion, increased amplitude, decreased amplitude, and changes in P-wave direction) under resting conditions, although similar findings have not been reported in other studies.

Epidemiological studies of exposure to carbon monoxide and cardiovascular outcomes have yielded mixed results, although, in general, the weight of evidence suggests that risks of certain specific outcomes are associated with increasing ambient carbon monoxide concentrations (hospitalizations and emergency room visits related to congestive heart failure, ischemic heart disease, myocardial infarction, and stroke). The corroborated observations of associations between carbon monoxide exposure and outcomes related to ischemic heart disease are particularly provocative in the context of results of human clinical studies in which carbon monoxide-induced hypoxia exacerbated ischemia symptoms in patients with coronary artery disease. Mean ambient air carbon monoxide concentrations reported in studies that have found carbon monoxide-associated cardiovascular outcomes ranged from 0.5 to 10 ppm, with maximum values ranging from 2 to 50 ppm. These values correspond to approximate steady-state blood COHb levels of <2% for the mean and <10% for the maximum.

Studies in animals have investigated adverse cardiovascular effects of carbon monoxide exposure over a much wider range of exposure conditions (e.g., exposure concentration and duration) and have evaluated additional outcome measures that are not possible to assess in humans. These studies provide further evidence of adverse cardiovascular effects of carbon monoxide exposure, including compensatory alterations in hemodynamics, cardiac hypertrophy, cardiac arrhythmias, and possibly atherosclerosis.

Human Clinical Studies

Studies of Patients with Cardiovascular Disease. The effects of acute-duration carbon monoxide exposure on cardiovascular function have been evaluated in several controlled clinical studies in patients with underlying cardiovascular disease (Adams et al. 1988; Allred et al. 1989, 1991; Anderson et al. 1973; Dahms et al. 1993; Hinderliter et al. 1989; Kleinman et al. 1989, 1998; Leaf and Kleinman 1996b; Sheps et al. 1987, 1990) and in healthy subjects (Adir et al. 1999; Davies and Smith 1980; Hausberg and Somers 1997; Horvath et al. 1975; Kizakevich et al. 2000; Morse et al. 2008; Resch et al. 2005; Vogel and Gleser 1972; Zevin et al. 2001). In general, similar effects of acute-duration carbon monoxide exposure were not observed in studies on healthy study subjects, due to functional cardiovascular compensatory mechanisms (e.g., vasodilation and increased cardiovascular output) that protect against COHb-induced tissue hypoxia. Mechanisms contributing to cardiac effects of carbon monoxide in patients with coronary artery disease have not been conclusively established. Under conditions of cardiac ischemia, tissue hypoxia secondary to elevated COHb levels is thought to be a contributing factor. However, other direct cellular effects of carbon monoxide on cardiac muscle may also be important. These include modulation of coronary arteriole calcium-activated potassium channels. These channels are required for membrane hyperpolarization and flow-induced dilation and are reversibly activated by carbon monoxide under normoxic conditions, but they are inhibited when hypoxic/ischemic conditions prevail (Wang and Wu 1997; Wang et al. 1997a, 1997b).

Controlled clinical studies conducted in small numbers of patients with underlying coronary artery disease have evaluated effects of carbon monoxide exposure on myocardial ischemia. Patients evaluated had a diagnosis of coronary artery disease with exertional angina (i.e., exercise-induced angina or stable angina) and/or ventricular arrhythmia. Most studies were conducted using a randomized, double-blind, crossover design, with cardiovascular function assessments conducted during exercise challenge, either during or immediately following carbon monoxide exposure. Carbon monoxide exposures (concentration and duration) were designed to reach target blood COHb levels between 2 and 6%; exposure durations ranged from 1 to 4 hours via an environmental chamber, with carbon monoxide concentrations ranging from 42 to 357 ppm. Effects of carbon monoxide exposure were evaluated by comparison to assessments conducted during room air exposure and at corresponding COHb concentrations of <1%. Additional experimental details for each study are summarized in Table 3-4.

Table 3-4. Effects of Acute-Duration Exposure to Carbon Monoxide on Cardiovascular Function in Patients with Coronary Artery Disease.

Table 3-4

Effects of Acute-Duration Exposure to Carbon Monoxide on Cardiovascular Function in Patients with Coronary Artery Disease.

Results of controlled clinical studies in patients with coronary artery disease show that acute-duration exposure to carbon monoxide at levels producing blood COHb levels between 2.4 and 5.9% exacerbates underlying cardiovascular disease, including enhanced myocardial ischemia and increased cardiac arrhythmias. In patients with exertional angina, carbon monoxide exposure exacerbated exercise-induced myocardial ischemia, including decreased time-to-onset of angina symptoms (Adams et al. 1988; Allred et al. 1989, 1991; Anderson et al. 1973; Kleinman et al. 1989, 1998; Leaf and Kleinman 1996b), increased duration of angina symptoms (Anderson et al. 1973; Kleinman et al. 1989), decreased time-to-onset of ST-segment depression (EKG change indicative of myocardial ischemia) (Allred et al. 1989, 1991; Kleinman et al. 1989), and decreased left ventricular ejection fraction (Adams et al. 1988). At the lowest blood COHb level evaluated in patients (i.e., COHb 2.4%), time-to-onset of angina symptoms and of ST-segment depression were significantly decreased by 4.2 and 5.1%, respectively (Allred et al. 1989, 1991). In contrast, no carbon monoxide-induced exacerbation of exertional angina, based on time-to-onset of angina symptoms and ST-segment depression and duration of angina symptoms, were observed at a blood COHb level of 4.1% (Sheps et al. 1987). Clinical evaluations in angina patients with ventricular arrhythmias have yielded mixed results on the effects of carbon monoxide-induced ventricular arrhythmias (Dahms et al. 1993; Hinderliter et al. 1989; Sheps et al. 1990). At a blood COHb level of 5.9%, but not 4.0%, single and complex premature ventricular depolarizations were significantly increased by 36 and 6.3%, respectively (Sheps et al. 1990). However, at this same blood COHb level, carbon monoxide exposure did not increase ventricular arrhythmias in other studies (Dahms et al. 1993; Hinderliter et al. 1989). Reasons for conflicting results in these studies are not readily apparent, although numbers of patients evaluated were small and the severity of underlying coronary artery disease may have varied in each study.

Studies of Healthy Subjects. Several controlled clinical studies in small numbers of healthy subjects have evaluated effects of acute-duration exposure to carbon monoxide on exercise performance, including cardiovascular function. In general, study protocols were similar to those used in clinical studies on patients with coronary artery disease, although typical carbon monoxide exposures were for shorter durations (a few minutes to ≤1 hour) to higher carbon monoxide concentrations (50–3,000 ppm) and produced slightly higher blood COHb levels (2.4–20.5%). However, one study exposed subjects continuously for 7 days (Davies and Smith 1980). Additional experimental details for each study are summarized in Table 3-5.

Table 3-5. Effects of Acute-Duration Carbon Monoxide Exposure on Cardiovascular Function in Healthy Subjects.

Table 3-5

Effects of Acute-Duration Carbon Monoxide Exposure on Cardiovascular Function in Healthy Subjects.

Results of several controlled clinical studies in healthy subjects show that carbon monoxide exposures that resulted in blood COHb levels between 3.35 and 20.5% decreased exercise performance, including decreased exercise duration (Adir et al. 1999); decreased maximal exercise effort or maximal aerobic capacity (Adir et al. 1999; Horvath et al. 1975; Vogel and Gleser 1972); decreased resistance to muscle fatigue (Morse et al. 2008); and induced compensatory increases in heart rate and cardiac output (Kizakevich et al. 2000; Vogel and Gleser 1972). The above effects of exercise performance are consistent with COHb-induced decreased O2 carrying capacity of the blood and normal physiological compensatory mechanisms. Evaluations on the potential for carbon monoxide exposure to produce exercise-induced cardiac arrhythmias similar to those observed in patients with coronary artery disease (i.e., ST-segment depression and premature ventricular contractions) have yielded negative results at blood COHb levels up to 5.1% (Adir et al. 1999; Horvath et al. 1975; Kizakevich et al. 2000), although the potential for carbon monoxide-induced cardiac arrhythmias during exercise challenge was not evaluated at higher blood COHb levels. In subjects (including smokers and nonsmokers) exposed continuously for 7 days and evaluated at rest, EKG showed several P-wave deviations (P-wave inversion, increased amplitude, decreased amplitude, and changes in P-wave direction) in 3/15 subjects exposed to 15 ppm carbon monoxide (COHb 2.4%) and in 6/15 subjects exposed to 50 ppm carbon monoxide (COHb 5.1%), compared to 0/14 in subjects exposed to room air (Davies and Smith 1980). In one subject, a heavy smoker, marked ST-depression also was observed. However, interpretation of these findings is limited due to lack of statistical analyses and of evaluation for potential confounding factors (e.g., smoking, medication use). Other studies evaluating effects of a single 1-hour exposure to carbon monoxide (Hausberg and Somers 1997) or repeated brief exposures to carbon monoxide (approximately once every 45 minutes to simulate smoking; Zevin et al. 2001) on cardiovascular function (blood pressure, heart rate, forearm blood flow) under resting conditions showed no effects at blood COHb levels up to 8.3%.

In conclusion, results of controlled clinical studies in patients with cardiovascular disease provide compelling evidence that acute-duration exposure to carbon monoxide resulting in COHb levels of 2.4–5.8% exacerbates cardiovascular morbidity (myocardial ischemia and cardiac arrhythmias). Effects at lower blood COHb levels have not been evaluated. However, it should be noted that patients with more severe cardiovascular disease may be more sensitive than patients evaluated in these clinical studies. In healthy subjects, acute exposure to carbon monoxide at concentrations producing blood COHb levels of 3.35 and 20.5% resulted in compensatory cardiovascular responses and decreased exercise performance consistent with COHb-induced decreased O2 carrying capacity of the blood; however, subjects did not exhibit exercise-induced adverse cardiovascular effects (e.g., evidence of myocardial ischemia or cardiac arrhythmias). Continuous exposure of healthy subjects, which included smokers, to carbon monoxide resulting in blood COHb levels of 2.4 and 5.1% produced P-wave deviations under resting conditions, although similar findings have not been reported in other studies.

Epidemiological Studies. Epidemiological studies of exposure to carbon monoxide and cardiovascular outcomes fall into two major categories: (1) studies that have assessed cardiovascular morbidity in cohorts or individuals (e.g., case-crossover, case-control) in association with ambient air carbon monoxide concentrations; and (2) studies that have evaluated associations between air carbon monoxide concentrations and the incidence of hospital admissions and/or emergency room visits related to cardiovascular disease (e.g., time-series studies, case-crossover). Collectively, these studies have yielded mixed results, although, in general, the weight of evidence suggests that risks of certain specific outcomes are associated with increasing ambient carbon monoxide concentrations (hospitalizations and emergency room visits related to congestive heart failure, ischemic heart disease, myocardial infarction, and stroke). Corroborated observations of association between carbon monoxide exposure and outcomes related to ischemic heart disease are consistent with results of human clinical studies that have shown that carbon monoxide-induced hypoxia exacerbates symptoms of coronary artery disease. Therefore, it is plausible that in certain susceptible populations, such as individuals who have ongoing cardiac impairment that would make them vulnerable to ischemia, exposure to carbon monoxide could contribute to triggering ischemic episodes and cardiac sequelae. While these studies provide some evidence in support of adverse cardiovascular effects of inhalation exposures to carbon monoxide, their utility for establishing dose-response relationships for these effects are limited by several factors: (1) reliance on area monitoring for estimating exposure levels; (2) uncertainty in knowledge of temporal correspondence between monitored exposure levels and outcomes; (3) relatively strong correlations between ambient air carbon monoxide concentrations and other air quality variables that can affect respiratory function; (4) tendency of outcomes investigated to be serious, life-threatening events (e.g., cardiac arrhythmia, myocardial, infarction, stroke, heart failure) and reflect the “late-stage” consequences of contributing pathophysiology that might be associated with exposures to lower levels of carbon monoxide for longer durations than indicated in the epidemiological studies; and (5) relatively low carbon monoxide exposures studied. Mean values of carbon monoxide exposure concentrations have ranged from 0.3 to 4.6 ppm, with the highest values ≤30 ppm. These values correspond to predicted steady-state COHb levels of <1% for the mean and <5% for the highest values (predicted from the CFK model). The latter range is consistent with COHb levels observed in a study of emergency department admissions for cardiopulmonary complaints (Leikin and Vogel 1986). At these COHb levels, it is possible that only highly susceptible individuals exhibit the serious cardiovascular outcomes that have been studied. Clinical studies described previously in this section of the Toxicological Profile have found that exposures to carbon monoxide that produce COHb levels of 2.4–5% can exacerbate underlying cardiovascular disease, including exercise-induced angina and cardiac arrhythmia; however, higher levels (3–20%) can be tolerated without producing such effects, even under exertion, in healthy individuals. For these and other reasons, epidemiologic studies have typically focused on highly susceptible populations. A typical design has been to examine associations between temporal trends in ambient air carbon monoxide concentrations and hospital admissions or emergency department visits for which the reported diagnosis was some form of cardiovascular disease of impairment.

The presentation of the epidemiology is organized by major categories of cardiovascular outcomes for which the studies were designed to evaluate, including studies of heart rate and heart rate variability, cardiac arrhythmia, ischemic heart disease and related subcategories (e.g., myocardial infarction), heart failure, stroke, and blood pressure. Study conclusions are presented in the text, with selected supporting details presented in tabular form. Although most of the studies explored various time lags between monitored air carbon monoxide concentrations and outcomes, as well as various sample strata, for the sake of brevity, only those indicative of the strongest associations to carbon monoxide are presented in the tables. Where co-pollutant models have been explored, these results are also presented.

Studies of Heart Rate (HR) and Heart Rate Variability (HRV). Studies of associations between exposure to carbon monoxide and various measures of HR and HRV have yielded mixed results (Table 3-6). Carbon monoxide exposure concentrations in these studies represent relatively narrow time periods (e.g., averages over 1–5 days prior to the end point assessment). Mean values of carbon monoxide exposure concentrations ranged from 0.3 to 4.6 ppm, with the highest values ≤30 ppm (predicted steady-state COHb%: <5). Significant associations between increasing carbon monoxide exposure and increasing HR were found in some studies (Liao et al. 2004; Peters et al. 1999) and not in others (Gold et al. 2000). Assessments of HRV found significant associations with increasing carbon monoxide exposure in some studies (Dales 2004); Riojas-Rodriguez et al. 2006; Schwartz et al. 2005; Tarkiainen et al. 2003; Timonen et al. 2006) and no associations with HRV in other studies (Chan et al. 2005; Gold et al. 2000; Holguin et al. 2003; Liao et al. 2004; Park et al. 2005b). In general, associations for HRV end points were not robust for various measures of HRV across studies. For example, assessments of standard deviation of the normal-to-normal interval (SDNN) yielded significant associations in some studies (Dales 2004); Schwartz et al. 2005; Timonen et al. 2006) and not in others (Chan et al. 2005; Gold et al. 2000; Liao et al. 2004; Park et al. 2005b; Tarkiainen et al. 2003).

Table 3-6. Selected Epidemiological Studies of Associations Between Ambient Carbon Monoxide Concentrations and Cardiovascular Disease.

Table 3-6

Selected Epidemiological Studies of Associations Between Ambient Carbon Monoxide Concentrations and Cardiovascular Disease.

Three studies are particularly notable for their relatively large size.Park et al. (2005b) assessed HRV in 497 male subjects (age range: 21–82 years) of the Boston Normative Aging Study and linked these data to local measurements of air pollutant concentrations (carbon monoxide, PM2.5, O3, NO2, SO2). The mean carbon monoxide exposure concentration was 0.50 ppm (24-hour average, range: 0.13–18 ppm). The study did not find significance between carbon monoxide exposure levels and HRV end points assessed (SDNN, low frequency [LF], high frequency [HF], LF/HF). A study conducted in Augsburg, Germany evaluated possible associations between exposures to total suspended particles (TSP), SO2, and carbon monoxide (mean: 4.5 mg/m3 [3.9 ppm], range: 2.4–6.8 mg/m3 [2.1–5.9 ppm) and HR in 2,681 adults (age range: 25–64 years; Peters et al. 1999). A 6.6 mg/m3 (5.8 ppm) increase in 24-hour average carbon monoxide concentration was associated with a small but significant increase in HR of approximately 1 beat per minute (0.97 beats/minute, 95% CI: 0.02, 1.91); however, carbon monoxide exposure was not significantly associated with HR when based on a 5-day average exposure concentration (0.7 per 3.3 mg/m3 [2.9 ppm], 95% CI: −0.09, 1.58). A subsample of 6,232 adults (age range: 45–64 years) from the Atherosclerosis Risk in Communities study (Washington County, Maryland; Forsyth County, North Carolina; and selected suburbs of Minneapolis, Minnesota and Jackson, Mississippi) examined possible associations between exposure to carbon monoxide (mean: 0.65 ppm, standard deviation [SD]: 0.44) and HR and HRV, as well as to other air pollutants (PM10, O3, NO2, SO2; Liao et al. 2004). An increase in carbon monoxide exposure concentration of 0.44 ppm was associated with a small but significant increase in HR of 0.4 beats per minute (standard error [SE]: 0.16, 95% CI: 0.09, 0.7 [calculated from SE]). Carbon monoxide exposure was not significantly associated with HRV.

Studies of Arrhythmia. Studies of associations between exposure to carbon monoxide and cardiac arrhythmia (e.g., fibrillation, tachycardia) have yielded mixed results (Table 3-6). Mean values of carbon monoxide exposure concentrations (i.e., averages over 1–5 days prior to the end point assessment) ranged from 0.2 to 1.7 ppm, with the highest values ≤8 ppm. Significant associations between increasing carbon monoxide exposure and arrhythmia episode incidence were found in some studies, particularly when the analysis was restricted to subjects with prior recent arrhythmia (Berger et al. 2006; Dockery et al. 2005; Peters et al. 2000a), but not in other studies (Metzger et al. 2007; Rich et al. 2004, 2005; Sarnat et al. 2006; Vedal et al. 2004). One of the larger studies examined associations between exposures to carbon monoxide (median: 0.8 ppm; 95th percentile: 1.37 ppm) and other air pollutants (e.g., PM10, PM2.5, O3, NO2, SO2) and ventricular arrhythmia episodes in 203 patients who were monitored with implanted defibrillators (ICDs) and who had prior episodes of ventricular tachycardia (Dockery et al. 2005; Rich et al. 2005). In the total sample, exposure to carbon monoxide was not significantly associated with ventricular arrhythmia; however, when restricted to patients who had experienced arrhythmias within the 3 days of the exposure assessment, increasing carbon monoxide exposure of 0.48 ppm was significantly associated ventricular arrhythmia (odds ratio: 1.65; 96% CI: 1.17, 2.33). A larger study examined 512 patients (age range: 15–88 years) exposed to a mean carbon monoxide concentration of 1.7 ppm (range: 0.1–7.7 ppm) and did not find a significant association with defibrillator discharges and carbon monoxide exposure concentration (Metzger et al. 2007).

Studies that have assessed arrhythmias from EKG measurements have also yielded mixed results (Berger et al. 2006; Sarnat et al. 2006). A study of 57 coronary heart disease patients (age range: 52–76 years) found a significant association between increasing carbon monoxide exposure and risk of supraventricular tachycardia (Berger et al. 2006). Mean carbon monoxide exposure concentration was 0.52 mg/m3 0.45 ppm; range: 0.11–1.93 mg/m3 [0.10–1.68 ppm]). The relative risk for a 0.27 mg/m3 (0.24 ppm) increase in carbon monoxide was 1.18 (95% CI: 1.04, 1.35). A significant association between carbon monoxide exposure and ventricular tachycardia was not evident.Sarnat et al. (2006) did not find a significant association between carbon monoxide exposure concentrations (mean: 0.2 ppm, range: 0.1–1.5) and arrhythmia in a sample of 32 adults (age range: 53–90 years). The study did find significant associations between increased blood pressure and other air pollutants (e.g., PM2.5, SO2, O3, back carbon).

Studies of Ischemic Heart Disease. Numerous studies have examined possible associations between carbon monoxide exposure and ischemic heart disease (Tables 3-6 and 3-7). Although results have been mixed, most studies, in particular those that examined hospital admissions or emergency room visits as the health outcome, have found increased risk association with increased ambient air carbon monoxide concentrations. A case-control study of myocardial infarction (1,397 cases, 1,870 controls) conducted in Sweden did not find a significant association with carbon monoxide exposure (Rosenlund et al. 2006). Exposure concentrations were estimated from emissions data and air dispersion modeling; the mean was 61.8 µg/m3 (0.067 ppm; 5th–95th percentile range: 14–296 µg/m3; predicted steady-state COHb%: <0.3). The odds ratio for myocardial infarct for a 300 µg/m3 (0.3 ppm) increase in 30-year average carbon monoxide concentration was 1.04 (95% CI: 0.89, 1.21). When restricted to out-of-hospital deaths, the odds ratio was 1.36 (95% CI: 1.01, 1.84). A smaller case-crossover study (772 cases) conducted in Boston, Massachusetts did not find a significant association between carbon monoxide exposure and myocardial infarction (Peters et al. 2001). The odds ratio for a 1 ppm increment in carbon monoxide exposure concentration, for a 2-hour period prior to infarct, was 1.22 (95% CI: 0.89, 1.67); the mean carbon monoxide exposure concentration was 1.09 (5th–95th percentile range: 0.49–1.78 ppm, predicted steady-state COHb%: ≤0.55).

Table 3-7. Selected Epidemiological Studies of Associations Between Ambient Carbon Monoxide Concentrations and Hospital Admissions and Emergency Department Visits Related to Cardiovascular Disease.

Table 3-7

Selected Epidemiological Studies of Associations Between Ambient Carbon Monoxide Concentrations and Hospital Admissions and Emergency Department Visits Related to Cardiovascular Disease.

Possible associations between carbon monoxide exposure (and other air pollutants; e.g., PM, NO2, O3) and ischemic heart disease have also been explored in studies of hospital admissions and/or hospital emergency department visits (D’Ippoliti et al. 2003; Hosseinpoor et al. 2005; Lanki et al. 2006; Lee et al. 2003b; Maheswaran et al. 2005b; Mann et al. 2002; Szyszkowicz 2007; von Klot et al. 2005). Although results have been mixed, most studies found significant associations between increasing carbon monoxide exposure concentration and ischemic heart disease-related admissions or emergency room visits (Table 3-7). These included increased odds (or relative risk) for hospital admissions for angina pectoris, myocardial infarction, and aggregated ischemic heart disease categories (D’Ippoliti et al. 2003; Hosseinpoor et al. 2005; Lanki et al. 2006; Lee et al. 2003b; Mann et al. 2002; Szyszkowicz 2007; von Klot et al. 2005). Mean air carbon monoxide concentrations in these studies ranged from approximately 0.4 to 2 ppm, with the upper percentiles (i.e., ≥75th) ranging from approximately 1 to 5 ppm, although one study reported a mean of 10.8 mg/m3 (9.40 ppm) and a maximum of 57.9 mg/m3 (approximately 50 ppm; Hosseinpoor et al. 2005). Associations with carbon monoxide tended to be stronger for more elderly males than for females or other age strata (e.g., Lee et al. 2003b; Szyszkowicz 2007). Risks were higher when ischemic heart disease was accompanied by a diagnosis of arrhythmia or congestive heart failure (Mann et al. 2002; see further discussion of carbon monoxide associations with congestive heart failure, below). Lee et al. (2003b) found a significant association (1997–1999) between ischemic heart disease-related hospital admissions in Seoul, Korea and carbon monoxide exposure concentrations among males ≥64 years of age, but not for females or other age strata. A 1 ppm increase in carbon monoxide concentration was associated with a relative risk of 1.07 (95% CI: 1.01, 1.1) for hospital admission. The mean daily 1-hour maximum carbon monoxide concentration was 1.8 ppm (25th–75th percentile range: 1.2–2.2 ppm). The association with carbon monoxide exposure was not significant after adjustment for exposure to PM10 in a two-pollutant model (1.04. 95% CI: 0.98, 1.11). Two-pollutant models with carbon monoxide and other gaseous pollutants were not reported, although significant associations were found when NO2, O3, or SO2 were considered alone. Szyszkowicz (2007) examined records of 4,979 emergency room visits in Montreal, Canada for the period 1997–2002. The percent increase in visits related to ischemic heart disease for a 0.2 ppm increase in carbon monoxide concentration was 5.4% (95% CI: 2.3, 8.5) for all patients, 7.5% (2.6–12.6) for males ≥64 years, and 2.4% (−3.0–0.0) for females ≥64 years. The mean carbon monoxide exposure concentration was 0.5 ppm (range: 0.1–3.1 ppm). Two other studies examined ischemic heart disease end points in multiple-pollutant models.Hosseinpoor et al. (2005) found that the relative risk for hospital admissions for angina pectoris (1,826 cases; Tehran, Iran; 1996–2001) was 1.044 (95% CI: 1.000, 0.089) when air carbon monoxide concentrations were considered in a carbon monoxide model and 1.0093 (95% CI: 1.0036, 1.0151) when the model was adjusted for PM10, O3, and NO2. Air carbon monoxide concentrations examined in this study were relatively high compared to other studies (mean: 10.8 ppm; range: 1.6–58 ppm). A study conducted in Rome, Italy examined records of 6,351 hospital admissions (1995–1997) for myocardial infarction and found a significant association between increasing air carbon monoxide concentration and increasing odds of hospital admission (odds ratio: 1.004, 95% CI: 1.000, 0.089), which did not persist when adjusted for total suspended solids (D’Ippoliti et al. 2003). The mean air carbon monoxide concentration was 3.8 ppm (25th–75th percentile range: 2.4–3.7 ppm).

Studies of Stroke. Possible associations between ambient air carbon monoxide concentrations and stroke have been examined in studies of hospital admissions and emergency room visits (Chan et al. 2006; Henrotin et al. 2007; Maheswaran et al. 2005b; Tsai et al. 2003b; Villeneuve et al. 2006a; Wellenius et al. 2005a; Table 3-7). Although results of these studies have been mixed, several studies have found significant associations between stroke admissions and/or emergency room visits and increasing air carbon monoxide concentrations (Chan et al. 2006; Tsai et al. 2003b; Villeneuve et al. 2006a; Wellenius et al. 2005a). Studies that explored two-pollutant models found that the associations persisted after adjustment for PM10, SO2, or O3 (Chan et al. 2006; Tsai et al. 2003b). Mean air carbon monoxide concentrations in these studies ranged from 0.4 to 1.7 ppm, and the highest reported values were ≤5 ppm. The largest study was a time-series analysis of ambient air carbon monoxide concentrations and 155,503 records of stroke-related emergency room visits in nine U.S. cities (Wellenius et al. 2005a). The study found that a 0.71 ppm increase in ambient air carbon monoxide concentration was associated with a 2.83% (95% CI: 1.23, 4.46) increase in daily rate of emergency room visits for ischemic stroke. Chan et al. (2006) conducted a time-series analysis of ambient air carbon monoxide concentrations in Taipei, Taiwan (1997–2002) and emergency room visits related to cerebrovascular disease, including stroke (n=7,341) and found a significant odds ratio for a 0.8 ppm increase in air carbon monoxide concentration for all cerebrovascular disease (1.03, 95% CI: 1.01, 1.06) and stroke (1.03, 95% CI: 1.01, 1.05), which remained significant after adjusting for O3, PM2.5 and PM10, or PM2.5 (1.034, 95% CI: 1.001, 1.067). A larger study conducted in Kaohsiung, Taiwan (1997–2000) examined records of 23,192 hospital visits for stroke in a case-crossover design and found significant odds ratios for a 0.8 ppm increase in carbon monoxide concentration (1.77, 95% CI: 1.31, 2.39), which also persisted after adjustment for SO2, O3, or PM10 (Tsai et al. 2003b).

