Chris Armstrong - Melbourne ME/CFS researcher, research updates and general chat

[Jonathan Edwards]

I think clinicians provide the first layer of evidence

Well, yes, they would do, even if it is taking notice when a patient mentions a surprising improvement. But things are a bit more complicated. The use of rituximab was a specific test of a hypothesis generated through the serendipitous observation of patients improving with B cell cytotoxics. The chances of benefit were lowish but credible. It was trialled with impeccable methodology and a negative result obtained.

Ampligen seems to me almost the opposite. It was made by a company that thought it might be marketable as a drug. Ten or more years later after a number of inconclusive trials in various diseases we still have no idea if it is any use for anything. It doesn't look as if it would be. If it was we would probably know. Instead large numbers of people have paid to have it without us being any the wiser.

Plausibility doesn't provide proof of efficacy but it goes a long way towards what one can reasonably bet on being likely to be any use. And plausibility has a lot to do with the way physicians play with things. If a physician seriously thinks a treatment is going to work they design a decent trial. If they know that it probably doesn't they either don't bother to do trials or design a bad trial that can be made to look positive - whether for GET or an 'immunomodulator'.

 

[Jaybee00]

Why try specific immunotherapies when these drugs have significant side-effects and we do not have any indication they would be helpful?

Why try specific immunotherapies when these drugs have significant side-effects and we do not have any indication they would be helpful?

Strongly disagree with your posts above. No one has proven that MECFS is NOT an immune mediated disease. People treated with cyclophosphamide have gone back to work.

Most patients are interested in significant improvements to their current health status. The maybe 2% improvement that might come from LDN isn’t it.

Almost anything that will significantly help you will almost certainly have significant side effects or risks.

 

[forestglip]

"Personalised Treatment Trials: Monitoring and Improving the Connection of Treatments to Patients"



Description from YouTube:

"The Melbourne ME/CFS Collaboration is conducting a study to closely watch how different treatments affect ME/CFS and Long COVID patients, using a very tailored approach. The team will collect samples from patients before, during, and after their treatment to see what biological changes happen and how these relate to whether they feel better or not. This method will help the researchers figure out which treatments work, why they work, and who they work best for.

The team plans to study 200 patients over three years. Each patient will be part of the study for about 3 to 4 months, during which they will try 2 to 3 different treatments. This includes three visits to the clinic and involves analysing many samples they provide, like blood and urine. This large collection of data from all the patients will help the researchers make better predictions about which treatments are most likely to help others in the future.

Watch this video to learn more about this study from Chris Armstrong, PhD, Director of the Melbourne ME/CFS Collaboration."

Additional important info in the video, text generated by AI:

"One key goal of this personalized treatment trials project is to develop a decision-making tool that doctors can use to more efficiently match patients with treatments that are likely to help them, based on their individual biology and medical history. By collecting a large amount of diverse data from many patients and applying machine learning techniques, the researchers aim to identify patterns that can predict which treatments will be most effective for specific subgroups of ME/CFS patients. This could help move the field beyond the current trial-and-error approach and allow patients to access helpful treatments earlier in their illness. Additionally, insights from this study about the Australian ME/CFS patient population and care pathways could inform future research and clinical practice."

 

[MelbME]

Yes this is a large scale project that we expect to take another 2-3 years to complete. The idea is to build an initial large dataset for a predictive model in this heterogenous population. And then continue to refine he model as its used. Will keep data scale large initially but will cut out assays and measures that don't seem useful to the algorithm so it gets cheaper and cheaper to monitor patient treatment outcomes this way.

Please feel free to ask questions on this project. I will answer any of them. And suggestions are good to as we haven't begun collecting data.

 

[Jonathan Edwards]

I am interested to know how this method would work @MelbME.

At present we do not know if any treatments are effective. If treatments are given and the person reports improvement and there are certain changes in biologic samples then we will not know whether or not either the changes in symptoms or samples relate to the treatment. I don't think we would even know whether they relate to ME/CFS status since intercurrent factors like respiratory infections could easily alter perceived levels of symptoms and also biological samples.

I can see that it work for changes in samples that could only be attributable to a treatment and were also not just normalisations of measures. If those who reported getting better on a vasopressor agent had consistently higher blood pressures then that would make sense. What I am not clear about is the rationale for screening across lots of different treatments, although that may be a misunderstanding.

