The FDA Fast-Tracked GSK's Rectal Cancer Drug and Left the Trial's Own Scorecard Out of Its Announcement

Kyle Diamantas wanted the public to notice the clock. On Monday the acting FDA commissioner announced accelerated approval of GSK’s immunotherapy Jemperli for a small group of rectal cancer patients. He said that finishing the review “nearly 4 months ahead of schedule” meant some patients “who might otherwise be preparing for surgery, radiation or chemotherapy may now have a different conversation with their doctor.” To support that conversation, the agency’s announcement quoted two encouraging results. It left out the result GSK’s trial was registered to measure.

For the patients it covers, the stakes are hard to exaggerate. The approval applies to adults with previously untreated stage II or III rectal cancer whose tumors are mismatch repair deficient or microsatellite instability-high, confirmed by an FDA-authorized test. The agency puts that group at about 3 percent of locally advanced rectal cancers. Their usual treatment is chemotherapy, pelvic radiation and surgery, and in the FDA’s own words the surgery can end “in some patients, permanent colostomy.” Angelo de Claro, director of the agency’s Oncology Center of Excellence, put it plainly: “Standard treatment for rectal cancer has major life-changing effects.” Dostarlimab offers those patients up to nine infusions, one every three weeks for six months. Michael Sapienza of the Colorectal Cancer Alliance noted in GSK’s release that “biomarker testing at diagnosis will be essential to identifying patients who may be eligible for this treatment,” because none of this helps anyone who is never tested.

The FDA’s two numbers went like this. In GSK’s AZUR-1 trial of 154 patients, after a median follow-up of more than 2 years, “almost three-fourths of patients had not needed surgery, radiotherapy or chemotherapy.” In an earlier study of 50 patients at Memorial Sloan Kettering, 93 percent maintained their cancer-free status for at least 12 months.

AZUR-1’s registry record lists one primary outcome: “Number of Participants with Sustained Complete Clinical Response for 12 Months (cCR12) as assessed by Independent Central Review (ICR)”. GSK’s release, issued the same day, gives the answer. Independent reviewers found that 47 percent of evaluable patients, 54 of 116, held a complete response for a year, with a 95 percent confidence interval of 37 to 56 percent. The FDA’s announcement does not contain the figure.

What it contains instead are measures that answer gentler questions. Against the independent reviewers’ 47 percent, the investigators at the trial sites, grading their own patients, counted 68 percent (98 of 144). An exploratory measure GSK calls “no evidence of cancer” reaches 76 percent (110 of 144) by adding 12 patients whose complete response was confirmed in tissue after they had surgery. That measure answers a reasonable question, but it is not the question of how many patients the drug cleared on its own.

The FDA’s “almost three-fourths” corresponds most closely to a different count, one reported by Targeted Oncology: 110 of 154 patients, or 71 percent, “had received dostarlimab alone, required no additional treatment, and experienced no disease progression.” It has the same numerator as GSK’s exploratory measure, a different denominator and different rules. It counts patients who had avoided further treatment by the data cutoff, not patients whose response independent reviewers confirmed had lasted a year. So there are two different 110s in the public record, and neither one is the registered endpoint.

GSK’s defense of the lower number is simple arithmetic, and it holds up. The company says another 38 participants who had responded were still working through the 12-month follow-up window, so the independently confirmed share may rise as they reach it. A response at the cutoff is not yet a 12-month response, though, and some of those 38 may not make it. GSK also reports no recurrences among responders to date, after a median follow-up of more than two years. For a cancer that usually means radiation and a surgeon, that is a remarkable run for one drug and no knife. It is also why an approval announcement should state the registered number and explain why it may grow, instead of reaching past it for a friendlier one.

The trial’s design makes that reach harder to excuse. AZUR-1 is open-label and single-arm, with no comparison group, and every investigator knew every patient was on the drug. Independent central review exists so that the people hoping a treatment works are not the only ones reading the scans. The site and central figures don’t share a denominator, so the gap between them is not a clean measure of investigator optimism, and GSK has not published a matched comparison. But the central read is the one the trial was built around. It is also the lowest of the three.

The drug’s costs belong next to its benefits. Like every checkpoint inhibitor, dostarlimab can turn the immune system against healthy organs, sometimes fatally. GSK’s release reports serious adverse reactions in 20 percent of patients and permanent discontinuation for side effects in 3 percent. The most common side effects were fatigue, rash, diarrhea and itching.

Then there is the 93 percent. It comes from NCT04165772, sponsored by Memorial Sloan Kettering with Tesaro, Inc. listed as industry collaborator. Tesaro is the GSK subsidiary that brought dostarlimab along. Andrea Cercek led that work and is also principal investigator of AZUR-1, and MSK’s disclosure, printed at the bottom of GSK’s release, reads: “Dr Cercek has financial interests related to GSK.” None of this sinks the MSK results. Cercek’s group published in the New England Journal of Medicine in 2025 that all 49 rectal cancer patients who completed treatment had a clinical complete response. That was the initial response. The 93 percent is the separate, sustained measure. The FDA reports one recurrence among the 27 MSK patients followed for at least three years. The data are strong, and the selection is still odd. The agency approved the drug on interim AZUR-1 results, then showed the public a smaller study led by an investigator with GSK financial ties, while the sponsor’s registered, independently read endpoint stayed in the sponsor’s own paperwork.

The speed came with a backstory. This was the 11th approval under the Commissioner’s National Priority Voucher pilot, and the drug already held Breakthrough Therapy designation and Priority Review. CancerNetwork reported that the original action date had been set for February 2027. In June the National Center for Health Research warned in a formal comment that when a product already qualifies for established expedited pathways, “the FDA should explain why an additional CNPV voucher is necessary rather than using existing tools.” The group added that the compressed timelines “place excessive strain on FDA staff, which remains weakened by the loss of employees.” Diamantas credited the Trump Administration with the timely access. Speed is a fair thing for an agency to boast about. It is a stranger boast to make in the same document that trims the trial’s scorecard.

Because the approval is accelerated, it stays conditional. GSK must keep following patients to show the responses translate into lasting benefit, and the agency concedes that longer follow-up is needed to learn how durable they are and how often the cancer returns. Full AZUR-1 interim data are due at the ESMO Congress this month. The registry estimates the trial’s primary completion date as November 2, 2026, two weeks after the approval, with the full study running to October 2029.

None of this was hidden. GSK printed the 47 percent, with its confidence interval, in its own release that morning. The FDA did not.

Sources

1. U.S. Food and Drug Administration: FDA Approves Immunotherapy That May Allow Certain Patients with Rectal Cancer to Avoid Chemotherapy, Radiation and Surgery (October 19, 2026)
2. GSK via Business Wire: Jemperli (dostarlimab-gxly) approved in US for adults with MSI-H/dMMR locally advanced rectal cancer (October 19, 2026)
3. ClinicalTrials.gov: AZUR-1, NCT05723562, registry record and primary outcome
4. Targeted Oncology: Dostarlimab Receives FDA Approval for MSI-H/dMMR Locally Advanced Rectal Cancer
5. ClinicalTrials.gov: Memorial Sloan Kettering study NCT04165772, sponsor and collaborator record
6. American College of Surgeons Brief: Is Neoadjuvant Checkpoint Blockade Sufficient to Treat Mismatch Repair-Deficient Rectal Cancer? (summary of Cercek et al., NEJM 2025)
7. CancerNetwork: FDA Approves Dostarlimab in Previously Untreated Locally Advanced Rectal Cancer
8. National Center for Health Research: Written Comment to FDA on the Commissioner’s National Priority Voucher Pilot Program (June 2026)