8
Application of Biologic Markers Of Immunotoxicity in Epidemiology

This chapter addresses the use of biologic markers of immunotoxicity in epidemiologic research on environmental health problems and how this use can contribute to traditional environmental epidemiology. The characteristics of biologic markers used in population research are distinguished from the characteristics that affect their use in laboratory studies. Appropriate epidemiologic study designs are reviewed, and four case studies are presented to illustrate strengths and weaknesses in methodology.

EPIDEMIOLOGY

Epidemiology is the study of disease patterns in human populations. One use of epidemiology is to determine whether environmental factors, such as toxicants, drugs, infectious microorganisms, parasites, physical stress, or psychologic stress, can cause or be associated with disease or dysfunction. This objective can be met by comparing groups of people to determine differences in the frequency of a particular disease or risk factor. What can biologic markers of immunotoxicity contribute to this study of disease patterns?

The most common problem in conducting epidemiologic studies is the problem of identifying and quantifying exposure to the putative cause(s) or suspect agent(s). Epidemiologic studies have been classified into three groups, or types:

  • Cross-sectional studies or snapshots in time.

  • Retrospective studies, usually involving attempts to establish the fact or amount of exposure so as to relate it to current disease or disability.

  • Prospective or cohort studies, in which the fact or amount of exposure is established at some specific time and cohorts of exposed and unexposed persons are followed over time to determine whether any disease appears more often among exposed than among unexposed or less exposed persons.

Each of the types of epidemiologic studies has some virtues and some drawbacks. A cross-sectional study can be done at an instant in time—comparing how frequently the suspected cause appears among the diseased persons with the relative frequency of the presence of the "cause" among the nondiseased. The major problem with cross-sectional studies is that it may be difficult to establish the temporal sequence necessary to



The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 113
Biologic Markers in Immunotoxicology 8 Application of Biologic Markers Of Immunotoxicity in Epidemiology This chapter addresses the use of biologic markers of immunotoxicity in epidemiologic research on environmental health problems and how this use can contribute to traditional environmental epidemiology. The characteristics of biologic markers used in population research are distinguished from the characteristics that affect their use in laboratory studies. Appropriate epidemiologic study designs are reviewed, and four case studies are presented to illustrate strengths and weaknesses in methodology. EPIDEMIOLOGY Epidemiology is the study of disease patterns in human populations. One use of epidemiology is to determine whether environmental factors, such as toxicants, drugs, infectious microorganisms, parasites, physical stress, or psychologic stress, can cause or be associated with disease or dysfunction. This objective can be met by comparing groups of people to determine differences in the frequency of a particular disease or risk factor. What can biologic markers of immunotoxicity contribute to this study of disease patterns? The most common problem in conducting epidemiologic studies is the problem of identifying and quantifying exposure to the putative cause(s) or suspect agent(s). Epidemiologic studies have been classified into three groups, or types: Cross-sectional studies or snapshots in time. Retrospective studies, usually involving attempts to establish the fact or amount of exposure so as to relate it to current disease or disability. Prospective or cohort studies, in which the fact or amount of exposure is established at some specific time and cohorts of exposed and unexposed persons are followed over time to determine whether any disease appears more often among exposed than among unexposed or less exposed persons. Each of the types of epidemiologic studies has some virtues and some drawbacks. A cross-sectional study can be done at an instant in time—comparing how frequently the suspected cause appears among the diseased persons with the relative frequency of the presence of the "cause" among the nondiseased. The major problem with cross-sectional studies is that it may be difficult to establish the temporal sequence necessary to

