Although machine learning (ML) has shown promise across disciplines, out-of-sample generalizability is concerning. This is currently addressed by sharing multi-site data, but such centralization is challenging/infeasible to scale due to various limitations. Federated ML (FL) provides an alternative paradigm for accurate and generalizable ML, by only sharing numerical model updates. Here we present the largest FL study to-date, involving data from 71 sites across 6 continents, to generate an automatic tumor boundary detector for the rare disease of glioblastoma, reporting the largest such dataset in the literature (n = 6, 314). We demonstrate a 33% delineation improvement for the surgically targetable tumor, and 23% for the complete tumor extent, over a publicly trained model. We anticipate our study to: 1) enable more healthcare studies informed by large diverse data, ensuring meaningful results for rare diseases and underrepresented populations, 2) facilitate further analyses for glioblastoma by releasing our consensus model, and 3) demonstrate the FL effectiveness at such scale and task-complexity as a paradigm shift for multi-site collaborations, alleviating the need for data-sharing.
- MeSH
- big data * MeSH
- glioblastom * MeSH
- lidé MeSH
- šíření informací MeSH
- strojové učení MeSH
- vzácné nemoci MeSH
- Check Tag
- lidé MeSH
- Publikační typ
- časopisecké články MeSH
- Research Support, N.I.H., Extramural MeSH
Neo-angiogenesis represents an important factor for the delivery of oxygen and nutrients to a growing tumour, and is considered to be one of the main pathodiagnostic features of glioblastomas (GBM). Anti-angiogenic therapy by vascular endothelial growth factor (VEGF) blocking agents has been shown to lead to morphological vascular normalisation resulting in a reduction of contrast enhancement as seen by magnetic resonance imaging (MRI). Yet the functional consequences of this normalisation and its potential for improved delivery of cytotoxic agents to the tumour are not known. The presented study aimed at determining the early physiologic changes following bevacizumab treatment. A time series of perfusion MRI and hypoxia positron emission tomography (PET) scans were acquired during the first week of treatment, in two human GBM xenograft models treated with either high or low doses of bevacizumab. We show that vascular morphology was normalised over the time period investigated, but vascular function was not improved, resulting in poor tumoural blood flow and increased hypoxia.
- MeSH
- bevacizumab farmakologie MeSH
- glioblastom patologie MeSH
- inhibitory angiogeneze farmakologie MeSH
- lidé MeSH
- myši nahé MeSH
- nádory mozku patologie MeSH
- patologická angiogeneze patologie MeSH
- xenogenní modely - testy antitumorózní aktivity MeSH
- zvířata MeSH
- Check Tag
- lidé MeSH
- mužské pohlaví MeSH
- ženské pohlaví MeSH
- zvířata MeSH
- Publikační typ
- časopisecké články MeSH
- práce podpořená grantem MeSH
BACKGROUND: Amplification of the epidermal growth factor receptor (EGFR) and its mutant EGFRvIII are among the most common genetic alterations in glioblastoma (GBM), the most frequent and most aggressive primary brain tumor. METHODS: In the present work, we analyzed the clonal evolution of these major EGFR aberrations in a small cohort of GBM patients using a unique surgical multisampling technique. Furthermore, we overexpressed both receptors separately and together in 2 patient-derived GBM stem cell lines (GSCs) to analyze their functions in vivo in orthotopic xenograft models. RESULTS: In human GBM biopsies, we identified EGFR amplification as an early event because EGFRvIII mutations emerge from intratumoral heterogeneity later in tumor development. To investigate the biological relevance of this distinct developmental pattern, we established experimental model systems. In these models, EGFR(+) tumor cells showed activation of classical downstream signaling pathways upon EGF stimulation and displayed enhanced invasive growth without evidence of angiogenesis in vivo. In contrast, EGFRvIII(+) tumors were driven by activation of the prototypical Src family kinase c-Src that promoted VEGF secretion leading to angiogenic tumor growth. CONCLUSIONS: The presented work shows that sequential EGFR amplification and EGFRvIII mutations might represent concerted evolutionary events that drive the aggressive nature of GBM by promoting invasion and angiogenesis via distinct signaling pathways. In particular, c-SRC may be an attractive therapeutic target for tumors harboring EGFRvIII as we identified this protein specifically mediating angiogenic tumor growth downstream of EGFRvIII.
- MeSH
- analýza přežití MeSH
- erbB receptory genetika metabolismus MeSH
- glioblastom diagnostické zobrazování genetika metabolismus patologie MeSH
- invazivní růst nádoru MeSH
- lidé MeSH
- molekulární evoluce MeSH
- mozek diagnostické zobrazování patologie MeSH
- multimodální zobrazování MeSH
- mutace MeSH
- nádorové buněčné linie MeSH
- nádory mozku diagnostické zobrazování genetika metabolismus patologie MeSH
- patologická angiogeneze metabolismus MeSH
- protoonkogenní proteiny pp60(c-src) metabolismus MeSH
- upregulace MeSH
- Check Tag
- lidé MeSH
- Publikační typ
- časopisecké články MeSH