Incidence of progressive hearing loss after the baseline visit in patients who were not already showing profound hearing loss. Kaplan-Meier curves are adjusted for baseline age and initial hearing loss. MV indicates M34T or V371; NT, nontruncating; T, truncating.
Level of hearing in the better ear at the baseline visit in 118 patients with 2 GJB2 mutations based on the 4-frequency pure-tone average (PTA). MV indicates M34T or V371; NT, nontruncating; T, truncating.
Progressive hearing loss in an 8-year-old patient with 2 truncating GJB2 mutations. Hearing loss is shown in the right ear (A) and the left ear (B).
Kenna MA, Feldman HA, Neault MW, Frangulov A, Wu B, Fligor B, Rehm HL. Audiologic Phenotype and Progression in GJB2 (Connexin 26) Hearing Loss. Arch Otolaryngol Head Neck Surg. 2010;136(1):81-87. doi:10.1001/archoto.2009.202
To document the audiologic phenotype of children with biallelic GJB2 (connexin 26) mutations, and to correlate it with the genotype.
Prospective, observational study.
Tertiary care children's hospital.
Infants and children with sensorineural hearing loss (SNHL).
Sequencing of the GJB2 (connexin 26) gene.
Main Outcome Measures
Degree and progression of SNHL.
From December 1, 1998, through November 30, 2006, 126 children with biallelic GJB2 mutations were identified. Of the 30 different mutations identified, 13 (43%) were truncating and 17 (57%) were nontruncating; 62 patients had 2 truncating, 30 had 1 truncating and 1 nontruncating, and 17 had 2 nontruncating mutations. Eighty-four patients (67%) initially had measurable hearing in the mild to severe range in at least 1 of 4 frequencies (500, 1000, 2000, or 4000 Hz). Of these 84 patients with residual hearing, 47 (56%) had some degree of progressive hearing loss. Patients with 2 truncating mutations had significantly worse hearing compared with all other groups. Patients who had 1 or 2 copies of either an M34T or a V37I allele had the mildest hearing loss.
Hearing loss owing to GJB2 mutations ranges from mild to profound and is usually congenital. More than 50% of patients will experience some hearing loss progression, generally gradually but occasionally precipitously. Hearing loss severity may be influenced by genetic factors, such as the degree of preserved protein function in nontruncating mutations, whereas hearing loss progression may be dependent on factors other than the connexin 26 protein. Genetic counseling for patients with GJB2 mutations should include the variable audiologic phenotype and the possibility of progression.
Sensorineural hearing loss (SNHL) is the most common congenital sensory impairment, with an incidence of 1 to 2 per 1000 for bilateral severe to profound losses (>70 dB) and up to 4 per 1000 if mild to moderate and unilateral losses are included. The most frequently identified causes of pediatric SNHL are divided into 3 categories: infectious, anatomic, and genetic. The most common infectious cause is congenital cytomegalovirus; the most common anatomic findings are the presence of enlarged vestibular aqueducts and other inner-ear anomalies, some of which have a genetic basis; and the most common genetic causes are mutations in the gap junction β 2 gene (GJB2) (OMIM 121011) encoding the connexin 26 (Cx26) protein. In developed countries, more than 50% of SNHL cases have a genetic origin; 30% are syndromic and the remainder are nonsyndromic. Of the nonsyndromic cases, approximately 80% are autosomal recessive, 15% to 17% are autosomal dominant, 2% to 3% are X-linked, and about 1% are mitochondrial. Of all nonsyndromic autosomal recessive cases of hearing loss, as many as half are caused by GJB2 mutations.1
In 1997, GJB2 was reported as the first autosomal recessive gene implicated in nonsyndromic SNHL.2,3 This gene encodes the Cx26 protein and segregates at the DFNB1 locus on chromosome 13q12. More than 100 mutations have been described for the GJB2 gene, with most associated with recessive hearing loss.4 One mutation, 35delG, is the most common, particularly among white populations.5 This mutation results in a frameshift and premature termination of the protein. A second mutation, 167delT, has a high frequency in the Ashkenazi Jewish population,6 and a third mutation, 235delC, is frequent among Asian populations.7 In addition, there are many other GJB2 mutations, including missense and nonsense mutations, small deletions and insertions, and several mutations for which the clinical implications are unknown.4 At least 9 dominant GJB2 mutations are associated with nonsyndromic SNHL and 8 with dominant syndromic SNHL.2,4GJB2 was first identified in patients with severe to profound bilateral SNHL, and, therefore, diagnostic clinical testing was initially offered to patients with that phenotype. However, it is now well recognized that GJB2 hearing loss ranges from mild to profound,1 with some cases demonstrating incomplete penetrance or delayed onset of hearing loss.8- 11 In addition, some mutations, including M34T and V37I, seem to be associated with a milder audiologic phenotype than others.1,10,12,13 Although there have been some reports of hearing loss progression,14- 17 there are no studies, to our knowledge, that have examined many patients over long periods to reliably identify the percentage of patients whose hearing loss progresses. Although 2 recent multicenter studies, 1 international1 and 1 US-based,18 described marked variability in the presenting audiologic phenotype, progression rate was not assessed because multiple sequential audiograms were not available for most participants.
