How long does the effect of exposure last?

Measure of interest Relative risks for specific intervals of time since exposure are the major measures used to address this question. The pattern of relative risks must be examined for the latest indication that the relative risk is greater than one.

Data requirements Dates of each start and stop of exposure are required to answer this question. These are needed to classify the subjects' time spent in each time since exposure category. If full exposure histories are available, more sophisticated analyses are possible. However, if the critical issue is "time since exposure stopped," such multiple starts and stops will be difficult to analyze.

Potential problems with this approach If exposure is protracted, time since exposure must be analyzed in the proper time-dependent fashion (Clayton and Hills, 1993). This would apply to an examination of time since exposure stopped, when exposure is intermittent. Finally, a study group must have sufficient numbers of subjects with long times since exposure. Much longer time periods are needed than for addressing the previous question.

How does the effect of exposure vary with the age at which it was received?

Measures of interest Relative risks for exposure beginning at various ages are the critical measures needed to address this question. One must examine the pattern of relative risks associated with exposure beginning at various ages and compare the patterns of relative risks by time since exposure across age at exposure categories.

Data requirements Dates of exposure and date of birth are the critical data needed to construct these measures. These are needed to classify subjects as exposed or unexposed in each age category. The date of birth of study participants is generally known in epidemiologic studies. If level of exposure information is available, it would be used in preference to the simple exposed/unexposed categorization. For the relative risks stratified by time since exposure, the data requirements include those described above.

Potential problems with this approach The problems with this approach parallel those for the previous questions. Large studies with long follow-up are more likely than a small study to detect differential age effects. Sample size becomes a practical problem, as analysis within age groups requires more data than pooling all age groups within exposure categories. For examining time since exposure within age group, the comments about investigation of relative risk by time since exposure apply here as well. We found no studies that report the results needed to address this question for herbicides and cancer risk.



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 265
--> How long does the effect of exposure last? Measure of interest Relative risks for specific intervals of time since exposure are the major measures used to address this question. The pattern of relative risks must be examined for the latest indication that the relative risk is greater than one. Data requirements Dates of each start and stop of exposure are required to answer this question. These are needed to classify the subjects' time spent in each time since exposure category. If full exposure histories are available, more sophisticated analyses are possible. However, if the critical issue is "time since exposure stopped," such multiple starts and stops will be difficult to analyze. Potential problems with this approach If exposure is protracted, time since exposure must be analyzed in the proper time-dependent fashion (Clayton and Hills, 1993). This would apply to an examination of time since exposure stopped, when exposure is intermittent. Finally, a study group must have sufficient numbers of subjects with long times since exposure. Much longer time periods are needed than for addressing the previous question. How does the effect of exposure vary with the age at which it was received? Measures of interest Relative risks for exposure beginning at various ages are the critical measures needed to address this question. One must examine the pattern of relative risks associated with exposure beginning at various ages and compare the patterns of relative risks by time since exposure across age at exposure categories. Data requirements Dates of exposure and date of birth are the critical data needed to construct these measures. These are needed to classify subjects as exposed or unexposed in each age category. The date of birth of study participants is generally known in epidemiologic studies. If level of exposure information is available, it would be used in preference to the simple exposed/unexposed categorization. For the relative risks stratified by time since exposure, the data requirements include those described above. Potential problems with this approach The problems with this approach parallel those for the previous questions. Large studies with long follow-up are more likely than a small study to detect differential age effects. Sample size becomes a practical problem, as analysis within age groups requires more data than pooling all age groups within exposure categories. For examining time since exposure within age group, the comments about investigation of relative risk by time since exposure apply here as well. We found no studies that report the results needed to address this question for herbicides and cancer risk.

OCR for page 265
--> Does the exposure appear to act at an early or late stage of the carcinogenic process? Measures of interest The key statistical measures needed to address this question are the relative risks by age at exposure and time since exposure or, alternatively, the parameters in one of several models of carcinogenesis. In the multistage model of carcinogenesis, a healthy cell is presumed to go through a series of stages before becoming a cancer cell (Armitage and Doll, 1961; Chu, 1987). This model predicts specific patterns of relative risks by age and time since exposure, depending on whether the agent acts on an early or late stage of the carcinogenic process (Whittemore, 1977; Thomas, 1988). Further, the parameters in the multistage model or other mechanistic models, such as the two-event "initiator-promoter" model of Moolgavkar and Venzon (1979), may be estimated from cohort data to distinguish early- and late-stage effects. Data requirements To construct these measures, complete exposure histories and the date of birth are required. The study group must include subjects with protracted exposures, and there must be variation with respect to exposure histories. Potential problems with this approach Estimation of the needed quantities requires large studies with high-quality exposure-history data. Results Of The Literature Review Of Herbicide Exposure And Cancer For the purposes of this discussion, the review of the literature on herbicide exposure and cancer was focused on cancers in the "sufficient" and "limited/suggestive" evidence of association categories in VAO—that is, those cancers for which there was some evidence of an association. These are soft-tissue sarcoma, non-Hodgkin's lymphoma, Hodgkin's disease, prostate cancer, respiratory cancer, and multiple myeloma. Although VAO and Chapter 7 of this report review the entire relevant literature on herbicide exposure, this chapter only discusses the articles that provide results that the committee believes reflect, with reasonable accuracy, the timing of herbicide exposure and that had sufficient cases to make some judgment about the patterns of relative risks reported. Limitations of the Literature Review Approach In Chapter 4, the committee considers the problem of using a literature review in order to determine whether an association exists between herbicides and disease. The committee concludes that for overall questions of association between exposure and disease, the published literature would adequately report

