Abstract
Purpose
Despite advances in treatments, nearly half of the patients with head and neck squamous cell carcinoma (HNSCC) experience recurrence. Early recurrence detection is critical for enabling timely curative interventions and improving survival outcomes. Here, we evaluated circulating tumor DNA (ctDNA) testing post-definitive treatment to detect molecular residual disease (MRD) and predict outcomes in patients with HNSCC.
Methods and Materials
This retrospective study included a real-world cohort of 97 patients with stage I to IV HNSCC undergoing ctDNA testing using a personalized, tumor-informed assay (Signatera, Natera, Inc). Longitudinal plasma samples (n = 500) were collected pretreatment, in the early postdefinitive local treatment setting (within 3 months of completing definitive local treatment, before adjuvant treatment), and during postdefinitive treatment surveillance (after adjuvant therapy [if given] or postcompletion of definitive local treatment until recurrence).
Results
Median patient follow-up was 22 (2-56) months. ctDNA-based MRD positivity in the early postdefinitive local treatment setting was significantly associated with inferior progression-free survival (PFS) (hazard ratio [HR], 6.39; P = .0002) and overall survival (OS) (HR, 25.72; P = .003). Longitudinally, MRD positivity during the postdefinitive treatment surveillance window was associated with significantly shorter PFS (HR, 38.87; P < .0001) and OS (HR, 14.37; P = .014) and preceded radiographic progression by a median of 4.3 (0.62-21.91) months. The increased risk of recurrence for patients who were MRD-positive was observed across all pathologic stages, regardless of p16 status. In multivariate analysis, MRD positivity postdefinitive treatment was the only clinicopathologic factor significantly associated with PFS (HR, 38.29; P < .001).
Conclusions
In patients with stage I to IV HNSCC, ctDNA-based MRD positivity after definitive local therapy and adjuvant treatment is associated with disease recurrence and poor survival outcomes.
Introduction
Head and neck squamous cell carcinoma (HNSCC), a diverse group of malignancies originating in the oral cavity, pharynx, and larynx, poses a major global health challenge.1 Although treatments such as surgery, radiation, and systemic therapies have advanced, recurrence is frequent (20%-30% for early-stage and 50%-60% for advanced-stage disease) and leads to poor long-term survival. The median survival after recurrence of HNSCC is 6 to 12 months, which can be extended to 15 to 30 months with aggressive salvage therapy in patients with resectable, isolated locoregional recurrence. Effective risk stratification and early detection of recurrence are crucial for enabling curative treatment strategies and improving survival rates. Although posttreatment surveillance protocols for HNSCC differ considerably between institutions, they generally rely on radiographic imaging approximately every 3 to 6 months and clinical examinations, including endoscopic examination, every 1 to 3 months during the first year, every 2 to 6 months during the second year, and every 4 to 8 months during years 3 to 5.1,2 However, these methods are not sensitive enough to detect microscopic residual disease. Altered anatomy and/or fibrosis from definitive treatment further pose challenges in surveillance using these methods, thus potentially delaying necessary interventions.
Circulating tumor DNA (ctDNA) has emerged as a promising prognostic and predictive biomarker across multiple cancer types, including colorectal, lung, and breast cancers.3, 4, 5, 6, 7, 8, 9, 10, 11 In HNSCC, where recurrences frequently occur within the first 2 years after treatment and yet standard surveillance guidelines are lacking, ctDNA monitoring can potentially detect molecular residual disease (MRD), risk-stratify patients, and guide therapeutic decision-making. Despite growing interest, data on the clinical validity and prognostic relevance of ctDNA monitoring in HNSCC remain limited. Most existing studies are small and/or lack longitudinal sampling.12, 13, 14 In this study, we evaluated the utility of longitudinal, tumor-informed ctDNA monitoring postdefinitive treatment for early detection of recurrence in a prospective cohort of patients with stage I to IV HNSCC. Our findings offer insight into the potential role of ctDNA as a prognostic biomarker for assessing postdefinitive treatment risk and early intervention in patients with HNSCC.
Methods and Materials
Study design and patients
This study retrospectively analyzed the results of Signatera (Natera, Inc.) testing performed in 97 patients (500 plasma samples) with stage I to IV HNSCC in routine clinical practice. All patients received standard-of-care treatment, and radiographic imaging and blood collection for ctDNA testing were performed at the clinician’s discretion according to established guidelines. The median time between 2 consecutive ctDNA tests was 2.90 months (range, 0.59-9.98 months), and the median frequency of radiographic imaging was 8.12 months (range, 0.72-35.19 months). The baseline was defined as any time point after the diagnostic biopsy/clinical presentation and before surgery or the initiation of definitive chemoradiation therapy (CRT)/radiation therapy (RT). The early postdefinitive local treatment setting included time points within 3 months (90 days) of completing definitive local treatment (defined as surgery, or definitive RT with or without concurrent chemotherapy [CRT/RT]), and before adjuvant treatment. The longitudinal, postdefinitive treatment surveillance analysis included time points after adjuvant therapy, if given, or after completion of definitive local treatment, until the last follow-up or disease progression. Patients with no imaging assessment available for >12 months from their last ctDNA test were excluded from the longitudinal postdefinitive treatment analysis. The study was conducted in compliance with Natera protocol 21-058, the Declaration of Helsinki, Title 21 of the United States Code of Federal Regulations (CFR) as applicable, good clinical practice guidelines, and International Conference on Harmonization guidelines. A waiver of informed consent was granted according to 45 CFR 46.116(d) and 45 CFR 46.117 (c)(2), respectively.
