GESTATIONAL TROPHOBLASTIC DISEASE (GTD) includes hydatidiform mole (HM) and choriocarcinoma. Hydatidiform mole complicates pregnancy and is the main factor of choriocarcinoma, a highly malignant cancer. Hydatidiform mole originates during gametogenesis or during fertilization. It is classified as complete or partial by histopathologic and cytogenetic criteria. Complete hydatidiform moles (CHMs) contain no embryo, umbilical cord, or amniotic membranes, and the chorionic villi of the trophoblastic tissues are all abnormal. its chromosomal constitution is usually 46XX. The origin of complete moles is androgenetic,[2-4] all chromosomes originating from the father. in contrast, partial moles are characterized by an embryo and by normal and hydropic chorionic villi. Their karyotype contains maternal chromosomes, usually with a triploid genotype of dispermic origin.[6,7] Complete moles constitute a high-risk group of choriocarcinoma, but whether partial moles become malignant has yet to be elucidated.[8-10]
The etiology of GTD remains almost unknown. Relatively high incidence rates of HM have been reported in some Asian countries and in other parts of the world. incidence rates of HM of approximately 2.5/1 000 pregnancies (i.e., two or three times higher than in Western Europe and the United States[14-16]) have been reported in Japanese studies published recently. Several genetic factors have been studied in association with an increased risk of choriocarcinoma or HM; the influence of consanguinity, suspected because of the high incidence rate of HM within some isolated populations, is not yet established. An excess of blood-group A and a deficit of blood-group O have been found in several series of choriocarcinoma cases, and there appears to be an increased risk of choriocarcinoma with the combinations of AxO or OxA blood groups between the woman and her male partner. Results of studies investigating the influence of Rhesus factors suggest that compatibility for the Rh-D factor between the hydatidiform trophoblast cells and the patient plays a role in the development of HM. It was suggested in a study on HLA markers20 that some HLA antigens might be related to gametogenesis-fecundation dysfunction and might be involved in early rejection of abnormal conceptus.
Whatever the geographic origin, late maternal age is the most firmly established risk factor[14,21,22] for CHMs, but not for partial moles.[5,23,24] Risk from late paternal age was reported by La Vecchia et al., but this was not confirmed by other studies.[25,26] In some studies, a significant protective effect of prior-term pregnancy has been reported for GTD[25,27] and for CHMs, the risk of HM decreasing with the number of previous births. A higher frequency of prior spontaneous abortions among cases of GTD was reported in some,[25,29] but not all,[26,30] studies.
In an effort to explain the geographic variability in incidence rates, socioeconomic and nutritional factors have been considered for a long time, but their influence, measured in general or indirect terms, is inconclusive, including in the United States.[5,14,25] A decreased risk of HM was reported in subjects who increased carotene intake. An increased risk of HM was associated with smoking and its duration in an Italian study; a similar result was not found in a U.S. study, in which the duration of smoking was not assessed.
An increased incidence of HM and choriocarcinoma, estimated from hospital data, in Vietnam has been claimed subsequent to Agent Orange sprayings during the war. The results of case-control studies, both published[33,34] or unpublished, suggest an association between maternal exposure and GTD. Between 1961 and 1971, the south of Vietnam was sprayed with large amounts of Agent Orange, a mixture of phenoxyherbicides (2,4,5-T and 2,4-D) contaminated with dioxin, corresponding to a contamination of approximately 170 kg of dioxin. Experimental studies with several animal species have revealed that after maternal exposure there are teratogenic and foetotoxic effects of 2,3,7,8-tetrachlorodibenzo-p-dioxin (2,3,7,8-TCDD). Chlorinated dioxins are ubiquitous in the environments of industrialized countries. Agent Orange is the main source of dioxin contamination in the South Vietnamese. During the past decade, several studies have shown that dioxin levels in human tissues from heavily Agent Orange-sprayed areas in the south of Vietnam were at least as high as in industrialized countries, and these levels contrast with the minimal levels found in unsprayed areas in the north of Vietnam. These high levels of dioxin may affect the reproductive system. The present case-control study was undertaken in Vietnam to assess the influence of known and suspected risk factors for GTD, especially the reproductive effects of environmental exposure to phenoxyherbicides and their contaminants.
