PrecisionLife’s Bold Attempt to Break the Code on and Beat ME/CFS and Long COVID

Geoff’s Narration

The GIST

This is the next in a series of blogs exploring ongoing projects in ME/CFS, FM and long COVID. The first featured the RECOVER project’s next set of long COVID clinical trials. 

PrecisionLife thinks they can do it all: explain these diseases, uncover subsets, produce biomarkers, enable “smart clinical trials,” and develop cheap diagnostic tests.

I gaped when I saw what PrecisionLife is attempting to do in ME/CFS and long COVID. PrecisionLife believes they can have it all. Not only is it trying to identify genetically based subsets, but it’s attempting to identify the mechanisms driving those subsets, identify biomarkers for those subsets, find drugs that move those mechanisms, and test them. It’s an all-in-one package.

Potentially, that means finding treatments that are targeted to specific biologically based subsets which, in ME/CFS, is basically what we’ve been hoping for, for quite a while.

Will they be able to pull it off? Time will tell. One thing, though, this company has been laser-focused on ME/CFS. I was astonished to learn that PrecisionLife fully funded its ME/CFS studies.

Why would any company, you might ask, go out of its way to focus on “the disease of a 1,000 names”, which has been described as a “wastebasket disease” and which is surely larded with subsets. Why would anyone take on what is arguably the most complex of the complex, chronic diseases?

Gardner acknowledged these diseases presented “a big challenge” but asserted that PrecisionLife can “deal with the complexity” and stated that complex, chronic diseases are PrecisionLife’s raison d’être – they’re its reason for being. For two, PrecisionLife isn’t simply being nice by taking on this neglected disease.

They believe their work in ME/CFS will help them out. If they can break the code on the disease others fear to touch, they can break the code on any disease. Referring to “the weird and wonderful mechanisms” at play in these diseases, it was easy to see that Gardner is simply fascinated by how a respiratory infection like COVID-19 can produce over 200 symptoms in long COVID.

Because the company’s work with pharmaceutical companies involves commercial interests, and cannot be published, they wanted a public test case. DecodeME’s publication of the first large genomic dataset on ME/CFS in 2020 gave them the data they needed. The fact that they think they’re ready enough for these diseases to devote funding for them says something in itself.

“Upon demonstrating success for ME/CFS and long Covid patients, we hope to also apply this precision medicine approach to multiple diseases of aging in respiratory, dementia, autoimmune, and metabolic diseases to benefit millions more people.” PrecisionLife

Why does PrecisionLife think it can jumpstart this disease into a new era? I talked with CEO and co-founder Steve Gardner about that.

“Respecting the Biological Complexity of Diseases” or, Genetics on Steroids…

You may have heard that 10-20% of the risk for a disease is due to genetics.  The recent DecodeME study, for instance, found that common genetic variants accounted for about 10% of the risk of developing ME/CFS.

They believe the techniques they use better reflect what’s happening in the body.

Not so fast, says Steve Gardner. That’s true if you assess the gene variants piece by piece. But what if instead of focusing on the gene variants, you analyze all the pathways those genes participate in? Once you assess the molecular pathways those genetic variants affect and how they interact, they say our genetic inheritance looms much larger.

Gardner calls this approach “respecting the biological complexity of chronic diseases”. In a 2021 opinion paper, “Combinatorial analytics: An essential tool for the delivery of precision medicine and precision agriculture“, he asserted that a gene-by-gene approach is only really effective in rare diseases caused by single-gene mutations and in cancer.

Take the APOE gene in Alzheimer’s. Even though it presents a major risk factor, clinical trials in Alzheimer’s have almost universally failed. Over 1,000 genetic association studies and 75 GWAS (genome-wide association studies) Alzheimer’s studies have identified just 29 risk genes, and even worse, have not produced any effective treatments. The lack of success led researchers to dig even deeper, seeking rarer and rarer gene variants – all to no avail. This is because, despite the clear genetic risk factors found in Alzheimer’s, at its heart, Alzheimer’s is a complex, multifactorial, polygenic disease; i.e., many genes contribute to the development of Alzheimer’s in ways that GWAS studies can’t hope to pluck out.

