Preimplantation genetic testing for aneuploidy versus no genetic testing in couples undergoing intracytoplasmic sperm injection for severe male infertility: multicentre, open label, randomised controlled trial.

Preimplantation genetic testing for aneuploidy (PGT-A) using next generation sequencing is used to enhance embryo selection by identifying chromosomal abnormalities alongside conventional morphologic assessment in assisted reproductive technologies. While PGT-A is widely used in couples using assisted reproductive technology to overcome infertility problems such as advanced maternal age, repeated implantation failure, and recurrent pregnancy loss, its effectiveness has not been proven in randomised clinical trials. 1 2 3 For infertility caused by male factors, mostly oligozoospermia, asthenozoospermia, teratozoospermia, or azoospermia, intracytoplasmic sperm injection (ICSI) has been adopted as the main therapeutic strategy since its introduction in 1992. 4 5 However, severe male factor infertility was correlated with an increased risk of genetic defects and chromosomal abnormalities in blastocyst embryos, due to sperm defects and the absence of natural elimination of abnormal sperm during fertilisation. 5 6 7 8 9 10 Recent retrospective studies showed that PGT-A was associated with a lower rate of early miscarriage and an increased live birth rate per transfer in couples with severe male factor infertility. 11 12 However, there are no randomised controlled trials assessing the effect of PGT-A in couples with male factor infertility problems undergoing ICSI. Data are lacking on the cumulative live birth rate for a given oocyte retrieval cycle, which is considered one of the most crucial patient centred outcomes in infertility treatment. 13 We performed a multicentre randomised controlled trial in couples with severe male factor infertility undergoing ICSI, and compared the live birth rate after ICSI in couples who had or had not also undergone PGT-A through a combination of morphological assessments and next generation sequencing.

We conducted this multicentre, open label randomised controlled trial at four reproductive medical centres in China (International Peace Maternity and Child Health Hospital, Renji Hospital, Obstetrics and Gynaecology Hospital of Fudan University, and Sir Run Run Hospital of Zhejiang University). All centres were qualified to perform PGT-A. The annual number of in vitro fertilisation cycles in the four centres is approximately 20 000, 8000 of which also included the use of ICSI. Before this randomised controlled trial, nearly 3000 of the ICSI cycles were performed with PGT-A. The andrology laboratories at these centres apply World Health Organization (WHO) 2010 criteria for semen assessment and are certified by the National Health Commission of China. 12 14 The study protocol was approved by the ethics committee at each trial site (GKLW2016-16 on 19 October 2016) and was published previously. 15 Data were collected by research staff at the participating centres, and data management and analyses were conducted at International Peace Maternity and Child Health Hospital in Shanghai. The study was performed in accordance with Good Clinical Practice and Declaration of Helsinki principles, including oversight by an independent Data Safety Monitoring Committee. The trial was registered at ClinicalTrials.gov on 21 October 2016 ( NCT02941965 ). A trial timeline, amendments in study protocol, and the registry are documented in the supplementary appendix. Couples were eligible if both partners were at least 20 years old, planning their first or second ICSI cycle, and if the male partner had severe male factor infertility, defined as a semen concentrate of less than 5×10 6 /mL and/or a progressive sperm motility of less than 10%. Only couples that developed at least four good quality cleavage stage embryos or at least one blastocyst of good or fair grade on day five or day six of the ICSI cycle were eligible. No donated sperm or oocyte was used in this study. Exclusion criteria included obstructive azoospermia, sexual dysfunction or immune infertility (defined as abnormal semen variables, a positive anti-sperm antibodies test, and a history of blood-testis barrier disruption), 16 17 the female partner being 38 years of age or older, uterine abnormalities or a history of recurrent miscarriages, abnormal chromosomal karyotypes, or other contraindications for assisted reproduction. Full eligibility criteria are listed in the supplementary appendix. Local