Studies of Other Cardiovascular Outcomes. Case-crossover studies examining possible associations between ambient carbon monoxide concentrations and cardiac arrest have either not found significant associations or found associations that were relatively small in magnitude (Dennekamp et al. 2010; Levy et al. 2001; Silverman et al. 2010; Sullivan et al. 2003; Table 3-6). In two studies that were conducted in Seattle, Washington, the mean carbon monoxide exposure concentration (both studies) was 1.7 ppm, with the highest value reported as 5.9 ppm. The study reported by Levy et al. (2001) included 362 cases out-of-hospital cardiac arrests (1988–1994); the relative risk for a 1-day lag between exposure estimate and outcome was 0.88 (95% CI: 0.83, 1.18). The Sullivan et al. (2003) study examined 1,542 cases of out-of-hospital cardiac arrest cases (1985–1994); the odds ratio for a 1.02 ppm increase in carbon monoxide concentration (1-day lag) was 0.97 (95% CI: 0.87, 1.08). The Silverman et al. (2010) study examined 8,216 cases of out-of-hospital cardiac arrests in New York City (2002–2006) and did not find a significant association; the relative risk for a 0.3-ppm increment in carbon monoxide was approximately 0.99 (95% CI: 0.96, 1.03). The Dennekamp et al. (2010) study examined 8,434 cases of out-of-hospital cardiac arrests in Melbourne, Australia (2003–2006) and found a small association; the relative risk was 1.031 (95% CI: 1.002, 1.046) per 0.25-ppm increase in carbon monoxide concentration.

Possible associations between air carbon monoxide concentrations and cardiac failure (primarily, congestive heart failure) have also been examined in studies of hospital admissions (Lee et al. 2007a; Symons et al. 2006; Wellenius et al. 2005b; Table 3-6). The larger of the these studies (Lee et al. 2007a) examined 13,475 hospital admissions for congestive heart failure in Kaohsiung, Taiwan (1996–2004) and found a significant odds ratios for hospital admission associated with a 0.31 ppm increase in air carbon monoxide concentration (odds ratio: 1.39, 95% CI: 1.24, 1.54). The odds ratio remained significant in two-pollutant models that adjusted for PM10, NO2, SO2, or O3.

Studies of possible associations between carbon monoxide exposure and abnormal blood pressure have not found significant associations (Ibald-Mulli et al. 2001; Zanobetti et al. 2004b; Table 3-6). The Ibald-Mulli et al. (2001) study included 2,607 men and women (age range 25–64 years) in the Augsburg, Germany area; the mean carbon monoxide exposure concentration was approximately 3.6 ppm and the upper end of the range was 7.2 ppm. The estimated increment in systolic blood pressure for a 4.9 ppm increase in carbon monoxide concentration was 1.06 (95% CI: −0.17, 2.29). The Zanobetti et al. (2004b) study measured blood pressure in repeated examinations of 62 subjects, and found no association between blood pressure and carbon monoxide exposure concentrations (90th percentile concentration <1.2 ppm).

Several studies have evaluated more generic cardiovascular disease outcomes, for example, hospital admissions or emergency room visits for any cardiovascular disease (Ballester et al. 2001, 2006; Barnett et al. 2006; Bell et al. 2009; Chang et al. 2005; Fung et al. 2005; Jalaludin et al. 2006; Koken et al. 2003; Linn et al. 2000; Metzger et al. 2004; Peel et al. 2007; Yang et al. 2004a; Table 3-7). Here again, although results have been mixed, most studies found significant associations between increased rate of hospital admissions or emergency room visits and increased air carbon monoxide concentrations. The association was stronger for older subjects (e.g., ≥64 years, Barnett et al. 2006), and in several studies that examined multi-pollutant models, the association persisted after adjustment for PM10, NO2, SO2, or O3 (Bell et al. 2009; Chang et al. 2005; Jalaludin et al. 2006; Yang et al. 2004a). The mean air carbon monoxide concentrations in these studies ranged from 0.8 to 6 ppm, with the highest values ≤18 ppm.

Animal Studies. Although clinical studies in patients with coronary artery disease provide compelling evidence of adverse cardiovascular effects of acute-duration exposure to carbon monoxide, these studies are limited with respect to exposure conditions and outcomes assessed. Studies in animals, however, have investigated adverse cardiovascular effects of carbon monoxide exposure over a much wider range of exposure conditions (e.g., exposure concentration and duration) and have evaluated additional outcome measures that are not possible to assess in humans. In general, animal studies evaluated carbon monoxide exposures resulting in higher COHb levels than those evaluated in clinical studies in patients and healthy volunteers. Results of animal studies provide evidence of adverse cardiovascular effects of carbon monoxide exposure, including compensatory alterations in hemodynamics, cardiac hypertrophy, cardiac arrhythmias, and possibly atherosclerosis (EPA 1991, 2000). For most outcomes evaluated in animal studies, conflicting evidence has been reported. Although the reasons for different results have not been established, differences in species sensitivity, exposure regimens, and experimental protocols are possible factors. Animal studies published prior to 2000 have been reviewed in detail by EPA (1991, 2000); therefore, these reviews largely form the basis of the following discussions for animals studies published prior to 2000.

Studies investigating the effects of acute-duration carbon monoxide exposure on hemodynamics have been conducted in several animal species, including monkeys, dogs, rats, and rabbits, under exposure conditions producing blood COHb levels ranging from 6.2 to 70% (EPA 1991, 2000). Results show that brief exposure (from a few minutes to approximately 3 hours) to carbon monoxide at concentrations of 80–20,000 ppm produces alterations in hemodynamics that are consistent with COHb-induced hypoxia and responsive compensatory mechanisms (e.g., vasodilation and increased cardiovascular output), including increased coronary blood flow, decreased myocardial O2 consumption, increased heart rate, and alterations in blood flow to various vascular beds (e.g., cerebral, limb muscular). Generally, carbon monoxide-induced effects on hemodynamics were observed at blood COHb levels ≥7.5%. With longer exposure durations at these same concentrations, COHb-induced tissue hypoxia and subsequent compensatory increases in heart rate and cardiac workload lead to the development of cardiomyopathy. Several studies confirm carbon monoxide-induced cardiomegaly in rats and rabbits exposed to 160–11,000 ppm carbon monoxide (COHb 12–58%) for intermediate durations (14 days to ~6 months) (EPA 1991, 2000). Cardiac remodeling, including left ventricular fibrosis of interstitial and perivascular tissue, was observed in rats exposed to 30 ppm, with five daily spikes of 100 ppm, for 12 hours/day for 4 weeks Andre et al. 2010). Under these same exposure conditions, carbon monoxide reduced myocardial perfusion reserved under β-adrenergic stress and alterations in contractile function (Meyer et al. 2011). In rats exposed chronically to 200 ppm carbon monoxide (COHb 14.7%) for 20 hours/day for 72 weeks, left and right ventricle weights increased by 20% (p<0.001) and 14% (p<0.001), respectively (Sørhaug et al. 2006). Studies evaluating developmental cardiovascular effects are reviewed in Section 3.2.6 (Developmental Effects).

Alterations in cardiac rhythm, possibly due to disturbances in cardiac conduction, have been observed in animals exposed to carbon monoxide for acute to intermediate durations (EPA 1991, 2000). Under conditions of simulated myocardial ischemia (i.e., coronary artery ligation), acute exposure (up to 15 minutes) of dogs to up to 5,000 ppm carbon monoxide (COHb 4.9–15%) enhanced myocardial ischemia, as indicated by ST-segment alterations, and severity of myocardial injury. Results of a recent study in rats exposed to 50 ppm carbon monoxide (COHb not reported) for 1 week indicate that rats with pulmonary hypertension may be more sensitive to carbon monoxide-induced myocardial ischemia (Gautier et al. 2007). However, in other studies simulating myocardial injury in dogs exposed to 100–500 ppm carbon monoxide for 6–120 minutes (COHb 5.1–20%), cardiac arrhythmias or alterations in cardiac conduction speed were not observed. In studies evaluating intermediate-duration (6–24 weeks) exposure of dogs and monkeys, 100 ppm carbon monoxide (COHb 2.6–12%) induced cardiac arrhythmias, including R-wave depression, ST-segment elevation, increased T-wave, preventricular contractions, and reduced threshold for stimulus-induced ventricular fibrillation. In contrast, no evidence of cardiac arrhythmias was observed in dogs exposed continuously to 50 ppm carbon monoxide (COHb 7.3%) for 3 months or in monkeys exposed intermittently to 500 ppm (COHb 21.6%) for 14 months.

Conflicting evidence has been reported regarding the potential for carbon monoxide exposure to induce or enhance atherosclerosis (EPA 1991, 2000). Results of most studies in animals (monkeys, baboons, rabbits, and pigs) fed normal diets were negative for inducing or enhancing atherosclerosis at carbon monoxide concentrations of 50–400 ppm (COHb ~20%) for exposure durations up to 14 months. However, atherosclerosis was increased in monkeys continuously exposed to 200 ppm carbon monoxide (COHb 20.6%) for 2 weeks (Thomsen 1974 and in rabbits and monkeys fed high cholesterol diets exposed to 100–300 ppm carbon monoxide (COHb 9–33%) for 2–7 months. Based on the weight of evidence, EPA (1991, 2000) concluded that the available data do not strongly support a relationship between atherogenic effects and carbon monoxide exposure.

Gastrointestinal Effects. Studies evaluating gastrointestinal effects of exposure to low levels of carbon monoxide (i.e., producing blood COHb levels <20%) were not identified. Although the gastrointestinal tract has not been identified as a specific target organ for carbon monoxide-induced toxicity, exposure to carbon monoxide at levels producing hypoxia would be expected to affect any tissue, in particular those tissues with high O2 utilization requirements (e.g., brain, liver, kidney, heart, small intestine).

Hematological Effects. Hematological effects of carbon monoxide include compensatory responses to tissue hypoxia resulting from binding of carbon monoxide to Hb (e.g., increased blood volume, Hb, hematocrit, and erythrocyte count and volume). Possible associations between ambient air carbon monoxide concentrations and hematologic biomarkers of coagulation and inflammation have been examined in epidemiological studies. Collectively, these studies provide some evidence for associations that may reflect effects of carbon monoxide in modulating these systems.

Hypoxia and Related Compensatory Responses. Inhalation exposure to carbon monoxide results in the formation of COHb, a stable complex between carbon monoxide and Hb. Because carbon monoxide has a much higher affinity for Hb than O2 (equilibrium constant for carbon monoxide is >200 times that of O2), O2 is displaced from Hb by relatively low partial pressures of carbon monoxide. Binding of carbon monoxide to Hb has two effects that contribute to impaired O2 delivery to tissues: (1) in the presence to carbon monoxide, the amount of O2 that can be stored on Hb for delivery to tissues decreases; and (2) binding of carbon monoxide to Hb impairs release of O2 from Hb for its diffusion into tissues (see Section 3.4.3 for further discussion of effects of carbon monoxide on O2 dissociation from Hb). At sufficient levels of COHb, the combined effects of impaired O2 storage and transport results in tissue hypoxia, the principal mechanism of many adverse effects of carbon monoxide exposure.

To maintain O2 delivery to tissues under conditions of hypoxia, compensatory hematological responses (e.g., increased erythrocyte count, hematocrit, and Hb) occur. The dose-response relationship for carbon monoxide-induced compensatory hematological responses in humans has not been established. Studies in animals show that compensatory hematological effects occur in response to the COHb-induced reduction in O2 delivery to tissues. Compensatory hematological effects, including increased blood volume, Hb, hematocrit, and erythrocyte count and volume, have been observed following acute- and intermediate-duration carbon monoxide exposure (Davidson and Penney 1988; Penney 1988; Penney et al. 1974a, 1974b; WHO 1999). Continuous exposure of rats to 500 ppm carbon monoxide (COHb 40%) for 5–42 days produced duration-dependent increases in hematocrit, blood Hb, erythrocyte count, and mean cell Hb concentration; mean cell volume was increased after 42 days of exposure (Davidson and Penney 1988). The largest increases (approximately 45–55%) were observed for hematocrit, blood Hb, and erythrocyte count, with increases reaching maximum levels after 30 and 42 days of exposure. Similarly, blood Hb concentration increased by 70% in rats exposed continuously to 500 ppm carbon monoxide (COHb approximately 12% after 42 days) for up to 42 days, with the most rapid increase occurring within the first 21 days of exposure (Penney et al. 1974b). A small, but statistically significant increase in Hb concentration (6.0%; p<0.001) was observed in rats exposed to 100 ppm carbon monoxide (COHb 9.25%) for 30 days (Penney et al. 1974b). In rats exposed to 200–1,300 ppm carbon monoxide (COHb not reported) for 10–17 days, dose-related increases in hematocrit and blood volume were observed (Penney et al. 1988).

Epidemiological Studies of Blood Markers Related to Coagulation. Studies of carbon monoxide mechanisms of action have provided evidence that carbon monoxide may participate in the regulation of vascular thrombi formation and inflammation (see Section 3.5.2, Mechanisms of Toxicity). Possible associations between ambient air carbon monoxide concentrations and biomarkers of coagulation and inflammation have been examined in epidemiological studies (Baccarelli et al. 2007; Liao et al. 2005; Pekkanen et al. 2000; Rückerl et al. 2006, 2007; Steinvil et al. 2008). Biomarkers that have been examined include inflammation markers, C-reactive protein (CRP), serum amyloid A (SAA), and WBC count; cell adhesion markers, E-selectin, von Willebrand factor, antigen (vWF), ICAM-1; and coagulation markers, fibrinogen, factor VII (FVII), prothrombin fragment 1+2, prothrombin time (PT), and activated partial thromboplastin time (APTT). Collectively, these studies provide some evidence for associations between ambient air carbon monoxide concentration and changes in hematological indicators of inflammation and coagulation. However, interpretation of the results of these studies, in particular quantitative estimates of the effect magnitudes on outcomes, is limited by several factors: (1) reliance on area monitoring for estimating exposure levels; (2) uncertainty in knowledge of temporal correspondence between monitored exposure levels and outcomes; (3) relatively strong correlations between ambient air carbon monoxide concentrations and other air quality variables that can affect respiratory function; and (4) relatively low carbon monoxide exposures studied. The mean values of carbon monoxide exposure concentrations have ranged from 0.3 to 3 ppm, with the highest values ≤12 ppm.

Major findings from the body of reported epidemiological studies of hematological outcomes are summarized below. Although most of the studies explored various time lags between monitored air carbon monoxide concentrations and outcomes, as well as various sample strata, for the sake of brevity, only those indicative of the strongest associations to carbon monoxide are presented. Where co-pollutant models have been explored, these results are also presented.

A cohort study conducted in London, United Kingdom (1991–1993) examined plasma fibrinogen concentrations in adult office workers (n=7,205) and found a significant association between increasing air carbon monoxide concentrations (mean 1.4 mg/m3 [1.2 ppm], maximum: 9 mg/m3 [8 ppm]) and increasing plasma fibrinogen concentrations (Pekkanen et al. 2000). For a 3-day lag between measured carbon monoxide concentration and fibrinogen assessment, the odds ratio for a plasma fibrinogen exceeding 3.19 g/L was 1.22 (p<0.02) per 1.6 mg/m3 (1.4 ppm) increase in air carbon monoxide concentration. Increasing air concentration of NO2 was also associated with increasing plasma fibrinogen concentration (odds ratio: 1.22, p<0.01), and when both were considered in two-pollutant models, the associations were no longer significant, reflecting the relative high correlation between air carbon monoxide and NO2 concentrations (r=0.81). When stratified by gender, the carbon monoxide association for a 1-day lag (3-day lag data not reported) was significant in males, but not in females.

In contrast to the Pekkanen et al. (2000) study, a smaller cohort study conducted in Tel-Aviv, Israel (2003–2006, n=3,659, mean age: 46 years) found a significant association between increases in air carbon monoxide concentration and decreasing plasma fibrinogen concentration among males, but not among females (Steinvil et al. 2008). The largest effect was estimated for the 1-week air carbon monoxide average (mean: 0.8 ppm, 75th percentile: 1 ppm), for which a 0.3 ppm increase in air carbon monoxide concentration was associated with a 7.7 mg/dL (95% CI: −12.1, −3.3) decrease in plasma fibrinogen in males, with no association among females (−1.6 mg/dL, 95% CI: −7.3, 4.1). Increasing air carbon monoxide concentration was also associated with a significant decrease in WBCs among males (−158 cell/μL, 95% CI: −298, −18), but not females (−182 cell/μL, 95% CI: −281, 61). However, as in the Pekkanen et al. (2000) study, the association between air NO2 and fibrinogen was stronger than that of carbon monoxide and when included in two-pollutant models, the association with carbon monoxide was no longer significant. Significant associations with air carbon monoxide concentrations were not found for CRP.

A cohort study conducted in three U.S. locations (1996–1998) examined various blood biomarkers of coagulation and inflammation (i.e., fibrinogen, factor VIII-C, von Willebrand factor, serum albumin, WBC) in subjects enrolled in an Atherosclerosis Risk in Communities study (n=10,208) and found a significant association between increasing air carbon monoxide concentrations (mean 1.4 ppm, SD: 0.6) and decreasing serum albumin concentrations (Liao et al. 2005). A 0.6 ppm increase in air carbon monoxide concentration corresponded to a 0.018 g/dL (SE: 0.003, p<0.01) decrease in serum albumin concentration. Other blood biomarkers evaluated were not significantly associated with air carbon monoxide concentration; results of multi-pollutant models were not reported.

A panel study conducted in Efert, German (2000–2001) examined air carbon monoxide concentrations (mean: 0.52 mg/m3 [0.45 ppm], range: 0.11–1.93 [0.10–1.68 ppm]) and several blood biomarkers of coagulation and inflammation in 57 nonsmoking male coronary heart disease patients (age range: 51–7 years; Rückerl et al. 2006). Biomarkers evaluated included inflammation markers, CRP and SAA; cell adhesion markers, E-selectin, vWF, ICAM-1; and coagulation markers, fibrinogen, FVII, and prothrombin fragment 1+2. Odds ratios for increase in air carbon monoxide concentration of 0.27 mg/m3 (0.23 ppm) (2-day lag) were: increasing CRP, 1.5 (95% CI: 1.1, 2.1); increasing ICAM-1, 1.7 (95% CI: 1.3, 2.3); and decreasing FVII, 2.8 (95% CI: 5.1, 0.4). Significant associations were not found for other blood biomarkers, including fibrinogen. A larger cohort study did not find significant associations between ambient air carbon monoxide concentrations (mean range: 0.3–1.48 mg/m3 [0.3–1.29 ppm]) and plasma interleukin-6 (IL-6), CRP, or fibrinogen among myocardial infarction survivors (n=1,003, age range: 31–87 years) in six European cities (2003–2004, Rückerl et al. 2007).

A panel study conducted in Milan, Italy (1995–2005) evaluated associations (mean range: 1.14–3.11 ppm; highest value: 11.4 ppm) between air carbon monoxide concentrations and several blood coagulation parameters, using prothrombin time (PT) and activated partial thromboplastin time (APTT) as outcome measures in 1,218 healthy adolescents and adults (age 11–84 years; Baccarelli et al. 2007). Increasing ambient air carbon monoxide concentration was associated with a decrease in PT (β coefficient: −0.11; 95% CI: −0.18, −0.05). Associations were also significant for PM10 and NO2; however, results from multi-pollutant models were not reported. Significant associations with carbon monoxide were not found for other blood biomarkers evaluated in the study, including carbon monoxide had no effect on fibrinogen, functional antithrombin, functional protein C, protein C antigen, functional protein S, or free protein S.

Musculoskeletal Effects. Studies evaluating effects of exposure to low levels of carbon monoxide (i.e., producing blood COHb levels <20%) on bone in humans or animals were not identified. A 26-year-old woman who attempted suicide by inhalation of charcoal smoke developed heterotopic ossification (formation of mature lamellar bone within soft tissue) of the femoral joints and proximal thighs within 3 months of exposure (Chen et al. 2010). Exposure levels of carbon monoxide were not reported; blood COHb level at the time of admission to the emergency room was 23.5%, although the time from exposure to emergency room admission was not reported. The study authors suggested that development of heterotopic ossification was secondary to ischemic reperfusion. In mice continuously exposed to 2,400 ppm for 180 days, increased bone mass was observed in parietal bone, sternum, lumbar vertebrae, and ribs (but not femur), and the marrow cavities of ribs, parietal bone, and femurs were expanded (Zebro et al. 1983).

A study of 25 patients with delayed sequelae following exposure to carbon monoxide reported persistent muscle weakness, lasting from 1 to 60 months following exposure (Huang et al. 2011). Evaluation by computed tomography (CT) imaging showed decreased technetium (99mTc) sestamibi in skeletal muscle. 99mTc sestamibi, a lipophilic cation used in nuclear medicine imaging, is retained within mitochondria of skeletal muscle. Decreased uptake of 99mTc sestamibi is indicative of muscle damage due to altered mitochondrial metabolism. Carbon monoxide exposure levels were not reported; blood COHb levels for the study cohort were reported as >10%, although COHb range was not reported. Rhabdomyolysis and myonecrosis have been reported in cases of carbon monoxide poisoning (Florkowski et al. 1992; Herman et al. 1988; Kuska et al. 1980; Wolff 1994). Damage to muscle tissue is consistent with hypoxia due to binding of carbon monoxide to Hb and myoglobin and with inhibition of cellular aerobic metabolism due to binding of carbon monoxide to intracellular cytochrome oxidases (Wolff 1994).

Hepatic Effects. Studies evaluating hepatic effects of exposure to low levels of carbon monoxide (i.e., producing blood COHb levels <20%) were not identified. Although the liver has not been identified as a specific target organ for carbon monoxide-induced toxicity, exposure to carbon monoxide at levels producing hypoxia would be expected to affect any tissue, in particular those tissues with high O2 utilization requirements (e.g., brain, liver, kidney, heart, small intestine).

Renal Effects. Studies evaluating renal effects of exposure to low levels of carbon monoxide (i.e., producing blood COHb levels <20%) were not identified. Although the kidney has not been identified as a specific target organ for carbon monoxide-induced toxicity, exposure to carbon monoxide at levels producing hypoxia would be expected to affect any tissue, in particular those tissues with high O2 utilization requirements (e.g., brain, liver, kidney, heart, small intestine). The kidney is the site of active transport processes involved in maintaining blood homeostasis and, next to the brain, is the greatest contributor to basal metabolic rate due to use of ATP-dependent active transport processes. Carbon monoxide-induced hypoxia would decrease the availability of oxygen to produce ATP in renal mitochondria, which would produce adverse effects to the kidneys. Acute renal failure secondary to rhabdomyolysis has been observed in cases of acute carbon monoxide poisoning (Florkowski et al. 1992; Kuska et al. 1980; WHO 1999; Wolff 1994).

Endocrine Effects. Studies evaluating endocrine effects of exposure to low levels of carbon monoxide (i.e., producing blood COHb levels <20%) were not identified. Although the endocrine system has not been identified as a specific target for carbon monoxide-induced toxicity, exposure to carbon monoxide at levels producing hypoxia would be expected to affect any tissue.

Dermal Effects. Studies evaluating dermal effects of exposure to low levels of carbon monoxide (i.e., producing blood COHb levels <20%) were not identified. Skin lesions have been observed in patients with severe carbon monoxide poisoning (Myers et al. 1985; Torne et al. 1991). Lesions are primarily observed in pressure areas and are characterized by blisters, subepidermal vesicles, and sweat gland necrosis.

Ocular Effects. Visual field deficits, temporary or permanent blindness, retinal venous congestion, retinal hemorrhage, papilledema, and optic atrophy have been associated with severe carbon monoxide poisoning in humans (Choi 2001). Numerous clinical studies have investigated the potential for carbon monoxide to induce visual effects (decreased visual tracking, visual vigilance, visual perception) at COHb levels <20% (Benignus et al. 1990; EPA 1991, 2000; Raub and Benignus 2002; Raub et al. 2000). Interpretation of results from most of these clinical studies is complicated by poor study design (single-blind or unblinded designs, small number of study subjects), inadequate reporting, inconsistent results, and inability to duplicate positive findings (Benignus et al. 1990; EPA 1991, 2000; Raub and Benignus 2002; Raub et al. 2000). No information was identified to indicate that carbon monoxide is an ocular irritant in humans or animals.

3.2.3. Immunological and Lymphoreticular Effects

Little information is available regarding the potential of carbon monoxide exposure to produce adverse immunological effects. Epidemiological studies regarding blood biomarkers of immunological status are summarized in the section on Hematologic Effects. Immunological biomarkers examined in these studies have included CRP, SAA, and WBC; however, other markers that can be influenced by the immune system have also been evaluated, including cell adhesion markers such as E-selectin, vWF, and ICAM-1; and coagulation markers such as fibrinogen, FVII, prothrombin fragment 1+2, PT, and APTT. Although results of these studies have been mixed, collectively, they provide some evidence for associations between ambient air carbon monoxide concentration and changes in hematological indicators of inflammation and coagulation. However, these studies do not distinguish between possible effects of carbon monoxide and/or other air pollutants on the immune system from indirect effects of carbon monoxide on blood biomarkers. Mechanistic studies have revealed evidence that carbon monoxide may participate in the regulation of immune function (see Section 3.5.2, Mechanisms of Toxicity).

Exposure of male Sprague-Dawley rats to 2,000 ppm carbon monoxide for 40 minutes followed by exposure to 3,000 ppm for 20 minutes activated and induced proliferation of microglia, immune effector cells, in the brain (Wang et al. 2011). In addition, expression of several immune makers, including MHCII, CD-4, VCAM-1, INF-γ, and OC-42, were increased in the hippocampus and cortex of the brain. Results of a study in guinea pigs investigating the effects of repeated brief bursts of carbon monoxide on systemic immune function were equivocal (Snella and Rylander 1979). Exposure of guinea pigs to 3-minute bursts of 10,000 ppm carbon monoxide 12 times/day for 4 weeks resulted in an 88% reduction in the number of splenic plaque-forming cells, compared to unexposed controls; however, the reduction did not reach statistical significance. Studies evaluating effects of gestational exposure on the developing immune system function are reviewed in Section 3.2.6 (Developmental Effects) (Giustino et al. 1993, 1994).

3.2.4. Neurological Effects

The literature reporting adverse nervous system effects, specifically central nervous system effects, of acute exposure of humans to carbon monoxide is extensive. This literature includes case studies of carbon monoxide poisoning, clinical studies of neurobehavioral effects of controlled exposures in humans, and experimental studies in a variety of animal models (e.g., dogs, rodents, nonhuman primates). Much of this literature has been summarized and analyzed in recent reviews and meta-analyses (Benignus et al. 1990; EPA 1991, 2000; Kao and Nañagas 2006; Raub and Benignus 2002; Raub et al. 2000; WHO 1999). This section summarizes the major findings from these analyses. Although the literature on central nervous system effects of acute experimental and accidental exposures of humans to carbon monoxide is extensive, reports of effects of longer-duration exposure of humans are not available. Effects of gestational and perinatal exposure on the developing nervous system are discussed in Section 3.2.6 (Developmental Effects).

