[Transcriptome] in PBMCs of [LC] at a median follow-up of 28 months [...] reveals upregulation of JAK/STAT [...], 2025, Fineschi et al

Comprehensive transcriptome assessment in PBMCs of post-COVID patients at a median follow-up of 28 months after a mild COVID infection reveals upregulation of JAK/STAT signaling and a prolonged immune response

Serena Fineschi, Joakim Klar, Juan Ramon Lopez Egido, Jens Schuster, Jonas Bergquist, Ren Kaden, Niklas Dahl

[Line breaks added]

Background
Post-acute sequelae of SARS-CoV-2 infection (PASC), also known as post-COVID-19 condition (here abbreviated as post-COVID) is an escalating global health issue. The aim of our study was to investigate the mechanisms and clinical manifestations of post-COVID following a mild SARS-CoV-2 infection.

Methods
We analyzed the gene expression profile in PBMCs from 60 middle-aged post-COVID patients and 50 age-matched controls at a median time of 28 months following a mild SARS-CoV-2 infection. The clinical assessments included intensity of post-COVID symptoms, physical and mental fatigue, depression and anxiety.

Sixty-seven participants performed a mild exertion ergometer test with assessment of lactate concentrations. Transcriptome analysis was performed on mRNA selected by poly-A enrichment and SARS-CoV-2 RNA fragments were analyzed using the ARTIC protocol.

Results
We identified 463 differentially expressed transcripts in PBMCs, of which 324 were upregulated and 129 downregulated in post-COVID patients.

Upregulated genes in post-COVID individuals were enriched for processes involving JAK-STAT signaling, negative regulation of ubiquitination, IL9 signaling, and negative regulation of viral process, suggesting chronic inflammation. Downregulated genes were enriched for processes involving mitochondrial ATP synthesis, and oxidative phosphorylation, suggesting mitochondrial dysfunction. No SARS-CoV-2 gene fragments were detected in PBMCs of patients with post-COVID and no IFN genes were found differentially expressed in post-COVID patients.

Post-COVID was associated with elevated lactate levels in blood, both at rest and after a short recovery phase following exertion, suggesting increased anaerobic activity in skeletal muscles. We did not find differences in the transcriptional profiles or clinical manifestations when comparing patients who contracted the infection from early SARS-CoV-2 variants with those who contracted the infection during the period when the Omicron variant was prevalent.

Conclusions
Our findings highlight molecular changes compatible with a persistent immune response in PBMCs of post-COVID subjects at a median follow-up of 28 months after a mild infection, supporting the hypothesis that post-COVID is a chronic inflammatory condition. The upregulation of JAK/STAT signaling suggests a potential therapeutic target in post-COVID.

Link | PDF (Frontiers in Immunology) [Open Access]

---- Requested to be posted by @Jesse.

 

Well, that would do nicely.

 

Spotted a potential big no-no right away:

The y axes seemed off, so I tried to check what the normalization method was (looking for CPM, TPM, etc.) and found...nothing. This isn't an issue for the actual differential expression and pathway analysis, since DESeq estimates distribution on a per-sample basis. However, if you're plotting expression levels for individual genes like this, it's absolutely vital, otherwise the read count will be extremely skewed by variable sequencing depth between samples.

So that'll give a skewed visual, but what ultimately matters is the p-values, and those also looked weird when I double checked them. It appears that they pulled these values from the DESeq results which were luckily included as a supplemental table. The p-vals for STAT1 and IL31RA appear to be the adjusted p-values from the spreadsheet, whereas the other 4 are the raw p-values (which would all still have been below the adj. p < 0.05 cut off so I think it's just the result of messily pasting the p-values manually for these plots instead of automatically pulling from the spreadsheet in the ggplot function).

I also really dislike overrepresentation analysis for gene ontology, if you're already using R why not do GSEA. But that's a personal preference I guess.

All in all this seems to point more towards messy plotting rather than a reason to throw out the whole analysis. I don't see any glaring issues with their reported methodology for the DEG analysis, that's pretty standard.

Anyways, it seems that none of these signatures correlated with any of their outcome measures for post-COVID deficits [edit: including "fatigue scale"], meaning that it's just a general signature of having had COVID. My team's analysis also did PBMC transcriptomics, but ranked analytes based on predictive capability for post-COVID physical function. Although many of these same genes were detected, they were not significantly associated with physical function. We also did not detect any significant association with lactate, though that was without any stress testing.

 

I suppose what's interesting in this analysis is the fact that all samples were taken at least a year after COVID infection, meaning that this signature persisted for that long. But I don't know off the top of my head whether that's comparable to other viral infections.

 

In what way?

 

Jak/Stat is a (the?) downstream signalling pathway of gamma IFN - I guess that would tie in with Jonathan's hypothesis.

 

The controls had also had Covid, though?

 

There are two trials of Jak/Stat inhibitors in long covid currently ongoing. Baricitinib and another one I can't quite recall.

Acknowledging the concerns raised above, if this were to be a genuine finding it would make those drugs much more compelling candidates.

I believe Jonathan and his co authors suggested Jak/Stat inhibition in their ME hypothesis paper as a possible treatment too.

