Predicting pregnancy and spermatogenesis by survival analysis during gonadotrophin treatment of gonadotrophin-deficient infertile men

Peter Y. Liu1, Val J. Gebski2, Leo Turner1, Ann J. Conway1, Susan M. Wishart1 and David J. Handelsman1,3

1 Department of Andrology and ANZAC Research Institute, Concord Hospital, and 2 NHMRC Clinical Trials Centre, University of Sydney, Sydney NSW 2139, Australia


    Abstract
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 Acknowledgements
 References
 
BACKGROUND: Predictors of fertility or spermatogenesis during gonadotrophin therapy of gonadotrophin-deficient men remain poorly defined. METHODS AND RESULTS: In order to evaluate potential predictors, this study evaluated 29 consecutive gonadotrophin-deficient men all desiring paternity who received 43 courses of therapy in one centre between 1982 and 1998. The Kaplan–Meier survival analysis estimates of median (SE) time to a sperm concentration of >0, >5 and >20x106/ml were 5.5 (1.1), 12.4 (2.3) and 29.1 (1.9) months respectively. Conception occurred in 22/43 cycles (with eight men achieving two pregnancies) with a median (SE) Kaplan–Meier estimate of 20.5 (4.7) months. The median sperm concentration at conception was 5.0 (SE 2.0; range 0.0–59.5) x106/ml. Multivariate correlated Cox proportional hazards models predicting these same sperm thresholds and conception were developed by forward stepwise variable selection with verification of the model by backward stepping. Larger testicular volume, prior gonadotrophin therapy, completion of puberty, older age, the absence of adverse fertility factors and the absence of multiple pituitary hormone deficiency predicted a favourable response. Multivariate modelling suggests that the two most important predictors of sperm output are testicular volume and pubertal status. The most important potentially modifiable predictor was prior gonadotrophin therapy. The efficacy of recombinant and urinary FSH were similar. Prior androgen therapy and partner's age did not appear to be significant. CONCLUSIONS: Since prolonged treatment may be required to induce spermatogenesis, attention to these predictors may allow appropriate early use of advanced reproductive technologies.

Key words: gonadotrophin deficiency/gonadotrophin therapy/men/survival analysis


    Introduction
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 Acknowledgements
 References
 
Gonadotrophin therapy, comprising HCG with or without FSH, for gonadotrophin-deficient men remains one of the few specific treatments for male infertility (Gordon Baker, 2001Go). GnRH is an alternative in men without pituitary disease, but is not often used due to the inconvenience of prolonged continuous wearing of an external pump. Although gonadotrophin therapy has been tried in non-gonadotrophin-deficient infertile men, it is no more effective than placebo (Kamischke et al., 1998Go). In sharp contrast, gonadotrophin therapy is highly effective at inducing spermatogenesis and fertility in gonadotrophin-deficient men. Effective regimens have been reported based on using HCG purified from urine of pregnant women together with FSH purified from human pituitaries (Burger and Baker, 1984Go), urine of menopausal women (Büchter et al., 1998Go) or recombinant (r)FSH (Liu et al., 1999Go). Pituitary extracts are no longer used due to the risk of transmitting Creutzfeld–Jakob disease (Cochius et al., 1990Go; Healy and Evans, 1993Go) and urinary preparations are increasingly supplanted by rFSH.

Few studies have reported the efficacy of gonadotrophin therapy in gonadotrophin-deficient infertile men. Most are small and none compare different forms of FSH. This is largely because the community prevalence of gonadotrophin deficiency is low and such men seek fertility on few occasions, so that even specialized centres accumulate small numbers. Even among the larger studies (Burger and Baker, 1984Go; Finkel et al., 1985Go; Okuyama et al., 1986Go; Burris et al., 1988Go; Liu et al., 1988Go; Mastrogiacomo et al., 1991Go; Saal et al., 1991Go; Schopohl et al., 1991Go; Okada et al., 1992Go; Vicari et al., 1992Go; Kirk et al., 1994Go; Kung et al., 1994Go; Burgues and Calderon, 1997Go; Anonymous, 1998aGo; Büchter et al., 1998Go), only three involved more than 20 men (Burgues and Calderon, 1997Go; Anonymous, 1998aGo; Büchter et al., 1998Go) and multi-centre studies (Burgues and Calderon, 1997Go; Anonymous, 1998aGo) may not maintain consistency of management compared with a single centre. Few studies have estimated time to achievement of pregnancy (Kung et al., 1994Go; Büchter et al., 1998Go) and none have previously employed correlated analysis. Therefore, quantitative estimates of important and basic data such as the expected time to sperm output thresholds and conception are not available. Hence, retrospective studies utilizing survival analysis techniques in large population groups are desirable. We report the largest series of treatment courses in gonadotrophin-deficient men desiring fertility within a single centre to estimate time to spermatogenic thresholds and conception using correlated multivariate survival analysis modelling. Survival analysis includes all available data, including treatment failures, so is less biased (Peto et al., 1976Go). The present study also provides the first comparison between the efficacy of rFSH and urinary (u)FSH for induction of spermatogenesis and male fertility.


    Materials and methods
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 Acknowledgements
 References
 
Patients
Men with established gonadotrophin deficiency (hypogonadotrophic hypogonadism) seeking fertility were treated by one of two physicians using a common protocol at a single centre. The diagnosis was established by standard medical history and physical examination (including testing for anosmia) and plasma hormone assays together with pituitary function tests and imaging of the hypothalamic–pituitary region as appropriate. The couple was counselled regarding timing of intercourse and the woman had tubal patency and ovulation evaluated by standard investigations prior to initiation of gonadotrophin therapy.

