2607 Background: Mechanistic evidence suggests opioid signaling to modulate anti-tumor immunity. Whether opioids exposure may affect outcome of patients (pts) treated with immune checkpoint inhibitors is unproven, with clinical evidence being flawed by the negative associative bias between pain and higher burden of disease. Methods: We conducted a post-hoc analysis of the phase 3 OAK (NCT02008227) and phase 2 POPLAR (NCT01903993) trials, which randomized (1:1) pts with advanced NSCLC to receive either atezolizumab or docetaxel. We assessed the differential impact of early opioids exposure (EOE), defined as a minimum of 7 days exposure to any systemic opioids in the time window ranging between -30 to +30 from treatment initiation on efficacy from immunotherapy vs chemotherapy. Pts with ≤30 days survival follow-up were excluded to avoid immortal time bias. Results: After the exclusion of 60 pts, 718 and 686 pts treated with atezolizumab and docetaxel were included, with 341 (47.5%) and 308 (44.9%) pts with EOE respectively (p=0.32) and a median follow-up of 25.5 months (95% CI: 22.8-25.9). Baseline characteristics was balanced across cohorts. In the pooled population, EOE was significantly associated with poorer ECOG-PS (p<0.01) and burden of disease (p<0.01). Univariable analysis showed EOE to be significantly associated with decreased objective response rate (ORR, 11.7% vs 19.8%, p<0.01) and progression free survival (PFS, 1.9 vs 4.2 months, p<0.01) in the atezolizumab cohort, whereas no effect on ORR (17.7% vs 14.4%, p=0.25) and a less pronounced effect on PFS (3.5 vs 4.1 months, p=0.02) was reported for the docetaxel cohort. The pooled multivariable backward stepwise Cox regression used to select variables for validating the results individuated ECOG-PS, histology, tumor burden and race. In multivariable models EOE was associated with decreased ORR (OR 0.58, 95%CI: 0.37-0.92), increased risk of progression (HR 1.47, 95%CI: 1.26-1.72) and death (HR 1.70, 95%CI: 1.42-2.03) in the atezolizumab cohort while no negative impact on ORR/PFS was confirmed among the docetaxel cohort, with a less strong effect on OS (HR 1.44, 95%CI: 1.21-1.72). Multivariable interaction tests confirmed the differential impact of EOE on ORR (p<0.01) and PFS (p<0.01) between treatment modalities (immunotherapy vs chemotherapy). The results were additionally confirmed using IPTW models including all the baseline variables with an optimal distribution (SMD <0.05). Conclusions: EOE is associated with worse response/PFS from immunotherapy but not from chemotherapy exposure, suggesting a possible immune-modulating effect of opioids signaling on anti-tumor immunity. Considering that systemic opioids are the cornerstone of pain management in oncology further research for mitigation strategies, such as the potential use of PAMORAs, is needed. Clinical trial information: NCT02008227 and NCT01903993 .