Authors:
Swan SH, Brazil C, Drobnis EZ, Liu F, Kruse RL, Hatch M, Redmon JB, Wang C, Overstreet JW, The Study For Future Families Research Group.

Title:
Geographic differences in semen quality of fertile U.S. males.

Source:
Environmental Health Perspectives. 111(4):414-20; 2003.

Summary:
An article published in 1992 by Carlsen et al. suggesting that semen quality was declining worldwide spawned numerous studies designed to measure semen quality in many countries. Carlsen suggested that the global decline in sperm quality may be associated with a concomitant increase in the use of man-made, estrogenic chemicals. Methodological and statistical questions regarding the Carlsen analysis of 61 papers on semen quality provided much controversy over this complex issue. Interlaboratory differences in semen collection, analysis, abstinence periods and subject selection made it difficult to assess global trends. As the Carlsen study depended on data obtained from many different countries over a 50 year period, it was suggested that geographical variations in semen quality may have contributed to the findings. Many subsequent studies have also detected geographical variation in semen quality, even between regions of an individual country.

To further investigate semen quality in the U.S., authors Swan et al. in collaboration with the Study for Future Families Research Group, recruited individuals from prenatal clinics in Los Angeles, California, Minneapolis, Minnesota, Columbia, Missouri, and New York City, New York. Recruitment was standardized such that subjects were male partners of a woman currently attending one of the prenatal clinics in the study locations. Exclusion criteria included medically assisted pregnancies and individuals less than 18 years of age. Laboratory technicians from each clinic attended a common training session prior to conducting semen analyses. Semen samples were provided at the clinic location following an advised period of 2-5 days of abstinence. Questionnaires were administered to the couples to ascertain race, education, smoking, body mass index (BMI), and fever in 3 months prior to onset of study, use of steroids, history of sexually transmitted diseases, cryptorchidism and other genital problems.

In total, data from 493 men was used for the statistical analyses: 176 from Missouri, 124 from California, 155 from Minnesota and 38 from New York. Race varied by center with only 23% 'white' subjects from California and 86% in Minnesota and Missouri. Subjects from California were less educated, while a greater number of smokers participated in the study from Missouri (13%).

Mean sperm concentration was lowest in Missouri (58.7 million/ml) and highest in New York City (102.9 million/ml). Semen volume and proportion of morphologically normal sperm did not vary much between centers. After multivariate modeling of the data in which race, age, smoking, fever, STD history and other parameters were incorporated into the model, mean sperm concentration was still lowest in Missouri and highest in New York. The proportion of motile sperm was similarly higher in New York compared to Missouri.

The study has several strengths including the prospective design of the study and attempts to standardize methodology and semen analysis between centers. However, the participation rate varied between centers with a relatively low number of subjects from the New York center (n=38) compared to the other three centers (Missouri n=176, California n=124, Minnesota n=155). As both California and Minnesota also had much higher sperm concentrations compared to Missouri, the geographical variation in sperm concentration is probably genuine. However, the variation may be related to differences in socioeconomic factors rather than environmental differences. Race varied significantly among centers with predominantly Hispanic subjects from Los Angeles and predominantly white participants from the other three centers. Racial differences in semen quality are not well understood, but may underlie the geographic variation in semen quality reported for different countries. A higher proportion of men from Missouri were smokers. Smoking is yet another variable that has been associated with lowered sperm counts.

The authors suggest that men from Columbia, Missouri, a rural region compared to New York City or Los Angeles may have lower sperm concentrations due to the agricultural industry and exposure to pesticides. Unfortunately the questionnaire given to study participants did not request information regarding occupational history or exposure to pesticides and other chemicals. Further study is required to attribute this geographical variation in sperm concentration to increased exposure to pesticides, rather than socioeconomic factors in rural regions.

Exposures to pesticides were analyzed for two exposure windows: the pre-conception (3 months before conception and the conception month) and post-conception periods (3 months post conception). Exposures after pregnancy loss but in the period of interest were ignored. Pregnancy specific variables were created for other time-related factors that may have influenced the pregnancy loss such as parental age, smoking status, farm activities, and alcohol and caffeine intake. Three types of comparisons were examined in the analysis: (1) Exposed pregnancies in the pre- and post-conception period versus non-exposed pregnancies, to the pesticide unit of interest during the time window; (2) Pre- verses post-conception exposures, using post-conception exposure as the referent group and; (3) Early (<12 weeks gestation) vs. late (12-19 weeks gestation) spontaneous abortions, using late spontaneous abortions as the referent group.

In total there were 395 spontaneous abortions from 3936 pregnancies. The results suggested that the critical window of exposure for spontaneous abortion was in the pre-conception period. A moderate increase in risk of early spontaneous abortion (<12 weeks gestation) was seen for pre-conception exposures to phenoxy acetic acid (OR=1.5, CI=1.1-2.1), and herbicides (OR=1.4, CI=1.1-1.9). For late spontaneous abortions (12-19 weeks gestation) an increased risk was seen for preconception exposures to glyphosate (OR=1.7, CI=1.0-2.9), thiocarbamate (OR=1.8, CI=1.1-3.0), and miscellaneous pesticides (OR=1.5, CI=1.0-2.4). For exposures in the post-conception period, only the miscellaneous pesticide exposure showed an elevated risk (OR=1.9, CI=1.2-3.0) of late spontaneous abortion. When comparing pre- vs. post-conception exposures an elevated risk for early spontaneous abortion was seen for pre-conception exposures to the active ingredients 2,4-D (OR=2.9, CI=1.1-8.0) and 2,4-DB (OR=7.8, CI=1.0-62.3). As well elevated risks were seen for the chemical families phenoxy acetic acid (OR=3.1, CI=1.4-6.4), triazine (OR=1.9, CI=1.0-3.2), organophosphate (OR=2.2, CI=1.0-4.8) and thiocarbamate (OR=2.5, CI=1.1=5.8) exposure in the pre-conception period. Comparisons of early vs. late spontaneous abortions revealed that exposure to phenoxy acetic acid elevated the odds of early spontaneous abortion (OR=1.9, CI=1.1-3.3).

The Classification and Regression Tree (CART) was used to explore interactions between various pesticide units and other risk factors. Advanced maternal age was significantly associated with increased risk of spontaneous abortion (OR=2.6, CI=1.7-3.9). Using the CART method, a strong interaction was found between maternal age and pesticide exposure. Women over 35 years of age exposed in the pre-conception period to carbaryl had a 4-fold increase in risk compared to women of the same age who were unexposed. As well, women over 35 exposed to both carbaryl and 2,4-D in the pre-conception period were 27 times more likely to have a spontaneous abortion compared to unexposed women of the same age.

Due to some methodological limitations in this study, the results should be interpreted with caution. First, there is a possibility that women who had spontaneous abortions would have had better recall of their pesticide use resulting in differential misclassification of pesticide exposures. However, the authors state that most of the reporting of pesticide use was done by the farm operator and/or husbands, which would have limited the potential for recall bias by women. Second, pesticide exposures were not directly measured, making the exposure measurements less precise. There are several factors that could have contributed to the delivered dose of pesticides to farmers including: the type of pesticide formulation, application methods and conditions, handling practices, differences in absorption, distribution, metabolism, and excretion of products or metabolites. Finally, the incidence of spontaneous abortion is estimated to be close to 50% of all pregnancies. Thus, several spontaneous abortions may not have been recognized by the farm women and accounted for in the study. Overall this paper contributes to our knowledge base and extends the literature suggesting a link between exposure to specific pesticide active ingredients and spontaneous abortion.