More Complex Studies and Additional Resources for Research

Differences between the sexes may be modest, yet the differences may still result in different outcomes or may point to significant and larger underlying mechanisms of sex differences. Detection of modest differences may require studies with more complex experimental designs, the use of more complex model systems, and the use of more subjects to achieve statistical power; and, thus, in some cases detection of modest differences may require additional financial resources.

A recent example is a large population-based study (O'Donnell et al., 1998) that demonstrates an association and genetic linkage of the angiotensin-converting enzyme (ACE) locus with hypertension and with diastolic blood pressure in men but not women. Existing data on the association of the ACE locus and blood pressure or hypertension have been conflicting. O'Donnell and colleagues have suggested that “many prior studies may have had inadequate power to detect the modest contribution that might be expected from an individual genetic factor to complex traits such as blood pressure” (O'Donnell et al., 1998, p. 1766). They also note that differences in the genetic makeup or environmental exposures of the different study populations may also play a role in the different results observed.

Another example relates to sex differences in sensitivity to noxious stimuli, which overall are relatively small but whose underlying mechanisms are wide-ranging and significant for many aspects of treatment (see Chapter 4 and Berkley [1997a,b]).

RECOMMENDATION 8: Support and conduct additional research on sex differences.

Because differences between the sexes are pervasive across all subdisciplines of biology, all research sponsors should encourage research initiatives on sex differences. Research sponsors and peer-review committees should recognize that research on sex differences may require additional resources.

Information from the Published Literature

Information on sex differences can be difficult to glean from the published literature. In 1999, the U.S. Congress asked GAO to assess the National Institutes of Health's (NIH's) progress in conducting research on women's health in the decade since publication of the 1990 GAO report on the inclusion of women in clinical trials. Over the course of its audit, GAO experienced difficulties in determining whether analysis of data by sex is occurring. Even when scientists do analyze data by sex, the results are not always published. Often, when an analysis reveals no difference in outcome by sex, journal editors omit this information and generally discourage researchers from including such information in the results (U.S. General Accounting Office, 2000). GAO noted that the deficiency in reporting the outcomes of studies by sex appeared to be due in part to a lack of documentation of negative findings. In addition, when sex differences are reported, they may not be appropriately indexed by abstracting and indexing services, making it difficult to identify articles of interest (Montgomery and Sherif, 2000).

In fact, very few journals have any information or guidelines regarding analysis of data by subpopulations, such as sex, in their instructions for authors. Many journals refer to the “Uniform Requirements for Manuscripts Submitted to Biomedical Journals” by the International Committee of Medical Journal Editors, which states that when writing about methods, researchers should describe their “selection of the observational or experimental subjects (patients or laboratory animals, including controls) clearly. Identify the age, sex, and other important characteristics of the subjects” (International Committee of Medical Journal Editors, 1997, p. 311). However, the uniform requirements do not say anything about “age, sex, or other important characteristics of subjects” in the requirements for the statistics, results, or discussion sections of articles.

The Journal of the National Cancer Institute (JNCI) has taken the lead in encouraging authors to include information on and analyses of subjects by sex. In October 2000, JNCI amended its Information for Authors to say, “Where appropriate, clinical and epidemiological studies should be analyzed to see if there is an effect of sex or any of the major ethnic groups. If there is no effect, that should be stated so in Results” (Arnold, 2000; Journal of the National Cancer Institute, 2000). On the basis of the recent GAO report and other related reports, encouraging the inclusion of “negative data” (the absence of a finding of a sexual dimorphism may be an indication of a similarity between the sexes) would be an important step. The phrase “where appropriate” regarding the need for analysis by sex, however, may be interpreted and applied quite differently by different researchers.

RECOMMENDATION 9: Make sex-specific data more readily available.

Journal editors should encourage researchers to include in their reports descriptions of the sex ratios of the research population and to specify the extent to which analyses of the data by sex were included in the study. If there is no effect (absence of a sex difference), that should be stated in the results and researchers should give an informative statistic (e.g., means and standard deviations, t-test value for comparison, exact P value) rather than simply indicating that the comparison was not statistically significant. Reporting of exact statistics allows the findings to be taken into account far more effectively in later meta-analyses. When designing or updating databases of scientific journal articles and other information, informatics developers should devise ways of reliably accessing sex-specific data.

