Post-Treatment ctDNA Positivity Is Associated With Recurrence in Locally Advanced HNSCC

Post-treatment circulating tumor DNA (ctDNA) positivity per the RaDaR assay was associated with disease recurrence, including in a subgroup of patients with p16-positive oropharynx cancer, in those with locoregionally advanced head and neck squamous cell carcinoma (HNSCC), according to phase 2 MERIDIAN trial (NCT05414032) findings presented at the 2026 AACR Annual Meeting.¹

The sensitivity of the RaDaR assay also improved with serial monitoring. Additionally, although the use of a human papillomavirus (HPV) DNA assay conferred higher sensitivity than RaDaR, it had lower specificity in patients with p16-positive oropharynx cancer.

“MERIDIAN is the first clinical trial to evaluate molecular residual disease [MRD] interception in head and neck cancer,” Enrique Sanz García, MD, PhD, said to OncLive^®. “Despite [the] study [being] paused due to lower than expected [numbers of] MRD-positive patients, the patient who underwent interception with rilvegostomig [AZD2936] had clearance of ctDNA and remains disease free [approximately] 2 years from the end of therapy.”

Sanz García is a staff medical oncologist at the Princess Margaret Cancer Centre and an assistant professor at the University of Toronto in Ontario, Canada.

What was the rationale for conducting the MERIDIAN trial in HNSCC?

Sanz García noted that patients with locally advanced HNSCC treated with curative-intent therapy have a 30% chance of recurrence within the first 3 years of treatment. Additionally, the observational PRE-MERIDIAN study (NCT04599309) showed that detecting MRD by ctDNA after definitive therapy was associated with inferior recurrence-free survival (RFS) and overall survival (OS) outcomes in patients with this disease compared with those with ctDNA-negative disease. Data from PRE-MERIDIAN showed that, at a median follow-up of 25 months (range, 5.1-32.0), 7 of 32 patients analyzed experienced clinical relapses.² Additionally, 88% of evaluable patients (n = 15/17) had ctDNA-positive disease at baseline per RaDaR, which was not associated with RFS, and the detection of ctDNA during follow-up was associated with shorter RFS (P < .001).

The MERIDIAN investigators hypothesized that treatment with rilvegostomig in patients with MRD-positive locally advanced HNSCC may clear ctDNA and lead to prolonged RFS and increased rates of cure after definitive therapy.¹

What was the design of the MERIDIAN HNSCC trial?

This randomized, single-center study enrolled patients with HNSCC of the oral cavity, oropharynx, larynx, hypopharynx, or unknown primary.

To enroll in parts A and B of the trial, patients needed to have stage II disease with radiological extranodal extension or stage III HPV-positive disease, stage III IVA-B and HPV-negative disease, or recurrent disease. Patients also needed to have adequate organ function and be candidates for curative-intent surgery followed by adjuvant therapy, definitive radiotherapy, or definitive chemoradiotherapy. Patients were not permitted to have contraindications for immunotherapy.

To enroll in parts C and D, patients needed to have ctDNA-positive disease at the end of part B, a PD-L1 combined positive score of at least 1, and no evidence of clinical or radiological progression at the end of part B.

Notably, enrollment was paused after 52 patients had enrolled because only 2 had MRD-positive disease, allowing enrollment in parts C and D.

In part A, all enrolled patients underwent definitive treatment for at least 7 weeks with radical radiotherapy with or without chemotherapy, or surgery followed by adjuvant radiotherapy with or without chemotherapy. In part B, over 12 to 18 weeks, patients were assessed for MRD using ctDNA. In part C, patients with MRD-positive disease at the end of part B were randomly assigned 3:1 to receive 6 cycles of rilvegostomig at 750 mg every 3 weeks or observation, and were then followed for progressive disease (PD; part D); patients with MRD-negative disease at the end of part B were followed for PD (part E).

