Putative environmental influence on male reproductive health have been discussed widely during the past 3 decades, stimulated in part by the findings of impaired testicular function in men exposed to dibromochloropropane.1,2 Later, a worldwide decline in sperms count was suggested by Carlsen et al3 but questioned by others.4–7 There are geographic differences in sperm counts and, in some countries, a possible decline.8–10 Environmental hormone-like exposures during fetal life,11 physical factors,12 and prenatal tobacco exposure13,14 have all been considered as possible influences.
Although sperm production, like all other biologic mechanisms, is under direct genetic control, the impact of genetic factors on male fertility is not established. Rare disorders such as Kallman disease15 and Klinefelter syndrome16 and microdeletions on the Y chromosome17,18 are known genetic causes of impaired sperm production. New evidence indicates that polymorphisms of the androgen receptor gene may be related to male subfertility19; longer CAG repeats in exon 1 of the androgen receptor seem related to male infertility, at least in Asian populations.
Sperm counts in Finland are among the highest in Europe, whereas low sperm counts have been found in Denmark.3 Sperm counts in Finland have remained unchanged from 1958 to 1992 but have possibly declined in Denmark.20,21 Because Danes and Finns have different ethnic origins, genetic differences could contribute to the geographic differences in sperm counts but not to the changes over time.
Twin studies can help to disentangle the role of nature and nurture. A difference in occurrence of phenotypic traits or diseases among genetically identical monozygotic twins (MZ) indicates an environmental effect, whereas similarity among MZ twins with differences in dizygotic twins (DZ) may express a genetic effect. An estimate of the prenatal environment can, to some degree, be evaluated by comparing twins with singleton brothers. Twins share the same prenatal environment; singleton brothers share the same uterus but at different points in time.
In a small study22 of 17 twin pairs (11 MZ and 6 DZ pairs), sperm density exhibited a strong familial effect, but a genetic component could not be confirmed. In contrast, sperm morphology seemed to exhibit a genetic effect (correlation for MZ was high but negative for DZ), although the modest size of the study allowed no firm conclusions. Christensen et al23 studied fecundability among twins. They found a correlation in time-to-pregnancy among MZ but not among DZ, indicating the importance of genetic factors. In a study of 120 pairs of MZ and 88 pairs of DZ, Meikle et al24 estimated that genes regulate 25% to 76% of the total variation of the plasma content of hormones such as testosterone, follicle-stimulating hormone (FSH), and luteinizing hormone (LH).
We investigated semen quantity and morphology, sexual hormone levels, and sperm chromatin integrity in 50 pairs of MZ, 51 pairs of DZ, and 51 pairs of singleton brothers.
The twins were selected from the Danish Twin Registry, a population-based registry established in 1954. The twin registry was extended in 1990 with a population-based cohort of younger twins born 1953 to 1982 ascertained by means of the Danish Civil Registration System. Definition of zygosity was based on a self-administrated questionnaire completed by the twins when they entered in the Twin Register. The misclassification rate of zygosity based on this questionnaire is estimated to be less than 5%.25
A selection of 500 MZ and 500 DZ (500 pairs) was carried out. Twins who were or had just been involved in other projects were not considered eligible. The remaining twins were sampled at random, but with a larger sampling fraction for those living close to the research center as identified by their postal codes.
The cohort of single-born brothers was extracted from The Danish Civil Registration System, which was established in 1968 and contains data on family ties, addresses, and vital status. All persons with a permanent address in Denmark are given a unique civil registration number at birth or at the time of immigration. We selected at random 500 pairs of brothers (1000 men), living in the county of Aarhus, who had a maximum age difference of 5 years. The single-born brothers were initially included because they study evaluated the impact of prenatal estrogen on sperm density during a twin pregnancy and a singleton pregnancy.26 Men were Danish-born and 20 to 45 years of age in all 3 groups.
Men who had undergone vasectomy or had known azoospermia were excluded. If one of the brothers was adopted or declined participation, neither of the brothers was included in the study. The study was approved by the regional ethics committees and participants provided written informed consent.