Treatment commenced by substituting androgen replacement therapy (if used) with HCG (Pregnyl; Organon or Profasi; Serono) at one ampoule (1500 or 2000 IU respectively) administered twice weekly, usually by self-injection under the skin of the abdomen or upper thigh. Adequacy of HCG dosage was evaluated after the first month. If the trough plasma testosterone (measured immediately before the next injection) remained subnormal and/or if androgenic effects were not well maintained, the same dosage was increased to three times or, rarely, four times weekly. HCG treatment was maintained alone for 3–6 months and, if no sperm had appeared by at most 6 months of adequate HCG treatment, FSH was added. The initial dose of uFSH (Pergonal; Serono or Humegon; Organon) was 75 IU three times weekly and for rFSH (Gonal F; Serono or Puregon; Organon) 150 IU three times weekly. When using both gonadotrophins, they were mixed and administered in the same syringe. If testis growth and sperm output was inadequate, the FSH dosage was increased to 150 IU three times weekly and, rarely, to 150 IU daily. The typical dose required was 150 IU three times weekly.

During gonadotrophin treatment, men were reviewed at 3 monthly intervals. At each visit clinical features (androgenic effects, increases in testis size) were monitored and blood (plasma testosterone) and semen samples were obtained. Testis size was measured using a Prader orchidometer. Semen samples were collected by masturbation and analysed according to the standard World Health Organization methods described in the then current World Health Organization manual (World Health Organization, 1980Go, 1987Go, 1992Go). Treatment was continued until completion of the first trimester when pregnancy was confirmed or when the couple decided to terminate therapy. No pregnancies included in this analysis involved IVF or related procedures.

Data analysis
The present study reviewed all data for gonadotrophin-deficient infertile men treated with gonadotrophin therapy between 1982 and 1998 at this centre. Data from all treatment cycles were collected prospectively according to a standard data sheet. This report includes data from a previously published report (Liu et al., 1999Go). Testis volume was analysed as the mean of the left and right testis volumes.

Conception dates were estimated to the nearest week, if possible, usually extrapolated from the date of positive urinary pregnancy tests, otherwise the 15th day of the estimated month was designated. Semen analysis at the time of conception was taken as the semen sample analysed closest to, and usually 1 month before, conception.

Statistical analysis
Data are expressed as median (SE; range). For continuous variables, differences between groups were tested by parametric or non-parametric tests as appropriate. Categorical data were analysed by exact methods for contingency tables. Two-tailed P-values of < 0.05 were considered statistically significant. Analyses were performed using the Statview 5.0 (Calderola et al., 1998Go) and StatXact 4.0 (Anonymous, 1998bGo) statistical software. Correlated survival analysis was performed using Accord 1.1 (Anonymous, 2000Go).

Since the goal of therapy for all subjects was pregnancy, some subjects may have ceased treatment due to their knowledge that their sperm count was not increasing. It is thus important to examine whether these subjects give rise to so-called `informed censoring' which may violate the assumptions of statistical methods used to analyse time to event data (Cox regression, logrank tests etc.). Examination of whether the censoring mechanism was related to outcome (time to pregnancy) will provide some idea as to whether standard statistical methods are still appropriate. The relationship was examined by fitting a logistic regression assuming an extreme case where all subjects who did not obtain a sperm concentration of 5x106/ml were classified as having withdrawn from treatment due to informed censoring. Survival analysis was then used to model underlying fertility estimated as the time to various sperm output thresholds (0, 5 or 20x106 sperm/ml) and to pregnancy. Kaplan–Meier product-limit estimates of median times to the various sperm concentration thresholds and to conception were also calculated.

Potential categorical covariates (all considered a-priori predictors of spermatogenesis or conception) were evaluated by log-rank (Mantel–Cox) test in a Kaplan–Meier model and continuous variables by Wald test in a Cox proportional hazards regression model. The categorical variables studied were the presence of adverse fertility factors (comprising cryptorchidism, female factors or poor compliance), multiple pituitary hormone deficiency (hypopituitarism versus isolated gonadotrophin deficiency), completion of puberty prior to diagnosis, treatment modality (HCG alone or with uFSH or rFSH) and prior androgen or gonadotrophin treatment. The continuous variables studied were testis volume, age and partner's age at the start of treatment. In addition, subgroup analyses were performed examining the effect of prior androgen or gonadotrophin therapy. These analyses were performed to examine the effect of prior gonadotrophin therapy, to allow comparison with the published literature which has generally treated multiple treatment courses in single individuals as independent. These were also used as an exploratory measure to determine potentially important predictors prior to performing correlated analyses. Log cumulative hazard and Kaplan–Meier plots of each variable were examined.

A model predicting sperm thresholds and conception using these variables was developed using forward stepwise variable selection with verification of the model by backward stepping in a correlated Cox proportional hazards model (Lee et al., 1992Go). The correlated Cox proportional hazards model allows analysis of time to event data using observations that are correlated, meaning that more than one observation per individual can be used. The P-value for inclusion and exclusion of the variables was 0.05. The model was checked manually at each step to examine for interactions, particularly for variables found to be significant from the exploratory Kaplan–Meier analysis. The unit of measure for continuous variables was years (for age) and millilitres (for testicular volume).