 

It mediates all interferon signaling, there will be slight differences in utilization of JAKs, STATs, and other downstream genes between type I, II, and III interferons.

 

As jnmaciuch says, whichever interferon you are rooting for, this would be nice if it were real.

 

Oh my bad—I misread the part where they were stating that controls and a couple cases did not have positive PCR results.

that’s what I get for trying to read papers quickly on my lunch break

 

It still seems weird that none of the identified signatures correlated with their outcome measure-specific subgroups, though, especially if those specific scales were used to define who has post-COVID deficits. Which tells me that this differential signature is really a differential signature for quite a mixed bag of various symptoms.

[edit: and, if I’m reading it right, seems like someone could have qualified for their post covid condition just by having anxiety/depression as long as it was “new onset” since infection]

 

Funding statement.

 

Email from Open Medicine Foundation:

Comprehensive transcriptome assessment in PBMC of post-COVID patients at two years after a mild COVID infection reveals upregulation of JAK/STAT signaling and chronic inflammation

OMF’s Collaborative Center at Uppsala, directed by Dr. Jonas Bergquist, recently published a paper on their work—in collaboration with the Science for Life Laboratory at Uppsala University—investigating molecular changes that occur in people with post-COVID lasting more than two years after a mild infection. Overall, the study found that these changes support the idea that post-COVID is a chronic inflammatory disease, with an upregulation of JAK/STAT signaling providing a potential avenue for therapeutic targets.

This study specifically examined the gene expression profile in peripheral blood mononuclear cells (PBMCs) in 50 controls and 60 people who had a mild COVID infection between March 2020 and April 2022 with symptoms lasting for over two years. Participants underwent sampling and clinical assessments—including intensity of symptoms, physical and mental fatigue, depression and anxiety—over two years after their acute infection. Sixty-seven participants also did a mild exertion ergometer test to measure lactate concentrations in the capillary blood at baseline, after exertion, and after resting. Transcriptome analysis was also performed on mRNA and SARS-CoV-2 RNA fragments were analyzed with sequencing-based screening.

From their analysis, 463 transcripts were differentially expressed between people with post-COVID and controls. There was an upregulation of genes in the JAK-STAT signaling pathway—making it a potential therapeutic target—and other genes suggesting that there is a persistent immune response in people with post-COVID. Additionally, the team found a downregulation of genes that suggested there was mitochondrial dysfunction. Overall, there was no correlation with the severity of symptoms. Finally, the study team found increased lactate levels at baseline and during recovery after exertion when compared to the control participants.

No SARS-CoV-2 gene fragments were detected in the PBMCs, which suggests that if viral persistence plays a role in post-COVID, the reservoir of these fragments is located elsewhere. Alternatively, the signaling pathways that remain active as though they are responding to an acute infection are similar to what you would find in autoimmune diseases.

Given the release of this manuscript, the project is in the “Publication” stage of the research process.

 

I’m not sure what that last point means—is it symptoms as defined by other scales or a list of other endorsed symptoms?

 

Regarding Neutrophils

From "Differentially Expressed Genes" section.

From recent Fluge, Mella, Trondstat paper thread.

Am I interpreting correctly that the trends are opposite between this Long Covid group and ME/CFS group of the other paper.?

 

Alright, rabbit hole time so I can get things straight. The study population was defined here:

https://pubmed.ncbi.nlm.nih.gov/39319311/

though they recruited ~20-30 more per group since that first publications. It looks like they defined “post-COVID” as having any new and persisting symptoms for 3 months after infection. I think that contributed to my original confusion, since they’re all post-COVID samples but their “post-COVID” group specifically means with-new-symptoms.

They were subjected to 5 scales (including cognitive fatigue, physical fatigue, 2 separate scales for depression, and anxiety) and then on top of that asked to endorse 13 other symptoms. Then, each individual symptom and each additional scale was converted to a separate 10 point scale (apparently the two depression scales were combined?), for a total of 17 items and a maximum total of 170 points.

Post-COVID group had a total score of 61.6 +/- 27, control was 9.1 +/- 13.1. So obviously a lot of variability, like I suspected it does seem like someone could have qualified for post-COVID by solely having a new case of depression.

[Edit: for their subgroup analysis, it looks like they set a cut off for each of the 4 (5?) scales they used, and then used that binary label as the outcome variable—though the text implies this was done only within patients. So I’m not really sure how meaningful that even is, honestly, I would’ve just created module scores for the top pathways and done a linear regression with the outcome scores]

 

Thanks for posting @forestglip, you were fast haha!

I was thinking there was some relevance to your hypothesis!

Post-COVID from what I've seen is a newer term to denote Long Covid with more emphasis on the fact that it's not "long-lasting Covid" but instead a Post-acute infection syndrome -> Post-COVID.

I honestly wish they would use some stricter criteria to define Long Covid, preferably even ME/CFS criteria. Because those people are probably the most afflicted and it would make it easier to compare ME/CFS and LC research.

 

They conclude it's 'inflammatory,' is it or not?

If it is, surely, treating it early would be important?

Thank you,