Shorter interval between consecutive testicular sperm aspirations and clinical pregnancy after intracytoplasmic sperm injection in obstructive azoospermia: a retrospective cohort study.

Azoospermia, defined as the complete absence of spermatozoa in the ejaculate, affects approximately 1% of the general male population and up to 15% of infertile men ( 1 ). This condition can arise from various etiologies, including primary or secondary testicular failure, and is often confirmed through clinical history, physical examination, hormone assessments, and testicular volume evaluation ( 2 - 4 ). For patients with azoospermia, testicular sperm aspiration (TESA) combined with assisted reproductive technology (ART) such as intracytoplasmic sperm injection (ICSI) has become a pivotal approach, enabling sperm retrieval and the potential for successful pregnancies ( 5 ). TESA is a fast, cost-effective, and reliable technique for obtaining viable sperm, requiring minimal microsurgical expertise, no routine surgical exploration, and causing only mild postoperative discomfort ( 6 , 7 ). However, initial TESA attempts may not always yield sufficient viable sperm, particularly in challenging cases, leading to low ICSI success rates ( 8 ). As a result, repeated TESA procedures are frequently necessary to provide additional opportunities for sperm acquisition and subsequent ICSI cycles ( 9 ). The interval between two TESAs may be biologically relevant considering recovery of testicular tissue and the spermatogenic cycle ( 10 - 13 ). However, evidence specific to obstructive azoospermia (OA) and repeated testicular retrievals remains limited, and most prior work focused on retrieval feasibility rather than pregnancy outcomes ( 5 , 14 - 17 ). This study examines whether different intervals are associated with ICSI outcomes within Reproduction Medical Center of West China Second University Hospital, Sichuan University’s practice. Our study retrospectively analyzes a large cohort of 751 obstructive azoospermic patients undergoing consecutive TESA procedures followed by ICSI. We examine whether the interval between these procedures significantly impacts clinical pregnancy outcomes. We present this article in accordance with the STROBE reporting checklist (available at https://tau.amegroups.com/article/view/10.21037/tau-2025-688/rc ).

This was a retrospective cohort study conducted at the Reproductive Endocrinology Department of Reproduction Medical Center of West China Second University Hospital, Sichuan University. The study was approved by the Ethics Committee of West China Second University Hospital, Sichuan University (IRB approval No. 2023374), which also granted a waiver of individual informed consent due to the retrospective use of clinical data and the absence of invasive procedures involving human subjects. Data were collected and analyzed in compliance with institutional guidelines and the Declaration of Helsinki and its subsequent amendments. We included men with OA who underwent an initial outpatient TESA followed by a second, retrieval-day TESA paired with ICSI at Reproduction Medical Center of West China Second University Hospital, Sichuan University between January 2015 and March 2023. OA was established and non-obstructive azoospermia (NOA) excluded using guideline-based, composite criteria integrating history, examination, laboratory testing, and imaging. Specifically, OA diagnosis required all of the following. ❖ Documented history of bilateral epididymitis and/or other obstructive risk factors (e.g., prior vasectomy or inguinal/scrotal surgery, genital tract infection), when present. ❖ Palpation of the vas deferens and epididymis; findings supportive of obstruction included absent vasa or epididymal enlargement/induration with otherwise normal testicular size and consistency. ❖ At least two semen analyses confirming azoospermia after centrifugation of the ejaculate pellet. ❖ Semen volume and pH assessed; patterns supportive of obstruction recorded (e.g., low volume with acidic pH and/or absent fructose for distal/ejaculatory duct obstruction; normal volume/pH for proximal obstruction). Serum follicle-stimulating hormone (FSH), luteinizing hormone (LH), and testosterone within reference ranges. ❖ Scrotal and ductal ultrasonography (and transrectal ultrasound when indicated) demonstrating features consistent with obstruction, including bilateral absence or obstruction of the vas deferens, epididymal ductal dilation, and/or seminal vesicle/ejaculatory duct dilation. ❖ In cases of suspected congenital bilateral absence of the vas deferens (CBAVD), cystic fibrosis transmembrane conductance regulator (CFTR) genotyping and renal ultrasound were performed when available/indicated. The initial outpatient TESA demonstrated preserved spermatogenesis with successful retrieval of testicular sperm. Inclusion criteria include: (I) documented primary or secondary azoospermia; (II) a previous TESA procedure; (III) attendance at the outpatient clinic of our reproductive endocrinology department between 1 January 2015 and 1 March 2023. Men with clinical or laboratory features consistent with non-obstructive azoospermia were excluded. To minimize female-factor bias, we excluded couples in which the female partner was older than 38 years, produced fewer than three oocytes, had an anti-Müllerian hormone (AMH) level <1 ng/mL, or had any other clearly identified condition likely to impair pregnancy outcomes. Additional exclusion criteria for the female partner included: (I) a history of recurrent miscarriage (≥2 consecutive clinical pregnancy losses); (II) diagnosed endometriosis or adenomyosis; and (III) major uterine abnormalities such as submucosal fibroids, uterine septa, or severe intrauterine adhesions (e.g. Asherman syndrome). In our clinical workflow, the initial outpatient TESA is primarily performed as a diagnostic and feasibility procedure to confirm the presence of testicular sperm and to inform subsequent treatment planning. For patients requiring a second, retrieval-day TESA, the predominant indication was an insufficient number of spermatozoa (typically <1×10 6 total sperm count) obtained at the first TESA to support multiple ICSI attempts or elective embryo cryopreservation. In keeping with Reproduction Medical Center of West China Second University Hospital, Sichuan University’s long-standing preference for using fresh rather than cryopreserved testicular sperm, routine sperm banking from the initial TESA is generally not performed. Consequently, when couples proceeded to an ICSI cycle, a repeat TESA on the day of oocyte retrieval was typically required to provide adequate fresh sperm for insemination. TESA was performed according to the method described by Esteves et al. ( 6 ). Under local anesthesia, the spermatic cord and epididymis were anesthetized. An 18-gauge needle connected to a 20-mL syringe was inserted through stretched scrotal skin into the upper pole of the testis. Negative pressure was generated by pulling the syringe plunger while obliquely maneuvering the needle tip within the testis to disrupt seminiferous tubules and sample different regions. After aspirating a small amount of testicular tissue, the needle was gently withdrawn while maintaining negative pressure. Microsurgical forceps were used to grasp exposed seminiferous tubules from the scrotal skin for specimen extraction. The prepared specimen was flushed into a tube containing 0.5–1.0 mL of warmed sperm culture medium and transferred to the laboratory. The side of puncture (same versus contralateral to the prior attempt) was determined by the operating surgeon and patient’s will. Throughout the study period, all TESA specimens were handled according to a standardized protocol. The specimen was carefully released into an outer dish. Under stereomicroscopy, seminiferous tubules were identified, and blood clots were efficiently removed using a tuberculin syringe. The tubules were then transferred to an inner dish with fresh sperm culture medium and mechanically dispersed repeatedly using two tuberculin syringes until no intact tubules remained. Within 10 minutes, the homogenate was examined under an inverted microscope at ×400 magnification to confirm sperm presence. The cell suspension was transferred to a sterile centrifuge tube, diluted with 3 mL of fresh sperm culture medium, and centrifuged at 300× g for 7 minutes. The supernatant was discarded, and the pellet was resuspended in 0.2 mL of sperm culture medium. If excessive red blood cell contamination persisted, the sample was diluted with erythrocyte lysis buffer and recentrifuged. Female partners underwent either a gonadotropin-releasing hormone (GnRH) agonist long protocol or an antagonist protocol. In the long protocol, pituitary down-regulation was initiated with triptorelin acetate (0.1 mg every 48 hours) during the mid-luteal phase of the previous cycle. After confirmation of down-regulation via ultrasound and baseline hormone levels, recombinant FSH was started at 150–225 IU [initial dose based on patient age and body mass index (BMI)]. Stimulation was monitored after 4 days, with dose adjustments based on follicular development (number and size of follicles). In the antagonist protocol, recombinant FSH was initiated on menstrual cycle day 2 following ultrasound and