The potential of exposure biomarkers in epidemiologic studies of reproductive health.

To further the development and application of exposure markers in field investigations in reproductive epidemiology, we have synthesized recent examinations of the issues surrounding exposure measurements in reproductive epidemiology. The specific goals of this paper are to define exposure biomarkers and explore their potential uses, particularly as screening tools. The tests for glucaric acid, thioethers, mutagenicity, and porphyrin patterns meet the general criteria for useful exposure screens. For certain xenobiotic agents, these tests accurately differentiate exposure levels, as demonstrated in occupational and environmental epidemiologic studies. As urinary screens, they are noninvasive and applicable on a large scale with current laboratory techniques. For short-term exposure, glucaric acid, thioethers, and mutagenicity tests are useful. Porphyrin patterns may measure cumulative effects as well as current exposure levels. The usefulness of these tests in epidemiologic studies of environmental effects on reproductive health has yet to be studied. To do so, the battery must be standardized for pregnant women, and test results must be correlated with measured adverse reproductive outcomes, such as gestational length and birthweight. This correlation is particularly important because maternal exposure rather than fetal exposure is being measured. The extent to which xenobiotic chemicals cross the placental barrier may vary greatly depending on the type of exposures, timing in pregnancy, and maternal detoxification capability. Without better exposure measures, epidemiologic studies of reproductive health probably will not successfully identify xenobiotic fetotoxic agents in the environment. However, with an adequate battery of nonspecific exposure biomarkers, prospective studies of environmental effects on pregnancy outcomes might be possible.(ABSTRACT TRUNCATED AT 250 WORDS)

halt probably Wm not ci ienobiotic fegotoc* agents in the environment. However, with an adeq battery ofnecf expsre blmarkers, prospective studies of environmental effects on pregnancy outcomes might be possible. lb narrow the lst of potentil exposures, these prospective studies could be foLowed by case-cntrol studies of more specific blomarks dired at upect exposures.

Introduction
The researchofHatch (1) hasgreatly increasedour understanding ofthe use and potential limitations ofbiological markers for adverse reproductive effects. Likewise, there are great potentials and pitfalls in the emerging field of exposure biomarkers for reproductive health. This area of research has lagged considerably behind the field ofdeveloping new effect markers, such as semen analysis and early pregnancy loss. However, its importance in quantitative risk assessment cannot be overemphasized. As Hulka has so clearly stated, "the most important current limitation in many epidemiological studies is the relative inaccuracy of methods for measuring the exposure variable" (2).
Exposure to a potential fetal health hazard can be estimated through ecological assessment (e.g., testing the community water supply), questionnaires (e.g., classifying residents according to whether they drink bottled or tap water), or biological markers (e.g., testing for exposures to chemicals or solvents found in tap water). In a community with an environmental factr suspected of adversely affecting reproductive outcome, measuring environmental contaminants provides a gross, ecological esimate ofthe exposure incurred by pregnant women and their fetuses. However, ecological estimates can lead to significant misclassification of individual exposure (3). Such misclassification, ifnondifferential, will underestimate the true effect of the exposure. If misclassification is differential, misleading results in either direction can occur.
Using questionnaires to assess the extent of an individual's potential for exposure may help to reduce misclassification bias. Yet reports of individual exposures can be erroneous in either direction (4). In addition, people are often unaware of their potential for exposure, and researchers may not know or be able to account for all the pathways ofexposure. For example, certain mothers in a Yugoslav community with a lead smelter had elevated blood-lead levels. Questionnaire data detennined which ofthese women were wives ofmen employed in the lead industry. However, these data could not distinguish between women with low blood-lead levels and women with elevated levels (4). In this example, a biological exposure marker (blood lead-level) was available for classifying mothers according to their exposure to lead. To date, such biological markers have not been widely available nor have they been widely used when they are available.
Several years ago, the Environmental Protection Agency cosponsored a National Research Council study on The Role of Biomarkers in Reproductive and Developmental Toxicology (5).

2HOGUEAND BREWSYER
After reviewing the situation, Longo described a "paradox" (6). Although several techniques for identifying individual exposures have been developed and tested, and although more and more xenobiotics have been recognized to have teratogenic and mutagenic potential, "essentially no specific biomarkers are currently available to indicate that exposure to a given xenobiotic is directly associated with a cellular, subcellular, or pharmacodynamic event" (6). The paradox continues, despite continuing advancements in laboratory science and the growing recognition of the need for biological markers to improve exposure measurement in the field ofenvironmental epidemiology (3,4,(7)(8)(9). To further the development and application ofexposure markers in studying the environmental hazards to reproductive health, we have attempted to synthesize recent examinations of the issues surrounding exposure measurements in reproductive epidemiology. The specific goals ofthis paper are to explore the potential uses ofbiomarkers as measures ofexposure, particularly as they may be used in an environmental setting as screening tools.

