The diagnosis before the cure: what an ME/CFS breakthrough actually requires

Consider two facts in sequence.

First: myalgic encephalomyelitis/chronic fatigue syndrome — ME/CFS — affects an estimated 65 to 71 million people worldwide. That is more than multiple sclerosis, more than Parkinson's disease, more than HIV.^[1] Post-COVID, some estimates place the number significantly higher, as a meaningful fraction of Long COVID cases meet ME/CFS diagnostic criteria. A quarter of patients are severely ill: bedbound or housebound, unable to work, often unable to tolerate light or sound. The disease typically strikes in the prime of adult life, and most patients do not recover.

Second: there are no FDA-approved treatments for ME/CFS. No validated diagnostic biomarker. No animal model. For most of the past forty years, the disease received less annual NIH funding than hay fever.^[3]

The gap between those two facts is not a scientific mystery. It is a systems failure.

If you've read the earlier posts in this series — about CAR-T therapy resetting autoimmunity in Erlangen, and a personalized CRISPR therapy built in six months for one child in Philadelphia — you already have most of the tools needed to understand both why ME/CFS is stuck, and what getting it unstuck would actually require. This is, in a real sense, the framework from the previous post applied to a single, very large, very neglected disease.

What the Syndrome Is

ME/CFS is currently diagnosed by symptom criteria, not by any objective biological test: post-exertional malaise — a severe worsening of symptoms after physical or mental exertion that can last days — unrefreshing sleep, cognitive impairment, and orthostatic intolerance, persisting for six months or more and not explained by another condition.^[1] There is no blood test, no imaging finding, no electrophysiological signature that confirms the diagnosis. A patient either fits the symptom pattern or they don't; nothing in a lab report settles the question either way.

The exhaustion at the center of this disease is not ordinary tiredness. It is a profound, whole-body fatigue that makes routine daily tasks — grocery shopping, household chores, holding down a job — genuinely difficult or impossible, not because the patient lacks willpower, but because the underlying biology limits how much exertion the body can tolerate before symptoms worsen. That creates a cruel feedback loop in the path to diagnosis itself: getting properly diagnosed requires appointments, travel, paperwork, and follow-up visits, and by the time a patient with severe ME/CFS is finally able to see a specialist, they may already be exhausted from the act of getting there. Telemedicine could meaningfully reduce this barrier — a video visit removes the physical toll of travel and waiting rooms that can set a patient back for days, and it deserves a more central role in how ME/CFS care is delivered, not as a lesser substitute for in-person care but as a genuinely better fit for a disease defined by exertion intolerance.

There is another layer to this that is easy to miss and that compounds the public health problem described below: most people with ME/CFS do not look sick. They are not visibly approaching death the way a late-stage cancer patient might be. There is no wheelchair, no IV pole, no obvious external marker of how disabled they actually are. The damage this disease does is largely invisible from the outside — it shows up as a collapsed quality of life, an inability to work, a withdrawal from relationships and activities that used to be normal, rather than as something a doctor, an employer, or a disability reviewer can see at a glance. That invisibility is itself a barrier: to taking patients seriously, to approving disability claims, to treating the disease with anything close to the institutional urgency that a more visibly devastating illness would receive.

That diagnostic gap is also, immediately, a public health gap. A disease with no objective test is a disease that is undercounted, underbilled, and underrepresented in the health statistics that drive funding decisions. Sixty-five million people is already a large number — and it is very likely an undercount, since an unknown fraction of people who meet the clinical picture are never formally diagnosed at all, particularly outside of specialty clinics that most patients never reach.

The Problems That Go Along With It

The bottlenecks in ME/CFS research are well documented individually. What's less often stated plainly is how they interact — how each one reinforces the others to produce a field that has been generating hypotheses for forty years without converting them into treatments.

