Quantitative susceptibility mapping at 7 T in COVID-19: brainstem effects and outcome associations

Abstract

Post-mortem studies have shown that patients dying from severe acute respiratory syndrome coronavirus (SARS-CoV-2) infection frequently have pathological changes in their CNS, particularly in the brainstem. Many of these changes are proposed to result from para-infectious and/or post-infection immune responses. Clinical symptoms such as fatigue, breathlessness, and chest pain are frequently reported in post-hospitalized coronavirus disease 2019 (COVID-19) patients. We propose that these symptoms are in part due to damage to key neuromodulatory brainstem nuclei. While brainstem involvement has been demonstrated in the acute phase of the illness, the evidence of long-term brainstem change on MRI is inconclusive. We therefore used ultra-high field (7 T) quantitative susceptibility mapping (QSM) to test the hypothesis that brainstem abnormalities persist in post-COVID patients and that these are associated with persistence of key symptoms. We used 7 T QSM data from 30 patients, scanned 93-548 days after hospital admission for COVID-19 and compared them to 51 age-matched controls without prior history of COVID-19 infection. We correlated the patients' QSM signals with disease severity (duration of hospital admission and COVID-19 severity scale), inflammatory response during the acute illness (C-reactive protein, D-dimer and platelet levels), functional recovery (modified Rankin scale), depression (Patient Health Questionnaire-9) and anxiety (Generalized Anxiety Disorder-7). In COVID-19 survivors, the MR susceptibility increased in the medulla, pons and midbrain regions of the brainstem. Specifically, there was increased susceptibility in the inferior medullary reticular formation and the raphe pallidus and obscurus. In these regions, patients with higher tissue susceptibility had worse acute disease severity, higher acute inflammatory markers, and significantly worse functional recovery. This study contributes to understanding the long-term effects of COVID-19 and recovery. Using non-invasive ultra-high field 7 T MRI, we show evidence of brainstem pathophysiological changes associated with inflammatory processes in post-hospitalized COVID-19 survivors.

Keywords: 7T MRI; brainstem; coronavirus disease of 2019; inflammation; quantitative susceptibility mapping.

Figure 1

3D projections of the quantitative susceptibility mapping χ maps on the rendered brainstem regions of interest extracted from the FreeSurfer segmentation for the healthy control group and the COVID group. The coronavirus disease 2019 (COVID) group shows increased χ in the brainstem, specifically in the medulla and pons (black arrows). A = anterior; HC = healthy control group; L = left; P = posterior; QSM = quantitative susceptibility mapping; R = right. 3D renderings were generated with Surf Ice.

Figure 2

Box plots of differences in the regional average χ between the COVID group and the healthy control group obtained from the brainstem. Group differences assessed with a linear model with age, gender and age by gender interactions added as explanatory variables of no interest. False discovery rate-corrected statistics represented on the box plots. ***P < 0.001, **P < 0.01, *P < 0.05, ns = not significant. COVID = coronavirus disease 2019; HC = healthy control; QSM = quantitative susceptibility mapping; SCP = superior cerebellar peduncle.

Figure 3

Voxel-wise analysis showing increased quantitative susceptibility mapping χ on the COVID group compared with healthy controls. Significant clusters determined with the ‘Randomise’ function in FSL [threshold free cluster enhancement (TFCE) corrected P < 0.01, cluster inference t = 2.5, cluster volume > 1 mm³]. 3D projection of the significant clusters on the brainstem region of interest. A = anterior; COVID = coronavirus disease 2019; L = left; P = posterior; R = right. 3D renderings were generated with Surf Ice.

Figure 4

Scatter plots of the average quantitative susceptibility mapping (QSM) χ obtained on the clusters from the voxel-wise group analysis with clinical and laboratory outcomes. (A) World Health Organization (WHO) score, (B) period of hospital admission, (C) highest C-reactive protein (CRP) during admission and (D) modified Rankin Score (mRS). The R² value is also displayed on each plot. B shows data without the outlier (Supplementary material section 5).