This case-control study was conducted in 1990 at the Obstetrical and Gynecological Hospital in Ho Chi Minh City; this is the main referral maternity hospital for the south of Vietnam. All cases with a new pathologic diagnosis of GTD who were admitted between March and May, 1990, were identified. All eligible cases had to meet the criteria of a histologically confirmed diagnosis of GTD. Patients with a diagnosis of partial hydatidiform mole were excluded (n = 7). All but l of the eligible cases admitted during the 2-mo recruitment period were included. Eighty-seven cases were finally included in our study, 71 of whom had complete moles and 16 of whom had choriocarcinomas.
For each case, we selected one control among women who were admitted during the same period for gynecological and obstetrical surgery (other than GTD). Only married women were selected because we wanted to exclude those who did not have the opportunity of being pregnant. Surgical controls were chosen because adipose tissues could be collected for biological analysis of dioxin (note: this analysis is in progress). In some cases, it was impossible to find adequate controls during the planned period; therefore, some controls were chosen a posteriori among women admitted for nonsurgical reasons (i.e., normal delivery). Controls were matched with cases for age ([+ or -] 5 y), and given the referral nature of the hospital, controls were also matched into two strata for place of residence: (1) Ho Chi Minh City and (2) outside Ho Chi Minh City. Eighty-seven control women were included. The reasons for their admissions were benign neoplasm of the ovary (n= 28), benign neoplasm of the uterus (n = 20), ectopic pregnancy (n = 3), caesarean section (n = 23), normal delivery (n = 3), nonmolar spontaneous abortion (n = 3), and other (n = 4).
Questionnaire. Patients were interviewed personally at the hospital in a separate office by the same physician. The physician was "blind" to case or control status and was not the physician who recruited the patients into the study. Patients who were interviewed with a standard questionnaire were questioned about sociodemographic characteristics, residence history from birth (including the names of provinces, districts, subdistricts, and villages, with the start and end dates of each period of residence); lifelong occupational history; medical, gynecological, and obstetrical histories; living children and their health status; duration of breastfeeding; possible herbicide exposure (i.e., during wartime and in agricultural practice); selected socio-economic indicators, such as the ownership of expensive consumer goods; smoking habits; and alcohol drinking habits. We assessed dietary habits semi-quantitatively to provide some indications on socio-economic status (meat consumption). Exposure to dioxins via food - mainly through fish consumption - was also determined. The women were also asked several questions about their husbands (e.g., dates and place of birth, occupational history, herbicide exposure during wartime, consanguinity). Data on clinical and pathological diagnoses and other characteristics (e.g., blood group) were obtained from medical records.
Estimation of cumulative exposure index to Agent Orange. The Vietnamese population could have been exposed to herbicides, either directly or indirectly via food, after sprayings occurred in their residential areas. For each subject, we calculated an exposure index to Agent Orange, using the whole residential history in accordance with a procedure described elsewhere.[39,40] Residential coordinates were determined with a Universal Transversal Mercator map (1:250 000). Residential coordinates were codified twice by two independent persons. When the name of the village was not found on the map, we recorded the corresponding subdistrict or district coordinates. The lists of sprayings were compiled from these coordinates and from the HERBS Tape for the area with a 10-km radius from each place of residence. The HERBS Tape contained data about the U.S. Air Force missions, which were conducted from August 1965 to February 1971, as well as an additional record of the spraying missions of other U.S. Army Corps stationed in Vietnam. With respect to each spraying, the HERBS tape contains the date, flight coordinates, and types and amounts of herbicides sprayed. We calculated exposures by taking into account the types of herbicides Used, the exposure intensity (i.e., amount of herbicide sprayed and proximity of spraying to place of residence), and duration of residence in a sprayed area. For each subject, we summed the exposure values that corresponded to her different places of residence.
