Abstract
PURPOSE: Observational studies suggest circulating tumor human papillomavirus (HPV) DNA may facilitate early detection of recurrent HPV-positive oropharynx cancer. We prospectively investigated whether biomarker-guided surveillance detects recurrence earlier than the standard-of-care methods. PATIENTS AND METHODS: We enrolled patients evaluated for HPV-positive oropharynx cancer at a single center from November, 2020, to April, 2023, undergoing curative-intent treatment in a single-arm cohort study. Pretreatment plasma and/or tumor tissue were tested for tumor tissue-modified viral (TTMV) HPV DNA from HPV subtypes 16/18/31/33/35 using a droplet digital PCR-based commercial assay. Posttreatment plasma TTMV was assessed periodically. Detectable/indeterminate tests prompted imaging. The primary outcome was the proportion of recurrences first detected by TTMV. RESULTS: Median follow-up was 23 months, with median six posttreatment TTMV tests for 155 subjects. Fifteen subjects (9%) experienced recurrence. Among these, six [40%, 95% confidence interval (CI) = 16%-68%] were "early true-positives," for whom TTMV detection predated and prompted the imaging and clinical workup that diagnosed recurrence (median lead time = 132 days; range = 47-280). Another five subjects (33%) were "confirmatory true-positives," for whom detectable TTMV confirmed suspicious standard-of-care imaging findings. Finally, four subjects (27%) with recurrence had undetectable TTMV at diagnosis ("false-negatives"). False-negatives had low or undetectable pretreatment TTMV, and 2/4 had non-HPV16 genotypes. Finally, three subjects had prolonged detectable TTMV without disease ("false-positives"); all had immunologic comorbidities. Overall, the sensitivity of TTMV for recurrence was 73% (95% CI = 45%-92%). Among 117 subjects with HPV16 and detectable pretreatment TTMV, sensitivity was higher (91%, 95% CI = 59%-100%). CONCLUSIONS: TTMV-guided surveillance facilitates early detection of many HPV-positive oropharynx cancer recurrences, with the highest sensitivity for HPV16 and detectable pretreatment TTMV. Clinical implementation should be carefully informed by the limitations described in this study. See related commentary by Califano, p. 1561.