Consumption of PCB-contaminated Freshwater Fish ...

2 downloads 0 Views 455KB Size Report
Sep 9, 1997 - Highly contaminated Lake Ontario sport fish represent an important human dietary exposure to polychlo- ..... metabolites. Role of hepatic mono-oxygenase in methoxy- ... Committee on the assessment of human health effect of ...
American Journal of Epidemiology Copyright O 1997 by The Johns Hopkins University School of Hygiene and Public Health All rights reserved

Vol. 146, No. 11

Printed in U.S.A

Consumption of PCB-contaminated Freshwater Fish and Shortened Menstrual Cycle Length

Pauline Mendola,1 Germaine M. Buck,1 Lowell E. Sever,2 Maria Zielezny,1 and John E. Vena1 Highly contaminated Lake Ontario sport fish represent an important human dietary exposure to polychlorinated biphenyls (PCBs) and other toxic contaminants that may disrupt endocrine pathways. New York State Angler Cohort women interviewed by telephone in 1993 provided menstrual cycle length (n = 2,223). Fish consumption at cohort enrollment in 1991 was categorized by duration and frequency and was used to calculate a PCB exposure index. Multiple regression analyses identified significant cycle length reductions with consumption of more than one fish meal per month (1.11 days) and moderate/high estimated PCB index (-1.03 days). Women who consumed contaminated fish for 7 years or more also had shorter cycles (-0.63 days). Am J Epidemiol 1997;146:955-60. environmental pollutants; menstruation; polychlorinated biphenyls; women

This secondary data analysis assessed the potential relation between menstrual cycle length and consumption of PCB-contaminated Lake Ontario sport fish in a cohort of New York State anglers. MATERIALS AND METHODS

Data were gathered as part of the New York State Angler Cohort Study, a population-based cohort of angler families from 16 counties in New York that surround Lake Ontario (28). In brief, the cohort was established in 1991 when mailed surveys were completed by 11,431 randomly selected licensed anglers aged 18-40 years (10,518 males and 913 females) and 6,651 spouses of male anglers (overall, 39 percent response). Species-specific fish consumption, medical and reproductive histories, sociodemographic characteristics, and other lifestyle behaviors were assessed. Telephone interviews with a random sample of 100 nonrespondents found sociodemographic differences but no differences with regard to fishing or fish consumption (28). Menstrual cycle data were collected from a subgroup of women who completed a telephone interview to update their reproductive history in 1993. Women who stated in 1991 that they were considering becoming pregnant in the following 3 years (n = 2,977, 40 percent of cohort women) were eligible for inclusion in the telephone follow-up study. Trained nurses successfully completed 2,445 interviews in 1993 (82 percent). The interview was designed to look crosssectionally at time to pregnancy and to assess reproductive health status. Self-reported menstrual cycle

Received for publication January 17, 1997, and in final form September 9, 1997. Abbreviation: PCB, polychlorinated blphenyi. 1 Department of Social and Preventive Medicine, State University of New York at Buffalo, Buffalo, NY. 2 Battelle Centers for Public Health Research and Evaluation, Seattle, WA. Reprint requests to Dr. Pauline Mendola, Department of Social and Preventive Medicine, State University of New York at Buffalo, 270 Farber Hall, Buffalo, NY 14214.

955

Downloaded from aje.oxfordjournals.org by guest on July 13, 2011

Menstrual cycle length and regularity are influenced by many factors, including age (1-3), contraception and smoking (4), and body size and exercise (5, 6). Menstrual function is hormonally mediated, and several environmental contaminants, including polychlorinated biphenyls (PCBs), are reported to disrupt endocrine pathways (7-13). A comprehensive review of the epidemiology of menstruation cited the need for research on environmental risk factors, particularly those with suspected hormonal effects (14). Consumers of Great Lakes sport fish are exposed to a complex mixture of chemicals, including persistent lipophilic compounds such as PCBs, dichlorodiphenyl dichloroethene, dioxin, other pesticides, and mercury (15-25). Sport fish consumption has been estimated to deliver a dose of PCBs that is 4,300 times higher than background drinking water or inhalation exposure (20). Lake Ontario sport fish have reported PCB concentrations that are more than twice as high as those in fish from the other Great Lakes (26). The New York State Department of Health recommends that women of childbearing age eat no fish from Lake Ontario because of the high levels of contaminants (27).

