Function

Vitamin K plays an essential role in the posttranslational conversion of specific glutamyl residues in a limited number of proteins to γ-carboxyglutamyl (Gla) residues (Suttie, 1993). These proteins include plasma prothrombin (coagulation factor II) and the plasma procoagulants, factors VII, IX, and X. Because under-γ-carboxylated forms of these proteins lack biological activity, the classical sign of a vitamin K deficiency has been a vitamin K-responsive increase in prothrombin time and, in severe cases, a hemorrhagic event. Two structurally related vitamin K-dependent proteins (Price, 1988), osteocalcin found in bone and matrix Gla protein originally found in bone but now known to be more widely distributed, have received recent attention as proteins with possible roles in the prevention of chronic disease (Ferland, 1998). No relationship between a decreased biological activity of any of the other vitamin K-dependent proteins and a disease-related physiological response has been postulated.

Physiology of Absorption, Metabolism, and Excretion

Phylloquinone, the major form of vitamin K in the diet, is absorbed in the jejunum and ileum in a process that is dependent on the normal flow of bile and pancreatic juice and is enhanced by dietary fat (Shearer et al., 1974). Absorption of free phylloquinone is nearly quantitative (Shearer et al., 1970), but recent studies (Garber et al., 1999; Gijsbers et al., 1996) suggest that the vitamin in food sources is less well absorbed. Absorbed phylloquinone is secreted into lymph as a component of chylomicrons and enters the circulation in this form. Circulating phylloquinone is present in the very low density triglyceride-rich lipoprotein fractions and chylomicrons (Kohlmeier et al., 1996; Lamon-Fava et al., 1998). A dependence of plasma phylloquinone concentrations (Kohlmeier et al., 1995) on the distribution of lipoprotein apoE isoforms suggests that the vitamin enters the liver through the endocytosis of chylomicron remnants. The liver rapidly accumulates ingested phylloquinone and contains the highest concentration. Skeletal muscle contains little phylloquinone, but significant concentrations are found in the heart and some other tissues (Davidson et al., 1998; Thijssen and Drittij-Reijnders, 1994). It is not known how or if hepatic phylloquinone is secreted and transported from the liver to peripheral tissues.

The vitamin is rapidly catabolized and excreted from the liver, mainly in bile. A smaller amount appears in urine (Shearer et al., 1974). The excretion products have not been extensively characterized but are known to proceed through the oxidative degradation of the phytyl side chain of phylloquinone, followed by glucuronide conjugation. Turnover in the liver is rapid and hepatic reserves are rapidly depleted when dietary intake of vitamin K is restricted (Usui et al., 1990).

The human gut contains a large amount of bacterially produced menaquinones, but their contribution to the maintenance of vitamin K status has been difficult to assess (Suttie, 1995). Although the content is extremely variable, human liver contains about 10 times as much vitamin K as a mixture of menaquinones than as phylloquinone (Shearer, 1992; Thijssen and Drittij-Reijnders, 1996; Usui et al., 1990). Absorption of these very lipophilic membrane-associated compounds from the distal bowel has been difficult to demonstrate (Ichihashi et al., 1992). Evidence of vitamin K inadequacy in normal human subjects following dietary restriction of vitamin K also suggests that this source of the vitamin is not utilized in sufficient amounts to maintain maximal γ-carboxylation of vitamin K-dependent proteins. One specific menaquinone, MK-4, appears to have a unique yet unidentified role. MK-4 can be formed from menadione (2-methyl-1,4-naphthoquinone) but is also formed in animal tissues from phylloquinone (Davidson et al., 1998; Thijssen and Drittij-Reijnders, 1994). It is present in much higher concentrations than phylloquinone in tissues such as pancreas, salivary gland, brain, and sternum, and its concentration in these tissues is to some degree dependent on phylloquinone intake.

Clinical Effects of Inadequate Intake

A clinically significant vitamin K deficiency has usually been defined as a vitamin K-responsive hypoprothrombinemia and is associated with an increase in prothrombin time (PT) and, in severe cases, bleeding. Spontaneous cases have been rare and have usually been associated with various lipid malabsorption syndromes (Savage and Lindenbaum, 1983). There are numerous case reports of bleeding episodes in antibiotic-treated patients, and these have often been ascribed to an acquired vitamin K deficiency resulting from a suppression of menaquinone-synthesizing organisms. However, these reports are complicated by the possibility of general malnutrition in this patient population and by the antiplatelet action of many of the same drugs (Suttie, 1995).

