Introduction

Intracytoplasmic sperm injection (ICSI) is an insemination method in which a single sperm is directly injected into an egg to achieve fertilization1. Initially developed for severe male factor infertility involving poor sperm quality or quantity2,3. ICSI is currently increasingly used in cases of non-male factor infertility, including previous failed conventional in vitro fertilization (IVF) cycles, unexplained infertility, or advanced maternal age. However, its efficacy and safety in these cases are still under investigation4,5. While some studies propose that ICSI enhances fertilization rates and embryo quality in these cases6, others report that ICSI does not improve pregnancy or live birth rates in these cases and may even increase the risk of birth defects7,8. The American Society for Reproductive Medicine (ASRM) guidelines emphasize insufficient evidence to universally recommend ICSI for non-male factor IVF cycles, advocating instead for individualized decision-making based on clinical context and patient preferences2.

Frozen-thawed embryo transfer (FET) is a treatment cycle in which cryopreserved embryos created in the oocyte pick-up cycle are thawed and transferred into a woman’s uterus9. FET offers multiple advantages, including higher success rates, scheduling flexibility, reduced ovarian hyperstimulation syndrome (OHSS) risk, multiple transfer opportunities, cost-effectiveness per live birth, and lower risk of preterm delivery, small for gestational age and low birth weight for singleton babies10,11,12,13,14.

In non-male factor infertility, prevailing discussions have focused primarily on patients undergoing fresh embryo transfer15,16,17. Notably, only the study by Li et al.7 has evaluated insemination method (ICSI vs. IVF) effects on embryos used in FET cycles, incorporating both fresh transfers and subsequent FETs. Insemination method may influence embryo quality, as fresh embryos generally exhibit higher implantation potential than frozen-thawed counterparts, irrespective of fertilization technique. However, cryopreservation processes may compromise embryo viability, with ICSI-inseminated embryos potentially showing marginally lower post-thaw survival rates compared to conventionally fertilized embryos7. To address this knowledge gap, this study aims to investigate the potential impact of insemination methods involving embryos transferred in FET cycles on the reproductive outcomes of couples with non-male factor infertility.

Methods

Study population and participants

The anonymized dataset was obtained from the Human Fertilisation and Embryology Authority (HFEA) (www.hfea.gov.uk), which is freely accessible for public use and covers the period from 1995 to 2018. Ethical approval was waived as this study utilized de-identified secondary data. For the purpose of this study, only cycles with non-male factor infertility and FET treatment were initially included. The detailed inclusive procedure of cycles of interest is shown in Fig. 1. Specifically, cycles prior to 1995 were not considered for inclusion because of the extremely low number of ICSI uses in non-male factor infertility; cycles involving embryos subjected to biopsy and genetic testing were excluded; either cycles of transferred embryos developed from donor gametes or inseminated with unknown insemination methods were excluded; Finally, the included cycles were divided into two groups (ICSI vs. IVF) based on the insemination methods of transferred embryos.

Fig. 1
figure 1

Flowchart of the included cycles.

Primary outcome measure

The primary outcomes were clinical outcomes such as the clinical pregnancy rate, miscarriage rate, live birth rate, and implantation rate. Clinical pregnancy was defined as the presence of one or more gestational sac(s) with evidence of foetal heart pulsations observed via ultrasound examination. Live birth was defined as the delivery of at least one live baby, which was recorded in the dataset. Miscarriage was defined as the loss of a clinical pregnancy that did not end up with termination, live birth or still birth (still birth: where a baby is born after 24 weeks gestation showing no signs of life). The implantation rate was calculated as the proportion of fetal sacs with fetal pulsation per transferred embryo. The pregnant women who were lost to follow-up were recorded as not experiencing any events. The clinical pregnancy rate and live birth rate were calculated per cycle of treatment, whereas the miscarriage rate was calculated per clinical pregnancy.

Secondary outcome measures

The secondary outcomes were neonatal outcomes, such as gestational age, birthweight, and sex ratio at birth, which were separately shown on the basis of the number of live births. Although the data are categorical variables, gestation weeks were further divided into three groups: very preterm birth (< 32 weeks), preterm birth (32–36 weeks), full-term birth (37–39 weeks), and post-term birth (≥ 40 weeks, which is different from the usual threshold of 42 weeks), and birth weight was divided into four groups: very low birth weight (< 1.5 kg), low birth weight (1.5–2.5 kg), normal birth weight (2.5–4.0 kg), and high birth weight (> 4.0 kg).

