KEY POINTS
- THUNDER-2 was a prospective single-centre phase II trial including 62 evaluable patients with locally advanced rectal cancer treated on a 0.35-T MR-Linac. Standard treatment delivered an SIB of 55 Gy to the high-dose target and 45 Gy electively in 25 fractions, with concurrent capecitabine or continuous-infusion 5-fluorouracil.
- Tumor volume was remeasured at fraction 10 to calculate the Early Regression Index. Patients with ERI ≤13.1 continued standard treatment, whereas those with ERI >13.1 underwent response-adapted escalation of residual disease to 60.1 Gy, corresponding to EQD2 62.2 Gy, using online adaptive replanning.
- Thirty patients (48.4%) were classified as predicted responders and 32 (51.6%) as predicted non-responders. Median tumor volume fell from 20.2 cc at baseline to 9.9 cc at fraction 10, illustrating the substantial early anatomical change used by the biomarker.
- Complete response at two years occurred in 60.0% of predicted responders and 21.9% of dose-escalated non-responders. The latter substantially exceeded the approximately 3% historical complete-response probability on which the trial design was based, but there was no concurrent untreated non-responder control.
- Overall, 25/62 patients (40.3%) achieved complete response. Non-operative management was used in 14/30 responders and 4/32 non-responders; two local regrowths occurred in the non-responder group and both were salvaged.
- Median follow-up was 27.2 months. Responders had higher TME-free survival than non-responders (38.7% vs 8.3%; HR 2.35, 95% CI 1.29–4.26; p=0.004, as reported), while overall and disease-free survival did not significantly differ. Eight patients developed distant metastases, evenly distributed between groups.
- ERI was the only variable independently associated with complete response (adjusted OR 0.912 per unit increase, 95% CI 0.840–0.990; p=0.028). Grade ≥3 gastrointestinal toxicity was 6.4%, and the paper reports no clear excess severe toxicity from the adaptive escalation strategy.
CLINICAL TAKEAWAY
THUNDER-2 moves adaptive radiotherapy beyond adapting to anatomy: it uses response observed during treatment to decide who receives more dose. The 21.9% complete-response rate in patients predicted to do poorly is compelling enough for further study, but without a randomized non-escalated comparator it cannot prove that escalation caused the improvement.