Lustig LR, Lin D, Venick H, Larky J, Yeagle J, Chinnici J, Polite C, Mhatre AN, Niparko JK, Lalwani AK. GJB2 Gene Mutations in Cochlear Implant RecipientsPrevalence and Impact on Outcome. Arch Otolaryngol Head Neck Surg. 2004;130(5):541-546. doi:10.1001/archotol.130.5.541
Copyright 2004 American Medical Association. All Rights Reserved. Applicable FARS/DFARS Restrictions Apply to Government Use.2004
To determine the prevalence of GJB2 gene mutations in patients undergoing cochlear implantation (CI) and their impact on rehabilitative outcome following implantation.
Prospective determination of GJB2 mutation by sequence analysis by denaturing high-performance liquid chromatography and its correlation with outcome following CI.
Two tertiary academic medical centers.
Subjects who have met the audiologic criteria and have undergone CI.
Of 77 cochlear implant recipients screened, 13 (18%) harbored a detectable sequence alteration in the GJB2 gene. Only 3 of these 13 patients had hearing loss clearly attributable to a biallelic GJB2 mutation. There were 2 patients with homozygous mutations, including a 35delG and a 167delT mutation, and a third with a compound heterozygous mutation. Of the remaining 10 patients, 8 had 1 deafness allele, while 2 had a normal polymorphism that was not believed to be implicated in the hearing loss. Six patients had the common 35delG mutation: 5 patients had heterozygous mutations, which are probably not related to the underlying hearing loss (a second deafness allele cannot be ruled out in these cases because of the screening methodology used), while 1 patient had a homozygous mutation, which was clearly implicated in the patient's deafness. Rehabilitative outcome among those with detectable sequence alterations, as well as the 3 patients with biallelic mutations, varied but were similar on average when compared with outcomes seen in our entire CI population.
A large percentage of implant candidates harbor mutations or sequence alterations in the GJB2 gene, although only a small number of these changes are biallelic and a clear cause of the hearing loss. These results demonstrate that patients with GJB2-related deafness clearly benefit from CI.
Congenital deafness is present in approximately 1 in 1000 births, while the incidence of profound hearing loss in infants may be as high as 1 in 250.1 Of these, it has been estimated that half are due to a hereditary cause. Mutations in the gene GJB2, with a carrier frequency of approximately 3% to 4%, have been established as the most common cause of nonsyndromic hearing loss, responsible for up to 20% of cases of childhood deafness.2- 6 Other studies of pediatric and adult patients with severe to profound hearing loss have reported an even higher incidence of GJB2 mutations.2,7 The GJB2 gene codes for the gap junction channel protein connexin 26 (Cx26). It has been suggested that gap junctions assist in potassium homeostasis in the cochlea and that mutations in the Cx26 protein impair the endolymph potassium concentrations that are required for auditory signal transduction.8- 11
Cochlear implants are now commonly used for the treatment of patients with severe to profound hearing loss. While cochlear implants are successful in many patients, there remains wide variability in the rehabilitative outcome among implant recipients. Underlying genetic mutations could contribute to this variability. In the present study, we investigated the prevalence of mutations and nonpathologic sequence alterations in the coding exon of GJB2 and their impact on the functional outcome in patients undergoing cochlear implantation (CI).
Adult and pediatric cochlear implant recipients from Johns Hopkins University (JHU), Baltimore, Md, and the University of California, San Francisco (UCSF), were recruited to participate in the present study to investigate the genetic determinants of rehabilitative outcome with CI. All patients were enrolled under the strict guidelines of the committees on clinical investigation and institutional review boards of JHU and UCSF. Following informed consent, blood samples for DNA extraction were obtained.
In each patient, a complete physical examination, including comprehensive otolaryngology examination, was performed. Additional information culled from history, physical examination, high-resolution temporal bone computed tomography, and medical chart review included age of onset of hearing loss and any potential environmental or neonatal factors contributing to the deafness, whether the hearing loss was fluctuating or progressive, the concurrent presence of vertigo or balance disorders, and the presence of any associated visual impairment, dysmorphic features, and cochlear anomalies.
