1205P - A retrospective machine learning-based analysis of nationwide cancer CGP data across cancer types to identify features associated with recommendation of mutation-based therapy

Background

The probability of discovering mutation-based therapy through comprehensive genomic profiling (CGP) remains low. To enhance the effectiveness and efficiency of precision medicine, it is crucial to identify patients who are likely to benefit from CGP tests. This study aimed to identify characteristics of patients in which mutation-based treatments are discovered by CGP tests.

Methods

We retrospectively analyzed data from 60,655 patients registered in the Center for Cancer Genomics and Advanced Therapeutics (C-CAT) database, which covers 99.7% of cancer patients who have undergone CGP tests in Japan. We developed an eXtreme Gradient Boosting (XGBoost) machine learning model, and used clinical information as input to predict the likelihood of discovering mutation-based drugs through CGP tests. SHapley Additive exPlanations (SHAP) was employed to extract significant features contributing to the model prediction.

Results

The prediction model achieved an area under the receiver operating characteristic curve of 0.819 for the overall cancer population. Positive SHAP values were observed for patients with breast (mean SHAP in breast cancer patients: 1.66), lung (1.19), prostate (0.81), and colorectal (0.17) cancers, while negative SHAP values were associated with pancreatic (-2.40), brain (-1.37), and biliary tract (-0.44) cancers. Positive SHAP values were also associated with the presence of distant metastases and advanced age. Similar trends were observed in cancer type-specific prediction models. Distant metastasis was also associated with discovery of mutation-based therapy in the breast cancer-specific model, even after excluding liquid CGP data. In the adolescent and young adult (AYA) group, brain and bone tumors were associated with negative SHAP values.

Conclusions

Our analysis identified features that predict cases in which mutation-based treatments are discovered by CGP tests, both in the overall cancer population and within specific cancer types and the AYA group. Expedited CGP testing is recommended for patients who match the identified profile to facilitate early targeted therapy interventions.

Clinical trial identification

Editorial acknowledgement

Legal entity responsible for the study

The authors.

Funding

Has not received any funding.

Disclosure

K. Watanabe, H. Kage: Financial Interests, Institutional, Research Funding: Konica Minolta. K. Oda: Financial Interests, Personal, Advisory Role: Chugai Pharma, Takeda, AstraZeneca/Merck, AstraZeneca, Konica Minolta; Financial Interests, Institutional, Research Funding: AstraZeneca/Merck, Konica Minolta. All other authors have declared no conflicts of interest.