To identify any chromosomal region that shows evidence for linkage to age-related hearing loss in humans.
Evaluation of genetic linkage using sibling-pair methods for hearing loss collected via self-report questionnaire and markers from a genome screening collected from a population-based representative sample of male fraternal twins born from 1917 to 1927.
Members of a group of 6108 World War II and Korean War veteran twins (2059 complete pairs) who completed a health history questionnaire at a mean age of 74.3 years (range, 69-82 years). A subset of 711 twins (343 complete pairs) later provided a blood sample for DNA extraction in a study of genetic factors in healthy aging. Among the complete pairs were approximately 160 fraternal twins; 50 of these pairs were concordant for age-related hearing loss with at least 1 co-twin reporting bilateral hearing loss and with marker data available for analysis.
A region suggestive of linkage was found on chromosome 3q, with a logarithm of the odds score of 2.5 in the same region of this chromosome where the DFNA18 locus resides, which has been reported to cause a form of progressive hereditary hearing loss.
To our knowledge, this is the first sample from the general population that has been used in a genome screening for qualitative hearing loss. The results, if confirmed, suggest that genetic variation in the region of DFNA18 may be responsible for hearing loss with age in the general population.
Hearing loss is the most common sensory disorder worldwide1 and can be classified as either syndromic or nonsyndromic. About 25% of hereditary hearing impairment can be attributed to a syndrome.2 The remainder of individuals with hearing impairment have no associated clinical features. Nonsyndromic hearing impairment and deafness have been attributed to a wide range of genes and loci. According to the most recent version of the Hereditary Hearing Loss homepage,3 at least 42 different autosomal dominant, 46 autosomal recessive, and 6 X-linked or mitochondrial forms of nonsyndromic hearing loss have been described. The genes involved in nonsyndromic hearing loss are responsible for the production of a wide assortment of proteins, including transcription factors, membrane bound receptors, myosin, extracellular matrix proteins, ion channels, and transmembrane proteins.1,4 Presbycusis, hearing loss associated with aging, is also one of the most common chronic ailments of the elderly population in the United States, affecting about 25% to 30% of individuals 65 years or older.5,6 However, at least 83% of individuals 60 years or older may actually experience hearing impairment at some level.7
In the course of a study looking for genes associated with healthy physical aging,8 genome-wide scanning with DNA microsatellite markers was performed. In this cohort, qualitative information was also available on hearing loss, and a reasonable number of fraternal (dizygotic [DZ]) twins were concordant for hearing loss in 1 or both ears. Linkage analysis was performed on these concordant siblings to see if any chromosomal regions could be linked to hearing impairment with age.
Subjects for this study were members of the National Academy of Sciences–National Research Council (NAS-NRC) twin panel. The twin registry is close to a population-based, representative sample of male twin births from 1917 to 1927 in the United States. A detailed description of the creation of the registry has been published9 and was more recently summarized by Page.10 Subjects for the present study were twins who completed a health history questionnaire (Q8) that was mailed in the fall of 1998, and they were subsequently recruited and participated by providing a blood sample for DNA extraction in a study of genes related to healthy aging. A total of 6108 Q8 questionnaires were returned, including those from 2059 complete pairs with a mean age of 74.3 years. In the study of genes related to aging, 1 or both co-twins met a definition of healthy physical aging because they had survived into their 70s without any cardiovascular disease, diabetes mellitus, or prostate cancer at the time of the Q8 survey. Blood was collected from 711 individuals participating in the aging study, including 343 complete pairs. Appropriate informed consent was obtained from all subjects, and the study was approved by the institutional review board of Indiana University, Purdue University, Indianapolis. Among the 343 complete pairs who participated by sending a blood sample, approximately 160 were DZ pairs with genotyping data available for linkage analysis from responses to questions regarding hearing loss in the Q8 survey.
