- Email: [email protected]
Male infertility and increased risk of diseases in future generations
target blood pressure, might have been a more realistic alternative to obtain the most reliable GFR evaluation (ie, independent of the functional drugspecific alteration of intraglomerular haemodynamics), therefore better reflecting the glomerular structural change. *A Pruna, N El Esper, R Makdassi, R Oprisiu, A Fournie Service de Néphrologie, CHU Sud, 80054 Amiens, France 1
Ruggenenti P, Perna A, Gherardi G, et al. Renoprotective properties of ACEinhibition in non-diabetic nephropathies with non-nephrotic proteinuria. Lancet 1999; 354: 359–64. 2 Wiggam MI, Bell PM, Sheridan B, Walmsley A, Atkinson AB. Low dose bendrofluazide (1·25 mg) effectively lowers blood pressure over 24 h. Results of a randomized, double-blind, placebocontrolled crossover study. Am J Hypertens 1999; 12: 528–31. 3 Zanad F, Matzinger A, Larché J. Trough/peak ratios of once daily ACE inhibitors and calcium antagonists. Am J Hypertens 1996; 9: 633–43. 4 Wentzel U, Helmchen U, Schoeppe W, Schwietzer G. Combination treatment of enalapril with nitrendipine in rats with renovascular hypertension. Hypertension 1994; 23: 114–22. 5 Mathiesen ER, Hommel E, Hansen HP, Schmidt UM, Parving MH. Randomized controlled trial of long term efficacy of captopril on preservation of kidney function in normotensive patients with insulin dependent diabetes and microalbuminuira. BMJ 1999; 319: 24–25.
Authors’ reply Sir—In stratum 1 of the REIN study, as A Pruna and colleagues correctly point out, ramipril reduced incidence of ESRF in patients with chronic nephropathies and non-nephrotic proteinuria more than the rate of GFR decline. This difference was due to the low rate of progression of patients with mild proteinuria that diluted the effect of ramipril on the rate of GFR decline rather than to imbalances in baseline characteristics of patients. The renoprotective effect of ramipril was higher in patients with baseline GFR less than 45 mL/min 1·73 m2 who had very similar GFRs at baseline (ramipril 30·3 [SE 1·3] mL min⫺1 1·73 m⫺2, conventional 30·4 [1·4] mL min⫺1 1·73 m⫺2). However, we have followed Pruna and colleagues’ suggestion and analysed baseline and final GFR (ramipril 29·4 [4·0] and 16·2 [2·2] mL min ⫺1 1·73 m⫺2, conventional 25·8 [2·0] and 16·5 [1·4] mL min⫺1 1·73 respectively) and serum m ⫺2, creatinine (ramipril 228·6 [30·5] and 556·3 [68·6] mmol/L, conventional 236·2 [15·2] and 502·9 [30·5] mmol/L, respectively) in patients who progressed to ESRF and have found
1906
that they are similar. Moreover, lower incidence of ESRF in ramipril-treated patients was not explained by better blood pressure control, as shown by similar baseline and final diastolic blood pressures (DBP) in ramipril (88·6 [4·4] and 92·6 [3·9] mm Hg) and conventionally (88·7 [2·8] and 89·4 [2·3] mm Hg) treated patients. DBP averaged throughout the whole study period was very similar in the two treatment groups (ramipril 87·2 [2·1] mm Hg, conventional 88·2 [1·2] mm Hg). In a large group of patients (50 on ramipril and 46 on conventional treatment) in the REIN study, we also had data on DBP for 10 h after study drug administration. We used these data to assess whether, in addition to pretreatment (trough) DBP, posttreatment DBP was also similar in the two treatment groups. DBP measured at each timepoint from 0 min (trough value) to 600 min and averaged throughout the study period, was very similar in the two treatment groups. The peak effect on DBP was at 360 min in both groups, and the peak/trough DBP ratios were very similar in ramipril (0·92) and conventionally (0·95) treated patients. These data support the possibility that the renoprotective effect of the study drug did not depend on an increased blood-pressure-lowering effect throughout the study. ACE inhibitors may induce a shortterm reduction in GFR that may mask, at least in part, their effect in decreasing GFR decline in the long term. Pruna and colleagues suggest that the extent of GFR recovery after withdrawal of ACE inhibitors should be assessed. However, to derive meaningful information from this approach, we have to assume that patients’ clinical conditions did not change over the 6 months from study conclusion to now. Obviously, this may not be the case. However, we do not think it reasonable to by-pass a confounding variable by introducing another one. We therefore compared GFR decline in the two treatment groups without considering baseline GFR. The monthly decline improved from 0·26 (0·05) to 0·21 (0·05) mL/min in ramipril patients, but did not change in controls. This resulted in larger monthly difference in GFR decline between the two groups (8 mL/min instead of 3 mL/min). The measured GFR decline (excluding baseline GFR) closely reflected the GFR decline that we assumed 1 at the time we estimated the sample size needed to detect as significant a 25% difference between the two groups