OCR for page 265
--> results, whether "positive" or "negative," with respect to association from the studies that have been carried out to date. Thus, there should be little "publication bias," the tendency for positive results to be published more frequently than negative, in the review of the literature for association. In a specific investigation of timing issues based on a review of the literature, the same question of publication bias needs to be addressed. That is, is it more likely that results of investigations of timing issues will be published depending on the outcome of those investigations? Unlike measures of association (in particular relative risk) that are universally reported, results of investigations of timing issues are not routinely reported. Indeed, although it is not possible to determine the reasons that timing was or was not reported, it is quite plausible that negative results (that is, no differential effect of timing) are less frequently reported than positive results. One likely scenario is that if no association is found between exposure and disease, then either timing issues were not investigated or were investigated and only "interesting" results (that is, large changes over time intervals) were reported, but "uninteresting" results (no association over all time intervals) were not reported. Thus, the committee recognizes that there is a potential for publication bias in our review. Overview of the Findings This focused review of the literature found few articles that were informative about timing of exposure. For soft-tissue sarcoma, non-Hodgkin's lymphoma, Hodgkin's disease, and multiple myeloma, the committee concluded that there was very little information about the timing of exposure and subsequent risk and therefore that no further discussion of latency issues and these cancers was warranted. The committee did find that there was enough information about timing of exposure and respiratory and prostate cancers to warrant reporting these results, with considerably more information about the former than the latter. Some of the available epidemiologic studies, particularly those of production workers, appear to have adequate variation in timing of exposure available, such that further analyses of timing issues may well provide additional insights about the relationship between herbicide exposure and cancer. However, even for these cancers, the reports of some potentially informative studies did not include latency results, so there is potential for publication bias. Also, both of these cancers are in the "limited/suggestive" evidence category, indicating that the committee believes that the evidence for association between herbicide exposure and these cancers is not conclusive. This view has not changed after this investigation of latency issues.

OCR for page 265
--> Respiratory Cancer Background There is a substantial body of literature that explores issues of timing of exposure and respiratory cancer, because of its relatively high incidence and because numerous carcinogenic agents have been identified. We summarize some of the studies here to provide a background for the examination of these issues. Gamma rays In an investigation of latency issues for radiation exposure in the atomic bomb survivors, it was found that the relative risk of respiratory cancer began to rise within five to ten years after exposure and reached a plateau about 15 years after exposure. After 30 years of exposure, there was no evidence of a decrease in relative risk (Land, 1987). Radon daughters For miners exposed to radon daughters, the relative risk of lung cancer was seen to peak within five to ten years after first exposure, then slowly decline, although the risk still appears to be elevated even 30 years after exposure (Lubin et al., 1994; Thomas et al., 1994). In addition, the effect of exposure varies with age at exposure: a given exposure level results in a lower relative risk in older workers than it does in younger workers. Smoking Based on an analysis of a large case-control study of lung cancer in five Western European countries, Brown and Chu (1987) reported that relative risks rise slowly after the start of cigarette smoking and fall quickly after quitting. Among ex-smokers, the relative risk continues to decline to about 50 percent of that of smokers by 12 years after cessation but then remains fairly constant (but still elevated relative to nonsmokers). Among continuing smokers, for the same cumulative amount smoked, the relative risk declines with age at start of smoking. Arsenic In a cohort of workers from a copper smelter in Montana, relative risks were observed to increase with time after exposure, reaching a maximum between 15 and 20 years after exposure, after which they slowly declined (Breslow and Day, 1987). There was little change in relative risk with age at first exposure. Asbestos In a cohort of workers exposed to high levels of asbestos only briefly during World War II, the relative risk rose sharply between five and ten years after exposure, after which the relative risk has remained constant up to 40 years after exposure (EPA, 1986). The relative risks are independent of age at exposure.

OCR for page 265
--> Nickel In a cohort study of nickel refiners in England and Wales, the relative risk for lung cancer peaks less than 20 years after exposure, then decreases sharply. After 50 years, however, the risk is still elevated, except in the low-exposure group. The relative risks are more or less constant across age at first exposure (Kaldor et al., 1986). It is interesting to note that in contrast to these results, the same author reported quite a different pattern of relative risks for nasal sinus cancer. It was found that the relative risks for nasal sinus cancer continued to increase slowly with time since exposure, but increase markedly with age at first exposure. Thus, for all of these exposures, increases in relative risk either reached a plateau or peaked within 20 years after exposure. This indicates that the first detectable increases occurred somewhat earlier than this. The pattern of relative risks after reaching the peak and the pattern with age at exposure vary greatly across the agents, probably reflecting different mechanisms of action. Review of the Herbicide Exposure and Respiratory Cancer Literature Five studies have reported timing effects related to herbicide exposure and respiratory cancer. The National Institute for Occupational Safty and Health (NIOSH) study of chemical production workers gives the most detailed account of timing effects and exposure to TCDD. Fingerhut et al. (1991) report standardized mortality ratios (SMRs) for lung cancer of 0.8 1.0, and 1.2, for 0-9, 10-19, and 20+ years since first exposure to TCDD based on a total of 85 cases. They further stratify time since exposure SMRs by duration of exposure category, as reproduced in Table 8.1. The increasing trend in SMR with increasing time since first exposure is somewhat consistent over the duration of exposure categories. The mortality study of Dutch production workers, a subset of the International Agency for Research on Cancer (IARC) study, similarly reports respiratory cancer SMRs of 0.4, 0.4, and 1.6 for 0-9, 10-19, and 20+ years since first TCDD exposure. These results are based on nine cases, however—too few cases to yield statistically significant results (Bueno de Mesquita et al., 1993). Assuming that the exposure to TCDD related to the Seveso accident was of relatively short duration, time since the accident is essentially time since exposure. The mortality study by Bertazzi et al. (1989a,b) provides some latency results for lung cancer mortality. Relative risks, compared to the surrounding unexposed population, are given by calendar periods 1976-81 and 1982-86 for males for Zones A, B, and R (see Chapter 7 for further details of this study). For those living in the area at the time of the accident, these correspond to 0-5 and 6-10 years since exposure to TCDD. Thus, assuming that the in-migrants represent a small proportion of the cohort, the calendar period relative risks will approximate time since exposure. The relative risks are summarized in Table 8.2. There are small, but consistent increases in relative risk with calendar period. In an 18-year follow-up of Finnish herbicide applicators, Asp et al., (1994)