 

[MelbME]

I am interested to know how this method would work @MelbME.

At present we do not know if any treatments are effective. If treatments are given and the person reports improvement and there are certain changes in biologic samples then we will not know whether or not either the changes in symptoms or samples relate to the treatment. I don't think we would even know whether they relate to ME/CFS status since intercurrent factors like respiratory infections could easily alter perceived levels of symptoms and also biological samples.

I can see that it work for changes in samples that could only be attributable to a treatment and were also not just normalisations of measures. If those who reported getting better on a vasopressor agent had consistently higher blood pressures then that would make sense. What I am not clear about is the rationale for screening across lots of different treatments, although that may be a misunderstanding.

There will be significant variance. It's relying on the treatment journey of hundreds of people. There will likely be a lot of overlap of treatments tried because of the network of GPs we will use, they at least have some idea of what ME/CFS is and they aren't pushing exercise. So it's not a seemingly endless list of treatment but there are 2 reasons for the way we designed this project:

- feasability: this is an observational trial, we aren't telling GPs how to treat patients, we are simply collecting large amounts of data before and after treatments. The ethics is possible, if we were dictating the trial the process would be 10x longer and 10x more expensive.
- the outcome is meant to be a tool used to help improve decisions on treatments for symptoms/co-morbidities based on patient data. It will be used in the same way that we collect the data (in uncontrolled fashion). So as I said, variance is expected, the solution to the variance is the scale of data collected. The data will be clinical and biological (surveys, functional tests, basic pathology and more detailed biological measures).

It's really leveraging new AI-based analytics to make decisions from data that we don't understand. Providing more data helps, we are just making sure the data is of high quality, reliable and consistent. The variability comes from the factors you highlight that we will try to track with patient diaries. It's certainly an ambitious project. At the very least we are going to deeply understand a number of people with the disease. We don't just expect the outcome to be the prediction tool. This project (or program) will form the basis of a lot of work in our lab for years.

 

[MelbME]

Imagine if everyone with ME on this site had a significant work up before each treatment they tried during their journey so far and were tracked during the trying of that treatment to monitor outcome.

Now imagine that this work up and monitoring was conducted the same way and was targeting measures that we know are likely important in ME among other data.

That's what this project is attempting to do.

 

[forestglip]

Imagine if everyone with ME on this site had a significant work up before each treatment they tried during their journey so far and were tracked during the trying of that treatment to monitor outcome.

Now imagine that this work up and monitoring was conducted the same way and was targeting measures that we know are likely important in ME among other data.

That's what this project is attempting to do.

What metric(s) will you use to track how bad a patient's ME is?

Edit: If it's not too expensive, maybe wearables could be incorporated? Tracking things like step count, number of sit-to-stands, time spent laying down.

 

[MelbME]

What metric(s) will you use to track how bad a patient's ME is?

Edit: If it's not too expensive, maybe wearables could be incorporated? Tracking things like step count, number of sit-to-stands, time spent laying down.

Do you mean how they feel? FUNCAP looks the best to use right now because it captures the patients perception of their functional capacity. Combine with function al information like from wearables seems like the best at the moment.

Yes wearables are actually a cheap way to produce a lot of information that is at least reliable to the individual and therefore delta change is useful.

We have been trialling these devices called activpal to look at position for the purpose you highlight. https://www.palt.com/why-activpal/
It's the only device we found that is reliable with position. Pretty underrated for this disease and we're doing a mini-trial with an at-home continuous blood pressure device to try capture OI problems in the home for patients. Looking at the feasibility of an all-in-one device for people to track if they might have OI. A lot of healthy people have OI but we never really study this and why is it different with disease but this is a critical question to answer.
Also we've been developing a completely novel symptom tracking app that I'll be happy to share details with you all. We had to make it from scratch really because it didn't exist and there is no code that could even be stitched together to make it work.

 

[forestglip]

FUNCAP seems like a pretty good way to measure. (S4ME thread about it)

https://twitter.com/user/status/1729488876660457662


Very cool about the Activpal! Too bad it looks like it's not easy to buy for personal use.

Also we've been developing a completely novel symptom tracking app that I'll be happy to share details with you all. I say novel because we had to make it because it doesn't exist and there is no code that could even be stitched together to make it work.

