Association between in vitro fertilization-embryo transfer and hearing loss: risk factors for hearing loss among twin infants in a cohort study

Assisted reproductive technologies (ART), including in vitro fertilization-embryo transfer (IVF-ET) and intracytoplasmic sperm injection (ICSI), are known to contribute a higher risk of birth defects; however, studies have rarely evaluated the association between IVF-ET and diagnostic hearing loss (HL). This study aimed to evaluate the prevalence of and risk factors for HL and to clarify the association between IVF-ET and HL among twinborn infants. We enrolled 1860 live-born twin neonates born at a hospital in China from January 2017 to December 2020. After multi-step hearing screening, participants were diagnosed with HL by pediatric audiologists at 6 months of age. The prevalence of hearing loss and the adjusted odds ratios (AORs) for specific risk factors were estimated using generalized estimation equation (GEE) models in twin-born infants. Characteristics and prevalence of failure for hearing screening and HL were measured in IVF-ET twin infants. IVF-ET conception and preterm birth conferred a higher risk of hearing loss, with increased adjusted odds ratios (AOR [95% confidence intervals (CI)] IVF-ET: 2.82 [1.17–6.80], P = 0.021; preterm birth: 6.14 [2.30–16.40], P < 0.001) than the control group, respectively. Among the 1860 twin infants, more IVF-ET twins failed in dual-step hearing screening (3.26%) and were diagnosed with hearing loss (2.21%) than those conceived by spontaneous pregnancy. Conclusion: IVF-ET conception and premature birth were associated with a higher risk of hearing impairment. Twin infants conceived by IVF-ET tended to fail in hearing screening and be diagnosed with hearing loss. These observations provide a more comprehensive approach for the prevention and management of deafness in twin-born children. What is Known: • IVF-ET technologies conferred a higher risk of birth defects. What is New: • Premature birth and IVF-ET conception were associated with a higher risk of hearing loss among twin infants. • Twin infants conceived by IVF-ET tended to fail in hearing screening and diagnosed with hearing loss. What is Known: • IVF-ET technologies conferred a higher risk of birth defects. What is New: • Premature birth and IVF-ET conception were associated with a higher risk of hearing loss among twin infants. • Twin infants conceived by IVF-ET tended to fail in hearing screening and diagnosed with hearing loss.


Introduction
Infertility is considered a global problem, and according to the latest World Health Organization statistics, 50-80 million people are affected worldwide [1,2]. Assisted reproductive technologies (ART), including in vitro fertilizationembryo transfer (IVF-ET), intracytoplasmic sperm injection (ICSI), gamete/zygote intrafallopian transfer (GIFT/ZIFT), and cryopreservation of embryos or gametes have been used extensively in the treatment of human infertility [3]. The proportion of IVF-ET births is between 1 and 3% of the total number of newborns [4,5]. Globally, IVF-ET newborns increase by > 1 million annually [6]. The annual number of IVF-ET births in China has been > 200,000 in recent years [7]. Although genetic factors are major contributors to anomalies, IVF-ET has been deemed associated with a higher rate of congenital birth defects, implicated in the procedures of ART procedures, infertility treatment, and paternal and maternal characteristics [8,9]. The presence of long-term complications affecting the urinary, digestive, and circulatory systems in IVF-ET offspring, associated with poor quality of life and higher healthcare costs, has attracted increasing medical attention [9,10]. Thus, prospective and epidemiological studies that systematically assess the potential risk factors for pediatric conditions are necessary to establish the risks associated with ART procedures.
Hearing loss (HL) is one of the most prevalent chronic disorders in children, affecting around 1.50 per 1000 newborns. HL might cause adverse effects on speech, language, developmental, educational, and cognitive outcomes in growth of children [11,12]. Early diagnosis of infantile hearing loss and hearing rehabilitation with appropriate intervention can alleviate these harmful effects. Newborn hearing screening (HS) is widely performed worldwide and has played an important role in the early detection, diagnosis, and intervention of hearing loss [13,14]. HL is confirmed by a structural criterion after multi-step hearing screening from birth to 6 months [15,16]. Multiple hearing screening approaches may detect patients with HL, guiding hearing rehabilitation and management for children's future quality of life.
To maximize the success rate of live births using ART procedures, double-embryo transfers are routinely performed, resulting in the increasing incidence of multiple pregnancies [17]. Several findings have indicated that twin pregnancies confer a higher risk of hearing loss compared with singleton pregnancies [18]. Some data have shown that IVF-ET technologies are probably associated with the failure of hearing screening during the neonatal period [19]. In consideration of the perspectives of HL guidelines, our study examined the possible relationship between IVF-ET twins and final diagnosis of hearing loss at the age of 6 months. Here we explored the risk factors for hearing impairment in 1860 twin infants and described the differences in hearing loss in IVF-ET twins using a prospective cohort study in China.