The Biomarker Problem Is Primary

Every other bottleneck flows from this one. Without a biomarker, clinical trials cannot reliably enroll a biologically homogeneous population. The fifteen patients in the Erlangen CAR-T trial discussed earlier in this series all had SLE confirmed by ACR/EULAR criteria with documented autoantibody profiles — the trial could be designed around a biologically coherent group from the start. A ME/CFS trial enrolling on symptom criteria alone almost certainly captures multiple biologically distinct subtypes with different underlying mechanisms bundled together. When such a trial produces a null result — as the Norwegian rituximab Phase III trial did in 2019 — it is impossible to know whether the therapy failed because the mechanism was wrong for everyone, or because it was right for some subtype that got diluted into statistical noise by the heterogeneous enrollment.^[2]

Without a biomarker, pharmaceutical companies also cannot perform the ordinary go/no-go decisions that structure drug development. There is no early signal to chase, and no patient stratification that would let a company design a trial it could actually win. This is where the public health consequence becomes concrete: an entire category of private investment simply never enters the field, because the basic infrastructure that would make that investment legible doesn't exist yet.

The Heterogeneity Problem Amplifies the Biomarker Problem

ME/CFS is almost certainly not one disease. The evidence increasingly points to several biologically distinct subtypes that converge on the same symptomatic presentation through entirely different upstream mechanisms. Some patients have documented autoantibodies against G protein-coupled receptors — specifically beta-adrenergic and muscarinic acetylcholine receptors — consistent with an autoimmune origin. Some show evidence of viral persistence, with elevated titers against herpesviruses or detectable SARS-CoV-2 components in the blood years after the original infection. Some show primarily mitochondrial dysfunction. Some show primarily neuroinflammation. Many patients show some combination of all of the above, in proportions that vary from person to person.^[1]

These are not minor variations on a single accepted explanation — they are genuinely competing theories of what ME/CFS actually is, each with its own research community, its own proposed mechanism, and its own candidate treatments. The autoimmune theory points toward therapies that target or remove autoantibodies. The viral persistence theory points toward antivirals and immune modulation aimed at clearing a lingering infection. The mitochondrial dysfunction theory points toward metabolic and energy-pathway interventions. The neuroinflammation theory points toward treatments that target the central nervous system and its immune signaling. No single theory has won out, because the evidence suggests that more than one of them is correct — just not for every patient, and not in the same proportion from one patient to the next.

This is precisely what makes the disease so difficult to treat in practice, beyond the research bottleneck described above. A patient without a biomarker has no way of knowing in advance which of these mechanisms is driving their own illness, and neither does their doctor. In the absence of that information, treatment becomes a slow process of trial and error: trying one drug aimed at one theory, waiting weeks or months to see whether it helps, often having to taper off carefully before trying something else, and repeating that cycle across a different theory entirely if the first approach fails. For a patient whose baseline energy is already severely limited, each of these trial-and-error cycles carries a real cost in time, money, and the physical toll of trying and failing — and for some patients, this process can stretch across years before anything resembling an effective treatment is found, if one is found at all.

Treating all of this as a single disease and designing one trial around it is roughly equivalent to running a cancer trial that enrolls all solid tumors regardless of histology or mutation status — a design oncology abandoned decades ago in favor of molecularly defined patient selection.^[2]

The Funding and Visibility Problems Compound Everything Else

This is where the public health failure becomes hardest to separate from the scientific one. The NIH invested roughly five to seven million dollars per year in ME/CFS research through most of the 2000s and 2010s — a figure that constrained not just the number of studies that could be run, but the ambition of any individual study.^[3] Deep phenotyping work of the kind Avindra Nath's team conducted at the NIH — the 2024 Nature Communications paper that enrolled seventeen patients for four weeks of intensive inpatient characterization — requires substantial infrastructure and sustained institutional commitment.^[1] That single study took years to design, fund, and execute. At five million dollars a year in total field funding, one study of that scale consumed a significant fraction of the field's entire annual budget. A disease affecting 65 million people was, for most of two decades, funded at a level that allowed for a small handful of well-designed studies per year, total, worldwide.

What the Idealized Breakthrough Path Looks Like

None of this means a breakthrough is impossible. It means the path is specific, and it differs in instructive ways from the paths that produced CAR-T and KJ's CRISPR therapy — both of which had the advantage of a single identifiable patient population and an existing platform to repurpose.

