Symptoms of coronavirus disease 2019 (COVID-19) can persist for months or years after infection, a condition called Post-Acute Sequelae of COVID-19 (PASC). Whole-brain white matter and cortical gray matter health were assessed using multi-shell diffusion tensor imaging. Correlational tractography was utilized to dissect the nature and extent of white matter changes. In this study of 42 male essential workers, the most common symptoms of Neurological PASC (n = 24) included fatigue (n = 19) and headache (n = 17). Participants with neurological PASC demonstrated alterations to whole-brain white matter health when compared to controls made up of uninfected, asymptomatic, or mildly infected controls (n = 18). Large differences were evident between PASC and controls in measures of fractional anisotropy (Cohen’s D=-0.54, P = 0.001) and cortical isotropic diffusion (Cohen’s D = 0.50, P = 0.002). Symptoms were associated with white matter fractional anisotropy (fatigue: rho = -0.62, P < 0.001; headache: rho = -0.66, P < 0.001), as well as nine other measures of white and gray matter health. Brain fog was associated with improved cerebral functioning including improved white matter isotropic diffusion and quantitative anisotropy. This study identified changes across measures of white and gray matter connectivity, neuroinflammation, and cerebral atrophy that were interrelated and associated with differences in symptoms of PASC. These results provide insights into the long-term cerebral implications of COVID-19.
Article
Neuroinflammatory imaging markers in white matter: insights into the cerebral consequences of post-acute sequelae of COVID-19 (PASC)
https://doi.org/10.21203/rs.3.rs-3760289/v1
This work is licensed under a CC BY 4.0 License
You are reading this latest preprint version
Post-Acute Sequelae of COVID-19
Neuroimaging
White Matter
Diffusion Tensor Imaging
Essential Workers
In 2019, a severe acute respiratory coronavirus (SARS-CoV-2) caused a pandemic. In some individuals, SARS-CoV-2 causes coronavirus disease (COVID-19), a potentially deadly disease that has caused millions of deaths worldwide predominantly through pathways linked to respiratory, cardiovascular, and neurological damage that the virus causes in vulnerable people 1, 2. Changes to central nervous system functioning are a common side effect of several viral infections 3, and are a central component of COVID-19 infection 4. Correspondingly, COVID-19 infections have been linked with increased depressive symptoms 5, cerebrovascular neuropathology 6, and increased risk of stroke 7. Research among individuals with mild to moderate COVID-19 shows changes in cerebral diffusivity and measures of white matter neuroinflammation 8, 9.
There is increasing interest in the lingering symptoms and functional changes that accompany some COVID-19 infections 10, especially because these long-term changes can impair activities of daily living 11. The Center for Disease Control and Prevention suggest that post-acute sequelae of COVID-19 infection (PASC) develops when acute COVID-19 symptoms last for at least four weeks after infection (https://www.cdc.gov), while the World Health Organization (https://www.who.int) further recognizes that PASC can include emergence of symptoms up to three months following the initial SARS-COV-2 infection that last at least two months without any other explanation 12. Pathogenesis of PASC has been linked to SARS-CoV-2 neurotropism even up to 8 months after infection, and with viral-induced coagulopathy, endothelial disruption, systemic inflammation, cytokine overactivation, and neuroglial dysfunction 13. The most common neurological complaints in a sample of 10,530 PASC patients at 12 weeks include fatigue (37%), brain fog (32%), memory problems (28%), attention deficit (22%), myalgia (28%), anosmia (12%), dysgeusia (10%), and headaches (15%) 13. Symptoms of fatigue, brain fog, executive dysfunction, and slowed response speed in PASC might reflect the presence of axonal 8, 14–18 or cortical 18–22, an injury that could linger even after COVID-19 has cleared. Follow-up neuroimaging studies of PASC have reported that gray and white matter alterations including changes in white matter mean and axial diffusivity were reported in patients with PASC for as long as 11 months 23. While these studies are strongly supportive of the view that PASC might signify the presence of white matter inflammation as measured using diffusion tensor imaging, the prognosis of the disease is unclear.
While ongoing research has used sensitive studies of white matter diffusivity using standard technologies, some symptoms of neuro-PASC could indicate the potential for changes in cortical diffusivity that have not yet been evaluated, while the use of quantitative anisotropy and isotropic diffusion measures have not been examined. The objective of the present study was to study white matter diffusion markers associated with neuroinflammation among individuals with mild to moderate COVID-19 who developed neurological PASC by comparing them to individuals who were never infected with COVID-19 or those who were infected but who never developed PASC.
Participants were recruited from a monitoring program for essential workers who work in construction and law enforcement agencies 24 with an existing neuroimaging program 25, 26. In this program, participants receive an annual personalized monitoring examination that assesses physical, cognitive, and mental health problems. In addition, during the COVID-19 pandemic, the program began to determine COVID-19 status using a barrage of text messaging, phone calls, and letters, and by taking both virtual and in-clinic infection and vaccination histories whenever possible 27. In 2020, it became obvious that some individuals were experiencing persistent symptoms for months after initial infections like brain fog and difficulty concentrating so we developed a clinical neuroimaging study to determine whether these neuropsychiatric symptoms being reported alongside PASC were associated with indicators of white matter health. Participants were asked to participate in the neuroimaging program if they fit the eligibility criteria.
