Background: The IMvigor130 primary analysis demonstrated statistically significant PFS benefit with 1L atezo + plt/gem (Arm A) vs placebo + plt/gem (Arm C) in pts with mUC (Galsky Lancet 2020). Interim data showed improved OS with Arm A vs C but did not cross the pre-specified threshold for significance (Galsky Lancet 2020; Galsky AACR 2021). In exploratory analyses, OS improved when atezo was combined with cisplatin (cis) vs carboplatin (carbo) regardless of PD-L1 status. Here, we report final OS data from Arms A and C. Methods: Pts were randomly assigned 1:1:1 to Arm A, B (atezo alone) or C. Arm A and C pts received cis or carbo per investigator (INV) choice. Co-primary endpoints were PFS per INV RECIST 1.1 and OS (Arm A vs C), and OS (Arm B vs C), tested hierarchically. Safety, ORR and DOR, disease control rate (DCR; confirmed CR, PR or [SD ≥ 6 mo]) and pre-specified exploratory OS data are also reported. Results: As of data cutoff Aug 31, 2022 (49 mo since last pt randomly assigned), in ITT pts, OS benefit was not statistically significant; in the cis subgroup, HR was 0.76 (95% CI 0.57, 1.01; Table). In ITT pts, DCR was 65% (290/447) in Arm A and 60% (239/397) in Arm C. 81% of safety-evaluable pts (370/454) in Arm A and 80% (312/389) in Arm C had a Gr 3/4 tx-related AE (TRAE). Gr 5 TRAEs occurred in 9 pts (2%) in Arm A and 4 (1%) in Arm C; Gr 3/4 AEs of special interest were seen in 41 pts (9%) in Arm A and 17 (4%) in Arm C. Conclusions: In this final analysis, improved OS with atezo + plt/gem vs placebo + plt/gem did not reach statistical significance in ITT pts with mUC. As seen with prior exploratory data, improved OS with atezo + plt/gem was greater when pts received cis vs carbo. No new safety signals were seen. Clinical trial information: NCT02807636.

95% CI are in parentheses. OS event rates: Arm A, 75%; Arm C, 78%. IC, tumor-infiltrating immune cells. ^a Stratified by PD-L1 status, Bajorin risk score, INV choice of cis vs carbo, enrollment stage. ^b 1-sided; not statistically significant (final OS efficacy boundary: P=0.021; adjusted from prior analyses).^c Stratified by enrollment stage.^d PD-L1–expressing IC on ≥5% (IC2/3) or <5% (IC0/1) of tumor area (VENTANA SP142 assay). ^e In pts with baseline measurable disease.