In a 2001 report15 about GJB2-related hearing loss, 4 of 19 participants with mild to severe hearing impairment had progression of their hearing loss. In this report, we describe the longitudinal audiologic features for a large cohort of pediatric patients with biallelic pathogenic GJB2 mutations, including documentation of progression of the hearing loss.
Starting December 1, 1998, children from birth to age 21 years with SNHL or mixed hearing loss who were seen in the outpatient clinics of the Department of Otolaryngology at Children's Hospital Boston were eligible for GJB2 testing. This study was approved by the hospital institutional review board.
Patients of both sexes and all races were offered testing. After April 1, 2002, all patients were also tested for the 309 kb GJB6-D13S1830 (Cx30) deletion described by del Castillo and colleagues in 2002.19 Initially, only children who had a bilateral, audiometrically profound, clinically nonsyndromic phenotype were offered GJB2 testing. Subsequently, testing was expanded to include patients with a less severe audiologic phenotype and to patients who had either other nonaudiologic clinical findings or other potential causes of the hearing loss.
GJB2 (Cx26) gene testing was performed in the Clinical Laboratory Improvement Amendments–approved Genetics Diagnostic Laboratory at Children's Hospital Boston. For all tests, genomic DNA was extracted from each patient's blood specimen according to the Gentra PureGene protocol (Qiagen Inc, Valencia, California). DNA was amplified by polymerase chain reaction (PCR) with 2 primer sets that amplified the entire open-reading frame of the GJB2 gene.20 The PCR product was then sequenced bidirectionally in 2 parts using the external PCR primers plus 2 internal sequencing primers on an ABI PRISM 377 DNA Sequencer (Applied Biosystems, Foster City, California). The 309-kb GJB6-D13S1830 deletion was detected by PCR, as previously published.20 This deletion has been found in some participants who were double heterozygous for both the Cx30 deletion and a GJB2 mutation or homozygous for the 309-kb Cx30 deletion.19 More recently, a 232-kb GJB6-D13S1854 deletion has been described21; however, in the present study, we did not look for this deletion.
All but 1 mutation identified had been previously reported. The exception is the 453_460del8ins9 mutation, which is presumed to be pathogenic because of its predicted truncation of the Cx26 protein. Although most variants in this study are generally accepted to be pathogenic, M34T and V37I have been reported to be both pathogenic and benign. We present data that support the more recent reports that these variants are pathogenic but result in a milder audiologic phenotype.1,10,13 All mutations were classified as truncating or nontruncating mutations. Truncating mutations include nonsense mutations, deletions and insertions that introduce a shift in the reading frame, or mutations predicted to affect translation initiation (eg, M1V). Nontruncating mutations contain amino acid substitutions. These groupings were made to differentiate mutations based on predicted severity, with truncating mutations generally resulting in transcript or protein degradation or a protein product that has either no, or very abnormal, function and therefore would be expected to result in a more severe hearing loss compared with nontruncating mutations. However, it is possible that some missense mutations could also lead to a protein with complete loss of function.
