Cochlear Implantation in Children with Labyrinthine Anomalies and Cochlear Nerve Deficiency: Implications for Auditory Brainstem Implantation

Transcription

1 The Laryngoscope VC 2011 The American Laryngological, Rhinological and Otological Society, Inc. Cochlear Implantation in Children with Labyrinthine Anomalies and Cochlear Nerve Deficiency: Implications for Auditory Brainstem Implantation Craig A. Buchman, MD; Holly F. B. Teagle, AuD; Patricia A. Roush, AuD; Lisa R. Park, AuD; Debora Hatch, AuD; Jennifer Woodard, AuD; Carlton Zdanski, MD; Oliver F. Adunka, MD Objectives/Hypothesis: Compare outcomes among children with inner ear malformations and/or cochlear nerve deficiency (CND) who have received a cochlear implant (CI). Study Design: Individual retrospective cohort study from 1993 to Methods: A select cohort of 76 children was identified. Imaging characteristics, operative findings, complications, mapping parameters, and performance were assessed. Comparisons among the different groups were undertaken. Results: Surgery was mostly uncomplicated. Nearly all children demonstrated behavioral responses to CI stimulation irrespective of inner ear morphology or the presence of CND. Children with CND had higher pure tone averages (PTAs) and required greater charge for stimulation than other malformation types. Open-set speech perception was achieved in 100% of children with incomplete partition-enlarged vestibular aqueduct (IP-EVA), 50% of those with hypoplastic malformations, and 19% of CND cases. Robust responses on eighth nerve compound action potential (ECAP) testing through the implant was associated with higher levels of speech perception. Manually supplemented communication strategies were more common among children with hypoplastic malformations (69%) and CND (95%) than those with IP-EVA (18%). Conclusions: Children with IP-EVA malformations have an excellent prognosis for developing open-set speech perception and using oral communication modes following CI. On the contrary, children with severe malformations or CND may have elevated charge requirements for attaining sound detection alone. These children s prognosis for obtaining open-set speech understanding, using exclusive oral communication, and participating in mainstream education is more limited. These findings have important implications for considering alternative forms of intervention such as auditory brainstem implantation and/or supplementation with visually based communication strategies. Key Words: Cochlear implant, cochlear nerve deficiency, inner ear malformation, auditory brainstem implant. Level of Evidence: 2b Individual retrospective cohort study. Laryngoscope, 121: , 2011 INTRODUCTION Developmental labyrinthine malformations are a relatively common finding among children with sensorineural hearing loss with a prevalence as high as 40%. 1,2 Cochlear nerve deficiency (CND) refers to a small or absent auditory nerve on high-resolution imaging and From the Department of Otolaryngology Head and Neck Surgery, University of North Carolina at Chapel Hill, Chapel Hill, North Carolina, U.S.A. Editor s Note: This Manuscript was accepted for publication May 8, Presented at the 114th Annual Meeting of the Triological Society, Chicago, IL, April 29, Craig A. Buchman serves as a surgical advisory board member for Advanced Bionics, Anspach, Cochlear, and MedEL Corporations, as well as receives funding from these companies for unrelated research. Holly F.B. Teagle serves as an audiology advisory board member for Advanced Bionics, Cochlear, and MedEL. Patricia A. Roush serves as an advisory board member for Phonak Corporation. Oliver F. Adunka receives research funding from Advanced Bionics, Cochlear, and MedEL Corporations for unrelated laboratory research. The remaining authors have no disclosures. The authors have no conflicts of interest to declare. Send correspondence to Dr. Craig A. Buchman, G190 Physicians Office Building, 170 Manning Drive, Department of Otolaryngology - Head & Neck Surgery, University of North Carolina at Chapel Hill, Chapel Hill, NC buchman@med.unc.edu DOI: /lary has been identified among patients with normal inner ear morphology as well as those with inner ear malformations, narrow internal auditory canals, and/or electrophysiologic characteristics of auditory neuropathy spectrum disorder. 3 6 Among cochlear implant (CI) candidates, the presence of either CND or an inner ear malformation has the potential to adversely affect outcomes by reducing both the organization and availability of neural populations to stimulate electrically. A variety of inner ear malformations have been previously described among patients with sensorineural hearing loss. Although numerous reports have focused on the surgical feasibility of CIs in these patients, only limited data exist regarding outcomes. As an example, the largest series of implanted children with enlarged vestibular aqueducts (EVA) currently numbers only In this important study, only 6 children have measurable open-set speech perception data available for analysis. With such small numbers, very few investigators have been able to make comparisons among the various malformations with regard to prognosis Moreover, only recently have reports of CI among children with CND been reported with generally poor results, although exceptions exist. 11,12 The present investigation was 1979

