Hearing loss in 3 classes of GJB2-GJB6 genotypes. Inactivating (I) mutations were frameshift and nonsense mutations, insertions, and deletions, with a disrupted open reading frame. Noninactivating (NI) mutations were missense mutations, deletions, or insertions of 3 or a multiple of 3 nucleotides. A significant difference (P = .001) was observed between the 2 genotype groups using the 2-tailed Fisher exact test. A reduced contingency table combines in one group profound and severe deafness and in another group moderate and mild hearing loss. The reference groups were those with two I mutations.
Evolution of the hearing defect in GJB2-GJB6 genotype groups. Inactivating (I) mutations were frameshift and nonsense mutations, insertions, and deletions, with a disrupted open reading frame. Noninactivating (NI) mutations were insertions and deletions without a disrupted open reading frame and missense mutations. The hearing loss evolution was different in patients with two I mutations and in those with two NI mutations or two I and one NI mutation: P = .01, 2-tailed Fisher exact test. The reference groups were those with two I mutations.
Marlin S, Feldmann D, Blons H, Loundon N, Rouillon I, Albert S, Chauvin P, Garabédian E, Couderc R, Odent S, Joannard A, Schmerber S, Delobel B, Leman J, Journel H, Catros H, Lemarechal C, Dollfus H, Eliot M, Delaunoy J, David A, Calais C, Drouin-Garraud V, Obstoy M, Goizet C, Duriez F, Fellmann F, Hélias J, Vigneron J, Montaut B, Matin-Coignard D, Faivre L, Baumann C, Lewin P, Petit C, Denoyelle F. GJB2 and GJB6 MutationsGenotypic and Phenotypic Correlations in a Large Cohort of Hearing-Impaired Patients. Arch Otolaryngol Head Neck Surg. 2005;131(6):481-487. doi:10.1001/archotol.131.6.481
To analyze the clinical features of hearing impairment and to search for correlations with the genotype in patients with DFNB1.
Collaborative study in referral centers, institutional practice.
A total of 256 hearing-impaired patients selected on the basis of the presence of biallelic mutations in GJB2 or the association of 1 GJB2 mutation with the GJB6 deletion (GJB6-D13S1830)del.
Main Outcome Measures
The prevalence of GJB2 mutations and the GJB6 deletion and audiometric phenotypes related to the most frequent genotypes.
Twenty-nine different GJB2 mutations were identified. Allelic frequency of 35delG was 69%, and the other common mutations, 313del14, E47X, Q57X, and L90P, accounted for 2.6% to 2.9% of the variants. Concerning GJB6, (GJB6-D13S1830)del accounted for 5% of all mutated alleles and was observed in 25 of 93 compound heterozygous patients. Three novel GJB2 mutations, 355del9, V95M, and 573delCA, were identified. Hearing impairment was frequently less severe in compound heterozygotes 35delG/L90P and 35delG/N206S than in 35delG homozygotes. Moderate or mild hearing impairment was more frequent in patients with 1 or 2 noninactivating mutations than in patients with 2 inactivating mutations. Of 93 patients, hearing loss was stable in 73, progressive in 21, and fluctuant in 2. Progressive hearing loss was more frequent in patients with 1 or 2 noninactivating mutations than in those with 2 inactivating mutations. In 49 families, hearing loss was compared between siblings with similar genotypes, and variability in terms of severity was found in 18 families (37%).
Genotype may affect deafness severity, but environmental and other genetic factors may also modulate the severity and evolution of GJB2-GJB6 deafness.
