NIOSHTIC-2 Publications Search

Re - total serum testosterone and gonadotropins in workers exposed to dioxin - reply.

Egeland-GM; Sweeney-MH; Fingerhut-MA; Wille-KK; Schnorr-TM; Halperin-WE
Am J Epidemiol 1995 Mar; 141(5):477-478
The conclusion of Egeland et al. that "current and half-life serum dioxin levels were positively and significantly related to luteinizing hormone and follicle-stimulating hormone and inversely related to testosterone" (1, p. 277) is not supported by the data they present. The paper reports results of a cross-sectional analysis of workers previously exposed to dioxin at two manufacturing sites. Egeland et al. pooled data from the two sites without providing information on homogeneity, and they do not inform readers about what proportion of the workers were from either site or how the groups compared with respect to dioxin exposure or other potentially relevant factors. The authors claim differences from referents in mean gonadotropin and testosterone levels, but these differences are within what they define as normal limits. These changes are reflected in excesses of individual estimates falling above the norm for luteinizing hormone (eight individual readings) and below the norm for testosterone (14 individuals). Since the authors tell us nothing about pathology that could be associated with these levels, we have no way of assessing their importance. Presumably, such hormonal abnormalities have been chronic, so dysfunction could easily have developed, but no data from medical examinations are presented. Details of hormone assays are not presented, except for the statement that they were "standard double antibody radioimmunoassay procedures" (1, p. 274). I know of no organization that standardizes such protocols. Information normally presented for endocrine assays is lacking - e.g., there is no information on the source of antibodies, conditions for storing samples, assay-to-assay consistency, senitivity, specificity, etc. No information is presented on the manner in which the normal limits were obtained. No assays appear to have been replicated, including those for the critical subjects outside of the normal limits. If the authors used commercial kits for the assays, they should so state. As was recently pointed out by Paul McDonough, an editor for Fertility and Sterility, kits may be subject to errors in hormone quantification as a result of contamination by other serum constituents (2). Dioxin assays are better defined by Egeland et aI., but again no replication is indicated. Analyses are presented by current values and estimates calculated back to last expose, 15-37 years earlier, using a half-life estimate for dioxin. There appear to have been no overall differences in hormone levels between exposed subjects and referents. When Egeland et al. divided dioxin levels into quartiles for comparison, the adjusted mean follicle-stimulating hormone values for the three higher quartiles were higher than those for the referents, but the standard errors broadly overlapped; the odds ratios for subjects above the norm were higher, but with 95 percent confidence intervals encompassing 1 in every case. The test for trend was barely significant at the 10 percent level. For luteinizing hormone, the standard error of the mean levels for the highest dioxin quartile did not overlap with that for the referents, although the standard errors did overlap for all of the other quartiles. Odds ratios encompassed 1 in all cases; the p value for the trend was 0.03. Standard errors for the mean testosterone estimates overlapped in all cases, while the 95 percent confidence interval for the odds ratio excluded 1 only for the next-to-lowest quartile. The p value for the trend was again 0.1. When the quartiles were based on serum dioxin levels extrapolated from a half-life estimate to the time at which exposure ceased, all odds ratios encompassed 1 except that for testosterone, but this time at the third quartile; the p value for the luteinizing hormone trend was 0.04, while significance was lost for follicle-stimulating hormone and testosterone, both at p = 0.12. Thus, the data are too inconsistent to support the authors' assertion that differences in hormone titers are related to occupational dioxin exposure; most critically, the authors present no evidence for reproductive dysfunction in these individuals. On a positive note, hormonal changes secondary to exposure to current environmental levels of dioxin would appear unlikely, since the half-life extrapolated serum dioxin levels in the exposed workers were about 140- fold higher than those in referents at one site and over 400 times higher at the other (3).
Biological-effects; Biological-systems; Cell-damage; Cellular-reactions; Chemical-hypersensitivity; Chemical-reactions; Exposure-assessment; Exposure-levels; Exposure-methods; Hormones; Occupational-exposure; Reproductive-effects; Reproductive-hazards; Reproductive-system; Reproductive-system-disorders; Risk-analysis; Statistical-analysis; Toxic-effects; Work-environment; Worker-health; Workplace-studies
National Institute for Occupational Safety and Health, Centers for Disease Control and Prevention, DSHEFS, 4676 Columbia Parkway, Cincinnati, OH 45226
Publication Date
Document Type
Journal Article
Fiscal Year
Issue of Publication
NIOSH Division
Source Name
American Journal of Epidemiology
Page last reviewed: September 2, 2020
Content source: National Institute for Occupational Safety and Health Education and Information Division