OCR for page 265
--> TABLE 8-1 NIOSH Study: Respiratory Cancer Relative Mortality by Time Since First Exposure and Duration of Exposure to TCDD   Duration of exposure to TCDD (Years) Time since first exposure <1   1-4   5-14   15+   Overall     Obs SMR Obs SMR Obs SMR Obs SMR Obs SMR 0-9 3 77 3 95 1 79 0 0 7 84 10-19 6 69 5 79 9 180 1 137 21 101 20+ 17 96 17 126 14 146 9 156 57 123 Total 26 86 25 109 24 151 10 154 85 112   SOURCE: Fingerhut et al., 1991, Table 4.

OCR for page 265
--> TABLE 8-2 Seveso Study: Lung Cancer Relative Mortality in Men by Calendar Period     Relative Risk   Zone Observed deaths 1976-1981a 1982-1986b A 2 (not givenc) 2.0 B 20 1.1 1.8 R 77 0.7 0.9 a Approximately 0-5 years since first exposure. b Approximately 6-10 years since first exposure. c Presumably 0 cases. SOURCE: Bertazzi et al., 1989. TABLE 8-3 Finnish Applicators Study: Respiratory Cancer Observed and Expected Deaths and SMRs for Men by Time Since First Exposure to Chlorophenoxy Herbicide Time since first employment 0-9 years     11-14 years     15+ years     Obs Exp SMR Obs Exp SMR Obs Exp SMR 4 6.4 0.6 11 8.1 1.4 22 21.0 1.0   SOURCE: Asp et al., 1994, Table 3. gives the respiratory cancer SMRs relative to the male Finnish age and calendar year specific rates by 0, 10, and 15 year ''latency intervals." These correspond to 0+, 10+, and 15+ years since first exposure to chlorophenoxy herbicides. There were a total of 37 respiratory cancer deaths. By subtracting the given observeds and expecteds, the SMRs by time since first exposure can be computed. These are given in Table 8.3. These data do not indicate a trend in the SMRs. Conclusions Perhaps because respiratory cancers are the most common type of cancer in all of the cohort studies, there is more latency information available for them than for any other cancer. However, based on the review of the evidence in VAO and Chapter 7 of this report, respiratory cancer is in the "limited/suggestive" evidence category, indicating that the committee believes that the evidence for association between herbicide exposure and these cancers is not conclusive. Although investigation

OCR for page 265
--> of latency effects could result in a change in association categorization, the fact that the committee reviewed the literature for latency effects does not imply an a priori belief on the part of the committee that the association is definitive. How long does it take after exposure to see an increase in disease risk? The evidence in the literature suggests that the time from exposure to TCDD to increased risk of respiratory cancer occurs within ten years. This conclusion is based primarily on the NIOSH study (Fingerhut et al., 1991), because this study is the most informative about the changes in risk of respiratory cancer with time since first exposure to TCDD. The Seveso study (Bertazzi et al., 1989a,b) is also suggestive of an increase in relative risk five years after exposure, but the number of exposed cases is too small to rule out chance as a likely explanation for the observed increases. The other relevant studies have relatively few respiratory cancer cases but are consistent with this conclusion. It is unlikely that chance can explain these patterns of relative risk, given the numbers of cases and the consistency of the pattern across duration of exposure category. Another explanation of the existing epidemiologic evidence is that TCDD exposure is not associated with respiratory cancer but is correlated with some other respiratory cancer risk factor, and that risk factor is related to the cancer and to TCDD exposure in a way that results in the observed pattern of relative risks with time since first exposure to TCDD. This explanation requires that the putative risk factor explain the observed increasing relative risk with duration as well as the time since first exposure pattern (see Table 8.1). Such a situation would occur if duration of TCDD exposure were correlated with duration of some other exposure that caused respiratory cancer, such as cigarette smoking or other chemical exposure. The committee is not aware of any evidence for this hypothesis but cannot discount it. How long do the effects of exposure last? If there is, in fact, a causal association between TCDD exposure and respiratory cancer, the literature suggests that the risk is elevated at least five years after exposure, but we cannot determine how long it takes before the relative risks return to one, if they ever do. However, chance cannot be ruled out as a possible explanation for the observed pattern of relative risk. This conclusion is based primarily on the Seveso mortality study (Bertazzi et al., 1989a,b) that reports a slight increase in the relative risk in the second five years after the accident (see Table 8.2), indicating that the risk continues to increase five years after exposure ends. All of the study groups that demonstrated latency results had protracted exposures. None of these report results of the type of analysis needed to address this question (see the previous section). However, although the NIOSH study does not directly address this issue, we can reasonably infer that the SMRs for time since exposure in the <1 year duration of exposure

