Reproductive Toxicology, Vol. 6, pp. 297-307, 1992
0890-6238/92 $5.00 + .00
Printed in the U.S.A. All rights reserved.
• Original Contributions
DOES EARLY EXPOSURE TO MATERNAL SMOKING AFFECT FUTURE FERTILITY IN ADULT MALES? JENNIFER M. RATCLIFFE,* BETH C. GLADEN,t ALLEN J. WILCOX,* a n d ARTHUR L. HERBST,J~ *Epidemiology Branch and tStatistics and Biomathematics Branch, Division of Biometry and Risk Assessment, National Institute of Environmental Health Sciences, Research Triangle Park, North Carolina; ~:Department of Obstetrics and Gynecology, The Chicago Lying-Hospital, The University of Chicago, 5841 South Maryland Avenue, Chicago, Illinois Abstract - - Animal data suggest that prenatal exposure to certain tobacco smoke components such as nicotine may affect the development of the male gonadal axis, which may in turn affect future adult fertility. There are no
previous epidemiologic studies on the potential effects of early (prenatal and childhood) exposure to maternal smoking on the reproductive system in adult male offspring. To investigate this question, we used data from a follow-up study of reproductive function and fertility among young adult sons of mothers who had participated in a randomized clinical trial of diethylstilbestrol use during pregnancy. We observed no significant effects of early exposure to maternal smoking on conventional semen characteristics, hormone levels (follicle stimulating hormone [FSH], luteinizing hormone [LH] and testosterone), urogenital abnormalities and diseases, or perceived infertility problems. Current active smoking by the men was, however, associated with a significant decrease in the percentage of sperm with normal morphology. Key Words: maternal smoking; smoking, male; infertility, male; semen quality; sperm; prenatal exposure, delayed effects; urogenital disorders.
Nicotine, cotinine, carbon monoxide (as carboxyhemoglobin), and other smoke components readily cross the human placenta and are found in cord blood in concentrations similar to those in maternal blood (1). There have been a large number of epidemiologic and toxicologic studies of the effects of prenatal exposure to maternal smoking or to tobacco smoke components on fetal development, including uterine growth retardation, prematurity, spontaneous abortion, perinatal mortality, postnatal growth, and behavioral and intellectual development. Comparatively little attention has been paid, however, to the question of whether prenatal exposure to maternal smoking (with or without subsequent passive postnatal exposure) might affect subsequent fertility in adult offspring. Evidence from animal studies suggests that such an effect in males may be biologically
plausible, but there are very few epidemiologic studies of this question to date.
Address correspondence to Dr. Jennifer M. Ratcliffe, National Institute of EnvironmentalHealth Sciences,Epidemiology Branch, Division of Biometry and Risk Assessment, Mail Drop A3-05, P.O. Box 12233, Research Triangle Park, North Carolina 27709 Received 16 July 1991;Revision received22 October 1991;Accepted 28 October 1991.
Benzo(a)pyrene. Exposure of mice to benzo (a)pyrene from gestation day (GD) 7 to 16 resulted in dose-dependent effects in male offspring, including a decrease in testicular weight, atrophy ofseminiferous tubules, altered spermatogenesis, and a significant in-
Animal st udies Very few investigators have attempted to expose pregnant animals to tobacco smoke by inhalation; most of the evidence comes from exposure (by oral administration or injection) to single tobacco smoke components. Tobacco smoke. An early study of prolonged exposure of male and female rats to tobacco fumes for successive generations resulted in a progressive decrease in the number of viable offspring from one generation to the next (2). Adult male rats prenatally exposed to tobacco fumes also showed a significant reduction in copulations and ejaculations (3).
297
298
Reproductive Toxicology
crease in both preimplantation and postimplantation fetal loss (4).
Dimethylbenzanthracene. Prenatal exposure to dimethylbenzanthracene (DMBA) had similar adverse effects on testicular function and fertility in male offspring (5,6). Serum LH levels were also found to be elevated 2 to 3-fold, but FSH levels were not affected. The most sensitive period of dosing of the pregnant mice was found to be GD 11 to 13, which corresponds to the time when migration of the primordial germ cells to the germinal ridges and initial sexual differentiation is occurring (7). Nicotine. Male rodents may be considerably more sensitive to the effects of a given dose of nicotine than females. In treated adults, the effect of nicotine on reproductive capacity is much greater in males than in females (8). The ratio of male to female offspring is reduced in a dose-dependent fashion after nicotine exposure of the mother (9-12); in addition, behavior (e.g., avoidance and learning behavior) and motor activity are often affected in exposed male but not female offspring (9,10,13). Hormone levels may also be affected. Fetal testosterone levels in rats peak at gestational day 18; this is thought to be important in determining the sexual differentiation of the rodent reproductive tract and brain. Prenatal exposure of rats to nicotine throughout pregnancy (by continuous subcutaneous infusion at a dose equivalent to approximately 2 to 4 times baseline levels seen in maternal and cord blood of smokers) suppresses this peak (14). Prenatal exposure of rats to 0.5 mg nicotine/kg/day ip on GD 3 to 21 was found to significantly decrease plasma testosterone in 70-day-old male offspring ( 15); this change was accompanied by a change in sexual performance (decreased number of mountings, intromissions, and ejaculations) in these males. The mechanisms by which nicotine and other tobacco smoke components may potentially affect the development of the fetal male reproductive system in animals have not been clearly elucidated. Nicotine may act prenatally by directly acting on the testes, pituitary, or hypothalamus or indirectly by altering the development and maturation of central and peripheral catcholamine neurotrasmitter systems ( 16-18). At least in some species, catecholamines are thought to play an important role in the development and maintenance of the male hypothalamo-pituitarytesticular axis, such as in the control of gonadotropin releasing hormone (GnRH) secretion from the hypothalamus (19). Catecholamines also affect fetal Leydig cell steroidogenesis (20); nicotine and cotinine
Volume 6, Number 4, 1992
have also been shown to directly reduce testosterone levels in isolated Leydig cells (21), but it is not clear whether this occurs in vivo. It has not been established whether other tobacco smoke components, such as benzo(a)pyrene and DMBA, affect fertility prenatally by direct action at the testicular site or by altering the development of the hypothalamo-pituitary-gonadal axis (4).
