KEY TAKEAWAYS
- The ESME-MBC research aimed to develop a deep learning model to predict individual outcomes and customize treatments for MBC.
- The study’s primary endpoint was OS.
- The models showed promising results in predicting OS for MBC pts, leading to better identification of those requiring more advanced treatment options.
The study focused on developing a deep-learning model to determine individual outcomes and create personalized treatments for metastatic breast cancer (MBC).
Researchers utilized data from the Unicancer large national multicenter real-world Epidemiological Strategy and Medical Economics (ESME) database to determine the overall survival (OS) of 27,855 women diagnosed with MBC between 2008 and 2020. To ensure accuracy in measuring this outcome and tracking the disease, treatment line (TL) initiation dates were used as crucial time points, and a dynamic deep-learning survival model was developed. This model extended a Long-Short Term Memory cell, a recurrent neural network that can handle sequential data with irregular timing. Through the use of the time-dependent concordance index (C-index) and Integrated Brier Score (IBS), the best models were selected.
The research cohort comprised 4,857 patients (pts) with triple-negative (TN), 5,027 with HER2+, and 17,971 with hormonal receptor positive (HR+)/HER2- MBC. The median follow-up period lasted 65.1 months (95%CI 63.8, 66.4). The first three TLs were evaluated using 5-fold cross-validation cohorts to determine the areas under the receiving operator characteristics (AUCs) for predicting OS at six months and one year, sensitivity at 90% specificity for 6-month OS prediction, and global C-index and IBS per subtype. The results were then averaged to overview the model’s performance comprehensively.
The models showed a significant breakthrough in predicting OS for MBC pts, which is crucial to identify those who require more advanced treatment options. This deep-learning survival model has a strong prognostic capacity that could inform treatment selection for MBC pts.
Clinical Trial: https://classic.clinicaltrials.gov/ct2/show/NCT03275311
Vuduc, L., Jacot, W., Frenel, J., Brain, E.G.C., Dieras, V.C., Bachelot, T., Mailliez, A., Dalenc, F., Cottu, P.H., Arnedos, M., Lefeuvre-Plesse, C., Gonçalves, A., Grinda, T., Antoine, A., Chevrot, M., Perol, D., Cournede, P., Delaloge, S. Annals of Oncology (2023) 8 (1suppl_4): 101223-101223. 10.1016/esmoop/esmoop101223.