baseline hormone assessment. GnRH antagonist (cetrorelix 0.25 mg) was administered daily from stimulation day 6 until the day of human chorionic gonadotropin (hCG) injection. When ≥2 follicles reached a diameter of 17 mm, hCG (5,000–10,000 IU) was administered, and oocyte retrieval was performed 36 hours later. Fertilization was assessed 16–18 hours post-ICSI. Embryo transfer occurred on day 3. ART clinical outcomes were compared between the non-pregnancy and pregnancy groups. Continuous variables are summarized as mean ± standard deviation (SD) or median [interquartile range (IQR)], categorical variables as counts and percentages. Between-group comparisons used Student’s t -test or Mann-Whitney U test for continuous data and χ 2 test for categorical data, as appropriate. The primary endpoint was clinical pregnancy in the index ICSI cycle paired with the second TESA. Multivariable binary logistic regression was used to estimate adjusted odds ratios (ORs) and 95% confidence intervals (CIs) for the association between TESA interval and clinical pregnancy, adjusting for male age, female age, baseline endocrine profile, ovarian stimulation protocol, number of oocytes retrieved, and embryo quality metrics where available. Propensity score analyses were conducted as sensitivity checks for different interval cutoffs. The score was estimated from potential covariates; weighted comparisons were performed with percentile truncation to limit extreme weights. Covariate balance was evaluated using standardized mean differences. Exploratory mediation analyses treated embryo quality metrics as potential mediators on the pathway from interval to clinical pregnancy. Estimates were obtained with nonparametric bootstrap for uncertainty. These analyses are observational and intended to assess robustness rather than establish causality, results are reported as associations. Exploratory threshold analyses evaluated different interval cutoffs and trends across interval groups. Missing data were handled using multiple imputation by chained equations (MICE). Two-sided P<0.05 was considered statistically significant. Analyses were performed in Python 3.11.

In this retrospective cohort of 751 patients with OA who underwent consecutive TESA procedures followed by ICSI, the overall clinical pregnancy rate was 51.7% (388/751). The mean male age was 30.6±6.2 years, while the mean female age was 28.9±4.1 years. The mean female BMI was 22.4 kg/m 2 (SD 1.7), and female AMH was 4.33±3.07 ng/mL. The median interval between TESA procedures was 4.0 months, with a mean of 5.8±6.3 months. Baseline hormone levels across the cohort included estradiol at 32.1±15.8 pg/mL, testosterone at 3.8±2.1 ng/L, LH at 4.3±5.0 IU/L, and FSH at 8.5±7.8 IU/L ( Table 1 ). Data are presented as mean ± standard deviation or n (%). TESA, testicular sperm aspiration. Patients were categorized into four groups based on the TESA interval: ≤3 months (n=339, 45.1%), 4–6 months (n=204, 27.2%), 7–12 months (n=123, 16.4%), and >12 months (n=74, 9.9%). There was no significant difference in female BMI, female AMH and baseline hormone levels between the clinical pregnancy and non-pregnancy groups (see Figures S1,S2 ). Pregnancy rates differed across these groups, with the highest rate observed in the ≤3 months group at 55.8%, followed by 54.1% in the >12 months group, 50.4% in the 7–12 months group, and the lowest at 44.1% in the 4–6 months group. When comparing the pregnancy group (n=388) and non-pregnancy group (n=363), no significant difference emerged in the mean TESA interval in unadjusted and descriptive analysis (5.65±5.96 vs. 5.98±6.62 months, P=0.47) (see Figure 1A-1C ). By inclusion, the second (retrieval-day) TESA yielded sperm for ICSI in the index cycle. Impact of TESA interval on pregnancy outcomes and embryo quality. (A) TESA interval group distribution. Numbers above bars indicate the count and proportion of index ICSI cycles in each prespecified interval group: ≤3 months, 4–6 months, 7–12 months, and >12 months (interval defined as the time between consecutive TESA procedures prior to the index cycle). (B) Pregnancy rate by interval groups. Bars show the clinical pregnancy rate per index cycle for each TESA-interval group; labels display the rate and sample size (n). Global between-group comparison was evaluated by a Chi-squared test. (C) Interval distribution by pregnancy outcome. Box-and-scatter plot of TESA intervals (months) for cycles with and without clinical pregnancy; boxes depict median and IQR, whiskers 1.5× IQR, dots are individual cycles. Mean values (μ) are annotated for each outcome group. (D) Embryo quality by pregnancy outcome. Mean