Annual variation in semen characteristics and plasma hormone levels in men undergoing vasectomy

Transcription

1 FERTILITY AND STERILITY Copyright 1988 The American Fertility Society Vol. 49, No. 2, February 1988 Printed in U.S.A. Annual variation in semen characteristics and plasma hormone levels in men undergoing vasectomy Alain Reinberg, M.D., Ph.D.* Michael H. Smolensky, Ph.D.t Michael Hallek, M.D.* Keith D. Smith, M.D.t Emil Steinberger, M.D.t Fondation A. de Rothschild and CNRS US 581, Chronobiologie-Chronopharmacologie, Paris, France, The University of Texas Health Science Center at Houston, School of Public Health, and Graduate School of Biomedical Sciences, and Texas Institute for Reproductive Medicine and Endocrinology, Houston, Texas Prevasectomy levels of plasma luteinizing hormone (LH), follicle-stimulating hormone (FSH), testosterone (T), estradiol (E 2), and 2a-dihydroprogesterone (2a-DHP), as well as semen analyses including semen volume, sperm count, and sperm motility from 26 healthy men were evaluated for annual changes. A statistically significant (P ;:::;.15) high-amplitude seasonal variation with the peak in April to May was detected in semen volume, sperm count, and sperm motility. A statistically significant (P ;:::;.4) annual change of moderate T to large FSH amplitude was detected in each of the five plasma endocrine variables as well. Plasma LH, T, and E 2 peaked in autumn, while FSH and 2a-DHP peaked in summer. Analysis ofpostvasectomy LH, FSH, E 2, 2a-DHP, and T blood levels for the 3 years following vasectomy revealed loss of seasonal rhythmicity as a group phenomenon in LH, E 2, and T. The amplitude ofthe seasonal variation in FSH was decreased and that in 2a-DHP was unchanged compared with before-vasectomy baselines. For those annual rhythms which persisted following vasectomy, the peak time was unchanged. Compared with the prevasectomy group annual mean, that for each of the endocrine values was unchanged, except for that of LH and T, which was slightly, yet statistically significantly, elevated. The existence of prominent annual variation implicates their consideration in the design of research protocols involving investigation of reproductive phenomena in human beings. Fertil Steril 49:39, 1988 In a prospective study, our group investigated the effect of vasectomy on testicular functions, i.e., plasma levels of follicle-stimulating hormone (FSH), luteinizing hormone (LH), testosterone Received March 3, 1987; revised and accepted October 6, * Fondation A. de Rothschild and CNRS UA 581, Chronobiologie-Chronopharmacologie. t The University of Texas Health Science Center at Houston, School of Public Health, and Graduate School of Biomedical Sciences. :j: Texas Institute for Reproductive Medicine and Endocrinology. Reprint requests: Keith D. Smith, M.D., Texas Institute for Reproductive Medicine and Endocrinology, P.A., 78 Fannin, Suite 5, Houston, Texas (T), estradiol (E 2), and 2a-dihydroprogesterone (2a-DHP). 1 2 Postvasectomy plasma hormone levels of the 27 participants were compared with prevasectomy baseline values using data from blood samplings at fixed spans of time after vasectomy, i.e., 1 and 6 weeks, 3 and 6 months, and 1, 2, and even 3 years. Although statistically significant increases were detected in both LH and T, they were small and within the limits of normal ranges. The findings of this investigation, 1 2 as well as others reviewed elsewhere, 2 indicate that vasectomy does not adversely affect the endocrine function of the testis. At the time when our earlier research was conducted, circannual or seasonal changes in hormonal variables related to human reproduction and sexual activities were just beginning to be re- Reinberg et al. Seasonal rhythms in men 39

