GRAIL's NHS-Galleri Trial Delivers a 27% Stage IV Reduction by Year Three — and Reveals a Hidden Epidemic of Undiagnosed Late-Stage Cancer
GRAIL ASCO 2026 Analyst Call, May 31, 2026 — Results from the 142,000-patient NHS-Galleri RCT and 35,000-patient PATHFINDER 2 study presented
The Headline Number That Matters Most
Lost beneath the February headlines about a missed primary endpoint is the data point investors and clinicians should be focused on: in the incident screening rounds of the NHS-Galleri randomized controlled trial — the rounds that most closely approximate a real-world steady-state screening program — Stage IV cancer diagnoses fell by more than 22% in year two and nearly 27% in year three. These reductions were nominally statistically significant and grew with each successive screening round. That trajectory, combined with a 16% increase in Stage I and II diagnoses and a greater than 25% reduction in emergency-presenting cancers, forms the core of GRAIL's clinical utility argument heading into FDA review.
GRAIL's Chief Scientific Officer Harpal Kumar was direct on the call: "While this slide shows the results for the 12 prespecified cancers, we saw reductions of greater than 20% in the incident rounds for all cancers, which is a remarkable decrease at a population level."
Why the Primary Endpoint Miss Is the Wrong Lens
The NHS-Galleri trial, which enrolled more than 142,000 participants aged 50 to 77 across England in a one-to-one randomized design, failed to meet its prespecified primary endpoint of a reduction in combined Stage III and IV cancer incidence across three annual screening rounds with 12 months of follow-up after the last blood draw. GRAIL disclosed this in February, and the negative media coverage that followed materially affected physician prescribing behavior and patient inquiries.
What the company is now explaining — with granular round-by-round data — is that the trial's structure created a mathematical headwind in the first, or prevalent, round that was not anticipated when the trial was designed approximately eight years ago. In the prevalent round, the Galleri test swept up a large bolus of already-existing but undiagnosed cancers in the intervention arm, including a disproportionate share of Stage III disease. This produced a 19% increase in Stage III and IV cancers in that first round, an increase large enough that the reductions seen in rounds two and three were insufficient to overcome it on a cumulative basis.
Kumar explained the mechanism plainly: "We believe the Stage III increase was driven in part by a number of Stage IV cancers being shifted to earlier stages, including at Stage III, and the fact that many more cancers overall were found earlier through screening in the intervention arm, while the equivalent cancers may not yet have been diagnosed in the control arm." In other words, the test worked — it found late-stage disease that was already present but invisible to standard care — and in doing so, temporarily inflated the Stage III count in ways that penalized the trial's aggregate endpoint.
Perhaps more consequentially, the trial exposed something no one had previously quantified: the sheer volume of undiagnosed late-stage cancer silently present in the general population. As Kumar noted, "One of the things that we've learned in this study is just how much undiagnosed relatively late-stage cancer exists in the population. We had no idea that was the case because it's never been studied before." That insight alone has significant implications for how future MCED trials should be designed and how long they need to run.
The Stage IV Survival Math Is Difficult to Dismiss
Pushback from some academic oncologists — that earlier detection may not improve outcomes — met a sharp rebuttal from the GRAIL team. Kumar cited colorectal cancer as an example: a Stage IV diagnosis carries a five-year survival rate of roughly 11%, while Stage III carries 64%. In prostate cancer, Stage IV five-year survival is 53% versus 97% at Stage III. The NHS-Galleri data showed a 34% reduction in Stage IV colorectal diagnoses and a 25% reduction in Stage IV prostate diagnoses across the trial. "I challenge anyone to say that's not meaningful. I fundamentally don't accept the premise that finding that cancer at Stage III rather than Stage IV is not beneficial for the patient," Kumar said.
President and CEO-elect Josh Ofman added a treatment landscape point that deserves attention from investors modeling long-term clinical adoption: "The survival cliff across most solid tumors is now between Stage III and Stage IV. Ten to fifteen years ago, it was between Stage II and Stage III. That has dramatically changed due to the treatment landscape evolution." This is not a trivial observation. As immunotherapy combinations have transformed outcomes in Stage III disease across lung, head and neck, and other cancer types, the clinical value of a test that reliably averts Stage IV presentation has increased materially compared to when the NHS-Galleri trial was conceived.
PATHFINDER 2 Completes the Picture
The full nearly 36,000-participant PATHFINDER 2 study, the largest MCED interventional study in North America, reinforced and in some respects exceeded the NHS performance metrics. The positive predictive value in PATHFINDER 2 reached 60.3%, specificity was 99.6%, and cancer signal origin accuracy exceeded 90%. More than half of newly detected cancers were Stage I or II, and more than 70% had no USPSTF Grade A or B recommended screening available.
The screening multiplication effect is striking. Adding Galleri to standard USPSTF Grade A and B screening recommendations produced 6.5 times as many screen-detected cancers. Even when added to the broader Grade B and C recommendations, it generated three times as many. As Kumar summarized: "Today, the standard of care finds only about 6% of cancers in the U.K. and around 14% of cancers in the U.S. Adding Galleri to the standard screening programs could increase this figure to greater than 50% in the U.S."