2 undertaken to detail the audiological outcomes of a large group of children with various types of inner ear malformations or CND that have received a CI. Comparing anatomic parameters and associated outcomes of these children reveals clear differences in sound awareness, speech perception abilities, and the mapping parameters necessary to create these results. The findings have significant implications for counseling prognosis, choosing communication modalities, and identifying those children that might be considered candidates for auditory brainstem implantation (ABI). MATERIALS AND METHODS Subjects This study is part of a retrospective review of over 900 children who have received over 1,000 cochlear implants at the study institution. The Biomedical Institutional Review Board approved this study. Between 1993 and 2010, 100 (11%) children with documented inner ear malformations and/or CND have received CIs at the study institution. Seventy-six of these had high-resolution imaging studies available for review. Before implantation, all children had severe to profound hearing loss and failed to make appropriate progress during a hearing-aid trial. Most children had participated in an intensive auditory-based intervention program and adopted manual communication modes only after unsuccessful attempts at auditory-only approaches following CI. Preoperative imaging characteristics, intraoperative findings, complications, mapping parameters, and performance are presented. Imaging Inner ear malformations and CND were diagnosed preoperatively based on the results of high-resolution magnetic resonance imaging (MRI), computed tomography (CT), or both. The imaging protocols and our classification scheme for the various malformations including CND have been detailed previously and are generally similar to those previously described. 8,9,13,14 In general, cochlear structure can be normal, absent (Michel malformation), cystic, incompletely partitioned (IP), or hypoplastic. IP refers to normal cochlear external dimensions with decreased or absent partitioning due to either absent internal divisions (type I); apical fusion with a present interscalar septum (type II Mondini type when EVA present); or X-linked stapes deafness type (type III). Hypoplastic cochleas have smaller external dimensions and are also subclassified as: type I bud-like; II cystic hypoplasia; and III cochlea with less than two turns. The vestibular aqueduct or endolymphatic duct and sac can be enlarged (EVA or LVA) in isolation or in association with cochlear IP. 15 This group is referred to as the IP EVA spectrum. If all cochlear partitioning was absent but there was differentiation into cochlear and vestibular labyrinthine compartments, this was termed a cystic cochleovestibular malformation (CCVA). 16 Common cavity (CC) was defined as no internal differentiation of the labyrinth into either a vestibular or cochlear compartment. Cochlear Implant Surgery A transmastoid facial recess approach with facial nerve monitoring was used for all surgeries. The cochlear vestige was implanted in all instances except for CC cases where a labyrinthotomy approach was used. In some revision cases, blind sac closure of the ear canal was undertaken where the ear canal obscured visualization or electrode extrusion or misplacement in to the ear canal was evident from a previous surgery performed elsewhere. In most cases, a round window-related cochleostomy was created. Electrode choice varied according to the malformation type and device availability at the time of implantation. In general, when a cochlear interscalar septum was lacking, modiolar conforming electrodes were avoided. Cerebrospinal fluid gushers from the labyrinth were managed intraoperatively by creating a larger cochlear opening and tightly packing the cochleostomy with connective tissue around the electrode. Cerebrospinal fluid diversion was not needed in any child in the series either intra- or postoperatively. Electrode array insertion was carried out to the point of first significant resistance. The size of the cochlea on preoperative imaging dictated the expected insertion depth for the array. When discrepancy between observed and expected insertion depth arose, the array was either redirected or imaging was undertaken. Following electrode insertion, intraoperative transorbital X-ray confirmed expected electrode coiling and impedances and eighth nerve compound action potential (ECAP) were carried out in most cases provided that this technology was available at the time. For the current study, two of the three available CI systems enabled the measurement of ECAPs using the commercial software (NRT 3.1VR and AutoNRTVR from Cochlear Corp, Englewood, CO, or NRIVR from Advanced Bionics Corp, Sylmar, CA). Intraoperative ECAP responses are collected remotely (via hospital intranet) using high stimulation levels to obtain recordings of waveforms quickly. Postoperatively, parameters are manipulated to achieve recording of the best waveform possible without causing discomfort. For most children, data were collected on at least four electrode pairs. Cochlear Implant Mapping Once implanted, determining stimulation parameters for each child was a dynamic process. Due to the presence of additional developmental delays and comorbidities in many children, speech processor mapping was often challenging. The potential number of electrodes to activate was established in reference to the surgeons report, intraoperative X-rays, and behavioral responses. Current level settings were based on a combination of behavioral and ECAP responses, if present. A conservative approach to programming was taken with current levels set low at the initial stimulation and increased with time and experience, and patient s tolerance. Given the knowledge of compromised labyrinthine anatomy, slower rates of stimulation were often used to account for possibly longer neural refractory periods and to enhance possibilities of stimulation. Usually a stable map could be achieved by 3 to 6 months after initial stimulation. In some cases when there was a lack of stimulation or very limited benefit, nonuse of the implant resulted either because the child rejected the device or the parents elected to discontinue device use. Behavioral Audiometry and Speech Perception Testing During programming sessions, observation and conditioned behavioral audiometry techniques were used to determine electrical thresholds and comfortable listening levels. When stimulation levels resulted in visible or painful facial contractions, higher levels were generally not attempted. After device fitting, speech perception assessment was attempted with children at 6-month intervals when possible. Standard tests included a hierarchical battery of measures as previously described. 17 Parental interview included the Infant Toddler 1980