Hearing loss is the most frequent sensory deficit. One child in 1000 is born with severe or profound deafness, and another 1 in 1000 becomes deaf before adulthood. Recent work in human genetics indicates that 60% to 80% of cases of congenital deafness are of genetic origin in developed countries1 and that nonsyndromic sensorineural hearing loss (NSSNHL) is mainly autosomal recessive. Since 1994, more than 80 loci for NSSNHL have been identified, and 35 different genes have been cloned.1 Despite extensive genetic heterogeneity, a single locus on chromosome band 13q12 has been recognized to account for a large proportion of cases of nonsyndromic autosomal recessive deafness (DFNB1). Furthermore, the gene that encodes connexin 26, GJB2, has been identified as causative in DFNB1.1 The GJB2 gene is also involved in an autosomal dominant form of deafness, DFNA3, and in syndromes such as Vohwinkel or keratitis-ichtyosis-deafness syndrome. The GJB6 gene, which encodes connexin 30, maps close to GJB2. The GJB6 gene was first described as causative in a rare dominant form of deafness, DFNA3, and its implication in NSSNHL was ascertained through the identification of a large deletion that includes the GJB6 5′-end noncoding region and most of the coding region. This genetic alteration was found in patients with hearing impairment who carry a GJB2 mutation in trans and at a homozygous state in some families.2- 4 In those patients, it is still not clear whether the deafness is due to a digenic pattern of inheritance or whether the deletion removes common GJB6 and GJB2 regulatory elements.5
More than 70 different GJB2 mutations have been reported in the connexin-deafness database6 for recessive deafness (DFNB1), and 6 have been associated with a dominant form of inheritance (DFNA3). Three variants are of major importance in terms of frequency. In white populations, GJB2 35delG was found to be the most frequent mutation in hearing-impaired children.7- 10 The 2 other mutations, 167delT and 235delC, are the most common pathogenic alleles in Ashkenazi Jews and Asians, respectively,11- 13 2 common deletions with carrier frequencies of 4% and 1%, respectively.12,13 The pathogenic role of GJB2 mutations has been clearly established for many variants but remains controversial for missense mutations such as M34T, V37I, and R127H. The M34T variation was first described as a dominant mutation,14 but the description of normal-hearing carriers abolished this hypothesis.15 Furthermore, the description of normal-hearing individuals with compound heterozygotes carrying an M34T allele and a GJB2 mutated allele in trans raised the possibility that M34T was a nonpathogenic variant in vivo.16- 18 Furthermore, the M34T allele has been observed in the general population with a high frequency of 1.2% to 1.5%.15,19 This high allele frequency was confirmed in a large study20 of the general population of the South of France (81 of 7032 chromosomes) and in a study18 of normal-hearing individuals (4 of 232 chromosomes). Because M34T allele frequency is not significantly higher in patients with hearing loss than in the general population, this suggests that M34T is a common polymorphism.18- 20 However, in vitro functional studies provide evidence that M34T could have an effect on deafness.1 Thus, some researchers supported the hypothesis that the phenotypic consequences of the M34T allele depend on the GJB2 mutated allele in trans or on the genetic background of the patients.15,21 Similarly, V37I and R127H mutations have also been described in patients with nonsyndromic sensorineural hearing loss, and in vitro studies have suggested that these variations could act as recessive mutations.1 However, there is no significant increase in V37I and R127H allele frequency in deaf populations compared with general populations.15,20 Furthermore, a high V37I allele frequency was recently observed in the general population of Taiwan (8.5%; 35 of 410 chromosomes).22 Also, a normal-hearing individual with 35delG/R127H was previously described.17 These observations suggest that M34T, V37I, and R127H could have no pathogenic effects.
Nonsyndromic autosomal recessive deafness (DFNB1) has been reported previously.10,23- 25 It is characterized by a prelingual onset. The severity of the deafness varies from mild to profound and may even vary among siblings. Hearing loss is generally stable but is occasionally progressive or fluctuant. Audiometric curves are either flat or sloping, and hearing loss affects all frequencies. Finally, high-resolution temporal bone computed tomography shows no inner ear malformation. Understanding the underlying causes of the variability in DFNB1 deafness is of major importance in terms of genetic counseling. One of these causes may be the GJB2-GJB6 genotype. Therefore, the first step is to study genotype-phenotype correlations, and a large series of patients is necessary to validate these correlations. Only 2 studies have been conducted to date, 1 in a small cohort of 31 patients with biallelic mutations26 and the second in a series of 277 patients of various geographic origins.27 These previous studies have shown, first, that homozygotes for truncating mutations are more likely to express more severe hearing loss than other genotypes and, second, that some rare genotypes could induce a less severe phenotype compared with 35delG homozygotes.