Clinical data
The study enrolled 1888 twinborn infants born in Changsha Hospital for Maternal & Child Health Care in China from January 1, 2017, to December 31, 2020, who visited the hearing testing center for multi-step hearing evaluations after birth (Fig. 1). After the exclusion of twin births who had non-biological parents (N = 6), stillbirths (N = 12), those who were withdrawn (N = 6), or transferred (N = 4), a total of 1860 twin newborns who underwent hearing screening were recruited after approval by the institutional review board of the hospital. The study was performed in accordance with the Declaration of Helsinki for human rights. All parents of newborns included in the study signed written consent authorizing their inclusion. The medical records and clinical observations including the perinatal conditions and conception mode of each newborn were collected. Corresponding maternal data including pregnancy conditions were also included in the baseline of this study.

Hearing screening and hearing loss diagnosis
The hearing evaluation adhered to the procedure guidelines for children aged birth to 6 months; it included multi-step hearing screening and diagnosis [15,16]. Newborn hearing screening consisted of two steps. Twin neonates underwent initial hearing screening using transient evoked otoacoustic emission (TEOAE) and/or automatic auditory brainstem Fig. 1 Flowchart of participants included in the study. Abbreviations: TEOAE, transient evoked otoacoustic emission; AABR, automatic auditory brainstem evoked response; C-ABR, click-evoked auditory brainstem response; T-ABR, tone burst ABR; DPOAE, distortion product otoacoustic emission ◂ evoked response (AABR) devices at 72 h after birth. The hearing screening devices automatically determine "Pass" or "Refer" responses based on an algorithm comparing the responses detected. For those who had a "Refer" response after initial testing or who were at high risk of hearing loss, TEOAE and AABR tests were repeated prior to discharge within the ages of 30-42 days.
In follow-up, for those at high risk of hearing loss and those referred after dual-step hearing screening, audiological diagnosis was performed using click-evoked auditory brainstem response (C-ABR), tone burst ABR (T-ABR), bone conduction ABR, distortion product otoacoustic emission (DPOAE), and acoustic conductance at the age of 3 months. A second diagnosis of HL with extended hearing screenings for those diagnosed with HL within 3 months was performed again at the age of 6 months. The final hearing loss diagnosis collected in this study was confirmed after extended hearing screenings at the age of 6 months.
Hearing loss was defined as a pure-tone average (PTA) above 25 dB, measured using thresholds of 500, 1000, 2000, and 4000 Hz, in the better of two ears. HL was categorized as mild, moderate, severe, and profound, with PTA at 26-30, 31-60, 61-80, and > 80 dB, respectively. HL was also classified as conductive, sensorineural, and mixed HL using comprehensive tympanometry evaluation and multiple ABRs/ OAEs. Full definitions and criteria of HL categories are available in the Supplementary materials.

Variables
Confounding variables that could potentially be associated with HL included the following: general newborn characteristics (sex, gestational age, birth weight, conception mode (IVF-ET/natural pregnancy)) and perinatal complications (chorionic properties, hyperbilirubinemia, infection, brain injury, cardiovascular disease, mechanical ventilation duration, and neonatal intensive care unit (NICU) stay). Among them, IVF-ET in our study was defined as a conventional term of IVF-ET and its derivatives (IVF-ET/ICSI). Maternal information was also incorporated into our recruitment with general information on mothers (maternal age) and abnormal pregnancy complications including pregnancy-induced hypertension (PIH), gestational diabetes mellitus (GDM), pelvic inflammatory disease (PID), intrahepatic cholestasis of pregnancy (ICP), fetal growth restriction (FGR), maternal anemia or infection, abnormal thyroid function, uterine and placental abnormalities, amniotic fluid condition, and premature rupture of membrane (PROM). The risk factors for HL in neonates collected in our study were summarized by the Joint Committee of Infant Hearing (JCIH) in 2007 and in previous literature on HL [15]. Information on the definitions used to classify the variates at baseline is presented in the Supplementary materials.

