key: cord-0789332-1pmd4b7f
authors: Silva, L. S.; Joao, R. B.; Nogueira, M. H.; Aventurato, I. K.; de Campos, B. M.; de Brito, M. R.; Alvim, M. K. M.; Ludwig, G. V. N.; Rocha, C.; Souza, T. K. A. S.; da Costa, B. A.; Mendes, M. J.; Waku, T.; Boldrini, V. d. O.; Brunetti, N. S.; Baptista, S. N.; Schmitt, G. d. S.; Sousa, J. G. D. d.; Cardoso, T. A. M. d. O.; Vieira, A. S.; Santos, L. M. B.; Farias, A.; Cendes, F.; Yasuda, C. L.
title: Functional and microstructural brain abnormalities, fatigue, and cognitive dysfunction after mild COVID-19
date: 2021-03-24
journal: nan
DOI: 10.1101/2021.03.20.21253414
sha: d930f6f7b051c31d3b339ef900a31cc05c5636f2
doc_id: 789332
cord_uid: 1pmd4b7f

Although post-acute cognitive dysfunction and neuroimaging abnormalities have been reported after hospital discharge in patients recovered from COVID-19, little is known about persistent, long-term alterations in people without hospitalization. We conducted a cross-sectional study of 87 non-hospitalized recovered individuals 54 days after the laboratory confirmation of COVID-19. We performed structured interviews, neurological examination, 3T-MRI scans with diffusion tensor images (DTI) and functional resting-state images (fMRI). Also, we investigated fatigue, anxiety, depression, somnolence, language, memory, and cognitive flexibility, using validated instruments. Individuals self-reported a high frequency of headache (40%) and memory difficulties (33%). The quantitative analyses confirmed symptoms of fatigue (68%), excessive somnolence (35%), anxiety (29%), impaired cognitive flexibility (40%) and language impairment (33%). There were widespread cerebral white matter alterations (mainly characterized by increased fractional anisotropy), which correlated with abnormal attention and cognitive flexibility. The resting-state fMRI networks analysis showed severely disrupted brain hyperconnectivity and loss of resting-state networks specificity.

Studies have consistently reported neurological manifestations of COVID-19 (Ellul et al., 2020; Mao et al., 2020) . However, little is known about the long-term neurological events associated with the novel coronavirus, severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2). (De Felice, Tovar-Moll, Moll, Munoz, & Ferreira, 2020) . While most individuals will recover from respiratory symptoms, the longterm course of post-covid fatigue and cognitive dysfunction is uncertain. One French study identified dysexecutive syndrome in 15/45 (33%) of patients with severe infection (Helms et al., 2020) . Another

Chinese study recruited 29 patients (after hospitalization) and reported cognitive dysfunction after their recovery (Hetong Zhou et al., 2020) . However, there is a limitation of the present understanding of the residual long-term neurological and cognitive dysfunctions (including the nature, duration, and pathophysiology) in individuals who recovered from COVID-19 (Alwan et al., 2020; Callard & Perego, 2020; Ritchie, Chan, & Watermeyer, 2020) , especially in those with mild infection who did not require hospitalization.

Although the neuroinvasion of COVID-19 has been demonstrated with the confirmation of the virus's presence in brain autopsies (Puelles et al., 2020) , the neural mechanisms underlying both neurological and neuropsychiatric symptoms (acute and chronic) remain unclear. One study with 60 patients (3 months after hospitalization) identified grey and white matter abnormalities with MRI analyses (Lu et al., 2020) but did not include cognitive tests. Given the lack of information about the long-term effects of COVID-19 after mild infection (Garg, Arora, Kumar, & Wig, 2020; Tenforde et al., 2020) we investigated the nature of persistent neurological symptoms in this subgroup of patients (without hospitalization), combining clinical data with cognitive tests, structural and functional MRI analyses.