Tumor-informed ctDNA testing
Formalin-fixed, paraffin-embedded tumor tissue and matched normal blood samples from each patient were subjected to whole-exome sequencing, as previously described.8 From the whole-exome sequencing results, a set of up to 16 patient-specific, somatic single-nucleotide variants were selected. The patient-specific and tumor-specific ctDNA assay was tested on associated cell-free DNA libraries via multiplex polymerase chain reaction, and subsequent sequencing was performed on an next-generation sequencing platform. Plasma samples with >2 single-nucleotide variants detected above a predefined algorithm confidence threshold were considered ctDNA-positive. ctDNA concentrations were reported as mean tumor molecules per milliliter of plasma (MTM/mL).
Statistical analysis
Progression-free survival (PFS) was defined as the duration between the start of definitive treatment (CRT/RT) or surgery and the date of radiographic progression or death because of any cause, with censoring at the latest radiologic assessment. Overall survival (OS) was defined as the duration between the start of definitive treatment (CRT/RT) or surgery and the date of death because of any cause, with censoring at the latest clinical follow-up. The χ2 and Fisher exact tests were used to compare categorical variables. The pairwise Wilcoxon test was used to compare the medians of numerical variables between groups defined by categorical variables. Survival analyses were conducted using R software version 4.5.0 (R Core Team, R Foundation for Statistical Computing), with the survival (version 3.7.0), survminer (version 0.4.9), and coxphf (version 1.13.4) packages (R Core Group). The exact P values from Cox regression hazard models were calculated using the Rmpfr package (version 0.9.5). The Kaplan-Meier method was used to estimate the survival distribution. Differences between groups were tested using the log-rank test. A multivariable Cox proportional hazards model was used to assess prognostic factors associated with PFS.
Lead time was defined as the interval between the first detection of molecular recurrence (first ctDNA positivity during the postdefinitive treatment surveillance window) and clinical recurrence/progression. A longer lead time was defined as an interval >6 months, reflecting the current surveillance monitoring cadence per National Comprehensive Cancer Network guidelines for HNSCC. To calculate the ctDNA growth rates, a log-linear regression was fitted to each patient with at least 2 plasma samples analyzed for ctDNA in the postdefinitive treatment surveillance window and before their clinical recurrence, based on the ctDNA level as a function of time before recurrence. The ctDNA growth rates were estimated from the slope of the regression lines.
Results
Patient characteristics
A total of 500 plasma samples from 97 patients with stage I to IV HNSCC were analyzed in this study (Fig. 1). The median follow-up after surgery or the end of definitive CRT/RT was 22 months (range, 2-56 months). Patient characteristics at baseline are summarized in Table 1. Patient characteristics and ctDNA status at baseline are summarized in Table E1. There were no significant differences in clinicopathologic features such as sex, primary location, clinical T and N stages, tumor location, histology, p16 status, and treatment regimen based on ctDNA status at baseline (Table E1). Treatment regimens, longitudinal ctDNA results, and clinical outcomes are depicted in Fig. 2.
Figure 1.
Flow chart depicting circulating tumor DNA (ctDNA) results by stage at baseline and in the molecular residual disease (MRD) and postdefinitive treatment surveillance windows.
Abbreviations: CRT = chemoradiation therapy; HNSCC = head and neck squamous cell carcinoma; RT = radiation therapy.
Table 1.
Patient and tumor characteristics.
