White matter microstructural abnormalities in neurological poste-acute sequelae of coronavirus disease: Imaging signatures consistent with persistent neuroinflammation Authors: Chuan Huang, Xinlan Zhang, Jia Ying, Ashley Fontana, Xiaohua Yang, Steven West, Paul Vaska, Sean AP Clouston, Benjamin J Luft Published: 2026 ABSTRACT AbstractBackgroundEssential workers, many of whom were infected with coronavirus disease (COVID-19) at work before vaccination, developed Neurological Post-Acute Sequelae of COVID-19 (N-PASC) at high rates.MethodTo characterize white matter abnormalities associated with N-PASC and to examine correlates of symptomatology, we recruited 54 participants with N-PASC and compared them to 26 participants with no COVID-19 history or who recovered from acute COVID-19 without PASC into a neuroimaging study to profile the cerebral connectome starting from 1/1/2021-12/31/2024. Diffusion parameters from whole-brain white matter and cortical gray matter analyses were examined with multi-shell diffusion MRI data. Correlational tractography was used to examine the nature and extent of white matter changes.ResultsParticipants with N-PASC lasting, on average, 2.7 years had global and tract-specific white matter alterations after adjusting for demographics across measures of white matter health. Correlational tractography located pronounced bilateral changes to the fornix and minor forceps, possibly providing indirect evidence of axonal injury or demyelination in these regions. But diffusion measures alone cannot determine the specific underlying biological process. Diffusion measures were associated with the presence of fatigue, executive impairment, anosmia and mood disorders, but not brain fog.ConclusionsResults identify changes in N-PASC consistent with neuroinflammation that help explain cognitive dysfunction, providing insights into the long-term cerebral implications of COVID-19 and suggesting that symptoms may reflect evidence of a cerebral changes persisting years after symptom onset.Keywords: Post-acute sequelae, COVID-19, Novel coronavirus, Neuroimaging, White matter, Diffusion tensor imaging, Essential workers Introduction In 2019, severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) caused a global pandemic and substantial morbidity from coronavirus disease (COVID-19) (Novel Coronavirus Pneumonia Emergency Response Epidemiology, 2020). The lingering symptoms and functional changes that accompany some SARS-CoV-2 infections have garnered increasing interest because these symptoms can persist after acute infections (Rogers et al., 2020) and cause long-term impairment to activities of daily living (Vanichkachorn et al., 2021). Explanations for persistence suggest that COVID-19 may cause changes to cerebrovascular neuropathology (Reichard et al., 2020) and increase the incidence of stroke (Avula et al., 2020). Referred to as post-acute sequelae of COVID-19 infection (PASC), this syndrome is defined by the persistence of COVID-19-related symptoms for months after acute infection has cleared. Changes in central nervous system (CNS) involvement is a recognized complication of several respiratory infections and can persist beyond resolution of acute illness (Cavallieri et al., 2022). One study reported that 22.5% of individuals testing positive for COVID-19 later developed neurological manifestations of PASC, including fatigue or headache (Wanga et al., 2021) and worse depressive symptoms (Rajkumar, 2020), though the risk of neurological subtypes of PASC (N-PASC) varies by COVID-19 vaccination status and COVID-19 severity (Pilotto et al., 2021). N-PASC is characterized by heterogeneous neuropsychological manifestations such as brain fog, persistent fatigue, ongoing loss of taste or smell, anxiety, and physical impairment (Takao and Ohira, 2023). Studies have also reported evidence of executive dysfunction and slowed response speed (Sekendiz et al., 2024) that are correlated with symptoms of N-PASC (Cahan et al., 2024) and could indicate COVID-19-related axonal degradation. Seeking to understand symptoms and functional changes, researchers recently reported evidence suggesting that individuals with mild to moderate COVID-19 appear to show changes in cerebral diffusivity and measures of neuroinflammation (Petersen et al., 2023). Concurrently, several small studies using positron emission tomography have reported increased glial activation in cortical and subcortical regions among individuals with N-PASC (Braga et al., 2023; Clouston et al., 2024; Visser et al., 2025). Few studies have examined a variety of abnormalities to the cerebral connectome among individuals who developed N-PASC after mild to moderate COVID-19. To fill this gap, we compared essential workers with mild-to-moderate COVID-19 infection who developed N-PASC to healthy controls who either never contracted COVID or had recovered fully from a documented case of COVID-19. Using whole-brain diffusion analyses and correlational tractography, we examined the relationships between white matter integrity and N-PASC status and duration. Second, we assessed