Predicting synapse counts in living humans by combining computational models with auditory physiology
Aging, noise exposure, and ototoxic medications lead to cochlear synapse loss in animal models. As cochlear function is highly conserved across mammalian species, synaptopathy likely occurs in humans as well. Indeed, temporal bone studies demonstrate loss of synapses with advancing age in humans. Synaptopathy is predicted to result in perceptual deficits including tinnitus, hyperacusis, and difficulty understanding speech-in-noise. However, there is currently no method for diagnosing synaptopathy in living humans. This prevents us from determining if noise-induced synaptopathy occurs in humans, identifying the perceptual consequences of synaptopathy, or testing potential drug treatments. Several physiological measures are sensitive to synaptopathy in animal models, including auditory brainstem response (ABR) wave I amplitude. However, it is unclear how to translate these measures to synaptopathy diagnosis in humans. In this study, a human computational model of the auditory periphery that can predict ABR waveforms and distortion product otoacoustic emissions (DPOAEs) was fit using Bayesian regression analysis to predict synapse counts in individual human participants based on their measured DPOAE levels and ABR wave I amplitudes. Lower predicted synapse numbers were associated with higher noise exposure history, increased likelihood of tinnitus, and poorer speech-in-noise perception. These findings illustrate the utility of this modeling approach in predicting synapse counts from physiological data.