For all participants, we collected history of clinical severity (asymptomatic, mild, moderate, and severe), types of acute symptoms during the index COVID-19 case linked with the neuro-PASC diagnosis, the length of time since COVID-19 infection when an infection was present, and the duration of neurological PASC. Participants self-reported their COVID-19 status during their initial phone screening, and we verified information to the best of our ability; we completed a blood draw for COVID-19 antibody testing with all participants. We acquired copies of other positive antibody, PCR, antigen, and molecular test results acquired outside of our clinic, most frequently from local urgent care facilities. We used the date of infections to calculate months between COVID-19 infection and neuroimaging among those who reported a COVID-19 diagnosis.
Neuroimaging analyses included participants who were recruited from three separate groups of participants: neurological PASC: a history of positive COVID-19 testing (antigen, molecular, PCR, or antibody) from 2020–2021 paired with neuropsychiatric symptoms of brain fog and difficulty concentrating lasting ≥ 6 months. Controls included both “uninfected controls” who had a history of negative COVID-19 test results and no COVID-19 symptoms before imaging and “acute COVID-19” controls who reported positive COVID-19 test results but who had an asymptomatic, mild, or moderate presentation of COVID-19 from 2020–2021 but had no evidence of PASC.
We linked COVID-19-related information to demographic and clinical data from our clinical database. All participants were male so demographics included age, education in years, and body mass expressed in kg/m2. We measured premorbid crystallized cognition using the Wide Range Achievement Test. As physical functional markers, we included measures of maximal handgrip strength, in 100s of pounds, comfortable walking speed over 12 feet averaged across two trials, and chair rise speed averaged over five trials expressed in rises per second. As measures of fluid cognition, episodic memory was measured using the total score on the Hopkin’s Verbal Language Test, processing speed was measured using the Trails B test and expressed in connections per second, language function was measured using the Boston naming test, working memory was measured using the Symbol Digit Modalities Test, and Verbal Fluency was measured using the Controlled Word Association Test.
Eligibility Criteria
Due to occupational inclusion, approximately 94% of program participants are male. Since COVID-19 was more severe in men, and because the measures used here are sensitive to participant sex, only males were eligible for this study. Participants were excluded if, during a pre-imaging screening visit, they met any additional exclusionary criteria as discussed below. We were able to verify infection and clinical details for 84.5% of potential COVID-19 cases. Participants whose case status was verified needed to have a body mass index ≤ 40 kg/m2 to fit comfortably into the MRI scanner. Participants also needed the capacity to provide informed consent, and a willingness to undergo a blood draw totaling about 100 ml of blood. Participants were fluent in English for neuropsychological testing, and there was no upper age restriction for inclusion, though participants need to be ≥ 18 years of age.
Participants undergoing magnetic resonance imaging (MRI) were excluded if they presented with a history of psychosis, a history of stroke, a history of serious head trauma as defined by loss of consciousness accompanied by confusion, slurred speech, or amnesia, or other neurological disorders such as epilepsy. We also excluded participants with a history of brain cancer, chronic autoimmune diseases, or heart failure, and participants currently in renal failure or receiving dialysis, those who had a myocardial infarction in the past year, and/or an indication of unmanaged diabetes, and those who had evidence of severe liver disease or hepatitis. Participants receiving cognitively active medications, or anti-inflammatory, or immunomodulatory drugs were also excluded. Participants with claustrophobia, embedded ferromagnetic metal implants, pacemakers, shrapnel, wires, and/or other MRI-unsafe surgical implants were excluded.
MRI Acquisition
All images were acquired on a 3T Siemens Biograph mMR scanner (V.VE11P) using the vendor-provided 20-channel head/neck coil. Diffusion MRI images were acquired using a state-of-the-art multi-band diffusion-weighted imaging sequence 28, 29 (obtained via C2P, Center for Magnetic Resonance Research, University of Minnesota) with TE/TR = 121.4/6300 ms, voxel size = 2x2x2 mm3, multi-band factor = 3. A multi-shell diffusion scheme was used with multiple b-values = 1000, 2000, 3000, and 4000 s/mm² with diffusion sampling directions 64, 32, 32, and 32, respectively. T1-weighted magnetization-prepared rapid gradient echo (MPRAGE) images were also acquired (TE/TR/TI = 2.49/1900/900 ms, flip angle = 9°, isotropic voxel size = 0.9x0.9x0.9 mm3, grappa factor = 2, Turbo factor = 192).
MRI image processing
Brain parcellation was performed with FreeSurfer (7.3.0) using T1-MPRAGE images. Two subjects were removed from the analysis because, upon visual inspection, there was evidence of chronic cerebral infarctions. The diffusion MRI images were processed using DSI Studio (V.06142023). FSL’s eddy was used to correct for eddy current distortion. We calculated fractional anisotropy (FA), mean diffusivity (MD), axial diffusivity (AD), and radial diffusivity (RD) using diffusion tensor imaging with diffusion images acquired with b-value = 0; 1,000 s/mm². To generate the spin distribution function 30, Q-space diffeomorphic reconstruction 31 was performed in the MNI (Montreal Neurological Institute) space with a diffusion sampling length ratio of 1.25. Output resolution in diffeomorphic reconstruction was an isotropic 2 mm. The model-free technique separates isotropic from anisotropic diffusion, thereby reducing partial volume effects and crossing fibers. This approach has been shown to have improved reliability against edema32 and crossing fibers33. It allows the quantification of isotropic diffusion (ISO, isotropic value of the spin distribution function). Quantitative anisotropy (QA) was extracted as a local connectome fingerprint 34, and used in connectometry analyses. Whole-brain average white matter diffusion parameters were calculated using masks generated by combining left and right cerebral white matter in the FreeSurfer DKT atlas. Cortical thickness and cerebral volume were extracted from the T1MPRAGE images using the standard pipeline in Freesurfer. All image analyses were performed on a Mac mini with Apple Silicon M2 Pro (32GB memory) running Ventura 13.5.