OCR for page 265
--> category closely track those of the time since first exposure. This is because subjects with less than one year of exposure were likely to have received all of their exposure in a single period of time. This will be less likely for subjects with longer duration of exposure because the possibility of "intermittent exposure" is greater. The SMRs, given in the first column of Table 8.1, appear to increase from <20 years to 20+ years since first exposure, but this is based on rather small numbers (nine cases in the <20 year group), and all the SMRs are less than one. How does the effect of exposure vary with the age at which it was received? None of the available studies provides information on the variation of the effect of exposure with age. Does the carcinogen appear to act at an early or late stage of the carcinogenic process? None of the available studies addresses this issue. Prostate Cancer Background There are no environmental exposures other than herbicides associated with prostate cancer for which latency issues have been investigated. Review of the Herbicide Exposure and Prostate Cancer Literature The NIOSH study of chemical production workers exposed to TCDD (Fingerhut et al., 1991) reports SMRs for prostate cancer for 20+ years since first exposure by duration of exposure <1 and 1+ years as well as for the entire cohort (Fingerhut et al., 1991). These SMRs, with observed and expected numbers of deaths, are given in Table 8.4. Because numbers of observed and expected deaths TABLE 8-4 NIOSH Production Workers Study: Prostate Cancer Observed, Expected and SMRs for Men by Time Since First Exposure and Duration of Exposure to TCDD   20+ years since first exposure Entire cohort   < 1 year exposure   1+ years exposure   Obs Exp SMR Obs Exp SMR Obs Exp SMR 17 13.9 122 2 3.0 67 9 5.9 152   SOURCE: Fingerhut et al., 1991, Table 2.

OCR for page 265
--> TABLE 8-5 NIOSH Production Workers Study: Prostate Cancer Observed and Expected Numbers of Deaths and SMRs by Time Since First Exposure to TCDD Time since first exposure < 20 years     20+ years     Obs Exp SMR Obs Exp SMR 6 5.0 120 11 8.9 123   SOURCE: Derived from Fingerhut et al., 1991, Table 2. are not given by duration of exposure for the entire cohort, it is not possible to compare the SMRs for <20 and 20+ years since firs exposure within duration of exposure categories. Thus, the most informative comparison is based on SMRs for time since first exposure categories without respect to duration of exposure. The SMR for 20+ years since first exposure is obtained by summing the observed and expected numbers for the two duration categories. The SMR for <20 years since first exposure is obtained by subtracting the 20+ year numbers from the total. These are given in Table 8.5. While both are slightly elevated, there is no difference in the SMRs between the <20 and 20+ categories. Assuming that the exposure to TCDD after the Seveso accident was of relatively short duration, time since the accident is essentially the same as time since exposure. The mortality study by Bertazzi et al. (1989a,b) provides results relevant to timing of exposure for prostate cancer mortality. Relative risks, compared to the surrounding unexposed population, are given by calendar periods 1976-81 and 1982-86 for males for Zones B and R. For those living in the area at the time of the accident, these time periods correspond to 0-5 and 6-10 years since exposure. Thus, assuming that inmigrants represent a small proportion of the cohort, the calendar period relative risks will approximate the time since exposure. The relative risks are summarized in Table 8.6. There are too few cases (three) in Zone B to draw any conclusions about time since first exposure and changing risk patterns. In Zone R, there is a decrease in the relative risk with calendar period, although the small number of cases and the fact that this is a mortality rather than an incidence study preclude strong statements about the actual pattern of relative risks. Conclusions Our review of the literature yielded only two articles on prostate cancer with latency-related results and sufficient numbers of cases for statistical analysis. Prostate cancer is in the limited/suggestive evidence category, so it is important

OCR for page 265
--> TABLE 8-6 Seveso Study: Prostate Cancer Relative Mortality in Men by Calendar Period     Relative risk   Zone Observed deaths 1976-1981a 1982-1986b Ac — — — B 3 2.8 1.5 R 16 1.9 1.2 a Approximately 0-5 years since first exposure. b Approximately 6-10 years since first exposure. c Not provided.   SOURCE: Bertazzi et al., 1989. to keep in mind that the committee believes that the evidence for association between herbicide exposure and prostate cancer is not conclusive. Although investigation of latency effects could result in a change in association categorization, the fact that the committee reviewed the literature for latency effects does not imply an a priori belief on the part of the committee that the association is definitive. How long does it take after exposure to see an increase in disease risk? The inconsistent, limited data from the NIOSH study (Fingerhut et al., 1991) and the Seveso study (Bertazzi et al., 1989a,b) do not indicate any increase in the relative risk of prostate cancer with time since exposure to TCDD. Because the NIOSH cohort was grouped into categories of <20 and 20+ years since first exposure, they give no indication about changes in risk of prostate cancer within 20 years of exposure. The Seveso study has too few cases of prostate cancer and too short an observation period (up to ten years) to be informative on this question. How long do the effects of exposure last? The available evidence is not informative on this issue. How does the effect of exposure vary with the age at which it was received? None of the studies provides information on the variation of the effect of exposure with age. Does the carcinogen appear to act at an early or late stage of the carcinogenic process? None of the studies addresses this issue.

OCR for page 265
--> 9 Reproductive Effects Introduction This chapter summarizes published scientific literature on exposure to herbicides and adverse reproductive and developmental effects. The literature discussed includes papers published since VAO. VAO included a number of environmental, occupational, and Vietnam veteran studies that evaluated herbicide exposure and the risk of adverse outcomes, including spontaneous abortion, birth defects, stillbirths, neonatal and infant mortality, low birthweight, and sperm quality and infertility. The report concluded that the evidence at that time was inadequate or insufficient to determine whether an association exists between exposure to herbicides and each of the above reproductive and developmental outcomes. The primary emphasis of VAO and the present review is on the potential adverse reproductive effects of herbicide exposure for males, because the vast majority of the Vietnam veterans are men. Nevertheless, a brief discussion of the epidemiologic findings pertaining to female exposure is warranted because of the Department of Veterans Affairs' planned study of female Vietnam veterans and their reproductive health. Additionally, there is growing evidence from experimental animal research on female reproductive toxicity and developmental toxicity via in utero exposure to dioxin. A number of studies have attempted to evaluate the potential association between herbicide exposure in women and the risk of adverse reproductive outcomes, including spontaneous abortion, stillbirth, preterm delivery, and birth defects (Hemminki et al., 1980; McDonald et al., 1987; Ahlborg et al., 1989;

