Intracytoplasmic sperm injection hampers fertilization rate and pregnancy per initiated cycle in patients with extremely poor ovarian response

The use of ICSI in patients with extremely poor responders is usually driven by fear of total fertilization failure. However, our data suggest that using ICSI in these patients further hampers fertilization.

In this study, a total of 2572 patients with fewer than three oocytes were analyzed, including 2159 patients receiving IVF and 413 patients receiving ICSI. After PS matching, a total of 401 patients for each group were included in the analysis. The basal characteristics before and after PS matching are shown in Table  1 . The patients with ICSI may have high OPU order and male age, lower sperm normal morphology rates, and TMC in unmatched cohort (absolute value of D  > 0.1). After matching, the two groups were similar in baseline characteristics (absolute value of D  < 0.1). Distributions of the PSs before and after PS matching are shown in Supplementary (Supplementary Figure S1 ). Table 1 Basic characteristics before and after PS matching for OPU cycles Variables Before matching D * After matching D * IVF ICSI IVF ICSI ( N  = 2159) ( N  = 413) ( N  = 401) ( N  = 401) Female age, year 35.0 [31.0, 39.0] 36.0 [32.0, 39.0] 0.115 36.0 [32.0, 40.0] 36.0 [32.0, 39.0] − 0.002 History of spontaneous miscarriage (%) 0 1790 (82.9%) 344 (83.3%) 0.004 333 (83.0%) 334 (83.3%) 0.003 1 303 (14.0%) 58 (14.0%) 0.000 60 (15.0%) 56 (14.0%) − 0.010 2 45 (2.1%) 6 (1.5%) − 0.006 3 (0.7%) 6 (1.5%) 0.008 ≧3 21 (1.0%) 5 (1.2%) 0.002 5 (1.2%) 5 (1.2%) 0 Parity ≧1 (%) 686 (31.8%) 123 (29.8%) − 0.020 121 (30.2%) 121 (30.2%) 0 OPU order (%) 1 1347 (62.4%) 95 (23.0%) − 0.394 88 (21.9%) 95 (23.7%) 0.018 2 521 (24.1%) 176 (42.6%) 0.185 188 (46.9%) 173 (43.1%) − 0.037 ≧3 291 (13.5%) 142 (34.4%) 0.209 125 (31.2%) 133 (33.2%) 0.020 Endometriosis (%) 282 (13.1%) 47 (11.4%) − 0.017 51 (12.7%) 47 (11.7%) − 0.010 PCOs (%) 45 (2.1%) 13 (3.1%) 0.011 12 (3.0%) 11 (2.7%) − 0.003 Duration of infertility, year 3.90 [2.00,6.00] 5.00 [2.70,7.50] 0.234 4.40 [2.00, 8.00] 4.80 [2.60, 7.50] − 0.042 Female BMI, kg/cm2 21.8 [20.1,23.2] 22.0 [20.0,23.1] 0.010 21.9 [20.3, 23.2] 21.9 [20.0, 23.1] − 0.027 Basal FSH, IU/l 8.90 [0.40, 1210] 9.32 [1.88, 698] 0.039 9.17 [2.31, 34.9] 9.32 [1.88, 35.0] − 0.002 Basal LH, IU/l 3.85 [0.01, 30.7] 3.78 [0.880, 62.6] − 0.007 3.80 [0.12, 30.4] 3.75 [0.88, 62.6] − 0.003 Antral follicle count 5.00 [3.00,7.00] 5.00 [3.00,7.00] − 0.021 5.00 [3.00, 7.00] 5.00 [3.00, 7.00] − 0.020 Male age, year 36.0 [32.0,40.0] 37.0 [33.0,41.0] 0.134 37.0 [33.0, 41.0] 37.0 [33.0, 41.0] − 0.015 Male BMI, kg/cm2 23.9 [21.8,25.9] 23.9 [21.9,26.0] − 0.000 23.9 [21.6, 26.0] 24.0 [21.9, 26.0] 0.021 Sperm normal morphology rate (%) 6.00 [4.00,9.00] 4.50 [3.00,8.00] − 0.132 5.00 [3.50, 8.00] 4.80 [3.00, 8.00] 0.020 Total motile sperm count(× 10 6 ) 55.9 [28.3,100] 27.6 [8.34,63.0] − 0.572 37.6 [20.0, 67.0] 29.2 [8.39, 63.2] − 0.074 Ovarian stimulation protocols (%) Luteal phase 87 (4.0%) 35 (8.5%) 0.044 36 (9.0%) 31 (7.7%) − 0.013 Mild stimulation 128 (5.9%) 21 (5.1%) − 0.008 20 (5.0%) 21 (5.2%) 0.003 GnRH antagonist 1029 (47.7%) 179 (43.3%) − 0.043 183 (45.6%) 177 (44.1%) − 0.015 GnRH agonist 725 (33.6%) 138 (33.4%) − 0.002 135 (33.7%) 133 (33.2%) − 0.005 Other protocols 12 (0.6%) 3 (0.7%) 0.002 2 (0.5%) 3 (0.7%) 0.002 Natural cycle 178 (8.2%) 37 (9.0%) 0.007 25 (6.2%) 36 (9.0%) 0.027 GN starting dose, IU 225 [0, 300] 225 [0, 300] − 0.076 225 [0, 300] 225 [0, 300] − 0.088 Oocyte yield 0.045 − 0.018 1 980 (45.4%) 169 (40.9%) 161 (40.1%) 168 (41.9%) 2 1179 (54.6%) 244 (59.1%) 240 (59.9%) 233 (58.1%) Extended culture (%) 118 (5.5%) 32 (7.7%) 0.023 26 (6.5%) 31 (7.7%) 0.013 Data were presented as median [first quartile, third quartile] and mean (SD) for continuous variables and n (percentage) for categorical variables OPU oocyte pickup; PCOs polycystic ovarian syndrome; BMI body mass index; FSH follicle-stimulating hormone, LH luteinizing hormone, GnRH gonadotropin, GN gonadotropin * D : standardized difference. The absolute value of D is less than 0.1, cohorts can be considered to be balanced with respect to the demographics being assessed Basic characteristics before and after PS matching for OPU cycles Data were presented as median [first quartile, third quartile] and mean (SD) for continuous variables and n (percentage) for categorical variables OPU oocyte pickup; PCOs polycystic ovarian syndrome; BMI body mass index; FSH follicle-stimulating hormone, LH luteinizing hormone, GnRH gonadotropin, GN gonadotropin * D : standardized difference. The absolute value of D is less than 0.1, cohorts can be considered to be balanced with respect to the demographics being assessed Table 2 shows the outcomes of the IVF group and ICSI group before and after PS matching. After matching, the TFF rate in the IVF group was significantly lower than that in the ICSI group ( P  < 0.05). On the other hand, the fertilization rate and MPN rate in the IVF group were significantly higher than those in the ICSI group ( P  < 0.05) in both the matched and unmatched cohorts (Table S1 ). There was no statistical difference in embryo availability between the IVF and ICSI groups in either matched or unmatched cohorts, as cleavage 2PN embryo rate, available embryo number, and high-quality embryo rate were comparable between cohorts. Concerning the cumulative outcomes taking freeze-all cycles into account, the cumulative pregnancy rate in the IVF group was significantly higher than in the ICSI group ( P   0.05) in the matched cohort. Table 2 Clinical outcomes before and after PS matching for OPU cycles Variables Before matching P value After matching P value IVF ICSI IVF ICSI ( N  = 2159) ( N  = 413) ( N  = 401) ( N  = 401) Canceled ET cycles (%) 744 (34.5%) 165 (40.0%) 0.037 142 (35.4%) 157 (39.2%) 0.307 Cycles got no embryo (%) 577 (26.7%) 117 (28.3%) 0.540 101 (25.2%) 114 (28.4%) 0.339 Freeze-all cycle (%) 167 (7.7%) 48 (11.6%) 0.012 41 (10.2%) 43 (10.7%) 0.908 TFF cycle (%) 187 (8.7%) 58 (14.0%)  < 0.001 31 (7.7%) 56 (14.0%) 0.006 Normal fertilization rate, (%)  Mean (SD) 64.2 (41.6) 67.7 (40.8) 0.114 65.7 (40.5) 67.6 (40.9) 0.426  Median [Min, Max] 100 [0, 100] 100 [0, 100] 100 [0, 100] 100 [0, 100] Cleavage 2PN embryo rate, (%)  Mean (SD) 73.9 (43.6) 77.1 (41.8) 0.158 76.2 (42.3) 76.9 (42.0) 0.769  Median [Min, Max] 100 [0, 100] 100 [0, 100] 100 [0, 100] 100 [0, 100] Available embryo number  Mean (SD) 0.722 (0.737) 0.826 (0.766) 0.012 0.800 (0.768) 0.813 (0.760) 0.787  Median [Min, Max] 1.00 [0, 2.00] 1.00 [0, 2.00] 1.00 [0, 2.00] 1.00 [0, 2.00] High-quality embryo rate, %  Mean (SD) 37.2 (46.4) 38.4 (45.9) 0.537 40.4 (46.9) 38.4 (46.1) 0.575  Median [Min, Max] 0 [0, 100] 0 [0, 100] 0 [0, 100] 0 [0, 100]  Clinical pregnancy per fresh ET (%) 490 (34.6%) 81 (32.7%) 0.596 99 (38.2%) 79 (32.4%) 0.202  Live birth per fresh ET (%) 389 (27.5%) 65 (26.2%) 0.733 74 (28.6%) 63 (25.8%) 0.553  Cumulative pregnancy (%) 563 (26.1%) 90 (21.8%) 0.077 115 (28.7%) 87 (21.7%) 0.028  Cumulative live birth (%) 444 (20.6%) 72 (17.4%) 0.165 85 (21.2%) 69 (17.2%) 0.179 Data were presented as mean (SD) and median [Min, Max] for continuous variables and n (percentage) for categorical variables TFF total fertilization failure, PN pronuclei Clinical