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

National Research Council (US) Panel on Race, Ethnicity, and Health in Later Life; Bulatao RA, Anderson NB, editors. Understanding Racial and Ethnic Differences in Health in Late Life: A Research Agenda. Washington (DC): National Academies Press (US); 2004.

Cover of Understanding Racial and Ethnic Differences in Health in Late Life

Understanding Racial and Ethnic Differences in Health in Late Life: A Research Agenda.

Show details

3Genetic Influences

The involvement of genetic factors in racial and ethnic differences in health and disease is currently the focus of intense scrutiny. There are numerous dimensions to this issue. Some have questioned the utility or meaningfulness of the race concept in addressing health issues (e.g., Wilson et al., 2001). Editors have echoed this concern (Anon., 2000, 2001; Schwartz, 2001) and required that submitted manuscripts explain and justify the study of particular ethnic groups or populations (Anon., 2000). At the same time, other researchers report that conventionally defined racial groups differ in genetic factors that affect risk for specific diseases or sensitivity to therapeutic drugs (e.g., Exner et al., 2001; Karter et al., 2002; Splawski et al., 2002; Taioli et al., 1995). Still others have presented the case that racial categories may have biological meaning (Crow, 2002) and can be valuable in biomedical research (Risch et al., 2002).


The term race is used in a great variety of ways. Common usage may differ from that of policy makers, and scholarly usage may also vary considerably. Genetic conceptualizations of race, in particular, make reference to differences between and among populations in gene frequencies. The subdiscipline of population genetics is explicitly concerned with such differences and with the dynamics of processes, such as mutation, differential survival, the reproduction of particular gene variants, gene flows between populations through migration, and similar matters. The models for conceptualizing and describing these dynamics are highly developed and are clearly central to the taxonomy of human beings.

A key concept is that of genetic distance, of which several measures have been proposed. Basic to these is the probability that a randomly selected allele (see below) from one population will be identical to another allele from the same population but not to one chosen from a different population (see, e.g., Hartl and Clark, 1997). Measures of distance provide metrics that can be used to decide if groups differ sufficiently to be regarded as separate races. But different criteria may be used with respect to how large the genetic distance must be to constitute a racial divide. Some racial classification schemes, therefore, feature many races with small interracial differences, and others identify only a few larger and more separated groups. Each scheme serves its own purposes, and no classification can be regarded as necessarily more correct than others.

In any scheme, the idea of racial identity is a probabilistic one. Different groups are not delineated by clear and unambiguous borders but shade into each other along gradients. Strict categories, exclusive and exhaustive, in which each human being is assignable without ambiguity to one and only one race on the basis of genetic characteristics, are unattainable. Thus, genetics cannot provide a single, definitive human classification scheme with which to address the many facets of health differences.

At the same time, race, as it is socially and politically defined, is a powerful social and political reality. Thus, it is worth asking whether there are identifiable gene differences among socially defined populations or racial groups. The answer to this question requires a distinction between the concepts of genetic loci and genetic alleles. The term locus had its origin at a stage of genetic science when it became evident that the hypothetical elements of Mendelian theory had a physical existence and were located at particular sites, or loci, on chromosomes. A basic postulate of Mendelian theory was that these hypothetical elements (later to be termed genes) could exist in different forms, which came to be termed alleles. Thus, a key question is whether socially defined populations differ in the number or types of loci they possess.

Although it is difficult to prove a universal negative, the evidence is persuasive that all humans possess the same array of genetic loci. The genetic differences among people as individuals consist in the particular array of alleles at these loci; the genetic differences between or among populations can be described in terms of the relative frequencies of the different alleles in the populations. Decades of research in population genetics have documented differences among populations in allelic frequencies for a large number of genetic loci, including many that are health- or disease-related. Of greatest salience for our present topic is the additional observation that the range of genetic differences among individuals within populations usually exceeds by far that between groups. For instance, Hartl and Clark (1997) present an analysis of data from Nei (1975) with respect to a group of enzyme variants, showing that only about 7 percent of the total genetic variation of Caucasoid, Negroid, and Mongoloid racial groups could be assigned to differences among the groups; 93 percent of the variation was within groups. The important consequence is that membership in a particular socially defined race is not a reliable or unambiguous marker for possession of any particular allele (except, perhaps, for such characteristics as skin color).