| Characteristic | N = 97⁎ |
|---|---|
| Sex | |
| Female | 17 (18%) |
| Male | 80 (82%) |
| Age (years) | 66 (29-95) |
| Smoking history | 63 (65%) |
| Tumor location | |
| Larynx/hypopharynx | 5 (5.2%) |
| Oral cavity | 16 (16%) |
| Oropharynx | 67 (69%) |
| Other (paranasal sinus and nasopharyngeal) | 9 (9.3%) |
| Clinical T stage | |
| T0 | 2 (2.1%) |
| T1 | 12 (12%) |
| T2 | 31 (32%) |
| T3 | 30 (31%) |
| T4 | 21 (22%) |
| TX | 1 (1.0%) |
| Clinical N stage | |
| N0 | 22 (23%) |
| N1 | 33 (34%) |
| N2 | 33 (34%) |
| N3 | 9 (9.3%) |
| Clinical M stage | |
| M0 | 93 (96%) |
| M1 | 4 (4.1%) |
| Histology | |
| Adenosquamous carcinoma | 1 (1.0%) |
| Basaloid squamous cell carcinoma | 6 (6.2%) |
| Epithelial myoepithelial carcinoma | 1 (1.0%) |
| Squamous cell carcinoma | 86 (89%) |
| Undifferentiated carcinoma | 3 (3.1%) |
| Clinical stage | |
| I/II | 49 (51%) |
| III/IVA/IVB | 45 (46%) |
| IVC | 3 (3.1%) |
| p16 status | |
| Negative | 43 (44%) |
| Positive | 54 (56%) |
| Definitive treatment | |
| Definitive CRT or RT | 69 (71%) |
| Surgery + CRT or RT | 24 (25%) |
| Surgery only | 1 (1.0%) |
| None (declined treatment) | 1 (1.0%) |
| None (hospice) | 2 (2.1%) |
| Progression during follow-up | |
| No progression | 64 (66%) |
| Progression | 33 (34%) |
| Status at last follow-up | |
| Alive | 81 (84%) |
| Deceased | 16 (16%) |
| Total follow-up (months) | 22 (2-56) |
Abbreviation: CRT = chemoradiation therapy; RT = radiation therapy.
n (%); Median (Min-Max).
Figure 2.
Overview plot depicting treatment regimen, longitudinal circulating tumor DNA (ctDNA) results, and clinical outcome for all patients in the cohort, grouped by clinical stage.
ctDNA detection and levels at baseline
At baseline (pretreatment), plasma samples for ctDNA analysis were collected for 63 of 97 patients, 88.9% (56/63) of whom were ctDNA-positive (Fig. 2). Although the median ctDNA levels at baseline increased significantly from stages N0 to N1 and N1 to N2, such a trend was not observed between stages N2 and N3 (Fig. 3a). Median baseline ctDNA levels were highest for patients with T3 and N3 disease (Fig. 3b). ctDNA levels at baseline were significantly higher in patients with p16-positive tumors than those with p16-negative tumors (Fig. 3c), whereas no significant variations were observed based on tumor location (Fig. 3d).
Figure 3.
(a) Box plot showing the median and distribution of circulating tumor DNA (ctDNA) levels at baseline in mean tumor molecules per milliliter of plasma (MTM/mL) for different clinical N stages. The heatmap shows P values for pairwise comparisons between different N stages as generated by the Wilcoxon nonparametric test. (b) Heatmap showing median baseline ctDNA levels at different T and N stages. (c, d) Box plot representing median and distribution of ctDNA levels (MTM/mL) at baseline based on p16 status (c) or tumor location (d). The heatmap in panel (d) shows P values for pairwise comparisons between tumor location and N stage. In box plots, the center line represents the median, the bottom and the top of the box represent the 25% and 75% quartiles of the distribution, respectively, with whiskers showing the distribution limits; data points represent outliers.
Abbreviations: LRX = larynx; OC = oral cavity; PRX = pharynx.
Association of MRD status in the early definitive local treatment setting with patient outcomes
ctDNA-based MRD status within 3 months of completing definitive local treatment (before adjuvant therapy) was available for 69 patients, 18.8% (13/69) of whom were ctDNA-positive (MRD-positive; Fig. 2). During follow-up, 61.5% (8/13) of patients who were MRD-positive experienced disease progression, in contrast to only 14.3% (8/56) of patients who were MRD-negative experiencing progression. Patients who were MRD-positive were found to be 6 times more likely to experience recurrence (hazard ratio [HR], 6.39; 95% CI, 2.39-17.09; P = .0002; Fig. 4a). Remarkably, the 24-month PFS rates for patients who were MRD-positive and MRD-negative were 32.3% (95% CI, 8.62%-59.4%) and 84.1% (95% CI, 70.5%-91.8%), respectively. A subanalysis restricted to patients who did not receive any adjuvant therapy revealed an even greater difference in PFS between MRD-positive and MRD-negative groups (24-month PFS: 14.3% [95% CI, 0.71%-46.5%] vs 84.7% [95% CI, 68.8%-92.9%], respectively; HR, 10.08; 95% CI, 3.23-31.46; P < .0001) (Fig. E1). The association of MRD positivity with a significantly increased risk for progression was noted across all disease stages (Fig. 2a, b) and in both p16-positive and p16-negative patients (Fig. E2c, d). Furthermore, compared with patients who were MRD-negative, MRD positivity was found to be associated with significantly inferior OS (HR, 25.72; 95% CI, 3.00-220.48; P = .003; Fig. 4b), demonstrating a 24-month OS of 56.10% (95% CI, 23.30%-79.50%) vs 98.20% (95% CI, 88.00%-99.70%), respectively. The subanalysis of patients treated with definitive CRT showed that end-of-treatment MRD positivity was strongly associated with inferior PFS and OS (Fig. E2e, f). A subanalysis in surgically treated patients could not be performed because of low patient numbers with available ctDNA time points within 3 months of surgery and before the start of adjuvant therapy, rendering the analysis underpowered.
