High-Frequency Sensorineural Hearing Loss in Children

The Laryngoscope VC 2015 The American Laryngological, Rhinological and Otological Society, Inc. High-Frequency Sensorineural Hearing Loss in Children Kaalan Johnson, MD; Meredith Tabangin, MPH; Jareen Meinzen-Derr, PhD; Aliza P. Cohen, MA; John H. Greinwald, MD Objectives/Hypothesis: Determine the prevalence of high-frequency sensorineural hearing loss (HFSNHL) in our hearing loss population and a diagnostic algorithm for these patients. Study Design: Retrospective case series. Methods: We identified patients diagnosed with sensorineural hearing loss (SNHL) at our pediatric tertiary care institution from 1981 to 2010. Based on audiometric profiles, these patients were subdivided into those with a flat SNHL configuration and those with HFSNHL. Imaging and genetic testing data and data regarding age at diagnosis, laterality, and risk factors were obtained for both groups. Comparisons were then made between the two groups. Results: Of 2,867 patients included in the study, 7.6% had HFSNHL. Age at diagnosis was significantly higher in HFSNHL patients (8.3 years vs. 6.1 years; P <.0001). These patients also had a significantly higher proportion of unilateral versus bilateral loss (49.1% vs. 26.1%; P <.0001); unilateral losses were also less severe. Genetic testing showed no significant difference between groups in the proportion of patients tested or in those who tested positive. Similarly, imaging data revealed no difference in the proportion of patients tested in the two groups; however, overall diagnostic yield was significantly higher in flat SNHL patients (29.5% vs.17.3; P 5.02). Conclusions: The positive predictive value of simple genetic testing is similar to that of imaging studies. However, given cost differences between genetic testing and imaging, it is prudent to perform genetic testing as the initial diagnostic test. Determination of whether high-throughput, multigene diagnostic platforms offer an added benefit in the evaluation of children requires further study. Key Words: High frequency, sensorineural hearing loss. Level of Evidence: 4. Laryngoscope, 126:1236 1240, 2016 INTRODUCTION Sensorineural hearing loss (SNHL) is the most common sensory deficit in children. 1 On both screening and diagnostic audiometric testing, patients with SNHL largely present with an audiometric profile characterized by a flat configuration (i.e., flat SNHL). 2 Unlike children with flat SNHL, those with isolated high-frequency sensorineural hearing loss (HFSNHL) present with a sloped audiometric configuration and may have subtle symptoms of hearing loss. From the Department of Otolaryngology Head and Neck Surgery, University of Washington School of Medicine and Seattle Children s Hospital, Seattle, Washington, U.S.A. (K.J.); Division of Biostatistics and Epidemiology, Cincinnati Children s Hospital Medical Center, Cincinnati, Ohio, U.S.A. (M.T., J.M.-D.); Ear and Hearing Center, Division of Pediatric Otolaryngology Head and Neck Surgery, Cincinnati Children s Hospital Medical Center, Cincinnati, Ohio, U.S.A. (J.M.-D., A.P.C., J.H.G.); and Department of Otolaryngology Head and Neck Surgery, University of Cincinnati College of Medicine, Cincinnati, Ohio, U.S.A. (J.H.G.). Editor s Note: This Manuscript was accepted for publication July 7, 2015. Presented at the Annual Meeting of the American Society of Pediatric Otolaryngology, Las Vegas, Nevada, U.S.A., May 1, 2010. The authors have no funding, financial relationships, or conflicts of interest to disclose. Send correspondence to John H. Greinwald, MD, Division of Pediatric Otolaryngology Head and Neck Surgery, Cincinnati Children s Hospital Medical Center, 3333 Burnet Ave., MLC 2018, Cincinnati, OH 45229-3039. E-mail: john.greinwald@cchmc.org DOI: 10.1002/lary.25544 1236 Estimates of the prevalence of HFSNHL vary significantly. In a survey of 6,581 school-aged children in Australia, Wake et al. 3 identified only 10 children with HFSNHL; however, an epidemiologic study reported by Lee et al. 4 estimated that as many as 3% of school-aged children in the United States may have mild HFSNHL. Given the known prevalence of HFSNHL among adults exposed to noise in the work environment, 5 there is a high likelihood that the ever-increasing use of personal listening devices both in young children and teens may be leading to an increase in the prevalence of pediatric HFSNHL. 