Relationship between major depressive disorder and [ME/CFS]: a two-sample mendelian randomization study analysis, 2025, Zhu et al

Abstract:
Major depressive disorder (MDD) and myalgic encephalomyelitis/chronic fatigue syndrome (ME/CFS) frequently occur together; yet their causal relationship remains unclear. To investigate the potential genetic causal link between these conditions, we conducted a two-sample Mendelian randomization (MR) analysis.

Summary data from Genome-Wide Association Studies (GWAS) for MDD were sourced from the UK Biobank and the Psychiatric Genomics Consortium, while GWAS data for ME/CFS were retrieved from the UK Biobank. Inverse-variance weighting (IVW), the MR-Egger method, and weighted median, simple and weighted modes were used to perform the MR analysis. In addition, Cochrane’s Q-test was used to detect heterogeneity among the MR results.

Horizontal pleiotropy was detected using the MR-Egger intercept and the MR pleiotropy residual sum and outlier (MR-PRESSO) tests. Leave-one-out analysis was performed to investigate the sensitivity of the association between MDD and ME/CFS.

The results of the MR analysis revealed no causal relationship between MDD and ME/CFS. The pleiotropy test revealed that causality bias was improbable, and no evidence of heterogeneity was found among the genetic variants. Finally, the leave-one-out test confirmed the stability and robustness of our findings.

Link | PDF (Nature Scientific Reports, open access)

 

I am a bit doubtful about there figure of 16% for major depression and the abstract statement that it and ME/CFS frequently occur together. My guess is that major depressive illness effects more like 2-5% of the population. I have not hear much about major depressive illness in ME/CFS. People are miserable but major depression or 'biological' depression is something pretty specific with early morning wakening and slowness of movement and thought.

Nevertheless, it is interesting to see no genetic link found.

 

"ME/CFS was confirmed based on patient-reported diagnoses."

 

The ME/CFS biobank GWAS data being used here appears to be the following, which is not the exactly same one as in the recently discussed paper by Chris Ponting (Replicated blood-based biomarkers for Myalgic Encephalomyelitis not explicable by inactivity), which consisted of 1,455 ME/CFS cases whilst here there are 2,076 ME/CFS cases. So what are the differences?

If the MDD biobank being used is as inaccurate as the biobank for ME/CFS probably is, it's hard to see how comparing the 2 is too sensible. But perhaps there is enough power in their sample sizes to handle the inaccuracies of the respective diagnoses.

At least it suggests that inaccurate diagnoses in these cases don't have much of an genetic overlap which is already some useful knowledge (or it's possible that the study is underpowered).

 

I’m actually really surprised no causal links were found, between lots of pwME being misdiagnosed with MDD as opposed to ME and lots of people with MDD being misdiagnosed with CFS back in the Oxford days…

 

I suspect it refers to lifetime diagnosis, not prevalence.

EDIT: the study it refers to writes:

The Epidemiology of Major Depressive Disorder: Results From the National Comorbidity Survey Replication (NCS-R) | Depressive Disorders | JAMA | JAMA Network

 

Yes, hard to see how misdiagnosis would bias the results towards no association, I would expect rather the opposite.

It also seems that the authors did not exclude many SNPs because they were associated with ME/CFS confounders, so that probably also isn't an explanation for the lack of association:

Not entirely clear though how they went from 69 to 57 SNPs in the end. They write:

 

On the basis of fatigue alone and conflating it as its own underlying cause. Which is incompetent, but they call that biopsychosocial these days.

I've read many descriptions of depression. I noticed two sets: one that fits the more common understanding of depression, and is easily differentiated from ME/CFS because it's basically the opposite: a deep feeling of worthlessness and sadness along with a complete loss of all motivation. The body works, but motivation is completely sapped away. They do feel better from doing things, including exercise, but they get zero motivation or joy out of doing it.

And then there's the rest: a grab-bag of obviously all sorts of things that have little in common other than medicine not understanding them, a lot of which is clearly just bad misdiagnosis. So all things considered, the self-diagnosed ME/CFS is probably not even the bigger problem here.

If depression is even one thing, misdiagnosis is rampant and probably not much better than self-diagnosis. Which should be expected, given the lack of diagnostic tests and decades of encouraging intentional misdiagnoses. In many cases people with ME and other chronic illnesses are misdiagnosed with depression because it's the only way to get any disability support. It's very polluted data.

Still this is about what would be expected out of comparing two grab bags of things medicine doesn't understand: not much correlation. Frankly it says far more about how deeply incoherent and out of its depth medicine is here, rather than anything having to do with the patients themselves. Which I don't think it really does, since any conflicting data is ignored by the ideologues anyway.

 

The conclusion of the paper

Our results confirm that there is no causal relationship between MDD and ME/CFS. While our study does not support a definitive causal link between them, the questions raised in this study are indeed worth exploring in future research to enhance our understanding of these complex disorders.

They were looking to see if there was a causal relationship, whereby having the genetic predisposition to MDD is a risk factor for ME/CFS. They didn't find a relationship. I am not surprised though there has been a lot of discussion in psychiatry about similar inflammatory pathways in the brain in rodent studies but nothing definitive.

This is different from a diagnosis of MDD (or misdiagnosis) which is due to the overlapping symptoms in the criteria, whereby ME/CFS symptoms are attributed to depression not a medical condition. Of course pwME may also present with MDD as well. The symptoms most indicative of MDD (and often seen in other chronic illnesses) are persistent depressed mood, anhedonia (decreased pleasure in activities that normally bring pleasure) excessive feelings of worthlessness and guilt which may become delusional, suicidal thoughts, plans or attempts.