numbers of usable embryos and good‑quality embryos in pregnant versus non‑pregnant cycles. Asterisks denote statistical significance as defined in “Methods” section (***, P<0.001). Definitions of “usable embryo” and “good‑quality embryo” are provided in “Methods” section. (E) Embryo quality by interval groups. Mean numbers of usable embryos and good-quality embryos across the four TESA-interval groups (definitions of embryo quality follow the laboratory criteria described in “Methods” section). (F) Pregnancy outcome by interval groups. Counts of pregnant and non-pregnant cycles within each interval group; the overall comparison yielded χ 2 =7.157, P=0.07. ICSI, intracytoplasmic sperm injection; IQR, interquartile range; m, months; TESA, testicular sperm aspiration. Embryo quality emerged as a key differentiator across groupings. In the pregnancy group, there were significantly higher numbers of usable embryos (4.96±2.97 vs. 2.70±2.19, P<0.001) and good-quality embryos (3.30±2.66 vs. 1.52±1.56, P<0.001) compared to the non-pregnancy group. Similarly, embryo quality varied by TESA interval, with shorter intervals (≤3 months) associated with more usable and good-quality embryos, suggesting a potential mediating role in pregnancy outcomes. This pattern prompted further investigation through causal inference and mediation analyses to quantify the interval’s impact via embryo quality (see Figure 1D-1F ). Analysis of operation side, available for 568 patients with complete data, showed no significant difference in pregnancy rates between same-side procedures (n=242, 50.8%) and opposite-side procedures (n=326, 54.9%; χ 2 =0.773, P=0.38) (see Figure 2 ). Distribution and impact of procedure side on pregnancy rates. (A) Procedure side distribution. Proportions of index ICSI cycles in which TESA was performed on the same testis as the most recent prior TESA (“Same side”) versus the contralateral testis (“Different side”); 57.4% different side and 42.6% same side. (B) Pregnancy rate by procedure side. Clinical pregnancy rate per index cycle for the two side categories; bar labels indicate the rate and sample size (n). (C) Procedure side effect on pregnancy rate. Counts of pregnant and non-pregnant cycles by side category. The overall difference was not statistically significant χ 2 =0.773, P=0.38. “Procedure side” was categorized relative to the most recent prior TESA of the same patient (same vs. contralateral testis). ICSI, intracytoplasmic sperm injection; TESA, testicular sperm aspiration. Propensity score-adjusted comparisons between short (≤3 months) and longer (>3 months) intervals yielded pregnancy rates of 56.0% vs. 47.9%, an absolute adjusted difference of 8.1 percentage points (95% CI: 1.0–15.2%; P=0.03), robust across 1,000 bootstrap resamples. These findings should be interpreted as associations rather than causal effects. Exploratory mediation analyses further elucidated that approximately 60% of the interval’s effect was mediated through embryo quality. For usable embryos, the path from interval to mediator showed an increase of +0.496 embryos (P=0.02), with the mediator-to-outcome path yielding an OR of 1.433; the indirect effect was 0.178 (95% CI: 0.034–0.360), accounting for a 60.4% mediation proportion (Sobel z =1.983, P=0.047). For good-quality embryos, similar patterns emerged with a path a of +0.393 (P=0.02), path b OR of 1.548, indirect effect of 0.172 (95% CI: 0.016–0.334), and 58.2% mediation (Sobel z =2.012, P=0.044) (see Figure 3 ). Causal inference and mediation analysis of TESA interval on pregnancy outcomes. The causal estimand is the absolute risk difference in clinical pregnancy between different intervals; positive values favor the short interval. Propensity scores were estimated from baseline covariates to adjust for confounding. (A) Bootstrap CI. Sampling distribution of the estimated effect size obtained by non-parametric bootstrap. The red line marks the point estimate (0.082), and the orange dashed lines indicate the 95% CI: 0.009–0.153. (B) Propensity score distribution. Overlap of propensity scores for the short- and long-interval groups, demonstrating common support for weighting/matching. (C) Optimal time-window analysis. Clinical pregnancy rates when progressively redefining the “short” interval as ≤1, ≤2, ≤3, ≤6, ≤9, and ≤12 months. Labels above bars show the rate for each window. (D) Threshold sensitivity analysis. Estimated causal effect sizes for alternative short-interval thresholds (e.g., ≤2 vs. >2 months, ≤3 vs. >3 months, etc.). Bars annotated “ns” were not statistically significant at α =0.05. CI, confidence interval; TESA, testicular sperm aspiration. Exploratory