Which features that you think would be useful are missing in all the available symptom tracking apps, if you can share that?

 

[MelbME]

FUNCAP seems like a pretty good way to measure. (S4ME thread about it)

https://twitter.com/user/status/1729488876660457662


Very cool about the Activpal! Too bad it looks like it's not easy to buy for personal use.



Which features that you think would be useful are missing in all the available symptom tracking apps, if you can share that?

It's not the features, it's the way they are tracked. We have the problem of reliability of a symptom experience being accurate based on a scale of 1-10 or existing scales. So that's the part we altered. We made a symptom tracker where each individual creates their own scale.

When you first enter the symptom tracker you are asked to list the symptoms you have.
Each symptom you have gets a slider to grade the experience of the symptom right now but there are no points along the slider except for a beginning (not experiencing) and an end (most extreme possible experience)
You put the first point on the slider wherever you feel is reflective of your current state and then you are prompted to type a short but colorful description.
Next time you enter a symptom update you will see your previous description at the spot you put it last time. If you feel that same way again then you simply tap on it. If you feel differently then you make a new point and description.
Eventually over time you will have created a scale that has points along the scale that resonate with you specifically meaning that you can accurately and quickly point to descriptions to track the severity of these symptoms.
This is the sort of platform that would enable you to pinpoint an experience months apart such that if we wanted multiple biofluid samples of a certain symptom experience then it's possible to capture them over several months or years from the same individual.
The data produced on the backend comes with a number (position of point along the slider), the date and time of the data entry, and the description.
We also built this because we believe individual symptom experience is far more complicated than the word it is often simplified to. And we think there is important information there.

 

[forestglip]

You put the first point on the slider wherever you feel is reflective of your current state and then you are prompted to type a short but colorful description.
Next time you enter a symptom update you will see your previous description at the spot you put it last time. If you feel that same way again then you simply tap on it. If you feel differently then you make a new point and description.
Eventually over time you will have created a scale that has points along the scale that resonate with you specifically meaning that you can accurately and quickly point to descriptions to track the severity of these symptoms.

That's a really good idea. I've always hated trying to track my symptoms with just a 1-10 because two months later, I'll almost certainly have no idea what exactly a "7" meant back then, and there will be drift. Using descriptions would really help.

I might recommend that patients put as much detail as they can manage for the descriptions, maybe with the ability to add more to the description later for that number.

For example, with anxiety, having both "I felt very wired at the beach but could still relax a little bit" and "I can't stop thinking about a job interview, but it's not sending me into any panic attacks" as descriptions for level 5 would be good if only one of those scenarios is relatable in the future.

I'm wondering if it'd be good to be able to rename symptoms. E.g. it starts as "anxiety", but later they notice that their feelings of being "wired" are improving, but their "dwelling thoughts" are not. Then you could split anxiety into two symptoms to track separately from then on.

 

[MelbME]

That's a really good idea. I've always hated trying to track my symptoms with just a 1-10 because two months later, I'll almost certainly have no idea what exactly a "7" meant back then, and there will be drift. Using descriptions would really help.

I might recommend that patients put as much detail as they can manage for the descriptions, maybe with the ability to add more to the description later for that number.

For example, with anxiety, having both "I felt very wired at the beach but could still relax a little bit" and "I can't stop thinking about a job interview, but it's not sending me into any panic attacks" as descriptions for level 5 would be good if only one of those scenarios is relatable in the future.

I'm wondering if it'd be good to be able to rename symptoms. E.g. it starts as "anxiety", but later they notice that their feelings of being "wired" are improving, but their "dwelling thoughts" are not. Then you could split anxiety into two symptoms to track separately from then on.

Very good points you raise. The slider isn't set 1-10, it is more like 1-100, so you could describe something at 55 and describe the other at 57 if you wanted.
All 3 options of anxiety, wired, dwellingthoughts would be options to use at the discretion of the person tracking. You could start with on and then break it up. You can or remove symptoms whenever you like.

Initially we have it set up that you can really use it how you wish, we would encourage an objective description, in your example this could be "I feel nervous but not the likelihood of panic attack is low" or "I feel like I need an environment change to shift my focus and relax"
I understand that objective descriptions can be hard when experiencing symptoms like anxiety.
We do offer a separate spot for more subjective info, like "a job interview is creating nervousness"

 

[Trish]

FUNCAP looks the best to use right now because it captures the patients prediction of their functional capacity. Combine with function al information like from wearables seems like the best at the moment.