3 Meaningful Auditory Integration Scale (IT-MAIS) or Meaningful Auditory Integration Scale (MAIS). Closed-set testing using the Early Speech Perception Test (ESP standard or ESP-low verbal) was attempted when the child was developmentally able to participate. Open-set measures for this report included the Phonetically Balanced Kindergarten word lists (PB-k words) and the Consonant Nucleus Consonant (CNC) word lists. The PB-k lists were administered when a child had achieved a score significantly above chance on closed-set tests. CNC words were given once a child scored above 80% correct on the PB-k word lists. Tests are administered in a sound controlled environment using live-voice or recorded stimuli presented at 60 db SPL. Due to age and attention restrictions, it was generally necessary to use live voice presentations for younger children to complete the tests. Communication mode and educational setting were identified and recorded by the managing audiologist and/or speech pathologist at our institution. Data and Analysis For the purposes of the analysis, preoperative audiologic data were used from the testing interval(s) immediately before surgery. In this analysis, no measureable response for threshold testing was considered to be 120 db HL. PTA was calculated using the thresholds from 500, 1,000, and 2,000 Hz. Performance metrics following surgery were chosen from the patient s most recent visit for speech perception testing. When speech perception testing was limited by short duration experience or young age, inattention, or the presence of significant comorbidities, these reasons were recorded. When inability to complete the test was evident despite significant implant experience, the patient was assigned the value for chance correct responses for the appropriate test (i.e., 0% for open-set tests or 33% for ESP pattern-[p] perception tests). For speech perception results, a modified cumulative speech perception index in quiet (SRI-Q) was created after Wang et al. 17 (2009) in an effort to simultaneously display the data from all tests in the speech recognition hierarchy as a function of time. ESP-P was considered easier than ESP-monosyllabic (M) words. PB-k word scores were then followed by CNC word scores. The measured percentage score for each test (0% 100% correct) was then hierarchically ranked and stacked in 100-point increments. Data for the best score on the hardest test are displayed visually as a function of duration of implant use to show peak performance. For charge analyses, formulas and conversion spreadsheets provided by the implant manufacturers were used to convert programming levels into units of charge (nc) per unit phase. This was determined from the middle electrode contact in the set of active electrodes for each child s map from the time speech perception testing was completed. 18 ECAP responses were characterized as absent, present but atypical, or robust. Absent responses were those where no repeatable waveforms were observed. Atypical waveforms were those that were present, but waveforms did not have a robust N1P2 pattern. Often this included some dysmorphic waveforms that were difficult to mark; in all cases of atypical ECAP, behavioral responses from the children indicated stimulus audibility. When waveforms were not present or atypical in morphology intraoperatively, the same was true in subsequent mapping sessions despite attempts to maximize the response with parameter manipulations. Descriptive statistics included frequency and proportion calculations for the categoric variables of interest and means, ranges, and standard deviations (SD) for the continuous variables. Where appropriate, individual t-tests and Fisher s Exact test were used for significance testing of the potential effects. The SRI-Q was not managed as a continuous variable due to TABLE I. Demographics of Implanted Children with Inner Ear Malformations and/or Cochlear Nerve Deficiency (CND). Variable Non-CND (%)* CND (%) Totals (%) Number of children 54 (71) 22 (29) 76 (100) Gender Female 25 (46) 8 (36) 33 (43) Male 29 (54) 14 (64) 43 (57) Syndrome or medical condition 26 (48) 15 (68) 41 (54) Failed newborn hearing screen Pass 3 (6) 1 (5) 4 (5) Refer Unilateral 4 (7) 1 (5) 5 (7) Bilateral 21 (39) 15 (68) 36 (47) Unknown 26 (48) 5 (22) 31 (41) Preop behavioral audiometry Mean unaided SAT (SD) 90 (17) 99 (20) 92 (20) Mean aided SAT (SD) 50 (38) 71 (27) 55 (36) *All patients with inner ear malformations that do not have cochlear nerve deficiency (CND). Branchio-oto-renal (BOR) S. (2), CHARGE (7), chromosome anomaly (7, 8, 13, 21), congenital hydrocephalus (3), Cornelia de Lange S. (1), isolated cerebral palsy (1), craniostenosis (1), Dwayne S. (1), ectrodactyly (1), isolated facial paralysis (1), fetal valproic acid S. (1), heterochromia iridum (2), hypotonia (1), Johanssson-Blizzard S. (1), Meningitis (1), Moebius S. (1), neonatal sepsis (1), pendred S. (3), seizures (2), spina bifida (1), tracheoesophageal fistula (1), VATER (1), Waardenberg S. (2), X-linked stapes gusher S. (1). One patient has unilateral CND with a normal contralateral ear without malformation. Visual reinforcement