Among the 2000 invited men, 778 agreed to participate (39%) (396 single-born brothers [40%], 197 DZ [39%], and 185 MZ [37%]). We excluded 147 men whose brother did not respond. Guided by statistical power calculations, we terminated the consecutive enrollment of the 631 men when we reached the target number of semen samples (105 single-born brothers, 104 MZ, and 107 DZ) (Fig. 1). 27 The final population comprised 304 semen samples from 50 pairs of MZ, 51 pairs of DZ, and 51 pair of single-born brothers.
Of the 304 men in the final sample, 272 gave us their consent to contact their mothers, of whom 257 provided information on birth weight and gestational age on their sons through a postal self-administered questionnaire. Eleven of the mothers had died, 11 sons declined to provide consent, mothers of 3 sons did not return the questionnaire, and for 9 mothers, only one of the 2 sons had delivered a sample.
Semen was collected by masturbation into a sterile 50-mL polyethylene jar and kept warm (close to the body) until delivery for analysis. The samples were analyzed by one of 2 trained staff members (technician or first author) within 30 minutes to 1 hour 30 minutes after masturbation. Information was obtained on a self-completed questionnaire for the following variables: reproductive, medical, and occupational history; frequency of saunas; spillage at collection; and number of days since previous ejaculation. Data collection started in November 1999 and ended in May 2000.
The semen samples were examined in a mobile laboratory at the participant's home (n = 110) or at a stationary laboratory at Aarhus University (n = 194). The samples were kept in a heat chamber at 37°C for liquefaction. Semen volume was measured in a graduated Falcon tube with an accuracy of 0.1 mL. The sperm concentration (number of spermatozoa per milliliter) of an appropriate dilution was counted in an improved Neubauer hemacytometer according to World Health Organization criteria.28 The samples were counted twice, and if the 2 counts differed by more than 10%, the sample was counted again twice (n = 48). If the 2 second counts also differed more than 10%, an average of the 4 counts was used (n = 23). An internal quality program was carried out during the sample collection (described by Storgaard et al13). Sperm morphology was classified according to the 1999 World Health Organization criteria (all performed by one technician who was unaware of sibling status).28 The smears were air-dried, fixed in 96% ethanol, and stained using a modification of Papanicolaou's stain.
Serum was stored at −80°C until hormone assay. Serum concentrations of LH, FSH, and total testosterone were measured using the VIDAS analysis, an enzyme-linked fluorescent assay (bioMérieux, Marcy l'Etoile, France). Dimeric inhibin-B was measured using a commercially available enzyme-linked immuno-sorbent assay (Oxford Bio-Innovation Ltd., Oxford, U.K.). The assay uses 2 monoclonal antibodies; the first is specific for the B-subunit of inhibin and the second is directed against the α-subunit and coupled to alkaline phosphatase. A standard of a mixture of inhibin forms from human follicular fluid was used, and the concentration was determined by calibration against recombinant 32 kD inhibin-B. The sensitivity of the assay was 15 pg/mL and the interassay variation of a sample containing 305 pg/mL inhibin-B was 9.5% (n = 8). All hormone analyses were performed without knowledge of sibling status.
Sperm Chromatin Structure Assay
The sperm chromatin structure assay was applied according to the procedure described initially by Evenson et al29 and later modified by Spano et al.30 Briefly, 0.2-mL aliquots of semen were kept at −80°C until analysis; one to 2 million sperm cells were treated with a low pH detergent solution (pH = 1.2) and then stained with acridine orange. The low pH solution partially denatures DNA in sperm with abnormal chromatin structure, whereas DNA in sperm with normal chromatin structure does not denature. Acridine orange that is intercalated into double-stranded DNA fluorescence red under a blue light. Sperm chromatin damage was quantified by measuring the red and green fluorescence intensity emitted from each of 10,000 sperm cells using a FACScan flow cytometer (Becton Dickinson, San Jose, CA). Each sample was measured twice. The extent of sperm DNA denaturation was expressed by the function αT obtained by the ratio of red to total (red plus green) fluorescence intensity, thus representing the amount of denatured, single-stranded DNA over the total cellular DNA. The results of the sperm chromatin structure assay are expressed in terms of mean αT and percentage of cells outside the main population (COMP αT). Samples with sperm showing increased susceptibility to DNA denaturation have broader αT distributions and larger COMP αT values. The sperm chromatin structure assay correlates with the probability to conceive in normal populations.30
Analysis and Statistical Methods
Monozygotic twins share all (or almost) all genes, whereas DZ, like ordinary siblings, share, on average, 50% of their genes. The classic twin study compares MZ and DZ intraclass correlations for a trait. A higher correlation in MZ than in DZ indicates that genetic factors (or intrauterine conditions shared to a higher degree among MZ than DZ) contribute to the variation.