Statistical analyses

Because the characteristics are shown as categorical variables, the data are displayed as relative frequencies. Comparisons of the general characteristics and reproductive outcomes between the IVF group and the ICSI group were performed via the chi-square tests. Owing to the feature of the anonymous data, we were unable to link the cycles that belonged to the same couples. It was thus impossible to calculate the cumulative clinical pregnancy rate or cumulative live birth rate to investigate the impact of insemination methods on reproductive outcomes via the generalized estimation equation. However, univariable or multivariable logistic regression could be used to determine the effects of ICSI compared with those of IVF on reproductive outcomes. Given that neonatal outcomes significantly differ between singleton and twin live births, the birthweight and gestation times of the ICSI and IVF groups were compared separately. Statistical analyses were carried out via IBM SPSS Statistics 22.0 (Armonk, NY: IBM Corp.). P values < 0.05 were considered statistically significant for the analysis.

Results

Patient characteristics

As shown in Fig. 1, a total of 57,907 cycles from 1995 to 2018 were included in the final analysis, with 12,976 cycles (22.4%) ICSI-inseminated embryos. Although the ages of the females in the ICSI and IVF groups were similar, the number of previous treatment cycles significantly differed. The proportions of women who achieved successful clinical pregnancy and delivery through ART were significantly greater in the ICSI group. Furthermore, women in the ICSI group were more likely to be diagnosed with unexplained infertility but less likely with infertility with tubal disease and endometriosis. The ICSI group had a significantly greater proportion of cycles with the single embryo transfer strategy (Table 1).

Table 1 Patient general characteristics.

Primary outcomes

The overall implantation rate, clinical pregnancy rate, and live birth rate were significantly greater in the ICSI group (19.9% vs. 15.8%, P < 0.001; 29.6% vs. 26.0%, P < 0.001; 21.5% vs. 19.1%, P < 0.001) than those in the IVF group, and the differences mostly remained consistent when subgroup analyses on the basis of stratified ages and numbers of transferred embryos were performed. For women without tubal disease, ovulatory disorders, or endometriosis, there were also significant differences between the two groups in terms of the abovementioned reproductive outcomes (Table 2). The logistic regression revealed higher odds of clinical pregnancy [crude OR: 1.20 (1.15, 1.25), P < 0.001] and live birth [crude OR: 1.26 (1.06, 1.50), P = 0.006] in the ICSI group compared with the IVF group (Table 3). Temporal trends (Fig. 2) showed marked improvements in clinical pregnancy and live birth rates from 1995 to 2018, likely reflecting advancements in ART techniques (e.g., ICSI optimization, vitrification protocols). The logistic regression analysis was then adjusted for treatment period, female age, causes of infertility, delivery history by ART treatment, number of transferred embryos, and transfer strategy. The results revealed that ICSI no longer significantly impacted on clinical pregnancy [adjusted OR: 1.01 (0.95, 1.05), P = 0.969] and live birth [adjusted OR: 1.05 (0.86, 1.28), P = 0.611]. The overall miscarriage rate was similar between the two groups (24.5% vs. 23.4%, P = 0.149), and the results remained similar when subgroup comparisons were performed. The logistic regression analysis confirmed that ICSI significantly increased the miscarriage rate with an adjusted OR of 1.15 (1.03, 1.29), P = 0.010, which was no longer the case in cycles transferring more than two embryos (Table 3).

Table 2 Clinical outcomes across the subgroups.
Table 3 Clinical outcomes between the ICSI and IVF groups.
Fig. 2
figure 2

Overall clinical pregnancy rate, miscarriage rate, and live birth rate from 1995–2018.