investigators screened couples scheduled for assisted reproductive technology treatment in each centre. Couples with severe male factor infertility who were scheduled for ICSI were informed about the study by a dedicated research staff member before or during ovarian stimulation. Eligibility was confirmed when the embryo cultivation results were obtained. All eligible couples who were willing to participate were required to provide written informed consent after the first screening and before randomisation. After we obtained written informed from the eligible couples with fitted embryo cultivation results, independent staff performed randomisation stratified by maternal age (20-29.9, 30-34.9, ≥35 years) and body mass index (<18.5, 18.5-24.9, ≥25) using an online system with a computer generated randomisation list. Eligible couples were randomly allocated at a 1:1 ratio to either the PGT-A group or the no PGT-A group. Because of the nature of the interventions, the participants, clinicians, embryologists, and investigators assessing the outcomes were not masked to the group allocation. All female participants underwent controlled ovarian stimulation treatment using either a gonadotrophin releasing hormone (GnRH) agonist long protocol, a gonadotrophin releasing hormone agonist short protocol, or a gonadotrophin releasing hormone antagonist protocol for controlled ovarian stimulation in each centre before randomisation. The initial dosage of gonadotrophin was 150-300 IU and was adjusted individually, and gonadotrophin treatment was continued until the day of ovulation trigger. After two or more follicles reached a diameter 18 mm or more, human chorionic gonadotrophin was injected once on the trigger day. Oocyte retrieval was performed 36 hours (±2 h) after human chorionic gonadotrophin injection, and semen samples were obtained on the day of oocyte retrieval by masturbation after 3-7 days of abstention from sexual intercourse. Detailed procedures of ovarian stimulation and oocyte retrieval are reported in the trial protocol. 15 Only oocytes that extruded the first polar body (ie, metaphase II oocytes) were microinjected with a single sperm directly into the cytoplasm. All embryos were frozen after cultivation to blastocyst stage or after biopsy. Based on the Gardner criteria, the selection of a single blastocyst gave priority to the score of the inner cell mass, and the score of the trophectoderm was also considered. 3 18 The blastocyst grading scale, in order from high quality to fair quality, was AA, AB, BA, BB, AC, and BC. We considered scores at the cleavage stage when two or more blastocysts were of equal grade. 19 After randomisation, blastocysts from couples allocated to the PGT-A group were biopsied before being frozen. Generally, three to 10 cells were removed from the trophectoderm by certified technicians for next generation sequencing in our cytogenetic laboratories. The funding of our study covered the genetic testing costs for three embryos in the PGT-A group. The next generation sequencing platforms used in our study included the Illumina NextSeq 550 or Ion PGM/Proton (Thermo Fisher Scientific, Waltham, MA, USA). Based on the estimated proportion of aneuploid cells in the trophectoderm biopsy, embryos were classified as follows: euploid (70% aneuploid cells). This proportion was derived from the fraction of biopsy DNA displaying aneuploid copy number profiles during PGT-A. 20 If amplification failure occurred, the embryo was regarded as a questionable embryo with no signal and was allowed to be transferred according to the same morphologic grading criteria as those used for the ICSI group, after the supply of euploid embryos was exhausted. The next generation sequencing procedures are detailed in the supplementary appendix. For couples receiving ICSI without PGT-A, we assessed the morphological features of the embryos before cryopreservation, with no additional genetic testing performed. According to our study protocol, only one blastocyst could be transferred in the first cycle in both groups. 