956

Mendola et a).

ing were forced into the model before exposure indices because of their a priori relation to menstrual cycle length. Exposure categories were entered as dummy variables. The results for each exposure measure (number of years, frequency of meals, and the PCB index) are presented for all women and for the group restricted to those with regular cycles. RESULTS

Characteristics of the women are shown in table 1. Women with regular cycles, the majority of all women in the study, are similar to all study women except for menstrual cycle length. Women who reported a regular cycle had a shorter average cycle with less variation than did all women combined. Given the large range of cycle lengths reported by women as regular, presumably, irregular cycles are based on inability to predict the onset of menses as well as cycle length. The mean age of the women is 31.2 years; approximately 50 percent reported ever having smoked. Despite the health advisory, 12 percent of the women eat Lake Ontario fish meals more than once a month; 13 to 14 percent have been eating these fish for 7 years or more. No significant relation was seen between age and menstrual cycle length in preliminary analyses, and smoking was of borderline significance (data not shown). Stratified analyses of women aged 35 years and older had less statistical power, but the results were of similar magnitude and direction as those for the full study group (data not shown). Results of the adjusted multiple regression analyses are shown in table 2. The pattern of results is the same for both groups of women, but the magnitude of the association is stronger in the analyses based on all women, largely because of the amount of variability in cycle length. Cycles appear shorter with more than 2 years of fish consumption among all women, but this is not statistically significant. Eating sport fish meals from Lake Ontario more than once a month in 1991 was associated with a statistically significant reduction in cycle length of greater than 1 day for the group that includes all women (—1.11 days, 95 percent confidence interval —1.87 to —0.35) and about half a day for women who reported regular cycles (—0.51 day; 95 percent confidence interval -0.92 to -0.10). On average, women in the PCB index moderate-to-high exposure group were observed to have cycles about 1 day shorter (—1.03 days; 95 percent confidence interval -1.88 to — 0.19), and about half that reduction was seen in the regular cycle group (-0.56 day; 95 percent confidence interval —1.01 to -0.09). These analyses were repeated, eliminating 5 percent of women at each extreme of menstrual cycle length to Am J Epidemiol

Vol. 146, No. 11, 1997

Downloaded from aje.oxfordjournals.org by guest on July 13, 2011

length, asked as "usual number of days," was analyzed for all women with complete data (n = 2,223; 91 percent of those interviewed) and also was restricted to women who reported "regular" cycles only (n = 2,080, 94 percent of those with cycle length information). The average number of days, rounded to the nearest whole number, was used for women who reported a range of days. For the most part, women defined their own cycles as regular or irregular. The only exceptions were women who reported "irregular" cycles that occurred approximately once a month with a small range of usual cycle length. After further questioning by the nurses, these women were considered to have regular cycles. Exposure measures used here are based on the mailed survey in 1991 (cohort enrollment) and reflect consumption of contaminated Lake Ontario fish only. Fish consumption was measured in terms of duration (total number of years eating fish) and monthly frequency of sport fish meals in 1991 for 12 species and size categories of fish. Since contaminant levels vary considerably by species and size of fish, a crude index was developed to estimate the potential Lifetime PCB exposure through sport fish consumption: PCB index = years X 2 species (frequency X portion size X PCB concentration), where years = total number of years eating Lake Ontario fish, 1955 to 1991; species = 0 (not eaten) or 1 (eaten) for each of 12 species/size measured; frequency = number of species-specific fish meals in 1991; portion size = usual grams per meal; and PCB concentration = milligrams of total PCBs (29, 30). Women who never ate Lake Ontario sport fish comprise the referent group. Women who ate these fish in the past but not during 1991 are characterized as "past consumers" because usual meal frequency information is not available for them. Past consumers are analyzed as a separate exposure category for frequency of meals and the PCB index. Categories for the number of years, frequency of meals, and PCB index are based on tertiles of exposure for all cohort women. The moderate and high PCB index groups are combined in this analysis. Smoking was assessed as "ever" versus "never" as reported upon cohort enrollment. Data on current smoking were not available in 1991 or 1993. Age was calculated from the woman's birth date given during the 1993 telephone interview. Multiple regression analysis was conducted with usual number of days per cycle as the dependent variable. The unit change (in days) at each exposure level is expressed, as well as the 95 percent confidence limits based on the standard error. Women who never ate sport fish were the referent group. Age and smok-

