A comparative study of auditory and non-auditory characteristics of congenital, early, and late-onset auditory neuropathy spectrum disorder

Abstract

Background

The study explored the differences in audiological and non-audiological characteristics between congenital, early-onset, and late-onset auditory neuropathy spectrum disorder (ANSD). Ninety-five individuals diagnosed with ANSD were included in the study. They were divided into three groups congenital ANSD—children (30 individuals, 60 ears), adults with early-onset ANSD (30 individuals, 56 ears), and adults with late-onset ANSD (35 individuals, 62 ears). The non-audiological characteristics (gender, laterality, and risk factors) and audiological characteristics (behavioral and electrophysiological measures) were compared between the three groups.

Results

Discriminant analyses showed that the pure tone average, audiogram configuration, and speech thresholds were the best auditory predictors of onset-based group differences in ANSD (congenital and early-onset versus late-onset ANSD). While the congenital and early-onset group showed poorer pure-tone and speech thresholds, along with flat configuration, the late-onset group demonstrated relatively better thresholds and other configurations (rising, tent-shaped, cookie-bite). In addition, long latency responses were delayed or absent in children with congenital ANSD, indicative of onset coding deficits at the cortical level.

Conclusions

The study highlights the audiological differences between congenital, early-, and late-onset ANSD groups. These differences could be because of variations in etiology, pathophysiology, site of lesion, or genetic variability between the groups, which needs to be explored further.

Background

Auditory neuropathy spectrum disorder (ANSD) is a retro-cochlear disorder with preserved otoacoustic emissions (OAE) and abnormal or absent auditory brainstem response (ABR) [1,2,3,4]. It is a clinical disorder in which the cochlear amplification is intact, but there is damage in the synchronous signal transmission in the auditory neural pathway. ANSD affects all age groups, with its onset reported in neonates, childhood, adolescence, and adulthood [1, 5, 6]. In Europe and America, the onset of ANSD is usually reported in children below ten years [1, 5, 7]. By contrast, the studies on the Eastern population, such as India, showed a late-onset of the disorder, which manifested itself during the second decade of life [6, 8], in the age range of 14 to 24 years. Along similar lines, Prabhu, Avilala, and Manjula [9] reported that the age of onset of ANSD was 15 to 25 years in their study, which they described as “late-onset auditory dys-synchrony.” This nomenclature was devised to emphasize the onset in such patients, who exhibited clinical signs of the disorder in adolescence or adulthood, with no reported high-risk factors in their neonatal or childhood stage. On the one hand, Narne, Prabhu, Chandan, and Deepthi [10] also reported the onset of ANSD symptoms was in adolescence and adulthood in 82% of their population (198 sample size). On the other hand, Shivashankar et al. [11] reported one of their 24 case reports with ANSD had reported onset at adulthood, i.e., at 44 years of age.

Other onset-based distinctions in ANSD can be traced to heterogeneity in its etiology [1, 12,13,14,15], symptoms, and pathophysiology [16, 17] of the disorder. In a classical study on ANSD, Berlin et al. [1] reported that audiological characteristics of the individuals seen varied across a continuum, with auditory abilities of the ANSD individuals ranging from mild to severe. In their research, they acknowledged that some subjects developed ANSD symptoms during adolescence or early adulthood. These subjects usually went unidentified or had no complaints about hearing in their childhood.

Timely detection of the onset of ANSD can help the audiologists determine appropriate rehabilitative measures. Yuvaraj and Jayaram [18] reported that early or late-onset ANSD could indicate the benefit of hearing aids. It is understandable that individuals with late-onset ANSD have extended periods of normal hearing before the onset of ANSD and are likely to benefit from hearing aids [18]. Despite the advantage of the application of onset-based differences in auditory rehabilitation, scanty literature with  specific focus on profiling the auditory and non-auditory characteristics of congenital, early, and late onset ANSD is available. The profiling of such traits can cluster factors that audiologists should pay attention to delineate ANSD onset-based group differences. The present study explored the differences in terms of auditory and non-auditory characteristics between children with congenital, early-onset adults (≥ 12 years with ANSD diagnosed in childhood) and late-onset ANSD groups (≥ 12 years with ANSD onset in adolescence or adulthood).

Methods

Study design and participants

A retrospective patient record review study design was adopted. The medical records of participants diagnosed with auditory neuropathy spectrum disorder (ANSD) in the Department of Audiology, All India Institute of Speech and Hearing, affiliated with the University of Mysore, India, between September 2015 and December 2019, were reviewed. A total of 95 individuals (178 ears) diagnosed with the condition by audiologists certified by the Rehabilitation Council of India (RCI) were considered for the study, who met the inclusion criteria. The criteria adopted to diagnose ANSD in the audiology clinic were those recommended by Starr et al. [7]: absent or abnormal ABRs (delayed in latency or attenuated in amplitude), presence (normal or robust amplitude) of OAEs, and absent middle ear reflexes. Based on the clinical record, the diagnosis of ANSD was confirmed by a neurologist using clinical examination and computerized axial tomography/magnetic resonance imaging. Only those participants in whom all the above tests were done were included in the study. Only those records containing information about the non-auditory factors were included in the study.