15 For patients in natural ovulation cycles, luteal phase support began from the day of ovulation with 10 mg oral dydrogesterone (Duphaston, Abbott Biologicals BV, Netherlands) two times per day until the pregnancy test. Patients using an artificial regimen were given vaginal progesterone gel at a dose of 90 mg once daily (Crinone, Merck Serono, Germany), or at a dose of 200 mg three times daily (Utrogestan, Belsins, Belgium), or progesterone injection 60 mg per day together with 10 mg oral dydrogesterone (Duphaston) two times per day until the pregnancy test. We assessed related outcomes resulting from up to three transfer cycles that were performed within 12 months after randomisation. The primary outcomes were the live birth rate in the first embryo transfer cycle and the cumulative live birth rate after three transfer cycles. The live birth rate was defined as the number of viable infants born more than 28 gestational weeks after the first embryo transfer. We defined the cumulative live birth rate as the proportion of live births resulting from up to three transfer cycles performed within 12 months after randomisation. Both live birth rates were calculated by dividing the number of participants obtaining their first live birth by the number of randomised participants. Secondary outcomes included biochemical pregnancy, implantation rate, clinical pregnancy, ongoing pregnancy, pregnancy loss, and time from randomisation to the confirmation of an ongoing pregnancy at 20 weeks’ gestation or later that subsequently resulted in live birth. Obstetric outcomes included gestational weeks, preterm delivery, gestational diabetes, hypertensive disorders of pregnancy, and postpartum haemorrhage. Neonatal outcomes included birthweight (g), low birthweight (<2500 g), macrosomia (≥4000 g), congenital anomaly, and stillbirth. Regarding maternal safety, we reported any discomfort or deterioration of existing discomfort, ectopic pregnancy, and ovarian hyperstimulation syndrome. 21 Definitions of the secondary outcomes are provided in the study protocol and the supplementary appendix. 15 Based on our previous institutional data, we estimated the live birth rate after the first embryo transfer with no PGT-A among couples with severe male factor infertility was 25%. We hypothesised that the minimal clinically relevant difference from PGT-A would be 14%, and that PGT-A would increase the live birth rate from 25% to 39% after the first transfer. We determined that a sample size of 436 women (218 per group) would provide a power of 80% to demonstrate or refute this 14% difference at a two sided α level of 0.05, with an estimated dropout rate of 20%. We ultimately recruited 450 women (225 per group) according to our study protocol, and evaluated the sample size (n=450) for power because of a co-primary endpoint design, using Hochberg’s procedure to account for correlated endpoints. 22 Assuming cumulative live birth rates in couples with severe male infertility of 46% for those who did not have PGT-A and 60% for those who had PGT-A, and a correlation coefficient (ρ) of 0.8 between the two outcomes, the sample size of 450 (225 per group) provided 80% power with a two sided alpha of 0.05 and an estimated dropout rate of 20%. We performed primary analyses on an intention-to-treat basis. Continuous variables are reported as the means (standard deviation (SD)) or medians (interquartile range (IQR)), and comparisons between groups were analysed using the Wilcoxon rank sum test because of the non-normality of the variables. We have reported categorical variables as frequencies and percentages and between-group comparisons were analysed using the χ 2 test. Primary outcomes were compared between groups by calculating the absolute differences and odds ratios, both with 95% confidence intervals (CIs), using a binomial regression model. The generalised linear mixed model was applied for the central effect analysis, with maternal age, body mass index, and trial centres adjusted as random effects for primary outcome. The time to ongoing pregnancy resulting in a live birth was assessed using Cox proportional hazard analysis, and hazard ratios and 95% CIs were calculated. We constructed Kaplan-Meier curves compared them using the log rank test. Worst case and best case imputation analyses were used for missing data. We also conducted a per protocol analysis for primary and secondary outcomes, and performed post hoc analysis for primary outcomes in prespecified subgroups categorised by maternal age, body mass index, type of male infertility factor, trial centre, and number of retrieved oocytes. We further compared primary outcomes among participants who had obtained different numbers of blastocysts in per protocol analysis. All analyses were performed with R software, version 4.0. No patients or public involvement was included in the study design, recruitment, conduct, or interpretation of the results, because the research was initiated before the widespread adoption of patient and public involvement.

