"Off-Label" Use of Checkpoint Inhibitors in Patients With Negative or Unknown PD-L1 Status in Advanced Head and Neck Cancer.

Abstract

Background: The KEYNOTE-048 study established the checkpoint inhibitor (CPI) pembrolizumab, with/without chemotherapy, as frontline treatment for recurrent/metastatic (R/M) head and neck squamous cell carcinoma (HNSCC). However, pembrolizumab monotherapy has limited efficacy in PD-L1-negative disease. Clinical practice patterns regarding PD-L1 combined positive score (CPS) testing and PD-L1-guided treatment selection remain unknown.

Patients and methods: This retrospective analysis included patients who initiated treatment for R/M HNSCC from 2011 to 2023 in a nationwide electronic health record-derived deidentified database. Frontline therapy was categorized as CPI monotherapy, CPI with chemotherapy, or chemotherapy ± cetuximab without CPI. A subset of patients treated in 2019 and beyond (2019+ cohort) were analyzed to investigate PD-L1 testing rates, treatment patterns following FDA approval of pembrolizumab, and the proportion receiving "off-label" CPI monotherapy (single-agent use in patients with metastatic HNSCC and negative/unknown PD-L1 status). Factors associated with "off-label" use were identified using multivariable logistic regression.

Results: The total cohort included 7,657 patients with a median age of 65 years (IQR, 58-72); 67% were White, 78% had a history of smoking, 66% had an ECOG performance status (PS) of 0-1, and 31% were HPV-positive. The 2019+ subset included 3,395 patients, of whom nearly half (47%) did not have a known PD-L1 CPS prior to systemic treatment initiation. The most common frontline treatment in the total cohort was CPI monotherapy (43%). CPI monotherapy use was even higher in patients aged ≥75 years (54%) and those with ECOG PS ≥2 (52%). Among the 2019+ subgroup with PD-L1 CPS negative/unknown tumors (n=1,926), 536 (28%) received CPI monotherapy "off-label." Factors associated with "off-label" use on multivariable regression included age ≥75 years (odds ratio [OR], 1.4), community practice setting (OR, 1.5), and earlier year of treatment (OR, 1.3 per year) (all P<.05).

Conclusions: Most US patients with R/M HNSCC are now receiving CPI-based therapy in the frontline setting; however, PD-L1 testing remains underutilized. "Off-label" use of CPI monotherapy in PD-L1-negative/unknown HNSCC is common, particularly among elderly patients.