Intracytoplasmic sperm injection (ICSI) was an extraordinary achievement in the field of assisted reproduction technology (ART). Introduced in 1992 as a modification of conventional in vitro fertilization (IVF), ICSI enables men with low sperm quantity and quality to father a child ^{1, 2} . Nowadays, ICSI has become not only the most commonly used method of fertilization in ART but also the method of choice for overcoming untreatable severe male factor infertility ³ .

ICSI is typically carried out with ejaculated sperm, which are generally regarded as having the highest fertilization potential since they have completed their transit through the male reproductive tract. By contrast, sperm retrieval methods—developed a few years after the introduction of ICSI—have been used to harvest sperm from the epididymides and testes of men with azoospermia-related infertility ^{4, 5} . After retrieval of epididymal or testicular sperm, ICSI is mandatory as the retrieved gametes are unable to fertilize the oocytes by conventional IVF.

However, as experience accumulated, reports of an association between semen quality and ICSI outcomes increased steadily ^{6– 8} . Concerns of a possible role of the paternal gamete on ICSI outcomes led Greco et al., in 2005, to investigate the utility of sperm retrieval in a group of 18 non-azoospermic patients with elevated sperm DNA fragmentation (SDF) on neat semen ⁹ . On the day of oocyte retrieval, the male partners underwent sperm retrieval using percutaneous or open methods to harvest sperm from the seminiferous tubules. In this series, ICSI with testicular sperm (Testi-ICSI) resulted in eight clinical pregnancies (44.5%) whereas only one pregnancy (5.6%) that ended in miscarriage had been obtained in previous ICSI cycles with the use of ejaculated sperm.

Given this information, the utility of sperm retrieval in indications other than azoospermia has been investigated. Here, the current support for these indications, including elevated SDF, severe oligozoospermia, and cryptozoospermia—denoted by very few spermatozoa (or none) in the fresh ejaculate but observed after microscopic examination of centrifuged pellet—will be summarized.

It is well established that sperm chromatin integrity is vital for the birth of healthy infants ¹⁰ . Fertilization of oocytes by sperm with DNA fragmentation might increase the risk of fertilization failure, embryo development arrest, implantation failure, miscarriage, congenital malformations, and perinatal and postnatal morbidity ^{11– 13} . Notably, infertile men often have elevated SDF rates in neat semen ^{14, 15} . Varicocele, systemic diseases, male accessory gland infections, advanced paternal age, obesity, lifestyle and environmental factors, radiation, and heat exposure are some of the conditions associated with SDF ^{16, 17} . These stressors have in common the trait of oxidative stress, which represents a significant cause of SDF ¹⁸ . The mechanisms involve reactive oxygen species (ROS) attack on sperm membranes and nuclear and mitochondrial DNA, mostly during sperm transit through the male reproductive tract ^{19– 21} .

Interestingly, data from human studies assessing paired testicular and ejaculated specimens of non-azoospermic men indicate that SDF is two to three times lower in testicular sperm than in ejaculated sperm ^{9, 22– 25} . A 2017 systematic review—followed by a meta-analysis—compiled the results of five studies including 143 patients and showed that the mean difference (MD) in SDF rates was −24.6% (95% confidence interval [CI] −32.5 to −16.6%, I ² = 92%, P <0.001) in favor of testicular sperm ²⁶ . In that report, SDF was measured by using the terminal deoxyribonucleotide transferase–mediated dUTP nick-end labeling (TUNEL) assay (four studies, pooled MD: −19.8%, 95% CI −22.3 to −17.2%, I ² = 15%, P <0.001) or the sperm chromatin dispersion (SCD) assay (one study, MD: −32.4%, 95% CI −34.85 to −29.95%, P <0.001).

After the report by Greco et al. ⁹ , several authors investigated the utility of sperm retrieval in non-azoospermic men with elevated SDF in neat semen ( Table 1) ^{19, 24, 25, 27– 32} . In a 2017 systematic review, we aggregated the evidence of five studies including 507 ICSI cycles ²⁶ . In total, 3,840 oocytes were injected with either ejaculated sperm or testicular sperm. Using meta-analysis, we showed higher clinical pregnancy rates (odds ratio [OR] 2.42, 95% CI 1.57 to 3.73, I ² = 34%, P <0.0001) and live birth rates (OR 2.58, 95% CI 1.54 to 4.35, I ² = 0%, P = 0.0003), and lower miscarriage rates (OR 0.28, 95% CI 0.11 to 0.68, I ² = 11%, P = 0.005) when comparing Testi-ICSI with ejaculated ICSI.