Acute exposure to high levels of carbon monoxide produces symptoms of central nervous system toxicity (EPA 1991, 2000; Ernst and Zibrak 1998; Raub and Benignus 2002; WHO 1999). Symptoms of central nervous system toxicity include headache, dizziness, drowsiness, weakness, nausea, vomiting, confusion, disorientation, irritability, visual disturbances, convulsions, and coma. Symptoms vary depending upon the degree of exposure; headache and dizziness are the most commonly reported symptoms (Dolan 1985; Ernst and Zibrak 1998). Neuroimaging evaluations of patients with acute carbon monoxide poisoning show lesions of the basal ganglia (primarily of the globus pallidus) and white matter (Hopkins et al. 2006; Kao and Nañagas 2006; Lo et al. 2007; Parkinson et al. 2002). Following acute-onset effects, delayed development of neuropsychiatric impairment may occur within 1–4 weeks of exposure, with symptoms including inappropriate euphoria, impaired judgment, poor concentration, memory loss, cognitive and personality changes, and psychosis. Delayed neuropsychiatric impairment has been estimated to occur in up to 68% of patients with acute carbon monoxide poisoning (Ernst and Zibrak 1998; Kao and Nañagas 2006; Raub and Benignus 2002; Raub et al. 2000; WHO 1999; Wolf et al. 2008). The relationship between initial symptom severity and the likelihood of developing delayed neuropsychiatric impairment has not been established. A cohort analysis of 256 patients with carbon monoxide poisoning suggests that initial symptom severity does not correlate to delayed-onset effects (Chambers et al. 2008). However, based on a review of available data, Kao and Nañagas (2006) concluded that patients with more severe initial symptoms are more likely to develop delayed neuropsychiatric impairment.

Mechanisms of acute and delayed adverse nervous system effects produced by carbon monoxide have not been conclusively established. Tissue hypoxia secondary to COHb formation may be a contributing factor, particularly in association with high levels of blood COHb (>60%); however, direct cellular effects of carbon monoxide (e.g., ATP depletion, excitotoxicity, oxidative stress, immunological responses) and postischemic reperfusion injury may also contribute to neurotoxicity (see Section 3.5.2, Mechanisms of Toxicity) (Ernst and Zibrak 1998; Gorman et al. 2003; Kao and Nañagas 2006). Under conditions of hypoxia, including that induced by COHb formation, cerebrovascular vasodilation and increased cardiac output occur as compensatory mechanisms to maintain O2 delivery to the brain (Gorman et al. 2003; Helfaer and Traystman 1996; Raub and Benignus 2002; WHO 1999). In healthy individuals, cardiovascular compensation effectively maintains whole brain O2 delivery up to COHb levels of approximately 60% (EPA 2000). However, based on studies conducted in animals, the elevated cerebral blood flow during carbon monoxide exposure is heterogeneous among different brain regions and may not be sufficient to satisfy O2 demand in all brain regions (Okeda et al. 1987; Sinha et al. 1991). However, despite these compensatory cardiovascular actions, cerebral O2 consumption declines as blood COHb levels increase, with statistically significant decreases at COHb levels of 30–50% (Raub and Benignus 2002). Results of an analysis conducted by EPA (2000) of data obtained in goats (Doblar et al. 1977) and sheep (Langston et al. 1996) show that the cerebral metabolic rate for O2 (CMRO2) decreased 10% at a blood COHb of approximately 23%. These results are consistent with data obtained from humans showing no statistically significant decline in CMRO2 at COHb levels up to 20% (Paulson et al. 1973).

Based on an extensive database of acute carbon monoxide poisoning, it is generally accepted that central nervous system symptoms are associated with acute exposures that produce blood COHb levels ≥20% (EPA 1991, 2000; Raub and Benignus 2002; Raub et al. 2000; WHO 1999). However, results of clinical studies of associations between carbon monoxide exposure and carbon monoxide-induced nervous system effects at blood COHb levels between 5 and 20% are more difficult to interpret (EPA 1991, 2000; Raub and Benignus 2002; Raub et al. 2000; WHO 1999). Numerous clinical studies have investigated the potential for carbon monoxide to induce adverse nervous system effects at COHb levels <20%, including visual and auditory sensory effects (decreased visual tracking, visual and auditory vigilance, visual perception), fine and sensorimotor performance, cognitive effects (altered time discrimination, learning, attention level, driving performance), and brain electrical activity (Benignus et al. 1990; EPA 1991, 2000; Raub and Benignus 2002; Raub et al. 2000). Interpretation of results from most of these clinical studies is complicated by poor study design (single-blind or unblinded designs, small number of study subjects), inadequate reporting, inconsistent results, and inability to duplicate positive findings (Benignus et al. 1990; EPA 1991, 2000; Raub and Benignus 2002; Raub et al. 2000). Therefore, although evidence has been provided for central nervous system effects in humans in association with carbon monoxide exposures that result in COHb levels <20%, dose-response relationships for these effects have not been firmly established. Effects of exposure to ambient environmental levels of carbon monoxide on neurological function and behavior have not been investigated in epidemiological studies.

Benignus (1994) conducted a meta-analysis of the clinical literature on carbon monoxide-induced behavioral effects (also see Benignus et al. 1990; EPA 2000; Raub and Benignus 2002). To minimize the introduction of bias from poorly designed studies, the Benignus analysis included only studies conducted using a double-blind design. The primary goal of these analyses was to compare the dose-response relationship of carbon monoxide-induced effects in humans to other extrapolated data, specifically, the dose-response functions for COHb-induced behavioral effects in rats and hypoxic hypoxia (HH)-induced behavioral effects in humans. Rat COHb dose-response data were converted to human COHb equivalents using a modification of the CFK human model, adjusted to predict exposure-COHb relationships in rats (Benignus and Annau 1994). Data on behavioral effects were also adjusted to account for carbon monoxide-Hb-related hypothermia that occurs in rats (Gordon 1990), but not in humans, assuming response additivity of hypothermia and hypoxia (Benignus 1994). Dose-response data for HH-induced behavioral effects were converted to COHb equivalents (COHb% that yields the equivalent arterial O2 content as a given level of HH) and were corrected for hyperventilation that occurs in HH and resulting changes in alveolar ventilation rate, based on estimates of blood CO2 (i.e., hypocapnia-induced by HH). For all data sets (i.e., carbon monoxide human, carbon monoxide rat, HH human), behavioral responses were transformed to percent of baseline response. Dose-response curves were fit to the extrapolated carbon monoxide rat and HH human data; curve-fitting was not conducted for carbon monoxide human data, due to the small effect level and low COHb levels. Comparison of the extrapolated carbon monoxide rat and HH human data showed very close agreement, with nearly superimposable dose-response curves (Figure 3-1); comparison of dose-response data from carbon monoxide in humans to extrapolated data from HH humans is shown in Figure 3-2. The dose-response relationship from this meta-analysis predicts a 10% decrement in behavioral outcomes in humans in association with COHb levels of approximately ≥20% (range: approximately 18–25%, based on 90% confidence limits) in healthy, sedentary individuals. As discussed by Benignus (1994) and others (Benignus et al. 1990; EPA 1991; Raub and Benignus 2002), the magnitude of behavioral effects that may occur at blood COHb levels ≤20% is expected to be small. Although most task performances are not likely to be affected, it is possible that even small effects (i.e., <10% decrement) could interfere with successful task performance for more difficult or demanding tasks. Therefore, potential minimal behavioral changes occurring at COHb levels ≤20% may be important under certain conditions or in sensitive individuals (e.g., individuals with compromised cardiovascular or central nervous system function).

Figure 3-1. Hypoxic Hypoxia Human and Carbon Monoxide Rat Curves*.

Figure 3-1

Hypoxic Hypoxia Human and Carbon Monoxide Rat Curves*. *Behavioral decrements in rats with elevated carboxyhemoglobin (COHb; dashed line) compared to behavioral effects in hypoxic humans (solid line) in which hypoxia has been expressed in equivalence (more...)

Figure 3-2. Hypoxic Hypoxia Human and Carbon Monoxide Human Curves*.

Figure 3-2

Hypoxic Hypoxia Human and Carbon Monoxide Human Curves*. *Behavioral decrements in humans (circles) with elevated carboxyhemoglobin (COHb) compared to behavioral effects in hypoxic humans (solid and dotted lines) in which hypoxia has been expressed in (more...)

Clinical studies evaluating exposure to low levels of carbon monoxide (i.e., producing blood COHb levels <20%) have focused on the potential for carbon monoxide to induce behavioral effects; however, very little information is available on the effects of carbon monoxide on sensory system function in humans.

Results of clinical studies in evaluating the effects of low carbon monoxide exposures on alterations in sensory perception (i.e., visual and auditory) are equivocal (EPA 1991, 2000). Epidemiological studies have not assessed the potential for carbon monoxide to produce alterations in sensory system function. Studies conducted in mature rats provide evidence that acute exposure (<10 hours) to higher levels of carbon monoxide (500–1,500 ppm) potentiates noise-induced hearing loss, including noise-induced elevation of compound action potential threshold (Chen and Fechter 1999) and auditory threshold shifts (Fechter et al. 1988; Young et al. 1987). In mature rabbits, exposure to 700 ppm for 9.5 hours/day for 5 days potentiated noise-induced increases in the auditory brainstem response threshold (Mortazavi et al. 2010). In these studies, exposure to carbon monoxide alone did not produce changes in auditory function. Noise-induced hearing loss was partially reversed within 4 weeks following exposure to noise alone, but not following exposure to noise and carbon monoxide (Chen and Fechter 1999). Results are consistent with carbon monoxide-induced impairment of mechanisms that repair noise-induced damage to outer hair cells. In marked contrast to results observed in adult rats, developmental studies in animals show that gestational and early postnatal exposure to low levels of carbon monoxide (i.e., 12–25 ppm carbon monoxide) may lead to altered development of the auditory system (Beltran-Parrazal et al. 2010; Lopez et al. 2003, 2008, 2010; Stockard-Sullivan et al. 2003; Webber et al. 2003). These data are discussed in Section 3.2.6.

Additional evidence for effects of carbon monoxide on depressed auditory thresholds and increased cochlear blood flow have been provided from studies in which adult rats received parenteral doses of carbon monoxide (Fechter et al. 1987b). However, in these studies, auditory thresholds assessed by measurement of the compound action potential recorded at the round window remained stable up to COHb levels of at least 30%. Transient impairments of the compound action potential threshold were observed among subjects that attained COHb levels of approximately 50%. Cochlear blood flow measured using laser Doppler flowmetry also increased following carbon monoxide exposure, reaching a maximum of 140% of control values at approximately 50% COHb. In those instances of high COHb production that did produce impairment of auditory threshold, recovery of function was observed as COHb levels declined.

3.2.5. Reproductive Effects

Fetal death in humans has been reported in cases of maternal carbon monoxide poisoning during pregnancy. Epidemiological studies have examined possible relationships between exposures to ambient air concentrations of carbon monoxide and fetal mortality (see Section 3.2.6). Studies in animals provide further evidence that maternal exposure during pregnancy can result in fetal death.

Case Reports. Several case reports have evaluated pregnancy outcomes following acute carbon monoxide poisoning (Brown et al. 1992; Caravati et al. 1988; Cramer 1982; Elkharrat et al. 1991; Farrow et al. 1990; Greingor et al. 2001; Hollander et al. 1987; Margulies 1986; Norman and Halton 1990; Silverman and Montano 1997; Yildiz et al. 2010). In all cases, the severity of carbon monoxide poisoning required emergency room treatment or hospitalization, suggesting that maternal COHb levels exceeded 20%. Pregnancy outcomes varied widely, from delivery of healthy infants at term to fetal death within first few days of exposure. Pregnancy outcome is most likely related to fetal age at time of exposure, with sensitivity increasing with fetal age, and severity of maternal poisoning, although maternal COHb levels are not a good indicator of fetal outcome (Greingor et al. 2001). Fetal death may be secondary to maternal hypoxia and/or fetal hypoxia. The available information from these cases is not adequate to define dose-response relationships for acute carbon monoxide exposure and pregnancy outcomes.

Animal Studies. Few animal studies have assessed the effects of carbon monoxide exposure on reproductive function. Continuous exposure of male mice to 50 ppm carbon monoxide for 2 weeks prior to mating with unexposed females or of male and female mice to 50 ppm carbon monoxide for 2 weeks prior to mating had no effect on pregnancy rate or live or dead fetuses/dam (Stupfel and Bouley 1970). Exposure of pregnant rats to carbon monoxide concentrations up to 200 ppm over the entire gestational period had no effect on number of liters, fetuses per litter, or number of live/dead pups per litter (Fechter and Annau 1977; Penney et al. 1983). In contrast, increased fetal death during the first 24 hours of the postnatal period was observed in rabbits following maternal exposure to 90 and 180 ppm carbon monoxide during pregnancy (9.9 and 35% mortality) (Astrup et al. 1972).

3.2.6. Developmental Effects

Epidemiological studies have examined possible associations between exposure to ambient air carbon monoxide concentrations and various developmental outcomes, including pre-term birth, birth weight, congenital anomalies, neonatal and infant death, and neurodevelopment. Results of these studies have been mixed and, collectively, do not provide strong evidence for developmental effects in association with exposures to ambient levels of carbon monoxide. Studies conducted in animals provide evidence of adverse developmental effects of gestational and early postnatal carbon monoxide exposure, including decreased fetal weight, adverse central nervous system development, altered peripheral nervous system development, cardiac effects, altered sexual behavior, immunological effects, and hematological effects. In addition, some studies showed that developmental effects persisted beyond the postnatal period.

Epidemiological Studies. Interpretation of the results of epidemiological studies of developmental effects of carbon monoxide, and in particular, quantitative estimates of the effect magnitudes on outcomes, is limited by several factors: (1) reliance on area monitoring for estimating exposure level that may not represent exposures that occurred in individuals during any particular period of gestation; (2) uncertainty in knowledge of temporal correspondence between monitored exposure levels and outcomes; (3) relatively strong correlations between ambient air carbon monoxide concentrations and other air quality variables that may affect developmental outcomes; and (4) relatively low carbon monoxide exposures studied. In general, these studies examined relatively low air carbon monoxide concentrations, typical of ambient levels (e.g., mean concentrations ranging from 0.5 to 3 ppm, with the highest reported values ≤10 ppm). The presentation of the epidemiological studies is organized by major categories of birth and neonatal outcomes. Study conclusions are presented in the text, with selected supporting details presented in tabular form (Table 3-8). Although most studies explored various time lags between monitored air carbon monoxide concentrations and outcomes, as well as various sample strata, for the sake of brevity, only selected representative time lags (usually those indicative of the strongest associations to carbon monoxide) are presented in the tables. Where co-pollutant models have been explored, these results are also presented.

Table 3-8. Selected Epidemiological Studies of Associations Between Ambient Carbon Monoxide Concentrations and Developmental Outcomes.

Table 3-8

Selected Epidemiological Studies of Associations Between Ambient Carbon Monoxide Concentrations and Developmental Outcomes.

Studies of Neurodevelopment. A prospective birth cohort study conducted in Guatemala (2002–2010), where exposure to wood smoke is common, examined cognitive development in a sample of 39 children (Dix-Cooper et al. 2011). Various cognitive and motor function tests were performed at age 6–7 years, and associations between test scores and maternal exposures to carbon monoxide during the third trimester were evaluated. Carbon monoxide exposures were measured with passive personal air monitors worn at the breathing zone during pregnancy. The weekly mean concentration during the third trimester was 3.8 ppm (range: 0.62–12.52 ppm). Cognitive and motor function tests performed at age 6–7 years included assessments of information processing speed, visual-spatial integration, short- and long-term memory, attention, fine motor speed, and coordination. In 4 of 11 tests administered, decreasing performance was associated with increasing third trimester maternal exposures. These included visual-spatial integration (β: −4.4 per log10 ppm, 95% CI: −9.5, 0.07), immediate recall (β: −0.3 per log10 ppm, 95% CI: −0.6, 0.01), delayed recall (β: −4.48 per log10 ppm, 95% CI: −9.8, 0.1), and finger tapping (β: −5.7 per log10 ppm, 95% CI: −9.7, −1.7). Significant associations were not found for carbon monoxide exposure measured with passive personal monitors worn by the infants during the first 9 postnatal months.

Studies of Pre-term Birth. Several epidemiological studies have examined possible associations between ambient air carbon monoxide concentrations and risk of pre-term birth (Huynh et al. 2006; Jalaludin et al. 2007; Leem et al. 2006; Liu et al. 2003; Ritz et al. 2000, 2007; Rudra et al. 2011; Wilhelm and Ritz 2005). Results of these studies have been mixed and have included estimates of significantly elevated risk or lower risks of pre-term birth (Jalaludin et al. 2007). Mean air carbon monoxide concentrations evaluated in these studies ranged from 0.8 to 2.7 ppm, with the highest values ≤10 ppm (Ritz et al. 2000).

A prospective cohort study conducted in the Seattle and Tacoma areas of Washington (1996–2006) examined possible associations between air carbon monoxide concentrations and pre-term birth and preeclampsia in 3,509 pregnancies (Rudra et al. 2011). Pre-term birth was not significantly associated with air carbon monoxide concentrations; the odds ratio was 0.98 (95% CI: 0.94, 1.01) per 0.1 ppm increase in carbon monoxide concentration. This study did find an association between carbon monoxide and preeclampsia (odds ratio: 1.07, 95% CI: 1.02, 1.13). However, the association was not significant when year of pregnancy was included in the model (0.98, 95% CI: 0.91, 1.06). A retrospective cohort study (n=97,158 births) conducted in southern California (1989–1993) found a significant association between increasing air carbon monoxide concentration (mean: 2.7 ppm, range: 0.6–9.12) and pre-term birth (Ritz et al. 2000). After adjustment for other risk factors as well as ambient air concentrations of NO2, O3, and PM10, the relative risk of pre-term birth was estimated to be 1.12 (95% CI: 1.04, 1.21) per 3 ppm increase in air carbon monoxide concentration during the last 6 weeks of pregnancy. Subjects were matched to ambient carbon monoxide measurements made at monitors that were <2 miles from the residence. In a follow-up study (1994–2000) that included 106,483 births, relative risk (adjusted for NO2 and O3) was estimated to be 1.10 (95% CI: 1.03, 1.08) per 1 ppm increment in air carbon monoxide concentration during the last 6 weeks of pregnancy (Wilhelm and Ritz 2005). The association was not significant when PM10 was included in the model along with NO2 and O3 (relative risk: 0.98; 95% CI: 0.83, 1.18). A subsequent analysis of southern California data included a nested case-control study consisting of a subset of 2,543 cases of pre-term birth (<37 weeks of gestation) or low birth weight (<2,500 g) and an equal number of randomly selected controls for a larger cohort of 58,316 births (Ritz et al. 2007). Detailed individual data on potential co-variables were obtained, including maternal age, race, education, marriage status, birth season, parity, active and passive smoking, and alcohol use during pregnancy. The adjusted odds ratio in the case-control study for pre-term birth (exposure to >1.25 ppm in first trimester compared to <0.58 ppm) was 1.21 (95% CI: 0.88, 1.65; adjusted for co-variables other than co-pollutants). The odds ratio in the larger cohort (n=58,316) was similar (odds ratio: 1.25, 95% CI: 1.12, 1.38; adjusted for maternal age, race, education, birth season, and parity). Results of multi-pollutant models were not reported. A larger case-control study, also conducted in southern California (1999–2000), did not find a significant association between ambient air carbon monoxide concentration and pre-term birth (Huynh et al. 2006). This study examined 10,673 cases of pre-term births (<37 weeks of gestation) and 32,119 full-term controls (matched for last menstrual period date within 2 weeks of cases). The adjusted odds ratios for pre-term birth were not significant when adjusted for maternal age, maternal race/ethnicity, maternal education, marital status, and parity (with or without inclusion of PM2.5 exposure). A 1 ppm increase in ambient carbon monoxide concentration during the first month of pregnancy was associated with an odds ratio of 1.10 (95% CI: 0.99, 1.20) when CO was considered alone, and 1.03 (95% CI: 0.83, 1.13) after adjustment for PM10. This study matched ambient carbon monoxide measurements to subjects whose residence was <5 miles of the monitoring station, whereas the Ritz et al. (2000, 2007) and Wilhelm and Ritz (2005) studies matched subjects to <2 miles from the monitoring station.

Other large retrospective cohort studies have examined pre-term births in association with ambient air carbon monoxide concentrations. Liu et al. (2003) examined a cohort of 229,085 births in Vancouver, Canada (1985–1998). The odds ratio for pre-term birth for a 1 ppm increase in ambient air carbon monoxide concentration measured during the first month of pregnancy was 1.08 (95% CI: 1.01, 1.15; adjusted for maternal age, parity, infant sex, birth weight, and season of birth) and remained elevated after adjustment for NO2 and O3 (1.08, 95% CI: 1.00, 1.20). The mean air carbon monoxide concentration in the study was 1 ppm, and the highest value was 12.8 ppm. A cohort study of 52,113 births in Incheon, Korea (2001–2002) also estimated elevated risks of pre-term birth in association with increasing ambient air carbon monoxide concentrations during the first trimester. The relative risk increased with increasing estimated exposure concentration and was 1.26 (95% CI: 1.11, 1.44) for the strata 0.91–1.27 mg/m3 (0.79–1.10 ppm) compared to the reference, >0.47–0.63 mg/m3 (0.40–0.55 ppm) (adjusted for maternal age, parity, sex, season of birth, and education level of father and mother; Leem et al. 2006). Results of multi-pollutant models were not reported. The ambient air carbon monoxide concentrations ranged from approximately 0.35 to 1.4 ppm and were matched to subjects by spatial kriging (a spatial statistical method for interpolating exposure levels between sampling locations), with spatial averaging on a scale of approximately 3.5 km2. A cohort study of 123,840 births in Sydney, Australia (1998–2000) did not find a significant association between ambient air carbon monoxide concentration and pre-term birth (Jalaludin et al. 2007). The highest odds ratio estimated was 1.24 (95% CI: 0.81, 1.91) or a 1 ppm increase in 8-hour maximum carbon monoxide concentration measured during first trimester, based on residence <5 km of the monitoring station. When exposure estimates were based on city-wide monitoring, the odds ratio was significantly less than 1 (0.89, 95% CI: 0.84, 0.95), suggesting that risk of pre-term birth decreased with a 1 ppm increase in air carbon monoxide concentration. The city-wide mean 8-hour maximum air carbon monoxide concentration was 0.9 ppm (SD: 0.68).

Studies of Low Birth Weight. Results of studies of low birth weight (i.e., <2,500 g or <10th percentile weight for gestation age) have been mixed, with most studies finding associations between increasing ambient air carbon monoxide concentrations and decreasing birth weight (Bell et al. 2007; Gouveia et al. 2004; Ha et al. 2001; Lee et al. 2003a; Liu et al. 2003, 2007; Maisonet et al. 2001; Mannes et al. 2005; Medeiros and Gouveia 2005; Morello-Frosch et al. 2010; Parker et al. 2005; Ritz and Yu 1999; Salam et al. 2005; Wilhelm and Ritz 2005). In some studies that found significant regression coefficients (i.e., β coefficients) relating air carbon monoxide concentration to birth weight (g), magnitudes of the changes in birth weight were not sufficient to produce elevated risk of low birth weight, where low birth weight is defined categorically as a birth weight <2,500 g. Furthermore, the relative contribution of co-exposures to other air pollutants has not been completely elucidated in epidemiology studies. In many, but not all, studies showing significant associations with birth weight, these associations were not significant after adjustments for exposures to other co-pollutants (e.g., NO2, O3, PM10). All of these studies described below estimated risks after adjustment for typical co-variables that are thought to affect birth weight (e.g., infant sex, maternal age, race/ethnicity, and education, interval since previous live birth, previous low birth weight or preterm infant, level of prenatal care, birth season, parity, and gestational age). Collectively, the reported studies of birth weight outcomes provide supportive evidence for an association between ambient air carbon monoxide concentrations and low birth weight.

A series of studies conducted in California found that such associations were stronger when carbon monoxide was the only pollutant considered in the model and attenuated when adjustments were made for co-pollutants (Morello-Frosch et al. 2010; Parker et al. 2005; Ritz and Yu 1999; Salam et al. 2005; Wilhelm and Ritz 2005). Mean air carbon monoxide concentrations in these studies ranged from 0.75 to 2.4 ppm and the highest value reported was 6.70 ppm (Wilhelm and Ritz 2005). Morello-Frosch et al. (2010) found significant associations between increasing carbon monoxide concentrations and decreasing birth weight and elevated risk of low birth weight in a sample of 3.5 million births (1996–2006). The β coefficient for birth weight was −5.4 g (95% CI: −6.8, −4.1) per 1 ppm increase in carbon monoxide concentration for the entire period of pregnancy and was also significant for exposures during the first or third trimesters. The odds ratio for low birth weight was 1.04 (95% CI: 1.02, 1.06). The association between exposure to carbon monoxide and birth weight persisted when adjusted for exposures to ozone and SO2, but was no longer significant when adjusted for PM10 or PM2.5. The Ritz and Yu (1999) study found elevated risk of low birth weight in a sample of 125,573 births (1989–1993). The odds ratio (>5.5 vs. <2.2 ppm during third trimester) was 1.22 (95% CI: 1.03, 1.44) when carbon monoxide alone was the only air pollutant considered and 1.38 (95% CI: 0.86, 2.22) when NO2, O3, and PM10 were included in the model. The odds ratios remained significant after adjustment for NO2, O3, and PM10, when the analysis was restricted to women <20 years of age (multi-pollutant model: 5.08, 95% CI: 1.77, 16.63; carbon monoxide alone: 1.54, 95% CI: 1.07, 2.22). In a follow-up study (Wilhelm and Ritz 2005; 1994–2000) that included 136,134 births, the odds ratio (>1.84 vs. <0.96 ppm) was estimated to be 1.36 (95% CI: 1.04, 1.76) for women who resided ≤1 mile from an air monitoring station; however, risk was no longer significant when NO2 and O3 (1.29, 95% CI: 0.92, 1.81) or NO2, O3, and PM10 (1.39, 95% CI: 0.77, 2.49) were included in the model. A smaller cohort study (1975–1987) of 3,901 births found significant associations between ambient air carbon monoxide concentrations and low birth weight and intrauterine growth retardation (IUGR, <15th percentile of weight at gestational age; Salam et al. 2005). The estimated odds ratio for IUGR was 1.2 (95% CI: 1.0, 1.4) for a 1.4 ppm increase in air carbon monoxide concentration in the first trimester (carbon monoxide considered alone). Although the odds ratio for low birth weight was not significant (1.0, 95% CI: 0.7, 1.5), a significant association between increasing air carbon monoxide concentration and lower birth weight was observed when birth weight was treated as a continuous variable (β coefficient: −21.7 g, 95% CI: −42.3, −1.1), which persisted when O3 was included in the model (−28.6, 95% CI: −50.4, −6.9). A cohort study of 18,247 full-term births in California (2000) did not find a significant association between ambient air carbon monoxide concentration and risk of SGA (i.e., small for gestational age, <10th percentile of weight for gestational age; Parker et al. 2005). The SGA odds ratio (>0.97 vs. <0.57 ppm for entire pregnancy) was 0.95 (95% CI: 0.81, 1.12) when carbon monoxide was considered alone and 0.82 (95% CI: 0.68, 0.99) when adjusted for PM2.5. When treated as a continuous variable, a significant association between increasing air carbon monoxide concentration and lower birth weight was found when carbon monoxide was modeled alone (β coefficient: −20.5 g, 95% CI: −40.1, −0.8); however, the association was not significant when adjusted for PM2.5 (2.6, 95% CI: −20.6, 25.8).

