Background: There is uncertainty when determining the optimal treatment for malignant polyps. Clinicians must balance the oncological risk of a malignant polyp with the risk of morbidity and mortality from surgery. This study developed an online risk-calculator using machine-learning techniques to predict the risk of an adverse outcome from a malignant colorectal polyp following polypectomy.
Methods: Retrospective data collection of a population-wide database of all malignant polyps from 2011 to 2020 was performed. Utilizing an artificial intelligence-based machine learning approach a generalized linear mixed (GLM) model was developed to predict the risk of an adverse outcome after polypectomy. The presence of an adverse outcome was determined by assessing for the presence of residual disease or lymphatic disease if colorectal resection was undertaken. Delayed disease recurrence was also assessed as an additional adverse outcome. Patient and pathological details from the Queensland Cancer Registry were collected and included in the calculator development.
Results: The following variables were included in the final model: age, gender, polyp location (right colon, left colon, rectum), depth of tumour invasion, lymphovascular space invasion, tumour grade, associated polyp type, mismatch repair immunohistochemistry status and margin status. Based on ROC analysis, the AUC for the final GLM model was 0.76. The mean accuracy of the GLM model was 0.76 (95% CI: 0.72-0.80).
Conclusion: This web-based nomogram will facilitate the discussion of whether individual patients with malignant colorectal polyps can be safely managed with polypectomy or whether the patient should undergo colorectal resection. This nomogram is now available at https://malignantpolyp.com/risk-calculator.
Keywords: colorectal carcinoma; malignant polyp; nomogram.
© 2025 Royal Australasian College of Surgeons.