Rethinking TNM: tumor deposit-based prognostic models may improve N-staging in colorectal cancer

Abstract Introduction Tumor deposits (TDs) are an important negative prognostic factor for long-term oncological outcomes in colorectal cancer patients, independently of lymph node status. Several novel models have been proposed to further integrate TDs into the TNM-staging system, but their comparative performance remains unclear. The aim of this study was to identify, compare and validate prognostic models incorporating TDs and/or TD-count for N-stage classification. Method A scoping literature review was conducted to identify novel prognostic models that further incorporated TDs or TD-count into N-staging. The identified models was validated using patient data from the Swedish Colorectal Cancer Registry, assessing overall survival, distant metastasis, and local recurrence. Prognostic performance was compared to the TNM N-staging using Kaplan-Meier curves for visual analysis, Harrell’s C-index for discriminative ability, and Bayesian information criterion for model fit and complexity. Result Of 792 articles, 17 studies met the inclusion criteria, resulting in 10 unique models in addition to TNM. For the patient cohort, 26,970 patients remained after exclusion, of whom 3,312 (12%) had TDs. Comparative analysis showed all models were superior to TNM, with two models standing out; an integrated model combining lymph node count and TD-count, and a ratio model considering number of TDs, positive lymph nodes, and total extracted nodal structures. All models provided prognostic value, but differences were modest. Discussion This study demonstrated that although all models outperformed TNM, prognostic differences between the models were small. While TDs provide valuable prognostic information for high-risk patients, additional risk variables need to be included to further enhance the staging system.