Sperm retrieval rates by micro-TESE versus conventional TESE in men with non-obstructive azoospermia—the assumption of independence in effect sizes might lead to misleading conclusions
Year: 2020
Type: letter
Abstract: Sir, We read with interest the article entitled ‘Sperm recovery and ICSI outcomes in men with non-obstructive azoospermia: a systematic review and meta-analysis’ (Corona et al., 2019), recently published in Human Reproduction Update. In this review, the authors attempted to summarize evidence from a variety of study designs on the effectiveness of sperm retrieval (SR) in patients with non-obstructive azoospermia (NOA), with a primary focus on the effectiveness of SR techniques, microdissection testicular sperm extraction (micro-TESE) and conventional TESE (cTESE). The main conclusions were that SR rates (SRR) of about 50% could be achieved in NOA patients, and such rates did not differ between micro-TESE and cTESE. We commend Corona and colleagues for their efforts in undertaking such an exhaustive review of the available data. While any effort to elucidate the treatment results in NOA is of potential value for the field of Andrology, we have identified a number of points of concern in their review. For reviews to be considered meta-analytic, specific guidelines for these studies must be followed. First, Corona et al. registered their study on PROSPERO (https://www.crd.york.ac.uk/PROSPERO/display_record.php?RecordID=92017), in which they stated that ‘all observational trials reporting SRR after cTESE and micro-TESE in subjects with NOA were included’. At first glance, it seemed that the authors included trials in which both techniques were used and therefore compared. However, in their manuscript, they used a different approach of including ‘…observational studies reporting SRR after cTESE or mTESE in subjects with NOA’ and therefore included not only trials that compared both techniques but also those that used data from case series with outcomes for cTESE or micro-TESE alone. Interestingly, they found that SRR by micro-TESE was significantly higher than cTESE (57 vs. 37%) when the analysis was limited to those studies directly comparing both techniques, thus corroborating other meta-analyses (Deruyver et al., 2014; Bernie et al., 2015a). By contrast, similar SRR was reported when non-comparative micro-TESE and cTESE studies were included in the analysis. Traditional meta-analytic techniques assume that effect sizes are independent (Cheung, 2019). However, the reported effect sizes are likely to be different for comparative and non-comparative trials as the studied populations will vary and likely be heterogeneous. Since the effect size relates to the specific population from the primary studies, this structure creates dependence, and the pooled results based on conventional meta-analytic methods might be misleading. Substantial evidence of bias in the reporting of these studies is noted in Corona et al.’s study, as up to 38% of included micro-TESE trials involved selected patient populations with an overall unfavorable prognosis, such as patients with previous failed SR, or men post-chemotherapy. By contrast, only ~6% of the cTESE trials included the so-called unfavorable patients, thus possibly over-estimating the SRR for cTESE (Table I and Supplementary Table SI). In routine practice, micro-TESE may be reserved for patients with worse clinical prognosis (Hsiao et al., 2011; Ashraf et al., 2013; Kalsi et al., 2015). mTESE: microdissection testicular sperm extraction; cTESE: conventional testicular sperm extraction; NNT: number needed to treat; NOA: non-obstructive azoospermia; SR: sperm retrieval; SCO: Sertoli cell-only; CI: confidence interval †Comparison of proportions by chi-square: (Campbell, 2007). ¶(Altman, 1998). ‡Patients with previous failed SR, SCO only, or men post-chemotherapy *Total number of patients with Sertoli cell-only reported in primary studies. mTESE: microdissection testicular sperm extraction; cTESE: conventional testicular sperm extraction; NNT: number needed to treat; NOA: non-obstructive azoospermia; SR: sperm retrieval; SCO: Sertoli cell-only; CI: confidence interval †Comparison of proportions by chi-square: (Campbell, 2007). ¶(Altman, 1998). ‡Patients with previous failed SR, SCO only, or men post-chemotherapy *Total number of patients with Sertoli cell-only reported in primary studies. Second, it is well-known that SRR depends on histopathology results, with poorer outcomes for Sertoli cell-only (SCO) patients than maturation arrest and hypospermatogenesis counterparts (Su et al., 1999; Esteves et al., 2014). The study by Corona et al. assessed the effect of some confounders and highlighted that SRR decreased as the number of