Preprint Persistent immune dysregulation and metabolic alterations following SARS-CoV-2 infection, 2025, Lage, Deeks, Henrich, Peluso+

Apparently this is getting a lot of attention on Xitter.

PolyBio has a thread on it:
https://threadreaderapp.com/thread/...2PCKb2oATDeCeKpSQg_aem_tTTjjqN8NZcZIpKHklrNOg

 

There is a brief mention of ME/CFS:

 

I haven’t had a chance to read the full paper but it’s interesting that the DHEA finding was replicated with my team’s LC paper, which also has come up in ME.

Also, I need to keep saying it because it continually annoys me in scRNA-seq analysis: I simply do not trust analyses that identify fine-grained immune cell subtypes on scRNA-seq alone. You simply will not get the necessary read depth to accurately differentiate half of those subsets, and I don’t see anything in the supplemental material that would make me more confident in their labeling. [edit: if you want to do that, do CITE-seq]

And I do not trust proportional differences in immune subtypes based off of that fine-grained scRNA-seq labeling whatsoever without concurrent flow cytometry data. I don’t know why papers keep doing it.

That being said, I think there are some things this paper did well, I’ll try to comment that when I have more time.

 

Okay, having had a chance to revisit the paper:
1) It looks like the actual clustering (Fig 7A) was in fact done on flow cytometry markers and not simply on RNA-seq. So I do trust it more. Though as always, it is hard to speculate what a difference in the proportions of some specific immune subsets actually means.

2) For their actual single-cell analysis I don't see any supplementary information showing markers for their clustering. However, the subsets that they define (CD4+ T cell, CD8+ T cell, and monocytes) are generally possible to differentiate on the basis of scRNA-seq alone. I just wished they would have shown evidence that the subsets are actually the subsets they claim--hopefully a reviewer will ask for this.

3) They used the NASEM LC definition, which is so broad it basically covers every possible symptom that emerges after COVID infection. I understand the justification for making an intentionally broad clinical definition, but I don't think it should be used as a research definition.

4) Potentially because of their broad LC definition, it seems like a lot of the findings in monocytes are just general post-COVID findings. Meaning that we don't know if they actually correlate to any clinically relevant outcome since it was uniform across "recovered" and "Long COVID" participants

5) Same goes with the tryptophan findings. It's interesting that it's different post-COVID and that the finding is consistent with ME/CFS findings, but this study has no indication of whether it's actually correlated with any specific post-COVID outcome.

 

The biggest worry I have is that anyone feeling dodgy for any reason in the months after Covid gets called Long Covid. So they may just be reporting the changes that occur when people feel dodgy - either because they were already a bit dodgy before they got Covid or they are dodgy for some other reason and happened to have Covid.

The 'results' in the abstract are too vague for me to want to look further.

 

That's my biggest complaint with the NASEM definition. It basically boils down Long COVID to "anything that could ever happen to you other than being perfectly healthy, accompanied by a positive COVID test at some point."

I think this paper does a decent job of showing differences correlated with prior COVID exposure. But without any stringent LC definition, or even specific outcome measures to correlate with their findings, that's all it shows. If the majority of your findings were consistent between "recovered" and "Long COVID" participants, why even mention Long COVID at all?