Human studies Evidence of an effect of prenatal exposure to maternal smoking on adult male fertility from human studies is sparse. In a prospective study of couples (22), prenatal exposure to maternal smoking among the wives significantly decreased fecundability (the per cycle probability of conception). Prenatally exposed husbands were not affected (C. Weinberg, personal communication), although data for husbands were available for only half the cohort. In a retrospective study of fecundability (23), no relationship between either the wives' or husbands' prenatal exposure and fecundability was observed. In a small subset of 26 men who reported having had a semen analysis, however, prenatal exposure was related to poor semen quality (23). In the absence of clinical verification of the semen analysis results and in view of the possibility of response bias, this finding, by itself, clearly provides insufficient evidence for an association, but suggests the possibility of some impairment of male reproductive function. In view of the paucity of epidemiologic data on this topic in comparison with experimental studies showing adverse effects, we used data from an available data set to investigate the hypothesis that early (prenatal and childhood) exposure to maternal smoking affects semen quality (as measured by conventional semen characteristics), hormone levels (FSH, LH, and testosterone), the prevalence of urogenital abnormalities and diseases, and perceived infertility problems in adult male offspring. We also investigated the secondary hypothesis that direct smoking by adult males affects these parameters. MATERIALS AND M E T H O D S Data were used from follow-up studies of the sons of women who participated in the first randomized clinical trial of diethylstilbestrol (DES) for pregnancy treatment conducted between 1950 and 1952, and from a follow-up study of the mothers themselves. The selection and characteristics of the original cohort of women who participated in the clinical trial of DES has been described by Dieckmann and colleagues (24). A total of 2162 women who regis-
Maternal smokingand male fertility • J. M. RATCLIFFEETAL. tered consecutively at the University of Chicago Lying-In Hospital at between 6 and 20 weeks of pregnancy was selected for the trial, with every other patient assigned to the control group. By the end of the trial, a total of 22% of the women had been excluded (due to moving, failure of compliance, abortion, etc.); 840 women had been given DES, starting when they registered and ending at the 35th week or delivery if earlier, and 806 women had been given a placebo over the same time course of pregnancy (24). As far as can be ascertained, no special instructions or advice were given to these women about the use of any other medications, tobacco, alcohol, or other substances during pregnancy. A total of 823 sons were born to these 1646 mothers (Figure 1). D a t a on s o n s
Starting in 1974 and ending in 1982, a follow-up study of the sons was conducted (Figure 1). Approximately 75% of the sons were eventually examined. The study was conducted in two phases. In the first phase, from 1974 to 1977, a physical examination of the urogenital tract was conducted and urine and prostate cytology was done (25). Serum assays were performed for luteinizing hormone (LH) follicle stimulating hormone (FSH), and testosterone (T) and a short questionnaire regarding sexual development, urogenital infections, and urogenital surgery was adN u m b e r o f Sons in Study
+
1950-1952
823
(420 DES Exposed 403 Unexposed)
1974-1982
615
Physical Exam
1974-1982
292
Semen Analysis
/ 161 UNEXPOSED
~
~ 68 EXPOSED
EarlyExposureto Maternal Smokln 63 UNCLASSIFIED
Fig. 1. Follow-up study of sons of mothers in clinical trial of DES in pregnancy.