Genetic counseling was provided through the Division of Genetics, Children's Hospital Boston, and, since February 2005, by a genetic counselor in the hospital's Department of Otolaryngology. Genetic counseling was offered to all patients before and after genetic testing.
All audiometric testing was performed in the Department of Audiology using age-appropriate techniques (visual reinforcement audiometry, conditioned-play audiometry, and/or conventional “hand-raising” audiometry) by audiologists seasoned in pediatric audiometric testing. Testing also included auditory brainstem-evoked response (ABR) testing in newborns, infants, and young children, otoacoustic emission testing to further confirm and characterize the hearing loss, and behavioral and frequency-specific testing in children who were old enough to participate. A combination of behavioral and electrophysiologic audiometric tests was often used to confirm the diagnosis of hearing loss. Hearing loss was categorized as mild (21-40 dB hearing level [dBHL]), moderate (41-55 dBHL), moderately severe (56-70 dBHL), severe (71-90 dBHL), or profound (>90 dBHL). Hearing loss was also classified as conductive, sensorineural, or mixed. The severity of hearing was noted for each ear separately.
Determination of progression of hearing loss or fluctuation of hearing was based on pure-tone behavioral audiograms. The first ear-specific pure-tone audiogram that was judged by the audiologist as having “good” reliability was compared with the most recently available audiogram with good reliability for each patient. Audiograms for patients for whom reliability of responses were in question were not included in the analysis of progression or fluctuation. According to clinical protocols, children with SNHL were seen every 3 months following new identification of SNHL or mixed hearing loss, so serial audiograms and/or electrophysiologic results were obtained for all patients. A 4-frequency pure-tone average (PTA) (500, 1000, 2000, and 4000 Hz) was calculated for each ear for each audiogram. Progression or fluctuation was documented if thresholds had worsened (or improved in the case of fluctuation) by 10 dB or more at 2 or more frequencies in the same ear and confirmed on retest, or by 15 dB at 1 or more frequencies in 1 ear compared with the prior audiogram and confirmed on retest, and provided that middle-ear immittance measures or pneumatic otoscopic testing did not suggest middle-ear fluid or significant negative middle-ear pressure. Similarly, definitive worsening or fluctuation of hearing was noted if the 4-frequency PTA decreased or improved by 10 dB. In patients younger than 3 years, if hearing loss progression or fluctuation was suspected based on behavioral testing, but the results were in question, ABR and otoacoustic emission were used for confirmation of the change. These definitions for progression (or fluctuation) of hearing loss were based on known reliability of pure-tone audiometry using the different test techniques. Conventional audiometry using the modified Hughson-Westlake procedure (as is standard audiologic practice and is used at Children's Hospital Boston) is known to have a test-retest reliability of 5 dB,22 whereas visual reinforcement audiometry is known to provide reliable responses in 90% of infants and toddlers.23 In addition, longitudinal studies of hearing thresholds in hearing-impaired infants with stable hearing loss showed no significant difference in hearing threshold levels across behavioral audiometric techniques,24 and retest reliability in hearing-impaired infants younger than 12 months has been shown to be better than 10 dB.25 For a child to be identified as having a significant change in hearing, thresholds must have changed by more than the known test-retest reliability of behavioral audiometry and confirmed on retest. Furthermore, averaging across frequencies eliminated the possibility that a child would be labeled as having a change in hearing based on a subtle change at a single frequency. Finally, including only those audiograms labeled as having “good” reliability decreases the likelihood of a child being misdiagnosed with progressive hearing loss based on less than optimal audiometric results.
These guidelines for progression were applied to the frequencies from 500 through 8000 Hz. Data for the 250-Hz frequency was not used to calculate progression or PTA because it can be difficult to separate vibrotactile from auditory thresholds at this frequency. In addition, we did not use any behavioral data that were not ear specific, which meant that no sound field data were used. However, if sound field behavioral data suggested a significant worsening of hearing, additional frequency-specific and ear-specific ABR testing was often performed in young children to rule out the possibility that the suspected worsening of hearing was caused by poor reliability of behavioral testing. This ABR data could then be compared with prior ABR data, and to support behavioral thresholds going forward.