The participants were divided into three groups congenital ANSD (< 5 years)—children (30 individuals; 60 ears), adults (age ≥ 12 years) with early-onset ANSD (30 individuals; 56 ears), and adults with late-onset ANSD (35 individuals; 62 ears). The mean age of the congenital children ANSD group was 3.17 ± 1.99 years, while adult early-onset and late-onset ANSD groups were 21.32 ± 7.75 years and 23.16 ± 7.55 years, respectively. The rationale for choosing 12 years as the criteria for the segregation of adults and children was based on the previous literature reports on the onset of ANSD adults in adolescence/teenage [6, 10, 11]. A recurring observation seen in these studies showed that ANSD tends to manifest more prominently or commonly from adolescence onwards. This could mean that individuals aged 12 years and below might exhibit different patterns of ANSD or might be more affected by it compared to older individuals. Hence, it was postulated that using 12 years as the cutoff age for segregating early and late onset in ANSD studies allows researchers to focus on distinct age groups where differences in symptomatology (auditory and non-auditory symptoms), etiology, or treatment responses may vary. It also helps in ensuring more homogeneous study groups, which can lead to clearer and more accurate conclusions regarding the effect of ANSD onset on auditory and non-auditory characteristics.

At the time of retrospective analyses of data, the language age of the congenital ANSD children and adults with early-onset ANSD was less than their chronological age by at least 6 months. This was ensured based on the language assessment reports in the case files. The age-specific normative [19, 20] of the linguistic profile test (LPT), conducted in the native language of the participant [21], was noted to obtain the language age of each participant. No such discrepancy between the language age and chronological age was reported in the late-onset ANSD group. This was done to differentiate between late-onset and late-diagnosed ANSD. The data collected was based on the first-time evaluation of the participants, before the prescription of any amplification devices such as hearing aids.

Procedure

Ethical approval and consent to participate

All the data collected and the testing procedures adopted in the present study were non-invasive techniques, with adherence to the institutional ethical guidelines for conducting bio-behavioral research [22]. The study was approved by the AIISH Review Board (SH/AIISH/ERB/RP/32). The data retrospectively collected comprised of test results of behavioral (pure tone and speech audiometry), physiological (Immittance, OAE, and reflexes), and electrophysiological (ABR, LLR, CM whichever is available) tests, which were carried out using a fixed protocol as detailed below.

Behavioral tests

Pure tone and speech audiometry

Based on the protocol for audiological evaluation in the department, pure-tone air-conduction and bone conduction thresholds were determined using the Modified Hughson and Westlake procedure [23] with a calibrated two-channel diagnostic audiometer. The speech audiometry data collected from the medical records was based on responses obtained for live/recorded voice presentation of bisyllabic native language words in adults [24] and speech detection thresholds for children. The pure tone audiometry data collected retrospectively from the case files were analyzed for the degree [25], symmetry, and configuration [26] of the hearing loss.

Physiological tests: Immittance, reflexes otoacoustic emissions

Immittance evaluation (tympanometry and acoustic reflexes) was done using a calibrated middle ear analyzer with a 226 Hz probe tone. Ruling out middle ear effusion is necessary for the pediatric population to confirm the thresholds are not elevated due to the conductive component. It was ensured that all the cases included in the study had an A-type tympanogram [27], which was also validated priorly by the otoscopic examination conducted by an otolaryngologist. The acoustic reflex thresholds were determined for 500 Hz, 1000 Hz, 2000 Hz, and 4000 Hz pure tones in ipsilateral and contralateral modes. Distortion product otoacoustic emissions (DPOAE) were recorded with tones F1 and F2 at the ratio of 1.22 and 65 dB SPL and 55 dB SPL intensities, respectively. OAE was considered present if the waveform reproducibility was more than 50%. The overall signal-to-noise ratio was more than 3 dB for TEOAE and 6 dB for DPOAE, at least at two frequency bands.

Electrophysiological tests

Auditory brainstem response

Click-evoked auditory brainstem response (ABR) was recorded using insert phones (ER-3A). Ag/AgCl disposable surface electrodes were placed on the vertex (Cz) (non-inverting), over the mastoid of the test ear (M1 or M2) (inverting), and non-test ear (M2 or M1 as ground). It was ensured that the electrode impedances were less than two kOhms each. ABR waveforms were recorded at 90 dBnHL at 11.1 clicks/s and at 90.1 clicks/s in both ears using both rarefaction and condensation polarities. ABR recordings were amplified 100,000 times and were recorded for 10 ms (click) post-stimulus. It was averaged over 2000 samples and filtered from 100 to 3000 Hz. The cochlear microphonic (CM), which had long ringing responses from the inner ear, was discernable on ABR recordings in a few ANSD participants. In accordance with the standard operating protocol in the clinic, whenever ABR was absent for clicks, the presence of peaks for 500 Hz tone burst (TB) was also checked as a measure to ensure the limitation of high-frequency hearing loss by click-evoked ABR. ABR waveforms recorded for tone bursts were obtained at an intensity of 80 dBnHL for alternating polarity stimulus at a repetition rate of 11.1 TBs/s.

Cochlear microphonic

Cochlear microphonics were recorded using click stimuli with a Tiptrode (non-inverting) placed in the participant’s test ear canal. The surface electrodes were placed over the non-test ear mastoid (inverting) and on the lower forehead (ground). Cochlear microphonic (CM) was amplified 100,000 times, and it was averaged over 1500 samples. CM was recorded for 5 ms post-stimulus, and it was filtered from 1 to 1500 Hz. The waveforms were recorded in both ears using click stimuli presented at 90 dBnHL. The stimuli were presented at a repetition rate of 11.1 clicks/s, using rarefaction and condensation polarities to check for the reversal of CM. By contrast, alternating polarity was used to check for its cancellation. CM was confirmed, and it was distinguished from stimulus artifact by recording the response with the insert tube clamped.

