Thank you for your comments in-thread too. Given your discussion in this paper and newer comments above, I'd like to follow-up with a question (probably naive, certainly non-expert, probably previously asked!). Qns at bottom after lead-in.
You don't need to qualify your questions before asking
. Thanks for engaging in such a great way with your thoughts and questions.
We can see these different, even contradictory, reported metabolic findings, sometimes from the same research group looking at different cohorts or longitudinally in cohorts. Does this imply that there might be a more generalised fuel-starvation issue? Ie something affecting substrate importing, but that can be variable in expression for involved substrate types or result in variable compensation mechanisms that increase the noise in our studies and obscure what's going on?
There are a few probable reasons for differences between studies and generally speaking I would say that they are not necessarily contradictory (at least I think not contradictory in the biology/disease sense).
1) difference in sample type (serum, plasma, urine, cells (or different cell types)).
2) within same group looking at eg: blood: metabolites are small molecules and pretty labile, so transport and handling differences can affect the results. I think this was actually even mentioned in one of the Hanson group's papers (finally met them a week ago, lovely people).
3) difference in measurement method. eg: NMR vs mass spec. Different sensitivity, quantification, and breadth of analytes captured.
4) cohort differences (whether due to criteria or demographics).
Basically this means (and this is my opinion, you don't need to believe it) that comparing broad strokes between studies and groups is fine for getting a taste of the direction of the field, but really, the value obtained can be a bit limited if comparing very specific details and measurements between groups, or even between studies within-group that eg: use different samples in scenarios where transit + handling, etc can affect the sample. (like metabolites). The comparisons are still worthwhile, they're just not absolute.
For those interested I have written about some particular metabolism study outcome comparisons in the titular paper of this thread and in my PhD thesis which is publicly available (and I would consider the thesis the director's cut of all of my papers plus extras, like proteomics and transcriptomics in cells working very well as disease classifiers, this is not published in the papers, it was done afterwards. I really recommend reading my thesis, it's just better in many ways).
As for your question about variability and your subsequent one about non-determinism, yes, it is possible. We just don't know enough yet. Generally speaking I also find it unlikely that every person with ME/CFS has the same problems going on and I am sure this is not a controversial view. Maybe this is also another contributor to variation between studies. This comes back to the need for subtyping when we make our group-group comparisons. My guess is that subtyping by clinical features is probably the best bet in the first instance. We can already cluster and potentially subtype by biological features seen in the lab, but I don't know what the real-world meaning of this is yet.
However, hypercoagulability keeps coming up, particularly (and maybe uniquely) in LC (which might represent early-onset ME vs the longer-duration patients typically enrolled in ME studies). Some ME studies report on plasma, others on serum. As I understand it simplistically: serum == blood fluid component with clotting factors removed; plasma == blood fluid component with clotting factors present but disabled (though measurable). Having "clotted off", fibrin-ogen (and stuff sticking to it) would be removed from serum, but would still be present in plasma, though not involved in active clotting. Could it be involved at the cell membrane, interfering with nutrient import? (This would imply something non-standard about the fibrin-ogen/attachments, perhaps wanting to stick more to cells and interact??).
This really isn't my area of expertise so I don't feel comfortable giving you an answer, but it's an interesting idea that I will think and read about. Thanks.
What determines the experimental design for using serum vs plasma when one doesn't particularly care about looking at the clotting components? Has anyone run identical cellular metabolism analyses in both serum and plasma to see if there's a difference in outcomes or does that prevent meaningful comparisons?
Again not my area of expertise but my understanding is that serum is more stable and has less noise. But it does require another step in preparation that usually involves chemicals. There may be other advantages or disadvantages.
One thing to note since you mentioned cellular metabolism, if we are talking about the effect of cells specifically (ie: aside from other factors less directly related to cells and their function), levels of metabolites in blood will be a steady-state that is resultant of a combination of cellular uptake (things leaving free circulation), intracellular activity (things changing outside of circulation), and efflux/rupture/lysis etc (things re-entering free circulation). So it's good that you are thinking about not just what's happening inside cells but on and around them, because measuring metabolites in biofluids isn't *only* a reflection of intracellular biochemical pathway reactions. I feel like this can get overlooked.