6,7 In view of these findings, we feel that the subpopulation of patients with HFSNHL warrants further investigation. In a previously published article by Preciado et al., 8 we described the development of an algorithm designed to improve the diagnostic efficiency of evaluating children with SNHL, but did not specifically study the subpopulation of patients with HFSNHL. Based on our findings, we recommended a stepwise approach in which patients with bilateral severe profound SNHL initially receive genetic testing, and patients with unilateral or less than severe SNHL initially receive high-resolution computed tomography (CT) imaging of the temporal bone. These recommendations were supported by our concurrent studies indicating that genetic testing and imaging had a greater positive predictive value in patients with severe profound SNHL as compared to those with mild to moderate SNHL. 9,10

TABLE I. Characteristics of Patients With Isolated High-Frequency Sensorineural Hearing Loss Versus Those With Flat Sensorineural Hearing Loss. Isolated High Frequency Loss, N 5 220 (7.62%)* Flat Loss, N 5 2,647 Age at diagnosis, yr, mean (SD) 8.3 (4), n 5 219 6.1 (4.5), n 5 2,597 <.0001 Male, n (%) 110 (50.9) 1,398 (53.9).39 Median follow-up, mo [IQR: 25%, 75%]* 5.7 [0, 44.4], n 5 220 9.7 [0, 48], n 5 2,639.24 Bilateral loss at 2 khz, n (%) 112 (50.9) 1,956 (73.9) <.0001 Unilateral loss at 2 khz, n (%) 108 (49.1) 691 (26.1) <.0001 Bilateral loss at 1 khz, n (%) 1,964 (74.2) PTA 4-tone (SD) 23.4 (7.8) 57.3 (30.8) <.0001 Severity of PTA 4-tone, n (%) <.0001 <20 74 (33.6) 66 (2.5) Mild, 20 40 137 (62.3) 908 (34.3) Moderate, 40 70 9 (4.1) 898 (33.9) Severe, 70 90 0 292 (11) Profound, 90 0 483 (18.3) High-frequency PTA (SD) 36.1(16.9) 59.1 (33.3) <.0001 PTA at 2 khz, mean (SD) 31.5 (12.7) 59.1 (32) <.0001 Risk factors, n (%) Prematurity 16 (7.3) 179 (6.8).77 NICU 5 (2.3) 39 (1.5).38 Congenital CMV 0 23 (.9).25 Family history 38 (17.3) 413 (15.6).51 Any risk factors (n, %) 52 (23.6) 582 (22).57 *Isolated 2 khz 5 hearing loss at 2 khz, normal at 1 khz. Fisher exact test; Pearson v 2 test was used for the remaining comparisons. CMV 5 cytomegalovirus; High-frequency PTA 5 PTA at 2 khz, 4 khz, and 6 khz; IQR 5 interquartile range; NICU 5 neonatal intensive care unit; PTA 5 pure-tone average; PTA 4-tone 5 PTA at 500 Hz, 1 khz, 2 khz, and 4 khz; SD 5 standard deviation. The aim of the present study was to determine the prevalence of HFSNHL in our SNHL patient population and to determine a diagnostic algorithm for these patients. We hypothesized that the positive predictive value of diagnostic testing in patients with HFSNHL would be similar to that in those with mild to moderate SNHL. MATERIALS AND METHODS We identified all patients with SNHL seen at our tertiary pediatric center from January 1981 to March 2010. These patients were divided into two groups, based on their audiometric profiles: those with flat SNHL and those with HFSNHL. Patients with HFSNHL were defined as having normal hearing at 1 khz and below, and SNHL 20 db at 4 khz and/or 6 khz, with or without this loss at 2 khz. To draw comparisons between the two groups, we documented data pertaining to age at diagnosis, laterality, and risk factors for hearing loss. We also documented audiometric data, including pure-tone average (PTA in db) at 500 Hz, 1 khz, 2 khz, and 4 khz and highfrequency PTA (HFPTA) at 2 khz, 4 khz, and 6 khz. Patients with incomplete audiometric data were excluded from the study. Audiometric data was gathered as previously described, 9 with pure-tone audiometry performed by pediatric audiologists in a standard soundproof audiologic booth. Genetic testing of the GJB2, GJB6, SLC26A4, and MTRNR1 genes was performed as previously described. 