over a follow-up of 5 years (0·22 vs 0·29 mL min ⫺1 1·73 m⫺2 per month). Piero Ruggenenti, Annalisa Perna, *Giuseppe Remuzzi Mario Negri Institute for Pharmacological Research, Clinical Research Centre for Rare Diseases “Aldo e Cele Daccò” Villa Camozzi, Ranica, and *Unit of Nephrology, Ospedali Riuniti, 24125 Bergamo, Italy (e-mail: [email protected]) 1
Gruppo Italiano Studi Epidemiologici in Nefrologia (GISEN). A long-term randomized trial to evaluate the effects of ramipril on the evolution of renal function in chronic nephropathies. J Nephrol 1991; 3: 193–202.
Male infertility and increased risk of diseases in future generations Sir—There have been concerns that the offspring conceived by intracytoplasmic sperm injection (ICSI) may have an increased risk of health problems. The concerns originate from the following: (1) the extent of heritable conditions among offspring, inherited from an affected parent, but which may have been associated with infertility before the introduction of ICSI; (2) the possible consequences of ICSI itself, which involves an invasive and traumatic procedure with potentially toxic chemical baths; and (3) the possible adverse consequences of the use of sperm that in many cases would not have been fit or mature enough to achieve fertilisation before the introduction of ICSI. The limited empirical evidence to test such concerns is still inadequate, especially because of insufficient offspring in published series. Aneta Dowsing and colleagues’ report (Aug 21, p 640)1 raises the possibility of an increased risk of neurodegenerative diseases among children conceived by ICSI. Confident assessment of this suggestion will require more data, as will other suggestions—eg, that such offspring may be at an increased risk of birth defects, paediatric developmental delay, or cancer. With respect to birth defects, consider conditions affecting 2% of births: to detect a doubling in that degree of risk with a 5% one-tailed test of significance, and to ensure an 80% chance of detecting such an excess, we would need 1000 ICSI offspring and 1000 control offspring. For a disorder that affects 1 in 1000 births, the corresponding number of offspring would be 20 000 in each group. One of the largest series investigated 423 such offspring; the data were subject to detailed independent review.2 The original investigators concluded
THE LANCET • Vol 354 • November 27, 1999
that there was no evidence of an excess of major birth defects among ICSI offspring, whereas the independent reviewers concluded that there was a two-fold excess. Mitchell 3 in an accompanying editorial suggested this discrepancy arose as a result of different definitions and methods of ascertainment of birth defects. Mitchell also discussed apparent differences between an expansion of the original 423 ICSI offspring to 877,4 and a further series of 578 ICSI offspring conceived in New York, and concluded that different methods of ascertainment of birth defects could explain a large element of the apparent discrepancy in the risk estimates of major birth defects. Therefore, methods of ascertainment of birth defects in ICSI and control offspring should be comparable, and the sample sizes should be sufficiently large to ensure results may be interpreted with confidence. Others have strongly advocated further large scale follow-up of ICSI offspring.5 To address the uncertainty that remains, carefully designed epidemiological systems to monitor these offspring should be established so that potential parents may consider the evidence of risk in a more informed way than is possible at present. An important step towards this objective is the establishment of a national register of offspring conceived by ICSI. With the encouragement of the British Fertility Society and the guidance of an Advisory Group we plan to investigate the risks of adverse health outcomes among offspring conceived by ICSI and compare them with those of offspring conceived by conventional IVF, and with those of offspring conceived naturally in the general population. We already have the cooperation of almost every infertility clinic in the UK that carries out ICSI. *M M Hawkins, C