299
ministered. Data from hospital records of the sons' condition at birth were also abstracted. In the second phase, from 1976 to 1981, semen analysis and, later on, a more detailed self-administered questionnaire regarding medical history, smoking status, marital status, and reproductive experience were added to the protocol, and more detailed data from the physical examination were recorded (26). A total of 615 men participated in at least one phase of the study (Figure 1). Subjects were initially asked to provide three consecutive semen samples; however, a majority of subjects provided only one or two, while some provided up to 8 samples over a period of one or more years. A total of 292 men (47.5% of those who underwent a physical examination) provided at least one semen sample; this was 40% of the original DES-exposed cohort and 30.8% of the control cohort. For the purpose of the present investigation, we confined our analysis to the 292 men who had at least one semen analysis, and we analyzed semen quality data for the first semen sample only to minimize the possibility of response bias. [Note: This sample includes 24 men examined after the last published report on this cohort (27).] The protocol for the semen analysis has been described by Gill and colleagues (25). Subjects were asked for 3 days' abstinence prior to collection of the sample by masturbation into a sterile glass jar during the clinic visit. Actual abstinence times and the length of time between sample collection and analysis were not recorded on the computerized forms; in the case of sample age, however, it is unlikely that small variations would affect viability and motility measurements since all samples were collected on site and were analyzed shortly thereafter (GFB Schumacher, personal communication). Sperm viability was evaluated by the method of Eliasson and Treichl (28). The percentage of motile sperm was measured by observing a minimum of 200 sperm under a phase contrast microscope. Motility grade was determined according to the method of Eliasson (29): sperm were graded from 0 (no movement of any sperm) through 1, 2, 3, and 4 (good forward progression); 2 + is considered the border between passable and poor motility. Eliasson's scoring system for the overall quality of the sample, which combines scores for sperm concentration, percent motile, motility grade, and morphology, was also used (29). Sperm morphology was classified according to the system of Amelar (30) and was performed by one technician throughout the study. Sperm concentration was determined according to Freund (31 ). In the case of the physical examinations, data for
300
Reproductive Toxicology
the first examination were used in the initial analysis; ifa conflicting diagnosis was recorded at a subsequent examination, further analyses were conducted incorporating these findings to determine if this altered the initial results. Data from the short questionnaire were available for 276 (95%) of the 292 men. Data from the more detailed questionnaire given in the second phase of the study were available for a total of 184 (63%) of the 292 men. In 55% of cases, questionnaires were administered on the same day as the semen analysis was performed; a further 9% of questionnaires were administered within one year and 34% within 1 to 7 years of the semen analysis. Data on son's smoking was available for only 52% of the men and was generally collected some time after the semen or hormone analysis. For these men, it was possible to determine smoking status at the time of semen analysis or physical examination for never smokers and for current smokers (from the age at which smoking started). No data on quitting dates were available for ex-smokers, however, so that some misclassification is possible ifa subject quit after rather than before his semen or physical examination. Data on mothers Data on the health status, socioeconomic status, parity, gravidity, age at pregnancy, and current smoking habits for each of the son's mothers were available from relevant medical records and a follow-up interview study of the mothers conducted from 1978 to 1981, 27 to 30 years after their clinical trial pregnancy. Smoking status during pregnancy was established for 229 (78%) of the sons' mothers. No data were collected on the amount or duration of smoking during pregnancy, or on exposures such as alcohol, caffeine, or occupational hazards. Only 15 women (6.5%) reported different smoking status during pregnancy and at follow-up. Thus, because most of the mothers did not change their smoking habits after pregnancy, we cannot distinguish between prenatal and childhood exposure of the sons to maternal smoking. No data on the smoking behavior of the sons' fathers were available. All data were analyzed first either by t tests, chi square tests and, for motility grade and Eliasson scores, by the Wilcoxon test for ordered scores. Linear regression analyses were also performed for each of the semen characterisitics and hormone data to examine any relationship between these and early exposure to maternal smoking, adjusting for sons' smoking staus at interview and his prenatal DES exposure status. Since it is biologically plausible that DES exposure might modify a maternal smoking effect, interactions betwen DES exposure and smoking
Volume 6, Number 4. 1992
(both mothers' and sons') were also examined for each outcome variable. RESULTS One hundred sixty-one men had been classified by the original investigators as having no prenatal smoke exposure and 68 as exposed. The prenatal smoke exposure status of the remaining 63 could not be ascertained. As an indirect verification of the classification of prenatal smoke exposure, we compared birth weights among the three groups. Exposed sons had a mean birth weight of 3187 g and unexposed sons a mean of 3456 g, consistent with the known effect of maternal smoking on birth weight (32). (Sons of unknown prenatal smoke exposure status had a mean birth weight of 3201 g, close to that of the exposed). There were no statistically significant differences between the exposed and unexposed groups for personal characteristics and medical status of either the mothers or the sons with respect to the pregnancy and birth (Table 1). The groups of mothers and sons were similar for most variables measured except for a higher prevalence of certain medical problems (high blood pressure, thyroid conditions, and one case of cancer) prior to pregnancy among nonsmoking mothers. There was only modest concordance between mothers' and sons' smoking habits. Early exposure to maternal smoking versus semen quality and hormone levels There were no significant differences for mean values of any of the conventional semen characteristics of sperm concentration, semen volume, sperm concentration per ejaculate, and percentage of viable and motile sperm between men who had early exposure to maternal smoking compared with those who were unexposed (Table 2). Also, no differences were observed between either of these groups and unclassified subjects. None of these findings was altered by adjusting for sons' smoking or DES exposure status. Analysis of interaction effects showed a significant interaction between DES exposure and exposure to maternal smoking only for sperm motility. Exposure to maternal smoking was associated with a significant decrease in average sperm motility among men not exposed to DES (61% compared with 67%; P = 0.020) but not among the DES-exposed men (63% compared with 59%; ns). Sons' smoking and a DESsons' smoking interaction term were also included in the model. The overall percentage of oligospermic men appeared higher among men exposed to maternal
301
Maternal smoking and male fertility • J. M. RATCLIFFE ET AL. T a b l e I. D e m o g r a p h i c characteristics a n d medical status at p r e g n a n c y / b i r t h o f m o t h e r s a n d sons
Characteristic Mothers Race (%) White Black Other Median family income ($100s) Mean gravidity Mean parity Average age of mother at delivery (yrs) Significant medical conditions prior to pregnancy
(%)a
Pregnancy complications (%)8 Sons Mean gestational age at delivery % Premature birth % Exposed to DES in preg. Adult smoking status c current smoker (%) ex-smoker (%) never smoker (%) Mean age, 1st semen analysis
Nonsmokers in pregnancy (n = 161)
Smokers in pregnancy (n = 68)
Unclassified (n = 63)
96 3 1 43 2 2 28 12
94 4 2 43 2 2 26 4
92 8 0 44 2 2 28 10
17
15
16
40 1 57
39 2 60
39 6 56
35 15 51 25
47 16 38 24
29 9 62 25
No significant differences (two-tailed t or chi square tests) found for any characteristic between any group. aOne or more of high blood pressure, thyroid condition, or cancer. bOne or more of spotting, pre-eclampsia, eclampsia, or "other complications." cSmoking data available for only 52% of subjects.
