Background: Atezolizumab (atezo) + bevacizumab (bev) + chemo (carboplatin + paclitaxel [CP]; ABCP) showed improved PFS and OS vs bev + CP (BCP) in pts with chemo-naive NSCLC (IMpower150). Benefit with ABCP vs BCP extended to key subgroups, including pts with baseline (BL) liver mets, which is a poor prognostic factor in metastatic NSCLC. Similar outcomes were not seen with atezo + chemo (IMpower150 [atezo + CP; ACP]; IMpower130; IMpower132), suggesting that the addition of bev to atezo + chemo is important for conferring clinical benefit in these pts. Here we further explore characteristics and responses of pts with BL liver mets in IMpower150. Methods: 1202 ITT pts were randomized 1:1:1 to receive ABCP, ACP or BCP. Doses were: A, 1200 mg; B, 15 mg/kg; C, AUC 6 mg/mL/min; P, 200 mg/m². Coprimary endpoints were OS and investigator-assessed PFS in ITT–wild-type pts. Exploratory analyses included efficacy and safety in pts with liver mets. Results: The data capture ≥ 20-mo follow-up in ITT pts (data cutoff: Jan 22, 2018). 162 pts had BL liver mets (ABCP, n = 52; ACP; n = 53; BCP, n = 57), with a median of 3 metastatic sites and median BL tumor SLD of 109 mm (range, 10-249). BL characteristics in these pts were generally balanced across study arms. PFS and OS were improved with ABCP vs BCP (Table). Gr 3-4 treatment-related AEs occurred in 52.1%, 36.5% and 54.5% of pts with liver mets in the ABCP, ACP and BCP arms, respectively. Conclusions: ABCP reduced the risk of death in pts with liver mets by 48% vs BCP and may represent an important new treatment option for this population. Clinical trial information: NCT02366143

^a Pts with measurable disease at BL