Figure 4.
(a, b) Kaplan-Meier estimates of progression-free survival (PFS)(a) and overall survival (OS) (b) stratified by molecular residual disease (MRD) status in the early postdefinitive local treatment setting among patients with head and neck squamous cell carcinoma. The bar plots depict the proportion of patients experiencing disease progression among the MRD-positive and MRD-negative groups. Hazard ratios (HRs) and 95% CIs were calculated using the Cox proportional hazards model. P values were calculated using the 2-sided log-rank test.
Abbreviation: ctDNA = circulating tumor DNA; mOS = median overall survival; mPFS = median progression-free survival; NR = not reached.
Association of MRD status during postdefinitive treatment surveillance with patient outcomes
Of the 79 patients with time points available during postdefinitive treatment surveillance (with a median of 4 [range, 1-20] time points per patient), 30.4% (24/79) were MRD-positive at 1 or more time points. Compared with patients who were serially MRD-negative, patients with MRD positivity at any time point during postdefinitive treatment surveillance were found to have significantly inferior PFS (HR, 38.87; 95% CI, 9.11-165.88; P < .0001; 24-month PFS: 20.8% [95% CI, 7.59%-38.5%] vs 96.4% [86.2%-99.1%]; Fig. 5a) and OS (HR, 14.37; 95% CI, 1.73-119.39; P = .014; Fig. 5b). Longitudinally, MRD positivity postdefinitive treatment had a median lead time of 4.3 (range, 0.6-21.9 months) months over clinical progression and a sensitivity of 91.7% (22/24) and a specificity of 96.4% (53/55) for detecting recurrence/progression. When a Cox regression analysis was performed with longitudinal time points as a time-dependent variable, MRD positivity during postdefinitive treatment surveillance was associated with a significantly shorter PFS (HR, 92.95; 95% CI, 21.57-400.63; P < .0001) and OS (HR, 25.81; 95% CI, 3.08-215.98; P = .003).
Figure 5.
(a, b) Kaplan-Meier estimates of progression-free survival (PFS) (a) and overall survival (OS) (b) stratified by molecular residual disease (MRD) status during postdefinitive treatment surveillance among patients with head and neck squamous cell carcinoma. The bar plot depicts the proportion of patients experiencing disease progression among the MRD-positive and MRD-negative groups. Hazard ratios (HRs) and 95% CIs were calculated using the Cox proportional hazards model. P values were calculated using the 2-sided log-rank test. (c) Forest plot depicting the multivariate analysis (including MRD status during surveillance and other clinicopathologic factors) for PFS in patients with head and neck squamous cell carcinoma. The unadjusted HRs (squares) and 95% CIs (horizontal lines) are shown for each prognostic factor; the vertical dotted line represents the null hypothesis. (d, e) Serial circulating tumor DNA (ctDNA) measurements for patients grouped based on MRD status during the postdefinitive treatment surveillance, color-coded based on outcomes: persistently ctDNA-negative (d) and ctDNA-positive at least at 1 time point during surveillance (e). Only patients with ≥2 time points available after surgery or after the start of definitive therapy were included. (f) Time to recurrence detection from surgery or end of definitive treatment for ctDNA and imaging among ctDNA-positive patients who experienced recurrence. Lead time in months is calculated between molecular recurrence (ctDNA positivity, indicated by the blue dot) and clinical recurrence (radiographic imaging positive, indicated by the red dot) during the postdefinitive treatment surveillance window (N = 22). Statistical significance was calculated using the paired Wilcoxon rank-sum test. A longer lead time (dashed line) was defined as >6 months between molecular and clinical recurrence. The median difference between molecular and radiologic recurrence was calculated using the Hodges-Lehmann estimator. (g) Linear regression lines of ctDNA levels at the time of clinical recurrence (radiographic imaging positive, time point zero) for patients having at least 2 plasma samples analyzed for ctDNA in the postdefinitive treatment surveillance window and before their clinical recurrence. The regression lines of slow-growing and fast-growing ctDNA levels are shown.
Abbreviation: AIC = Akaike Information Criteria; mPFS = median progression-free survival; mOS = median overall survival; MTM/mL = mean tumor molecules per milliliter of plasma; NR = not reached;.
The association of MRD positivity with a significantly increased risk of progression was observed across all pathologic stages (Fig. E3a, b) and regardless of p16 status (Fig. E3c, d). Furthermore, in a subgroup of patients (N = 64) with time points available during both early postdefinitive local treatment and postdefinitive treatment surveillance windows, significantly inferior PFS was observed among patients with MRD positivity during surveillance than those who remained MRD-negative (HR, 62.87; 95% CI, 8.26-478.32; P < .0001; Fig. E4a). To evaluate the prognostic value of MRD status at the time of standard posttreatment imaging, we performed a landmark analysis at 3 months posttreatment. MRD positivity at 3 months remained a strong predictor of inferior PFS (HR, 17.77; 95% CI, 2.90-108.84; P = .002), with progression occurring in all patients who were MRD-positive (100%, 3/3) versus 8.7% (2/23) of patients who were MRD-negative (24-month PFS: 90.2%; Fig. E4b).