9,11 13 Imaging data were reviewed for all groups as per the method previously described by Boston et al. 10 and Johnson et al. (unpublished data). This study was approved by our institutional review board. Statistical Analysis Continuous variables were reported as means with standard deviation. Categorical variables were reported as frequencies and proportions. t tests were used to test differences in continuous measures between the two groups (HFSNHL vs. flat SNHL). Differences in proportions were evaluated using either the Pearson v 2 test or the Fisher exact test, as appropriate. Statistical significance was set at P <.05. Statistical analysis was performed using SAS version 9.1 (SAS Institute, Inc., Cary, NC). RESULTS Of 2,889 patients with SNHL, 2,867 met our inclusion criteria; 2,647 (93.3%) had flat SNHL, whereas 220 (7.6%) had HFSNHL. Audiometric data comparing these two groups is presented in Table I. As shown, the age at diagnosis was significantly higher in patients with HFSNHL than in those with flat SNHL (8.3 years vs. 6.1 years, respectively; P <.0001). Patients with HFSNHL also had a significantly higher proportion of unilateral versus bilateral hearing loss (49.1% vs. 26.1%, respectively; P <.0001). Overall, unilateral losses were also less severe, with a mean threshold of 31.5 db versus 59.1 db, respectively at 2 khz; a mean PTA of 23.4 db versus 57.3 db, respectively; and a mean HFPTA of 36.1 db versus 59.1 db, respectively; P <.0001). We found no difference between the two groups in the prevalence of risk factors 1237

Fig. 1. Representative audiogram of a patient with high-frequency sensorineural hearing loss. This patient is a 17-year old male diagnosed at age 7 years. (Table I). Figure 1 illustrates a representative audiogram in a patient with HFSNHL. Genetic testing showed no difference between the HFSNHL and flat SNHL groups in the proportion of patients tested or in those who tested positive for the GJB2, GJB6, SLC26A4, or MTRNR1 genes (Table II). Also, the diagnostic yield for genetic testing in unilateral HFSNHL was 7/46 (15.2%) versus bilateral HFSNHL, which was 11/60 (18.3%) (P 5.67). This difference was not significant. As with genetic testing, temporal bone imaging data revealed no difference between the two groups in the proportion of patients tested; 75 (34.1%) patients with HFSNHL had images available for review, whereas 973 (36.8%) patients with flat SNHL had available images. As seen in Figure 2, the overall diagnostic yield for imaging was significantly higher in patients with flat SNHL than in those with HFSNHL (29.5% vs. 17.3%, respectively; P 5.02). The most common abnormality in the both groups was enlarged vestibular aqueduct (EVA), followed TABLE II. Genetic Testing Results. High-Frequency SNHL, n 5 220 Flat SNHL, n 5 2,647 Any genetic testing 106 (48.2) 1,272 (48.1).97 Abnormal (diagnostic yield) 18 (17) 248 (19.5).53 GJB2 tested 103 (46.8) 1,208 (45.6).74 GJB2 positive 15 (14.6) 198 (16.4).63 GJB6 tested 53 (24.1) 626 (23.7).88 GJB6 positive 1 (1.9) 3 (.48).29* Pendrin (SLC26A4) tested 9 (4.1) 186 (7).10 SLC26A4 positive 1 (11.1) 46 (24.7).69* MTRNR1 tested 52 (23.6) 591 (22.3).65 MTRNR1 positive 2 (3.9) 10 (1.7).25 *Fisher exact test. Pearson s chi-square test was used for all other comparisons. SNHL 5 sensorineural hearing loss. Fig. 2. Temporal bone imaging findings. EVA 5 enlarged vestibular aqueduct; SNHL 5 sensorineural hearing loss. [Color figure can be viewed in the online issue, which is available at www.laryngoscope.com.] by cochlear dysplasia. There was no significant difference in the presence of each of these abnormalities between those with HFSNHL and those with flat SNHL. EVA (unilateral or bilateral) was present in 10 of 75 (13.3%) patients with HFSNHL who received imaging studies versus 160 of 973 (16.4%) patients with flat SNHL (P 5.48). Cochlear dysplasia was present in five of 75 (6.7%) patients with HFSNHL versus 93 of 973 (9.6%) patients with flat SNHL (P 5.30). As shown in Table II and Figure 2, there was a difference in the diagnostic yield for our flat hearing loss group in genetic versus imaging testing (19.5% vs. 29.5%, respectively). Additionally, the diagnostic yield for imaging in unilateral HFSNHL was 8/35 (22.9%) versus bilateral HFSNHL, which was 5/40 (12.5%) (P 5.24). This difference was not significant. There were no significant differences in age, risk factors, genetic testing, or imaging data among patients with isolated SNHL at 2 khz, 4 khz, or 6 khz; nevertheless, the mean pure-tone threshold for an isolated loss at 4 khz was slightly greater than for an isolated loss at 2 khz (38 db vs. 31.5 db, respectively; P <.0001). Additionally, there was no significant difference in the rate of hearing loss progression between patients with flat SNHL and those with HFSNHL (41.7% vs. 39.4%, respectively; P 5.61). There was no significant difference in the rate of progression between patients with flat SNHL and those with HFSNHL when analyzed by specific frequencies (Table III). Additionally, the diagnostic yield of imaging studies in patients with hearing loss progression was 31.7% (154/486), whereas the diagnostic yield of genetic testing in these patients was 22.7% (85/375). DISCUSSION Although past studies have noted a distinct group of children with isolated HFSNHL, to our knowledge, this is the first study that has focused on the medical evaluation of these children. We found a relatively high prevalence (7.6%) of HFSNHL in our pediatric hearing 1238

