Cell-Free Plasma DNA for Disease Stratification and Prognosis in Head and Neck Cancer.

Background: Clinicians face many challenges in disease stratification and outcome prediction in head and neck squamous cancer cell (HNSCC) patients. Given the limitations of currently used clinical scoring, repetitive biopsies, and imaging techniques, liquid biopsy approaches may provide valuable additional diagnostic and prognostic information.

Methods: A noninterventional, single-center observational study was performed with clinical data and plasma samples from HNSCC patients. Cell-free tumor DNA-derived copy number aberrations (CNAs) were determined in 116 patients by low-coverage next-generation sequencing (NGS). Significant CNAs were combined in a genome-wide copy number instability score (CNI), which was evaluated with respect to conventional clinical staging and patient outcome.

Results: Receiver-operating characteristic (ROC) curve analysis comparing the presurgery CNI in patients (n = 103) with that in tumor-free controls (n = 142) yielded an area under the ROC curve of 87.2% (95% CI, 79.4%-93.3%). At a specificity of 95%, the sensitivity to detect tumors varied between 46% (pT1) and 94% (pT4). A CNI above the median (i.e., >72) had a positive predictive value of 90% (95% CI, 79%-96%) for lymph node involvement (LNI), while the negative predictive value was 57% (95% CI, 43%-70%). For a CNI >72, overall survival (OS) was worse (hazard ratio, 4.89; 95% CI, 1.39-17.17; = 0.01) with 62% and 90% survivors 3 years after surgery for a CNI >72 and ≤72, respectively. In multivariable models, the CNI was a superior predictor of OS compared to established disease features, including LNI.

Conclusions: The CNI may assist in predicting LNI and prognosis in HNSCC with direct therapeutic implications concerning the need for neck dissection or more aggressive treatment.