Statistical Analyses
For gross comparisons between group characteristics, the Fisher exact test was utilized for categorical variables, while Welch’s t-test was utilized for continuous variables. To report effect sizes after adjustment for covariables we used generalized linear modeling and estimated standardized beta coefficients (b). All modeling assumptions were tested, and models were adjusted for potential unmatched confounders. When examining symptoms-based correlations within COVID-19-infected individuals, models adjusted for age, and intracranial volume. When examining cognitive outcomes, we additionally adjusted for schooling and estimated premorbid cognitive ability. Heat maps were generated, and estimated effect sizes were overlaid to aid in interpretation.
Correlational tractography was used to examine white matter tractography in COVID-19-infected versus uninfected participants and to compare those with acute COVID-19 to those with PASC. We used a nonparametric Spearman partial correlation to derive the correlation because of the small sample size and the potential for non-Gaussian distributions. As an effort to show differences in effect size, we reported images with a length threshold of 15 voxels (where T = 1), that were tracked using a deterministic fiber tracking algorithm with whole brain seeding. Generated tracts were then filtered by topology-informed pruning with 16 iterations. We use the area under the receiver-operating curve (AUC) to indicate the extent to which neuroimaging measures differed between diagnostic categories.
Since cerebral atrophy can indicate increased severity and poorer prognoses in neurological conditions, we examined correlations between gray and white matter parameters to determine if any white matter changes were related to neurodegenerative differences. Volumetric estimates were adjusted for total intracranial volume in all models.
Statistical significance was determined using a two-tailed p-value (a = 0.05) and adjusted for the false discovery rate (FDR) 35. FDR in tractography analyses was estimated using a total of 4,000 randomized permutations applied to the case label to obtain the null distribution of the tract length; tracts significant after adjusting for multiple comparisons (FDR = 0.05) were reported. Group-wise analyses were performed using Stata MP/17 (StataCorp), and Tractography was analyzed using DSI-Studio (dsi-studio.labsolver.org).
Sensitivity Analyses
Some factors might influence vulnerability to other cerebrovascular or neurodegenerative diseases, so we examined the potential for diffusion and volumetric parameters to relate to vaccination status or APOE4 status.
Ethics
The [Institution] internal review board reviewed all study procedures. Participants provided informed written consent to participate in this study.
Participant Characteristics
Only 70 of 788 PASC patients met the study's eligibility requirements, and 42 participants (24 participants with PASC and 18 controls), were included in this study (Figure 1). As shown in Table 1, the average age, education, and cognitive ability metrics were matched between PASC participants and controls, but PASC participants had more severe clinical disease and more variable symptoms during their acute infections. Differences between infected and uninfected controls are shown in Supplemental Table 1. The average COVID-19 antibody levels were similar between PASC and COVID-19 acutely infected controls. Not shown in Table 1, individuals with PASC developed a range of common symptoms, with more than half of all PASC participants reporting persistent fatigue (n=19), either cough or shortness of breath (n=18), headache (n=17), joint or muscle pain (n=16), chest pain (n=14), fever (n=13), or either brain fog or loss of taste or smell (n=12). The least common persistent symptoms in this study were nausea (n=1) and weight loss (n=4).
Whole-brain analysis
In Table 2, we tabulated and identified whole-brain white matter averages of MD, RD, and ISO were found to be significantly increased in participants with PASC, and that average FA was significantly lower in participants with PASC. Whole brain gray and white matter FA and ISO, cortical volume and thickness, and white matter RD and MD values survived FDR correction.
Between-group comparisons of FA, MD, AD, ISO, QA, and RD diffusion parameters the whole white matter was stratified by PASC status (Supplemental Figure 1). These results suggested that diffusion parameters differed between groups, even at the whole-brain level, and that accuracy to detect PASC was highest for FA with an AUC of 0.80 and accuracy of 71.4%. Inter-measure correlations are shown in Supplemental Table 1.
Correlational Tractography
To examine the scope of white matter changes, we performed correlational tractography to identify white matter tracts with a diffusion parameter correlated with the group variable (i.e., PASC versus controls). Figure 2 shows that diffusion measures displayed global changes in white matter tractography globally (increases shown in red, decreases in blue) that survived FDR adjustment. Whereas most diffusion measures (Figure 2, Panels A-D) identified regional changes in the cingulate, cerebellum, and parietal lobes as well as in the subcortical regions, ISO was found to increase throughout the entire white matter (Figure 2, Panel E). Whole-brain cortical gray matter averages of MD, RD, and AD were also studied but none had tractography results that were found to be significantly different between the PASC and control groups.
COVID-19 symptoms
Next, we examined Spearman’s correlation coefficients to test the hypothesis that specific acute COVID-19 symptoms in cases would be associated with neuroimaging measures (Table 3). After adjusting for intracranial volume and age, we found that neurological symptoms including nausea, headache, and fatigue were associated with differences in white matter connectivity. The presence of brain fog was associated with evidence of reduced ISO whereas nausea, joint pain, and fatigue were associated with increased ISO. Lower cortical thickness was associated with headache, chest pain, wheezing, shortness of breath, and nausea.