L R Barratt, A G Sutcliffe, I D Cooke *Centre for Childhood Cancer Survivor Studies, Department of Public Health and Epidemiology, University of Birmingham, Birmingham B15 2TT, UK; Reproductive Biology and Genetics Group, Department of Obstetrics and Gynaecology, Birmingham Women’s Hospital, Birmingham; Royal Free and University College Medical School, University College London; and Department of Obstetrics and Gynaecology (Jessop), Division of Surgical and Anaesthetic Sciences, Jessop Hospital for Women, Sheffield,UK Dowsing AT, Yong EL, Clark M, et al. Linkage between male infertility and trinucleotide repeat expansion in the androgen-receptor gene. Lancet 1999; 354: 640–43. 2 Kurinczuk JJ, Bower C. Birth defects in infants conceived by intracytoplasmic sperm injection: an alternative interpretation. BMJ 1997; 315: 1260–65. 3 Mitchell AA. Intracytoplasmic sperm injection: offering hope for a term pregnancy
and a healthy child? BMJ 1997; 315: 1245–46. 4 Bonduelle M, Wilikns A, Buysse A, et al. Prospective follow-up study of 877 children born after intracytoplasmic sperm injection (ICSI), with ejaculated epididymal and testicular spermatozoa and after replacement of cryopreserved embryos obtained after ICSI. Hum Reprod 1996; 11 (suppl 4): 131–55. 5 te Velde ER, van Baar AL, va Kooij RJ. Concerns about assisted reproduction. Lancet 1998; 351: 1524–25.
Sir—Aneta Dowsing and colleagues1 propose a relation between the length of the poly-glutamine stretch (CAG repeat) in the human androgen receptor and the risk of defective spermatogenesis based on the study of 35 infertile men and 32 controls. They make the assumption that transmission of elongated CAG repeats through assisted reproductive techniques may lead to an increased risk of spinal and bulbar muscular atrophy (Kennedy’s disease) in future generations. We have investigated a cohort of 180 white patients aged 20·9 to 54·4 years who were referred between 1993 and 1997 from the Department of Andrology at the University Hospital Hamburg-Eppendorf. Clinical examinations excluded overt virilisation disorders and other genetic, endocrine, infectious, or obstructive causes of male infertility. Ejaculates were investigated according to WHO guidelines.2 110 patients showed oligoteratozoospermia (<106 sperm/mL ejaculate, <30% normal forms), 57 had teratozoospermia (30% normal sperm), and 13 patients had azoospermia. 53 white men served as controls, all of whom had fathered at least one child. Informed consent was obtained from all individuals and EDTA-blood samples were taken for DNA analysis of the human androgen receptor gene. PCRgenerated amplification products containing the variable CAG-repeat region were electrophoresed on nondenaturing polyacrylamide gels with
adequate size markers and visualised with silver staining. 3 Additionally, selected samples were sequenced to assess the exact number of CAG repeats, and these control samples were included in electrophoresis. CAG repeat sizes were determined by a computerised analysis system (ImageMaster, Pharmacia, Freiburg, Germany). In the patient group, the mean CAG repeat length was 23 repeats (range 13–30), whereas in the control group the mean was 24 repeats (17–39; figure). Wilcoxon and Mann-Whitney tests were used for statistical analysis and showed no significant differences between the two groups (p=0·471). This finding agrees with Giwercman and colleagues’ report on the CAG repeat length in a Swedish population of 33 infertile men.4 Thus, we cannot confirm the assumptions of Dowsing and colleagues and conclude that polyglutamine repeat expansion within the androgen-receptor gene in infertile men does not play a significant part in male infertility. This study was supported by grants from the German Ministry of Education and Research (BMBF 01KY9301/1 to OH and GHGS) and German Research Foundation (DFG Hi 497/3-3 to OH).
*Olaf Hiort, Thorsten Horter, Wolfgang Schulze, Britta Kremke, Gernot H G Sinnecker *Departments of Paediatrics, Medical University of Lübeck, D-23538 Lübeck, Germany; City Hospital, Wolfsburg; Children’s Hospital, Hamburg-Altona; and Department of Andrology, University Hospital, Hamburg-Eppendorf (e-mail: [email protected]) 1
2
Dowsing AT, Young EL, Clark M, McLachlan RI, de Kretser DM, Trounson AO. Linkage between male infertility and trinucleotide repeat expansion in the androgen-receptor gene. Lancet 1999; 354: 640–43. WHO. WHO laboratory manual for the examination of human semen and spermcervial mucus interaction, 3rd edn. Cambridge: Cambridge University Press, 1992.