Long latency responses

Based on the clinical records, the latency of long latency response (LLR) peaks recorded using a /da/ stimulus was noted. The stimulus was presented through insert phones (ER-3A), and the response was obtained using surface electrodes, placed on the vertex (Cz) (non-inverting), and over the mastoid of the test ear (inverting) and non-test ear (ground). The stimulus level was maintained at 90 dBnHL with a repetition rate of 1.1 stimuli/s. LLR recordings were then amplified 50,000 times, recorded in a 500 ms time window, averaged over 300 stimuli, and filtered at 1 Hz to 30 Hz. LLR recorded in the current study was not used to diagnose ANSD as it was not recommended by Starr et al. [7]. However, LLRs were used as a part of the test battery in previous studies [10, 18]. The summary of non-auditory and auditory characteristics profiled using the above audiological test battery is shown in Table 1.

Table 1 Summary of parameters assessed, for profiling individual's status

Statistical analyses

The raw scores of the retrospectively collected data were subjected to statistical analysis using IBM Statistical Package Social Sciences, version 20.0 (SPSS Inc., Chicago). Descriptive statistics (mean, median, standard deviation, and interquartile) were done for all the measures. Following this, the Shapiro–Wilk test of normality was administered. On the one hand, the group differences (if any) for categorical data (non-auditory and few auditory behavioral measures and physiological measures) were explored statistically using the chi-square (χ ²) test, which, when significant, were subjected to follow-up using residual analyses [28, 29]. On the other hand, for the quantitative/numerical data (dependent variable: PTA threshold SDT, LLR, and CM latency), a non-parametric Kruskal–Wallis test was conducted to investigate the group differences (independent variable). If a significant difference among these three groups was present in the Kruskal–Wallis test, the post hoc Dunn-Bonferroni test was used to analyze each inter-group significance.

Furthermore, Fisher’s linear discriminant function analysis (FLDA) was carried out for group classification based on the auditory behavioral data recorded from all the participants in the study. FLDA is a multivariate analysis technique that attempts to categorize groups based on measures obtained from the same set of variables [30]. The FLDA generates a mathematical operation (Di = a + b ₁ x ₁ + b ₂ x ₂ + … + b _n x _n; Di = predicted discriminant score; a = constant, x = predictor; and b = discriminant coefficient), based on weights generated for each test variable that maximizes the differences in performance for each group [31]. Based on the weights obtained in the FLDA, the best tests for group segregation were determined. In addition, the error rate for accurate group prediction by FLDA was calculated by performing a casewise tally of predicted membership (discriminant function) from the original (pre-determined) membership. The FLDA-derived group assignment was compared with the otherwise original membership of the individual to determine the overall error rate for group segregation.

Results

The Shapiro-Wilks test of normality carried out on auditory (behavioral and electrophysiological) across the three groups showed that data did not adhere to the normal distribution (p < 0.05). The results of the study are discussed under the following headings.

Non-auditory measures

The percentage of gender bias and laterality effect of each group showed a strong gender bias in the late-onset adult group, as indicated in Table 2. The females in the late-onset group were more affected than males by a ratio of 2.1: 1. Chi-square (χ ²) test revealed no association of groups with gender [χ ²(2) = 3.58, p = 0.17] and laterality [χ ²(2) = 6.44, p = 0.21].

Table 2 Effect of non-auditory factors on ANSD onset.The number count is tabulated, with the numerator representing the count and denominator showing sample size (no. of participants) for each group. The percentage for a given factor is shown in parentheses

Patient history and risk factor information

Patient history and risk factors for congenital ANSD participants in the study are shown in Table 3. A retrospective review of the patient records revealed that significant pre-, peri-, and post-natal history information was available for 47 patients out of 95 diagnosed with ANSD, which are only analyzed and tabulated in Table 3. It is important to note that 47 out of 95 patients have “no available history” in their medical records and not mistake it for “no significant history.” Overlap may be observed in percentage data (Table 3) as most of one or more risk factors were associated simultaneously in a patient.

Table 3 High-risk factors in three groups of participants with ANSD. The number count is tabulated, with the numerator representing the count and denominator showing the total number of participants for whom the data was available. The denominator in the last row represents the total number of participants in each group, while numerator stands for the participants in whom the data was not available. The percentage for a given risk factor is shown in parentheses

Behavioral audiological measures

The first level of analysis entailed comparing the group medians and variability separately for the behavioral measures: overall pure tone average (PTA), degree-, configuration-, symmetry of hearing loss, and speech measures.

Overall pure-tone hearing sensitivity and degree of hearing loss

The average pure-tone (PTA) hearing sensitivity of the adults with early-onset ANSD (degree: moderately severe) was comparable to the PTA of the congenital ANSD group, as shown in Fig. 1 (top panels). By contrast, the mean PTA of the adults with late-onset ANSD fell in a relatively lesser degree (moderate) of hearing loss. The Kruskal–Wallis test confirmed the significant main effect of the group [χ ²(2) = 52.04, p < 0.001]. Results of Dunn-Bonferroni post hoc comparisons revealed that there was no significant difference (p > 0.05) in the PTA of the former two groups (congenital ANSD and adults with early-onset ANSD). By contrast, the PTA of both these groups (congenital ANSD and adults with early-onset ANSD) was significantly higher (p < 0.001) than the adults with late-onset ANSD.

Fig. 1

Cluster plots showing median (center line) and interquartile range (error bars) along with individual (symbols) data for overall (top panels) and degree-wise (bottom panels) pure tone thresholds of the three participant groups. The-axis represents the hearing level, while the y-axis reflects the overall hearing sensitivity (top panels) or degree of hearing loss (bottom panels). The asterisks on the sides of the plot (top panels) denote paired comparisons along with the level of significance (***p < 0.001), while the numerals on the side of each error bar (bottom panels) represent the proportion (in %) of individuals within each group

The proportion of ANSD participants in each of the three groups across different degrees of hearing loss is depicted in Fig. 1 (bottom panels). The degree-wise distribution of participants in each group (Fig. 1, bottom panels) shows that 74.99% (45/60 ears) of the participants were clustered in a greater degree of hearing loss (moderately severe to profound) for the congenital group. A similar distribution (66.07% or 37/56 ears, exhibited moderate to a severe degree) was also seen in adults with early-onset ANSD. In striking contrast to the above two groups, most of the participants (70.90%, 44/62 ears) in the late-onset group were skewed towards lower degrees of loss (minimal to moderate).