Statistical analysis. Two analyses were performed. First, all cases were compared with their controls. Second, choriocarcinomas were excluded, and only cases of CHMs were compared with their respective controls. However, no important differences were observed between the results of both analyses. The results of the aggregate analysis of all cases and controls are presented here.
Potential confounding by matching variables was accounted for by stratified analysis of age (in two strata: < 35 y and [greater than or equal to] 35 y) and area of residence, strata being sufficiently wide not to require individually matched analysis. For all risk factors under study, we estimated adjusted odds ratios (ORs) and 95% confidence intervals (95% CIs) with unconditional logistic regression (BMDP software), including matching variables and potential confounders and, when appropriate, testing for trend. A multivariate logistic regression was performed, and it included matching variables and all the variables found to be significant in the univariate comparisons.
There were no statistically significant differences between cases and controls with respect to matching variables (i.e., age and place of residence), distribution of residence area between town and countryside, and occupational status (Table 1). The occupational status of the women was estimated to be occupation at the time of the interview or the last job held.
[TABULAR DATA OMITTED]
All cases included in the study had GTD for the first time. Choriocarcinoma cases were always preceded by a HM, which was not accounted for in the number of previous pregnancies. Analysis of gynecological and reproductive factors revealed no significant differences between cases and controls for age at menarche, menstrual irregularity, age at first pregnancy, duration of breast-feeding per live birth, and number of stillbirths or spontaneous abortions (Table 2). The mean number of previous pregnancies and previous live births was higher among cases (2.8 [+ or -] 0.3 versus 2.3 [+ or -] 0.3, respectively; 2.3 [+ or -] 0.3 versus 1.6 [+ or -] 0.2, respectively). The risk was increased as the number of previous pregnancies and previous live births increased. The trend in risk was significant (p = .01) for previous live births. Previous induced abortions occurred significantly less frequently among cases than controls (8.0% versus 20.7%, respectively). Given the limited number of subjects, we were unable to evaluate the role of oral contraceptives (i.e., 1 case, 3 controls) and intrauterine devices (i.e., 3 cases, 7 controls).
[TABULAR DATA OMITTED]
The comparisons of some selected nutritional variables showed that the percentage of cases who consumed at least five meat dishes per week was less than that of controls (OR = 0.4, 95% CI = 0.2-0.9). Although no significant associations were observed, cases had fewer daily meals and fruit dishes than controls, but cases ate more fish dishes.
Relative risk was evaluated relative to number of markers of socioeconomic status. Family income per person per year (in local currency) was analyzed in two classes: (1) below the control median and (2) above the control median. Incomes of cases were more often below the control median, and the difference was borderline statistically significant (OR = 0.6; 95% CI = 0.3-1.0). Although no statistically significant differences were observed, owners of some expensive consumer goods (e.g., television, refrigerator, motorbike) were less numerous globally among cases than controls.
The number of women who had participated in agricultural activities during their lifetimes or who had used pesticides did not differ significantly between cases and controls; as well, the mean number of years spent in agricultural activity did not differ significantly between the groups. in the subgroup of women who used pesticides in agricultural practice, cases and controls did not differ in number or in mean number of pesticide applications per year. However, cases spent more years in agricultural activities, but this finding was not significant. We were unable to obtain precise information about the type of pesticides used. When agricultural workers were compared with nonagricultural workers according to type of crops cultivated, we found no significant differences. Animal husbandry was compared between the two groups. The percentage of women who bred at least three pigs was higher among cases than controls (11.6% versus 2.3%, respectively; OR = 5.7, 95% Cl = 1.2-27.6). The risk increased as the number of pigs bred increased (p for trend = .04). No significant differences were observed between the two groups with respect to other animals (i.e., buffalo, oxen, cattle, and poultry).
