Preprint Insights into Pathophysiological Pathways in ME/CFS Through Genetic Correlation and Mendelian Randomization, 2026, Wielscher et al

[forestglip]

Insights into Pathophysiological Pathways in ME/CFS Through Genetic Correlation and Mendelian Randomization

Spoiler: Authors

Wielscher, Matthias; Vincenzi, Leonardo; Weninger, Wolfgang P; Schernhammer, Eva S

Abstract
Myalgic encephalomyelitis/chronic fatigue syndrome (ME/CFS) and post-acute infection syndromes (PAIS) are multisystem disorders involving immune, vascular, neuroinflammatory, and metabolic abnormalities, yet the causal relevance of these processes remains unclear.

Using genome-wide summary statistics from DecodeME (15,579 cases), we performed genetic correlation, pleiotropic heritability, and Mendelian randomization analyses.

Across 22 auxiliary traits spanning five mechanistic domains, cellular energetics, neurovascular regulation, and barrier–microbiome function showed the strongest genetic overlap with ME/CFS, with migraine and irritable bowel syndrome contributing most to shared pleiotropy. Immunothrombotic related and inflammatory traits showed smaller but measurable genetic correlations. Energetics-related traits, including type 2 diabetes and blood lactate, displayed consistent genetic correlation but relatively low shared pleiotropy, suggesting that metabolic dysfunction may act through broader physiological networks.

Mendelian randomization identified three biomarkers with evidence for causal effects on ME/CFS risk: higher mitochondrial DNA copy number was protective, whereas increased glycoprotein acetyls and mean platelet volume increased risk.

Together, these findings indicate that ME/CFS susceptibility reflects interacting pathways involving barrier–microbiome dysfunction, neurovascular instability, inflammation, platelet activation, and impaired cellular energetics

Web | DOI | PDF | Research Square | Preprint

 

[Hoopoe]

This reminds that the other day I heard of a patient that had episodes where, in order to recover, he was forced to lie in a darkened quiet room for half or a whole day. The illness was migraines. I don't know what these episoders where triggered by; it sounded like they didn't occur often enough to interfere much with daily life, but were crippling when they occurred.

 

[forestglip]

Migraine, a common comorbidity of POTS reflecting impaired neurovascular reactivity (10), showed a clear positive genetic correlation with ME/CFS (rg = 0.45, p < 10⁻⁶). PHBC analyses mirrored this result: migraine exhibited substantial single-trait shared pleiotropy with ME/CFS, and its removal from the multi-trait model produced a marked reduction in shared pleiotropy, indicating that neurovascular dysregulation represents a key pleiotropic axis underlying ME/CFS risk.

I recently was interested in migraine because I found a large correlation using Google Trends scores, showing that [edit: US] states that search more for "chronic fatigue syndrome" tend to also search more for "migraine" and "migraine aura".

I previously also did genetic correlations using DecodeME summary stats against all the conditions in the UK BioBank and FinnGen using the Bigagwas platform. I just went back to check what the correlations were for ME/CFS with traits that relate to migraine:

Source	Trait code	Trait name	Genetic correlation (rg)	P value
UK BioBank	G43.gwas.imputed_v3.both_sexes	Diagnoses - main ICD10: G43 Migraine	0.4207	0.0063
UK BioBank	20002_1265.gwas.imputed_v3.both_sexes	Non-cancer illness code, self-reported: migraine	0.2132	6.34E-05
FinnGen	G6_MIGRAINE	Migraine-VI Diseases of the nervous system (G6_)	0.4425	1.62E-23
FinnGen	MIGRAINE_TRIPTAN	Migraine, single triptan purchase ok & required. ICD-code if available is included-Neurological endpoints	0.3206	7.48E-17

The correlation values with these traits seem fairly consistent with this study's correlation with migraine (rg = 0.45, p < 10⁻⁶).

The supplemental file says that the study used for migraine was Diversity and scale: Genetic architecture of 2068 traits in the VA Million Veteran Program, which was done on a USA cohort. So the positive genetic correlation of ME/CFS with migraine seems consistent when comparing to migraine cases from three different countries: USA, UK, and Finland.

