Notice: Users may be experiencing issues with displaying some pages on stanfordhealthcare.org. We are working closely with our technical teams to resolve the issue as quickly as possible. Thank you for your patience.
 

Learn about the flu shot, COVID-19 vaccine, and our masking policy »

Stanford Health Care

Optimal policy tree to assist in adjuvant therapy decision-making after resection of colorectal liver metastases. Surgery Endo, Y., Alaimo, L., Moazzam, Z., Woldesenbet, S., Lima, H. A., Yang, J., Munir, M. M., Shaikh, C. F., Azap, L., Katayama, E., Rueda, B. O., Guglielmi, A., Ruzzenente, A., Aldrighetti, L., Alexandrescu, S., Kitago, M., Poultsides, G., Sasaki, K., Aucejo, F., Pawlik, T. M. 2023

Abstract

Although systemic postoperative therapy after surgery for colorectal liver metastases is generally recommended, the benefit of adjuvant chemotherapy has been debated. We used machine learning to develop a decision tree and define which patients may benefit from adjuvant chemotherapy after hepatectomy for colorectal liver metastases.Patients who underwent curative-intent resection for colorectal liver metastases between 2000 and 2020 were identified from an international multi-institutional database. An optimal policy tree analysis was used to determine the optimal assignment of the adjuvant chemotherapy to subgroups of patients for overall survival and recurrence-free survival.Among 1,358 patients who underwent curative-intent resection of colorectal liver metastases, 1,032 (76.0%) received adjuvant chemotherapy. After a median follow-up of 28.7 months (interquartile range 13.7-52.0), 5-year overall survival was 67.5%, and 3-year recurrence-free survival was 52.6%, respectively. Adjuvant chemotherapy was associated with better recurrence-free survival (3-year recurrence-free survival: adjuvant chemotherapy, 54.4% vs no adjuvant chemotherapy, 46.8%; P < .001) but no overall survival significant improvement (5-year overall survival: adjuvant chemotherapy, 68.1% vs no adjuvant chemotherapy, 65.7%; P = .15). Patients were randomly allocated into 2 cohorts (training data set, n = 679, testing data set, n = 679). The random forest model demonstrated good performance in predicting counterfactual probabilities of death and recurrence relative to receipt of adjuvant chemotherapy. According to the optimal policy tree, patient demographics, secondary tumor characteristics, and primary tumor characteristics defined the subpopulation that would benefit from adjuvant chemotherapy.A novel artificial intelligence methodology based on patient, primary tumor, and treatment characteristics may help clinicians tailor adjuvant chemotherapy recommendations after colorectal liver metastases resection.

View details for DOI 10.1016/j.surg.2023.06.045

View details for PubMedID 37778970