whether diffusion abnormalities were associated with changes in measures of cognitive and functional outcomes among individuals with N-PASC. Finally, we explored whether diffusion markers could discriminate N-PASC cases from controls. Methods During the COVID-19 pandemic, our research program conducted systematic surveillance of COVID-19 status through regular communication and annual in-clinic evaluations (Morozova et al., 2021). In 2020, some individuals were found to have experienced persistent symptoms for months after initial infections, particularly brain fog and difficulty concentrating for months after the initial infection. We therefore performed a cross-sectional neuroimaging study to determine whether these symptoms reported alongside N-PASC might be associated with indicators of white matter health. Participants were asked to participate in the neuroimaging program if they fit the eligibility criteria. Owing to occupational eligibility criteria, approximately 94% of program participants are male. Participants We included three participant groups. N-PASC cases (n = 54) were defined on a history of positive COVID-19 testing (documented antigen, molecular, PCR, or in three cases with symptom onset prior to March 2020, using antibody testing) and self-reported neurological symptoms including brain fog or difficulty concentrating lasting ≥3 months. Unexposed controls (n = 10) were those who had no history of COVID-19 symptoms and had a history of negative COVID-19 test results. COVID-19 recovered controls (n = 16) were those with a documented infection but no evidence of PASC (COVID-19 recovered controls, n = 16). In all cases, trained research staff collected the symptoms using a standard protocol and a trained clinician collated information together and made the final diagnosis. Inclusion criteria Participants with a valid case status were eligible if they had a body mass index (BMI) ≤ 40 kg/m2 to fit into the MRI scanner, had the capacity to provide informed consent, and were willing to undergo a blood draw. Participants had to clear a standard MRI screening questionnaire, and any metal medical implants needed to be determined MRI safe to facilitate scanning. Eligibility for the study included being fluent in English, to facilitate neuropsychological testing and there were no upper age restrictions for inclusion. Exclusion criteria Participants were excluded if had a history of psychosis, a history of stroke, a history of serious head trauma (e.g., loss of consciousness with confusion, slurred speech, or amnesia), epilepsy, brain cancer, renal failure or receiving dialysis, severe liver disease, autoimmune disease, heart failure or myocardial infarction in the past year, and/or an indication of unmanaged diabetes. 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 (Version: 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(Cauley et al., 2014; Setsompop et al., 2012) (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 multishell 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 = 0.9x0.9 × 0.9 mm3, accerleration factor = 2, Turbo factor = 192). Clinical measures For all participants, we collected a detailed history of clinical severity (none, mild, moderate, and severe), types of acute symptoms during the index COVID-19 case linked with the N-PASC diagnosis, the length of time since COVID-19 infection when an infection was present, and the duration of N-PASC. We acquired copies of other positive antibody, polymerase chain reaction, antigen, and molecular test results from other health care facilities when the initial diagnosis was made outside of this clinic. Cognitive, functional, and physical measures During the screening examination, trained neuropsychological and clinical research staff measured cognitive and physical functioning using a wide variety of tests. General mental status was measured using the Montreal Cognitive Assessment (MoCA (Nasreddine et al., 2005)). MoCA was collected at the post-COVID study visit, was not used for study inclusion or exclusion, and should not be interpreted here as a diagnostic threshold for cognitive impairment. Verbal fluency was measured over the course of 1 min using the controlled word association task. The Hopkins verbal learning test was used to measure verbal total and delayed recall, retention, and recognition (Brandt, 2001). We measured speed of completion on the Trails A and B tasks to measure psychomotor and choice-response speed (Arbuthnott and Frank, 2000). We used the Wide Range Achievement Test's reading task (WRAT) test that is a proxy for pre-morbid aptitude (Johnstone et al., 1996). We used tests from a validated (Mielke et al., 2015) computer-assisted battery (CogState (Maruff et al., 2009)) to complete the controlled paired associate learning task to measure episodic memory (accuracy), the one card learning task to measure visual learning and recognition (accuracy) and cognitive throughput (accuracy of one-card learning/average response time), the detection task to measure response speed (number of tasks/response time), the speed of card