Real-time ctDNA analysis was performed using the tumor-informed, whole exome sequencing–based, patient-specific RaDaR tumor-informed ctDNA assay, which targets a maximum of 48 variants, or an HPV DNA sequencing assay using ultrasensitive next-generation sequencing. ctDNA analyses were performed at the following time points:

• Pretreatment in all patients
  • Assay: retrospective HPV DNA
• Part A: baseline and post-surgery (for patients who underwent surgery)
  • Assay: real-time RaDaR
• Part B: weeks 4 to 6 and weeks 8 to 12 (as well as weeks 14 to 16 if prior findings were equivocal)
  • Assay: real-time RaDaR
• Part C: week 10
  • Assay: real-time RaDaR
• Part D: weeks 2 and 10
  • Assay: real-time RaDaR
• Part D months 6, 12, 18, 24, and 36; and at the development of PD
  • Assay: retrospective RaDaR
• Part E: 2 to 6 months at the first in-person assessment, and at the development of PD
  • Assay: retrospective RaDaR
• Follow-up samples in patients with p16-positive oropharynx cancer
  • Assay: retrospective HPV DNA

“The most important time point was at 8 to 12 weeks post the end of definitive therapy,” Sanz García explained in the presentation of the data. “This was the [time point] that helped us to decide whether the patient was MRD positive, provided there was no radiological or clinical progression.”

The primary end point was ctDNA clearance in MRD-positive patients, defined as the absence of detectable ctDNA in part D weeks 2 and 10. Additional end points included RFS, OS, the safety profile of rilvegostomig, HPV DNA outcomes, quality of life, and health economics.

What were the baseline characteristics of patients enrolled in MERIDIAN?

Among 46 enrolled patients, the median age at diagnosis was 63 years (range, 36-80), and most were male (76%). Among the 39 patients with tumor in situ, 97.4% had ctDNA-positive disease by RaDaR at baseline.

Primary tumor sites included the oropharynx (56.5%), oral cavity (23.9%), larynx (15.2%), hypopharynx (2.2%), and unknown (2.2%). In patients with oropharynx cancer, the concordance rate between HPV DNA plasma and p16 status at baseline was 96%.

TNM stages included III HPV positive (41.3%), III HPV negative (13%), IVA (21.7%), and IVB (23.9%). Definitive treatments included chemoradiotherapy (52.2%), radiotherapy (17.4%), surgery followed by radiotherapy (15.2%), surgery followed by chemoradiotherapy (13%), and surgery alone (2.2%).

What additional efficacy findings were reported with the use of ctDNA testing in MERIDIAN?

Among the 42 patients evaluable for ctDNA analysis following definitive treatment, 5 had ctDNA-positive disease, 2 had equivocal ctDNA results, and 35 had ctDNA-negative disease. At additional follow-up, 1 patient each in the equivocal group had ctDNA-positive or -negative disease, yielding a total ctDNA-positive detection rate of 14% at the end of part B.

All 6 patients with ctDNA-positive disease were screened for trial continuation in parts C and D; screening failed in 4 patients, leaving 1 patient in each group to receive treatment vs undergo observation. The patient who received rilvegostomig remained without evidence of disease, whereas the patient assigned to observation experienced PD.

All 35 patients with ctDNA-negative disease were screened for trial continuation in part E, and 35 were enrolled. At the end of part E, 30 patients were alive with no recurrence, 3 patients were alive with recurrence, 1 patient was dead with recurrence, and 1 patient was dead without recurrence. Overall, PD occurred in 23% of evaluable patients.

Sanz García explained that 3 patients had notable lead times from the detection of ctDNA positivity to PD, at 76, 190, and 235 days. Additionally, only 1 patient had positive ctDNA results during surveillance but did not experience radiological progression. However, Sanz García reported that because the follow-up time for this patient was below 1 year at data cutoff, more time is needed to determine whether this indicates a false-positive ctDNA result.

Sanz García also stated that all tumor samples collected at the time of PD were ctDNA positive. Additionally, he highlighted the correlation between ctDNA detection and worse outcomes.

What were the trial outcomes in the patient who received rilvegostomig?

This patient was diagnosed with stage III HPV-positive oropharyngeal cancer and received definitive treatment with standard-of-care chemoradiation. Their ctDNA levels decreased following this definitive treatment, then increased slightly at the end of part B with no evidence of radiological or clinical progression, making them eligible for the interception portion of the trial. The patient then received 5 of the 6 planned doses of rilvegostomig; the patient then requested treatment discontinuation due to the development of grade 2 tongue pain and grade 2 neck edema.

Shortly after rilvegostomig initiation, the patient cleared their ctDNA; this lasted for 12 months after rilvegostomig discontinuation. At data cutoff, the patient remained disease free at 21.9 months from the end of definitive treatment.

How did the RaDaR assay perform in MERIDIAN?