Between 15 July 2018 and 6 January 2023, 1347 couples were screened for eligibility, of whom 1005 met the inclusion criteria, and 450 provided informed consent and were randomly allocated to the PGT-A group or no PGT-A group (n=225 couples per group). All participants were followed up for up to three transfer cycles performed within 12 months after randomisation ( fig 1 ). Trial profile. CONSORT (consolidated standards of reporting trials) diagram of all participants who were assessed for eligibility, randomised, and followed up. ISCI=intracytoplasmic sperm injection; PGT-A=preimplantation genetic testing for aneuploidies The baseline characteristics were comparable between the groups ( table 1 ). In the PGT-A group, initial semen analysis identified 56 (24.9%) couples with severe oligozoospermia only, 66 (29.3%) with severe asthenozoospermia only, 80 (35.6%) with both oligozoospermia and asthenozoospermia, and 23 (10.2%) with azoospermia. Corresponding figures in the no PGT-A group were 60 (26.7%) couples with severe oligozoospermia, 64 (28.4%) with severe asthenozoospermia, 76 (33.8%) with oligozoospermia and asthenozoospermia, and 25 (11.1%) with azoospermia. The median numbers of retrieved oocytes were 14 (IQR 9-20) in the PGT-A group and 12 (9-16) in the no PGT-A group ( table 1 ). Baseline characteristics of all randomised participants Data are number (%) unless otherwise stated. GnRH=gonadotrophin releasing hormone; PGT-A=preimplantation genetic testing for aneuploidies; HCG=human chorionic gonadotrophin; MII=metaphase II oocytes; SD=standard deviation; IQR=interquartile range. The baseline level of steroid hormones were measured at the early follicular phase, mostly on days 2-5 of the menstrual cycle. Data were missing for the following steroid hormones: anti-Müllerian hormone (42 women in PGT-A group; 38 women in no PGT-A group), follicle stimulating hormone (eight; four); luteinising hormone (nine; eight); oestradiol (five; 10); and total testosterone (48; 43). One participant from the no PGT-A group did not receive controlled ovarian stimulation treatments within 12 months after randomisation and withdrew from the study. Embryos were rated according to the Istanbul criteria, with good defined as grade I and grade II, a cell number of 7-9, even cell size, <25% fragmentation, and no multinucleation. Of 1101 embryos obtained in the PGT-A group versus 872 in the no PGT-A group, 1047 and 829 were blastocysts, respectively. In general, the numbers of good quality blastocysts, including AA, AB, BA, and BB grades, were similar between the groups (supplementary appendix). Among the 664 (60.3%) blastocysts tested in the PGT-A group, 49.5% (329) were euploid, 7.1% (47) were chromosomal mosaic, and 42.6% (283) were aneuploid ( table 2 ). One questionable embryo with no copy number signal for chromosomes or subchromosomal regions was transferred on the participant’s request after exhaustion of euploid embryos, and achieved a live birth. Twelve couples underwent 16 transfer cycles of 16 mosaic embryos when no available euploid embryos were left. Of these 12 couples, four had live births, three had clinical pregnancy loss, and five did not conceive successfully. Outcomes of ovarian stimulation, embryo culture, and transfer (intention-to-treat analysis) Data are number (%) unless otherwise stated. PGT-A=preimplantation genetic testing for aneuploidy; SD=standard deviation. 22 women in the no PGT-A group who had no blastocysts available for transfer after scanning and seven women in each group whose blastocysts were not obtained were included. Two women allocated to the PGT-A group were excluded because they required only intracytoplasmic sperm injection and fresh embryo transfer. At 12 months after randomisation, 191 (84.9%) participants in the PGT-A group and 211 (93.8%) participants in the no PGT-A group underwent embryo transfers. Eighty nine couples (39/191 (20.4%), PGT-A; 50/211 (23.7%), no PGT-A) underwent two transfer cycles, and 29 couples (13/191 (6.8%), PGT-A; 16/211 (7.6%), no PGT-A) underwent three transfer cycles. The PGT-A group had 256 cumulative