Correlations among Distortion Product Otoacoustic Emissions, Thresholds and Sensory Cell Impairments

Distortion product otoacoustic emissions (DPOAE) are increasingly used as an objective test for noninvasive hearing screening. When two pure tones with frequencies f1 and f2 are sent to the cochlea, the most prominent DPOAE is the cubic one produced at 2f1-f2, and this presentation will mainly emphasize its properties. DPOAEs are undoubtedly generated by cochlear nonlinearities. It is widely held that they arise from certain stages of sound processing by the outer hair cells (OHC) and that OHCs ensure normal cochlear sensitivity and tuning. Thus, DPOAEs should provide a privileged tool for monitoring the harmful effects of loud sound because OHCs are known to be one of the main targets of NIHL. Although DPOAEs provide the clinicians with a reliable screening limit of about 30 dB HL around f2, no reliable relationship has been found thus far between possible residual DPOAEs and either hearing loss or amount of impaired sensory cells. Furthermore, puzzling contradictory findings have been reported as to the presence of DPOAEs despite a large hearing loss (i.e. >30-40 dB) notably with high-level stimuli. These observations raise the following issues. What is the generation site of DPOAEs in a normal or pathological cochlea (OHCs, basilar membrane, place tuned to f2, 2f1-f2, places basal to f2.)? Is it necessary to account for interferences between several discrete sources, arising from different locations or different mechanisms and possibly exhibiting differential susceptibility to sensory cell damage? Do DPOAE changes depend on the nature of OHC pathology (NIHL, anoxia, ototoxic drugs, genetics.)? Once a source of DPOAE is characterized, is there any means of modelling the physiological process of its generation and deriving what might quantitatively relate DPOAE amount to sensory cell activity and thresholds? The goal of this presentation is to examine these issues, review the available data and propose a comparatively simple model.