Functional MRI (fMRI) is used extensively for studying neural correlates of brain functioning. FMRI is an indirect measure of neural activity as it measures changes in blood oxygenation level. Most fMRI studies assume a standard canonical hemodynamic response function (HRF) during analysis. However, recent advances show variability in HRF for different brain regions across subjects [1]. This challenges the interpretation of fMRI results since it is unclear if observed changes are due to neural activity or HRF variability. In this work, we identified HRF differences in Soldiers with posttraumatic stress disorder (PTSD) and postconcussion syndrome (PCS, chronic outcome associated with mild traumatic brain injury [mTBI]).

Recent evidences using Doppler ultrasound and infrared spectroscopy suggested alterations in cerebrovascular reactivity in mTBI [2]. Although neurochemical alterations in PTSD are well established [3], it is important to explore if these changes affect cerebrovascular reactivity. We hypothesized that the HRF, which depends on cerebrovascular reactivity and neurovascular coupling, may be altered in PTSD and mTBI. We tested the hypothesis by estimating the underlying HRF by performing blind hemodynamic deconvolution of resting-state fMRI data obtained from these populations.

Resting-state fMRI data was obtained from 87 male U.S. Army Soldiers (17 having PTSD, 42 having comorbid PTSD and PCS [PCS+PTSD] and 28 matched combat controls) in a 3T Siemens Verio scanner using T2*-weighted multiband-EPI sequence, with TR=600ms, TE=30ms, FA=55°, voxel size=3×3×5mm³, 2 sessions and 1000 volumes. Standard pre-processing was performed (realignment, normalization to MNI space, detrending, regressing out nuisance-covariates). Each voxel timeseries was then subjected to blind hemodynamic deconvolution; “blind” because it estimates both the HRF and underlying latent neuronal variables from just the observed fMRI data. We employed the deconvolution technique proposed by Wu et al. [4] which considers resting-state fMRI as a spontaneous pseudo-event-related signal and uses a variant of Weiner deconvolution. The HRF at each voxel in each subject was characterized by three parameters – response height (RH), time-to-peak (TTP), and full-width at half-wax (FWHM). Whole-brain two-sample t-tests were performed separately on the three parameters to obtain group-wise voxel-specific differences in HRF parameters (p<0.05, cluster-level thresholded at 400mm³, controlled for age, race, education and head-motion). This was done separately for the three pairwise comparisons between groups, and resulting binary maps were overlapped (with cluster-level threshold of 50mm³, in overlapped map) to obtain final maps.

In all the regions with altered HRF, we found that RH increased in the disorders compared to controls, while TTP and FWHM decreased in the disorders. This indicates that PTSD and PCS+PTSD are characterized by taller, quicker and narrower HRF in affected regions. We first elucidate the differences for Control vs Disease comparison, which refers to an overlap of Control vs PTSD and Control vs PCS+PTSD comparisons. Differences in RH were found in (see Fig.1) thalamus, midbrain, precuneus, posterior cingulate cortex (PCC), secondary visual areas and parts of insula (anterior and posterior). Given that PTSD is an anxiety disorder, prior work shows abnormal GABAergic and glutamatergic systems relate to anxiety [5], with thalamus being anatomically well situated to produce the experience of anxiety [6], and serotonin in the midbrain playing a key role in anxiety disorders [7]. Further, alterations in TTP (Fig.2) and FWHM (Fig.3) largely overlapped, with key default-mode network (DMN) regions being disrupted (PCC and precuneus) along with secondary visual areas. Additionally, when the results of RH, TTP and FWHM were overlapped, we found common differences in PCC as well as precuneus (see Fig.4). Earlier studies have reported neurochemical alterations in these key areas [8, 9]. Taken together, these results corroborate with earlier findings of disrupted neurochemistry, and show that fMRI studies need to exercise caution in interpreting results arising from these regions if, for example, they employ a canonical HRF.

Comparing all the three groups, we found FWHM to be significantly different between all three groups in PCC and precuneus (see Fig.5). This shows that the hemodynamic response in PCC and precuneus are affected by both PTSD and mTBI. This is a substantial result given that neural underpinnings of comorbid PTSD and mTBI are poorly understood [10]. PCC and precuneus showed altered HRF between all three groups with all the three HRF parameters.

In summary, we showed that PTSD and mTBI cause overlapping and distinct HRF alterations in subcortical structures and the DMN. Our findings also corroborate with prior neurochemical findings. Given these findings, future studies on PTSD and mTBI, and fMRI studies in general must exercise caution in interpreting their results. We encourage researchers to employ hemodynamic deconvolution during data pre-processing to mitigate the issue.