Background: In May 2017, atezolizumab and pembrolizumab (IO) received accelerated approval for first-line treatment of cisplatin-ineligible patients with mUC, irrespective of PDL1 test status. In June 2018, the FDA and EMA restricted IO to cisplatin-ineligible patients with PDL1 positive tumors based on early review of data from confirmatory trials which suggested decreased overall survival in patients with PDL1 negative tumors treated with IO. We assessed the impact of the FDA label change on clinical outcomes of mUC patients in routine care. Methods: We conducted a controlled interrupted time series analysis using the US Flatiron Health electronic health record-derived de-identified database. The study sample included patients from 280 cancer clinics nationwide diagnosed with mUC and compared patients potentially impacted by the label change (cisplatin ineligible patients initiating first-line IO or carboplatin-based chemotherapy) to a comparator group who would have been unaffected by the label change (patients initiating first-line cisplatin-based chemotherapy) from 01 April 2017 to 17 May 2018 (pre-label change) and 20 June 2018 to 01 March 2020 (post-label change), excluding a 30-day wash out period encompassing the time-period between the initial FDA safety alert (18 May 2018) and the official FDA label change (19 June 2018). We used Cox regression to estimate adjusted pre-/post-label change related mortality differences in patients receiving carboplatin-chemotherapy or IO, accounting for secular changes in survival through comparison with the cisplatin comparator group. Results: The study included 829 patients with mUC initiating treatment in the pre-label change period (582 IO or carboplatin, 247 cisplatin) and 1,184 patients in the post-label change period (849 IO or carboplatin, 336 cisplatin), respectively. The use of IO, carboplatin, and cisplatin was similar across time-periods (pre-label change: 44.4%, 25.8%, and 29.8%; post-label change: 48%, 23.6%, 28.4%); while PD-L1 testing increased (6.6% to 28.1%). In adjusted models, there were no differences in survival in any of the groups following the FDA label change policy (table). Conclusions: The U.S. FDA label restriction on first-line immunotherapy was associated with increased PD-L1 testing but was not associated with changes in treatment patterns or improved mortality among patients with mUC.

*Interaction HR, corresponding to the ratio of ratios on the HR scale, an estimate of the multiplicative difference in the pre/post change between treatment groups.