    Results
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 Acknowledgements
 References
 
Patients
Twenty-nine consecutive gonadotrophin-deficient men all desiring paternity received 43 courses of therapy comprising HCG alone (six courses), HCG with uFSH (26 courses) or HCG with rFSH (11 courses). Isolated gonadotrophin deficiency was present in 20 men [13 idiopathic hypogonadotropic hypogonadism (IHH) and seven Kallmann's syndrome (KS)] none of whom completed puberty spontaneously. The remaining nine men had multiple pituitary hormone deficiencies (hypopituitarism) due to craniopharyngiomas (two) or pituitary disease (seven: five adenomas, one trauma, one agenesis) and required corticosteroids, thyroxine or vasopressin. All of these men (except for one with craniopharyngioma) had completed puberty spontaneously. Adverse fertility factors were associated with men with isolated gonadotrophin deficiency alone, and comprised cryptorchidism (five), female factors (four endometriosis or irregular ovulation) and poor compliance (one) affecting 12 courses of treatment.

The median age of men was 36 years (range 26–52) with no difference according to diagnosis, pubertal status, treatment, previous androgen treatment or adverse fertility factors. As expected, men receiving their second course (n = 16, median age 39 years) were marginally but significantly (P < 0.05) older than those treated for the first time (n = 26, median age 35 years). The female partners were younger than the men with a median age of 30 years (range 20–41).

The median pre-treatment testis volume was 8 ml (range 1–20). Cryptorchid men had significantly lower testis volume (median 2, range 2–4 ml) compared with non-cryptorchid men (median 8, range 1–20 ml, P < 0.005). Prior androgen therapy (n = 29 courses) was associated with smaller testis volume (median 6 versus 12.5 ml in those who had not received prior androgen therapy, P = 0.03). Most men (except one) who did not receive prior androgen therapy also did not fully complete puberty, were often diagnosed as having `delayed puberty' and were subsequently found to have hypothalamic disease (IHH or KS). There was a non-significant trend towards increased testicular volume among those who completed puberty (median 6 versus 10.5 ml, P = 0.07) and, to a lesser extent, in those previously treated with gonadotrophin therapy (median 6 versus 8 ml, P = 0.38). There were no consistent differences according to diagnosis or treatment.

Treatment
The details of the courses of treatment are summarized in Table IGo. Each course for any man was separated by at least 6 months and was considered independently. Three of the 29 men were treated elsewhere for their first course of therapy and details of their first course were not available, and hence there were 26 first, 16 second and one third course of therapy included in this study. Since these same three men had their second course of therapy at our institute, complete details for two courses of treatment were available for 13 men, although 16 had previously received a course of gonadotrophin therapy. Seven men had not received androgens prior to their first course of gonadotrophin therapy. One man initially treated with HCG and uFSH for 2 years was switched to pulsatile GnRH for <4 weeks before conception was confirmed and hence was included in the analysis. No other patients received pulsatile GnRH. There was no association between treatment type, diagnosis and course number (P = 0.60, extended Fisher's exact test).


View this table:
[in this window]
[in a new window]
 
Table I. Treatment, diagnosis and course
 
FSH was added after a median time of 3.8 months (SE 0.5; range 3–6) at a usual maintenance dose of 450 IU/week (SE 50; range 150–1050). Half received a maintenance dose of 450 IU/week, with one-quarter each receiving 225 and 1050 IU/week. This median time was not different between uFSH (median 3.3; SE 0.7; range 0.0–10.3) or rFSH (median 4.6; SE 0.4; range 3.0–6.1; P = 0.34), nor was the median maintenance dose different (450 IU/week for both). The median maintenance HCG dose did not differ among those treated with HCG alone (4250 IU/week), or with uFSH (4500 IU/week) or rFSH (6000 IU/week). The treatment was well tolerated, and no patient discontinued for medical reasons.

In the subgroup of 13 men with two evaluable cycles of gonadotrophin treatment, there was a median time interval between cycles of 16.2 months (SE 3.2; range 7.9–43.5). In this subgroup, median testis volume remained larger at the start of the second course compared with that at the start of the first course of treatment (median 11 versus 6 ml; P < 0.01, Wilcoxon signed rank test).

Among the subgroup of 26 men who had never previously received gonadotrophin therapy, those who had received prior androgen therapy tended to be younger (median 34 versus 38 years) and to have smaller initial testicular volumes (median 6 versus 12 ml), but these differences were not significant.

Time to thresholds: sperm output and conception
The Kaplan–Meier survival analysis estimate of median time to appearance of sperm was 5.5 months (SE 1.1), to a sperm concentration of 5x106/ml was 12.4 months (SE 2.3) and to a sperm concentration of 20x106/ml was 29.1 months (SE 1.9) (Figure 1Go). Six men were not azoospermic at the start of therapy. All had sperm concentrations of <1x106/ml, most with only occasional sperm, except one man with a sperm concentration of 28x106/ml. This man had not been able to impregnate his wife for >3 years and was markedly androgen deficient.



View larger version (19K):
[in this window]
[in a new window]
 
Figure 1. Kaplan–Meier survival plots of the proportion of the total number of courses of gonadotrophin therapy which successfully resulted in sperm concentrations of >0, >5 and >20x106/ml (closed symbols) and conception (open symbol). The total number of courses is 43 for each plot. n = number of successful courses. The median time (months) to success is plotted as a dashed line and indicates when half the courses had been successful.

 
Conception occurred in 22/43 cycles (with eight men achieving two pregnancies) with a Kaplan–Meier estimate of 20.5 months (SE 4.7). Excluding patients who failed to achieve a pregnancy produced a (biased) estimate of median time to conception of 14.1 months (SE 1.6). At conception, the median sperm concentration was 5.0x106/ml (SE 2.0; range 0.0–59.5) and median sperm output was 11.8x106 (SE 7.3; range 0.0–238.0) per ejaculate.

Univariate predictors of spermatogenesis and fertility
The Kaplan–Meier product limit estimates are summarized in Table IIGo.


