Commentary: Trends in Sperm Counts

The Saga Continues

Almost 20 years ago, a longstanding debate over possible declines in sperm counts was reignited by a paper in BMJ, claiming that sperm counts had declined worldwide by 50%.¹ Despite its rather weak documentation, this paper by Carlsen et al had a strong impact in the public media, and has been cited in more than 1000 scientific papers—perhaps in part because the authors were bold enough to include a linear regression line pointing forward toward continuing declines and a doomed society. Since the publication of that paper, numerous studies have reported secular trends and geographical shifts in sperm counts, with conflicting findings and no emerging consensus.^2,3 However, recent developments may be changing this picture.

Population studies of sperm parameters have been highly problematic. Nonresponse in population-based semen studies is invariably high, and there is lack of compliance with the requirement of an abstinence time before sampling. The fundamental problem in secular-trend studies is a lack of comparability of various study populations from different time periods—a violation of the first principle of trend analysis. Add to this the problems of poor quality assurance in laboratory methods for counting sperm, and we probably will never be able to know how much the quality of sperm might have changed during the late 20th century.

Although we cannot change the data quality of the past, we can improve studies prospectively. Toward this end, a Danish research project was started in 1996 with support from Danish government agencies to prospectively monitor semen quality of young Danish military draftees.^4–6 Healthy 18-year-old men who attend a compulsory examination of fitness for military service in 2 Danish cities are encouraged to provide semen and blood samples. Each year, from 16% to 30% have agreed,⁶ and a total of 5000 men have provided semen samples. These data were publicly presented for the first time in March of this year, when they were posted on a government agency Web site by the Danish National Board of Health.⁷ These data provide no indication that semen quality has changed during the past 15 years (Fig.). Throughout the years of surveillance, the crude median sperm concentration has fluctuated around a median value of 40–45 million/mL, with the lowest value (35 million/mL) in 2006 and the highest (50 million/mL) in 2007.

FIGURE.: Sperm concentrations for Danish military draftees from 1996 to 2010 based on information from the Danish monitoring program (Department for Growth and Reproduction, Rigshospitalet, Copenhagen, Denmark). This figure was first presented in March 2011 on a government agency website by the Danish National Board of Health.⁷

As it happens, 2 other population-based studies have also recently appeared—one from Sweden⁸ showing no changes over time, and the other from Finland showing declines in various semen parameters.⁹ Both studies are based on much smaller samples and cover shorter time periods. (Needless to say, it was the Finnish study that was featured in a BBC news report—“Sperm quality fall ‘due to chemicals’.”)

The Danish data, collected annually over a 15-year period, provide the best longitudinal semen data yet available. Such prospectively collected data from a well-defined source population and examined according to a uniform protocol offer a much better basis for evaluation of secular trends than retrospective data. Still, there are important concerns. Were findings similar in the 2 cities? Were sperm counts related to the fluctuating participation rates? Would adjustment for period of sexual abstinence or season affect the results? Might findings depend on the statistical model for adjustment and trend analysis? Unfortunately, the Danish data are presented without any of this information. Earlier analyses from this project have shown that the serum level of inhibin B (a marker of Sertoli cell function and sperm production) is the same among the men providing a semen sample and those providing only a blood sample.⁶ Thus, there is no evidence of selection by fecundity among these young men.⁴ Still, a proper analysis of these data would be needed before we could have full confidence in their interpretation.

Until such analyses are provided, let us assume that the reported crude sperm count data of Danish army draftees are a valid reflection of the fecundity of young Danish men. This result raises new and important questions.

First, the Danish data provide a new perspective on the claim that human sperm counts are declining. Alleged links between declining sperm counts and exposure to environmental toxicants have generated immense media and public interest, while fuelling research programs to address the “estrogen hypothesis,”¹⁰ and more recently the broader “endocrine disruption hypothesis.”¹¹ The Danish data obviously do not reflect semen quality in earlier years, but it is a puzzling coincidence that a decline in sperm counts would end exactly at the time when a proper monitoring program is initiated.