2 ported in males. This was the case for plasma T, 3 4 LH, and FSH, 5 melatonin 6 and corticosteroids, 7 although not necessarily for prolactin. 8 Moreover, annual changes in sex-related activities, such as frequency of sexual intercourse and masturbation, 9 incidence of rape and sexually transmissible diseases, as well as birth were validated as reviewed elsewhere. 1 With regard to these findings, it was of interest to reanalyze the hormonal values obtained after vasectomy because not only the span of time after surgery but also the time of year of sampling hormone levels could be important for properly interpreting findings. The reanalysis of before-vasectomy reference values also was of interest because annual rhythms in 2a-DHP in healthy mature human males had yet to be reported. Finally, data on semen characteristics-volume of ejaculate, sperm count, and motility-obtained prior to vasectomy in men of this sample remained unanalyzed. Thus, the aims of the present project were to examine the data for annual change in sperm characteristics and plasma hormones (LH, FSH, T, E 2, and 2a-DHP) as a group phenomenon in men studied before vasectomy and to reanalyze hormonal time series of up to 3 years' duration following vasectomy. Subjects MATERIALS AND METHODS Participants for the earlier-conducted vasectomy study were solicited from the vasectomy clinic at Planned Parenthood of Houston (Houston, TX). Men completing prevasectomy screening interviews were informed of the study and offered the opportunity to participate. Informed consent was obtained from all subjects, and no monetary rewards were offered. Three months after vasectomy, 27 of the 26 men entering the study remained in the investigation; 182, 147, and 128 remained, respectively 6, 12, and 24 months postvasectomy. Only 65 of the original group provided samples 3 years postoperatively. Data pertaining to endocrine levels of the 53 subjects who did not remain in the study a minimum of 3 months after surgery were excluded from evaluation. Conversely, data pertaining to their prevasectomy semen characteristics were included. Almost two thirds of the men were 25 to 34 years old (median age, 32. years); 9.% were Caucasian, 7.7% Mexican-American, and 2.1% Negro. Approximately one half of the participants re- 31 Reinberg et al. Seasonal rhythms in men ported having finished 12 years of school or less. The median was 12. years of education. More than 92% of the men were currently married; and, with one exception, all unmarried men were Anglo American. The sample included only slightly more white-collar (53.6%) than blue-collar workers ( 45.5%; not stated.9%) according to the classification scheme of the United States Department of Labor. More than 7% reported having two children or fewer; the mean number of children was Other particulars about the sample can be found in another publication. 11 Experimental Design and Data Gathering In principle, each subject was to submit two semen samples prior to vasectomy and additional semen samples 3, 6, 12, and 24 months after surgery. Such samples were collected by masturbation after a minimum continence time of 2 days. Every subject was requested to bring a semen sample with him to the first screening interview to document the need for vasectomy. The second sample was provided on the day of vasectomy. Overall, 26 men provided 355 semen samples collected in the standardized fashion as specified by the protocol. Preoperative blood samples were taken upon entrance into the study, the following day, and on the day of vasectomy, just prior to surgery. In total, 592 plasma samples from 27 subjects were obtained. Postoperative samples were taken 1 week, 6 weeks, 3 months, 6 months, 1 year, 2 years, and 3 years after surgery with both time of day (clock hour) and date recorded. Heparinized blood samples (3 ml each) were centrifuged immediately after collection, the plasma separated, frozen, and stored in a freezer at -7 C until assay. Laboratory Determinations Semen Analyses Sperm counts were performed by the hemocytometer technique in the same laboratory and under the same conditions, always by the same technician. The results were expressed in terms of the sperm count per milliliter as well as per ejaculate (total sperm count). Motility was estimated by the counselor conducting the prevasectomy screening interview and recorded as to motility relative to the time since ejaculation. Semen volume was measured in a graduated cylinder. Fertility and Sterility

