Hemodynamic Response Variability and its Relationship to the BOLD signal in Younger and Older Adults

Studies have shown age-related differences in blood-oxygen-level-dependent signal (BOLD) variability, specifically amplitude variability. However, results have been mixed. Little remains known about the sources contributing to this variability. Identifying these sources would have implications for underlying mechanisms contributing to BOLD measurement. Changes in BOLD yield a characteristic hemodynamic response function (HRF) that reflects a combination of blood flow and oxygenation changes that follow neural activity. In healthy aging, multiple components of the HRF (e.g., time-to-peak, rise slope, peak amplitude, full-width half-maximum, peak-to- trough, time-to-trough, fall slope, and trough amplitude) are susceptible to the mediating effects of age-related cerebrovascular alterations and underlying processes. Additionally, several studies have demonstrated that neuro-vascular coupling (NVC) differences in older adults are mirrored in HRF differences. To further explore these phenomena, the current study utilized the publicly available Cambridge Center for Aging and Neuroscience (CamCAN) dataset to estimate HRF variability in a visual-auditory task in 80 younger (18-30 years old; 44 Female/36 Male) and 212 older adults (54-74 years old; 100 Female/112 Male). The proposed study was carried out according to three aims: (1) examine intra-individual HRF variability in younger and older adults, (2) examine inter-individual HRF variability in younger and older adults, and (3) determine the relationship between HRF variability and cognitive performance in younger and older adults. Linear mixed models were used to assess individual and age-related differences in HRF features. I hypothesized that individuals, regardless of age, would have increased HRF variability in higher frequency task conditions compared to lower frequency conditions. For age- related differences, I hypothesized that older adults would have increased HRF feature variability, and that their HRF variability would be inversely related to canonical-derived BOLD voxel extent. Finally, I hypothesized that there would be an interaction between HRF variability, age, and cognitive performance such that low-performing older adults would have increased HRF variability compared to high-performing older and younger adults. For group differences in HRF feature variability, I found that increased/decreased HRF feature variability was associated with increasing auditory frequencies depending on the region examined. For group differences in mean HRF features, I found that increased mean HRF features were associated with increasing auditory frequencies, with the exception of fall slope which exhibited an inverse relationship. Older adults had increased HRF feature variability and mean HRF features, primarily in the precentral and temporal ROIs, compared to younger adults. Older adults’ increased voxel extent was associated with decreased variance of their rise slopes, full-width half-maxima, peak-to- troughs, and time-to-troughs. Finally, younger adults exhibited a significant relationship between their reaction times and mean HRF features in the highest frequency condition while the older adults did not. My results showed that HRF feature variability exhibits region- and task- dependent differences that need to be accounted for when performing age-group comparisons, the latter-half of the HRF evolution and underlying mechanisms are potential sources of additional variability in older adults, and the difference between HRF features in the precentral cortex and other sensory cortices may serve a mediatory role between age and processing speed ability. This study assessed features of BOLD HRF shape as a proxy of NVC to identify potential sources of altered age-related variability and their relationships to behavior.