outcomes before and after PS matching for OPU cycles Data were presented as mean (SD) and median [Min, Max] for continuous variables and n (percentage) for categorical variables TFF total fertilization failure, PN pronuclei Given ICSI’s invasive nature, we examined whether its higher TFF rate resulted from increased oocyte degeneration (Supplementary Table S1 ). The number of oocytes degenerated was recorded after denudation (D0) and during pronuclei checking (D1) for both ICSI and IVF. However, neither the number of oocytes degenerated nor the proportion of cycles with all oocytes degenerated significantly differed between the two groups. Table 3 shows the results of multivariate analyses for TFF, cumulative pregnancy, and CLBR with adjustment of the aforementioned covariates. The details of the models are also shown in Supplementary Table S 2 – 4 . It is shown that ICSI insemination was significantly associated with about 60% higher odds of TFF, before matching (OR:1.62, 95% CI: 1. 12–2. 33) and after matching (OR: 1. 65; 95% CI: 1.12–2.43). The odds of clinical pregnancy per complete cycle were also 35% lower in the ICSI group compared to the IVF group (adjusted OR: 0.65; 95% CI: 0.46–0.91). However, the association between insemination protocols and CLBR remained insignificant. Table 3 Association between insemination protocols and outcomes in patients with one or two oocytes Characteristic Before matching P value After matching P value OR 1 95% CI 1 OR 1 95% CI 1 Total fertilization failure  IVF Ref Ref Ref Ref  ICSI 1.62 1.12, 2.33 0.009 1.65 1.12, 2.43 0.011 Cumulative pregnancy  IVF Ref Ref Ref Ref  ICSI 0.77 0.58, 1.02 0.071 0.65 0.46, 0.91 0.013 Cumulative live birth  IVF Ref Ref Ref Ref  ICSI 0.82 0.60, 1.11 0.200 0.76 0.55, 1.04 0.094 Models were adjusted for female age, female BMI, female basal FSH, LH, AFC, previous maternal history, PCOS, endometriosis, OPU order, male BMI, male age, sperm normal morphology rate, total motile sperm number, stimulation protocol, oocyte yield, and extended culture 1   OR odds ratio, CI confidence interval Association between insemination protocols and outcomes in patients with one or two oocytes Models were adjusted for female age, female BMI, female basal FSH, LH, AFC, previous maternal history, PCOS, endometriosis, OPU order, male BMI, male age, sperm normal morphology rate, total motile sperm number, stimulation protocol, oocyte yield, and extended culture 1   OR odds ratio, CI confidence interval To detail the potential effect of other indicators for decision-making of insemination protocols, namely the TMC and maternal age, on the outcomes of interest, we compared the splines where TMC and maternal age were fitted to TFF or CLBR using GAM (Supplementary Figure S 2 , S 3 ) in the matched cohort. It shows that with the range of TMC evaluated in the study, ICSI appeared to lead to a higher TFF. On the other hand, however, the TFF in IVF patients may increase with increasing maternal age, where ICSI patients are less likely to be affected. IVF patients also appeared to have a higher adjusted CLBR, when TMC < 200 million or maternal age between 30 and 40, according to the confidence interval of the splines. To further compare our study with previous evidence, we used the “meta” package to pool the available data on this topic [ 5 , 16 , 25 – 30 ]. For both fertilization rates per oocyte and pregnancy per initiated cycle, a high degree of heterogeneity was observed, regardless of whether our data were included or not. However, the pooled effect of the previous study may slightly favor ICSI, if our data were not included (Supplementary Figure S4 ).