I'm liking very much your development of ways to track changes and potential for use as outcome measures. I like the mix of the three aspects, FUNCAP, wearable activity tracker and personalised symptom tracking.

Is your app for symptom tracking available for others to use?

On FUNCAP, I don't think 'prediction of functional capacity' is an accurate description, as it seems to imply predicting into the future, though I think I get what you mean.

I prefer to describe is as a 'realistic assessment of current functional capacity based on recent experience'. So that means we are answering based on experience of performing each activity now and the past few weeks, but also assessing those things we haven't actually done on the basis of experience with similarly or less strenous exertions.

 

[Trish]

@MelbME, are you planning to use any cognitive testing as part of your outcome assessments?

 

[Kitty]

In terms of symptoms, @MelbME, it may be useful to have some reference to time of day. I'm not sure quite what (can't think clearly at the moment), but possibly a way to indicate whether there were points in the day where symptoms were worse than at other times? [Edited to add: Or only present at all at some times?]

There's a phenomenon that seems to be common in at least a subset of pwME, where they feel at their worst in the two to three hours after they wake up. It may not mean anything, but on the other hand it might—and there could be other phenomena or trends hiding in there that we haven't yet picked up.

 

[MelbME]

I'm liking very much your development of ways to track changes and potential for use as outcome measures. I like the mix of the three aspects, FUNCAP, wearable activity tracker and personalised symptom tracking.

Is your app for symptom tracking available for others to use?

On FUNCAP, I don't think 'prediction of functional capacity' is an accurate description, as it seems to imply predicting into the future, though I think I get what you mean.

I prefer to describe is as a 'realistic assessment of current functional capacity based on recent experience'. So that means we are answering based on experience of performing each activity now and the past few weeks, but also assessing those things we haven't actually done on the basis of experience with similarly or less strenous exertions.

Yes it will be free for others to use once ready for release. We will beta test with patients very soon.

Because of it's novelty we are actually making it fit to track any disease and putting in a research portal. We think this is a good opportunity to link in research from other disorders. I'd really like to see how people from many different diseases define their fatigue.

Sorry I meant perception of functional capacity, I will edit that. Had prediction on my brain I think but it wasn't far off I guess. FUNCAP isn't asking people what happened when they did tasks, it's asking what they think would happen if they attempted those tasks. Which I think is very important because perception matters.

I think recent experience might be variable for different people but that's certainly part of how people estimate the consequences of the activities listed in the FUNCAP.

Another positive of the FUNCAP is the 7 options is a nice split for a clinical trial outcome of ~15% improvement, since each step up in option is a ~15% change (100/7).

 

[MelbME]

@MelbME, are you planning to use any cognitive testing as part of your outcome assessments?

Yeah Braincheck but any tests that assess executive function or processing speed are fine.

 

[Trish]

Thanks for the replies.
Your work on all this looks important. Are you planning to publish anything soon about your plans for a package of tracking and outcome measures for ME/CFS? This work looks really useful, especially as you will be testing it on patients along with biological testing.

 

[MelbME]

In terms of symptoms, @MelbME, it may be useful to have some reference to time of day. I'm not sure quite what (can't think clearly at the moment), but possibly a way to indicate whether there were points in the day where symptoms were worse than at other times? [Edited to add: Or only present at all at some times?]

There's a phenomenon that seems to be common in at least a subset of pwME, where they feel at their worst in the two to three hours after they wake up. It may not mean anything, but on the other hand it might—and there could be other phenomena or trends hiding in there that we haven't yet picked up.

Yeah you can track symptoms as many times s you like over the day. I was just using an example in my description of day to day. But certainly it would be great for people to track over day. I find this area fascinating.

We are interested in developing a morning questionnaire that is predictive of the day. We want to do this because morning fasted samples are a great baseline and it would be useful if the participant in a study could tell if the rest of the day was going to be relatively bad or good before they ate breakfast.

So it's interesting that you say that some patients are especially bad in the morning but improve over the day, they would obviously struggle with a morning day predictor.

Do you think you can predict the day based on how you feel when you wake?