16% of the global population having MDD is the prevalence over a lifetime. It could be one episode of major depression (MDE) or more than one (MDD - recurrent type)

You do not have to have genetic predisposition to MDD, to develop MDD, it only increases the risk of having MDD.

People with a family history of MDD, more than one episode of MDD, episodes of MDD developing earlier in life - have a greater risk of developing MDD

The current established contribution of genetic factors to MDD is about 40-50%. The other established non-genetic risk factors are severe childhood physical and sexual abuse, severe childhood emotional and physical neglect and severe life stress. Loss of a parent in early life may also have an effect.

For severe depression (used to be called biological depression vs reactive depression but now called MDD - Melancholic type), there is an even stronger genetic predisposition. Non-melancholic depression can also be severe but the Melancholic type is very notable on observation - as Jonathan indicated - severe psychomotor retardation - slowing of thought and movement, usually blank facies. They have diurnal variation of mood, significant weight loss, early morning waking and excessive quilt as well as the other required criteria for MDD. Melancholic depression is the most responsive to ECT.

 

This seems like quite an important study. Hopefully, it will get some more discussion and analysis.
@Jonathan Edwards @Simon M @Chris Ponting @chillier

The major limitation is the risk of misdiagnosis but the results look so strong that they might be immune to even large rates of misdiagnosis. I've asked a friend who knows a lot about genetics and math and he made the following reasoning:

It would be interesting if the authors could do a robustness check to calculate exactly how much misdiagnosis the study permits for there to remain strong evidence of no effect. But based on the tight confidence interval, even if 75% of ME/CFS cases did not have ME/CFS, the results would pretty much be the same.

 

Trying to understand where they extracted the data from. The paper refers to reference 8, a meta-analysis by Howard et al. 2019 of the three biggest GWAS studies on depression. The supplementary material of this study shows the 102 SNPs that reached statistical significance (P < 5 × 10-8).
Genome-wide meta-analysis of depression identifies 102 independent variants and highlights the importance of the prefrontal brain regions | Nature Neuroscience

But when I looked at the supplementary data of the current paper on ME/CFS, none of the 57 SNP identifiers matched the 102 of Howard et al. In contrast 41 out of 57 SNPs match with the 79 on obesity mentioned in this paper:
The Causal Relationship Between Body Mass Index and the Risk of Osteoarthritis - PMC

Could this be an error by the authors? Anyone able to check this? If it's an error, I hope it's only the names that are wrong and not the data....

 

Haven't spent a lot of time on this but, yes, none of the SNPs appear in the supplementary data table from Howard et al (41593_2018_326_MOESM3_ESM.xlsx column B - variants with a p-value < 5x10^-8 for an association with depression). Running a few of the rs-numbers through different search engines I see that many of them also appear as obesity or BMI-associated SNPs in these two different papers:

https://www.cghjournal.org/article/S1542-3565(22)00104-5/pdf
https://www.bmj.com/sites/default/files/attachments/bmj-article/pre-pub-history/Original article 23.4.18_0.pdf

 

Also checked and none of the SNPs match with Supplementary table 1. I also checked Supplementary Table 2 from Howard et al just in case since it lists SNPs, but none of those match Zhu et al's 57 SNPs either. I checked the first few SNPs shown in figure 2 A and D from Zhu just in case those were the correct ones, but they look to be the same as the paper's supplementary data SNPs.

 

Looking at Mendelian Randomization Analysis Reveals No Causal Relationship Between Nonalcoholic Fatty Liver Disease and Severe COVID-19, Li et al.

In "Supplementary Table 7. The Association of IVs With Exposure and Severe COVID-19" it lists a lot of SNPs.

Interestingly, the first five listed for BMI are the same as the first five in the thread's paper, in the same order. 42 of the 57 SNPs match those for BMI in Li et al. 46 if looking at all 2553 SNPs for all conditions in Li.

All 42 of the ones that match with those in the BMI section are in the same order, with 15 extra SNPs scattered in between for Zhu. I highlighted all identical clusters of at least three SNPs in both papers (not all matching SNPs are highlighted, only clusters):



Edit:

The order of the first six matches this paper's first six as well. Might be a standard order these papers all use.

 

Of the 15 from Zhu that weren't also associated with BMI in the Li paper, I found that at least 13 of them have been associated with BMI in other papers:

One of the last two is listed in a table for "obesity" but might mean BMI. The other one I couldn't find an association with BMI through a cursory search.

So at least 56 out of 57 of the SNPs in the thread's paper have been associated with BMI/obesity.

Edit: 57 out of 57, see next post.

 

For the one (rs2049045) that you couldn't find an association with either BMI or obesity - "Association of the BDNF Val66Met variation with obesity in women". It is not open access but there is this paragraph:

Another reference for rs2303223, "Identifying loci affecting trait variability and detecting interactions in genome-wide association studies"

 

Of the 42 SNPs that match those for BMI in Mendelian Randomization Analysis Reveals No Causal Relationship Between Nonalcoholic Fatty Liver Disease and Severe COVID-19, the effect allele and the alternative/other allele listed are exactly the same in both papers for all SNPs.

 

Thanks a lot @forestglip and @Nightsong

Will send a message to the authors asking for a clarification. Let's hope it just some minor error that they uploaded the wrong Excel sheet or used the wrong SNP identifiers or something like that. Everything else in the paper seems plausible: they cite the correct literature, databases, statistics, etc.

On the other hand the authors haven't published before on ME/CFS or mendelian randomisation from what I can see while the senior author has published a lot on acupuncture.
https://www.researchgate.net/profile/Luwen-Zhu/research