The assignment for concordance for hearing loss in pairs of DZ twins was established if both siblings answered yes to the question, “Do you have hearing loss?” Pairs in which both siblings answered affirmatively were considered to be concordant for hearing loss, and they answered some additional questions about the hearing deficit. One item asked whether the hearing loss was bilateral or only in the left or right ear. Among the 50 concordant pairs selected for the linkage analysis, at least 1 of the co-twins had to have bilateral hearing loss. Twenty-nine pairs were concordant for bilateral hearing loss (79% of the 100 individual co-twins with any hearing loss). In the Q8 responses from all DZ twins, the overall percentage of bilateral hearing loss was 73.4% (1200 of 1634). The twins were also asked if they ever had worn a hearing aid. In all of the DZ subjects with hearing loss according to the Q8 survey responses, only 42.6% (701 of 1647) indicated that they had worn a hearing aid. In the 50 pairs who were concordant for hearing loss, 49.5% admitted to wearing a hearing aid (49 of 99). Among the concordant pairs for hearing loss, if any co-twin reported an onset of hearing loss before age 16 years, the pair was excluded from analysis. For all DZ individuals in the Q8 survey, the mean (median) age of the subjects when hearing loss began was 58.5 (64) years. In the sample of 50 concordant pairs for the linkage analysis, the mean (median) age of hearing loss was 56.9 (60) years (range, 17-79 years). The average difference in onset age between siblings was 14.3 years (range, 0-53 years).
Genotyping and error detection
A genome screening was completed using 400 dinucleotide repeat markers from the ABI Prism Linkage Mapping Set (Applied Biosystems, Foster City, Calif) with an average heterozygosity of 79% and an average intermarker spacing of 8.6 centiMorgans (cM) as previously described.8 Genotypes were determined using ABI DNA analyzers (models 3100, 3700, or 3730) and GENEMAPPER 2.0 software (Applied Biosystems). All genotypic data were evaluated for Mendelian inheritance of marker alleles with the computer program PedCheck.11 The genotypic data were then used to verify that the individuals were in fact DZ twins using the computer program RELPAIR.12,13
Marker allele frequencies were estimated from all the individuals genotyped. Marker order and map positions were obtained from the Marshfield electronic database (available at: http://research.marshfieldclinic.org/genetics/). Multipoint affected sibling pair linkage analysis was performed using the computer program Mapmaker/SIBS.14 This program analyzes the extent of allele sharing among sibling pairs. Because each sibling receives 1 allele from each parent, siblings can share 0, 1, or 2 alleles identical by descent (IBD) at any given autosomal locus. Under the null hypothesis of no linkage, 25% of siblings should share 0 alleles IBD, 50% of siblings should share 1 allele IBD, and 25% of siblings should share 2 alleles IBD. Analysis of the affected sibling pairs reveals the extent to which there is deviation from the expected proportion of 0.25:0.50:0.25 alleles shared IBD. Because the fraternal twin siblings are concordant for the trait of interest, if a marker is near a gene influencing that trait, then a greater percentage of pairs should share 2 alleles IBD at that marker, with a corresponding decrease in the percentage of pairs who share 0 alleles IBD. Logarithm of the odds (LOD) scores were computed at 1-cM intervals, and the information content at each interval was recorded.
After the analysis of 400 markers, strong evidence was found for a hearing impairment susceptibility locus on chromosome 3 (3q22) near marker D3S1292. A LOD score of 2.5 for hearing loss was observed in the 50 affected sibling pairs, indicating increased allele sharing in the marker region. Most of this evidence can be attributed to the subset of 29 sibling pairs concordant for bilateral hearing impairment (LOD = 2.2 at the same locus). This marker, D3S1292, maps to the DFNA18 locus, a form of hereditary deafness with progressive hearing loss with age. Four additional markers were added to the region, and the linkage evidence remained robust (Figure; LOD = 2.5); however, the maximum LOD score shifted slightly to one of the new markers, D3S1558. The linkage interval, which is roughly between D3S2496 and D3S3637, is 23-cM long and contains 255 genes, 189 of which have been found expressed in the brain. Although low information content increases the suspicion that a linkage result might be spurious, this was not an issue here because the average information content between markers in this region (average information content, 0.79) was higher than the rest of the chromosome (average information content, 0.66) owing to the higher density of typed markers (Figure).