Generally, the women were very young at the time of the sprayings; therefore, many were unable to remember whether they had witnessed sprayings during the war. This explains the large number of missing values in the answers to direct questions about sprayings. The Agent Orange cumulative exposure index ranged from 0 to 62 000. Four classes were constituted, and the reference class was formed by women who had an exposure index of [less than] 1. With respect to all classes, Ors did not differ significantly from l (OR = 0.7, 95% Cl = 0.2-1.8 [highest group]). Exposure in utero of the subject was of special interest and was assessed by potential exposure at the place of birth at the time the woman was born. The percentage of women potentially exposed in utero did not differ between cases and controls (35.6% versus 31.0%, respectively).
Husbands were seldom present at the interviews ([greater than] 22%) of cases and controls. Answers to questions about husbands were given most often by the spouse, a fact that explains the large number of missing data, particularly about occupational history and exposure to herbicides. The average age, as well as occupational characteristics of the male partner, did not differ between cases and controls.
The blood-group distribution did not differ between cases and controls, but the number of missing values was high (19.5%). Only one control reported having a consanguineous relationship with her husband (i.e., cousin).
Regular consumption of tobacco or alcohol was seldom reported. Less than 4% of the subjects admitted smoking (2 cases, 4 controls), and less than 6% drank alcohol (3 cases, 6 controls). Whenever the women reported regular cigarette smoking or drinking of alcohol, the quantities were very small.
The number of previous live births, history of induced abortion, and number of meat dishes consumed per week might be interrelated. Confounders among the variables previously found to be significant were examined, using multivariate logistic regression. Having had more than five live births was associated with a relative risk of 8.4 (95% Cl = 2.3-31.7 [adjusted for area of residence, age, history of induced abortion, number of meat dishes, and number of pigs bred]). A history of induced abortion was associated with an adjusted risk of 0.3 (95% Cl = 0.1-0.8). Having had five or more meat dishes per week was associated with an adjusted risk of 0.4 (95% Cl = 0.1-0.9).
Our results did not support any association between the index of exposure to Agent Orange and GTD. This absence of association could have resulted from our choice of control group, particularly the inclusion of controls with diseases that could have been related to dioxin exposure. Nonetheless, no evidence of a relationship between benign gynecologic tumors and herbicides was found in animal and human studies, although an increased risk of ovarian mesotheliomas was reported with herbicide exposure. Thus, it is unlikely that the inclusion of benign gynecologic tumors in the control group accounts for the similarity between cases and controls with respect to the Agent Orange exposure index. To avoid selection bias, we selected controls from the same recruitment area as cases, controls also had to have a disease, like GTD, to be transferred to a specialized center.
Cumulative exposure to Agent Orange was estimated on the basis of residence history, from the beginning of the sprays to the time of study. The model for calculating the cumulative exposure index we used is described by Verger et al. For 27 male subjects who resided in the south of Vietnam, Verger et al. showed a significant correlation between levels of 2,3,7,8-TCDD measured in the subjects, fat tissues and the exposure index. In a future report, dioxin levels in fat tissues will be compared between cases and controls of the present study, and a measure of correlation will be performed between dioxin isomers and the exposure index. The model for calculating the exposure index had some limitations. For example, we sometimes failed to identify a village on the map because its name had been changed, and this index did not account for dietary exposure via foods produced in contaminated areas and transported for sale in areas free of Agent Orange (e.g., Ho Chi Minh City). However, our approach was somewhere between the more accurate dioxin analysis of subjects'tissues and the more simple index that was based on current residence or on the answers to direct auestfons about contact with sprays, the answers for whick may have been difficult to recall. We compared the two types of exposure measurements. Only approximately 25% (average) of the women who had an exposure index [greater than or equal to] 1 000 indicated that they had witnessed tde sprayings or falling vegetation during wartime or thit they had ever inhabited the sprayed areas (i.e., sensivity of womens report). The percentage was higher Or cases (41.2%) than for controls (14.8%). This difference might have led to an overestimation of the risk if the women's answers had been used as a measure of exposure.