Audiometric data collected on each patient was age appropriate. For each adult, audiometric data included pure-tone audiometry, CNC (consonant-nucleus-consonant) for words, CNC for phonemes, HINT (Hearing in Noise Test) in quiet, and HINT for sentences +10-dB signal to noise. For children, audiometric data varied depending on age, but broadly included ESP (Early Speech Perception test), NU-CHIPS (Northwestern University Children's Perception of Speech test), PBK (Phonetically Balanced Kindergarten test) for words, PBK for phonemes, LNT (Lexical Neighborhood Test) for words, LNT for phonemes, MLNT (Multisyllabic and LNT) for words, MLNT for phonemes, GASP (Glendonald Auditory Screening Procedure) for words, GASP for sentences, HINT-Children (HINT-C) in quiet, HINT-C in noise (+10 dB), MAIS (Meaningful Auditory Integration Scale), IT-MAIS (infant-toddler version of MAIS), MUSS (Meaningful Use of Speech Scale), and Common Phrases.
Genomic DNA was extracted from whole blood with Puregene reagents using the manufacturer's protocol (PUREGENE DNA Isolation Kit; Gentra Systems, Minneapolis, Minn). Mutation screening of the GJB2 gene was carried out using previously described denaturing high-performance liquid chromatography, followed by sequencing of all samples.12 Each sample was analyzed blindly without the availability of clinical information. Briefly, after genomic DNA was extracted from whole blood, the single coding exon of GJB2 (GenEmbl accession No. M86849) was amplified using an established primer pair (5′ GCATTCGTCTTTTCCAGAGC and 3′ TGAGCACGGGTTGCCTCATC). Samples were amplified using a touchdown program (94°C × 3 minutes; 3 cycles of 94°C × 10 seconds, 65°C × 20 seconds, 72°C × 40 seconds; 3 cycles of 94°C × 10 seconds, 62°C × 20 seconds, 72°C × 40 seconds; 3 cycles of 94°C × 10 seconds, 59°C × 15 seconds, 72°C × 40 seconds; 3 cycles of 94°C × 10 seconds, 56°C × 15 seconds, 72°C × 40 seconds; 33 cycles of 94°C × 10 seconds, 55°C × 15 seconds, 72°C × 40 seconds, and a final extension of 72°C × 5 minutes). Denaturing high-performance liquid chromatography was performed on a Wave Nucleic Acid Fragment Analysis System (Model D7000IV with UV detector; Transgenomic Inc, Omaha, Neb); initial settings were established using the coding sequence amplified as 1 fragment or 2 overlapping fragments to determine the optimal melting temperature (Tm) for detecting at least 2 variants (a heterozygous G79A polymorphism and a homozygous 35delG mutation). The polymerase chain reaction (PCR) samples were hybridized to control DNA to form heteroduplexes in an approximately 1:1 ratio by heating mixed samples at 95°C for 8 minutes, then cooling to 25°C by decreasing 0.5°C every 19 seconds. The Tm was then estimated to be 61°C using the WaveMaker proprietary software, and final runs were performed at 62°C. Additional analyses at 58°C, 60°C, 63°C, and 64°C were performed on samples from patients in whom mutations were not detected.
In those samples where a mutation was detected, DNA sequencing of the complete GJB2 coding region was performed. The PCR samples were purified using 2 µL of exonuclease I (USB Corporation, Cleveland, Ohio) and 1 µL of shrimp alkaline phosphatase (Life Technologies, Rockville, Md) per reaction. Sequencing reactions used the ABI Prism BigDye Terminator Cycle Sequencing Ready Reaction Kit (Applied Biosystem, Foster City, Calif), and the products were resolved on an ABI 377-XL sequencer in the Genomics Core Facility at UCSF. The PCR samples were bidirectionally sequenced using the primers described in the previous section. All nucleotide changes were confirmed by repeating PCR and sequencing reactions. The sequences were analyzed using the Sequencher 3.0 software (Gene Codes Corporation, Ann Arbor, Mich).