In comparison with previous studies, the strength of the study may include a larger sample size, adjustment for a set of covariates using PS matching, and the reporting of the cumulative outcomes per complete cycle. However, the study is also limited by the retrospective design. In addition, due to the low reproductive potential of poor responders, the current sample size may lack sufficient power to detect the difference in CLBR between the groups, while the confidence interval may suggest a trend favoring IVF. Considering that the latest level 1 evidence suggested that ICSI may slightly decrease the CLBR in patients with an adjusted risk ratio of 0.89 [ 51 ], a similar difference would be expected in poor responders with a larger sample size.

A retrospective study of infertile patients treated at the Reproductive Medicine Center of Xiamen University Affiliated Chenggong Hospital, Xiamen, China, from 2013 to 2022, was included and analyzed. Inclusion criteria were poor responders with one or two oocytes retrieved in a cycle, patients with a complete cycle where live births had been achieved or no surplus frozen embryos left, and patients with a total motile sperm count (TMC) higher than 2 × 10 6 in their male counterparts. Exclusion criteria were severe male factor cases, frozen–thawed sperm, and surgical sperm collection. The reasons for ICSI in the included patients were borderline or suboptimal semen parameters. The suboptimal semen parameters were considered when the semen parameters were above the ICSI indications in our clinic (TMC ≤ 2 × 10 6 ), but lower than the WHO criteria [ 19 ] (concentration ≥ 15 million/ml, motility ≥ 40%, morphology ≥ 4%). Institutional review board approval for this study was obtained from the Ethical Committee of Xiamen University Affiliated Chenggong Hospital. Informed consent was not necessary because this retrospective research was based on non-identifiable records.

Our data show that for poor responders with only one or two oocytes retrieved, ICSI cannot avoid TFF. Alternatively, it reduces fertilization and cumulative pregnancy rates. It is suggested that ICSI is not recommended for poor responders, as it does not prevent TFF and may reduce cumulative pregnancy rates. Furthermore, the technical complexity, the cost, the potential damage to oocytes, and possible long-term health issues are also potential concerns of this invasive technique [ 14 , 34 , 51 ].

In this study, we focused on infertile couples with extremely poor response and non-severe male factors. The results confirmed that canceled ET cycle rates, the rates of cycles with no embryo, embryo quality, and CLBR were similar between IVF and ICSI, indicating that the insemination method did not affect embryo development ability and pregnancy rate. However, ICSI also led to an increased TFF, suggesting against using ICSI to secure the fertilization of these patients. According to the Vienna Consensus, the TFF of IVF and ICSI cycles should be less than 5% [ 31 ]. However, poor responders may suffer from a much higher incidence of TFF than the general population [ 32 ]. Krog et al. reported that when less than four oocytes were retrieved, the TFF rate increased by three times compared with the normal responders [ 33 ]. A possible reason that drives the increasing use of ICSI is the virtual elimination of cases further complicated by TFF, even though the cumulative evidence did not support the use of the procedure in non-male factor infertile patients [ 34 ]. ICSI grants the consistent ability of a viable spermatozoon to activate an oocyte. However, when the problem is oocyte origin, even ICSI may fail [ 35 ]. Our previous study also showed that patients with less than four oocytes had an unexpected TFF rate of 21% in their first treatment and semen parameters were not reliable predictors for TFF for these patients [ 10 ]. It might suggest that for poor responders, the problems associated with oocytes are the major detriments to hamper fertilization. One of the most common reasons for TFF is the nucleus–cytoplasmic maturation asynchrony [ 35 ]. Cytoplasmic maturation includes a series of events that are key to oocyte competence, such as the rearrangement of organelles and the accumulation of mRNA, proteins, substrates, and nutrients [ 36 ]. The calcium store content in the endoplasmic reticulum (ER), which is essential for the fertilization event, also increases markedly during maturation, along with the rearrangement of ER [ 37 ]. Immature cytoplasm may lead to a failed calcium oscillation following fertilization. In this case, even if a single sperm was injected, no sign of fertilization would be observed. For poor responders, the nucleus–cytoplasmic maturation asynchrony might be more frequent due to the possible alternated follicle growth dynamic, abnormal endocrine response, or other stress and anxiety [ 38 ]. The function of cumulus cells surrounding the oocyte which promotes cytoplasm/nucleus maturation through an autocrine/paracrine