Secondary outcomes

There were similar proportions of singletons in the ICSI and IVF groups (85.3% vs. 84.4%, P = 0.235). Unadjusted analyses revealed that ICSI was associated with reduced odds of very preterm birth, preterm birth, and post-term birth, with crude ORs of [0.65(0.43, 0.97), P = 0.037; 0.81 (0.67, 0.98), P = 0.033; and 0.87 (0.79, 0.96), P = 0.005, respectively] in singletons. After adjusting for confounders, the impact of ICSI became nonsignificant. Although ICSI significantly decreased the proportion of low weight and overweight individuals [OR: 0.52 (0.32, 0.84), P = 0.008; 0.85 (0.73, 0.98), P = 0.021], the confounders significantly attenuated the effect of ICSI on normal weight [adjusted OR: 0.90 (0.77, 1.04), P = 0.144]. Finally, ICSI had a nonsignificant effect on the sex ratio at birth with an adjusted OR of [0.91 (0.83, 1.01), P = 0.052] (Table 4). For multiple live births, ICSI was associated with a greater proportion of low birthweight, with a crude OR of [1.28 (1.01, 1.51), P = 0.003]. After adjustment for confounders, ICSI no longer impacted the birthweight. Furthermore, there were no significant differences in terms of gestational age or sex ratio at birth between the ICSI and IVF groups (Table 4).

Table 4 Neonatal outcomes between the ICSI and IVF groups.

Discussion

This retrospective study analyzed data from the HFEA registry spanning 1995–2018, focusing on FET cycles involving non-male factor infertility. After adjusting for confounders, ICSI demonstrated no significant advantages over IVF in clinical pregnancy rates (29.6% vs. 26.0%) or live birth rates (21.5% vs. 19.1%). Neonatal outcomes, including birthweight and gestational age, showed no statistically significant intergroup differences.

This retrospective study provides solid evidence on the impact of ICSI compared with IVF on the reproductive outcomes of infertile couples with non-male factor infertility, which is consistent with the evidence from studies that included patients mostly undergoing fresh embryo transfer. Both Tannus et al. and Farhi et al. investigated the impact of ICSI on reproductive outcomes in women of advanced maternal age with non-male factor infertility18,19. However, their conclusions contradict each other, which could be explained by the study design and definition of advanced age. Another two retrospective studies including patients with poor oocyte yields20,21, together with the RCT performed by Bhattacharya et al., which was the first high-quality study challenging the use of ICSI in couples with non-male factor infertility16, all of which demonstrated that ICSI did not improve reproductive outcomes in the absence of male factor infertility. However, it should be noted that the abovementioned studies all included patients with fresh embryo transfer treatment, the conditions of which could differ from those of FET treatment. For this reason, more evidence on the impact of ICSI on reproductive outcomes in couples with non-male factor infertility is warranted.

Four population-based studies recording the details of ART treatments in local places, the data of which were collected by local authorities, shed light upon the effectiveness of ICSI on reproductive outcomes in the absence of male factor infertility7,15,22,23. The studies performed by Schwarze et al. and Supramaniam et al. included cycles with fresh embryo transfer. Although both studies reported no benefits of ICSI for couples with non-male factor infertility, conflicting data exist between the two studies. While data from the Latin American Registry revealed a significantly lower live birth rate in the ICSI group than in the IVF group (22.99% vs. 28.76%)22, HFEA data paradoxically indicated increased clinical pregnancy rates (31.4% vs. 37.3%) and live birth rates (27.2% vs. 32.5%) in the ICSI group23. However, the other two population-based studies, which were able to link subsequent fresh and frozen embryo transfers to the initial retrieval cycle, made the cumulative pregnancy rate or cumulative live birth rate calculable7,15. It should be noted that although both studies demonstrated that the cumulative live birth rate of ICSI was similar to that of IVF, the reported cumulative live birth rates were mixed by fresh and frozen-thawed embryo transfer, and subgroup analyses based on fresh and frozen-thawed embryo transfers were not carried out. The freezing and thawing process can stress the embryos’ cellular structures (24). Specifically, embryos created via ICSI may be more vulnerable due to their unique developmental origins. The injection of sperm directly into the egg during ICSI could impact the embryo’s membrane integrity or cytoskeletal structure, making them potentially more sensitive to the stresses of freezing and thawing. Conventionally fertilized embryos, on the other hand, undergo natural sperm-egg interaction, which might better preserve their ability to handle cryopreservation. Their cellular machinery may be better equipped to cope with the freezing process. As a consequence, it was necessary to investigate the potential impact of insemination methods on women undergoing the FET treatment only.