smoking than among unexposed men (15% compared with 9%), but this difference was not significant before or after adjustment for DES exposure and sons' smoking (Table 2). Again, a similar interaction between DES exposure and maternal smoking was observed. Exposure to maternal smoking was associated with a significant increase in oligospermia among men not exposed to DES (22% compared with 3%; P = 0.030) but not among DES-exposed men (10% compared with 13%; ns). It was not possible to fit sons' smoking in this model. Men unclassified with respect to exposure to maternal smoking had a significantly lower percentage of sperm with normal morphology compared with both exposed and unexposed men, both before and after adjustment for son's smoking status and DES exposure status and interaction effects (Table 3). This was not due to a significant prevalence of any one type of abnormality and is uninterpretable in terms of possible biologic significance. The percentage of men with less than 40% normal morphology (using Eliasson's cut-off level for "severely pathological" specimens [29]) were closely comparable (Table 3); further, there were no significant differences between groups on the overall Eliasson scores. No differences were observed between groups for mean levels of LH, FSH, or testosterone before or after adjustment for DES exposure, sons' smoking, and interactions (Table 4); there were no significant
differences in the percentages of men classified by the Chicago investigators as having abnormal values. No differences between groups were observed with respect to abnormal fructose levels in seminal plasma or findings on urinary tract cytology before and after prostatic massage; a total of 4 men had unusual findings on prostate fluid cytology. No testicular biopsies were performed.
Early exposure to maternal smoking versus urogenital abnormalities and diseases With respect to first physical examinations, there were no significant differences between groups on testicular size, the prevalence of capsular induration of the testis, appendix of the testis, or history of cryptorchidism (Table 5). The low prevalence of abnormalities did not permit exploration of the effects of confounders and interactions. Similarly, there were no differences between groups for epididymal abnormalities (epididymal nodules, appendix, cysts, or other abnormalities), the presence of hydrocele, varicocele, or other scrotal abnormalities or prostrate abnormalities. External genitalia were generally normal, with 3 subjects exhibiting abnormal breast development and none abnormal body hair or pubic hair development; only 3% of all subjects had either hypospadias, stenotic meatus, or other abnormalities of the penis. The data were also analyzed with respect to abnormalities noted on any (that is, including re-
302
Reproductive Toxicology
Volume 6, Number 4, 1992
T a b l e 2. S e m e n characteristics in a d u l t m e n in relation to early e x p o s u r e to m a t e r n a l s m o k i n g
Characteristic Semen volume (mL) Mean (SD) Range Sperm concentration Mean (SD) (million/mL) Range Sperm count per ejaculate (millions) Mean (SD) Range % Mean with _< = 20 million sperm/mL (oligospermia) % Viable sperm Mean (SD) a'b Range % Motile sperm Mean (SD) b Range % Men with _< = 30% motile sperm b Total motile sperm b per ejaculate (millions) Mean (SD) Range Motility grade d (%) 2-2.5 (poor) 3-3.5 (moderate) 4 (good)
Unexposed (n = 161)
Exposed (n = 68)
Unclassified (n = 63)
3.26 (1.45) 0.2-8.0
3.18 (1.75) 0.7-9.9
2.89 (1.83) 0.4-9.9
102.5 (80.6) 0-500
100.1 (86.6) 4-470
117.3 (145.0) 0-825
315.7 (242.9) 0-1250.0 8.7
310.2 (278.0) 7-1275.0 14.7
285.3 (329.1) 0-1750 9.5
70.4 (12.1) 32-96
68.0 (13.0) 26-89
66.7 (13.2) 16-89
62.3 (14.6) 4-85 4.4
62.2 (13.8) 18-87 4.4
59.4 (14.9) 6-86 4.8
208.9 (165.9) 1-888
203.1 (191.4) 2-931
173.6 (181.2) 3-1022
12.0 39.0 49.1
8.8 41.2 50.0
9.7 54.8 35.5
an = 132 (unexposed), n = 59 (exposed), n = 52 (unclassified). bExcluding 2 unexposed and 1 unclassified azoospermic men. All differences not significant on two-tailed t tests or chi square; overall motility grade not significantly different between groups using Wilcoxon test for ordered scores.
peat) examinations or questionnaires, since there were cases in which abnormalities were diagnosed on later examinations. Again, no significant differences were found between the groups. The groups did not differ significantly with respect to their history of specific urogenital infections of the scrotum, epididymis, prostate, kidney, and bladder or gonorrhea, syphilis, or urinary tract infections, with the exception of penile infections, which were reported by approximately 8% of the unexposed and unclassified men but none of the exposed men. We consider it unlikely that this association has biologic significance, however. Reported age at puberty (at approximately 13 years), at first ejaculation (at approximately 14 years), and age at first intercourse (at approximately 18 years) did not differ among groups. Son's smoking status versus semen quality and hormone levels Current smokers had a significantly lower percentage of sperm with normal morphology compared with nonsmokers and men with unknown smoking status, both on a crude t test and after adjusting for DES exposure, exposure to maternal smoking, and
interaction terms (Table 6). Ex-smokers had an average percentage of normal forms that was intermediate between that of current and nonsmokers. The proportion of men with
0890-6238/92 $5.00 + .00
Printed in the U.S.A. All rights reserved.