View this table:
[in this window]
[in a new window]
 
Table II. Kaplan–Meier product limit estimates of median times
 
Prior gonadotrophin therapy was associated with shorter median time to first appearance of sperm (Figure 2Go). There was a non-significant trend for shorter times to conception and sperm concentrations of >5 or >20x106/ml with a second course of treatment. These relationships were replicated in the subgroup of 13 men in whom details from both courses of gonadotrophin therapy were available (Table IIIGo). As can be seen from this table, only eight of these 13 men had successfully impregnated their partners during the first course of treatment.



View larger version (15K):
[in this window]
[in a new window]
 
Figure 2. Kaplan–Meier survival plots comparing the proportion of the total number of courses of gonadotrophin therapy which successfully resulted in first appearance of sperm between those who had or had not previously received gonadotrophin treatment. The total number of courses is 43 for each plot. n = number of successful courses. The median time (months) to detectable sperm is plotted as a dashed line.

 

View this table:
[in this window]
[in a new window]
 
Table III. Prior gonadotrophin and prior androgen therapy
 
Prior androgen use was associated with a longer time to a sperm concentration of 20 mol/l/ml, but not any other sperm threshold or pregnancy. These results were replicated in the subgroup of 26 men who had never received prior gonadotrophin therapy (Table IIIGo). However, the number of censored observations were high.

There was no consistent or significant effect of cryptorchidism, different treatment regimens, hypopituitarism (versus isolated gonadotrophin deficiency), or completion of puberty prior to treatment.

Larger pre-treatment testis volume was a significant (P < 0.002) predictor of shorter time to all sperm thresholds. Pre-treatment testis volume did not significantly predict conception unless other adverse fertility factors were excluded (P = 0.048). Age of partner was not a predictor for any sperm threshold or conception.

Multivariate predictors of spermatogenesis and fertility and variable interaction
Using forward stepwise inclusion, testicular volume (P < 0.005) was selected in all models of sperm thresholds and pregnancy. Post-pubertal status was a signficant predictor of first sperm appearance, achieving a sperm concentration of 5x106/ml and pregnancy. No other variables consistently predicted sperm thresholds and pregnancy. However, larger testis volume, post-pubertal onset of hypogonadism, increasing age, no adverse fertility factors and the absence of multiple pituitary hormone deficiency predicted a better response. Prior androgen use and partner's age were not significant predictors for any model. The best models are shown in Table IVGo. Backward stepwise variable exclusion confirmed stability and log cumulative hazard and Kaplan–Meier plots (not shown) confirmed validity of these models. Logistic regression of time to pregnancy on sperm output failed to show a significant relationship, suggesting that informed censoring (if it was present) did not invalidate the results of the log-rank and Cox regression analyses.


View this table:
[in this window]
[in a new window]
 
Table IV. Optimal correlated Cox proportional hazards models
 

    Discussion
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 Acknowledgements
 References
 
Conventionally, gonadotrophin deficiency is treated with androgen replacement therapy to induce and maintain virilization as this is more cost-effective and convenient than gonadotrophin therapy. The one exception is that testis development, spermatogenesis and fertility cannot be induced by testosterone alone (Schaison et al., 1993Go), so that gonadotrophin therapy is required in gonadotrophin-deficient men who seek fertility. Thus, gonadotrophin treatment constitutes one of the few treatable causes of male infertility.

Our data show that during gonadotrophin replacement therapy, testis volume is the single most important predictor of sperm output and a significant predictor of pregnancy. This is because testis volume is largely physically determined by the amount of spermatogenic tissue present and it is this tissue which controls spermatogenesis. The effect size (hazard ratio) was remarkably consistent for all spermatogenic endpoints, being 1.27–1.28 (see Table IVGo), and comparable for conception and not necessarily small since increases in testicular volume of >1 ml result in proportionally greater increases in absolute hazard. The pre-eminence of testis volume as an explanatory factor for sperm output thresholds is supported by the published literature, which is summarized graphically as a bivariate plot of initial testicular volume versus duration to induce spermatogenesis (Figure 3Go). Given the heterogeneity of the data (varying measures of central tendency and differing sample sizes), it is not appropriate to perform regression analysis. Ten studies of gonadotrophin treatment for gonadotrophin-deficient men that reported baseline testis volume and estimates (mean or median) of time to appearance of sperm in at least seven men were found by a Medline review of the literature published since 1966, supplemented by hand searching (Ley and Leonard, 1984Go; Liu et al., 1988Go; Saal et al., 1991Go; Schopohl et al., 1991Go; Okada et al., 1992Go; Jones and Darne, 1993Go; Kliesch et al., 1994Go; Kung et al., 1994Go; Burgues and Calderon, 1997Go; Anonymous, 1998aGo; Büchter et al., 1998Go). One (Kliesch et al., 1994Go) was subsequently incorporated into a larger analysis (Büchter et al., 1998Go) and was excluded from further analysis. All these studies used gonadotrophin dose regimens similar to those described in this study except for one study, which appeared to use much lower doses (Okada et al., 1992Go).



View larger version (8K):
[in this window]
[in a new window]
 
Figure 3. Bivariate bubble plot of central estimates (mean or median) of time to appearance of sperm (months) against initial testicular volume (ml). Each point represents a single published study of at least seven men. The area of each point is proportional to the sample size of that study. The results from this study are indicated by the closed symbol.

 
The critical testis volume of 4 ml is consistent with previous studies, indicating that this testis size is the best determinant of complete gonadotrophin deficiency. Below this level, induction of spermatogenesis requires protracted therapy although conception can rarely occur (Vicari et al., 1992Go). Within individual studies, some (Burris et al., 1988Go; Vicari et al., 1992Go; Kung et al., 1994Go; Büchter et al., 1998Go) but not all (Ley and Leonard, 1984Go; de Sanctis et al., 1988Go; Saal et al., 1991Go; Jones and Darne, 1993Go; Burgues and Calderon, 1997Go) large studies have reported a significant relationship between testicular volume and spermatogenic response. These negative studies may be due to the presence of confounding variables that are not accounted for in a univariate analysis.