In this article, we report on the genotypes and phenotypes of 256 hearing-impaired patients with biallelic GJB2 or GJB6 alterations analyzed and documented between January 1, 1995, and March 1, 2004, following homogenous guidelines through a prospective research collaborative study conducted in France.
Clinical data and samples were obtained from a prospective collection compiled between January 1, 1995, and March 1, 2004, through a broad national program on deafness. A standardized database was developed using a software program (FileMaker Pro 5; FileMaker Inc, Santa Clara, Calif). Clinical features included age at onset; hearing thresholds; audiometric configuration; evolution within 2-, 5-, and 10-year periods; balance symptoms and tests; pedigree; inner ear computed tomographic scans; and genetic assessment. All data were reviewed by a clinical geneticist (S.M.) who specializes in the treatment of patients with hearing impairments. A subgroup of 256 patients with biallelic alterations in GJB2 or GJB6 as described in the introduction was extracted from the database for the following study. Hearing loss was sporadic in 158 families and familial (2 affected siblings in the same family) in 49 families. After medical questioning to determine age at deafness onset and to exclude environmental causes, individuals with hearing impairments underwent an otoscopic examination of the ear and a nose, throat, and general examination, with routine assessment for signs suggestive of syndromic deafness performed by a clinical geneticist. Participants also underwent an ophthalmologic evaluation, including funduscopy and investigation for hematuria and proteinuria. An electrocardiogram was recorded. High-resolution temporal bone computed tomography was performed. Hearing-impaired children and their parents underwent pure-tone audiometry with a diagnostic audiometer in a soundproof room, with recording of pure-tone air and bone conduction thresholds. Air conduction pure-tone average (ACPTA) thresholds in the conversational frequencies (0.5, 1, 2, and 4 kHz) were calculated for each deaf ear and were used to grade the severity of deafness. Four levels were defined: mild (20 dB<ACPTA≤39 dB), moderate (40 dB<ACPTA≤69 dB), severe (70 dB<ACPTA≤89 dB), and profound (≥90 dB). The severity of deafness was defined by the degree of hearing loss of the better ear. In accordance with the European Working Group on Genetics of Hearing Impairment criteria, hearing loss was considered to be progressive when the patient lost more than 15 dB in the ACPTA thresholds in the conversational frequencies as determined by comparing the results of 2 reliable audiometric tests performed at least 10 years apart, and we added to this criterion a more than 8-dB loss in tests performed 5 years apart. We considered deafness to be fluctuant when the hearing level had risen by more than 10 dB between 2 successive audiograms.
For intrafamilial comparisons, we considered that an intrafamilial variation was present when 1 sibling had severe or profound deafness and the other had moderate or mild deafness with a minimum 15-dB difference between ACPTA thresholds. Blood samples on EDTA and informed consent were obtained from the patients and their parents for subsequent genetic analysis. The study was approved by the Committee for the Protection of Individuals in Biochemical Research, as required by French legislation.
Genomic DNA was isolated from whole blood using various extraction methods. Analysis of GJB2 was performed on some of the samples using direct sequencing23 and on some using denaturing high-performance liquid chromatography followed by sequencing. The GJB2 coding exon was amplified in 2 fragments. The following primer pairs were used: CX26ex2S: 5′GCA TTC GTC TTT TCC AGA GCA and CX26DG2AS: 5′GAG CCT TCG ATG CGG ACC TT for the first fragment, CX26 2A: 5′GAG CCT TCG ATG CGG ACC TT and CX26ex2AS: TCA TCC CTC TCA TGC TGT CT for the second fragment. Fragments were amplified using AmpliTaq Gold (Applied Biosystems, Courtaboeuf, France). The first fragment was amplified using the following program: 95°C for 8 minutes, 35 cycles at 95°C for 30 seconds, 60°C for 30 seconds, and 72°C for 1 minute. The second fragment was amplified using touchdown polymerase chain reaction: 95°C for 8 minutes, 10 cycles at 95°C for 25 seconds, 64°C (–1°C) for 25 seconds, 72°C for 30 seconds, 25 cycles at 95°C for 25 seconds, 54°C for 25 seconds, and 72°C for 30 seconds.