Interviews were conducted by the same person, and that person was "blind" control status. In only a few cases was the interviewer able to guess the status of the subject (e.g., some cases of caesarean sections). However, bias appears unlikely for answers about most of the variables, especially history of residence.
Another reason for the absence of an association between exposure to Agent Orange and the risk of GTD could be the lack of statistical power of our study. For our sample, we estimate a probability of 70% ([alpha] = .05) to detect a twofold relative risk of GTD relative to an exposure index [greater than or equal to] 1 000.
We review additional risk factors, and we found that cases had more previous pregnancies and histories of fewer interpreted abortions than controls. Our results, which contradicted those of some previous studies, [26-28] should be irterpreted with caution because of the manner in which we recruited control subjects. Our controls might have included women whose gynecological, obstetrical or hormonal diseases prevented them from having any future pregnancies. After iterative exclusion of one control group and its corresponding cases, we noted that the significant association between high parity and GTD persisted.
The results of some studies suggest that social class distribution may also play a role in this disease, at least in some high-incidence areas. In our study, cases had a lower family income, owned fewer consumer goods, ate fewer meat dishes each week, consumed more fish dishes per week, and bred more pigs than did controls. These factors might indicate a lower socioeconomic status and might also be related to higher dioxin intake via greater fish consumption. Breeding domestic pigs occurs frequently among the Vietnamese who live throughout the countryside. Contact with infected pigs may be related to a higher prevalence of some parasites among humans. A multivariate analysis was performed, and we found an independent effect of the frequency of meat consumption and number of live births with respect to the risk of GTD.
The present results should be explored further, with a focus on low socioeconomic status and chronic infections via contact with animals, both of which may be associated with poor health and mild immunodeficiency. These factors may allow for the development of a pathologic conceptus, which, in other circumstances, would be eliminated rapidly.
The authors thank all the technical staff from the 10-80 Committee and the medical staff from the Obstetrical and Gynecological Hospital, all of whom made this study possible. Special thanks are expressed to Mrs. Lan (midwife) and to Dr. F. Lert and ML Nguyen Anh Tuan.
This study was supported by a grant from INSERM (Reseau NordSud 490 NS2).
Submitted for publication December 22, 1994; revised; accepted for publication April 24, 1996.
Requests for reprints should be sent to Dr. Sylvaine Cordier, Institut National de la Sante et de la Recherche Medicale, Inserm U.170, 94807 Villejuif Cedex, France.