mechanism may also be impaired in poor responders [ 39 – 41 ]. Compared with conventional IVF, ICSI requires an earlier denudation, leading to a shorter in vitro incubation of the cumulus–oocyte complex. Several studies have shown that the interval between oocyte retrieval and denudation may affect the outcomes of ICSI [ 42 , 43 ] and some authors suggested that a longer interval might be associated with a better outcome [ 40 , 44 ]. A longer interval may allow the oocytes to reach full maturity with the somatic cell compartment [ 45 , 46 ]. For oocytes with a potential nucleus–cytoplasmic maturation asynchrony problem, such extended culture before insemination might rescue the immaturity. A possible interpretation for the lower TFF in conventional IVF in our cohort may be that a longer denudation interval may partially reserve the synchronized nucleus–cytoplasmic maturation. On the other hand, however, immature or over-mature cytoplasm might also lead to polyspermy when conventional IVF is used [ 47 , 48 ]. With conventional IVF, a higher multiple pronuclei (MPN) rate was observed in the cohort. It is generally accepted that MPN formation is due to the abnormal extrusion of the second polar body or to abnormal fertilization with multiple sperm [ 49 ]. With the sign of the extrusion of the second polar body, polyspermy is suspected. Polyspermy could be due to the aging of oocytes or a high number of capitated sperm at the fertilization site [ 47 ]. Practically, these two factors may be additive. Standard sperm concentrations work well for competent oocytes, but may cause polyspermy in those with impaired cortical granule function. Reducing the sperm concentration at the fertilization site with methods such as droplet insemination may be a possible alternative to avoid polyspermy [ 50 ]. Although there have been several previous studies concerning the use of ICSI in patients with low oocyte yield (1, 2) and non-severe male infertility[ 5 , 15 , 16 , 25 , 27 – 30 ], considerable heterogeneity was found when comparing these data (Supplementary Figure S4 ). While the majority of the studies [ 15 , 16 , 25 , 27 , 30 ] suggested a fertilization rate favored ICSI, the work of Wu et al. and Sfontouris et al. suggested similar fertilization rates between IVF and ICSI in patients with non-severe male infertility [ 28 , 29 ]. The source of heterogeneity may be from the patients’ characteristics or practices of laboratories. Considering the relatively small sample size of these studies [ 5 , 29 ], the confounders may lead to considerable fluctuation of outcomes. As our data have shown, the differences in TFF became diminished when the age of patients increased (Supplementary Figure S 2 , S 3 ). On the other hand, however, since the studies were retrospective, inclusion bias may also be considerable. Imbalances in oocyte maturity rates between IVF and ICSI groups were found in several studies [ 25 , 28 , 29 ]. The proportion of MII was usually higher in ICSI patients. For instance, Wang et al. showed a significantly higher MII rate in the ICSI group than in the IVF group in patients with one oocyte and also a higher cumulative live birth rate per cycle [ 25 ]. Since ICSI could only be given to mature oocytes following denudation, it is less clear whether cycles with immature or degenerated oocytes were excluded in these studies. Considering the clinical scenarios, the intention-to-treat inclusion rather than per-protocol inclusion may be more meaningful to the decision-making for patients with very low oocyte yield. The heterogeneity calls for higher-quality evidence on this issue. Our data contribute to the issue by giving evidence in indication-matched, intention-to-treat patients. Similar to fertilization, the reports concerning the effect of ICSI on pregnancy following an initiated cycle in patients with extremely low responses are heterogeneous [ 5 , 16 , 25 – 30 ]. Wang et al. [ 25 ] suggested a cumulative pregnancy rate favoring the ICSI group, which conflicts with our study and Xi et al. [ 30 ]. While low oocyte yield leads to a very limited chance of transfer, earlier studies only reported pregnancies following fresh transfer [ 5 , 16 , 26 – 29 ]. With the increasing use of FET and ovarian stimulation protocols such as the luteal phase stimulation, more recent studies reported cumulative pregnancy or live birth following a complete cycle. Considering the cumulative chance of pregnancy, the latest evidence in normal responders with non-severy infertility has implied that outcomes following complete cycles would be hampered by the ICSI procedure [ 51 ]. A similar effect would be expected in patients with very low oocyte yield, although the small sample size, low number of events, and high uncertainty are also expected in these patients. Current evidence-based guidelines explicitly discourage the routine use of ICSI in cases of unexplained infertility, diminished ovarian response, or advanced maternal age, given the absence of