Cell-Based Blood Biomarkers for Myalgic Encephalomyelitis/Chronic Fatigue Syndrome: Missailidis et al Feb 2020

[Sly Saint]

Abstract
Myalgic Encephalomyelitis/Chronic Fatigue Syndrome (ME/CFS) is a devastating illness whose biomedical basis is now beginning to be elucidated.

We reported previously that, after recovery from frozen storage, lymphocytes (peripheral blood monocytic cells, PBMCs) from ME/CFS patients die faster in culture medium than those from healthy controls. We also found that lymphoblastoid cell lines (lymphoblasts) derived from these PBMCs exhibit multiple abnormalities in mitochondrial respiratory function and signalling activity by the cellular stress-sensing kinase TORC1.

These differences were correlated with disease severity, as measured by the Richardson and Lidbury Weighted Standing Test. The clarity of the differences between these cells derived from ME/CFS patient blood and those from healthy controls suggested that they may provide useful biomarkers for ME/CFS.

Here we report a preliminary investigation into that possibility using a variety of analytical classification tools, including linear discriminant analysis, logistic regression and Receiver Operating Characteristic (ROC) curve analysis.

We found that results from three different tests, lymphocyte death rate, mitochondrial respiratory function and TORC1 activity could each individually serve as biomarker with better than 90% sensitivity but only modest specificity vís a vís healthy controls.

However, in combination they provided a cell-based biomarker with sensitivity and specificity approaching 100% in our sample. This level of sensitivity and specificity was almost equalled by a suggested protocol in which the frozen lymphocyte death rate was used as a highly sensitive test to triage positive samples to the more time consuming and expensive tests measuring lymphoblast respiratory function and TORC1 activity.

This protocol provides a promising biomarker that could assist in more rapid and accurate diagnosis of ME/CFS.

https://www.preprints.org/manuscript/202002.0029/v1

 

[John Mac]

From the same researchers last year

https://www.s4me.info/threads/an-is...me-cfs-patients-missailidis-et-al-2019.11121/

 

[alex3619]

The real test will be specificity with respect to immune and mitochondrial disorders. However, though on the odds this test will fail like all the others, this is what we might see on what will be the eventual successful test. We have to wait for replication and further testing.

 

[borko2100]

We found that results from three different tests, lymphocyte death rate, mitochondrial respiratory function and TORC1 activity could each individually serve as biomarker with better than 90% sensitivity but only modest specificity vís a vís healthy controls.

However, in combination they provided a cell-based biomarker with sensitivity and specificity approaching 100% in our sample.

Wow, if true, this is groundbreaking.

 

[Andy]

A sensible note of caution from @Simon M

When you combine three different measures, it’s not that hard to get close to 100%. At least not until it comes to replication in an independent cohort.

 

[Milo]

Where are these authors from, and who are they affiliated with?

 

[Andy]

Where are these authors from, and who are they affiliated with?

Latrobe University in Australia.

From the same researchers last year

https://www.s4me.info/threads/an-is...me-cfs-patients-missailidis-et-al-2019.11121/

Paul Fisher, one of the authors, presented on their work at the Emerge symposium last year, https://www.s4me.info/threads/video...nsatory-changes-in-me-cfs-patient-cells.9177/

 

[Jonathan Edwards]

It is worth remembering that a test with 100% sensitivity and specificity is useless, because it recognises a group that you have already recognised clinically.

Tests are only useful if they indicate some underlying process that might be recognised better by the test than by clinical assessment. If the test is better it must have a specificity and sensitivity of less than 100% because at least some of the clinical assessments must be wrong (for there to be something better).

 

[Londinium]

This line jumps out at me, unfortunately:

Testing was carried out at participant homes when the severity of illness precluded patients from travelling.

This implies patient samples and control samples were not treated equally. If, say, patient samples traveled further before being frozen that might explain a high lymphocyte death rate upon being unfrozen, might it not?

I hope that, upon coming out of pre-print, the peer review process will have necessitated the authors clarifying as to whether patient and HC samples were treated identically.

 

[rvallee]

A sensible note of caution from @Simon M

I'm not sure I understand the problem with that. Many diseases are diagnosed this way. It's not ideal but nothing ever is.

 

[alex3619]

If the test is better it must have a specificity and sensitivity of less than 100% because at least some of the clinical assessments must be wrong

One of my concerns is that ME might turn out to be some kind of spectrum disorder, and that we share hundreds of potential biomarkers but no biomarker is core to the disease.

 

[Ryan31337]

I thought it interesting that they made reference specifically to co-morbid POTS in the CCC-guided patient selection.

Would seem to me like including those with POTS makes drawing any conclusions harder, given the many known diseases and issues that present with POTS beyond just CFS/ME. I doubt they were experts in POTS too and able to confidently exclude all of these.

Even if they did, POTS research is showing secondary mito dysfunction in many of the patients, not all of which have a CFS/ME diagnosis...

 

[Hutan]

It is worth remembering that a test with 100% sensitivity and specificity is useless, because it recognises a group that you have already recognised clinically.

