Gibson, William Peter Rea; Sanli, Halit

For many years clinicians have been aware that a few hearing impaired children have a disproportionate loss of speech perception leading to a poor performance using conventional hearing aids. The term “auditory neuropathy” was originally coined in 1995 (^{Sininger, Hood, & Starr, 1995}) after the first longitudinal study of 10 subjects who were discovered to have absent or grossly abnormal auditory brainstem responses (ABR), but present otoacoustic emissions (OAE). These findings led to the concept that the dysfunction primarily affected the neural auditory pathway (^{Starr, Picton, Sininger, et al., 1996}); however, this concept failed to account for the unexpectedly good performance of most of these ears using a cochlear implant (^{Miyamoto, Kirk, Renshaw, et al., 1999}; ^{Trautwein, Sininger, & Nelson, 2000}; ^{Shallop, Peterson, Facer, et al., 2001}; ^{Gibson & Sanli, 2002}; ^{Madden, Hilbert, Rutter, et al., 2002}; ^{Buss, Labadie, Brown, Grosset, et al., 2002}; ^{Mason, De Michelle, & Stevens, 2003}). This paper reviews the electrophysiological findings obtained from children affected by AN.

OAE are assumed to be derived from the motor activity of outer hair cells (OHC) and do not appear to depend in any important way of the functioning of inner hair cells (IHC) (^{Mills, 2006}). IHC are not connected to the overlying tectorial membrane and cannot exert an effective force back on the organ of Corti and cannot be a significant source of OAE (^{Trautwein, Hofstetter, Wang, et al., 1996}). Abnormal positive potentials (APP) are an early positive summating potential (SP) with respect to an active electrode placed in the round window niche (^{O’Leary, Mitchell, Gibson, et al., 2000}). The APP or early positive SP is probably generated by a slight asymmetry of production of CM.

Materials and Methods

Ears with OAE and absent or grossly abnormal ABR (a broad N1 component only) were investigated using round window electrocochleography (RW ECochG) and electric auditory brainstem responses (EABR). The electrophysiological data were collected prospectively between 1994 and April 2005 from 435 severely and profoundly deaf children who received cochlear implants. The majority of the children were referred with OAE results obtained during the neonatal period. Preoperatively these children underwent RW ECochG, ABR, and often transtympanic EABR testing. A specially constructed ‘golf club’ electrode was placed into the round window niche during general anesthesia (^{Aso & Gibson, 1994}). Electrodes placed on the vertex and ipsilateral earlobe recorded the ABR. Data were collected using the Medelec® Sensor apparatus until 1998 and afterwards using the Medelec Synergy apparatus.

All of the children received a Nucleus® Cochlear implant. Only patients who had full insertion of 22 electrodes were included in the study. The EABR were obtained intraoperatively. Software from the Cochlear® Corporation, Australia was used to generate the EABR via the Nucleus 22 and 24 devices. EABR were sought from all 22 electrodes. The normal values for the latency of eV were established in deaf ears that had no abnormal positive potentials (APP) and achieved speech perception scores over 50% using a cochlear implant (Melbourne category 7) (^{Ray & Gibson, 2004}). The EABR waveform was judged to be abnormal according to the following criteria: either no discernible wave eV or a delay of eV greater than 2 msec on at least 5 of the 22 electrodes tested.

Three categories were compared: the first two categories were children with auditory neuropathy sub classified according to the EABR findings. The third category was a control group.

• Group A Large CM and APP on RWECochG and normal EABR
• Group B Large CM and APP on RWECochG and abnormal EABR
• Group C No large CM or APP on RWECochG and normal EABR

The outcome after cochlear implantation of the study groups was measured using the Melbourne speech perception categories (Table 1) at 1 yr and 2 yr postsurgery (^{Dowell, Blamey, & Clark, 1995}). All data were statistically evaluated using standard statistical package (Statistica, Release 6.1). Mann-Whitney U Test for nonparametric data was used to test for significant difference at significance level p < 0.005 (two tailed).

TABLE 1

Results

39 of the 435 children that received a cochlear implant had present OAE during the neonatal period and had absent or abnormal ABR. On RW ECochG testing, all of these 39 children showed a large cochlear microphonic (CM) and an abnormal positive potential (APP) in both ears. It is probable that the APP is derived from CM which has not cancelled with phase reversal of the stimulus. Examples of traces are shown in Figure 1.

Fig. 1. RW ECochG recordings from 5 different subjects showing APP alternating 8 kHz tone burst at 100 dBnHL.

There were a further 21 children who had large CM and APP who had not been tested for OAE during the neonatal period. Thus a total of 60/435 (13.8%) children had the RW ECochG abnormality. The abnormality was most commonly encountered in children who had been born prematurely and had suffered a period of hypoxia (^{Rea & Gibson, 2003}).

