Brain imaging before and after COVID-19 in UK Biobank
Douaud G., Lee S., Alfaro-Almagro F., Arthofer C., Wang C., McCarthy P., Lange F., Andersson JLR., Griffanti L., Duff E., Jbabdi S., Taschler B., Winkler AM., Nichols TE., Collins R., Matthews PM., Allen N., Miller KL., Smith SM.
AbstractThere is strong evidence for brain-related pathologies in COVID-19, some of which could be a consequence of viral neurotropism, or of neuroinflammation following viral infection. Most brain imaging studies have focused on qualitative, gross pathology in moderate to severe cases, most typically carried out on hospitalised patients. It remains unknown however whether the impact of SARS-CoV-2 infection can be detected in milder cases, in a quantitative and automated manner, and whether this can reveal possible mechanisms for the spread of the disease. UK Biobank scanned over 40,000 participants before the start of the COVID-19 pandemic, making it possible in 2021 to invite back hundreds of previously-imaged participants for a second imaging visit. Here, we studied the possible brain changes associated with the coronavirus infection using multimodal MRI data from 785 adult participants (aged 51–81) from the UK Biobank COVID-19 re-imaging study, including 401 adult participants who tested positive for SARS-CoV-2 infection between their two scans. We used structural, diffusion and functional brain scans from before and after infection, to compare longitudinal changes between these 401 SARS-CoV-2 cases and 384 controls who had either tested negative to rapid antibody testing or had no COVID-19 medical and public health record, and who were matched to the cases for age, sex, ethnicity and interval between scans. The controls and cases did not differ in blood pressure, body mass index, diabetes diagnosis, smoking, alcohol consumption, or socio-economic status. Using both hypothesis-driven and exploratory approaches, with false discovery rate multiple comparison correction, we identified respectively 68 and 67 significant longitudinal effects associated with SARS-CoV-2 infection in the brain, including, on average: (i) a more pronounced reduction in grey matter thickness and contrast in the lateral orbitofrontal cortex (min P=1.7×10-4, r=-0.14) and parahippocampal gyrus (min P=2.7×10-4, r=-0.13), (ii) a relative increase of diffusion indices, a marker of tissue damage, in the regions of the brain functionally-connected to the piriform cortex, anterior olfactory nucleus and olfactory tubercle (min P=2.2×10-5, r=0.16), and (iii) greater reduction in global measures of brain size and increase in cerebrospinal fluid volume suggesting an additional diffuse atrophy in the infected participants (min P=4.0×10-6, r=-0.17). When looking over the entire cortical surface, these grey matter thickness results covered the parahippocampal gyrus and the lateral orbitofrontal cortex, and extended to the anterior insula and anterior cingulate cortex, supramarginal gyrus and temporal pole. The increase of a diffusion index (mean diffusivity) meanwhile could be seen voxel-wise mainly in the medial and lateral orbitofrontal cortex, the anterior insula, the anterior cingulate cortex and the amygdala. These results were not altered after excluding cases who had been hospitalised. We further compared hospitalised (n=15) and non-hospitalised (n=386) infected participants, resulting in similar findings to the larger cases vs control group comparison, with, in addition, a marked reduction of grey matter thickness in fronto-parietal and temporal regions (all FDR-significant, min P=4.0×10-6). The 401 SARS-CoV-2 infected participants also showed larger cognitive decline between the two timepoints in the Trail Making Test compared with the controls (both FDR-significant, min P=1.0×10-4, r=0.17; and still FDR-significant after excluding the hospitalised patients: min P=1.0×10-4, r=0.17), with the duration taken to complete the alphanumeric trail correlating post hoc with the cognitive and olfactory-related crus II of the cerebellum (FDR-significant, P=2.0×10-3, r=-0.19), which was also found significantly atrophic in the SARS-CoV-2 participants (FDR-significant, P=6.1×10-5, r=-0.14). Our findings thus relate to longitudinal abnormalities in limbic cortical areas with direct neuronal connectivity to the primary olfactory system. Unlike in post hoc cross-sectional studies, the availability of pre- infection imaging data mitigates to some extent the issue of pre-existing risk factors or clinical conditions being misinterpreted as disease effects. We were therefore able to demonstrate that the regions of the brain that showed longitudinal differences post-infection did not already show any difference between (future) cases and controls in their initial, pre-infection scans. These brain imaging results may be the in vivo hallmarks of a degenerative spread of the disease — or of the virus itself — via olfactory pathways (a possible entry point of the virus to the central nervous system being via the olfactory mucosa), or of neuroinflammatory events due to the infection, or of the loss of sensory input due to anosmia. Whether this deleterious impact can be partially reversed, for instance after improvement of the hyposmic symptoms, or whether these are effects that will persist in the long term, remains to be investigated with additional follow up.