In the present study, first appearance of sperm occurred more rapidly (3.5 versus 7.1 months) among men with prior exposure to gonadotrophin therapy. Pregnancy also occurred more rapidly, especially when adverse fertility factors were taken into account. Furthermore, the effect of prior gonadotrophin therapy was sustained, with a 5 ml increase in median testis volume persisting even after cessation of gonadotrophin therapy for >16 months. Three other studies have examined the effect of prior gonadotrophin therapy, but have not been conclusive. One found that pregnancy occurred more rapidly in those previously treated with gonadotrophins, but four out of the 16 pregnancies reported in that study were achieved via assisted reproduction, and surprisingly, no significant effect on spermatogenesis was seen (Büchter et al., 1998Go) consistent with an earlier, but smaller analysis from the same group (Kliesch et al., 1994Go). Another noted that men previously exposed to HCG for induction of puberty subsequently required a lower dose of HCG for normalization of testosterone levels and induction of testicular growth (Kung et al., 1994Go). The third study reported a non-significant (P = 0.3) trend for improved response rate (Burgues and Calderon, 1997Go). It may be that earlier gonadotrophin exposure, whether due to natural puberty or prior gonadotrophin treatment, may prime the testis and enhance spermatogenic response to subsequent gonadotrophin therapy.

Our study confirms that the beneficial effect of natural completion of puberty on spermatogenesis is not mediated by larger testis volume, since the effect was apparent even when testis volume was accounted for. Furthermore, these spermatogenic effects translated into beneficial effects on conception. This is consistent with an additional priming effect of gonadotrophins at puberty, confirming most (Finkel et al., 1985Go; Mastrogiacomo et al., 1991Go; Burgues and Calderon, 1997Go) but not all (Kung et al., 1994Go) reports. Long-term gonadotrophin therapy has also been successfully used to induce spermatogenesis and androgenization in adolescent males (Barrio et al., 1999Go; Bouvattier et al., 1999Go), but this approach has been little used due to the limited availability, cost and inconvenience of gonadotrophins. Given the beneficial effects of prior gonadotrophin therapy as well as spontaneous completion of puberty, it is an intriguing but untested possibility that pubertal gonadotrophin treatment culminating in chronologically appropriate testis growth and spermatogenesis may facilitate later induction of spermatogenesis and/or fertility (Bouvattier et al., 1999Go). This concept is further supported by the greater delay in inducing spermatogenesis reported when treatment of acquired gonadotrophin deficiency is delayed by >2 years (Tachiki et al., 1998Go). It remains to be established whether gonadotrophin therapy administered to complete puberty can mimic the beneficial effect of naturally occurring puberty.

Prior androgen therapy has an apparently deleterious effect on attainment of sperm concentration of 20x106/ml in the exploratory Kaplan–Meier analysis, but not on lower sperm output thresholds or on pregnancy, and importantly, no effect was detected by correlated Cox analysis. Furthermore, this exploratory analysis was confounded because prior androgen therapy was associated with significantly lower testis volume. The replication of these results in gonadotrophin-naïve men suggests that this quantitative effect on the rapidity and extent of induction of spermatogenesis could be due to the prior use of androgens instead of gonadotrophins. It is likely that men with more severe gonadotrophin deficiency and lower pre-treatment testis volumes, may also be more overtly androgen deficient and hence more likely to have previously received androgen therapy. The prolonged times may therefore reflect selection bias rather than the prior use of androgens per se. Since a marked effect on time to conception was not shown and others have not found any relationship between the duration of androgen therapy and the time to detect sperm (Okada et al., 1992Go) nor any difference between prior androgen or prior gonadotrophin therapy (Kliesch et al., 1994Go), androgen use per se is not likely to be detrimental.

Our data show that multiple pituitary hormone deficiency is deleterious to spermatogenesis. However, this effect appeared only to be important for the induction of spermatogenesis and, critically, did not appear important for conception. The presence of multiple pituitary hormone deficiency (hypopituitarism) has been variously reported as being advantageous (Okuyama et al., 1986Go) or detrimental (Ley and Leonard, 1984Go) in small studies (<=5 men). Larger studies have reported no significant effect on spermatogenesis (Kung et al., 1994Go; Burgues and Calderon, 1997Go) although a trend favouring those with hypopituitarism (Büchter et al., 1998Go) has been noted. Since hypopituitarism tends to occur post-pubertally and is therefore also associated with larger testicular volume (Kliesch et al., 1994Go), these other confounders must be adjusted for. Our data confirm this since the categories of completion of puberty and the presence of hypopituitarism were virtually identical, differing by only one man.

The presence of adverse fertility factors, including cryptorchidism, was a negative prognostic predictor of conception. Due to small numbers, the effect of cryptorchidism alone could not be studied, although there was a clear relationship with smaller testicular volume. In a study of 13 cryptorchid men, matched with 13 non-cryptorchid men with similar testicular volume (<4 ml), a statistically significant reduction in testicular growth and spermatogenesis was reported (Kirk et al., 1994Go) suggesting an additional detrimental effect not accounted for by testicular volume. This is consistent with our data. The remainder of the studies have examined fewer men, have not controlled for testicular volume and have variably reported significant (Ley and Leonard, 1984Go; Finkel et al., 1985Go), equivocal (Saal et al., 1991Go; Büchter et al., 1998Go) or no (Jones and Darne, 1993Go) association.