transfers and a total of 274 total embryo transfers; the no PGT-A group had 293 cumulative transfers and a total of 353 embryo transfers ( table 2 ). At 12 months’ follow-up, 34 couples in the PGT-A group and 13 in the no PGT-A group had not undergone embryo transfer because no blastocyst had been obtained (seven in each group), because of personal factors (five in each group did not wish to undergo embryo transfer, one in the no PGT-A group did not wish to receive controlled ovarian hyperstimulation treatment), or because no blastocyst was available for transfer after PGT-A (22 women in the PGT-A group; table 2 ). Of 225 couples in each group, live birth after the first embryo transfer from the initiated cycle occurred in 109 (48.4%) in the PGT-A group and in 104 (46.2%) in the no PGT-A group (absolute difference 2.2% (95% CI −7.0% to 11.4%); odds ratio 1.09 (95% CI 0.76 to 1.58); P=0.64; table 3 ). Obstetric and perinatal outcomes (intention-to-treat analysis) Data are number (%) unless otherwise stated. PGT-A=preimplantation genetic testing for aneuploidies; CI=confidence interval; SD=standard deviation. The obstetric outcomes and neonatal outcomes of three newborn babies (including one in the PGT-A group and two in the no PGT-A group) were not obtained because of ongoing pregnancy (with >30 gestational weeks). Positive pregnancy test=serum human chorionic gonadotrophin ≥5 mIU/mL. Clinical pregnancy=detection of a gestational sac in the uterine cavity. Twin pregnancies=detection of two gestational sacs via ultrasound scan five weeks after embryo transfer. Monozygotic twin pregnancies=detection of two fetal heartbeats five weeks after single embryo transfer. Ongoing pregnancy=detection of a viable fetus with a fetal heartbeat at 20 weeks of gestation. For the implantation rate, the numerator was the total number of gestational sacs, and the denominator was the total number of embryos transferred. Two women in the PGT-A group and three women in the no PGT-A group had more than one pregnancy loss during their transfer cycles within 12 months after randomisation. Two women in the no PGT-A group experienced more than one clinical pregnancy loss during their transfer cycles within 12 months after randomisation. After the first embryo transfer from the initiated cycle, the rates of clinical pregnancy and ongoing pregnancy were comparable between the groups. The implantation rates per embryo after initial transfer were comparable with PGT-A and with ICSI alone (116/197 (58.9%) v 130/224 (58.0%)); absolute difference 0.85% (95% CI −8.6% to 10.3%); odds ratio 1.04 (95% CI 0.70 to 1.53); P=0.86; table 3 ). The percentage of biochemical pregnancy was significantly higher among women who conceived from blastocysts derived from ICSI without PGT-A than those who had undergone PGT-A (147/225 (65.3%) v 122/225 (54.2%); −11.1% (−20.1% to −2.1%); 0.63 (0.43 to 0.92); P=0.02; table 3 ). However, the data showed that couples undergoing PGT-A were less likely to have pregnancy loss (13/225 (5.8%) v 43/225 (19.1%); −13.3% (−19.3% to −7.4%); 0.26 (0.14 to 0.50); P<0.001), biochemical pregnancy loss (7/225 (3.1%) v 19/225 (8.4%); −5.3% (−9.6% to −1.1%); 0.35 (0.14 to 0.85); P=0.02), and first trimester pregnancy loss (5/225 (2.2%) v 21/225 (9.3%); −7.1% (−11.3% to −2.8%); 0.22 (0.08 to 0.60); P<0.01; table 3 ). The cumulative live birth rate up to three transfers (at 12 months after randomisation) was 136 (60.4%) in the PGT-A group and 137 (60.9%) in the no PGT-A group (odds ratio 0.98 (95% CI 0.67 to 1.43); P=0.92; table 3 ), with an absolute difference of −0.4% (95% CI −9.5% to 8.6%). The cumulative implantation rate per embryo was also higher with PGT-A than with ICSI alone (153/274 (55.8%) v 177/353 (50.1%); absolute difference 5.7% (95% CI −2.2% to 13.6%); 1.26 (95% CI 0.92 to 1.73); P=0.16; table 3 ). We also recorded corresponding values (PGT-A v no PGT-A) for total pregnancy loss at 12 months’ follow-up (25/225 (11.1%) v 51/225 (22.7%); −11.6% (−18.4% to −4.7%); 0.43 (0.25 to 0.72); P=0.001), biochemical pregnancy loss (11/225 (4.9%) v 22/225 (9.8%); −4.9% (−9.7% to −0.1%); 0.47 (0.22 to 1.00); P=0.05), and pregnancy loss in the first trimester (14/225 (6.2%) v 29/225 (12.9%); −6.7% (−12.1% to −1.3%); 0.45 (0.23 to 0.87); P=0.02; table 3 ). For the remaining embryos at 12 months after randomisation, we found that in participants who did not achieve a live birth, the PGT-A group had fewer remaining embryos (n=68, median 0 (IQR 0-1) per participant) than the no PGT-A group (n = 115, 0 (0-1) per participant; P=0.08) (supplementary appendix). We found no significant differences in primary outcome between the groups after adjustment for central effect (supplementary appendix). After the worst case and best case imputation for missing data, the differences of primary outcomes between groups remained non-significant, whether unadjusted or adjusted for maternal age, body mass index, and study centres (supplementary appendix). The time to ongoing pregnancy resulting in a live birth within 12 months after randomisation was comparable between the groups in terms of the intention-to-treat principle (hazard ratio 1.00 (95% CI 0.79 to 1.27); P=0.99) and the per protocol analysis (0.93 (0.72 to 1.21); P=0.59)( fig 2 ). Despite the crossing Kaplan-Meier curves, the proportional hazards assumption was supported by Schoenfeld residuals analysis (χ 2 =0.23, P=0.63) and segmented time analysis (P interaction =0.92) (supplementary appendix). Kaplan-Meier graphs of cumulative ongoing pregnancy resulting in live birth in intention-to-treat (top) and per protocol analyses (bottom). PGT-A=preimplantation genetic testing for aneuploidies; shaded areas=95% confidence intervals In the neonatal outcomes, birthweight did not differ significantly among the newborns of both groups, but was lower in the PGT-A group neonates than in the no PGT-A group neonates among singleton pregnancies (pregnancies with one fetus only) after the first embryo transfer (mean 3369.8 g (SD 490.6 g) v 3494.7 g (434.2 g); P=0.03; table 3 ). We observed no other significant differences in maternal or neonatal outcomes between the groups in either the intention-to-treat or per protocol analyses ( table 3 ; supplementary appendix). By 31 May 2024, one woman in the PGT-A group and two women in the no PGT-A group had an ongoing pregnancy at more than 30 weeks’ gestation. In the per protocol analysis, we compared the baseline characteristics and primary outcomes of 176 couples who underwent ICSI with additional genetic testing and 197 couples who underwent ICSI only (supplementary appendix). In general, the live birth rate after the first embryo transfer or cumulative live birth rate 12 months after randomisation were consistent with those based on the intention-to-treat analyses (supplementary appendix). In post hoc analyses for primary outcomes in different subgroups, women who were overweight (body mass index ≥25) in the PGT-A group had a significantly higher live birth rate (22/35 (63%) v 11/31 (36%); odds ratio 3.08 (95% CI 1.13 to 8.41); P=0.03) and cumulative live birth rate (28/35 (80%) v 15/31 (48%); 4.27 (95% CI 1.44 to 12.66); P=0.01) than those in the no PGT-A group (supplementary appendix). Therefore, we further compared baseline characteristics in women who were overweight (35 in PGT-A group, 31 in no PGT-A group) and observed a higher number of retrieved oocytes per cycle from the PGT-A group (median 15 (IQR 10-20) v 14 (6-16), P=0.03) as well as a higher number available blastocysts from the PGT-A group (4 (IQR 2-7) v 2 (1-4), P<0.01; supplementary appendix). For women with one to three blastocysts, live birth rates were lower in the PGT-A group than in the no PGT-A group, although the differences were not significant for live birth rates (29/77 (37.7%) v 46/101 (45.5%); odds ratio 0.72 (95% CI 0.40 to 1.32); P=0.29) and cumulative live birth rates (33/77 (42.9%) v 57/101 (56.4%); 0.58 (95% CI 0.32 to 1.05); P=0.07). For women who had obtained more than three blastocysts, post hoc analysis showed no significant differences comparing live birth and cumulative live birth between groups, regardless of whether three or more than three blastocysts were tested (supplementary appendix).