The Nature of Genetic Influence

The literature also presents a picture of great complexity in the relationships of genes to racial and ethnic health differences. First, it is essential to understand that genetic influence is mediated through a causal nexus comprising the anatomy, biochemistry, physiology, immunology, endocrinology—all of the complex of structures and functions—of the human organism. This is, of course, the same causal nexus through which environmental influences are mediated.

Such complex causal networks are usually characterized by converging and diverging processes derived from multiple inputs from both genetic sources and environmental sources. Multiple redundant pathways with feedback loops may provide for compensatory mechanisms, in which one pathway adjusts for inadequacies in another pathway. Consequences of differences or changes in one of the input elements can be enormously influenced by both genetic and environmental context. The dynamics of these complex causal nexuses underlying health and disease phenotypes is frequently manifest as interactions of genes with genes, genes with environments, or environments with environments.

The resulting causal field is not homogeneous, of course. Some of the inputs and some of the mediating pathways may have a greater impact than others. Some genetic and some environmental variables are sufficiently potent that their influence is apparent in all or most system contexts. In systems terms, there may be nodes within the network that are describable as having a broad “span of control” (Simon, 1973) or as being “soft spots” (Waddington, 1977). In genetic parlance, a gene providing input to a node of this kind may be referred to as a major gene, or a Mendelian gene, or, recklessly, as “the gene” for this or that phenotype. Even the effects of such a major gene, however, may be subject to fine tuning by its companion genes or by the environmental milieu in which all of them operate. Some long-recognized genetic phenomena represent such effects: variable expressivity, in which individuals with the same genotype at a particular locus display different levels of the phenotype associated with the locus; incomplete penetrance, such that some individuals with the genotype do not show the phenotype at all; and variable time of onset, in which individuals with the same disease genotype vary widely in the age at which symptoms appear.

In respect to health and disease, genetic loci for which particular allelic configurations result with high reliability in a diagnosable disease outcome are often described as “disease genes.” However, particular genotypes at other genetic loci may affect the same general categorical disease state. Thus, many disease genes, though they may be sufficient, are not necessary for the expression of the disease phenotype. In other cases, as in the association between the ε4 allele of the apoE locus and Alzheimer's disease (discussed below), the identified allele is neither necessary nor sufficient. In different populations, the same disease state may be achieved by different genotypes affecting different mediating mechanisms.

Other disease states may not be categorically distinct from the normal range of variability of some anatomic structure or physiological process but may constitute extremes of such a dimension. The location of an individual in continuously distributed phenotypes is typically due to the action of many genetic loci, as well as many environmental variables. Any particular value of such a continuously distributed phenotype can be achieved by numerous combinations of input variables. The genetic input is described as polygenic. Analysis of such systems is generally statistical, with a central theme being the attribution of proportions of the measured phenotypic variability into two broad categories, genetic and environmental, and, depending on the research design, into more detailed categories, such as additive or nonadditive genetic effects and shared or nonshared environmental influences.

Gene Interactions

In considering ethnic identity and health, genes may be relevant in two broad senses. First, the gene pools of different ethnic groups may contain different frequencies of alleles at some loci that are pertinent to health status or to disease processes. However, such differences by themselves are unlikely to account for broad and pervasive health differences among socially identified racial and ethnic groups.

Second, the phenotype consequent on a given genotype may vary between ethnic groups because of interactions with environmental factors. The environment, in this context, is defined by exclusion, as all influences not coded in DNA. It thus covers all the other factors noted in Chapter 2, including prenatal effects, nutritional influences, the preventive consequences of health care, peer group pressures, educational level, religious instruction, toxins in homes and in the air and water, occupational hazards, job stress, and exposure to infectious agents, among many, many others.

Much is known about the etiological significance of a vast array of such environmental factors; much also is known about the influence of major genes and of polygenic systems. Conceptually, the possibility of interactions within and between these two broad domains has long been recognized. For various reasons, research emphasizing and characterizing these interactions has been less well developed than might be expected. Their implications for health differences are not yet known, though the accumulated literature, both from human and animal model research, is substantial. Only a few examples are cited here, but they should illustrate the great complexity and power and the sometimes astonishing subtlety of these interactions.