. CC-BY-NC-ND 4.0 International license It is made available under a perpetuity.

is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted March 24, 2021. ; https://doi.org/10.1101 https://doi.org/10. /2021 

We conducted a cross-sectional analysis of data from a longitudinal observational study designed to evaluate post-acute neurological alterations (clinical and neuroimaging) related to the COVID-19. We used social media to advertise our study with an online questionnaire (Rayhan et al., 2013) (the questionnaire is presented in the Supplementary Table 13 ). We successively recruited the first 87 responders (who did not require hospitalization and presented a confirmed diagnosis of COVID-19 (by a polymerase chain reaction [PCR] test or confirmed presence of IgM or IgG antibodies) to visit our center and perform the four steps of the complete protocol: a personal structured interview and neurological examination (performed by certified neurologists), 3T MRI acquisition, neuropsychological testing, and blood sample collection at our University Hospital (University of Campinas, Campinas, Brazil).

Due to the uncertainties related to cognitive impairment associated with the new coronavirus, we performed an exploratory neuropsychological evaluation of recovered individuals. We intentionally selected tests to evaluate different cognitive domains (detailed description is presented in the supplementary material), including language (Verbal Categorical Fluency Test (S. D. Brucki, et al. , 1997; S. M. Brucki & Rocha, 2004) and Phonemic Verbal Fluency Test (Tombaugh, Kozak, & Rees, 1999) ), episodic memory (Logical Memory subtest from the Wechsler Memory Scale (WMS-R) (Bolognani et al., 2015; Weschsler, 1987) ) and cognitive flexibility (Trail Making Test (TMT) (Campanholo et al., 2014; Strauss E, 2006) with parts A and B).

We calculated z-scores for the results of the neuropsychological tests based on Brazilian standard and scaled scores. We controlled the effects of the age or schooling in a separate analysis using multiple linear regression residuals when normative data covered only one of these variables. For each test, the function was categorized as preserved (z-score > -0.66), including average, high average, above average, and exceptionally high scores), . CC-BY-NC-ND 4.0 International license It is made available under a perpetuity.

is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted March 24, 2021. ; https://doi.org/10.1101 https://doi.org/10. /2021 low average score (z-score between -0.7 and -1.26); below-average score (z-score between -1.32 and -1.82); and exceptionally low score (z-score < -1.96) (Beauchamp et al., 2015; Guilmette et al., 2020) .

We quantified anxiety symptoms with BAI (Beck Anxiety Inventory), and symptoms of depression with BDI-II (Beck Depression Inventory II); additionally we investigated fatigue with Chalder Fatigue Questionnaire (CFQ-11) (Chalder Fatigue Questionnaire) (Jackson, 2015; Townsend et al., 2020) and excessive daytime sleepiness with ESS (Epworth Sleep Scale) (Walker, Sunderram, Zhang, Lu, & Scharf, 2020) . (Details of these tests are described in the supplementary file).

We acquired all images (patients and controls) in a 3T Philips Achieva scanner using the same protocols: i) structural 3D T1-weighted images with isotropic voxel (1x1x1 mm³), acquired in the sagittal plane with 180 slices, TE = 3.2 ms, TR = 7 ms, matrix = 240x240, flip angle = 8 and FOV = 240x240mm2) ii) Resting-state: echo planar functional images with isotropic voxel (3x3x3 mm³), acquired on the axial plane with 40 slices, no gap, matrix of 80x80, flip angle = 90º, FOV = 240x240mm 2 , TR=2s, TE=30ms and 180 dynamics, resulting in a six minutes scan; iii) DTI was acquired with a single-shot EPI technique (voxel size = 2×2×2 mm³, 32 gradient directions, max b-factor = 1000 s/mm², FOV = 256x256 mm²) (Garcia et al., 2019; Hatton et al., 2020) . Additional details of MRI protocol acquisitions for post-COVID individuals are presented in the supplementary file.