Regional Fractional Anisotropy Signature
To further examine which regions of the brain were most affected we performed correlational tractography results using different thresholds (Supplemental Figure 2 and 3). These results show that the tracts consistently found to have the strongest effect included the left and right fornix, Forceps Minor, and Tapetum. We further estimated tract FA averages derived from the tractographic analyses shown in Figure 2, Panel A, and termed that measure “CoreFA”. Core FA had the highest effect size than any of the whole-brain measures shown in Table 2 (b = -0.59, P<0.001, AUC=0.81). Next, we examined the hypothesis that clinical indicators of COVID-19 severity were associated with changes to Core FA (Figure 3). Results showed that Core FA was increased in individuals with more clinical severity of acute COVID-19 (Panel A), the presence of PASC (Panel B), lengthier periods between infections and ongoing symptoms (Panel C), and lengthier duration of infections (Panel D). Variability around Core FA was highest when participants were further away from their diagnosis, while those whose neuro-PASC symptoms persisted for ≥3 years showed the strongest changes in Core FA. Finally, we examined the association between Core FA and functional symptoms in PASC (Figure 4, with overall results in Panel D). Results suggested that Core FA was associated with slower walking speed (Panel A), while results with possible trends are shown in Figure 4, Panels B and D.
Since prognosis is a question of some pressing importance, we examined the degree of association between white matter alterations and gray matter structural outcomes in COVID-19-infected individuals (Table 4). These results identified moderate to strong associations between reduced white matter FA and reduced cortical and subcortical volume and cortical thickness. Increases in white matter ISO, QA, and gray matter ISO were also associated with reduced cortical thickness. Additionally, white and gray matter MD, RD, and AD were associated with cortical volume, subcortical gray matter volume, whole-brain gray matter volume, and whole-brain volume among those who reported COVID-19.
Sensitivity Analyses
Whole-brain diffusion parameters and white/gray matter volumes were not found to have a noticeable association with vaccination status, or APOE4 status aside from associations with COVID-19 severity (data not shown). Comparisons of the two control subgroups used multiple regression analysis to adjust for the effects of age and school years in all study participants, and the group residuals were used for t-tests, which did not identify any difference in diffusion parameters (data not shown).
The presence of neurological PASC is a new problem emerging after acute COVID-19 and may persist for years after initial infections. Replicating and expanding on prior neuroimaging work among COVID-19 patients, the present study identified pronounced changes in white matter diffusion parameters in PASC patients compared to both COVID-19 naive and COVID-19 infected participants without PASC. Using correlational tractography, we elucidated the presence of widespread neuroinflammation, with findings indicative of symmetric axonal injury. Elevated values of white matter MD, RD, and ISO coupled with diminished FA values and insignificant AD and QA changes observed in PASC patients are strongly suggestive of axonal injuries with neuroinflammation 36–38. Indeed, the robustness of the FA parameter, particularly the whole-brain white matter average FA value, in differentiating PASC from controls (AUC > 0.8) underscores its potential clinical significance. For example, when considering the severity of the acute COVID-19 infection, individual measures were able to differentiate COVID-19 controls including those who were never infected and those who were only acutely infected from those with neurological PASC. The robustness of the AUC values and logistic regression prediction accuracy highlights their potential diagnostic utility, hinting at a promising avenue for more precise and early detection of this condition. Our results bridge the current knowledge gap by highlighting the pervasive inflammatory landscape for individuals with neurological PASC.
Our work replicates and expands on analyses showing that COVID-19 can induce changes in white matter health. Prior work has reported reduced axonal density in hospitalized COVID-19 patients up to one year after recovery 39, while others have noted changes consistent with neuroinflammation in hospitalized COVID-19 patients 16. In the largest study to date, researchers found that individuals with mild to moderate COVID-19 appear to show changes in cerebral diffusivity and measures of neuroinflammation 8. The present work expands upon this earlier research and suggests that the presence of white matter dysfunction likely indicates the potential for PASC in patients infected with COVID-19, and further notes that the presence of FA dysfunction, in particular, is associated with a wide array of neurological PASC symptoms including fever, headache, fatigue, and pain, while we also found that white matter QA and white and gray matter ISO were strongly associated with brain fog symptoms.
We also found that decreases in whole-brain white matter FA was associated with reduced cortical thickness. Similar results showed that increases in white matter ISO, QA, and gray matter ISO were also associated with cortical thickness. Similarly, we found that several white matter and gray matter measures were also associated with cortical, subcortical, and whole-brain gray matter volume, as well as with reduced whole-brain volume and increased cerebrospinal volume. Cortical thickness and cerebral brain volume are established biomarkers for neurodegenerative diseases including Alzheimer’s and related dementias 40. Future work should examine the location of structural damage to help determine the potential for chronic functional limitations.
Controls with a history of acute COVID-19 and individuals lacking COVID-19 exposure did not show marked differences in the diffusion parameters evaluated here. This observation strengthens the premise that the neurocognitive sequelae in PASC, often recognized by the presence of brain fog, are intricately tied to neuroinflammation. The absence of notable differences between never-infected controls and those recovered from COVID-19 further accentuates the pivotal role of neuroinflammation in PASC-related cognitive deficits.