Step One: A Validated Biomarker, Not a Treatment

This sounds counterintuitive. The field has been trying to find treatments for forty years; how does prioritizing a diagnostic tool accelerate that? The answer is that without a biomarker, no treatment can be reliably tested. Every dollar spent on a clinical trial in the absence of patient stratification is a dollar spent on a study that cannot produce an interpretable result. The single highest-leverage investment in ME/CFS right now is not a drug trial — it is a large-scale, deeply phenotyped biomarker study.

The raw material for this study already exists, scattered across several labs. Avindra Nath's NIH intramural work identified consistent abnormalities in catecholamine metabolism, autonomic function, and immune activation.^[1] Ron Davis's group at Stanford has been developing a nanoneedle-based diagnostic assay that can distinguish ME/CFS patient samples from healthy controls with high accuracy in preliminary data. Carmen Scheibenbogen's laboratory at Charité Berlin has validated a specific autoantibody profile in a subset of patients. The open question is whether these signals — each developed in a small, well-characterized cohort — can be validated in a large, multi-site, geographically and demographically diverse patient population. A validation study run at the scale of the NIH RECOVER initiative's Long COVID cohorts, which have enrolled hundreds of thousands of patients, would transform the field's capacity to do everything that follows.^[3]

Step Two: Subtyping, Not Treating

Once a validated biomarker — or, more likely, a panel of biomarkers corresponding to distinct biological subtypes — exists, the field can do what oncology already did with cancer: define molecularly distinct patient groups and design a separate trial for each one. The autoantibody-positive subgroup is probably the most tractable near-term target. BC007, an aptamer that neutralizes G protein-coupled receptor autoantibodies, produced statistically significant improvement in a Phase II trial, and larger randomized controlled trials are now advancing.^[4] Immunoadsorption — physically removing autoantibodies from the blood, as Scheibenbogen has done at Charité — works within this subtype specifically. Applied to an unselected ME/CFS population, these approaches produce noise. Applied to the autoantibody-positive subtype alone, they may produce a real signal.

For the viral persistence subtype — patients with evidence of ongoing SARS-CoV-2 or herpesvirus replication — the therapeutic question is different: targeted antivirals, potentially combined with immune modulation to clear chronically infected reservoirs. The RECOVER Long COVID trials are already generating data on exactly this approach, and ME/CFS research can and should leverage that existing investment rather than duplicating it from scratch.

Step Three: A Regulatory Pathway Designed for the Problem

The FDA's expanded drug development toolkit — Breakthrough Therapy Designation, accelerated approval based on biomarker endpoints, adaptive trial designs — exists precisely for diseases with unmet need and evolving scientific understanding. ME/CFS qualifies on every criterion. What it has lacked is the validated surrogate endpoint — the biomarker change that reliably predicts clinical improvement — that would allow accelerated approval to be granted on something other than years-long functional outcome measures. Step one, the biomarker validation work, is what unlocks step three. The regulatory tools already exist; they simply have nothing to attach to yet.

What This Requires That the Current System Doesn't Provide

The path described above is coherent. It is also not happening at the necessary scale or speed, for reasons that are structural rather than scientific — which is exactly the public health dimension of this problem.

The NIH RECOVER initiative has brought unprecedented funding to Long COVID research, and by extension to ME/CFS research. But RECOVER's mandate is broad, and its enrollment criteria include many patients who would not meet formal ME/CFS diagnostic criteria. The ME/CFS-specific research infrastructure — the specialized clinical centers, the patient registries, the biobanks with deeply phenotyped samples — remains badly undersized relative to a disease affecting 65 million people.^[3]

Pharmaceutical industry engagement is correspondingly limited. CAR-T therapy attracted commercial investment because the oncology proof of concept was already established and the addressable market was clear from day one. KJ's CRISPR therapy attracted in-kind support from industry partners because the underlying platform components — base editing, LNP delivery — were already developed, and the reputational upside of participating was high. ME/CFS offers neither an established clinical proof of concept nor a clearly segmented patient population that industry can target yet. The commercial pathway only becomes visible after the biomarker and subtyping work is complete, which means the early-stage investment has to come from public and philanthropic sources almost entirely — there is no private capital waiting in the wings to take over once the science gets interesting, because the science can't get interesting at scale until that early investment happens first.