TABLE III. Frequency-Specific Progression in Patients With High-Frequency Sensorineural Hearing Loss and Flat Hearing Loss With 90 Days Follow-up (n 5 1,718). Frequency N HFSNHL, n 5 122, n (%)* Flat SNHL, n 5 1,590, n (%) 500 Hz 1,701 45 (36.9) 636 (40.3).46 1 khz 1,712 48 (39.3) 634 (39.9).91 2 khz 1,712 48 (39.3) 671 (42.2).54 4 khz 1,694 51 (42.2) 650 (41.3).86 6 khz 1,175 25 (23.6) 227 (21.2).57 8 khz 1,471 43 (36.4) 410 (30.3).17 PTA 3-tone (500, 1 khz, 2 khz) 1,712 36 (29.5) 514 (32.3).52 PTA 4-tone (500, 1 khz, 2 khz, 4 khz) High frequency PTA (2 khz, 4 khz, 6 khz) 1,712 39 (32) 517 (32.5).90 1,711 47 (38.5) 706 (44.4).21 HFSNHL 5 high-frequency sensorineural hearing loss; PTA 5 pure-tone average; SNHL 5 sensorineural hearing loss loss population. Interestingly, these children had a somewhat different clinical profile than their counterparts with flat SNHL. They were more likely to have unilateral hearing loss and less- severe hearing loss. They also showed a lower yield on imaging studies than their counterparts with flat SNHL. As would intuitively follow, patients with HFSNHL were diagnosed at a later age. Overall, they constituted a less-severe hearing phenotype. Particularly important, there was no difference in the positive predictive value of GJB2 testing for patients with HFSNHL and those with flat SNHL. This suggests that it is clinically prudent for both hearing loss groups to undergo genetic testing. Although positive results for GJB6 and 12SrRNA testing were higher in children with HFSNHL, the total number of positive tests in our cohort was too low to be statistically significant. There was also a higher positive predictive value with SLC26A4 testing in patients with flat SNHL; however, this was not statistically significant. In view of the fact that SLC26A4 mutations are associated with patients who have abnormalities in temporal bone scans, 11 patients with HFSNHL would be expected to have a significantly lower prevalence of these mutations. Further study with a larger cohort may clarify this issue. Both Lim et al. 2 and Preciado et al. 8 investigated the diagnostic yield of genetic and imaging studies in patients with mild to moderate SNHL (PTA <55 db). Lim et al. reported that this loss was associated with the presence of missense (i.e., nontruncating) mutations, and that biallelic mutations were present in 9% of these patients; however, patients with HFSNHL were not specifically studied. Preciado et al. were the first to propose a sequential diagnostic algorithm for the evaluation of children with SNHL based on audiometric presentation. These authors reported positive predictive values of 15% and 20% for GJB2 testing and temporal bone imaging, respectively. Although a small cohort (n 5 36) of patients with HFSNHL was identified, only 12 patients received imaging studies; of these patients, four showed abnormalities. Genetic testing data for the HFSNHL cohort were not reported. For patients with HFSNHL, imaging tests had a positive predictive value similar to that of genetic testing (17.3% vs. 17.0%, respectively). Another consideration in the evaluation of children with HFSNHL is the cost of these diagnostic tools. Our data show that the majority of positive findings on genetic tests come from mutations at the DFNB1 locus (GJB2 and GJB6), with a small proportion coming from mutations in MTRNR1. Because mutations in SLC26A4 are found only in patients with temporal bone anomalies, they are not useful as a screening test in all patients. Based on the current charges at our institution for genetic tests ($950 for GJB2, GJB6 and MTRNR1) compared to imaging ($1,900 for CT and $3,100 for magnetic resonance imaging), genetic testing is a more cost-effective diagnostic tool in patients with HFSNHL. Our study spanned a period of nearly 3 decades (1981 2010). As is often the case in a retrospective study over such a long period of time, we did not have complete study data on all