Reports of experimentally induced, clinically significant vitamin K deficiencies are scant. Udall (1965) fed 10 healthy subjects a diet that probably contained less than 10 μg /day of phylloquinone. After 3 weeks, a statistically significant increase in the PT was observed, but it was still within the normal range. In another study, Frick and coworkers (1967) administered a parenteral nutrient solution to a small number of neomycin-treated adults for 4 weeks and observed prolonged PT that responded to the parenteral administration of phylloquinone. They concluded that the minimal daily requirement was between 0.3 and 1.05 μg per kg body weight of phylloquinone. In more recent studies (Allison et al., 1987; Ferland et al., 1993), feeding healthy individuals diets containing 5 to 10 μg/day of phylloquinone for 14- to 16-day periods failed to induce any change in PT measurements.

These limited studies, conducted over a number of years, indicate that the simple restriction of vitamin K intake to levels almost impossible to achieve in any nutritionally adequate, self-selected diet do not impair normal hemostatic control in healthy subjects. Although there is some interference in the hepatic synthesis of the vitamin K-dependent clotting factors that can be measured by sensitive assays, standard clinical measures of procoagulant potential are not changed.

Prothrombin Time

The classical PT used to measure the procoagulant potential of plasma is not a sensitive indicator of vitamin K status because plasma prothrombin concentration must be decreased by approximately 50 percent before a value is outside of the “normal” range (Suttie, 1992). Furthermore, studies conducted thus far clearly indicate that PT does not respond to a change in dietary vitamin K in healthy subjects (Allison et al., 1987; Bach et al., 1996; Binkley et al., 1999; Booth et al., 1999a; Suttie et al., 1988).

Factor VII

On the basis of its relatively short half-life (approximately 6 hours), factor VII activity has been used to assess vitamin K status. Allison and colleagues (1987) maintained 33 healthy subjects, some given antibiotics, for 2 weeks on a low vitamin K diet (less than 5 μg/day of phylloquinone) and observed a decrease from the normal range of plasma factor VII in seven of the subjects. However, in the absence of antibiotic treatment, factor VII activity is not a sensitive indicator of vitamin K status as it does not usually respond to changes in vitamin K intake in healthy individuals (Bach et al., 1996; Ferland et al., 1993).

Plasma and Serum Phylloquinone Concentration

Both phylloquinone and the menaquinones have been used to assess status, with phylloquinone as the vitamer usually studied because it is the primary source of dietary vitamin K in western countries (Booth and Suttie, 1998). Serum or plasma phylloquinone concentration reflects recent intakes and has been shown to respond to changes in dietary intake within 24 hours (Sokoll et al., 1997). However, given the distribution of vitamin K in the food supply, a single day plasma (serum) phylloquinone concentration may not reflect normal dietary intake. Positive correlations between circulating phylloquinone concentration and dietary intake have been reported, but the strength of this association has varied according to studies, possibly due to differences in intake assessment methodology (i.e., number of diet record days) (Booth et al., 1995, 1997b). In healthy individuals, phylloquinone concentrations are higher in older subjects than in younger subjects, irrespective of dietary intake (Booth et al., 1997b; Ferland et al., 1993; Sokoll and Sadowski, 1996). Strong positive correlations between plasma (serum) phylloquinone and triglyceride concentrations have been reported (Kohlmeier et al., 1995; Sadowski et al., 1989; Saupe et al., 1993), a finding that likely explains the higher vitamin K concentrations observed in older individuals (Sadowski et al., 1989). Normal ranges for plasma phylloquinone concentration in healthy adults aged 20 to 49 years (n = 131) was 0.25 to 2.55 nmol/L; for those aged 65 to 92 years (n = 195), 0.32 to 2.67 nmol/L (Sadowski et al., 1989).

Urinary γ-Carboxyglutamyl Residues

After protein catabolism, γ-carboxyglutamyl (Gla) residues contained in the vitamin K-dependent proteins are not further metabolized and are excreted via urine (Shah et al., 1978). As a result, urinary Gla excretion has been used as an indicator of vitamin K status. Urinary Gla responds to alterations in dietary intake, but periods of several days are needed before any change can be observed (Ferland et al., 1993; Suttie et al., 1988). In a study by Suttie and coworkers (1988), 10 college-age men were asked to eliminate the major sources of vitamin K from their diet, thereby reducing their intake to less than 40 μg/day of phylloquinone. Urinary Gla excretion decreased 22 percent after 3 weeks and returned to baseline values 12 days after supplementation with 50 or 500 μg of phylloquinone. In a recent study, increasing phylloquinone intakes from 100 μg/day to a range of 377 to 417 μg/day for 5 days did not induce significant changes in urinary Gla (Booth et al., 1999a).

