PURPOSE The regimens approved for the treatment of advanced head and neck squamous cell carcinoma are accessible to only 1%-3% of patients in low- and middle-income countries because of their cost. In our previous study, metronomic chemotherapy improved survival in this setting. Retrospective data suggest that a low dose of nivolumab may be efficacious. Hence, we aimed to assess whether the addition of low-dose nivolumab to triple metronomic chemotherapy (TMC) improved overall survival (OS). METHODS This was a randomized phase III superiority study. Adult patients with recurrent or newly diagnosed advanced head and neck squamous cell carcinoma being treated with palliative intent with an Eastern Cooperative Oncology Group performance status of 0-1 were eligible. Patients were randomly assigned 1:1 to TMC consisting of oral methotrexate 9 mg/m2 once a week, celecoxib 200 mg twice daily, and erlotinib 150 mg once daily, or TMC with intravenous nivolumab (TMC-I) 20 mg flat dose once every 3 weeks. The primary end point was 1-year OS. RESULTS One hundred fifty-one patients were randomly assigned, 75 in TMC and 76 in the TMC-I arm. The addition of low-dose nivolumab led to an improvement in the 1-year OS from 16.3% (95% CI, 8.0 to 27.4) to 43.4% (95% CI, 30.8 to 55.3; hazard ratio, 0.545; 95% CI, 0.362 to 0.820; P = .0036). The median OS in TMC and TMC-I arms was 6.7 months (95% CI, 5.8 to 8.1) and 10.1 months (95% CI, 7.4 to 12.6), respectively ( P = .0052). The rate of grade 3 and above adverse events was 50% and 46.1% in TMC and TMC-I arms, respectively ( P = .744). CONCLUSION To our knowledge, this is the first-ever randomized study to demonstrate that the addition of low-dose nivolumab to metronomic chemotherapy improved OS and is an alternative standard of care for those who cannot access full-dose checkpoint inhibitors.

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.

No abstract

9001 Background: Standard first-line therapy for EGFR mutant advanced non-small cell lung cancer (NSCLC) is an EGFR-directed oral TKI. We evaluated whether adding pemetrexed-carboplatin to oral TKI would improve outcomes. Methods: Phase III randomized trial in advanced chemotherapy-naïve NSCLC harboring EGFR sensitizing mutation (exon 19, 21 or 18) with performance status (PS) 0 to 2 planned for palliative therapy. Patients were stratified for PS and EGFR mutation and randomly assigned (computer-generated randomization by independent biostatistician) 1:1 to gefitinib 250 mg orally daily (gef) or gefitinib 250 mg orally daily with pemetrexed 500 mg/m2 IV and carboplatin AUC 5 IV every 3 weeks for 4 cycles, followed by maintenance pemetrexed 500 mg/m2 IV every 3 weeks (gef+C). Restaging was every 2 to 3 mths; therapy continued until progression or intolerable toxicity. Primary end point was progression-free survival (PFS); secondary end points included overall survival (OS), toxicity and response rate. Survival endpoints were assessed in the intention-to-treat population. Results: Between Aug 2016 and Aug 2018, 350 patients were randomly assigned to gef (n = 177) and gef+C (n = 173). Median age was 54 yrs, 48% were females, 84% never-smokers, 21% were PS 2 and 18% had brain metastases. Median follow-up in surviving patients was 17 months (range, 7 to 30). Radiologic response rates were 81% and 69% in gef+C and gef respectively, P = 0.012. 234 patients (67%) have had events for PFS, 98 in gef+C and 136 in gef. Estimated median PFS was significantly longer with gef+C than gef (16 months, [95% CI, 13.7 to 18.3] vs. 8 months [95% CI, 7.1 to 8.9]; hazard ratio for disease progression or death, 0.5; 95% CI, 0.39 to 0.65; P < 0.001). 120 patients (34%) have died, 42 in gef+C and 78 in gef. Estimated median OS was significantly longer with gef+C than gef (not reached vs. 18 months [95% CI, 14.28 to 21.72]; hazard ratio for death, 0.45; 95% CI, 0.31 to 0.66; P < 0.001). Clinically relevant ≥ grade 3 toxicities occurred in 51% and 25% of patients in gef+C and gef arms respectively, P < 0.001. Conclusion: Adding pemetrexed-carboplatin chemotherapy to gefitinib significantly prolonged progression free and overall survival but also increased toxicity. Pemetrexed-carboplatin-gefitinib represents a new standard first-line therapy for EGFR mutant NSCLC. Clinical trial information: CTRI/2016/08/007149.