144P - Integrated clinical and genomic models using machine-learning methods to predict the efficacy of paclitaxel-based chemotherapy in patients with advanced gastric cancer from K-MASTER project

Background

Paclitaxel is commonly used as second-line therapy in advanced gastric cancer (AGC). The decision to proceed with second-line chemotherapy and select a chemotherapy regimen may be critical in vulnerable AGC patients after progression with first-line chemotherapy. However, there are no predictive biomarkers to identify patients with AGC who benefit from paclitaxel-based chemotherapy.

Methods

This study included 288 patients with AGC receiving second-line paclitaxel-based chemotherapy between 2017 and 2022 from K-MASTER project, a nationwide, government-funded precision medicine initiative. The data included clinicogenomic factors: clinical (age [young-onset vs. others], sex, histology [intestinal vs. diffuse type], prior trastuzumab use, duration of first-line chemotherapy, etc.) and genomic factors (pathogenic or likely pathogenic variants). The data were randomly divided into training and test sets (0.8:0.2). Three machine-learning methods, including random forest (RF), logistic regression (LR), and artificial neural network with genetic embedding (ANN) models, were used to develop the prediction model and were validated in the test sets.

Results

The median age was 64 years (range, 25-91) and 65.6% were male. A total of 288 patients were divided into training (n=230) and test sets (n=58). There were no significant differences in baseline characteristics between training and test sets. In the training set, the AUC for prediction of progression-free survival (PFS) with paclitaxel-based chemotherapy was 0.51, 0.73, and 0.75 in RF, LR, and ANN models, respectively. In the test set, the Kaplan-Meier curves of PFS were separated according to the three models: 2.8 vs. 1.5 months (P=0.07) in RF, 2.3 vs. 6.5 months (P=0.07) in LR, and 2.1 vs. 7.6 months (P=0.02) in ANN models.

Conclusions

These machine-learning models integrated clinical and genomic factors and can guide the selection of patients with AGC with a greater likelihood of a benefit from second-line paclitaxel-based chemotherapy. Further studies are necessary to validate and update these models in independent datasets in future.

Clinical trial identification

Editorial acknowledgement

Legal entity responsible for the study

The authors.

Funding

Has not received any funding.

Disclosure

All authors have declared no conflicts of interest.