A similar effort to understand COVID-19 occurred early on. A global GWAS effort involving almost 14,000 patients with severe disease and over 2 million controls identified four genome-wide significant loci (genetic regions that increase risk) associated with SARS-CoV-2 infection and 11 genetic regions associated with severe COVID-19. The genetic regions identified were not surprising – most had to do with lung, autoimmune, and inflammatory diseases. The results hardly matched the range of symptoms and problems found in long COVID and provided few new treatment approaches.

Compare those results to a much, much (much) smaller (a few hundred people) COVID-19 study, which used a combinatorial approach to identify 156 severe-disease-associated loci mapping to 68 protein-coding genes (proteins do the work in cells) across a range of mechanisms. They were subsequently validated and opened the door to new treatments.

That, Gardner says, is why using something called “combinatorial analytics” is so much more effective.

Combinatorial Analytics

The FM and DecodeME GWAS studies sought “disease architectures”; i.e., genomic regions containing genetic variants that appear to increase the risk of these diseases.

The disease architectures PrecisionLife is looking for exist on a whole other scale. It’s like they’re playing 3-dimensional chess. PrecisionLife is looking for gene combinations whose pathways correlate strongly with ME/CFS and/or long COVID. The disease signatures get grouped into disease “communities” (patients sharing overlapping combinations).

One goal – using combinatorial analytics to uncover the hidden molecular pathways driving symptoms.

The 15 communities identified by their combinatorial analysis make perfect sense given ME/CFS research results. They include communities like neural-autonomic / urea cycle / stress signaling / mitochondrial-metabolic / vascular-endothelial/immune-inflammatory / metabolic-stress response / vascular-endothelial.

The similarities are intriguing, given that these communities are based solely on the genetic variants and pathways found in ME/CFS. Despite PrecisionLife not using any research findings to develop these communities, their findings essentially mirrored what the research shows.

In fact, Steven Gardner said they didn’t know much about ME/CFS when they did their first study. When he went to DecodeME to show them their findings, Chris Ponting promptly accused them of confirmation bias.

If PrecisionLife is right, then the genes we are born with set us up in a major way for ME/CFS.

Next Step – Identify the Mechanisms

Next, in its mechanostics platform, PrecisionLife seeks to find the mechanisms underlying these disease signatures. It’s as if the engine in your car seized. You can see that the engine will not crank, but you don’t know why. You have to get into the guts of the engine to understand what happened.

Next – identify the mechanisms driving those molecular pathways.

PrecisionLife’s mechanostic approach is part of an emerging paradigm of diagnosing diseases by mechanism rather than “manifestation“, such as symptoms or even clinical findings. Ultimately, its work should yield new disease labels that reflect the mechanisms underlying ME/CFS, FM, long COVID, and related diseases.

Because a process called pleiotropy can cause the same gene variant to produce very different outcomes, a mechanistic view of disease could redefine disease boundaries.

In our talk, Gardner referred to the “Victorian or Edwardian, clinical observations” that have produced many of the labels we currently use to refer to diseases.

Combinatorial analytics has found, for instance, that multiple pathways can produce wheezing in asthma. One molecular subset with high T-helper cell type 2 (eosinophilic/T2) expression has drugs that can help. Another molecular subset without high T-helper cell type 2 does not.

The genes associated with the second group are entirely different and are even involved in non-immune pathways (!), including fatty acid synthesis, LDL oxidation, and modulators of GABA, purinergic, and glutamate signaling, suggesting a different type of drug should help them.

Pleiotropic Diseases

The high levels of pleiotropy found in the recent fibromyalgia GWAS study suggest that diseases like FM, ME/CFS and long COVID can benefit from combinatorial analysis. (Note that many of the loci found in FM also showed up in the ME/CFS GWAS study.)