The degree of baseline hearing loss was compared among genotypes by 1-way analysis of variance, with the significance level for pairwise comparisons adjusted by the Tukey-Kramer method. The time course of progressive hearing loss was compared across genotypes by the log-rank χ2 statistic for homogeneity of product-limit estimates (Kaplan-Meier curves), with baseline age and hearing level as covariates. All computations were carried out with SAS statistical software, version 9.2 (SAS Institute, Cary, North Carolina).
Between December 1, 1998, and November 30, 2006, a total of 126 patients with pathogenic biallelic Cx26 mutations were identified (Table 1). There were 50 boys (40%) and 76 girls (60%), with ages ranging from birth to 21 years; 104 were white, 17 Asian, 2 African American and white, 2 Asian and white, and 1 of unknown race. Ten participants were Hispanic (all white). The sample included 11 sets of 2 siblings each and 1 set of identical twins. A total of 30 mutations were identified and are listed, with their frequencies, in Table 1. Only 1 mutation has not been reported, 453_460del8ins9, which is presumed to be pathogenic owing to its predicted truncation of the Cx26 protein.
There were 62 patients with homozygous mutations, of which 43 were 35delG/35delG, and 64 who were compound heterozygous. The other homozygous mutations were 167delT/167delT (3 patients), 235delC/235delTC (2 patients), M34T/M34T (3 patients), V37I/V37I (9 patients), 313_326del14/313_326del14 (1 patient), and V84L/V84L (1 patient). Ninety-four of 126 patients (75%) had at least 1 35delG mutation. Of the 30 types of GJB2 mutations among 126 patients, 13 (43%) were truncating and 17 (57%) were nontruncating (Table 2).
Of the 126 patients, 42 (33%) could not be evaluated for progression because they presented with bilateral profound SNHL at all frequencies (n = 25) or had insufficient follow-up data to evaluate progression (n = 17). Eighty-four (67%) had measurable hearing in at least 1 of 4 frequencies (500, 1000, 2000, or 4000 Hz) in the mild to severe range on their initial audiogram. One patient with 35delG/35delG had normal hearing documented by ABR at birth but progressed to a mild sloping to moderate SNHL by age 15 months. Table 3 compares the initial hearing loss in the better ear among patients with either 2 truncating mutations, 1 truncating and 1 nontruncating, or 2 nontruncating mutations. Patients with the nontruncating M34T or V37I mutations were treated as a separate group. Patients with 2 truncating mutations had significantly worse hearing when compared with all other groups (Figure 1). Conversely, the patients who had 1 or 2 copies of either an M34T or a V37I allele had the mildest hearing loss when compared with patients who had either truncating/truncating or the truncating/nontruncating mutations. Thresholds in the left and right ear were compared similarly by genotype, with similar results (data not shown); accordingly, the better ear was used for all other comparisons.
Patients were considered to have progression of hearing loss if thresholds had worsened by 10 dB or more at 2 or more frequencies in the same ear or by 15 dB at 1 or more frequencies in 1 ear compared with the prior audiogram. Table 4 compares the patients who did and did not have progressive hearing loss. Of the original 126 patients, 84 (67%) had enough residual hearing in at least 1 of 4 frequencies (500, 1000, 2000, or 4000 Hz) to allow for calculation of progression. To calculate progression, the patient had to have at least 2 audiograms from which an accurate PTA could be calculated; the median number of visits (beyond baseline) from which PTA could be calculated was 4 (range, 1-18). Of the 84 patients, 47 (56%) had some degree of progressive hearing loss. The median time to the onset of progressive hearing loss was 13 months after initial presentation, but there was great variability across the group (range, 1-110 months). Although in most cases the hearing loss progressed fairly gradually, there was 1 case of an 8-year-old patient with 2 truncating GJB2 mutations (35delG/299_300delAT) who had rapid progression of hearing loss from mild to profound over several months. Although audiograms that preceded clinical evidence of hearing loss were not available, the patient had completely normal speech and language and above-average academic achievement, suggesting that until shortly before we saw the child the hearing had been normal or only mildly impaired (Figure 2). None of the demographic or genetic classifications shown in Table 4 differed significantly between those with and without progressive loss.