Other studies of low birth weight conducted in the United States have yielded mixed results. A cohort study of 89,557 full-term births conducted in five northeastern cities (1994–1996) estimated the odds ratio for low birth weight to be 1.31 (95% CI: 1.06, 1.62) for a 1 ppm increase in air carbon monoxide concentration during the third trimester (Maisonet et al. 2001). When stratified by race, the effect was only significant among African Americans for the first trimester (odds ratio: 1.43, 95% CI: 1.18, 1.74) and third trimester (odds ratio: 1.75, 95% CI: 1.50, 2.04). Results of multi-pollutant models were not reported. A cohort study of 358,504 births in Connecticut and Massachusetts found a significant association between increasing ambient air carbon monoxide concentration (for entire pregnancy) and reduced birth weight (β coefficient: −16.2, 95% CI: −19.7, −12.6), which persisted when adjusted for NO2, O3, PM10, PM2.5, or SO2 (Bell et al. 2007). However, the magnitude of the change in birth weight did not result in a significant odds ratio (odds ratio: 1.028, 95% CI: 0.983, 1.074). Mean air carbon monoxide concentrations in the above studies ranged from 0.65 to 0.95 ppm.

Two studies examined low birth weight or IUGR in Canada (Liu et al. 2003, 2007). The Liu et al. (2003) study included 229,085 births in the Vancouver area (1985–1998) and found a significant association between ambient air carbon monoxide concentration (during fist month of pregnancy) and IUGR risk per 1 ppm increase in carbon monoxide (odds ratio: 1.06, 95% CI: 1.01, 1.10), which persisted when adjusted for NO2, O3, or SO2 (OR estimates not reported). In a larger area study (Liu et al. 2007) that included 386,202 births (1995–2000), odds ratios were significant for IUGR when ambient air carbon monoxide concentrations were considered during the first trimester (1.18, 95% CI: 1.14, 1.22), second trimester(1.14, 95% CI: 1.10, 1.18), or third trimester (1.19, 95% CI: 1.14, 1.24). The odd ratios remained significant after adjustment for exposures to NO2 and PM2.5. The mean air carbon monoxide concentration in both studies was 1 ppm and the highest value reported was 5.6 ppm.

Two studies conducted in Seoul, South Korea (1996–1997, 1996–1998) found significantly increased risk of low birth weight in cohorts of 276,763 births (Ha et al. 2001) and 388,105 births (Lee et al. 2003a). In the Ha et al. (2001) study, the relative risk of low birth weight for a 0.42 ppm increase in ambient air carbon monoxide concentration during the first trimesters was 1.08 (95% CI: 1.04, 1.12). In the Lee et al. (2003a) study, the odds ratios for a 0.5 ppm increase in carbon monoxide concentration were 1.04 (95% CI: 1.01, 1.07) for the first trimester, 1.03 (1.00–1.06) for the second trimester carbon monoxide, 0.96 (0.93–0.99) for the third trimester carbon monoxide, and 1.05 (1.01–1.09) for carbon monoxide concentration during the entire pregnancy. Results of multi-pollutant models were not reported. The mean air carbon monoxide concentration was 1.2 ppm (range: 0.4–3.4 ppm).

A cohort study conducted in Sydney, Australia (1998–2000) included 138,056 births and found a significant association between increasing ambient air carbon monoxide concentration and decreasing birth weight (Mannes et al. 2005). For carbon monoxide exposure during the second trimester, an increase of 1 ppm was associated with a 28.7 g reduction birth weight (95% CI: 7.76, 50.98). The association did not persist when the regression models were adjusted for exposure to NO2 or PM10 in two-pollutant models. Furthermore, the magnitude of the effect did not result in a significant risk of SGA (<2 SD of mean weight for gestational age). The mean ambient carbon monoxide concentration was 0.8 ppm (range: 0–4.6).

Two cohort studies conducted in Sao Paulo, Brazil also found a significant association between increasing ambient air carbon monoxide concentration and decreasing birth weight (Gouveia et al. 2004; Medeiros and Gouveia 2005). In the Gouveia et al. (2004) study, in a cohort of 179,460 births (1997), a 1 ppm increase in ambient air carbon monoxide concentration during the first trimester was associated with a 23.1 g reduction in birth weight (95% CI: 4.9, 41.3); however, the effect was not of sufficient magnitude to result in elevated risk of low birth weight. The Medeiros and Gouveia (2005) study included 311,735 births (1998–2000) and estimated a reduction in birth weight of 11.9 g (95% CI: 8.2, 15.5) per 1 ppm increase in first trimester carbon monoxide concentration. Here again, odds ratios for low birth weight were not significant. The ambient carbon monoxide concentrations in these studies ranged from approximately 1 to 11 ppm.

Studies of Congenital Anomalies. Possible associations between exposures to carbon monoxide and developmental anomalies have been evaluated in four case-control studies (Dadvand et al. 2011; Gilboa et al. 2005; Hwang and Jaakkola 2008; Ritz et al. 2002). Three of the studies found significant associations between carbon monoxide and heart anomalies (i.e., ventricular septic defect (Ritz et al. 2002), conotruncal defects (Gilboa et al. 2005), malformation of cardiac septa (Dadvand et al. 2011), and pulmonary valve stenosis (Dadvand et al. 2011). The evaluation in the Hwang and Jaakkola (2008) study was limited to cleft lip and palate and found no association with carbon monoxide, consistent with the Ritz et al. (2002) study. Collectively, these studies do not provide convincing evidence for associations between exposure to carbon monoxide and congenital anomalies.

The Dadvand et al. (2011) study included 2,769 cases and 14,256 controls born during the period 1993– 2003 in northeast England. The study examined heart anomalies, including malformations of the cardiac chambers and connections, cardiac septa, aortic and mitral valves, and great arteries and vein; atrial septal defect; coarctation of aorta; congenital pulmonary stenosis; tetralogy of Fallot; and ventricular septal defect. Of these, significant associations with carbon monoxide concentrations (odds ratio per 1 mg/m3 [0.87 ppm] increase in carbon monoxide concentration) were observed for malformation of cardiac septa (2.330, 95% CI: 1.748, 3.102), pulmonary valve stenosis (2.682, 95% CI: 2.682, 5.534), and ventricular septal defect (2.634, 95% CI: 1.871, 3.707). The Ritz et al. (2002) study included 3,549 cases and 10,649 controls born during the period 1987–1993 in southern California. The study included various categories of anomalies including heart and pulmonary, conotruncal, cleft lip and palate, and chromosomal defects. Of these, significant odds ratios were found for ventricular septal defects with increasing ambient air carbon monoxide concentration during the second month of pregnancy. Odds ratios increased with increasing ambient air carbon monoxide concentration during the second month of pregnancy (<1.14 ppm reference): 1.14–1.60 ppm, odds ratio 1.62 (95% CI: 1.05, 2.48); 1.60– <2.47 ppm, odds ratio 2.09 (95% CI: 1.19, 3.67); and ≥2.47 ppm, odds ratio 2.95 (95% CI: 1.44, 6.05). Significant odds ratios for ventricular septal defects persisted in multi-pollutant models.

A case-control study conducted in Texas (1996–2000) included 4,594 cases and 3,667 controls (Gilboa et al. 2005). Anomalies considered in the study included heart and pulmonary, conotruncal, and cleft lip and palate. Of these, a significant dose trend was evident for tetralogy of Fallot (a conotruncal anomaly of the heart), with a significant odds ratio (>0.7 vs. <0.4 ppm) for ambient air carbon monoxide concentration during the 3rd–8th week of pregnancy (2.04, 95% CI: 1.26, 3.29). The odds ratio for any conotruncal anomaly was also significant at the highest exposure category (>0.7 vs. <0.4 ppm) (1.46, 95% CI: 1.03, 2.08). Results of multi-pollutant models were not reported.

A case-control study conducted in Taiwan (2001–2003) included 653 cases of cleft lip and/or palate and 6,530 controls and examined associations with ambient air carbon monoxide concentrations measured during the first trimester of pregnancy (Hwang and Jaakkola 2008). Air carbon monoxide concentration was not associated with odds ratio for either anomaly in single-or multiple-pollutant models (O3, PM10, or SO2).

Studies of Neonatal and Infant Mortality. Studies of possible associations between ambient air carbon monoxide concentrations and neonatal and infant mortality have yielded mixed results (Ha et al. 2003; Lin et al. 2004a; Ritz et al. 2006; Tsai et al. 2006a; Woodruff et al. 2008; Yang et al. 2006). Although increased risk of infant mortality in association with increasing ambient air carbon monoxide concentrations has been reported (Ha et al. 2003; Ritz et al. 2006), these outcomes have not been rigorously examined for the possible confounding by birth weight and/or gestational age, two relatively influential variables in predicting infant mortality. In one study that did examine the effect of stratification by gestational age or birth weight, the association between air carbon monoxide concentration and mortality from all causes or sudden infant death syndrome (SIDS) persisted in single-pollutant models, but not in models that adjusted for NO2, PM10, and O3 (Ritz et al. 2006). Collectively, these studies do not provide convincing evidence for an association between ambient air carbon monoxide concentrations and neonatal or infant mortality.

A time-series analysis conducted in Sao Paulo, Brazil (1998–2000) did not find a significant association between ambient air carbon monoxide concentrations and neonatal mortality (age 1–28 days; Lin et al. 2004a). The mean air carbon monoxide concentration was 2.8 ppm (range: 0.5–10 ppm). A case-control study conducted in California (1989–2000) included 13,146 cases of infant death (age 28 days–1 year) and 151,015 controls (Ritz et al. 2006). Results varied depending on the temporal window for averaging air carbon monoxide concentration. The odds ratio for infant death for a 1 ppm increase in ambient air carbon monoxide concentration, averaged for 2 months prior to death, was 1.11 (95% CI: 1.06, 1.16), with higher risk for deaths attributable to SIDS (odds ratio: 1.19; 95% CI: 1.10, 1.28). However, risks were not significant after adjustment for co-pollutants (NO2, PM10, and O3). For an averaging period of 2 weeks prior to death, risks were significant for all respiratory-related deaths that occurred between 28 days and 1 year (odds ratio: 1.14, 95% CI: 1.03, 1.25) or from age 28 days to 3 months (odds ratio: 1.20, 95% CI: 1.02, 1.40). Significant respiratory deaths persisted in multi-pollutant models. The association between air carbon monoxide concentration and SIDS mortality risk was persistent when the data were stratified by gestational age and birth weight; however, in these strata, the associations were not significant after adjustment for exposure to NO2, PM10, and O3. The mean air carbon monoxide concentration in the study was 1.6 ppm (range: 0.4–3.4 ppm).

A retrospective cohort study analyzed deaths that occurred between age 28 days and 1 year in a cohort of 3,590,134 births (6,939 deaths) across U.S. counties having a population >250,000 residents (1989–2000; Woodruff et al. 2008). Ambient air carbon monoxide concentration (median: 0.70 ppm, 25th–75th percentile range: 0.48–0.87 ppm) during the first 2 months postpartum was not associated with infant death of any category, including all causes (odds ratio: 1.01, 95% CI: 0.95, 1.07), respiratory (odds ratio: 1.14, 95% CI: 0.93, 1.40), or SIDS (odds ratio: 0.88, 95% CI: 0.76, 1.03). The latter values were estimated from single-pollutant models; however, similar results were obtained in multi-pollutant models that adjusted for O3 and SO2.

A time-series study conducted in Seoul, South Korea (1995–1999) found a significant risk of respiratory-related mortality (age 1 month–1 year) in association with increasing ambient air carbon monoxide concentration measured on the day of death (odds ratio: 1.388, 95% CI: 1.009, 1.911, per 0.57 ppm increase in carbon monoxide concentration), but not for deaths from all causes (Ha et al. 2003). Two case-crossover studies conducted in Taiwan (1994–2000) did not find a significant association between ambient air carbon monoxide concentration and infant mortality (Tsai et al. 2006a; Yang et al. 2006). The Tsai et al. (2006a) study included 206 cases (age 28 days–1 year), and the mean ambient air carbon monoxide concentration was 0.83 ppm (range: 0.23–1.77 ppm). The Yang et al. (2006) study included 471 cases (age 27 days–1 year); the mean ambient air carbon monoxide concentration was 1.58 ppm (range: 0.3–4.8 ppm).

Animal Studies. Numerous studies on developmental effects of gestational and early postnatal exposure to carbon monoxide have been conducted in animals. Study details are summarized in Table 3-9. In general, most studies evaluated effects of relatively low carbon monoxide concentrations (i.e., ≤300 ppm), with exposure concentrations selected to produce maternal COHb levels typically associated with smoking; however, studies did not consistently report maternal or fetal COHb levels. Studies in animals have examined effects of carbon monoxide exposure on numerous developmental outcomes, including several outcomes that have not been assessed in epidemiological studies (e.g., auditory and immune system development). Results of animal studies show adverse developmental effects of gestational and early postnatal carbon monoxide exposure, including decreased fetal weight, adverse central nervous system development, altered peripheral nervous system development, cardiac effects, altered sexual behavior, immunological effects, and hematological effects. In addition, some studies showed that developmental effects persisted beyond the postnatal period (Carratu et al. 1993, 2000a, 2000b; De Salvia et al. 1995; Mactutus and Fechter 1985; Stockard-Sullivan et al. 2003; Webber et al. 2003), although persistence of effects was not examined in all studies. The lowest LOAEL values for developmental effects were obtained in studies evaluating effects of carbon monoxide on the developing auditory system (i.e., LOAEL 12–25 ppm carbon monoxide); however, since other developmental outcomes were not assessed at this range of low carbon monoxide concentrations, it is not possible to determine if the developing auditory system is more sensitive to carbon monoxide exposure than other systems, or to identify a biological basis for such susceptibility.

Table 3-9. Effects of Gestational or Perinatal Exposure on Developmental Outcomes in Animals.

Table 3-9

Effects of Gestational or Perinatal Exposure on Developmental Outcomes in Animals.

Birth Weight. Several studies have evaluated effects of gestational exposure to carbon monoxide on birth weight in rats (Carmines and Rajendran 2008; Fechter and Annau 1977, 1980; Fechter et al. 1980, 1987a; Penney et al. 1983; Prigge and Hochrainer 1977; Singh 2008; Storm and Fechter 1985b), rabbits (Astrup et al. 1972), and guinea pigs (Tolcos et al. 2000b). Results consistently show that gestational exposure to carbon monoxide significantly decreased birth weight and/or pre-weanling weight, with LOAEL values ranging from 60 ppm carbon monoxide (maternal COHb 15%) for continuous exposure during gestation (Fechter and Annau 1977) to 600 ppm (maternal COHb approximately 30%) for intermittent (2 hours/day) exposure (Carmines and Rajendran 2008). Decreases in birth weight exhibited exposure concentration-dependence (Astrup et al. 1972; Fechter et al. 1987a; Prigge and Hochrainer 1977) and were exacerbated when dams were fed protein-deficient diets (Singh 2008).

Central Nervous System Development. Several studies have observed adverse effects on the developing central nervous system, including altered control of respiratory function, behavioral effects, changes in brain neurotransmitters levels, alteration of the cerebellum, and auditory development. Gestational exposure of guinea pigs to 200 ppm carbon monoxide produced abnormal respiratory responses to asphyxia and hypercapnia, indicative of altered brain stem development (McGregor et al. 1998). In rat or mouse pups exposed to 60–150 ppm carbon monoxide during gestation, behavioral changes included decreased motor activity and response to stimulation (Fechter and Annau 1977, 1980), impaired righting reflexes (Fechter and Annau 1980; Singh 1986), and impaired homing and memory acquisition behaviors (Giustino et al. 1999; Mactutus and Fechter 1985). Effects on acquisition and reacquisition behavior persisted up to 18 months of age in rats exposed to 150 ppm carbon monoxide during gestation (De Salvia et al. 1995; Mactutus and Fechter 1985), although similar effects were not observed in pups exposed to 150 ppm carbon monoxide during the early postnatal period (Tattoli et al. 1999). Gestational exposure of rats or guinea pigs to 60–300 ppm carbon monoxide produced altered levels of dopamine and norepinephrine in several regions of the brain (Cagiano et al. 1998; Fechter and Annau 1977; Fechter et al. 1987a; Storm and Fechter 1985a, 1985b; Tolcos et al. 2000b). Gestational and postnatal exposure of rat pups to low levels of carbon monoxide (25 ppm) produced changes consistent with oxidative stress in the cerebellum, including increased expression of neuroglobin and cytoglobin and up-regulation of genes related to generation of ROS and lipid metabolism; changes were more pronounced in pups exposed during gestation compared to those exposed only during the postnatal period (Beltran-Parrazal et al. 2010; Lopez et al. 2009). Gestational or postnatal exposure to low levels of carbon monoxide (i.e., 12–25 ppm) has been shown to have effects on the developing auditory system. Gestational and/or early postnatal exposure of artificially-reared rat pups (i.e., reared on formula to allow exposure to the pups without exposure to the dams during nursing) to 25 ppm carbon monoxide produced morphological changes in the developing auditory system, including swelling, cytoplasmic vacuolization and atrophy of nerve terminals innervating inner hair cells, “distorted myelin” with vacuolization in the 8th cranial nerve at the level of the internal auditory canal, decreased immunoreactivity of the enzymes cytochrome oxidase, NADH-TR, and calcium-mediated myosin ATPase, and decreased immunostaining of neurofilament and myelin basic protein in the organ of corti (Lopez et al. 2003, 2008). Limited physiological and acoustic testing of pups suggested potential functional impairment in the auditory system. Neuroglobin was decreased in spiral ligament cells and spiral ganglia neuron of the cochlea of rat pups exposed to 25 ppm carbon monoxide during gestation (Lopez et al. 2010). The study authors suggested that decreased cochlear neuroglobin could result in development of auditory deficits. Exposure of artificially-reared rat pups during the early postnatal period decreased action potential amplitude of the 8th cranial nerve at ≥12 ppm carbon monoxide, although the magnitude of the effect was not dose-dependent. Otoacoustic emissions showed a small reduction in amplitude (3 dB at two of three frequencies tested and one loudness level) at ≥50 ppm carbon monoxide, with effects on action potential amplitude of the 8th nerve persisting through age 73 days (Stockard-Sullivan et al. 2003). The reduction in amplitude of otoacoustic emissions observed in this study was relatively small. Larger effects are more typically associated with hearing loss (e.g., 10–29 dB across five or more frequencies). Gestational exposure on gestation days (GDs) 8–20 of dam-reared rat pups to ≥12.5 ppm carbon monoxide significantly decreased neuronal activation of the central inferior colliculus (the area or the brain with auditory integrative function), with effects observed through postnatal days 75–77 (Webber et al. 2003). Results indicate that carbon monoxide-induced effects on brain auditory integrative neurons can persist into adulthood.

Peripheral Nervous System Development. Gestational exposure to carbon monoxide produced effects on peripheral nervous system development. In sciatic nerves of pups exposed during gestation, myelin sheath thickness was decreased and sphingomyelin composition was altered at 75–150 ppm carbon monoxide, with changes persisting through age 90 days (Carratu et al. 2000a, 2000b). Altered nerve transmission, indicative of slowed inactivation kinetics of inactive sodium channels, was observed in sciatic nerves from pups exposed to 75 ppm carbon monoxide during gestation, with changes persisting to age 270 days (Carratu et al. 1993).

Cardiac Effects. Dose-related increases in heart weight were observed in rat pups exposed to 60–157 ppm carbon monoxide during gestation (Penney et al. 1983; Prigge and Hochrainer 1977) or to 500 ppm carbon monoxide during the early postnatal period (Penney et al. 1982). The development of cardiomyopathy is consistent with a pathophysiological compensatory response to hypoxia. In ventricular cells isolated from pups exposed to 150 ppm carbon monoxide (but not 75 ppm) during gestation, age-related decreases in action potential duration were seen (Sartiani et al. 2004). Maturation of cardiac cells was impaired, as indicated by decreased expression of sarcomeric proteins (actin, myosin, and troponin I), Ca2+ transporters (Ca2+ transporting ATPase), and enzymes (aldolase), in rat pups exposed to 150 ppm carbon monoxide during gestation (Sartiani et al. 2010)

Sexual Behavior. In male pups exposed to 150 ppm carbon monoxide during gestation, altered development of sexual behavior was observed at age 80 days (Cagiano et al. 1998). Effects included increased mount to intromission latency, decreased mount to intromission frequency, and decreased ejaculation frequency.

Immune System Effects. In studies conducted by Giustino et al. (1993, 1994), gestational exposure produced reversible alterations in immune system function. Decreases in splenic macrophage phagocytosis of Candida albicans was observed in pups exposed to 150 ppm carbon monoxide, and decreased splenic macrophage killing was observed in pups exposed to 75 and 150 ppm carbon monoxide (Giustino et al. 1993); effects did not persist at age 60 days. Rats exposed to 150 ppm carbon monoxide, but not 75 ppm, showed an alteration in splenic immune cell populations, with a significant decrease in leukocyte common antigen cells; effects did not persist to age 540 days (Giustino et al. 1994).

Hematological Effects. Dose-related decreases in hematocrit and Hb were observed in pups exposed to 250 and 500 ppm carbon monoxide during gestation (Prigge and Hochrainer 1977). Effects are consistent with physiological compensatory responses to COHb-induced hypoxia.

3.2.7. Cancer

Epidemiological studies have examined possible associations between exposure to ambient air concentration of carbon monoxide and cancer mortality (described in Section 3.2.1, Death). A prospective study examined mortality in a cohort of 552,138 adults in 151 U.S. metropolitan areas enrolled in the study in 1982 and followed through 1998 (Jerrett et al. 2003; Pope et al. 1995, 2002). Based on estimates of average ambient air carbon monoxide concentration during the period 1982–1998, relative risk for mortality from lung cancer was approximately 0.90 (95% CI: 0.83, 0.96) per 1 ppm increase in carbon monoxide concentration (estimated from Figure 5 of Pope et al. 2002). A time-series analysis conducted in New York City, New York (1985–1994) evaluated data on cancer deaths with or without contributing respiratory diseases (De Leon et al. 2003). Estimates of cancer mortality risk were significant only in the strata with contributing respiratory disease (risk ratio was approximately 1.18, the 95% lower confidence limit >1) and did not persist when estimates were adjusted for ambient concentrations of PM10 (risk ratio approximately 1.06, 95% CI: ~0.97, 1.17). Collectively, the epidemiological data do not show an increase in cancer risk in association with exposure to carbon monoxide.

Carbon monoxide has not been assessed for carcinogenicity using animal models.

3.3. GENOTOXICITY

Few studies have evaluated the genotoxic effects of carbon monoxide exposure. Results of available studies are summarized in Table 3-10. An in vitro bacterial replication assay in Escherichia coli reported that exposure to carbon monoxide inhibited deoxyribonucleic acid (DNA) synthesis (Cairns and Denhardt 1968); however, interpretation of study results is compromised by poor reporting of experimental and analytical methods (e.g., carbon monoxide concentration, exposure conditions, and statistical analyses). Exposure of pregnant mice to carbon monoxide during gestation produced dose-related increases in micronuclei and sister chromatid exchanges in maternal bone marrow and fetal blood (Kwak et al. 1986).

Table 3-10. Genotoxicity of Carbon Monoxide.

Table 3-10

Genotoxicity of Carbon Monoxide.

Pregnant mice were exposed to 0, 1,500, 2,500, or 3,500 ppm carbon monoxide for 10 minutes on GDs 5, 11, or 15 or to 0 or 500 ppm carbon monoxide for 1 hour on GDs 0–6, 7–13, or 14–20; maternal bone marrow and fetal blood were examined for micronuclei formation and sister chromatid exchanges on GD 21. Increases in micronuclei and sister chromatid exchanges in maternal bone marrow and fetal blood were observed under all carbon monoxide exposure conditions, compared to unexposed controls.

The magnitude of effect was similar in dams and fetuses. Although blood COHb levels were not measured in this study, a 1-hour exposure to 500 ppm carbon monoxide would be expected to produce a maternal blood COHb concentration of approximately 20–25% (based on CFK model prediction for rats).

3.4. TOXICOKINETICS

Carbon monoxide exists in the environment as a gas (Henry’s law constant >50,000 atm/mol fraction, 25 °C, see Table 4-2). As a result, humans can be exposed to carbon monoxide from breathing and/or skin contact with carbon monoxide in air. No information is available on the dermal absorption of carbon monoxide resulting from exposures to gaseous carbon monoxide. However, as is the case for other gases that are avidly absorbed from the lung (e.g., O2), dermal absorption of carbon monoxide through intact skin would be expected to make a minor contribution to absorbed carbon monoxide, relative to inhalation. Although carbon monoxide can dissolve in water (23 mL CO/L water, 20 °C), appreciable concentrations in drinking water would occur only at very high partial pressures of carbon monoxide in air, conditions in which inhalation would be the dominant absorption pathway. Since the only relevant pathway of exposure to humans is the inhalation pathway, oral and dermal exposures are not considered further in the discussion of the toxicokinetics of carbon monoxide,

3.4.1. Absorption

Inhaled carbon monoxide is rapidly and extensively absorbed into blood (Benignus et al. 1994; Burge and Skinner 1995; Peterson and Stewart 1970, 1975; Tikuisis et al. 1987b). Inhaled carbon monoxide is transported to lung alveoli as a result of convective forces in the respiratory tract and diffusion. At the alveolar gas-blood interface, carbon monoxide dissolves into pulmonary capillary plasma and, from plasma, diffuses into erythrocytes and other tissues (Figure 3-3). Binding of carbon monoxide by erythrocyte Hb (to form COHb) contributes to maintaining relatively low concentrations of dissolved carbon monoxide in erythrocyte cytosol and a partial pressure gradient to drive carbon monoxide transfer from alveolar air to blood. Diffusion of carbon monoxide into erythrocytes and binding of carbon monoxide to Hb is sufficiently rapid that near equilibrium is achieved between carbon monoxide partial pressures in alveolar air and alveolar end capillary (arterial) blood (Chakraborty et al. 2004). Continued uptake of carbon monoxide results from systemic elimination processes that lower mixed venous blood carbon monoxide concentrations returned to alveolar capillaries (Benignus et al. 1994). The primary elimination processes include diffusive transfer of carbon monoxide to extravascular tissues and binding of carbon monoxide to intracellular heme proteins (e.g., muscle myoglobin), and oxidative metabolism of carbon monoxide to CO2. Seconds after initiating inhalation exposures to carbon monoxide, COHb is detected in arterial blood. During 2–6 minutes of exposure, arterial COHb concentrations were approximately twice venous COHb concentrations and approached unity within 3–5 minutes of cessation of exposure (Benignus et al. 1994). With continued exposure to a constant air carbon monoxide concentration, a steady-state would be achieved within 200–500 minutes (Figure 3-4; Bruce and Bruce 2003; Peterson and Stewart 1970, 1975). At that point, net absorption diminishes to a value determined primarily by metabolic elimination of carbon monoxide (see Section 3.4.3, Metabolism).

Figure 3-3. Summary of Carbon Monoxide Uptake, Distribution, and Elimination Pathways.