audiometry or play audiometry. the nature of its differing tests being represented within a common index scale. RESULTS Patient Characteristics One hundred patients (n ¼ 100) with inner ear malformations and/or CND have been implanted to date. Only those 76 children implanted prior to age 18 with imaging studies available for review were considered for this report. Demographics are shown in Table I. For the 45 children with available newborn infant hearing screening results, most (n ¼ 41; 91%) failed (or referred) in at least 1ear(n¼ 36; 80% bilateral) whereas only 4 (9%) passed. More than half (54%) of the children in the study had an identified syndrome, chromosomal anomaly, or major medical condition that is known to be associated with hearing loss. Preoperative diagnostic auditory brainstem response (ABR) testing identified electrophysiologic evidence for auditory neuropathy spectrum disorder (ANSD) in at least 1 ear in 8 (11%) children with appropriate testing available. Interestingly, 75% of the ANSD ears had evidence of CND on radiographic imaging studies. The mean age at implantation was 5 years (SD ¼ 4; range ¼ ) and 70% of children received their implants before age 6. When three outlier patients implanted after the age of 16 are not included in the analysis (one fluctuating, two stable), the mean age at implantation was significantly later for those children 1981

4 Fig. 1. Inner ear malformation categories. The top series shows examples of incomplete partition (IP) malformations (I III) corresponding to A C. IP refers to normal cochlear external dimensions with decreased or absent partitioning due to either absent internal divisions (type I); apical fusion with a present interscalar septum (type II-Mondini type when EVA present); or X-linked stapes deafness type (type III). The middle series shows the various hypoplastic malformations (I III) corresponding to D F. Hypoplastic cochleas have smaller external dimensions and are also subclassified as: type I bud-like, II cystic hypoplasia, and III cochlea with less than two turns. The lowest row shows a cystic cochleovestibular malformation (CCVA) (G) and a common cavity (CC) malformation (H). If all cochlear partitioning was absent but there was differentiation in to cochlear and vestibular labyrinthine compartments, this was termed a cystic cochleovestibular malformation (CCVA). Common cavity (CC) was defined as no internal differentiation of the labyrinth into either a vestibular or cochlear compartment. with progressive or fluctuating hearing loss (6 6 3 years) when compared to those with stable preoperative hearing (4 6 3 years; P ¼.01). To date, there are five (7%) children who are nonusers of their device, one of whom was explanted. The mean duration of implant use for the entire group was 52 months (SD ¼ 43; range ¼ 1 168). The mean age at the most recent test interval was 9 years (SD ¼ 5; range ¼ 2 22). Imaging CT was used exclusively in 31 patients, MRI in 14 patients, and the remaining 31 patients had both imaging modalities. Imaging examples are shown in Figure 1, and the classifications for the implanted ear are shown in Table II as described in the Methods section above. The most common (n ¼ 35; 46%) malformation type was the IP-EVA spectrum. Among these, the classic Type II malformation, as described by Carlo Mondini in 1791, 1982 was most prevalent (n ¼ 21; 28%). Three children among this group had homozygous mutations in the PDS (SLC26A4) gene identified. Twenty-one ears of children with apparent CND have received a CI in their affected ear. In all but four (18%) instances, a coexisting inner ear malformation was evident. The most common labyrinthine malformation associated with CND in the present study was the hypoplastic spectrum with a type III cochlea. Among the 13 children with CND and a hypoplastic malformation, 10 (77%) had a known or were highly suspected of having a syndromic etiology including: CHARGE (n ¼ 4), BOR (n ¼ 1), Moebius (n ¼ 1), VATER (n ¼ 1), congenital facial paralysis (n ¼ 1), congenital hydrocephalus (n ¼ 1), and tracheoesophageal fistula (n ¼ 1). Overall, a syndromic association was much more prevalent among children with hypoplastic malformations (77%) and those with CND (69%) when compared to those with IP-EVA spectrum (29%) findings (P <.05 for both). Despite small

5 TABLE II. Imaging Findings in the Implanted Ears. Labyrinthine anomaly* Non-CND (%) CND(%) Totals (%) Number of patients 54 (71) 22 (29) 76 (100) IP-EVA spectrum 34 (63) 1 (5) 35 (46) Cochlear structure Normal (isolated EVA) 7 (13) 0 (0) 7 (9) Type I 5 (9) 1 (5) 6 (8) Type II 21 (39) 0 (0) 21 (28) Type III (X-linked stapes) 1 (2) 0 (0) 1 (1) Hypoplastic spectrum 13 (24) 13 (59) 26 (34) Cochlear structure Type I 1 (2) 2 (9) 3 (4) Type II 7 (13) 2 (9) 9 (12) Type III 5 (9) 9 (41) 14 (18) CCVA 1 (2) 1 (5) 2 (3) CC þ 2 (4) 1 (5) 3 (4) Isolated SSCA 4 (7) 2 (9) 6 (8) Normal 0 (0) 4 (18) 4 (5) *See Methods for definitions; CND ¼ cochlear nerve deficiency; IP-EVA ¼ incomplete partitioning-enlarged vestibular aqueduct; CCVA ¼ cystic cochleovestibular malformation; CC ¼ common cavity; SSCA ¼ semicircular canal atresia. Malformations were bilateral and symmetric in all but 8 (11%) children. numbers, children with CC, CCVA, and SCC anomalies also had a high preponderance of syndromic associations. One of two children with CC had a chromosome 7 abnormality. One child with CCVA had congenital hydrocephalus. Three of four children with SCC anomalies had either a chromosome 13 deletion, Waardenberg syndrome (SCC) or craniostenosis (SCC). Surgery Implantation was generally undertaken in the ear with the greatest degree of cochlear partitioning when cochlear nerve presence was confirmed. Ears with CND were avoided where possible (n ¼ 2). In cases where CND was