3 Plasma Hormone Determinations Plasma concentrations oft, E2, FSH, and LH were measured in duplicate by radioimmunoassay (RIA). The details of procedures used fort, FSH, LH, and 2a-DHP have been reported previously The E2 RIA was a modification of the procedure of Smith et al.1 and Hotchkiss et al.14 Except for 2a-DHP, the first six samples from each subject were placed in the same assay run. Subsequent samples from the same subject were assayed with the first six samples from other subjects entering the study at a later date. Determinations of 2a-DHP were not performed until at least nine samples, the three prior to vasectomy plus the six accumulated during the 2 years after surgery, had been obtained. At that time, all samples from the subject were included in the same assay run. Statistical Analyses During the conduct of the investigation, new participants were entered into the study at all months of the year. Availability of pre- and postvasectomy data approximately equally distributed as a function of the time of year enabled the study of annual changes in the sampled variables, both before and after surgery. Several statistical methods were used, each one having both its merits and limitations. Therefore, conclusions are based on the concordance of results provided by all the methods. Time series were first conventionally analyzed. Raw data were plotted initially as a function of time of year on a monthly basis (annual chronogram). This approach provided a visualization of the temporal patterns, including the timing of the peak(s). Analysis of variance (ANOVA) was used to detect changes between the monthly means of the time series; however, it did not provide an answer with regard to difference in the means at the time of the peak and trough. Differences between highest and lowest monthly means also were t-tested when the ANOV A suggested statistical significance (P <.5). The cosinor method15 was used to evaluate the time series data for 1-year (365-day) and other period rhythms by testing the significance of the amplitude. Also, the cosinor method served to provide an estimate of parameters used to characterize rhythms, i.e., mesor (M; the rhythm-adjusted mean), amplitude (A; one half of the total peak-totrough variability) and the acrophase ( ; peak time) of the best-fitting cosine function approximating all data of a variable. Because in actuality the variation in data over the year in most cases is not a perfect cosine function, the cosinor-computed acrophase may sometimes differ slightly from the peak time observed in graphs. By means of this least-squares method, each parameter was given with its 95% confidence limits when rhythmicity was detected, i.e., when the cosinor-derived amplitude was not equal to zero at P <.5. The program we used provided residuals and allowed us to verify the normality of their distribution. Paired correlation tests were performed also to assess the strength of the association between selected variables. Finally, power spectrum analyses were conducted on the data for those 65 subjects for whom longitudinal participation in the study was for the entire 3 years. The method proposed by De Prins et al.16 was used since it was developed for time series analyses with missing data and/or unequally spaced intervals between time point measurements, which was the situation with regard to the sampling protocol used. RESULTS Annual Changes in Semen Characteristics Statistically significant changes were detected (with P <.1 top<.2) between the respective peak and trough of the three considered variables -total sperm count, sperm motility, and ejaculate volume (Fig. 1). Statistically significant annual variation was not detected in sperm count expressed per milliliter semen. Major peak times occurred in April for semen volume and in May for both total sperm count and sperm motility. However, for both of the latter variables, a second peak occurred respectively in December and November. The results of the cosinor analyses (Table 1) confirmed the existence of statistically significant annual large-amplitude changes in the three variables. The peak-to-trough difference (double A) was 25% for semen volume, 45.2% for total sperm count, and 33.8% for motility, with reference to the respective annual group mesor value. Display of the raw data and the findings of the cosinor analyses verified the same peak time (April) for semen volume, while these methods revealed a slight difference of 2 months for the other two variables. This discrepancy apparently results from the fact that the annual temporal pattern in total sperm count and motility does not resemble as strongly a sinusoidal shape as that in semen volume. In this regard, a 6-month component rhythm, in addition to the 1-year rhythm, was validated (P <.1) for Reinberg et al. Seasonal rhythms in men 311

