Association between the p.V37I variant of GJB2 and hearing loss: a pedigree and meta-analysis

Pathogenic variants in the gap junction protein beta-2 (GJB2) gene are the most common cause of hearing loss. Of these, the p.V37I variant of GJB2 has a high allele frequency (up to 10%) in East Asians. Characterization of the phenotypic spectrum associated with p.V37I, as well as the role of this variant in the onset of hearing loss could have a remarkable effect on future diagnostic strategies. Here, we performed a pedigree analysis of unrelated families exhibiting various hearing phenotypes, and then conducted a meta-analysis to comprehensively assess the association between the p.V37I and the risk of hearing loss. Pedigree analyses showed that both homozygous p.V37I variants, as well as compound heterozygous p.V37I with other GJB2 pathogenic variants, contributed to various phenotypes of hearing loss. Meanwhile, meta-analysis demonstrated that, compared with those in the wild type group, both p.V37I homozygotes and compound heterozygous p.V37I variants were at significantly higher risk of developing hearing loss (odds ratios = 7.14 and 3.63; 95% confidence intervals = 3.01-16.95 and 1.38–9.54, respectively). Conversely, heterozygous p.V37I variants alone did not increase the risk of hearing loss. Given the high allele carriage rate of p.V37I (up to 10%) within the general population, our work not only provides information that might influence future genetic screening policies, but also offers insight into clinical risk evaluation and genetic counseling regarding hearing loss.


INTRODUCTION
Hearing loss (HL) is the most sensory defect that affects 1-3 in every 1,000 newborns worldwide, and half of these cases are attributed to genetic factors [1]. Notably, while a large number of HL-related genes have been identified, the gap junction protein beta 2 (GJB2) gene accounts for nearly 20% of all cases of HL, as well as 50% of autosomal recessive non-syndromic HL, in many populations [2,3]. GJB2 encodes the connexin 26 protein, which comprises a critical component of cochlear gap junctions, and is important to cell communication. To date, greater than 300 variants within GJB2 have been found to be associated with HL (http://www.hgmd.cf.ac.uk/ac/), including c.35delG, c.235delC, and c.176_191del16.
Given the high allelic frequency of the p.V37I variant (up to 10%), it is estimated that greater than five million East Asians suffer from HL due to homozygous or compound heterozygous carriage of this allele [6]. It is therefore imperative to evaluate the risks associated with p.V37I for clinical genetic counseling and public health assessment purposes. In this study, we performed a pedigree analysis of families with probands exhibiting various HL phenotypes, and then conducted a metaanalysis of sporadic HL to comprehensively evaluate the role of p.V37I in the risk of HL.

Pedigree analyses
Seven families carrying the p.V37I variant were included for pedigree analyses (Table 1 and Figure 1). There were five male and two female probands. Six probands were children (7 months-9 years) and one proband was an adult (33 years). These probands exhibited both congenital and delayed-onset non-syndromic HL, and one member had sudden deafness. Bilateral and unilateral non-syndromic HL were also observed, with HL degrees of mild to moderate. Families 1-5 had probands that were homozygous for p.V37I (S1-S5), while the probands of Families 6 and 7 had compound heterozygous p.V37I variants (H1-H2). In this study, compound heterozygous p.V37I variants were defined as the p.V37I allele in a trans configuration with another pathogenic mutant allele of GJB2 gene. Pedigree analyses revealed that p.V37I was transmitted from heterozygous parents to their children, who suffered HL if he/she inherited two affected alleles (p.V37I homozygotes or compound heterozygous p.V37I variants). Meanwhile, siblings that inherited one affected allele retained normal hearing. These analyses strongly suggest that p.V37I increases the risk of HL through an autosomal recessive inheritance pattern.
Furthermore, we evaluated the association between compound heterozygous p.V37I variants on HL risk. Ten eligible studies comprising of 6,762 cases and 4,211 controls were included (Table 3). Notably, our results suggest that people harboring compound heterozygous p.V37I variants have a 3.63-fold higher risk of HL than those without these variants ( Figure 3; OR = 3.63; 95% CI = 1.38-9.54; P heterogeneity = 0.060; I 2 = 44.9%). These results were strongly supported by sensitivity analyses (Supplementary Figure 3), and no publication bias was found (P = 0.152).