bilateral, implantation was carried out on the side with the lowest behavioral thresholds or where the cochlear nerve canal appeared patent. Surgery was mostly uncomplicated across the study group (Table III). There were no facial nerve injuries despite nine cases where the nerve was displaced anteriorly adjacent to or within the round window niche. This was common in children with CHARGE and VATER. Despite 32 cases of intraoperative CSF gusher or oozer, there were no postoperative CSF leaks. Gushers were more common among ears with IP-EVA (n ¼ 21; 68%) than all other malformations (n ¼ 9; 23%) (P ¼.0002). Device failure occurred in 14 (18%) cases requiring replacement. Two children with CND had revision surgery; one for suspected device malfunction that ultimately resulted in no clinical improvement and one for device removal following lack of stimulation. There was one revision case initially implanted elsewhere of a misplaced electrode array that tracked from the basal turn into the internal auditory canal in a child with modiolar deficiency. Fourteen (18%) cases had a limited electrode insertion. Nine (64%) of these 14 had hypoplastic cochleas, two had CCVA malformations, one CC malformation, and two with IP-EVA (type I and II). In all but one case was a limited insertion anticipated based on the preoperative assessment of cochlear morphology. Audibility, Charge Requirements, and Speech Perception Testing Data are shown in Table IV. Using 120 db HL for the cases where no responses were measurable, the mean preoperative unaided and aided best ear speech awareness thresholds (SAT) were 92 (SD ¼ 20) and 55 (SD ¼ 36) db HL, respectively. Children with IP-EVA had significantly lower unaided SATs than those with hypoplastic malformations or CND, whereas children with CND had significantly higher aided thresholds than those with hypoplastic and IP-EVA malformations (P <.05 for these comparisons). Following implantation, the most recent behavioral testing revealed a mean PTA and SAT of 37 (SD ¼ 20) and 20 (SD ¼ 22) db HL, respectively. Figure 2a shows the most recent PTA by malformation category with CND cases considered separately. With rare exception (n ¼ 3; 4%), nearly all children demonstrated behavioral responses to CI stimulation irrespective of inner ear morphology and/or the presence of CND (graph excludes explanted child). Children with CND did have significantly higher mean PTAs than those with IP-EVA (P ¼.02) and hypoplastic (P ¼.009) malformations. Figure 2b and c shows the PTA as a function of charge per unit phase (nc) by malformation category and ECAP morphology, respectively. Children with CND as well as those with abnormal or absent TABLE III. Surgical Characteristics and Findings. Variable Non-CND (%) CND(%) Totals (%) Number of patients 54 (71) 22 (29) 76 (100) Age at Implantation 0 1 year 0 (0) 0 (0) 0 (0) 1 2 years 11 (20) 9 (40) 20 (26) 2 3 years 5 (9) 6 (27) 11 (15) 3 5 years 17 (31) 5 (23) 22 (29) 5 10 years 12 (22) 1 (5) 13 (17) >10 years 9 (17) 1 (5) 10 (13) Implant laterality Right 25 (46) 7 (32) 32 (36) Left 19 (35) 13 (59) 32 (36) Bilateral 10 (19) 2 (9) 12 (28) Full electrode insertion 43 (83)* 17 (77) 60 (81)* Cerebrospinal fluid gusher Observed 29 (54) 3 (14) 32 (42) Unknown 4 (7) 2 (9) 6 (8) *Two cases operated elsewhere had an unknown surgical insertion depth; CND ¼ cochlear nerve deficiency. 1983

6 TABLE IV. Utilization, Fitting, and Audibility and Speech Perception. Variable IP-EVA Hypoplastic CND Other Totals No. of patients (%) 34 (45) 13 (17) 22 (29) 7 (9) 76 (100) Nonusers (%) 0 (0) 0 (0) 4 (18) 1 (14) 5 (7) Duration of use (%) 0 1 ear 7 (21) 0 (0) 2 (9) 0 (0) 9 (12) 1 2 years 2 (6) 2 (15) 7 (32) 1 (14) 12 (16) 2 3 years 4 (12) 2 (15) 2 (9) 3 (43) 11 (14) 3 5 years 10 (29) 3 (23) 6 (27) 0 (0) 19 (25) >5 years 9 (26) 6 (46) 5 (23) 3 (43) 23 (30) Unknown 2 (6) 0 (0) 0 (0) 0 (0) 2 (3) Mean (SD) % 94 (7) 74 (14) 71 (15) 86 (18) 82 (16) Active Electrodes Mean (SD) Charge U. (nc) 19 (13) 21 (11) 51 (45) 32 (27) 31 (31) Mean (SD) PTA 32 (16) 30 (9) 49 (27) 34 (8) 37 (20) ECAP (%)* Robust 19 (83) 3 (33) 4 (22) 2 (50) 28 (52) Abnormal 3 (13) 3 (33) 3 (17) 1 (25) 10 (19) Absent 1 (4) 3 (33) 11 (61) 1 (25) 16 (29) Not available perception as a milestone (SRI-Q >300, >20 month duration of use), 100% (n ¼ 21) of children with IP-EVA and available data have achieved this level compared to only 50% (n ¼ 5) of children with hypoplastic malformations and 19% (n ¼ 4) of those with CND. A comparison of these proportions is significant for all groups (P <.05) *Percent of those with results available. CND ¼ cochlear nerve deficiency; IP ¼ EVA ¼ incomplete partitioningenlarged vestibular aqueduct; CCVA ¼ cystic cochleovestibular malformation; CC ¼ common cavity; SSCA ¼ semicircular canal atresia; PTA ¼ pure tone average; ECAP ¼ eighth nerve compound action potential testing; nc ¼ units of charge per unit phase. ECAPs required greater charge per unit phase to generate their respective PTAs. Between groups analysis revealed the differences in mean charge per unit phase to be significant (CND vs. IP-EVA [P ¼.006] and CND vs. hypoplastic [P ¼.01]). Similarly, those with robust ECAPs had lower mean charge per unit phase values than those with absent or abnormal ECAPs (P ¼.005). Robust ECAPs were much more common among children with IP-EVA spectrum findings (83%) than both hypoplastic (33%) and CND (22%) malformation groups. CC, CCVA, and isolated SCC cases were too few to make valid statistical comparisons. Overall, 37 (61%) of 61 children with available data demonstrated some degree of open-set speech perception skills (PB-k score >0%) at the last test interval. Of the children who could not perform speech perception testing, 6 (25%) had short durations of device usage (<7 months), 10 (42%) did not achieve scores above