4 - 3.5 E.!: IJJ 3.. ::E :::;)...J g 2.5 z IJJ 11i 2. t/1 coo ~ z :::;) 2 ::E 16 a: IJJ :); 12...J <( 1-8 t- 11' 7 > 1- ::! 6 t- o ::E 5 ::E a: IJJ 4 D. t/1 TIME OF YEAR (MONTHS) J F M A M J J A S N D Figure 1 Annual changes in semen characteristics. From top to bottom, chronograms of semen volume (in ml), total sperm count (1 6 ), and sperm motility (%). Monthly means ± SEM. Number of samples for calculating each monthly mean "" 28. both total sperm count and sperm motility. No component rhythm with a period equal to 6 months was detected for semen volume. No statistically significant correlation was found between the annual change in total sperm count and motility or semen volume. Finally, differences in semen volume before and 1 week to 12 months after vasectomy were not statistically significant. Annual Change in Plasma Hormones 13efore \!asecto~y Displays of raw data (Fig. 2) revealed statistically significant (with P <.1 top<.1) annual changes with regard to peak-to-trough differences in each of the five hormones. The temporal pattern over the year was roughly monophasic for FSH, LH, and 2a-DHP, and presumably biphasic for E 2 and T. Cosinor analyses confirmed (Table 1) annual rhythmicity for the five variables; most exhibited rather large-amplitude changes. The peak-totrough difference (equal to the cosinor-derived amplitude value multiplied by two) expressed rela- tive to its annual group mean ranged from 14.8% for T to 67.4% for FSH. Taking into account the fact that temporal patterns of the monthly means exhibited large swings, there was rather good agreement between the conventional and cosinor methods in determining the occurrence of the annual peak. Plasma LH, T, and E 2 peaked in autumn, while plasma FSH and 2a-DHP peaked in summer. The major trough occurred in winter for plasma E 2 and in autumn for 2a-DHP. Cross-correlation tests were performed between all the variabies using prevasectomy data. A positive correlation was found between 2a-DHP and sperm motility (r =.6, P <.5), while a negative correlation was found between LH and semen volume (r = -.579, P <.5), as well as between E2 and sperm motility (r = -.617, P <.5). After \!asecto~y Comparisons were conducted on the annual mesor values using data gathered before and during the 1-year span, as well as during the second and third-year spans after vasectomy. No statistically significant changes (P <.5) were observed in plasma FSH, E 2, and 2a-DHP. On the contrary, statistically significant postvasectomy increases were observed in plasma LH ( + 14% during first year and +22% last 2 years with P <.5) as well as in plasma T ( + 13% during first year with P <.1 and +8.9% during the last 2 years with P <.5). A circannual rhythm was no longer detected in plasma T during the 1-year postvasectomy period. Also, rhythmicity was no longer detected in LH, T, or E 2 using the data obtained during the entire 3- year period after vasectomy. Annual rhythms in the other variables were validated by the cosinor method as well as the power spectrum (e.g., longitudinal study over 3 years of 65 subjects). However, the respective amplitude of detected rhythms, based on the data obtained during the 1-year span after vasectomy, was reduced to almost one-half of that found before vasectomy (Table 1). Only plasma 2a-DHP maintained its large-amplitude circannual rhythmicity during the 3-year span after vasectomy. For detected rhythms as well as rhythms close to detection (.5 < P ;:;;.1), the respective cps remained the same. Taking into account both the time of year and span of time after surgery, vasectomy was associated with the following: (1) in consideration ofthe annual means (M), no change in FSH, E 2, and 2a-DHP and slight, but statistically significant, increase in LH and T; (2) in consideration of the 312 Reinberg et al. Seasonal rhythms in men Fertility and Sterility