OCR for page 113
Biologic Markers in Immunotoxicology a causal relationship. Did the cause actually precede the effect, or did the effect really appear earlier and thus possibly give rise to the presumed cause? Retrospective studies, or case-control studies, deal better with the temporality of a presumed cause-effect relationship. Persons are identified today as having or not having developed the disease of interest, and then the attempt is made to discover whether the presumed cause occurred earlier (or more often) among the diseased than among the nondiseased. Some of the early work relating cigarette smoking to lung cancer proceeded along these lines. Persons with lung cancer were identified "cases." Persons similar in what were considered characteristics possibly associated with the disease—but currently not known to have the disease—were identified "controls." Smoking histories in cases were compared with smoking histories in controls—and cigarette smoking was found to be much more common among the cases. What was sought was the development of disease among smokers in contrast to development of disease among nonsmokers. What was found was sort of an inverse: smoking among diseased in contrast to smoking among nondiseased. For rare events, some simple and easily examined assumptions permitted the inversion of the data into the form necessary to make a cause-effect inference. This demonstration prompted a great growth in use of case-control studies for rare diseases, in which logistical problems were easily and inexpensively overcome. What the inversion demonstration did not overcome was the potential for recall bias—unequal remembering of exposure to putative excesses by persons with a disease (looking for a reason for the disease), compared with the recollections of persons without the disease, who had no special reasons to try to remember something. A variety of techniques and devices have been developed to reduce the recall bias (Ozonoff, 1987), but none has been completely successful. Cohort studies have been proposed as an answer to the recall-bias problem. Following persons known to have been exposed and persons known (or believed to have been) unexposed over time should reduce or eliminate the recall bias. However, even these studies have their problems. They may take a very long time, particularly for studies of diseases with a long interval between exposure and disease development. Rare diseases require that very large cohorts be followed. For causes that have consequences other than the disease of interest (deaths from other diseases), exposed persons will be differentially removed from the populations followed, making the estimation of a cause-effect relationship more difficult to demonstrate or quantify. Each of these types of studies can have difficulties in establishing or estimating exposure. In studying industrial populations in particular, establishing exposure or quantifying peripheral exposure may be particularly difficult. It is here that biologic markers may become useful to the study of exposure response in human populations. CONTRIBUTION OF BIOLOGIC MARKERS TO EPIDEMIOLOGY What biologic markers in general and immune markers in particular can contribute to epidemiology is increased accuracy (e.g., no recall bias) and perhaps some better quantification. With appropriate biologic markers, analysts may be able to establish not only the fact of exposure but also the quantity of a foreign material that reaches a target tissue. Instead of studying cases of overt disease only, analysts could study the early manifestations of exposure. This would allow research to proceed more quickly, thus making possible early intervention or prevention. Biologic markers also are useful in identifying whether there are host or susceptibility factors that predispose some groups of people to disease or confound assessments of exposure-disease relationships

OCR for page 113
Biologic Markers in Immunotoxicology (Hulka et al., 1990). The accurate classification by exposure or disease is of primary importance in epidemiology. Misclassification usually leads to underestimating a cause-effect relationship and leads to a false-negative result, although sometimes it can lead to a false-positive result (i.e., that there is a relationship when none really exists). As an example, instead of assuming persons were exposed to a carcinogen because of their proximity to a landfill, an analyst can determine whether a person was exposed to that carcinogen as shown by DNA adducts in peripheral lymphocytes. Many immune-system biologic markers can be used in epidemiologic studies as indicators of exposure, early effect, or susceptibility. If immune-system markers are to be used this way, they need to be shown to be valid indicators or surrogates. Some characteristics of validity in the broadest terms have been described in other chapters. For epidemiologic studies, immunotoxic markers must have laboratory validity and population validity. A marker may be accurate in the laboratory in representing a biologic event, but be of uncertain meaning because of wide population variability or low predictive value (Schulte, 1987, 1989). VARIABILITY IN REFERENCE POPULATIONS Variability in immune-system markers within a population can be attributable to genetic, environmental, or biorhythmic influences rather than to the exposure that is the object of a study. The analytic techniques used to measure immune-system markers are rarely rigidly reproducible and also contribute to variability. Therefore, evaluation of the differences observed in epidemiologic studies will need to take variability into account before ascribing observed health effects solely to exposure to toxicants. Current data suggest that most markers of immunotoxicity will show considerable overlap between exposed and unexposed populations. Valid data on the sources of variability are obtainable only from carefully designed and executed studies (Edwards et al., 1989; Shopp et al., 1989). Determining the sources of variability and improving the precision of the marker measurements are among the highest priorities of researchers who use biologic markers in public-health studies. SENSITIVITY, SPECIFICITY, AND PREDICTIVE VALUE The use of the terms sensitivity and specificity as they relate to populations under epidemiologic study must be distinguished from their use with respect to laboratory methods (Griffith et al., 1989). Laboratory sensitivity indicates the lowest level of an analyte that can be measured reliably by the analytic technique. Laboratory specificity indicates the ability of the analytic technique to exclude identification of substances other than the desired analyte. Sensitivity and specificity as used in population studies are measures of the accuracy of a test. Sensitivity is the ability to identify a true positive correctly. If the event (exposure) did occur, sensitivity is the measure of the proportion of the cases that the test (or marker) indicates did occur? Specificity is the ability of a test or marker to identify a true negative correctly. If the event did not occur, in what proportion of the cases does the test say that the event did not occur? Predictive value is a measure of the potential usefulness of the test in identifying an exposed individual in the population (positive predictive value) or identifying an unexposed individual in the population (negative predictive value). Several examples (derived from Ozonoff, 1987) are given in Table 8-1. For rare conditions (low prevalence in the population), even very good measures (markers) give poor positive prediction, as shown in Example 1. When there is high prevalence in a population, even substantially