To estimate the heritability of semen quality and sex hormones (ie, the proportion of the population variance attributable to genetic variation), the twin data were analyzed using structural equation biometric models.31 The simple assumptions of a classic twin study (no gene-environment interaction or correlation, no assortative mating or epistasis, and environmental similarity among MZ and DZ) were also applied in this study, although violations to these assumptions are expected.32,33
We used an “ACE” model for all variables. This model estimates the “Additive” genetics variation (A, the variance contribution of inherited genetics), the variation accounted for by the “Common” environment (C, the influence on a variable from shared environmental factors in twins reared together), and the variation accounted for by the nonshared “Environment” (E, the influence of the environment experienced by the individual). Figure 2 shows the 3 variance components that can be modeled (A, C, and E). The additive genetic correlation (A) for MZ is fixed at 1.0 (ie, MZ are 100% genetically identical), and for DZ and singleton brothers, the correlation is 0.5 (ie, the brothers share on average 50% of their genes). By definition, nonshared environmental (E) factors are uncorrelated, and shared environmental factors (C) are perfectly correlated for the twins. This assumption was also applied for singletons, but analyses were carried out both with and without including singletons. Because single-born brothers do not share prenatal and perinatal environment to the same extent as twins, analyses were also carried out setting the correlation of shared environment to 0.5 (data not shown). Thus, 2 models were analyzed.
Model 1, which included MZ and DZ only (the traditional twin method), and model 2, which included MZ (shared environment correlation = 1.0) and DZ plus singleton brothers (shared environment correlation = 0.5). Only the results from model 1 are shown (Table 1).
To obtain an approximation to a Gaussian distribution, the following transformations were carried out: sperm density by the cubic root; inhibin B by the square root; FSH, LH, and mean αT by the logarithm; morphology by logit; and COMP αT by the logarithm of the logit. Pearson correlation coefficients, variances, and covariances were first calculated for each of the 3 types of pairs (SAS; SAS Institute, Cary, NC) procedure Proc Corr. Calculations were also carried out on adjusted values using the residuals from multiple regression models (SAS procedure Proc. GLM). The following potential confounders were included in the regression model: duration of sexual abstinence (0–2 days, 3–6 days,; 7+ days); season (January and February vs the rest of the year34); high alcohol intake (more than 21 alcoholic drinks/week: yes/no); cryptorchidism (one or both testicles not in scrotum or having undergone treatment of a nondescended testicle); current smoker (yes/no); sauna (during the last 5 weeks: yes/no; birth weight (dummy variable using the highest birth weight as a reference); and age (20–29, 30–39, 40–45 years). In addition, for analysis of sex hormones, adjustment was done for time of the day when the blood sample was collected (before 9:00 am; 9:00 am-12:00 pm, and after 12:00 pm). For missing values of birth weight (n = 37), median values were applied for twins and singletons.