The only other region with a LOD score higher than 1.5 for the 50 families was at D1S249 (LOD = 2.1). Again, most of the evidence came from the 29 siblings with bilateral hearing loss (LOD = 2.2). The nearest known Mendelian deafness gene (DFNA7) is more than 20 cM away and is unlikely to be the cause of this peak. There was also 1 region near D15S153 that showed evidence of linkage in the bilateral subset (LOD = 2.3) but did not show evidence in the full sample (LOD = 0.0). There are 2 deafness genes on chromosome 15q – STRC, which codes for stereocilin, and the locus DFNA30; however, the peak is 15 to 20 cM away from the former and 40 cM away from the latter. One early hypothesis we considered was that carriers of the relatively more common connexin-26 (GJB2) mutations causing autosomal recessive congenital deafness might have increased presbycusis with age. No evidence existed for linkage to this region of chromosome 13 in our sample. There was also no evidence for linkage in the chromosomal regions warranting further investigation of linkage from family data assessed via quantitative assessment of hearing ability.15
Nonsyndromic hearing loss has been found to segregate as autosomal dominant, autosomal recessive, X-linked, and mitochondrial forms of inheritance. Age-related hearing loss has been studied less extensively. It was, however, reported to have a heritability of at least 40% in Danish twins over 75 years.16 An analysis of hereditary hearing loss based on all responses to Q8 by the members of the NAS-NRC twin panel (mean age, 74.3), using the same analytical method of analysis as in the Danish twin study,16 had a heritability of 61%,17 but subjects were not excluded on the basis of reported age at onset or whether the deficit was unilateral or bilateral. The present study was undertaken because of the moderately high heritability from our Q8 analysis and because, to our knowledge, no other reported data exist on genome screens of qualitative assessment of hearing loss with age in the general population.
The region we found linked to chromosome 3 in our sample is corroborated by Bonsch et al,18 who studied a large German family with an autosomal dominant, nonsyndromic form of bilateral progressive hearing impairment with onset in the first decades of life. The locus causing the hearing loss was mapped to 3q22, with the peak LOD score at marker D3S1292. Bonsch et al18 named this locus DFNA18 and excluded all other known deafness loci as possible causes of the hereditary hearing impairment in this family.
The DFNA18 locus overlaps with the DM2 (myotonic dystrophy 2) locus on chromosome 3 (3q22). An occasional clinical feature of DM is hearing loss, and hearing loss has been the only clinical finding reported in some DM1 families. The DNFA4 locus, which is responsible for autosomal dominant hearing impairment, is adjacent to the DM1 locus on chromosome 19.19 It has been hypothesized that the CTG repeat expansion in the DMPK genes that are responsible for the DM phenotype can affect the expression of nearby genes, such as those in the DFNA18 and DFNA4 loci.20 This ability to affect neighboring genes may be a source of the variability in the DM phenotype. Other deafness-related genes are also located on 3q22, such as DFNB15 and USH3. Although USH3 is approximately 20 cM away from the 1-LOD interval, DFNB15 is within the margin of error of linkage analysis (Figure). However, the phenotypes corresponding to these loci are more profound, with earlier ages of onset than found in either our twins or in the subjects in the German family.18
Although the family in the study by Bonsch et al18 had early-onset autosomal dominant hearing loss, it should be noted that the hearing loss was initially with high tones and progressively involved lower tones over decades. Variability in phenotype could depend on the type and position of the gene mutation. The severity could depend on the degree to which the mutation affects the native activity of the gene product. This variation in phenotype should not undermine the possibility that the 2 studies indicate that a single hearing loss gene resides within the DFNA18 locus. If our results can be replicated and the same locus involved with DFNA18 is also related to susceptibility to hearing loss with age in the general population, it would not be the first instance in which linkage analysis led to the recognition of a relatively common Mendelian form of hereditary deafness. The same was found for the now well-recognized connexin-26 mutations and autosomal recessive deafness21 accounting for up to 50% of hereditary deafness.1,22 The gene(s) responsible for hearing impairment within the DFNA18 locus has yet to be elucidated.3,18
As noted, we found no evidence of linkage for any of the 4 chromosomal regions with LOD scores over 1.5 reported in the Framingham study.15 There are considerable differences between the 2 studies. First, quantitative audiologic measures of low- and medium-frequency hearing loss was the phenotype in the Framingham families,15 despite the admission that high-frequency loss is the most common form of presbycusis. Our cases were self-reported and likely include a substantial proportion with high-frequency hearing loss. High-frequency hearing loss was also the characteristic initial finding in the DFNA18 family reported by Bonsch et al.18 In the Framingham families in the study by DeStefano et al,15 audiologic examinations were collected at a relatively wide age range, 32 to 89 years. The twins in our study were all born within the same decade, and questionnaires were completed at a mean age of 74 years with most between the ages of 70 and 80 years. Our subjects had to have at least 1 co-twin with bilateral hearing loss. Unilateral hearing loss was not excluded in the study by DeStefano et al,15 which used data from the better ear for linkage analysis.