To detect homozygous mutations, all amplification products were mixed and denatured for 5 minutes at 95°C with normal control polymerase chain reaction to allow the generation of heteroduplex. Both mixes were run on a Wave 2100A apparatus (Transgenomic; Courtaboeuf) at 63°C. Additional analyses were performed at 64°C to analyze from codon 71 to codon 121 and at 57°C to analyze from codon 200 to codon 227 for the first and the second fragments, respectively. Denaturing high-performance liquid chromatography variant profiles were sequenced using a Big Dye Terminator V.2 kit, following the manufacturer’s instructions, on an ABI310 (Applied Biosystems) application. The GJB2 exon 1 was only sequenced when a monoallelic mutation was found in exon 2. Exon 1 was amplified using the Expand Long Template polymerase chain reaction system (Roche, Meylan, France) with forward and reverse primers CX26 E1F 5′TCC GTA ACT TTC CCA GTC TCC GAG GGA AGA GG and CX26 ER1R 5′CCC AAG GAC GTG TGT TGG TCC AGC CCC. Sequencing was run as already described. Concerning GJB6, specific polymerase chain reaction assay using the method described by del Castillo et al3 was performed to detect the presence of the (GJB6-D13S1830)del deletion when no mutation was identified or when a monoallelic mutation was identified in GJB2. This alteration was then ascertained by sequencing. Mutation segregation was confirmed by molecular analysis of other members of the family to study the linkage between mutations and disease.
Patients with GJB2 variations considered to be polymorphisms or having an unclear pathogenic nature, that is, G4D, V27I, M34T, V37I, F83L, R127H, E114G, V153I, c.-15C→T, c.-34C→T, and c.558G→C, were excluded from the study.
All statistical comparisons between groups were performed using the Mantel-Haenszel χ2 test, or the Fisher exact test when numbers were small. For genotype-phenotype correlation analysis, patients were classified into 2 groups: those with severe or profound defects and those with mild or moderate defects. This distribution is medically justified in terms of implications of the hearing impairment on patient care.
A total of 29 different mutations of GJB2 and the GJB6 deletion (GJB6-D13S1830)del were identified in 207 unrelated patients (Table 1). As expected, the 35delG mutation was the most frequent, accounting for 69% of the alleles. Six other mutations demonstrated an allele frequency higher than 2%: the GJB6 deletion (5%), 313del14 (2.9%), E47X (2.9%), L90P, and Q57X (2.6%).
Three new mutations were identified: 355del9 was observed in 2 siblings associated in trans with 35delG, M151R was present in a sporadic case associated in trans with V95M, and 573delCA was identified in a sporadic case associated in trans with 35delG. These mutations were not observed in 100 chromosomes of normal-hearing individuals. The 355del9 mutation leads to the deletion of 3 amino acids, glutamic acids 119 and 120, and isoleucine 121, located in the second intracytoplasmic domain. Because it does not disrupt the open reading frame, 335del9 probably does not modify the rest of the protein. The M151R (c.452T→G) is located in the third transmembrane segment of the protein. Methionine 151 is evolutionarily conserved in all the species studied: Homo sapiens, Mus musculus, Cavia porcellus, Cricetus griseus, Macaca mulatta, and Pongo pygmaeus. The 573delCA mutation induces the creation of a stop codon at position 195 and modifies the amino acids between 191 and 195.
In this study, genotypes and phenotypes of 256 patients were analyzed. Genotyping of the 256 patients revealed that 131 were 35delG homozygotes and 72 were compound heterozygous for 35delG and a rare GJB2 mutation. Seven patients were homozygous for a rare GJB2 mutation and 21 were compound heterozygous for 2 rare GJB2 mutations. Twenty-five patients were identified as (GJB6-D13S1830)del and GJB2 compound heterozygotes. Not included in the 256 patients, we identified 16 patients as compound heterozygotes with M34T and a GJB2 mutation, 2 as homozygous for M34T, 5 as compound heterozygous for V37I and a GJB2 mutation, and 4 as homozygous for V37I, with 1 patient homozygous for R127H.