Recent studies providing live birth data corroborate the effectiveness of testicular sperm for ICSI in men with high SDF ^{29– 31} . Thus, despite the limited evidence and lack of randomized controlled trials, data from seven retrospective studies and three prospective studies, including a total of 830 patients and 902 ICSI cycles, suggest that Testi-ICSI is superior to ICSI with ejaculated sperm to overcome infertility among non-azoospermic men with elevated SDF in semen. Testi-ICSI has been postulated to bypass post-testicular sperm chromatin damage caused by oxidative stress during sperm transit through the epididymis ³³ . As a result, the chances of oocyte fertilization by genomically intact spermatozoa and formation of a normal embryonic genome are increased, thus positively impacting the likelihood of achieving a live birth. Notably, a single study ³² including 110 couples with sperm DNA damage data failed to corroborate the latter findings; however, in that study, SDF thresholds of 15% (by sperm chromatin structure assay, or SCSA) were used to select couples eligible for Testi-ICSI; those thresholds are not fully consistent with the 30% SCSA thresholds reported to be associated with adverse pregnancy outcomes in ART ³⁴ . Thus, the inclusion of ~30% of men with SDF values between 15% and 30% in the above study might have diluted the positive effect of Testi-ICSI.

Weissman et al., in 2008, reported the first series of Testi-ICSI in patients with severe oligozoospermia (<5 million sperm/mL) ³⁵ . The authors performed testicular sperm injections in four couples with a history of multiple failed IVF/ICSI cycles after the use of poor-quality ejaculated sperm. The male partners had sperm counts ranging from 0.2 million/mL to 2.0 million/mL. On the day of oocyte retrieval, sperm retrieval was performed, and in all cases, motile spermatozoa were retrieved from the testis. All couples achieved embryo implantation and delivery of healthy offspring after embryo transfers.

Given the success reported by Weissman et al. ³⁵ , many authors sought to investigate the utility of sperm retrieval for ICSI in non-azoospermic patients with severe oligozoospermia or cryptozoospermia ( Table 2) ^{36– 39} . These studies report an overall better pregnancy outcome with the use of testicular than ejaculated sperm. But surprisingly, in 2016, a systematic review and meta-analysis aggregating the data of the above studies concluded that sperm retrieval should not be recommended in men with severe oligozoospermia or cryptozoospermia ⁴⁰ . In that report, the relative risk (RR) of achieving pregnancy (272 cycles, RR = 0.53, 95% CI 0.19 to 1.42) with the use of testicular or ejaculated sperm for ICSI was not different. However, we performed a careful examination of the authors’ data and discovered that they inadvertently inverted the number of pregnancies reported in the study by Bendikson et al. ³⁶ concerning the group of patients undergoing ICSI with testicular and ejaculated sperm. This critical mistake inflated the total number of pregnancies in the ejaculate sperm group, thus leading to an erroneous RR calculation. We reassessed the pregnancy results of the meta-analysis by Abhyankar et al. ⁴⁰ —after correcting the incongruency mentioned above—and found a significantly higher pregnancy rate with the use of testicular sperm than with ejaculated sperm in men with cryptozoospermia and severe oligozoospermia (272 cycles, RR = 3.21, 95% CI 1.70 to 6.05, I ² = 42%, P = 0.0003) ( Figure 1; unpublished data).

Recently, additional reports and systematic reviews on the matter concerned were published ^{41– 45} . In a 2018 systematic review and meta-analysis, Kang et al. pooled the data of six studies including a total of 578 patients and 761 ICSI cycles ⁴³ . The authors showed that sperm retrieval and Testi-ICSI improved the likelihood of achieving good-quality embryos (RR = 1.17, 95% CI 1.05 to 1.30, P = 0.005), implantation (RR = 1.52, 95% CI 1.02 to 2.26, P = 0.04), and pregnancy (RR = 1.74, 95% CI 1.20 to 2.52, P = 0.004). These results were corroborated by Ku et al., who pooled the evidence of studies that provided miscarriage and live birth data ⁴⁴ . The authors included a total of 331 patients and 479 ICSI cycles. In that report, miscarriage rates were not affected by the use of testicular or ejaculated sperm for ICSI (RR = 1.06, 95% CI 0.48 to 2.35), but live birth rates per initiated cycle were increased among couples that had undergone Testi-ICSI (RR = 1.77, 95% CI 1.28 to 2.44, P = 0.0005).

Collectively, evidence from seven retrospective studies and one prospective study, including a total of 613 patients and 799 ICSI cycles, suggests that Testi-ICSI is superior to ICSI with ejaculated sperm to overcome infertility among non-azoospermic men with severe oligozoospermia or cryptozoospermia ( Table 2). Likewise, Testi-ICSI has been postulated to bypass post-testicular sperm damage during sperm transit through the genital tract. However, no randomized controlled study has been published yet to support the routine use of sperm retrieval and testicular sperm for ICSI to non-azoospermic men with low sperm count undergoing ICSI.