• Original Contributions
DOES EARLY EXPOSURE TO MATERNAL SMOKING AFFECT FUTURE FERTILITY IN ADULT MALES? JENNIFER M. RATCLIFFE,* BETH C. GLADEN,t ALLEN J. WILCOX,* a n d ARTHUR L. HERBST,J~ *Epidemiology Branch and tStatistics and Biomathematics Branch, Division of Biometry and Risk Assessment, National Institute of Environmental Health Sciences, Research Triangle Park, North Carolina; ~:Department of Obstetrics and Gynecology, The Chicago Lying-Hospital, The University of Chicago, 5841 South Maryland Avenue, Chicago, Illinois Abstract - - Animal data suggest that prenatal exposure to certain tobacco smoke components such as nicotine may affect the development of the male gonadal axis, which may in turn affect future adult fertility. There are no
previous epidemiologic studies on the potential effects of early (prenatal and childhood) exposure to maternal smoking on the reproductive system in adult male offspring. To investigate this question, we used data from a follow-up study of reproductive function and fertility among young adult sons of mothers who had participated in a randomized clinical trial of diethylstilbestrol use during pregnancy. We observed no significant effects of early exposure to maternal smoking on conventional semen characteristics, hormone levels (follicle stimulating hormone [FSH], luteinizing hormone [LH] and testosterone), urogenital abnormalities and diseases, or perceived infertility problems. Current active smoking by the men was, however, associated with a significant decrease in the percentage of sperm with normal morphology. Key Words: maternal smoking; smoking, male; infertility, male; semen quality; sperm; prenatal exposure, delayed effects; urogenital disorders.
Nicotine, cotinine, carbon monoxide (as carboxyhemoglobin), and other smoke components readily cross the human placenta and are found in cord blood in concentrations similar to those in maternal blood (1). There have been a large number of epidemiologic and toxicologic studies of the effects of prenatal exposure to maternal smoking or to tobacco smoke components on fetal development, including uterine growth retardation, prematurity, spontaneous abortion, perinatal mortality, postnatal growth, and behavioral and intellectual development. Comparatively little attention has been paid, however, to the question of whether prenatal exposure to maternal smoking (with or without subsequent passive postnatal exposure) might affect subsequent fertility in adult offspring. Evidence from animal studies suggests that such an effect in males may be biologically
plausible, but there are very few epidemiologic studies of this question to date.
Address correspondence to Dr. Jennifer M. Ratcliffe, National Institute of EnvironmentalHealth Sciences,Epidemiology Branch, Division of Biometry and Risk Assessment, Mail Drop A3-05, P.O. Box 12233, Research Triangle Park, North Carolina 27709 Received 16 July 1991;Revision received22 October 1991;Accepted 28 October 1991.
Benzo(a)pyrene. Exposure of mice to benzo (a)pyrene from gestation day (GD) 7 to 16 resulted in dose-dependent effects in male offspring, including a decrease in testicular weight, atrophy ofseminiferous tubules, altered spermatogenesis, and a significant in-
Animal st udies Very few investigators have attempted to expose pregnant animals to tobacco smoke by inhalation; most of the evidence comes from exposure (by oral administration or injection) to single tobacco smoke components. Tobacco smoke. An early study of prolonged exposure of male and female rats to tobacco fumes for successive generations resulted in a progressive decrease in the number of viable offspring from one generation to the next (2). Adult male rats prenatally exposed to tobacco fumes also showed a significant reduction in copulations and ejaculations (3).
297
298
Reproductive Toxicology
crease in both preimplantation and postimplantation fetal loss (4).
Dimethylbenzanthracene. Prenatal exposure to dimethylbenzanthracene (DMBA) had similar adverse effects on testicular function and fertility in male offspring (5,6). Serum LH levels were also found to be elevated 2 to 3-fold, but FSH levels were not affected. The most sensitive period of dosing of the pregnant mice was found to be GD 11 to 13, which corresponds to the time when migration of the primordial germ cells to the germinal ridges and initial sexual differentiation is occurring (7). Nicotine. Male rodents may be considerably more sensitive to the effects of a given dose of nicotine than females. In treated adults, the effect of nicotine on reproductive capacity is much greater in males than in females (8). The ratio of male to female offspring is reduced in a dose-dependent fashion after nicotine exposure of the mother (9-12); in addition, behavior (e.g., avoidance and learning behavior) and motor activity are often affected in exposed male but not female offspring (9,10,13). Hormone levels may also be affected. Fetal testosterone levels in rats peak at gestational day 18; this is thought to be important in determining the sexual differentiation of the rodent reproductive tract and brain. Prenatal exposure of rats to nicotine throughout pregnancy (by continuous subcutaneous infusion at a dose equivalent to approximately 2 to 4 times baseline levels seen in maternal and cord blood of smokers) suppresses this peak (14). Prenatal exposure of rats to 0.5 mg nicotine/kg/day ip on GD 3 to 21 was found to significantly decrease plasma testosterone in 70-day-old male offspring ( 15); this change was accompanied by a change in sexual performance (decreased number of mountings, intromissions, and ejaculations) in these males. The mechanisms by which nicotine and other tobacco smoke components may potentially affect the development of the fetal male reproductive system in animals have not been clearly elucidated. Nicotine may act prenatally by directly acting on the testes, pituitary, or hypothalamus or indirectly by altering the development and maturation of central and peripheral catcholamine neurotrasmitter systems ( 16-18). At least in some species, catecholamines are thought to play an important role in the development and maintenance of the male hypothalamo-pituitarytesticular axis, such as in the control of gonadotropin releasing hormone (GnRH) secretion from the hypothalamus (19). Catecholamines also affect fetal Leydig cell steroidogenesis (20); nicotine and cotinine
Volume 6, Number 4, 1992
have also been shown to directly reduce testosterone levels in isolated Leydig cells (21), but it is not clear whether this occurs in vivo. It has not been established whether other tobacco smoke components, such as benzo(a)pyrene and DMBA, affect fertility prenatally by direct action at the testicular site or by altering the development of the hypothalamo-pituitary-gonadal axis (4).