PCB-contaminated Fish and Shortened Menstrual Cycle

957

TABLE 1 . Menstrual cycle length, age, smoking, Lax* Ontario sport flsh consumption, and related PCB* exposure characteristics, New York State Angler Cohort, established in 1991 and reinterviewed in 1993 Characteristic

Mean

Al womenf (n = 2,223) (SO*)

Regular cyde3 ortyt (n - 2,080) Range

Mean

(SO)

Range

Menstrual cycle length (days)

28.69

(5.8)

14-105

27.99

(3.0)

14-60

Age (years)

31.20

(4.2)

20-47

31.20

(4.2)

20-47

%

No.

%

1,120

50.4

1,050

50.5

1,234

56

1,156

56

368 322 299

17 15 14

341 305 276

16 15 13

Frequency of mealsH Past consumers £1 per month >1 per month

412 303 275

19 14 12

381 287 256

18 14 12

PCB index* Past consumers Low (51 mg) Moderate/nigh (>1 mg)

412 365 212

19 16 10

381 348 195

18 17 9

Ever smoked Sport fish exposures Not exposed No. of years§ 1-2 3-6

* PCB, polychlorinated biphenyl; SO, standard deviation. t All women with cyde length information and no other missing information. $ Women with cyde length information and no other missing information who reported having 'regular" cydes. § Number of years during which Lake Ontario fish was eaten. | Frequency of Lake Ontario fish meals in 1991; past consumers have unknown frequency. # PCB index is based on a cumulative spedes-spedfic model; past consumers have unknown levels. TABLE 2. Menstrual cycle length (days), Lake Ontario sport fish consumption, and related PCB* exposure characteristics, multiple regression analysis, New York State Angler Cohort, 1991-1993

floh iRMl

exposures

Regular cyctes only} (n = 2,080)

Al womenf (n = 2,223)

Sport

Cycle days

Not exposed

96%CI»5

Cycle days

Referent

85%CI§ Referent

No. of yearsU 1-2 3-6 Z.7

Frequency of meals# Past consumers 51 per month >1 per month PCB index** Past consumers Low (51 mg) ModerateJhigh (>1 mg)

-0.13 -0.65 -0.63

-0.80 to 0.54 -1.36 to 0.62 -1.36 to 0.10

-0.13 -0.26 -0.22

-0.49 to 0.23 -0.64 to 0.12 -0.62 to 0.18

-0.03 -0.41 -1.11

- 0 . 6 8 to 0.62 -1.14 to 0.3 -1.87 to-0.35

0.10 -0.32 -0.51

-0.25 to 0.45 -0.71 to 0.07 -0.92 to-0.10

-0.03 -0.57 -1.03

- 0 . 6 8 to 0.62 -1.25 to 0.1 -1.88 to-0.19

0.10 -0.34 -0.56

-0.25 to 0.45 -0.70 to 0.02 -1.01 to-0.09

* PCB, polychtorinated biphenyl; Cl, confidence interval. t AD women with cycle length information and no other missing information. $ Women with cycle length information and no other missing information who reported having 'regular* cydes. § Adjusted for age and ever smoked, change in the number of cycle days at each exposure level, and the 95% confidence intervals based on the standard error. H Number of years during which Lake Ontario fish was eaten. # Frequency of Lake Ontario fish meals in 1991; past consumers have unknown frequency. ** PCB index is based on a cumulative species-specific model; past consumers have unknown levels.

