Thread by @ZdenekVrozina

New Mayo Clinic study.
Brain hypometabolism in long COVID still showing up 2 years post-infection. This finding keeps replicating. It matters clinically. But there’s a lot worth unpacking. 🧵

https://twitter.com/harryspoelstra/status/2064282131304485117

Reduced brain metabolic activity in LC isn’t a one-lab quirk. Guedj 2021, a French multicenter study across three centers (n=143), pediatric case series - it keeps showing up across countries and cohorts.

And unlike standard MRI, which usually comes back normal in LC patients, PET is actually catching something. That gap - normal MRI, abnormal PET - is exactly why this modality matters here.

What does hypometabolism actually mean? Lower glucose uptake = lower neuronal activity. In other studies it maps directly onto symptoms - cingulate cortex onto cognitive dysfunction, brainstem/cerebellum onto overall symptom burden. This isn’t abstract.

What this study specifically adds?
A well phenotyped cohort split explicitly by PEM.
And the signal concentrates almost entirely in the PEM subgroup.
The non-PEM patients look close to normal - Z-scores near zero, some even positive.

The whole effect is driven by the PEM phenotype. That’s actually the stronger claim. Not LC = hypometabolism but - PEM specifically correlates with it.
For anyone tracking the LC-as-spectrum debate, this is a meaningful data point!

Median time from infection to scan 62 weeks. Max 149 weeks - nearly 3 years out. Still showing up.

No longitudinal scans. Persists here means - even late stage patients still show the finding. Not that we watched it fail to resolve in the same person. Real limitation. But it doesn’t make the observation less serious.

If anything it sharpens the question - what fraction of patients normalize, what fraction don’t, and what predicts which?

The most robust finding - surviving even strict multiple comparison correction - is bilateral primary visual cortex hypometabolism.
Not the cingulate, not the hippocampus. The visual cortex. Surprising at first glance.

Primary visual cortex is one of the most metabolically hungry regions in the brain. And photosensitivity, visual fatigue, sensory overload - these are well documented in LC. There’s a recurring hypothesis in research around impaired sensory gating. The brain losing its ability to efficiently filter incoming stimuli, making sensory processing disproportionately costly.

Patients who crash after reading or screen time aren’t imagining it. Whether this is cause or effect is genuinely unclear - but the anatomy fits and it deserves dedicated follow-up.

One methodological flag. The study normalizes everything to the pons. Standard for neurodegeneration.
But Guedj 2021 and others describe pons/brainstem hypometabolism as part of the LC pattern itself. Normalizing to a region that may itself be compromised shifts every single Z-score derived from it.

It could partly explain why the Mayo pattern (sensorimotor, parietal, visual cortex) looks different from Guedj’s (olfactory, limbic, brainstem). Different methods, different cohorts, or a normalization artifact inflating some regions and masking others. That needs direct comparison in future work.

Sum:
Brain hypometabolism in LC is real, replicated, and clinically significant. This study adds PEM specific signal and persistence data. The bilateral visual cortex finding is the most robust result and deserves follow-up. The pons normalization question is open.

The scans span 4 months to almost 3 years post-infection. Nobody in this dataset looks like they’ve bounced back. We’ve known since the HIV/HAND era that viruses can leave lasting neurological footprints. That lesson took too long to learn the first time.

Ganesh at al., Persistent Cerebral 18-FDG PET Changes in Patients With Long COVID Presenting With Fatigue and Post Exertional Malaise. journals.sagepub.com/doi/10.1177/21…