The etiological evaluation of sensorineural hearing loss in children

The audiometric evaluation was performed by the referring Audiology Center or by the VUmc Audiology Center. The audiometric evaluation consisted of pure tone audiometry (PTA) if possible, auditory brainstem response using clicks (ABR), or both.

The degree of hearing loss was determined on the first available audiometric test and summarized by an average threshold at 500, 1000, 2000 and 4000 Hz on PTA or the estimated hearing threshold around 3 kHz on ABR. Children were diagnosed with bilateral SNHL if the sensorineural hearing threshold at the best hearing ear was 30 dB or more. Asymmetric bilateral SNHL was defined as 1 or more frequencies with greater than a 30 dB difference, 2 or more frequencies with greater than a 15 dB difference in threshold or 3 or more frequencies with greater than a 10 dB difference in threshold between the 2 ears (10,22). Unilateral SNHL was defined as a hearing threshold at the worst hearing ear of 30 dB or more, and a hearing threshold of 20 dB or less at the contralateral ear. Hearing loss was categorized as a slight impairment (26-40 dB), moderate impairment (41-60 dB), severe impairment (61-80 dB) and profound impairment (81 dB or greater) according to the classification of the World Health Organization (WHO) [10] . A patients’ hearing loss was graded according to the worst hearing side. In case of mixed type hearing loss, the inclusion and consecutive analyses were based on the sensorineural component only. Patients with pure conductive hearing loss were excluded from this study.

The age at detection was defined as the age at which the hearing loss was first diagnosed by the Audiology Center, either by ABR or PTA. Patients were categorized in 4 age groups: 0-1 year old, 1-6 years old, 6-12 years old and 12-18 years old.

Patient charts were reviewed for demographic data, audiometry, and the results of dysmorphologic, pediatric, ophthalmologic, and neurologic evaluation. Furthermore, the use and results of imaging, molecular genetic testing, and laboratory tests were reviewed. Imaging consisted of computed tomography (CT) of the temporal bone and/or MR imaging of the inner ear, cerebellopontine angle, and brain. The decision to obtain imaging and the choice of the imaging modality was individualized per patient and made by the multidisciplinary team. Molecular genetic testing consisted at first of Sanger sequencing of one or several single genes (usually GJB2, other genes based on clinical suspicion), 201 children in our population underwent this type of genetic testing. Whole-exome-sequencing (WES) became available in a diagnostic setting in 2013. WES targeting a panel of hearing loss–related genes was performed in 204 children [29]. A congenital CMV infection was detected with polymerase chain reaction (PCR) using the dried blood spots on Guthrie cards, which are preserved for 5 years after birth in Netherlands. After this period, Guthrie cards are destroyed and reliable detection of congenital CMV is no longer possible. During the period reviewed in this study, an unrelated nationwide study into the occurrence of congenital CMV infections in children with congenital SNHL in Netherlands was taking place [10]. Some children were already evaluated for the occurrence of congenital CMV by this study before presentation for etiologic diagnosis of SNHL, in these cases, the CMV status as determined was used for the analysis in this study, and testing was not repeated at our center. Additional tests were performed when indicated by the multidisciplinary team and included metabolic screening and DNA testing for copy number variations by single-nucleotide polymorphism (SNP) array, urine screening for hematuria and proteinuria (in case of childhood onset hearing loss in boys), ECG, and evaluation of congenital infections other than CMV.

The etiology of SNHL was divided into different diagnostic categories: genetic, suspected genetic, structural anomalies, acquired, miscellaneous, and unknown (Fig. 1). A genetic cause was established if it was confirmed with DNA testing [29]. The SNHL was considered to be of “suspected genetic” origin if there was a strong suspicion of syndromic SNHL because of a patients’ dysmorphic features or comorbidities, a positive family history for SNHL, a gene variant of unknown pathogenicity in a gene known to be associated with hearing loss, or in some cases with a single autosomal recessive pathogenic gene variant and a specific phenotype associated with that gene. A diagnosis was categorized as “structural anomaly” if a causative abnormality was found on imaging, and no further genetic or syndromic diagnosis could be established. Acquired causes included congenital TORCH infections, meningitis, hyperbilirubinemia requiring exchange transfusion, asphyxia, neonatal intensive care stay longer than 5 days, prematurity, trauma, ototoxic drugs, and others. These risk factors were deemed causative of SNHL after exclusion of other possible causes.