Configuration of hearing loss

The congenital-ANSD group showed flat (86.66%) and falling configurations, predominantly demonstrating flat configuration. The adults with the early-onset ANSD also showed similar audiometric configurations as that of congenital ANSD, with the flat being the most predominant (60.71%) configuration in them too. On the other hand, of the participants in the late-onset ANSD group, only 40.32% exhibited flat configuration, while the other configurations (rising, tent-shaped, cookie-bite) were predominant in them. Approximately 50.00% of the late-onset adults showed other configurations. Another configuration seen in the adults with late-onset ANSD was Cookie-bite configuration (observed in 6.45%), which otherwise was absent in the other two ANSD groups. The distribution (%) of different audiogram configurations within each of the three groups is shown in Fig. 2.

Fig. 2

Distribution (%) of different audiogram configurations (a flat, b sloping, c rising, d tent-shaped, e inverted trough, f cookie bite) within each of the three groups. The x-axis represents frequency (kHz), while the y-axis reflects hearing sensitivity (dB HL). The horizontal error bars represent interquartile deviation (SD) with the median (indicated by symbols) in its center. The numerals on the side of each error bar represent the proportion (in %) of individuals within each group across configurations

On statistical inspection, the chi-square (χ ²) test revealed a significant association of groups with configurations [χ ²(10) = 51.33, p < 0.01]. The post hoc test of residuals done separately for each configuration substantiated the above observations of group similarities for congenital and adults with early-onset ANSD in at least two configurations: flat and sloping, as indicated in Table 4.

Table 4 Comparison of group differences across audiometric configurations

While group differences were apparent in descriptive data (Fig. 2) for flat and sloping configurations, only flat configurations withstood the statistical tests (Table 4). The flat audiometric configuration best segregated ANSD groups. The congenital ANSD demonstrated a significantly higher (p < 0.05) distribution of flat hearing configuration compared to the early onset adults. The latter, in turn, had shown higher distribution (p < 0.05) than the late-onset group. On the other hand, rising and inverted trough configurations seen only in adult groups highlighted a different trend. The rising configuration significantly (p < 0.01) demarcated both early and late-onset adult groups from the congenital ANSD group. By contrast, the inverted trough configuration helped differentiate the two adult groups, i.e., late-onset from early-onset adults.

Symmetry of hearing sensitivity

Among the 95 participants, pure tone thresholds were symmetrical in 56 participants, while the remaining 39 participants reported asymmetrical hearing sensitivity between the ears. When the group-wise distribution of the hearing thresholds was analyzed, 83.33% of congenital ANSD (25/30 participants), 46.66% of early-onset adults (14/30 participants), and 48.57% of late-onset adults (17/35 participants) exhibited symmetrical hearing sensitivity. The chi-square (χ ²) test revealed a significant association of groups with symmetry [χ ²(2) = 9.41, p = 0.009]. The follow-up test performed using the test of residuals showed no group differences for both symmetrical and asymmetrical hearing sensitivities.

Speech thresholds

Speech recognition threshold (SRT) for bisyllabic words in quiet were available for the adult participants, whereas speech detection thresholds (SDT) were obtained in children due to inadequate language skills. Figure 3 reflects the median scores of speech audiometry (SRT/SDT) results obtained from 95 participants (178 ears). Higher speech thresholds were recorded in the children with ANSD and adults with early-onset ANSD, which was in contrast to the lower speech thresholds recorded for the late-onset adult group.

Fig. 3

Box plots show the median (center line) and interquartile range along with individual (symbols) data for speech thresholds of the three participant groups. The box whiskers in the plot represent maximum and minimum speech thresholds in data. The asterisk denoted beside the box plot represents the significance level of paired comparisons (***p < 0.001)

The Kruskal–Wallis test statistically confirmed these group differences as it revealed a significant main effect of the group [χ ²(2) = 44.35, p < 0.001] on speech thresholds. The follow-up test using Dunn-Bonferonni pairwise comparisons showed that the speech thresholds of adults with late-onset were significantly lower (p < 0.001) than adults with early onset and congenital ANSD groups. By contrast, the latter two groups showed no statistical difference (p > 0.05) in their speech thresholds.

Relative contributions of behavioral audiological measures in group segregation

The second approach to data analysis was to identify a selected set of auditory behavioral measures that most effectively distinguished the groups based on the parameters that can be tracked (PTA, degree, configuration, and SRT/SDT scores) on all the participants of the study. This was not carried out for the physiological and electrophysiological data, as OAE was present and acoustic reflexes/ABR was absent. LLR was not recorded on all the participants. Two canonical discriminant functions (DF) were generated using discriminant function analysis. The DF, which was statistically significant best clustered behavioral measures that segregated ANSD groups. Out of the two functions, DF1 was significant (Wilks lambda, λ(8) = 0.66, χ ² ₌ 71.24, p < 0.001), accounting for 97.8% of the cumulative variance. However, an examination of the weights for each test indicated that PTA and configuration were heavily weighed (canonical coefficients) on DF1, as reflected in Table 5, indicative of the higher sensitivity of these measures in predicting group membership. The corresponding structure matrix, shown in Table 6, indicates the pooled within-group correlations between discriminating variables and standardized canonical DFs. The canonical DFs obtained in the study based on the weights (Table 5) are summarized below:

Table 5 Contribution (weights) of behavioral auditory tests for group membership prediction on auditory tests

Table 6 Structure matrix of the canonical discriminant function coefficients

• DF1: (1.06 × PTA) – (0.44 × configuration) – (0.39 × degree) + (0.27 × SRT/SDT).