OCR for page 113
Biologic Markers in Immunotoxicology TABLE 8-1 Three Examples of the Relationship Between Exposed Subjects and the Presence (+) or Absence (-) of Markers Illustrating the Interaction of Prevalance, Sensitivity, Specificity, and Predictive Value EXAMPLE 1: Highly accurate test, used when there is low population prevalence (10 in 1,000 = 1%): Sensitivity = 100% = 10/10 Specificity = 95% = 940/990   Exposed   Yes No   + 10 50 60 - 0 940 940   10 990 1000 Positive predictive value (% of persons with markers present who were actually exposed): 10/60 = 16.7% EXAMPLE 2: Highly accurate test, used when there is high population prevalence (100 in 1,000 = 10%): Sensitivity = 100% = 10/10 Specificity = 95% = 940/990   Exposed   Yes No   + 100 45 145 - 0 855 855   100 900 1000 Positive predictive value 100/145 = 69% EXAMPLE 3: Less accurate test, used when there is a higher prevalence (100 in 1,000 = 10%): Sensitivity = 100% = 10/10 Specificity = 95% = 940/990   Exposed   Yes No   + 100 180 280 - 0 720 720   100 900 1000 Positive predictive value 100/280 = 35.7% Source: Adapted from Ozonoff (1987).

OCR for page 113
Biologic Markers in Immunotoxicology poorer tests (markers) give higher positive predictive values, as shown in Example 2. However, in Example 3, the predictive value is less, despite high prevalence, because of the less accurate test applied. In all three examples, the negative predictive value is 100%, because the sensitivity is 100%, meaning that ''all negatives" are true negatives. Population sensitivity and specificity are determined by both intrinsic and extrinsic factors. Intrinsic factors include the differences in the distribution of the marker in the unexposed population and in the exposed population. An ideal biologic marker is absent in all unexposed persons and easily detectable in all those who have been exposed to the toxicant. Unfortunately, no such marker is known to exist for any organ system, and, because of the variability of immune-system responses within and between individuals, the distributions of most markers are likely to show considerable overlap in unexposed and exposed individuals. Extrinsic factors also influence population specificity and sensitivity. People who have not been exposed to a toxicant could have other conditions that cause the appearance of the same biologic marker seen in people who have been exposed. Errors in measurement also influence population specificity and sensitivity. Analytic imprecision blurs the intrinsic distinction in distribution of a marker between exposed and unexposed populations, and analytic inaccuracy can lead to misclassification of individuals as positive or negative for the marker. The particular decision rules or criteria used to determine the event or condition status (especially for health effects) can also have a profound influence on determining population sensitivity and specificity. If, for example, the event or condition is a disease, the "case definition" is often a subjectively determined collection of signs and symptoms. Adding or excluding a particular symptom can result in the inclusion in the group under study of a larger or smaller number of persons with the disease. This will in turn have an influence on estimates of sensitivity and specificity. For a given test, where the definition of a positive on the test is a value above (or below) some index value, or threshold, the sensitivity of the test can sometimes be increased by adjusting the cutoff or threshold value. However, because sensitivity and specificity are linked, for a given test increasing one will inevitably result in decreasing the other. There are circumstances, however, in which adding another test (or tests) can lead to increasing both the sensitivity and specificity. There the cost is not decreased sensitivity (overspecificity), but rather the actual dollar cost in conducting the additional test(s). A simple inexpensive test with poor specificity can sometimes be followed by a more complex and more expensive test with greater specificity that will weed out many (if not all) false positives. In developing a test, an estimate is usually made of sensitivity and specificity as a result of a trial on a known or (often) an easily acquired population, such as medical students or nurses. The measures of the test (i.e., sensitivity and specificity) may often shift with the population tested, so that persons using the test (or marker or system of markers) need to be aware that the first published sensitivities and specificities are almost certain to be different from the same measures computed from a different population. The interaction of prevalence, specificity, and sensitivity are illustrated in Table 8-2. AUTHENTICATION OF THE EVENT STATUS Accurately assigning the presence or absence of a specified event in reference populations used to estimate the sensitivity and specificity of a biologic marker is critically important to the overall validation process. Unless the status of the event itself can be accurately determined in each member