This is, in the end, the same argument that runs through every disease that hasn't yet had its breakthrough. The scientific questions are real and difficult. But the primary constraint is not the biology. It is the decision to fund, or not fund, the infrastructure that makes biological progress possible in the first place: the biobanks, the deep phenotyping studies, the data-sharing agreements, the specialized clinical networks that can enroll patients at the scale needed to produce an interpretable result.

ME/CFS has sixty-five million patients. It has Avindra Nath, Ron Davis, Carmen Scheibenbogen, and a growing international research community that has spent decades building toward the moment when the tools of modern immunology and genomics are finally applied to this problem at the necessary scale. What it needs is a decision — institutional, political, financial — that this problem is worth solving on a timeline commensurate with its size.

That decision has not yet been made. The science is ready for it.

A Dream: The Pittsburgh Model

What follows is not a policy proposal. It is an attempt to think seriously about what a breakthrough research effort would look like if you built it from scratch, with the full framework from the previous post in mind — not just the science, but the human conditions, the community, the funding structure, and the cultural environment that the science grows inside of. I am placing it in Pittsburgh because Pittsburgh is what I know, and because the Hill District specifically offers something that most research neighborhoods don't: a community that genuinely needs what this kind of institution could offer, and that could offer something back in return.

The Living Structure: A Shipping Container Community in the Hill District

The team lives together — not in a dormitory, not in separate apartments scattered across the city, but in a purpose-built community designed to reduce the friction of daily life while building the kind of social density that research environments need. The structure is made from shipping containers, stacked and arranged into a multilevel hexagonal form that houses roughly twenty individual adults, each with a two-bedroom, two-bathroom unit that includes a small kitchen, dining area, and living room. The hexagonal layout encloses a shared interior space — an outdoor grill, a small playground, a food garden, a fire pit, and a projector screen for warm months. For winter, two containers are configured as a large common room with couches and screens large enough to actually host a group gathering. Two additional containers are reserved for exercise equipment. Three containers serve as dedicated work and study space, separate from the living quarters — a place to think that isn't a bedroom.

The structure is fully accessible: two elevator points at each level, glass walkways beginning at the third story that connect the levels and allow natural light into the interior. The design is done by Pitt engineering and architecture students as part of their capstone projects — not as charity toward the research team, but as a genuine design challenge that produces a real building in the real world, in a neighborhood that needs investment. The students who design it learn something that no studio project can teach. The researchers who live in it get a home that was built with intention.

The Research Team

The team is built around a set of principles that run against several default assumptions about how academic research groups are assembled.

It is fifty percent women and fifty percent men. It draws from a genuine range of socioeconomic backgrounds — lower income through upper middle class — because the diversity of lived experience in a research team affects the questions the team thinks to ask, and because a team that looks like a cross-section of the country is better positioned to build trust with patients who also look like a cross-section of the country. Ages range from 24 to 55, deliberately spanning the gap between early-career researchers who are closer to the methodological frontier and senior researchers who carry the institutional and relational knowledge that takes decades to accumulate.

The scientific composition is deliberately multidisciplinary: biologists, immunologists, computational biologists, chemists, physicists, mathematicians, and engineers all on the same team, because ME/CFS is the kind of problem that has resisted single-discipline approaches for forty years and may only yield to a genuinely cross-domain effort. Each researcher is also required to see patients once a month — not as a clinical obligation but as an orientation device, a way of keeping the lab grounded in the reality of what the disease actually does to a person's life rather than allowing the research to drift into abstraction.

Two roles in the team are explicitly structural rather than purely scientific. One person serves as the political face of the team: they attend conferences, go to University of Pittsburgh events, represent the group at mixers and public functions, build the external relationships that make collaboration and funding possible. One person is dedicated almost entirely to grant writing and funding strategy — translating the science into the language that funding bodies respond to, tracking deadlines, and pursuing every available source. Both people maintain small active roles in the lab itself, because credibility in both directions requires it — the grants person needs to be able to speak to the work from the inside, and the public face needs to know the science well enough to answer real questions. Other team members can join events and conferences when it makes sense, but the consistent external presence of one person makes networking easier for everyone else, because there is always a familiar face that the broader community knows how to find.

