Persistent serum protein signatures define an inflammatory subcategory of long COVID, 2023, Talla et al.

That's a big difference in the number of proteins that were differentially expressed between the PASC group (55 adults:21 men; 34 women); recovered group (24 adults: 9 men; 15 women); uninfected group (22 adults: 12 men; 10 women). The recovered group and the uninfected group are both relatively small, and the uninfected group differs in the sex ratio. The recovered group had a confirmed infection but no lingering symptoms.

They next did an unbiased clustering analysis across all of participants. I really like that sort of analysis, as it removes some of the bias. If a difference pops out and it correlates with the group differentiation, then there is a chance that the difference is characteristic. I'm not quite so keen on them using the first sample from the PASC participants and the last samples from the recovered participants as that sort of negates the point of having infected controls, but I understand why they would do that. The analysis doesn't sound to have been simply doing a PCA with all the proteins though.

They looked for pathways with affected proteins: 85 significantly affected pathways. They then accounted for overlapping genes, reducing the pathways down to 54 proteomic modules. The results are shown in Fig 1a, which is worth a look. The image is too big to copy, but here is the top part of it.




Have a look at Cluster 1 on the left. There's a high percentage of males and uninfected participants, no symptoms and low levels of enrichment of the pathway modules listed at the bottom right (denoted by the purple squares). In contrast, look at Cluster 4. There's a higher proportion of females and almost all of them have persisting symptoms. There's a high incidence of enrichment of the pathway modules.

It's worth noting though that not all of the participants with persisting symptoms reported fatigue (as shown in the section with white, grey and black squares. In fact, quite a few did not.

 

This was interesting. Participants in the PASC group mostly stayed in the same cluster over time.

 

They didn't find any difference between the identified clusters in SARS-CoV-2 specific IgG levels, suggesting that an inadequate immune response wasn't the problem.

Clusters 4 and 5 (which they characterised as the two inflammatory clusters and the ones with most PASC participants) had more older people and more people with a high BMI, but they also included people with healthy BMIs and ages at the younger end of the range. Cluster 1, the healthiest cluster, didn't have anyone with a BMI over 30 and had relatively young participants.

 

Here's Figure 2a which summarises the pathways with enriched expression in clusters 4 and 5. Surely there must be some clues in there somewhere, even though the PASC cohort is very noisy with respect to ME/CFS-like illness. There are lots of interesting-sounding pathways e.g. viral acute myocarditis, biomarkers for urea cycle disorders

The Leishmania pathway is enriched in both clusters. I'm assuming that pathway is the hsa05140 pathway: Leishmania is an intracellular protozoan parasite that changes host cell immune responses

The Anthrax pathway is enriched in both clusters too. Information on how the host protein response might be affected by anthrax here:

There do look to be substantial differences between the groups and clusters with respect to the pathways - for example:

 

Here is part of 3b which shows the levels of selected proteins for each participant - this chart is for IFN-y (interferon gamma). The pink dots are for the PASC participants, black dots are the infected recovered, and the grey dots are for uninfected participants. IFN-y does look different, but there are a lot of people labelled as having PASC with normal levels.

It seems that a lot of the differentially expressed proteins are downstream of interferon.

It's suggested that interferon is driving the protein signature of cluster 5 too. The authors say that it is very hard to quantify systemic type 1 IFN, but suggest that it is producing the elevation of the downstream proteins they did find.

Just a bit on interferons from Wikipedia:
These are cytokines released by cells in the presence of viruses. They are called interferons because they interfere with viral replication.

So, they cause lots of downstream impacts as part of a response to a pathogen, and they could be potentially causing flu-like ME/CFS symptoms. There are lots of varieties of interferons, each with their own particular effects.

I'm pretty sure that interferons have been looked at hard in proteomic studies of ME/CFS. And there was the work on the side-effects of interferon therapy (including fatigue) for viral hepatitis. If the release of interferons is very localised in tissue, perhaps they are hard to find in blood.

The authors note the evidence from other studies for persistent SARS-CoV-2 virus and its proteins, noting that this might be causing the interferon response. That's the big question - if interferons are the drivers of symptoms, what is causing them to be released?