At the first follow-up time point in part B, RFS outcomes favored patients with ctDNA-negative disease (n = 42; HR, 13.8; 95% CI, 4.3-44.7; P < .001), and the RaDaR analysis showed the following:

• Sensitivity rate: 64% (95% CI, 31%-89%)
• Specificity rate: 100% (95% CI, 89%-100%)
• Positive predictive value (PPV) rate: 100% (95% CI, 59%-100%)
• Negative predictive value (NPV) rate: 89% (95% CI, 73%-97%)
• Accuracy rate: 90% (95% CI, 77%-97%)

Between the second and third follow-up time points (MRD window) in part B, RFS outcomes still favored patients with ctDNA-negative disease (HR, 16.3; 95% CI, 4.3-62; P < .001), and the RaDaR analysis showed the following:

• Sensitivity rate: 50% (95% CI, 19%-81%)
• Specificity rate: 100% (95% CI, 89%-100%)
• PPV rate: 100% (95% CI, 48%-100%)
• NPV rate: 86% (95% CI, 71%-95%)
• Accuracy rate: 88% (95% CI, 74%-96%)

At any follow-up time between parts B and E, RFS outcomes continued to favor patients with ctDNA-negative disease (HR, 11.8; 95% CI, 3.5-40.3; P < .001), and the RaDaR analysis showed the following:

• Sensitivity rate: 73% (95% CI, 39%-94%)
• Specificity rate: 97% (95% CI, 83%-100%)
• PPV rate: 89% (95% CI, 52%-100%)
• NPV rate: 91% (95% CI, 76%-98%)
• Accuracy rate: 90% (95% CI, 77%-97%)

The patient who received rilvegostomig was excluded from this RFS analysis because they received interception therapy.

How did the HPV DNA assay perform in patients with p16-positive oropharynx cancer in the MERIDIAN trial?

Per the HPV DNA analysis in patients with p16-positive oropharynx cancer, 18 patients were evaluable for disease recurrence. HRs were not reported because there were insufficient events to estimate a meaningful HR. The patient who received rilvegostomig was also excluded from this RFS analysis since they received interception therapy.

At the first follow-up time point in part B, RFS outcomes favored patients with ctDNA-negative disease (P = .043), and the HPV DNA analysis showed the following:

• Sensitivity rate: 67% (95% CI, 9%-99%)
• Specificity rate: 87% (95% CI, 60%-98%)
• PPV rate: 50% (95% CI, 7%-93%)
• NPV rate: 93% (95% CI, 66%-100%)
• Accuracy rate: 83% (95% CI, 59%-96%)

Between the second and third follow-up time points (MRD window) in part B, the RFS outcomes continued to favor patients with ctDNA-negative disease (P = .020), and the HPV DNA analysis showed the following:

• Sensitivity rate: 100% (95% CI, 29%-100%)
• Specificity rate: 73% (95% CI, 45%-92%)
• PPV rate: 43% (95% CI, 10%-82%)
• NPV rate: 100% (95% CI, 72%-100%)
• Accuracy rate: 78% (95% CI, 52%-94%)

At any follow-up time between parts B and E, the RFS outcomes still favored patients with ctDNA-negative disease (P = .109), and the HPV DNA analysis showed the following:

• Sensitivity rate: 100% (95% CI, 29%-100%)
• Specificity rate: 53% (95% CI, 27%-79%)
• PPV rate: 30% (95% CI, 7%-65%)
• NPV rate: 100% (95% CI, 63%-100%)
• Accuracy rate: 61% (95% CI, 36%-83%)

“MERIDIAN has been amended and [is] currently recruiting patients using a potentially more sensitive assay, longitudinal analysis (every 3 months), and removing randomization to allow more patients to benefit from interception strategies,” Sanz García concluded in his comments to OncLive.

Disclosures: Sanz García reported receiving consulting fees from GSK; receiving speaker’s bureau fees from Incyte; and receiving grant/research support from Regenta Therapeutics, GSK, and AstraZeneca.

References

1. Sanz García E, Spreafico A, Hosni A, et al. Molecular residual disease (MRD) interception in locoregionally advanced head and neck squamous cell carcinoma (LA-HNSCC): the MERIDIAN phase II trial. Cancer Res. 2026;86(suppl 8):CT173-CT173. doi:10.1158/1538-7445.AM2026-CT173
2. Sanz-Garcia E, Zou J, Avery L, et al. Multimodal detection of molecular residual disease in high-risk locally advanced squamous cell carcinoma of the head and neck. Cell Death Differ. 2024;31(4):460-468. doi:10.1038/s41418-024-01272-y