chance on the closed-set tests despite prolonged ( months ) use (i.e., poor performers), and 8 (33%) had significant comorbidities. Figure 3a shows speech perception testing results (SRI-Q) by malformation category at the most recent test with CND cases considered separately. The figure clearly demonstrates that despite similar durations of device use between groups, children with IP-EVA achieved higher SRI-Q levels than those with either hypoplastic spectrum malformations or those with CND. Children with hypoplastic malformations comparatively performed better than those with CND as well. When considering the development of open-set level speech Fig. 2. Pure tone average (PTA) (a) by malformation category, and as a function of charge per unit phase (nc) by (b) malformation category or (c) eighth nerve compound action potential (ECAP) morphology. CND ¼ cochlear nerve deficiency; IP-EVA ¼ incomplete partition-enlarged vestibular aqueduct; SCC ¼ isolated semicircular canal atresia; CC ¼ common cavity; nc ¼ indicates units of charge per unit phase. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.] 1984

7 of device use. Children with the less common malformations (isolated SCC, CC, CCVA) had numbers that were too small to make valid comparisons. Figure 3b demonstrates the relationship between ECAP on the SRI-Q. High SRI-Q scores were more common among children with robust ECAPs than those with abnormalorabsentecaps.only4(27%)of15childrenwith abnormal or absent ECAPs and more than 20 months of device use have developed any open-set speech perception abilities (albeit very limited in all but one case) compared to 17 (81%) of 21 children with robust responses (P ¼.0019). Communication Mode and Educational Setting A variety of communication modes and educational settings are in use at the time of this report (Table V). Among this very select patient population, 56% of the children are using either manual supplementation or primarily a manual communication mode. A closer look at the data reveals that the oral communicators (OC) are significantly more likely to have open-set speech perception abilities when compared to the children using total communication (TC) or sign language (SL). Figure 3c shows this graphically. Only 8 (33%) children using TC (>20 month duration of use) had any open-set speech perception abilities when compared to 24 (100%) children using exclusive OC (P ¼.0001). Also, SRI-Q levels were higher among OC children when compared to the TC group. By malformation classification, strategies that included some form of manual communication (TC, Cued Speech, SL, OC with AugCom) were substantially more common among children with hypoplastic malformations (69%) and CND (95%) than those with IP-EVA (18%) (P.05 for all comparisons). Educational setting is similarly diverse across study patients (Table V). As with communication mode, TABLE V. Communication Mode and Educational Setting. Variable Non-CND (%)* CND (%) Totals (%) Fig. 3. Speech perception index in quiet (SRI-Q) as a function of duration of use by (a) malformation category, (b) eighth nerve compound action potential (ECAP) morphology, and (c) communication mode. The dashed line represents a logarithmic trendline. CND ¼ cochlear nerve deficiency; IP-EVA ¼ incomplete partition-enlarged vestibular aqueduct; SCC ¼ isolated semicircular canal atresia; CCVA ¼ cystic cochleovestibular malformation; CC ¼ common cavity; IT-MAIS ¼ infant toddler meaningful auditory integration scale; ESP ¼ early speech perception test pattern (P) or monosyllables (M); PBK ¼ phonetically balanced kindergarten word lists; CNC ¼ consonant nucleus consonant word list. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.] except between hypoplastic and CND (P ¼.3973). Importantly, for the children with CND, all have relatively limited open-set performance (PB-k ¼ 8, 12, 28, and 36%) at their last follow up despite very long durations Number of children 54 (71) 22 (29) 76 (100) Communication mode Oral Only 28 (52) 1 (4) 29 (38) w/augcom 1 (2) 1 (4) 2 (3) Total communication (TC) 16 (30) 12 (55) 28 (37) Cued speech (CS) 1 (2) 3 (4) 4 (5) Sign language (SL) 1 (2) 4 (18) 5 (7) Unknown 7 (12) 1 (4) 8 (10) Educational setting Preschool 3 (6) 4 (18) 7 (9) Mainstream 25 (46) 5 (23) 30 (40) Home school 9 (17) 4 (18) 13 (17) Self-contained classroom 2 (4) 2 (9) 4 (5) School for the deaf 1 (2) 1 (4) 2 (3) None 5 (9) 2 (9) 7 (9) Unknown 9 (16) 4 (18) 13 (17) CND ¼ cochlear nerve deficiency. 1985

8 Fig. 4. Speech perception index in quiet (SRI-Q) as a function of age at testing by educational setting. Other settings includes: preschool, home school, selfcontained classroom, school for the deaf, and none. The dashed line represents a logarithmic trendline. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.] mainstream children had better speech perception abilities than those children in other educational setting. Figure 4 shows this graphically by age at testing. It is clear that educational setting tracks SRI-Q scores with higher levels of speech perception being associated with the use of a mainstream educational setting irrespective of malformation. However, when excluding those children in preschool, there is a trend for a higher proportion of children with IP-EVA (65%) to use the mainstream educational settings than those with hypoplastic malformations (50%) or CND (36%), although these differences do not reach significance DISCUSSION Inner ear malformations are common among children with sensorineural hearing loss. Although the surgical feasibility of placing a CI in a child with an inner ear malformation has been described for some time, there remains limited data regarding performance measures in such children. The reasons for this lack of data are probably related to the fact that most