Next, we performed a multivariate analysis in the postdefinitive treatment surveillance window to further assess the prognostic value of MRD status when compared with other clinicopathologic risk factors. MRD positivity was observed to be the only significant prognostic factor associated with inferior PFS (HR, 38.29; 95% CI, 8.75-167.6; P < .001; Fig. 5c). We further evaluated the clinical outcomes among patients with distinct ctDNA dynamic patterns in the postdefinitive treatment surveillance window. Disease progression or death events were considerably lower among patients who were serially MRD-negative during postdefinitive treatment surveillance (3.6%, 2/55; Fig. 5d) compared than those with MRD positivity (92.9%, 22/24; Fig. 5e). Among patients who were MRD-positive who experienced recurrence (N = 22), a median lead time of 4.3 months (range, 0.6-21.9 months) for molecular recurrence (MRD positivity) over clinical recurrence was observed (P < .001; Fig. 5f). Furthermore, among patients with at least 2 ctDNA test results available after completion of definitive therapy and before the clinical evidence of recurrence, we investigated the rate of ctDNA increase as a proxy for tumor growth. All patients with rapidly growing ctDNA levels (higher slopes of linear regression lines of ctDNA levels) experienced clinical recurrence with a shorter lead time (<6 months) than those with a slower increase in ctDNA levels (Fig. 5g).
Impact of ctDNA testing on clinical management of patients with HNSCC
To evaluate the clinical utility of ctDNA-based MRD monitoring in HNSCC, we examined longitudinal ctDNA kinetics in representative patients from this real-world cohort. We identified clinical scenarios in which longitudinal ctDNA testing after definitive treatment informed patient management, such as imaging escalation, early detection of recurrence, and timely initiation of salvage treatment.
In the first case (UNM-069, Fig. 6a), rapid ctDNA clearance was observed following definitive CRT, consistent with an observed radiographic treatment response. During surveillance, MRD positivity was noted approximately 18 months after definitive treatment, with ctDNA levels rising over subsequent months, prompting escalated imaging surveillance. Although imaging and biopsy initially showed no evidence of disease, subsequent imaging demonstrated suspicious findings, with biopsy confirming metastatic disease. The patient received RT for recurrent disease, resulting in ctDNA clearance. In this case, MRD positivity anticipated clinical recurrence nearly 21 months ahead of radiographic imaging.
Figure 6.
(a, b) Patient-specific changes in circulating tumor DNA (ctDNA) levels over time. (a) In a patient with stage I head and neck squamous cell carcinoma (UNM-069), ctDNA positivity postdefinitive CRT prompted escalated surveillance and anticipated recurrence nearly 21 months ahead of radiologic imaging. (b) In a patient with stage II head and neck squamous cell carcinoma (UNM-031), escalated surveillance imaging prompted by ctDNA positivity postdefinitive CRT led to detection of disease progression and timely intervention, resulting in long-term disease control.
Abbreviations: CRT = chemoradiation therapy;MTM/mL = mean tumor molecules per milliliter of plasma; SBRT = stereotactic body radiation therapy.
In the second case (UNM-031, Fig. 6b), transient ctDNA clearance was noted after definitive CRT, which was concordant with no evidence of disease at approximately 3 months after the end of CRT. During follow-up, the patient turned back MRD-positive at approximately 6 months. This prompted an early radiographic scan, resulting in the detection of disease progression. The patient subsequently underwent stereotactic body RT, after which ctDNA clearance was noted. The patient remained serially MRD-negative throughout surveillance until the end of follow-up (over 22 months) with no clinical evidence of disease. This case highlights the utility of ctDNA testing for timely detection of disease progression and initiation of salvage treatment, which led to long-term and ongoing disease control.
Discussion
This study evaluated the utility of serial ctDNA monitoring for MRD detection after definitive local treatment and for surveillance in a prospective cohort of patients with stage I to IV HNSCC. We demonstrate the feasibility and clinical utility of serial ctDNA-based MRD evaluation to predict recurrence risk, identify early disease recurrence, and predict patient outcomes. Ultimately, our data provide evidence that ctDNA is a reliable biomarker for predicting recurrence risk and ongoing surveillance following definitive therapy in HNSCC.
The clinical utility of ctDNA in HNSCC has previously been explored, particularly through studies of circulating tumor Epstein-Barr virus DNA (ctEBV DNA) in nasopharyngeal carcinoma and circulating tumor human papillomavirus DNA (ctHPV DNA) in oropharyngeal cancers. In a recent meta-analysis evaluating 22 studies, ctDNA positivity—evaluated using ctEBV DNA, ctHPV DNA, and methylation-based assays—was associated with worse OS (HR, 2.00; 95% CI, 1.35-2.96) and poor PFS/recurrence-free survival (HR, 3.54; 95% CI, 1.05-11.85).15 Subgroup analyses revealed that ctEBV DNA and ctHPV DNA positivity were both associated with poorer outcomes (ctEBV DNA [OS: HR, 2.86]; ctHPV DNA [OS: HR, 1.38]).15 Emerging evidence from clinical trials aligns well with these findings. In a secondary analysis of the DART phase 3 trial, Routman et al16 reported a 24% difference in 24-month PFS based on postsurgical ctHPV DNA–based MRD status in HPV-positive patients, highlighting the potential role of ctDNA for patient selection in de-escalating adjuvant therapy approaches.
Recent studies have adopted personalized, tumor-informed platforms for ctDNA detection and have demonstrated prognostic value in HNSCC.12, 13, 14,17 In a retrospective analysis of 100 patients with locoregionally advanced HPV-negative HNSCC, Hanna et al13 found that ctDNA positivity posttreatment was strongly predictive of recurrence (HR, 7.33; 95% CI, 3.12-17.2; P < .001). Similarly, Honoré et al14 reported that patients with ctDNA positivity within 12 weeks of completing curative-intent treatment had significantly inferior recurrence-free survival and OS (P ≤ .05) compared with ctDNA-negative patients. Another recent study by Honoré et al18 demonstrated that favorable ctDNA kinetics in response can predict response to anti-PD1 therapy efficacy in patients with recurrent and metastatic HNSCC. Particularly, ctDNA clearance was significantly associated with prolonged PFS and OS (PFS, P = .0006; OS, P = .027).
In our study, we performed ctDNA testing using a personalized, tumor-informed assay and detected MRD in the early postdefinitive local treatment setting, that is, within 3 months postsurgery (prior to adjuvant CRT/RT) or within 3 months postdefinitive CRT/RT. We observed that 61.5% (8/13) of patients who were MRD-positive experienced disease progression, compared with 14.3% (8/56) of patients who were MRD-negative (HR, 6.39; 95% CI, 2.39-17.09; P = .0002). After excluding patients who received adjuvant therapy postsurgery/ after definitive treatment, 85.7% (6/7) of patients who were MRD-positive experienced disease progression, compared with only 13.6% (6/44) of patients who were MRD-negative (HR, 10.08; 95% CI, 3.23-31.46; P < .0001). Moreover, compared with patients who were MRD-negative, MRD positivity was associated with profoundly inferior OS (HR, 25.72; 95% CI, 3.00-220.48; P < .0001), demonstrating a 24-month OS of 56.10% versus 98.20%, respectively. Although the timing for ctDNA testing posttreatment varies across studies, our findings reinforce the utility of ctDNA-based MRD assessment within 3 months following postdefinitive therapy for identifying patients at high risk of recurrence. This opens avenues for personalized treatment planning: escalation for high-risk patients or deintensification for those with low recurrence risk.
Importantly, our study also reveals the potential of longitudinal ctDNA-based MRD monitoring for surveillance beyond the early postdefinitive local treatment setting. During postdefinitive treatment surveillance, compared with patients who were serially MRD-negative, those with MRD positivity at any time point were 38 times more likely to experience recurrence (HR, 38.87; 95% CI, 9.11-165.88; P < .0001; 24-month PFS: 20.8% vs 96.4%). Moreover, longitudinally during postdefinitive treatment surveillance, where effects of adjuvant treatment efficacy are not confounding, the tumor-informed assay exhibited a specificity of 96.4% (53/55). MRD positivity preceded radiographic progression by a median of 4.3 months (range, 0.62-21.91 months). There were 2 patients (UNM-089 and UNM-092; 2/97, 2%) who were MRD-positive during surveillance despite being recurrence-free at the end of follow-up, both of whom had a single positive test with ctDNA levels close to the limit of detection, followed by serial MRD-negativity. Notably, in the p16-positive subgroup, we observed a sensitivity of 100% (9/9) and a specificity of 97% (32/33) in the surveillance setting. Similar findings were reported in a prospective biomarker clinical trial among patients (N = 45) with nonmetastatic p16-positive oropharyngeal squamous cell carcinoma, where ctHPV DNA detection during surveillance identified patients with clinical recurrence with 100% sensitivity and specificity and a median lead time of 6.6 months.19 Although it is important to note that 100% specificity was observed in that study only after applying 2 consecutive ctHPV DNA–positive tests as the criteria for detecting recurrence, because 3 of 45 (8%) patients had spontaneous clearance of ctHPV DNA without any treatment.