We showed that in couples with severe male infertility, undergoing ICSI with the addition of PGT-A did not result in a higher live birth rate compared to undergoing ICSI alone. Although PGT-A reduced total pregnancy loss by limiting the transfer of unqualified embryos, this difference did not translate into a higher live birth rate after the initial embryo transfer, nor a higher cumulative live birth rate within 12 months after randomisation. The strengths of our study include its multicentre and randomised design, large sample size, and our reporting of the comprehensive range of pregnancy outcomes assessed. We also reported the cumulative live birth rates rather than focusing on the success rate per transfer. 23 However, limitations remain: firstly, male infertility should be diagnosed based on a comprehensive evaluation, including reproductive history, physical and ultrasonic examination, semen testing, and additional tests if necessary. 24 25 Tests that are not routinely recommended in clinical practice, such as hormone profiles, sperm DNA fragmentation evaluations, and genetic testing, were not included as diagnostic criteria in our study. Our study results need to be confirmed in couples from whom more embryos could have been obtained and tested. Secondly, the generalisability of the study was limited owing to the specific eligibility criteria and the one year follow-up period after randomisation. Thirdly, approximately 8.4% (38/450) of the patients deviated from the protocol, but the results of per protocol analysis with exclusion of this population were consistent with those of intention-to-treat analysis. In addition, randomisation was not stratified by number of retrieved oocytes, but post hoc analysis results of baseline characteristics and primary outcomes in women with different number oocytes were comparable with intention-to-treat analysis (supplementary appendix). Also, the funding of our study only covered the costs of genetic testing for one to three blastocysts per couple, which could influence patient decisions and the study results; however, the post hoc analysis showed no significant differences in primary outcomes (supplementary appendix). Furthermore, our study was not powered to detect or refute between group differences in live birth rate after the first transfer smaller than 14%. Therefore, a smaller increase in live birth rates cannot be excluded. However, none of the other randomised controlled trials on PGT-A reporting on (cumulative) live birth rates have shown a benefit from PGT-A, with some studies showing harm to participants. 26 The 42.6% aneuploidy rate identified in our study of severe male infertility patients provides a rationale for evaluating the potential benefits of PGT-A for this specific population. In 2015, Coates et al 8 showed that severe male factor infertility was associated with a substantial increase in the occurrence of sex chromosome abnormalities in blastocyst embryos, but no pregnancy outcome was reported in their study. However, Mazzilli et al’s 27 observational cohort study involved 1219 ICSI cycles with PGT-A and demonstrated that severe male factor infertility may decrease only the fertilisation rate and interrupt blastocyst development but not the percentage of euploid blastocysts, which is not significantly associated with semen quality. A retrospective cohort study analysing 35 797 embryos also found no significant associations between embryo ploidy and male age or sperm quality. 28 In another observational study, live birth rates of 55.6% and 51.1% were reported in the PGT-A and non-PGT-A groups, respectively, among 266 cases of severe male factor infertility, which indicates that the use of PGT-A is not likely to be an independent factor associated with the live birth per transfer in couples with severe male factor infertility. 