Tumor-informed ctDNA for recurrence and survival prediction in HPV-independent head and neck squamous cell carcinoma.

Abstract

e18115 Background: Over half of patients with locally advanced, HPV-independent head and neck squamous cell carcinoma (HNSCC) experience recurrence. The detection of minimal residual disease (MRD) and early detection of recurrence have the potential to significantly improve patient outcomes. However, current disease surveillance methods have poor sensitivity and diagnostic accuracy. Circulating tumor DNA (ctDNA)-based MRD detection may predict recurrence-free survival (RFS) and overall survival (OS) in several cancers, such as colorectal and lung cancer, but their clinical utility in HPV-independent HNSCC remains understudied. Methods: A prospective observational study was performed to evaluate a personalized, tumor-informed ctDNA-based MRD assay (PredicineBEACON) in patients with newly diagnosed, locally advanced HNSCC treated with surgery followed by risk-adjusted post-operative treatment to test the primary hypothesis that MRD detection post-adjuvant therapy is associated with worse RFS and OS. MRD was assessed before surgery, within 1.5 months after the completion of adjuvant treatment (MRD-TC), and during the surveillance period (MRD-S) (1.5 to 6 months after completion of adjuvant therapy). We used Kaplan-Meier survival analyses to compare RFS and OS between patients who were MRD positive during the MRD-TC or MRD-S windows, and those who were MRD negative. We used multivariable logistic regressions, including extranodal extension (ENE) and margin status, to assess the association between MRD status and three-year OS and RFS while controlling for established risk factors of recurrence. Results: We processed 142 samples from 40 patients, with an average of 3.6 samples per patient (range: 2–6), over a median follow-up of 32 months (range: 3.8-41). Thirty-six (90%) of the patients had HPV-independent disease, 67% had oral cavity tumors, 75% had stage III or nonmetastatic IV disease, and 56% underwent surgery followed by chemoradiotherapy. Fifty percent (20/40) of patients experienced recurrence. The ctDNA detection rate at enrollment was 97% (35/36). Patients who were MRD positive in the MRD-TC window had worse three-year OS (HR= 22.2; 95% CI: 2.72-182.62) and RFS (HR= 11.6; 95% CI: 3.1-43.9). Patients who were MRD positive in the MRD-S window had worse three-year RFS (HR=5.89; 955 CI: 1.64-21.11). The median lead time from first MRD detection to clinical detection of recurrence was 5 months (range 0.2-21.6). In multivariable analyses, the detection of MRD-TC (OR= 35; 95% CI 1.19–1055.4; p = 0.03) or MRD-S (OR= 25.6; 95% CI 2.20–298.5; p = 0.009) remained significant independent predictors of recurrence after adjustment for ENE and positive margins. Conclusions: Tumor-informed ctDNA MRD detection early after completion of adjuvant treatment was highly predictive of recurrence and overall survival in patients with locally advanced HPV-independent HNSCC.