Response of urinary Gla to vitamin K intake alterations appears to be age-specific. In a study by Ferland and coworkers (1993), 32 subjects were divided into four groups of eight (men or women, 20 to 40 or 60 to 80 years old) and housed in a metabolic research unit. They were fed 80 μg of phylloquinone for 4 days followed by a low vitamin K diet (approximately 10 μg phylloquinone/day) for 16 days. At the end of the depletion period, urinary Gla excretion had decreased significantly in the younger, but not the older subjects. Short-term supplementation with 45 μg/day of phylloquinone reversed the decline to near baseline values. In another study involving 263 healthy individuals (127 men, 136 women) aged 18 to 55 years, urinary Gla/creatinine excretion ratios increased significantly with age in both men and women with values 20 percent higher in women over the age span (Sokoll and Sadowski, 1996). To date, there are insufficient data for using urinary Gla excretion for estimating an average requirement.

Undercarboxylated Prothrombin

In humans, an insufficiency of vitamin K leads to the secretion into plasma of biologically inactive, under-γ-carboxylated forms of the vitamin K-dependent clotting factors. These proteins are referred to as protein induced by vitamin K absence or antagonism (PIVKA). In reference to prothrombin (factor II), the term used is PIVKA-II. This protein has been measured by specific immunoassay (Blanchard et al., 1981), by thrombin generation after the removal of normal prothrombin by adsorption to barium or calcium salts (Francis, 1988), or by an indirect assay that compares biologically active prothrombin to the amount of thrombin that can be generated by a nonphysiological activator (Allison et al., 1987). A number of immunochemical assays, which are very sensitive and are capable of measuring very small increases of this indicator of vitamin K insufficiency, are now commercially available. Typically, these kits will detect changes of a few ng/mL whereas plasma prothrombin concentration averages 100 μg/mL.

Concentrations of PIVKA-II vary little with aging in healthy subjects (Sokoll and Sadowski, 1996) but respond to dietary alterations. In two independent studies using immunological assays (Booth et al., 1999b; Ferland et al., 1993), intakes of 10 μg/day of phylloquinone were associated with abnormal PIVKA-II concentrations (greater than 2 ng/mL) in the great majority of subjects, whereas an intake of 100 μg/day was associated with normal (less than 2 ng/ mL) PIVKA-II concentrations in 15 of 16 subjects (Booth et al., 1999b). In older studies that used indirect colorimetric assays, abnormal PIVKA-II concentrations were observed with diets containing 40 to 60 μg/day of phylloquinone but were normal when intakes were approximately 80 μg/day (Jones et al., 1991; Suttie et al., 1988).

Although it is clear from these data that PIVKA-II concentrations can be influenced by vitamin K intake, results from these studies cannot be used to set dietary vitamin K recommendations. This is because there have been no studies to compare the immunoassay and colorimetric studies for determining whether the data given above can be used collectively. Therefore, at the present there are inadequate dose-response data from a single procedure. Intervention studies using graded intakes of vitamin K and protocols of longer duration need to be conducted before this indicator can be used to establish dietary recommendations for vitamin K.

Under-γ-carboxylated Osteocalcin

Small amounts of the bone protein, osteocalcin, circulate in plasma, and like PIVKA-II, under-γ-carboxylated osteocalcin (ucOC) has been considered an indicator of suboptimal vitamin K status. Assays for measuring the degree of carboxylation of osteocalcin have been indirect and have relied on the lower affinity of ucOC for hydroxy-apatite (Knapen et al., 1989) or barium sulfate (Sokoll et al., 1995). Only recently has direct assessment of ucOC been possible with the development of a monoclonal antibody specific for the undercarboxylated form of osteocalcin (Vergnaud et al., 1997).

As discussed below, a number of reports have correlated decreased bone mineral density (BMD) or increased fracture rate with a five-to eight-fold increase in ucOC. Concurrently, it has been observed that vitamin K intakes similar to those reported for the general population did not ensure complete carboxylation of osteocalcin (Bach et al., 1996; Sokoll and Sadowski, 1996) and that ucOC could be decreased by increasing vitamin K intake (Binkley et al., 1999; Booth et al., 1999b; Douglas et al., 1995; Knapen et al., 1989, 1993). These reports have led to the suggestion that vitamin K requirements for bone function are probably much higher than those needed to maintain normal hemostasis and that the recommendation for vitamin K should be much higher than current recommendations (Weber, 1997).

