Predicting response to immune checkpoint blockade in recurrent/metastatic head and neck squamous cell carcinoma using personalized circulating tumor DNA.

6051 Background: Immune checkpoint blockade (ICB) is the standard of care for recurrent/metastatic head and neck squamous cell carcinoma (R/M HNSCC) yet, response rates are poor. Currently there are no real-time, non-invasive biomarkers of treatment response. Circulating tumor DNA (ctDNA) has shown promising results as a non-invasive response prediction tool in several solid malignancies. In this retrospective study we applied a tumor-informed next-generation sequencing liquid biopsy assay (RaDaR, NeoGenomics Laboratories, Inc.) to track the dynamics of ctDNA in HNSCC patients treated with ICB to test the primary hypothesis that ctDNA detection across treatment predicts disease progression and the secondary hypothesis that ctDNA clearance during treatment positively impacts clinical outcomes in HNSCC. Methods: Whole exome sequencing was performed on pre-ICB archival tumor sections from 16 R/M HNSCC patients. A median sequencing coverage of 306x was obtained. Personalized assays targeting a median of 48 tumor-derived somatic variants were used in serial plasma samples for the detection and monitoring of ctDNA during ICB. Response was assessed using RECIST 1.1 or clinical assessment when imaging was unavailable. Multivariable logistic regression was performed to measure the effect of clinicopathologic features on achieving ctDNA-clearance. Multivariate Cox regression was performed to measure the effect of achieving ctDNA-clearance on survival. Results: 137 plasma samples were collected from 16 patients with a median of 6.5 samples per patient. 14/16 patients had a pre-ICB plasma sample available and ctDNA was detected in 12 (85.7%) (median estimated variant allele frequency: 0.28%, range: 0.004%-1.8%). In the cohort there were 2 patients with complete response (CR), 4 patients with partial response (PR), 2 patients with stable disease (SD), 7 patents with progressive disease (PD), and 1 patient whose response could not be assessed. In patients who cleared ctDNA (n=9) during ICB, there were 2 CR, 4 PR, 1 SD, and 1 PD. In patients who did not clear ctDNA (n =7) there was 1 SD and 6 PDs. ctDNA-clearance at any point after ICB start, including beyond PD on ICB, was associated with 33-fold increased odds of achieving CR, PR, or SD (OR 33.6, 95% CI 2.4-1400.6, p=0.022) independent of PD-L1 status, ICB regimen, and if ICB was administered as first or later line. Achieving ctDNA-clearance was associated with improved overall survival (OS) (HR 0.052, 95% CI 0.005-0.493, p=0.0101) after controlling for ICB regimen, virus status, and PD-L1 status. Median OS from start of ICB was 7.1 months in patients who remained ctDNA-positive and 28.5 months in those who became ctDNA-negative at any point after starting ICB. Conclusions: Undetectable levels of ctDNA at any point during ICB treatment were strongly predictive of achieving CR, PR, or SD in patients with R/M HNSCC.