demonstrable clinical benefits over conventional IVF [ 52 ]. However, the lack of significant outcome disparities may also be interpreted as non-inferiority. The justification for ICSI utilization may shift from medical necessity to pragmatic considerations, including operator-dependent technical preferences or patient anxiety regarding TFF. Patients or clinicians may weigh the fear of TFF over cost-effectiveness and use ICSI as a safeguard technically and psychologically. However, the trade-off should be reevaluated if the attempt of ICSI leads to risks such as increased TFF and decreased cumulative pregnancy rates. On the other hand, the concerns of TFF in poor responders might be addressed by other less invasive add-ons or technologies. For instance, recent advances in microfluidic sperm sorting may increase the sperm capacity for fertilization, improving fertilization, embryo morphology, and implantation [ 53 , 54 ]. This novel technology may provide a viable option for patients who are worried about TFF. Future studies may examine whether the technology benefits patients with low oocyte yield.

Oocyte retrieval was performed by transvaginal ultrasound, and follicles were aspirated by a 17 G needle (Cook Medical). Repeated follicular flushing was performed when the oocyte was not retrieved after the initial puncture to maximize the chances of recovery. On the same day, semen was collected in sterile containers and evaluated for sperm density, motility, and morphology according to the World Health Organization (WHO) criteria [ 19 ]. Sperm was prepared by density gradient centrifugation, combined with the swim-up method, then it was incubated in an incubator (C200, Labotect) at 37 °C, 6% CO 2 , and 5% O 2 in a humidified atmosphere. Consultation and decision-making were carried out following OPU for patents with suboptimal semen parameters. IVF is generally recommended unless the patient insists on ICSI [ 20 ]. According to the patient’s prior informed consent, the oocytes were assigned to conventional IVF or ICSI. Insemination was performed 4–6 h after oocyte retrieval, and the sperm concentration was adjusted to 1.5 ~ 3 million/ml for IVF insemination. For ICSI cycles, oocytes were denuded 2 h after oocyte retrieval and ICSI was scheduled 42 h following triggering. Fertilization was judged according to whether the pronucleus appeared at 17 ± 1 h after insemination. Normal fertilization is defined as the observation of two pronuclei (2PN), and multiple pronuclei (MPN) is defined as the presence of ≥ 3 pronuclei. To compare the fertilization rate between IVF and ICSI cycles, the total fertilization rate was defined as fertilized oocyte per retrieved oocyte in both IVF cycles and ICSI cycles. Normal fertilization rate was defined as 2PN zygotes per oocyte. A cycle was considered TFF when no sign of fertilization was observed in all oocytes. Embryos were cultured in cleavage medium (K-SICM, COOK) on days 1–3 after fertilization, and then in blastocyst medium (K-SIBM, COOK) on days 4–7. According to the Istanbul Consensus embryo evaluation criteria [ 21 ], embryos on day 3 were evaluated. Embryos with very poor morphology, arrested embryos, and embryos with all blastomere degenerated or lysed were considered not available for ET. The patients received extended culture to blastocyst stage according to their preference. For blastocyst assessment, we used the Gardner grading system [ 22 ], The score of high-quality blastocysts was ≥ 4BB. Blastocysts with poor morphological score (≤ CC) or low expansion grade (grade 1, 2) were not cryopreserved or transferred. For cryopreservation, a vitrification protocol employing 15% dimethyl sulfoxide, 15% ethylene glycol, and 0.6 M sucrose as cryoprotectants was used for cryopreserved. All embryos were preserved in an open vitrification device (Lifecarrier, Sunlight, Shishou, China). In either fresh cycles or FET cycles, the embryo transfer was scheduled on day 3 if cleavage-stage embryos were transferred and scheduled on day 5 if blastocysts were transferred. Embryo transfer was performed using a COOK catheter (COOK, IN, USA) under trans-abdominal ultrasound guidance. Fourteen days after embryo transfer, serum β-HCG level was measured to determine whether pregnancy occurred. Clinical pregnancy was defined by the presence of a gestational sac, and live birth was defined as the live birth of a fetus over 28 weeks of gestation. Cumulative live birth rate (CLBR) refers to the proportion of individuals achieving at least one live birth from a complete cycle, including all subsequent fresh and frozen embryo transfers [ 23 ]. Similarly, the cumulative pregnancy rate measures the proportion of individuals achieving at least one clinical pregnancy per complete cycle. A complete cycle is defined as a cycle that achieves live birth or all embryo transfer. The primary outcomes were CLBR and cumulative pregnancy rate. The secondary outcomes were total fertilization failure (TFF) and normal fertilization rate.