Information on Sex of Origin of Cell and Tissue Culture Material

All somatic cells have a full complement of chromosomes, including the sex chromosomes. Despite this, useful information on the sex of origin of cell lines or tissue cultures is often lacking in the literature. Data suggest that many cells throughout the body display sex specificity. For example, human primary osteoblasts (a type of bone cell) isolated from preand postmenopausal women show age-dependent changes in their expression of biochemical markers and responsiveness to hormones that are not observed in the same bone cells from younger or older men (Katzburg et al., 1999). Receptors for sex hormones are present on a wide variety of cells. Arterial wall macrophages derived from circulating blood monocytes are known to take up lipoproteins and to form large foam cells. Such foam cell formation is an early event in atherosclerosis, and recent research suggests that the earlier onset and higher incidence of atherosclerosis in men may be related to the presence of androgen receptors on the immune cells and androgen-mediated lipid loading (McCrohon et al., 2000).

A great deal of research on human cell lines involves newborn foreskin cells, which are obligatorily male. There is no equivalent source of female newborn tissue. Reviewers often ask whether the genital location of foreskin fibroblasts or keratinocytes modifies their behavior and makes them invalid models for the behaviors of cells derived from a nongenital location, but reviewers do not ask if male versus female sex is similarly relevant. The fetal fibroblast-like cell lines commonly used in aging research (e.g., WI-38 and IMR-90 cells) are nowhere characterized as male versus female, nor has anyone questioned the relevance of cellular sex in studies with these lines. (The large number of malignant immortal cell lines available from the American Type Culture Collection and similar sources are rarely identified by sex, but even if they were it would be extremely difficult to interpret differences among donor lines, as the expression of many genes is presumably abnormal by virtue of the malignant transformation process.)

Few to no data are available for determination of whether cell sex matters when studying such functions as cell life span and the response to stresses. Information about the sex of origin of cells and tissues, however, could be very useful and even critical for interpretation of results and for comparison of results across disciplines. Because the sex of the source of eukaryotic cells used in research can readily be determined by DNA sequencing, it need not remain an unknown. Many published studies, however, do not report such information even when it is known.

RECOMMENDATION 10: Determine and disclose the sex of origin of biological research materials.

The origins and sex chromosome constitutions of cells or tissue cultures used for cell biological, molecular biological, or biochemical experiments should be stated when they are known. Attempts should be made to discern the sex of origin when it is unknown. Journal editors should encourage inclusion of such information in Materials and Methods sections as standard practice.

(The committee acknowledges that inclusion of people, animals, or cells and tissues of or from both sexes in all studies is not always feasible or appropriate. Rather, the committee is urging researchers to regard sex, that is, being male or female, as an important basic human variable that should be considered when designing, analyzing, and reporting findings from studies in all areas and at all levels of biomedical research. Determining and disclosing the sex of origin of biological research materials are important steps in that direction.)

Lack of Data on Sex Differences Across the Life Span from Longitudinal Studies

The health status of males and females can vary considerably, both within and between the sexes, across the life span—from intrauterine development to old age. Several longitudinal studies, some spanning more than 40 years, have provided vitally important data that demonstrate sex differences, ranging from genetic differences to differences in diagnostic and therapeutic interventions, as individuals age (such as the Framingham Study and the Bogalusa Heart Study [Berenson et al., 1992; Dawber et al., 1951; Kannel et al., 1976; O'Donnell et al., 1998]). These studies have provided such information not only about the original participants but about their offspring as well. Those studies, however, were designed with specific disease end points, such as the risk factors for and the development of coronary artery disease, thereby precluding consideration of many other relevant developmental issues and other diseases, disorders, and conditions. Unfortunately, few such longitudinal studies exist. As a result, the lack of longitudinal studies has limited understanding in particular of sex differences throughout the life span.

RECOMMENDATION 11: Longitudinal studies should be conducted and should be constructed so that their results can be analyzed by sex.

