Male subfertility and the risk of major birth defects in children born after in vitro fertilization and intracytoplasmic sperm injection: a retrospective cohort study

We conducted a registry-based, retrospective cohort study using a Japanese national ART registry database assembled by the Japan Society of Obstetrics and Gynecology (JSOG). All ART clinics and hospitals are required to contribute to this database, which has previously been reported in detail [11]. The data comprised all cycle-specific information, including infertility diagnosis, IVF or ICSI, and pregnancy and obstetric outcomes. JSOG requires all participating clinics or hospitals with delivery facilities to report delivery outcomes to the ART registry. ART clinics without delivery facilities usually receive delivery reports from referral delivery hospitals or clinics and then report these to the registry. JSOG strongly recommends that ART clinics or hospitals that cannot obtain such delivery reports contact the mothers directly to obtain self-reported obstetrical outcomes. JSOG prohibits the use of donor gametes or embryos, and all embryos transferred were therefore autologous. Similarly, preimplantation genetic testing (PGT) for chromosomal aneuploidy is not allowed in Japan, and PGT for monogenic or structural rearrangements is only allowed after approval by both the institutional review board at each facility and the JSOG board of ethics. A total of 671 cycles of oocyte retrieval involving PGT for monogenic or structural rearrangements were reported between September 2004 and September 2012 in Japan, resulting in 65 singleton births [12].

Fertilization-related information, such as IVF or ICSI status, was only available for fresh cycles. We therefore included all singleton live births and stillbirths after 22 weeks of gestation, or those with a birth weight > 500 g in cases of unknown gestational age, after fresh embryo transfer cycles. We also included pregnancies that were terminated because of birth defects. Overall, 69,346 cycles were eligible for analysis in this study (68,832 cycles for live births, 284 for stillbirths, and 230 for selectively terminated cases). Among these, we excluded cycles using previously frozen oocytes (n = 24), gamete intra-Fallopian transfers (n = 7), split-ICSI (n = 6493) or ICSI using non-ejaculated sperm (n = 1298), unknown fertilization methods (n = 262), and other embryo stages at transfer (n = 1291). A total of 59,971 cycles (59,527 live births, 248 stillbirths, and 196 selectively terminated pregnancies) were included in the final analysis.

This study was approved by the Institutional Review Board of the National Center for Child Health and Development (Institutional Review Board Approval no. 34), Saitama Medical University (Institutional Review Board Approval no. 873), and the JSOG Ethics Committee (Approval no. IM007). After receipt of approval, JSOG provided the required data without any personal identifying information. We analyzed data from 2007 to 2014 because JSOG implemented mandatory online reporting for all institutions from 2007, and the most recent data available at the time of this study were from 2014.

We included major birth defects that were identified before the end of the neonatal period (within 28 days). We defined major birth defects according to the US Centers for Disease Control and Prevention (CDC) guidelines [13]. These birth defects were classified based on blinded review of abstraction forms by a medical doctor who had completed residencies and was board-certified in obstetrics and pediatrics (J.J.). Fetuses and infants with minor anomalies as defined by the CDC, with complications caused by prematurity, and with unspecified conditions or suspected anomalies without definite diagnoses were defined as having no major anomalies. Classification of major birth defects according to organ systems does not suitably express the overall panorama of major birth defects of different natures. We therefore investigated the most common specific major birth defects that occurred in sufficient numbers to examine each organ system. These birth defects included ventricular septal defect, atrial septal defect, tetralogy of Fallot, omphalocele, gastroschisis, diaphragmatic hernia, polydactyly, syndactyly, hypospadias, alimentary atresia (esophageal, small intestine, and rectal and large intestinal atresia), anencephaly (including acrania), spina bifida, and cleft lip with and without cleft palate. To create more uniform case categories, cases with chromosomal abnormalities (n = 279) were excluded from the major birth defects, including 271 (97.1%) cases of chromosomal aneuploidy, four cases of deletion, two cases of inversion, one case of unbalanced translocation, and one case of additional chromosomal material. Among the eight cases of chromosomal abnormalities other than chromosomal aneuploidy, there were only two cases of major congenital anomalies. Conditions that were not defined by the CDC were discussed by a board-certified obstetrician and pediatrician (J.J) and a board-certified obstetrician and clinical geneticist (S.C.J.), blinded to the methods of conception. These conditions were classified as major birth defects based on the presence of structural abnormalities causing functional impairment and requiring surgical correction.

Infertility diagnosis was classified in the registry as tubal factor, male subfertility, endometriosis, antisperm antibody, unexplained, or other, with multiple answers allowed. Male subfertility included abnormal results of semen analysis or sexual or ejaculation dysfunction, and was diagnosed in each institution. Patients who failed to meet any of the specific infertility diagnoses including male subfertility were defined as having unexplained infertility. Sperm parameters, including total sperm motility and sperm concentrations used for fertilization, were available for 52,968 cycles (88.3% of the total sample). Total sperm motility (%) was recorded in the database to the nearest integer, and sperm concentrations (× 10⁶/mL) were calculated to two decimal places. We defined oligozoospermia as a sperm concentration < 15 × 10⁶/mL, asthenozoospermia as total sperm motility < 40%, and oligoasthenozoospermia as both of the above, according to the World Health Organization classification of subfertility [14].

We first compared baseline characteristics between fertile and subfertile men in IVF/ICSI cycles using χ² or Student’s t-tests. Second, we assessed the effect of male subfertility on major birth defects by comparing the prevalence of major birth defects in fertile and subfertile men within IVF or ICSI cycles. We calculated ORs and 95% CIs using generalized estimating equations with robust variance estimation, adjusting for the effect of clustering of births within clinics or hospitals. Our potential confounders included for adjusted analysis were maternal age (categorized into 5-year age groups), calendar year, embryo stage at transfer, and fetal sex. Analysis was restricted to male infants for the outcome of hypospadias. Third, the effect of treatment (i.e., IVF/ICSI) was assessed by comparing the prevalence of major birth defects between IVF and ICSI among fertile and subfertile men, and ORs of ICSI compared with IVF were calculated for major birth defects. Finally, we performed subgroup analyses of cycles for which sperm parameters were available by calculating ORs of different thresholds of sperm concentrations/motility for major birth defects with ICSI and IVF. These thresholds were defined as oligozoospermia (< 15 × 10⁶/mL) and severe oligozoospermia (< 5 × 10⁶/mL), and sperm motility was defined as asthenozoospermia (total sperm motility < 40%) and severe asthenozoospermia (total sperm motility < 25%). Because the number of cases of oligozoospermia was small in IVF cycles (n = 868), we did not stratify the category further into severe oligozoospermia. We also investigated the risk of major birth defects in ICSI among cycles with normal semen concentrations and motility. Adjustments for multiple comparisons were not applied because of the exploratory nature of this study. All analyses were performed using the STATA SE statistical package, version 12.1 (Stata, College Station, TX, USA). A two-tailed value of P < 0.05 was considered statistically significant.