Statistical analysis
Statistical analysis was performed using R software version 4.1.2 (R Development Core Team, Vienna, Austria). Results of the unadjusted odds ratios (ORs) with 95% confidence intervals (CIs) were calculated using generalized estimation equation (GEE) models. GEE models considering a twin pair as a cluster were recognized as appropriate regression models for twin studies [20,21]. Associations between multiple variables and HL were examined using univariate GEE analyses. We further analyzed the associations using a multivariate GEE model, adjusting for confounders previously found to be significantly associated with HL. The chi-squared test was used to compare categorical data. We considered a P value less than 0.05 (two-sided) to be statistically significant.

Prevalence of hearing loss and characteristics of twin infants
In total, 1860 live-born twin infants for whom complete information had been provided and who underwent hearing screening were included in this analysis. One hundred fortynine (8.01%) and 49 (2.63%) participants were referred after the first and second hearing screening, respectively. A total of 30 (1.61%) twin newborns who were diagnosed with HL at the age of 6 months were included in the final analyses ( Fig. 1). Table 1 containing the baseline information of our study shows that twin infants with HL are more likely to be IVF-ET twins than those without HL. Neonates with HL tended to have lower birth weights and higher rates of premature delivery than the reference group.

Risk factors for hearing loss among twin infants
Univariate GEE regression analysis identified various factors associated with the diagnosis of HL at 6 months among twin infants. The relationships between numerous maternal or obstetric characteristics and hearing impairment were analyzed in the twin cohort (  Table 3). To investigate whether the association between HL and IVF-ET twin infants was partially mediated by preterm birth or low birth weight, we added sensitivity analyses by re-running the models without adjusting for premature birth or low birth weight.     Figure 2 shows the prevalence of failure in dual-step hearing screening and diagnoses of HL among IVF-ET and natural pregnancy (control group) twin infants. The distribution of failure for HS and diagnoses of HL among IVF-ET and control group infants is shown in Supplementary Fig. 1 Fig. 2) than those in the control group. Figure 3 shows the differences in prevalence of HL subtypes, including subtypes derived from mechanism and severity in IVF-ET and control twins. Sensorineural hearing loss (SNHL), bilateral HL, and moderate HL have considerable proportions in different classifications. Of the 30 twin infants diagnosed with HL at the age of 6 months, 21 were conceived using IVF-ET. Among the IVF-ET infants with HL, 20 were diagnosed with SNHL, 12 had bilateral HL, and 12 had moderate HL. More IVF-ET twin infants had SNHL and moderate hearing loss among all subtypes than natural pregnancy infants (SNHL: IVF-ET vs. control: 95.2% vs. 66.7%; moderate HL: 57.1%, vs. 55.6%, Fig. 3).

Discussion
In our single-center cohort study, the data of 1860 twin infants were collected and the prevalence of failures in hearing screening and the diagnoses of HL were described. After univariate GEE regression analyses, we found that the presence of IVF-ET, premature birth, and low birth weight were associated with a higher risk of hearing loss. IVF-ET and prematurity showed significant correlations with hearing loss after adjustment for correlative variables. More infants in the IVF-ET group failed in hearing screening and suffered from HL than those in the natural pregnancy group. Our findings suggest that ART confers a higher risk of hearing impairment, highlighting the need for a comprehensive approach for the prevention of detrimental hearing outcomes in ART twins.

