Machine Learning Algorithm-Based Tumor-Stroma Ratio Can Stratify the Prognosis of Intrahepatic Cholangiocarcinoma

Area of Science:

• Oncology
• Computational Pathology
• Biomarker Discovery

Background:

• Intrahepatic cholangiocarcinoma (iCCA) prognosis after radical surgery requires refined stratification.
• Accurate quantification of tumor microenvironment components is crucial for understanding iCCA behavior.

Purpose of the Study:

• To quantify tumor components using machine learning for iCCA prognosis.
• To identify effective biomarkers for stratifying iCCA patient outcomes post-surgery.
• To explore predictive values for adjuvant chemotherapy response in iCCA.

Main Methods:

• A cohort of 237 iCCA patients undergoing radical resection was analyzed.
• A semiautomated pipeline quantified tumor, lymphocyte, and stromal cells.
• Tumor-stroma ratio and tumor-infiltrated lymphocytes ratio were calculated and correlated with survival (OS, DFS).

Main Results:

• High stroma and low tumor-infiltrated lymphocytes ratio % correlated with shorter DFS and OS in iCCA patients.
• Machine learning-based nomograms outperformed existing staging systems (AJCC, LSG-Japan).
• Patients with a low-stroma ratio showed greater benefit from adjuvant chemotherapy for DFS and OS.

Conclusions:

• The tumor-stroma ratio, quantified via machine learning, is a potent prognostic biomarker for iCCA post-radical surgery.
• This ratio may predict therapeutic response to adjuvant chemotherapy in iCCA patients.
• Machine learning offers novel tools for iCCA biomarker development and patient stratification.