• DF2: (3.62 × degree) – (3.48 × PTA) + (0.31 × configuartion) + (0.21 × SRT/SDT).

Each participant’s score on the discriminant function was calculated by multiplying the standardized canonical DF coefficient by the test score of each individual on the four associated measures and summing these products. Thus, calculated individual and mean scores for each group (group centroids) on the two discriminant functions are shown in Fig. 4. It is clear from Fig. 4 that the DF1 separates the childhood and early-onset ANSD from the late-onset ANSD group, which emerged as two distinct clusters that are concentrated on either side of the reference line (Fig. 4, dotted line).

Fig. 4

Grouping participants based on canonical discriminant scores derived for behavioral auditory measures. The reference line (dotted) indicates the cutoff scores on function 1 for group segregation. The green circles, red triangles, and blue diamonds represent individual participant scores for each group (see legend), while squares of the corresponding colors represent the group centroid scores

The error rate in the DF analysis (indicating the accuracy of classification) was carried out by comparing case-wise statistics of participant’s DF scores against their original pre-verified condition, as shown in Table 7. An overall correct classification rate of 60.10% was observed. Group-wise analysis of the group membership errors shows that adults with the late-onset ANSD were the most correctly (72.68%) classified group. On the other hand, the group membership predictivity of the early-onset ANSD was the poorest, with approximately 50% (i.e., 48.21%, less than chance) of overall errors. The majority of errors in the classification (41.90%) occurred between the congenital onset and adults with early onset. Errors in classification involved predicting group membership for 28.57% (16/56 ears) of adults with early-onset ANSD who were instead classified as the congenital onset. On the other hand, 23.33% (14/60 ears) of congenital ANSD were misclassified into the early-onset ANSD adult group., indicative of the overlap in auditory characteristics between the two groups. In striking contrast, the group prediction error was minimal for adults with late-onset ANSD, with 72.68% of the group’s participants assigned to the same target group on FLDA analyses.

Table 7 Accuracy of discriminant function analyses comparing predicted and original group memberships. Total ears (number count) is tabulated with the corresponding percentage in parentheses

Physiological measures: Immittance, otoacoustic emissions and middle ear muscle reflexes

As part of the routine audiological test battery, all the physiological tests, namely tympanograms, acoustic middle ear muscle reflex (MEMR), and the OAEs, were recorded on all participants with ANSD. “A” type tympanogram was obtained in all the participants, indicative of normal middle ear functioning. The MEMRs were absent for all frequencies and conditions for all the participants in all the ears tested. The analyses of OAE data collected from 178 ears revealed the presence of OAEs in approximately 93.26% of total ears, whereas 6.74% reported partial or absent OAE responses. However, in these 6.74% ears in whom OAEs were partial/absent, cochlear microphonic was present. The presence of cochlear microphonic and abnormal ABR was used in these participants (in whom OAEs are otherwise partially/totally absent) to diagnose ANSD. The group-wise distribution of OAEs in the study is given in Table 8. Chi-square test also reflected no association of group with OAE responses, i.e., its presence or absence [χ ²(2) = 4.10, p > 0.05]. Test of residuals also substantiated no statistical differences (p > 0.05) in OAE responses between groups.

Table 8 Distribution (number of ears) of OAEs across the three ANSD groups. The percentage corresponding to the number count is given in parentheses

Electrophysiological audiological measures

Auditory brainstem response

ABRs were absent for all the 60 ears in children with ANSD, while for the adults with early-onset ANSD and late-onset ANSD. ABR was absent in all but one ear (55/56) for the former group and all but two ears (60/62) in the latter group. For the individuals where ABR was present, it was delayed in latency. The chi-square test showed no significant association of groups with ABR presence [χ ²(2) = 7.95, p > 0.05].

Cochlear microphonic

Cochlear microphonic (CM) was recorded in a total of only 30 ears. Out of these, the maximum number was recordable in the congenital group (60.00% or 36/60 ears), followed by adults with late-onset (23.33% or 14/62 ears) and adults with early-onset (16.7% or 10/56 ears). The chi-square test revealed a significant association of groups with CM presence [χ ²(2) = 28.29, p < 0.001]. Test of residuals showed that CM recorded from congenital ANSD was significantly higher (p < 0.001) in frequency than the other two adult groups. The analyses of CM latencies (in participants whom CM was recordable) compared using Kruskal–Wallis test revealed significant main effect of the group [χ ²(2) = 13.26, p < 0.001], which when followed by pairwise Bonferroni Dunn’s test revealed that congenital ANSD group (mean latency: 2.11 ± 0.45 ms; median: 2.09 ± 0.00 ms) and adults with early-onset ANSD (mean latency: 2.86 ± 0.90 ms; median: 2.87 ± 1.00 ms) exhibited statistically earlier (p < 0.01) CM than those recorded in late-onset adults (mean latency: 2.97 ± 0.88 ms; median: 3.00 ± 0.5 ms), with the former two groups showing statistically similar (p > 0.05) CM latencies.

Late latency response

LLRs were recorded in only 65 ears in the database, which amounts to approximately 36.52% (65/178 ears). The group-wise data on the percent distribution of recordable LLR peaks (P1, N1, P2, and N2) within each group is shown in Fig. 5.

