AZUR-1 NCT05723562
FordMMR/MSI-H locally advanced rectal cancer, treatment-naive
TL;DRcCR12 68% (investigator) and cCR12+pCR 76% with dostarlimab alone in dMMR/MSI-H LARC, no recurrences to date at 26.3mo median f/u.
The RT-omission read hinges on the assessor: investigator cCR12 is 68%, central review 47% (54/116) with 38 pts still pending. Either figure exceeds the 21-41% post-TNT cCR range the source cites, which supports MMR testing before any pelvic RT is planned in LARC.
In dMMR/MSI-H LARC, this supports checkpoint-first, organ-preservation discussion before committing to chemoRT or TNT; it does not extend to pMMR/MSS rectal cancer.
Central review cCR12 is 47% (54/116) against investigator 68%, with 38 pts pending; both exceed the 21-41% post-TNT cCR range cited. In dMMR/MSI-H LARC, pelvic RT is shifting toward a salvage role, so MMR status should be known before simulation.
Nine cycles of dostarlimab 500 mg q3w delivered 68% cCR12 (INV) with 88% completing therapy and G≥3 immune-mediated AEs in 4 pts. This positions PD-1 monotherapy as the first move in dMMR/MSI-H LARC, ahead of oxaliplatin-based TNT.
Only 15 pts underwent resection after dostarlimab alone and 12 (80%) were pCR, so most surgical specimens showed no residual tumour. The surgical question becomes timing of salvage and surveillance intervals, not upfront TME, in dMMR/MSI-H LARC.
10 details 2 trials watching
Global single-arm phase II, registrational. Interim analysis, N=154 enrolled, median f/u 26.3 mo (range 2-36).
dMMR/MSI-H locally advanced rectal cancer. 144 pts had ≥12 mo f/u for investigator-assessed cCR12.
Dostarlimab 500 mg IV q3w for 9 cycles as monotherapy; chemotherapy, radiation and surgery reserved for non-responders.
Key endpoints: sustained cCR for 12 months (cCR12) after end of treatment by investigator and independent central review, and composite cCR12 (INV) + pCR.
cCR12 68% (INV) vs 47% (54/116) (ICR); composite cCR12+pCR 76%. Of 15 resected pts, 12 (80%) had pCR; no recurrences to date after cCR or pCR.
| Assessor | Evaluable | cCR12 |
|---|---|---|
| Investigator | 144 | 68% |
| Independent central review | 116 | 54/116 (47%) |
88% completed all 9 cycles. G≥3 immune-mediated AEs in 4 pts.
Post-TNT cCR is cited at 21-41% in LARC broadly; AZUR-1 exceeds that in a biomarker-selected group. It scales up the earlier small dostarlimab study that first showed durable cCR in dMMR LARC.
Investigator and central review diverge on cCR12, and ICR evaluability lagged because cCR was confirmed later. cCR12 is a surrogate for durable organ preservation; local regrowth risk beyond 3 years is unmeasured.
Durable cCR with no recurrences so far makes RT omission a realistic default question in dMMR LARC. Open: how long surveillance must run, and what salvage looks like for late regrowth.
Single-arm interim analysis without a randomised comparator; investigator vs central review cCR12 diverge, and durability beyond 26.3mo median f/u is unproven.
- Will ICR cCR12 converge with investigator rate as pending pts mature?
- Late local regrowth risk beyond 3 years without RT or surgery not yet Non-Operative Management and Following Immunotherapy for Colorectal Cancer and Other GI Cancersn=50 · primary completion 2029-06 · dMMR/MSI-H W&W registry after neoadjuvant IO
- Role of RT for dMMR pts without cCR after dostarlimab recruiting Study of Induction PD-1 Blockade in Subjects With Locally Advanced Mismatch Repair Deficient Solid Tumors Phase 2n=212 · primary completion 2026-11 · dostarlimab then capecitabine-IMRT in dMMR LARC
📚 Sources · 📄 2 papers
Abstract
Abstract
The longer read
AZUR-1 moves the dMMR/MSI-H rectal question from a single remarkable cohort to a multinational registrational dataset, and the central finding holds at scale: most pts given nine cycles of dostarlimab and nothing else are sitting in sustained clinical complete response at a median of 26.3 months. For a radiation oncologist, this is the most direct challenge yet to reflexive pelvic chemoRT or TNT in this biomarker group. The comparison the source draws, 21-41% cCR after TNT, is not a like-for-like benchmark because it comes from unselected LARC, but the gap is wide enough that selection alone does not plausibly explain it.
The methodological detail that should move confidence most is the split between investigator and central review. Investigator cCR12 is 68%; independent central review is 47% (54/116), with 38 pts still in cCR and awaiting the 12-month mark. The investigators attribute the gap to central reviewers confirming cCR later than the initial end-of-treatment assessment, which would shift pts into a pending bucket rather than into failure. That explanation is plausible and the 98% durability figure among responders supports it, but it is also exactly the pattern one would see if central readers were stricter about what counts as a complete response on endoscopy and MRI. Until the pending 38 resolve, the honest range for this endpoint sits between the two figures, and the registrational read will likely rest on the ICR number.
The pathology data are small but informative. Only 15 pts went to surgery after dostarlimab alone, and 12 of them had pCR. That tells us the clinical non-responders are not uniformly true non-responders, which raises the question of whether some resections in this setting were avoidable with longer observation, a timing issue the trial does not settle.
What the trial cannot answer is late behaviour. Watch-and-wait experience after chemoRT has taught the field that local regrowth can appear after the first two years, and the absence of recurrence to date is bounded by follow-up that ranges from 2 to 36 months. A single-arm design also leaves no estimate of what RT would add for the minority who do not reach cCR, or whether sequencing RT after incomplete response preserves the organ as often as upfront TNT would have.
Practically, the result reinforces two things: MMR status belongs at the start of LARC workup, before RT simulation, and in dMMR disease the RT decision is increasingly a salvage decision rather than a definitive one. What would make this wrong is a late regrowth signal, or central review cCR12 failing to recover toward the investigator figure as the pending pts mature.