Considering that there is an increased risk for females for both migraine and ME/CFS, and that both conditions display light/sound sensitivity, this might be a good lead.

I haven't read a lot about it yet, but it seems that research on migraine is a bit further along in identifying mechanisms, such as the involvement of the trigeminal nerve and CGRP proteins.

 

[DMissa]

to the mere humans among us (such as myself) what specifically from the genetic information are the traits being correlated against?

 

[forestglip]

to the mere humans among us (such as myself) what specifically from the genetic information are the traits being correlated against?

Genetic correlations are basically showing how similar the genetic influences are for two traits. If completely different variants were associated with ME/CFS compared to migraine, then the genetic correlation would be 0. If the exact same variants were associated in the same direction with both, the genetic correlation would be 1.

The higher the value, the more similar the underlying genetics of two conditions. A high correlation might suggest that they share causal pathways (though that's not the only possible reason for seeing a correlation.)

 

[DMissa]

Genetic correlations are basically showing how similar the genetic influences are for two traits. If completely different variants were associated with ME/CFS compared to migraine, then the genetic correlation would be 0. If the exact same variants were associated in the same direction with both, the genetic correlation would be 1.

The higher the value, the more similar the underlying genetics of two conditions. A high correlation might suggest that they share causal pathways (though that's not the only possible reason for seeing a correlation.)

Right, so what I mean is, specifically, what are the genetic correlations or influences being tested? Is it some sort of singular aggregated value describing genetic variation in the individual? Or is it against specific gene variants? I couldn't tell from reading

(thanks for helping)

 

[Murph]

Migraine prevalence by age and sex in the United States: a life-span study - PubMed

The present study assessed age- and sex-specific patterns of migraine prevalence in a US population of 40,892 men, women, and children who participated in the 2003 National Health Interview Survey. Gaussian mixture models characterised the relationship between migraine, age, and sex. Migraine...

pubmed.ncbi.nlm.nih.gov

The present study assessed age- and sex-specific patterns of migraine prevalence in a US population of 40,892 men, women, and children who participated in the 2003 National Health Interview Survey. Gaussian mixture models characterised the relationship between migraine, age, and sex. Migraine prevalence was 8.6% (males), 17.5% (females), and 13.2% (overall) and showed a bimodal distribution in both sexes (peaking in the late teens and 20s and around 50 years of age).

Migraine prevalence by age and sex in the United States: a life-span study - PubMed

The present study assessed age- and sex-specific patterns of migraine prevalence in a US population of 40,892 men, women, and children who participated in the 2003 National Health Interview Survey. Gaussian mixture models characterised the relationship between migraine, age, and sex. Migraine...

pubmed.ncbi.nlm.nih.gov

 

[forestglip]

Right, so what I mean is, specifically, what are the genetic correlations or influences being tested? Is it some sort of singular aggregated value describing genetic variation in the individual? Or is it against specific genes? I couldn't tell from reading

No, it is comparing GWAS summary stats from two different conditions. For example, the DecodeME summary stats give us an idea of how much each variant is associated with ME/CFS on a population level. They obtained summary stats for migraine and several other conditions as well, and compared each condition's summary stats to those for ME/CFS.

Basically, how well do the variant associations in the two traits line up? Do the summary stats for the two conditions show the same variants associated with both conditions? That would increase the correlation.

 

[DMissa]

No, it is comparing GWAS summary stats from two different conditions. For example, the DecodeME summary stats give us an idea of how much each variant is associated with ME/CFS on a population level. They obtained summary stats for migraine and several other conditions as well, and compared each condition's summary stats to those for ME/CFS.

Basically, how well do the variant associations in the two traits line up? Do the summary stats for the conditions show consistent direction of effect for many variants? That would increase the correlation.

Okay, so there's a summary statistic from the GWAS. Got it, thanks! I have basically 0 exposure to this stuff.

 

[forestglip]

Okay, so there's a summary statistic from the GWAS.