However, a number of issues must be considered before a minimal ucOC concentration can be used as an indicator to estimate an average requirement for vitamin K. Because osteocalcin is used clinically as a marker of bone turnover, there are a number of commercial kits currently marketed. Although they may all be internally reproducible, they react with different epitopes and have different reactivity with osteocalcin degradation fragments. Therefore, they do not give the same “normal” values (Delmas et al., 1990a, 1990b; Gundberg et al., 1998). Because of this, most investigators interested in the influence of vitamin K status on bone have expressed measurements of ucOC as percent ucOC. In apparently healthy subjects, ucOC has ranged from 3 to 45 percent, depending on the assay. The basis for these higher values has not been established but in many cases may reflect the fact that the assay is recognizing some osteocalcin fragments that do not contain potential Gla sites as ucOC. This interpretation of the data is supported by the high ucOC values that have been seen in some studies after vitamin K supplementation (Booth et al., 1999a; Douglas et al., 1995; Knapen et al., 1993). Other investigators have reported nearly complete elimination of ucOC by vitamin K supplementation. Bach and coworkers (1996) reduced ucOC from 8 to 3 percent and from 2 to 1 percent in small groups (n = 9) of younger and older subjects, respectively, with 1 mg phylloquinone for 5 days. Binkley and coworkers (1999) supplemented a larger (n = 107) group of both younger and older subjects by supplementation with 1 mg phylloquinone for 2 weeks and reduced ucOC from 8 to 3 percent and from 7 to 3 percent, respectively.

The wide variations in percent ucOC reported for vitamin K-sufficient subjects have made it essentially impossible to compare studies. The emphasis that investigators have placed on ucOC, an indicator of a nonfunctional protein, has also influenced thinking in this field. If percent ucOC in the apparently healthy population is as low as indicated in the more recent studies, about 90 to 95 percent of osteocalcin is in its biologically active form. Whether it is reasonable to assume that an increase in this value to 100 percent would be expected to have any physiological significance is a question that must be considered.

Although there is little doubt that vitamin K intake affects the degree of osteocalcin λ-carboxylation, the technical problems associated with the current assays and the uncertainty surrounding the physiological significance of diet-induced changes prevent the use of ucOC for estimating an average requirement for vitamin K.

Relationship of Vitamin K Intake to Chronic Disease

Bioavailability

The predominant form of vitamin K in the North American diet is phylloquinone from green leafy vegetables, and the available data on the vitamin K content of foods have been reviewed (Booth and Suttie, 1998). These data are comprehensive, but little information on the relative bioavailability of phylloquinone from various foods in human subjects is available. Gijsbers and colleagues (1996) have compared the relative bioavailability, measured as area under an absorption curve, of 1,000 μg of phylloquinone from a synthetic preparation and from a food matrix. Phylloquinone in the form of cooked spinach was reported to be 4 percent as bioavailable as that from a phylloquinone supplement. Three times as much phylloquinone was absorbed when butter was consumed with the spinach. Garber and coworkers (1999) observed that when 500 μg of phylloquinone was consumed with a 400 kcal (27 percent energy from fat) meal, the relative absorption was between five and six times lower from spinach than from a phylloquinone tablet. In this study, phylloquinone absorption from fresh spinach, broccoli, or romaine lettuce did not differ and was highly variable between subjects. It is apparent from these limited studies that until more data are available, the bioavailability of phylloquinone from vegetable sources should not be considered to be more than 20 percent as available as phylloquinone consumed as a supplement.

Approximately 34 percent of phylloquinone in the American diet is consumed from fats and oils (Booth et al., 1995). It might be expected that phylloquinone dissolved in oil would be more available than from a food matrix, but this may not be true. Vitamin K is not well absorbed by patients exhibiting lipid malabsorption syndromes (Savage and Lindenbaum, 1983), and efficient absorption of this fat-soluble vitamin from the digestive tract does require dietary fat. Although direct measures of bioavailabilty have not been reported, a recent study reported no difference in fasting plasma phylloquinone concentrations when 400 μg of phylloquinone as broccoli or as phylloquinone-fortified oil was added to a diet containing 100 μg of phylloquinone (Booth et al., 1999a). Hydrogenated fats contain significant amounts of 2',3'-dihydrophylloquinone formed from phylloquinone during processing. Dietary intake of this form of the vitamin in the United States is estimated to be about 20 percent of phylloquinone (Booth et al., 1999c). Neither the biological activity nor the bioavailability of this form of vitamin K is known. The amount of dietary fat needed for optimal absorption has not been determined.