[1.] Vassilakos P, Riotton G, Kajii T. Hydatidiform mole: two entities. Am J Obstet Gynecol 1977; 127:167-70. [2.] Kajii T, Ohama K. Androgenetic origin of hydatidiform mole. Nature 1977; 268:633-34. [3.] Jacobs PA, Hassold TJ, Matsuyama AM, Newlands IM. Chromosome constitution of gestational trophoblastic disease. Lancet 1978; 2:49. [4.] Wake N, Takagi N, Sasaki M. Androgenesis as a cause of hydatidiform mole. J Natl Cancer Inst 1978; 60:51-53. [5.] Jacobs PA, Hunt PA, Matsuura J, et al. Complete and partial hydatidiform mole in Hawaii: cytogenetics, morphology and epidemiology. Br J Obstet Gynaecol 1982; 89:258-66. [6.] Lawler SD, Fisher RA, Pickthall VJ, et al. Genetic studies on hydatidiform moles. I. The origin of partial moles. Cancer Genet Cytogenet 1982; 5:309-20. [7.] Jacobs PA, Szulman AE, Funkhouse J, et al. Human triploidy: relationship between parental origin of the additional haploid complement and the development of partial hydatidiform mole. Ann Hum Genet 1982; 46:223-31. [8.] Szulman AE, Surti U. Strict clinicopathological criteria in the diagnosis of partial hydatidiform mole: a plea renewed. Am J Obstet Gynecol 1985; 152:1107-08. [9.] Lawler SD, Fisher RA. Genetic studies in hydatidiform mole with clinical correlations. Placenta 1987; 8:77-88. [10.] Chen R-J, Huang S-C, Chow S-N, et al. Persistent gestational trophoblastic tumour with partial hydatidiform mole as the antecedent pregnancy. Br J Obstet Gynaecol 1994; 101:330-34. [11.] Bracken MB, Brinton LA, Hayashi K. Epidemiology of hydatidiform mole and choriocarcinoma. Epidemiol Rev 1984; 6:52-75. [12.] Nakano R, Sasaki Y, Yamoto M, Hata H. Trophoblastic disease: analysis of 342 patients. Gynecol Obstet Invest 1980; 11:237-42. [13.] Takeuchi S. Incidence of gestational trophoblastic disease by regional registration in Japan. Hum Reprod 1987; 2:729-34. [14.] Hayashi K, Bracken MB, Freeman DH, Hellenbrand K. Hydatidiform mole in the United States (1970-1977): a statistical and theoretical analysis. Am J Epidemiol 1982; 115:67-77. [15.] Bagshawe KD, Dent J, Webb J. Hydatidiform mole in England and Wales 1973-83. Lancet 1986; 88:673-77. [16.] Flam F, Lundstrom-Lindstedt V, Rutqvist LE. Incidence of gestational trophoblastic disease in Stockholm County, 1975-1988. Europ J Epidemiol 1992; 8:173-77. [17.] Martin PM. High frequency of hydatidiform mole in native Alaskans. Int J Gynaecol Obstet 1978; 15:395-96. [18.] Bagshawe KD, Lawler SD. Choriocarcinoma. In: Schottenfeld D, Fraumeni JF Jr. (Eds). Cancer Epidemilogy and Prevention. Philadelphia, PA: WB Saunders, 1982; pp 909-24. [19.] Tomoda Y, Kaseki S, Goto S, et al. Rh-D factor in trophoblastic tumors: a possible cause of the high incidence in Asia. Am J Obstet Gynecol 1981; 139:742-43. [20.] Couillin P, Ravise N, Affoutou JM, et al. HLA et grossesses molaires (triploidies, moles hydatiformes, choriocarcinoma). Ann Genet 1987; 30:197-208. [21.] Matsuura J, Chiu D, Jacobs PA, Szulman AE. Complete hydatidiform mole in Hawaii: an epidemiological study. Genet Epidemiol 1984; 1:271-84. [22.] La Vecchia C, Parazzini F, Decarli A, et al. Age of parents and risk of gestational trophoblastic disease. J Natl Cancer Inst 1984; 73: 639-42. [23.] Parazzini F, La Vecchia C, Pampallona S. Parental age and risk of complete and partial hydatidiform mole. Br J Obstet Gynaecol [24.] Graham IH, Fajardo AM, Richards RL. Epidemiologic study of complete and partial hydatidiform mole in Abu Dhabi: influence of maternal age and ethnic group. J Clin Pathol 1990; 43:661-64. [25.] Messerli ML, Lilienfeld AM, Parmley T, et al. Risk factors for gestational trophoblastic neoplasia. Am J Obstet Gynecol 1985; 153:294-300. [26.] Brinton LA, Wu B-Z, Wang W, et al. Gestational trophoblastic disease: a case-control study from the People's Republic of China. Am J Obstet Gynecol 1989; 161:121-27. [27.] Parazzini F, La Vecchia