We are all familiar with the extraordinary range of responses to treatments found in these diseases. Getting at the mechanistic pathways underlying these responses and the symptoms of these diseases is particularly important for diseases with vague, difficult-to-quantify symptoms like fatigue or post-exertional malaise.

Check out some of the provisional mechanistic clusters PrecisionLife has identified in ME/CFS and long COVID. (ChatGPT 5.0 derived – names may be different)

• Viral/Bacterial Susceptibility: key genes S100PBP; CDON; USP6NL – pathways involved; entry/adhesion and innate immune signaling.
• Metabolic Dysfunction: key gene AKAP1 – pathways involved – mitochondrial scaffolding, respiration,  post-exertional recovery.
• Autoimmune- neuro-immune signaling: Key genes GPC5; PHACTR2 + more; pathways involved – immune-neuro.

Gardner reported that while eight mechanisms may be in play in an ME/CFS or long-COVID patient, only one or two mechanisms may be the principal drivers of their symptoms. For instance, in ME/CFS, mitochondrial genes were highlighted in people with severe post-exertional malaise. People displaying genetic issues with neurotransmitters tended to experience more severe brain fog.

In long COVID, PrecisionLife uncovered critical genetic variants in two major cohorts: 86 in the “severe cohort” and 84 in the “fatigue dominant” cohort. The fact that at least 74 unique gene variants were found in each cohort again suggested that PrecisionLife had discovered distinct genetically based subsets.

Interestingly, though. PrecisionLife found that, while the variants differed, similar genes were also highlighted in both cohorts. Twenty-eight of 43 genes highlighted in the severe cohort were also associated with the fatigue-dominant cohort. Twenty-five of the 35 genes highlighted in the fatigue-dominant cohort were also found in the severe cohort.

One or two mechanisms may be the main drivers of symptoms in most ME/CFS and long-COVID patients.

As in ME/CFS, genes uniquely associated with the severe long-COVID cohort aligned remarkably well with what we know about it. They impacted fat metabolism, autophagy (mitochondria), insulin resistance/metabolic syndrome, viral resistance, and the innate immune response.

So too were the genes that were uniquely associated with the fatigue-dominant cohort. They affected acetyl-CoA signaling (mitochondria), muscle function and lipid metabolism, NADH dehydrogenase (mitochondria), oxidative stress, mitochondrial muscle activity, and monocytes.

While there were differences between the cohorts, it was also interesting to see similar general themes (innate immune response, energy production, lipid metabolism) driven by distinct genes or gene variants in each cohort.

PrecisionLife also found that many of the long-COVID gene variants appeared in their ME/CFS analysis. The two diseases are close enough that Gardner believes “ME is a sort of endpoint for something that starts through long COVID”.

Super-Responders and Smart Clinical Trials

PrecisionLife’s mechanostics program then asks two questions: “What exact mechanism (molecular pathway) is broken, and can they measure it?” If they can assess the “brokenness” of a molecular pathway, they can biologically assess how effective treatments are – exactly what the drug companies are waiting for. If drugs exist that specifically target those broken pathways (and clinical trials can be funded), the game is on.

A major goal – identify the “super-responders” who are most likely to benefit – and get them into targeted clinical trials.

The ability to now target precise subsets of patients – PrecisionLife calls them “super-responders” – who are likely to do well should result in “smart clinical trials” and better results than current clinical trials which simply throw everyone with ME/CFS into the mix.

It’s the throw-everyone-into-the-mix approach in this very heterogeneous disease that has made pharma take a hands-off approach. Gardner’s goal is to give pharma the confidence it needs to enter the ME/CFS and long-COVID space. As a bonus, if those trials succeed, a low-cost genotypic test to identify patients would be the next step.

PrecisionLife has identified nine generic drug candidates it wants to test in ME/CFS and/or long COVID and has honed in, in particular, on TLR4-targeting drugs. (Low-dose naltrexone is a TLR4-targeted drug.)

