9517 Background: Initial results from ADAPT-IT (NCT03122522) demonstrated that adaptive dosing of nivolumab (nivo) and ipilimumab (ipi) based on early radiographic assessment was associated with comparable response rates and toxicity compared to historical data for conventional nivo + ipi. Here, we report the final clinical analysis, including median progression-free survival (PFS), and present the first correlative analyses of blood-based biomarkers of response. Methods: ADAPT-IT was a multicenter, phase 2 clinical trial with a planned sample size of 60 patients with unresectable melanoma. Patients received two doses of nivo (1 mg/kg) + ipi (3 mg/kg) followed by a computed tomography scan at week 6. Patients with early favorable antitumor effect (FATE; defined as the absence of new lesions and ≤ 4% index lesion tumor growth), discontinued ipi and continued nivo alone. Patients without FATE at week 6 received third and fourth doses of nivo + ipi followed by nivo alone. The primary endpoint was investigator-assessed overall response rate (ORR; defined as the percent of patients with complete or partial responses by RECIST v1.1) at week 12. A Simon two-stage design was employed where a 43% 12-week ORR was considered not promising, a 60% 12-week ORR was considered promising, and the probabilities of type I and type II error were each set at 0.10. Pre- and on-treatment peripheral blood samples were assayed for 1) ten cytokines using a multiplex immunoassay (Meso Scale Diagnostics) and 2) circulating tumor DNA (ctDNA) using the MSK-ACCESS sequencing platform. The Wilcoxon rank sum test and Cox regression modeling were used to evaluate associations between biomarker levels with radiographic response and PFS, respectively. Results: 60 patients were enrolled. Week 12 and best ORR were 48% (95% CI, 35-61%) and 58% (95% CI, 45-71%), respectively. Median follow-up was 34 months (interquartile range: 21-36) in survivors, median PFS was 21 months (95% CI: 11-not reached) and the median overall survival was not reached. There were no new Grade 3-5 treatment-related adverse events since the initial report. Higher baseline IL-6 was observed in non-responders compared to responders (p = 0.03; n = 26) and was associated with shorter PFS (HR for progression or death 1.24; 95% CI: 1.01-1.52; p = 0.04). Among 20 patients with detectable ctDNA at baseline (of 25 patients with available samples), an increase in mean adjusted variant allele fraction from baseline to week 6 was associated with shorter PFS (HR 1.14; 95% CI: 1.02-1.27; p = 0.02). Conclusions: In this updated analysis, now with long-term follow-up, the efficacy of adaptively dosed nivo + ipi resembles that of historical data for standard dose nivo + ipi. Baseline IL-6 and on-treatment changes in ctDNA warrant further prospective study as biomarkers of response to ipi + nivo. Clinical trial information: NCT03122522 .