We used a semi-automated tractography method based on a deterministic approach implemented in ExploreDTI (http://www.exploredti.com) and previously described in detail. As an exploratory study, we used tractography to delineate different tracts, including the commissural tracts (corpus callosum divided into three parts: Genu, body (BCC) and splenium (SPL)), association tracts (inferior longitudinal fasciculus (ILF), inferior fronto-occipital fasciculus IFO and uncinate fasciculus (UNC)), limbic tracts, dorsal . CC-BY-NC-ND 4.0 International license It is made available under a perpetuity.

is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted March 24, 2021. ; https://doi.org/10.1101 https://doi.org/10. /2021 cingulum (Dorsal-CING) and parahippocampal cingulum (Parahippocampal-CING)) and one projection tract (corticospinal tract (CST)) (Campos et al., 2015; Lebel et al., 2012; Liu, Concha, Lebel, Beaulieu, & Gross, 2012) . The values of diffusivities (FA, fractional anisotropy; MD, mean diffusivity; RD, radial diffusivity; AD, axial diffusivity) for each tract of each subject were averaged across all voxels in the entire tract, ensuring that each voxel was counted only once Liu et al., 2012) . Only tracts with at least five streamlines were included in the study (Lebel & Beaulieu, 2009 ). For bilateral tracts, independent right and left values were obtained separately; for the midline tracts (subdivisions of corpus callosum) a single value was calculated for each segment.

Given the exploratory nature of this study, we aimed towards a global analysis of functional connectivity and opted to perform a whole-brain study of this group of recovered individuals. Therefore, we did not restrict our analysis to a single or specific resting-state network; instead, we intentionally selected the ROIto-ROI approach and included the twelve functional resting-state networks (anterior salience, posterior salience, auditory, basal ganglia, dorsal default mode network (DMN), ventral DMN, language, left executive control network (LECN), right executive control network (RECN), sensorimotor, visual and visuospatial) to evaluate how these networks would deviate from the typical pattern observed in healthy volunteers.

We performed the functional connectivity (FC) analysis with the UF²C toolbox (https://www.lniunicamp.com/uf2c) within SPM12 (http://www.fil.ion.ucl.ac.uk/spm/, using MATLAB 2019b)(Brunno Machado de Campos, Casseb, & Cendes, 2020; B. M. de Campos, Coan, Lin Yasuda, Casseb, & Cendes, 2016) . For this analysis, we initially included 217 subjects (134 controls and 83 patients).

The preprocessing steps followed the UF²C standard pipeline and (briefly) were based on functional image realignment, normalization to the MNI-space (Montreal Neurologic Institute Standard template), coregistration with T1WI image, framewise displacement (FD) estimation, and smoothing with a kernel of 6x6x6 mm (FWHM)(Brunno Machado de Campos et al., 2020) . is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted March 24, 2021. ; https://doi.org/10.1101/2021.03.20.21253414 doi: medRxiv preprint Supplementary Figure 2) . The final number of subjects included in the analysis was 197 (120 controls and 77 patients, balanced for age (p=0.62) and sex (p=0.49)).

We performed a region of interest (ROI) study, with ROI-to-ROI FC analysis using 70 ROIs from the 12 functional networks (Supplementary table 11) . These ROIs are a sub-set of the Shirer et al. 2012 (Shirer, Ryali, Rykhlevskaia, Menon, & Greicius, 2012) atlas available on https://findlab.stanford.edu/functional_ROIs.html. The time-series extracted from each ROI followed homogenization procedures, excluding voxels non-functionally representatives, CSF and high probability WM voxels. The individual connectivity matrices were estimated using Pearson's correlation between each possible pairs of ROIs, resulting in a 70x70 matrix for each volunteer. We performed the second-level analysis (group inference) using a two-sample T-test (p<0.05, FDR corrected) including the connectivity matrices of each group converted to Z-score (Fischer's z-transformation).

Clinical data were analyzed with SPSS 22. We used Chi-square and Fisher´s exact test for categorical data.

For continuous variables, we used non-parametric Kruskal-Wallis tests (for variables with non-normal distribution) and Student´s T-test for those with a normal distribution.