OCR for page 113
Biologic Markers in Immunotoxicology TABLE 8-2 The Interrelationship Among Prevalence, Sensitivity, and Specificity Positive predictive values in a test with 95% specificity and varying sensitivity for four possible exposure prevalences Specificity = 95%     Prevalence     0.0001% 0.001% 0.01% 0.1% Sensitivity 95% 0.19 1.9 16.1 68   90% 0.18 1.8 15.4 67   85% 0.17 1.7 14.7 65   80% 0.16 1.6 13.9 64   75% 0.15 1.5 13.2 63   70% 0.14 1.4 12.4 61   65% 0.13 1.3 11.6 59   60% 0.12 1.2 10.8 57 Positive predictive values in a test with 95% sensitivity and varying specificities for four possible exposure prevalences Sensitivity = 95%     Prevalence     0.0001% 0.001% 0.01% 0.1% Specificity 95% 0.19 1.9 16.1 68   90% 0.09 0.9 8.8 51   85% 0.06 0.6 6.0 41   80% 0.05 0.5 4.6 35   75% 0.04 0.4 3.7 30   70% 0.03 0.3 3.1 26   65% 0.03 0.3 2.7 23   60% 0.02 0.2 0.2 21   Source: Ozonoff (1987). Reprinted with permission; copyright 1987, Gordon and Breach Science Publishers.

