A Comparison of Regional Brain Volumes in Older Adults With and Without History of COVID-19

BACKGROUND: Recent studies have shown that SARS-CoV-2 can have neuropsychiatric consequences and has the ability to penetrate the blood-brain barrier. If SARS-CoV-2 has a specific route of entry into the brain, it may leave imprints in the form of specific changes in brain morphology. Older individuals are most vulnerable to the neuropsychiatric COVID-19 complications. This study aims to compare regional brain volumes in older adults individuals with and without COVID-19 history (COVID+ and COVID-, respectively). METHODS: Individuals over 65 years old who applied for treatment to the Memory Clinic (Mental-Health Clinic No. 1 named after N.A. Alexeev, Moscow, Russia) were assessed between October 2020 and April 2021. Their COVID-19 history was determined by the self-report and COVID-19 certificate. Individuals with severe neuropsychiatric or acute or severe chronic somatic or infectious disease and those taking medications potentially affecting cognitive functioning were excluded. All participants underwent MRI examinations followed by image segmentation and morphometric quantitative analysis. Regional brain volumes were compared in COVID+ and COVID- people. RESULTS: 207 participants were included in the study. The COVID+ group consisted of 24 participants. The comparison between groups revealed statistically significant differences in left amygdala area (median 1199.3 mm 3 in COVID+ vs. 1263.7 mm 3 in COVID-) and right postcentral gyrus volumes (median 8055.5 mm 3 in COVID+ vs. 8434.0 mm 3 in COVID-). Then case-control analysis was performed in individuals matched for gender, age and common somatic causes of structural brain changes (hypertension and/or diabetes mellitus type 2) for 22 subjects in each group. Statistically significant differences in regional brain volumes between groups were absent.


INTRODUCTION
Studies have shown that advanced age is one of the greatest risk factors for higher severity and worse outcome of COVID-19 [1][2][3][4] with neurological and psychiatric symptoms affecting 33-62% of patients within six months of recovery [5]. Some researchers suggest that SARS-CoV-2 may have an ability to invade the brain via the olfactory tract, circumventricular organs, leaky blood-brain barrier due to inflammation or direct damage of brain vascular endothelium or with migrating immune cells [6].
If specific routes of SARS-CoV-2 to enter the brain do exist than it may be associated with the distinct patterns of brain morphology changes. A recent article systematically reviewed brain imaging case series, casecontrol and cohort studies in patients with  and found that alterations associated with COVID-19 predominated in the olfactory brain network, limbic and prefrontal structures [7]. About half of these studies used only visual evaluation of MRI scans, and among studies that used image statistical processing approach there were none that used matched control of age, gender and comorbidity (hypertension and/or diabetes mellitus type 2) in brain morphology analysis in older population affected and non-affected by COVID-19.
Keywords: COVID-19; SARS-CoV-2; elderly people; MRI; brain morphology; regional brain volumes Image processing and segmentation were performed using Freesurfer v6.0 software package and morphological indices (thickness and volume of gray matter, volume of white matter, gyrification index, volume of gray matter by subcortical structures, etc.) were derived. Regional brain volumes (gray matter and white matter volumes) were selected for the purpose of this work.

Statistical processing
Database was extracted on 9 September 2021. The primary endpoint was finding the differences in MRI volumes between those who experienced COVID-19 (COVID+) and those who had not (COVID-) in a whole study population.
Mann-Whitney test was used to compare continuous variables while participant's distribution by the categorical variables between groups was performed using Fisher Moreover, none of these studies evaluated brain changes specifically in older (65+ years old) population. Therefore, case-control study of MRI scans using image statistical processing approach and matched case-control analysis of elderly individuals with and without COVID-19 history is relevant.
Since October 2020 a longitudinal cohort study of patients with mild cognitive impairment (MCI) compared to healthy control was initiated in Psychiatric clinical hospital No 1 (Moscow, Russia) to identify unmet needs of MCI patients during COVID-19 pandemic. One of the aims of this study is a MRI-morphometry of brain scans in patients with mild cognitive impairment (MCI) and healthy control with and without reported COVID-19 history.
Our research question was: are there specific brain alterations in older people with reported history of COVID-19 infection compared to those without COVID-19 history?

Study population
This study is a part of multidisciplinary project 'Impact of the COVID-19 pandemic on the mental health of the elderly' and is supported with grant of RFFI 20-04-60546.
Individuals over 65 years old were eligible in the study.

Study population
Overall, 207 participants had processed Visit 1 MRI scan information in the study database as of 9 th September 2021.
Among them 24 indicated that they had a history of COVID-19 (either outpatient or hospitalization) before Visit 1 of the study (before October, 2020), with 4 of 24 (16.7%) were hospitalized with COVID-19. Thus, 24 participants were categorized in COVID+ group and n=183 in COVID-group. Their socio-demographic and medical characteristics are summarized in the Table 1. exact test. All regional brain volumes were n-1 normalized to standardize the variables using the unbiased standard deviation. Group profiles of brain volumes means and medians of received z-scores were than additionally visually analyzed using parallel coordinates plots. In all statistical tests two-tails p <0.05 considered as statistically significant. We didn't use multiple p correction because it would preclude finding of any significant differences given large number of regions (109 regions) to compare highly variable volumes. To compare regional brain volumes, we also used case-control approach in subpopulations of COVID+ and COVID-groups matched by age, gender and history of hypertension and type II diabetes. The following strategy was used to search for matched subjects: exact matches on gender, hypertension stage and type II diabetes and fuzzy search for age within 2 years range.