Am J Epidemiol

Vol. 146, No. 11, 1997

Downloaded from aje.oxfordjournals.org by guest on July 13, 2011

No.

958

Mendola et al.

remarkably similar, given the differences in sample size.

assess the potential effect of outliers on the results. Cycle length ranged from 22 to 34 days for the remaining women. The magnitude of the reductions was smaller, but the pattern of results was similar (data not shown). As in the total group, the strongest finding was with greater than one fish meal per month (—0.28 days; 95 percent confidence interval —0.53 to —0.03). In addition, the authors compared women who reported regular cycles versus those with irregular cycles to determine whether having an irregular cycle itself was related to exposure (table 3). Women who characterized their cycles as either regular or irregular were included in these categorical analyses, whether or not they provided data on cycle length. Usual cycle length data were available for about three quarters of the women with irregular cycles. Women with irregular cycles had more variability and significantly longer cycles than did women with regular cycles. Women with irregular cycles were about 7 months younger than women with regular cycles (p = 0.07). No statistically significant differences were seen with regard to smoking, and none of the exposure measures appeared to vary by cycle regularity. In fact, the proportions of women at each exposure level was

DISCUSSION

TABLE 3. Select characteristics by regularity of manstrual cycle*, New York State Angler Cohort, 1991-1993 Irregular}

Regutart Characteristic

Moan

(SO§)

No.

Mean

(SD)

No.

Menstrual cycle length (days)*

27.99

(3.0)

2,190

38.48

(16.2)

153

Age (years)

31.25

(4.2)

2^21

30.69

(4.2)

211

No.

%

No.

%

1,066

50.4

102

51.0

1,186

56

112

56

356 315 283

17 15 13

39 24 27

19 12 13

Frequency of meals# Past consumers £1 per month >1 per month

392 293 261

18 14 12

40 27 22

20 13 11

PCB index** Past consumers Low(51 mg) Moderate/high (>1 mg)

392 355 199

18 17 9

40 28 21

20 14 11

Ever smoked Sport fish exposures Not exposed No. of yearsi] 1-2 3-6

* p< 0.001. t AD women with cycle length information and no other missing information. t Women with cycle length information and no other missing information who reported having "regular" cycles. § SO, standard deviation. U Number of years during which Lake Ontario fish was eaten. # Frequency of Lake Ontario fish meals in 1991; past consumers have unknown frequency. ** PCB index is based on a cumulative species-specific model; past consumers have unknown levels.

Am J Epidemiol

Vol. 146, No. 11, 1997

Downloaded from aje.oxfordjournals.org by guest on July 13, 2011

This study has several strengths, including a population-based design and the use of trained nurseinterviewers. Considerable information was obtained on specific fish consumption. Of particular importance is the fact that this cohort represents a group of women with low-level environmental exposure to contaminants of public health concern, avoiding the problems of extrapolating dose-effects from occupationally exposed individuals or animal models (31, 32). However, these preliminary findings need to be interpreted cautiously because of the inherent limitations of a secondary data analysis, including lack of information on potential confounders, e.g., current smoking, stress, use of contraceptives, body mass index, and physical exercise. The extent to which these unmeasured variables may be related to contaminated sport fish consumption is unknown. Both the exposure and the outcome are self-reported and subject to error, but exposure was ascertained in a 1991 mailed survey

PCB-contaminated Fish and Shortened Menstrual Cycle

ACKNOWLEDGMENTS

This study was funded in part by the Great Lakes Protection Fund (RM791-3021) and the Agency for Toxic Substance and Disease Registry (H75/ATH298338). The authors acknowledge the contributions of the New York State Angler Cohort Group: Dr. Edward Fitzgerald, New York State Department of Health; Drs. Hebe Greizerstein, Paul Kostyniak, and James Olson, Toxicology Research Center, State University of New York at Buffalo; and Dr. James McReynolds, Biophysics, State University of New York at Buffalo; as well as Mya Swanson for her computer programming assistance in calculating the PCB index.