OCR for page 113
Biologic Markers in Immunotoxicology of the population, the predictive value of any biologic marker is likely to be less than estimated. Before an event or condition can be confirmed, however, it must be clearly defined. This is often the most crucial factor in estimating the predictive value of any biologic-marker result (Radford, 1981). Event definitions should be considered separately for biologic markers of exposure, effect, and susceptibility. For markers of exposure, the definition of an event may depend on the detection limits of analytic techniques. As these limits become lower and lower, the possibility exists that nearly all persons in industrialized societies will be shown to have some exposure to many, if not most, environmental pollutants. Thus, an arbitrary level of "excessive" exposure might have to be used to define "exposed" and "unexposed" populations. For biologic markers of susceptibility, the condition must be defined in terms of the increased risk of a specified health effect (or of any step in the continuum between exposure and disease) for some defined exposure. For biologic markers of effect, the effect must be clearly defined within the wide spectrum of potential immune-related effects, ranging from those associated with overt disease to those which have no known or apparent health consequences. Once specific events or conditions have been defined, the definitions must be tested and confirmed. Instances of exposure as defined by biologic markers should be confirmed by physical or chemical measurements of toxicant levels. In some workplaces, epidemiologic exposure indexes could provide legitimate surrogate measures of exposure (Fingerhut et al., 1989). However, verification of exposure status, especially when the researcher is seeking dose-response relationships, usually demands actual measurement of toxicant levels. Where intensity of biologic markers can be measured, these should correlate positively with other accepted measures of exposure. Susceptibility can be validated by the evidence (usually epidemiologic) of an increased relative risk for a specified health event. Prospective studies are often informative for separating the contributions of normal biologic variation, low-level toxicant effects, and truly increased susceptibility attributable to exposure at a specific time to a given toxicant, but other epidemiologic techniques have a useful place. Effects or disease status should be confirmed similarly to susceptibility. For well-defined effects, including death and diagnosable diseases, confirmation through medical records and epidemiologic studies should be possible. For subclinical, immune-related effects, confirmation is more difficult and could depend solely on supporting evidence from other biologic markers. For vague, nonspecific self-reported complaints, confirmation is usually not possible; these present a very difficult problem for epidemiologic studies. STUDY DESIGN Immunologic markers can be used as dependent (outcome) or independent (risk-factor) variables in exploratory epidemiologic studies of the relationship between exposure and effect. Much of the literature on the immunotoxic effects of occupational and environmental exposure to toxic chemicals consists of use of cross-sectional studies, in which the outcome variables are markers of effect. The clinical significance of these markers is often unclear, particularly for markers that indicate immunosuppression (Trizio et al., 1988). This unknown clinical significance comes about as a product of the one-time assessment of effect in a cross-sectional study. A single assessment of a marker of immunotoxicity will usually not provide definitive information as a true indication of exposure. Multiple measurements of a marker over time should provide a more accurate representation. Investigators have used immunologic

OCR for page 113
Biologic Markers in Immunotoxicology markers to define disease in only a few studies. The usefulness of markers should extend to studies of preclinical disease. Immunologic markers, such as human leukocyte antigen (HLA), have been used as the risk factors or independent variables, as for example in case-control designs (Masi, 1979). Because such immunologic markers as the HLA system have been linked to numerous diseases, they could have potential for use in field studies as markers of susceptibility. There is strong support for the use of immunologic and immunotoxicity markers in prospective studies as early indicators of effect of xenobiotic exposure. In the classical prospective design as indicated earlier, two disease-free groups, distinguished by exposure, are followed for a given period, the incidence of disease is assessed, and the disease rates of the groups are compared. One or more (usually different) immunologic markers can be used to identify the resulting disease or dysfunction. In prospective studies, researchers must confirm that both groups are free of the disease(s) of interest at the inception of the study. If disease is defined as deviation from the norm of some condition identified by or correlated with an immunologic marker, the level of the marker at the study's beginning must be measured (Schulte, 1987). In prospective studies, the research is more likely to be productive if the researchers know the natural history of the disease and the marker, because "the effects of a chemical may be directly related to the temporal relationship between the chemical exposure and antigenic challenge" (Bick, 1985). This knowledge of disease progression and immune-marker expression will aid researchers in determining when to evaluate the groups in a study. Evaluation too soon after exposure could lead to false-negative results. Periodic assessments that yield time-series data are preferable to a single evaluation. These data often need to be analyzed by procedures not well known to epidemiologists and immunologists. It is not sufficient to analyze such data by using repeated cross-sectional analyses at each observation point. A more suitable approach is to use a model in which the dependent variable in a linear regression is compared with earlier dependent-variable values and an attempt is made to see whether the relationship changes with time. Generally, however, in prospective studies the appearance of an (immunologic marker of) effect can be estimated by the standard hazard rate function and analyzed by the relative-risk regressions method. REFERENCE POPULATIONS Occupational exposures often present the best instances in which to test markers of immunotoxicity in chronically exposed individuals, because exposure and health status are often documented. Such populations are ideal for coordinated, cost-effective prospective studies. There are several sources of groups who have been exposed in the workplace. Some companies have established inhouse medical monitoring or worker registries (Tamburro and Liss, 1986). The National Institute for Occupational Safety and Health periodically conducts health-assessment studies that involve immune-system markers (Shopp et al., 1989). Occupational-health and environmental-health clinics can at times be a good source of groups of exposed persons. Many of the clinics are members of the Association of Occupational and Environmental Clinics (Vogt et al., 1990). Superfund sites offer researchers both an opportunity and a mandate for using immune-system markers in health-effects and exposure studies. The 1987 Superfund Amendment and Reauthorization Act requires the Agency for Toxic and Substances and Disease Registry (ATSDR) to conduct health-assessment studies at the toxic-waste sites on the National Priority List. Studies are to include the use of immune-system