In human beings, interaction between two major genes is implicated in the etiology of the large and burgeoning health problem of Alzheimer's disease. Three different alleles—ε2, ε3, and ε4—have been described at the apoE locus on chromosome 19. In general, possession of one ε4 allele is associated with an increased risk of developing Alzheimer's disease, and possession of two confers a greater risk than possession of one. This latter outcome, however, depends on the genotype at another locus, ACT. In the case of one genotype at that locus, there is no difference in risk of having one or two ε4 alleles at the apoE locus; for another ACT genotype, the risk is somewhat elevated; and for the third, the difference in risk between one and two ε4 alleles is fivefold. Clearly, when considering differences in allelic frequency in different populations, it may be necessary to be concerned with dyads, triads, or larger collectives of loci.

A classic animal model study showing that the effect of different genotypes at a major locus can be modified by the polygenic background of the organism is the work of Coleman and Hummel (1975). Two copies of a particular allele at a specific locus generally lead to some manifestation of diabetes in mice, but in two different but related strains the resulting syndromes are strikingly different, with blood glucose levels and body weight differing twofold, large differences in lifespans, and Islet hypertrophy in one strain and atrophy in the other.

Perhaps the prototypic illustration of interaction between polygenes and the environment is that of Cooper and Zubek (1958), who measured the maze learning ability of two lines of rats reared under environmental conditions that differed in the variety of stimuli that the animals could experience. The two strains had been selectively bred for maze performance (Heron, 1935); the resulting “maze-bright” and “maze-dull” lines differed strikingly in the number of errors committed in learning the maze pattern, and, by strong inference, in terms of allelic configurations at an unknown number of polygenic loci pertinent to maze performance. The results of differential rearing were that the bright lines did not profit from enrichment, but the dull ones did; the dull rats were not adversely affected by impoverishment, but the bright ones were. Numerous other studies have shown similar differential responses in a variety of phenotypes to environmental manipulations by groups of mice or rats of differing genotypes.

Another striking recent example of gene-environment interaction is provided by the study of quantitative trait loci (QTLs) affecting longevity in Drosophila flies. QTLs are loci that remain anonymous at present, but whose approximate chromosomal locations are known. Vieira et al. (2000) sought evidence for the effect of such loci on the length of life under five different environmental conditions of rearing. The extraordinary result was that 17 QTLs were identified, but none was pertinent to all environments. Some were effective in one sex only and in one environment; others were effective in both sexes in a specific environment, but the same allele was associated with longer lives in one sex and shorter lives in the other sex; some were effective in one sex in two environments, but with the same allele associated with longer lives in one environment and shorter lives in the other. All of the genetic variance was involved in genotype x sex interactions, genotype x environment interactions, or both.

Within the general domain of coaction of genes and environmental factors, there are several lines of investigation that convincingly demonstrate that environments not only can interact in a statistical sense with genetic factors, but can also actually influence which genes are expressed. In an oversimplified explanation, some environments can turn genes on and off. Certain subdomains of this research are of particular potential relevance for the present topic, dealing with the effects of stress of various sorts on gene expression. For instance, an extensive body of literature (summarized, for example, by Hoffman and Parsons, 1991) describes observations that suggest that stressful environments often increase the heritability—the proportion of phenotypic variance attributable to the collective influence of a polygenic system—of a wide variety of phenotypes in a wide array of organisms. A major body of data dealing with specific genes concerns the “heat-shock” proteins that are produced in Drosophila after exposure to a high-temperature environment. These proteins appear to protect other proteins in the organism from damage or destruction by the stressful environment. An example from mammals is the increase in the levels of specific RNAs in the adrenal glands of rats following immobilization stress (McMahon et al., 1992). Biobehavioral influences are clearly implicated by a study showing that classical Pavlovian conditioning—pairing foot-shock and an auditory stimulus—can result in a previously neutral feature of the environment acquiring the capability of eliciting a stress-related expression of a particular mRNA in regions of the brain of rats (Smith et al., 1992). These lines of research are perhaps particularly relevant to hypotheses concerning the role of discrimination stress, such as those of Thayer and Friedman (2004).

