2023 IIMEC15 - Alain Moreau

I agree with this on Moreau's last slide, even with the exclamation mark.
Unfortunately I didn't find this presentation to be part of the roadmap. There may be some true findings in there, but the presentation was not very convincing.

For example, Slide 10 15 mins Random Forest Classes with micro-RNA patterns
tldr: there are inconsistencies in the reported percentages, and they are making classes with ridiculously small numbers of members.

Spoiler: boring analysis of the reported percentages and group size

The data consists of 66 people, 14 of whom are designated SC, which is Short Covid, presumably they have a Covid infection and recover. The 66 people (including those recovered people) are divided into 6 broad predefined categories: ME, fibromyalgia, ME-FM, respiratory diseases, other neurological diseases, severe allergies. The idea is to match patterns of 11 micro-RNA to each category. But, it's not clear where those Short Covid people go - they must go in there somewhere, because the percentages in the pie chart are based on 66 people, not the 55 Long covid.

ME is noted as having 35% of the 66 people - that's 23. The ME groups is divided into a further 4 groups. But the percentages reported for the ME groups don't work for a total of 23 people. For example, one group has 5% of the total people in the broad ME group, but one person would be 4.3%. It looks as though 1 person accounts for 5.5% of the total number of people in the ME group. That's 18 people, not 23.

The broad fibromyalgia group has 6% of the 66 people. That would be 4 people. That group is divided into two groups, Severe and Moderate, so it might be 2 people in each group.

But, there's also a group called ME+Fibromyalgia. That has 23% of the 66 people - that's 15 people. This is further divided into two groups, with 55% on one group and 45% in the other group. But, you can't have 55% of 15 people.

Putting the problems with the inconsistent percentages aside, it is really stretching things to make 11 groups out of the 66 people, with a number of groups only having one or two people in them and suggest that you have found an important pattern.

And, there's no information there about how unique the micro-RNA patterns are to each identified group.
This work hasn't been published yet. Petre et al, 2023. There might be something good there, but, if so, the explanation does the work no favours.



For example, Slide 11 17 mins about SMPDL3B:
This continues the trademark of Moreau research, dividing sample populations into often very small sub-groups in order to claim that there is something diagnostic. The 7 people out of 46 people who performed worst on the cognitive tests were reported as having a 'severe elevation of SMPDL3B'. The people in this group had a mean level of SMPDL3B of about 90 ng/ml. The people in the other two groups had a level of about 38 ng/ml.

Slide 12:Moreau then tells us that high or normal levels could both indicate a problem.

Slide 13:
Chart A - shows a relationship between SMPDL3B and ME/CFS severities. Levels increase from about 50 (mild) to about 170 ng/ml (severe).

Chart B - shows the relationship between age and SMPDL3B levels. It isn't clear if these people are healthy or not. what is clear is the enormous variation in SMPDL3B levels with age. People aged 18 to 30 years had levels of about 170 ng/ml. People aged 30 to 50 years had levels of about 75 ng/ml. People aged over 50 years had levels of about 50 ng/ml. These levels don't seem to fit with the idea of people with memory problems having higher levels that the cognition testing seemed to suggest. Moreau explains that the reason for the dramatic reduction with age is because of the decline in oestrogen levels in females over time. But, we aren't given any details about the sex or age of the people in the cognitive testing or the severity chart. Also, the 'severe elevation' reported in the cognitive testing (90 ng/ml) isn't looking very severe after all.

Chart C - shows levels for males and females. We aren't told if they have ME/CFS or not. Females have about 105 ng/ml, males have about 50 ng/ml.

Chart D - people with ME have about 40 ng/ml; people with POTS and ME have about 70 ng/ml; and finally we have a level for clearly noted healthy controls - 58 ng/ml.

Charts E and F - have levels between 40 and 80 ng/ml, with a story supposedly explaining why people with low levels of "PEM" have SMPDL3B levels similar to those of the people with high levels of "PEM" but not to the levels of people with moderate "PEM". A "u-shaped response" maybe is a thing, but when the differences in the means is so small compared to the massive differences shown for age or sex, it doesn't seem credible.

The levels are all over the place and are probably better explained by sex and age differences. Maybe there is something there, but the presentation was not at all convincing. It would take a much more careful analysis, teasing out the confounders.

The problems continue on Slide 14. See for example, chart F, which divides participants by whether they have SMPDL3B levels of more of less 27 ng/ml. People with more than 27 ng/ml are reported to have more problems remembering things than people with less than 27 ng/ml. But the differences in the 'trouble remembering things' scores are tiny. And why use 27 ng/ml as the separation point? Does that produce the biggest separation - which is still very underwhelming?



In conclusion - All the nice coloured charts and science-y words are superficially impressive, but I think there are all sorts of problems. I haven't looked at the cerebral blood flow findings yet.