patients meeting inclusion criteria. Incomplete data reflected variations in diagnostic interventions and clinical practice among subspecialists over this duration. Moreover, genetic testing did not become routinely available at our institution until 1999. Despite this drawback, the impressively large number of patients (n 5 2,647) included in the study, together with the fact that there was no difference in the proportion of patients who received temporal bone imaging and genetic testing, add validity to our findings. Clinicians should, however, keep in mind that these findings do not de-emphasize the importance of history and physical exam, both of which are important in ruling out causal or related factors such as meningitis, identifiable syndromes (e.g., Waardenburg, Down), and temporal bone trauma. CONCLUSION For patients with no definable etiology of hearing loss indicated by history and physical examination, there was no significant difference in the positive predictive value of genetic tests between those with a flat audiometric configuration and those with a high-frequency audiometric configuration, regardless of hearing loss severity. Imaging findings were present in 17.3% of patients with HFSNHL, a finding comparable to that reported for patients with mild to moderate SNHL in the cited study by Preciado et al. 8 ; it is important to note, there was no overlap in HFSNHL patients between these two studies. Overall, data do not support our hypothesis that the diagnostic findings in patients with HFSNHL are similar to those in patients with bilateral mild-to-moderate SNHL, but neither are they similar to those with bilateral severe-to-profound SNHL. The positive predictive value of simple genetic testing for 1239

mutations at the DFNB1 locus (GJB2 and GJB6) and MTRNR1 is similar to that of imaging studies. However, in light of cost differences between these two diagnostic tools, it is prudent to perform genetic testing as the initial diagnostic test. Determination of whether highthroughput, multigene diagnostic platforms offer an added benefit in the evaluation of children will require further study. BIBLIOGRAPHY 1. Morton CC, Nance WE. Newborn hearing screening silent revolution. N Engl J Med 2006;354:2151 2164. 2. Lim LH, Bradshaw JK, Guo Y, et al. Genotypic and phenotypic correlations of DFNB1-related hearing impairment in the Midwestern United States. Arch Otolaryngol Head Neck Surg 2003;129:836 840. 3. Wake M, Tobin S, Cone-Wesson B, et al. Slight/mild sensorineural hearing loss in children. Pediatrics 2006;118:1842 1851. 4. Lee DJ, Gomez-Marin O, Lee HM. Prevalence of childhood hearing loss: the Hispanic Health and Nutrition Examination Survey and the National Health and Nutrition Examination Survey II. Am J Epidemiol 1996;144:442 449. 5. Daniel E. Noise and hearing loss: a review. J Sch Health 2007;77:225 231. 6. Berg AL, Serpanos YC. High frequency hearing sensitivity in adolescent females of a lower socioeconomic status over a period of 24 years (1985 2008). J Adolesc Health 2011;48:203 208. 7. Vogel I, Brug J, van der Ploeg CP, Raat H. Strategies for the prevention of MP3-induced hearing loss among adolescents: expert opinions from a Delphi study. Pediatrics 2009;123:1257 1262. 8. Preciado DA, Lawson L, Madden C, et al. Improved diagnostic effectiveness with a sequential diagnostic paradigm in idiopathic pediatric sensorineural hearing loss. Otol Neurotol 2005;26:610 615. 9. Lee KH, Larson DA, Shott G, et al. Audiologic and temporal bone imaging findings in patients with sensorineural hearing loss and GJB2 mutations. Laryngoscope 2009;119:554 558. 10. Boston M, Halsted M, Meinzen-Derr J, et al. The large vestibular aqueduct: a new definition based on audiologic and computed tomography correlation. Otolaryngol Head Neck Surg 2007;136:972 977. 11. Madden C, Halsted M, Meinzen-Derr J, et al. The influence of mutations in the SLC26A4 gene on the temporal bone in a population with enlarged vestibular aqueduct. Arch Otolaryngol Head Neck Surg 2007; 133:162 168. 12. Lu J, Li Z, Zhu Y, et al. Mitochondrial 12S rrna variants in 1642 Han Chinese pediatric subjects with aminoglycoside-induced and nonsyndromic hearing loss. Mitochondrion 2010;10:380 390. 13. Putcha GV, Bejjani BA, Bleoo S, et al. A multicenter study of the frequency and distribution of GJB2 and GJB6 mutations in a large North American cohort. Genet Med 2007;9:413 426. 1240