Objective To explore the long-term survival impact of traditional Chinese medicine (TCM) syndrome differentiation and treatment in patients with stage Ⅲ-Ⅳ colorectal cancer and its therapeutic efficacy for such patients.

Methods This retrospective cohort study included 201 patients with stage Ⅲ-Ⅳ colorectal cancer who were treated at Yangzhou City Hospital of Traditional Chinese Medicine from January 2015 to December 2019. Based on whether they received TCM syndrome differentiation and treatment, the patients were divided into control group (not receiving TCM syndrome differentiation and treatment, n=61) and treatment group (receiving TCM syndrome differentiation andtreatment, n=140). According to the duration of TCM syndrome differentiation and treatment, the treatment group was further divided into treatment subgroup 1 (1- < 6 months, n=51) and treatment subgroup 2 (≥6 months, n=89). The associations between TCM syndrome differentiation and treatment and survival outcomes were analyzed using the Cox proportional hazards model and inverse probability weighting in both the propensity score-matched cohort and the full cohort. Potential subgroups benefiting from TCM treatment were identified through clinical stratified analysis and an unsupervised machine learning method based on K-medoids consensus clustering.

Results Inverse probability weighting analysis of the full cohort showed a reduced risk of death in treatment subgroup 2 (HR=0.29, 95%CI, 0.18 to 0.47, P < 0.001). Treatment subgroup 2 (with 86 patients matched) was matched with the control group, resulting in 86 patients in each of the matched TCM treatment group and the matched control group. In the propensity score-matched cohort, the univariate Cox proportional hazards regression model showed a reduced probability of death in the TCM treatment group compared with the matched control group (HR=0.69, 95%CI, 0.41 to 1.17, P=0.168). After adjustment, the probability of death in the TCM treatment group wasalso lower than that in the matched control group (HR=0.31, 95%CI, 0.16 to 0.60, P < 0.001). Further Cox proportional hazards model analysis indicated that in the propensity score-matched cohort, the 3-year and 5-year survival rates in the TCM treatment group were 92.5% and 72.4%, respectively, which were better than those in the matched control group (73.5% and 60.7%, respectively). The propensity score-matched cohort results showed that in patients with low Ki67 expression, TCM syndrome differentiation and treatment did not improve survival (HR=0.57, 95%CI, 0.10 to 3.37, P=0.540); in patients with high Ki67 expression, TCM syndrome differentiation and treatment significantly reduced the risk of death (HR=0.53, 95%CI, 0.29 to 0.96, P=0.037); in patients with poorly differentiated pathology, TCM syndrome differentiation and treatment significantly reduced the risk of death (HR=0.41, 95%CI, 0.24 to 0.70, P=0.001); in patients with moderately to well-differentiated pathology, TCM syndrome differentiation and treatment reduced the risk of death, but the difference was not statistically significant (HR=0.52, 95%CI, 0.26 to 1.02, P=0.057). After Bonferroni correction, TCM syndrome differentiation and treatment significantly improved survival only in patients with poorly differentiated pathology (corrected P=0.002). Using an unsupervised machine learning algorithm, the patient population in the propensity score-matched cohort was divided into two subgroups: Cluster 1 (n=93) and Cluster 2 (n=79). After Bonferroni correction, TCM syndrome differentiation and treatment significantly improved survival in the Cluster 1 patient subgroup (corrected P=0.023).

Conclusion TCM syndrome differentiation and treatment is significantly associated with improved survival in patients with stage Ⅲ-Ⅳ colorectal cancer, and continuous treatment for more than 6 months yields better results. Patients with poorly differentiated tumors or those with characteristics such as low inflammation levels and no distant metastasis may be the preferred treatment population.