identification task to measure processing speed (number of tasks/response time) and attention (accuracy on identification), and the Groton maze task to measure visuospatial learning (accuracy on the training task) and visuospatial memory (accuracy on a delayed Groton maze recall task). We measured physical functioning using a 12-foot natural walking speed test, the 5-rise chair-rise task, and maximal handgrip strength. Due to the large number of measures, we then used factor analysis with varimax rotation to create six domain-specific measures of functioning and labelled these based on the main testing domain being identified (Supplemental Table 1). We diagnosed mood disorders using the structured clinical interview for DSM-5 (First et al., 2015). We identified pre-COVID mood disorders as those disorders emerging prior to COVID symptom onset, while post-COVID mood disorders were those that emerged afterwards. Current mood disorders were those that were considered active at the time of screening. We also asked participants to describe how many days a week they drank, on average, and how many drinks would they normally have when they drank alcohol and defined a heavy drinker as someone who was drinking at least 4 drinks per week. We recorded cigarette smoking behavior as the number of pack-years smoked. Finally, we used a validated task called the Brief Smell Identification Test (Double et al., 2003) to measure olfactory function, and identified anosmia as being unable to identify ≥5 scents. Symptoms of executive cognitive impairment were identified as having a <1.5 SD when compared to the control sample on a scale indicating executive dysfunctional that uses an index created from averaging standardized measures of trails, verbal fluency, cognitive throughput, processing speed, and response speed (Cronbach's α = 0.78). 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 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), axonal diffusivity (AD), and radial diffusivity (RD) using diffusion tensor imaging with diffusion images acquired with b-value = 0; 1000 s/mm2. To generate the spin distribution function (Yeh et al., 2010), Q-space diffeomorphic reconstruction (Yeh and Tseng, 2011) was performed in Montreal Neurological Institute space with a diffusion sampling length ratio of 1.25. Output resolution in diffeomorphic reconstruction was isotropic 2 mm. Whole-brain average white matter diffusion parameters were calculated using masks generated by combining left and right cerebral white matter in the DKT atlas. Cortical thickness and cerebral volume were extracted from the T1MPRAGE images using the standard pipeline. Image analyses were performed on a Mac mini with Apple Silicon M2 Pro running Ventura 13.5 (Apple Inc.). Statistical analyses For gross comparisons between group characteristics, we used the Fisher exact test for categorical variables, while Welch's t-test was utilized for continuous variables. Primary analyses tested associations between N-PASC status and global white matter diffusion metrics using generalized linear modeling and estimated standardized beta coefficients (Std. Beta). Models variably adjusted for demographics, behavioral measures including pre-COVID-19 MDD, smoking status, and whether a participant was a heavy drinker. Correlational tractography was conducted as a complementary localization analysis to identify white matter pathways associated with case status. Statistical significance for tractography was determined using permutation-based inference on tract length. Diffusion metrics extracted from tractography-identified clusters were subsequently summarized to describe the magnitude of diffusion differences within those pathways. These ROI-derived analyses were considered exploratory because the clusters were defined in the same dataset. FDR in tractography analyses was estimated using 4000 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. We used a non-parametric Spearman partial correlation to derive the correlation because of the small sample size and the potential for non-Gaussian distributions. Generated streamlines were filtered using topology-informed pruning implemented in DSI Studio. This pruning iteratively removes singular streamlines that lack topological support from neighboring fibers, thereby reducing false-positive tracts. We used 16 iterations as commonly applied in DSI Studio pipelines. False discovery rate (FDR) correction was pre-specified and applied within each analysis family, including tractography analyses, subgroup or symptom regression models, and functional-outcome models using a two-tailed FDR = 0.05 (Benjamini and Hochberg, 1995). Multivariable adjusted analyses used generalized linear models, and adjusted for demographics (age, gender, educational attainment), and pre-COVID behavioral factors (mood disorders, heavy drinking, smoking status). Sensitivity analyses further examined stratified results when stratifying by sex/gender. To facilitate comparison between studies and across measures, we expressed differences in terms of standardized regression coefficients. Group-wise