Fig. 5

Box plots showing median (center line) and inter-quartile range along with individual (symbols) data for the LLR peak latencies of (a) P1, (b) N1, (c) P2, and (d) N2 for the three participant groups. The box whiskers in the plot represent maximum and minimum latency values in data. The numerals on the side of each whisker represent the proportion (in %) of individuals within each group in whom LLR peaks were recorded

It is evident that the most robust LLR peaks (P1, N1, P2) are recordable in a high proportion (approx. 65–70%, 42–43/62 ears) of adults with late-onset ANSD, although the other adult group with early-onset demonstrated considerably lesser frequency (14.28%, 8/56 ears) of recordable LLR peaks. On the other hand, the congenital ANSD group exhibited a relatively higher proportion (22.66%, 16/60 ears) of recordable LLR peaks compared to adults with early-onset ANSD. Data on the distribution of LLR presence across three groups analyzed using chi-square test revealed significant association of the group with LLR peaks [P1: χ ²(2) = 37.90, p < 0.001; N1: χ ²(2) = 42.66, p < 0.001; P2: χ ²(2) = 40.25, p < 0.001; N2: χ ²(2) = 10.74, p < 0.01]. Test of residuals showed that LLR recorded from adults with late-onset was significantly higher (p < 0.001) in frequency than adults with early-onset, who in turn had marked significantly greater presence (p < 0.05) of P1, N1, and P2 peaks compared to congenital ANSD group. However, the N2 was only prominent for adults with late-onset ANSD (16.13%), who demonstrated a higher occurrence of N2 than the other two groups. Kruskal–Wallis test revealed significant main effect of group [P1: χ ² (2) = 11.79, p < 0.01; N1: χ ²(2) = 12.65, p < 0.01; P2: χ ²(2) = 11.89, p < 0.01; N2: χ ²(2) = 10.74, p < 0.05] on LLR latencies. The pairwise Bonferroni Dunn’s test revealed that adults with late-onset exhibited significantly earlier peak latencies (p < 0.01) compared to adults with early-onset ANSD and congenital ANSD (Fig. 5). The latter two groups showed statistically similar (p > 0.05) delayed LLR latencies (relative to the adults with late onset).

Discussion

The study aimed to profile the auditory and non-auditory differences between late-onset and early-onset ANSD. A respective data analysis of 95 participants (178 ears) belonging to three groups (congenital ANSD, adults with early onset, and adults with late onset) highlighted differences in both non-auditory and auditory characteristics across the groups. It is important to note that the absence or poor implementation of universal newborn screening in developing or under-developed countries can result in late diagnosis of ANSD, though the individual may manifest hearing-related deficits in the childhood itself. There is also a scope of misdiagnosing late-reported congenital patients as late-onset ANSD, due to the lack of early detection facilities in underdeveloped countries. In such cases, a detailed speech and language evaluation can assist in differentiating late-onset and late-diagnosed patients with ANSD, wherein language abilities in the latter group will be lesser than their chronological age. The effect of ANSD onset on the non-auditory and auditory features is discussed as follows:

Effect of ANSD onset on non-auditory characteristics

Gender effect

The study reported a female-to-male ratio of 1.14:1, 1.06:1, and 2.11:1 for the congenital ANSD, adults with early-onset, and adults with late-onset ANSD, respectively. The gender factor can therefore be used to delineate the early-onset with late-onset ANSD groups, with females showing significantly higher ANSD prevalence rates compared to males. Many studies demonstrated gender differences in ANSD prevalence among adolescents and adults in the Indian population [6, 9, 32]. While Kumar and Jayaram [6] reported a female-to-male ratio of 2:1, Prabhu et al. [13] reported a ratio of 2.33:1. Similarly, Jijo and Yathiraj [32] reported a ratio of 1.66:1 for females: males in their study on 64 ANSD participants. These reports were similar to those obtained in the study showing females are affected more than males in adult groups with late onset.

Contrary to these findings in the late-onset group, no such gender difference was seen in the early-onset groups (Table 1). This finding is in consensus with the literature on ANSD children by Berlin et al. [1] and Sininger and Oba [5]. They found the distribution across the two genders to be almost similar. The more significant number of female ANSD patients and the onset of the disorder during adolescence/puberty can be linked to the X-linked dominant inheritance pattern of AIFM1-related gene mutation, affecting females [33]. The females with the dominant trait of AIFM1gene were reported to have no hearing problems in their childhood and expressed onset of ANSD in later childhood or adolescence (6 to 20 years). The role of hormonal influence during puberty in females can also be hypothesized as a likely etiology for late-onset ANSD, which needs to be explored further.

Laterality effect

The study results show that almost 98–100% of early-onset ANSD groups (97.77% adult-early onset and 100% congenital onset) had bilateral ANSD, and only 3.33% in adult-early onset had unilateral pathology. By contrast, the unilateral condition was slightly higher in the adult-late onset group (Table 1); although in general, the bilateral cases were the majority in this group, as well. The literature review also shows only a few case reports of unilateral ANSD [34,35,36,37], with the majority of research documents showing bilateral cases. Berlin et al. [1] multi-centered study on ANSD with 260 patients documents only 7.3% of unilateral cases, while 92.7% of patients demonstrated bilateral ANSD.