The Future

On the plus side, PrecisionLife has been able to generate its findings quickly and cost-effectively. Its focus on repurposed drugs means that, if funded, clinical trials can occur quickly. Its technology appears robust; it’s been doing this for 10 years and is working across more than 60 disease groups. It presents the potential of cheap diagnostic tests.

Because virtually everyone recognizes that precision medicine – the ability to link a treatment to a specific biological state – is the future, if PrecisionLife can deliver, its future should be bright. Gardner stated that a new, greatly expanded ME/CFS paper will be out soon.

I asked ChatGPT 5.0 what could derail this effort. Not enough good data could. Because data is the lifeblood of PrecisionLife’s work, the more high-quality data PrecisionLife has, the better it should do.

PrecisionLife has analyzed more high-quality data. Will it be able to uncover the disease signatures at play in ME/CFS, long COVID, and fibromyalgia?

Replicated or validated datasets provide very high-quality data, and DecodeME appears to be the only major dataset currently available that meets that criterion in ME/CFS. PrecisionLife obtained its first results without using DecodeME (and replicated most of its results in the UK DecodeME and the US “All of Us” cohorts). Ian Lipkin’s current NIH-funded US GWAS study should provide more high-quality data.

PrecisionLife is now analyzing the full spread of DecodeME data available to it (@ 11,500 samples), and not surprisingly, is finding many more genes associated with ME/CFS. Gardner said that because the new analyses validate past findings and open up new ones,

Gardner reported in the interview that PrecisionLife has “high confidence” that it’ll be able to find targets that are “going to be beneficial for patients”. The rubber meets the road for the company in the clinical trials, where its findings are put to the test. It’s in the clinical trials that we’ll learn if PrecisionLife can accurately identify biological subsets and the mechanisms driving them, and find treatments that help a particular group of patients.

Citing work by Action for ME, DecodeME, the Complex Disorders Alliance, and ChronicleBio (an R&D partner), as well as the steadily growing patient databases, Gardner said he was encouraged by the ecosystem emerging around ME/CFS.

Gardner noted, though, the “very big computational challenges” PrecisionLife’s work faces. False positives can be a problem in diseases with generally weak genetic signals like ME/CFS. PrecisionLife has a way to guard against these, but ChatGPT reported that the best bulwark against them is larger, validated, genetic studies such as DecodeME and Lipkin’s study.

Another question is how well the field will be able to test PrecisionLife’s findings and whether it can withstand less-than-stellar early results. PrecisionLife can do the analytics, but they can’t put on the clinical trials needed to validate them. (One is underway – see below).

We ended the interview with Gardner saying that while much remains to be done, “I do feel confident saying that we understand the disease much better now and will do, particularly with the next set of studies that come out, than we ever have in the past.”

Metrodora / Complex Disease Alliance ME/CFS Studies

“Our partnership with Metrodora has the potential to transform the lives of tens of millions of patients affected by these and many other debilitating diseases.” PrecisionLife, Sept, 2024

The Mello study is dramatically expanding the datasets used.

In September 2024, PrecisionLife shared insights gathered from a $1 million grant from the Metrodora Foundation, which aimed to recruit up to 1,000 ME/CFS and long-COVID patients.

Metrodora appears to have dropped its clinical arm and has morphed into a research institute since then. Metrodora is now the Complex Disorders Alliance (“CODA”) and continues to work with PrecisionLife. The new website states it has:

• collaborated with PrecisionLife to confirm key genetic drivers of Long COVID in a diverse U.S. cohort.
• funded the MELO Study, in collaboration with PrecisionLife, to uncover the genetic drivers of ME/CFS and Long COVID, using deep phenotyping and multi-omic analysis to identify subtypes and treatment targets. The MELO study is particularly interesting because it combines PrecisionLife’s genetic mechanostic factors with multi-omic (genomic, transcriptomic, proteomic, and metabolic profiling) and immune factors (immune cell types, cytokines, markers of inflammation, and blood work). Gardner said 400 people have been recruited in the Mello study, and a genetic swab test is being developed.

Support Health Rising and Keep the Information Flowing!

Health Rising is not a 501 c (3) non-profit