analysis was performed using Stata MP/17 (StataCorp). Sensitivity analyses We examined sensitivity of results to stratification by vaccination status at time of imaging and by vaccination status at the time of initial infection. Next, we examined results stratified by vaccination status at the time of infection at which point 84% of participants reported being fully vaccinated. To consider the potential for change in biomarkers over time, we examined correlations between time since infection and neurological biomarkers shown here. Ethics All participants provided informed written consent to participate. The Stony Brook University internal review board reviewed all study procedures. Results Participant characteristics After applying inclusion/exclusion criteria we examined data for 80 participants (Supplemental Fig. 1). Comparisons between participants who completed imaging and to those who did not indicated that non-completers had fewer years of schooling (median [interquartile range] = 15 [14-16] versus 16 [15-17], p = 0.033) but did not differ in age, gender, mental status, COVID-19 severity, or the number of years since index COVID-19 infection (mean = 2.69 [SD = 0.90] years across the sample who developed COVID-19 prior to scanning versus 2.44 [SD = 0.90] years in those who did not complete scanning). Sensitivity analyses (Supplemental Table 2) examined whether there were differences in COVID-19-recovered controls as compared to controls with no history of COVID-19 infections, but identified no significant differences between groups, justifying the combined approach used here. As shown in Table 1, the average age, education, and premorbid cognitive ability metrics were matched between PASC participants and controls. However, PASC participants had more severe clinical disease and more variable symptoms during the acute phase of COVID-19 infection. Participants with N-PASC reported a median of two symptoms, with the most common being brain fog. Upset plots (Supplemental Fig. 2) indicated that the most common symptom patterns all included both reported brain fog and fatigue though many also reported additional symptom (n = 14 [20.0%; 7 women, 36 men]). Intriguingly, both working memory and fluency were significantly lower in participants with N-PASC. Table 1Sample characteristics for all neuroimaged participants, stratified into post-acute sequelae of COVID-19 and control groups.Table 1CharacteristicsWhole Sample (N = 80)N-PASC (n = 54)Controls (n = 26)PAge, years59.8 (8)60.7 (8.3)57.8 (7)<0.001Female Sex/Gender8 (10%)6 (11.1%)2 (7.7%)0.635Education, years16 (2.1)15.9 (1.8)16.4 (2.7)0.668Mood Disorder Pre-COVID-1912 (15%)11 (20.4%)1 (3.8%)0.054Heavy Drinker20 (25%)12 (22.2%)8 (30.8%)0.411Smoking Pack-Years3.8 (8.8)4.5 (9.9)2.3 (5.7)0.206Intracranial Volume, mm31573.9 (136.2)1574 (132.8)1573.7 (145.9)0.993Brain Matter, %ICV2.7 (67.2)2.7 (67.2)2.7 (69.7)0.473Premorbid Cognition100.8 (11.4)100.6 (10.5)101.2 (13.4)0.847COVID-19 severity<0.001 Asymptomatic/Never10 (12.5%)-10 (38.5%) Mild28 (35%)20 (37%)8 (30.8%) Moderate32 (40%)25 (46.3%)7 (26.9%) Severe10 (12.5%)9 (16.7%)1 (3.8%)Years since Initial Infection2.7 (0.9)2.7 (1)2.8 (0.5)0.426N-PASC Symptoms Fatigue48 (60%)39 (72.2%)9 (34.6%)0.001 Brain Fog44 (55%)38 (70.4%)0 (0%)<0.001 Olfactory Impairment11 (13.8%)10 (18.5%)1 (3.8%)0.076 Executive Impairment26 (32.5%)22 (40.7%)4 (15.4%)0.024 Mood Disorder14 (17.5%)14 (25.9%)0 (0%)0.005 Other Mixed14 (17.5%)14 (25.9%)0 (0%)0.054Functional Testing Verbal Memory−0.3 (0.86)−0.35 (0.9)−0.19 (0.79)0.410 Visual Memory0.02 (0.88)0.05 (0.91)−0.04 (0.83)0.680 Cognitive Throughput−0.03 (0.75)−0.04 (0.84)−0.02 (0.53)0.929 Physical Functioning0.17 (0.87)0.07 (0.9)0.37 (0.76)0.138 Fluency0.07 (0.74)−0.11 (0.72)0.45 (0.64)0.001 Working Memory−0.25 (0.7)−0.2 (0.74)−0.35 (0.62)0.001Note: COVID-19: coronavirus disease 2019; N-PASC: Neurological post-acute sequelae of COVID-19 infection. MoCA: Montreal Cognitive Assessment.Columns report Mean (SD) for continuous variables, and frequency (percentages, %) for categorical variables.Mental status scores reflect post-COVID scores for participants had COVID and was not used for exclusion or inclusion. Sample characteristics for all neuroimaged participants, stratified into post-acute sequelae of COVID-19 and control groups. Note: COVID-19: coronavirus disease 2019; N-PASC: Neurological post-acute sequelae of COVID-19 infection. MoCA: Montreal Cognitive Assessment. Columns report Mean (SD) for continuous variables, and frequency (percentages, %) for categorical variables. Mental status scores reflect post-COVID scores for participants had COVID and was not used for exclusion or inclusion. Gross cerebral analyses We confirmed the absence of differences between uninfected and COVID-19-recovered controls. Correlational tractography revealed statistically significant white matter tract abnormalities in participants with N-PASC relative to controls (Fig. 1A–D). After FDR correction, prominent clusters of altered diffusion parameters were localized bilaterally in the fornix and the minor forceps. The