The hearing defect was profound in 146 patients, severe in 52, moderate in 48, and mild in 10. To evaluate the impact on the deafness phenotype of a non-35delG mutation, we compared the phenotype of 35delG compound heterozygous with 35delG homozygous patients (Table 2). No difference in deafness severity was observed with the GJB6 deletion or the GJB2 mutations Q57X, 313del14, E120del, and 167delT. In contrast, the L90P/35delG and N206S/35delG compound heterozygotes were statistically associated with a less severe phenotype than 35delG homozygotes.
The hearing loss of 7 patients homozygous for a rare mutation is reported in Table 3. Mutations were classified according to their possible effect on the connexin 26 protein. Frameshifts, nonsense mutations, deletions, and insertions causing a disruption of the open reading frame were considered to be inactivating mutations. Missense mutations, deletions, or insertions of 3 or a multiple of 3 nucleotides were considered to be noninactivating mutations. Moderate or mild hearing loss was more frequent in patients with 2 noninactivating mutations or with 1 inactivating and 1 noninactivating mutation (20 of 38 patients) compared with patients with 2 inactivating mutations (38 of 218 patients; P = .001) (Figure 1).
We compared hearing loss between 2 siblings bearing the same genotype in 49 families. In 18 families (37%), the severity of the defect differed between siblings, that is, profound or severe in one sibling and moderate or mild in the other. Intrafamilial variability was present in 13 (35%) of 37 families with 2 inactivating mutations and in 5 (42%) of 12 families with 1 or 2 noninactivating mutations, and this difference was not statistically significant (P = .74).
The evolution of the hearing defect was studied using tonal audiometric tests in 96 patients. The hearing level was stable in 73 patients (76%), progressive in 21 (22%), and fluctuant in 2 (2%). There was no statistically significant difference between the ages of patients with stable hearing loss and those with progressive hearing loss: 20 years (range, 7-50 years) vs 19.3 years (range, 6-56 years), by Mann-Whitney-Wilcoxon test. Deafness evolution was found to significantly depend on genotype (P = .01) (Figure 2).
To date, the most frequent form of NSSNHL is DFNB1, due to mutations in GJB2 and to GJB6 deletion. DFNB1 represents up to 40% of NSSNHL cases in some populations.28 In France, it has recently been shown that DFNB1 accounts for 32% of NSSNHL.29 In the present study, we report molecular and phenotypic analyses of 256 patients with DFNB1, the largest cohort of patients with DFNB1 reported in any single country using standardized data collection. Thirty different GJB2-GJB6 mutations are reported; 35delG is the most frequent mutation observed, with an allele frequency of 69% (287 of 414). The 35delG mutation is the most frequent in white patients with DFNB1, but the allelic frequencies reported vary by country: 88% in Italy, 55% in Spain, and 58.3% in the South of France, which is a lower figure than in the present study.20,30,31 The second most frequent allele is the large deletion involving GJB6, (GJB6-D13S1830)del, which accounts for 5% of the alleles. The (GJB6-D13S1830)del allele frequency is high in Spain (7.6%-9.7%), Israel (6%-7.1%), and the United Kingdom (5.9%) but low in Italy and Belgium (1.4%).32,33 The highest (GJB6-D13S1830)del frequency was observed in the South of France, where it accounts for 15% (9 of 60) of the mutated alleles.20 Analysis of haplotypes associated with the deletion suggests a common founder for Western European countries,32 possibly originating in the South of France or in Spain, where the frequency is the highest. Four other mutations were recurrent, with an allele frequency greater than 2%: 310del14, E47X, L90P, and Q57X. Interestingly, 310del14 and L90P have frequently been found in Eastern European populations.26,34- 38 In our French cohort, 310del14 and L90P represented 3% and 2.5% of the alleles, respectively, possibly owing to Eastern migrations. In Ashkenazi Jews and Palestinians, the most common GJB2 mutation is the 167delT mutation.11,39,40 In our French cohort, this variant represented only 6 of the 414 alleles. The 235delC Asiatic mutation12,13 was present with a 167delT deletion in trans in 1 family of Chinese origins. We did not identify the African R143W GJB2 mutation41 in any of our patients. In addition to the prevalence of GJB2-GJB6 mutations, we described 3 new mutations most likely to have pathogenic consequences: 355del9, M151R, and 573delCA.