Studies have noted that some symptoms of PASC, like anosmia, anxiety, and brain fog, eventually dissipate in a majority of PASC patients in the months or years after initial infections though others, like fatigue, might show a pattern of relapse potentially with additional 41. Indeed, the brain's capacity for neuroplasticity raises the question of whether observed changes are permanent, or instead reflect a potential for recovery or adaptation over time. The potential that the resolution of symptoms reflects a reversal of diffusion parameters upon symptom resolution could provide some indication that the brain is resilient, in the long term, against these effects. Longitudinal studies will be paramount to tracking the progression or potential resolution of these neural changes over time, setting the stage for more comprehensive therapeutic interventions and a deeper understanding of PASC's long-term implications.
Findings inherent here might help to isolate the mechanisms of neuro-PASC. Changes in FA are often thought to measure the degree of axonal injury and decreased FA often indicates the presence of neuroinflammation. Increased AD often indicates an increase in axonal density consistent with an increased axonal inflammation. Increased RD in white matter suggests that neuroinflammation might be contributing to axonal demyelination. Changes in cortical MD have been associated with cerebral astrocytosis in familial Alzheimer’s disease 42. Gray matter evaluation presented a distinct landscape. Significant differences in ISO and cortical gray volume in PASC echo the potentially lasting severe findings in prior PASC studies 22, and might reflect the significant correlations found between ISO and FA with cortical and whole-brain volumetry reported here. Cortical and subcortical atrophy are central indicators of neurodegenerative disease that often result in cognitive, behavioral, and physical functional changes and can reduce independence. The observed alterations in diffusion parameters, especially within the white matter, pave the way for a deeper understanding of the possible cellular or molecular mechanisms responsible for chronic PASC. However, collectively, these observations imply that inflammatory changes in neuro-PASC could lead, at least for some individuals with neuro-PASC, to discernible cerebral atrophy. The presence of demyelination and neurodegeneration likely belies the potential for neuro-PASC symptoms to persist and raise the potential for symptoms to worsen with time.
Limitations
This study has several important limitations. First, one of the most salient limitations of this study is the relatively small sample size, which reduces the statistical power to detect anything but the largest associations. While the focus on outcomes among males is important because COVID-19 is more severe in males 43, a lack of females in our cohort limits the generalizability of these findings. Second, the cross-sectional nature of this study limits our ability to conclude causality, or the temporal progression of changes observed in PASC patients. Longitudinal studies are needed to monitor the trajectory of these brain alterations over time both among individuals with mild to moderate acute COVID-19, and among individuals who develop PASC could help to elucidate the persistence of changes. Third, our study alludes to neuroinflammation as a potential cause for the observed brain changes. However, we lack any direct measures identifying specific systemic or CNS-specific inflammatory markers in the brain, making this labeling speculative. Follow-up work including work using molecular neuroimaging is now emerging and may help to determine the specific types of neuroinflammation identified here 44.
Implications
Currently, neurological PASC is a syndrome that includes several symptoms including fatigue or brain fog. In the past, conditions with these symptoms have been difficult to identify or to understand using both biomarkers, structural neuroimaging, and cognitive or physical functional changes. Our study similarly showed that cognitive and physical functional symptoms were mild or lacking for a wide range of test outcomes, despite evidence of a substantial neuroinflammatory signature. The results of this study imply that we might be able to positively diagnose neurological PASC using a neuroimaging-based signature in the white matter with a high degree of accuracy. Indeed, the presence and importance of FA and ISO imply the potential for widespread neuroinflammation as a central mediator of these symptoms. Given the changes evident here, therapeutic strategies targeting neuroinflammation might prove beneficial for PASC patients. Interventions that counteract neuroinflammatory processes could be a significant step forward in managing and alleviating the syndrome's debilitating symptoms.
AD: axial diffusivity; AUC: area under the receiver-operating curve; COVID-19: coronavirus disease, 2019; FA: fractional anisotropy; FDR: false discovery rate; ISO: isotropic diffusion; MD: mean diffusivity; MRI: Magnetic Resonance Imaging; PASC: Post-Acute Sequelae of COVID-19; QA: Quantitative anisotropy; RD: radial diffusivity; T1-MPRAGE: T1-weighted magnetization-prepared rapid gradient echo.
Funding
This study was supported, in part, by the National Institutes of Health (NIH/NIA R01 AG049953) and the Centers for Disease Control and Prevention (CDC/NIOSH U01 OH012275).
Competing Interests
None to disclose.
Supplementary Material
Supplementary Material available online.
- Arabi YM, Murthy S, Webb S. COVID-19: a novel coronavirus and a novel challenge for critical care. Intensive care medicine 2020: 1-4.
- Holshue ML, DeBolt C, Lindquist S, Lofy KH, Wiesman J, Bruce H et al. First case of 2019 novel coronavirus in the United States. New England Journal of Medicine 2020; 382(10): 929-936.
- Koyuncu OO, Hogue IB, Enquist LW. Virus infections in the nervous system. Cell host & microbe 2013; 13(4): 379-393.
- Wu Y, Xu X, Chen Z, Duan J, Hashimoto K, Yang L et al. Nervous system involvement after infection with COVID-19 and other coronaviruses. Brain, behavior, and immunity 2020; 87: 18-22.
- Rajkumar RP. COVID-19 and mental health: A review of the existing literature. Asian journal of psychiatry 2020; 52: 102066-102066.
- Reichard RR, Kashani KB, Boire NA, Constantopoulos E, Guo Y, Lucchinetti CF. Neuropathology of COVID-19: a spectrum of vascular and acute disseminated encephalomyelitis (ADEM)-like pathology. Acta Neuropathologica 2020: 1-6.