High-risk factors and etiology. Based on the available information in the present study on high-risk factors, the significant factors in early-onset groups (congenital and adults with early-onset) were linked to the family history of hearing loss and consanguinity (Table 2). In addition, the congenital group exhibited hyperbilirubinemia and pre-mature delivery also as significant risk factors. Literature evidence points to case reports of patients with early-onset were associated with hyperbilirubinemia, ototoxic drug regimen, low birth weight, low APGAR scores, anoxia, and family history of deafness [38]. However, these risk factors were not associated with late-onset ANSD in the study, who did not have any pre-, peri, or post-natal causes. Instead, some predisposing factors such as exposure to drugs and epileptic attacks were seen in these individuals. Manchaiah et al. [15] reported genetic markers linked to non-syndromic ANSD. They reported AUNA1 and PCDH9 as the gene markers for the autosomal dominant type of inheritance, OTOF, DFN895, and GJB2 for autosomal recessive inheritance, and AUNX1 for the X-linked type of inheritance. They have pointed out the importance of genetic screening for early identification of the disorder and deciding the best treatment option. The detected mutation can help identify the pathology site, for instance, if the mutation is in the otoferlin gene, which encodes protein Otoferlin which is, in turn, expressed in inner hair cells. This inner hair cell damage can be managed better with a cochlear implant than hearing aids [39].

Prabhu, Avilala, and Manjula [9] reported that late-onset ANSD cases could be caused due to exposure to toxic chemicals (pesticides), low socioeconomic status, and hormonal variations. Along similar lines, Draper and Bamiou [40] noted Xylene exposure as a cause for late-onset ANSD in their case study. The literature on late-onset ANSD cites associated neuropathies such as optic atrophy, sensorimotor neuropathy, and other peripheral neuropathies as possible etiologies for the disorder. In addition, the available information in the present study also revealed 40.00% involvement of family history as a genetic cause for late-onset ANSD. Among the late-onset cases, AIFM1 is reported to be the most common genetic cause of all noninfant-onset ANSD cases [41]. Thus, probing into the risk factors and etiology of the disorder can help the understanding of the underlying pathogenic mechanisms, which can differentially affect the onset of the disorder. However, the authors strongly suggest that readers exercise caution in this regard, as data from only 47 participants (out of 95) in total was extractable retrospectively from medical records. The missing information of the other 48 participants in medical records could be associated with factors such as lack of collection (e.g., the patient was never asked about the risk-factors) or lack of documentation (e.g., the patient was asked about associated risk-factor, but the response was never recorded in the medical record).

Effect of ANSD onset on auditory characteristics

Overall pure-tone hearing sensitivity and degree of hearing loss

The findings from the study showed that children and adults with early-onset ANSD showed a higher degree of hearing loss compared to the adult-late onset group. The lesser degree of hearing loss (mild to a moderate degree) noted in the late-onset ANSD group can be an indication of compromised neural functioning in the presence of preserved inner hair cell (IHC) functioning [42]. The delayed onset could have spared the functional inner hair cells, which primarily relay the sound to the auditory nerve. This is reflected in its clinical correlate, i.e., better auditory thresholds. On the other hand, the site of lesion in early-onset ANSD groups (both in congenital children and adults) is most probably the inner hair cell loss, reflected in the pure tone audiometry as severe to profound hearing loss [43]. It is well reported in the literature that mutations of OTOF lead to a reduction of synaptic vesicle exocytosis at ribbon synapse and damage the inner hair cell [44]. Most patients with OTOF mutation have profound hearing loss since inner hair cells are damaged [45]. Thus, these findings highlight the possibility of neuronal impairment in the late-onset group while the early-onset groups predominantly have overall IHC damage.

Configuration of hearing loss

The configuration of hearing loss was flat in the congenital ANSD group and adults with early-onset ANSD. By contrast, other configurations such as rising, tent-shaped, and cookie bite were more prominent in the late-onset ANSD group. It is reported in the literature that low-frequency cues are coded by the phase-locked responses of the auditory nerve fibers. ANSD in the auditory nerve affects the phase-locking abilities and can result in a predominantly low-frequency hearing loss [46]. Thus, it can be hypothesized that late-onset ANSD typically affects the auditory nerve leading to poor phase-locking abilities resulting in a rising audiogram configuration. However, a flat audiogram configuration in the congenital-ANSD group and adults with early-onset ANSD suggests a lesion more at the level of IHC or synapse leading to hearing loss at all the frequencies.

In addition, the cookie bite and tent-shaped audiogram configuration can be attributed to the length of the auditory nerve fibers [47]. The higher degree of loss at low frequencies is attributed to a longer course of low-frequency auditory fibers [5, 8]. The auditory nerve fibers are the shortest, which codes for the mid frequencies and a thus lesser degree of loss at these frequencies [48]. This can explain the different audiogram patterns obtained in the late-onset ANSD group suggestive of auditory nerve damage. However, more recent studies are essential to substantiate the claim.

OAEs and reflexes

OAEs were present in 93.26% of the population considered for the study, indicating normal outer hair cell functioning. Thus, OAE helps to diagnose neural pathology from a cochlear disorder differentially. The presence of OAE with absent ABR indicates a possible ANSD [4]. It was also suggested that OAE is lost in a few patients over time [4], and hence, CM was used along with ABR to diagnose such patients.

The middle-ear muscle reflexes were found to be absent in all the participants in all three groups. Similar results are reported in the literature where absent middle ear reflexes are reported in individuals with ANSD [1, 4, 6, 10]. The dyssynchronous neural transduction is reported to be the reason for absent reflexes in individuals with ANSD [10]. The poor neural conduction of the sound may fail to activate the stapedius muscle, which leads to the absence of acoustic reflexes in the ANSD population.