OCR for page 113
Biologic Markers in Immunotoxicology markers to assess baseline immune function, toxicant exposure, and possible susceptibility. The toxicants of greatest concern—which include heavy metals, volatile organic compounds, pesticides, and polyaromatic hydrocarbons—have been listed by ATSDR and the Environmental Protection Agency (EPA, 1987). Registries of persons exposed to certain toxicants are being established through Superfund by ATSDR. The toxicants studied are chosen on the basis of their anticipated public-health importance. The first such subregistry is for trichloroethylene (Burg, 1990), which has been implicated as a cause of immune dysregulation (Byers et al., 1988) and leukemogenesis (Lagakos et al., 1986). Registries could provide researchers an opportunity to conduct prospective studies in which they could test the predictive value of some immune-system markers. Studies conducted by state public-health agencies of nonoccupational toxicant exposure have included tests for immune-system markers (Bekesi et al., 1987; Fiore et al., 1986; Stehr-Green et al., 1988). Exposure registries have been established to permit longitudinal studies. The third National Health and Nutritional Evaluation Survey began in 1989; the first phase will run through 1992; the second phase will go through 1995. About 40,000 persons constituting a representative sample of the U.S. population will receive medical examinations and a battery of laboratory tests. This sampling should provide normative reference data on immune-system markers and perhaps allow researchers to relate the data to toxicant exposures. A subset (perhaps 4,000) will be tested for evidence of exposure to volatile organic compounds, phenolic compounds, and pesticides. CASE STUDIES Four field studies have been selected to illustrate the use of biologic markers, especially as related to immunotoxicology. The summaries presented are not meant to be comprehensive; they are intended to provide information on how a study was or could be designed to use markers of immunotoxicity. The fact, however, is that epidemiologic studies that have made full use of the marker concepts have not been completed. Case 1: Incidence of immune hyper-sensitization in workers using hexamethylene diisocyanate (HDI) and its trimer (THDI) (Grammer et al., 1988). Background Isocyanates can induce symptoms of respiratory disease through immune mechanisms (inducing antibody products) or as nonimmune irritants. A group of 150 workers who used HDI and THDI were evaluated to determine the prevalence of immune sensitization and its relationship to work-related respiratory disease. Study Design An 18-month prospective study was conducted on 150 current workers in a factory where truck cabs had been sprayed with paint that contained HDI and THDI. The researchers administered a medical history and symptom questionnaire to workers with a minimum of 2 months of employment. On the basis of the questionnaire responses, two physicians (who did not know any subject's work history) identified workers with asthma, allergic rhinitis, allergic conjunctivitis, and hypersensitivity pneumonitis. They then distinguished symptoms of these conditions from those attributable to HDI and THDI. Exposures to HDI and THDI were measured several times in each work area and were expressed as time-weighted averages.

