Background: In the SAUL study (NCT02928406), clinical outcomes in a broad population were similar to findings from phase 3 trials of anti-PD-(L)1 in highly selected pts. We report the final analysis 4 y after enrolling the last pt. Methods: Eligible pts had locally advanced/metastatic urothelial or non-urothelial UTC that had progressed during or after 1–3 prior lines of therapy for advanced UTC (or within 12 months of [neo]adjuvant therapy). Unlike most phase 3 immunotherapy trials, pts with autoimmune disease (AID), ECOG PS 2, creatinine clearance (CrCl) ≥15–30 mL/min, and/or stable CNS metastases were eligible. Pts took atezo 1200 mg q3w until disease progression or unacceptable toxicity. The primary endpoint was safety. Secondary endpoints included overall survival (OS) and duration of response (DoR). Results: Among 997 treated pts, 10% had ECOG PS 2, 5% had CrCl <30 mL/min, 4% had a history of AID, and 7% had 2–3 prior lines of therapy. At data cutoff (Dec 20, 2022; median follow-up 55 mo), 78% of pts had died. Median treatment duration was 2.8 mo (range 0–62 mo; mean 9.5 mo), 96 pts (10%) took atezo for >3 y, and 68 (7%) for >4 y. Grade ≥3 adverse events (AEs) occurred in 51%; AEs led to atezo discontinuation in 8%. We show outcomes overall and in key subgroups. OS was >4 y in 136 pts (14%); 132 responders had OS >2 y. Conclusions: After 55 mo median follow-up, median OS was 8.6 mo and 3-y OS was 21% in a real-world population including important understudied subgroups. Long-term safety data continue to support the tolerability of atezo in pts with complex comorbidities. AESIs were not increased in pts typically considered to be at increased risk of immune-related AEs. OS was similar to trials in highly selected pts. Ongoing analyses are characterizing pts with exceptional long-term outcomes. Clinical trial information: NCT02928406.

^an=997. ^bn=264. ^cn=360.

AESI = AE of special interest; PD-L1 IC2/3 = ≥5% PD-L1-expressing tumor-infiltrating immune cells; NE = not estimable.