1
THE LANCET • Vol 354 • November 27, 1999
Variation of CAG repeat sizes in 180 infertile men and 53 controls
1907
Ruggenenti P, Perna A, Gherardi G, et al. Renoprotective properties of ACEinhibition in non-diabetic nephropathies with non-nephrotic proteinuria. Lancet 1999; 354: 359–64. 2 Wiggam MI, Bell PM, Sheridan B, Walmsley A, Atkinson AB. Low dose bendrofluazide (1·25 mg) effectively lowers blood pressure over 24 h. Results of a randomized, double-blind, placebocontrolled crossover study. Am J Hypertens 1999; 12: 528–31. 3 Zanad F, Matzinger A, Larché J. Trough/peak ratios of once daily ACE inhibitors and calcium antagonists. Am J Hypertens 1996; 9: 633–43. 4 Wentzel U, Helmchen U, Schoeppe W, Schwietzer G. Combination treatment of enalapril with nitrendipine in rats with renovascular hypertension. Hypertension 1994; 23: 114–22. 5 Mathiesen ER, Hommel E, Hansen HP, Schmidt UM, Parving MH. Randomized controlled trial of long term efficacy of captopril on preservation of kidney function in normotensive patients with insulin dependent diabetes and microalbuminuira. BMJ 1999; 319: 24–25.
Authors’ reply Sir—In stratum 1 of the REIN study, as A Pruna and colleagues correctly point out, ramipril reduced incidence of ESRF in patients with chronic nephropathies and non-nephrotic proteinuria more than the rate of GFR decline. This difference was due to the low rate of progression of patients with mild proteinuria that diluted the effect of ramipril on the rate of GFR decline rather than to imbalances in baseline characteristics of patients. The renoprotective effect of ramipril was higher in patients with baseline GFR less than 45 mL/min 1·73 m2 who had very similar GFRs at baseline (ramipril 30·3 [SE 1·3] mL min⫺1 1·73 m⫺2, conventional 30·4 [1·4] mL min⫺1 1·73 m⫺2). However, we have followed Pruna and colleagues’ suggestion and analysed baseline and final GFR (ramipril 29·4 [4·0] and 16·2 [2·2] mL min ⫺1 1·73 m⫺2, conventional 25·8 [2·0] and 16·5 [1·4] mL min⫺1 1·73 respectively) and serum m ⫺2, creatinine (ramipril 228·6 [30·5] and 556·3 [68·6] mmol/L, conventional 236·2 [15·2] and 502·9 [30·5] mmol/L, respectively) in patients who progressed to ESRF and have found
1906
that they are similar. Moreover, lower incidence of ESRF in ramipril-treated patients was not explained by better blood pressure control, as shown by similar baseline and final diastolic blood pressures (DBP) in ramipril (88·6 [4·4] and 92·6 [3·9] mm Hg) and conventionally (88·7 [2·8] and 89·4 [2·3] mm Hg) treated patients. DBP averaged throughout the whole study period was very similar in the two treatment groups (ramipril 87·2 [2·1] mm Hg, conventional 88·2 [1·2] mm Hg). In a large group of patients (50 on ramipril and 46 on conventional treatment) in the REIN study, we also had data on DBP for 10 h after study drug administration. We used these data to assess whether, in addition to pretreatment (trough) DBP, posttreatment DBP was also similar in the two treatment groups. DBP measured at each timepoint from 0 min (trough value) to 600 min and averaged throughout the study period, was very similar in the two treatment groups. The peak effect on DBP was at 360 min in both groups, and the peak/trough DBP ratios were very similar in ramipril (0·92) and conventionally (0·95) treated patients. These data support the possibility that the renoprotective effect of the study drug did not depend on an increased blood-pressure-lowering effect throughout the study. ACE inhibitors may induce a shortterm reduction in GFR that may mask, at least in part, their effect in decreasing GFR decline in the long term. Pruna and colleagues suggest that the extent of GFR recovery after withdrawal of ACE inhibitors should be assessed. However, to derive meaningful information from this approach, we have to assume that patients’ clinical conditions did not change over the 6 months from study conclusion to now. Obviously, this may not be the case. However, we do not think it reasonable to by-pass a confounding variable by introducing another one. We therefore compared GFR decline in the two treatment groups without considering baseline GFR. The monthly decline improved from 0·26 (0·05) to 0·21 (0·05) mL/min in ramipril patients, but did not change in controls. This resulted in larger monthly difference in GFR decline between the two groups (8 mL/min instead of 3 mL/min). The measured GFR decline (excluding baseline GFR) closely reflected the GFR decline that we assumed 1 at the time we estimated the sample size needed to detect as significant a 25% difference between the two groups
over a follow-up of 5 years (0·22 vs 0·29 mL min ⫺1 1·73 m⫺2 per month). Piero Ruggenenti, Annalisa Perna, *Giuseppe Remuzzi Mario Negri Institute for Pharmacological Research, Clinical Research Centre for Rare Diseases “Aldo e Cele Daccò” Villa Camozzi, Ranica, and *Unit of Nephrology, Ospedali Riuniti, 24125 Bergamo, Italy (e-mail: [email protected]) 1
Gruppo Italiano Studi Epidemiologici in Nefrologia (GISEN). A long-term randomized trial to evaluate the effects of ramipril on the evolution of renal function in chronic nephropathies. J Nephrol 1991; 3: 193–202.