Yes, but as Simon and others have pointed out, they have essentially only tested their 3 biomarker model on their initial dataset. So, having a model that perfectly predicts diagnosed ME/CFS in that initial dataset is a promising start. It's better than if they combined their 3 best factors and there was only a 70% specificity.

Lymphoctye death rate propensity (after freezing and thawing) does show pretty good separation.


It sounds as though they have proved to their satisfaction that the length of time frozen doesn't affect lymphocyte viability upon thawing. But yes, as @Londinium said, there is the possibility that the samples from severe patients waited longer before being frozen. Still, I would have thought the majority of samples would have been from patients who could come in to the lab to give blood, just as the controls did. And the separation between CFS and controls is pretty good.

It would be nice to have some confirmation from the researchers that time from collection to freezing isn't significantly different between groups.

edited to add:

The ME/CFS and control sample sizes were 57 and 33 individuals respectively....
(cyan shading). ... The “best” threshold for the propensity score (0.59) is shown, together with the specificity (0.76) and sensitivity (0.84) at that threshold (in parentheses).

 

[Hutan]

Given this is a pre-print and maybe the authors are reading the thread:
Page 8, this title looks like a typo, should be 'ME/CFS and control blood samples'

Immortalized lymphocytes from ME/CFS and patient blood samples can be distinguished by mitochondrial and cellular respiratory dysfunction

 

[Hutan]

So they looked at the ability of a score relating to 5 mitochondrial respiratory measures derived from the seahorse machine using lymphoblasts. Their conclusion was that this score worked about as well as the lymphocyte death related score.


Again, pretty nice separation.

At the “best” threshold of the regression propensity score, the false positive error rate was 30% and the false negative error rate was 10%.

So, not bad. Incidentally, another typo I think, 'regression' should be 'respirometry'.

 

[Hutan]

Same story with the third parameter - a measure of TORC1 activity - good separation of values between ME/CFS and controls, very similar values as a biomarker.

The “best” threshold for the propensity score (0.59) is shown, together with the specificity (0.77) and sensitivity (0.89) at that threshold (in parentheses).

So, when they combine the three:

Combining measures of frozen lymphocyte death rate, lymphoblast mitochondrial dysfunction and lymphoblast TORC1 signalling discriminates ME/CFS and patient blood samples with high accuracy.

(another typo - ME/CFS and controls).

At the “best” threshold of 0.61 in the ROC analysis, the specificity was 100% and the sensitivity was 97%.

And then they talk about the possibility of a 2-step test, with the seahorse analysis only being done if the lymphocyte death rate is bad.

These outcomes suggest a testing protocol in which an initial assay of lymphocyte death rate would be followed, in the case of positive results by lymphoblast isolation, Seahorse respirometry and TORC1 activity assay. Because it involves seven different biochemical, mitochondrial and cellular parameters, such a combination of tests may provide greater specificity vís a vís other conditions (eg. thyroxin deficiency) that may cause chronic fatigue.

This was interesting in relation to JE's concern:

Of the 29 ME/CFS patients for whom we have data for all three assays, 27 would have been classed as “ME/CFS suspected” based on lymphocyte death rates and all of these would have been subsequently confirmed by the combined tests (Fig. 7c, Table 5). Of the remaining two patients classed as “ME/CFS not suspected”, one was “ME/CFS positive” in the combined tests. This individual would have been misclassified, if the combined tests had not been done and thus would have been missed in a protocol using the lymphocyte death rate alone as a screen. The remaining individual was misclassified as not having ME/CFS even using the combined tests.

So the test didn't pick up all of the people diagnosed with ME/CFS. The abstract only claims 'sensitivity and specificity approaching 100% in our sample'.

And this was interesting given my earlier impression that they had only had a training set of data, although of course the numbers involved are small.

To confirm that the proposed biomarkers would be effective in correctly diagnosing ME/CFS in new samples that were not part of the original analysis, we randomly selected 80% of our sample for use as a training set while the remaining 20% was used as a test data set. Despite the fact that the smaller training set makes the parameter estimates and threshold determination less accurate, very similar results to those obtained with the full data set were obtained in all of our assays and all performed as well with the test data set as with the training set

 

[Hutan]

Sorry to be going on a bit, but I feel a bit hopeful about this team's work. They have the best opening sentence in an ME/CFS paper that I have seen, I think.

Myalgic Encephalomyelitis/Chronic Fatigue Syndrome (ME/CFS) is a devastating illness whose biomedical basis is now beginning to be elucidated.

There is quite a bit in the discussion that is interesting.

They say that it would be useful to compare the viability of frozen lymphocytes with other diseases.

It has been documented previously that frozen PBMC viability is also reduced in paediatric Dengue fever

But they think the Step 2 Seahorse tests would differentiate ME/CFS from the rest.

To be confused with ME/CFS in the confirmatory tests of mitochondrial respiratory function and TORC1 activity, other illnesses would not only need to cause reduced viability of frozen lymphocytes, they would also need to confer upon the derived lymphoblasts the same pattern of molecular abnormalities – decreased Complex V efficiency, elevated proton leak as a proportion of basal metabolic rate, as well as increases in maximum respiratory capacity, Complex I activity, nonmitochondrial oxygen consumption and TORC1 activity.

