A comparison of genome-wide association analyses of persistent symptoms after Lyme disease, fibromyalgia, and ME/CFS, 2025, Hirsch et al

Full title: A comparison of genome-wide association analyses of persistent symptoms after Lyme disease, fibromyalgia, and myalgic encephalomyelitis – chronic fatigue syndrome

Annemarie G. Hirsch, Anne E. Justice, Amy Poissant, Cara M. Nordberg, Navya S. Josyula, John Aucott, Alison W. Rebman & Brian S. Schwartz

Abstract

Background
Up to 20% of Lyme disease cases experience post-treatment Lyme disease syndrome (PTLDS). The biological basis for PTLDS is poorly understood and no evidence-based treatment has been identified. Genetic studies have the potential to elucidate PTLDS pathophysiology and identify treatment targets.

Methods
We used electronic health record data (EHR) and genetic data from a linked biorepository to conduct a genome-wide association study (GWAS) for PTLDS among patients from a Pennsylvania health system. We evaluated the validity of the GWAS results in two separate conditions that have hypothesized overlapping pathophysiology, fibromyalgia and myalgic encephalomyelitis – chronic fatigue syndrome (ME/CFS). GWAS analyses were performed using logistic regression in SUGEN, assuming an additive genetic model, and adjusting for age, sex, array, and the first 10 principal components calculated from whole genome genotyping to adjust for ancestry, and accounting for relatedness including all 1st degree relationships. The functional mapping and annotation analysis (FUMA) tool was used to explore top findings from our GWAS.

Results
Among the 161,875 eligible MyCode participants with genotyping, there were 3,585 who met the criteria for treated Lyme disease. A subset of 695 (19.4%) of these patients met the criteria for PTLDS and the remaining 2890 were classified as controls. We identified two PTLDS loci that reached the suggestive significance threshold (P < 5 × 10− 7), with lead variants rs77857587, near IRX1, and rs10833979, near GAS2. Our top index single nucleotide polymorphism (SNP), rs77857587, is in high linkage disequilibrium with a long-range protein quantitative locus SNP, rs111774530, for the MARC2 (Mitochondrial Amidoxime Reducing Component 2) protein. We identified 5,041 cases of fibromyalgia (150,599 controls) and 2,268 cases of ME/CFS (151,594 controls) among the MyCode participants. Neither of the two suggestively significant loci were associated with fibromyalgia or ME/CFS.

Conclusion
We identified two PTLDS loci that reached a suggestive significance threshold. Our top index SNP is associated with the MARC2 protein, a protein that has been linked to multiple immune checkpoints. Further study is needed in a larger population to evaluate whether there is genetic evidence of the role of immune response in the occurrence of PTLDS.

Open access: https://bmcinfectdis.biomedcentral.com/articles/10.1186/s12879-024-10238-x

 

It would be good to get the thoughts of someone a bit more familiar with this sort of study. The numbers look quite low for getting much useful from GWAS from what I’ve read around DecodeME.

But they’ve used CDC diagnostic codes for diseases, and MyCode looks like a proper healthcare system rather than the sketchy DNA services some others have used
https://www.nature.com/articles/gim2015187

 

This is where the data came from:

But for ME/CFS it seems that they used questionable ICD codes:

ICD-10-CM R53.82 refers to 'Chronic fatigue, unspecified'. ME/CFS is listed at G93.3.

 

The normal threshold for statistical significant in GWAS is 5 x 10^-8. Because the sample size here was very low (only 695 PTLDS cases), the authors only highlight SNPs that had a p-value below 5 x 10^7

 

So they actively excluded those with actual ME/CFS?

Maybe something to ask the authors about

 

Looks like they do have some interesting data for Persistent Lyme disease, but the samples are too small annd the ME cases are meaningless given those selection codes.

Also, it is super easy to get false positives (or very low P values near significance) due to artefacts in the Data and even subtle issues with study design. Put another way, the study needs to be done by experts (maybe these people are, I haven’t looked ) and with bigger/better samples.

All the same, it will be interesting to compare the decodeme results with those for persistent Lyme disease.