Our study has a number of limitations that should be addressed. First, our sample was relatively small and included only male subjects. However, men do show greater frequency of presbycusis than women in the general population. Next, this study relied on self-reporting of hearing loss by the subjects. Hearing was never confirmed by follow-up testing or by medical record review. The subjects were not asked whether their hearing impairment had ever been confirmed by a medical professional. It has been pointed out that clinical and population-based studies show consistent epidemiological patterns for hearing loss whether the hearing is ascertained by questionnaire or audiometric testing.16 The percentage of male DZ twins indicating that they had a hearing loss (55.4%) was quite similar to the percentage of men with reduced hearing ascertained by questionnaire at roughly similar ages in the Danish twins.16 Our percentage of subjects admitting to ever wearing a hearing aid (42.6%) was nearly double the Danish twins, but the hearing aid question in the latter group indicated ownership of a hearing aid and might not be comparable.
Clark et al23 studied the accuracy of self-reported hearing loss in women aged 60 to 85 years and found that the more profound the hearing loss, the more accurate the self-reporting. The sensitivity of self-reporting was found to be 90% in these women. Based on positive predictive power calculations, the authors suggested that a single question asking about hearing loss should be used with caution if the prevalence of hearing loss in the population is less than 30%. It is likely that some individuals excluded themselves from our study by not recognizing or admitting to the fact that they were hearing impaired. There was a question asking whether the co-twin had hearing loss. In the 50 pairs in our study who both answered yes to the question that they themselves had hearing loss, their co-twin concurred on the hearing loss in 82.3% (65 of 79) of those who answered the question.
For most bilateral traits with genetically determined abnormalities, there is a higher percentage of affected individuals with bilateral expression. For this study, we included concordant pairs only if at least 1 of the co-twins had bilateral hearing loss. Results were similar when the analysis was restricted to the 29 pairs in which both co-twins had bilateral disease. There was variability between co-twins with regard to reported onset of hearing loss (median difference, 10 years). Some of this variability may relate to the interpretation of the question, “About how old were you when you first noticed hearing loss?” To eliminate true early-onset congenital hearing loss, we eliminated any pair if either co-twin answered this question as age 15 years or less. It is also possible that a small percentage might have answered the question as if the question were asking, “About how long ago did you first notice hearing loss?” Terminal digit bias was noted, with ages ending in 0 or 5 cited in higher percentages.
Finally, the effect of the military service that all the men in the sample shared on the hearing loss experienced by these individuals also should be noted. Many of the men were likely to have been exposed to loud noises such as munitions explosions, gun fire, and machinery clamor. Other than the striking effect of age on hearing loss, in the Framingham study,15 noise exposure was a significant risk factor for men.7 This exposure may be in large measure the cause of the hearing impairment in some of the subjects. The military service and exposure to loud noises may also represent a variable that is needed for gene-environment interaction. It is possible that subjects with a genetic susceptibility will only exhibit hearing loss late in life if they are exposed to significant noise earlier in life.
