NEW YORK – Several different molecular methods for detecting lingering or reemerging tumor DNA in the bloodstream or saliva of patients with head and neck cancers (HNC) are being advanced, according to data presented at the American Society of Clinical Oncology's annual meeting on Monday.
HNC is revealing itself to be a newly favored target for the application of minimal residual disease testing, due to the technology's potential to improve outcomes in a cancer type where initial curative-intent surgery often fails to prevent relapse, and physicians believe earlier detection could help them more successfully employ adjuvant treatment.
During a webcast meeting session on progress in circulating biomarkers, two studies showcased two very different molecular approaches to MRD detection: one a tumor-informed patient specific technology and the other a tumor agnostic methylation-based approach.
Yale University School of Medicine's Barbara Burtness, the session's discussant, said she hoped attendees would leave the session convinced of the analytical value of both approaches. However, she stressed that prospective validation is required before such assays can be used to guide therapy outside of the clinical trial setting.
In one presentation, Susanne Flach, a clinician-scientist at the University Hospital of Ludwig Maximilian University of Munich, shared updated data from the LIONESS trial. The single-center prospective cohort study is aimed at determining whether a personalized, tumor-informed liquid biopsy test technology called Radar — developed by Inivata and now owned by NeoGenomics — can differentiate patients who have cleared their ctDNA versus those that have not, and whether it can detect emerging relapse before it appears clinically.
"Current standards of care for locally advanced HNC are surgery followed by adjuvant treatment," Flach said. "However, around 30 percent of patients with HNC go on to develop local [or] regional recurrences or a second primary tumor, often within two years after [the] completion of treatment."
Flach shared data from 76 patients recruited to the study, all of whom received primary surgery with curative intent. About a third of the patients had stage I or II disease, and two-thirds stage III or IV. The team collected plasma from all, plus saliva samples from a subset of 54 patients, and used archived tumor tissue to design personalized Radar assays for each individual.
Overall, researchers studied 617 longitudinal plasma and 128 saliva samples with a median patient follow-up time of 805 days. Of 23 cases with confirmed clinical recurrences, 96 percent had increased ctDNA levels detected in their plasma after surgery, with a median lead time of 116 days.
Of the remaining 54 patients with no clinically confirmed relapse, ctDNA was detected immediately post-op in seven. In five patients, adjuvant therapy resulted in persistent ctDNA clearance, while two patients died, the team reported.
Flach said that the plasma and saliva versions of the tests could detect ctDNA in 96 percent of patients prior to treatment. The overall concordance of the two analytes was 69 percent.
"I think that the 96 percent positivity pretreatment when you combine plasma and saliva is an extremely potent finding," said Burtness, in discussing the results. "It outperforms many of the other assays people have been looking at."
Burtness added that the predictive value for recurrence of a positive ctDNA signal "really stands out." That said, she noted that the need for tumor tissue sequencing is a limitation.
Flach highlighted three cases from the cohort. For one, a patient with a laryngeal tumor who received laser resection and bilateral neck dissection, the tumor board recommended follow-on endoscopy at eight weeks. At that point, the recurrence of disease was systematically confirmed. The patient declined surgery and received definitive radiotherapy for the local recurrence.
When they examined the patient's ctDNA, Flach and her team saw continued presence of ctDNA postoperatively, indicating residual disease that lingered until the patient received the definitive radiotherapy, rather than a true recurrence. The patient now remains ctDNA-negative, which is "interesting," Flach said, "when you think about what would have happened if the patient had had adjuvant radiotherapy" right away, referring to the observation of immediate post-surgical, lingering ctDNA.
A second laryngeal cancer patient who had endoscopy with negative biopsies initially and a negative CT scan during their post-surgery follow-up later had endoscopy and biopsy confirm a large loco-regional recurrence. The patient had a laryngectomy and has been recurrence-free to date.
In this patient, the researchers were able to detect ctDNA with a lead time of 500 days.
Finally, Flach described a patient who had a hypopharyngeal tumor that was resected with a laryngectomy and who underwent adjuvant chemotherapy. During follow-up, a chest scan showed a pulmonary lesion that was initially suspected to be a second primary tumor. But by comparing the initial HNC panel for the patient to the mutational profile of this new lung tumor, the researchers found them to be "more or less identical."
