Year: 2025
Type: article
Abstract: Human papilloma virus (HPV)-associated oropharyngeal carcinoma has been termed an epidemic. Even with steady prophylactic vaccine adoption, the incidence is predicted to continue to rise for the upcoming decades and account for up to 70% of head and neck cancers.1-3 Unlike cervical cancer, no screening paradigm or dysplastic signature exists for oropharyngeal carcinoma. Early detection paradigms are urgently needed, and in this issue of the Journal, Das and colleagues demonstrate detectability of HPV-associated oropharyngeal cancer up to 10 years before diagnosis using a plasma cell-free DNA HPV sequencing assay, HPV-DeepSeek, with direct comparison to antibody testing and digital droplet (dd) PCR.4 This study advances our understanding of very early detection of oropharyngeal cancer while leaving open questions of how to best consider implementation in the setting of extended lead times before clinical manifestation and in the multicancer early detection (MCED) testing era. The authors analyzed frozen plasma samples from 28 patients collected 1 to 10 years before oropharyngeal cancer diagnosis and compared these to 28 matched controls. HPV-DeepSeek detected evidence of malignancy in 22 patients (79% sensitivity) up to 7.8 years before diagnosis with 100% specificity.4 HPV-DeepSeek outperformed the authors’ in-house ddPCR assay as well as a commercial ddPCR assay with regard to overall early detection sensitivity and lead time to clinical cancer diagnosis, consistent with prior analyses.4-6 The authors then went on to develop a cell-free DNA-based machine learning model relying on percentage of the HPV genome covered by sequencing reads, number of unique HPV sequencing reads, and fragment length characteristics using an independent plasma cohort from 153 patients and 153 controls. When this externally trained and locked algorithm was then applied to the early detection case-control cohort, cancer detection sensitivity increased to 96% with a lead time of up to 10.3 years.4
DOI: 10.1093/jnci/djaf302
Language: English
Topic: Head and Neck Cancer Studies
Subfield: Otorhinolaryngology
Field: Medicine
Domain: Health Sciences
Open Access status: bronze
OpenAlex ID: W4416122671