REPRODUCIBILITY AND RELIABILITY OF REPEATED SEMEN ANALYSES IN MALE PARTNERS OF SUBFERTILE COUPLES

Transcription

1 REPRODUCIBILITY AND RELIABILITY OF REPEATED SEMEN ANALYSES IN MALE PARTNERS OF SUBFERTILE COUPLES Esther Leushuis Jan Willem van der Steeg Pieternel Steures Sjoerd Repping Patrick M.M. Bossuyt Marinus A. Blankenstein Ben Willem J. Mol Fulco van der Veen Peter G.A. Hompes FERTILITY AND STERILITY 21 DEC;947:2615

2 IMPACT OF THE MALE FACTOR ON THE PREDICTION OF NATURAL CONCEPTION ABSTRACT OBJECTIVE To determine the precise degree of variability that is represented by the reproducibility and reliability of semen analysis. The general assumpon is that semen analyses need to be repeated because of a high degree of within-individual variability. However, the precise degree of variability is not well established in male partners of subferle couples. DESIGN Retrospecve cohort study. SETTING Two university hospitals in the Netherlands, which rounely perform two semen analyses in the male partner of subferle couples. PATIENTS Male partners of subferle couples. MAIN OUTCOME MEASURES We assessed the test-retest reproducibility, by calculang the coefficient of variaon (CV W ) for five semen parameters. The CV W expresses, on a relave scale, the degree of closeness of repeated measurements taken in the same subject. We also esmated the reliability of these semen parameters, in terms of the intraclass correlaon coefficient, which express the rao of the between-subject variability over the total variability. RESULTS We analysed the data of 5,24 men and found that the CV W of all semen parameters ranged from 28% to 4%. The intraclass correlaon coefficients of these semen parameters were moderate to high: volume:.7; concentraon:.89; molity:.58; morphology:.6; total mole count:.7. CONCLUSIONS This study affirmed the presumed large within-subject variability and the limited reproducibility of semen analyses in subferle men. Whether this degree of variability within men jusfies one or more repeons of the semen analysis in view of consequences for clinical management should be the topic of future studies. Unl then it seems reasonable to perform two semen analyses. 5

3 CHAPTER REPRODUCIBILITY AND RELIABILITY OF REPEATED SEMEN ANALYSES IN MALE PARTNERS OF SUBFERTILE COUPLES INTRODUCTION Semen analysis is the cornerstone of the laboratory evaluaon of the male partner of subferle couples. 1;2 It is well known that semen analyses of the same man vary from one ejaculate to the other. -7 The results of repeated semen analyses can vary in principle because of three factors: (1) preanalyc influences, such as duraon of absnence or seasonality; (2) analyc variaon in the method of analysis and/or the evaluator; and () inherent biologic variability. 5;8-1 The majority of reports on the reproducibility and reliability of the semen analysis involve only healthy men, but it may well be that the variability in a subferle populaon differs from that in the general populaon. 6 This is an important disncon, because the subferle populaon is the target group for performing a semen analysis during the ferlity workup. One of the earliest reports on the quanficaon of the variability of semen parameters within men from a subferle populaon did not take agreement beyond chance into account. 1 Another study reported high variability of semen parameters within one man in terms of coefficients of variaon, but these samples were collected specifically for IUIs. The authors of the laer study also reported that the fluctuaons in the samples of one man were substanally smaller than the fluctuaons between men. 9 In an aempt to achieve a more reliable approximaon of the true values of individual semen parameters, several guidelines recommend to repeat the semen analysis once or twice in the basic ferlity workup. ;11;12 Yet, the first step in deciding whether or not to repeat the semen analysis in the basic ferlity workup would be to assess the actual degree of within-subject variability that is represented by the reproducibility and reliability in male partners of subferle couples. This was the aim of the study reported in this arcle, which was based on two rounely performed semen analyses in a large cohort of male partners of subferle couples. MATERIALS AND METHODS Paents We searched the consecuve electronic records of male partners of subferle couples at two university hospitals in the Netherlands (Vrije Universiteit medical center [VUmc] and Academic Medical Center [AMC], Amsterdam). For one hospital (AMC) records were available of men vising the hospital between January 1998 and 28. For the other hospital (VUmc) data were available for men between January 24 and August 28. In all men, two semen analyses had been performed rounely as part of the basic ferlity workup. Morphology was registered in only one center (AMC). Because the data were collected during standard paent treatment and were handled anonymously in our study, no Instuonal Review Board approval was required according to Dutch law. Semen analysis Semen analyses were performed according to the World Health Organizaon (WHO) 51