OCR for page 113
Biologic Markers in Immunotoxicology Workers were classified according to job class, tobacco use, and other behavioral and demographic characteristics. The immunologic markers used included IgE and IgG antibodies to HDI-HSA (human serum albumin) and THDI-HSA determined by enzyme-linked immunosorbent assay. Results Approximately 21% of the workers had a positive antibody result (generally low-level IgG). One worker had symptoms that appeared to be caused by exposure to HDI. Generally, there was no difference among job classifications or between smokers and nonsmokers for antibody levels. The number of workers whose antibody levels increased roughly equaled the number whose antibody levels decreased. The investigators concluded that the low-level presence of antibodies in the absence of clinically observed disease could indicate exposure but not current clinical disease. Strengths A prospective design was used to determine whether exposure to HDI or THDI resulted in the formation of a biologic marker of exposure and whether the marker was correlated with clinical disease. Limitations The lack of measurements of the toxicants in the breathing zone (i.e., personal sampling) limited the researcher's ability to make individual exposure-response determinations. Researchers were unable to distinguish job categories on the basis of immunologic markers. Case 2: Workers exposed to 2,3,7,8- tetrachlorodibenzo-p-dioxin (TCDD) (Sweeney et al., 1989). Background TCDD is a potent toxicant in some small mammals and produces selective immunotoxicity in mice (Luster et al., 1987). Human exposure to TCDD can cause chloracne. There are no reported cases of death or serious illness (Filippini et al., 1981; Suskind and Hertzberg, 1984). Some reports suggest human immunotoxic effects (Hoffman et al., 1986; Knutsen et al., 1987; Jennings et al., 1988), but no convincing evidence has emerged from these studies. Some of the data have been challenged (Evans et al., 1988). Study Design A cross-sectional study was conducted on production workers previously exposed to TCDD-contaminated chemicals. Some workers were exposed to high doses over several years (Sweeney et al., 1990). A total of 541 persons (281 exposed and 260 control subjects) were brought to a single, central clinical research facility for evaluation. Control subjects were matched to study subjects by age, gender, and community of residence. The study included an extensive history and medical examination, a large array of immune tests (Shopp et al., 1989), and measurement of serum TCDD levels (D.G. Patterson, Jr., et al., 1989). Results Results for immune-system markers from the study are now under analysis (Sweeney et al., 1989). Measurements of serum TCDD levels, a marker of exposure, and a

OCR for page 113
Biologic Markers in Immunotoxicology good estimate of the dose in terms of body burden could be used to determine whether there is a dose-response relationship for immune-system markers of effect (Fingerhut et al., 1989). Strengths The size and chronic high-level exposure of the study population will allow researchers to make valid conclusions about whether TCDD affects the immune system. Serum TCDD has a half-life of several years and can be detected long after exposure has ceased, allowing researchers to determine long-past exposure by measuring TCDD levels (Pirkle et al., 1989). The duration of employment in TCDD-contaminated areas has been found to be highly correlated with serum TCDD levels (Sweeney et al., 1990). The well-matched control population will allow the researchers to determine valid reference ranges for this population of older, mostly white males. The array of immune assays was comprehensive and well controlled; it was tested repeatedly on a small group of laboratory volunteers (Shopp et al., 1989). The full extent of variability throughout the immunologic assessments can be determined from this control population. Detailed medical histories and medical examinations will allow researchers to explore correlates of exposure, markers of immunotoxicity, and health status. Because members of this study population are approaching an age at which immune-system function declines, a fairly susceptible group is available for study. Participation rates were high in both the exposed and the control populations (Sweeney et al., 1989). Limitations The length of the study might permit seasonal variations to affect the distribution of results. The required long-distance travel for some participants could have affected immune response because of stress or changes in circadian rhythm. A study of this kind is very costly. Unfortunately, this well-designed study is not yet completed. Case 3: Effect of solvent-contaminated well water on the immune system (Byers et al., 1988). Background In this study, four kinds of markers were used to assess the immunologic effect of drinking solvent-contaminated water. The study took place because of a statistically significant excess of leukemia in a census tract that received some of its potable water from two contaminated wells. Study Design A modified case-control design with repeated evaluations was used to determine whether persons exposed to contaminated water exhibited immunologic effects different from effects found in those who were not exposed. The case subjects were family members (those with leukemia probands) and were residents from the census tracts receiving water from the contaminated wells. Control subjects were selected by probability sampling techniques and were matched to case subjects on age, sex, and social habits. The groups were compared for absolute numbers of T cells (CD3), CD4 cells, and CD8-positive cells; the ratio of CD4 to CD8; the presence of autoantibodies; and the incidence of infections and rashes. Results Changes in the immunologic system were