Age, but not partner's age, was found to be a significant predictor of attainment of only high sperm concentration (20x106/ml) and pregnancy. Furthermore, the effect is small given that the unit of measurement is years, and the age range was 26–52 years.

This study found no significant difference in spermatogenic or pregnancy outcomes when comparing rFSH with uFSH. Importantly, univariate estimates of median time to sperm output thresholds and conception differed by only ~20%, nor were there any consistent differences in the multivariate analyses. Therefore, despite the limitations of these retrospective data, including differing initial FSH dosage (although maintenance dosage was identical) and small numbers in the rFSH group, any differences are considered likely to be small.

Our Kaplan–Meier estimate of 5.5 months of treatment to induce spermatogenesis is in close agreement with published estimates in men with similar testicular volumes. Figure 2Go shows that spermatogenesis should occur after 3–15 months of gonadotrophin therapy. This is consistent with other studies of gonadotrophin therapy (de Sanctis et al., 1988Go; Mastrogiacomo et al., 1991Go; Tachiki et al., 1995Go) or GnRH (Mortimer et al., 1974Go; Aulitzky et al., 1988Go).

Our Kaplan–Meier estimate that 20.5 months of treatment is required before pregnancy occurs is in agreement with the limited published data. A median treatment duration of 20 months (range 4–78) was found in a study of 12 successful pregnancies (Kung et al., 1994Go). Median treatment durations of 23 (range 7–54) (Burris et al., 1988Go) and 43 (range 19–64) months (Vicari et al., 1992Go) have been reported in two other studies, each reporting successful pregnancies in seven couples and both employing long-term HCG treatment. Although the duration of therapy is longer in the latter study, it is not clear whether gonadotrophin therapy was discontinued at conception. Shorter mean treatment durations of 9 (range 1–21) and 8 (range 2–46) months have also been reported in a study where 26 (out of 36) courses of treatment resulted in pregnancy (Büchter et al., 1998Go). However, this is likely to be an underestimate since it included three couples who achieved pregnancy through ICSI and two further pregnancies that occurred after 40 months of therapy were not included in the above estimates.

Conception occurred at a median sperm concentration of 5x106/ml which is in keeping with many reports of 3–8x 106/ml (Burger and Baker, 1984Go; Burris et al., 1988Go; Vicari et al., 1992Go; Kung et al., 1994Go; Büchter et al., 1998Go). The Kaplan–Meier survival plots indicate that, since conception occurs after ~20 months and a sperm concentration of 5x106/ml after ~12 months, pregnancy can be predicted on average 8 months after a sperm concentration of 5x106/ml is detected. Interestingly, our Kaplan–Meier survival plot (Figure 1Go) also showed that the risk of pregnancy seemed to parallel the risk of achieving a sperm concentration of >5, rather than 0 or 20x106/ml. This is consistent with World Health Organization male contraception studies where this sperm output threshold appeared to demarcate subnormal from normal pregnancy rates (World Health Organization Task Force on Methods for the Regulation of Male Fertility, 1996Go). This is at variance with a recent prospective study of planned pregnancies which claimed a higher threshold (40x106/ml); however, that study population may be subfertile in that it excluded the most fertile couples who had accidental pregnancies (Bonde et al., 1998Go).

In summary, the present study in conjunction with the published literature suggests that when gonadotrophin-deficient men are treated with gonadotrophin therapy, sperm can be expected to appear within 6 months for a testis volume of >=4 ml and over 9 months for smaller testis volumes. These durations may become more significant if the goal of gonadotrophin therapy is to induce enough sperm for ICSI. Furthermore, if pregnancy does not occur after 20 months, or 8 months after achieving a sperm concentration of 5x106/ml, assisted reproductive technologies may be considered time-effective. Cryostorage of sperm may be possible, but may not be of sufficient quality post-thaw to allow subsequent insemination without the risk of ovulation induction required for IVF. Larger testicular volume, prior gonadotrophin therapy, completion of puberty, the absence of adverse fertility factors and possibly the absence of multiple pituitary hormone deficiency predicted a favourable response. The efficacy of rFSH and uFSH appeared similar. Although our study was retrospective, no significant association between the type of treatment and diagnosis, course number, age or initial testicular volume was found, suggesting that there was no systemic bias in our study population. The recent commercial availability of rFSH, HCG and LH (Laml et al., 1999Go) in clinical trials suggests that the previous limitations on gonadotrophin supply may be overcome, and that it is timely to re-evaluate the role of gonadotrophin therapy for induction of testis growth and spermatogenesis in adolescence, particularly in the light of recent advances in reproductive technologies.


    Acknowledgements
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 Acknowledgements
 References
 
The authors wish to thank the staff of the Department of Andrology, Concord Hospital (formerly Andrology Unit, Royal Prince Alfred Hospital) for the dedication and commitment to high quality service without which this study could not have been completed. This study was supported in part by the National Health and Medical Research Council of Australia (postgraduate medical research scholarship to P.Y.L.). This work was presented in part during the President's abstract session of the 7th International Congress of Andrology, June 15–19, 2001, Montreal, Canada.


    Notes
 
3 To whom correspondence should be addressed at: ANZAC Research Institute, Sydney, NSW 2139, Australia. E-mail: djh{at}med.usyd.edu.au Back

Submitted on May 10, 2001; resubmitted on September 21, 2001


    References
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 Acknowledgements
 References
 
Anonymous (1998a) Efficacy and safety of highly purified urinary follicle-stimulating hormone with human chorionic gonadotropin for treating men with isolated hypogonadotropic hypogonadism. European Metrodin HP Study Group. Fertil. Steril., 70, 256–262.[ISI][Medline]

Anonymous (1998b) StatXact-4. Cytel Software Corporation.