This study reports clinical findings for 256 patients with DFNB1. Profound or severe hearing loss was present in 198 patients (77%) and in 107 (82%) of 131 patients homozygous for 35delG. To study the effect of rare mutations on the phenotype, we compared the phenotype of patients carrying a rare mutation associated with 35delG with the phenotype of 35delG homozygous patients. Indeed, the phenotype of recessive disease usually depends on the less severe mutation.42 The distribution of patients with profound and severe or moderate and mild hearing loss did not differ from that of the 35delG homozygote group for the following mutations: GJB6 deletion, Q57X, 313del14, E102del, and 167delT. Conversely, L90P and N206S were associated with a higher proportion of moderate or mild hearing loss. L90P has already been described as being associated with nonsyndromic sensorineural hearing loss43 in many studies. Because L90P is not observed in the general population,19,20 this mutation is considered to be deleterious. This study did not include patients who carry mutations with controversial effects, such as M34T, V37I, and R127H, which are frequent and mostly associated with a mild phenotype. Nevertheless, we observed a significant milder hearing loss in patients with at least 1 noninactivating mutation. This could be explained by a residual activity of some connexin 26 mutants.
In addition, we showed that the severity of the hearing defect can differ between siblings. Indeed, an intrafamilial deafness variation was demonstrated in 36% of the families carrying various genotypes. This is in agreement with previous study results,23 and our study found an intrafamilial hearing loss variation even in cases of mild mutations. This observation demonstrates that the DFNB1 phenotype is determined not only by the GJB2-GJB6 genotype but also by environmental factors and genetic background.
Finally, we observed that DFNB1 hearing loss is mostly stable (76%) but can be progressive and, in rare cases, fluctuant. Although Cryns et al27 found no evidence of progressive hearing loss, this pattern has been observed in patients with DFNB1.23,26 Also, we observed that progressive hearing loss was more frequent in patients with mild genotypes. It would be useful to examine the effect of environmental factors, such as infections or medications, on hearing loss increase in patients with DFNB1.
This is the first homogenous study of GJB2 genotype-phenotype correlations. It reviews the spectrum and distribution of mutations in France, confirms previous study results, and broadens the spectrum of mild mutation. Indeed, genotype is linked to phenotype for specific mutations, but the role of genetic background and environmental factors cannot be discounted, as shown by intrafamilial variability. Many factors remain to be studied to help with the genetic counseling and treatment of patients with NSSNHL.
Correspondence: Françoise Denoyelle, MD, PhD, and Sandrine Marlin, MD, PhD, Service d'ORL et de Chirurgie Cervico-faciale, Unité de Génétique Médicale, Hôpital d'Enfants Armand-Trousseau, AP-HP, 26 Avenue du Dr Arnold Netter, 75012 Paris, France (firstname.lastname@example.org and email@example.com).
Submitted for Publication: February 8, 2005; accepted February 12, 2005.
Previous Presentation: This study was presented as a poster at the 54th Meeting of the American Society of Human Genetics; October 26-30, 2004; Toronto, Ontario.
Funding/Support: This work was supported by the Fondation pour la Recherche Medicale, the Association Française Contre Les Myopathies, the Institut National de la S ante et de la Recherche Medicale (INSERM), and the Association “S’entendre,” Paris, France.
Acknowledgment: We thank Catherine Magnier, DETAB, Corinne Chauve, DETAB, Isabelle Sargis, BTS; Christina Das Neves, DETAB, and France Laroze, DETAB, for their excellent technical assistance.