Tertiary Lymphoid Structures as Predictors of Recurrence in Colorectal Cancer: Development and Validation of a...

Area of Science:

• Oncology
• Immunology
• Computational Biology

Background:

• Tertiary lymphoid structures (TLS) are immune aggregates in the tumor microenvironment with prognostic value.
• The predictive significance of location-specific TLS features for colorectal cancer (CRC) recurrence is not well-established.
• This study aimed to develop and validate a machine-learning (ML) model integrating TLS features for CRC recurrence prediction.

Purpose of the Study:

• To develop and internally validate a machine-learning (ML) model for predicting recurrence in stage II-III colorectal cancer (CRC).
• To assess the value of location-specific tertiary lymphoid structures (TLS) features in predicting CRC recurrence.

Main Methods:

• Retrospective analysis of 224 stage II-III CRC patients, split into training (n=156) and validation (n=68) cohorts.
• Semi-quantitative scoring of intratumoral, invasive-front, and peritumoral TLS by pathologists.
• Machine-learning model development using LASSO selection, five algorithms, stratified cross-validation, and SHAP interpretation.

Main Results:

• The ML model, incorporating peritumoral TLS score, invasive-front TLS score, and CEA, achieved an AUC of 0.718 in internal validation.
• Invasive-front TLS score was identified as the most influential predictor by SHAP analysis.
• The model effectively stratified patients into high- and low-risk groups with significantly different disease-free survival (log-rank p < 0.0001).

Conclusions:

• A TLS-based ML model demonstrated moderate internal validation performance for predicting CRC recurrence.
• Location-specific TLS features show potential for postoperative risk stratification in CRC patients.
• External, multicenter, and molecularly integrated validation is necessary prior to clinical implementation.