For the analyses of diffusion tensor imaging (DTI) data between patients and controls, we used separated models for each diffusivity (FA, MD, RD and AD (covaried for age)), considering the multivariate analysis of variance (MANOVA) for segments of the corpus callosum and separated, repeated measures ANOVA (RM-ANOVA) for the analyses of bilateral tracts. We applied FDR (Genovese, Lazar, & Nichols, 2002) adjustments with R to adjust for multiple comparisons in each model, (R Core Team, 2020).

Pearson correlations were performed between the neuropsychological data and DTI parameters for specific tracts according to a priori hypothesis for each test, reducing the need to correct for multiple comparisons.

For language, we investigated the correlation between phonological fluency and FA from ILF (Del Tufo, Earle, & Cutting, 2019); for the Trail Making Test, we investigated the association between the scores and the FA of right ILF, as previously investigated in healthy-aging individuals (Perry et al., 2009 ). However, we were aware of some dissociations of DTI parameters and cognitive impairment (for example, alterations . CC-BY-NC-ND 4.0 International license It is made available under a perpetuity.

is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted March 24, 2021. ; https://doi.org/10.1101/2021.03.20.21253414 doi: medRxiv preprint of mean diffusivity without abnormality of FA (Goldstein et al., 2009) ) and therefore explored the correlations between DTI indices (all four measurements) of both parahippocampal cingulae (right and left sides) and episodic memory.

This study was approved by the Research Ethics Committee of the University of Campinas (Certificate of Ethical Appreciation Presentation -CAAE 31556920.0.0000.5404), and all subjects signed a consent form to participate.

We evaluated 87 individuals (64 women, median age 36 [range 18-71]), with a median interval (between diagnosis and personal interview) of 54 days (range 16-120 days). During the acute phase, patients reported a median of 4 symptoms (range 0-10, with five asymptomatic individuals), while in the post-acute stage (reported during the structured interview), they informed a median of 2 symptoms (range 0-13, with 25 asymptomatic individuals (28.7%)). The most frequent post-acute symptoms were fatigue (43.7%), headache (40%), memory difficulties (40%), anosmia (31%), and somnolence (18%) (The types and percentages of these symptoms are displayed in Supplementary Figures 3-4) . Interestingly, fatigue was reported by 38 individuals (43.7%) and was mostly combined with other symptoms such as headache (25 subjects), memory difficulties (17 subjects) and somnolence (14 subjects).

In terms of post-COVID-19 neurological examination and interview, we identified abnormalities in 11 (12.6%) individuals (described in Supplementary Table 1), although visual inspection of MRI was normal (appropriate for age) for eighty-six (98.9%) individuals. One radiologist (JGDS) visually inspected all the structural MRIs and identified one hemangioma in the temporo-occipital region (the subject confirmed to have it before the infection); we, therefore, excluded this subject from the MRI post-processing studies.

In addition to the structured interview, sixty-five subjects answered the fatigue questionnaire (CFQ-11) with a median of 15 points (range 0-32) and the Epworth sleepiness scale (ESS), with a median of 9 points (range 0-21). Differently from the proportion of symptoms reported during the interview (fatigue in 43.7%, and . CC-BY-NC-ND 4.0 International license It is made available under a perpetuity.

is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted March 24, 2021. ; https://doi.org/10.1101/2021.03.20.21253414 doi: medRxiv preprint somnolence in 18%), the binary classification (presence or absence of symptoms) resulting from the scores showed a higher proportion of symptoms of fatigue in (44/65, 68% of individuals) and excessive daytime sleepiness (23/65, 35%).

The Pearson correlation between CFQ-11 and ESS was moderate (r=0.44, p<0.001). Besides, excessive daytime sleepiness was more frequent in individuals with fatigue (20/44) than in those without fatigue (3/21), p= (0.025).

. CC-BY-NC-ND 4.0 International license It is made available under a perpetuity.

is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint 

We performed the neuropsychological evaluation for a subset of 78 individuals (59 women, with a median age of 36 years (range 18-70)). Nine individuals did not perform the evaluation due to lack of time; no other reasons were reported. Approximately 18% of subjects presented depression symptoms (BDI-II >13), and 29% showed anxiety symptoms (BAI > 10). We identified a correlation between symptoms of depression with the scores of CFQ-11 (BDI-II r=0.47, p<0.001, controlled for BAI scores).