A total of 498 children with SNHL were retrospectively evaluated. Seventy-five children were excluded from the study because further evaluation was not performed at their parent’s request, the hearing loss was not of sensorineural origin, or the sensorineural component did not exceed 30 dB. A total of 423 children were reviewed, 239 males and 184 females. Age at detection of hearing loss ranged from 1 week to 16.9 years (median age of 0.9 year). The mean age of children with bilateral SNHL was 0.7 years old, and the mean age of children with USNHL was 3.3 years old. One hundred and ninety-seven children were diagnosed with hearing loss before the age of six months (47%). They were referred directly after newborn hearing screening. Two hundred and twenty-six children (53%) were referred at an older age.

Two hundred fifty-one (59%) children were referred by audiology centers, 99 (23%) by otolaryngologists, 24 (6%) by other medical specialists including pediatricians, 18 (4%) children by NICUs, 20 (5%) by child health and care service, and 11 (3%) by general practitioners.

Bilateral SNHL was diagnosed in 300 patients (71%) and unilateral SNHL in 123 children (29%). The hearing loss was most frequently profound in nature, both in uni- and bilateral SNHL (48% and 39%, respectively) (Table 1). The hearing loss was detected by automated auditory brainstem response (AABR) in 242 children (57%), at a mean age of 9 months old. The remaining children underwent pure-tone audiometry (PTA).

The etiological evaluation was performed using a stepwise protocol in 67% of the children with USNHL and in 61% of the children with bilateral SNHL (Fig. 2). Reasons to deviate from the protocol were medical indications (including meningitis, neurodevelopment disorders, and syndromic features), a cochlear implant procedure, or parental request.

The etiology of the SNHL could be established in 67% of the children. The distribution of etiologic diagnoses is presented in Fig. 1. Among the children with an established etiology, the cause was most frequently genetic (n = 87, 31%) or acquired (n = 75, 26%) (Supplementary Table 3).The probability of identifying an etiologic diagnosis is significantly higher in the youngest age group (74% vs. 60%, p < 0.01). We also found a significantly higher diagnostic yield in the most severe hearing loss category (p = 0.01). As suspected, there was a significant association between age and the level of hearing loss (p = 0.03), i.e., the most severe hearing loss category is overrepresented in the youngest age group. Using a likelihood ratio test, the level of hearing loss was found to be of significant added value to a logistic regression model that included age only (p = 0.01), and vice versa (p = 0.04). This suggests that both severity of hearing loss and age are independent prognosticators for finding an etiologic diagnosis.

The probability of identifying a cause for SNHL did not differ between children with uni- and bilateral hearing loss (67% and 67%, respectively; p = 0.92). Nevertheless, the distribution of the etiologies was different for these two groups: a genetic etiology was found more frequently in the children with bilateral SNHL (27%), whereas a structural temporal bone abnormality was detected more often in children with USNHL (25%) (Table 2 A and B).

A clinical evaluation by the geneticist was completed in 330/423 children. In 307 of these children, a DNA analysis was performed; 57/123 (46%) children with USNHL and 250/300 (83%) with bilateral SNHL. Reasons for genetic testing in children with USNHL included abnormalities found on CT and/or MR imaging, suspicion of syndromic SNHL, or a strong positive family history of genetic SNHL. A genetic abnormality associated with SNHL was identified in 87/307 children overall (28%). The diagnostic yield of genetic evaluation is higher in children with bilateral SNHL compared with children with USNHL (52% vs. 33% of the tested children). Genetic evaluation also revealed a significant higher proportion of genetic causes in the youngest age group (56% vs. 39%, p < 0.01). In contrast, the probability of finding a genetic cause was comparable between the different hearing loss categories (p = 0.07). Over time, the protocols for genetic evaluation have changed, from single-gene testing to WES. Especially since the introduction of WES, the diagnostic yield has increased (in this study from 26 to 36%) and will probably increase even more in the future due to improved protocols.

