Autoimmunity-Related Risk Variants in PTPN22 and CTLA4 Are Associated With ME/CFS With Infectious Onset (2020) Scheibenbogen et al.

https://www.frontiersin.org/articles/10.3389/fimmu.2020.00578/full

Excerpt:

In conclusion, our study shows that the PTPN22 rs2476601 and CTLA4 rs3087243 autoimmunity risk variants are more frequent in patients with ITO ME/CFS. This finding provides further evidence that there is a genetic predisposition for ME/CFS. The associations with PTPN22 and CTLA4 SNPs point to a role of autoreactive T and B cells in the pathomechanism of ME/CFS. In contrast, the lack of association of CTLA4 and PTPN22 SNPs and the lower frequencies of IRF5 and TNF risk variants in ME/CFS patients without ITO suggest that the pathomechanism is distinct. These associations need to be confirmed in other ME/CFS patient cohorts. Our findings prompt us to intensify research into autoimmune mechanisms and perform clinical studies with drugs targeting autoreactive B cells.

 

Autoimmunity-Related Risk Variants in PTPN22 and CTLA4 Are Associated With ME/CFS With Infectious Onset


Sophie Steiner1, Sonya C. Becker1†, Jelka Hartwig1, Franziska Sotzny1, Sebastian Lorenz1, Sandra Bauer1, Madlen Löbel2, Anna B. Stittrich3,4, Patricia Grabowski1 and Carmen Scheibenbogen1,3*

This was a candidate gene association study, with candidates identified by the following:

https://www.frontiersin.org/files/A...11-00578-HTML/image_m/fimmu-11-00578-t002.jpg

Just be aware that candidate gene association studies are notoriously unreliable.

 

why are they unreliable?

 

Can these SNPs be replicated? The answer is "no". I went back to the UK Biobank CFS findings (far more than 305 ME/CFS patients) and asked what is the association (what are the p-values?) of these SNPs to CFS status? The answers are: p = 0.43, 0.41, 0.56, 0.92 and 0.17. None of the 5 are replicated. So the paper's results are not replicated with a far larger number of CFS cases.

 

Were the associations checked against all UK Biobank patients or only those who reported an infectious onset (if the biobank records this data)?

From the abstract (bolding mine):

 

An infectious onset is not reported for this UK Biobank cohort. Nevertheless, I would expect these SNPs to skew towards zero. But they do not. Genetics is most powerful an instrument when it is applied evenly across the human genome without prejudice as to what could be relevant to ME. Otherwise it easily can misinform and divert attention from scientific leads with stronger evidence.

 

That makes sense. The fact that the small sample size brought out these irrelevant associations goes to show how important carrying out a wide scale GWAS is. Thank you for your work!

ETA: Would it be worth it to send a reply to the editor to point out your finding?

 

The Editor would request proof that these SNPs are not associated. Trying to prove a negative is a fool's errand!
I do hope that the MEBiomed GWAS is funded. For its objectivity, statistical power and cohort data it is sorely needed.

 

Inappropriate statistical thresholds combined with publication bias.

 

Full paper is on sci-hub now, here is one link.

EDIT: the paper is open access now, just follow @Ron 's link in the first post.

 

https://twitter.com/user/status/1250884851923853317