Structural equation modeling was carried out on both unadjusted and adjusted data using the software package Mx for a univariate model.31 To correct for unequal variances between twin 1 and twin 2, data were entered twice and the degrees of freedom adjusted accordingly.35
Categorical variables were generated for sperm density (≤20 million/mL vs higher), FSH (≥75th percentile vs lower) and inhibin B (≤25th percentile vs higher). We calculated probandwise concordance rates (ie, the probability that the cotwin to an affected twin will also be affected) for the 3 categorical variables. For sperm density, analyses were also carried out after direct adjustment to sperm concentration after 3 days of sexual abstinence. For a period of abstinence less than 3 days, we added 14.1 million/mL per day to the measured sperm concentration values. For abstinence between 3 and 8 days, we subtracted 14.1 million/mL per day of abstinence, and for abstinence of more than 8 days, we subtracted 53 million/mL. For inhibin B, a correction factor of 7.4 pg/mL per hour was added for those samples collected after 10:00 am. The correction factors were derived from analysis in the present data of period of sexual abstinence and the time of the day when the blood sample was collected.
Analyses were also carried out for all data with exclusion 12 of cases with cryptorchidism. Results of these analyses are shown only for probandwise concordance rates.
Birth weight and gestational age were similar for both types of twins but lower than for singletons (Table 2). Dizygotic twins had more cases with cryptorchidism.
Table 3 shows correlation coefficients for the 3 groups of pairs. The correlations between MZ were highest, especially for inhibin B, FSH, and COMP αT.
The components for heritability (h), common environment (c), and nonshared environment (e) are presented for all variables in Table 1. Only a small proportion of the variation in sperm density was accounted for by genetic factors. This is in contrast to a high heritability factor for inhibin B and FSH, both of which reflect Sertoli cell function. For Leydig cell function (as expressed by the hormones LH and testosterone), there was a smaller heritability. For morphology and sperm chromatin integrity, expressed as mean αT and COMP αT, the heritability was relatively large. The common environment component was low for all variables except sperm count. We found no change in the estimates when pairs of single-born brothers were included in the model.
Overall, there is a notable familial component (familial resemblance) for all outcomes (this is shown in the fact that the confidence interval for the E component never includes 1.0). However, the statistical power of the twin studies is too low to separate common genetic factors from common rearing environment. The fact that C converts to zero is not related to this power issue, but rather to the fact that there is no indication of an influence of common environment in cases in which the MZ correlation is more than twice the DZ correlation. All the similarity is under the model assumption due to genetic factors, which is why C converts to zero in these cases.
We found support for major genetic contribution to variations in inhibin B and FSH, but not in sperm density. Inhibin B and FSH reflect Sertoli cell function and thus the production of germ cells (approximately 10 germ cells per Sertoli cell). The production of semen in adulthood is therefore related to the number and function of Sertoli cells.36 The “common environment” was related only to sperm density, which may seem contradictory because the number of Sertoli cells is mainly established in early childhood. These findings indicate that the model assumptions are too simplistic or that sperm counts are subject to variability within individuals over short time periods and to measurement errors.28,37 Thus, the heritability factor for sperm density may be underestimated. For the hormones we analyzed, the within-subject variability is relatively small apart from the circadian dependence on blood levels.38,39 The probandwise concordance rate for oligospermia was at a similarly high level among both MZ and DZ, but very low among single-born brothers. For both twin groups, these rates were in the same range (0.46 and 0.62), whereas for singletons, it was considerably lower (0.15) (Table 4).
The equal concordance rates among MZ and DZ suggest, consistent with the ACE analysis, that genetic factors are of minor importance. If we assume that the major factors that distinguish DZ and singleton brothers are the intrauterine environment and early postnatal factors, then the finding is compatible with an effect of the intrauterine environment on sperm counts. However, there is a limit to how much the single-born brothers contribute to the analyses of the impact of pre- or postnatal environment. Correlations for sperm density indicated no such strong effects.
Some of the results from the twin analysis (Table 1) have broad confidence intervals, and interpretation should be made with caution.
The common environmental factors captured by C are the environmental factors shared by twins. These factors would typically be mother's age and lifestyle during pregnancy, rearing conditions (the few twins reared apart has almost no impact on the overall analysis), and factors that might be shared by twins later in life (such as socioeconomic status). In contrast, E represents nonshared environmental factors. These could be contrasting lifestyle habits (eg, smoking) and job exposures as well any measurement error and stochastic component.