4 IMPACT OF THE MALE FACTOR ON THE PREDICTION OF NATURAL CONCEPTION guideline. Men were advised to maintain days of sexual absnence before semen analysis. The interval of the me of absnence was recorded when the sample was handed in. Semen samples were collected at the hospital or at home by masturbaon directly into a 5-mL polyethylene jar. Semen samples were analyzed within 1 hour aer ejaculaon. Aer liquefacon, semen volume was measured in a graded tube with.1-ml accuracy. Sperm concentraon was counted and molity was assessed in a Makler counng chamber at a magnificaon of x2. Sperm concentraon, molity and morphology assessment were performed by trained laboratory technicians. Total mole count (TMC) (prewash) was calculated as semen volume mulplied by sperm concentraon mulplied by the percentage of sperm with progressive molity (WHO grade A). Data analysis We calculated the mean difference between the two analyses for all semen parameters and the associated 95% confidence intervals (CI), to see whether there was a systemac difference between measurements. We first calculated the reproducibility of semen parameters by evaluang the degree of within-subject variability between the two analyses. We did so by calculang within-subjects coefficient of variaon (CV W ) for each parameter. The CV W expresses the within-subjects SD relave to the subject s true score. 1 It expresses, on a relave scale, the degree of closeness of repeated measurements taken in the same subject, on different occasions under the same condi- ons. CV W values close to zero indicate beer reproducibility and less variability within subjects. With a CV W of, repeated measures are always idencal and no within-subject variability exists. As not all semen parameters are normally distributed, we applied a natural logarithmic transformaon to sperm concentraon and TMSC and a square root transformaon to sperm morphology. In addion to the CV W, we calculated the probability (p) that the two semen analyses of one man differ by a factor (k) or more given a certain CV W, to provide a more clinically applicable stasc (Figure 1). 14 Figure 1. This formula (1) calculated the probability that two semen analyses differ by a factor k or more for a specified value of the CV. For example (2), the frequency of two semen analyses that differ by 1% or more (so that k = 1.1) with a CV of % is 45%. p( k) 2 log e ( k) 2 2log e ( CV 1) (1) log (1.1) e p (1.1) 2.45 (2) 2 2log e ( /1) 1) 52

5 CHAPTER REPRODUCIBILITY AND RELIABILITY OF REPEATED SEMEN ANALYSES IN MALE PARTNERS OF SUBFERTILE COUPLES We also esmated the implicaons of the within-subject variability for the reliability of semen analysis parameters, the extent to which these parameters can be used to disnguish between men. For this purpose, we calculated the intraclass correlaon coefficient for each parameter (ICC). The ICC expresses the proporon of the total variability that can be aributed to the genuine variability between men, the remainder being within-subject fluctuaons and random error. 15 This means that, where the CV W looks at within-subject fluctuaons only, the ICC looks at the consequences of these fluctuaons liming reproducibility for disnguishing between men. The ICC takes values in the range to 1 with values close to 1 indicang that the contribuon of the within-subject fluctuaons is relavely small compared with the genuine differences between men. With a perfect ICC of 1 all differences in semen parameters between men are genuine, and a repeated semen analysis will be idencal to the first semen analysis of that man. With an ICC of, all differences between two semen analyses within a man are due to within-man fluctuaons or error. The ICC is interpreted similar to κ, with values >.8 indicang outstanding reliability, and values between.6 to.79 presenng substanal reliability. 16 The agreement between repeated semen analyses was also presented graphically with use of scaer plots. The cutoff values according to the WHO reference values were displayed in the plots with vercal lines on the reference value for the first semen analysis and a horizontal line on the second semen analysis. The number of concordant or disconcordant pairs was reported in percentages of the total cohort. In addion to obtaining measures of within-subject fluctuaons, we examined a potenal determinant of these. The effect of the duraon of absnence on the different semen parameters was assessed by a repeated-measurements procedure with use of mixed-effects models. Calculaons were performed with use of SPSS 16. program (SPSS Inc., Chicago, IL) and SAS System 9.1 (SAS Instute, Cary, NC). RESULTS We analysed the data of 5,24 men with two rounely performed semen analyses: 2,941 from the AMC and 2,299 from the VUmc. The results for the first and second semen analysis and the average differences between the two samples are summarized in Table 1. The mean male age at the first semen analysis was 6. years (5 th to 95 th percenle: 27 to 47 years). The mean number of days of absnence for the first semen analysis was 4 days (5 th to 95 th percenle: 2 to 7 days). The median number of days between the two semen analyses was 19 days (5 th to 95 th percenle: 1 to 292 days). There was a significant but small difference between the semen analyses for semen volume, and sperm molity (all grades). No differences between the two semen analyses were found for the other semen parameters. The CV W were comparable and high for all semen parameters, ranging from 28% to 4% (Table 2). The probability to detect a difference of 1% or more between two semen analyses for the same man was.45, given the corresponding CV W of %. There was more variability in the corresponding ICC. The ICC for sperm concentraon was high (ICC:.89), indicang excellent reliability despite its limited reproducibility. 5