The Research Approach: The Whole Human

The scientific strategy is built around a principle that has guided this entire post: ME/CFS is a multi-system disease, and the only research approach with a real chance of producing a breakthrough is one that monitors the whole human rather than isolating one variable at a time in the hope of finding the one lever that explains everything.

The team works with approximately fifty patients, monitored continuously through a combination of wearable technology and regular clinical contact. Each patient wears two devices. The first is a pair of glasses with embedded cameras trained to recognize food — specifically, the camera begins saving data only when it can identify a plate, fork, or knife in the visual field, so that patients are not being recorded continuously and privacy is protected. The only data retained from the camera feed is meal composition: the food items visible, estimated calories, and macronutrient breakdown. Patients can supplement this with voice memos recorded through the second wearable, a watch, adding context about how they felt before or after eating, medications they took, or anything else they want to log. The watch itself tracks heart rate, blood oxygen, and any additional biometrics that become available as wearable sensor technology advances — blood glucose monitoring is an obvious near-term candidate, and white blood cell tracking, if it ever becomes feasible at the wrist, would be transformative for this particular patient population.

From this continuous stream of data — meals, sleep, activity, physiological signals, medication timing, and patient-reported experience — the team tries to identify the patterns that precede crashes, the dietary factors that correlate with symptom changes, and the treatment responses that vary by patient in ways that might map onto the underlying subtype distinctions described earlier in this post.

The four core scientific focuses are gut microbiome health, brain health and neuroinflammation, mitochondrial health and activity, and viral load monitoring across the body. These are not separate parallel projects — they are four lenses on the same fifty patients, generating data that the computational biologists and mathematicians on the team can look for cross-system patterns within. The goal is not to prove that one of the existing theories of ME/CFS is right and the others are wrong. It is to find the combinations of factors — the multi-system signatures — that characterize each patient's illness, and from those signatures, work backward toward the interventions most likely to help.

The Lab and the Community

The lab itself is built in the Hill District, and that choice is not incidental. The Hill District is a historically Black neighborhood in Pittsburgh that has absorbed the negative side of university expansion for decades without receiving proportionate benefit. Building a research center there, on those terms, would normally reproduce that pattern — bringing educated outsiders into a community, raising costs, and leaving when the grant runs out.

The design here is different. The lab building is itself a Pitt capstone project: engineers and architects design it as functional research space, with high ceilings capable of housing large computational equipment and industrial freezers, easily cleanable surfaces throughout, and flexible interior configurations that can be reconfigured as the science evolves. The electrical infrastructure is built to handle the power demands of serious laboratory equipment from day one. Outside, a large garden produces fresh food for the team and surplus for the neighborhood — not as a symbolic gesture but as a genuine local food resource in a community that is frequently underserved by fresh produce access.

The funding comes from everywhere it can be found: Pitt, the Commonwealth of Pennsylvania, NIH, and — because this is Pittsburgh — the Steelers, the Pirates, and the Penguins, whose charitable foundations would support youth programs that give Hill District high school students meaningful roles in the lab. Not as mascots, not as tour visitors, but as people who are doing real work, winning awards that translate into college scholarships, and building the kind of documented scientific experience that opens doors. When breakthrough researchers live in the Hill District, eat at neighborhood restaurants, and have high school kids from the neighborhood over for dinner and a game, those kids can see a path that wasn't visible before — not because someone told them about it, but because it is standing in their neighborhood and the people doing it look like them.

When the research eventually produces commercially valuable discoveries — and the framework suggests it will, if the investment is made and sustained — the returns are structured to feed back into the community rather than out of it: scholarships, training pipelines, and the institutional capacity to bring the next generation of researchers up from inside the neighborhood that made the work possible.

This is what gentrification looks like when it goes right instead of wrong. The university doesn't arrive as a colonizing force that raises rents and displaces people. It arrives as an institution that has decided its success and the community's success are the same thing — and then builds the structures that make that true. If it works in the Hill District, it is a model. If it works as a model, it is a template for how research institutions can generate scientific breakthroughs and socioeconomic mobility at the same time, in the same place, from the same investment.

The science is ready. The neighborhood is ready. What it requires is an institution willing to do something harder than publish a paper — willing to plant itself in a community and mean it.

Key Sources