For PS matching, the covariates include female age, history of spontaneous miscarriage, parity, OPU order, diagnosis of endometriosis or polycystic ovary syndrome, the duration of infertility, female BMI, basal FSH, basal LH, AFC, male age, male BMI, sperm normal morphology rate, TMC, ovarian stimulation protocol, gonadotropin starting dose, oocyte yield, and extended culture. After the two groups were matched by the propensity score matching 1:1 ratio with a caliper of 0.2, case and control discards were permitted. The matching was carried out with a MatchIt package. Standard differences ( D ) were calculated to evaluate the balance of the distribution of the baseline characteristics between the two groups before and after PS matching using a cobalt package. D  < 0.1 was used as the threshold to indicate a negligible difference in the mean or prevalence of a covariate between groups. Multivariate analyses were also carried out in the unmatched and matched cohorts adjusted for the important covariates mentioned above to assess the association between the outcomes (TFF, cumulative pregnancy, CLBR) and insemination protocols. Continuous variables were represented as median [first quartile, third quartile] or median [Min, Max], and mean (SD), while categorical variables were represented as N (percentage). Wilcoxon test was used for continuous variables, and the Chi-square test was used for categorical variables. P  < 0.05 was considered to be statistically significant. All analyses were based on R software [ 24 ].

Since first described in 1992, intracytoplasmic sperm injection (ICSI) has become the gold standard for treating severe male factor infertility in assisted reproductive technologies (ART) [ 1 ]. Nowadays, over half of the ART cycles are inseminated with this technique according to the data of ICMART [ 2 ]. The fact that the number of ICSI cycles outnumbers the ART cycles involving male factor infertility [ 3 ] and varies across geographical regions[ 2 ] suggests that the use of ICSI may be driven by factors beyond male infertility, such as clinic-specific policy [ 4 ] or patients’ attitudes [ 5 , 6 ]. Current evidence shows that ICSI does not improve pregnancy rates or live birth rates in non-male factor infertility [ 7 – 9 ]. The popularity of ICSI is possibly driven by the fear of an expected total fertilization failure during conventional IVF treatment. The fear might be irrational in patients with a normal ovarian response, as the absolute incidence of TFF is rare in the general population [ 10 ]. For patients with only one or two oocytes retrieved, however, the chance of fertilization might be a game of all or none. An intracytoplasmic sperm injection could bypass several pre-fertilization hindrances, confirming the one selected spermatozoon reaches the oocyte, and be deemed as a method to overcome the frustrating TFF [ 11 , 12 ]. However, successful fertilization is not only determined by the entry of the spermatozoon into the oocyte in ICSI cycles [ 13 ]. The developmental competence of the spermatozoon selected by the embryologist cannot be guaranteed, as it also bypasses the natural selection barriers. Additionally, mechanical damage to the oocytes is also a potential concern [ 14 ]. Practically, using ICSI in patients with one or two oocytes does not always promise a better fertilization rate. Poorer or equalized fertilization chances were reported along with improved fertilization rates among previous studies comparing ICSI and conventional IVF in patients with one or two oocytes [ 5 , 15 , 16 ]. The heterogeneity could be due to the low statistical power and high patient heterogeneity in such populations. Furthermore, meta-analyses focusing on advanced maternal age populations indicate that ICSI may not significantly enhance fertilization outcomes when oocyte yield is suboptimal [ 17 , 18 ]. We postulate that the utilization of ICSI in patients with one to two oocytes may offer little substantial clinical advantages. Since a major reason for ICSI overuse is the pressure on the clinicians from the patients` intolerance to TFF, an informed consult would be essential. The present study aimed to provide further evidence concerning the association between ICSI and TFF in patients with one or two oocytes, with a larger sample size and matched indications. In addition, previous evidences were summarized and pooled for further reference.