threshold analyses supported a favorable interval at ≤3 months, where the pregnancy rate was 56.0%—comparable to ≤1 month (57.1%) and ≤2 months (56.0%), but higher than ≤6 months (51.6%). Sensitivity testing across various thresholds (2–6 months) confirmed consistent effects (see Figure 4 ). Time-threshold and dose-response analyses of TESA interval on pregnancy rates. (A) Pregnancy rate by time threshold. For each month-level threshold t on the x-axis, the curve shows the cumulative clinical pregnancy rate among cycles with a TESA interval ≤t months. The early decline from 1–5 months indicates that adding cycles with longer intervals lowers the observed rate, after which the curve plateaus near ~0.51. This exploratory scan motivated the short-interval definition used in the main analyses. (B) Dose-response relationship. Bin‑specific clinical pregnancy rates across increasingly longer TESA-interval groups {(0,1], (1,2], (2,3], (4,5], (5,6], (6,9], (9,12], 12+ months}. Labels above markers indicate the sample size for each bin (n). The connecting line is for visual guidance only; formal trend was evaluated as described in “Methods” section (Cochran-Armitage test/logistic regression). TESA, testicular sperm aspiration. Multivariate logistic regression on the 568 complete cases identified the TESA interval as a significant predictor, with each additional month associated with a 2.7% decline in pregnancy probability (OR =0.973; 95% CI: 0.949–0.997; P=0.03).

The principal finding of this retrospective study is that a repeated TESA procedure within short interval is not associated with adverse ICSI outcomes in men with OA. While a specific contrast analysis suggested a marginal increase in clinical pregnancy rates with shorter intervals, the collective evidence across descriptive, four-group, and sensitivity analyses remained inconsistent and the observed differences were of small magnitude. A key strength of this study is its large sample size of the OA cohort, which provides sufficient statistical power for subgroup and sensitivity analyses. The application of propensity score matching and exploratory mediation analysis represents a rigorous attempt to mitigate confounding and elucidate potential causal pathways within the constraints of a retrospective design. Nevertheless, several limitations must be acknowledged. First, the single-center design inherently limits external validity, as the findings may be influenced by region-specific patient demographics and institutional protocols. Second, the retrospective nature, despite statistical adjustments, is susceptible to residual confounding from unmeasured variables, such as lifestyle factors or subtle variations in laboratory techniques over time. Third, while we implemented strict exclusion criteria to minimize female-factor confounders, our assessment of oocyte quality was limited to conventional morphological grading recorded in routine clinical charts. More granular embryological variables—such as detailed fertilization kinetics, blastulation dynamics, and time-lapse imaging parameters—that more directly reflect oocyte and embryo competence were not prospectively collected in this retrospective clinical cohort. Fourth, although we restricted inclusion to cycles using standardized agonist or antagonist stimulation protocols, subtle variations in gonadotropin dosing, duration, or trigger timing inevitably occurred over the study period. These nuances were not analyzed in detail and may have exerted a modest, unmeasured influence on oocyte yield and embryo development, although we expect any such impact to be small within our tightly selected population. Finally, our primary endpoint was clinical pregnancy. While clinically relevant, this metric does not capture downstream outcomes such as live birth rates or neonatal health, and may therefore overestimate the true clinical benefit. Furthermore, the mediation analysis, while providing intriguing hypothesis-generating data, relies on statistical assumptions that cannot definitively establish causality. Advanced quasi-experimental methods were not feasible with the available data. Consequently, only a prospective, randomized study design can provide conclusive evidence to confirm these exploratory findings. Direct studies on how the spacing between repeat TESA procedures affects ICSI OA are scarce; however, one report indicates that repeating TESA in the same testis does not reduce recovery of motile sperm and that elapsed time per se does not appear to impair outcomes ( 18 ). Evidence from repeat testicular retrievals—largely in non-obstructive or mixed cohorts—shows that subsequent procedures remain