Lack of Consideration of Hormonal Variability

Data on cycles (menstrual, circadian, etc.) are often lacking. Most studies with women do not define which part of the cycle the participants were in at the time of study, note the participants only by age and not whether they are pre- or postmenopausal, or are based on only one cycle.

Furthermore, characterization of the phases of estrous and menstrual cycles across studies is inconsistent and can lead to considerable confusion. For example, in studies with humans some investigators assess a phenomenon or condition during the luteal phase and take their measures near the end of that phase, when progesterone levels are dropping or are at low levels, whereas others take their measures near the midpoint of the luteal phase, when progesterone levels are rising or are at their peak.

Problems with timing the menstrual cycle for research are not insurmountable, and several studies (including studies determining timing of the menstrual cycle for surgery for breast cancer prognosis) are under way (Hagen and Hrushesky, 1998; Jatoi, 1998; Macleod et al., 2000).

RECOMMENDATION 12: Identify the endocrine status of research subjects (an important variable that should be considered, when possible, in analyses).

Uniformity in Application of Federal Regulations

Interpretation and application of federal regulations regarding protection of human subjects are at the discretion of university and industry institutional review boards (IRBs) and can be highly variable. For example, current federal guidance describes policies for inclusion of women in research protocols that may be applied more restrictively at the discretion of local IRBs. Similarly, the policies for the use of research animals set by institutional animal care and use committees (IACUCs) vary by institution. Thus, the ability to study sex differences is not uniform across institutions, and there is a question of fairness, as some oversight bodies allow procedures that others do not. This lack of uniformity can inhibit collaborative work between institutions with different IRB and IACUC regulations.

Uncertainties about how best to protect women in their childbearing years contribute significantly to the inconsistencies in the application of guidelines. A1999 statement by the Committee on Ethics of the American College of Obstetricians and Gynecologists (ACOG) notes the valid concerns for the well-being of the fetus but states that “the resulting sense of restraint has limited the growth of knowledge in some areas of normal physiology, pathophysiology, and therapeutic approaches during pregnancy” (Chervenak and McCullough, 1999, p. 206). The ACOG committee affirmed the need for research with pregnant women and offered guidelines for selecting, informing, and caring for pregnant women in clinical research studies.

IRBs are a topic of great interest to the members of the Institute of Medicine (IOM) Clinical Research Roundtable (www.iom.edu/crr). In addition, the Office of Human Research Protection of the U.S. Department of Health and Human Services has commissioned IOM to do a two-part study on the future of IRBs. The first report will be a 6-month fasttrack study that addresses accreditation standards for IRBs (expected release date, summer 2001). The second part will be a full-length study of the overall structure and functioning of activities used to protect human subjects and the criteria used to evaluate the performance of activities used to protect human subject (expected release date, fall 2002).

The committee is encouraged by the attention that the problems of IRBs will receive from these two IOM activities and looks forward to their published proceedings and recommendations.

Opportunities for Interdisciplinary Collaboration

The need for greater use of interdisciplinary collaboration in not unique to the study of sex differences. The committee refers the reader to the recent IOM report Bridging Disciplines in the Brain, Behavioral, and Clinical Sciences (Institute of Medicine, 2000). Although the report is primarily concerned with brain research, the discussion of the issues and barriers related to interdisciplinary research and training and to translational research is relevant to the study of sex differences as well. The committee that prepared that report concluded that researchers sometimes perceive interdisciplinary research as impure or “second rate,” with each discipline having a sense of superiority and believing others to be less rigorous. Moreover, students fear that interdisciplinary training will not properly prepare them for a career.

The National Research Council's Committee on National Needs for Biomedical and Behavioral Sciences noted in its recent report (National Research Council, 2000) that the number of researchers in the basic biomedical workforce is sufficient to meet the nation's needs. It also encouraged NIH to gradually shift the focus of its predoctoral programs from single-discipline to interdisciplinary training and increase opportunities for postdoctoral training through interdisciplinary training grants.