fornix showed a consistent pattern of decreased FA and increased MD, RD, and AD. Visualizations highlight tract-level differences: reductions in anisotropy (red) and increases in diffusivity metrics (blue) mapped across these tracts. Correlational tractography shows tracts with increased (red) and decreased (blue) tractography when comparing N-PASC to controls. Statistically significant tracts (T threshold = 3, FDR<0.05) after adjusting for age and sex/gender. Blue indicates lower values in N-PASC, red indicates higher values in N-PASC. (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.) Table 2 presents adjusted regression analyses of global white matter diffusion metrics comparing participants with N-PASC to controls. The cluster identified in Fig. 1 is termed the “core ROI”. Overall, these showed robust group differences with moderate effect sizes that remained after covariate adjustment. Correlational tractography localized permutation-significant white matter clusters associated with N-PASC status bilaterally in the fornix and forceps minor (Fig. 1). Table 2Standardized beta coefficients providing the degree of association between the presence of N-PASC and measures of white matter health, stratified by N-PASC symptomatology.Table 2Gross Cerebral ParametersN-PASCN-PASC FatigueN-PASC Brain FogN-PASC AnosmiaN-PASC Executive ImpairmentN-PASC Mood DisorderNumber of SymptomsFractional Anisotropy−0.43−0.72∗0.03−0.34−0.53−0.46-0.65∗Mean Diffusivity0.59∗0.50−0.110.480.450.250.66∗Axial Diffusivity0.57∗0.39−0.130.490.380.220.58∗Radial Diffusivity0.54∗0.54−0.120.460.470.270.64∗Core Cerebral ParametersFractional Anisotropy−0.81∗−0.55−0.18−0.31−0.360.13-0.69∗Mean Diffusivity0.67∗0.300.080.170.27-0.170.52Axial Diffusivity0.69∗0.330.090.190.27-0.190.54Radial Diffusivity0.70∗0.340.100.190.27-0.200.55∗Note: Gross white matter metrics represent the primary whole-brain analyses. Core tract metrics summarize diffusion properties within clusters identified by tractography in the same sample and should be interpreted as exploratory summaries rather than independent confirmatory tests. N-PASC: Neurological post-acute sequelae of COVID-19; COVID-19: coronavirus disease 2019. Results reported standardized coefficients derived from generalized linear modeling and adjust for demographics and behavioral factors. ∗Statistically significant after adjusting for the false discovery rate (FDR = 0.05). Nominally significant results (two-tailed p-value <0.05) are shown in bold typeface. Standardized beta coefficients providing the degree of association between the presence of N-PASC and measures of white matter health, stratified by N-PASC symptomatology. Note: Gross white matter metrics represent the primary whole-brain analyses. Core tract metrics summarize diffusion properties within clusters identified by tractography in the same sample and should be interpreted as exploratory summaries rather than independent confirmatory tests. N-PASC: Neurological post-acute sequelae of COVID-19; COVID-19: coronavirus disease 2019. Results reported standardized coefficients derived from generalized linear modeling and adjust for demographics and behavioral factors. ∗Statistically significant after adjusting for the false discovery rate (FDR = 0.05). Nominally significant results (two-tailed p-value <0.05) are shown in bold typeface. To summarize the clinical correlates of diffusion differences, we extracted diffusion measures from the tractography-defined cluster and report them as secondary exploratory summaries (Table 2). These analyses suggested that the number of symptoms was a potential proxy for imaging severity. Additionally, we found that gross WM FA was strongly associated with fatigue (β = −0.72), executive impairment (β = −0.53), and mood disorder (β = −0.46). Focusing on the core WM markers, we found even stronger associations between N-PASC and core FA (β = −0.81), but symptoms analysis revealed that this was due to a strong association with fatigue in N-PASC (β = −0.70). Next, we examined correlations between gross WM markers and functioning in participants with N-PASC (Fig. 2). These analyses showed statistically significant relationships between WM FA, AD, MD, and RD, and domains of visual memory (standardized β = 0.78-0.90), and working memory (standardized β = −0.75). Heat map showing positive (blue) and negative (gold) associations between neuroimaging markers (columns) and functional measures (rows) in participants with N-PASC (n = 54). Standardized regression coefficients adjusting for demographics and behavioral factors. Coefficients shown in black typeface show results that are statistically significant (p < 0.05). Results shown in bold italicized typeface are statistically significant after adjusting for the false discovery rate (FDR = 0.05). (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.) Sensitivity analyses examined the influence of results to the absence of vaccination at time of infection and found that findings were very similar in both subgroups. We examined analyses specifically in males (n = 72) and only N-PASC values with