Drug-Nutrient Interactions

Oral 4-hydroxycoumarin derivatives such as warfarin are widely prescribed anticoagulants for the prevention of thrombotic disorders. These drugs function through the inhibition of a hepatic vitamin K-epoxide reductase. This enzyme reduces the coproduct of the λ-glutamyl carboxylase reaction, the vitamin K 2,3-epoxide, to the hydronaphthoquinone form of the vitamin, which is the substrate for the enzyme. The result is an acquired cellular vitamin K deficiency and a decrease in the synthesis of the vitamin K-dependent plasma clotting factors. Alterations in vitamin K intake can, therefore, influence warfarin efficacy, and numerous case reports of these occurrences have been reviewed (Booth et al., 1997a). Short-term, day-to-day variations in vitamin K intake do not appear to alter anticoagulant status, and there are few data on the extent to which long-term differences in dietary vitamin K intake modulate the response to warfarin. Lubetsky and coworkers (1999) studied a population of 46 patients with an estimated (by food frequency recall) median intake of 179 μg/day of phylloquinone. Patients with intakes greater than 250 μg/day were maintained at the targeted international normalized ratio with 5.8 mg/day warfarin, while patients with an intake of less than 250 μg/day of phylloquinone were maintained on a lower warfarin intake of 4.4 mg/day. These data suggest that alterations in vitamin K intake might influence warfarin dosage. As an effective warfarin dose varies widely within individuals, patients are closely monitored. Once a dose has been established, patients can avoid any complications resulting from variations in vitamin K intake by continuing to follow their normal dietary patterns.

Nutrient-Nutrient Interactions

The ability of elevated intakes of vitamin E to antagonize vitamin K action has been clearly established. Woolley (1945) first demonstrated that increased dietary or parenteral α-tocopherol or α-tocopherol quinone could induce a hemorrhagic syndrome in the rat, and vitamin K administration was demonstrated to reverse this response (Rao and Mason, 1975). Studies of the microsomal vitamin K-dependent carboxylase have demonstrated that the enzyme can be inhibited by α-tocopherol and that it is even more sensitive to α-tocopherol quinone (Bettger and Olson, 1982; Dowd and Zheng, 1995).

Increased intakes of vitamin E have not been reported to antagonize vitamin K status in healthy humans. However, in one study, oral supplementation of anticoagulated patients (50 percent plasma prothrombin concentrations) with approximately 360 mg/day (400 IU/day) of α-tocopherol resulted in nonstatistically significant decreases in prothrombin concentrations over a 4-week period, and a statistically significant decrease in the ratio of biologically active prothrombin to prothrombin antigen (Corrigan and Ulfers, 1981). More sensitive measures of vitamin K status are now available and should be used to assess the potential impact of vitamin E supplementation in anticoagulated patients or subjects with low vitamin K intakes.

The metabolic basis for vitamin E antagonism of vitamin K function has not been completely elucidated. Recent data from a study using a rat model have demonstrated an adverse effect of dietary α-tocopherol on phylloquinone absorption (Alexander and Suttie, 1999), and it is likely that both this response and cellular interactions are responsible for the antagonism that has been observed in both animals and human subjects.

Antagonism of vitamin K action in animal models by retinoids (retinyl acetate, 13-cis retinoic acid, and N-(7-hydroxyphenyl retinamide) has been reported by a group of investigators (McCarthy et al., 1989). High doses of these compounds have been used in animal studies, and adverse responses in humans have not been reported. The metabolic basis for this interaction has not been determined.

Infants Ages 0 through 12 Months

Vitamin K is poorly transported across the placenta, which puts newborn infants at risk for vitamin K deficiency (Greer, 1995). Concentrations of vitamin K in cord blood are usually less than 0.1 nmol/L or undetectable (Mandelbrot et al., 1988; Widdershoven et al., 1988), and elevated concentrations of undercarboxylated prothrombin (PIVKA-II) have been reported (Greer, 1995). Poor vitamin K status added to the fact that the concentrations of most plasma clotting factors are low at the time of birth increases the risk of bleeding during the first weeks of life, a condition known as hemorrhagic disease of the newborn (HDNB). Because HDNB can be effectively prevented by administration of vitamin K, infants born in the United States and in Canada routinely receive 0.5 to 1 mg of phylloquinone intramuscularly or 2.0 mg orally within 6 hours of birth. Compared to oral prophylaxis, intramuscular (IM) treatment has been shown to be more efficacious in the prevention of HDNB (Greer, 1995). In light of this and because an oral dosage form of vitamin K has not been readily available in North America, newborns have typically been administered vitamin K via the IM route. Studies published in the early nineties reporting an association between IM prophylaxis and childhood cancer created some concern and questioned the safety of this practice. In two studies, Golding and coworkers (1990, 1992) reported an increased risk (odds ratio, 1.97–2.6) of leukemia and other forms of cancer in children who had received vitamin K intramuscularly at birth. Subsequent studies conducted in the United States and European countries (Ansell et al., 1996; Ekelund et al., 1993; Klebanoff et al., 1993; Olsen et al., 1994; von Kries et al., 1996) have failed to confirm Golding's findings and quieted the debate. Recently, both the American and the Canadian pediatric societies reaffirmed their confidence in the IM prophylaxis, encouraging its general use (AAP, 1993; CPS, 1998).

