Can rollover be used to assess cochlear synaptopathy in listeners with normal audiograms?

Animal studies indicate that noise exposure may lead to cochlear synaptopathy, primarily affecting low-spontaneous-rate fibers in the auditory nerve. Consequently, it can be hypothesized that in humans, the effects of cochlear synaptopathy will be more pronounced for high-intensity stimuli compared to low-intensity ones. For instance, cochlear synaptopathy may result in a reduced strength of the middle-ear muscle reflex (MEMR), or poorer speech intelligibility (SI) at high presentation levels compared to low levels. Recently, Shehorn et al. (2020) (DOI: 10.1016/j.heares.2020.107982) has reported that lower speech recognition at very high input levels (104 dBA) were associated with reduced MEMR magnitude (measured with an advanced, research protocol), and the reduced MEMR magnitude was associated with higher estimated lifetime noise exposure.
While that study focused on SI at single presentation levels, SI changes across levels can also be assessed. SI declines with increasing level, so-called rollover (RO), have been observed in individuals with normal hearing thresholds, and may also be CS-related. However, broadened auditory filters or a reduced ‘sharpness’ of the neural speech representation at higher levels could also lead to RO.
Thus, RO occurrence in individuals with normal hearing thresholds does not necessarily indicate CS. Instead, ‘normal’ and ‘abnormal’ contributions to RO would need to be disentangled. Since there are currently no clinically established tools for detecting CS, this cannot be done a priori (e.g., by testing groups with and without CS and then comparing RO for the two groups). Instead, it can be hypothesized that in case of ‘abnormal’ contributions RO will be larger and correlated with other experimental measures, such as extended high-frequency audiometry, that presumably reflect CS, or MEMR strength.
To test these hypotheses, we measured SI in speech-like noise presented at 65 and 80 dB SPL. in 25 young adults with normal hearing thresholds. Extended high-frequency audiometry, acoustic-reflex measurements, and structured interviews regarding noise exposure were also conducted, using clinically available protocols.
The negative correlation between noise exposure history and MEMR strength was significant, and in a good agreement with Shehorn et al. Further, majority of the listeners showed some degree of RO (median RO = 9%). However, no simple relations were found between RO and other measures. We argue that the majority of the listeners were normal-hearing and thus the 9% median can be considered ‘normal’ RO.

We expect to provide estimates of ‘normal’ RO magnitude and, based on the relations with the other experimental measures, to identify individuals with a high probability of abnormal RO and, thus, CS.