6 IMPACT OF THE MALE FACTOR ON THE PREDICTION OF NATURAL CONCEPTION Table 1. Characteriscs of the first and second semen analysis. First semen analysis Second semen analysis Systematic differences between the two semen analyses percentile percentile percentile n mean 5th 95th n mean 5th 95th n mean 5th 95th Semen parameters Volume (ml) 5, , , Concentration (1*6/mL) (median) 5, , , Motility (%) WHO grade A 5, , , WHO grade B 5, , , WHO grade A+B 5, , , Morphology (% normal) a 2, , , TMC b (1*6) (median) 4, , , a According to WHO b Calculated as volume * concentraon * molity/1. a Table 2. Intraclass correlaon coefficients and CVW of semen parameters in two semen samples for each man (N=5,24). Semen parameter CVw 95% confidence interval ICC 95% confidence interval lower limit upper limit lower limit upper limit Semen volume 28% 27% 28% Sperm concentration 29% % 28% Sperm motility 4% % 4% Sperm morphology 29% 28% % Total motile count % 1% 29%

7 CHAPTER REPRODUCIBILITY AND RELIABILITY OF REPEATED SEMEN ANALYSES IN MALE PARTNERS OF SUBFERTILE COUPLES In contrast, the reliability of sperm molity and morphology was moderate, represented by ICCs of.59 and.6, respecvely. The ICC of semen volume was.7 (Table 2). The percentage of concordant results between the two semen analyses according to the WHO reference values between the first and second semen analysis ranged from 61% tot 7% and the percentage with discordant results ranged from 6% to 21% (Figure 2). The duraon of absnence had an effect on the results of all semen parameters (all P values <.), except on morphology (F value <.1, P value.99) (Table ). Increasing duraon of absnence was associated with decreasing sperm molity and morphology, but with increasing semen volume, sperm concentraon and TMSC. Table. The effect of the duraon of absnence and seasonality on the within-subject variability. Effect of the duration of abstinence Semen parameter ß a df F P value Semen volume.81 2, <.1 Sperm concentration.5 2, Sperm motility -.9 2, <.1 Sperm morphology -.1 1, Total motile count.74 2, <.1 a A posive ß represents an increase in the variability of the semen parameters within men with increasing days of 2 absnence or seasonality. A negave ß represents a negave associaon. DISCUSSION In this study, we described the within-individual variability of semen analyses in male partners of subferle couples in a large retrospecve cohort. This study affirmed the presumed large within-subject variability of semen analyses in subferle men, represented by the limited reproducibility. The low reliability of sperm molity and morphology indicated that the within-subject variability is high relave to the between-subject variability, making it difficult to classify men on the basis of a single measurement. In contrast, the high reliability of sperm concentraon provided beer classificaon potenal based on a single measurement. The reliability of semen volume was moderate. The strength of this study is that we analysed an unselected large cohort of male partners of subferle couples that had two semen analyses rounely performed. Numerous studies have evaluated the within-subject variability of various semen parameters, but no study has assessed this variability in >5, male partners of subferle couples. 55

8 IMPACT OF THE MALE FACTOR ON THE PREDICTION OF NATURAL CONCEPTION Figure 2(a). Scaerplot of the results of the first semen analysis versus the second for semen volume. A N=412 8% 12 N=,687 7% Semen volume (second semen analysis) (ml) N=68 1% N=458 9% Semen volume (first semen analysis) (ml) Figure 2(b). Scaerplot of the results of the first semen analysis versus the second for sperm concentraon. B N=48 7% N=,429 65% 25 Sperm concentration (second semen analysis) (milj/l) N=1,12 21% N=4 7% Sperm concentration (first semen analysis) (milj/l) 56

9 CHAPTER REPRODUCIBILITY AND RELIABILITY OF REPEATED SEMEN ANALYSES IN MALE PARTNERS OF SUBFERTILE COUPLES Figure 2(c). Scaerplot of the results of the first semen analysis versus the second for sperm molity. C N=575 11% 1 N=,214 61% Sperm motility (second semen analysis) (%) N=721 14% Sperm motility (first semen analysis) (%) N=58 1% Figure 2(d). Scaerplot of the results of the first semen analysis versus the second for sperm morphology. D N=412 14% 1 N=1,866 6% 8 Sperm morphology (second sperm analysis) (%) N=19 6% Sperm morphology (first semen analysis) (%) N=6 2% 57

