Translating resting state fMRI to determine invariant, replicable biomarkers of the heterogeneous tinnitus patient population

In recent years, the use of resting state functional connectivity (RS-FC) has gained prominence as a tool for understanding neural correlates of varying disorders and diseases, including Alzheimer’s disease, mild cognitive impairment, depression, and schizophrenia. This tool has also been used to study the neural correlates of subjective tinnitus, defined as the perception of sound without an external source. Tinnitus patients are often most bothered by their tinnitus when they are at rest, with no external stimuli to distract them from their percept, such as when falling asleep. As such, RS-FC assesses neural correlates of the disorder in the circumstance when the tinnitus percept is likely most intrusive and bothersome. The tool therefore has the potential to be a useful tool in identifying biomarkers of tinnitus, designing treatments, and tracking treatment efficacy. Work towards this goal has revealed general trends in RS-FC alterations in tinnitus; these can broadly be divided into increased connectivity between limbic regions and the auditory network, altered connectivity in various networks of attention, and alterations in the functional connectivity of the default mode network. Notably, however, there has been very little one-to-one agreement between the various RS-FC studies. This dissertation explores the sources for the discrepant results within the RS-FC tinnitus literature by assessing two possible sources of this variation: the heterogeneity of the tinnitus population and the reliability of the tool of RS-FC in tinnitus patients. Chapter 2 examined tinnitus subgroups that varied in tinnitus severity and in the length of time patients had had their tinnitus. We found that in patients with long-term tinnitus, there was decreased correlations between seed regions in the default mode network and the precuneus that became more prominent with increased severity of tinnitus. Similarly, connectivity between the precuneus and dorsal attention network was increased in these same participants. Chapter 3 attempted to replicate the findings associated with tinnitus severity in a large cohort across two different study sites. However, we did not replicate our findings from Chapter 2, and further, found differing results in the subgroups at each study site. The discrepant results were more noticeable in subgroups of patients with mild tinnitus rather than in those with bothersome tinnitus. These results made the direct examination of reliability performed in Chapter 4 even more important. Here, two resting state scans were collected for both tinnitus and control participants at two different study sites. Reliability was assessed using Intraclass Correlation Coefficients, or ICCs. Overall, RS-FC analyzed using seed-to-seed correlations was reliable in tinnitus and comparable to that of controls at each study site as assessed via the ICCs. ICCs for specific seed-pairs were less reliable in both tinnitus and controls, suggesting that while RS-FC in tinnitus is globally reliable, caution is warranted when considering specific connections. In general, correlations between regions in the dorsal attention network, default mode network, and across the two networks demonstrated high reliability. This builds confidence in the reliability of alterations to these networks found in tinnitus patients, and focus on these networks in future RS-FC studies of tinnitus is worthwhile. Alternatively, correlations between the auditory and limbic areas demonstrated poor reliability, which helps to explain disagreement between studies that examine these regions. Taken together, my dissertation provides evidence to support the reliability of RS-FC studies of tinnitus and suggests that the tool can reveal biomarkers of tinnitus that are critical to the development of diagnostics and treatments. It also emphasizes the role of attention systems in tinnitus perception, suggesting that these networks should be a focus in the development and evaluation of treatment options for the disorder. However, it also found that the heterogeneity of the tinnitus population does influence RS-FC, and that careful consideration of tinnitus characteristics such as severity and duration are important when interpreting findings across studies.U of I OnlyAuthor requested U of Illinois access only (OA after 2yrs) in Vireo ETD syste