Will Changes in the COVID‐19 Brain MRI Signature Region Affect Subsequent Dementia Outcomes? Linking Pre‐Pandemic Brain MRIs to Long‐Term Dementia Incidence in UK Biobank Participants

Abstract

AbstractBackgroundNeurocognitive sequelae of SARS‐CoV‐2 infection have been suggested as potentially leading to a higher incidence of dementia, but the underlying mechanisms remain unclear. By comparing brain magnetic resonance imaging (MRI) between participants with and without a history of SARS‐CoV‐2 infection, prior work identified infection‐related structural changes, including reduced cortical thickness and gray‐white matter contrast. We characterized the COVID‐19 signature region (COVIDsig) in MRIs taken prior to the pandemic and evaluated whether changes in this region predicted incident dementia.MethodUsing pre‐pandemic data on UK Biobank participants aged 55+, we examined the association between the COVID‐19 brain MRI signature region and dementia incidence up to March 2020 (N = 39,119 dementia‐free participants with a median of 1.8 years of follow‐up). The COVIDsig comprised the lateral orbitofrontal cortex and parahippocampal gyrus and the exposures included COVIDsig cortical thickness and COVIDsig gray‐white matter contrast. Dementia diagnoses were obtained from linked hospital, primary care, and death registry data. We used Cox proportional‐hazards models and adjusted for sex, APOE‐ε4, education, material deprivation, body mass index, and assessment center. To test if COVIDsig predicts dementia risk independently of global cerebral atrophy, we further adjusted for an Alzheimer’s disease signature region (ADsig) or a non‐COVID‐19 signature region (cerebral cortex except COVIDsig). In sensitivity analysis, we repeated the primary analysis on each component of COVIDsig (lateral orbitofrontal cortex and parahippocampal gyrus) separately.ResultThe mean age of the participants was 64.7 years (SD = 7.4; Table 1) and 65 dementia cases were identified during follow‐up. The age‐adjusted dementia incidence rate was 2.5 per 1,000 person‐years. One standard deviation decrease in COVIDsig cortical thickness and gray‐white matter contrast was associated with a 49% and a 52% increase in the hazard of dementia diagnosis (HR for cortical thickness = 1.49, 95% CI = 1.19‐1.88; HR for gray‐white matter contrast = 1.52, 95% CI = 1.20‐1.94; Table 2), respectively. After adjusting for ADsig or non‐COVID‐19 signature region, associations were attenuated (Figure 1).ConclusionThese results should be evaluated in independent samples and the persistence of post‐COVID‐19 structural brain changes assessed. If confirmed, findings suggest SARS‐CoV‐2 infection may increase dementia incidence via structural brain changes.