It should be noted that neither the splice donor mutation in intron 1 nor the connexin 30 (Cx30) deletion were assessed. It is thus possible that mutations in these regions may have been responsible for the hearing loss detected in some of these individuals.
A total of 77 patients undergoing CI were included in the present study. The average patient age in this group was 33 years, with a range of 18 months to 84 years. There were 41 male (53%) and 36 female (47%) patients. Approximately one half of patients (n = 40) were younger than 18 years. These ratios were representative of the entire CI database at JHU (n = 680), indicating a representative sample.
Partial results from the genomic sequencing have been previously included in an article detailing the method of GJB2 screening using denaturing high-performance liquid chromatography and are given in Table 1.12 Of the 77 patients screened, 3 patients (approximately 4%) had biallelic mutations that were believed to be the cause of the hearing loss: 2 patients had classic homozygous mutations (patients 5 and 8), while the third patient had a compound heterozygous mutation (patient 10). The remaining 10 patients harbored nonpathologic polymorphisms (patients 1 and 3) or a single heterozygous deafness allele. However, a second deafness allele could not be ruled out in these cases because of the screening methodology used, which did not identify Cx30 mutations or alterations in the noncoding region of GJB2.
Of the 3 patients with biallelic mutations, 1 had the classic 35delG homozygous mutation, 1 had a 167delT homozygous mutation, and 1 had a compound (235delC + 504InsAACG) mutation. (The patient with the 167delT mutation was not an Ashkenazi Jew, an ethnic group with a known predominance of this mutation; however, there was an Ashkenazi Jew in the study who had a heterozygous 35delG mutation [patient 6].) Demographics of this group of 3 patients are given in boldface type in Table 1. There were no major clinical differences noted in this group of 3 patients compared with the larger groups of 10 patients with heterozygous sequence alterations, as well as the entire group of 77 patients who were screened. All 3 patients with biallelic mutations had a congenital hearing loss, while 7 of 10 of the heterozygous group had congenital hearing loss. Additionally, in these 3 patients with biallelic mutations, there was no family history of hearing loss, vertigo, visual impairment, dysmorphic features, and tinnitus and all had normal findings on temporal bone computed tomography, without evidence of anomalies or dysplasias. Surgery for these 3 patients was routine with no difficulties encountered and no complications noted.
The group of 10 patients with heterozygous sequence alterations in whom GJB2 could not be implicated as the cause of deafness should be clinically and audiometrically indistinct from our CI population as a whole. In this group, 1 patient (patient 4; heterozygous 35delG) had a family history of notable sensorineural hearing loss; this was an adult whose hearing loss was first diagnosed during childhood, which progressed to a profound loss by age 30 years. Two patients had visual impairment (patients 1 and 3), 2 patients had associated dysmorphic features that were nonsyndromic (patients 1 and 2), and 2 patients had associated tinnitus (patients 3 and 4). As with the biallelic group, all patients with heterozygous sequence alterations had normal findings on temporal bone computed tomography, and in all cases the CI was uneventful without no intraoperative complications. However, 1 patient (patient 4) had excessive facial nerve stimulation following implantation that could not be satisfactorily ameliorated by reprogramming; she ultimately underwent explantation and reimplantation with a modiolar-hugging electrode that significantly reduced the facial nerve stimulation.