Not just one statistic. One p-value for every one of the millions of genetic locations that were tested in the GWAS, based on how often there was a variant at that location in ME/CFS versus the controls. The summary stats are often publicly available, including for DecodeME. There are a few million lines in the file they provide.

For example, one line looks like this, and shows that the variant on chromosome 17 at position 52183006 is strongly associated with ME/CFS at a log10p of 8.67.

CHROM GENPOS ID ALLELE0 ALLELE1 A1FREQ A1FREQ_CASES A1FREQ_CONTROLS N N_CASES N_CONTROLS TEST BETA SE CHISQ LOG10P EXTRA
17 52183006 17:52183006:C:T C T 0.329679 0.34585 0.349439 275488 15579 259909 ADD 0.0805953 0.0134577 35.8658 8.67492 NA

You might pull up a summary stats file for migraine and look at that same location to find that it was also strongly associated with migraine, which might suggest that this same variant has the potential to increase risk for both conditions. (Just an example. I don't know what the migraine data actually shows at this location.) The correlation algorithm looks at all the millions of locations at the same time to see how well the variant associations match between the two conditions.

Edit: Sorry if you already understood, but maybe helpful for others anyway.

 

[DMissa]

Thanks so much

 

[forestglip]

Migraine prevalence by age and sex in the United States: a life-span study - PubMed

The present study assessed age- and sex-specific patterns of migraine prevalence in a US population of 40,892 men, women, and children who participated in the 2003 National Health Interview Survey. Gaussian mixture models characterised the relationship between migraine, age, and sex. Migraine...

pubmed.ncbi.nlm.nih.gov

The present study assessed age- and sex-specific patterns of migraine prevalence in a US population of 40,892 men, women, and children who participated in the 2003 National Health Interview Survey. Gaussian mixture models characterised the relationship between migraine, age, and sex. Migraine prevalence was 8.6% (males), 17.5% (females), and 13.2% (overall) and showed a bimodal distribution in both sexes (peaking in the late teens and 20s and around 50 years of age).

Interesting!

Maybe another similarity, from Wikipedia:

Migraine episodes tend to diminish during the 2nd and 3rd trimesters of pregnancy, when elevated estrogen levels are stable. They are likely to return following the abrupt drop in estrogen levels after childbirth.[25]

A paper discussing whether ME/CFS might improve during pregnancy:

Chronic fatigue syndrome: implications for women and their health care providers during the childbearing years (Journal of Midwifery & Women's Health, 2008)

The most comprehensive study to date investigating the reciprocal relationship between CFS and pregnancy was published by Schacterle and Komaroff in 2004.1 Through a retrospective self-reported questionnaire, the investigators compared outcomes for 86 women with a cumulative total of 252 pregnancies that occurred before or after the onset of CFS. In “many cases,” pregnancy predated the time of data collection “by several years.”1

The investigators found that 41% of women experienced no change in CFS symptoms during pregnancy, 30% noted improvement of symptoms, and 29% experienced a worsening of CFS symptoms during pregnancy.1 The researchers were unable to identify factors that may influence whether an individual woman with CFS will improve or worsen during pregnancy.

CFS experts Nancy Klimas, Lucinda Bateman, and Charles Lapp report slightly different findings in clinical practice. Their reports are based on relatively small numbers of women with CFS whom they have followed throughout pregnancy, accentuating the current scarcity of evidence-based information.

In the approximately 20 women Dr. Klimas followed throughout pregnancy, improvement in CFS symptoms during pregnancy was almost universal, in some cases to the point of total remission, despite typically more severe early pregnancy nausea and vomiting requiring antiemetics used during chemotherapy.

In Dr. Bateman’s clinical observations of the approximately 6 women she followed throughout pregnancy, women commonly report feeling less ill with CFS symptoms during pregnancy despite experiencing typical pregnancy discomforts.