5 Table 1 Estimates of Circannual Rhythms Parameters Cosinor Summary Rhythm No. of detection Situation Variable data with P" Before vasectomy Semen Volume 355 <.2 Total count 354 <.12 Motility 343 <.3 Pl. hormones FSH 592 <.1 LH 592 <.1 Testosterone 575 <.4 Estradiol 598 <.1 2aDHP 586 <.1 Semen Volume 135 >.5 ;\ to 12 months Pl. hormones after vasectomy FSH 991 <.1 LH 99 >.25 Testosterone 94 <.5 Estradiol 999 <.1 2aDHP 97 < to 3 years after Pl. hormones vasectomy FSH 36 <.14 LH 339 >.5 Testosterone 36 >.5 Estradiol 378 >.5 2aDHP 272 <.2 a Probability that the rhythm amplitude does not differ from zero. beach M and A is given with its respective units: semen, volume in ml, total sperm count (1 6 ), sperm motility in percentage; plasma hormones: FSH and LH in miu /ml, testosterone, estradiol, and 2a-dihydroprogesterone in ng/1 mi. c A (amplitude of rhythm) represents '12 of total variability, Annual adjusted Amplitude Acrophase </Jd Mean:Mesor Amplitude Ab,c,d ±for months± M ± Mesor A as% M 3 95% CI days 2.35 ± ±.23 Apr 1 ± 47 d ± ± 29.4 Mar 8 ± 61 d 44.7 ± ± 5.48 Mar 14 ±51 d 3.12 ± ±.39 Sep 2 ± 25 d 2.35 ± ±.15 Oct 26 ± 14 d 47.3 ± ± 33. Nov 16 ± 75 d 3.49 ± ±.21 Sep 12 ± 34 d 67.4 ± ± 5.4 Jun 6 ± 18 d 2.54 ± ± ±.26 Nov 9 ± 25 d 2.68 ± ±.15 Jan 31 ± 62 d 5.32 ± 1.4 Nov 3.32 ± ±.16 Dec 6 ± 61 d 69.6 ± ± 4.38 Jun 16 ± 39 d 3.2 ± ±.46 Nov 7 ±55 d 2.87 ±.7 Oct ± 12.7 Nov 3.35 ± ± ± 8. Jun 9 ± 35 d i.e., the peak-to-trough difference. d A and each are given with their respective 95% confidence intervals (CI) when rhythm detection is at P <.5; </> is the peak time of the best fitting cosine curve. Note: A is given only when rhythm detection is statistically significant; </> estimates without confidence limits are provided when.5 < P <.1. amplitude A, zero A (no rhythm detection) fort, LH, and E 2, and a decreased annual A for FSH and a comparable A for 2a-DHP; (3) in consideration of the acrophase (</>), unchanged</> location for annual rhythms, which remained detectable or close to statistical significance. DISCUSSION The first point to be discussed relates to the reanalyses of the time series on endocrine variabies. Both conventional and cosinor methods provided similar results. On a group basis, mean plasma T and LH levels exhibited small but statistically significant increases during the 3 years following vasectomy, while the mean levels of plasma FSH, E 2, and 2a-DHP were unaltered. However, when annual rhythms were considered, new facts emerged from cosinor analyses. In particular, the plasma T, LH, and E 2 group annual rhythms were obliterated after vasectomy, while the amplitude of the annual rhythm of plasma FSH was reduced by almost 48%. Such large decreases in the annual amplitude of rhythms resulted in a diminution of the variation in values over the year. This effect of vasectomy appears to explain why conclusions, which were based only on the consideration of the span of time after surgery/ 2 do not differ from those which take into account, in addition, the time of year of sampling. We confirm that changes in T and LH levels after vasectomy remained within the respective normal ranges. A new issue arising from our reanalyses of the data for bioperiodicities relates to the clinical importance of the obliteration of annual rhythms, such as in plasma T and LH. Circannual rhythms in T have been reported in samples of nonvasectomized normal males Also, such a rhythm of plasma gonadotropins has been detected in both elderly male and female subjects. 18 However, lack of rhythmicity also has been reported in plasma T in nonvasectomized healthy adult Reinberg et al. Seasonal rhythms in men 313

6 Ill z O ffi~ '- 1;&'45 Ill E ~ 3.5 at Cl c a: :; 2.5 Ill i1oo..... ' at 8 D. :z: Cl! 4 Cll J TIME OF YEAR (MONTHS) F M A M J J A S N D Figure 2 Annual changes in plasma hormones. From top to bottom, chronograms of FSH and LH (in miu/ml); T, E 2, and 2a-DHP (in ng/1 ml). Number of samples for calculating each monthly mean ""' 49. males.19 At present, the obliteration of the annual rhythm of T and LH is not considered an adverse effect of vasectomy. The second point deals with annual changes in plasma hormone levels before vasectomy. The annual c/j of plasma T occurred in autumn in the present study; this is in agreement with the findings of other studies that relied on either single4 or several (six) samples/24 hours daily at regular intervals over a 1-year period. 