DISCUSSION
In this study, we demonstrated that bi-allelic or compound heterozygous p.V37I variants are associated with increased risk of various HL phenotypes, and quantified the risk associated with these variants and development of HL.
Our pedigree analyses indicated that p.V37I can cause HL as either a homozygous variant or as compound heterozygous with other pathogenic variants in GJB2. Functional studies performed in cells and mouse models support this conclusion [40][41][42]. In consistent with previous reports [6,[12][13][14], we observed that the HL       phenotypes associated with p.V37I varied by onset type, disease site, and degree of HL, implying that other causes, especially environmental factors, influence p.V37Imediated onset of HL. Notably, we also found that homozygous p.V37I might give rise to sudden deafness in children. Indeed, similar results were reported in a recent study [39]. As such, p.V37I might be considered as a potential cause of sudden onset of deafness in future cases.
To quantify the pathogenic association between p.V37I and HL risk, we performed a meta-analysis. In keeping with previous cohort studies and functional experiments, our results supported the conclusion that the p.V37I variant significantly increases an individual's risk of developing HL. Compared with wild type individuals, the p.V37I homozygote group, but not the heterozygote group, showed a significantly greater likelihood of developing HL. This could explain why p.V37I heterozygotes are prevalent among the general healthy population, while bi-allelic p.V37I variants are found predominantly in patients with HL. Notably, our results also indicate that the compound heterozygous p.V37I variants are associated with a nearly four-fold greater risk of developing HL, compared to wild type individuals. In view of the high prevalence of p.V37I and the large number of other GJB2 pathogenic variants, our findings indicate that previous reports have likely underestimated HL risk in human populations. Compared with the previous meta-analysis [16], our work included more eligible studies and quantified the risky effects of p.V37I on HL in more detail.
The major strengths of our work were the variable phenotypic spectrum associated with p.V37I and that a large population (23,097 participants) was used to evaluate the association between this variant and HL risk. However, our results should be interpreted with caution. First, only commonly known HL-related variants (variants within the GJB2, SLC26A4, 12S rRNA, and GJB3 coding regions) were screened in our pedigree analysis. Second, the number of studies regarding compound heterozygous p.V37I variants evaluated herein was insufficient to further explore the concrete effects of particular compound variant types on HL risk. Third, although both our study and previous reports found that p.V37I is associated with various HL phenotypes, the specific mechanism by which this variant promotes HL, such as interactions between p.V37I and other genetic or environmental factors, remains unclear. Further studies are therefore needed to address these issues.
In summary, our work strongly suggests a pathogenic role for p.V37I in various HL phenotypes and provides a quantitative assessment of the risk associated with carriage of this variant and development of HL. Considering the high carriage rate of p.V37I within the general population, these findings provide compelling information that should influence future genetic screening policy and that offer insights into clinical risk evaluation and genetic counseling.

Study participants and ethical statement
For this study, we recruited seven unrelated probands with non-syndromic, sensorineural hearing loss, and their family members. Each participant provided written informed consent. For underage participants, written informed consent was obtained from their parents. Our study was approved by the Medical Ethics Committee of Tongji Hospital, Tongji Medical College, Huazhong University of Science and Technology. All procedures were performed according to the ethical guidelines for human subjects research.

Clinical and audiometric evaluation
All participants were subjected to physical and neurological examinations to exclude syndromic deafness. Each participant's level of hearing was assessed via a comprehensive auditory evaluation, such as by otoscope examination, tympanometry, or pure-tone audiometry (PTA). For very young participants, auditory brainstem responses (ABR) and/or auditory steady-state responses (ASSR) were recorded. Degree of hearing loss was estimated as the average hearing levels at 0.5, 1.0, 2.0, and 4.0 kHz for the better ear. Severity of hearing loss was categorized as normal (<25 dB), mild (26-40 dB), moderate (41-70 dB), severe (71-95 dB), or profound ( >95 dB). In addition, we defined sudden deafness as over 30 dB of sensorineural hearing loss involving at least three frequencies occurring less than 72 hours [43].

DNA extraction and mutation analysis
Genomic DNA was extracted from anticoagulant peripheral blood by the QIAamp DNA blood mini kit (Qiagen, Germany). For mutation screening, the entire coding region and flanking sequences of four commonly mutated genes, GJB2, SLC26A4, 12S rRNA, and GJB3, were polymerase chain reaction (PCR) amplified and then subjected to bidirectional sequencing with an ABI 3500 DNA sequencer (Applied Biosystems). The primers and PCR conditions used for these analyses are described in detail in Supplementary Table 1. Sanger sequencing results of pedigree analyses were shown in Supplementary Figure 4.

Meta-analysis on sporadic HL
We further performed a meta-analysis to determine the detrimental effects of p.V37I on sporadic HL, according to the guidelines of Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) [44]. A structured literature search was conducted of articles published through November 2016 using the PubMed, Web of Science, and EMBASE databases and the following search items: "GJB2 OR connexin 26", "polymorphism OR variant OR mutation", "hearing loss OR deafness", and "case-control OR cohort OR population". Only English language articles were included. The inclusion criteria were: (i) studies that investigated the association between p.V37I and HL under a case-control or cohort design; (ii) studies that reported genotypes or allelic data for p.V37I to estimate OR and 95%CI. In cases of overlapping populations, the study with the largest sample size was included. We used OR as the effect measure and combined data using a randomeffects model. Hardy-Weinberg equilibrium (HWE) in the controls was checked by χ 2 test. Meanwhile, Cochran χ 2 test and I 2 values were used to evaluate the heterogeneity between studies. Sensitivity analyses and publication bias assessment were also conducted. All statistical analyses were performed using Stata 12.1 software (StataCorp, College Station, TX, USA), and P ≤ 0.05 was considered significant for all tests.

Author contributions
A.L. and Y.L. conceived and designed this study. N.S., A.L. and Y.L. recruited the study subjects and collected their characteristics and examination results. Y.Z. and W.L. extracted DNA. N.S. and J.P. conducted mutation analyses. J.P. and X.W. performed literature search, study selection and data extraction. X.W. and Y.Z. conducted statistical analysis. J.P. and N.S. prepared tables and figures. N.S., A.L. and Y.L. wrote the manuscript. All these authors completely consented with all the data in the study and approved the final manuscript. A.L. and Y.L. had the primary responsibility for final manuscript.

CONFLICTS OF INTEREST
The authors declare no competing financial interests.