OCR for page 113
Biologic Markers in Immunotoxicology manifested by altered ratios of T-lymphocyte subpopulations and by an increased incidence of autoantibodies, infections, and rashes. Strengths The findings were consistent with earlier reports in the scientific literature. They occurred about 5 years after exposure, possibly demonstrating a persistent effect. The findings were consistent with the use of multiple markers over multiple periods. The analysis used appropriate sophisticated statistical techniques. Limitations The study design did not allow researchers to discern whether the illnesses found resulted from genetic or environmental factors. The exposure information was on an area basis, rather than an individual basis, and required that the researchers assume that all case subjects were exposed and that all control subjects were not. Because the study was done several years after the exposure occurred, intervening factors, such as unrelated exposures, could have affected the results. Case 4: Tight-building syndrome (Levin and Byers, 1987), study of environmental illness. Background Immunologic effects have not been the primary focus of investigations of the "sick-building" syndrome. Immune-system responses have been determined to be consistent with the range of symptoms reported, and they are believed to be confined to immunologically mediated respiratory diseases, occasionally involving the skin or mucous membranes (Samet et al., 1988). Immune-system markers have rarely been used, except to assess antibody responses to infectious organisms, for example, in Legionnaires' disease, or to assess specific allergies. The work of Levin and Byers (1987), however, describes the use of immunologic markers in studying several tight-building situations in which people were exposed to airborne pollutants. Levin and Byers suggest that immunologic responses resulting from chronic low levels of exposure to toxic chemicals can be assessed by evaluating a constellation of signs, symptoms, and laboratory findings. In this case study, they reviewed the use of T-helper/suppressor-cell (CD4:CD8) ratios to evaluate exposed and control populations. Study Design Levin and Byers (1987) cite data from four studies with different exposure scenarios in which the helper/suppressor ratios were plotted according to the percentage of the population in each category. The comparison was between exposed patients and asymptomatic, apparently unexposed control subjects. Results The ratios of T-helper to suppressor cells were found to be statistically lower in exposed than in control subjects in the four studies cited. These changes were supported by a similarity in symptomatology of the patients evaluated. Strengths This approach allows researchers to quantitate the immunologic markers of a

OCR for page 113
Biologic Markers in Immunotoxicology series of diverse conditions to characterize better the potential health effects of unknown origin. It also allows researchers to quantitate immunologic changes and validate the markers by review of reported symptoms. Limitations In each of the reports of tight-building syndrome reviewed by Levin and Byers (1987), each study included only a small number of exposed subjects, and the analytic methods were not published for detailed review. With small numbers of subjects, differences in the way cases and controls are assessed can contribute to an apparent effect in the small exposed group. The possibility of bias in the selection of the exposed cases cannot be ruled out. The findings have not been replicated and published in the scientific literature. RECOMMENDATIONS Immunotoxic markers can be used to identify exposure, effect, or susceptibility. For epidemiologic studies, markers should be selected for use on the basis of their predictive value in human populations, validated animal models, and the underlying biology of the markers. Biologic markers of immunotoxicity, to be used in epidemiologic studies, are subject to several limitations: The predictive value of the markers should be known, and it should be high. Study populations must be large enough to yield firm estimates. The comparison of results from exposed populations with those from unexposed populations should be controlled for known confounders. Standardized assays should be used to allow for the confirmation and interpretation of markers. Multiple immune markers should be assessed, and multiple periods should be considered for specimen collection. Biologic markers of immunotoxicity can best be tested in occupationally exposed populations. The use of biologic markers of immunotoxicity in environmental epidemiologic research requires interdisciplinary collaboration among laboratory scientists, field scientists, and clinicians in planning, implementing, and interpreting studies. Cooperative interchange among laboratories that use markers of immunotoxicity to monitor other kinds of exposure to immune-related disease (such as HIV infection) should be encouraged, so as to standardize some of the more promising new assays. The possibility of banking specimens from persons with documented exposures to chemicals should be explored, so that new assays can be investigated quickly. Attention should be given to how the results of uncertain assays of immunotoxicity markers will be interpreted and communicated to study subjects. The legal and ethical implications of labeling groups or individuals on the basis of altered marker frequencies should be considered. Surveys of the prevalence of immunotoxicity and immunologic markers should be conducted to provide baseline and reference values. Pilot studies should be performed before large-scale epidemiologic studies are begun.

OCR for page 113
Biologic Markers in Immunotoxicology This page in the original is blank.