- Avula A, Nalleballe K, Narula N, Sapozhnikov S, Dandu V, Toom S et al. COVID-19 presenting as stroke. Brain, Behavior, and Immunity 2020.
- Petersen M, Nägele FL, Mayer C, Schell M, Petersen E, Kühn S et al. Brain imaging and neuropsychological assessment of individuals recovered from a mild to moderate SARS-CoV-2 infection. Proceedings of the National Academy of Sciences 2023; 120(22): e2217232120.
- Caroli A, Capelli S, Napolitano A, Cabrini G, Arrigoni A, Pezzetti G et al. Brain diffusion alterations in patients with COVID-19 pathology and neurological manifestations. NeuroImage: Clinical 2023; 37: 103338.
- Wu Y, Xu X, Chen Z, Duan J, Hashimoto K, Yang L et al. Nervous system involvement after infection with COVID-19 and other coronaviruses. Brain, Behavior, and Immunity 2020.
- Post–COVID-19 syndrome (long haul syndrome): description of a multidisciplinary clinic at Mayo Clinic and characteristics of the initial patient cohort. Proceedings of the Mayo clinic proceedings2021. Elsevier.
- Pilotto A, Cristillo V, Cotti Piccinelli S, Zoppi N, Bonzi G, Sattin D et al. Long-term neurological manifestations of COVID-19: prevalence and predictive factors. Neurological Sciences 2021; 42(12): 4903-4907.
- Leng A, Shah M, Ahmad SA, Premraj L, Wildi K, Li Bassi G et al. Pathogenesis underlying neurological manifestations of long COVID syndrome and potential therapeutics. Cells 2023; 12(5): 816.
- Huang S, Zhou X, Zhao W, Du Y, Yang D, Huang Y et al. Dynamic white matter changes in recovered COVID-19 patients: a two-year follow-up study. Theranostics 2023; 13(2): 724-735.
- Huang S, Zhou Z, Yang D, Zhao W, Zeng M, Xie X et al. Persistent white matter changes in recovered COVID-19 patients at the 1-year follow-up. Brain 2021; 145(5): 1830-1838.
- Rau A, Schroeter N, Blazhenets G, Dressing A, Walter LI, Kellner E et al. Widespread white matter oedema in subacute COVID-19 patients with neurological symptoms. Brain 2022; 145(9): 3203-3213.
- Huang Y, Ling Q, Manyande A, Wu D, Xiang B. Brain Imaging Changes in Patients Recovered From COVID-19: A Narrative Review. Front Neurosci 2022; 16: 855868.
- Douaud G, Lee S, Alfaro-Almagro F, Arthofer C, Wang C, McCarthy P et al. SARS-CoV-2 is associated with changes in brain structure in UK Biobank. Nature 2022; 604(7907): 697-707.
- Abbasi J. Even Mild COVID-19 May Change the Brain. JAMA 2022; 327(14): 1321-1322.
- Besteher B, Machnik M, Troll M, Toepffer A, Zerekidze A, Rocktäschel T et al. Larger gray matter volumes in neuropsychiatric long-COVID syndrome. Psychiatry Res 2022; 317: 114836.
- Bendella Z, Widmann CN, Layer JP, Layer YL, Haase R, Sauer M et al. Brain Volume Changes after COVID-19 Compared to Healthy Controls by Artificial Intelligence-Based MRI Volumetry. Diagnostics (Basel) 2023; 13(10).
- Rothstein TL. Cortical Grey matter volume depletion links to neurological sequelae in post COVID-19 “long haulers”. BMC neurology 2023; 23(1): 1-11.
- Díez-Cirarda M, Yus M, Gómez-Ruiz N, Polidura C, Gil-Martínez L, Delgado-Alonso C et al. Multimodal neuroimaging in post-COVID syndrome and correlation with cognition. Brain 2023; 146(5): 2142-2152.
- Luft BJ, Schechter C, Kotov R, Broihier J, Reissman D, Guerrera K et al. Exposure, probable PTSD and lower respiratory illness among World Trade Center rescue, recovery and clean-up workers. Psychol Med 2012; 42(5): 1069-1079.
- Clouston S, Deri Y, Horton M, Tang C, Diminich ED, Pellecchia A et al. Reduced cortical thickness in World Trade Center responders with cognitive impairment: Neuroimaging/differential diagnosis. Alzheimer's & Dementia 2020; 16: e039996.
- Deri Y, Clouston SAP, DeLorenzo C, Gardus JD, 3rd, Bartlett EA, Santiago-Michels S et al. Neuroinflammation in World Trade Center responders at midlife: A pilot study using [(18)F]-FEPPA PET imaging. Brain Behav Immun Health 2021; 16: 100287.
- Morozova O, Clouston SAP, Valentine J, Newman A, Carr M, Luft BJ. COVID-19 cumulative incidence, asymptomatic infections, and fatality in Long Island, NY, January-August 2020: A cohort of World Trade Center responders. Plos One 2021; 16(7): e0254713.
- Setsompop K, Cohen-Adad J, Gagoski BA, Raij T, Yendiki A, Keil B et al. Improving diffusion MRI using simultaneous multi-slice echo planar imaging. Neuroimage 2012; 63(1): 569-580.