Figure 3-3

Summary of Carbon Monoxide Uptake, Distribution, and Elimination Pathways.

Figure 3-4. Temporal Profile of Blood Carboxyhemoglobin (%) for Hypothetical Continuous Exposures to Air Carbon Monoxide at Concentrations of 25, 200, or 1,000 ppm.

Figure 3-4

Temporal Profile of Blood Carboxyhemoglobin (%) for Hypothetical Continuous Exposures to Air Carbon Monoxide at Concentrations of 25, 200, or 1,000 ppm. Source: Peterson and Stewart 1975

Major factors that control absorption of carbon monoxide include those that affect delivery of inhaled air to the alveolar region of the lung and those that affect lung carbon monoxide diffusion capacity. Lung diffusion capacity refers to the driving force for diffusion of carbon monoxide across the alveolar air-blood interface and is reflected in the carbon monoxide partial pressure difference between inspired and exhaled air. In general, factors that increase alveolar ventilation, cardiac output, and gas diffusing capacity of the lung, while maintaining alveolar ventilation perfusion matching (i.e., proportional increases in alveolar ventilation rate and cardiac output) tend to increase absorption of inhaled carbon monoxide. Factors that may increase net carbon monoxide absorption include exercise, supine position, age (increased in infancy and childhood and declines in adults with age), altitude, increasing blood Hb concentration, and decreasing partial pressure of O2 in inhaled air or blood (Castillo et al. 2006; Horvath et al. 1988; Joumard et al. 1991; McGrath et al. 1993; Peterson and Stewart 1975; Tikuisis et al. 1992).

3.4.2. Distribution

Inhaled carbon monoxide distributes to blood and extravascular tissues. In blood, carbon monoxide rapidly distributes into erythrocytes where is exists primarily as a complex with Hb (COHb). The binding equilibrium constant for COHb is approximately 750 µM−1 (Chakraborty et al. 2004; see further discussion in Section 3.4.3, Metabolism). At a blood Hb concentration of 150 g/L (approximately 2.3 mM), the molar concentration ratio of COHb would be approximately 1.7×106 times greater than the molar concentration of dissolved carbon monoxide. The relative affinity of Hb for carbon monoxide and O2 greatly favors binding of carbon monoxide to Hb over O2. The CO/O2 affinity ratio (Haldane constant) is approximately 234 (Chakraborty et al. 2004). This means that a partial pressure of O2 (pO2) approximately 234 times that of pCO is needed to achieve an equilibrium in which equal amounts of O2 and carbon monoxide are bound to Hb (i.e., %COHb = % oxyhemoglobin [O2Hb]; Engel et al. 1969; Rodkey et al. 1969; Roughton 1970). In general, typical levels of COHb in nonsmokers are <2% (Adams et al. 1988; Allred et al. 1991; Anderson et al. 1973; Hinderliter et al. 1989; Kleinman et al. 1989, 1998; Sheps et al. 1987, 1990). Of this, COHb levels of approximately 0.2–1.0% are thought to be derived from endogenous production of carbon monoxide (Coburn et al. 1963; Delivoria-Papadopoulos et al. 1974; Longo 1977). Levels of COHb can (rarely) achieve 10% immediately following cigarette smoking (Kao and Nañagas 2006; Landaw 1973).

Although bound extensively to Hb in blood, distribution of dissolved carbon monoxide out of the vasculature can occur in response to diffusion gradients for carbon monoxide created by binding of carbon monoxide to Hb, myoglobin, and other heme proteins in extravascular tissues (e.g., muscle, spleen), as well as removal of carbon monoxide from extravascular tissues by oxidative metabolism (Bruce and Bruce 2006; Bruce et al. 2008; Luomanmäki and Coburn 1969). Myoglobin, the dominant binding protein for carbon monoxide in cardiac and skeletal muscle, has a lower affinity for carbon monoxide than Hb (COMb equilibrium constant of ≈20 µM−1 and Haldane constant of approximately 23) and a single binding site for carbon monoxide (Gibson et al. 1986); however, its abundance in muscle can result in transfer of substantial amounts of inhaled carbon monoxide into muscle tissue. At a skeletal muscle myoglobin concentration of 4.7 g/kg muscle (approximately 0.26 mmole/kg; Moller and Sylven 1981), the molar concentration ratio of COMb would be approximately 7.4×103 times greater than the molar concentration of dissolved carbon monoxide. Binding of carbon monoxide to Hb in blood and myoglobin in muscle decreases the concentrations of dissolved carbon monoxide in both tissues and limits the concentration gradient for diffusion of carbon monoxide into muscle (Bruce and Bruce 2006; Bruce et al. 2008). Exercise can increase the rate of transfer of carbon monoxide from blood to skeletal muscle (Richardson et al. 2002; Werner and Lindahl 1980). The effect of exercise is thought to be related, in part, to increased O2 consumption and lower pO2 in muscle, which favors binding of carbon monoxide to myoglobin, and increased muscle blood flow. These effects a more pronounced at low COHb concentrations (≤2%) than at higher concentrations (20% COHb; Richardson et al. 2002). Hypoxia appears to increase transfer of carbon monoxide from blood to muscle (Coburn and Mayers 1971; Richardson et al. 2002).

Binding of carbon monoxide to intracellular proteins, in particular, heme proteins, contributes to maintaining a favorable gradient for carbon monoxide diffusion from plasma into cells. Measurements of total carbon monoxide concentrations in tissues obtained from human autopsies showed the highest concentrations in blood, spleen, lung, kidney, and skeletal muscle, with detectable levels also in brain and adipose tissue (Table 3-11, Vreman et al. 2006). Higher concentrations of carbon monoxide in blood, heart, skeletal muscle, and spleen observed in human autopsy studies reflect the abundance of the major carbon monoxide binding proteins in these tissue, Hb in erythrocytes and spleen (site of destruction and storage of erythrocytes), and myoglobin in cardiac and skeletal muscle. If the values presented in Table 3-11 are converted to tissue carbon monoxide mass (pmol), based on assumed tissue masses for a 70-kg adult (Davies and Morris 1993; ILSI 1994), carbon monoxide in blood and muscle accounted for approximately 60–70 and 20–30% of total body carbon monoxide burden, respectively. Relatively large contributions of blood and skeletal muscle to total body carbon monoxide content is consistent with the contributions that these tissues make to total body Hb and muscle myoglobin (Bruce and Bruce 2003). In a typical adult, blood contains approximately 12 mmoles of Hb with a total binding capacity of approximately 48 mmole carbon monoxide when completely saturated and muscle contains approximately 8 mmoles of myoglobin with a total binding capacity of 8 mmole carbon monoxide when completely saturated.

Table 3-11. Post-Mortem Tissue Distribution of Carbon Monoxide in Humans.

Table 3-11

Post-Mortem Tissue Distribution of Carbon Monoxide in Humans.

The distribution of carbon monoxide residues observed in human autopsy studies is similar to that observed in rodents. In mice and rats that had not been intentionally exposed to air carbon monoxide (i.e., sacrificed during steady-state conditions), blood and muscle accounted for most the total body carbon monoxide burden (Vreman et al. 2005). Following 30-minute exposures to carbon monoxide (5,000 ppm), blood made a much larger contribution (approximately 90%) to carbon monoxide body burden, reflecting the longer period needed for carbon monoxide to distribute from blood to extravascular tissues (Vreman et al. 2005).

Carbon monoxide in the maternal system distributes to fetal tissues. Measurements of steady-state COHb concentrations in fetal and maternal blood of nonsmoking women have found fetal COHb concentrations to be approximately 10–15% higher than maternal blood (fetal/maternal ratio=1.1–1.15; Longo 1977). Carbon monoxide binding to fetal Hb is analogously similar to maternal Hb. Both Hbs have the same binding capacity of 4 moles CO/mole Hb and both exhibit cooperative binding of carbon monoxide, whereby binding affinity increases as carbon monoxide molecules are successively added to Hb (see Section 3.4.3, Metabolism). However, the binding affinity of fetal Hb is approximately twice that of maternal Hb (Di Cera et al. 1989). The Haldane coefficient, reflecting the relative affinities of carbon monoxide and O2 (CO/O2) for fetal Hb, appears to be approximately 20% lower than adult Hb (Di Cera et al. 1989; Engel et al. 1969; Hill et al. 1977). The higher binding affinity of fetal Hb as well as the relatively small diffusion gradients for carbon monoxide between maternal and fetal blood contribute to slower kinetics of COHb in fetal blood compared to maternal blood. This has been experimentally verified in studies conducted in pregnant sheep (Longo and Hill 1977). The half-time for approach to steady-state in the human fetus has been predicted (from modeling) to be approximately 7.5 hours in the fetus, compared to approximately 4 hours in maternal blood (Hill et al. 1977). As a result, with continuous exposure to a constant level of carbon monoxide in air, fetal blood COHb concentration would be expected to lag behind maternal levels and reach a steady-state that is approximately 10–15% higher than maternal (Longo 1977). Similarly, following cessation of exposure, elimination kinetics of carbon monoxide from fetal blood would be expected to lag behind that of maternal blood (Hill et al. 1977).

3.4.3. Metabolism

Metabolism of carbon monoxide consists of three major processes: (1) metabolic production of carbon monoxide from endogenous and exogenous precursors; (2) binding of carbon monoxide to heme proteins (e.g., Hb, myoglobin, cytochromes); and (3) oxidative metabolism of carbon monoxide to CO2.

Metabolic Production of Carbon Monoxide. Carbon monoxide is produced from the enzymatic degradation of endogenous and exogenous precursors. The major endogenous source of carbon monoxide production is the oxidative metabolism of heme by the enzyme heme oxygenase (HO). The net rate of endogenous production of carbon monoxide (i.e., total production minus metabolism) has been estimated in healthy adults to be approximately 0.42 mL CO/hour (16.4 µmole/hour) or approximately 400 µmole/day (Coburn et al. 1963). Of this, approximately 75–80% is thought to be derived from Hb metabolism, and the remaining portion from metabolism of other heme proteins, such as myoglobin, cytochromes, peroxidases, and catalase (Berk et al. 1976). Endogenous carbon monoxide production contributes to a blood COHb concentration of approximately 0.4–0.7% (Benignus 1995; Coburn et al. 1965). Carbon monoxide is produced in all tissues that express HO; however, the major tissues contributing to carbon monoxide production are also major sites of heme metabolism, including liver, spleen, and reticuloendothelial system (Berk et al. 1974).

Numerous physiological and disease factors affect the rate of endogenous production of carbon monoxide. Carbon monoxide production increases by approximately 2-fold during the post-ovulatory (progesterone) phase of the menstrual cycle (Delivoria-Papadopoulos et al. 1974; Mercke and Lundh 1976). Carbon monoxide production also increases during pregnancy, reaching values that are 2–5 times that of the estrogen phase of the menstrual cycle at partum and returning to pre-pregnancy levels within 4 days following delivery (Longo 1977). Contributing factors to the increase in carbon monoxide production during pregnancy include the contribution of the growing fetus to carbon monoxide production, as well as the increase in maternal erythrocyte production and related changes in Hb metabolism (Longo 1977). The rate of production of carbon monoxide is accelerated in conditions that increase catabolism of Hb or other heme proteins, including hemolysis, hematomas, hemolytic anemias, thalassemia, and Gilbert’s syndrome (Berk et al. 1974; Coburn et al. 1964, 1965; Hampson 2007; Meyer et al. 1998; Solanki et al. 1988). Carbon monoxide production was estimated to be 2–3 times higher in patients with hemolytic anemia compared with healthy individuals (Coburn et al. 1965).

HO is the rate-limiting step in heme degradation, and its expression and activity are modulated by numerous factors. The HO-1 isozyme (one of three isozymes) and expression of the enzyme is induced by heme and heme derivatives, oxidative stress, hypoxia (including altitude-induced hypoxia), various metals, various cytokines, and exogenous carbon monoxide (Wu and Wang 2005). These factors would tend to promote endogenous carbon monoxide production.

In addition to endogenous carbon monoxide production from metabolism of heme, carbon monoxide is produced from oxidative metabolism of certain exogenous hydrocarbons, including dichloromethane (DCM) and other dihalomethanes, and carbon tetrachloride (Agency for Toxic Substances and Disease Registry 2000, 2005; Stevens et al. 1980). The oxidative metabolism of DCM by isoform CYP2E1 of cytochrome P450 results in the production of carbon monoxide and increased blood COHb concentrations in blood (Andersen et al. 1991). Elevated blood COHb levels have been observed following DCM poisoning (Leikin et al. 1990). Other sources of exogenous carbon monoxide production include the heme oxidase catalysis of products of auto-oxidation of phenols, photo-oxidation of organic compounds, and lipid peroxidation of cell membrane lipids (Rodgers et al. 1994).

Binding of Carbon Monoxide to Heme Proteins. Carbon monoxide binds to heme and competes for binding of O2. Binding of carbon monoxide to heme proteins (e.g., blood Hb, muscle myoglobin) has profound effects on the kinetics of distribution and elimination kinetics of both carbon monoxide and O2. Affinities of heme for carbon monoxide for Hb and myoglobin are sufficiently high to result in concentrations of dissolved carbon monoxide being 103–106 times lower than the concentrations of COHb or COMb. This substantially lowers the concentration of dissolved carbon monoxide in blood and tissues, decreases the diffusion gradients for carbon monoxide movement of carbon monoxide from tissues and blood to the lungs, and substantially prolongs the elimination time for carbon monoxide than would otherwise occur in the absence of binding. Furthermore, the rates of dissociation of carbon monoxide from Hb and myoglobin are substantially slower than that of O2 and sufficiently slow, relative to erythrocyte capillary transit times, to limit rates of exchange of carbon monoxide between tissues and capillary blood. Binding of carbon monoxide to Hb also increases the affinity of Hb for O2, decreasing the availability of O2 to tissues.

Although carbon monoxide binds to many different heme proteins (see Section 3.5.2), the affinities and abundance of Hb and myoglobin in the body results in most of the body stores of carbon monoxide being associated with these two heme proteins. Estimated kinetics and equilibrium parameters for carbon monoxide and O2 binding to Hb and myoglobin are presented in Table 3-12. Each of the four heme moieties in Hb can bind O2 or carbon monoxide. The overall affinity (i.e., equilibrium constant, Keq) is substantially higher for carbon monoxide (≈750 µM−1) compared to that of O2 (≈3.2 µM−1). As a result, at similar partial pressures of the two gases (pCO=PO2), binding of carbon monoxide will occur at the expense of binding of O2, and the O2 carrying capacity of blood will be compromised. The ratio of the Keq values (CO/O2) is referred to as the Haldane constant (≈234). This value gives the partial pressure ratio (pO2/pCO) needed to achieve equal amounts of bound COHb and O2Hb (i.e., %COHb = %O2Hb).

Table 3-12. Hemoglobin and Myoglobin Binding Kinetics and Equilibrium Constants for Oxygen and Carbon Monoxide.

Table 3-12

Hemoglobin and Myoglobin Binding Kinetics and Equilibrium Constants for Oxygen and Carbon Monoxide.

At an ambient air pO2 of approximately 160 Torr (21% O2), the equilibrium condition of equal %COHb = %O2Hb would be achieved at a carbon monoxide exposure concentration of approximately 900 ppm. The affinity of Hb for carbon monoxide results in an equilibrium condition in which nearly all carbon monoxide in blood is bound to Hb. For example, based on the value of Keq of 750 µM−1, at a blood Hb concentration of 150 g/L (≈2.3 mM), the equilibrium ratio of bound carbon monoxide to dissolved carbon monoxide in blood would be approximately 1.7×106 (i.e., > 106 times as much carbon monoxide is bound to Hb than is circulating in blood as dissolved carbon monoxide). The carbon monoxide dissociation rate from Hb (kd≈0.008 s−1; t1/2≈87 s) is slower than from O2 (ke≈20 s−1, t1/2≈0.035 s). As a result, exchange of O2 bound to Hb in blood with alveolar air and tissues is much more efficient than carbon monoxide exchange (i.e., fast vs. slow dissociation relative to erythrocyte transit times in capillaries). These two characteristics (high affinity and slow dissociation) contribute to relatively long retention times of carbon monoxide in blood following exposures to carbon monoxide.

Binding of O2 to the four heme moieties of Hb is cooperative (Alcantara et al. 2007; Roughton 1970). The associative reaction rate (ka) becomes faster with successive additions of O2, increasing the affinity (Keq) of Hb for O2 (values for ka shown in Table 3-12 are for the high affinity, R, state of fully loaded Hb). Cooperative binding of O2 to Hb serves an important physiological function. Under conditions of relatively high pO2 in arterial blood entering the lung (pO2=100 mmHg), cooperative binding promotes the loading of Hb with O2; with successive binding of O2 molecules, the associative rate and affinity increases, promoting binding of a subsequent molecule of O2. Conversely, at lower pO2 in venous blood (pO2≈20 mmHg), cooperative binding promotes unloading of O2 for utilization; dissociation of a molecule of O2 decreases the affinity of Hb for O2, promoting the dissociation of a subsequent molecule of O2. Cooperative binding contributes to the sigmoid shape of the O2 dissociation curve for Hb (Figure 3-5).

Figure 3-5. Oxyhemoglobin Dissociation Curve for Normal Human Blood Containing 50% Carboxyhemoglobin and for Blood of an Anemic Human.

Figure 3-5

Oxyhemoglobin Dissociation Curve for Normal Human Blood Containing 50% Carboxyhemoglobin and for Blood of an Anemic Human. Source: EPA 2000

Binding of carbon monoxide to Hb is also cooperative. Binding of carbon monoxide increases the associative rate of binding and affinity of both carbon monoxide and O2. As a result, the presence of carbon monoxide has several effects on the dissociation curve for O2 (Figure 3-5): (1) decreases the partial pressure of O2 in blood and tissues (i.e., by an amount equal to pCO), shifting the pO2 axis downscale at each point in the arterial-venous circuit; (2) displaces O2 from Hb, decreasing the maximum %COHb achieved in arterial blood; and (3) shifts the O2 dissociation curve downscale so that less O2 dissociates O2 from Hb (more will remain bound, %O2Hb will be higher) at the lower pO2 values in venous blood. The combined result of these changes is that less O2 in blood is available for utilization in tissues.

The binding affinities of carbon monoxide and O2 to myoglobin are lower than that for Hb; however, the higher affinity for carbon monoxide (Haldane constant ≈23), results in preferential binding of carbon monoxide over O2 and displacement of O2 of myoglobin at similar partial pressures of both gasses (Table 3-12). In addition to displacing O2 from myoglobin, binding of carbon monoxide to myoglobin has a pronounced effect on the distribution of carbon monoxide in the body by limiting the dissolved carbon monoxide concentration in muscle and, thereby, the rates of diffusion of carbon monoxide from muscle to blood. The affinity of myoglobin (Keq≈27) results in an equilibrium condition in which dissolved carbon monoxide concentrations in muscle are very low relative to COMb concentrations. For example, based on a value for Keq of 27 µM−1, at a muscle myoglobin content of approximately 4.7 g/kg (≈0.28 mM), the equilibrium ratio of bound carbon monoxide to dissolved carbon monoxide in muscle would be approximately 7.4×103 (approximately 104 times as much carbon monoxide is bound to myoglobin than is dissolved muscle tissue). Furthermore, the carbon monoxide dissociation rate from myoglobin (kd≈0.019 s−1; t1/2≈36 s) is slower than O2 (ke≈12 s−1, t1/2≈0.06 s). As a result exchange of carbon monoxide bound to myoglobin in muscle with blood is much less efficient than O2 exchange (i.e., slow vs. fast dissociation relative to erythrocyte transit times in capillaries). These two characteristics (high affinity and slow dissociation) contribute to relatively long retention times of carbon monoxide in muscle following exposures to carbon monoxide.

Oxidative Metabolism of Carbon Monoxide. Carbon monoxide can be oxidized to CO2. The rate of oxidation has been estimated in human subjects who inhaled 14CO to be approximately 0.015 mL CO/hour, which corresponds to approximately 0.6 µmol/hour; 14 µmol/day, and approximately 3% of endogenous carbon monoxide production (Luomanmaki and Coburn 1969). Similar measurements made in dogs yielded estimates for carbon monoxide metabolism that were approximately 8% of endogenous production (Luomanmaki and Coburn 1969). The major pathway for metabolism of carbon monoxide to CO2 is oxidation by mitochondrial cytochrome oxidase (Fenn 1970; Young and Caughey 1986).

3.4.4. Elimination and Excretion

Absorbed carbon monoxide is eliminated from the body by exhalation and oxidative metabolism. Oxidative metabolism of carbon monoxide has been estimated to be a relatively small fraction (<10%) of endogenous carbon monoxide production (Luomanmaki and Coburn 1969). Under most conditions, the dominant route of elimination of absorbed carbon monoxide is exhalation. Rates of elimination of carbon monoxide in humans who have experienced inhalation exposures to carbon monoxide have been estimated from kinetic analyses of the time course of blood %COHb (Bruce and Bruce 2006; Joumard et al. 1991; Landaw 1973; Levasseur et al. 1996; Peterson and Stewart 1970; Shimazu et al. 2000; Weaver et al. 2000). The decline in blood %COHb following cessation of an inhalation exposure to carbon monoxide exhibits at least two kinetic phases (Bruce and Bruce 2006; Shimazu et al. 2000). The fast phase is thought to reflect a combination of exhalation of carbon monoxide along with slower distribution of blood carbon monoxide to tissues that continues after cessation of exposure (Bruce and Bruce 2006). The elimination half-times for both phases appear to increase with exposure intensity and/or duration; a contributing factor to this effect may be the amount of carbon monoxide accumulated in extravascular stores. In human subjects exposed to 2% carbon monoxide for 1–3 minutes (peak COHb%=30–40), values for elimination half-lives were 5.7±1.5 and 103±20.5 minutes for the fast and slow phases, respectively (Shimazu et al. 2000). The fast phase rate increased when subjects were exposed to 500 ppm for 5–10 hours (peak %COHb=34–37): 21.5±2.1 and 118±11.2 minutes, respectively (Shimazu et al. 2000). Rates have been reported for the slow phase of elimination. In subjects whose peak %COHb ranged from 2 to 15, the half-life was 320 minutes (range: 128–409 minutes, Peterson and Stewart 1970). Bruce and Bruce (2006) measured the time to achieve one-half of peak %COHb in subjects whose peak %COHb values ranged from 15 to 20 and estimated a half-life of 248 minutes (range: 206–328). Clinical studies conducted on healthy individuals exposed to ambient air carbon monoxide concentrations have found that elimination half-time increases with age, with the most pronounced increase occurring from age 2 to 20 years, and is approximately 6% longer in males compared to females (Joumard et al. 1991). The gender effect may be related to relative differences in body size and/or muscle mass. Exercise decreases carbon monoxide elimination half-time (Joumard et al. 1991).

The rate of elimination of absorbed carbon monoxide is accelerated by inhaling high concentrations of O2 (Landaw 1973; Peterson and Stewart 1970; Weaver et al. 2000). Elimination half-lives of approximately 20 minutes have been observed in subjects who inhaled 100% O2 following cessation of exposure to carbon monoxide. An analysis of elimination rates observed in 93 acute carbon monoxide poisoning patients who were treated with 100% O2 did not find significant associations between elimination half-lives and age, gender (<40 years, ≥40 years), smoke inhalation, history of loss of consciousness, concurrent tobacco smoking, degree of initial metabolic acidosis (base excess), or initial %COHb level (Weaver et al. 2000).

3.4.5. Physiologically Based Pharmacokinetic (PBPK)/Pharmacodynamic (PD) Models

Physiologically based pharmacokinetic (PBPK) models use mathematical descriptions of the uptake and disposition of chemical substances to quantitatively describe the relationships among critical biological processes (Krishnan et al. 1994). PBPK models are also called biologically based tissue dosimetry models. PBPK models are increasingly used in risk assessments, primarily to predict the concentration of potentially toxic moieties of a chemical that will be delivered to any given target tissue following various combinations of route, dose level, and test species (Clewell and Andersen 1985). Physiologically based pharmacodynamic (PBPD) models use mathematical descriptions of the dose-response function to quantitatively describe the relationship between target tissue dose and toxic end points.

PBPK/PD models refine our understanding of complex quantitative dose behaviors by helping to delineate and characterize the relationships between: (1) the external/exposure concentration and target tissue dose of the toxic moiety, and (2) the target tissue dose and observed responses (Andersen and Krishnan 1994; Andersen et al. 1987). These models are biologically and mechanistically based and can be used to extrapolate the pharmacokinetic behavior of chemical substances from high to low dose, from route to route, between species, and between subpopulations within a species. The biological basis of PBPK models results in more meaningful extrapolations than those generated with the more conventional use of uncertainty factors.

The PBPK model for a chemical substance is developed in four interconnected steps: (1) model representation, (2) model parameterization, (3) model simulation, and (4) model validation (Krishnan and Andersen 1994). In the early 1990s, validated PBPK models were developed for a number of toxicologically important chemical substances, both volatile and nonvolatile (Krishnan and Andersen 1994; Leung 1993). PBPK models for a particular substance require estimates of the chemical substance-specific physicochemical parameters, and species-specific physiological and biological parameters. The numerical estimates of these model parameters are incorporated within a set of differential and algebraic equations that describe the pharmacokinetic processes. Solving these differential and algebraic equations provides the predictions of tissue dose. Computers then provide process simulations based on these solutions.

The structure and mathematical expressions used in PBPK models significantly simplify the true complexities of biological systems. If the uptake and disposition of the chemical substance(s) are adequately described, however, this simplification is desirable because data are often unavailable for many biological processes. A simplified scheme reduces the magnitude of cumulative uncertainty. The adequacy of the model is, therefore, of great importance, and model validation is essential to the use of PBPK models in risk assessment.

PBPK models improve the pharmacokinetic extrapolations used in risk assessments that identify the maximal (i.e., the safe) levels for human exposure to chemical substances (Andersen and Krishnan 1994). PBPK models provide a scientifically sound means to predict the target tissue dose of chemicals in humans who are exposed to environmental levels (for example, levels that might occur at hazardous waste sites) based on the results of studies where doses were higher or were administered in different species. Figure 3-6 shows a conceptualized representation of a PBPK model.

Figure 3-6. Conceptual Representation of a Physiologically Based Pharmacokinetic (PBPK) Model for a Hypothetical Chemical Substance.

Figure 3-6

Conceptual Representation of a Physiologically Based Pharmacokinetic (PBPK) Model for a Hypothetical Chemical Substance. Note: This is a conceptual representation of a physiologically based pharmacokinetic (PBPK) model for a hypothetical chemical substance. (more...)

If PBPK models for carbon monoxide exist, the overall results and individual models are discussed in this section in terms of their use in risk assessment, tissue dosimetry, and dose, route, and species extrapolations.

The first physiologically-based mechanistic model of carbon monoxide kinetics was reported in Coburn et al. (1965). This model, which subsequently became known as the Coburn-Forster-Kane (CFK) model, was developed to predict relationships between inhalation exposures to carbon monoxide and blood COHb levels in humans. Numerous modifications and extensions of the original CFK model have been reported that more rigorously address nonlinearities resulting from interdependence of COHb and O2Hb levels in blood (Tikuisis et al. 1987b), simulation of arterial and venous blood compartments (Smith et al. 1994), simulation of maternal-fetal transfer of carbon monoxide (Hill et al. 1977), extrapolation to nonhuman species (Benignus and Annau 1994; Hill et al. 1977; Longo and Hill 1977), linkage of the CFK model to more complex respiratory gas exchange models (Abram et al. 2007; Benignus 1995), and use in simulating the elimination kinetics of carbon monoxide produced from exogenous precursors (Andersen et al. 1991). A multi-compartment model was also developed that simulates carbon monoxide absorption and distribution kinetics in muscle (Bruce and Bruce 2003; Bruce et al. 2008).