3 However, the existence of a second peak in March (Fig. 2) suggests the possibility of other rhythmic components in addition to the circannual one, although this was not validated by cosinor analyses with a trial period of 6 months. As in other studies, annual changes were validated for both LH and FSH ; however, the acrophase differed from that found here. 314 Reinberg et al. Seasonal rhythms in men In our sample, it occurred in late autumn to early winter, rather than in late winter and early spring This difference might be associated with the geographic location where the study was conducted in relationship to the local photoperiod.1 2 This point will be reconsidered with regard to annual changes in semen characteristics. An annual rhythm of 2a-DHP in human beings, regardless of sex, has never been reported, while that of E2 has been reported only once before in a study conducted in Finland.19 In this investigation, highest levels occurred between April and June. It seems that, in human males, a prominent fraction of E2 is derived from the enzymatic conversion oft in peripheral tissues. 21 This is consistent with the fact that the annual change in T and E2 both exhibited their major peak in autumn. Large-amplitude annual changes in 2a-DHP, with its peak in June, differ clearly from the other investigated variables. We have no explanation for this rather prominent annual rhythmicity. This leads to the third point of the discussion. Annual changes in animal species, including humans, are viewed as adaptive phenomena related to reproduction According to Boissin et al.,22 it seems that different reproductive "strategies" exist in mammalian species to achieve birth during the spring. The best candidate for a synchronizing signal of the mating season seems to be the photoperiod, whether mating occurs in spring or autumn. The photoperiod may convey important signals to our species as well. This point of view is based on indirect evidence, such as annual cycles in human births with peak times that vary according to the latitude of the population considered2 as well as annual cycles of T and sex-related activities of human males.9 1 Annual changes in semen characteristics, therefore, could represent an aspect of the broad adaptive phenomena of our species to seasonal changes in environmental conditions. In the present study, for both total sperm count and motility, the major peak occurred in spring and a secondary peak during autumn. Total sperm count was found to peak at the same time of year by Mortimer et al.,23 while MacLeod and Heim24 found a peak in winter and Hotchkiss25 in spring. Based on the analyses of more than 4 samples from Houston, Tjoa et al. 26 found sperm count to be lowest during August and greatest in February. Although it is tempting to ascribe the seasonal variation in sperm count to seasonal variation in ambient temperature, such an explanation may not be valid. For example, with regard to the Houston study, in particular, the summertime reduction in sperm count is not believed to be due to environ- Fertility and Sterility

7 mental heat, even though it is especially great in this southern city, since most participants in this study resided, commuted, and worked in temperature-controlled environments achieved by air conditioning in the summer and heating in the winter It is likely that the production (and presumably the motility) of sperm varies seasonally with two peaks, a larger one in spring and a smaller one in fall. With regard to the present study, the peak time of sperm production (total count) and motility, as well as plasma T and E 2, occurred at the same time of the year. Despite the fact that a coincidence in time cannot be taken on its own as evidence of a causal relationship, the hypothesis of a common factor driving annual changes in both sperm and androgen production has to be considered. Whatever the case, the existence of such annual changes requires consideration not only for a better understanding of predictable variability in human reproductive phenomena, but for an improved design of experimental protocols involving the investigation of reproductive phenomena. Acknowledgments. We are grateful to Drs. Paul Robel (Directeur des Recherches, CNRS, Paris) and Yvan Touitou (Professor, Department Biological Chemistry, University of Paris VI) for their advice and comments in discussing the data and findings