- Cauley SF, Polimeni JR, Bhat H, Wald LL, Setsompop K. Interslice leakage artifact reduction technique for simultaneous multislice acquisitions. Magnetic resonance in medicine 2014; 72(1): 93-102.
- Yeh F-C, Wedeen VJ, Tseng W-YI. Generalized ${q} $-sampling imaging. IEEE transactions on medical imaging 2010; 29(9): 1626-1635.
- Yeh F-C, Tseng W-YI. NTU-90: a high angular resolution brain atlas constructed by q-space diffeomorphic reconstruction. Neuroimage 2011; 58(1): 91-99.
- Zhang H, Wang Y, Lu T, Qiu B, Tang Y, Ou S et al. Differences between generalized q-sampling imaging and diffusion tensor imaging in the preoperative visualization of the nerve fiber tracts within peritumoral edema in brain. Neurosurgery 2013; 73(6): 1044-1053.
- Yeh F-C, Verstynen TD, Wang Y, Fernández-Miranda JC, Tseng W-YI. Deterministic diffusion fiber tracking improved by quantitative anisotropy. Plos One 2013; 8(11): e80713.
- Yeh F-C, Vettel JM, Singh A, Poczos B, Grafton ST, Erickson KI et al. Quantifying differences and similarities in whole-brain white matter architecture using local connectome fingerprints. PLoS computational biology 2016; 12(11): e1005203.
- Benjamini Y, Hochberg Y. Controlling the false discovery rate: a practical and powerful approach to multiple testing. Journal of the Royal statistical society: sreies B (Methodological) 1995; 57(1): 289-300.
- Budde MD, Kim JH, Liang HF, Schmidt RE, Russell JH, Cross AH et al. Toward accurate diagnosis of white matter pathology using diffusion tensor imaging. Magnetic Resonance in Medicine: An Official Journal of the International Society for Magnetic Resonance in Medicine 2007; 57(4): 688-695.
- Kono K, Inoue Y, Nakayama K, Shakudo M, Morino M, Ohata K et al. The role of diffusion-weighted imaging in patients with brain tumors. American journal of neuroradiology 2001; 22(6): 1081-1088.
- Shenton ME, Hamoda H, Schneiderman J, Bouix S, Pasternak O, Rathi Y et al. A review of magnetic resonance imaging and diffusion tensor imaging findings in mild traumatic brain injury. Brain imaging and behavior 2012; 6: 137-192.
- Huang S, Zhou Z, Yang D, Zhao W, Zeng M, Xie X et al. Persistent white matter changes in recovered COVID-19 patients at the 1-year follow-up. Brain 2022; 145(5): 1830-1838.
- Wang L, Davis PB, Volkow ND, Berger NA, Kaelber DC, Xu R. Association of COVID-19 with New-Onset Alzheimer’s Disease. Journal of Alzheimer's Disease 2022; 89: 411-414.
- Kim Y, Bae S, Chang H-H, Kim S-W. Long COVID prevalence and impact on quality of life 2 years after acute COVID-19. Scientific Reports 2023; 13(1): 11207.
- Vilaplana E, Rodriguez-Vieitez E, Ferreira D, Montal V, Almkvist O, Wall A et al. Cortical microstructural correlates of astrocytosis in autosomal-dominant Alzheimer disease. Neurology 2020; 94(19): e2026-e2036.
- Jin J-M, Bai P, He W, Wu F, Liu X-F, Han D-M et al. Gender differences in patients with COVID-19: focus on severity and mortality. Frontiers in public health 2020: 152.
Table 1 Sample characteristics for all neuroimaged participants, stratified into post-acute sequelae of COVID-19 and control groups.
|
Sample Characteristics |
Whole Sample N=42 |
Post-Acute Sequelae of COVID-19 n=24 |
COVID-Infected and Uninfected Controls N=18 |
D, P |
|
Age |
59.8 (7.9) |
61 (8.5) |
58.9 (6.8) |
0.27, 0.392 |
|
Education, Years |
16.2 (2.1) |
16.1 (1.8) |
16.4 (2.7) |
-0.14, 0.683 |
|
Premorbid Cognition |
99.6 (11.3) |
99.3 (9.8) |
101.2 (13.1) |
-0.16, 0.625 |
|
Body Mass, kg/m2 |
29.4 (3.9) |
29.3 (4.4) |
30.1 (3) |
-0.21, 0.496 |
|
COVID-19 Indicators |
|
|
|
|
|
Months Since Infection* |
30.1 (9.4) |
28.8 (9.9) |
33.5 (7.5) |
-0.53, 0.213 |
|
COVID-19 Severity |
1.5 (1) |
2 (0.7) |
0.8 (0.9) |
1.5, 0 |
|
PASC Symptom Burden |
5 (4.1) |
7.3 (3.5) |
1.9 (2.9) |
1.69, 0 |
|
COVID-19 Antibody Levels |
3.1 (3.1) |
3.4 (3.3) |
2.5 (2.8) |
0.31, 0.33 |
|
Physical Functioning |
|
|
|
|
|
Grip Strength |
10.1 (2.2) |
98.2 (25.3) |
107.4 (15.7) |
-0.43, 0.167 |
|
Walking Speed |
1 (0.2) |
1.1 (0.2) |
1 (0.1) |
0.25, 0.429 |
|
Chair Rise Speed |
5.1 (1.4) |
49.6 (13.7) |
52.6 (15.3) |
-0.21, 0.523 |
|
Cognitive Functioning |
|
|
|
|
|
Episodic Memory |
21.7 (4.8) |
21.4 (0.9) |
21.7 (21.4) |
-0.06, 0.855 |
|
Processing Speed |
1.4 (0.5) |
1.3 (0.4) |
1.6 (0.5) |
-0.57, 0.093 |
|
Language |
29.5 (1) |
29.4 (1.2) |
29.7 (0.6) |
-0.28, 0.366 |
|
Working Memory |
43 (9) |
41.3 (9.3) |
44.3 (8.2) |
-0.34, 0.294 |
|
Verbal Fluency |
41.3 (11.2) |
39.8 (9.5) |
44.2 (12) |
-0.41, 0.225 |
Note: COVID-19: coronavirus disease 2019; Std.: standardized; Diff.: difference; P: p-value.