C, Pampallona S, Franceschi S. Reproductive patterns and the risk of gestational trophoblastic disease. Am J Obstet Gynecol 1985; 152:866-70. [28.] Parazzini F, Mangili G, La Vecchia C, et al. Risk factors for gestational trophoblastic disease: a separate analysis of complete and partial hydatidiform moles. Obstet Gynecol 1991; 78:1039-45. [29.] Acaia B, Parazzini F, La Vecchia C, et al. increased frequency of complete hydatidiform mole in women with repeated abortion. Gynecol Oncol 1988; 31:310-14. [30.] La Vecchia C, Franceschi P, Parazzini F, et al. Risk factors for gestational trophoblastic disease in Italy. Am J Epidemiol 1985; 121:457-64. [31.] Berkowitz RS, Cramer DW, Bernstein MR, et al. Risk factors for complete molar pregnancy from a case-control study. Am J Obstet Gynecol 1985; 152:1016-20. [32.] Huong LD, Phuong NTN, Thuy TT, et al. An estimate of the incidence of birth defects, hydatidiform mole and fetal death in utero between 1952 and 1985 at the obstetrical and gynecological hospital of Ho Chi Minh City, Republic of Vietnam. Chemosphere 1989; 18:805-10. [33.] Phuong NTN, Thuy TT, Phuong PK. An estimate of differences among women giving birth to deformed babies and among those with hydatidiform mole seen at the Ob@gyn Hospital of Ho Chi Minh City in the south of Vietnam. Chemosphere 1989; 18:801-03. [34.] Phuong NTN, Thuy TT, Phuong PK. An estimate of reproductive abnormalities in women inhabiting herbicide sprayed and non-herbicide sprayed areas in the south of Vietnam, Chemosphere 1989; 18:843-46. [35.] Constable JD, Hatch MC. Reproductive effects of herbicide exposure in Vietnam: recent studies by the Vietnamese and others. Teratogenesis Carcin Mutagen 1985; 85:516-22. [36.] Westing AH. Herbicides in war: past and present. In: Westing AH (Ed). Herbicides in War, the Long-Term Ecological and Human Consequences. London, UK and Philadelphia, PA: Taylor and Francis, 1984; pp 3-24. [37.] Friedman JM. Does Agent Orange cause birth defects? Teratology 1984; 29:193-221. [38.] Schecter A, Dai LC, Thuy LTB, et al. Agent Orange and the Vietnamese: the persistence of elevated dioxin levels in human tissues. Am J Public Health 1995; 85:516-22. [39.] Verger P, Cordier S, Thuy LTB, et al. Correlation between dioxin levels in adipose tissue and estimated exposure to agent orange in South Vietnamese residents. Environ Res 1994; 65:226-42. [40.] Stellman SD, Stellman JM. Estimation of exposure to Agent Orange and other defoliants among American troops in Vietnam: a methodological approach. Am J Ind Med 1986; 9:305-21. [41.] Smith C, Watkins D (Eds). The Vietnam Map Book: A Self-Guide to Herbicide Exposure. Berkeley, CA: Winter Soldier Archives, 1981. [42.] Rothman KJ, Modern Epidemiology. Boston, MA: Little, Brown, 1986; pp 237-83. [43.] World Health Organization, Environmental Health Criteria 88. Polychlorinated Dibenzo-para-dioxins and Dibenzofurans. Geneva, Switzerland: World Health Organization, 1989. [44.] Donna A, Betta PG, Robutti F, et al. Ovarian mesothelial tumors and herbicides: a case-control study. Carcinogenesis 1984; 5: 941-42. [45.] Svensson BG, Nilsson A, Hansson M, et al. Exposure to dioxins and dibenzofurans through the consumption of fish. N Engl J Med 1991; 324:8-12. [46.] Seuri M, Koskela P. Contact with pigs and cats associated with high prevalence of toxoplasma antibodies among farmers. Br J Ind Med 1992; 49:845-49.
COPYRIGHT 1996 Heldref Publications
COPYRIGHT 2004 Gale Group