12 Additionally, we found no other significant differences in obstetric or neonatal outcomes between the groups, despite four outcome values missing due to ongoing pregnancy by the end of the planned follow-up. These results further suggest that PGT-A is not recommended as a priority strategy for couples with severe male factor infertility, despite a possibly increased rate of aneuploidy, specifically in women with a low number of blastocysts. However, higher live birth rates and cumulative live birth rates observed in women with a body mass index of 25 or more in the PGT-A group suggested that PGT-A could possibly benefit this group of women, although the result was exploratory and warrants further investigation in future studies. The live birth rate has been recommended as a patient centred outcome in randomised controlled trials of infertility treatment. PGT-A is purported to help with the birth of a healthy child and reduce the economic and psychological burden of implantation failures and miscarriages. 29 Our results revealed a higher implantation rate per embryo and a lower rate of early pregnancy loss when PGT-A was included in couples undergoing ICSI. These findings indicate that embryos selected by PGT-A are more likely to have better developmental potential in the first trimester and are more favourable for people who have recurrent miscarriages. However, the clinical pregnancy rate per woman was lower after PGT-A, and no differences were found in the live birth rates or cumulative live birth rates between the groups, which is consistent with the findings of several previous studies. 3 12 30 We first considered the possibility that the results of trophectoderm biopsy could not completely represent the genetic composition of the inner cell mass of the blastocyst, which is the precursor to the embryo. 31 Whether an embryo is able to achieve a live birth not only depends on an assumed normal chromosome constitution, but also on other factors (eg, steady morphokinetic patterns influenced by gene expression, cellular mechanisms, laboratory conditions). 32 Trophectoderm biopsy was conducted only in women who underwent PGT-A; to date, there is no sound evidence from prospective randomised controlled trials regarding the impact of embryo biopsy on maternal and neonatal health, and the safety of PGT-A for women undergoing infertility treatment remains a concern. 33 PGT-A could reduce the chance of a live birth, but this harm is balanced by an increased chance of an euploid embryo being transferred. The lower birthweight of newborns from singleton pregnancies in the PGT-A group in our study also suggested that potential adverse effects of PGT-A did not necessarily lead to pregnancy loss, but did likely lead to a less optimal pregnancy outcome. Therefore, we further suggested that the need for PGT-A should be carefully considered to avoid unnecessary procedures because of the potential risks of invasive procedures, in short term and long term outcomes, for offspring. 34 In addition to participant safety and effectiveness of the treatment, cost is an essential factor that needs to be considered during the decision making process when discussing treatment strategies. Given the current lack of insurance coverage available for assisted reproductive technology in China, the average cost per ICSI cycle is approximately $4000 (£3060; €3470) to $7000. For PGT-A, extra testing fees of $700 per embryo are required in our study centres, increasing the economic burden on patients and the healthcare system. These high costs, plus the lack of beneficial effect of PGT-A found in our study to couples with infertility problems, should be considered carefully by decision makers. In conclusion, we found that the use of PGT-A alongside ICSI did not improve the live birth rate compared with routine ICSI alone among couples with severe male factor infertility, but potentially lowered the risk of pregnancy loss. Further studies are needed to evaluate more precise indications for PGT-A in assisted reproductive technology practices. Preimplantation genetic testing for aneuploidies (PGT-A) is increasingly used worldwide as a method of embryo selection in assisted reproductive technologies Previous retrospective studies found no improvement in live birth rate in the PGT-A group compared with no PGT-A in patients with severe male factor infertility Evidence of prospective studies is lacking In couples with severe male factor infertility undergoing ICSI, PGT-A did not improve live birth rate after the first embryo transfer or the cumulative live birth rate up to three transfer cycles, at 12 months after randomisation PGT-A was observed to reduce rates of pregnancy loss PGT-A should be carefully considered in patients undergoing ICSI, for reasons such as high cost and potential maternal and neonatal risks