Some Racial and Ethnic Differences

These various studies offer conceptual lenses for considering evidence of genetic influences on ethnic differences in health and disease. Several recent examples illustrate the pertinent evidence being presented.

Differential responsiveness to therapy has been identified by Exner et al. (2001), who compared black and white patients with left ventricular dysfunction on response to Enalapril therapy. Briefly, black patients displayed no significant reduction of risk for hospitalization for heart failure within 36 months of therapy, while white patients experienced a 44 percent reduction in risk. Other such racial differences have been identified, such as greater resistance among black kidney and liver transplant patients to immunosuppression (Nagashima et al., 2001). The mechanisms underlying these physiological differences may, of course, be environmental, genetic, or some combination of these effects.

Research that more explicitly implicates genetic factors can be illustrated by that of Splawski et al. (2002), who found a particular allele (described as Y1102) of a gene (SCN5A) to be associated with arrhythmia in blacks. This allele is found in about 13 percent of blacks and in about 19 percent of West Africans and Caribbeans, but it has not yet been identified in whites or Asians. Y1102 increases the risk of arrhythmia but is not a sufficient cause—most carriers of the allele never display arrhythmia. The conjecture is that the gene operates in the context of other risk factors, including possibly the use of certain medications.

A further example is provided by Karter et al. (2002), who examined the complications experienced by diabetics who participated in the same medical care program. The results differed by type of complication: blacks, Asians, and Hispanics had a substantially lower incidence of myocardial infarction than did whites; blacks experienced more strokes than whites; Asians had many fewer lower extremity amputations; and whites had lower incidence of end-stage renal disease. Complication susceptibility is clearly not the exclusive attribute of any one group.

Contextual scrutiny of genetic factors thus can lead to discoveries of important gene-gene and gene-environment interactions. For instance, there is evidence that the apoE genotype is less of a risk factor in Hispanic or black populations than in Caucasian populations (Sahota et al., 1997). Whether these differences are due to gene-gene interactions, gene-environment interactions, or both remains to be demonstrated.

Increasingly, reference is made to the prospects of individualized medicine, in which the preventive or intervention strategies will be tailored to the relevant genotype of the individuals. The foregoing discussion about interactions has emphasized that it will be necessary for the genotypic information to be joined with environmental information. The identification of all of the relevant input variables for all health and disease issues will be a formidable undertaking. Fortunately, benefits will be derived from the partial knowledge that will be generated on the way. Increasingly, it will be possible for medical decisions to be informed by information about DNA as well as about environmental circumstances of the individual. Racial identity, at best a weak surrogate for these genetic and environmental factors, will become increasingly irrelevant in treatment of disease.


In broad terms, both genetic and environmental factors, particularly in interaction, play a role in racial and ethnic differences in health and disease. These factors are active subjects of investigation by the genetic and social sciences separately.

Research Need 5: Assess genetic and environmental factors in racial and ethnic differences in health simultaneously, in designs that permit identification of both main effects and interactions.

This research focus necessarily would emphasize the identification and assessment of environmental features, both physical and sociocultural, that are pertinent to ethnic differences in health outcomes. The range of environmental variables should be considered in a life-course perspective, with attention to the possibility of the existence of critical periods for environmental impact. Attractive research opportunities are offered by populations of particular ethnic groups in similar settings and life circumstances.

The potential value of concentration of research efforts on interactions is self-evident. If the incidence or severity of a disease related to a particular genotype is dependent on the environment, detailed examination of the mechanism through which environmental influence is mediated may suggest preventive as well as ameliorative measures. The advantages would accrue to individuals of all racial and ethnic groups.

Genetic comparisons among populations of course need to continue. Characterization of gene frequency differences among populations should be strongly oriented toward the detection of epistatic (interactive) gene networks, and the newly available molecular procedures for identifying and characterizing genes and evaluating the level of gene expression should be melded with quantitative methods for dealing with complex phenotypes.

As in all domains of biomedical research, although human beings are our targeted concern, some aspects of complex systems are most efficiently approached through animal models.

Copyright © 2004, National Academies.
Bookshelf ID: NBK24694


  • PubReader
  • Print View
  • Cite this Page
  • PDF version of this title (790K)

Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...