infection onset preceding vaccination dates (n = 74) and found that results remained very similar to those shown here (Supplemental Table 3). Additionally, analyses examining the correlation between time since infection and markers of white matter health identified no statistically significant results (Supplemental Fig. 3). Discussion The presence of N-PASC is an emerging concern following acute COVID-19 infection, with symptoms often persisting for months or years. Building on prior neuroimaging studies, this work applied diffusion MRI analysis in a cohort of individuals who experienced mild-to-moderate acute COVID-19, rather than severe or hospitalized cases. While earlier imaging research has often relied on global white matter metrics, our use of tract-specific analyses provides enhanced anatomical precision and replicates increases in core results across MD, AD, and RD, alongside decreased FA. These changes, evident at both the global and tract-specific levels, may provide indirect evidence consistent with the presence of axonal injury and demyelination consistent with other neuroinflammatory studies (Budde et al., 2007; Kono et al., 2001; Shenton et al., 2012). Notably, correlational tractography revealed bilateral and symmetric alterations in tracts such as the fornix and minor forceps—regions critically involved in memory and interhemispheric connectivity—suggesting N-PASC might be associated with specific, functionally important white matter circuits. Because the tractography-defined cluster summaries were derived from the same imaging dataset, they are presented as exploratory secondary analyses intended to localize the whole-brain findings rather than as independent confirmatory tests. Our focus on essential workers—including individuals employed in construction and law enforcement—offers a distinctive contribution to the literature. Occupationally defined cohorts are usually healthier than the general population. However, during the COVID-19 pandemic, many essential workers experienced sustained risk of COVID-19 exposures at work. These were monitored through routine surveillance and structured follow-up, enabling accurate classification of infection history and symptom status. These participants represent a population with substantial public health relevance who are often underrepresented in neuroimaging research. The richness of clinical and demographic data, coupled with rigorous eligibility screening and standardized imaging protocols, enhances the internal validity of the findings and establishes a strong foundation for future longitudinal investigation. Our findings align with prior studies showing COVID-19-related white matter alterations. One study reported reduced axonal density in 22 hospitalized COVID-19 patients up to one year after recovery (Huang et al., 2022), while another noted changes consistent with neuroinflammation in 20 hospitalized COVID-19 patients with neurological symptoms(Rau et al., 2022). A third study (N = 36) reported both alterations in axial and mean diffusivity alongside functional connectivity changes that were evident 11 months after infection (Díez-Cirarda et al., 2023). Finally, in the largest study (N = 223) researchers found that individuals with mild to moderate COVID-19 appear to show changes in cerebral diffusivity and measures of neuroinflammation (Petersen et al., 2023). Our study confirms and extends these observations to individuals experiencing chronic neurological symptoms, supporting the hypothesis that N-PASC may reflect ongoing or unresolved neuroinflammatory processes. The absence of notable differences between never-infected controls and those who recovered from COVID-19 exposure indicates that diffusion abnormalities observed here are more closely related to N-PASC status than with prior infection alone. The absence of notable differences between never-infected controls and those recovered from COVID-19 suggests that the observed imaging abnormalities may be more closely related to N-PASC than to prior infection alone. Although some PASC symptoms improve over time, the enduring presence of diffusion abnormalities raises concerns about the long-term impact on brain health. Whether these changes resolve with symptom remission or persist as latent neurobiological alterations remains unknown. Future longitudinal studies employing repeated neuroimaging assessments are essential to determine if observed diffusion abnormalities persist, resolve, or evolve over time, thereby clarifying their clinical significance and progression. The observed alterations in FA, RD, AD, and MD are indicative of diffuse white matter microstructural abnormalities. Decreased FA is commonly interpreted as a marker of compromised axonal integrity, while increased RD suggests inflammatory demyelination or loss of myelin sheath coherence. Elevated AD and MD may reflect broader axonal degeneration and increased extracellular space (Vilaplana et al., 2020), potentially due to interstitial edema or chronic neuroinflammatory processes though the extent to which these factors account for evidence