centers have not accumulated sufficient experience with such cases (<5 10/year in most centers) and the protocols needed to capture the necessary objective measures for making such comparisons require expertise in testing and prolonged stability. Since the mid-1990s speech perception measures have been tested, recorded, and tracked in our program as a matter of clinical routine. The speech perception abilities, communication modes and educational setting among a group of 76 CI children that have developmental inner ear malformations and/or CND are detailed here in an effort to provide long-term outcomes for this unique cohort. Results of the present study demonstrate that CI in children with inner ear malformations is safe and many of these children gain significant benefit from their devices. Importantly, the type of inner ear malformation as seen on imaging is significantly correlated with the development of certain levels of achievement for speech perception ability and communication mode and there was a trend for educational setting. Children with IP- EVA spectrum abnormalities frequently achieve good performance with 100% of these developing open-set speech perception skills (>20 month use), 82% using an exclusively oral communication mode and 65% attending a mainstream educational setting at their last follow-up. By contrast, children with hypoplastic malformations or CND achieve open-set test speech perception abilities much less frequently (50% and 19%, respectively). The degree of open-set speech perception for these groups is also significantly less than those children with IP-EVA malformations. Moreover, the communication mode for children with hypoplastic malformations and CND includes some form of visual supplementation in 69% and 95% of cases, respectively. The one child with CND that uses an exclusively oral communication mode has received a contralateral CI in his ear that does not have CND and uses this device only. The educational setting for these two groups was other than mainstream in 50% of children with hypoplastic malformations and 64% of children with CND. One important finding in the present study was that ECAP testing results were associated with the development of speech perception abilities. Specifically, only 27% of children with abnormal or absent ECAPs developed open-set speech perception skills compared to 81% of children with robust results. Although most studies have not shown such an association, the utility of this tool in patients with suspect (or limited) neural populations, such as those with auditory neuropathy spectrum disorder, has only recently been demonstrated. 19,20 In that study, no child with absent or abnormal ECAPs developed open-set speech perception

9 abilities. 20 The findings in the present work again suggest utility for this technology in this select patient population. It seems prudent to further develop electrical stimulation technologies for use prior to the placement of a CI as suggested previously. 12,20,21 Results of the present study as well as those of others have shown that some children with CND can benefit from CI. Warren et al. 12 recently reported on three children who demonstrated auditory awareness following CI in ears with apparent CND. This finding is in agreement with the present study where 18 (90%) of 20 children with CND and available testing had PTAs of less than 120 db HL. A closer look at the charge units needed to generate the behavioral responses reveals significantly greater levels for children with CND. Moreover, absent or very abnormal ECAP responses were present in 78% of these ears tested. None of the children in the Warren et al. 12 study have developed open-set speech understanding to date. The IT-MAIS scores they report for two of their patients are subjective parental assessments rather than objective measures of speech perception ability. In fact, many of the questions in the IT-MAIS might be answered positively based on sound awareness abilities alone, an achievable outcome even with poor neural populations. Clearly further follow-up of these children is needed to demonstrate a true speech and language benefit. The implications of this study are that children with CND (and severe inner ear malformations) have a relatively poor prognosis for the development of open-set speech perception abilities necessitating the addition of nonoral communication modes and school placement outside of the mainstream in most instances. With this in mind, it seems prudent to recommend to the families of these children the use of some form of visual augmentation for communication early, irrespective of CI placement. If sound awareness is the goal for the child and family, then a CI has a very high likelihood of providing this benefit. The fact that the implanted children with CND in the present study had poor speech perception abilities is consistent with the constellation of findings that includes: elevated sound detection thresholds, high charge levels, and abnormal or absent ECAPs, and suggests that the peripheral neural populations are insufficient for the development of synchronized auditory stimulation in most instances. Further work is clearly needed to better understand the mechanisms underlying the auditory performance observed in these children. In some instances, nonauditory (vestibular or somatosensory) stimulation might be responsible. Decision making for children with CND and/or severe inner ear malformations remains difficult when spoken language development is the goal. The relatively high association with multiple other challenges further complicates the situation. For the four implanted children at our institution with CND that gain limited openset speech perception with their device, such performance remains insufficient for the purposes of exclusive spoken language communication. These children all use supplementary sign language (i.e., total communication) for daily communication. ABI has recently been considered for children with CND and/or severe inner ear malformations. 