REFERENCES 1. Treoloar AE, Boynton RE, Behn BG, et al. Variation of the human menstrual cycle throughout reproductive life. Int J Fertil 1967;12:77-126. Am J Epidemiol

Vol. 146, No. 11, 1997

2. Matsumoto S, Nogami Y, Ohkuri S. Statistical studies on menstruation: a criticism on the definition of normal menstruation. Gunma J Med Sci 1962;11:294-318. 3. Chiazze L, Brayer FT, Macisco JJ, et al. The length and variability of the human menstrual cycle. JAMA 1968;203: 89-92. 4. Brown S, Vessey M, Stratton I. The influence of contraception and cigarette smoking on menstrual patterns. Br J Obstet Gynecol 1988;95:905-10. 5. Cooper GS, Sandier DP, Whelan EA, et al. Association of physical and behavioral characteristics with menstrual cycle patterns in women aged 29-31 years. Epidemiology 1996;7: 624-8. 6. Cumming DC, Wheeler GD, Harber VJ. Physical activity, nutrition and reproduction. Ann N Y Acad Sci 1994;709: 55-76. 7. Nelson A. Effects of dichlorodiphenyltrichloroethane (DDT) analogs and polychlorinated biphenyl (PCB) mixtures on 17b[3H]estradiol binding to rat uterine receptor. Biochem Pharmacol 1974;23:447-51. 8. Korach K, Sarver P, Chae K, et al. Estrogen receptor-binding activity of polychlorinated hydroxybiphenyls: conformationally restricted structural probes. Mol Pharmacol 1987;33: 120-6. 9. Safe S, Astroff B, Harris M, et al. 2,3,7,8-Tetrachlorodibenop-dioxin (TCDD) and related compounds as antioestrogens: characterization and mechanism of action. Pharmacol Toxicol 1991;69:400-9. 10. Soontornchat S, Li MH, Cooke PS, et al. Toxicokinetic and toxicodynamic influences on endocrine disruption by polychlorinated biphenyls. Environ Health Perspect 1994; 102: 568-71. 11. Jansen HT, Cooke PS, Porceli J, et al. Estrogenic and antiestrogenic actions of PCBs in the female rat: in vitro and in vivo studies. Reprod Toxicol 1993;7:237-48. 12. Bitman J, Cecil H. Estrogenic activity of DDT analogs and polychlorinated biphenyls. J Agr Food Chem 1970;18: 1108-12. 13. Bulger WH, Muccitelli RM, Kupfer D. Studies on the in vivo and in vitro estrogenic activities of methoxychlor and its metabolites. Role of hepatic mono-oxygenase in methoxychlor activation. Biochem Pharmacol 1977,27:2417-23. 14. Harlow SD, Ephross SA. Epidemiology of menstruation and its relevance to women's health. Epidemiol Rev 1995; 17: 265-86. 15. Kimbrough RD. Laboratory and human studies on polychlorinated biphenyls (PCBs) and related compounds. Environ Health Perspect 1985;59:99-106. 16. Fiore BJ, Anderson HA, Hanrahan HP, et al. Sport fish consumption and body burden levels of chlorinated hydrocarbons: a study of Wisconsin anglers. Arch Environ Health 1989;44: 82-8. 17. International Joint Commission. Status report on organic and heavy metal contaminants in Lake Erie, Michigan, Huron and Superior basins. Windsor, Ontario, Canada, 1978. 18. International Joint Commission. Committee on the assessment of human health effect of Great Lakes water quality. Annual report Washington, DC: International Joint Commission, 1981. 19. Cordle F, Locke R, Springer J. Assessment of risk associated with human consumption of some species of fish contaminated with polychlorinated biphenyls (PCBs). Environ Health Perspect 1982;45:171-82. 20. Humphrey HEB. Population studies of PCBs in Michigan residents. In: D'ltri FM, Kamrin MA, eds. PCB's: human and environmental hazards. Ann Arbor, MI: Ann Arbor Science, 1983:299-310. 21. Humphrey HEB. The human population—an ultimate receptor for aquatic contaminants. Hydrobiologia 1987; 149:75-80. 22. Mohammed A. Polychlorinated biphenyl congeners in blood of Wisconsin sport fish consumers. Arch Toxicol 1990;64: 567-71.