Coburn-Forster-Kane Model

The CFK model, as originally described in Coburn et al. (1965), is a single-compartment model of instantaneous equilibrium between carbon monoxide, O2, and blood Hb, and assumes that blood O2Hb concentration is constant (i.e., independent of COHb concentration). This linear form of the CFK model is expressed in the following differential equation which is integrated over time to solve for COHb concentration (Equation 1):

Vbd[COHb]tdt=VCO[COHb]t1PcO2[O2Hb]M(11DLCO+1VA)+(PICO1DLCO+1VA)
Eq. (1)
where t represents time at the end of the integration time step and t-1 represents the time at the beginning of the time step; Vb is blood volume in mL; [COHb] is the COHb concentration in mL carbon monoxide per mL blood (STPD); VCO is the endogenous carbon monoxide production rate in mL/minute (STPD); [O2Hb] is the oxyhemoglobin concentration in mL O2 per mL blood (STPD); PcO2 is the average partial pressure of O2 in lung capillaries in Torr; M is the Haldane constant; VA is the alveolar ventilation in mL/minute (STPD); DLCO is the lung diffusing capacity of carbon monoxide in mL/minute-Torr (STPD); and PICO is the carbon monoxide partial pressure in inhaled air in Torr.

The linear form of the CFK model is acceptable for simulating low COHb concentrations (<6%; Smith 1990). However, at higher COHb levels, appreciable errors are introduced into the simulation by ignoring the interdependence of COHb and O2Hb levels that result from limited binding capacity of Hb. This interdependence can be introduced into the CFK model by replacing the constant term for [O2Hb] with the following expression relating COHb and O2Hb (Tikuisis et al. 1987b; Equation 2):

[O2Hb]=1.38Hb[COHb]
Eq. (2)
where 1.38·Hb represents the maximum binding capacity of Hb for O2 or carbon monoxide in mL O2/mL blood (STPD). The resulting nonlinear form of the CFK model is expressed in Equations 3 and 4:
d[COHb]tdt=VcoVb+1Vbβ(PICO[COHb]t1PcO2[O2Hb]t1M)
Eq. 3
β=1DLCO+PB47VA
Eq. 4
where PB is the barometric pressure (Torr) and the value 47 represents the partial pressure of water in water saturated air (Torr). Typical parameters values used in the CFK model are presented in Table 3-13. Parameter values for application of the nonlinear CFK model for predicting COHb levels in rats were reported by Benignus and Annau (1994) and are also shown in Table 3-13.

Table 3-13. Parameter Values for the Coburn-Forster-Kane (CFK) Model.

Table 3-13

Parameter Values for the Coburn-Forster-Kane (CFK) Model.

Validation of the model. Numerous studies have examined the accuracy of exposure-blood COHb relationships predicted from the CFK model (Benignus et al. 1994; Coburn et al. 1965; Hauck and Neuberger 1984; Joumard et al. 1991; Peterson and Stewart 1970, 1975; Stewart et al. 1970, 1973; Tikuisis et al. 1987a, 1987b). In general, the model performs well for predicting near equilibrium conditions; however, it performs less well in predicting the early uptake phase of carbon monoxide kinetics and COHb levels associated with rapidly changing exposures (Benignus et al. 1994; Tikuisis et al. 1987a, 1987b). The model also underestimates the arterial blood COHb level and overestimates the venous blood COHb level, an expected outcome of the simplistic assumption of a single well-mixed blood compartment. Nevertheless, the model predicts quasi-steady-state blood COHb levels that agree closely with observations (Benignus et al. 1994; Bruce and Bruce 2003; Tikuisis et al. 1987a, 1987b).

McCartney (1990) reported a sensitivity analysis of the CFK model and found that parameter value sensitivities varied considerably in magnitude at different points in the exposure simulation (Figure 3-7). At steady-state, the model is most sensitive to the values assigned to the parameters representing the input value, the inhaled carbon monoxide concentration (PICO), as well as the total Hb level (Hb) and the Haldane coefficient (M). Increases in values for any of these parameters result in higher predicted COHb levels. However, during the accumulation phase of the simulation, the predicted COHb level is also highly sensitive to alveolar ventilation rate (VA) and the lung carbon monoxide diffusing capacity (DLCO), with increases in these parameters resulting in higher COHb levels, and to the value assigned to the blood volume (Vb), with increases in the parameter value resulting in lower predicted COHb levels.

Figure 3-7. Plot of Standardized Sensitivity Coefficients for Carboxyhemoglobin (COHb) Levels Predicted from the Coburn-Forster-Kane (CFK) Model for a 100 ppm Exposure to an Individual Engaged in Vigorous Exercise (Alveolar Ventilation Rate=30 L/Minute)*.

Figure 3-7

Plot of Standardized Sensitivity Coefficients for Carboxyhemoglobin (COHb) Levels Predicted from the Coburn-Forster-Kane (CFK) Model for a 100 ppm Exposure to an Individual Engaged in Vigorous Exercise (Alveolar Ventilation Rate=30 L/Minute)*. *Each curve (more...)

Risk assessment. The CFK model has been used in establishing the National Ambient Air Quality Standard for carbon monoxide (EPA 2000) and has been used in analyses that have supported development of Threshold Limit Values for occupational safety (Tikuisis et al. 1987b).

Target tissues. The CFK model predicts COHb levels in blood.

Species extrapolation. Although originally developed to simulate blood COHb levels in humans, the model can be used to predict COHb levels in other mammalian species if physiological and chemical-specific parameter values are available for the species of interest. The latter include the Haldane coefficient (M), the binding capacity of Hb, and the lung carbon monoxide diffusing capacity (DLCO). Parameter values for the rat have been reported by Benignus and Annau (1994; Table 3-13).

Interroute extrapolation. The CFK model simulates inhalation exposures.

Smith et al. (1994) Multi-compartment Model

The nonlinear CFK model, described above, represents blood as a well-mixed compartment, an acceptable assumption for simulating near equilibrium conditions. However, during uptake and elimination phases of carbon monoxide kinetics, appreciable differences have been observed between arterial and venous blood (Benignus et al. 1994). Improved prediction of arterial COHb levels was achieved with an expansion of the CFK model to include slowly and rapidly perfused compartments and an arm compartment (representing vascular sampling sites) that includes compartments representing the major arterial and venous vasculature (Smith et al. 1994, Table 3-14; Figure 3-8).

Table 3-14. Parameter Values for the Smith et al. (1994) Model.

Table 3-14

Parameter Values for the Smith et al. (1994) Model.

Figure 3-8. Structure of the Smith et al. (1994) Model*.

Figure 3-8

Structure of the Smith et al. (1994) Model*. *Diffusive exchange carbon monoxide (CO) and oxygen (O2) occurs between lung alveolar space alveolar capillary blood. Instantaneous equilibrium is assumed between gases and blood hemoglobin, governed by the (more...)

Validation of the model. The Smith et al. (1994) models was evaluated with data collected from 15 healthy adults subjects who inhaled carbon monoxide under controlled conditions (Benignus et al. 1994). Each subject (15 adult males) inhaled for approximately 5 minutes from a bag that had an initial carbon monoxide concentration of approximately 7,000 ppm. The exposures achieved end exposure blood COHb concentrations ranging from 10 to 15%. The Smith et al. (1994) model provided improved predictions of COHb levels in samples taken from radial artery and antecubital vein.

Risk assessment. Risk assessment applications of the Smith et al. (1994) were not identified in the available literature.

Target tissues. The Smith et al. (1994) model predicts COHb levels in arterial and venous blood.

Species extrapolation. Although originally developed to simulate blood COHb levels in humans, the model could be used to predict COHb levels in other mammalian species provided that physiological and chemical-specific parameter values are available for the species of interest.

Interroute extrapolation. The Smith et al. (1994) model simulates inhalation exposures.

Hill et al. (1977) Maternal-Fetal Model

Hill et al. (1977) developed a multi-compartment model for simulating maternal-fetal transfer of carbon monoxide. This model includes simulations of maternal, placental, and fetal compartments, as well as maternal and fetal contributions to endogenous production of carbon monoxide (Table 3-15; Figure 3-9).

Table 3-15. Parameter Values for the Hill et al. (1977) Model.

Table 3-15

Parameter Values for the Hill et al. (1977) Model.

Figure 3-9. Structure of the Hill et al. (1977) Maternal-Fetal Model*.

Figure 3-9

Structure of the Hill et al. (1977) Maternal-Fetal Model*. *Average blood carboxyhemoglobin (COHb) levels in maternal and fetal tissues are calculated in the Body Gas Equilibration Model, based on inputs of PO2/O2Hb from lung, maternal tissue, placental (more...)

The model assumes instantaneous equilibrium of O2 and carbon monoxide with Hb in maternal, placental, and fetal tissues, with delivery to placental and fetal tissues governed by blood flow rates.

Validation of the model. The Hill et al. (1977) model was evaluated with data collected in experiments conducted in pregnant sheep (Longo 1977; Longo and Hill 1977). Pregnant sheep were exposed to 30, 50, or 100 ppm carbon monoxide for 24 hours or to 300 ppm for 4 hours, and maternal and fetal arterial blood was sampled from indwelling catheters for measurement of COHb levels. These exposures achieved maternal blood COHb levels ranging from approximately 4 to 24% (300 ppm). Steady-state blood COHb levels in the fetus, achieved during the exposures to 30–100 ppm, were approximately 70% higher than maternal levels. The model predicted maternal and fetal arterial blood COHb levels that agreed with observations (e.g., were within ±1 SD of observations). The model predicts slower accumulation and elimination of COHb in the human fetus compared to the maternal system, with steady-state values in fetal blood that are approximately 10–12% higher in than maternal blood. The latter prediction agrees with observations made in cord blood (Longo 1977).

Risk assessment. Risk assessment applications of the Hill et al. (1977) model were not identified in the available literature. However, Hill et al. (1977) present and discuss predictions made with the model regarding effects of various factors on maternal and fetal blood COHb levels, including air carbon monoxide concentration, air O2 concentration, altitude, exercise, fetal O2 utilization, and fetal blood flow. Of particular relevance to risk assessment is the prediction of slower accumulation and elimination of COHb in the human fetus compared to the maternal system. Based on model predictions, the elimination half-times were approximately 4 hours for maternal and 7.5 hours for fetal. Furthermore, the model predicted a more pronounced effect of breathing 100% O2 on carbon monoxide elimination from the maternal system (5-fold decrease in half-time) compared to the effect on fetal half-time (2-fold decrease).

Target tissues. The Hill et al. (1977) model predicts COHb levels in maternal and fetal blood.

Species extrapolation. The Hill et al. (1977) model has been applied to predicting maternal and fetal blood COHb levels in humans and sheep (Hill et al. 1977; Longo 1977; Longo and Hill 1977).

Interroute extrapolation. The Hill et al. (1977) model simulates inhalation exposures.

Bruce et al. (2008) Blood-Muscle Model

Bruce et al. (Bruce and Bruce 2003; Bruce et al. 2008) developed a multi-compartment model that simulates kinetics of carbon monoxide, blood COHb, and muscle COMb (Table 3-16; Figure 3-10). The model includes compartments for blood, muscle, and other non-vascular tissues. In blood, carbon monoxide exists as dissolved carbon monoxide in equilibrium with COHb. Binding of carbon monoxide and O2 to Hb is assumed to achieve instantaneous equilibrium of carbon monoxide with Hb, defined by the Haldane equation, with adjustment for cooperative binding of carbon monoxide and O2 to Hb and myoglobin. Simulation of arterial and venous blood includes a time delay to account for the time required to achieve mixing of arterial blood leaving the lung and mixed venous blood. The muscle compartment consists of two subcompartments, between which dissolved carbon monoxide exchanges by diffusion. Subcompartment 1 represents tissue that exchanges dissolved carbon monoxide with the arterial and venous vasculature (representing arterial-venous shunting). Subcompartment 2 represents tissue that exchanges with capillary blood. Dissolved carbon monoxide in muscle is in equilibrium with myoglobin, with the equilibrium defined by the Haldane equation adjusted for cooperative binding of O2. Carbon monoxide in the other tissue compartment is assumed to consist entirely of dissolved carbon monoxide.

Table 3-16. Parameter Values for the Bruce et al. (2008) Model.

Table 3-16

Parameter Values for the Bruce et al. (2008) Model.

Figure 3-10. Structure of the Bruce et al. (2008) Model*.

Figure 3-10

Structure of the Bruce et al. (2008) Model*. *The model simulates diffusive exchange of carbon monoxide (CO) and oxygen (O2) between alveolar space and capillaries and tissue capillary blood and tissues. Competitive cooperative binding of carbon monoxide (more...)

Validation of the model. The Bruce et al. (2008) model was evaluated with data collected in experiments various experiments conducted in volunteers (Benignus et al. 1994; Burge and Skinner 1995; Peterson and Stewart 1975; Tikuisis et al. 1987a). The evaluations are reported in Bruce and Bruce (2003, 2006) and Bruce et al. (2008). The model predicted observed arterial and venous blood COHb differences and kinetics during and following single or multiple brief exposures to carbon monoxide (Benignus et al. 1994). The model predicted continued uptake of carbon monoxide into muscle during the first 300 minutes following cessation of exposure, with subsequent return of muscle carbon monoxide to blood (Bruce and Bruce 2006). Blood-muscle carbon monoxide exchange, in combination with elimination of carbon monoxide from blood to exhaled air, resulted in multi-phase blood elimination kinetics, with the early faster phase dominated by transfer of carbon monoxide from blood to muscle, and the later slower phase contributed by return of carbon monoxide from muscle to blood and elimination from blood to exhaled air. The multi-phase kinetics predicted from the model agreed with post-exposure elimination kinetics of carbon monoxide from blood observed in Benignus et al. (1994). The model also predicted blood COHb kinetics observed in 1- and 5-minute exposures observed in Tikuisis et al. (1987a).

Model predictions match the CFK model when muscle diffusion is set to zero, and the model predicts lower blood COHb levels than the CFK model when the muscle carbon monoxide diffusion is enabled. This difference reflects the blood-muscle kinetics and storage of carbon monoxide in muscle myoglobin that is not accounted for in the CFK model.

Risk assessment. Risk assessment applications of the Bruce et al. (2008) model were not identified in the available literature. However, the model has been applied to predicting the outcome of breathing 100% O2 on blood COHb and muscle COMb kinetics (Bruce and Bruce 2006). The model predicted a decrease in blood and muscle elimination half-times. Of particular relevance to risk assessment is the prediction of continued uptake of carbon monoxide from blood to muscle that occurs for several hours after cessation of exposure, which contributes to slowing the elimination kinetics of carbon monoxide from the body.

Target tissues. The Bruce et al. (2008) model predicts blood COHb and muscle COMb levels and can also be used to predict blood and muscle carbon monoxide burden.

Species extrapolation. The Bruce et al. (2008) model has been applied only to predicting blood COHb and muscle COMb levels in humans. The model could be applied to other mammalian species providing that physiological and chemical-specific parameter values are available for species of interest.

Interroute extrapolation. The Bruce et al. (2008) model simulates inhalation exposures.

3.5. MECHANISMS OF ACTION

3.5.1. Pharmacokinetic Mechanisms

Pharmacokinetic mechanisms of carbon monoxide action are related to its binding to heme proteins. Kinetics and equilibria of binding to the major heme proteins that have the greatest influence on carbon monoxide pharmacokinetics (Hb and myoglobin) are discussed in the Section 3.4, Toxicokinetics.

Absorption. Inhaled carbon monoxide is rapidly and extensively absorbed into blood (Benignus et al. 1994; Burge and Skinner 1995; Peterson and Stewart 1970, 1975; Tikuisis et al. 1987b). Inhaled carbon monoxide is transported to lung alveoli as result of convective forces in the respiratory tract and diffusion. At the alveolar gas-blood interface, carbon monoxide dissolves into pulmonary capillary plasma and, from plasma, diffuses into erythrocytes and other tissues. Binding of carbon monoxide by erythrocyte Hb (to form COHb) contributes to maintaining relatively low concentrations of dissolved carbon monoxide in erythrocyte cytosol and a partial pressure gradient to drive carbon monoxide transfer from alveolar air to blood. Diffusion of carbon monoxide into erythrocytes and binding of carbon monoxide to Hb is sufficiently rapid that near-equilibrium is achieved between carbon monoxide partial pressures in alveolar air and alveolar end capillary (arterial) blood (Chakraborty et al. 2004). Continued uptake of carbon monoxide results from systemic elimination processes that lower mixed venous blood carbon monoxide concentrations returned to alveolar capillaries (Benignus et al. 1994).

Distribution. The distribution of absorbed carbon monoxide is determined, in large part, by its affinity for binding to heme. Binding of carbon monoxide to intracellular heme proteins contributes to maintaining gradients for carbon monoxide diffusion from plasma into erythrocytes and extravascular tissues. The dominant heme proteins that influence carbon monoxide distribution are erythrocyte Hb and muscle myoglobin. The distribution of carbon monoxide in the body is largely a reflection of the distribution of Hb and myoglobin, with the largest carbon monoxide burden found in blood, heart, skeletal muscle, and spleen (Vreman et al. 2005, 2006). Relatively large contributions of blood and skeletal muscle to total body carbon monoxide content is consistent with the contributions that these tissues make to total body Hb and muscle myoglobin (Bruce and Bruce 2003). In a typical adult, blood contains approximately 12 mmoles of Hb, with a total binding capacity of approximately 48 mmole carbon monoxide when completely saturated, and muscle contains approximately 8 mmoles of myoglobin, with a total binding capacity of 8 mmole carbon monoxide when completely saturated.

Metabolism. Metabolism of carbon monoxide consists of three major processes: (1) metabolic production of carbon monoxide from endogenous and exogenous precursors; (2) binding of carbon monoxide to heme proteins; and (3) oxidative metabolism of carbon monoxide to CO2.

Metabolic Production of Carbon Monoxide. Endogenous carbon monoxide is produced primarily from the enzymatic degradation of heme by the enzyme HO. Carbon monoxide is produced in all tissues that express HO; however, the major tissues contributing to carbon monoxide production are also major sites of heme metabolism, including liver, spleen, and reticuloendothelial system (Berk et al. 1974). Three isoforms of HO have been identified, of which HO-1 is inducible, whereas HO-2 and HO-3 are constitutive. In general, factors that induce HO-1 increase carbon monoxide production, whereas inhibition of HO activity or expression of HO decreases carbon monoxide production. Numerous stimuli have been shown to up-regulate HO-1 expression (Wu and Wang 2005): (1) those related to heme metabolism (e.g., heme and heme derivatives, endogenous carbon monoxide); (2) stimuli that produce or are associated with physiological or oxidative stress such as hyperoxia, heat shock factors, oxidized lipids, lipopolysaccharides, hydrogen peroxide, radiation (including ultraviolet light), endotoxin, heavy metals and arsenite, and various cytokines (e.g., interleukins, TNF-α); (3) substances that are mediators of vascular resistance (NO, angiotensin II); and (4) mitogens and other factors that regulate cell growth (e.g., growth factors, phorbol ester). Expression of the constitutive isoform, HO-2, is increased by estrogen, glucocorticoids, stimuli that inhibit HO, and stimuli that down-regulate NO production (e.g., inhibition of NO synthase). In human cells (e.g., grown in culture), expression of HO-1 is down-regulated during hypoxia, heat shock factors, and interferon-γ. The response to the above factors is not uniform across species. For example, hypoxia and heat shock have been shown to induce HO in rodents, whereas the dominant effect in human tissues is down-regulation (Nakayama et al. 2000; Shibahara et al. 2003; Wu and Wang 2005).

Carbon monoxide can be produced from oxidative metabolism of certain exogenous hydrocarbons (e.g., DCM) by isoform CYP2E1 of cytochrome P450 (Andersen et al. 1991) and from heme oxidase catalysis of products of auto-oxidation of phenols, photo-oxidation of organic compounds, and lipid peroxidation of cell membrane lipids (Rodgers et al. 1994).

Carbon Monoxide Binding to Heme Proteins. Carbon monoxide competes for O2 for binding to the iron site in heme. Kinetic constants and equilibrium constants for the binding of O2 and carbon monoxide to Hb and myoglobin are provided in Table 3-12.

Oxidative Metabolism of Carbon Monoxide. The mechanism of oxidative metabolism of carbon monoxide to CO2 is thought involve catalysis by cytrochrome c oxidase (Fenn 1970; Young and Caughey 1986).

Excretion. Absorbed carbon monoxide is eliminated from the body by exhalation and oxidative metabolism. Oxidative metabolism of carbon monoxide has been estimated to be a relatively small fraction (<10%) of endogenous carbon monoxide production (Luomanmaki and Coburn 1969). Under most conditions, the dominant route of elimination of absorbed carbon monoxide is exhalation. The mechanism of elimination of carbon monoxide by exhalation is diffusion, with the driving forces at the alveolar interface being the partial pressure difference between carbon monoxide in alveolar air and in alveolar capillary blood. The latter is maintained at a relatively low level by binding of carbon monoxide to Hb. Diffusion is also the mechanism of release of carbon monoxide from intracellular stores to blood, with the driving forces influenced by binding of carbon monoxide to extravascular heme proteins (e.g., myoglobin) and blood Hb.

3.5.2. Mechanisms of Toxicity

Carbon monoxide exerts effects on cell metabolism through both hypoxic and non-hypoxic mechanisms. Both types of effects are thought to be largely (if not entirely) the result of the ability of carbon monoxide to bind to heme and alter function and/or metabolism of heme proteins. Carbon monoxide may also act through mechanisms unrelated to heme binding, although evidence for this is currently more contentious.

Hypoxic Mechanisms. Carbon monoxide produces tissue hypoxia by binding to, and displacing O2 from, Hb. The formation of COHb decreases the O2 carrying capacity of blood and impairs release of O2 from Hb for its utilization in tissues (see Section 3.4.3, Metabolism for further discussion of kinetics of binding of carbon monoxide and O2 to Hb). Through similar mechanisms, carbon monoxide decreases O2 storage in muscle cells by binding to, and displacing O2 from, myoglobin. The principal mechanisms underlying the hypoxic mechanism of carbon monoxide are: (1) higher affinity of carbon monoxide for Hb than O2 and (2) increased binding affinity for O2 from COHb. The overall affinity of Hb for carbon monoxide is approximately 230 times greater than that for O2 (Chakraborty et al. 2004). As a result, at a partial pressure of carbon monoxide that is approximately 230-fold lower than ambient O2 concentration (21%/230≈900 ppm), O2 and carbon monoxide will bind to Hb in an equimolar ratio (i.e., COHb≈50%). Binding of carbon monoxide and O2 to the four heme moieties of Hb is cooperative (Alcantara et al. 2007; Perrella and Di Cera 1999; Roughton 1970). The associative reaction rate (ka) becomes faster with successive additions of carbon monoxide or O2, increasing the affinity (Ka) of Hb for O2 and impairing the release of O2 from Hb for utilization in tissues. Overall, the affinity increase is 588-fold due to cooperative events as more carbon monoxide binds to hemoglobin; however, this is not a monotonic function with each ligation. Once the first hemoglobin chain binds a carbon monoxide molecule, the affinity increases 16.5-fold when the second carbon monoxide molecule binds, 4.6-fold at the third ligation step, and 7.7-fold in the fourth ligation step (Perrella and Di Cera 1999).

Although all tissues are vulnerable to carbon monoxide-induced hypoxic injury, those having the highest O2 demand are particularly vulnerable, including the brain and heart. Carbon monoxide-induced hypoxia triggers compensatory cardiovascular responses that include increased cardiac output (heart rate and stroke volume) and dilation of cardiac and cerebral vasculature. During exercise, cardiac work and O2 consumption increase and the vulnerability of the heart to carbon monoxide-induced hypoxic injury increases. In patients who have underlying coronary artery or myocardial disease, the increased O2 demand of exercise, in combination with decreased O2 carrying capacity of blood and impaired release of O2 from Hb, can produce cardiac toxicity (e.g., myocardial ischemia, exertional angina, arrhythmia) at lower carbon monoxide exposure levels than experienced in healthy individuals (Adams et al. 1988; Allred et al. 1989, 1991; Anderson et al. 1973; Kleinman et al. 1989, 1998; Leaf and Kleinman 1996b).

Brain hypoxia induced by carbon monoxide can result in various symptoms of impaired central nervous system function, including headache, dizziness, drowsiness, weakness, nausea, vomiting, confusion, disorientation, irritability, visual disturbances, convulsions, and coma (Dolan 1985; Ernst and Zibrak 1998). Delayed development of neuropsychiatric and neurological impairment may occur within 1–4 weeks of exposure, with symptoms including inappropriate euphoria, impaired judgment, poor concentration, memory loss, cognitive and personality changes, psychosis, and Parkinsonism (Choi 2002; Ernst and Zibrak 1998; Kao and Nañagas 2006; Klawans et al. 1982; Raub and Benignus 2002; Raub et al. 2000; Ringel and Klawans 1972; WHO 1999; Wolf et al. 2008).

Non-Hypoxic Mechanisms. Most of the non-hypoxic mechanisms of action of carbon monoxide have been attributed to binding of carbon monoxide to heme in proteins other than Hb. A list of heme proteins that have been shown to undergo functional modification in response to carbon monoxide is presented in Table 3-17. Notable targets of carbon monoxide include components of several important physiological regulatory systems, including brain and muscle O2 storage and utilization (myoglobin, neuroglobin), nitric oxide NO cell signaling pathway (e.g., nitric oxide synthase [NOS], guanylyl cyclase [GC]), prostaglandin cell signaling pathway (cyclooxygenase, prostaglandin H synthase), energy metabolism and mitochondrial respiration (cytochrome c oxidase, cytochrome c, NADPH oxidase), steroid and drug metabolism (cytochrome P450), cellular redox balance and ROS (catalase, peroxidases), and various transcription factors (e.g., neuronal PAS domain protein, NPAS2, implicated in regulation of circadian rhythm).

Table 3-17. Heme Proteins Modulated by Carbon Monoxide.

Table 3-17

Heme Proteins Modulated by Carbon Monoxide.

Endogenously produced carbon monoxide may participate in the physiological regulation of some, if not all, of these systems. This has potentially important implications for the understanding of carbon monoxide toxicology and dose-response relationships for the following reasons: (1) carbon monoxide modulation of physiological processes may underlie some aspects of the toxicity of exogenous carbon monoxide; (2) exogenous carbon monoxide may disrupt physiological regulation of those systems that are responsive to endogenous carbon monoxide (e.g., vascular resistance); and (3) exposures to exogenous carbon monoxide may affect carbon monoxide-mediated physiological responses at levels that approach those resulting from endogenous production. Whole-body endogenous carbon monoxide production rate has been estimated to be approximately 16.4 µmol/hour (0.007 mL/minute STPD; Coburn 1970a). Local rates of production of carbon monoxide in tissues are influenced by many tissue-specific factors, including those that regulate intracellular heme oxygenase activity (e.g., some areas of brain have relatively high HO-2 activities). A similar whole-body rate of delivery of carbon monoxide to the alveolar region of the lung would be achieved by inhaling air at a carbon monoxide concentration of approximately 1–2 ppm (i.e., assuming that alveolar ventilation rate is approximately 4,000 mL/minute; 0.007 mL/minute/4,000 mL/minute=1.75·10–6 mL CO/mL air, STPD=1.75 ppm). Such levels are only modestly above national average ambient air carbon monoxide concentrations in the United States (approximately 0.5 ppm, see Chapter 6) and well within variability observed in measurements made with personnel air monitors (Chang et al. 2000). One implication of this is that, if a dose threshold for effects of exogenous carbon monoxide exists at all, it may lie near or below ambient air carbon monoxide concentrations. Although currently available toxicology and epidemiological studies provide evidence for increasing severity of adverse carbon monoxide effects in association with increasing levels of exposure, these studies do not provide strong evidence supporting or refuting a dose threshold for adverse effects.