Biomarkers As Measures of Exposure
Hulka defines biological markers in environmental epidemiology as "cellular, biochemical or molecular alterations which are measurable in biological media such as human tissues, cells or fluids and are indicative of exposure to environmental chemicals" (2). Biomarkers are not environmental measures in air, soil, water, or food; nor reports from research subjects; nor results of physical, anthropometric, or mental examinations. Rather, they are material measures obtained from physical samples (4).
Biomarkers used to estimate environmental exposues must be distinguished from those used to estimate the effects ofthose exposures. Biological markers ofeffects can be subclassified into biologically effective doses (such as DNA adducts) and biological responses (2). Examples ofthe biological response include sensitive tests of early pregnancy loss and serum alphafetoprotein to detect etal neurl tube defect. The National Research Council study prinarily dealt with effect markers (5). Although both exposure and outcome measures are necessary for an epidemiologic investigation, we will focus on measuring exposures. In Hulka's classification, these are "internal dose markers" (2).
Internal dose-exposure markers may be useful to improve the quality ofexposure measurement in an epidemiologic investigation ofa known environmental hazard; to serve as the "gold standard" for other infrmation sources; to provide a screening tool for environmental exposures to a target tissue (in this case, the fetus); and to provide quantification ofthe biological load from a known exposure (4). To be useful in epidemiologic investigations of reproductive health an exposure marker should be better than the woman's ability to recall an exposure; allow for differentiation between exposure levels, at least qualitatively; allow the use ofnoninvasive procedures that are applicable on a large scale; and provide interprable data for short-term or cumulative exposure regarding time, dose, and duration (4).
For the environmental Sherlock Holmes, internal dose markers offer strong circumstantial evidence that the perpetator xenobiotic has invaded the human victim. This evidence is very specific if the chemical is retrieved unaltered. However, sub-stantial circumstantial evidence can be gleaned from metabolically altered chemicals. The metabolic outcome can be very specific (e.g., urinary cotinine for nicotine in cigarette smoke) or nonspecific (e.g., thioethers for cigarette smoking).
Nonspecific markers measure a biochemical pathway affected by a variety of xenobiotic agents.

Exposure Markers As Screening Tools
In addition to possessing the characteristics of all useful exposure markers, biomarkers used as exposure screens should be able to detect subtoxic exposures and be nonspecific (8). Nonspecificity ofthe marker is important because the environment commonly includes complex and unknown chemical mixtures, such as those found in drinking water, that could be misclassified by selecting a few specific markers for a screening battery.
For epidemiologic research, nonspecific markers tend to be held in lower esteem than specific markers, since it is impossible, without further evidence, to identify which chemical has triggered the metabolic response being measured. However, as screening tools, nonspecific markers hold some promise. A biomarker that can be used to detect that one or more ofa class of xenobiotic agents to which the pregnant woman has been exposed and may have exposed her fetus could be useful for targeting a subset of women for further investigation and follow-up. First, however, the fiat that the metabolic pathway has been altered must be correlated with adverse human reproductive outcomes so that such alteration can be shown to reflect fetotoxicity.
We have previously proposed three nonspecific urinary biomarkers as potential screening tools for reproductive epidemiology (8): glucaric acid, thioethers, and porphyrin patterns. Vainio et al. (10) proposed mutagenic activity in bacteria as another nonspecific urinary screening tool, and Brewster (9) has added that biomarker to her proposed battery forpreantm. Let us briefly review the usefulness oftiese biomarkers.
Bacterial Urinary Assay for Mutagenic Activity Vainio et al. (10 critically reviewed the bacterial bioassay procedu as a test ofmutagenic activity in urine. Like the ohertests proposed, it has the advantage of demonstrating biological activity rather than the mere presence ofxenobiotic substances. However, this procedure poses problems that lead the reviewers t recommend that it be used in conjunction with other screening measures. We agree. For example, a bacterial bioassay cannot detect cumulative exposures and may react to substances normally present in urine (such as amino acids). Ikble 1 gives examples of the numerous studies of reported alterations in muagenicity by extrnal agents . Occupational exposures have been associated with alterations in mutagenicity; however, not all studies are positive. For example, oncology nurses handling cytotoxic drugs have been studied by several investigators. i00 studies were positive for altered mutagenicity (13,17), but two later reports were negative (25,42). This difference may reflect changes in routines for handling these drugs. Tire plant workers (possible smoking synergism); tetramethyl (thiuram disulfide, poly-p-dinitrosobenzene, diaryl-p-phenylendiamies) Oncology nurses handling cytotic drugs Anode plant workers exposed to coal tar and pitch Oncology nurses handling cytostatic drugs   (61). Stimulation of the glucuronidation pathway results in increased ecretionofglucaric acid, primarily thughreducedproduction ofglycogenandpossiblythroughdirectinductionofpathwayenzymes (9,62). Although pregnancy (63) and estrogen therapy (64) produce a nonsignificantincreaseinglucaricacideretion, anumberofexogenouschemicalshave produced significant changes in glucaric acid excretion (Table 2) (65-71). Intraindividual variation, which can be as high as 50%, may reflect daily variations in environmental exposures (72). Such variations would suggest that daily measures during vulnerable points in pregnancy would be required to assure accurate classification ofexposure through the pathways that increase glucaric acid excretion. Laboratory methods for measuring urinary glucaric acid have been developed and stadardized (9). However, standards for pregnant women do not yet exist.