of neurodegeneration emerging in some studies remains unknown (Carbone et al., 2026; Yang et al., 2026). Still, the convergence of these diffusion abnormalities across multiple white matter pathways, particularly in the fornix and minor forceps, suggests distributed white matter involvement in N-PASC. Correlational tractography revealed bilateral alterations in the fornix and forceps minor. The fornix is the principal efferent pathway of the hippocampus and plays a key role in limbic circuitry supporting memory and affective regulation. The forceps minor connects homologous frontal regions through the genu of the corpus callosum and contributes to interhemispheric communication involved in executive and attentional processes. Disruption of these pathways could plausibly relate to cognitive and executive symptoms reported in N-PASC, although diffusion metrics alone cannot determine the precise biological mechanism underlying these tract-specific findings. An alternative or complementary explanation for the white matter changes observed may reflect COVID-19-associated microvascular injury. SARS-CoV-2 infection has been linked to endothelial dysfunction, micro-thrombosis, and compromised cerebral microcirculation, potentially affecting the integrity and functionality of brain tissue (Fekete et al., 2025; Pelle et al., 2022). Microvascular damage could impair blood-brain barrier integrity and induce subtle ischemic injury within cerebral white matter. Such vascular-related pathology might manifest as increased diffusion parameters (MD, RD, AD) and reduced FA, resembling findings typically attributed to inflammatory or degenerative processes. Importantly, these microvascular disturbances might coexist with inflammatory processes, compounding the detrimental effects on neuronal and axonal integrity. Given the recognized impact of COVID-19 on the vascular system, future research should also explore vascular mechanisms as contributors to persistent cognitive symptoms in N-PASC. Noting that the reduction in post-COVID-19 cognitive functioning in essential workers with PASC (Sekendiz et al., 2024), we measured several markers associated with central functioning. While we used a wide array of markers, we found that computer-assisted measures of cognitive functioning were the most sensitive to white matter dysregulation. Results were moderate in size, and statistically significant, implying that evidence of changes to white matter health may help to explain cognitive dysregulation. Future work should examine the extent to which neuroimaging could be used alongside functional testing to determine either the presence of N-PASC or is instead indicative of a dysfunctional subtype of severe N-PASC. Biological context from ancillary plasma and PET measures Diffusion MRI is sensitive to white matter microstructural change and is not specific to neuroinflammation, so we examined whether diffusion measures in this cohort were associated with ancillary biological markers available from related nested sub-studies. Plasma measures of inflammation from the epidemiologic cohort were available in 41 participants and included β-amyloid-40 (Barber et al., 2025), glial fibrillary acidic protein (Natale et al., 2023), and neurofilament-light (Kritikos et al., 2023). Translocator protein (TSPO) distribution volume (Vt) was measured in a 18F-FEPPA positron emission tomography (PET) study among eight N-PASC participants (Clouston et al., 2025). In these exploratory analyses, diffusion abnormalities were associated with several plasma and PET markers in directions broadly consistent with greater white matter injury or inflammatory burden. For example, white matter FA was positively associated with TSPO Vt (WM_FA: rho = 0.79, P = 0.021; Supplemental Table 4). These measures were not used to define case status and cannot be interpreted as diagnostic validation of N-PASC, but provide some convergent context for interpreting diffusion findings. Limitations The primary limitation of this study is a modest sample size, which limits power and generalizability. The male-predominant cohort, while scientifically justified given COVID-19's increased severity in males (Jin et al., 2020), precludes the assessment of sex-specific differences. Second, the cross-sectional design restricts causal inference and prevents examination of symptom or biomarker trajectories over time. Third, although the results suggest a role for neuroinflammation, none of these measures is established as a proxy for neuroinflammation. Prior work has already established that N-PASC is associated with the presence of increased glial activation (Clouston et al., 2024) and cytokine dysregulation though, interestingly, only mildly increased levels of glial fibrillary acidic protein (Peluso et al., 2022), a marker of astrocyte activation. Validation efforts in this sample suggested that diffusion imaging was associated with other indicators of neuroinflammation in participants with N-PASC, thereby supporting our conclusions. Diffusion tensor–derived indices should be interpreted as indirect markers of