Should iCPET patients be genotyped?

[tralfamadorian97]

I am interested in the relationship of two ME/CFS research programs :

1. Chris Ponting et al.'s work on DecodeME. See paper and discussion

Advantages:

• Large sample
• As a genetic study, can uncover causal biology.

Disadvantages:

• While more precisely specified than most ME/CFS cohorts, there is still ambiguity resulting from use of a questionnaire to define phenotype.

2. Invasive cardiopulmonary exercise testing (iCPET) studies performed by David Systrom's group at Harvard, I. Singh/ P. Joseph's group at Yale, and others. These groups report two primary iCPET abnormalities: reduced peak right atrial pressure, and reduced peak arterio-venous oxygen difference, and several secondary abnormalities. See Paper 1, Paper 2 (Discussion), Paper 3 (Discussion), Paper 4, Paper 5 (Discussion), and Paper 6 (Discussion).

Advantages:

• Instead of defining phenotype via questionnaire, these studies report abnormalities that are precisely measured via techniques established for the study of heart and lung disease

Disadvantages:

• Since the studies are non-genetic, it is unclear where the measured abnormalities lie in the pathological causal chain.
• Smaller samples.

I would like to know how these two research programs relate.

• Are the two groups studying patients with same underlying biological pathomechanism?
• If so, what is the causal chain by which the genetic changes reported by DecodeME would cause iCPET abnormalities?

I think answering these questions requires more data. The obvious path would be for the iCPET groups to ask patients undergoing iCPET to consent to genotyping and inclusion in genetic research. One could then run genetic studies of reduction in peak arterio-venous O2 difference, reduction in peak right atrial pressure, and other iCPET abnormalities. There would be a loss of power compared to DecodeME due to reduced sample size, but a gain in power from using a quantitative objectively-measured phenotype instead of a binary questionnaire-based phenotype. One could examine the extent to which the genetic signal from these studies coincides or does not coincide with that of DecodeME.

 

[forestglip]

These groups report two primary iCPET abnormalities: reduced peak right atrial pressure, and reduced peak arterio-venous oxygen difference, and several secondary abnormalities. See Paper 1, Paper 2 (Discussion), Paper 3 (Discussion), Paper 4, Paper 5 (Discussion), and Paper 6 (Discussion).

I just did a quick look for right atrial pressure, looking at the papers specifically about ME/CFS.

Paper 3 is about ME/CFS, and it does say this in the abstract:

During exercise, the 160 ME/CFS patients averaged lower RAP [right atrial pressure] (1.9 ± 2 vs 8.3 ± 1.5; P < .0001) [...] than control subjects.

But this seems misleading to me. For the ME/CFS group, they specifically excluded those with RAP >= 6.5.

Data from 1,516 clinically indicated iCPETs performed between 2011 to 2018 at the Brigham and Women’s Hospital Dyspnea Clinic (Boston, MA) were analyzed. To eliminate potentially confounding submaximal effort and intrinsic heart and lung disease, we excluded the following:

[...]

• 5. RAP ≥6.5 mm Hg, which defines our lower limit of normal11

For controls they say all have normal iCPET results:

A control cohort assembled for comparison purposes comprised 36 patients who had undergone iCPET for exertional intolerance but who had normal results.

Does this mean they excluded controls with abnormal RAP and excluded cases with normal RAP? It seems that would guarantee seeing a significant difference in RAP.

---

Paper 5 is also about ME/CFS, but shows p=0.804 for right atrial pressure. Paper 6 is an abstract about ME/CFS but doesn't have a control group.

Edit: Fixed paper numbering.

 

[tralfamadorian97]

Good point about the right atrial pressure in paper 3. I had forgotten about the cohort filtering.

The impaired systemic oxygen appears to be the stronger finding, since multiple groups report it in various versions of long covid or ME/CFS. So it would be interesting to investigate the genetics of this iCPET-measured reduced systemic oxygen extraction, and see if it overlaps with DecodeME

 

[Hutan]

There's a problem if ME/CFS is largely the result of random processes in response to environmental triggers. That would mean that many people with ME/CFS would not have a genetic predisposition.

DecodeME found some genetic signals yes, but I think many of the participants would have had none of the identified genome regions of interest issues. There are much smaller studies that have identified genetic differences, but at least one seems to have had some sort of interpretational problem.

I think there would be a good chance that there would be no useful genetic signals from the small number of people who have undergone iCPET. Perhaps I am wrong. But, perhaps it would be better to do an analysis further downstream from genes - maybe an analysis of mRNA or protein expression of genes or pathways of possible interest?

 

[forestglip]

The impaired systemic oxygen appears to be the stronger finding, since multiple groups report it in various versions of long covid or ME/CFS.

I admittedly have not looked deeply at any of these papers. Maybe you could point me to the strongest ones.