In terms of cognitive performance, we identified abnormal performance (below low and exceptionally low scores) in 33% of phonological fluency, 30% of TMT-A, and 40% of TMT-B tests. ( is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint 

We obtained paired neuropsychological and DTI data from 71 participants. The right ILF-FA was mildly correlated with TRAIL-B (r=0.3; p=0.015). However, we did not identify correlations between ILF and Phonological fluency, neither between the logical memory scores (immediate and delayed recall) and diffusivities in the parahippocampal cinguli. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted March 24, 2021. ; https://doi.org/10.1101 https://doi.org/10. /2021 

We detected persistent headache, fatigue, excessive somnolence, anxiety, and cognitive dysfunction along with structural and functional brain MRI abnormalities two months after the acute COVID-19 infection of non-hospitalized individuals. Cerebral white matter alterations were widespread and correlated with abnormal attention and cognitive flexibility. Brain connectivity was severely disrupted, with excessively and chaotically connected regions associated with significant loss of specificity of eleven (out of twelve investigated) individual standard resting-state networks (anterior and posterior salience, basal ganglia, ventral and dorsal DMN, language, left and right executive control networks, sensorimotor, visual and mostly, the visuospatial network).

Two months after the acute period, the most frequently reported symptoms were fatigue (43.7%), headache (40%), and memory difficulties (33%), similar to the results from a preprint meta-analysis (Lopez-Leon et al., 2021) . Interestingly, the quantitative analysis of the CFQ-11 and ESS revealed a higher proportion of symptomatic individuals (68% with fatigue and 35% with excessive daytime somnolence), compared to the proportions of symptoms reported during the structured interview. While the presence of depression symptoms (18% of subjects, according to the BDI-II scores) was higher than reported (1%), the frequency of anxiety symptoms (29% of subjects) according to the BAI scores was close to those self-reported (23%).

These findings are in accordance with one previous study (Townsend et al., 2020) , as we observed a similar median score on CFQ-11 (median of 16 points) in our group of non-hospitalized participants. Besides, while we detected a positive correlation between fatigue scores and intensity of depression symptoms (BDI-II scores), they also described the elevated proportion of individuals with fatigue in participants with a history of anxiety/depression. The proportion of women responders was higher than men, probably due to our method of recruitment nature. There is as yet no clear evidence if sex is a risk factor for developing or perpetuating long-term effects of COVID-19 (Lopez-Leon et al., 2021) ; however, some studies have demonstrated that post-. CC-BY-NC-ND 4.0 International license It is made available under a perpetuity.

is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted March 24, 2021. ; https://doi.org/10.1101 https://doi.org/10. /2021 infection fatigue seems to be more frequent in women (Townsend et al., 2020; Xiong et al., 2021) . As we observed high rates of fatigue in our study group, one possible hypothesis to justify this disproportion is that more affected women were more interested in enrolling than men.

Recent studies have confirmed long-term cognitive dysfunction in survivors of COVID-19 (from ICU and ward hospitalization) (Almeria, Cejudo, Sotoca, Deus, & Krupinski, 2020; H. Zhou et al., 2020) It was surprising in our data that a high proportion of subjects (with a median education of 15 years) performed poorly, specifically in the Trail Making Test. It can be partially explained by our convenience sample, which may have attracted more symptomatic participants. Besides, the poor performance on Trail

Making Test -B test correlated with abnormalities in the WM structure (reduced FA) of the right inferior longitudinal fasciculus (as previously reported in one study of healthy aging (Perry et al., 2009 ), which has . CC-BY-NC-ND 4.0 International license It is made available under a perpetuity.

is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted March 24, 2021. ; https://doi.org/10.1101 https://doi.org/10. /2021 been associated with visual processing (Shin et al., 2019) . It also correlated with the thinner cortex in the rectus gyrus (submitted data).