Method Used to Set the Adequate Intake

No functional criteria of vitamin K status have been demonstrated that reflect response to dietary intake in infants. Thus, recommended intakes of vitamin K are based on an Adequate Intake (AI) that reflects a calculated mean vitamin K intake of infants principally fed human milk and provided vitamin K prophylaxis.

Although vitamin K prophylaxis at birth offers good protection with respect to HDNB during the first few weeks of life, infants become increasingly dependent on vitamin K intake in subsequent weeks. Though not a major concern in the United States or Canada, late HDNB occurs between 3 and 8 weeks of life and is usually associated with breast-feeding (Lane and Hathaway, 1985; von Kries et al., 1993). Milk intake appears to be an important factor in the etiology of late HDNB as inverse correlations have been reported between human milk intake and undercarboxylated prothrombin (Motohara et al., 1989; von Kries et al., 1987a).

Vitamin K concentrations in mature human milk have ranged from 0.85 to 9.2 μg/L with a mean concentration of 2.5 μg/L (Table 5-1) (Canfield et al., 1990, 1991; Greer et al., 1991, 1997; Haroon et al., 1982; Hogenbirk et al.,1993; von Kries et al., 1987b). Vitamin K content of colostrum is slightly higher than that of mature milk, but concentrations do not vary significantly through the first 6 months of lactation (Canfield et al., 1991; Greer et al., 1991; von Kries et al., 1987b). Vitamin K content of human milk can be increased by maternal intakes of pharmacological doses of vitamin K (Greer et al., 1991; Haroon et al., 1982). In a study by Greer and coworkers (1997), supplementing mothers with a dose of 5 mg/day of phylloquinone for 12 weeks increased the vitamin K concentration of milk by 70-fold (82.1 versus 1.17 μg/mL in unsupplemented mothers). In another study, supplementing one mother with 0.1 mg/day of phylloquinone, an amount that can be obtained in the diet, raised milk concentrations from 2.5 to 4.9 ng/L (von Kries et al., 1987b).

TABLE 5-1

Vitamin K in Human Milk.

Ages 0 through 6 Months. The AI for infants 0 through 6 months of age is based on a reported average intake of milk of 0.78 L/day (Chapter 2) and on an average phylloquinone concentration of 2.5 mg/L in human milk. This gives an AI of 2.0 μg/day after rounding. The AI assumes that infants also receive prophylactic vitamin K at birth in amounts suggested by the American and Canadian pediatric societies. The AI agrees with intakes of 0.6 to 2.3 μg/day reported for infants exclusively fed human milk (Canfield et al., 1991; Greer et al., 1991; Pietschnig et al., 1993).

The AI is significantly lower than recently reported intakes based on the Food and Drug Administration Total Diet Study of 77 μg/ day (Booth et al., 1996b) and 111 μg/day (Appendix Table E-1). However, these values exclude intakes from human milk and include only intakes from other food sources. Booth and coworkers (1996b) reported that 87 percent of the estimated intake of 77 μg/ day was attributed to infant formulas, which on average contain 50 to 100 mg/L of phylloquinone (Greer, 1995; Haroon et al., 1982).

Ages 7 through 12 Months. Using the method described in Chapter 2 to extrapolate from the AI for infants ages 0 through 6 months, the AI for older infants is 2.5 μg/day after rounding. Because complementary foods become a more important part of the infant diet in the second 6 months of life, vitamin K intake for this age category is expected to be higher than the AI based solely on human milk consumption. However, data concerning the vitamin K content of weaning foods and their contribution to daily vitamin K intake are not presently available. Alternatively, if the adult AI of 80 μg/day is extrapolated down by means of the method described in Chapter 2, the AI would be 23 μg/day. Because older infant vitamin K intakes of 2.5 μg/day have not been associated with adverse clinical outcomes (Greer et al., 1991), the AI is set at the level obtained by extrapolating up from young infants.

Children and Adolescents Ages 1 through 18 Years

Adults Ages 19 Years and Older

Method Used to Set the Adequate Intake

Clinically significant vitamin K deficiency is extremely rare in the general population, with cases being limited to individuals with malabsorption syndromes or those treated with drugs known to interfere with vitamin K metabolism. The recent development of indicators sensitive to vitamin K intake, though useful to describe relative diet-induced changes in vitamin K status, were not used for establishing an EAR because of the uncertainty surrounding their true physiological significance and the lack of sufficient dose-response data.