Of the children with a confirmed genetic cause (n = 87) for SNHL, 47% presented with syndromic etiology and 53% with non-syndromic etiology (Table 3; Supplementary Table 1). The most frequent genetic cause was a mutation in the GJB2 gene (27 patients, 30%), encoding Connexin 26. The most common syndromic cause was Usher syndrome (6 patients, 7%), followed by Stickler syndrome (5 patients, 6%). Ten patients (10%) were identified with Pendred syndrome.

Children with a suspected genetic cause (n = 70), presented in 43% with a positive family history for SNHL, 39% presented with a suspected syndrome associated with SNHL, and 18% had a gene mutation of unknown pathogenicity, or a single heterozygous variant in a gene (predominantly in the SLC26A4 gene) known to cause SNHL with autosomal recessive inheritance. Fourteen (20%) of these children had USNHL and 56 (80%) bilateral SNHL (Supplementary Table 2).

Radiologic imaging was performed in 321 children (76%), of which 112 children had USNHL and 209 had bilateral SNHL. Ninety children (28%) underwent CT as a single modality, 110 children (34%) underwent MR as a single modality, and 122 children (38%) underwent both modalities. The overall prevalence of relevant findings on imaging was 38%. Of all identified abnormalities, 60% was located within the labyrinth, 15% in the cochlear nerve, and 25% in the brain. Detailed description of the type of abnormality has been reported elsewhere [22, 23].

The diagnostic yield of imaging is higher in children with USNHL than in children with bilateral SNHL (48% vs. 32%). Profound hearing loss is associated with the highest chance of finding a radiological abnormality (p < 0.01). In contrast, the probability of finding an abnormality with CT or MR imaging was comparable between the different age groups (42% vs. 34%, p = 0.13).

Congenital CMV was diagnosed in a large proportion of the tested patients (35/165; 21%). Two children had a negative CMV PCR but clear and specific clinical signs and MR findings associated with CMV infections, and were therefore diagnosed as patients with congenital CMV by the pediatric neurologist. Twenty children with a congenital CMV infection had bilateral SNHL (20/37; 54%) (Fig. 2). We identified a congenital CMV infection in a significantly higher proportion of children with USNHL compared with children with bilateral SNHL (37% vs. 17%, p < 0.01). In children with a congenital CMV infection, the severity of the hearing loss is usually profound, milder hearing loss was significantly less frequently observed (p < 0.01).

Metabolic screening tests were performed upon indication in 57/423 (13%) children. Three out of 423 (1%) patients had a metabolic disorder (mucopolysaccharidosis type I (n = 1), mucopolysaccharidosis type II (n = 1), and Niemann-Pick disease (n = 1). Urinalysis was performed in 19 (4%) children, one of whom had an abnormality which contributed to the diagnosis of Alport syndrome. An electrocardiogram was performed in 53 (12%) children, in two patients an abnormality was identified (in one case related to CHARGE syndrome and the other related to the disease of Niemann-Pick). Ophthalmologic examination was performed in 243 children (57%), of which 38 had USNHL and 205 bilateral SNHL. Abnormalities were identified in 84 (35%) children. Twelve children (5%) had eye abnormalities related with syndromic SNHL (i.e., retinitis pigmentosa and coloboma) or congenital CMV infection (chorioretinitis). The remaining children had refractive disorders or strabismus.

The majority of children that were referred for etiological diagnostics suffered from bilateral SNHL (71%). Whereas, the chance of identifying an etiological diagnosis is comparable for USNHL and bilateral SNHL (67% vs. 67%), the distribution of etiologies differs between both groups. The most frequent etiology in children with USNHL was an isolated structural anomaly of the temporal bone, while in children with bilateral SNHL the most common cause was a genetic variant affecting gene function (in short, variant) [20, 26].