Figure 3 uses Kaplan-Meier curves to show the development of hearing loss progression after the baseline audiologic evaluation. In contrast to the initial hearing loss, the incidence of progression did not differ significantly according to genotype (log-rank χ2 = 2.28; P = .51). In addition, neither baseline age (P = .33) nor initial level of hearing loss (P = .77) was significantly associated with time to onset of progressive loss.
GJB2 is the most common recessive genetic cause of SNHL. GJB2 mutations were first identified in individuals with profound bilateral SNHL, and, therefore, diagnostic clinical testing was only offered to patients with that audiologic phenotype.2,3 Most initial reports described the audiologic phenotype as moderate to profound and/or were based on single audiograms so that progression was difficult to judge.2,3,26 In a 2001 report15 of GJB2-related hearing loss, 42 children aged 1 week to 16 years who had SNHL and biallelic GJB2 mutations were described. Of the 42, 23 (55%) were congenitally deaf and 19 (45%) were hearing impaired. Four of the 19 hearing-impaired children had progression of their hearing loss, with dramatic progression to bilateral profound hearing loss in 2 patients and a much milder progression in 2 others. Of the 23 congenitally deaf children, 18 (78%) were biallelic for 35delG and/or 167delT mutations. In contrast, of the 19 hearing-impaired children, only 3 (16%) were biallelic for 35delG and/or 167delT. These results suggest that the audiologic phenotype varied with the genotype, with some genotypes being more likely to present with a severe phenotype than others, and that progression of the hearing loss was more common than originally suspected.
The present report of 126 patients includes these initial 42 patients. Of the 126, 42 patients (33%) had bilateral profound SNHL when initially identified, most of them as infants. Eighty-four (67%) presented with mild to severe SNHL. Of these 84, 47 (56%) have had progression of their hearing loss. Three of the 47 patients who have had progression have other diagnoses that may have contributed to their hearing loss. One had congenital cytomegalovirus (235delC/235delC), and a second (35delG/M34T) has cystic fibrosis and has received many courses of systemic and inhaled aminoglycosides; this second patient has been tested for 1555A→G and 1494C→T mutations in the 12SrRNA mitochondrial gene, and results were negative. The third patient (35delG/35delG) had congenital syphilis and was treated effectively in infancy. Two of these 3 have progressed to bilateral profound hearing loss and have received cochlear implants. For all 3 of these patients, the medical diagnosis was thought to be the cause of their hearing loss before GJB2 testing was performed. In all 3 cases, however, the degree or configuration of the hearing loss was not believed to be consistent with the medical diagnosis, so genetic testing was pursued and the results are included in the final analysis of this article. Although it is impossible to say with 100% certainty that these 3 patients had progression of SNHL owing to GJB2 mutations, it seems probable; the patient with syphilis was definitively treated in infancy, and the patient with cystic fibrosis did not have the sharply downward sloping audiogram of most patients with aminoglycoside ototoxicity. The patient with congenital cytomegalovirus had a very symmetrical hearing loss, which is less characteristic of cytomegalovirus and more so of GJB2. An additional important point for these patients is that they have a definite genetic cause of hearing loss, which previously would not have been considered given their medical diagnoses.
The degree of severity of the hearing loss at initial presentation is statistically related to the type of mutation. Patients who had 1 or 2 nontruncating mutations had a milder audiologic phenotype than those who had 2 truncating mutations (Figure 1). In addition, patients who had at least 1 M34T or V37I mutation had less severe hearing loss than those patients with truncating or nontruncating mutations not involving M34T or V37I. These results support our initial findings, as well as those of others.1,14,15 The 2005 report by Snoeckx et al1 represents a 1531-participant, multicenter, international study of GJB2 audiologic phenotype-genotype in which our research group participated. In this study, 79% of patients had mild to moderate hearing loss. The current data, as well as the larger aggregate data in the Snoeckx et al study, demonstrate that the degree of hearing loss was generally more severe in the patients with 2 truncating mutations compared with those with 2 nontruncating mutations or 1 of each. Furthermore, this larger multicenter study also found milder hearing loss in those with at least 1 M34T or V37I mutation.