22 That some of the children in this and other studies have demonstrated auditory awareness and limited speech perception abilities using a CI seems to justify placement prior to considering ABI. The low risk profile for this intervention when compared to the ABI further bolsters this approach. 23 CONCLUSIONS Children with IP-EVA malformations have an excellent prognosis for developing open-set speech perception and using oral communication modes following CI. On the contrary, children with severe malformations or CND may have elevated charge requirements for attaining sound detection alone. These children s prognosis for obtaining open-set speech understanding, using exclusive oral communication, and participating in mainstream education is more limited. With this in mind, it seems prudent to recommend to the families of these children the use of some form of visual augmentation for communication early, irrespective of CI placement. If sound awareness is the goal for the child and family, then a CI has a very high likelihood of providing this benefit. That some children with CND and/or severe malformations in this and other studies have demonstrated auditory awareness and limited speech perception abilities using a CI seems to justify placement prior to considering ABI. Acknowledgments The authors acknowledge Harold C. Pillsbury for his departmental leadership, unwavering support, and his expert care of some of the children in this cohort. We also acknowledge Vincent Carassco, Sarah Martinho, Corrine Macpherson, Paula Johnson, and Jill Ritch for their expertise in managing many of these children. We acknowledge Emily Buss for her helpful insight in the data management. Finally, we thank Carolyn J. Brown for her foresight in demanding objective measures of performance for all children seen in our program over the years. BIBLIOGRAPHY 1. Coticchia JM, Gokhale A, Waltonen J, Sumer B. Characteristics of sensorineural hearing loss in children with inner ear anomalies. Am J Otolaryngol 2006;27: McClay JE, Booth TN, Parry DA, Johnson R, Roland P. Evaluation of pediatric sensorineural hearing loss with magnetic resonance imaging. Arch Otolaryngol Head Neck Surg 2008;134: Buchman CA, Roush PA, Teagle HF, Brown CJ, Zdanski CJ, Grose JH. Auditory neuropathy characteristics in children with cochlear nerve deficiency. Ear Hear 2006;27: Adunka OF, Roush PA, Teagle HF, et al. Internal auditory canal morphology in children with cochlear nerve deficiency. Otol Neurotol 2006;27: Adunka OF, Jewells V, Buchman CA. Value of computed tomography in the evaluation of children with cochlear nerve deficiency. Otol Neurotol 2007;28: Nelson EG, Hinojosa R. Aplasia of the cochlear nerve: a temporal bone study. Otol Neurotol 2001;22: Lee KH, Lee J, Isaacson B, Kutz JW, Roland PS. Cochlear implantation in children with enlarged vestibular aqueduct. Laryngoscope 2010;120: Buchman CA, Copeland BJ, Yu KK, Brown CJ, Carrasco VN, Pillsbury HC 3rd. Cochlear implantation in children with congenital inner ear malformations. Laryngoscope 2004;114:

10 9. Papsin BC. Cochlear implantation in children with anomalous cochleovestibular anatomy. Laryngoscope 2005;115(1 Pt 2, Suppl 106): Dettman S, Sadeghi-Barzalighi A, Ambett R, Dowell R, Trotter M, Briggs R. Cochlear implants in forty-eight children with cochlear and/or vestibular abnormality. Audiol Neurootol 2010;16: Kang WS, Lee JH, Lee HN, Lee KS. Cochlear implantations in young children with cochlear nerve deficiency diagnosed by MRI. Otolaryngol Head Neck Surg 2010;143: Warren FM 3rd, Wiggins RH 3rd, Pitt C, Harnsberger HR, Shelton C. Apparent cochlear nerve aplasia: to implant or not to implant? Otol Neurotol 2010;31: Sennaroglu L. Cochlear implantation in inner ear malformations a review article. Cochlear Implants Int 2009; Apr 8 [Epub ahead of print]. 14. Roche JP, Huang BY, Castillo M, Bassim MK, Adunka OF, Buchman CA. Imaging characteristics of children with auditory neuropathy spectrum disorder. Otol Neurotol 2010;31: Campbell AP, Adunka OF, Zhou B, Qaqish BF, Buchman CA. Large vestibular aqueduct syndrome: anatomic and functional parameters. Laryngoscope 2011;121: Khan AM, Levine SR, Nadol JB Jr. The widely patent cochleovestibular communication of Edward Cock is a distinct inner ear malformation: implications for cochlear implantation. Ann Otol Rhinol Laryngol 2006; 115: Wang NY, Eisenberg LS, Johnson KC, et al. Tracking development of speech recognition: longitudinal data from hierarchical assessments in the Childhood Development after Cochlear Implantation Study. Otol Neurotol 2008;29: Zwolan TA, O sullivan MB, Fink NE, Niparko JK,CDACI Investigative Team. Electric charge requirements of pediatric cochlear implant recipients enrolled in the Childhood Development After Cochlear Implantation study. Otol Neurotol 2008;29: Cosetti MK, Shapiro WH, Green JE, et al. Intraoperative neural response telemetry as a predictor of performance. Otol Neurotol 2010;31: Teagle HF, Roush PA, Woodard JS, et al. Cochlear implantation in children with auditory neuropathy spectrum disorder. Ear Hear 2010;31: Kim AH, Kileny PR, Arts HA, El-Kashlan HK, Telian SA, Zwolan TA. Role of electrically evoked auditory brainstem response in cochlear implantation of children with inner ear malformations. Otol Neurotol 2008;29: Colletti V, Shannon RV. Open set speech perception with auditory brainstem implant? Laryngoscope 2005;115: Colletti V, Shannon RV, Carner M, Veronese S, Colletti L. Complications in auditory brainstem implant surgery in adults and children. Otol Neurotol 2010;31: