Postoperative ctDNA May Predict Recurrence Risk in HPV-Negative Head and Neck Cancer

For patients with head and neck squamous cell carcinoma (HNSCC), the battle does not end when the tumor is removed. In fact, for the majority of patients with HPV-negative disease, the most dangerous period begins immediately after surgery, when microscopic cancer cells may already be circulating silently in the bloodstream, invisible to every scanner and clinical examination. Now, a new meta-analysis published in Frontiers in Oncology suggests that a simple blood test measuring circulating tumor DNA (ctDNA) after surgery could identify which of these patients are truly at risk of recurrence, potentially sparing many from unnecessary adjuvant therapy while flagging those who need intensified treatment.

The research team, led by Mohammad Rehan Asad of Majmaah University in Saudi Arabia and Atta Ullah Khan of the University of Campania Luigi Vanvitelli in Italy, systematically reviewed eight studies encompassing 325 patients with HPV-negative head and neck squamous cell carcinoma. Their findings are striking: patients with detectable postoperative ctDNA faced a pooled hazard ratio of 10.07 for diminished recurrence-free survival, meaning they experienced roughly a ten-fold elevated risk of cancer returning compared to those whose blood showed no trace of tumor DNA. The association extended to overall survival as well, with a pooled hazard ratio of 8.98, suggesting that ctDNA positivity may influence mortality through mechanisms beyond recurrence alone.

Head and neck squamous cell carcinoma is the seventh most common malignancy worldwide, accounting for approximately 890,000 new cases annually. While human papillomavirus drives a subset of oropharyngeal cancers, the majority of HNSCC cases remain HPV-negative, predominantly linked to tobacco and alcohol consumption. These HPV-negative tumors carry a poorer prognosis than their HPV-positive counterparts, and despite advances in multimodal treatment, disease recurrence occurs in 50 to 60 percent of patients following curative-intent treatment, with most relapses occurring within two years. Current surveillance relies on clinical examinations and radiological imaging, which lack the sensitivity to detect microscopic residual disease before it becomes clinically apparent.

Circulating tumor DNA has emerged as a promising liquid biopsy biomarker in precision oncology. While ctDNA utility is well-established in HPV-positive HNSCC, its application in HPV-negative disease presents unique challenges due to genetic heterogeneity and high mutational burden. Recent advances in tumor-informed personalized ctDNA assays using next-generation sequencing and digital droplet PCR have enhanced sensitivity and specificity in non-viral solid tumors. Tumor-informed assays demonstrate superior performance compared to tumor-agnostic approaches, with reported sensitivities exceeding 90 percent and specificities approaching 99 percent. Emerging evidence demonstrates that postoperative ctDNA positivity predicts disease recurrence and inferior survival outcomes in colorectal, breast, and lung cancers, with preliminary studies suggesting similar prognostic value in HPV-negative HNSCC.

The eight included studies employed a remarkable diversity of ctDNA detection architectures, encompassing NGS-based platforms, commercial tumor-informed assays such as Signatera and RaDaR, and emerging pooled methodologies like MAESTRO-Pool. The temporal relationship between surgical intervention and biomarker assessment ranged considerably, from immediate postoperative periods within one to three days to delayed sampling intervals up to three months after surgery. Five of the eight studies used MRD or residual-disease terminology, framing post-treatment ctDNA detection as a minimal residual disease application, while the remaining studies evaluated postoperative ctDNA primarily in the contexts of relapse monitoring, ctDNA clearance, or PET-adjunct surveillance.

Clinical performance among the MRD-focused studies was encouraging but heterogeneous. In the study by Honoré and colleagues, ctDNA was detected in four of 42 evaluable patients within 12 weeks after curative-intent treatment; three of these patients relapsed, and the median lead time before clinical recurrence was 7.0 months. Sim and colleagues reported a positive predictive value of 92.9 percent and a negative predictive value of 80 percent for immediate postoperative ctDNA detection in relation to recurrence or death. Karasik and colleagues reported lower actuarial one-year recurrence-free survival among ctDNA-positive patients at 38.2 percent compared to 87.4 percent among ctDNA-negative patients. In the LIONESS cohort, Flach and colleagues reported 91.3 percent sensitivity, 94.2 percent specificity, 87.5 percent positive predictive value, and 96.1 percent negative predictive value for recurrence detection using plasma samples collected at least 14 days after surgery.

Subgroup stratification by patient cohort composition revealed comparable effect estimates across HPV status representation. Three studies enrolling exclusively HPV-negative populations yielded a pooled hazard ratio of 10.33, while five studies enrolling mixed cohorts including both HPV-negative and HPV-positive patients yielded a pooled hazard ratio of 9.91. Statistical heterogeneity remained negligible within both subgroups, with I² values of 0 percent, suggesting that the observed association between postoperative ctDNA and recurrence-free survival was not substantially influenced by HPV status representation in the study population. Individual study estimates demonstrated uniform directional concordance, ranging from 5.64 to 22.28, with all confidence intervals excluding the null.

However, the researchers urge caution in interpreting these striking effect magnitudes. Leave-one-out sensitivity analysis revealed that sequential omission of individual studies yielded pooled hazard ratios ranging from 2.22 to 2.43, considerably lower than the primary pooled estimate of 10.07. While all sensitivity estimates remained statistically significant, this discrepancy suggests that the magnitude of the primary effect may reflect the combined influence of all eight studies rather than representing a stable, platform-independent association. The observed hazard ratios substantially exceed those reported in other solid tumors, such as colorectal cancer at 2.5 to 4.0 and breast cancer at 2.0 to 3.5, suggesting potential overestimation. Multiple factors may contribute, including heterogeneous detection methodologies with variable sensitivity and specificity, small individual study sample sizes susceptible to selection bias, variable plasma collection timing affecting ctDNA kinetics, and confounding bias in one-third of studies where treatment decisions may have been influenced by ctDNA status.

Publication bias testing with only eight studies has inherent power limitations. While Egger’s regression-based test and Begg’s rank correlation test were non-significant and the funnel plot appeared symmetric, these findings do not definitively exclude bias. The LIONESS trial, using personalized ctDNA assays in predominantly HPV-positive HNSCC, reported a hazard ratio of approximately 2.5, substantially lower than the 10.07 observed here, highlighting the need for comparative studies. Additionally, 62.5 percent of studies enrolled mixed HPV-positive and HPV-negative cohorts, limiting applicability to purely HPV-negative disease. All studies restricted to resectable disease, reducing generalizability to unresectable populations. The small overall sample size of 325 patients, with approximately 180 HPV-negative, constrains statistical power for meaningful subgroup analyses.

The clinical utility of postoperative ctDNA detection for guiding treatment decisions or surveillance strategies in HPV-negative HNSCC remains incompletely established. Until larger, prospective, multi-institutional studies employing standardized methodologies and enrolling purely HPV-negative populations are conducted, postoperative ctDNA detection should be considered an investigational biomarker rather than a clinically validated tool. Randomized controlled trials examining whether ctDNA-guided interventions improve clinical outcomes are essential before recommending routine clinical implementation. Future research should prioritize standardization of detection methodologies and prospective validation in well-characterized HPV-negative HNSCC populations, potentially transforming how clinicians stratify recurrence risk and personalize adjuvant therapy decisions for this challenging malignancy.

Subject of Research: Postoperative ctDNA as a prognostic biomarker for recurrence in HPV-negative head and neck squamous cell carcinoma

Article Title: Postoperative ctDNA could potentially spare patients from unnecessary adjuvant therapy in HPV-negative head and neck squamous cell carcinoma: a meta-analysis

Article References: Asad, M. R., Pusarico, A. F., Reji, N., AlRamadneh, T. N., & Khan, A. U. (2026). Postoperative ctDNA could potentially spare patients from unnecessary adjuvant therapy in HPV-negative head and neck squamous cell carcinoma: a meta-analysis. Frontiers in Oncology, 16, Article 1902328. https://doi.org/10.3389/fonc.2026.1902328

Image Credits: AI Generated

DOI: 10.3389/fonc.2026.1902328

Keywords: ctDNA, head and neck cancer, HPV-negative, liquid biopsy, minimal residual disease, meta-analysis, recurrence, adjuvant therapy, Postoperative, potentially, spare, patients