OCR for page 265
--> Savitz et al., 1989; Fenster and Coye, 1990; Restrepo et al., 1990; Goulet and Theriault, 1991; Nurminen et al., 1994). The quality and results of these studies have been mixed. A major limitation of nearly all the studies is the determination of specific exposures. Many studies have defined exposure solely based on employment in agricultural occupations. Exposure to specific chemicals and other agents in these agricultural settings is usually not ascertained. Further, problems such as incomplete ascertainment of the outcome of interest, selection of inappropriate or no control groups, and failure to account for confounding factors have plagued some of this work. Improvements in study design, especially exposure assessment, should allow for a more definitive evaluation of the relationship between herbicide exposure and adverse reproductive outcomes among women. The following sections will separately discuss specific categories of reproductive effects: fertility, spontaneous abortion, stillbirth, birth defects, and childhood cancer. Childhood cancer is discussed in this chapter rather than in Chapter 7, since many such cancers are related to preconceptual and in utero exposures. For most outcomes, a brief summary of the scientific evidence in VAO is presented, followed by an update of the recent scientific literature. A complete discussion of the evidence is presented for birth defects, because the committee has changed its assessment of this literature since VAO. Fertility Background Male reproductive function is a complex system under the control of several components whose proper coordination are important for normal fertility. There are several components or endpoints related to male fertility, including reproductive hormones and sperm parameters. Only a brief description of male reproductive hormones will be given here; more detailed reviews can be found elsewhere (Yen and Jaffe, 1991; Knobil et al., 1994). The reproductive neuroendocrine axis involves the central nervous system, the anterior pituitary gland, and the testis. The hypothalamus integrates neural inputs from the central and peripheral nervous systems and regulates gonadotrophins (luteinizing hormone and follicle-stimulating hormone). Both of these hormones are necessary for normal spermatogenesis. Luteinizing hormone (LH) and follicle-stimulating hormone (FSH) are secreted in episodic bursts by the anterior pituitary gland into the circulation. LH interacts with receptors on the Leydig cells, which leads to increased testosterone synthesis. FSH and testosterone from the Leydig cells interact with the Sertoli cells in the seminiferous tubule epithelium to regulate spermatogenesis. Several agents, such as lead and dibromochloropropane (DBCP), have been shown to affect the neuroendocrine system (Ng et al., 1991; Whorton et al., 1979).

OCR for page 265
--> Summary of VAO Only one occupational epidemiologic study is available for assessing the association between herbicide exposure and altered sperm parameters (sperm count, motility, morphology). This study of 2,4-D exposure did indicate an association with abnormal sperm morphology; however, given the small sample size and lack of additional studies, the evidence for determination of an association is considered inadequate. No studies were identified that examined occupational or environmental exposure and impaired fertility. One study of veterans reported an association with altered sperm measures (reduced sperm concentration and increased percentage of abnormal sperm), although there was no relationship to the number of children fathered, self-reported herbicide exposure, or the extent of combat experience. The paucity of occupational studies, lack of consistent findings in veterans studies, and methodologic problems in the studies reviewed do not permit a valid assessment of an increased infertility. Update of the Scientific Literature The Centers for Disease Control and Prevention (CDC) Vietnam Experience Study (VES) did not find any association between service in Vietnam and alterations in FSH, LH, and testosterone (Centers for Disease Control, 1989). The recent analysis of the Ranch Hand data indicated that FSH was not associated with estimated initial dioxin level (Roegner et al., 1991). There was a pattern of decreasing testosterone with increasing serum dioxin, although this was statistically nonsignificant. NIOSH researchers conducted a cross-sectional study to evaluate the relationship between serum dioxin and serum testosterone and gonadotrophins in men previously occupationally exposed to dioxin and in a referent group (Egeland et al., 1994). The exposed group consisted of men who were either current or former employees at two of the 12 plants that are part of the NIOSH cohort study of dioxin-exposed workers. The plants manufactured 2,4,5-T from 1951 to 1969 in New Jersey and from 1968 to 1972 in the Missouri plant. A cross-sectional medical study was conducted in 1987, with a total of 586 workers identified. Among these men, 400 were considered eligible (143 died, 43 were not located). A total of 357 workers completed the interview, and 281 completed a medical exam (70 percent of eligible, 48 percent of original group). For comparison, age-, sex-, and race-matched neighborhood referents were identified and contacted. A total of 325 referents were interviewed, and 260 completed the medical exam. The original number of referents contacted was not stated. The interview included questions on medical history, demographics, and lifestyle factors. A random sample of the comparison subjects was chosen (N = 99), and their mean dioxin value (6.08 picograms/gram) was assigned to the 161 referent subjects without a serum measurement. A half-life decay model (7.1 year half-life; steady-state