Below is the link to the electronic supplementary material. Supplementary file1 Figure S1. Distribution of propensity scores (distance) before and after matching (JPG 63 KB) Supplementary file2 Figure S2. Change of cycle outcomes across TMC levels in different insemination protocols. (A) Crude TFF across TMC levels in different insemination protocols. (B) Adjusted TFF across TMC levels in different insemination protocols; (C) Crude cumulative live birth across TMC levels in different insemination protocols. (D) Adjusted cumulative live births across TMC levels in different insemination protocols. Models were adjusted for female age, female BMI, female basal FSH, LH, AFC, previous maternal history, PCOS, endometriosis, OPU order, male BMI, male age, sperm normal morphology rate, total motile sperm number, stimulation protocol, oocyte yield and extended culture in GAM models (TIFF 21793 KB) Supplementary file3 Figure S3. Change of cycle outcomes across MATERNAL AGE levels in different insemination protocols. (A) Crude TFF across MATERNAL AGE levels in different insemination protocols. (B) Adjusted TFF across MATERNAL AGE levels in different insemination protocols; (C) Crude cumulative live birth across MATERNAL AGE levels in different insemination protocols. (D) Adjusted cumulative live births across MATERNAL AGE levels in different insemination protocols. Models were adjusted for female age, female BMI, female basal FSH, LH, AFC, previous maternal history, PCOS, endometriosis, OPU order, male BMI, male age, sperm normal morphology rate, total motile sperm number, stimulation protocol, oocyte yield and extended culture in GAM models (TIFF 21793 KB) Supplementary file4 Figure S4. Pooled analysis of existing evidence of the effect of ICSI on patients with 1–2 oocytes. (A) Pooled analysis of the evidence before our study concerning the fertilization rate in patients with 1-2 oocytes; (B) Pooled analysis of the evidence including our study concerning the fertilization rate in patients with 1-2 oocytes; (C) Pooled analysis of the evidence before our study concerning the pregnancy per cycle in patients with 1-2 oocytes; (D) Pooled analysis of the evidence including our study concerning the pregnancy per cycle in patients with 1-2 oocytes (JPG 319 KB) Supplementary file5 (DOCX 30 KB) Supplementary file1 Figure S1. Distribution of propensity scores (distance) before and after matching (JPG 63 KB) Supplementary file2 Figure S2. Change of cycle outcomes across TMC levels in different insemination protocols. (A) Crude TFF across TMC levels in different insemination protocols. (B) Adjusted TFF across TMC levels in different insemination protocols; (C) Crude cumulative live birth across TMC levels in different insemination protocols. (D) Adjusted cumulative live births across TMC levels in different insemination protocols. Models were adjusted for female age, female BMI, female basal FSH, LH, AFC, previous maternal history, PCOS, endometriosis, OPU order, male BMI, male age, sperm normal morphology rate, total motile sperm number, stimulation protocol, oocyte yield and extended culture in GAM models (TIFF 21793 KB) Supplementary file3 Figure S3. Change of cycle outcomes across MATERNAL AGE levels in different insemination protocols. (A) Crude TFF across MATERNAL AGE levels in different insemination protocols. (B) Adjusted TFF across MATERNAL AGE levels in different insemination protocols; (C) Crude cumulative live birth across MATERNAL AGE levels in different insemination protocols. (D) Adjusted cumulative live births across MATERNAL AGE levels in different insemination protocols. Models were adjusted for female age, female BMI, female basal FSH, LH, AFC, previous maternal history, PCOS, endometriosis, OPU order, male BMI, male age, sperm normal morphology rate, total motile sperm number, stimulation protocol, oocyte yield and extended culture in GAM models (TIFF 21793 KB) Supplementary file4 Figure S4. Pooled analysis of existing evidence of the effect of ICSI on patients with 1–2 oocytes. (A) Pooled analysis of the evidence before our study concerning the fertilization rate in patients with 1-2 oocytes; (B) Pooled analysis of the evidence including our study concerning the fertilization rate in patients with 1-2 oocytes; (C) Pooled analysis of the evidence before our study concerning the pregnancy per cycle in patients with 1-2 oocytes; (D) Pooled analysis of the evidence including our study concerning the pregnancy per cycle in patients with 1-2 oocytes (JPG 319 KB) Supplementary file5 (DOCX 30 KB)