The relatively low testicular sperm positivity for DNA damage might explain the better reproductive outcomes with the use of testicular sperm rather than ejaculated sperm for ICSI. Nevertheless, it is important to acknowledge that the evidence concerning the superiority of Testi-ICSI relies overwhelmingly on cohort studies with few patients, in which confounding factors, such as maternal and paternal age, etiology of male factor infertility, use of medication with possible gonadotoxic effect, and lifestyle factors, to cite a few, were not properly controlled. For instance, it has been suggested that the adverse effect of sperm DNA damage on reproductive outcomes is modulated by female age because of the intrinsic (albeit limited) capacity of oocytes from young women to repair the DNA damage ^{46– 48} . On the other hand, women of advanced reproductive age have significantly fewer euploid embryos available for transfer, which will reduce ART success irrespective of the type of sperm used ⁴⁹ . Since not all sperm DNA damage is repairable, it seems sound to suggest that surgically retrieved sperm should not be used as a last resort after years of treatment with ejaculated sperm because the oocyte apparatus to repair sperm DNA damage is less efficient as both ovarian reserve and maternal age increase ⁴⁸ . These observations highlight the importance of controlling for confounders in future studies evaluating the clinical utility of testicular sperm in non-azoospermic men.

Both percutaneous and open sperm retrieval procedures can be used to harvest sperm from the seminiferous tubules in non-azoospermic men ( Figure 2) ^{50– 52} . The testicle rather than the epididymis is the target organ because of the reported lower SDF rates in the former ^{53– 55} . In such patients, the reported sperm retrieval success rates are close to 100% with the use of testicular sperm aspiration (TESA), testicular sperm extraction (TESE), or microdissection TESE (micro-TESE) ( Table 1 and Table 2). Our choices are TESA for men with elevated SDF and TESE or micro-TESE for cryptozoospermic patients ^{24, 56} . In our hands, these methods are carried out on an outpatient basis on the same day of oocyte retrieval ^{50– 52, 56– 58} . The reason relates to the fact that prolonged sperm incubation—in particular, at 37°C—and sperm freezing might negatively affect sperm chromatin integrity ^{21, 59, 60} .

In the context of non-azoospermic men, sperm retrieval is associated with few complications (less than 5%) as minimal tissue extraction yields sufficient numbers of sperm for ICSI ^{26, 33, 51, 52} . Nevertheless, given the potential risk for complications and adverse effects on testicular function, sperm retrieval should be performed by well-trained urologists.

The use of sperm retrieval in non-azoospermic men has raised concerns about the health of resulting offspring because of the reports of increased sperm aneuploidy rates in testicular sperm (versus ejaculated sperm) ^{23, 61– 64} . On the one hand, ICSI has been associated with possible increased risks of congenital malformations, epigenetic disorders, chromosomal abnormalities, infertility, cancer, delayed psychological and neurological development, and impaired cardiometabolic profile compared with naturally conceived children and this is probably due to the influence of parental subfertility ³ . On the other hand, data concerning risks and sequelae to offspring health with the use of surgically retrieved gametes from azoospermic men are overall reassuring albeit limited ^{3, 65– 70} . However, no study has yet examined whether ICSI with testicular instead of ejaculated sperm (when both are available) affects the risk of malformations and long-term health of offspring.

Nevertheless, new data generated by whole-exome sequencing molecular karyotype suggest that sperm aneuploidy in testicular specimens is not a major concern ⁷¹ . In this series, paired assessments in ejaculated and surgically retrieved testicular samples of non-azoospermic patients with elevated SDF in semen showed that the rates of aneuploidy (1.3% versus 8.4%, respectively, P = 0.02) were lower in testicular sperm than in ejaculated sperm. Along these lines, Weng et al. showed that the origin of sperm used for ICSI had no marked influence on embryo aneuploidy rates ⁷² . Moreover, a 2019 ICSI study from our group—using 24-chromosome genetic testing—revealed that euploid blastocyst rate per metaphase II oocyte was not differently affected whether ejaculated or testicular sperm retrieved from men with elevated SDF was used for ICSI (18.7% versus 18.2%, respectively) ⁷³ . These observations corroborate the safe utilization of sperm retrieval in non-azoospermic men, but owing to limited data concerning the health of resulting offspring, continuous monitoring is warranted.

A growing body of evidence supports sperm retrieval for ICSI in non-azoospermic men with elevated SDF, severe oligozoospermia, and cryptozoospermia. In these scenarios, Testi-ICSI instead of ICSI with ejaculated sperm seems to be associated with improvements in pregnancy outcomes. Percutaneous aspiration and open TESE (with and without the aid of microsurgery) are the methods that have been applied, with high success rates and few complications, to harvest sperm from the seminiferous tubules of non-azoospermic men. However, it is essential to acknowledge the limitations of existing evidence. First, most of the data summarized derive from small observational studies in which confounder factors were not properly controlled. Thus, level 1 evidence in support of Testi-ICSI is still lacking. Second, sperm retrieval is an invasive procedure with potential complications. Thus, identification and treatment of the male factor associated with high SDF, oligozoospermia, and cryptozoospermia are essential to potentially avoid the use of surgical retrieval. We recommend a holistic approach to improve paternal health—whenever there is an opportunity—to patients embarking on any type of ART treatment. Lastly, there are limited data concerning the health of resulting offspring with the use of sperm retrieval and ICSI in cases where both ejaculated and testicular sperm are available. Keeping these limitations in mind, by summarizing the current literature, this article might guide health-care providers in presenting available evidence to patients to help them make informed decisions.