Human studies Evidence of an effect of prenatal exposure to maternal smoking on adult male fertility from human studies is sparse. In a prospective study of couples (22), prenatal exposure to maternal smoking among the wives significantly decreased fecundability (the per cycle probability of conception). Prenatally exposed husbands were not affected (C. Weinberg, personal communication), although data for husbands were available for only half the cohort. In a retrospective study of fecundability (23), no relationship between either the wives' or husbands' prenatal exposure and fecundability was observed. In a small subset of 26 men who reported having had a semen analysis, however, prenatal exposure was related to poor semen quality (23). In the absence of clinical verification of the semen analysis results and in view of the possibility of response bias, this finding, by itself, clearly provides insufficient evidence for an association, but suggests the possibility of some impairment of male reproductive function. In view of the paucity of epidemiologic data on this topic in comparison with experimental studies showing adverse effects, we used data from an available data set to investigate the hypothesis that early (prenatal and childhood) exposure to maternal smoking affects semen quality (as measured by conventional semen characteristics), hormone levels (FSH, LH, and testosterone), the prevalence of urogenital abnormalities and diseases, and perceived infertility problems in adult male offspring. We also investigated the secondary hypothesis that direct smoking by adult males affects these parameters. MATERIALS AND M E T H O D S Data were used from follow-up studies of the sons of women who participated in the first randomized clinical trial of diethylstilbestrol (DES) for pregnancy treatment conducted between 1950 and 1952, and from a follow-up study of the mothers themselves. The selection and characteristics of the original cohort of women who participated in the clinical trial of DES has been described by Dieckmann and colleagues (24). A total of 2162 women who regis-
Maternal smokingand male fertility • J. M. RATCLIFFEETAL. tered consecutively at the University of Chicago Lying-In Hospital at between 6 and 20 weeks of pregnancy was selected for the trial, with every other patient assigned to the control group. By the end of the trial, a total of 22% of the women had been excluded (due to moving, failure of compliance, abortion, etc.); 840 women had been given DES, starting when they registered and ending at the 35th week or delivery if earlier, and 806 women had been given a placebo over the same time course of pregnancy (24). As far as can be ascertained, no special instructions or advice were given to these women about the use of any other medications, tobacco, alcohol, or other substances during pregnancy. A total of 823 sons were born to these 1646 mothers (Figure 1). D a t a on s o n s
Starting in 1974 and ending in 1982, a follow-up study of the sons was conducted (Figure 1). Approximately 75% of the sons were eventually examined. The study was conducted in two phases. In the first phase, from 1974 to 1977, a physical examination of the urogenital tract was conducted and urine and prostate cytology was done (25). Serum assays were performed for luteinizing hormone (LH) follicle stimulating hormone (FSH), and testosterone (T) and a short questionnaire regarding sexual development, urogenital infections, and urogenital surgery was adN u m b e r o f Sons in Study
+
1950-1952
823
(420 DES Exposed 403 Unexposed)
1974-1982
615
Physical Exam
1974-1982
292
Semen Analysis
/ 161 UNEXPOSED
~
~ 68 EXPOSED
EarlyExposureto Maternal Smokln 63 UNCLASSIFIED
Fig. 1. Follow-up study of sons of mothers in clinical trial of DES in pregnancy.