Downloaded from aje.oxfordjournals.org by guest on July 13, 2011

and menstrual cycle characteristics were determined in a 1993 telephone interview, limiting the likelihood that concern over fish consumption was related to reporting shorter menstrual cycles. While the small decreases in menstrual cycle length observed are not likely to be clinically relevant or of major public health concern per se, they may indicate potential endocrine effects on a population level. The mechanism by which contaminated fish consumption and, presumably, exposure to PCBs alters menstrual function is unknown, but growing evidence suggests that PCBs may be estrogenic (7-10, 12, 13). Higher levels of estrogen have been associated with a shorter follicular phase and menstrual cycle (33). In conclusion, consumption of contaminated sport fish was associated with shorter menstrual cycles in this secondary data analysis. Frequency of consumption and an index of lifetime PCB exposure were observed to have a stronger relation with cycle length than duration (number of years of fish consumption). At the highest exposure levels, the effect size is approximately one-half day for women who reported regular cycles and 1 day for all women who report a usual cycle length. The effect does not appear to be mediated through irregular cycles since exposure levels are similar for women who reported regular or irregular cycles. These preliminary findings underscore the need for further study of environmental factors in association with hormonally mediated reproductive endpoints. Future studies should include menstrual calendars, full assessment of potential confounders, and measured body burdens of PCBs and other persistent contaminants to address this question adequately.

959

960

Mendola et al.

23. Smith J. PCB levels in human fluids: Sheboygan case study. Sheboygan, WI: University of Wisconsin Sea Grant Institute, 1983. 24. Sonzogni W, et al. Polychlorinated biphenyl congeners in blood of Wisconsin sport fish consumers. Arch Environ Contarn Toxicol 1991;20:56-6O. 25. Kuwabara K, Yakuskiji T, Watanable I, et al. Increase in the human blood PCB levels promptly following ingestion of fish containing PCBs. Bull Environ Contain Toxicol 1979;21: 273-8. 26. Clark JR, DeVault D, Bowden RJ, et al. Contaminant analysis of fillets from Great Lakes Coho salmon, 1980. Great Lakes Res 1984;10:38-47. 27. New York State Department of Health (NYSDOH). 1995-1996 health advisories: chemicals in sportfish and game. Albany, NY: New York State Department of Health Center for Environmental Health, 1995. 28. Vena JE, Buck GM, Kosytniak P, et al. The New York State Angler Cohort Study: exposure characterization and reproduc-

29. 30.

31. 32.

33.

tive and developmental health. Toxicol Indust Health 19%; 12:327-34. Cox C. Ministry of the Environment database. Toronto, Ontario, Canada: Ministry of the Environment Eater Resources Branch, 1988-1990. Skinner, L. New York State Department of Environmental Conservation (NYSDEQ database 1980-1990. Albany, NY: New York State Department of Environmental Conservation, 1996. Winneke G, Lilienthal H. Extrapolation from animals to humans: scientific and regulatory aspects. Toxicol Lett 1992; 64-65:239-46. Windham GC. Design of environmental epidemiologic studies of reproductive outcomes. Host factors in environmental epidemiology. Proceedings of the Conference and Workshop, Cracow, Poland, June 11-14, 1995. Landgren BM, Unden AL, Diezfalusy E. Hormonal profile of the cycle in 68 normal menstruating women. Acta Endocrinol (Copenh) 1980;94:89-98.

Downloaded from aje.oxfordjournals.org by guest on July 13, 2011

Am J Epidemiol

Vol. 146, No. 11, 1997