In the study of sex differences, synergy is needed between and among basic scientists, epidemiologists, social scientists, and clinical researchers. There is a need for better translational—or bench-to-bedside—research. Data on interlevel integration (cellular, to animal, to human) are often limited. At the human level, collaboration across medical specialties is also needed.

As noted in a 1994 IOM report, Women and Health Research, such collaborations can be difficult as researchers from different disciplines look at different points along a causal pathway, use different tools, and obtain different results (Institute of Medicine, 1994). Strategies for more effective communication and cooperation are needed.

As an example of mutual benefits that can come from interdisciplinary research, consider epidemiology and basic biological research. Epidemiological studies can be used to identify areas in which basic research is needed. Information about environmental exposures that affect disease risk and pathogenesis could point to genes and biological mechanisms that modify risk. In the opposite and equally beneficial scenario, understanding the biological basis of differences or similarities could help identify unrecognized environmental risk factors (Hoover, 2000). Collection of better data on exposures, including multiple exposures, could enhance analysis.

RECOMMENDATION 13: Encourage and support interdisciplinary research on sex differences.

Interdisciplinary research is generally accepted as valuable and important. Opportunities for interdisciplinary collaboration to enhance the understanding of sex differences, however, have not been fully realized. The committee recommends the continued development of interdisciplinary research programs and strategies for more effective communication and cooperation to achieve the following goals:

• synergy between and among basic scientists, epidemiologists, social scientists, and clinical researchers;
• enhanced collaboration across medical specialties; and
• better translational—or bench-to-bedside—research and interlevel integration of data (cellular, to animal, to human).

Lack of Awareness That the Consequences of Genetics and Physiology May Be Amenable to Change

A problem with discussing sex differences is that such differences, when shown to have a genetic or physiological influence, are thought to be immutable. Such sex differences are interpreted as being a direct result of chromosomes. Since an individual's sex chromosomes, either XX or XY, cannot be changed, people tend to think that differences between the sexes are also unchangeable (Valian, 1998). Valian (1998) notes that “it is odd that we should interpret sex differences as immutable, when we do not accept biology as destiny in other aspects of human existence” (p. 67) and cites aging as an example. The human life span is set by biology, but people do not accept the average life span as their fate. Biology incorporates mutually interacting factors, from genetic to psychosocial, across the life span. Great amounts of time and many resources are put into research on understanding mechanisms of disease and developing cures. People eat certain foods, take vitamins, exercise, and take preventive treatments such as vaccines, all in attempts to prolong life.

Discriminatory Practices Based on Sex Differences

Historically, studies on race, ethnicity, age, nationality, religion, and sex have led to discriminatory practices. Participants in the Research Designs and Gender working group at the NIH Scientific Meeting and Public Hearing on Influences of Sex and Gender on Health in 1996 noted that in the past it has been difficult to publish findings on sex differences. “The paucity of studies on neuroendocrine aspects of female functions may be attributable to a belief that it was not politically correct to acknowledge different functioning at different times of the month for fear of discrimination based on the differences” (National Institutes of Health, Office of Research on Women's Health, 1999d, p. 97).

Scientific research is not separate from other practices of society. Bias in interpretation or reporting by observers is well documented in the social sciences but is less obvious in the basic and clinical sciences. Yet, scientific theories, questions, and data are also subject to biased interpretation (Institute of Medicine, 1994). Such biases and the tendency to perceive one sex (classically, males), race, age group, or other subpopulation to be the norm can portray differences displayed by others as “deviant” (Institute of Medicine, 1994). To acknowledge and explore nonreproductive aspects of sex differences in health and illness requires a view of males and females as different but equally “legitimate” biological entities without respect to whether there is (or should be) equivalence or equality in all other domains.

RECOMMENDATION 14: Reduce the potential for discrimination based on identified sex differences.

The committee noted that, historically, studies on race, ethnicity, age, nationality, religion, and sex have sometimes led to discriminatory practices. The committee believes, therefore, that these historical practices should be taken into consideration so that they will not be repeated. The past should not limit the future of research but should serve as a guide to its use. Ethical research on the biology of sex differences is essential to the advancement of human health and should not be constrained.