feasible, with higher success when sperm are retrieved previously, suggesting that prior testicular intervention does not inherently diminish future retrieval potential ( 19 ). In OA-focused practice, clinical decision reviews note that once viable sperm are available, ICSI outcomes are generally robust across commonly used sperm sources, and repeated percutaneous or open retrievals have been reported as safe without clear deterioration across cycles ( 20 ). Against this backdrop, our observation that a short interval between a diagnostic/outpatient TESA and treatment-cycle TESA did not confer a measurable disadvantage is consistent with prior data showing stable motile sperm recovery and acceptable outcomes with repeated TESA over short intervals ( 18 ). Several biological mechanisms may underlie the observed trends. Shorter intervals between procedures might contribute to the preservation of sperm quality and yield by limiting the duration of testicular inactivity, which could otherwise promote a functional decline in spermatogenesis ( 21 ). An interval of approximately 3 months may represent an optimal balance, allowing for sufficient tissue recovery and resolution of acute inflammation from the initial procedure while avoiding potential testicular atrophy or inefficiency associated with prolonged disuse ( 22 ). Moreover, minimizing the interval could indirectly mitigate the impact of age-related declines in the female partner’s ovarian reserve or oocyte quality, although this hypothesis remains speculative as these dynamics were not directly measured in our study ( 23 ). Our exploratory mediation analysis provides preliminary quantitative support for these biological hypotheses. The model suggests that approximately 60% of any observed effect of the TESA interval on pregnancy likelihood is mediated indirectly through the pathway of embryo quality. The statistical significance of the direct effect (interval → pregnancy) and the indirect effect (interval → embryo quality → pregnancy), supported by bootstrap CIs that exclude zero, lends credence to this framework. However, given the retrospective data, these findings should be interpreted as exploratory and hypothesis-generating rather than confirmatory. From a clinical perspective, our findings suggest that when a repeat TESA is necessary for men with OA, a short interval is a reasonable and safe option that is not associated with inferior outcomes. This provides valuable flexibility for cycle planning. It is important to contextualize this finding within Reproduction Medical Center of West China Second University Hospital, Sichuan University’s specific clinical workflow, which historically favors percutaneous TESA as an initial, less invasive approach, reserving conventional or microTESE for cases of failed retrieval. This differs from programs that may opt for microTESE as a primary intervention. Additionally, our practice of performing a fresh TESA on the day of oocyte retrieval, largely due to scheduling logistics and a preference for fresh rather than cryopreserved testicular sperm, is not universal; many centers favor elective cryopreservation from an initial diagnostic TESA. In our cohort, repeat TESA was most commonly undertaken because the first outpatient procedure yielded an insufficient sperm count to support multiple ICSI attempts or elective embryo banking, and routine cryostorage of TESA specimens was not implemented. Therefore, the generalizability of our findings to settings with different procedural preferences and cryopreservation strategies should be considered with caution. To resolve the limitations of this study and build upon its findings, future research should prioritize multicenter, prospective randomized controlled trials (RCTs). Such trials are essential to rigorously test the causal relationship between TESA interval, embryo quality, and live birth rates, while controlling for both measured and unmeasured confounders. Incorporating molecular biomarkers of sperm DNA integrity and embryonic developmental potential within these trials would offer deeper mechanistic insights and pave the way for more refined, evidence-based clinical guidelines in male fertility treatment.

In men with obstructive azoospermia undergoing ICSI, repeating TESA within 3 months was not associated with inferior clinical pregnancy rates. Taken together with the absence of consistent between-group differences, these findings support the feasibility of proceeding to the ICSI cycle shortly after an outpatient TESA when clinically appropriate. Confirmation in prospective, ideally randomized, studies is needed.

The article’s supplementary files as