Because there were only 3 patients with biallelic mutations that were believed to be clearly implicated in the hearing loss, combined with the varying lengths of follow-up for each patient and the variability of tests that must be used to measure hearing during the different ages during childhood and adulthood, meaningful statistical analyses could not be made. However, for a given patient at a given postoperative time point, some generalizations can be made by comparing their results with those within our entire implant database (n = 680). Audiometric results for the 2 pediatric patients at varying postimplantation time points are given in Table 2. Patient 5, a congenitally deaf child with a homozygous 167delT mutation, underwent CI at age 1 year. At her 5-year follow-up (Table 2), she scored above our center average on WIPI (Word Intelligibility by Picture Identification), PBK for words/phonemes, LNT for words/phonemes, GASP for sentences, and Common Phrases. Patient 10, a congenitally deaf child with compound (235delC + 504InsAACG) mutation underwent CI at age 3 years. He scored below average on most of the measured tests at 1 year, but had an average score on the 2-year MAIS and MUSS tests (Table 3). Thus, overall, the 2 children with biallelic mutations performed on average similar to our entire center average, while the adult performed worse than expected, most likely because of the prolonged time between the onset of deafness (birth) and age at implantation (36 years).
The audiometric data for patient 8, the 1 adult patient with a biallelic mutation (homozygous 35delG) is not shown. She is a prelingually deaf adult who did not undergo implantation until age 36 years; based on her length of deafness she was expected to achieve simple sound awareness and lip-reading assistance only with implant use. As predicted, she scored a 0 for all postoperative closed-set audiometric speech tests (monosyllable words and phonemes, CID (Central Institute for the Deaf) for sentences, and HINT in quiet and 10-dB noise). Despite these low scores, she enthusiastically uses the implant on a daily basis for assistance with lip-reading cues and environmental sound awareness.
Excluding the 2 patients with simple polymorphisms (patients 1 and 3), a similar comparison was made with the 8 patients who carried a single heterozygous deafness allele (Table 3 and Table 4). The performance scores in this group were similar to the average in our implantation center. In the adult group (Table 3), the scores for patient 4, a 19-year-old woman who had been completely deaf only several years prior to implantation, were equal to or better than the center averages for all audiometric tests. Long-term follow-up was only available for patient 7; he transferred to our institution and there were no preoperative audiometric scores available for comparison. However, compared with our center audiometric scores, his scores were better for the monosyllable words/phonemes, CID for sentences, and HINT in quiet tests. At present, he continues to have difficulty, however, on the HINT in noise (+10 dB) test.
Similarly, the performance scores for our pediatric group with a single heterozygous deafness allele (excluding patient 1 who had a simple polymorphism) were also similar to our center averages (Table 2 and Table 4). Patient 2 had both physical and mental disabilities and could not be adequately tested for any of the tests except the WIPI, in which she scored well below average. At the 1-year follow-up, patient 6 scored at or above average on the ESP, NU-CHIPS, and IT-MAIS tests, but scored below average on LNT for words/phonemes and MLNT for words/phonemes tests. At the 3-year follow-up, patient 9 scored below average on all measured tests. Patient 11 scored above average on all tests at the 2-year follow-up, with the exception of the MLNT for words. Patient 12, like patient 2, had a nonsyndromic mental disability and could not be tested adequately. Patient 13, at her 1-year follow-up, performed above average on the ESP test, but below average on LNT for words/phonemes tests.
Advances in the genome project and our understanding of the molecular genetics of deafness affords otolaryngologists a real opportunity to harness this newly discovered knowledge to impact clinical practice. Defining the genetic determinants of CI outcome is one such application to provide better information for individuals as they contemplate various therapeutic options for themselves or their children. In a study by Kenna et al,7 30% of 99 probands had GJB2 mutations, which included biallelic mutations in 18% (9% homozygous and 9% compound heterozygous) and single mutations detected in 12%. Though much lower than the homozygous mutation rate in the series by Kenna et al,7 our detection of a 4% homozygous mutation rate is within the range that has been previously described.1- 4,6 This slightly lower rate is not surprising given that CI candidates of all ages and ages at onset of deafness were offered inclusion in the study and not just those with congenital deafness, which was characteristic of these prior reports.