Although our linkage peak in the region of the DFNA18 locus is only suggestive of linkage using the criteria of Lander and Kruglyak24 (LOD score, 2.2-3.6), the findings were robust. Despite a relatively small sample size (50 pairs), there was consistent linkage when analysis was restricted to an even smaller number of pairs in which both co-twins reported bilateral hearing impairment. Genotyping of additional markers in the area of linkage on chromosome 3 did not eliminate evidence for linkage. The marker with the strongest evidence for the linkage peak (D3S1292) before fine mapping was the same marker at the peak of linkage in the family reported with the DFNA18 gene, and the phenotype is a progressive hearing loss that may have onset over decades. We await with interest the eventual cloning of the DFNA18 gene. If these results can be confirmed, it would suggest that other variations in sequence at the DFNA18 gene locus might be responsible for a substantial percentage of hearing loss with aging in the general population.
Correspondence: Terry Reed, PhD, 975 W Walnut St, IB 130, Indianapolis, IN 46202 (email@example.com).
Submitted for Publication: September 28, 2005; final revision received December 28, 2005; accepted January 24, 2006.
Financial Disclosure: None.
Funding/Support: This study was funded by grant R01AG18736.
Acknowledgment: We thank Danielle M. Dick, PhD, for sharing her marker files from the healthy aging linkage study with H.J.G., and Shannon A. Rinehart, RN, for managing the files in the database.
JR Hearing and Aging. New York, NY: Grune & Stratton; 1979
MC Prevalence estimates of communicative disorders in the U.S. language, hearing, and vestibular disorders. ASHA
1981;23229- 237PubMedGoogle Scholar
PM Hearing loss in the elderly: an epidemiologic study of the Framingham Heart Study cohort. Ear Hear
1985;6184- 190PubMedGoogle ScholarCrossref
WC Genome-wide scan for a healthy aging phenotype provides support for a locus near D4S1564 promoting healthy aging. J Gerontol A Biol Sci Med Sci
2004;59B218- B226PubMedGoogle ScholarCrossref
GF The NAS-NRC twin panel: methods of construction of the panel, zygosity diagnosis, and proposed use. Am J Hum Genet
1967;19133- 161PubMedGoogle Scholar
DE PedCheck: a program for detection of genotype incompatibilities in linkage analysis. Am J Hum Genet
1998;63259- 266PubMedGoogle ScholarCrossref
ES Complete multipoint sib-pair analysis of qualitative and quantitative traits. Am J Hum Genet
1995;57439- 454PubMedGoogle Scholar
CT Genomewide linkage analysis to presbycusis in the Framingham heart study. Arch Otolaryngol Head Neck Surg
2003;129285- 289PubMedGoogle ScholarCrossref
HJ Genetic and environmental influences on self-reported hearing in the old and oldest old. J Am Geriatr Soc
2001;491512- 1517PubMedGoogle ScholarCrossref
WH Self-reported health history survey (Q8) and genetic analyses in the NAS-NRC aging twin panel cohort. Am J Hum Genet
2000;67(suppl 2)215Google Scholar
et al. A novel locus for autosomal dominant, non-syndromic hearing impairment (DFNA18
) maps to chromosome 3q22 immediately adjacent to the DM2 locus. Eur J Hum Genet
2001;9165- 170PubMedGoogle ScholarCrossref
et al. Linkage of a gene for dominant non-syndromic deafness to chromosome 19. Hum Mol Genet
1995;41073- 1076PubMedGoogle ScholarCrossref
E Myotonic dystrophy: molecular windows on a complex etiology. Nucleic Acids Res
1998;261363- 1368PubMedGoogle ScholarCrossref
et al. Connexin26 mutations associated with the most common form of non-syndromic neurosensory autosomal recessive deafness (DFNB1
) in Mediterraneans. Hum Mol Genet
1997;61605- 1609PubMedGoogle ScholarCrossref
A Relation between choice of a partner and high frequency of connexin-26 deafness. Lancet
2000;356500- 501PubMedGoogle ScholarCrossref
C The accuracy of self-reported hearing loss in women aged 60-85 years. Am J Epidemiol
1991;134704- 708PubMedGoogle Scholar
L Genetic dissection of complex traits: guidelines for interpreting and reporting linkage results. Nat Genet
1995;11241- 247PubMedGoogle ScholarCrossref