299
ministered. Data from hospital records of the sons' condition at birth were also abstracted. In the second phase, from 1976 to 1981, semen analysis and, later on, a more detailed self-administered questionnaire regarding medical history, smoking status, marital status, and reproductive experience were added to the protocol, and more detailed data from the physical examination were recorded (26). A total of 615 men participated in at least one phase of the study (Figure 1). Subjects were initially asked to provide three consecutive semen samples; however, a majority of subjects provided only one or two, while some provided up to 8 samples over a period of one or more years. A total of 292 men (47.5% of those who underwent a physical examination) provided at least one semen sample; this was 40% of the original DES-exposed cohort and 30.8% of the control cohort. For the purpose of the present investigation, we confined our analysis to the 292 men who had at least one semen analysis, and we analyzed semen quality data for the first semen sample only to minimize the possibility of response bias. [Note: This sample includes 24 men examined after the last published report on this cohort (27).] The protocol for the semen analysis has been described by Gill and colleagues (25). Subjects were asked for 3 days' abstinence prior to collection of the sample by masturbation into a sterile glass jar during the clinic visit. Actual abstinence times and the length of time between sample collection and analysis were not recorded on the computerized forms; in the case of sample age, however, it is unlikely that small variations would affect viability and motility measurements since all samples were collected on site and were analyzed shortly thereafter (GFB Schumacher, personal communication). Sperm viability was evaluated by the method of Eliasson and Treichl (28). The percentage of motile sperm was measured by observing a minimum of 200 sperm under a phase contrast microscope. Motility grade was determined according to the method of Eliasson (29): sperm were graded from 0 (no movement of any sperm) through 1, 2, 3, and 4 (good forward progression); 2 + is considered the border between passable and poor motility. Eliasson's scoring system for the overall quality of the sample, which combines scores for sperm concentration, percent motile, motility grade, and morphology, was also used (29). Sperm morphology was classified according to the system of Amelar (30) and was performed by one technician throughout the study. Sperm concentration was determined according to Freund (31 ). In the case of the physical examinations, data for
300
Reproductive Toxicology
the first examination were used in the initial analysis; ifa conflicting diagnosis was recorded at a subsequent examination, further analyses were conducted incorporating these findings to determine if this altered the initial results. Data from the short questionnaire were available for 276 (95%) of the 292 men. Data from the more detailed questionnaire given in the second phase of the study were available for a total of 184 (63%) of the 292 men. In 55% of cases, questionnaires were administered on the same day as the semen analysis was performed; a further 9% of questionnaires were administered within one year and 34% within 1 to 7 years of the semen analysis. Data on son's smoking was available for only 52% of the men and was generally collected some time after the semen or hormone analysis. For these men, it was possible to determine smoking status at the time of semen analysis or physical examination for never smokers and for current smokers (from the age at which smoking started). No data on quitting dates were available for ex-smokers, however, so that some misclassification is possible ifa subject quit after rather than before his semen or physical examination. Data on mothers Data on the health status, socioeconomic status, parity, gravidity, age at pregnancy, and current smoking habits for each of the son's mothers were available from relevant medical records and a follow-up interview study of the mothers conducted from 1978 to 1981, 27 to 30 years after their clinical trial pregnancy. Smoking status during pregnancy was established for 229 (78%) of the sons' mothers. No data were collected on the amount or duration of smoking during pregnancy, or on exposures such as alcohol, caffeine, or occupational hazards. Only 15 women (6.5%) reported different smoking status during pregnancy and at follow-up. Thus, because most of the mothers did not change their smoking habits after pregnancy, we cannot distinguish between prenatal and childhood exposure of the sons to maternal smoking. No data on the smoking behavior of the sons' fathers were available. All data were analyzed first either by t tests, chi square tests and, for motility grade and Eliasson scores, by the Wilcoxon test for ordered scores. Linear regression analyses were also performed for each of the semen characterisitics and hormone data to examine any relationship between these and early exposure to maternal smoking, adjusting for sons' smoking staus at interview and his prenatal DES exposure status. Since it is biologically plausible that DES exposure might modify a maternal smoking effect, interactions betwen DES exposure and smoking
Volume 6, Number 4. 1992
(both mothers' and sons') were also examined for each outcome variable. RESULTS One hundred sixty-one men had been classified by the original investigators as having no prenatal smoke exposure and 68 as exposed. The prenatal smoke exposure status of the remaining 63 could not be ascertained. As an indirect verification of the classification of prenatal smoke exposure, we compared birth weights among the three groups. Exposed sons had a mean birth weight of 3187 g and unexposed sons a mean of 3456 g, consistent with the known effect of maternal smoking on birth weight (32). (Sons of unknown prenatal smoke exposure status had a mean birth weight of 3201 g, close to that of the exposed). There were no statistically significant differences between the exposed and unexposed groups for personal characteristics and medical status of either the mothers or the sons with respect to the pregnancy and birth (Table 1). The groups of mothers and sons were similar for most variables measured except for a higher prevalence of certain medical problems (high blood pressure, thyroid conditions, and one case of cancer) prior to pregnancy among nonsmoking mothers. There was only modest concordance between mothers' and sons' smoking habits. Early exposure to maternal smoking versus semen quality and hormone levels There were no significant differences for mean values of any of the conventional semen characteristics of sperm concentration, semen volume, sperm concentration per ejaculate, and percentage of viable and motile sperm between men who had early exposure to maternal smoking compared with those who were unexposed (Table 2). Also, no differences were observed between either of these groups and unclassified subjects. None of these findings was altered by adjusting for sons' smoking or DES exposure status. Analysis of interaction effects showed a significant interaction between DES exposure and exposure to maternal smoking only for sperm motility. Exposure to maternal smoking was associated with a significant decrease in average sperm motility among men not exposed to DES (61% compared with 67%; P = 0.020) but not among the DES-exposed men (63% compared with 59%; ns). Sons' smoking and a DESsons' smoking interaction term were also included in the model. The overall percentage of oligospermic men appeared higher among men exposed to maternal
301
Maternal smoking and male fertility • J. M. RATCLIFFE ET AL. T a b l e I. D e m o g r a p h i c characteristics a n d medical status at p r e g n a n c y / b i r t h o f m o t h e r s a n d sons
Characteristic Mothers Race (%) White Black Other Median family income ($100s) Mean gravidity Mean parity Average age of mother at delivery (yrs) Significant medical conditions prior to pregnancy
(%)a
Pregnancy complications (%)8 Sons Mean gestational age at delivery % Premature birth % Exposed to DES in preg. Adult smoking status c current smoker (%) ex-smoker (%) never smoker (%) Mean age, 1st semen analysis
Nonsmokers in pregnancy (n = 161)
Smokers in pregnancy (n = 68)
Unclassified (n = 63)
96 3 1 43 2 2 28 12
94 4 2 43 2 2 26 4
92 8 0 44 2 2 28 10
17
15
16
40 1 57
39 2 60
39 6 56
35 15 51 25
47 16 38 24
29 9 62 25
No significant differences (two-tailed t or chi square tests) found for any characteristic between any group. aOne or more of high blood pressure, thyroid condition, or cancer. bOne or more of spotting, pre-eclampsia, eclampsia, or "other complications." cSmoking data available for only 52% of subjects.