Auditory brainstem response and long latency response

The results of the study showed that ABR was absent in all the participants considered in the study. The above result is consistent with previous studies on ANSD, which report absent ABR due to dyssynchronous firing of the auditory nerve [4, 10]. The results of the study indicated that the LLR responses were recorded in a higher proportion in adults with late-onset ANSD. In addition, the LLR latencies were relatively shorter in the late-onset ANSD group compared to the other two groups. The presence of LLR with absent ABR indicates that the dys-synchrony is more localized to the auditory nerve, and the cortical structures are spared. This indicates a lesser severity of ANSD in those with the presence of LLR [10]. Thus, it can be assumed that adults with late-onset ANSD have relatively lesser processing issues compared to the congenital ANSD group. ANSD could be at the early stage affecting only the auditory nerve in the late-onset ANSD group. However, in the congenital ANSD group, there could be dyssynchronous neural firing both at the neural and the cortical level.

The results of the study also showed a higher presence of N2 in adults with late-onset ANSD compared to adults with congenital ANSD. It is well reported in the literature that the amplitude of N2 is dependent on the duration of hearing loss [49]. This result is supported by Yuvaraj and Jayaram [18], who reported that N2 amplitude reduces to a lower representation of acoustic features, which is essential for generating the potential. Thus, it can be speculated that adults with congenital ANSD have relatively more poor representation of acoustic cues leading to the absence of N2 potential. Hence, it is clear that LLR reflects different neural synchrony compared to ABR [50,51,52], and it is less affected in adults with late-onset ANSD.

In addition to establishing group differences through inferential statistics (chi-square, Kruskal–Wallis), the presence of group differences and identification of best predictors measuring such differences were achieved using Fisher’s linear discriminant function analysis (FLDA, Fig. 4). The results of FLDA showed that errors in group prediction were maximum for the early-onset ANSD adult group (Table 7), whose group predicted membership overlapped considerably with the congenital-onset group. This finding indicated that many ANSD individuals in the early-onset adult group could indeed be a sub-set of the congenital ANSD group, comprising neonates and children who were missed in school screening/newborn hearing screening. This finding is supported by inadequate and under-utilization of universal newborn hearing in developing nations like India [53, 54], where the present study was conducted. Gupta et al. [54] report that significant neonates screened in a tertiary care center of their research were lost due to challenges in follow-up, limited public awareness, and inappropriate rehabilitation services.

In contrast to the early-onset ANSD adult group, the group membership error for the late-onset ANSD was minimal (Table 7), with 72.68% being correctly categorized. In conjunction with the previous one (confusion in early-onset ANSD group membership), this finding strongly suggests the actual existence of the late-onset group, in whom the ANSD characteristics emerge during or after post-pubertal age. In addition to strengthening the finding of the actuality of late-onset ANSD presence, the FLDA analyses also revealed that PTA and configuration were the best predictors of the group differences. This adds higher diagnostic value to the PTA and configuration. Their manifestations can be used as a valuable tool to verify the onset of the disorder, which is often retrospectively reported by the patient (especially in early-onset ANSD). While adults with late-onset have a lesser degree (mild to moderately severe, Fig. 1) of hearing loss and a rising audiometric configuration (Fig. 2), the early onset ANSD and congenital groups often demonstrate a greater degree of hearing loss (severe to profound, Fig. 2) and flat audiometric configuration (Fig. 3). The presence of such indicators in pure-tone audiometry should alert audiologists to reflect on the possible onset of the disorder, which in turn can facilitate their rehabilitation choice. While applications of cochlear implants in early onset may be advisable, the utility of hearing aids or assistive listening in late-onset groups can be advocated as the first line of rehabilitation.

As the study involved retrospective analyses of information from the medical reports, many factors emerged as challenges for the direct interpretation of study findings. In the present study, non-audiological factors such as laterality and gender were considered. Future studies can explore the influence of other non-audiological factors such as the presence or absence of peripheral neuropathy of other senses and vestibular symptoms in the differential diagnosis of onset-based group differences in ANSD. Thus, future prospective studies and longitudinal studies on ANSD are warranted in the direction of entailing such group differences.

The authors would like to acknowledge such limitations in factors like linking pathophysiology with onset, experimenter bias in marking peaks of electrophysiological responses, and correlating onset-based audiological findings with findings related to domains (speech and language deficits). Recent studies suggest that neither audiogram configuration nor word recognition scores show a significant correlation with cellular damage in the cochlea [55]. Lobarinas, Salvi, and Ding [56], in their experiments in chinchillas, have demonstrated that an extensive loss of inner hair cells (exceeding 80%) produces only minimal effects on audiometric thresholds. Hence, the correlation between the degree of hearing loss and loss of inner hair cells should be further verified using minimally invasive tools such as cochlear endoscopy. Since the study was retrospective, there could be variations in the audiological testing and ABR/LLR waveform identification and marking of the peaks between the examiners. However, since the clinic used a standardized protocol for audiological evaluation, it might have reduced the inter-tester variability. In addition, the participants may have neuropathies of other senses, too, which is not explored in detail since it was a retrospective study. Future studies can also explore longitudinally if the prolongation of P1 latency of LLR in congenital ANSD is due to a maturational effect or is it pathological in nature. In addition, the current study also warrants differential rehabilitative options for the congenital and late-onset ANSD groups. Future studies can explore differences in benefits derived from amplification devices such as hearing aids or cochlear implants between these onset-based ANSD groups.

Conclusions

The present study attempted to differentiate congenital and late-onset ANSD based on audiological and non-audiological characteristics. The study results clearly show differences in audiological profiles, which can be used as possible indicators to separate congenital and late-onset ANSD. The degree of hearing loss (pure tone average) and configuration are good predictors that distinguish between early-onset groups (congenital and early-onset adult ANSD) from those with late-onset ANSD. While the early-onset ANSD groups showed severe to profound hearing loss with a flat configuration, the late-onset group showed lower degrees of hearing loss with a raising or cookie-bite configuration.