The 2023 EAU annual meeting included a session on locally advanced kidney cancer, featuring a presentation by Dr. Zine-Eddine Khene discussing artificial intelligence to predict recurrence after surgical resection of non-metastatic renal cell carcinoma (RCC). Predictive tools can be useful to adapt surveillance or include patients in adjuvant trials after surgical resection of non-metastatic RCC. Current models have been built using traditional statistical modeling and prespecified variables, which limits their performance. The aim of this study was to investigate the performance of machine learning framework to predict recurrence after RCC surgery and compare them with current validated models.
In this observational study, Dr. Khene and colleagues derived and tested an ensemble of machine learning-based models (Random Survival Forests [RSF], Survival Support Vector Machines [S-SVM], and Extreme Gradient Boosting [XG boost]) of patients who underwent radical or partial nephrectomy for a non-metastatic RCC, between 2013 and 2020, at 21 French medical centers using standard clinicopathological variables. The primary endpoint of prediction was disease-free survival. Model discrimination was assessed using the concordance index (c-index), and calibration was assessed using the Brier score. machine learning models were compared with four conventional prognostic models, using decision curve analysis.
There were 4,067 patients included in this study. Machine learning models obtained higher c-index values than conventional models. The RSF yielded the highest c-index values (0.794) followed by S-SVM (c-index 0.784) and XG boost (c-index 0.782). In addition, all models showed good calibration with low integrated Brier scores (all integrated brier scores <0.1). However, they found calibration drifted over time for all models, albeit with a smaller magnitude for machine learning models. Finally, decision curve analysis showed an incremental net benefit from all machine learning models compared to conventional models currently used in practice.
Dr. Khene concluded this presentation discussing artificial intelligence to predict recurrence after surgical resection of non-metastatic renal cell carcinoma with the following take-home messages:
- Applying machine learning approaches to predict recurrence following surgical resection of RCC resulted in better prediction than that of current validated models available in clinical practice
- However, there is still room for improvement that may come from the integration of novel biological and/or imaging biomarkers
Presented by: Zine-Eddine Khene, MD, PhD, Rennes, Dept. of Urology, Rennes, France
Co-Authors: Bigot P.2, Doumerc N.3, Albiges L.4, Bernhard J-C.5, Ouzaid I.6, Rioux Leclercq N.7, Roupret M.8, Bensalah K.1, Uro-CCR9
Affiliations: 1Rennes, Dept. of Urology, Rennes, France, 2Angers University Hospital, Dept. of Urology, Angers, France, 3Toulouse, Dept. of Urology, Toulouse, France, 4Gustave Roussy, Dept. of Oncology, Paris, France, 5Bordeaux, Dept. of Urology, Bordeaux, France, 6Bichat, Dept. of Urology, Paris, France, 7Rennes, Dept. of Pathology, Rennes, France, 8Pitié Salpêtrière Hospital, Dept. of Urology, Paris, France
Written by: Zachary Klaassen, MD, MSc – Urologic Oncologist, Assistant Professor of Urology, Georgia Cancer Center, Augusta University/Medical College of Georgia, @zklaassen_md on Twitter during the 2023 European Association of Urology (EAU) Annual Meeting, Milan, IT, Fri, Mar 10 – Mon, Mar 13, 2023.