Prognostic Significance of Tumor-Infiltrating Lymphocytes Determined Using LinkNet on Colorectal Cancer Pathology...

Area of Science:

• Oncology
• Computational Pathology
• Artificial Intelligence

Background:

• Tumor-infiltrating lymphocytes (TILs) are crucial prognostic markers in various cancers.
• Automated, deep learning-based methods for TIL quantification in colorectal cancer (CRC) are scarce.

Purpose of the Study:

• To develop an automated deep learning workflow for quantifying TILs in CRC.
• To evaluate the prognostic value of automated TIL scores for disease progression and overall survival (OS).

Main Methods:

• A multiscale LinkNet workflow was developed for TIL quantification using H&E-stained CRC images.
• The model's performance was validated on two independent datasets (TCGA and MCO) comprising 1,684 CRC patients.

Main Results:

• The LinkNet model achieved high performance with an F1 score of 0.9347.
• High TIL abundance was significantly associated with reduced risk of disease progression (up to 75%) and improved OS (30-54%) in both cohorts.
• TIL scores demonstrated prognostic significance across various patient subgroups.

Conclusions:

• The automated LinkNet workflow provides a reliable tool for TIL quantification in CRC.
• Automated TIL scores are independent predictors of disease progression and OS, offering value beyond existing clinical factors.