Background: Cabozantinib inhibits multiple tyrosine kinases, including MET, VEGFR, RET, and TAM kinases (Tyro3, AXL, MER), involved in tumor growth and angiogenesis, and whose mutations and expression are associated with prostate cancer aggressiveness and poor prognosis. Targeting these kinases with cabozantinib may promote an immune permissive tumor environment and may enhance response to immune checkpoint inhibitors. In the ongoing phase 1b COSMIC-021 study of pts with solid tumors, cabozantinib plus the PD-L1 inhibitor atezolizumab, showed preliminary meaningful clinical activity in soft tissue disease and a tolerable safety profile for 44 pts with mCRPC (Agarwal et al., ASCO 2020; abstract 5564). We present the study design of a phase 3 trial of cabozantinib plus atezolizumab versus second NHT in pts with mCRPC. Methods: This randomized, open-label, controlled phase 3 study (NCT04446117) evaluates the efficacy and safety of cabozantinib plus atezolizumab versus second NHT (abiraterone or enzalutamide) in pts with mCRPC who previously received one NHT to treat metastatic castration-sensitive PC (mCSPC), non-metastatic CRPC (M0 CRPC), or mCRPC. Additional eligibility criteria include histologically or cytologically confirmed adenocarcinoma of the prostate, measurable visceral disease or measurable extrapelvic adenopathy per RECIST 1.1 by investigator, prostate specific antigen progression and/or soft-tissue disease progression, ECOG 0 or 1, and age ≥18 years. Key exclusion criteria include prior nonhormonal therapy for mCRPC and uncontrolled significant illness. Eligible pts (N = 580) are randomized 1:1 to receive cabozantinib (40 mg PO QD) + atezolizumab (1200 mg IV Q3W) vs abiraterone (1000 mg PO QD) + prednisone (5 mg PO BID) or enzalutamide (160 mg PO QD). Designated NHT will differ from previous NHT taken. Randomization is stratified by: liver metastasis (yes, no), prior docetaxel treatment for mCSPC (yes, no), and disease stage for which the first NHT was given (mCSPC, M0 CRPC, mCRPC). Treatment will continue until there is no longer clinical benefit as determined by the treating investigator, unacceptable toxicity, or consent withdrawal. The multiple primary endpoints are progression-free survival per RECIST 1.1 by blinded independent radiology committee (BIRC) and overall survival. Additional endpoints include objective response rate per RECIST 1.1 by BIRC, safety, correlation of biomarkers with outcomes, quality of life and pharmacokinetics. Patient enrollment is ongoing. Clinical trial information: NCT04446117