The mechanism of the varying phenotype (including severity and presence or absence of progression) in patients with GJB2 is unclear. In 1995, Kikuchi and colleagues27 showed immunochemical localization of Cx26 to 2 groups of cells in the cochlea: nonsensory epithelial cells and connective tissue cells. Gap junction channels, which in the cochlea are comprised of Cx26 and other connexin proteins, are thought to help maintain the endocochlear potential, which is essential for hair cell excitation and function, by facilitating passage of potassium ions between cells. A 2002 study28 using a targeted knockout mouse model of Cx26 showed normal development of the inner ear with cell death occurring soon after the onset of hearing. These investigators suggest that Cx26 deficiency may lead to an increase in extracellular potassium, which in turn may cause death of supporting cells through oxidative stress. However, it is now understood that different mutations have varying effects on the function of gap junctions, with some mutations leading to protein absence and others leading to expressed proteins with altered function. These mutation-specific differences as well as differences in other genetic and environmental modifiers are likely to explain the variability in the severity and progression of hearing loss.
In the present study and in the one by Snoeckx et al,1 truncating mutations included nonsense mutations as well as deletions, insertions, and duplications that introduced a shift in the reading frame. The nontruncating mutations included amino acid substitutions. For the truncating mutations, the protein product is generally not made, is quickly degraded, or is extremely abnormal. For the nontruncating mutations, it is possible that a protein product with partial functional activity may be made. Therefore, the difference in the presenting audiogram between truncating and nontruncating mutations makes sense. However, there was no statistical difference between truncating and nontruncating groups in terms of incidence of progression of the hearing loss. In addition, progression could not be linked to any 1 particular mutation. Possible explanations include as yet unidentified modifiers to the genotype, including modifier genes or environmental factors. Length of follow-up may also be a factor; it is possible that if all patients with biallelic GJB2 mutations are followed up long enough, they will exhibit progression. Even this thought, however, does not explain the variability in the rate of progression.
Mutations in GJB2 are the most common cause of recessive hearing loss. The hearing losses range from mild to profound and are usually congenital. Of the two-thirds of our patients who presented with mild to severe SNHL owing to biallelic GJB2 mutations, 56% experienced progressive hearing loss. In most cases, this was a progressive loss that occurred over a period of years; however, 1 patient had precipitous bilateral hearing loss that progressed from mild to profound in a period of months. The exact causes of the varying audiologic phenotype and audiologic progression remain unclear. Hearing-loss severity may be influenced by genetic factors, such as the degree of preserved protein function in nontruncating mutations, whereas hearing-loss progression may be dependent on factors other than the Cx26 protein. Regardless of the mechanism, genetic counseling for patients with GJB2 mutations needs to include the variability of the audiologic phenotype as well as the possibility of progression.
Correspondence: Margaret A. Kenna, MD, MPH, Department of Otolaryngology and Communication Enhancement, Children's Hospital Boston, 300 Longwood Ave, LO-367, Boston, MA 02115 (firstname.lastname@example.org).
Submitted for Publication: March 9, 2009; final revision received June 9, 2009; accepted July 21, 2009.
Author Contributions: All authors had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Kenna, Feldman, Fligor, and Rehm. Acquisition of data: Kenna, Neault, Frangulov, Wu, Fligor, and Rehm. Analysis and interpretation of data: Kenna, Feldman, Neault, Wu, Fligor, and Rehm. Drafting of the manuscript: Kenna, Feldman, Fligor, and Rehm. Critical revision of the manuscript for important intellectual content: Kenna, Feldman, Neault, Frangulov, Wu, Fligor, and Rehm. Statistical analysis: Feldman. Administrative, technical, and material support: Kenna, Neault, Frangulov, Fligor, and Rehm. Study supervision: Kenna, Fligor, and Rehm.
Financial Disclosure: None reported.
Funding/Support: This study was supported by grant NIDCD R01 DC05248 from the National Institute for Deafness and Other Communication Disorders (Dr Kenna).
Additional Contributions: Jessica Guidi, BS, provided significant help with manuscript preparation.