Sperm morphology and sperm chromatin structure seem to be largely determined by hereditary factors, which is consistent with the small within-subject variation.40–42
Selection bias due to nonresponse constitutes a well-known problem in semen studies from healthy men, and low participation rates have resulted in oversampling of men with low fertility.43 We obtained information from the Danish Civil Registration System on the number of children for all men initially selected for the study; we found no differences between responders and nonresponders in fertility rate. Even if the subfecund were more likely to respond, it is unlikely that this would be differential among the 3 groups of brothers. Further analysis demonstrated that nonresponders were slightly older (30 years vs 29 years).
It is well established that the length of sexual continence correlates with semen variables (volume, count, chromatin structure).40,44 The duration of sexual abstinence was slightly longer for DZ and singletons in comparison with MZ. This aspect was, however, taken into account in the adjusted values together with other potential confounders, including seasonal variation in semen sampling.34
In conclusion, our study indicates a substantial hereditary component in Sertoli cell function as expressed by plasma levels of hormones and in sperm cell chromatin stability and morphology. In contrast, there appears to be a large environmental contribution, including prenatal environment, to sperm counts. Thus, the findings of this study reinforce the search for environmental factors that influence semen quality and, in particular, sperm counts.
The authors are indebted to support from Professor Gorm Dancher, research technician Kirsten Lunding, and technicians at the laboratory at the Institute of Anatomy, Aarhus University.
1. Whorton D, Krauss RM, Marshall S, et al. Infertility in male pesticide workers. Lancet
2. Whorton D, Milby TH, Krauss RM, et al. Testicular function in DBCP exposed pesticide workers. J Occup Med
3. Carlsen E, Giwercman A, Keiding N, et al. Evidence for decreasing quality of semen during past 50 years. BMJ
4. Becker S, Berhane K. A meta-analysis of 61 sperm count studies revisited. Fertil Steril
5. Handelsman DJ. Sperm output of healthy men in Australia: magnitude of bias due to self-selected volunteers. Hum Reprod
6. Olsen GW, Bodner KM, Ramlow JM, et al. Have sperm counts been reduced 50 percent in 50 years? A statistical model revisited. Fertil Steril
7. Fisch H, Goluboff ET, Olson JH, et al. Semen analyses in 1,283 men from the United States over a 25-year period: no decline in quality. Fertil Steril
8. Fisch H, Ikeguchi EF, Goluboff ET. Worldwide variations in sperm counts. Urology
9. Fisch H, Goluboff ET. Geographic variations in sperm counts: a potential cause of bias in studies of semen quality. Fertil Steril
10. Bujan L, Mansat A, Pontonnier F, et al. Time series analysis of sperm concentration in fertile men in Toulouse, France between 1977 and 1992. BMJ
11. Sharpe RM, Skakkebaek NE. Are estrogens involved in falling sperm counts and disorders of the male reproductive tract? Lancet
12. Hjollund NH, Bonde JP, Jensen TK, et al. Diurnal scrotal skin temperature and semen quality. Int J Androl
13. Storgaard L, Bonde JP, Ernst E, et al. Does smoking during pregnancy affect sons' sperm counts? Epidemiology
14. Jensen TK, Jorgensen N, Punab M, et al. Association of in utero exposure to maternal smoking with reduced semen quality and testis size in adulthood: a cross-sectional study of 1,770 young men from the general population in five European countries. Am J Epidemiol
15. Hardelin JP, Levilliers J, Blanchard S, et al. Heterogeneity in the mutations responsible for X chromosome-linked Kallmann syndrome. Hum Mol Genet
16. Hargreave TB. Genetics and male infertility. Curr Opin Obstet Gynecol
17. Krausz C, Rajpert-De Meyts E, Frydelund-Larsen L, et al. Double-blind Y chromosome microdeletion analysis in men with known sperm parameters and reproductive hormone profiles: microdeletions are specific for spermatogenic failure. J Clin Endocrinol Metab.