Dr. Lapp reports that 25 out of 27 patients in his practice who became pregnant while they had CFS felt better during pregnancy.48 Dr. Lapp proposes that lessening in severity of CFS symptoms during pregnancy may be related to immune system and hormonal changes of pregnancy.48

There's a members only S4ME thread about this: Did pregnancy affect your ME?

Also an old discussion from Phoenix Rising with some more anecdotes: https://forums.phoenixrising.me/threads/research-on-me-cfs-and-pregnancy.57296/

 

[obeat]

Migraine attacks as a result of hypothalamic loss of control - PMC

Hypothalamo-limbic connectivity reflects the cyclic nature of migraine. Hypothalamo-limbic connectivity is largest just before the attack. Hypothalamo-limbic connectivity is collapsing during the attack. Limbic perfusion is increasing and has a ...

pmc.ncbi.nlm.nih.gov

There is alsso research ongoing on PACAP pituitary adenyl cyclase polypeptides.

 

[DHagen]

Migraine attacks as a result of hypothalamic loss of control - PMC

Hypothalamo-limbic connectivity reflects the cyclic nature of migraine. Hypothalamo-limbic connectivity is largest just before the attack. Hypothalamo-limbic connectivity is collapsing during the attack. Limbic perfusion is increasing and has a ...

pmc.ncbi.nlm.nih.gov

From the paper's conclusions:

We observed an increasing connectivity between the hypothalamus and limbic areas over the interictal migraine interval towards the next attack that dropped during the headache. Given the predominant role of the hypothalamus as a biological clock, our data indicate that headache attacks are the result of the hypothalamus losing control over the limbic system. Potential therapeutic interventions are therefore suggested to strengthen limbic circuits with psychological approaches, such as relaxation techniques, cognitive interventions, or biofeedback.

*sigh*

 

[obeat]

From the paper's conclusions:



*sigh*

Yes I agree they've drawn the wrong conclusion but the changes in connectivity and cerebral perfusion are interesting

 

[Clio]

I used to describe PEM as a migraine-like state without the headache. Unfortunately the neurologist just looked at me like I was crazy and sent me away instead of trying to see if common migraine meds would provide any relief for feeling this way.

Migraine is a very good comparison because PEM to me involves a lot of vestibular symptoms (just looking at a simple video on Instagram will trigger instant motion sickness) which don't happen outside of PEM to me. Then there's the nausea, low appetite, slow digestion but also just feeling pain everywhere (but the head and legs for some reason), worse sleep issues than usual. But I never had any aura and the ocassional headaches I get are very mild.

Since migraine has triggers, what if ME/CFS is just a subset of a migraine where the trigger is literally everything? Would explain PEM and different severities nicely since migraines can also vary in triggers and severity between sufferers.

 

[obeat]

OI is a symptom in 40%of migraineurs

I had chronic migraine for over 20 years. 2 years ago over a three month period my migraines came to an end and my severe/severe ME symptoms improved to moderate/severe. My hunch is that a third factor is involved in both because although Triptans were very useful in stopping migraine symptoms, they never relieved ME symptoms.

 

[Alinda]

Since migraine has triggers, what if ME/CFS is just a subset of a migraine where the trigger is literally everything? Would explain PEM and different severities nicely since migraines can also vary in triggers and severity between sufferers.

Personally my migraine triggers are the exact same as my PEM triggers. And taking a triptan often reduces my ME/CFS symptoms a lot.
However, migraines are not known to cause limb pain! All of my symptoms, besides the limb pain, can be explained by several other conditions I am lately finding out I have.
Even the way the pain can migrate from left to right is the same as how migraine headache can switch sides mid attack.

 

[DHagen]

Yes I agree they've drawn the wrong conclusion but the changes in connectivity and cerebral perfusion are interesting

Oh yes, I agree as well and am glad you linked the paper - I apologize, I did not mean to dismiss the work in its entirety,I had simply just finished reading the paper and was hit with a wave of frustration on hitting those final lines.

 

[Verity]

I used to describe PEM as a migraine-like state without the headache.

At some point when I developed more serious cognitive symptoms, it was like they merged with migraine aura. It was hard to distinguish them.