Conditional survival rates after neoadjuvant chemoradiotherapy combined with surgery in intermediate-low locally advanced rectal cancer based on two-centre retrospective analysis

Abstract To evaluate the temporal changes in 5-year conditional survival (CS5) among patients with intermediate-low locally advanced rectal cancer (LARC) treated with neoadjuvant chemoradiotherapy (nCRT) and surgery, and to develop predictive nomograms for individualized CS5 estimation. A retrospective analysis was conducted on patients with intermediate-low LARC treated at two colorectal centres in China between 2012 and 2023. A total of 1,589 eligible patients were enrolled, with a median follow-up of 46.8 months (range, 1–144 months). Conditional survival probabilities were calculated using the formula CS (y|x) = OS(x + y)/OS(x). Cox regression analyses were performed to identify time-dependent prognostic factors influencing CS5, and nomograms predicting CS5 after surviving 1, 2, 3, and 4 years were constructed. CS5 increased steadily with additional years survived. After 1, 2, 3, 4, and 5 years of survival, CS5 increased to 83.5%, 84.4%, 85.2%, 88.8%, and 92.4%, respectively. At baseline, distance to the anal verge (DTAV), No.253 lymph node positive, post-nCRT CEA, tumour regression grade, and ypTNM stage were identified as independent prognostic factors for overall survival. However, the influence of several factors diminished with accumulating survival time, whereas the prognostic impact of ypTNM stage remained stable and progressively strengthened (HR = 1.81, 1.85, 1.89, and 2.27, all P < 0.001). Notably, patients with yp0–I stage achieved a CS5 of 100% after 3 years of survival, suggesting a potential statistical cure phenomenon. The nomograms predicting CS5 after 1, 2, 3, and 4 years of survival demonstrated acceptable discrimination, with AUCs of 0.731 (95% CI 0.676–0.786), 0.738 (95% CI0.669–0.807), 0.641 (95% CI0.543–0.740), and 0.701 (95% CI0.556–0.847), respectively. CS5 improves substantially over time in patients with intermediate-low LARC after nCRT and surgery, particularly in the yp0–I subgroup, in whom a statistical cure phenomenon may be achieved after 3 years. Time-dependent prognostic dynamics highlight the limitations of traditional survival estimates, while the developed CS5 nomograms offer a practical tool for individualized, stage-adapted follow-up planning. External validation in independent cohorts is warranted.