Machine learning-based prognostic subgrouping of glioblastoma: A multicenter study
Hamed Akbari, Spyridon Bakas, Chiharu Sako, Anahita Fathi Kazerooni, Javier Villanueva-Meyer, José García, Elizabeth Mamourian, Fang Liu, Quy Cao, Russell T. Shinohara, Ujjwal Baid, Alexander Getka, Sarthak Pati, Ashish Singh, Evan Calabrese, Susan Chang, Jeffrey D. Rudie, Aristeidis Sotiras, Pamela LaMontagne, Daniel S. Marcus, Mikhail Milchenko, Arash Nazeri, Carmen Balañá, Jaume Capellades, Josep Puig, Chaitra Badve, Jill S. Barnholtz‐Sloan, Andrew E. Sloan, Vachan Vadmal, Kristin Waite, Murat Ak, Rivka R. Colen, Yae Won Park, Sung Soo Ahn, Jong Hee Chang, Yoon Seong Choi, Seung‐Koo Lee, Gregory S. Alexander, Ayesha Ali, Adam P. Dicker, Adam E. Flanders, Spencer Liem, Joseph Lombardo, Wenyin Shi, Garima Shukla, Brent Griffith, Laila Poisson, Lisa R. Rogers, Aikaterini Kotrotsou, Thomas C. Booth, Rajan Jain, Matthew Lee, Abhishek Mahajan, Arnab Chakravarti, Joshua D. Palmer, D.J. DiCostanzo, Hassan M. Fathallah‐Shaykh, Santiago Cepeda, Orazio Santo Santonocito, Anna Luisa Di Stefano, Benedikt Wiestler, Elias R. Melhem, Graeme F. Woodworth, Pallavi Tiwari, Pablo A. Valdés, Yūji Matsumoto, Yoshihiro Otani, Ryoji Imoto, Mariam Aboian, Shinichiro Koizumi, Kazuhiko Kurozumi, Toru Kawakatsu, Kimberley L. Alexander, Laveniya Satgunaseelan, Aaron Rulseh, Stephen Bagley, Michel Bilello, Zev A. Binder, Steven Brem, Arati Desai, Robert A. Lustig, Eileen Maloney, Timothy J. Prior, Nduka Amankulor, Mac Lean P Nasrallah, Donald M. O’Rourke, Suyash Mohan, Christos Davatzikos, Christos Davatzikos
Abstract
BACKGROUND: Glioblastoma (GBM) is the most aggressive adult primary brain cancer, characterized by significant heterogeneity, posing challenges for patient management, treatment planning, and clinical trial stratification. METHODS: We developed a highly reproducible, personalized prognostication, and clinical subgrouping system using machine learning (ML) on routine clinical data, magnetic resonance imaging (MRI), and molecular measures from 2838 demographically diverse patients across 22 institutions and 3 continents. Patients were stratified into favorable, intermediate, and poor prognostic subgroups (I, II, and III) using Kaplan-Meier analysis (Cox proportional model and hazard ratios [HR]). RESULTS: The ML model stratified patients into distinct prognostic subgroups with HRs between subgroups I-II and I-III of 1.62 (95% CI: 1.43-1.84, P < .001) and 3.48 (95% CI: 2.94-4.11, P < .001), respectively. Analysis of imaging features revealed several tumor properties contributing unique prognostic value, supporting the feasibility of a generalizable prognostic classification system in a diverse cohort. CONCLUSIONS: Our ML model demonstrates extensive reproducibility and online accessibility, utilizing routine imaging data rather than complex imaging protocols. This platform offers a unique approach to personalized patient management and clinical trial stratification in GBM.