The Number Behind GSK's Rectal Cancer Approval Is 47 Percent, Not 76

Sascha Roth was 38 when she was diagnosed with rectal cancer, and she went to Memorial Sloan Kettering in New York for a second opinion. There she learned her tumor was driven by Lynch syndrome and was mismatch-repair deficient, the kind of cancer less likely to respond to chemotherapy and surgery, and she agreed to join a trial of an immunotherapy drug given on its own. By the hospital’s own account of her case, her doctors afterward could find no sign of cancer, and she has now passed five years cancer free. “I was stunned and ecstatic,” she said.

Her case, and the run of others like it, made dostarlimab famous as the drug that erased rectal tumors in every patient who completed treatment. On Monday the FDA granted it accelerated approval, sold by GSK as Jemperli, for adults with stage II or III rectal cancer that is mismatch-repair deficient (dMMR) or microsatellite-instability high (MSI-H). The evidence at the center of that approval comes from a different and much larger trial with a stricter test, and on that test the independent reviewers confirmed a response in 47 percent of patients. The company’s announcement also carried a second, more flattering number: 76.

One trial, three answers

AZUR-1, GSK’s registrational study, treated 154 previously untreated patients with 500 mg of dostarlimab every three weeks for nine cycles, about six months. Its primary objective was a clinical complete response, no detectable tumor, held for 12 months and judged by independent central review.

The GSK press release reports three answers to that question, each with its own denominator. The independent reviewers could evaluate 116 patients and confirmed a year-long complete response in 54 of them, which is 47 percent. The trial’s own investigators, grading 144 patients, put the same endpoint at 68 percent, or 98 of 144. Then there is an exploratory company analysis of “no evidence of cancer” at a year or more, 110 of 144 patients or 76 percent, which takes those 98 and adds 12 patients who went to surgery and whose removed tissue turned out to hold no cancer. Only the first number is the one the trial was built to answer to.

Memorial Sloan Kettering, whose trial launched the legend, has updated its 2022 story to acknowledge as much. The FDA also reviewed interim AZUR-1 data from a group “who were not investigators on the study,” the hospital now writes, and that review found tumors disappeared in about half of patients.

The gap could narrow. At the data cutoff, 38 of the independently evaluable patients had a complete response that had not yet reached the 12-month mark, and after a median follow-up of 26.3 months no responder had progressed. A complete response at seven months, though, is not a complete response at twelve, and nobody outside the company has seen the full dataset that would settle it. GSK has scheduled the full interim results for the ESMO Congress in October, which puts the public presentation after the approval rather than before it.

The 12 surgical patients are their own uncomfortable detail. MSK’s update says doctors at some AZUR-1 hospitals “less experienced in this kind of surveillance” decided to remove the rectum after immunotherapy and later found the operation was not needed, because the drug had already worked. GSK’s 76 percent counts the patients whose surgery turned up no cancer as successes. For the patients themselves, it meant a major operation with a long recovery that the drug may well have spared them.

The hospital that started it

The FDA also weighed Memorial Sloan Kettering’s own phase 2 trial, NCT04165772, in 50 patients. It reported a 12-month complete response rate of 93 percent and an overall response rate, a broader measure that also counts partial and near-complete responses, of 100 percent. Of 27 patients followed for at least three years, one had a local recurrence, treated by removing lymph nodes. An earlier readout presented at AACR in 2025 and published in the New England Journal of Medicine reported a complete clinical response in all 49 rectal cancer patients, with 92 percent disease-free at two years.

Those results were graded by the trial’s investigators at the center that pioneered the approach. AZUR-1 took the protocol to hospitals in 10 countries and handed the primary call to outside reviewers, and the confirmed rate came in at roughly half. The two trials measured differently and enrolled differently, so the gap between them proves nothing on its own about who read the scans right. It does show how far a single center’s result can travel from what a multinational trial confirms.

The trial that built the legend was not independent of the manufacturer, either. MSK lists GSK among its funders, alongside the Simon and Eve Colin Foundation, Stand Up To Cancer, Swim Across America and the National Cancer Institute. Andrea Cercek, who led it, lists advisory roles with 14 companies including GSK, and institutional research grants from GSK and Pfizer. GSK’s approval release carries the hospital’s disclosure that she “has financial interests related to GSK.”

Neither trial had a control arm. Accelerated approval exists for that situation: the FDA approves on an endpoint it judges reasonably likely to predict clinical benefit, here the 12-month complete response, and the label states that continued approval “may be contingent upon verification” of that benefit in confirmatory trials. GSK’s release does not name one. The application carried Breakthrough Therapy and Fast Track designations, priority review and one of the agency’s new National Priority Vouchers, and it was approved on October 19 against a goal date of February 2027. Before the agency granted that speed, someone there presumably asked which trial would verify the benefit, against what comparison and by when. The public record does not say.

None of this erases what the drug offers the patients who respond. Standard treatment for locally advanced rectal cancer is chemotherapy, radiation and surgery, and Cercek’s description of the aftermath is blunt: “life-altering bowel and bladder dysfunction, incontinence, infertility, sexual dysfunction.” Of two young women with large tumors treated with the drug alone, she notes that radiation “would have induced immediate menopause”; both have since had babies. Organ preservation is not new, and a 2023 review in Clinical Oncology traced watch-and-wait from controversy to common clinical practice. What dostarlimab changes is the route, skipping radiation for a group whose tumors are unusually visible to the immune system.

That group is small. MSK puts dMMR and MSI-H tumors at 5 to 10 percent of the nearly 50,000 rectal cancers diagnosed in the US each year, across all stages, and the label covers only stage II and III disease. Finding those patients means testing tumors at diagnosis. “Biomarker testing at diagnosis will be essential to identifying patients who may be eligible for this treatment,” said Michael Sapienza of the Colorectal Cancer Alliance in GSK’s release.

The drug carries risks of its own. In AZUR-1, 20 percent of patients had serious adverse reactions; the most common side effects were fatigue (32 percent), rash (22 percent), diarrhea (18 percent) and itching (17 percent), and 3 percent stopped treatment permanently. The label carries the standard PD-1 warnings about immune attacks on organs. Patients who keep their rectum sign up for endoscopy, rectal MRI and imaging every 4 months for the first 2 years, then every 6 months.

GSK’s release gives no price. When the first results made headlines in 2022, the drug cost about $11,000 per dose; nine doses at that price come to about $99,000, before infusion fees and the years of scans.

The label is plain about what remains unknown: “The long-term risk of relapse is unknown, with additional follow-up expected.” GSK printed that sentence in the same GSK press release that carried the 76 percent.

Sources

1. GSK press release
2. Oncology Nursing News: FDA grants accelerated approval to dostarlimab in dMMR rectal cancer
3. OncLive: FDA approves dostarlimab for dMMR/MSI-H locally advanced rectal cancer
4. Memorial Sloan Kettering: Rectal cancer disappears after experimental use of immunotherapy (updated Oct. 19, 2026)
5. MedPage Today: “Groundbreaking Option” approved for rectal cancer
6. The ASCO Post: Nonoperative management of mismatch repair-deficient tumors (AACR 2025 / NEJM)
7. BioPharm International: FDA accepts priority review for GSK’s Jemperli in rectal cancer
8. Clinical Oncology: Watch-and-wait is an option in rectal cancer patients (2023)
9. BioSpace: Small study of GSK rectal cancer drug yields “unheard of” results (2022)