One recent study of six patients with severe disease (imaged 19 days after admission) identified overall reduced FA values associated with increased MD values (Newcombe et al., 2020) , which typically suggest microstructural impairment of white matter (possibly resultant from inflammation and edema). On the contrary (after two months of acute diagnosis), we observed higher FA values in non-hospitalized post-covid subjects, similar to the findings reported by (Lu et al., 2020) that included 60 hospitalized patients evaluated three months after discharge. However, while they (Lu et al., 2020) identified reduced values for MD, RD and AD, we found increased values of these parameters, mostly axial diffusivity (AD). The disparities between these results may be explained by some factors, including the timing of image acquisition, as the different findings may reflect different stages of white matter insult. While the reduced FA values in the acute phase may suggest overall white matter damage, elevated FA values in the post-acute stage may suggest neuroplasticity phenomenon (as observed in neurodegenerative disease (Mole et al., 2016) ).

Contrary to the original hypothesis of a "mild disease" our results suggest that the new coronavirus may negatively impact the white matter microscopically (as the visual inspection of MRI was normal for all the "mildly infected" participants) in a widespread pattern. Besides, the concomitant identification of neurocognitive impairment in these individuals with elevated FA values indicates that increased Fractional anisotropy may not necessarily signify better function (Alba-Ferrara & de Erausquin, 2013) .

Identifying structural brain alterations in both mildly and severely affected individuals suggests a more specific effect of the new coronavirus in the white matter than the consequences of sepsis or severe illness.

Another intriguing question we intend to analyze with the longitudinal data is the duration of such abnormalities; to this point, we cannot classify these alterations as transient or permanent. Larger groups of patients with longitudinal data are necessary to further understanding the neurological impact of the new coronavirus.

. CC-BY-NC-ND 4.0 International license It is made available under a perpetuity.

is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted March 24, 2021. ; https://doi.org/10.1101 https://doi.org/10. /2021 

Although visual inspection of structural MRI excluded macrostructural abnormalities in these mildly infected patients, the examination of functional connectivity revealed an intriguing pattern of increased functional connectivity among regions of several resting-state networks. The functional hyperconnectivity of resting-state networks here identified has been portrayed as a paradoxical response to neural injury (Hillary & Grafman, 2017) and previously described in other neurological conditions (Parkinson's disease (Gorges et al., 2015) , traumatic brain injury (Hillary et al., 2014) , mild cognitive impairment (Gour et al., 2011) ). One hypothesis is that such a strong cerebral response attempts to restore the composition of original networks lost due to the neural insult; another possibility is that after the neurological injury, the hyperconnectivity is required to recruit new, alternative pathways to replace the damaged ones (Hillary & Grafman, 2017) .

Besides, it seems that the hyperconnectivity pattern represents an early stage response for some neurodegenerative processes (Gorges et al., 2015; Hawellek, Hipp, Lewis, Corbetta, & Engel, 2011) , and declines over time as the disease advances (Hillary & Grafman, 2017; Olde Dubbelink et al., 2014) . The uncertainties related to the abnormal hyperconnectivity in post-infected patients require longitudinal data to confirm (or not) whether the new coronavirus can trigger a neurodegenerative process.

From previous studies, hyperconnectivity in the default mode network has been associated with cognitive impairment (Whitfield-Gabrieli et al., 2009; Yasuda et al., 2013) , as well as the deposition of beta-amyloid (Buckner et al., 2005) (116), a marker of Alzheimer disease. Interestingly, individuals with schizophrenia present brain functional hyperconnectivity associated with cognitive impairment in different domains (Krukow, Jonak, Grochowski, Plechawska-Wojcik, & Karakula-Juchnowicz, 2020; Whitfield-Gabrieli et al., 2009) . We have previously associated impaired deactivations of the DMN with poor performance on verbal fluency task, in individuals using topiramate (an anti-seizure medication, also used for migraine prophylaxis) (Yasuda et al., 2013) . While some studies have demonstrated a negative impact of is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted March 24, 2021. ; https://doi.org/10.1101 https://doi.org/10. /2021 we explored the relationship between several networks and detected hyperconnectivity among distinct resting-state networks. Unfortunately, we cannot explain the underlying physiopathological mechanisms.