Therefore, the AI for adults is based on reported vitamin K dietary intakes in apparently healthy population groups. In a recent paper, Booth and Suttie (1998) reviewed 11 studies in which phylloquinone intakes ranged from 61 to 210 μg/day with average intakes of approximately 80 μg/day for adults younger than 45 years and approximately 150 μg/day for adults older than 55 years (Table 5-2). NHANES III data (Appendix Table C-10) indicate that median vitamin K intakes of adults varied between 82 and 117 μg/day.

TABLE 5-2

Dietary Phylloquinone Intake in Healthy Men and Women.

Studies have demonstrated that abnormal PIVKA-II concentrations were observed in individuals consuming 40 to 60 μg/day of vitamin K but were normal when intakes were approximately 80 μg/ day (Jones et al., 1991; Suttie et al., 1988). Healthy individuals with phylloquinone intakes approaching 80 μg/day have been investigated and have shown no signs of a deficiency, a finding that suggests this level of intake is probably adequate for the majority of the adult population (Bach et al., 1996; Ferland et al., 1993; Suttie et al., 1988). Reported vitamin K intakes are slightly lower for women than men (Booth et al., 1996b; Appendix Table C-10).

Reported phylloquinone intakes of older adults have generally been higher than those of younger individuals, a finding explained by their higher intakes of vegetables (Booth et al., 1996b). Older subjects have been found to be more resistant to vitamin K deficiency than younger adults (Ferland et al., 1993).

The AI is based on median intake data from NHANES III (Appendix Table C-10). Because dietary intake assessment methods tend to underestimate the actual daily intake of foods, the highest intake value reported for the four adult age groups was used to set the AI for each gender; numbers are rounded up to the nearest 5 μg.

Pregnancy

Lactation

Food Sources

Early data obtained by chick bioassay on the vitamin K content of foods were unreliable and a limited number of foods were assayed. Over the last decade, rapid and reliable chromatographic procedures for vitamin K have been developed, and data for the phylloquinone content of most commonly consumed foods are available (Booth et al., 1995). Only a relatively small number of food items (Table 5-3) contribute substantially to the dietary phylloquinone intake of most people. A few green vegetables (collards, spinach, and salad greens) contain in excess of 300 μg of phylloquinone/100 g, while broccoli, brussels sprouts, cabbage, and bib lettuce contain between 100 and 200 μg of phylloquinone/100 g. Other green vegetables contain smaller amounts. Plant oils and margarine are the second major source of phylloquinone in the diet. The phylloquinone content of plant oils is variable, with soybean and canola oils containing greater than 100 μg of phylloquinone/100 g. Cottonseed oil and olive oil contain about 50 μg/100 g, and corn oil contains less than 5 μg/100 g. Prepared foods contain variable amounts of vitamin K depending on their content of green vegetables and the source and amount of oil used in their preparation. Information relative to the important food sources of vitamin K for infants and children of various age groups and for adults by gender and age group are available from data obtained from the U.S. Food and Drug Administration Total Diet Study (Booth et al., 1996b). Spinach, collards, broccoli, and iceberg lettuce are the major contributors of vitamin K in the diet of U.S. adults and children.

TABLE 5-3

Phylloquinone Concentration of Common Foods.

Hydrogenation of plant oils to form solid shortenings results in some conversion of phylloquinone to 2′,3′-dihydrophylloquinone (Davidson et al., 1996). This form of the vitamin is most prevalent in margarines, infant formulas, and prepared foods (Booth et al., 1996a) and can represent a substantial portion of the total vitamin K in some diets. The bioavailability and the relative biological activity of dihydrophylloquinone have not been determined with any certainty. The long-chain menaquinones, which are produced in substantial amounts by intestinal microorganisms, can also serve as active forms of vitamin K, but they are not widely distributed in commonly consumed foods. Green vegetables and plant oils, the major dietary sources of vitamin K, do not contain menaquinones, and only small amounts are found in animal products. Relatively large amounts (40–80 μg/100 g) can, however, be obtained from some cheeses (Schurgers et al., 1999).

Dietary Intake

The availability of reliable data on the vitamin K content of foods has now made it possible to obtain reasonable estimates of the dietary phylloquinone intake of the North American population. The results of a number of studies on phylloquinone intake that used dietary records, with or without recall or a food frequency questionnaire, have been summarized by Booth and Suttie (1998) and are presented in Table 5-2. These data are somewhat variable but indicate a mean phylloquinone intake of about 150 μg/day for older (above 55 years) and 80 μg/day for younger men and women. Gender differences were not apparent in these studies.