OCR for page 265
--> background = 6.08 picograms/gram) was used to estimate past dioxin levels. A single measurement of total serum testosterone, FSH and LH was taken for each subject. The strength of the relationship between serum dioxin and serum levels of testosterone, FSH, and LH were estimated by linear regression (continuous dioxin) and logistic regression (using quartiles of dioxin), adjusting for age, body-mass index (BMI), diabetes, alcohol, smoking, and race. The results of the linear regression analysis indicated that current serum dioxin was related to FSH, LH, and testosterone levels. Serum dioxin was positively related to FSH (b = .04) and LH (b = .03), and inversely related to testosterone (b = -.02). The regression coefficient b represents the unit increase or decrease in gonadotrophins and testosterone per unit increase in serum dioxin. The magnitude of the increases or decreases in hormones was thus rather small, compared to the normal range for these hormones in humans. The logistic regression analysis used serum dioxin categorized into quartiles and ''high" LH and FSH and "low" testosterone. High LH was defined as >28 IU/liter (laboratory standard for normal range = 5-28 IU/liter) and high FSH as >31 IU/liter (normal range = 3-31 IU/liter). Low testosterone was defined as <10.4 nmol/liter (normal range = 9.4-34.7 nmol/liter). These cutoffs were at the 8th percentile of the LH and FSH distributions. The cutoff for low testosterone was set at the 8th percentile (<10.4 nmol/liter) to be consistent with FSH and LH. There was an association found between high LH and current serum dioxin (2nd dioxin quartile OR = 1.9; 3rd OR = 2.5; 4th OR = 1.9; p for trend = .03). For FSH, a pattern of increasing risk with increasing serum dioxin was also found, but the test for trend was not statistically significant (p = .10). The adjusted odds ratios for low testosterone were more elevated (2nd dioxin quartile = 3.9; 3rd = 2.7; 4th = 2.1), but again the trend test was not significant (p = .10). Similar estimates were obtained for half-life serum dioxin extrapolated to the time at which occupational exposure ended. The results of this study indicated that estimated dioxin exposure levels were positively associated with LH and FSH levels and negatively associated with serum testosterone. The strengths of the study included the use of an occupational cohort with documented dioxin exposure and an attempt to control for potentially confounding factors. However, the cross-sectional nature of the study, the fact that only one serum hormone measurement was obtained, the imprecision of the effect estimates, the failure to detect a dose-response gradient, and the relatively low proportion of participating workers included in the analyses are of concern. The implications of some of these limitations on the interpretation of their findings are not straightforward. The authors argued that their single measurement results are likely to be conservative. Although this issue is not directly addressed in the original paper, the authors have argued in a Letter to the Editor in the American Journal of Epidemiology that any laboratory errors in the measurement of serum dioxin and hormones are unlikely to produce the results they obtained and that within- and between-assay variation was acceptable (Egeland

OCR for page 265
--> et al., 1995). As to possible selection bias due to the poor response, no data exist to directly evaluate possible biasing effects on the odds ratio estimates. As the authors correctly noted, the magnitude of the differences in hormone concentrations were small. Some of the men did have hormone levels that were either above (LH and FSH) or below (testosterone) the cutoff of the 8th percentile. A major issue in the interpretation of these findings, if real, is whether these changes in hormone concentration ascribed to dioxin have any implications for reproductive failure. Clearly, the hormonal changes are rather subtle and are well below levels expected to result in gonadal failure. Although the one study reviewed here suggested an association between TCDD exposure and changes in male reproductive hormones, there were a number of methodologic concerns with the study that do not permit definitive conclusions to be drawn. Conclusions Strength of Evidence in Epidemiologic Studies There is inadequate or insufficient evidence to determine whether an association exists between exposure to the herbicides considered in the report and altered sperm parameters or infertility. The evidence regarding association is drawn from occupational and other studies in which subjects were exposed to a variety of herbicides and herbicide components. Biologic Plausibility Experimental animal evidence supports the notion that dioxin can alter testosterone synthesis, generally at relatively high doses, but does not provide direct clues as to the reproductive significance of hormone disregulation of the magnitude found in available studies. Spontaneous Abortion Background Spontaneous abortion (or miscarriage), according to the World Health Organization (1977), is a "nondeliberate fetal death of an intrauterine pregnancy before 22 completed weeks of gestation, corresponding to a fetal weight of approximately 500 grams or more." Pregnancy losses prior to implantation (preimplantation) are not clinically detectable with current diagnostic procedures. The rate of early (postimplantation) detectable (but often unrecognized) pregnancy losses has been estimated to be approximately 30 percent (Wilcox et al., 1988). Because preimplantation and early postimplantation losses are difficult to ascertain for epidemiologic studies of pregnancy loss, the appropriate epidemiologic

OCR for page 265
--> endpoint for these studies is not all spontaneous abortions but rather all clinically recognized spontaneous abortions—those that come to the attention of a woman or her physician. All subsequent discussions of pregnancy loss, miscarriage, or spontaneous abortion refer to clinically recognized outcomes unless otherwise specified. Approximately 10 to 15 percent of all clinically recognized pregnancies end in a clinically recognized loss. Of these clinically recognized pregnancy losses, 35 to 40 percent are losses of chromosomally abnormal embryos and fetuses (Kline et al., 1989). A wide range of maternal characteristics and exposures has been linked to miscarriage. However, two major risk factors have been established—advanced maternal age and history of previous miscarriage (Kline et al., 1989). Summary of VAO The studies involving occupational and environmental herbicide exposure generally reported no association with spontaneous abortion; however, these studies were inadequate with respect to sample size, elimination of potential bias, and assessment of exposure. The available epidemiologic studies of veterans are generally limited by inadequate sample size, potential bias, and other methodologic problems. There are some suggestive findings indicating an increased risk for Vietnam veterans, including a possible dose-response gradient of increasing risk with increasing estimated (self-reported or inferred) Agent Orange exposure. Nonetheless, the inconsistency with environmental and occupational studies, the uncertainty of the methods of exposure determination, the marginal magnitude of the increased risk, and the failure to exclude chance are of enough concern that the evidence can be considered insufficient. Future analyses of the data for Ranch Hands, the Air Force personnel involved in handling and spraying herbicides, may contribute important evidence regarding an increased risk for spontaneous abortion among exposed Vietnam veterans. Update of the Scientific Literature The general introduction and review of the recently published Ranch Hand study of reproductive outcomes (Wolfe et al., 1995) can be found in the section on birth defects below. Overall, the conception and birth rates were similar for the Ranch Hands and the comparison veterans. There was a total of 157 (16 percent) spontaneous abortions among Ranch Hands and 172 (14 percent) among comparison veterans. The results by dioxin level showed only a weak increased relative risk estimates for the background (RR = 1.1; CI 0.8-1.5) and low-level (RR = 1.3; CI 1.0-1.7) categories and no association for the high-level category (RR = 1.0; CI 0.7-1.3).