Of the 13 patients with detectable sequence alterations identified in our series, 10 patients (13% of the entire population of 77 patients screened) had heterozygous alterations or polymorphisms. Those harboring the well-described heterozygous Cx26 mutation 35delG (patients 4, 6, 7, 9, and 11) can be considered carriers whose mutation may have had little impact on hearing status or may have contributed to the loss along with other mutations that were not characterized in this study. Similarly, the other patients with detectable heterozygous sequence alterations (patients 2, 12, and 13) or polymorphisms (patients 1 and 3) in all likelihood had other genetic or environmental factors that contributed to their deafness. Interestingly, we identified a higher incidence of heterozygous alterations (18%) in our CI population compared with the 3% to 4% carrier frequency that has been documented for the general population.7,13 The higher incidence of sequence alterations is not surprising given that this was a population of deaf individuals who were screened. It should be noted that neither the splice donor mutation in intron 1 nor the Cx30 deletion were assessed. It is thus possible that mutations in these regions may have been responsible for the hearing loss detected in some of these individuals.
In our study, only 3 patients (approximately 4%) had biallelic mutations; 2 patients had homozygous mutations and 1 patient had a compound heterozygous mutation. All 3 patients had congenital profound hearing loss, suggesting that the cause of the deafness was related to the GJB2 mutation identified. Again, however, our sequencing method did not rule out the possibility that the deafness in these individuals did not have a contribution from a concurrent mutation in the noncoding region of GJB2 or an additional DFNB1 mutation in the gene encoding for Cx30.
To date, many types of mutations in the GJB2 gene have been described, but the 35delG mutation is the most common in patients of Western European ancestry, with a carrier frequency of 1:51.2,5 Similarly, the most common GJB2 alteration seen in our study was the 35delG mutation, identified in 6 (46%) of the 13 patients with detectable sequence alterations (5 heterozygous and 1 homozygous), which probably reflects a Western European ancestry bias in our patient population as well.
Unfortunately, meaningful statistical analysis to determine functional outcome in this limited data set was problematic given the small number of true biallelic GJB2 mutations and the wide array of audiologic tests required for varying ages. However, more meaningful comparisons were made by comparing the outcome results in our study subjects and our implantation center as a whole. This comparison suggests that our pediatric patients with GJB2 mutations can be expected to perform as well, on average, as cochlear implant recipients with deafness from all causes combined. The single adult identified with a biallelic 35delG mutation (patient 8) was not truly representative of our CI population; this patient was prelingually deaf and did not undergo implantation until age 36 years. We thus expected only simple environmental sound awareness and assistance with lip-reading with implant use, which she achieved. As expected, however, she did not achieve open-set speech discrimination.
The few prior studies that have examined the relationship between GJB2 mutations and CI outcomes would suggest that these patients perform slightly better than implant recipients on average. In a study by Green et al,14 cochlear implant recipients with GJB2-related deafness within 1 SD of hearing controls performed better than other congenitally deaf cochlear implant recipients and noncochlear implant recipients. In a Japanese study, 4 prelingually deaf patients with the 233delC mutation were shown to clearly benefit from CI.15 Further, these patients performed better than implant recipients who were deaf from another cause. In another Japanese study, better speech performance was noted in 3 children with GJB2-related deafness compared with 4 children with deafness unrelated to GJB2, as measured by the number of vocabulary words and tests of cognitive ability.16 These prior findings, combined with the results in our study, demonstrate that patients with GJB2-related deafness clearly benefit from CI.
Corresponding author: Lawrence R. Lustig, MD, Department of Otolaryngology–Head & Neck Surgery, Johns Hopkins University, JHOC 6241, 601 N Caroline St, Baltimore, MD 21287 (e-mail: email@example.com).
Submitted for publication January 15, 2004; final revision received January 30, 2004; accepted February 10, 2004.
This study was presented at the Ninth Symposium on Cochlear Implants in Children; April 24-26, 2003; Washington, DC.