Two other well-designed RCTs also investigated the impact of ICSI compared with that of IVF on the reproductive outcomes of couples with non-male factor infertility. There may be little difference with regard to early embryo development potential; however, the clinical and neonatal outcomes between the ICSI and IVF groups were similar in both studies, which led to the conclusion that the routine use of ICSI in the absence of male factor infertility was not suggested25,26. Nevertheless, the comparisons of reproductive outcomes between the ICSI and IVF groups were mixed by fresh and frozen-thawed embryo transfer in both RCTs.

Although our data were extracted from the same dataset used in the study by Supramaniam et al.23 and both studies included cycles with non-male factor infertility, the present study reported overall clinical pregnancy rates of 29.6% and 26.0% for embryos inseminated by ICSI and IVF, respectively, and overall live birth rates of 21.5% and 19.1% for embryos inseminated by and IVF, respectively, which were significantly lower than the overall clinical pregnancy rate and live birth rate reported by Supramaniam et al. These discrepancies may be explained by the fact that the freezing method was changed from slow freezing to vitrification, which is associated with better embryo development potential after the freeze‒thaw procedure, and that the ICSI procedure has been promoted since its application in clinical use, which is associated with a decreased oocyte degeneration rate. Furthermore, the study by Supramaniam et al. did not compare neonatal outcomes between the ICSI and IVF groups. These are also other important indices for assessing the efficacy and safety of insemination methods. In fact, few retrospective studies, with the exception of the study by Schwarze et al., reported neonatal outcomes22. The results revealed that ICSI was associated with significantly greater birthweight in the ICSI group when singletons were born [3064.6 (512.8) vs. 2993 (479.5), difference 71.58 (44.15 to 99.01)]. In the present study, although neonatal outcome data were used as categorical variables, comparisons between the ICSI and IVF groups were possible. Compared with IVF, ICSI was not associated with higher birthweight or full-term birth after adjustment for confounders, regardless of whether singleton or multiple live births were delivered. The inconsistent results regarding birthweight could be explained by several factors. First, the choice of ICSI as the insemination method varied dramatically between the two studies, which could be explained by the opposite proportion of cycles that were inseminated by ICSI. Under such conditions, there may be significantly different characteristics that could affect neonatal outcomes. Second, the present study included patients with FET treatment, whereas the study by Schwarze et al. included patients with fresh embryo transfer. It is well known that neonatal outcomes differ between fresh and frozen-thawed embryo transfer, which might mask the impact of ICSI on neonatal outcomes12,14. Finally, the present study included cycles spanning two decades. The large sample size made the statistical analysis more powerful.

Several population-based studies from China have demonstrated that ICSI is associated with a lower rate of male live births27,28,29,30,31. The present study confirmed that compared with IVF, ICSI is associated with a lower proportion of male live births. However, after adjusting for confounders, ICSI no longer showed significant impact on sex ratio at birth. However, the present study and the study by Supramaniam et al. included different study populations from UK reproductive centers. These two studies yielded a similar conclusion: the number of male offspring was greater in the IVF groups than in the ICSI groups during FET cycles. The possible explanation may be that embryologists select sperm based on morphology and motility, which may inadvertently favor X-bearing sperm, especially in cases of male infertility where Y sperm might be more fragile or have structural abnormalities. However, in the present study, the determined impact of ICSI on sex ratio was weakened after controlling for the confounders. On one hand, the study included couples with non-male factor infertility, which leads to embryologist’s equal selection of X- and Y-bearing sperm, in turn leading to a sex ratio closer to natural or conventional IVF levels. On the other hand, we could not divide the cycles into subgroups based on the stages of transferred embryo, which were reported to significantly impact the sex ratio.

There were several limitations of the present study. First, these retrospective HFEA data were anonymized, making it impossible to analyze the cumulative clinical pregnancy rate and cumulative live birth rate per woman being treated with FET. Nevertheless, the large sample size made the comparison of interest more powerful. Second, detailed information about the transferred embryos, such as quality and developmental stage, was not recorded, both of which would surely impact reproductive outcomes, and subgroup analyses based on the embryos were not possible. Furthermore, confounders such BMI, smoking status, and endometrial thickness were not recorded. However, considering the large sample size, the abovementioned confounders and their impacts were randomly distributed between the two groups. Third, most of the data of interest were reported as categorical variables, which limits comparisons with other studies.

In conclusion, in non-male factor FET cycles, ICSI confers no clinical advantage over IVF in reproductive or neonatal outcomes. These results underscore the need for judicious ICSI application, reserving its use for clear male factor indications.