Data from nationally representative U.S. surveys are available to estimate vitamin K intakes (Appendix Tables C-10, C-11, E-1). Data from the Third National Health and Nutrition Examination Survey (NHANES III) shows that median intakes of dietary vitamin K ranged from 79 to 88 μg/day for women and 89 to 117 μg/day for men (Appendix Table C-10). Because of the relatively small number of foods that provide significant amounts of phylloquinone in the diet, the daily variation in intake is high, and Booth and coworkers (1995) have estimated that a 5-day record of intake is needed to get a true measure of dietary intake. Data on phylloquinone intake have recently been calculated (Booth et al., 1999c) from 14-day food diaries collected by the Market Research Corporation of America. These data reflect the intake of nearly 4,000 men and women aged 13 years or older with a demographic profile similar to that of the U.S. census. These data clearly demonstrate the large daily variation in phylloquinone intake and indicate an average intake of 70 to 80 μg/day for the U.S. adult population. The same data provide an estimate of the dihydrophylloquinone intake of this population (19 μg/day for men and 15 μg/day for women) that is about 20 to 25 percent as much as the intake of phylloquinone.

The adult phylloquinone intake in The Netherlands has been reported to be two to three times higher than that of the U.S. population (Schurgers et al., 1999). Whether this represents a real difference in the consumption of phylloquinone-rich foods or differences in methods used to estimate foods consumed is not known at the present time. This study has also provided an estimate of the average intake of long-chain menaquinones of 21 μg/day. Comprehensive data on menaquinone intake are not available for the U.S. population.

Intake from Supplements

The median intakes of vitamin K from food and supplements for the four adult age groups was 93 to 119 μg/day for American men who took supplements (Appendix Table C-11). The median vitamin K intake from food and supplements for women who reported consuming supplements was 82 to 90 μg/day.

Hazard Identification

No adverse effects associated with vitamin K consumption from food or supplements have been reported in humans or animals. Therefore, a quantitative risk assessment cannot be performed and a UL cannot be derived for vitamin K.

A search of the literature revealed no evidence of toxicity associated with the intake of either the phylloquinone or menaquinone forms of vitamin K. A synthetic form of vitamin K, menadione, has been associated with liver damage (Badr et al., 1987; Chiou et al., 1998) and therefore is no longer used therapeutically.

One study showed a significant association between intramuscularly (IM) administered vitamin K and childhood cancer, particularly leukemia (Golding et al., 1992). This study compared 195 children diagnosed with cancer between 1971 and 1991 and born in one of two major hospitals (between 1965 and 1987) with 558 controls. Golding and coworkers (1992) reported a significant association between IM vitamin K and cancer incidence (p = 0.002; observed risk, 1.97; 95 percent confidence interval, 1.3–3.0). No significantly increased risk was reported for children who had been given oral vitamin K. These findings on IM vitamin K doses have limited relevance to ULs based on oral intake.

Furthermore, evidence from other population studies fails to confirm an association between vitamin K and cancer (Ansell et al., 1996; Klebanoff et al., 1993; McKinney et al., 1998; Parker et al., 1998; Passmore et al., 1998). In a nested case-control study that used data from a large, multicenter prospective study (54,795 children), Klebanoff and coworkers (1993) found no association between vitamin K exposure and an increased risk of any childhood cancer or of all childhood cancers combined. Ansell and coworkers (1996) assessed associations between leukemia and prenatal and neonatal exposures and failed to show an increased risk of childhood leukemia in neonates receiving IM-administered vitamin K. The findings of Ansell and coworkers (1996) were confirmed by three similar case-control studies (McKinney et al., 1998; Parker et al., 1998; Passmore et al., 1998).

Data from animal models have shown no toxicity of vitamin K (NRC, 1987). No adverse effects were reported with administration of up to 25 g/kg of phylloquinone either parenterally or orally to laboratory animals (Molitor and Robinson, 1940).

Dose-Response Assessment

The data on adverse effects from high vitamin K intakes are not sufficient for a quantitative risk assessment, and a UL cannot be derived.

Intake Assessment

The highest intake of dietary vitamin K reported for the U.S. population was 340 μg/day in women aged 19 through 30 years (Appendix Table C-10). The highest intake of vitamin K from food and supplements was 367 μg/day, also in women aged 19 through 30 years (Appendix Table C-11).

Risk Characterization

No adverse effects have been reported with high intakes of vitamin K.