We further illustrate the impact of longitudinal MRD testing on clinical management of patients with HNSCC. MRD positivity postdefinitive treatment was found to be an early indicator of molecular relapse and prompted escalated surveillance. In some patients, such as UNM-069, an extensive lead time of ∼21 months over radiographic imaging was observed in detecting recurrence. However, in patients such as UNM-031, earlier surveillance scanning enabled early detection of recurrence within 9 months from the start of definitive CRT. This facilitated timely salvage therapy and provided a second chance at cure to the patient.
Overall, to our knowledge, we present the first real-world study to evaluate the role of longitudinal ctDNA-based MRD evaluation in HNSCC, regardless of p16 status, for surveillance beyond the early postdefinitive local treatment period. Although high metabolic tumor burden measured by 18F-fluorodeoxyglucose positron emission tomography/computed tomography has been associated with ctDNA positivity and high maximum variant allele frequency,20 ctDNA detection has been found to be a more sensitive predictor for recurrence when compared with radiologic imaging.21 Our results suggest that surveillance strategies incorporating ctDNA could complement traditional imaging and clinical examinations, especially as surveillance intensity typically decreases beyond the first 2 years posttreatment.2
Nonetheless, this study has limitations, including its retrospective design, potential selection bias, variability in treatment regimens, and heterogeneity in the cadence of ctDNA testing. Despite these limitations, the robustness of our multivariate models and the consistent prognostic value of ctDNA across stages and p16 statuses lend strong support to its clinical utility. Importantly, we observed that ctDNA-based MRD status outperformed all conventional clinicopathologic risk factors in predicting disease progression during surveillance.
Overall, this real-world study provides evidence that longitudinal ctDNA monitoring, using a tumor-informed platform, enables early detection of recurrence in patients with HNSCC—irrespective of HPV status. Our data support the integration of ctDNA MRD into postdefinitive treatment patient management to inform individualized care. Future prospective studies are needed to evaluate the clinical impact, safety, and cost-effectiveness of MRD-guided treatment strategies in this population.
Disclosures
George Laliotis, Charuta C. Palsuledesai, Giby V. George, Michael Krainock, and Minetta C. Liu are employees of Natera, Inc, receive a salary, and may own stock and/or stock options. George Laliotis also reports receiving an honorarium from Tegus and providing medical advisory support for DocusAI. Abhinand V. Peddada reports stock ownership in Natera Inc. Sydney F. Denney declares no conflicts of interest.
Footnotes
Sources of support: This work had no specific funding.
All relevant data used to conduct the analyses are available within the article. Complete clinical information and ctDNA results are presented in the overview plot (Fig. 2). Patient characteristics and demographic information are available in Table 1 and Table E1. To ensure compliance with the ethics committee and to protect the privacy and confidentiality of patients in this study, the sequencing data and source data supporting the findings are not made publicly available. However, they can be requested from the corresponding author for academic use only, within the limitations of the provided informed consent. Data will not be made available for commercial use. Any requests will be reviewed within a timeframe of 2 to 4 weeks to verify whether the request is subject to any intellectual property or confidentiality obligations. All data shared will be deidentified and provided to researchers with access limited for scientific verification purposes, subject to strict prohibitions on secondary use. Applying researchers will be required to sign a data usage agreement.
The fully documented code for the R statistical computing environment for analyses related to this manuscript is deposited in the GitHub repository and can be accessed at the GitHub registry (https://github.com/Natera-TMED/Althoff-et-al_CLIA-HNSCC-UNM).
Supplementary material associated with this article can be found in the online version at doi:10.1016/j.adro.2026.102205.
Appendix. Supplementary materials
Supplementary Materials
Supplementary material associated with this article can be found in the online version at [comp will supply].