reported in this manuscript. REFERENCES 1. Smith KD, Tcholakian RK, Chowdhury M, Steinberger E: An investigation of plasma hormone levels before and after vasectomy. Fertil Steril 27:145, Smith KD, Tcholakian RK, Chowdhury M, Hsi BP: Endocrine studies in vasectomized men. In Vasectomy, Edited by IH Lepow, R Crozier. New York, Academic Press, 1979, p Reinberg A, Lagoguey M, Chauffournier JM, Cesselin F: Circannual and circadian rhythms in plasma testosterone in five healthy young Parisian males. Acta Endocrinol (Copenh) 8:732, Smals NH, Kloppenborg PWC, Benroad TJ: Annual cycle in plasma testosterone levels in man. J Clin Endocrinol Metab 42:979, Reinberg A, Lagoguey M, Cesselin F, Touitou Y, Legrand JC, de la Salle A, Antreassian J, Lagoguey A: Circadian and circannual rhythms in plasma hormones and other variables of five healthy young human males. Acta Endocrinol (Copenh) 88:417, Touitou Y, Fevre M, Bogdan A, Reinberg A, De Prins J, Beck H, Touitou C: Patterns of plasma melatonin with aging and mental condition: stability of nyctohemeral rhythms and differences in seasonal variations. Acta Endocrinol (Copenh) 16:145, Reinberg A, Lagoguey M: Annual endocrine rhythms in healthy young adult males. In Environmental Endocrinology, Edited by I Assenmacher, DS Farner. Berlin, Springer Verlag, 1978, p Touitou Y, Carayon A, Reinberg A, Bogdan A, Beck H: Differences in the seasonal rhythmicity of plasma prolactin in elderly human subjects: detection in women but not in men. J Endocrinol 96:635, Reinberg A, Lagoguey M: Circadian and circannual rhythms in sexual activity and plasma hormones (FSH, LH, testosterone) of five human males. Arch Sex Behav 7:13, Smolensky MH, Reinberg A, Bicakova-Rocher A, Sanford J: Chronoepidemiological search for circannual changes in the sexual activity of human males. Chronobiologia 8:217, Mumford SD: Vasectomy: The Decision Making Process. A Guide for Promoters. San Francisco, San Francisco Press, 1977, p Smith KD, Tcholakian RK, Chowdhury M, Steinberger E: Rapid oscillations in plasma levels of testosterone, luteinizing hormone and follicle-stimulating hormone in men. Fertil Steril 25:965, Rodriguez-Rigau LJ, Tcholakian RK, Smith KD, Steinberger E: In vitro steroid biosynthetic studies in human testis. I. Effect of estrogen on progesterone metabolism. Steroids 3:729, Hotchkiss J, Atkinson LE, Knobil E: Time course of serum estrogen and luteinizing hormone (LH) concentrations during the menstrual cycle of the rhesus monkey. Endocrinology 89:177, Nelson W, Tong YK, Jueng-Kuen L, Halberg F: Methods for cosinor rhythmometry. Chronobiologia 6:35, De Prins J, Cornelissen G, Malbecq W: Statistical procedures in chronobiology and chronopharmacology. Ann Rev Chronopharmacol 2:27, Nicolau GY, Lakatua D, Sackett-Lundeen L, Haus E: Circadian and circannual rhythms of hormonal variables in elderly men and women. Chronobiol Int 1:31, Touitou Y, Lagoguey M, Bogdan A, Reinberg A, Beck H: Seasonal rhythms of plasma gonadotropins: their persistence in elderly men and women. J Endocrinol 96:15, Martikainen H, Tapanainen J, Vakkuri, Leppaluoto HI: Circannual concentrations of melatonin, gonadotrophins, prolactin and gonadal steroids in males in a geographical area with a large annual variation in daylight. Acta Endocrinol (Copenh) 19:446, Aschoff J: Annual rhythms in man. In Handbook of Behavioural Neurology, Vol4, Edited by J Aschoff. London, Plenum, 1981, p Horton R: Testicular steroid secretion, transport and metabolism. In Endocrinology, Edited by LJ De Groot. New York, Grune & Stratton, 1979, p Boissin J, Jallageas M, Assenmacher I: The annual cycle of testicular activity in birds and mammals. In Rhythmes et Reproduction, Edited by R Ortavant, A Reinberg. Paris, Masson, 198, p Mortimer D, Templeton AA, Lenton EA, Coleman RA: Annual patterns of human sperm production and semen quality. Arch Androl 1:1, MacLeod J, Heim LM: Characteristics and variations in semen specimens in 1 normal young men. J Urol54:474, Hotchkiss RS: Factors in stability and variability of semen observations on 64 successive samples from 23 men. J Urol 45:875, Tjoa WK, Smolensky MH, Hsi BP, Steinberger E, Smith K: Circannual rhythms in human sperm count revealed by serially independent sampling. Fertil Steril 38:454, 1982 Reinberg et al. Seasonal rhythms in men 315