Anonymous (2000) Accord. Boffin Software, Sydney.

Aulitzky, W., Frick, J. and Galvan, G. (1988) Pulsatile luteinizing hormone-releasing hormone treatment of male hypogonadotropic hypogonadism. Fertil. Steril., 50, 480–486.[ISI][Medline]

Barrio, R., de Luis, D., Alonso, M., Lamas, A. and Moreno, J.C. (1999) Induction of puberty with human chorionic gonadotropin and follicle-stimulating hormone in adolescent males with hypogonadotropic hypogonadism. Fertil. Steril., 71, 244–248.[ISI][Medline]

Bonde, J.P., Ernst, E., Jensen, T.K., Hjollund, N.H., Kolstad, H., Henriksen, T.B., Scheike, T., Giwercman, A., Olsen, J. and Skakkebaek, N.E. (1998) Relation between semen quality and fertility: a population-based study of 430 first-pregnancy planners [see comments]. Lancet, 352, 1172–1177.[ISI][Medline]

Bouvattier, C., Tauber, M., Jouret, B., Chaussain, J.L. and Rochiccioli, P. (1999) Gonadotropin treatment of hypogonadotropic hypogonadal adolescents. J. Pediat. Endocrinol. Metab., 12 (Suppl. 1), 339–44.[ISI][Medline]

Büchter, D., Behre, H.M., Kliesch, S. and Nieschlag, E. (1998) Pulsatile GnRH or human chorionic gonadotropin/human menopausal gonadotropin as effective treatment for men with hypogonadotropic hypogonadism: a review of 42 cases. Eur. J. Endocrinol., 139, 298–303.[ISI][Medline]

Burger, H.G. and Baker, H.W.G. (1984) Therapeutic considerations and results of gonadotropin treatment in male hypogonadotropic hypogonadism. Ann. NY Acad. Sci., 438, 447–453.[ISI][Medline]

Burgues, S. and Calderon, M.D. (1997) Subcutaneous self-administration of highly purified follicle stimulating hormone and human chorionic gonadotrophin for the treatment of male hypogonadotrophic hypogonadism. Spanish Collaborative Group on Male Hypogonadotropic Hypogonadism. Hum. Reprod., 12, 980–986.[ISI][Medline]

Burris, A.S., Rodbard, H.W., Winters, S.J. and Sherins, R.J. (1988) Gonadotropin therapy in men with isolated hypogonadotropic hypogonadism: the response to human chorionic gonadotropin is predicted by initial testicular size. J. Clin. Endocrinol. Metab., 66, 1144–1151.[Abstract]

Calderola, J., Dilmaghani, A., Gagnon, J., Haycock, K.A., Roth, J., Soper, C. and Wasserman, E. (1998) Statview. SAS Institute Inc., Cary.

Cochius, J.I., Burns, R.J., Blumbergs, P.C., Mack, K. and Alderman, C.P. (1990) Creutzfeldt–Jakob disease in a recipient of human pituitary-derived gonadotrophin. Aust. NZ J. Med., 20, 592–593.[ISI][Medline]

de Sanctis, V., Vullo, C., Katz, M., Wonke, B., Nannetti, C. and Bagni, B. (1988) Induction of spermatogenesis in thalassaemia. Fertil. Steril., 50, 969–975.[ISI][Medline]

Finkel, D.M., Phillips, J.L. and Snyder, P.J. (1985) Stimulation of spermatogenesis by gonadotropins in men with hypogonadotropic hypogonadism. New Engl. J. Med., 313, 651–655.[Abstract]

Gordon Baker, H.W. (2001) Male Infertility, Vol. 3, Endocrinology. Saunders, Philadelphia.

Healy, D.L. and Evans, J. (1993) Creutzfeldt–Jakob disease after pituitary gonadotrophins [editorial]. Br. Med. J., 307, 517–518.[ISI][Medline]

Jones, T.H. and Darne, J.F. (1993) Self-administered subcutaneous human menopausal gonadotrophin for stimulation of testicular growth and the initiation of spermatogenesis in hypogonadotropic hypogonadism. Clin. Endocrinol., 38, 203–208.[ISI][Medline]

Kamischke, A., Behre, H.M., Bergmann, M., Simoni, M., Schafer, T. and Nieschlag, E. (1998) Recombinant human follicle stimulating hormone for treatment of male idiopathic infertility: a randomized, double-blind, placebo-controlled, clinical trial. Hum. Reprod., 13, 596–603.[Abstract]

Kirk, J.M.W., Savage, M.O., Grant, D.B., Bouloux, P.M.G. and Besser, G.M. (1994) Gonadal function and response to human chorionic and menopausal gonadotrophin therapy in male patients with idiopathic hypogonadotrophic hypogonadism. Clin. Endocrinol., 41, 57–63.[ISI][Medline]

Kliesch, S., Behre, H.M. and Nieschlag, E. (1994) High efficacy of gonadotrophin or pusatile gonadotrophin-releasing hormone treatment in hypogonadotrophic hypogonadal men. Eur. J. Endocrinol., 131, 347–354.[ISI][Medline]

Kung, A.W., Zhong, Y.Y., Lam, K.S. and Wang, C. (1994) Induction of spermatogenesis with gonadotrophins in Chinese men with hypogonadotrophic hypogonadism. Int. J. Androl., 17, 241–247.[ISI][Medline]

Laml, T., Obruca, A., Fischl, F. and Huber, J.C. (1999) Recombinant luteinizing hormone in ovarian hyperstimulation after stimulation failure in normogonadotropic women. Gynecol. Endocrinol., 13, 98–103.[ISI][Medline]

Lee, E.W., Wei, L.J. and Amato, D.A. (1992) Cox-type regression analysis for large numbers of small groups of correlated failure time observations. In Klein, J.P. and Goel, P.K. (eds), Survival Analysis: State of the Art. Kluwer, Dordrecth, pp. 237–247.