They note that lymphoblasts from Parkinson's patients look quite different to those of ME/CFS in these tests.

They discuss why the ME/CFS thawed lymphocytes might be dying quicker. They say it could be due to the poorer mitochondrial function:

Impaired mitochondrial respiratory function, including Complex V impairment, has long been known to result in apoptotic cell death in ex vivo lymphoid cells [45]. However, we have not demonstrated that the lymphocyte death we observe is apoptotic or whether it is one of the other known forms of eukaryotic cell death.

They say that whatever the cell death pathway is, it may reflect the inability of the cells to respond normally to cell damage or stress (in this case freezing). They note that freezing causes damage, but also has been documented to affect PBMC expression of stress response genes.

They noted this:

Compared with controls, the number of dead ME/CFS lymphocytes continued to increase at a faster rate than the controls over multiple days in
culture. This suggests that underlying and ongoing cytopathological processes could be contributing towards cell death in culture of previously frozen lymphocytes.

I'm not sure what they have in mind. But they seem to be suggesting that there could be something pathological going on not directly related to the freezing. I wonder what the death rate is in ME/CFS lymphocytes that aren't frozen.

Anyway, thanks to Daniel and the team for remaining interested in ME/CFS and their good work. And the funders, including the Mason Foundation which I have given a bit of a hard time previously for some of their funding choices.

I look forward to replication of the work by this team and others. I note that the team in Otago have had some trouble getting consistent differences between people with ME/CFS and controls in the Seahorse studies; they thought they had some technical issues to sort out.

 

[borko2100]

I really think they are on to something. Their main idea is that Complex V inefficiency is the root cause of ME/CFS. I don't understand exactly what Complex V is and what it does, however it seems that ME/CFS might be a milder form of a genetic mitochondrial disease called Complex V deficiency.

Mitochondrial complex V deficiency can cause a wide variety of signs and symptoms affecting many organs and systems of the body, particularly the nervous system and the heart. The disorder can be life-threatening in infancy or early childhood. Affected individuals may have feeding problems, slow growth, low muscle tone (hypotonia), extreme fatigue (lethargy), and developmental delay. They tend to develop elevated levels of lactic acid in the blood (lactic acidosis), which can cause nausea, vomiting, weakness, and rapid breathing.

The bold part describes ME/CFS very well, or at least a subset. Namely, extreme lactic acid build up and muscle fatiguability after trivial / minor activities, such as walking, staying upright, etc. I recall reading several papers confirming this lactic acid issue.

I wonder though if Complex V deficiency also has PEM as a symptom?

PS:

Another mitochondrial disease affecting the complexes, Leigh's disease seems to be triggered in certain cases by a sudden stressor. From wikipedia:

Symptoms are often first seen after a triggering event that taxes the body's energy production, such as an infection or surgery. The general course of Leigh syndrome is one of episodic developmental regression during times of metabolic stress. Some patients have long periods without disease progression while others develop progressive decline.

If this mitochondrial disease can be triggered in such a way, then maybe ME/CFS could be triggered in a similar manner? Very interesting.

 

[Milo]

I thought it interesting that they made reference specifically to co-morbid POTS in the CCC-guided patient selection.

Would seem to me like including those with POTS makes drawing any conclusions harder, given the many known diseases and issues that present with POTS beyond just CFS/ME. I doubt they were experts in POTS too and able to confidently exclude all of these.

Even if they did, POTS research is showing secondary mito dysfunction in many of the patients, not all of which have a CFS/ME diagnosis...

The world of POTS is messy as well. Some of the internationally known POTS specialists do not believe in ME and believe those who have ME have primary POTS. Then their POTs patients are ‘tired’ too. Many if not all display several symptoms which overlaps with ME, without ever being diagnosed with ME.

Other than Dr Rowe, POTS and ME specialists are not typically mingling and in my view, they should. They should be invited to ME conference and vice versa.

To add to the complexity of just being diagnosed, first line physicians are not necessarily able to identify patients with POTs that come to their office. In my case, when i initially went in because i was feeling worse upright and slightly better reclining, i was deemed lazy and depressed and told i needed anti-depressants. I had to go 7 hours in a plane to a doctor who knew exactly what was wrong. And when i got back the doctor was still in disbelief, though willing to prescribe me atenolol which had been started in the US. It was another 8 years until she was willing to do a NASA lean test so I could be included in a study.

 

[butter.]

It is worth remembering that a test with 100% sensitivity and specificity is useless, because it recognises a group that you have already recognised clinically.

Tests are only useful if they indicate some underlying process that might be recognised better by the test than by clinical assessment. If the test is better it must have a specificity and sensitivity of less than 100% because at least some of the clinical assessments must be wrong (for there to be something better).

This is just plain wrong. Your perfectly identified clinical diagnosis of me/cfs still makes you a psychiatric patient in 99% of all cases.