18. Fujisawa M, Shirakawa T, Kanzaki M, et al. Y-chromosome microdeletion and phenotype in cytogenetically normal men with idiopathic azoospermia. Fertil Steril
19. Gottlieb B, Lombroso R, Beitel LK, et al. Molecular pathology of the androgen receptor in male (in)fertility. Reprod Biomed Online
20. Vierula M, Niemi M, Keiski A, et al. High and unchanged sperm counts of Finnish men. Int J Androl
21. Suominen J, Vierula M. Semen quality of Finnish men. BMJ
22. Handelsman DJ. Estimating familial and genetic contributions to variability in human testicular function: a pilot twin study. Int J Androl
23. Christensen K, Kohler HP, Basso O, et al. The correlation of fecundability among twins: evidence of a genetic effect on fertility? Epidemiology
24. Meikle AW, Bishop DT, Stringham JD, et al. Quantitating genetic and nongenetic factors that determine plasma sex steroid variation in normal male twins. Metabolism
25. Kyvik KO, Christensen K, Skytthe A, et al. The Danish Twin Register. Dan Med Bull
26. Storgaard L, Bonde JP, Ernst E, et al. Effect of prenatal exposure to estrogen on quality of semen: comparison of twins and singleton brothers. BMJ
27. Bonde JP, Giwercman A, Ernst E. Identifying environmental risk to male reproductive function by occupational sperm studies: logistics and design options. Occup Environ Med
28. World Health Organization. WHO Laboratory Manual for the Examination of Human Semen and Sperm-Cervical Mucus Interaction
, 4th ed. Cambridge University Press; 1999.
29. Evenson DP, Jost LK, Baer RK, et al. Individuality of DNA denaturation patterns in human sperm as measured by the sperm chromatin structure assay. Reprod Toxicol
30. Spano M, Bonde JP, Hjollund HI, et al. Sperm chromatin damage impairs human fertility. The Danish First Pregnancy Planner Study Team. Fertil Steril
31. Neale M, Boker SM, Xie G, et al. Mx: Statistical Modeling
, 6th ed. Department of Psychiatry; 2002.
32. Neale M, Maes H. Methodology for Genetic Studies of Twins and Families.
Dordrecht, The Netherlands: Kluwer Academic Publisher BV; 2002.
33. Plomin R, DeFries J, McClearn GE, et al. Behavioral Genetics
, 4th ed. In: Bayne J, ed. New York: Worth Publisher; 2001.
34. Levine RJ, Mathew RM, Chenault CB, et al. Differences in the quality of semen in outdoor workers during summer and winter. N Engl J Med
35. Sham P. Statistics in Human Genetics
. New York: John Wiley & Sons; 1998;214.
36. Orth JM, Gunsalus GL, Lamperti AA. Evidence from Sertoli cell-depleted rats indicates that spermatid number in adults depends on numbers of Sertoli cells produced during perinatal development. Endocrinology
37. Neuwinger J, Behre HM, Nieschlag E. External quality control in the andrology laboratory: an experimental multicenter trial. Fertil Steril
38. Yamaguchi M, Mizunuma H, Miyamoto K, et al. Immunoreactive inhibin concentrations in adult men: presence of a circadian rhythm. J Clin Endocrinol Metab
39. Weitzman ED. Circadian rhythms and episodic hormone secretion in man. Annu Rev Med
40. Spano M, Kolstad AH, Larsen SB, et al. The applicability of the flow cytometric sperm chromatin structure assay in epidemiological studies. Hum Reprod
41. Schrader SM, Turner TW, Breitenstein MJ, et al. Longitudinal study of semen quality of unexposed workers. I. Study overview. Reprod Toxicol
42. Wyrobek AJ, Gordon LA, Burkhart JG, et al. An evaluation of the mouse sperm morphology test and other sperm tests in nonhuman mammals. A report of the US Environmental Protection Agency Gene-Tox Program. Mutat Res
43. Larsen SB, Abell A, Bonde JP. Selection bias in occupational sperm studies. Am J Epidemiol
44. Schwartz D, Laplanche A, Jouannet P, et al. Within-subject variability of human semen in regard to sperm count, volume, total number of spermatozoa and length of abstinence. J Reprod Fertil