The probability of finding an etiologic diagnosis is significantly higher in children with profound SNHL [11, 26]. In the current study, we also find a higher diagnostic yield in the youngest age group, and both age and severity of hearing loss are independent prognosticators for finding an etiologic diagnosis. However, the diagnostic yield in older pediatric patients and patients with milder hearing losses is still considerable (Table 2).

CT and/or MR imaging was performed in 76% of the patients. Imaging is an essential part of the etiologic analysis of USNHL because of the high prevalence of causative abnormalities that can be identified with radiology (38% in our population). In agreement with previous reports, we find a higher ratio of causative abnormalities in children with USNHL (49%) compared with children with bilateral SNHL (32%), indicating a higher diagnostic yield of imaging in children with USNHL [3, 8, 15, 27]. Because of this, we recommend to perform radiology as the first diagnostic step in patients with USNHL, and genetic evaluation as the first step in bilateral SNHL. We recommend performing CT imaging as the first modality of choice in USNHL, followed by MR imaging if CT results are negative (Supplemental Fig. 1), and MR imaging in bilateral SNHL [22, 23]. By performing the modality with the highest diagnostic yield first, additional diagnostics may be avoided, minimizing the impact on the patient and reducing costs.

We found that the diagnostic yield of DNA testing is considerably higher in children with bilateral SNHL than in children with USNHL (27% vs. 5%). In agreement with previous reports, variants of the GJB2 gene were the most prevalent genetic cause (30%) [1, 8, 11, 13, 14, 16, 20, 26]. We found an equal distribution of syndromic and non-syndromic genetic etiologies, in contrast with the reported dominance of non-syndromic diagnoses [1, 18, 25]. The Pendred spectrum was the most common syndromic diagnosis in our cohort (10%). While this is in accordance with some previous reports, others find that Waardenburg or Usher syndromes are more prevalent [16, 25].

Congenital CMV infection is by far the most prevalent acquired cause of congenital SNHL in this study, an observation that is in line with previous reports [6, 12]. In our study population, congenital CMV infection was found in 9% of all included children, and in 21% of children tested for CMV. With the growing recognition of congenital CMV infections as a cause of SNHL, CMV tests are nowadays performed in all children that present with SNHL, but in the first years of this study, this evaluation was not standard protocol. In addition, a neonatal screening program for congenital CMV infections has not yet been introduced in Netherlands, and the diagnosis relies on CMV PCR using the Guthrie card. As this is available until the age of 5 years in Netherlands, children that present after the age of 5 cannot be reliably tested for congenital CMV infections. The hearing loss of most of the children with congenital CMV in this cohort was classified as profound, which is in line with prior studies [7, 18]. We found that the diagnostic yield of the screening for congenital CMV infections is higher in children with USNHL than in children with bilateral SNHL (37% vs. 17%). With the possible advent of newborn screening programs for congenital CMV infections, the diagnosis of congenital CMV may become even more prevalent [7, 10].

The chance of identifying the cause of SNHL in children is high. Using our stepwise diagnostic approach, we found a diagnostic yield of 67%, both in children with uni- and bilateral SNHL (Fig. 3). Bilateral SNHL often has a genetic cause, whereas in unilateral SNHL, structural abnormalities of the labyrinth are the dominant etiologic factor. Based on these results we start the etiologic diagnostic work-up with genetic evaluation in children with bilateral SNHL, and with radiology in children with USNHL (Supplementary Fig. 1). Congenital CMV infections are a cause for both uni- and bilateral SNHL, and we recommend evaluation of congenital CMV infections in all children that present with SNHL. The highest proportion of causative abnormalities is found in children younger than 1 year and in children suffering from profound hearing loss. However, the diagnostic yield in older pediatric patients and patients with milder hearing losses is still considerable. We therefore offer etiological diagnostics to all pediatric patients with SNHL exceeding 30 dB, irrespective of age or degree of hearing loss.