To test, or not to test, that is the question: A real-world analysis of PD-L1 expression testing patterns in recurrent/metastatic head and neck squamous cell carcinoma (R/M HNSCC).

6033 Background: PD-L1 expression testing plays an important role in identifying patients who benefit from immunotherapy when treating R/M HNSCC. We aimed to understand how PD-L1 expression testing evaluated by Combined Positive Score (CPS) is used in clinical practice and how testing results impact treatment decision-making. Methods: A retrospective cohort study was conducted using the Flatiron Health Advanced Head and Neck database. We included adult R/M HNSCC patients who initiated first-line (1L) treatment between 07/01/19 and 12/31/21 with follow-up until 06/30/22. Patients were excluded if they received platinum therapy within 6 months prior to 1L treatment, had other primary cancers before advanced diagnosis, were treated on protocol, or were tested but did not have any record of date of specimen collected, received, or results returned. Multivariable logistic regression was conducted to assess factors associated with PD-L1 testing decisions before 1L therapy start and the association between testing results and choice of 1L treatment. Results: A total of 1,762 patients with R/M HNSCC were included, with 32.0% tested for PD-L1 prior to 1L therapy start, 25.6% tested after, and 42.4% never tested. Most patients were tested using the IHC 22C3 pharmDx assay, an FDA-approved companion diagnostic. Among patients tested before 1L treatment initiation, the most prescribed regimens were pembrolizumab or pembrolizumab + platinum + 5-FU for patients with PD-L1(+) (55.9%, 14.2%) and PD-L1(-) tumors (25.3%, 32.0%), respectively. Patients tested after 1L treatment initiation regardless of CPS and patients never tested were most often prescribed platinum monotherapy for 1L. In a multivariable model, patients who were older, treated in academic settings, HPV(+), or had an ECOG≥2 were more likely to get PD-L1 testing before 1L treatment initiation. Compared to CPS<1, CPS≥20 and 1≤CPS<20 were associated with a higher likelihood of receiving immunotherapy (OR=3.36, 95% CI, 1.69-6.75, p<0.001; OR=3.11, 95% CI, 1.59-6.14, p<0.001, respectively). Conclusions: As an important part of treatment decision-making, PD-L1 testing by CPS should guide 1L single-agent immunotherapy use, but there are many patients not tested before initiating 1L therapy. Age and fitness, clinical practice setting, and HPV status appear to impact testing patterns.[Table: see text]