Analysis of data from 500k individuals in UK Biobank shows an inherited component to ME/CFS (Ponting blog)

Has there ever been a meaningful study of the male to female ratio under the different criteria?

What makes me ask is that I was recently reading a paper by Pelosi and Lawrie of epidemiological findings around Glasgow . I think it was 1994. it suggested equal numbers of males and females affected (which seems out of line with the modern view). Does a lot depend on the criteria, or are there other reasons for the different findings?

 

Thanks. The bit I had not understood was more basic but having read your answer and re-read the blog I think I understand: This study suggests people with ME are more likely to have valine instead of aspartic acid in a particular part of the P4HA1 protein. However, other species (such as the nematode) have this same amino acid substitution without it appearing to cause disease. If this substitution was not normal in some other species, the probability that it may be associated with disease would be higher. Is that correct?

 

OK, I still don't understand the word soup that I quoted above but this section from earlier in the blog helps.

So the assumption that it is heritable is because it's a DNA sequence, and therefore passed down from our parents.

 

Further to my question... Am I correct in thinking that a SNP is not necessarily a disease-causing mutation and could equally be silent, but could also indicate the possibility of further variation in that section of the genome (linkage disequilibrium)?

(I did study molecular medicine at uni, but it was 25 years ago!)

 

I think that is an interesting question. Maybe Luis Nacul, (and others) who assessed patients for various criteria, could extract that information.

 

To a first approximation I think yes. But as indicated, there may be caveats.

 

That was certainly true ten years ago. What I am not sure about is whether the technology is now so powerful that it would be relied on to pick up the disease gene in disequilibrium too. Simon may know. Chris will obviously know. Interesting point.

 

I think Ramsay and Hyde noted this too. Epidemics only seemed to have more women because of where they struck (schools, hospitals). Maybe this is something to do with a particular (ostensibly post-viral) subgroup?

 

I think I saw at least one study where less stringent criteria resulted in a male to female ratio that was higher than that seen with more stringent criteria.

That is typically interpreted by the patient community as evidence that the less stringent criteria result in more miadiagnoses, but a different interpretation could be that men are on average less affected than women. Or if we want to go with the two different diseases with different gender ratios, that the disease that is more common in men is the one that's usually not as bad.

 

Or that symptoms present differently in men than in women, like in for instance heart disease.

 

If the illness is perceived as being more common in women it could also lead to doctors not looking for it as much in men.

 

Sort of open-ended question on this research -

I'm thinking (please correct if wrong) this sort of bioinformatic analysis is relatively recent, and that typically it would be done on a disease with established etiology. It then adds another layer of understanding by sifting out the alleles that predispose.

Has this sort of analysis been successfully used to go in the opposite direction (be it in a disease or maybe just in answering a question about the function of some obscure bacterium - or roundworm) - fishing for mutations to then look for an explanatory pathology?
Just trying to clarify what the thought process going on here is.

 

I think people have been misdiagnosed. And that is pretty much how I viewed this biobank and ME/CFS, patients misdiagnosed resulting in bad data.

 

I was curious about this, too. I would assume that it would have to be quite high in order to get a p-value of 2.5x10^-12 ( 1 in 400 billion compared to the typical significance threshold of 1 in 20 ), but maybe not with the huge number of subjects without CFS.

The total number of subjects seems to be have been 408,455... so, 1829 with self-reported CFS and 406,626 without CFS*. I'm not familiar enough with statistics to know if just those numbers and the p-value are enough information to get the two groups' proportional values on the binary test for P4HA1, but someone else may know.

*[ ETA: This would be where the prevalence rate of 0.448% comes from: 1829 / 408,455 = 0.004477 ]

 

23 and me dosn' t cover this ( checked my aunt' s data from earlier thus year)

 

Not diagnosed - " self reported"

 

Actually it is more extreme than that. The paradigm of genetic searching in unravelling disease is ankylosing spondylitis. Around 1974 Bluestone's research lab decided to see if there was a link between gout and tissue type. Nobody knew what tissue the meant much at the time except that it seemed to be something to do with immune responses. The scientists sensibly included some disease controls. One was ankylosing spondylitis - which nobody knew much about. Derrick Brewerton heard that the Bluestone group had found by surprise that all the bank spend patients were HLA-27. Bremerton had a hunch that this told us that an spend was an immunological disease and repeated the study and published in the Lancet.

For about twenty years nobody really knew what the link to HLA B27 meant but gradually it became clear that it must have something to do with cytotoxic lymphocytes (which use tissue type proteins as signalling proteins). Now we know that ank spond and related diseases are mediated through this type of cell using certain cytokine pathways - and they are successfully teated with drugs targeting the cytokines.

The link between narcolepsy and DR2 was also completely unexpected.