Of the many physiological systems that may contribute to non-hypoxic mechanisms of carbon monoxide action, the following are thought to be particularly important and are the subject on ongoing intense research interests: modulation of cell signaling pathways and generation of ROS.

Carbon Monoxide Modulation of Cell Signaling and Ion Channel Activity. Cell signaling pathways that are thought to be modulated by carbon monoxide are depicted in Figure 3-11. These pathways contribute to the regulation of various physiological process, including inflammatory responses, apoptosis, cell proliferation, thrombosis (i.e., platelet aggregation), smooth muscle tone (including vasodilation and relaxation of extravascular smooth muscle), and synaptic neurotransmission in the central and peripheral nervous systems (e.g., glutamate). The nitric oxide radical (NO•) plays an important signal transduction role in regulating smooth muscle relaxation, synaptic neurotransmission (e.g., glutamate), platelet function, and other physiological processes (Denninger and Marletta 1999). One of the major transduction pathways mediated by NO• is the NO•-cGMP pathway, in which NO• produced by the enzyme NOS binds to the heme moiety of soluble guanylate cyclase (sGC), which catalyzes conversion of guanosine triphosphate (GTP) to cyclic guanosine monophosphate (cGMP). The latter (cGMP) activates various downstream signaling cascades that regulate cell function (e.g., cGMP-dependent protein kinase, cGMP-regulated cation channels, and cGMP-regulated phosphodiesterase). Carbon monoxide binds to the heme moieties of both NOS and sGC, and in doing so, inhibits NOS activity and activates sGC activity (Stone and Marletta 1994; Thorup et al. 1999). The potency of carbon monoxide for activating sGC is approximately 20-fold less than that of NO•; as a result, in the presence of NO•, carbon monoxide may modulate sGC activity downward by displacing NO• from sGC (Stone and Marletta 1994). The NO•-cGMP signaling pathway has been implicated in the mechanism by which carbon monoxide produces relaxation of smooth muscle (Ny et al. 1996) and inhibition of platelet homotypic aggregation (Brune and Ullrich 1987). Carbon monoxide increases platelet-neutrophil interactions (heterotypic aggregation) through a free-radical mechanism (Thom et al. 2006).

Figure 3-11. Potential Signaling Pathways Modulated by Carbon Monoxide.

Figure 3-11

Potential Signaling Pathways Modulated by Carbon Monoxide. + = stimulates pathway; – = inhibits pathway; cGMP = cyclic guanosine monophosphate; CO = carbon monoxide; COX = cylcooxygenase; CRF = corticotropin releasing factor; CNS = central nervous (more...)

Carbon monoxide modulates mitogen activated protein kinase (MAPK) signaling pathways, either through its interaction with sGC (Ryter et al. 2006) or through secondary effects on ROS generation (see below). MAPK pathways are important regulators of cytokine production and release in inflammatory responses and in regulating apoptosis and proliferation.

Carbon monoxide exposure can perturb heme protein binding by nitric oxide; induce a variety of proteins including heme oxygenase (HO), superoxide dismutase, and nitric oxide synthase; alter mitochondrial function; and modify production of ATP and ROS (Brown and Piantadosi 1992; Davutoglu et al. 2006; Kim et al. 2007; Lee et al. 2006; Maulik et al. 1996; Nakao et al. 2008; Piantadosi et al. 1997; Thom et al. 1994, 1997, 2000; Zuckerbraun et al. 2003). Carbon monoxide increases production of ROS from mitochondria and also some non-mitochondrial sources, which can activate one or more protein kinases (e.g., p38 MAPK, c-Jun N-terminal kinase, ERK), signal transducers (e.g., peroxisome proliferator-activated receptor-γ), and transcriptional regulators (e.g., nuclear factor-κB, AP-1), as well as guanylate cyclase, and lead to secondary events such as vasodilation and fibrinolysis (Arruda et al. 2004; Fujita et al. 2001; Mishra et al. 2006; Ryter and Otterbein 2004; Soares et al. 2004; Wijayanti et al. 2004).

Carbon monoxide also binds to the heme moieties and inhibits cytochrome P450 and cyclooxygenase (COX) and, thereby, can modify physiological processes regulated by products of these enzyme systems. Inhibition of cytochrome P450 catalyzed production of eciosanoids (e.g., 20-hydroxyeicostetraenoic acid, [20-HETE]) and COX-mediated production of prostaglandins have been postulated as a potential mechanisms by which carbon monoxide may contribute to the regulation of vascular smooth muscle tone and vasodilation (Botros et al. 2002; Volti et al. 2008).

Carbon monoxide has been shown to modulate activity of several ion channels, including calcium-sensitive and voltage-gated K+ channels, cardiac L-type and intestinal smooth muscle L-type Ca2+ channels, and epithelial Na+ channels (Peers 2011; Peers and Steel 2011; Peers et al. 2009). Carbon monoxide may augment or inhibit channel activity, depending on ion channel type and carbon monoxide concentration. Several mechanisms for ion channel modulation have been proposed, including increased production of ROS (cardiac L-type Ca2+ channels and voltage-gated K+ channels), protein kinase G activation (K+ channels), and increased nitric oxide and cGMP (intestinal smooth muscle L-type Ca2+ channels).

Carbon Monoxide Modulation of ROS Generation. Carbon monoxide may promote and/or modify ROS generation through several different mechanisms. Binding to the heme moiety and inhibition of mitochondrial cytochrome c oxidase disrupts mitochondrial respiration and promotes ROS generation (Alonso et al. 2003; D'Amico et al. 2006; Favory et al. 2006a; Iheagwara et al. 2007; Zuckerbraun et al. 2007). Carbon monoxide also appears to promote a more general disruption of the regulation of iron and ferritin levels, at least in part through promotion of heme protein degradation (Cronje et al. 2004; Ghio et al. 2008; Iheagwara et al. 2007).

3.5.3. Animal-to-Human Extrapolations

Studies conducted in nonhuman primates, dogs, and various rodent species have demonstrated the major health effects of carbon monoxide that have been observed in humans, including COHb-related hypoxia, cardiovascular responses, and neurological effects (Benignus 1994; Benignus et al. 1990; EPA 1991, 2000; Raub and Benignus 2002). These studies also provide support for the concept of non-hypoxic modes of action of carbon monoxide that operate on a variety of physiological systems (e.g., Wu and Wang 2005; also, see Section 3.5.2 for discussion and relevant references). These systems include physiological regulatory systems that function in all mammalian species, such as brain and muscle oxygen storage and utilization; nitric oxide cell signaling pathway; prostaglandin cell signaling pathway; energy metabolism and mitochondrial respiration; steroid and drug metabolism; cellular redox balance and ROS; and various transcription factors. Therefore, results from mechanistic studies conducted in typical laboratory animal models are likely to be directly relevant to understanding mechanisms of action of carbon monoxide in humans. Manifestations of these various modes of action may differ qualitatively and quantitatively across species. For example, rats exhibit pronounced hypothermia in response to exposures to carbon monoxide, whereas at low levels of exposure (e.g., COHb <20%), this response is not evident in humans (Benignus 1994; Gordon 1990). Differences in dose-response relationships between humans and animal models can also be expected, in part due to differences in physiological and chemical determinants of carbon monoxide uptake, elimination, and binding kinetics with major heme proteins, (Benignus and Annau 1994; Longo and Hill 1977) as well as interspecies differences in physiological regulation of systems that are non-hypoxic targets of carbon monoxide.

3.6. TOXICITIES MEDIATED THROUGH THE NEUROENDOCRINE AXIS

Recently, attention has focused on the potential hazardous effects of certain chemicals on the endocrine system because of the ability of these chemicals to mimic or block endogenous hormones. Chemicals with this type of activity are most commonly referred to as endocrine disruptors. However, appropriate terminology to describe such effects remains controversial. The terminology endocrine disruptors, initially used by Thomas and Colborn (1992), was also used in 1996 when Congress mandated the EPA to develop a screening program for “…certain substances [which] may have an effect produced by a naturally occurring estrogen, or other such endocrine effect[s]…”. To meet this mandate, EPA convened a panel called the Endocrine Disruptors Screening and Testing Advisory Committee (EDSTAC), and in 1998, the EDSTAC completed its deliberations and made recommendations to EPA concerning endocrine disruptors. In 1999, the National Academy of Sciences released a report that referred to these same types of chemicals as hormonally active agents. The terminology endocrine modulators has also been used to convey the fact that effects caused by such chemicals may not necessarily be adverse. Many scientists agree that chemicals with the ability to disrupt or modulate the endocrine system are a potential threat to the health of humans, aquatic animals, and wildlife. However, others think that endocrine-active chemicals do not pose a significant health risk, particularly in view of the fact that hormone mimics exist in the natural environment. Examples of natural hormone mimics are the isoflavinoid phytoestrogens (Adlercreutz 1995; Livingston 1978; Mayr et al. 1992). These chemicals are derived from plants and are similar in structure and action to endogenous estrogen. Although the public health significance and descriptive terminology of substances capable of affecting the endocrine system remains controversial, scientists agree that these chemicals may affect the synthesis, secretion, transport, binding, action, or elimination of natural hormones in the body responsible for maintaining homeostasis, reproduction, development, and/or behavior (EPA 1997). Stated differently, such compounds may cause toxicities that are mediated through the neuroendocrine axis. As a result, these chemicals may play a role in altering, for example, metabolic, sexual, immune, and neurobehavioral function. Such chemicals are also thought to be involved in inducing breast, testicular, and prostate cancers, as well as endometriosis (Berger 1994; Giwercman et al. 1993; Hoel et al. 1992).

No studies were located regarding endocrine disruption in humans and/or animals after exposure to carbon monoxide. Several cell signaling pathways that are thought to be modulated by endogenous carbon monoxide that could play a role in the regulation of activity of the hypothalamic-pituitary-adrenal (HPA) axis include cGMP and NO• signaling (Mancuso et al. 1997, 2010; Snyder et al. 1998; Wu and Wang 2005). Modulation of endogenous carbon monoxide levels and/or production has been shown to modify various HPA axis processes, including release of arginine vasopressin (AVP), gonadotropin, and corticotropin releasing hormone from the hypothalamus, release of adrenocorticotropic hormone (ACTH) from the pituitary, and response of the adrenal gland to ACTH (Wu and Wang 2005). The toxicological significance of these various actions of carbon monoxide has not been established.

3.7. CHILDREN’S SUSCEPTIBILITY

This section discusses potential health effects from exposures during the period from conception to maturity at 18 years of age in humans, when all biological systems will have fully developed. Potential effects on offspring resulting from exposures of parental germ cells are considered, as well as any indirect effects on the fetus and neonate resulting from maternal exposure during gestation and lactation. Relevant animal and in vitro models are also discussed.

Children are not small adults. They differ from adults in their exposures and may differ in their susceptibility to hazardous chemicals. Children’s unique physiology and behavior can influence the extent of their exposure. Exposures of children are discussed in Section 6.6, Exposures of Children.

Children sometimes differ from adults in their susceptibility to hazardous chemicals, but whether there is a difference depends on the chemical (Guzelian et al. 1992; NRC 1993). Children may be more or less susceptible than adults to health effects, and the relationship may change with developmental age (Guzelian et al. 1992; NRC 1993). Vulnerability often depends on developmental stage. There are critical periods of structural and functional development during both prenatal and postnatal life, and a particular structure or function will be most sensitive to disruption during its critical period(s). Damage may not be evident until a later stage of development. There are often differences in pharmacokinetics and metabolism between children and adults. For example, absorption may be different in neonates because of the immaturity of their gastrointestinal tract and their larger skin surface area in proportion to body weight (Morselli et al. 1980; NRC 1993); the gastrointestinal absorption of lead is greatest in infants and young children (Ziegler et al. 1978). Distribution of xenobiotics may be different; for example, infants have a larger proportion of their bodies as extracellular water, and their brains and livers are proportionately larger (Altman and Dittmer 1974; Fomon 1966; Fomon et al. 1982; Owen and Brozek 1966; Widdowson and Dickerson 1964). The infant also has an immature blood-brain barrier (Adinolfi 1985; Johanson 1980) and probably an immature blood-testis barrier (Setchell and Waites 1975). Many xenobiotic metabolizing enzymes have distinctive developmental patterns. At various stages of growth and development, levels of particular enzymes may be higher or lower than those of adults, and sometimes unique enzymes may exist at particular developmental stages (Komori et al. 1990; Leeder and Kearns 1997; NRC 1993; Vieira et al. 1996). Whether differences in xenobiotic metabolism make the child more or less susceptible also depends on whether the relevant enzymes are involved in activation of the parent compound to its toxic form or in detoxification. There may also be differences in excretion, particularly in newborns who all have a low glomerular filtration rate and have not developed efficient tubular secretion and resorption capacities (Altman and Dittmer 1974; NRC 1993; West et al. 1948). Children and adults may differ in their capacity to repair damage from chemical insults. Children also have a longer remaining lifetime in which to express damage from chemicals; this potential is particularly relevant to cancer.

Certain characteristics of the developing human may increase exposure or susceptibility, whereas others may decrease susceptibility to the same chemical. For example, although infants breathe more air per kilogram of body weight than adults breathe, this difference might be somewhat counterbalanced by their alveoli being less developed, which results in a disproportionately smaller surface area for alveolar absorption (NRC 1993).

The fetus may be particularly vulnerable to maternal carbon monoxide exposure. Carbon monoxide in the maternal system distributes to fetal tissues. Measurements of steady-state fetal COHb concentrations in fetal and maternal blood of nonsmoking women have found fetal COHb concentrations to be approximately 10–15% higher than maternal blood (fetal/maternal ratio=1.1–1.15; Longo 1977). Binding affinity of fetal Hb is approximately twice that of maternal Hb (Di Cera et al. 1989). The higher binding affinity of fetal Hb, as well as the relatively small diffusion gradients for carbon monoxide between maternal and fetal blood, contribute to slower kinetics of COHb in fetal blood compared to maternal blood. As a result, with continuous exposure to a constant level of carbon monoxide in air, the kinetics of fetal blood COHb concentration would be expected to lag behind maternal kinetics and to reach a steady-state that is approximately 10–15% higher than maternal (Longo 1977). Similarly, following cessation of exposure, elimination kinetics of carbon monoxide from fetal blood would be expected to lag behind that of maternal blood (Hill et al. 1977).

Epidemiological studies have examined possible associations between exposure to ambient air carbon monoxide concentrations and various developmental outcomes, including pre-term birth, reduced birth weight, congenital anomalies, and neonatal and infant death (see Section 3.2.6, Table 3-8). Results of these studies have been mixed and, collectively, do not provide strong evidence for developmental effects, including neonatal or infant mortality in association with exposures to ambient levels of carbon monoxide (<10 ppm). However, studies conducted in animals provide evidence of adverse developmental effects of gestational and early postnatal carbon monoxide in association with maternal exposure concentrations as low as 12–25 ppm. Observed effects have included decreased fetal weight, adverse central nervous system development, altered peripheral nervous system development, cardiac effects, altered sexual behavior, immunological effects, and hematological effects (see Table 3-9). In addition, some studies showed that developmental effects persisted beyond the postnatal period (Carratu et al. 1993, 2000a, 2000b; De Salvia et al. 1995; Lopez et al. 2003, 2008; Stockard-Sullivan et al. 2003; Webber et al. 2003).

Although no studies have directly compared the sensitivity of adults and children to the effects of carbon monoxide, several epidemiological studies have examined pulmonary function, asthma morbidity, and exacerbation of asthma symptoms in children (Chen et al. 1999; Fischer et al. 2002; Mortimer et al. 2008; Yu et al. 2000). Although results have been mixed, these studies provide some evidence for increased vulnerability of asthmatic children to carbon monoxide-associated respiratory effects (see Section 3.2.2, Respiratory Effects).

3.8. BIOMARKERS OF EXPOSURE AND EFFECT

Biomarkers are broadly defined as indicators signaling events in biologic systems or samples. They have been classified as markers of exposure, markers of effect, and markers of susceptibility (NAS/NRC 1989).

The National Report on Human Exposure to Environmental Chemicals provides an ongoing assessment of the exposure of the U.S. population to environmental chemicals using biomonitoring. This report is available at http://www.cdc.gov/exposurereport/. The biomonitoring data for carbon monoxide from this report is discussed in Section 6.5. A biomarker of exposure is a xenobiotic substance or its metabolite(s) or the product of an interaction between a xenobiotic agent and some target molecule(s) or cell(s) that is measured within a compartment of an organism (NAS/NRC 1989). The preferred biomarkers of exposure are generally the substance itself, substance-specific metabolites in readily obtainable body fluid(s), or excreta. However, several factors can confound the use and interpretation of biomarkers of exposure. The body burden of a substance may be the result of exposures from more than one source. The substance being measured may be a metabolite of another xenobiotic substance (e.g., high urinary levels of phenol can result from exposure to several different aromatic compounds). Depending on the properties of the substance (e.g., biologic half-life) and environmental conditions (e.g., duration and route of exposure), the substance and all of its metabolites may have left the body by the time samples can be taken. It may be difficult to identify individuals exposed to hazardous substances that are commonly found in body tissues and fluids (e.g., essential mineral nutrients such as copper, zinc, and selenium). Biomarkers of exposure to carbon monoxide are discussed in Section 3.8.1.

Biomarkers of effect are defined as any measurable biochemical, physiologic, or other alteration within an organism that, depending on magnitude, can be recognized as an established or potential health impairment or disease (NAS/NRC 1989). This definition encompasses biochemical or cellular signals of tissue dysfunction (e.g., increased liver enzyme activity or pathologic changes in female genital epithelial cells), as well as physiologic signs of dysfunction such as increased blood pressure or decreased lung capacity. Note that these markers are not often substance specific. They also may not be directly adverse, but can indicate potential health impairment (e.g., DNA adducts). Biomarkers of effects caused by carbon monoxide are discussed in Section 3.8.2.

A biomarker of susceptibility is an indicator of an inherent or acquired limitation of an organism's ability to respond to the challenge of exposure to a specific xenobiotic substance. It can be an intrinsic genetic or other characteristic or a preexisting disease that results in an increase in absorbed dose, a decrease in the biologically effective dose, or a target tissue response. If biomarkers of susceptibility exist, they are discussed in Section 3.10, Populations That Are Unusually Susceptible.

3.8.1 Biomarkers Used to Identify or Quantify Exposure to Carbon Monoxide

Measurement of blood COHb is the principal biomarker for identifying exposure to carbon monoxide. Nearly all carbon monoxide in the body exists as complexes with Hb and other heme proteins. Approximately 60–70% of the body burden exists as blood COHb (Bruce and Bruce 2003; Vreman et al. 2006). Therefore, measurement of COHb provides a useful indicator of the elevated carbon monoxide body burden, although the utility for blood COHb measurements for large-scale screening for elevated exposures or poisoning is questionable (Heckerling et al. 1990). The relationship between COHb levels and exposure is more complex because of the numerous physiological factors that influence carbon monoxide uptake and elimination. The elimination half-time of absorbed carbon monoxide has been estimated to be approximately 300 minutes (Bruce and Bruce 2006; Peterson and Stewart 1970); therefore, a value for COHb measured at a given time following exposure will reflect both the exposure and elimination rate of carbon monoxide. Several pharmacokinetic models have been developed that can be used to predict blood COHb levels that would correspond to a given exposure concentration and duration in a typical adult and in the fetus resulting from maternal exposures (see Section 3.4.5). These models can be used to reconstruct possible carbon monoxide exposure scenarios (e.g., concentrations, duration and gap between measurement of COHb and cessation of exposure) that could result in a given measured COHb level.

3.8.2 Biomarkers Used to Characterize Effects Caused by Carbon Monoxide

Although blood COHb reflects current carbon monoxide body burden, measurement of blood COHb has not been shown to be a reliable predictor of severity of acute toxicity (Hampson and Hauff 2008). Furthermore, due to the time elapsed between cessation of CO exposure to measurement of COHb, coupled with effects of medical intervention (see Section 3.11, Methods Reducing Toxic Effects), blood COHb does not provide reliable estimates of the level of CO exposure. Typical levels of COHb in nonsmokers are <2% (Adams et al. 1988; Allred et al. 1991; Anderson et al. 1973; Hinderliter et al. 1989; Kleinman et al. 1989, 1998; Sheps et al. 1987, 1990). Levels of ≥10% can be observed immediately after smoking a cigarette (Kao and Nañagas 2006). In general, signs and symptoms of acute carbon monoxide poisoning can present at COHb levels ranging from 3 to 24% (Hampson and Hauff 2008; Kao and Nañagas 2006). Levels ranging from 2 to 6% have been shown to exacerbate underlying cardiovascular disease, including enhanced myocardial ischemia and increased cardiac arrhythmias (Adams et al. 1988; Allred et al. 1989, 1991; Anderson et al. 1973; Kleinman et al. 1989, 1998; Leaf and Kleinman 1996b). More severe signs of carbon monoxide poisoning are poorly correlated with blood COHb, with loss of consciousness occurring at mean levels of 24.3% (range: 2–70%) and fatality at mean levels of 32.1% (range: 3.0–60%; Hampson and Hauff 2008). Exposures that result in COHb levels >50% are usually fatal (Kao and Nañagas 2006).

Diagnosis and characterization of carbon monoxide toxicity include both the measurement of blood COHb and assessment of signs. In order of increasing severity, these include: (1) headache, nausea, dilation of cutaneous vasculature, vomiting, dizziness, and blurred vision; (2) confusion, syncope, chest pain, dyspnea, weakness, tachycardia, and tachynea rhabdomyolysis; and (3) palpitations, cardiac dysrhythmias, hypotension, myocardial ischemia, cardiac arrest, respiratory arrest, pulmonary edema, seizures, and coma (Kao and Nañagas 2006). Neuropsychological testing and neuroimaging have been considered for evaluating central nervous system effects (Tomaszewski 2006).

Cardiac enzyme markers are associated with myocardial dysfunction and an elevated risk for long-term cardiac mortality following carbon monoxide poisoning; biochemical markers for brain injury, such as neuron-specific enolase and S-100 beta protein, have not been found to reliably correlate with severity of poisoning or clinical outcome (Brvar et al. 2004; Davutoglu et al. 2006; Kalay et al. 2007; Rasmussen et al. 2004; Satran et al. 2005).

3.9. INTERACTIONS WITH OTHER CHEMICALS

The effects of combined exposure of carbon monoxide with other chemicals (e.g., pollutants, drugs) were reviewed by the EPA (1991, 2000). In general, research has mainly focused on effects of combined chemical exposures on the cardiovascular, respiratory, and central nervous systems. As discussed in Section 3.10 (Populations that Are Unusually Susceptible), in addition to interactions with other chemicals, exposure to carbon monoxide at a higher altitude may result in higher body burdens of carbon monoxide and, therefore, there may be higher vulnerability to carbon monoxide at higher altitudes than at lower altitudes (McGrath et al. 1993), in particular for people who have not physiologically adapted to higher altitudes (Horvath et al. 1988; Hsia 2002).

Oxygen. The chemical interaction of highest toxicological and clinical significance is that between carbon monoxide and O2. Carbon monoxide and O2 compete for binding to heme (Chakraborty et al. 2004; Gibson et al. 1986). Binding to, and displacing O2 from, heme proteins, including blood Hb and muscle myoglobin, has profound effects on the kinetics of distribution and elimination kinetics of both carbon monoxide and O2, and is also the principal mechanism underling the hypoxic effects of carbon monoxide (Bruce and Bruce 2006; Bruce et al. 2008; Vreman et al. 2005, 2006). Decreasing levels of O2 in inspired air facilitates carbon monoxide binding to Hb and myoglobin and exacerbates carbon monoxide-induced hypoxia. This interaction can contribute to a higher vulnerability to carbon monoxide toxicity at lower O2 pressures that occur at higher altitudes (Horvath et al. 1988; Hsia 2002; McGrath et al. 1993). Increasing levels of O2 in inspired air facilitates displacement of carbon monoxide from Hb and myoglobin, increases the rate of carbon monoxide elimination from the body, and increases O2 availability to tissues (Landaw 1973; Peterson and Stewart 1970; Weaver et al. 2000). These actions of O2 form the basis for the use of 100% O2 and hyperbaric O2 therapy in the treatment of carbon monoxide poisoning (Lavonas 2007; Tomaszewki 2006).

Chemicals that Bind to Hemoglobin. Several nitrate compounds, including dinitrotoluenes (Ellis et al. 1985; U.S. Army 1979) and nitroanilines (NTP 1993), bind to hemoglobin and induce anemia. Although studies evaluating the interaction between CO and other chemicals that bind hemoglobin were not identified, any chemical that binds to hemoglobin, and thereby reduces the oxygen carrying capacity of the blood, could potentiate the effects of CO.

Cyanide. Carbon monoxide and cyanide poisonings can occur concomitantly in victims of smoke inhalation (Lundquist et al. 1989; Shusterman et al. 1996; Wetherell 1966). Experimental studies conducted in animal models provide evidence that carbon monoxide and cyanide can produce synergistic toxicity (Moore et al. 1987; Norris et al. 1986; Pitt et al. 1979).

Drugs Affecting the Central Nervous System. Due to effects of carbon monoxide on the central nervous system, it is anticipated that combined exposure to drugs with activity in the central nervous system, including therapeutic agents and drugs of abuse, could enhance the central nervous system toxicity of carbon monoxide (EPA 1991, 2000). However, the available data are not sufficient to determine if effects of combined exposures are additive or exhibit other types of interaction relationships (e.g., synergistic). Enhanced carbon monoxide-induced central nervous system toxicity has been reported to occur with concomitant exposures to alcohol, barbiturates, amphetamine, chlorpromazine, nicotine, diazepam, and morphine (EPA 1991).

Drugs Affecting the Cardiovascular System. In response to carbon monoxide-induced tissue hypoxia, alterations in hemodynamics and responsive compensatory mechanisms (e.g., vasodilation and increased cardiovascular output) occur, including increased coronary blood flow, decreased myocardial O2 consumption, increased heart rate, and alterations in blood flow to various vascular beds (e.g., cerebral, limb muscular). Thus, drugs used to treat patients with coronary artery disease (e.g., beta-blockers, calcium-channel blockers, nitrates) potentially may affect susceptibility to carbon monoxide. This is of particular concern, as patients with cardiovascular disease are more susceptible to carbon monoxide-induced cardiovascular toxicity than are healthy individuals. However, little data evaluating potential interactions with therapeutic agents used to treat cardiovascular disease are available (EPA 1991, 2000).

