onc brain

The longer read

The question STAR-TREC answers at 12 months is narrower than organ preservation versus surgery, and more useful for it: in small, MRI node-negative rectal cancers where primary TME remains a reasonable default, which radiation backbone should a response-adapted pathway be built on. The answer favours long-course chemoradiotherapy with little ambiguity. TME-free survival was 78.5% with LCCRT-OP against 60.6% with SCRT-OP, the posterior probability of LCCRT superiority exceeded 99.5%, and the response data point the same way, with clinical complete response at the second assessment in 104 pts after LCCRT and 61 after SCRT. SCRT pts were more often sent to TAE for near-complete response, so the pathway was already compensating for the weaker response rather than manufacturing the gap.

How far to trust the magnitude is the harder question. The phase 2 estimate that prompted this report, HR 3.7, came from a small unmasked interim, and once phase 3 pts were pooled the estimate fell to 1.90 with a far narrower interval. The direction held and the magnitude contracted, which is how an early extreme estimate usually behaves, and the decision to publish at 12 months was taken after the committees had seen it. The Bayesian design removes the alpha-spending objection but not the concern about when to look. The endpoint also deserves scrutiny. TME-free survival is a clinician decision in an open-label trial, so any lower threshold for abandoning organ preservation after a short course would widen the gap without reflecting a difference in eventual tumour control. The authors themselves note that downstaging is time dependent.

The comparison with surgery is weaker evidence than the comparison between radiation schedules. Only phase 2 randomised pts to primary TME; in phase 3 the TME arm was self-selected, and 86% of phase 3 pts chose organ preservation. The surgical morbidity is still instructive: grade 3 or worse SAEs in 18% of TME recipients, intraperitoneal perforation from anastomotic leak in 15%, and one postoperative death. But it frames the stakes rather than measuring a randomised benefit. On the other side of the ledger, SCRT's advantage is acute, with grade 3 or worse SAEs within 4 weeks of radiation in 6% after LCCRT and 1% after SCRT. That is a real trade, but it is paid once, whereas a TME the schedule failed to avoid carries the morbidity the trial exists to escape.

Much of the organ-preservation experience, OPRA among it, comes from watch-and-wait after neoadjuvant therapy in locally advanced disease; STAR-TREC moves the question to tumours for which the historical reference is surgery alone. That shift carries a staging risk the 12-month data cannot resolve. Nodal disease was found in 22% of TME specimens despite MRI excluding it, and organ-preserving pts never have those nodes sampled. Whether that becomes regional failure, and whether regrowth after complete response differs by schedule, belongs to the prespecified 30-month endpoint, which also counts stoma and local recurrence. The result would be wrong in the way that matters if organ preservation converges by then, or if function and late toxicity favour SCRT enough to offset the TMEs it fails to prevent early.