Pathological Response Depth and Pre-Treatment Risk Stratification After Neoadjuvant Chemoradiotherapy for Rectal...

Neoadjuvant chemoradiotherapy (nCRT) is standard care for locally advanced rectal cancer, yet whether benefit derives from treatment receipt or from response depth remains debated. We examined nCRT receipt and tumour regression against outcome, and whether poor response is predictable beforehand.

Eighty-nine consecutive patients undergoing resection for rectal or rectosigmoid adenocarcinoma (2020-2022; median follow-up 16 months) were analysed; outcomes were described in patients receiving nCRT (n = 60) and in those undergoing upfront surgery (n = 29), two groups whose allocation depended on stage and location and which are therefore presented as distinct clinical populations rather than as directly comparable arms. Response was graded by modified Ryan tumour regression grade. An exploratory four-item Pre-treatment Response Risk (PRR) score was derived and internally validated by bootstrap optimism correction. Comparisons used the Student t-test or Mann-Whitney U test and the chi-square or Fisher exact test, with logistic (maximum likelihood and Firth-penalised) and Cox regression and receiver operating characteristic analyses.

nCRT exposure was not statistically associated with mortality during follow-up (21.7% vs. 24.1%, p = 0.792), metastatic recurrence (21.7% vs. 20.7%, p > 0.999) or textbook outcome (35.0% vs. 34.5%, p > 0.999); the association with mortality remained non-significant after covariate adjustment in the whole cohort (Cox hazard ratio 0.66, 95% CI 0.24-1.83) and in a sensitivity analysis restricted to rectal tumours (Charlson-adjusted Firth odds ratio 0.52, 95% CI 0.08-3.94). The median follow-up of 16 months was too short to assess disease-free survival. Among patients who received nCRT, mortality varied with the depth of pathological response rather than with treatment receipt itself, from 0% in complete responders to 62% in poor responders (p < 0.001), and each one-grade worsening of the tumour regression grade multiplied the odds of death (OR 6.26, 95% CI 1.74-22.5, p = 0.005), although response grade, residual ypT/ypN stage and comorbidity were closely collinear. Comorbidity was associated with mortality (Charlson index, Firth-penalised OR 3.41 per point, 95% CI 1.89-6.15, p < 0.001) whereas nCRT was not (OR 0.38, 95% CI 0.08-1.72). In the derivation cohort, the PRR score discriminated poor response (apparent AUC 0.95, optimism-corrected 0.89) and was associated with major morbidity, which rose from 0% in the lowest to 93% in the highest risk band, and with mortality, which rose from 0% to 71% (p < 0.001 for each outcome).

Within this cohort, nCRT exposure was not statistically associated with the evaluated short-term outcomes, which does not establish equivalence between strategies; depth of pathological response and host comorbidity carried the prognostic information, and routine pre-treatment variables showed promise for anticipating poor response. External validation is required before any clinical use.