Machine Learning Applications in Gallbladder Cancer Prognosis

By: Julia Fiederlein
Posted: Friday, September 30, 2022

Gallbladder cancer has been found to be associated with a high risk of recurrence and mortality. Thus, Timothy M. Pawlik, MD, PhD, MPH, of The Ohio State University Wexner Medical Center, Columbus, and colleagues used a machine-based learning approach to preoperatively stratify patients into distinct prognostic groups. The results of this analysis were published in the journal HPB.

“Characterizing patient prognosis with machine learning tools may help physicians provide more patient-centered care,” the investigators concluded.

Using a multi-institutional database, the investigators identified patients undergoing curative-intent resection of gallbladder cancer. Based on a classification and regression tree analysis, tumor size, biliary drainage, carbohydrate antigen 19-9 (CA 19-9) levels, and neutrophil-lymphocyte ratio were identified as the factors most strongly associated with overall survival. Approximately two-thirds of patients (66.9%) were female, and 72% were White. In addition, 86.8% of patients had cholecystectomy, with 84.9% having achieved an R0 margin status. Almost all patients (95.9%) had not received neoadjuvant chemotherapy.

Machine learning classified patients into four prognostic groups: neutrophil-lymphocyte ratio ≤ 1.5, CA 19-9 level ≤ 20 U/mL, no drainage, and tumor size < 5.0 cm (group 1; n = 109); neutrophil-lymphocyte ratio > 1.5, CA 19-9 level ≤ 20 U/mL, no drainage, and tumor size < 5.0 cm (group 2; n = 88); CA 19-9 level > 20 U/mL, no drainage, and tumor size < 5.0 cm (group 3; n = 46); tumor size < 5.0 cm with drainage or tumor size ≥ 5.0 cm (group 4; n = 77). The median duration of overall survival decreased incrementally with classification and regression tree group designation (59.5, 27.6, 20.6, and 12.1 months, respectively; P < .0001).

“The use of artificial intelligence and machine learning to develop risk stratification models may help identify high-risk patients who may benefit the most from neoadjuvant and adjuvant chemotherapy,” the study authors concluded.

Disclosure: For full disclosures of the study authors, visit hpbonline.org.

HPB

---