Regional brain volumes in COVID+ and COVID-groups
Comparison of brain regional volumes in whole study population (n=207) revealed differences only in two

Figure 1. Significant differences between COVID+
and COVID-groups on regional brain volumes (Mann-Whitney test) in whole study population (n=207).

Figure 2. Standardized brain volumes profiles in COVID+ and COVID-participants (n=207).
Note: All means (solid line) or medians (dashed line) of residuals of regional brain volumes in both COVID+ (red) and COVID-(green) participants were within 1 standard deviation from mean values. Olfactory tract projections are considered as a possible gateway of SARS-CoV-2 invasion into the brain. Amygdala None of means or medians of z-scores exceed 1, reflecting that there were no differences in regional brain volumes larger than one standard deviation of study population means, though, generally most of z-scores are slightly decreased in COVID+ against COVID-subjects.

DISCUSSION
Our study revealed differences in regional brain volumes between COVID+ and COVID-groups: right postcentral Note: All means (solid line) or medians (dashed line) of residuals of regional brain volumes in both COVID+ (red) and COVID-(green) participants were within 1 standard deviation from mean values.

Mean
Median COVID+ COVID-type II diabetes that also associated with brain structure changes. Moreover, these comorbid disorders are known to be associated with increased risk of COVID-19 complications [15,16]. Thus, underestimation of these medical conditions may lead to bias in the results.
In our study we attempted to control these conditions.
When compared subpopulations matched on gender, age and comorbid medical condition (hypertension and/ or type II diabetes) regional brain volume differences disappeared. This may be due to that controlled conditions account for more of the variability in regional brain volumes than COVID-19 history. The median MMSE score before matching was mathematically lower in the COVID-compared to COVID+ group and in opposite to our results one could expect that if COVD-19 is associated with brain leisure than the matching on MMSE score will make differences even larger. Nevertheless, as expected, total MMSE (and MoCA) scores in whole study population showed significant positive correlation with total grey matter volume and negative correlations with total CSF volume showing adequate association between structural brain alterations and cognitive functions. This may indicate that an alternative explanation can take place: the decrease in sample size reduces the statistical power to detect brain differences between groups.

Study limitations
One of the study limitations was that COVID-19 status was not confirmed in the laboratory at the study entry but is a part of limbic system that receives projections from olfactory bulbs [11,12]. According to recent research [13] a comparison of MRI scans before and after COVID-19 found that patients after COVID-19 had greater grey matter loss in the central nucleus of the amygdala than those who had no history of COVID-19. Another study [14] found hypometabolism in the right temporal lobe, including amygdala in patients with long COVID-19.
Contrary, in our study we did not find significant morphological changes in other brain regions. One possible explanation is that our study included population who experienced mild forms of COVID-19 (only 4 of 24 reported that they were hospitalized) that did not associate with brain tissue lesions. This explanation is supported by the fact that we did not find bilateral morphological changes that can be expected from olfactory route of virus penetration through bloodbrain barrier. Also, this may result from a selection bias while enrolling patients into the study, those who had more severe forms of COVID-19 may not had applied for treatment to Memory clinic due to either restriction on transit for people older than 65 years or precaution/incapacity to move across the city.
Many MRI studies of brain structural changes in patients experienced COVID-19 reported alteration in different brain areas other than olfactory system [7]. Nevertheless, most of these studies did not include control sample and studied patients with COVID-19 severe enough to be hospitalized. The recent brain imaging study before and after COVID-19 included patients with second MRI scan after 35-407 days after recovery from in most cases mild forms of the disease in comparison to healthy control [13].
This study revealed decrease of cortical thickness in the lateral orbitofrontal cortex, generally greater brain size reduction, increase in diffusion indices and CSF volume.
Though, a comparison of hospitalized cases with either non-hospitalized or control subgroups failed to detect marked differences due to decrease in the sample size.
While this study that was conducted on a population that was very close to our study revealed significant differences in brain morphology, it didn't account control for comorbid disorders like hypertension and

CONCLUSION
We didn't find definite associations of any regional brain volumes differences with COVID-19 history in people older than 65 years. Our study results are based on a population exposed to relatively mild forms of COVID-19. Thus, given study limitations, these results can't be generalized to other people who recovered from COVID-19. Further better balanced and controlled and larger studies on an association of brain morphology with COVID-19 experience stratified by the severity in older people would help to disentangle relationships between COVID-19 severity and brain morphology changes.

Supplementary data
Supplementary material related to this article can be found, in the online version, at doi: 10.17816/CP145