Ley, S.B. and Leonard, J.M. (1984) Male hypogonadotropic hypogonadism: factors influencing response to human chorionic gonadotropin and human menopausal gonadotropin, including prior exogenous androgens. J. Clin. Endocrinol. Metab., 61, 746–752.[Abstract]

Liu, L., Banks, S.M., Banres, K.M. and Sherins, R.J. (1988) Two year comparison of testicular responses to pulsatile gonadotropin-releasing hormone and exogenous gonadotropins from the inception of therapy in men with isolated hypogonadotropic hypogonadism. J. Clin. Endocrinol. Metab., 67, 1140–1145.[Abstract]

Liu, P.Y., Turner, L., Rushford, D., McDonald, J., Gordon Baker, H.W., Conway, A.J. and Handelsman, D.J. (1999) Efficacy and safety of recombinant human follicle stimulating hormone (Gonal-F) with urinary human chorionic gonadotrophin for induction of spermatogenesis and fertility in gonadotrophin-deficient men. Hum. Reprod., 14, 1540–1545.[Abstract/Free Full Text]

Mastrogiacomo, I., Motta, R.G., Botteon, S., Bonanni, G. and Schiesaro, M. (1991) Achievement of spermatogenesis and genital tract maturation in hypogonadotropic hypogonadic subjects during long term treatment with gonadotropins or LHRH. Andrologia, 23, 285–289.[ISI][Medline]

Mortimer, C.H., McNeilly, A.S., Fisher, R.A., Murray, M.A. and Besser, G.M. (1974) Gonadotrophin-releasing hormone therapy in hypogonadal males with hypothalamic or pituitary dysfunction. Br. Med. J., 4, 617–621.[Medline]

Okada, Y., Kondo, T., Okamoto, S. and Ogawa, M. (1992) Induction of ovulation and spermatogenesis by hMG/hCG in hypogonadotropic GH-deficient patients. Endocrinol. Japon., 39, 31–43.[Medline]

Okuyama, A., Nakamura, M., Namiki, M., Aono, T., Matsumoto, K., Utsunomiya, M., Yoshioka, T., Itoh, H., Itatani, H., Mizutani, S. and Sonoda, T. (1986) Testicular responsiveness to long-term administration of hCG and hMG in patients with hypogonadotrophic hypogonadism. Hormone Res., 23, 21–30.[ISI][Medline]

Peto, R., Pike, M.C., Armitage, P., Breslow, N.E., Cox, D.R., Howard, S.V., Mantel, N., McPherson, K., Peto, J. and Smith, P.G. (1976) Design and analysis of randomised clinical trials requiring prolonged observation of each patient: I. Introduction and design. Br. J. Cancer, 34, 585–612.[ISI][Medline]

Saal, W., Happ, J., Cordes, U., Baum, R.P. and Schmidt, M. (1991) Subcutaneous gonadotropin therapy in male patients with hypogonadotropic hypogonadism. Fertil. Steril., 56, 319–324.[ISI][Medline]

Schaison, G., Young, J., Pholsena, M., Nahoul, K. and Couzinet, B. (1993) Failure of combined follicle-stimulating hormone–testosterone administration to initiate and/or maintain spermatogenesis in men with hypogonadotropic hypogonadism. J. Clin. Endocrinol. Metab. 77, 1545–1549.[Abstract]

Schopohl, J., Mehltretter, G., von Zumbusch, R., Eversmann, T. and von Werder, K. (1991) Comparison of gonadotropin-releasing hormone and gonadotropin therapy in male patients with idiopathic hypothalamic hypogonadism. Fertil. Steril., 56, 1143–1150.[ISI][Medline]

Tachiki, H., Kumamoto, Y., Itoh, N., Maruta, H. and Tsukamoto, T. (1995) [Testicular findings, endocrine features and therapeutic responses of men with idiopathic hypogonadotropic hypogonadism]. Nippon Naibunpi Gakkai Zasshi–Folia Endocrinol. Japon., 71, 605–622.

Tachiki, H., Ito, N., Maruta, H., Kumamoto, Y. and Tsukamoto, T. (1998) Testicular findings, endocrine features and therapeutic responses of men with acquired hypogonadotropic hypogonadism. Int. J. Urol., 5, 80–85.[Medline]

Vicari, E., Mongioi, A., Calogero, A.E., Moncada, M.L., Sidoti, G., Polosa, P. and D'Agata, R. (1992) Therapy with human chorionic gonadotrophin alone induces spermatogenesis in men with isolated hypogonadotrophic hypogonadism—long-term follow-up. Int. J. Androl., 15, 320–329.[ISI][Medline]

World Health Organization (1980) Laboratory Manual For The Examination of Human Semen and Sperm–Cervical Mucus Interaction. Press Concern, Singapore.

World Health Organization (1987) WHO Laboratory Manual For The Examination of Human Semen and Sperm–Cervical Mucus Interaction. Cambridge University Press, Cambridge.

World Health Organization (1992) WHO Laboratory Manual For The Examination of Human Semen and Sperm–Cervical Mucus Interaction. Cambridge University Press, Cambridge.

World Health Organization Task Force on Methods for the Regulation of Male Fertility (1996) Contraceptive efficacy of testosterone-induced azoospermia and oligozoospermia in normal men. Fertil. Steril., 65, 821–829.[ISI][Medline]

accepted on November 13, 2001.