OCR for page 265
--> The Ranch Hand study had some clear strengths. The AFHS cohorts were well-defined and systematically followed, and health outcomes were carefully monitored. The Ranch Hands represent veterans with high potential for herbicide and dioxin exposure, and the use of serum dioxin assays has provided a direct estimate of exposure that is superior in some respects to the retrospective exposure assessment methods based on troop movement and spraying data. Several aspects of how the study was conducted and how the data were analyzed deserve consideration. First, the difference in the number of original cohort subjects and those that were included in the final analyses raises the possibility of selection bias. The authors noted that those who volunteered for the study were more likely to have children with birth defects than those who did not volunteer. The impact on the risk ratio is not likely to be important, because this volunteerism was not associated with exposure. Nonetheless, there is no discussion of a comparison of the original cohort subjects and their children with those who were included in the final analyses, not just those who agreed to the serum dioxin assay. The complete exclusion of the subjects with serum dioxin sample measurements below the level of detectability, rather than their inclusion with the background or referent categories, does not appear justified. Conclusions Strength of Evidence in Epidemiologic Studies There is inadequate or insufficient evidence to determine whether an association exists between exposure to the herbicides considered in this report and spontaneous abortion. The evidence regarding association is drawn from occupational and other studies in which subjects were exposed to a variety of herbicides and herbicide components. Stillbirth Background The use of the terms "stillbirth" and "neonatal death" can be confusing and has differed in various epidemiologic studies. Stillbirth (or late fetal death) is typically defined as the delivery of a fetus occurring at or after 28 weeks of gestation and showing no signs of life at birth, although a more recent definition includes deaths among all fetuses weighing more than 500 grams at birth, regardless of gestational age at delivery (Kline et al., 1989). Neonatal death is usually defined as the death of a liveborn infant within the first 28 days of life. Because there are no clear biological differences between late fetal deaths (stillbirths) and deaths in the early neonatal period, these are commonly referred to together as

OCR for page 265
--> perinatal deaths (Kallen, 1988). Stillbirths occur in approximately 1 to 2 percent of all births (Kline et al., 1989). Among low birthweight live- and stillborn infants (500-2,500 grams), placental and delivery complications such as abruptio placentae, placenta previa, malpresentation, and umbilical cord complications are the most common causes of perinatal mortality (Kallen, 1988). Among infants weighing more than 2,500 grams at birth, the most common causes of perinatal death are lethal congenital malformations and placental complications (Kallen, 1988). Summary of VAO A statistical association with stillbirth has not been reported in the available occupational and environmental epidemiologic studies. The majority of studies did not have adequate statistical power, and the assessment of exposure was incomplete. Some studies of veterans have reported an increased risk, whereas others have indicated no statistical association. Interpretation of these veteran studies is constrained by limited statistical power and most importantly, by uncertainty of correctly assigning herbicide exposure to study groups. Update of the Scientific Literature The general introduction and review of the recently published study of reproductive outcomes among the Ranch Hands is provided in the section on spontaneous abortion (Wolfe at al., 1995). With respect to stillbirth, a total of 14 (1.4 percent) Ranch Hand conceptions resulted in a stillbirth, compared with 13 (1.1 percent) comparison veteran conceptions. Elevated risk ratios were found for the background (RR = 1.8; CI 0.7-4.5) and low (RR = 1.8; CI 0.7-4.7) levels, although both estimates were nonsignificant and relatively imprecise. The risk ratio for the high level was 0.3 (CI 0.0-2.3), based on one stillbirth among Ranch Hands. Thus, study results indicated an association between TCDD levels and the risk of stillbirth among those Ranch Hand subjects with either background or low exposure levels. No association was seen among veterans in the high exposure category. The evidence from the previous studies reviewed in VAO was mixed: some occupational, environmental, and Vietnam veteran studies suggested an increased risk of stillbirth, while other studies did not report an association. Conclusions Strength of Evidence in Epidemiologic Studies There is inadequate or insufficient evidence to determine whether an association exists between exposure to the herbicides considered in this report and

OCR for page 265
--> stillbirth. The evidence regarding association is drawn from occupational and other studies in which subjects were exposed to a variety of herbicides and herbicide components. Birth Defects Background The March of Dimes defines a birth defect as "an abnormality of structure, function or metabolism, whether genetically determined or as the result of an environmental influence during embryonic or fetal life" (Bloom, 1981). Other terms often used interchangeably with birth defects are "congenital anomalies" and "congenital malformations." Major birth defects are usually defined as those abnormalities that are present at birth and severe enough to interfere with viability or physical well-being. Major birth defects are seen in approximately 2 to 3 percent of live births (Kalter and Warkany, 1983). An additional 5 percent of birth defects can be detected with follow-up through the first year of life. Given the general frequency of major birth defects of 2 to 3 percent and the number of men who served in Vietnam (2.6 million), and assuming that they had at least one child, it has been estimated that 52,000 to 78,000 babies with birth defects have been fathered by Vietnam veterans, even in the absence of an increase due to exposure to herbicides or other toxic substances (Erickson et al., 1984a). Epidemiologic Studies of Birth Defects Because the publication of new data from the Ranch Hand study has caused the committee to change its conclusion about the strength of the evidence regarding the association between exposure to herbicides used in Vietnam and birth defects, the following material was included from VAO to present a complete picture about the evidence for the committee's conclusions. The section entitled "Ranch Hand Study," however, is based on the new information. Occupational Studies Four occupational epidemiology studies have examined the potential association between herbicide exposure of male workers and birth defects. The Townsend study (Townsend et al., 1982) of workers with potential dioxin exposure at a Dow Chemical plant did not find an increased risk of birth defects among dioxin-exposed workers (30 births with anomalies; 47/1,000 births) compared to unexposed workers (87 births with anomalies; 49/1,000 births; OR = 0.9, CI 0.5-1.4). A major limitation of this study is its limited statistical power to detect an elevated odds ratio for specific defects. The authors noted that the study had 26 percent power to detect a doubling of risk due to exposure for a group of