microstructural change rather than specific indicators of neuroinflammation. Advanced multishell diffusion models, such as neurite orientation dispersion and density imaging, may provide more specific biophysical characterization of neurite density and extracellular water and could complement DTI-based analyses in future studies of N-PASC. Still, future work should determine which immune cells are activated in N-PASC. Fourth, we collected individuals with test-confirmed mild to moderate COVID-19 that occurred, for the most part, prior to vaccination. Once available, individuals increasingly used at-home testing for mild-to-moderate COVID-19, and thus testing positivity and date are not as easily confirmed and the burden of asymptomatic cases also increased. Fifth, we recruited essential workers with new-onset N-PASC symptoms in individuals with a known pre-COVID status, while individuals with pre-COVID depression sufficient to explain N-PASC were excluded from this diagnosis. Many of the symptoms in N-PASC, however, are sensitive to other psychological conditions. Results may not generalize to individuals in the general population who are attending a PASC clinic and have severe, but potentially undocumented, psychological comorbidities. Sixth, while essential workers reported being mostly infected while at work during the pandemic, they are not uniformly infected at work. Additionally, because this is an occupational cohort the participants have been previously exposed to both workplace stressors and other occupational exposures that could influence results. Our efforts to exclude results with pre-COVID depression and other psychiatric conditions helped to mitigate this source of bias, but there is a potential for residual confounding. Clinical impact We reported diffusion abnormalities in the fornix and corpus callosum in participants with verified COVID who reported persistent neurological symptoms ≥ 6 months after infection that were not observed among participants who had not had COVID-19 or those who fully recovered. Inflammatory biomarkers were intercorrelated and were associated with the presence of functional dysregulation in measures of executive function and psychomotor functioning. In combination with several prior studies, these results imply that white matter tractography affected in N-PASC may guide our understanding of associated symptoms in N-PASC and could also improve targeted interventions to preserve cognitive function or reduce symptom burden. Indeed, the overlap between regions of microstructural disruption in the current analyses appears to proxy regions of increased TSPO binding and is, in a small sample, correlated with TSPO distribution. But this observation requires confirmation in larger samples. This convergence of molecular and structural imaging evidence may strengthen the pathophysiological link between markers of white matter health on diffusion imaging and N-PASC. Multi-study collaboration is necessary to provide guidelines on a diagnostic protocol for diffusion imaging-supported N-PASC diagnosis. Funding The authors received funding from the NIH/NIA (R01 AG049953, R01 AG067590) and from the CDC/NIOSH (U01 OH012275; CDC-75D30122c15522). CRediT authorship contribution statement Chuan Huang: Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Validation, Visualization, Writing – original draft, Writing – review & editing. Xinlan Zhang: Data curation, Formal analysis, Validation, Visualization, Writing – review & editing. Jia Ying: Formal analysis, Visualization, Writing – review & editing. Ashley Fontana: Data curation, Project administration, Supervision, Validation, Writing – review & editing. Xiaohua Yang: Investigation, Methodology, Project administration, Resources, Writing – review & editing. Steven West: Formal analysis, Project administration, Resources, Validation, Writing – review & editing. Paul Vaska: Data curation, Formal analysis, Methodology, Project administration, Supervision, Writing – review & editing. Sean A.P. Clouston: Conceptualization, Data curation, Formal analysis, Funding acquisition, Investigation, Methodology, Project administration, Resources, Software, Supervision, Validation, Visualization, Writing – original draft, Writing – review & editing. Benjamin J. Luft: Conceptualization, Data curation, Funding acquisition, Investigation, Methodology, Project administration, Resources, Validation, Visualization, Writing – original draft, Writing – review & editing. Declaration of competing interest The authors have no conflicts of interest to disclose. Footnotes This article is part of a special issue entitled: Post-COVID19 condition published in Brain, Behavior, & Immunity - Health. Supplementary data to this article can be found online at https://doi.org/10.1016/j.bbih.2026.101250. Supplementary data The following is the Supplementary data to this article: Multimedia component 1 Data availability Data will be made available on request. References Associated Data Supplementary Materials Multimedia component 1 Data Availability Statement Data will be made available on request.