The safety profile across both studies also held up. Across three screening rounds in the 70,000-participant intervention arm of NHS-Galleri, there were 864 false positives — a false positive rate of 0.45%. For context, physician Eric Sue, who has performed 1,277 Galleri screens in his Los Angeles internal medicine practice, put the comparative PPV in perspective on the call: "The positive predictive value of an abnormal mammogram actually meaning someone has breast cancer is 4.4%. The positive predictive value of a positive Cologuard stool test actually meaning someone has colon cancer is 3.7%. So 60% positive predictive value of a positive Galleri test is an order of magnitude better than what we're doing as standard of care."
The Cancer Signal Origin Feature Is Proving Clinically Decisive
One of the more underappreciated aspects of the Galleri test — its cancer signal origin, or CSO, prediction — emerged as a central clinical theme during the physician panel discussion. Radiation oncologist Nima Nabavizadeh of Oregon Health and Science University described a case where a signal-positive result for gastroesophageal cancer in a patient with prior gastric bypass surgery led to a biopsy of an area of the stomach inaccessible by routine outpatient endoscopy. The biopsy, performed under general anesthesia using a specialized endoscope, confirmed early-stage gastric cancer. "Without that cancer signal origin, we would have completely overlooked it, and this patient would have presented with Stage IV cancer," Nabavizadeh said.
He described a second case where an anorectal signal led to an early-stage anal cancer diagnosis in an asymptomatic patient, enabling curative treatment with substantially reduced radiation and chemotherapy. His broader point was that CSO accuracy — above 90% in both studies — allows physicians to pursue otherwise ambiguous or indeterminate imaging findings with a level of clinical conviction they would not otherwise have.
GRAIL's Andrew Partridge noted that the FDA advisory panel on MCEDs had itself identified CSO as an important component of any approvable MCED test, which suggests the feature is not merely a commercial differentiator but a regulatory consideration.
Mortality Data, FDA Pathway, and What Comes Next
On the question of mortality endpoints, Ofman was categorical: "I don't believe a true mortality study is going to ever be done for MCED." He argued that the presupposition underlying such a demand — that finding early cancer may not be beneficial — is clinically and ethically untenable given decades of evidence across existing screening programs. GRAIL has committed to passive mortality follow-up comparing the two NHS-Galleri arms at future time points, and Kumar referenced a "nested mortality analysis" targeted for the 2028 timeframe, though he declined to detail the methodology on the call.
More near-term, GRAIL is seeking to extend follow-up of the NHS-Galleri population by at least another 12 months. Kumar indicated the company is "hopefully close to getting that agreed with the NHS," with data analysis expected by Q1 or Q2 of next year. That additional data cohort would be the first opportunity to observe whether the Stage III and IV reduction trend in the incident rounds continues to steepen — and whether the control arm begins catching up on the undiagnosed late-stage cancers that the intervention arm already found.
On FDA and reimbursement, Ofman confirmed GRAIL submitted its premarket approval application on January 29 and noted the recently signed Medicare Multi-Cancer Early Detection and Screening Coverage Act, which establishes a Medicare reimbursement pathway conditional on FDA approval. He was measured on the FDA question specifically: "We believe that they're going to be focused on clinical validation and clinical performance. We don't believe that they're going to be particularly focused on clinical utility and stage shift." CMS and private payers, by contrast, are expected to be highly interested in the stage shift data presented this weekend.
Competitive Moat Is Real, But Physician Communication Remains a Work in Progress
The competitive positioning argument made repeatedly on the call — that no other MCED test has real-world interventional prospective trial data at this scale — is factually accurate and matters for how physicians and payers will eventually frame their choices. Sue stated it bluntly: "The other competitors, I don't think have any real-world interventional prospective trials that prove what they are showing in case-control trial data actually works in the real world intended patient use population."
The more immediate challenge GRAIL faces is the communication gap created by the February headline. Management acknowledged on the call that physician prescribing behavior softened following the "cancer screening fails major trial" coverage and that the company has not yet been able to provide the detailed data clarification to its sales and medical teams. That training is described as imminent, with the ASCO data providing the foundation. Whether the nuanced story of prevalent-round bolus effects and incident-round improvement trajectories can be translated into a thirty-second conversation between a sales rep and a busy primary care physician remains an open and genuinely difficult question.
Galleri has now detected cancers across more than 150 distinct cancer types across the combined study population of nearly 180,000 participants — more than triple the 50 types identified in the original study. Roughly 60% of the cancers detected have no current screening paradigm, rising to more than 70% when accounting for the limited availability of lung screening at the time the NHS trial ran. That breadth, combined with a test that finds four to six and a half times as many cancers as standard of care alone, is the core value proposition. The data now exist to defend it rigorously. The work ahead is making sure the people who write the prescriptions understand it.