OCR for page 265
--> indicator malformations (anomalies thought to be easily recognized and reported by the mother, such as an oral cleft, spina bifida, and Down's syndrome). An additional problem is that despite the use of these "indicator malformations," without medical records, validation of the accuracy of maternal self-report of birth defects is questionable for many conditions. Two studies of workers from a 2,4,5-T plant in Nitro, West Virginia, did not report an association with birth defects among offspring (Moses et al., 1984; Suskind and Hertzberg, 1984). The relative risk estimates for any birth defect were 1.3 (CI 0.5-3.4) for Moses et al. and 1.1 (CI 0.5-2.2) from the Suskind and Hertzberg study. Both studies had limited statistical power, given the small number of subjects (204 exposed workers in the Suskind and Hertzberg study; 117 exposed workers in the Moses study). This is especially problematic for the evaluation of most specific birth defects. Both studies also relied on self-reports for the ascertainment of birth defects. A study of 2,4,5-T sprayers found only a slightly elevated odds ratio for congenital anomalies (OR = 1.2, CI 0.5-3.0) associated with the spraying group (Smith et al., 1982). The study used self-administered questionnaires to determine outcomes. Like the other studies, it had limited power for the analysis of individual birth defects. Environmental Studies A variety of environmental studies have examined the relationship between herbicide exposure and prevalence of birth defects (Nelson et al., 1979; Gordon and Shy, 1981; Hanify et al., 1981; Mastroiacovo et al., 1988; Stockbauer et al., 1988; White et al., 1988; Fitzgerald et al., 1989; Jansson and Voog, 1989). Some studies reported a statistical association with specific birth defects (clubfoot, Fitzgerald et al., 1989; cleft lip with or without cleft palate, Gordon and Shy, 1981; heart, hypospadias, clubfoot, Hanify et al., 1981; oral clefts, Nelson et al., 1979), although others have not reported an association (Stockbauer et al., 1988; Fitzgerald et al., 1989; Jansson and Voog, 1989), including the Seveso study (Mastroiacovo et al., 1988). Interpretation of the results of these environmental studies is difficult, because most of the studies were inconsistent, were based on ecologic correlations, had inadequate statistical power, did not validate birth defects recorded from vital statistics or self-reports, and included both male and female exposures. A recently published study from Vietnam evaluated the risk of birth defects among the offspring of mothers who resided in a village in the southern part of the country that had been sprayed during the conflict (Phuong et al., 1989); 81 cases of birth defects (diagnosis not specified) were identified. No differences were reported between cases and controls for the potentially confounding factors investigated. Strong associations were found for birth defects (calculated from data presented; OR = 3.8, CI 1.1-13.1). The paper is difficult to evaluate given

OCR for page 265
--> the sparse details presented. Study design factors such as how birth defects were diagnosed and what types were detected, the size of the original case and control groups from which the final groups were sampled, the pattern of patient accrual for this hospital, the method of data collection, and how the potential herbicide spraying histories were determined were not specified. Finally, to put the study in the context of this review, the potential exposure 17 to 22 years earlier pertains to both the mother and the father. Results from a number of other studies from Vietnam, both of sprayed villages in the southern part of the country and of veterans returning to the unsprayed northern regions, have been reported, mostly in a review by Constable and Hatch (1985). These studies indicate an increased risk of birth defects, including anencephaly, oral clefts, and a variety of other anomalies. Nonetheless, these studies generally suffer from poor reporting and a variety of methodologic problems such as limited control of confounding factors, use of a referral hospital, lack of comparison groups, uncertainty of exposure classification, and no validation of reported birth defects. Although the findings are suggestive of an association between herbicide spraying and birth defects, the available studies are insufficient to draw firm conclusions. Vietnam Veterans Studies As part of the CDC Vietnam Experience Study (1989), the reproductive outcomes and the health of children of male veterans were examined. The VES assessment included a telephone interview, a review of hospital birth defect records for a subsample of veterans who underwent a medical examination, and a review of the medical records of selected birth defects for all study subjects. The interview data revealed that Vietnam veterans reported more birth defects (64.6 per 1,000 total births) among offspring than did non-Vietnam veterans (49.5 per 1,000 total births). The adjusted odds ratio estimate for congenital anomalies as a group was 1.3 (CI 1.2-1.4). When examined by specific defect category, elevated adjusted odds ratios were found for defects of the nervous system (OR = 2.3, CI 1.2-4.5); ear, face, neck (OR = 1.6, CI 0.9-2.8); and integument (OR = 2.2, CI 1.2-4.0). A small but statistically significant odds ratio of 1.2 (CI 1.1-1.5) was found for musculoskeletal defects. An analysis of specific defects considered by the investigators to be relatively common and reliably diagnosed was also conducted. Elevated (crude) odds ratios were reported for hydrocephalus (OR = 5.1, CI 1.1-23.1), spina bifida (OR = 1.7, CI 0.6-5.0), and hypospadias (OR = 3.1, CI 0.9-11.3). Vietnam veterans also reported having more children with multiple defects (OR = 1.6, CI 1.1-2.5) than non-Vietnam veterans. An analysis of Vietnam veterans' self-reported herbicide exposure found a dose-response gradient, with an adjusted odds ratio for birth defects of 1.7 (CI 1.2-2.4) at the highest level of exposure. The VES also examined serious health problems in the veterans' children;