"Looking back at the longitudinal ctDNA data, levels had been rising up until the point that the patient received, and responded well, to chemoimmunotherapy," Flach said.
Still, as with other technologies, Burtness noted that the "utility for patient management has not been demonstrated."
In another presentation during the session, Geoffrey Liu, a senior scientist at Princess Margaret Cancer Center in Toronto, presented data from his and his colleagues' exploration of a different approach to predict and detect early relapse in head and neck cancers: a methylation sequencing method developed by Liu's colleague Daniel De Carvalho called cell-free methylated DNA immunoprecipitation and high-throughput sequencing (cfMeDIP-seq).
The technology has been licensed to a spinout company, Adela, which is exploring cfMeDIP-seq for a variety of applications, but with most of its focus on multi-cancer early detection.
In their HNC study, Liu and colleagues tested more than 1,000 biobanked samples from 325 individuals diagnosed with stage I to stage IV tumors. The team split these samples into distinct training and validation sets, looking at the performance of their sequencing assay both at a landmark time point three months after surgery and longitudinally.
"There are advantages to blood-based approaches to detect recurrences, particularly if they do not require tumor tissue," Liu said. "With head and neck cancer [in particular], we usually end up with quite small biopsies to start off with," which limits the tumor DNA available. A blood-only approach could also mean a quicker turnaround time, he added.
Liu's presentation at ASCO was limited to the study's training cohort, which consisted of 130 patients with 432 samples. Among these individuals, 45 had a recurrence and 85 did not.
Overall, the team concluded that MRD detection using cfMeDIP-seq correlated strongly with reported recurrence-free survival with hazard ratios consistent with what has been described previously for tumor-informed assays.
Focusing on the three-month post-surgery landmark time point, the investigators calculated that assay positivity was associated with significantly lower recurrence-free survival. The hazard ratio improved even further when they considered test-positivity any time during longitudinal testing.
The test could detect molecular signs of recurrence up to 19.3 months before it manifested clinically, with a mean of 5.1 months across the cohort.
Liu noted, though, that the analysis had a relatively low frequency of surveillance blood draws. "We started collecting these samples years before we had any idea that these assays would be available. And so there's a potential for underestimation for both the detectability and also for potentially underestimation of the lead time, as well," he said.
The readout in the study's holdout independent validation cohort will also be crucial to confirm the current findings.
Burtness, the session discussant, said that it is encouraging that developers are continuing to refine technologies, but the field will need to take a hard look at how to ultimately validate the role of MRD tests in head and neck cancer patient management, as has begun to be done in other areas like early-stage colorectal cancer.
The head and neck cancer space already has a commercially available assay for blood-based detection of disease, Naveris' digital PCR-based NavDx test for HPV-associated tumors. There is much more substantial literature for that test's prognostic ability, and the company is now moving into the kind of studies that would validate it for specific clinical uses. For instance, a Phase II trial in partnership with Memorial Sloan Kettering Cancer Center is to use NavDx to monitor patients treated for HPV-driven head and neck tumors. In those showing molecular recurrence, the study will test the efficacy of the investigational treatment HB-200.
Burtness also cited growing interest in boosting assay sensitivity even further, highlighting another abstract presented at a separate session that day, which shared data on a whole-genome, hybrid-capture, next-generation sequencing approach called HPV-DeepSeek, being advanced by a team at Harvard Medical School.
"With this assay, first, it [has] much more sensitive detection at baseline. And second, they showed a very dramatic prediction of two-year progression-free survival, segregating a 60 percent from a 97 percent population," said Burtness.
The Harvard team behind HPV-DeepSeek believes the platform could also serve as a screening test, detecting these tumors before they present clinically.
"I see a lot of challenges as we take these technologies forward," Burtness said. "How do we translate the sensitivity and specificity that are observed in small trials to large trials in a time frame and at a cost that they have a hope of influencing patient management? How do we demonstrate actionability?"
"Moreover, are we able to use intensified therapies that actually produce better functional outcomes? Are we improving cure rates?" she added.
"We saw in these studies that the tissue agnostic methylation assays strongly correlate with risk of recurrence and detection of circulating tumor DNA using a personalized assay is very strongly associated with risk of recurrence," Burtness concluded. "But these assays are not ready for clinical use until we have studied them in trials where treatment paradigms are driven by ctDNA."