Background: Poorly differentiated, extrapulmonary small cell NEC are rare cancers with median overall survival (OS) < 1 year. Treatment is extrapolated from small cell lung cancer (SCLC) with use of P (cisplatin or carboplatin) + E. More effective treatment regimens and predictive biomarkers are needed to improve outcomes. In SCLC, induction therapy with combination of P/E + PD-L1 checkpoint inhibitor atezolizumab and maintenance atezolizumab improved OS (12.3 months vs 10.3 months; HR 0.70, 95% CI 0.54 – 0.91, P = 0.007) vs P/E alone (Horn L, et al. N Engl J Med 2018). No study to date has compared PD-1/PD-L1 inhibition during induction only vs during induction and maintenance therapy. In SCLC, distinct molecular subtypes can be identified by the presence of specific transcription factors (e.g., ASCL1, NEUROD1, POU2F3) or an Inflamed gene signature (SCLC-I), with SCLC-I pts more likely to benefit clinically from the addition of immunotherapy (Gay CM, et al. Cancer Cell 2021). In this study we plan to test the benefit of adding atezolizumab to induction P/E plus maintenance vs P/E alone, as well as the role of adding maintenance atezolizumab vs observation after induction chemo-immunotherapy. We also plan to correlate tumor- and blood-based subtype biomarkers with response to therapy. Methods: Eligible pts ≥18 years old have evaluable, histologically confirmed extrapulmonary small cell NEC, Zubrod PS ≤2, and are allowed to have up to 1 cycle of P/E prior to enrollment. P (cisplatin 75 mg/m2 or carboplatin AUC 5, iv) on day 1, E 100 mg/m2 iv on days 1-3, and atezolizumab 1200 mg iv on day 1 of q21 day cycles. Treatment consists of an induction phase x 4 cycles, and if no disease progression, a maintenance/observation phase given until disease progression for up to 1 year. Pts are randomized to 1 of 3 arms: A) Induction P/E + atezolizumab → maintenance atezolizumab B) Induction P/E + atezolizumab → observation C) Induction P/E → observation. The primary endpoint is to compare the OS (from randomization) between arms in a fixed sequence: A vs C → B vs C → A vs B. Secondary endpoints include comparing OS (from start of maintenance/observation), progression free survival, response rate, duration of response, and safety/tolerability across arms. Tumor and blood samples will be banked for future biomarker analyses, including immunohistochemistry of transcription factors on tissue and whole exome sequencing on tumor and circulating tumor DNA. With 189 pts, the study is powered to detect an improvement in 12-months OS from 35% to 57.5% (HR 0.53). Both phase 2 and phase 3 portions include interim analyses. Accrual will not pause for phase 2 analysis, expected early 2024. This study was activated December, 2021 and is open to accrual across the NCI National Clinical Trials Network (NCTN). Clinical trial information: NCT05058651.