Air Pollution. Numerous epidemiological studies have examined associations between exposure to carbon monoxide and other air pollutants (e.g., NO2, O3, particular matter, SO2). In general, these studies have found evidence for multiple air pollutant contributors to outcomes observed, including cardiovascular, respiratory, and mortality end points (see discussion of epidemiology in Section 3.2). The interactive aspects of these associations have not been fully explored or elucidated. However, evidence is provided from several studies of increased risks of adverse cardiovascular and respiratory outcomes in the context of ongoing cardiovascular disease (e.g., ischemic heart disease) and respiratory diseases (e.g., asthma). These studies suggest that it is both plausible and likely that interactions contribute to outcomes observed in populations that experience co-exposures to carbon monoxide, NO2, O3, particular matter, and SO2.

Tobacco Smoke. In addition to exposures to carbon monoxide, tobacco smoking results in exposures to a variety of agents that act directly on the heart to increase cardiac O2 demand (e.g., nicotine) (Benowitz 1997) and that contribute to respiratory and coronary vascular disease, both of which can render people more vulnerable to carbon monoxide toxicity (see Section 3.10, Populations that are Unusually Susceptible).

Noise. Studies conducted in animals provide evidence that acute exposure to carbon monoxide potentiates noise-induced hearing loss, including noise-induced elevation of compound action potential threshold (Chen and Fechter 1999; Mortazavi et al. 2010) and auditory threshold shifts (Fechter et al. 1988; Young et al. 1987). However, in these studies, exposure to carbon monoxide alone did not produce changes in auditory function. Additional details are provided in Section 3.2.4.

3.10. POPULATIONS THAT ARE UNUSUALLY SUSCEPTIBLE

A susceptible population will exhibit a different or enhanced response to carbon monoxide than will most persons exposed to the same level of carbon monoxide in the environment. Reasons may include genetic makeup, age, health and nutritional status, and exposure to other toxic substances (e.g., cigarette smoke). These parameters result in reduced detoxification or excretion of carbon monoxide, or compromised function of organs affected by carbon monoxide. Populations who are at greater risk due to their unusually high exposure to carbon monoxide are discussed in Section 6.7, Populations with Potentially High Exposures.

Cardiovascular Disease. Studies of the cardiovascular effects of inhalation exposures to carbon monoxide conducted in controlled human clinical studies, epidemiology studies, and various animal models (monkeys, dogs, rats, and rabbits) provide convincing evidence for adverse cardiovascular effects in association with carbon monoxide exposures that result in blood COHb levels of ≥2.4%, with effects occurring at the lowest levels in subjects who have compromised cardiovascular function (e.g., coronary artery disease). Therefore, exposure to carbon monoxide appears to exacerbate underlying cardiovascular disease, and people with cardiovascular disease that renders them vulnerable to myocardial ischemia (e.g., coronary artery disease) represent a population that would be unusually susceptible to carbon monoxide. Epidemiological studies of exposure to ambient concentrations of carbon monoxide and cardiovascular outcomes also provide supporting evidence for higher vulnerability of people with pre-existing heart conditions, such as ischemic heart disease, people with a history of cardiac arrhythmia, and the elderly (Barnett et al. 2006; D’Ippoliti et al. 2003; Dockery et al. 2005; Hosseinpoor et al. 2005; Lanki et al. 2006; Lee et al. 2003b; Mann et al. 2002; Rich et al. 2005; Szyszkowicz 2007; von Klot et al. 2005).

Asthma and COPD. Several studies have examined possible associations between ambient air carbon monoxide concentrations and pulmonary function in subjects who had ongoing lung disease (e.g., asthma, chronic obstructive lung disease) and who might be more sensitive to agents that affect pulmonary function (Rabinovitch et al. 2004; Silkoff et al. 2005; Timonen et al. 2002), as well as asthma morbidity and exacerbation of asthma symptoms (Park et al. 2005a; Rabinovitch et al. 2004; Schildcrout et al. 2006; Silkoff et al. 2005; Slaughter et al. 2003; von Klot et al. 2002; Yu et al. 2000). Results of these studies have been mixed; however, collectively, these studies provide evidence for higher vulnerability of people who have ongoing respiratory disease, including asthma and COPD. This subpopulation may also include a subpopulation of former tobacco smokers (Silkoff et al. 2005). As noted in Section 3.7, Children’s Susceptibility, the fetus may be particularly vulnerable to exposures that occur during pregnancy.

Anemia. Conditions that produce anemia (e.g., hemolysis, hemorrhage, reduced hematopoiesis, or iron, vitamin B-6, folate, and/or vitamin B-12 deficiency) decrease the oxygen carrying capacity of blood. Hemolytic anemia increases heme metabolism and endogenous carbon monoxide production. These factors would be expected to render individuals more vulnerable to hypoxic effects of carbon monoxide.

Environmental and Behavioral Factors. In addition to ongoing diseases, other environmental and/or behavior factors may increase vulnerability to carbon monoxide toxicity. In general, factors that increase alveolar ventilation and cardiac output, while maintaining alveolar ventilation perfusion matching (i.e., proportional increases in alveolar ventilation rate and cardiac output), tend to increase absorption of inhaled carbon monoxide. Factors that increase carbon monoxide absorption include moderate exercise, supine position, age (increased in infancy and childhood and declines in adults with age), altitude, increasing blood Hb concentration, and decreasing partial pressure of O2 in inhaled air or blood.

These factors could contribute to higher body burdens of carbon monoxide and, therefore, higher vulnerability to carbon monoxide at higher altitudes than at lower altitudes (McGrath et al. 1993), in particular for people who have not physiologically adapted to higher altitudes (Horvath et al. 1988; Hsia 2002). Exercise increases oxygen demand, absorption of carbon monoxide, and transfer of carbon monoxide from blood to muscle, and decreases elimination of carbon monoxide (Joumard et al. 1991; Richardson et al. 2002; Werner and Lindahl 1980); these factors may contribute to higher vulnerability during strenuous exercise.

3.11. METHODS FOR REDUCING TOXIC EFFECTS

This section describes clinical practice and research concerning methods for reducing toxic effects of exposure to carbon monoxide. However, because some of the treatments discussed may be experimental and unproven, this section should not be used as a guide for treatment of exposures to carbon monoxide. When specific exposures have occurred, poison control centers and medical toxicologists should be consulted for medical advice. The following texts provide specific information about treatment following exposures to carbon monoxide:

  • Dart RC, ed. 2004. Medical toxicology. 3rd ed. Philadelphia, PA: Lippincott Williams & Wilkins, 1208.
  • Lavonas EJ. 2007. Carbon monoxide poisoning. In: Haddad and Winchester's clinical management of poisoning and drug overdose. 4th ed. Philadelphia, PA: Saunders, 1297-1307
  • Tomaszewski C. 2006. Carbon monoxide. In: Goldfrank's toxicologic emergencies. New York, NY: McGraw-Hill, 1689-1704.

3.11.1. Reducing Peak Absorption Following Exposure

Human exposure to carbon monoxide occurs by the inhalation route. General recommendations for reducing absorption of carbon monoxide following acute high-level inhalation exposure have included immediate removal of the exposed individual from the contaminated area and administering 100% oxygen through a face mask that does not allow rebreathing of expired air or by endotracheal tube for a minimum of 4 hours. For patients with life-threatening carbon monoxide exposure, hyperbaric oxygen therapy has also been suggested, although this remains a controversial method for treatment (Buckley et al. 2011; Dart 2004; Lavonas 2007; Tomaszewski 2002; Weaver et al. 2000, 2009; Wolf et al. 2008). Hyperbaric oxygen treatment is optimal within 6 hours of exposure, as patients treated after 6 hours have displayed a higher percent of delayed sequelae (30 vs. 19%) and mortality (30 vs. 14%) compared with patients treated within 6 hours of carbon monoxide exposure (Tomaszewski 2006).

Studies conducted in animal models have provided evidence for a potential influence of isocapnia or hypercapnia (i.e., normal or elevated CO2 levels) in the effect of O2 therapy in promoting the reduction in carbon monoxide body burden. Studies utilizing laboratory animals (dogs and sheep) exposed to carbon monoxide have demonstrated that the carbon monoxide washout time can be significantly reduced if the subjects receive a mixture of oxygen and carbon dioxide gas designed to maintain normal PCO2 values (Fisher et al. 1999; Kreck et al. 2001). This outcome has been replicated in experimental trials using volunteers exposed to carbon monoxide concentrations producing COHb levels of 10–12%. As compared to the subjects treated with hyperoxia alone, those treated with hyperoxia and carbon dioxide who maintained isocapnia showed accelerated clearance of carbon monoxide (Rucker et al. 2002; Takeuchi et al. 2000). These studies identify a potential means to enhance the treatment of acute carbon monoxide poisoning.

3.11.2. Reducing Body Burden

Absorbed carbon monoxide binds to, and displaces oxygen (O2) from, blood Hb and other heme proteins (e.g., muscle myoglobin). Nearly all of carbon monoxide in the body exists as complexes with Hb or other heme proteins. Binding to heme proteins limits diffusion of carbon monoxide from tissues to blood and elimination of carbon monoxide from blood to exhaled air. At ambient oxygen pressures of approximately 0.2 atm, elimination of carbon monoxide occurs with a half-time of approximately 100–300 minutes (Bruce and Bruce 2006; Peterson and Stewart 1970). However, the elimination rate can be increased substantially by administering 100% oxygen (Tomaszewski 2006; Weaver et al. 2000). As noted above, the general recommendation for reducing the body burden of carbon monoxide is administration of 100% oxygen through a face mask that does not allow rebreathing of expired air, at ambient pressure, and/or hyperbaric oxygen therapy. The longer time period required for the elimination of carbon monoxide in the fetus (Hill et al. 1977; Longo 1977) has prompted a concern about the premature cessation of oxygen therapy to a pregnant woman based upon her COHb level.

3.11.3. Interfering with the Mechanism of Action for Toxic Effects

Binding of carbon monoxide to Hb impairs transfer of oxygen to tissues and can produce tissue hypoxia. Tissues that have a high oxygen utilization rate, including brain and heart, are particularly vulnerable to hypoxia. Carbon monoxide-induced hypoxia is treated by administering 100% oxygen, which increases the dissolved concentration of oxygen in blood, facilitates displacement of carbon monoxide from Hb, and increases the rate of elimination of carbon monoxide from the body. Hyperbaric oxygen therapy may further reduce the risk of brain and heart injury from hypoxia. Intravenous fluids and inotropic agents may be administered to alleviate hypotension and myocardial depression (Lavonas 2007; Tomaszewski 2006).

3.12. ADEQUACY OF THE DATABASE

Section 104(I)(5) of CERCLA, as amended, directs the Administrator of ATSDR (in consultation with the Administrator of EPA and agencies and programs of the Public Health Service) to assess whether adequate information on the health effects of carbon monoxide is available. Where adequate information is not available, ATSDR, in conjunction with the National Toxicology Program (NTP), is required to assure the initiation of a program of research designed to determine the health effects (and techniques for developing methods to determine such health effects) of carbon monoxide.

The following categories of possible data needs have been identified by a joint team of scientists from ATSDR, NTP, and EPA. They are defined as substance-specific informational needs that if met would reduce the uncertainties of human health assessment. This definition should not be interpreted to mean that all data needs discussed in this section must be filled. In the future, the identified data needs will be evaluated and prioritized, and a substance-specific research agenda will be proposed.

3.12.1. Existing Information on Health Effects of Carbon Monoxide

The existing data on health effects of inhalation, oral, and dermal exposure of humans and animals to carbon monoxide are summarized in Figure 3-12. The purpose of this figure is to illustrate the existing information concerning the health effects of carbon monoxide. Each dot in the figure indicates that one or more studies provide information associated with that particular effect. The dot does not necessarily imply anything about the quality of the study or studies, nor should missing information in this figure be interpreted as a “data need”. A data need, as defined in ATSDR’s Decision Guide for Identifying Substance-Specific Data Needs Related to Toxicological Profiles (Agency for Toxic Substances and Disease Registry 1989), is substance-specific information necessary to conduct comprehensive public health assessments. Generally, ATSDR defines a data gap more broadly as any substance-specific information missing from the scientific literature.

Figure 3-12. Existing Information on Health Effects of Carbon Monoxide.

Figure 3-12

Existing Information on Health Effects of Carbon Monoxide. ● Existing Studies

Human clinical studies of acute exposures have investigated cardiovascular, respiratory, and neurobehavioral end points. Human epidemiology studies have examined possible associations between ambient air carbon monoxide concentrations and various health end points, including mortality, cardiovascular, respiratory, blood biomarkers of coagulation and inflammation, and developmental (preterm birth, birth weight, congenital anomalies, neonatal and infant mortality). Studies conducted in animals have included acute, intermediate, and chronic exposures that have largely focused on cardiovascular, respiratory, and reproductive (e.g., fetal mortality) end points. Bioassays of carcinogenicity of carbon monoxide have not been reported. Carbon monoxide has been assessed for genotoxicity in an in vivo mouse model of micronuclei formation and sister chromatid exchange.

3.12.2. Identification of Data Needs

Acute-Duration Exposure. Most information on effects of acute exposures to carbon monoxide derive from clinical case studies of carbon monoxide poisoning, in which blood COHb levels were >20%. Although these studies provide information on effects resulting from severe carbon monoxide-induced toxicity, they do not provide information on low dose-response relationships for carbon monoxide. Experimental clinical studies providing low dose-response information have focused on evaluating organ systems that are expected to be vulnerable to or contribute to hypoxia (e.g., brain, heart, respiratory tract). Effects of acute exposures to carbon monoxide on other organ systems have been studied far less thoroughly. Studies of low-dose carbon monoxide exposure in humans or animals have not identified NOAELs for sensitive organ systems and have not provided data that would enable low-dose response modeling (e.g., benchmark dose). Such studies would be beneficial for analyses in support of deriving MRLs.

Experimental clinical studies have examined cardiovascular effects in humans, including a major vulnerable subpopulation, people with ongoing heart disease; however, these studies have not identified NOAELs in cardiovascular disease patients (Adams et al. 1988; Allred et al. 1989, 1991; Anderson et al. 1973; Kleinman et al. 1989, 1998; Leaf and Kleinman 1996b). Studies investigating the effects of acute-duration carbon monoxide exposure on hemodynamics have been conducted in several animal species, including monkeys, dogs, rats, and rabbits, under exposure conditions producing blood COHb levels ranging from 6.2 to 70% (EPA 1991, 2000). These studies have shown that brief exposure (from a few minutes to approximately 3 hours) to carbon monoxide at concentrations of 80–20,000 ppm produces alterations in hemodynamics that are consistent with COHb-induced hypoxia and responsive compensatory mechanisms (e.g., vasodilation and increased cardiovascular output), including increased coronary blood flow, decreased myocardial O2 consumption, increased heart rate, and alterations in blood flow to various vascular beds (e.g., cerebral, limb muscular).

A few clinical experimental studies have evaluated adverse respiratory effects of carbon monoxide exposure in relatively small numbers of healthy subjects (Chevalier et al. 1966; Fisher et al. 1969; Koike et al. 1991; Ren et al. 2001; Vesely et al. 2004). Results of these studies indicate that brief exposure to relatively higher levels (e.g., ≥500 ppm) of carbon monoxide may decrease ventilatory performance. No studies were identified that examined effects of carbon monoxide on respiratory function in patients with underlying respiratory diseases, who would be expected to have a higher sensitivity to hypoxic effects of carbon monoxide. Studies of dose-response relationships of respiratory function at low levels of carbon monoxide exposure (e.g., COHb <20%) in patients with respiratory disease (e.g., asthma) would be beneficial for quantifying vulnerability of this subpopulation to carbon monoxide toxicity.

Intermediate-Duration Exposure. Controlled experimental studies in humans of effects of intermediate-duration exposures to carbon monoxide were not identified. Intermediate-duration studies have examined cardiovascular and respiratory effects of exposure to carbon monoxide in various animal models, including monkeys, dogs, and rodents (EPA 2000). The observation of cardiomegaly in rats exposed chronically to carbon monoxide (Sørhaug et al. 2006) suggests that prolonged exposure to carbon monoxide may produce effects on the heart that might not be observed with acute exposures. Additional intermediate-duration exposure studies in animals that provide dose-response relationships for cardiovascular effects and other end points (e.g., respiratory, neurological) would be beneficial for analyses in support of deriving MRLs.

Chronic-Duration Exposure and Cancer. Epidemiological studies have examined health outcomes in the context of chronic exposures or acute variations in exposure concentrations that occur during chronic exposures. Outcomes that have been assessed include mortality, including cancer mortality, cardiovascular effects, respiratory effects, effects on blood biomarkers of coagulation and inflammation, and developmental effects (pre-term birth, birth weight, congenital anomalies, neonatal and infant mortality). A chronic study conducted in rats found cardiomegaly, which may have represented long-term adaptive changes to carbon monoxide-induced hypoxia (Sørhaug et al. 2006). This finding suggests that prolonged exposure to carbon monoxide may produce effects on the heart that might not be evident from acute- or intermediate-duration studies. Additional chronic-duration exposure studies in animals would that provide dose-response relationships for cardiovascular effects and other end points (e.g., respiratory, neurological) would be beneficial for analyses in support of deriving MRLs.

Carbon monoxide has not been assessed for carcinogenicity using animal models. Cancer bioassays in animals would provide a more confident basis for establishing whether or not carbon monoxide has carcinogenic potential in humans.

Genotoxicity. In a bacterial replication assay with Escherichia coli, exposure to carbon monoxide inhibited DNA synthesis (Cairns and Denhardt 1968). Gestational exposure of mice to carbon monoxide increased micronuclei formation and sister chromatid exchange in dams (bone marrow) and fetuses (blood) (Kwak et al. 1986). Additional studies conducted on the mechanisms of genotoxicity in mammalian cells would provide a more confident basis for establishing whether or not carbon monoxide has carcinogenic potential in humans.

Reproductive Toxicity. Pregnancy outcomes, including fetal death in humans, have been reported in cases of maternal carbon monoxide poisoning during pregnancy (Brown et al. 1992; Caravati et al. 1988; Cramer 1982; Elkharrat et al. 1991; Farrow et al. 1990; Greingor et al. 2001; Hollander et al. 1987; Margulies 1986; Norman and Halton 1990; Silverman and Montano 1997). Epidemiological studies have examined possible relationships between exposures to ambient air concentrations of carbon monoxide and fetal mortality (see Table 3-8). Few studies of reproductive effects in animals were identified, and these have been limited to assessments of effects of maternal exposure during pregnancy on fetal mortality (Astrup et al. 1972; Fechter and Annau 1977; Penney et al. 1983; Stupfel and Bouley 1970). The available data do not allow a confident determination of the relative sensitivity of reproductive systems to carbon monoxide. Greater confidence would be achieved from studies that examine more comprehensive end points of reproductive toxicity in females and males, in association with exposures prior to mating and pregnancy and during pregnancy.

Developmental Toxicity. Epidemiological studies have examined developmental outcomes in the context of chronic exposures or acute variations in exposure concentrations that occur during chronic exposures (see Table 3-8). Outcomes that have been assessed include pre-term birth, birth weight, congenital anomalies, neonatal, infant mortality, and neurodevelopment. Studies conducted in animals provide evidence of adverse developmental effects of gestational and early postnatal carbon monoxide exposure, including decreased fetal weight, adverse central nervous system development, altered peripheral nervous system development, cardiac effects, altered sexual behavior, immunological effects, and hematological effects. In addition, some studies showed that developmental effects persisted beyond the postnatal period (see Table 3-9). Two studies found functional impairments in the auditory system of rats in association with exposures to ≥12 ppm carbon monoxide during the neonatal periods or during gestation (Stockard-Sullivan et al. 2003; Webber et al. 2003). These values are the lowest LOAELs for external exposure that have been reported and, as a result, have a substantial impact on analyses conducted to derive MRLs. Additional studies to establish greater confidence in the dose-response relationship for neurodevelopmental effects on the auditory system would be beneficial for establishing MRLs for carbon monoxide based on this end point. Studies conducted to understand the mechanism for this effect would provide additional information about the relevance of these effects to humans exposed at similar levels.

Immunotoxicity. Epidemiological studies have examined effects on blood biomarkers of inflammation. Biomarkers examined have included been inflammation markers (CRP, SAA, and WBC) and cell adhesion markers (E-selectin, vWF, and ICAM-1) (Baccarelli et al. 2007; Liao et al. 2005; Pekkanen et al. 2000; Rückerl et al. 2006, 2007; Steinvil et al. 2008). A study conducted in guinea pigs examined effects of carbon monoxide exposure on plaque forming cells in spleen (Snella and Rylander 1979). Effects of gestational exposure on the developing immune system have been studied in rats (Giustino et al. 1993, 1994). Studies conducted in cell culture and in vivo models of sepsis have examined effects of endogenous and exogenous carbon monoxide on inflammatory responses and cytokine and signaling pathways involved in inflammation (Ryter et al. 2006). These studies suggest a potential for carbon monoxide to exert effects on the immune system. Additional studies to establish modes of action and dose-response relationships would be beneficial for determining the sensitivity of the immune system to carbon monoxide, relative to other targets of toxicity, and for establishing MRLs.

Neurotoxicity. Experimental clinical studies (Benignus et al. 1994), clinical case studies (Chambers et al. 2008; Dolan 1985; Ernst and Zibrak 1998; Hopkins et al. 2006; Kao and Nañagas 2006; Lo et al. 2007; Parkinson et al. 2002; Raub and Benignus 2002), and studies conducted in animals (Benignus et al. 1994; Chambers et al. 2008; Chen and Fechter 1999; Dolan 1985; Ernst and Zibrak 1998; Fechter et al. 1988; Hopkins et al. 2006; Kao and Nañagas 2006; Lo et al. 2007; Parkinson et al. 2002; Raub and Benignus 2002; Young et al. 1987) have examined effects of acute carbon monoxide exposures on behavior cognitive function. Although these studies provided strong evidence for impairments in association with exposures that result in blood COHb levels >20%, dose-response relationships for lower levels remain controversial (Benignus 1994; Benignus et al. 1990; Raub and Benignus 2002). Further research on dose-response relationships of neurobehavioral and neurosensory effects at low levels of carbon monoxide exposure (e.g., COHb<20%) utilizing double-blind designs would be beneficial for analyses in support of deriving MRLs.

Epidemiological and Human Dosimetry Studies. Epidemiological studies have examined health outcomes in the context of chronic exposures or acute variations in exposure concentrations that occur during chronic exposures (see Tables 3-3, 3-6, 3-7, and 3-8). Outcomes that have been assessed include mortality, including cancer mortality, cardiovascular effects, respiratory effects, blood biomarkers of coagulation and inflammation, and developmental effects (pre-term birth, birth weight, congenital anomalies, neonatal and infant mortality). Human dosimetry studies have examined relationships between carbon monoxide exposure levels and durations and blood COHb levels in resting subjects, during exercise, and at various altitudes (see Tables 3-4 and 3-5). Studies of elimination kinetics of carbon monoxide in humans have explored effects of exercise, age, gender, and oxygen therapy on elimination half-time (Bruce and Bruce 2006; Joumard et al. 1991; Landaw 1973; Levasseur et al. 1996; Peterson and Stewart 1970; Shimazu et al. 2000; Weaver et al. 2000).

Biomarkers of Exposure and Effect.

Exposure. Measurement of blood COHb is the principal biomarker for identifying exposure to carbon monoxide. Several pharmacokinetics models have been developed that can be used to predict blood COHb levels that would correspond to a given exposure concentration and duration in a typical adult and in the fetus resulting from maternal exposures. These models can be used to reconstruct possible carbon monoxide exposure scenarios (e.g., concentrations, duration, and gap between measurement of COHb and cessation of exposure) that could result in a given measured COHb level.

Effect. Although blood COHb reflects current carbon monoxide body burden, measurement of blood COHb has not been shown to be a reliable predictor of severity of acute toxicity (Hampson and Hauff 2008). Therefore, diagnoses and characterization of toxic effects of carbon monoxide relies on both the measurement of blood COHb and the assessment of signs and symptoms of carbon monoxide toxicity. Studies conducted to identify biomarkers of non-hypoxic mechanisms of carbon monoxide (those unrelated to blood COHb levels) would be beneficial for internal dosimetry in epidemiological and clinical experimental studies, for development of PBPK/PD models of non-hypoxic effects of carbon monoxide, and for improving diagnosis of carbon monoxide poisoning.

Absorption, Distribution, Metabolism, and Excretion. Pharmacokinetics of carbon monoxide has been extensively studied in humans and in animal models, including nonhuman primates (see Section 3.4). Studies conducted in humans have provided information on absorption, metabolism, distribution, and elimination kinetics, as well as physiological factors (e.g., age, gender) and environmental factors (e.g., altitude, exercise) that affect carbon monoxide kinetics. These studies have provided the bases for pharmacokinetic models that simulate carbon monoxide absorption and elimination kinetics and blood COHb and muscle COMb levels in humans (see Section 3.4.5). Several studies have provided estimates of rates of whole-body endogenous production of carbon monoxide in adult males and females during and following pregnancy (Coburn et al. 1963; Delivoria-Papadopoulos et al. 1974; Longo 1977; Mercke and Lundh 1976). However, measurement of local tissue production rates would also be valuable for studying mechanisms by which endogenously produced carbon monoxide modulates cellular metabolism and cell signaling pathways, as well as the relative contributions of exogenous carbon monoxide and endogenously produced carbon monoxide contribute to these processes.

Comparative Toxicokinetics. Major physiological and biochemical determinants of blood and skeletal muscle carbon monoxide kinetics that would be expected to contribute to interspecies differences in carbon monoxide toxicokinetics have been identified. This information provides a basis for extrapolating pharmacokinetics models across mammalian species, for which physiological (e.g., ventilation rates, blood Hb levels) and chemical-specific parameters are known (e.g., binding affinities and capacities of Hb and myoglobin, long carbon monoxide diffusion capacity). Models developed to simulate carbon monoxide kinetics in humans have been adapted for applications to animal models (e.g., rats; Benignus and Annau 1994) and to simulate maternal-fetal transfer of carbon monoxide during pregnancy in sheep (Longo and Hill 1977). Adaptation of these models to a wider range of species would be facilitated by studies that provide estimates of model parameters in these species (e.g., Haldane coefficient, binding coefficients of Hb and Mb, and the lung carbon monoxide diffusing capacity).

Methods for Reducing Toxic Effects. Efficacies of normobaric and hyperbaric oxygen therapy for hastening the elimination of carbon monoxide from the body and for treating carbon monoxide-induced hypoxia have been studied. Although hyperbaric oxygen therapy is considered to be a therapeutic option, its efficacy, over normobaric therapy, remains controversial (Wolf et al. 2008). Further research regarding benefits and limitations of hyperbaric oxygen therapy would improve the clinical management of treatment of carbon monoxide poisoning.

Children’s Susceptibility. Data needs relating to both prenatal and childhood exposures, and developmental effects expressed either prenatally or during childhood, are discussed in detail in the Developmental Toxicity subsection above.

Child health data needs relating to exposure are discussed in Section 6.8.1, Identification of Data Needs: Exposures of Children.

Mechanisms of Toxicity. Although hypoxic mechanisms of action of carbon monoxide are well established (i.e., those related to formation of COHb), a better understanding is needed of non-hypoxic mechanisms, their contribution to observed adverse health effects of carbon monoxide, and related dose-response relationships.

3.12.3. Ongoing Studies

Ongoing studies pertinent to information of health effects and/or mechanisms of action of carbon monoxide are listed in Table 3-18.

Table 3-18. Ongoing Studies on the Health Effects of Carbon Monoxide.

Table 3-18

Ongoing Studies on the Health Effects of Carbon Monoxide.