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.
Paraneoplastic syndromes can commonly occur due to lung cancer, especially small cell lung cancer. Frequently paraneoplastic syndromes can precede the diagnosis of the neoplasm or present with limited stage disease. However, these syndromes can also occur at the time of recurrence or metastasis of disease. This review focuses on the epidemiology, pathogenesis, clinical features, and current management of the most common paraneoplastic syndromes encountered in patients with small cell lung cancer. Manifestations of paraneoplastic syndromes in small cell lung cancer include endocrine syndromes with secretion of excess hormones, and neurologic syndromes due to the production of antibodies causing an autoimmune condition. Recent advances have allowed for greater understanding of these syndromes and for the development of improved diagnostic as well as therapeutic tools. Awareness of paraneoplastic syndromes in small cell lung cancer can lead to an earlier diagnosis and recognition of both the condition and in some cases the disease potentially improving the overall survival and prognosis for patients. Further research examining effective methods to improve recovery from neurologic deficits in patients with a paraneoplastic neurologic illness is warranted.
Glioblastoma is the most common primary malignant brain tumor in adults, but effective therapies are lacking. With the scarcity of positive phase III trials, which are increasing in cost, we examined the ability of positive phase II trials to predict statistically significant improvement in clinical outcomes of phase III trials. A PubMed search was conducted to identify phase III clinical trials performed in the past 25 years for patients with newly diagnosed or recurrent glioblastoma. Trials were excluded if they did not examine an investigational chemotherapy or agent, if they were stopped early owing to toxicity, if they lacked prior phase II studies, or if a prior phase II study was negative. Seven phase III clinical trials in newly diagnosed glioblastoma and 4 phase III clinical trials in recurrent glioblastoma met the inclusion criteria. Only 1 (9%) phase III study documented an improvement in overall survival and changed the standard of care. The high failure rate of phase III trials demonstrates the urgent need to increase the reliability of phase II trials of treatments for glioblastoma. Strategies such as the use of adaptive trial designs, Bayesian statistics, biomarkers, volumetric imaging, and mathematical modeling warrant testing. Additionally, it is critical to increase our expectations of phase II trials so that positive findings increase the probability that a phase III trial will be successful.