Dijkstra 2023 - RAPIDO 5-Year Locoregional Failure

Full citation: Dijkstra EA, Nilsson PJ, Hospers GAP, Bahadoer RR, Meershoek-Klein Kranenbarg E, Roodvoets AGH, Putter H, Berglund Å, Cervantes A, Crolla RMPH, et al., and Collaborative Investigators. Locoregional Failure During and After Short-course Radiotherapy Followed by Chemotherapy and Surgery Compared With Long-course Chemoradiotherapy and Surgery: A 5-Year Follow-up of the RAPIDO Trial. Ann Surg. 2023;278(4):e766-e772.

Raw PDF: raw/Dijkstra 2023 - RAPIDO 5-Year Locoregional Failure.pdf

Study type / design

Planned secondary-endpoint analysis of the RAPIDO trial (international, multicenter, phase III RCT; original results in Bahadoer 2021 - RAPIDO Trial) at extended follow-up (median 5.6 years, vs the ~3-year follow-up of the original primary publication). 906 eligible patients (460 experimental/EXP = short-course RT → chemotherapy → TME; 446 standard-care/STD = long-course CRT → TME → optional adjuvant chemo). Primary focus: locoregional failure (LRF), split into early LRF (eLRF — no resection or R2 resection) and locoregional recurrence (LRR — recurrence after an R0/R1 resection). Also reports a 5-year update of RAPIDO’s original endpoints (disease-related treatment failure [DrTF], distant metastases [DM], overall survival [OS]).

Key findings

• This is the paper that resolves the cross-trial locoregional-recurrence question flagged as unresolved elsewhere in this wiki. With longer follow-up, the RAPIDO EXP (TNT) arm shows a statistically significant increase in locoregional recurrence that was not apparent at the trial’s original ~3-year analysis:
  • Overall LRF: 12% (EXP) vs 8% (STD), P=0.07 (borderline, not itself significant).
  • LRR after R0/R1 resection: 10.2% (EXP) vs 6.1% (STD), P=0.027 — significant.
  • LRR after R0 resection specifically: 7.2% vs 3.9%, P=0.049 — significant.
  • Multivariate Cox regression: EXP treatment (HR 1.87, P=0.014) and enlarged lateral lymph nodes (HR 1.79, P=0.042) independently predicted LRR.
• Mechanism explored, not settled: intraoperative mesorectal breach was more common in the EXP group (11% vs 6%, P=0.022), and LRR was markedly more common when the mesorectum was breached (21% EXP vs 4% STD in that subgroup) — the authors speculate the longer EXP treatment time (~40 weeks vs ~25 weeks to surgery) may yield a more fragile/fibrotic mesorectum and lower specimen quality, but this is explicitly framed as a hypothesis, not proven. Radiation technique (3D-CRT vs IMRT/VMAT) also showed an unexplained interaction with LRR risk in the EXP group specifically. Positive CRM, tumor deposits, and higher ypN-stage were also independently associated with LRR at pathology.
• 5-year update of RAPIDO’s primary endpoints — the systemic-control benefit is durable:
  • DrTF: 27.8% (EXP) vs 34.0% (STD), HR 0.79, P=0.048 — still significant.
  • DM: 23.0% (EXP) vs 30.4% (STD), HR 0.73, P=0.011 — still significant.
  • OS: 81.7% (EXP) vs 80.2% (STD), HR 0.91, P=0.50 — still no significant difference, consistent with the original 3-year analysis.
• Overall survival after an LRF was comparable between arms (HR 0.76, P=0.29) — an LRR, once it happens, is not more lethal in one arm than the other.
• eLRF (the rarer, poor-prognosis “never resectable” subset) occurred at similar rates in both arms (10/460 EXP vs 10/446 STD) and was strongly associated with pre-existing/synchronous distant metastases in both groups — a distinct, high-risk subpopulation rather than a TNT-specific effect.

Limitations

• RAPIDO was powered for its primary endpoint (DrTF), not for this secondary LRF/LRR analysis — the authors note this explicitly.
• No central review of MRIs, RT target volumes, dose-volume histograms, or histopathology specimens had been completed at time of publication (data mostly from case report forms; central review was “currently being revised”).
• Restaging MRI was not performed in all patients.
• The mesorectal-breach/specimen-quality explanation for the EXP arm’s excess LRR is exploratory, not confirmed by this study design.

Relevance

This directly resolves TNT.md’s and Rectal Cancer.md’s previously-flagged “unresolved” open item about whether RAPIDO’s locoregional trade-off would strengthen at longer follow-up — it does, and significantly so. At 5 years, RAPIDO now shows a genuine, statistically significant increase in locoregional recurrence with TNT, while the systemic-control benefit (DrTF, DM) persists and OS remains unchanged. This sharpens rather than resolves the cross-trial disagreement with Conroy 2024 - PRODIGE 23 Long-Term Results and Jin 2022 - STELLAR Trial (both of which show numerically better local control with TNT): the RAPIDO-specific locoregional trade-off is now confirmed as a real, durable finding within RAPIDO, not a fluke of short follow-up — but it still does not generalize to the other two trials, so “TNT trades local for distant control” remains a RAPIDO-specific finding, not a class effect, per this wiki’s contradiction-flagging convention. See Total Neoadjuvant Therapy (TNT) for the updated cross-trial synthesis.