Neither can we predict the duration of such states in the post-infected individuals. Given the widespread pattern of alterations, we speculate that these changes may be related to neurological symptoms (such as excessive sleepiness, headache and fatigue) and cognitive dysfunction.

One intriguing finding is the severe impairment of visuospatial network (which included the most impaired ROI in our analysis (ROI 64, left superior parietal lobule/supramarginal gyrus/postcentral gyrus/angular gyrus). As previous studies have confirmed the importance of visuoperceptual abilities (along with working memory and other high-order processes (Sánchez-Cubillo et al., 2009; Varjacic, Mantini, Demeyere, & Gillebert, 2018) ) for the performance of Trail Making Test, we raise the hypothesis that the dysfunction of the visuospatial network may be associated with the lower scores produced for both Trail Making Tests (A and B) by our group of individuals with 15 years of education. This idea is supported by the neural correlates of TMT performance (which suggest the involvement of "large-scale networks including prefrontal and parietal structures" (Varjacic et al., 2018) ), which coincides with the localization (parietal region) of the most compromised ROI in our FC analysis.

Our longitudinal analyses will allow us not only to comprehend how transient (or permanent) these changes are but will provide more evidence related to possible undesired (and unexpected) consequences of chronic hyperconnectivity in the case of persistent changes of brain dynamics networks.

This cross-sectional study with a convenience sample restrains the generalization of our findings. However, in facing the limited understanding of the new coronavirus's neurological impact, we believe our initial observations raise concerns about possible long-term impairment in mildly infected patients. Our longitudinal analyses with a larger sample will answer some of the unsolved questions related to these longterm, persistent symptoms related to mood, cognition and brain connectivity. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted March 24, 2021. ; https://doi.org/10.1101 https://doi.org/10. /2021 

Our findings suggest the new coronavirus affects the brain in individuals who did not require hospitalization, with persistent fatigue, headache, memory problems and somnolence even after two months of their diagnosis. We detected cognitive impairment in these patients, along with white matter abnormalities and hyperconnectivity among distinct resting-state networks. The degree of brain alterations and the severity of cognitive dysfunction raises the attention for the necessity of extensive longitudinal studies of chronic neuropsychiatric symptoms in post-COVID-19 infected individuals. Specific treatment for symptoms and neurorehabilitation strategies may be necessary to improve the quality of life and cognitive function for those with persistent limitations after the acute phase.

The authors declare they have no competing financial interests related to this study.

Data collected and analyzed for this study will be available at the University database. Neuroimaging raw data will be available upon reasonable request.

We thank the team of assistants who worked on the consecutive recruitment of patients: Andrea Ismara de Araújo Ruas, Lilian Cristina dos Santos Capelli and Sonia Neves Romeu Silva. We also are grateful for the work of the Radiology's team. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted March 24, 2021. ; https://doi.org/10.1101 https://doi.org/10. /2021 

What does anisotropy measure? Insights from increased and decreased anisotropy in selective fiber tracts in schizophrenia

Cognitive profile following COVID-19 infection: Clinical predictors leading to neuropsychological impairment

From doctors as patients: a manifesto for tackling persisting symptoms of covid-19

Empirical Derivation and Validation of a Clinical Case Definition for Neuropsychological Impairment in Children and Adolescents

Development of alternative versions of the Logical Memory subtest of the WMS-R for use in Brazil

Dados normativos para o teste de Fluência Verbal categoria animais em nosso meio

Category fluency test: effects of age, gender and education on total scores, clustering and switching in Brazilian Portuguese-speaking subjects

Molecular, structural, and functional characterization of Alzheimer's disease: evidence for a relationship between default activity, amyloid, and memory