The EABR results were obtained from all 60 children at the time of their cochlear implant surgery. Electrically evoked ABR (EABR) were normal in 45 (75%) (Fig. 2) and abnormal in 15 (25%) (Fig. 3). The 39 children who had OAE and absent or abnormal ABR showed a similar result: EABR were normal in 32 children (82%) and abnormal in 7 (18%).

Fig. 2. Normal implant evoked electrical auditory brainstem response recorded from channels 22–1 in MP1 + 2 mode of stimulation at CL: 228 and PW: 25 μs.

Fig. 3. Implant evoked electrical auditory brainstem response traces recorded from an ear with APP and abnormal EABR.

Outcomes after Cochlear Implant Surgery

The Melbourne Speech perception scores of group A and group B at 1 yr and 2 yr postimplant are given in Table 2. The mean speech perception scores in the control (Group C) group at 1 and 2 yr are 3.93 (SD 2.36) and 5.37 (SD 2.1). It is interesting to note that the speech perception outcome in group A are statistically better than the control group (C) (two tailed p value <0.005, Mann Whitney U Test). Analysis of the 39 children with OAE rather than all 60 children with APP did not alter the conclusions of this paper but it was decided not to include ears with APP in the control group C.

TABLE 2

Further statistical analysis was performed ANOVA comparing the three groups which shows that the speech perception scores of the three groups are significantly different, but overall ANOVA does not give the order of the three groups. One problem with ANOVA and t-test is that they require a normal distribution of samples (or near normal). Our distributions do not fit these criteria as our samples are not normally distributed. Because ANOVA does not give any specific ordering, a multiple comparisons approach was employed. To perform multiple comparisons procedure, t statistics for all pairs of means were computed and compared with the critical t value for that pair. In these computations pooled standard deviation of all populations was used which increases the power of the tests. Table 3 shows the multiple comparisons t statistics of paired groups in the study.

TABLE 3

From the table one can safely conclude that, speech perception scores of group A (APP normal EABR) are better than speech perception scores of groups B (APP abnormal EABR) and C (No APP normal EABR). Also it is clear that speech perception scores of group C are better than speech perception scores of group B.

Power calculations of t statistics in comparing groups was undertaken as there are only 15 subjects in category B which showed that despite this discrepancy, the statistical results are valid (Table 4).

TABLE 4

Conclusions

Forty five of the 60 ears that had APP and abnormal or absent ABR had normal EABR on intraoperative testing and the outcome of cochlear implantation was statistically better than a control group of deaf ears which did not have APP. The EABR findings suggest that for these children, the afferent conduction in the auditory pathway to the level of the inferior colliculus was not disordered—in other words, there was no brainstem neuropathy. The finding that the 45 children with APP and normal EABR performed better using cochlear implants than a group of children with no APP and normal EABR suggests that the higher cortical areas are not adversely affected. Fifteen of the 60 ears (25%) which had APP and absent or abnormal ABR had abnormal EABR and these children only gained limited benefit from a cochlear implant. It would appear that these 15 children did have a neuropathy.

The presence of OAE and the absence of ABR can be explained by the presence of OHC when there is a significant loss of IHC. Animal studies support this theory. Studies performed in chinchillas (^{Harrison, 1998}) with carboplatin (an anticancer drug from the same group as cisplatin) have shown an extensive loss of IHC while the OHC remain intact. In these animals, the preserved outer hair cells produced both an electrical signal recorded as the cochlear microphonic (CM), and an acoustic signal recorded as the OAE while the lack of IHC produced a reduced or absent ABR. There are also genetic animal models: the Bronx waltzer mouse (^{Bock, Yates, & Deol, 1982}) and the Beethoven mouse (^{Bussoli, Kelly, & Steel, 1997}) are mouse mutants which have extensive loss of IHC in the presence of OHC. Chronic hypoxia can also produce a similar loss of IHC in guinea pig (^{Harrison, 2001}). This is of interest as many premature infants with auditory neuropathy have suffered periods of hypoxia (^{Rea & Gibson, 2003}) and may explain why a few infants pass OAE screening tests when they have a profound hearing loss on ABR testing (^{Davis & Wood, 1992}).

An alternative suggestion is that there is a disorder of the synapse rather than loss of IHC. There is a recent discovery of hair cell synaptic ribbons required for faithful synaptic transmission (^{Khimich, Nouvain, & Pujol, et al., 2005}). Mouse mutants (Basoon and Piccolo mutants) with disordered hair cell synaptic ribbons have been studied but the ABR findings are not similar to auditory neuropathy as the wave V is present but delayed in a manner similar to when the nerve is affected by an acoustic neuroma.

References