Male infertility and increased risk of diseases in future generations Sir—There have been concerns that the offspring conceived by intracytoplasmic sperm injection (ICSI) may have an increased risk of health problems. The concerns originate from the following: (1) the extent of heritable conditions among offspring, inherited from an affected parent, but which may have been associated with infertility before the introduction of ICSI; (2) the possible consequences of ICSI itself, which involves an invasive and traumatic procedure with potentially toxic chemical baths; and (3) the possible adverse consequences of the use of sperm that in many cases would not have been fit or mature enough to achieve fertilisation before the introduction of ICSI. The limited empirical evidence to test such concerns is still inadequate, especially because of insufficient offspring in published series. Aneta Dowsing and colleagues’ report (Aug 21, p 640)1 raises the possibility of an increased risk of neurodegenerative diseases among children conceived by ICSI. Confident assessment of this suggestion will require more data, as will other suggestions—eg, that such offspring may be at an increased risk of birth defects, paediatric developmental delay, or cancer. With respect to birth defects, consider conditions affecting 2% of births: to detect a doubling in that degree of risk with a 5% one-tailed test of significance, and to ensure an 80% chance of detecting such an excess, we would need 1000 ICSI offspring and 1000 control offspring. For a disorder that affects 1 in 1000 births, the corresponding number of offspring would be 20 000 in each group. One of the largest series investigated 423 such offspring; the data were subject to detailed independent review.2 The original investigators concluded
THE LANCET • Vol 354 • November 27, 1999
that there was no evidence of an excess of major birth defects among ICSI offspring, whereas the independent reviewers concluded that there was a two-fold excess. Mitchell 3 in an accompanying editorial suggested this discrepancy arose as a result of different definitions and methods of ascertainment of birth defects. Mitchell also discussed apparent differences between an expansion of the original 423 ICSI offspring to 877,4 and a further series of 578 ICSI offspring conceived in New York, and concluded that different methods of ascertainment of birth defects could explain a large element of the apparent discrepancy in the risk estimates of major birth defects. Therefore, methods of ascertainment of birth defects in ICSI and control offspring should be comparable, and the sample sizes should be sufficiently large to ensure results may be interpreted with confidence. Others have strongly advocated further large scale follow-up of ICSI offspring.5 To address the uncertainty that remains, carefully designed epidemiological systems to monitor these offspring should be established so that potential parents may consider the evidence of risk in a more informed way than is possible at present. An important step towards this objective is the establishment of a national register of offspring conceived by ICSI. With the encouragement of the British Fertility Society and the guidance of an Advisory Group we plan to investigate the risks of adverse health outcomes among offspring conceived by ICSI and compare them with those of offspring conceived by conventional IVF, and with those of offspring conceived naturally in the general population. We already have the cooperation of almost every infertility clinic in the UK that carries out ICSI. *M M Hawkins, C L R Barratt, A G Sutcliffe, I D Cooke *Centre for Childhood Cancer Survivor Studies, Department of Public Health and Epidemiology, University of Birmingham, Birmingham B15 2TT, UK; Reproductive Biology and Genetics Group, Department of Obstetrics and Gynaecology, Birmingham Women’s Hospital, Birmingham; Royal Free and University College Medical School, University College London; and Department of Obstetrics and Gynaecology (Jessop), Division of Surgical and Anaesthetic Sciences, Jessop Hospital for Women, Sheffield,UK Dowsing AT, Yong EL, Clark M, et al. Linkage between male infertility and trinucleotide repeat expansion in the androgen-receptor gene. Lancet 1999; 354: 640–43. 2 Kurinczuk JJ, Bower C. Birth defects in infants conceived by intracytoplasmic sperm injection: an alternative interpretation. BMJ 1997; 315: 1260–65. 3 Mitchell AA. Intracytoplasmic sperm injection: offering hope for a term pregnancy
and a healthy child? BMJ 1997; 315: 1245–46. 4 Bonduelle M, Wilikns A, Buysse A, et al. Prospective follow-up study of 877 children born after intracytoplasmic sperm injection (ICSI), with ejaculated epididymal and testicular spermatozoa and after replacement of cryopreserved embryos obtained after ICSI. Hum Reprod 1996; 11 (suppl 4): 131–55. 5 te Velde ER, van Baar AL, va Kooij RJ. Concerns about assisted reproduction. Lancet 1998; 351: 1524–25.
Sir—Aneta Dowsing and colleagues1 propose a relation between the length of the poly-glutamine stretch (CAG repeat) in the human androgen receptor and the risk of defective spermatogenesis based on the study of 35 infertile men and 32 controls. They make the assumption that transmission of elongated CAG repeats through assisted reproductive techniques may lead to an increased risk of spinal and bulbar muscular atrophy (Kennedy’s disease) in future generations. We have investigated a cohort of 180 white patients aged 20·9 to 54·4 years who were referred between 1993 and 1997 from the Department of Andrology at the University Hospital Hamburg-Eppendorf. Clinical examinations excluded overt virilisation disorders and other genetic, endocrine, infectious, or obstructive causes of male infertility. Ejaculates were investigated according to WHO guidelines.2 110 patients showed oligoteratozoospermia (<106 sperm/mL ejaculate, <30% normal forms), 57 had teratozoospermia (30% normal sperm), and 13 patients had azoospermia. 53 white men served as controls, all of whom had fathered at least one child. Informed consent was obtained from all individuals and EDTA-blood samples were taken for DNA analysis of the human androgen receptor gene. PCRgenerated amplification products containing the variable CAG-repeat region were electrophoresed on nondenaturing polyacrylamide gels with
adequate size markers and visualised with silver staining. 3 Additionally, selected samples were sequenced to assess the exact number of CAG repeats, and these control samples were included in electrophoresis. CAG repeat sizes were determined by a computerised analysis system (ImageMaster, Pharmacia, Freiburg, Germany). In the patient group, the mean CAG repeat length was 23 repeats (range 13–30), whereas in the control group the mean was 24 repeats (17–39; figure). Wilcoxon and Mann-Whitney tests were used for statistical analysis and showed no significant differences between the two groups (p=0·471). This finding agrees with Giwercman and colleagues’ report on the CAG repeat length in a Swedish population of 33 infertile men.4 Thus, we cannot confirm the assumptions of Dowsing and colleagues and conclude that polyglutamine repeat expansion within the androgen-receptor gene in infertile men does not play a significant part in male infertility. This study was supported by grants from the German Ministry of Education and Research (BMBF 01KY9301/1 to OH and GHGS) and German Research Foundation (DFG Hi 497/3-3 to OH).
*Olaf Hiort, Thorsten Horter, Wolfgang Schulze, Britta Kremke, Gernot H G Sinnecker *Departments of Paediatrics, Medical University of Lübeck, D-23538 Lübeck, Germany; City Hospital, Wolfsburg; Children’s Hospital, Hamburg-Altona; and Department of Andrology, University Hospital, Hamburg-Eppendorf (e-mail: [email protected]) 1
2
Dowsing AT, Young EL, Clark M, McLachlan RI, de Kretser DM, Trounson AO. Linkage between male infertility and trinucleotide repeat expansion in the androgen-receptor gene. Lancet 1999; 354: 640–43. WHO. WHO laboratory manual for the examination of human semen and spermcervial mucus interaction, 3rd edn. Cambridge: Cambridge University Press, 1992.
1
THE LANCET • Vol 354 • November 27, 1999
Variation of CAG repeat sizes in 180 infertile men and 53 controls
1907