But just looking at Paper 4 (Singh 2022), it looks like it might be a similar kind of selection bias responsible for the difference. From abstract:

The patients who had recovered from COVID-19 exhibited markedly reduced peak exercise aerobic capacity (oxygen consumption [VO2]) compared with control participants (70 ± 11% predicted vs 131 ± 45% predicted; P < .0001). This reduction in peak VO2 was associated with impaired systemic oxygen extraction (ie, narrow arterial-mixed venous oxygen content difference to arterial oxygen content ratio) compared with control participants (0.49 ± 0.1 vs 0.78 ± 0.1; P < .0001)

On selection of controls:

we identified 10 age- and sex-matched control participants from our iCPET database. This cohort consisted of symptomatic patients without a prior history of COVID-19 who previously underwent iCPET for clinical investigation of exertional intolerance, but who exhibited a normal physiological limit to exercise defined by a peak oxygen uptake (peak oxygen consumption [VO2]) and peak CO of ≥ 80% predicted.

The main result is about difference in peak VO2, but they selected controls who had normal VO2. So expected, right?

I don't know how peak VO2 or peak CO related to systemic oxygen extraction, but it seems plausible that they are correlated, and thus the observed difference in that parameter may have more or less also been expected based on their selection criteria.

I feel like I must be missing something, because otherwise the abstracts of these papers are very misleading.

 

[forestglip]

I think there would be a good chance that there would be no useful genetic signals from the small number of people who have undergone iCPET.

Yes, maybe I'm wrong, but I think even continuous, objective traits typically require pretty large cohorts for GWAS. This test sounds pretty invasive, so I'm not sure the sample size will be there.

I'm also still not sure how much iCPET relates to ME/CFS. The two papers I looked at above didn't fill me with confidence.

 

[Eddie]

I'm also still not sure how much iCPET relates to ME/CFS. The two papers I looked at above didn't fill me with confidence.

I am having a hard time trying to understand how the reached the conclusion that:

RAP ≥6.5 mm Hg, which defines our lower limit of normal

It comes from this paper: https://onlinelibrary.wiley.com/doi...getft_integrator=sciencedirect_contenthosting

They took 619 people that had a clinical indication for an iCPET, so already not people who are healthy. They removed 542 of those people due to other explanations for the exercise intolerance being found or a problem with the test itself.

Of the 77 people left they separate them into two groups, people with VO2 max over 80% predicted and those under 80% predicted. They then say:

impaired patients were found to have decreased absolute biventricular filling pressures at maximum exercise (Fig. 2) as well as less augmentation of RAP and PCWP with exercise (ΔRAP: 4 [2–5] vs. 6.0 [4–7] mmHg,

So they claim to show that RAP is related to lower V02 max in a group of people with unexplained exercise exertion. But I don't think that shows that lower RAP are related to the exercise intolerance, because it seems like all these people had exercise intolerance regardless of their RAP. The 80% VO2 number is also an arbitrary cutoff, although there might be some justification for that in the exercise testing world.

The only justification for the 6.5 mmHg number I can find is:

From these data, a diagnosis of preload insufficiency should be considered in patients with RAPmax < 6.5 mmHg, PCWPmax <12.5 mmHg (Fig. 3), ΔRAP < 5.5 mmHg, or ΔPCWP < 6.75 mmHg based on receiver-operator characteristic analyses of the single-test cohort (data not shown)

I understand why they didn't or couldn't subject healthy controls to an intensive CPET. But I don't think we have good justification to think that 6.5 mmHg is the appropriate cutoff for healthy filling pressures or that it has been shown to related to exercise intolerance. I think the test could be useful, but we really need to know what RAP looks like in actually healthy people.

 

[Eddie]

Paper 5 is also about ME/CFS, but shows p=0.804 for right atrial pressure. Paper 6 is an abstract about ME/CFS but doesn't have a control group.

Paper 5 is potentially interesting but the sample size is far too small. There are only 11 controls from two different sources.

The control arm comprised of two groups; one evaluated in our pulmonary hypertension clinic for dyspnea or echocardiographic evidence of PH and a subsequent normal iCPET and individuals in good health recruited from a study in Belgium.

It would be interesting to know how many of the 11 control patients were from the pulmonary hypertension clinic and how many were recruited in good health.

The p=0.804 number is the RAP at rest. Controls had an average RAP of 3 and ME/CFS of 5 at rest.

At peak exercise control RAP increased to 4 and in ME/CFS declined to 3. So maybe there is something in the fact that control RAP increased from 3 to 5 and ME/CFS RAP declined from 5 to 3. But very small sample size and issue with potentially unhealthy controls.

 

[Jonathan Edwards]

I don't think we have good reason to think that CPET studies of any sort pick out some special ME/CFS pathology. The results have been very variable and where they show something it is a statistical shift in a population, not a qualitative difference. I find Systrom's cohort selection very difficult to interpret.

And as Hutan says, if the genetic component in ME/CFS is only 10% or so and scattered over about fifty genes in the main (the eight were picked out as statistically robust but the plots suggest there are lots more) and likely many more genes to a small extent, trying to link genes to physiology tests is doomed to failure.

And I don't think brain fog, disturbed sleep rhythms, light sensitivity or even PEM are ever going to be explained by shifts in right atrial pressure. So I tend to see the CPET findings as likely to be rather unreliable epiphenomena that distract people from the actual problem.