smoking than among unexposed men (15% compared with 9%), but this difference was not significant before or after adjustment for DES exposure and sons' smoking (Table 2). Again, a similar interaction between DES exposure and maternal smoking was observed. Exposure to maternal smoking was associated with a significant increase in oligospermia among men not exposed to DES (22% compared with 3%; P = 0.030) but not among DES-exposed men (10% compared with 13%; ns). It was not possible to fit sons' smoking in this model. Men unclassified with respect to exposure to maternal smoking had a significantly lower percentage of sperm with normal morphology compared with both exposed and unexposed men, both before and after adjustment for son's smoking status and DES exposure status and interaction effects (Table 3). This was not due to a significant prevalence of any one type of abnormality and is uninterpretable in terms of possible biologic significance. The percentage of men with less than 40% normal morphology (using Eliasson's cut-off level for "severely pathological" specimens [29]) were closely comparable (Table 3); further, there were no significant differences between groups on the overall Eliasson scores. No differences were observed between groups for mean levels of LH, FSH, or testosterone before or after adjustment for DES exposure, sons' smoking, and interactions (Table 4); there were no significant
differences in the percentages of men classified by the Chicago investigators as having abnormal values. No differences between groups were observed with respect to abnormal fructose levels in seminal plasma or findings on urinary tract cytology before and after prostatic massage; a total of 4 men had unusual findings on prostate fluid cytology. No testicular biopsies were performed.
Early exposure to maternal smoking versus urogenital abnormalities and diseases With respect to first physical examinations, there were no significant differences between groups on testicular size, the prevalence of capsular induration of the testis, appendix of the testis, or history of cryptorchidism (Table 5). The low prevalence of abnormalities did not permit exploration of the effects of confounders and interactions. Similarly, there were no differences between groups for epididymal abnormalities (epididymal nodules, appendix, cysts, or other abnormalities), the presence of hydrocele, varicocele, or other scrotal abnormalities or prostrate abnormalities. External genitalia were generally normal, with 3 subjects exhibiting abnormal breast development and none abnormal body hair or pubic hair development; only 3% of all subjects had either hypospadias, stenotic meatus, or other abnormalities of the penis. The data were also analyzed with respect to abnormalities noted on any (that is, including re-
302
Reproductive Toxicology
Volume 6, Number 4, 1992
T a b l e 2. S e m e n characteristics in a d u l t m e n in relation to early e x p o s u r e to m a t e r n a l s m o k i n g
Characteristic Semen volume (mL) Mean (SD) Range Sperm concentration Mean (SD) (million/mL) Range Sperm count per ejaculate (millions) Mean (SD) Range % Mean with _< = 20 million sperm/mL (oligospermia) % Viable sperm Mean (SD) a'b Range % Motile sperm Mean (SD) b Range % Men with _< = 30% motile sperm b Total motile sperm b per ejaculate (millions) Mean (SD) Range Motility grade d (%) 2-2.5 (poor) 3-3.5 (moderate) 4 (good)
Unexposed (n = 161)
Exposed (n = 68)
Unclassified (n = 63)
3.26 (1.45) 0.2-8.0
3.18 (1.75) 0.7-9.9
2.89 (1.83) 0.4-9.9
102.5 (80.6) 0-500
100.1 (86.6) 4-470
117.3 (145.0) 0-825
315.7 (242.9) 0-1250.0 8.7
310.2 (278.0) 7-1275.0 14.7
285.3 (329.1) 0-1750 9.5
70.4 (12.1) 32-96
68.0 (13.0) 26-89
66.7 (13.2) 16-89
62.3 (14.6) 4-85 4.4
62.2 (13.8) 18-87 4.4
59.4 (14.9) 6-86 4.8
208.9 (165.9) 1-888
203.1 (191.4) 2-931
173.6 (181.2) 3-1022
12.0 39.0 49.1
8.8 41.2 50.0
9.7 54.8 35.5
an = 132 (unexposed), n = 59 (exposed), n = 52 (unclassified). bExcluding 2 unexposed and 1 unclassified azoospermic men. All differences not significant on two-tailed t tests or chi square; overall motility grade not significantly different between groups using Wilcoxon test for ordered scores.
peat) examinations or questionnaires, since there were cases in which abnormalities were diagnosed on later examinations. Again, no significant differences were found between the groups. The groups did not differ significantly with respect to their history of specific urogenital infections of the scrotum, epididymis, prostate, kidney, and bladder or gonorrhea, syphilis, or urinary tract infections, with the exception of penile infections, which were reported by approximately 8% of the unexposed and unclassified men but none of the exposed men. We consider it unlikely that this association has biologic significance, however. Reported age at puberty (at approximately 13 years), at first ejaculation (at approximately 14 years), and age at first intercourse (at approximately 18 years) did not differ among groups. Son's smoking status versus semen quality and hormone levels Current smokers had a significantly lower percentage of sperm with normal morphology compared with nonsmokers and men with unknown smoking status, both on a crude t test and after adjusting for DES exposure, exposure to maternal smoking, and
interaction terms (Table 6). Ex-smokers had an average percentage of normal forms that was intermediate between that of current and nonsmokers. The proportion of men with
