Related Experiment Video
Updated: Aug 7, 2026
Competing-Risk Nomogram for Predicting Cancer-Specific Survival in Multiple Primary Colorectal Cancer Patients after Surgery
Published on: September 27, 2024
NTCP models predicting bowel toxicity after total neoadjuvant treatment or chemoradiation for rectal cancer in the
Max D Tanaka1, Tomas M Janssen2, Bastiaan W K Schipaanboord2
1Department of Radiation Oncology, Netherlands Cancer Institute, Amsterdam, Netherlands;.
Purpose:
Radiation-induced bowel toxicity is a frequently observed adverse event after rectal cancer treatment. In the phase III RAPIDO trial, patients with locally advanced rectal cancer received either short-course radiation therapy (SCRT, 5×5 Gy) with consolidation chemotherapy (TNT arm) or chemoradiation with optional postoperative chemotherapy (CRT arm). We developed normal tissue complication probability (NTCP) models for bowel toxicity after SCRT-based TNT and CRT, using two different bowel contour definitions.
Methods And Materials:
NTCP models were developed for each arm separately using EMBRACE and RTOG bowel contours. Evaluated dose-parameters were the V10Gy, V15Gy, and V20Gy in the TNT and V15Gy, V30Gy, and V40Gy in the CRT arm. Endpoints included acute diarrhea grade ≥ 2, non-compliance with preoperative chemotherapy, late diarrhea grade ≥ 1, and patient-reported late diarrhea. Associations were assessed with uni- and multivariable logistic regression models, and model performance with the area-under-the-curve (AUC) and calibration plots.
Results:
The numbers of eligible patients in the TNT arm were 309 (acute toxicity), 221 (late diarrhea grade ≥ 1) and 209 (patient-reported late diarrhea), whereas in the CRT arm respective numbers were 279, 191 and 190. Acute diarrhea grade ≥ 2 was reported in 38% (TNT) and 28% (CRT) of the patients. All dose-parameters (from both bowel contours) showed comparable associations with the endpoint acute diarrhea grade ≥ 2. In the TNT arm, p-values ranged from .003 to .08 and cross-validated AUCs from 0.48 to 0.61. In the CRT arm p-values ranged from <.001 to .03 and cross-validated AUCs from 0.56 to 0.67. All models were reasonably well calibrated. In addition, associations were observed between dose-parameters and late diarrhea grade ≥ 1 in the CRT arm.
Conclusions:
The evaluated dose-parameters were able to predict acute diarrhea after SCRT-based TNT and CRT, and late diarrhea after CRT. Model performance was better in the CRT arm than the TNT arm. Both the EMBRACE bowelloop and RTOG bowelbag contour can be used to guide treatment plan optimization.
