Real-world adoption of PD-L1 testing in metastatic urothelial carcinoma.

e19184 Background: The role of PD-L1 testing in locally advanced (LA) or metastatic urothelial carcinoma (mUC) is not currently well-defined, except in cisplatin-ineligible patients (pts). IMVigor130 trial data are the first to suggest cisplatin-eligible PD-L1 positive pts may also benefit from immunotherapy (IO) (ESMO LBA14_PR, 2019). As clinical data accumulate on the influence of PD-L1 expression on LA/mUC pt outcomes with IO, we assessed medical oncologists’ (ONCs) PD-L1 testing perceptions and practices. Methods: ONCs completed a longitudinal, online self-report survey fielded from August 22 to September 23, 2019, prior to release of IMVigor130 data. Baseline results are reported. Descriptive statistics and bivariate comparisons by practice setting were conducted. Results: ONCs (N = 203) had a mean of 15 years in practice; a majority (64%) practice in the community. ONCs tested most LA/mUC pts, but initiated treatment for about a third before receiving results (Table). Half believe PD-L1 testing in LA/mUC will increase over time; this view was more common for ONCs who do not (n = 55) vs. do (n = 148) order PD-L1 testing as standard for LA/mUC pts (55% vs. 49%). Most ONCs reported that pts in an IO + chemotherapy trial arm should be stratified by PD-L1 status (74%), but not platinum agent (69%). Common barriers to more LA/mUC pts (n = 92) getting a PD-L1 test were limited biopsy sample (54%), lack of influence on treatment decisions (54%), and treatment initiation urgency (30%). ONCs varied little by practice setting. Conclusions: PD-L1 testing has been adopted by ONCs for most of their LA/mUC pts. However, ONCs report limited biopsy sample and lack of influence on treatment decisions as barriers to testing more LA/mUC pts, supporting their initiation of therapy prior to receipt of results and desire for trial data to be stratified by PD-L1 status. Results suggest PD-L1 expression is not yet a key driver in treatment decisions. Given the variability in self-reported responses, results should be interpreted with this limitation in mind. [Table: see text]