Previous studies
The prevalence of failure in dual-step hearing screening and HL diagnosis in our cohort of twin-born infants was 2.63% and 1.61% respectively, which was considerably higher than the estimates reported for the general population [22]. Meanwhile, among IVF-ET twin infants, the prevalence of HL of 2.21% was higher than with spontaneous conception. These findings were consistent with the increase in prevalence of HL in multi-gestational infants in previous prospective studies [23][24][25][26]. However, the rise of prevalence varied with different diagnostic standards and risk factors [18,19]. A higher risk of birth defects is not only manifested in multiple gestations compared with single gestations, but also in ART newborns compared with newborns with natural gestation [10,18]. However, in some literature, only abnormal infants with hearing impairment were recruited and the prevalence was compared with that in other studies rather than with that in control group in the same cohort. Approximately 4% of ICSI infants in a Turkish retrospective study failed in hearing screening tests, but no data were shown for the natural pregnancy group [18,27]. A Korean cohort study collected 159 preterm twin neonates and showed that IVF was associated with hearing screening failure among preterm twin neonates. However, there was no correlation between IVF-ET and HL diagnosis outcome in this study [19]. Our study recruited > 1000 Chinese twin infants, focused on the diagnostic outcome of HL by following the 1-3-6 (at birth-age of 3 months-6 months) structural standard, and revealed the independent risk factors for the definitive diagnosis of HL.

Possible mechanisms
The etiology of infantile hearing loss occurs with a combination of factors including genetics, the intrauterine environment, and perinatal and postnatal factors [28,29]. Genetic factors are considered to account for half of the cases and the most frequent genetic cause of HL is mutation in the gap junction genes [30]. Mutations can involve any component of the hearing pathway, in particular inner Fig. 3 Distribution of subtypes of diagnostic hearing loss in three different classifications in the IVF-ET group vs. control group. Percentages were used to identify the relative contribution of each subtype of HL among twin infants. Etiological classification: sensorineural HL, conductive HL, mix HL; location: unilateral, bilateral HL; severity: mild, moderate, severe, profound HL. Abbreviations: IVF-ET, in vitro fertilization-embryo transfer; HL, hearing loss ear homeostasis and mechano-electrical transduction [22]. It is postulated that the process of ovarian stimulation, the agents used for the procedure, culture, laboratory factors, and frozen-thawed embryo transfer, might interfere in the reprogramming, genetic stability, and epigenetic processes resulting in congenital disorders [3,4]. Moreover, ICSI also evades natural selection and allows genetically and structurally abnormal sperm to fertilize eggs and pass detrimental mutations to the children [10]. In addition to genetic factors, concomitant issues associated with the manual interventions of ART that could be explained by systemic hypoperfusion, pressure variations, and an increase in reactive oxygen species (ROS) are able to damage the inner ear [22,23].
Development of the fetal hearing system begins after 25 weeks of gestation [31]. Hearing function is not fully developed in preterm infants and is more likely to be complicated by infection, hyperbilirubinemia, asphyxia, and hypoglycemia. Premature infants are more sensitive to genetic change and ototoxic drugs, which contribute to the increased incidence of hearing impairment [19].
Congenital infections, particularly cytomegalovirus, are also a common risk factor for hearing loss due to neural damage [22]. We had hearing-impaired infants with maternal or neonatal infection in our study, but no significant associations were identified during statistical analyses. Further animal models and molecular studies are needed to elucidate the potential mechanisms underlying the increased risk of hearing impairment in IVF-ET twins.

Strengths and limitations
This was a novel single-center study of pediatric HL in Chinese twin infants. We focused on 1-3-6 guidelines for early hearing testing and intervention rather than sole hearing screening, allowing reliability and specificality in HL diagnosis, but at the expense of smaller sample size compared with other twin studies [15]. The participants of this research were mostly Chinese twinborn infants, likely to provide more conservative estimates than the wider population of multicenter study. The limited number of HL participants due to the study population might have created bias in our results. A wide range of confounding variates were introduced in this study. As the data came from observational confounders, the findings may have been altered by unobserved variates.

Conclusion
In summary, the prevalence of both failure in hearing screening and diagnostic hearing loss was higher in IVF-ET twin infants than in infants conceived naturally. IVF-ET and preterm birth were independently associated with a higher risk of hearing loss among twinborn infants. Natural conception and avoidance of prematurity would be beneficial in lowering the risk of hearing impairment. Hearing screening needs to be measured after the evaluation of extended risk factors, providing clues for innovative prediction tools for hearing impairment.