Table 2 Means and standard deviations for diffusition parameter stratified by post-acute sequelae of novel coronavirus 2019 as compared to all controls, rank-ordered by multivariable-adjusted standardized regression coefficients
|
Neuroimaging Indicators |
Post-Acute Sequelae of COVID-19 Mean (SD) |
COVID-Infected and Uninfected Controls Mean (SD) |
Std. Coef., P |
AUC, Sensitivity, Specificity |
|
White Matter Fractional Anisotropya |
0.31 (0.01) |
0.32 (0.01) |
-0.54, 0.001 |
0.80, 0.75, 0.67 |
|
Gray Matter Isotropic Diffusiona |
0.33 (0.04) |
0.29 (0.04) |
0.50, 0.002 |
0.72, 0.63, 0.72 |
|
Gray Matter Volumeb |
0.63 (0.04) |
0.66 (0.05) |
-0.47, 0.003 |
0.64, 0.71, 0.61 |
|
White Matter Isotropic Diffusiona |
0.38 (0.05) |
0.35 (0.04) |
0.46, 0.004 |
0.71, 0.54, 0.83 |
|
Cortical Volumeb |
0.46 (0.03) |
0.48 (0.04) |
-0.45, 0.004 |
0.63, 0.71, 0.67 |
|
White Matter Radial Diffusivityc |
0.73 (0.05) |
0.69 (0.04) |
0.45, 0.004 |
0.75, 0.67, 0.78 |
|
White Matter Mean Diffusivityc |
0.86 (0.05) |
0.83 (0.04) |
0.41, 0.008 |
0.73, 0.63, 0.78 |
|
Cortical Thicknessd |
2.41 (0.07) |
2.45 (0.07) |
-0.34, 0.021 |
0.67, 0.67, 0.67 |
|
Cerebrospinal Fluid Volumeb |
0.13 (0.03) |
0.11 (0.03) |
0.31, 0.031 |
0.66, 0.63, 0.78 |
|
White Matter Axial Diffusivityc |
1.12 (0.05) |
1.10 (0.04) |
0.26, 0.054 |
0.66, 0.54, 0.72 |
|
Whole Brain Volumeb |
0.73 (0.03) |
0.74 (0.03) |
-0.22, 0.088 |
0.62, 0.67, 0.50 |
|
White Matter Quantitative Anisotropya |
0.15 (0.02) |
0.15 (0.02) |
0.19, 0.120 |
0.61, 0.46, 0.83 |
|
Gray Matter Radial Diffusivityc |
1.11 (0.09) |
1.07 (0.11) |
0.14, 0.202 |
0.63, 0.63, 0.61 |
|
Gray Matter Mean Diffusivityc |
1.17 (0.09) |
1.14 (0.11) |
0.13, 0.217 |
0.62, 0.63, 0.61 |
|
Subcortical Volumeb |
0.06 (0.01) |
0.06 (0.01) |
-0.11, 0.244 |
0.57, 0.63, 0.61 |
|
Gray Matter Axial Diffusivityc |
1.30 (0.09) |
1.26 (0.11) |
0.11, 0.253 |
0.61, 0.63, 0.61 |
aunitless, bliter, c*10-3, dmillimeters
Note: Std.: standardized; Coef.: Coefficient; AUC: Area under the receiver operating curve.. Standardized regression coefficient comparing PASC to control groups as estimated using generalized linear modeling to account for intracranial volume and age. Nominal p-values are reported; bold typeface shows which tests are nominally significant; bold italicized typeface shows which tests remained statistically significant upon adjusting for the false discovery rate.
Table 3 Heatmap correlation table showing standardized regression coefficients linking COVID-19 symptoms with neuroimaging parameters
Note: Standardized regression coefficients reported comparing PASC to control groups as estimated using generalized linear modeling to account for intracranial volume and age. Nominal p-values are reported in black typeface when the p-value exceeds 0.10; bold typeface shows which tests are nominally significant (two-tailed a=0.05); bold italicized typeface shows which tests remained statistically significant upon adjusting for the false discovery rate.
Table 4. Heatmap showing standardized regression coefficients from generalized linear models examining correlates of white and gray matter health linked with volumetric measures in COVID-19-infected participants with and without Post-Acute Sequelae of COVID-19
Note: Standardized regression coefficients estimated using generalized linear modeling to account for intracranial volume and age. Nominal p-values are reported in black typeface when tests are nominally significant (two-tailed a=0.05); bold italicized typeface shows which tests remained statistically significant upon adjusting for the false discovery rate. The colored bar figure shows the standardized regression coefficient (β) and shows its direction such that red shows negative associations, while blue shows positive relationships.
The authors have declared there is NO conflict of interest to disclose