10 IMPACT OF THE MALE FACTOR ON THE PREDICTION OF NATURAL CONCEPTION The two studies that previously reported CV W on semen parameters, respecvely of men from an IUI program 9 and a retrospecve analysis of 74 inferle men and 65 normal men producing five or more ejaculates each 6, reported CV W similar to the ones obtained in our study, arriving at similar conclusions about the high within-subject variability. These CV W were 27.1% and 29.6% for semen volume, compared with 28% in our study. Results were also similar for sperm molity with CV W of respecvely 4.4% and 4.2%, versus 4% in our study. The results on sperm concentraon were more contrasng, with CVw of 54.2% and 56.8% in the two other studies, compared with 29% in our study. Sperm morphology was not analysed in these two other studies. These previous studies reported ICCs that were higher than in our study:.84 and.92 for semen volume, compared to.7 in our study;.78 and.88 for sperm molity, compared with.58 in our study; and.92 for both studies, compared with.89 in our study. The high ICC in these studies might be explained by a more heterogeneous study populaon compared with our study populaon. The same semen parameter will have a higher ICC in a heterogeneous populaon compared with a homogenous populaon, because the proporon of differences between men will be larger relave to the overall variability. In our study, all semen parameters, except sperm morphology, were affected significantly by the duraon of absnence. A number of previous studies have invesgated the effect of absnence and seasonality on semen parameters. 9;17;18 Although there are reports in the literature on the effect of absnence on the variability of the semen analysis, the studies differ greatly in aim, study design and populaon, and the results differ accordingly. The results of our study indicate that is preferable to standardize duraon of absnence to lower the within-subject variability caused by this confounder and to increase reproducibility and reliability. Although this study affirms the presumed large variability of semen analyses in male partners of subferle couples, a problem that could be remedied by repeang the semen analysis, it is not enrely clear how the two semen analyses results, when different, should be combined and interpreted. The ulmate trial to evaluate this would be to correlate combinaons of results with capacity to predict spontaneous pregnancy. Unl then it seems reasonable to perform two semen analyses. 58

11 CHAPTER REPRODUCIBILITY AND RELIABILITY OF REPEATED SEMEN ANALYSES IN MALE PARTNERS OF SUBFERTILE COUPLES REFERENCES 1. The Male Inferlity Best Pracce Policy Commiee of the American Urological Associaon and The Pracce Commiee of the American Society for Reproducve Medicine. Report on opmal evaluaon of the inferle male. Ferlity and Sterility 26; 86(5, Supplement 1):S22-S de Kretser DM. Male inferlity. The Lancet 1997; 49(954): World Health Organizaon. WHO laboratory manual for the examinaon of human semen and semen-cervical mucus interacon. 4th ed. Cambridge, England: Cambridge University Press; Auger J, Eustache F, Ducot B, Blandin T, Daudin M, Diaz I et al. Intra- and inter-individual vari ability in human sperm concentraon, molity and vitality assessment during a workshop involving ten laboratories. Hum Reprod 2; 15(11): Alvarez C, Caslla JA, Marnez L, Ramirez JP, Vergara F, Gaforio JJ. Biological variaon of seminal parameters in healthy subjects. Hum Reprod 2; 18(1): Keel BA. Within- and between-subject variaon in semen parameters in inferle men and normal semen donors. Ferl Steril 26; 85(1): Mallidis C, Howard EJ, Baker HW. Variaon of semen quality in normal men. Int J Androl 1991; 14(2): Dawson B, Trapp BG. Basic & clinical biostascs. rd ed. Lange Medical books/mcgraw-hill; Francavilla F, Barbone A, Necozione S, Santucci R, Cordeschi G, Macerola B et al. Within-subject variaon of seminal parameters in men with inferle marriages. Int J Androl 27; (): Tielemans E, Heederik D, Burdorf A, Loomis D, Habbema DF. Intraindividual variability and redundancy of semen parameters. Epidemiology 1997; 8(1): Naonal Collaborang Centre for Women s and Children s Health. Clinical guideline 11: Ferlity: Assessment and treatment for people with ferlity problems Dutch Society of Obstetrics and Gynaecology. Guideline: basic ferlity work-up. NVOG-richtlijn nr 1 [1], Quan H, Shih WJ. Assessing reproducibility by the within-subject coefficient of variaon with random effects models. Biometrics 1996; 52(4): Reed GF, Lynn F, Meade BD. Use of coefficient of variaon in assessing variability of quantave assays. Clin Diagn Lab Immunol 22; 9(6): Fleiss JL. Stascal Methods for Rates and Proporons. New York: John Wiley; Landis JR, Koch GG. The measurement of observer agreement for categorical data. Biometrics 1977; (1): Levitas E, Lunenfeld E, Weiss N, Friger M, Har-Vardi I, Koifman A et al. Relaonship between the duraon of sexual absnence and semen quality: analysis of 9,489 semen samples. Ferlity and S Sterility 25; 8(6): de Jonge C., LaFromboise M, Bosmans E, Ombelet W, Cox A, Nijs M. Influence of the absnence period on human sperm quality. Ferl Steril 24; 82(1):