References
- 1.Kang H., Kennedy T.A., Yu E. In: Diseases of the Brain, Head and Neck, Spine 2024-2027. Hodler J., Kubik-Huch R.A., Roos J.E., editors. Springer; 2024. Head and neck squamous cell cancer: Approach to staging and surveillance; pp. 251–264. [PubMed] [Google Scholar]
- 2.National Comprehensive Cancer Network . Invasive Breast Cancer (Version 3.2025); 2025. NCCN Clinical Practice Guidelines in Oncology (NCCN Guidelines®)https://www.nccn.org Accessed November 01, 2025. [Google Scholar]
- 3.Martin T.K., Dinerman A., Sudhaman S., et al. Early real-world experience monitoring circulating tumor DNA in resected early-stage non-small cell lung cancer. J Thorac Cardiovasc Surg. 2024;168:1349–1359.e2. doi: 10.1016/j.jtcvs.2024.01.017. [DOI] [PubMed] [Google Scholar]
- 4.Oh Y., Yoon S.M., Lee J., et al. Personalized, tumor-informed, circulating tumor DNA assay for detecting minimal residual disease in non-small cell lung cancer patients receiving curative treatments. Thorac Cancer. 2024;15:1095–1102. doi: 10.1111/1759-7714.15281. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Lebow E.S., Shaverdian N., Eichholz J.E., et al. ctDNA-based detection of molecular residual disease in stage I-III non-small cell lung cancer patients treated with definitive radiotherapy. Front Oncol. 2023;13 doi: 10.3389/fonc.2023.1253629. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Bratman S.V., Yang S.Y.C., Iafolla M.A.J., et al. Personalized circulating tumor DNA analysis as a predictive biomarker in solid tumor patients treated with pembrolizumab. Nat Cancer. 2020;1:873–881. doi: 10.1038/s43018-020-0096-5. [DOI] [PubMed] [Google Scholar]
- 7.Nakamura Y., Watanabe J., Akazawa N., et al. ctDNA-based molecular residual disease and survival in resectable colorectal cancer. Nat Med. 2024;30:3272–3283. doi: 10.1038/s41591-024-03254-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Reinert T., Henriksen T.V., Christensen E., et al. Analysis of plasma cell-free DNA by ultradeep sequencing in patients with stages I to III colorectal cancer. JAMA Oncol. 2019;5:1124–1131. doi: 10.1001/jamaoncol.2019.0528. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Foldi J., Oesterreich S., Tin A., et al. Personalized circulating tumor DNA testing for detection of progression and treatment response monitoring in patients with metastatic invasive lobular carcinoma of the breast. JCO Precis Oncol. 2025;9 doi: 10.1200/PO-24-00577. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Magbanua M.J.M., Brown Swigart L., Ahmed Z., et al. Clinical significance and biology of circulating tumor DNA in high-risk early-stage HER2-negative breast cancer receiving neoadjuvant chemotherapy. Cancer Cell. 2023;41:1091–1102.e4. doi: 10.1016/j.ccell.2023.04.008. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Coombes R.C., Page K., Salari R., et al. Personalized detection of circulating tumor DNA antedates breast cancer metastatic recurrence. Clin Cancer Res. 2019;25:4255–4263. doi: 10.1158/1078-0432.CCR-18-3663. [DOI] [PubMed] [Google Scholar]
- 12.Flach S., Howarth K., Hackinger S., et al. Liquid bIOpsy for miNimal rESidual diSease detection in head and neck squamous cell carcinoma (LIONESS)—A personalised circulating tumour DNA analysis in head and neck squamous cell carcinoma. Br J Cancer. 2022;126:1186–1195. doi: 10.1038/s41416-022-01716-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Hanna G.J., Dennis M.J., Scarfo N., et al. Personalized ctDNA for monitoring disease status in head and neck squamous cell carcinoma. Clin Cancer Res. 2024;30:3329–3336. doi: 10.1158/1078-0432.CCR-24-0590. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Honoré N., Laliotis G., Aushev V., et al. Tumor-informed ctDNA assay to predict recurrence in locally advanced squamous-cell carcinoma of the head and neck (SCCHN) ESMO Open. 2025;10 doi: 10.1016/j.esmoop.2025.104534. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Yang R., Li T., Zhang S., Shui C., Ma H., Li C. The effect of circulating tumor DNA on the prognosis of patients with head and neck squamous cell carcinoma: A systematic review and meta-analysis. BMC Cancer. 2024;24:1434. doi: 10.1186/s12885-024-13116-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Routman D.M., Van Abel K.M., Price K.A., et al. ctDNA and recurrence risk for adjuvant de-escalation in HPV-positive oropharyngeal carcinoma: A secondary analysis of the DART phase 3 randomized clinical trial. JAMA Otolaryngol Head Neck Surg. 2025;151:665–672. doi: 10.1001/jamaoto.2025.0903. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Kogo R., Manako T., Iwaya T., et al. Individualized circulating tumor DNA monitoring in head and neck squamous cell carcinoma. Cancer Med. 2022;11:3960–3968. doi: 10.1002/cam4.4726. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Honoré N., Laliotis G., Aushev V.N., et al. Tumor-informed ctDNA assay to predict outcome in recurrent and/or metastatic squamous cell carcinoma of the head and neck treated with PD-1 inhibitor. Clin Cancer Res. 2025;31:4790–4801. doi: 10.1158/1078-0432.CCR-25-1309. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Chera B.S., Kumar S., Shen C., et al. Plasma circulating tumor HPV DNA for the surveillance of cancer recurrence in HPV-associated oropharyngeal cancer. J Clin Oncol. 2020;38:1050–1058. doi: 10.1200/JCO.19.02444. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Silvoniemi A., Laine J., Aro K., et al. Circulating tumor DNA in head and neck squamous cell carcinoma: Association with metabolic tumor burden determined with FDG-PET/CT. Cancers (Basel) 2023;15:3970. doi: 10.3390/cancers15153970. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Chikuie N., Urabe Y., Ueda T., et al. Utility of plasma circulating tumor DNA and tumor DNA profiles in head and neck squamous cell carcinoma. Sci Rep. 2022;12:9316. doi: 10.1038/s41598-022-13417-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Supplementary Materials
Supplementary material associated with this article can be found in the online version at [comp will supply].