How do the new results fit with the hypothesis that sperm quality and testicular cancer share etiologic factors operating during fetal life and early childhood?¹⁰ There has been a well-documented increase in the incidence of testis cancer in Denmark and most other affluent countries since World War II.¹² As army draftees are 18–20 years when examined, and the incidence of testis cancer peaks at 30–35 years of age, an attenuation of the increase in testis cancer would be expected to lag some 10–15 years behind attenuation of a sperm-count decline. The fact that the rise in testis cancer in Denmark has slowed (and perhaps even reversed) since the end of the 1990s is not incompatible with the Danish semen data.¹³

How do the new findings fit with concerns regarding environmental effects on male reproduction? A substantial research effort in the past 15 years has uncovered numerous environmental chemicals that weakly interfere with endogenous hormonal regulation through a variety of mechanisms related to steroid receptors and various enzymes. The estrogen hypothesis proposed in the early 1990s has been largely abandoned by one of its fathers,¹⁴ and is not supported by a wide range of epidemiologic data.¹⁵ Meanwhile, there is only very limited epidemiologic evidence to support the broader endocrine disruption hypothesis (delayed effects on the male reproductive system following imbalance of estrogenic and androgenic actions in early fetal life, for instance, as might be caused by phthalate exposure¹⁶).

Environmental chemicals should not be our sole concern. Advances in our understanding of reproductive health in light of the fetal programming hypothesis suggest many other potential threats.¹⁷ Maternal alcohol intake has been associated with lower sperm counts in sons,¹⁸ and almost half of pregnant women in Denmark have reported alcohol intake throughout pregnancy.¹⁹ Maternal smoking during pregnancy has also been associated with lower sperm counts.^20–22 Danish women started smoking in large numbers after World War II, and for many years had the highest smoking prevalence in Northern Europe.²³ Smoking among pregnant women peaked at about 40% in the 1980s²⁴ and has since declined to 22% in 1997 and 16% in 2005.²⁵ If mother's smoking causes lower sperm counts, the decline in maternal smoking should be reflected in a slower decline (or even an increase) in sperm counts over time.

Although there is no evidence of a recent decline in sperm concentration in Denmark and Sweden, the proportion of young men with low sperm counts is surprisingly large.²⁶ It is somehow hard to think it has always been so. If sperm quality were truly higher in the past, what are possible reasons (besides bias, or a possible increased sexual activity over time)? If a decline had been produced by environmental exposures, these exposures may have reached a plateau, or may have exceeded a threshold level above which higher exposures produce no further effects. Male obesity is one candidate for a harmful exposure operating either at the time of sperm production or the time of organogenesis—although the obesity epidemic probably has come too late to explain these observations. Exposure to environmental chemicals or drugs may be a better candidate, and has certainly been the preferred explanation. Some of these exposures have been reduced over time, but if the causal effect is taking place during prenatal life, we may not see improvements for a generation. The same delayed detection would be true for effects of a newly introduced reproductive toxicant. It may be too early for society to dismiss the concerns so graphically presented in the book and film, “The Children of Men.”

Recent developments in the sperm-count story emphasize the continuing need for good prospective data—not only of semen quality and reproductive hormones, but also of indicators of female fecundity and couple fecundity, such as time to pregnancy.²⁷ In searching for possible etiologies of reproductive disorders, we need to pay attention not just to environmental toxicants, but to the wide range of behavioral, medical, and other factors that have potential to damage human reproduction.

ABOUT THE AUTHORS

JENS PETER BONDE is a professor in occupational medicine at the Copenhagen University Hospital of Bispebjerg, Denmark. CECILIA HØST RAMLAU-HANSEN is an assistant professor in epidemiology at Aarhus University, Denmark. JØRN OLSEN is a professor in epidemiology at Aarhus University, Denmark, and UCLA, LA. All authors have worked extensively in reproductive epidemiology.