Thioethers
Alkylating agents can be detoxified by reaction with glutathione or other sulfhyryl compounds. These conjugates frequently appear in urine as mercapturic acids or other thioether (R-S-R) products. Xenobiotics known to be detoxified through this sequence include aromatic hydrocarbons, arylamines, and many other chemical agents (9). Seutter-Berlage et al. (73) first posed using urine thioedhers as a screening tool for xenobiotic exposures. Sincethen, numeros studieshavedocumentedelevations in urinary excretions related to occupational, enironmental, and behavioral (i.e., smoking) exposures (  (105) and Van Doornet al. (76) have critically reviewed the literature on the urinary thioether assay as an exposure screening tool. They note that positive results reflect true exposures, but negative results may not reflect lackofexposure. This false negativity occurs because thioethers measure shortterm exposure and thus may miss past exposures that could have a future biological effect. This limitation, although prhaps not as great for exposures in pregnancy (because short-term expo-sures may be the most valid for measuring fetal exposure), suggests prospective urine collection at several points during pregnancy.
Presumably, all xenobiotics dtat produce chronic changes would also show the urinary patten at early stages prior to overt toxicity, but this assumption has not been tested in all circumstances.

Strengths and Limitations of Proposed Screening Battery
These four tests discussed previously generally meet the criteria for useful exposure screens. For certain xenobiotic agents, they accurately differentiate exposure levels, as demonstrated in occupational and environmental epidemiologic studies. As urinary screens, they are noninvasive and applicable on a large scale with current laboratory techniques. For shortterm exposure, glucaric acid, thioethers, and mutagenicity tests are useful. Pbrphyrin patterns may measure cumulative effects as well as current exposure levels.
The potential for this battery to identify groups of pregnant women at risk ofeavironenal insults to their ftuses can be illustraedby cite smoking. Glucaric acid may (23) or may not (125) be elevated by cigaette smoke. However, thioethers are elevated by cigarette smoke (73,80,90,91). Also, mutagenicity tests are very sensitive to cigarette smoking (10). Although cotinine is a specific marker for cigaette smoking (and thus would be the biomarker of choice if cigarette smoking is the specific exposure ofinterest), the factthatthis battery is responsive to a known feltotoxic agent lends credence to its potential value for detecting other, as yet unknown, fetotoxins. However, as we have previously coented (8), to be effective screening   (124) and embryos in ovo t Copro III Lead in rats Tohiba and Tomokumi (115) a(t) Increased concentration. Uro, uroporphyrin; hepta, heptacarboxygloporphyrin; copro, coproporphyrin; Copro III, coproporphyrin m. bCEL, chick embryo liver cells in culture. tools for adverse reproductive health exposures, several steps have yet to be taken.
First, tests must be standardized for pregnant women. Although there is little evidence to suggest that pregnancy itself can alter these test outcomes, it is important to establish standard levels for pregnant women with normal pregnancy outcomes. Second, tests should be administered to women with known exposures, such as maternal smoking, so that patterns ofalterations can be correlated with reported exposures. Third, the tests must be associated with adverse pregnancy outcomes, such as reduced birthweight or gestational length. This last element in the validation research is particularly important since maternal exposure rather than fetal exposure is being measured. The extent to which xenobiotic chemicals cross the placental barrier may vary greatly, depending on the type ofexposures, timing in pregnancy, and maternal detoxification capability. Ifthe battery ofscreening tests proves useful, further field investigations would be warranted to determine the tests' ability to measure environmental exposures that adversely affect fetal development.
In reproductive epidemiology, we may be at a unique point for implementing this validation process. Because a number of studies of early pregnancy loss are collecting serial urines during pregnancy, the moment may be opportune to begin examining these urines for metabolic alterations, as tests of the potential usefulness ofthese nonspecific biomarkers to predict adverse pregnancy outcomes. Progress is being made in learning about these tests' response to specific environmental chemicals, but more research needs to focus on the quantitative relationship of these agents to body burdens. It would be helpful if this battery oftests were routinely applied to pregnant women in known exposure situations. Also, ifpregnant wvmen with abnormal tests (with and without adverse outcomes) were investigated further, much could be learned about the metabolic functions that are affected and the specific chemicals that are creating the effect.

Conclusion
Without better exposure measures, epidemiologic studies of reproduction will probably fail to identify xenobiotic fetotoxic agents in the environment. However, with an adequate battery of nonspecific exposure biomarkers, prospective studies of environmental effects on pregnancy outcomes might be possible. A proposed battery of nonspecific biomarkers should be tested to determine their usefulness for predicting adverse pregnancy outcomes.