patients with Klinefelter syndrome increased. However, the authors should have also looked at the prevalence of SCO patients among included studies. In our analysis of Corona et al. data, we found that the proportion of patients with SCO or tubular atrophy was higher in micro-TESE trials than cTESE trials (57.3 vs. 46.8%, relative risk [RR]: 1.11, 95% CI 1.01–1.21; P < 0.0001; Table I and Supplementary Table SI). We, therefore, reassessed the SR estimates of non-comparative studies as a function of histopathology and found that when the analysis was limited only to patients with SCO, micro-TESE resulted in a significantly higher SRR of 34.7% versus 31.2% with cTESE (P = 0.019; Table II and Supplementary Table SI). mTESE: microdissection testicular sperm extraction; cTESE: conventional testicular sperm extraction; NNT: number needed to treat; NOA: non-obstructive azoospermia; SR: sperm retrieval; SCO: Sertoli cell-only; CI: confidence interval †Comparison of proportions by chi-square: (Campbell, 2007) ¶(Altman, 1998) *Total number of patients with Sertoli cell-only reported in primary studies. mTESE: microdissection testicular sperm extraction; cTESE: conventional testicular sperm extraction; NNT: number needed to treat; NOA: non-obstructive azoospermia; SR: sperm retrieval; SCO: Sertoli cell-only; CI: confidence interval †Comparison of proportions by chi-square: (Campbell, 2007) ¶(Altman, 1998) *Total number of patients with Sertoli cell-only reported in primary studies. Third, the diagnosis of NOA relies on a diagnosis of impaired sperm production that is most reliably made with histopathology data (Esteves, 2015). Men with normal FSH and normal testis volume could either have maturation arrest or obstructive azoospermia and, therefore, might not be distinguished solely based on clinical and laboratory characteristics (Bernie et al., 2015b). In our analysis of Corona et al.’s data, we found that histopathology figures were reported in ~65% of included trials (Table I and Supplementary Table SI). When the analysis was limited only to non-comparative studies providing confirmatory diagnostic histopathology details, micro-TESE again resulted in significantly higher SRR than cTESE (50.3 vs. 47.4%, RR: 1.06, 95% CI 1.02–1.10; P = 0.0026; Table I). These results are consistent with those of comparative studies, including the findings reported by the only randomized controlled trial (RCT) published so far on the matter concerned (Colpi et al., 2009). It is important to note that Colpi et al. reported that results were better for micro-TESE with SRR of 27.5% (n = 40) vs. 12.5% (n = 40) with the use of cTESE in patients with SCO. We, therefore, disagree with the conclusions of the Corona et al. study, as published. The SRR is affected by the type of surgical technique used for SR, provided strict diagnostic criteria are applied to identify the NOA patient. The higher the study’s quality, the higher the magnitude of the effect size. Although our analysis of Corona et al. data—based on the pooled estimates from non-comparative studies—indicate that the numbers needed to treat (NNT) by micro-TESE vs. cTESE to achieve one additional positive SR are 35 and 29 in the overall patient population and SCO patients, respectively (Table I), these results are not valid because of the heterogeneity of treated populations. By contrast, the NNT estimates from comparative studies are 7.5 and 4.4, respectively (Table II). As highlighted above, SR success depends on many different factors, including patient selection criteria, surgeon’s experience, embryologist’s experience and laboratory technique to process retrieved specimens, to cite a few. Therefore, only studies directly comparing SR techniques can be considered in a meta-analysis. Lastly, given the heterogeneity of published studies and lack of large RCTs, we agree with Corona et al. that well-designed comparative trials are needed to confirm the superiority of micro-TESE over cTESE definitely. Given the existent Level 1 evidence in well-performed meta-analyses (Deruyver et al., 2014; Bernie et al., 2015a), there is little question that micro-TESE provides optimized SR results in expert hands. Clinicians should be judicious in interpreting the existing data to deliver the best care to infertile men with NOA seeking biological parenthood. Extended data hosted in https://figshare.com/s/858ea5c25ddc32d2dbf9. Data are available under the terms of the Creative Commons Zero ‘No rights reserved’ data waiver (CC0 1.0 Public domain dedication). None.
Source: Human Reproduction Update
Language: English
Subfield: Reproductive Medicine
Field: Medicine
Domain: Health Sciences
Open Access status: hybrid
Estimated APC paid: $4,490
OpenAlex ID: W3028172529