For years, the working theory was straightforward: SARS-CoV-2 triggers lasting inflammation throughout the brain, and that inflammation explains the fog, the fatigue, the memory lapses that millions of long-COVID patients describe. A PET imaging study published in May 2026 in the Journal of Neurology now directly contradicts that narrative. Researchers scanned the brains of long-COVID patients and healthy controls using TSPO PET, a technique that detects activation of glial cells, the brain’s resident immune responders. Across every major brain region, there was no significant difference between the two groups.
No widespread neuroinflammation. No smoldering immune fire consuming the cortex. The result pulls the rug from under several proposed anti-inflammatory treatments designed to calm a brain-wide blaze that, according to these scans, does not appear to exist in most patients.
But the scans were not entirely quiet. Within the long-COVID group, patients reporting higher levels of anxiety and depression showed elevated glial activation in one specific territory: the limbic system, the tightly wired network of structures including the hippocampus and amygdala that governs emotion, memory formation, and stress response. That targeted signal tracked with severe fatigue and diminished quality of life. The pattern suggests that what looks like “brain inflammation” in long COVID may actually be a stress- and mood-driven disturbance concentrated in the circuits people rely on to form memories and regulate emotions.
Structural evidence points to the same circuits
The PET findings do not stand alone. An earlier MRI study published in 2024 in Nature Communications had already identified microstructural brain changes in post-COVID patients tied to cognitive impairment, loss of smell, and fatigue. Those changes clustered in mesiotemporal and orbitofrontal regions, areas that overlap heavily with the hippocampus, amygdala, and entorhinal cortex flagged in the May 2026 PET work. Two different imaging modalities, two independent research teams working years apart, and both landed on the same anatomical neighborhood: the brain’s memory and emotion architecture, not a diffuse inflammatory storm.
Cerebrospinal fluid analysis adds another layer. A study from researchers at Yale School of Medicine examined spinal fluid from long-COVID patients experiencing brain fog and reported no classical markers of neuroinflammation or blood-brain barrier breakdown. The investigators instead identified immune signatures and non-inflammatory mechanisms that could disrupt cognition without producing the kind of global glial activation many clinicians had expected. The specific findings and methodology of that cerebrospinal fluid work have been summarized in institutional communications from Yale, though a direct link to the primary publication is not available here. Taken together, the fluid data and the imaging data converge on the same uncomfortable conclusion: the dominant theory was too simple.
Why earlier studies told a different story
The inflammation question is not fully closed. An earlier PET study, published in Brain, Behavior, and Immunity, used a different tracer called [11C]PBR28 and did report elevated TSPO binding in long-COVID patients compared to controls. That study also found correlations between the heightened signal and markers of vascular disease, raising the possibility that a subset of patients carries neuroinflammation linked to cardiovascular risk rather than direct viral damage.
So two PET studies measured the same biomarker and reached opposite conclusions. Several factors could account for the discrepancy. The tracers have different binding properties and sensitivities. The patient populations differed in symptom profiles and time elapsed since acute infection. Analytic methods for quantifying TSPO expression varied. And full quantitative binding data from the [11C]PBR28 study are not extensively reported in its public summary, making a precise head-to-head comparison difficult.
Both cohorts were also relatively small and drawn from specialty clinics, which tend to attract patients with more severe or persistent symptoms. Without larger, multicenter imaging efforts using harmonized tracers and protocols, researchers cannot yet determine which result better represents the typical long-COVID trajectory.
Correlation is not causation, but it still matters
Readers should be clear-eyed about what the limbic findings do and do not prove. The association between anxiety scores and glial activation in memory-related brain regions is correlational. It shows that two things track together within the long-COVID group. It does not establish that anxiety causes glial activation, or that glial activation causes anxiety. Both could stem from a shared upstream process: autonomic nervous system dysfunction, chronic stress signaling, or persistent immune activation outside the brain that indirectly disrupts mood circuits.
Raw participant-level data linking anxiety scores to limbic PET signal have not been released publicly. Without those data, independent researchers cannot verify the strength of the correlation, test alternative statistical models, or determine whether patients with both high anxiety and elevated limbic signal might respond differently to treatment than those with anxiety alone.
Still, correlational findings carry real clinical weight. They highlight which systems deserve closer study and help clinicians identify symptom clusters that tend to travel together. In this case, the overlap between fatigue, mood disruption, and limbic involvement suggests that interventions targeting emotional regulation and sleep quality could influence cognitive complaints, even if they do not directly alter whatever subtle glial changes might be present.
What changes at the bedside
If widespread brain inflammation is absent in most long-COVID cases, then treatments designed to suppress neuroinflammation across the board may be targeting a problem that is not there. The limbic findings redirect clinical attention toward mood, stress regulation, and memory circuit function. Cognitive behavioral therapy, structured anxiety management, sleep restoration, and rehabilitation programs focused on attention and working memory could offer more direct benefit than broad anti-inflammatory regimens whose effectiveness for long-COVID brain symptoms remains unproven.
That does not mean inflammation is irrelevant for every patient. The conflicting PET results and the vascular correlations from the PBR28 study leave room for biological heterogeneity. Some patients, particularly those with significant cardiovascular risk factors or evidence of small-vessel disease, may harbor pockets of neuroinflammation that contribute to their symptoms. Until larger studies clarify who falls into which subgroup, clinicians will likely need a personalized approach: screening for mood disorders, sleep disruption, and autonomic dysfunction while also monitoring cardiovascular health, blood pressure, and metabolic status.
Overlapping disturbances, not a single culprit
For people living with long COVID, these findings cut both ways. The absence of widespread inflammatory damage is reassuring; the brain is not universally under siege. But the complexity underscores that post-COVID symptoms defy any single, tidy explanation. The emerging picture is less about a uniform injury and more about overlapping disturbances in emotion, memory, sleep, and circulation that vary from person to person.
As more imaging and biomarker data accumulate through the rest of 2026 and beyond, researchers should be better positioned to sort patients into biologically meaningful categories and match targeted therapies to each. For now, the strongest studies point to a consistent message: long-COVID brain fog is real, but it is unlikely to be driven by one monolithic wave of neuroinflammation. Addressing mental health, restoring daily structure, and managing cardiovascular and autonomic risks may do as much for recovery as any yet-unproven drug aimed at quieting a brain-wide fire that these scans suggest is not burning.
More from Morning Overview
*This article was researched with the help of AI, with human editors creating the final content.
