Proteomic signatures in cerebrospinal fluid and their clinical associations in patients with ME/CFS, 2026, Bragee et al

[Nightsong]

Abstract:
This study evaluated the cerebrospinal fluid (CSF) proteomes from 31 patients diagnosed with myalgic encephalomyelitis/chronic fatigue syndrome (ME/CFS). We quantified 902 proteins, each expressed in at least eleven samples, and systematically categorized clinical factors relevant to ME/CFS symptoms—including autonomic dysfunction, neuroinflammation and metabolic disturbances.

Differentially expressed protein and pathway analyses evaluated protein features associated with both postural orthostatic tachycardia syndrome (POTS) status and disease severity among the patients, while ratio-based analysis further explored associations with severity ratings. Data are available via ProteomeXchange with identifier PXD076216. Neutrophil degranulation and platelet activation were enriched in patients with POTS, and several pathways, such as the complement cascade, coagulation-related pathways and IGFBP‑mediated insulin-like growth factor transport, were enriched in severe cases.

Ratio-based analysis identified four biologically interpretable severity-associated protein ratios related to cellular stress, extracellular remodelling and immune–neuronal interaction. Together, these findings provide insight into the biological processes associated with clinical heterogeneity in ME/CFS and generate hypotheses for future validation in larger independent cohorts.

Link | PDF (Nature Scientific Reports, April 2026, open access)

 

[ME/CFS Science Blog]

Looks like there was no control group.

I also thought they already published similar data in this paper:
Cerebrospinal fluid immune phenotyping reveals distinct immunotypes of myalgic encephalomyelitis/chronic fatigue syndrome - PubMed

 

[forestglip]

A recent study by Oh et al. (2025) identified the ratio of the levels of two proteins in the CSF—the ratio of YWHAG to NPTX2—as a marker of synaptic resilience and cognitive impairment in Alzheimer’s disease. 18 To assess its relevance in ME/CFS, this ratio was evaluated across the severity-defined subgroups. As shown in Figure 4, the YWHAG/NPTX2 ratio increased with clinical severity, reflecting a similar trend to that seen in Alzheimer’s disease.

The paragraph above is referring to the top left plot. Seems like it could be interesting, but the p-value isn't very low, and from the plot it looks like there were only two people in the mild group and two people in the severe group for this test.

The other three plots were based on a more unbiased scan of ratios of many different proteins, looking for significant ratios after multiple test correction. Except I think it was not an unbiased analysis. From methods:

To explore coordinated protein expression changes associated with ME/CFS severity, a systematic analysis of pairwise protein abundance ratios derived from DEPs was conducted. DEPs were pre-filtered based on nominal statistical criteria (p < 0.05 and fold change > 1.5) from the severity-based differential expression analysis. All unique pairwise combinations of these DEPs were generated, and the abundance ratio for each protein pair was calculated per participant. CSF samples were grouped into three clinician-defined ME/CFS severity categories (mild, moderate, severe), which were encoded as an ordinal numeric variable (1 = mild, 2 = moderate, 3 = severe) to facilitate trend analysis

As I understand it, first they tested which individual proteins significantly differ between severity groups (without multiple test correction). Then they only used these proteins when testing if ratios between any two proteins increase or decrease with increasing severity, and then did multiple test correction.

If so, this makes the p-value correction in the ratio analysis unreliable, since they only tested the proteins they already knew had large differences.

 

[Ravn]

Looks like there was no control group.

They give this reason

Healthy control CSF samples were not collected due to ethical constraints restricting lumbar puncture to symptomatic individuals.

Having had an awful experience with a lumbar puncture myself that sounds fair enough. But they also say this

Ultimately, validation in larger, independent cohorts will be essential to confirm these observations and assess their clinical relevance. This would also imply the incorporation of appropriate control groups (POTS without ME/CFS, other orthostatic intolerance phenotypes, healthy controls, and/or disease controls)

So that sounds like they could have used disease control samples but for some reason didn’t. Would suitable CSF disease control samples - e.g. from suspected MS cases where lumbar punctures are common - be hard to come by?

 

[Utsikt]

Would suitable CSF disease control samples - e.g. from suspected MS cases where lumbar punctures are common - be hard to come by?

Don’t most countries have biobanks for MS? Norway started on almost 30 years ago.