How and why patients made Long Covid

Performance of an adult Brazilian sample on the Trail Making Test and Stroop Test

White matter abnormalities associate with type and localization of focal epileptogenic lesions

UF2C -User-Friendly Functional Connectivity: A neuroimaging toolbox for fMRI processing and analyses. SoftwareX

Large-scale brain networks are distinctly affected in right and left mesial temporal lobe epilepsy

Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2) and the Central Nervous System

The impact of expressive language development and the left inferior longitudinal fasciculus on listening and reading comprehension

Neurological associations of COVID-19

Anxiety and depression symptoms disrupt resting state connectivity in patients with genetic generalized epilepsies

The "post-COVID" syndrome: How deep is the damage

Thresholding of statistical maps in functional neuroimaging using the false discovery rate

To rise and to fall: functional connectivity in cognitively normal and cognitively impaired patients with Parkinson's disease

Basal functional connectivity within the anterior temporal network is associated with performance on declarative memory tasks

American Academy of Clinical Neuropsychology consensus conference statement on uniform labeling of performance test scores

White matter abnormalities across different epilepsy syndromes in adults: an ENIGMA-Epilepsy study

Increased functional connectivity indicates the severity of cognitive impairment in multiple sclerosis

Neurologic Features in Severe SARS-CoV-2 Infection

Injured Brains and Adaptive Networks: The Benefits and Costs of Hyperconnectivity

The rich get richer: brain injury elicits hyperconnectivity in core subnetworks

The Chalder Fatigue Scale (CFQ 11)

Resting-state hyperconnectivity within the default mode network impedes the ability to initiate cognitive performance in first-episode schizophrenia patients

Lateralization of the arcuate fasciculus from childhood to adulthood and its relation to cognitive abilities in children

Diffusion tensor imaging of white matter tract evolution over the lifespan

Mesial temporal sclerosis is linked with more widespread white matter changes in temporal lobe epilepsy

More than 50 Long-term effects of COVID-19: a systematic review and metaanalysis. medRxiv

Cerebral Micro-Structural Changes in COVID-19 Patients -An MRI-based 3-month Follow-up Study. EClinicalMedicine

Neurologic Manifestations of Hospitalized Patients With Coronavirus Disease

Increased fractional anisotropy in the motor tracts of Parkinson's disease suggests compensatory neuroplasticity or selective neurodegeneration

Neuroanatomical substrates of generalized brain dysfunction in COVID-19

Functional connectivity and cognitive decline over 3 years in Parkinson disease

White matter tracts associated with set-shifting in healthy aging

Multiorgan and Renal Tropism of SARS-CoV-2

R: A language and environment for statistical computing

The cognitive consequences of the COVID-19 epidemic: collateral damage? Brain Communications

Construct validity of the Trail Making Test: role of task-switching, working memory, inhibition/interference control, and visuomotor abilities

Inferior Longitudinal Fasciculus' Role in Visual Processing and Language Comprehension: A Combined MEG-DTI Study

Decoding subject-driven cognitive states with whole-brain connectivity patterns

A compendium of neuropsychological tests. Administration, norms and commentary

Symptom Duration and Risk Factors for Delayed Return to Usual Health Among Outpatients with COVID-19 in a Multistate Health Care Systems Network -United States

Normative data stratified by age and education for two measures of verbal fluency: FAS and animal naming

Persistent fatigue following SARS-CoV-2 infection is common and independent of severity of initial infection

Neural signatures of Trail Making Test performance: Evidence from lesion-mapping and neuroimaging studies

Clinical utility of the Epworth sleepiness scale

Manual for the Wechsler Memory Scale -Revised

Hyperactivity and hyperconnectivity of the default network in schizophrenia and in first-degree relatives of persons with schizophrenia

Clinical sequelae of COVID-19 survivors in Wuhan, China: a single-centre longitudinal study

The effect of topiramate on cognitive fMRI

The landscape of cognitive function in recovered COVID-19 patients

The landscape of cognitive function in recovered COVID-19 patients