The treatment of male factor infertility, due to non-obstructive azoospermia (NOA), by testicular sperm extraction (TESE)–ICSI is a major breakthrough in assisted reproduction. However, although the first patient series on this approach were published more than 15 years ago (Tournaye et al., 1995), to date, it still remains difficult to provide candidate patients with truthful data about their chances to eventually father their genetically own child. In this issue, Boitrelle et al. (2011) published a study trying to determine predictive parameters to improve patient counselling. Retrieval rates after testicular surgery reported in the literature differ considerably. While in unselected NOA patients, the surgical approach seems to have a limited effect as long as multiple biopsies are taken whenever needed, the extended processing of the wet preparations of the testicular tissue may have a greater impact on sperm recovery. Unfortunately, large well-designed controlled trials to support both assertions are lacking. But sperm retrieval rates are also subject to the (pre)selection of patients and can be biased either by re-allocating successful patients in case series, either by including patients showing almost normal spermatogenesis, e.g. mild-to-moderate hypospermatogenesis, or by inclusion of patients with no testicular histology available. As a result of the above, retrieval rates reported in the literature for NOA men may vary from about 30% to even more than 80%. Larger case studies in well-defined NOA populations report sperm recovery rates after a first TESE attempt around 50%. However, given the invasive character of TESE, NOA patients want to have a better prediction of their chances to retrieve testicular sperm than tossing a coin. Because testicular volume and serum FSH are routinely assessed in azoospermic men, these parameters are often used in studies on prediction either as a stand-alone parameter or in combination with other parameters. Unfortunately, their predictive power remains limited and is subject to the heterogeneity of the population of NOA patients studied. Idem dito for the predictive parameters published in the study by Boitrelle et al. (2011). The positive likelihood ratios for the stand-alone parameters are below 2 and hence not of a great diagnostic value in predicting testicular sperm recovery. With a positive likelihood ratio of 3, a predictive score combining testicular volume, FSH and inhibin-B looks more promising in their setting. But again, is this a robust predictive model applicable to every population of NOA men? All NOA men with all parameters included in the score available were selected in our Brussels database. Although we refrained from assessing inhibin-B routinely, after concluding that this parameter had a poor predictive value (Vernaeve et al., 2002), the score could be calculated in a population of 110 NOA men that had their first TESE attempt and showed neither normal spermatogenesis nor hypospermatogenesis on their testicular histologies. Compared with the study by Boitrelle et al. (2011), our sperm retrieval rates were, respectively, 50.0% compared with 77.4% in men with a score below 18.5 (n 1⁄4 38) and 57.1% compared with 39.7% in men with a score between 18.5 and 3700 (n 1⁄4 72). Our ‘test’ population did not include men with a score higher than 3700. This ‘test’ emphasizes that any blind application of a prediction model on a population of patients, other than the one it originated from, remains difficult due to patient heterogeneity. Whether out-of-routine assessmentsmay improve the prediction of a successful testicular recovery also remains to be proven. Doppler ultrasoundor even invasive assessments such as ‘testicularmapping’ by puncture provide low sensitivities. Since many NOA men do not accept donor insemination as an alternative for their fatherhood and because current predictive tests cannot exclude that any sperm will eventually be retrieved, to date, reliable counselling towards sperm retrieval remains problematic for all TESE candidates suffering from NOA. Counselling becomes even more difficult when an NOA patient would like to know what his chances are to have a child, which in the end is the true outcome of interest to candidate TESE–ICSI patients. At present, only fragmentary data exist either reporting on retrieval rates after TESE or reporting on the outcome of ICSI once testicular spermatozoa were obtained in different subsets of patients. Boitrelle et al. (2011) show that in their setting, the take-home baby rate in patients with a score over 3700 was low (7.7%) compared with lower-score patients (48.8%), a finding that still may suffer from bias because only a subset of patients was included in the subsequent ICSI analysis (89 out of 149 men in which testicular sperm were retrieved, i.e. 59%, eventually proceeded to ICSI). Again, the score can probably not be applied blindly in other TESE–ICSI programmes
Read full abstract
Nov 20, 2012
Taku Sato 
Open Access
