Surveillance and Monitoring Techniques for HPV-Related Head and Neck Squamous Cell Carcinoma: circulating tumor DNA

Opinion Statement

Human papilloma virus (HPV) related head and neck cancer is rising in prevalence, preferentially affecting young patients and imparting long term toxicities. Despite this, there are no screening tests or clinical biomarkers for treatment monitoring. HPV circulating tumor DNA (HPV ctDNA) represents a novel circulating biomarker which may provide real-time assessment of tumor response to therapy and recurrence. Early work suggests the promise of this assay as a predictive biomarker in numerous clinical settings, namely risk of recurrence after chemoradiation in locally advanced disease. Advancement of these findings to the clinic will require a collaborative effort in the field, including technical harmonization of assay testing characteristics, understanding of the normal kinetics in patients being treated with standard of care therapies, and appropriately designed phase III trials prior to implementation in the clinic. If successful, HPV ctDNA has the potential to revolutionize clinical trial treatment paradigms and transform patient care.

Introduction

There is an increasing incidence of human papilloma virus (HPV) related cancers of the oropharynx [1]. Estimates predict that HPV-positive oropharyngeal cancer (OPC) will account for the majority of all head and neck cancers within a decade [2]. HPV-positive head and neck squamous cell cancers (HNSCC) may arise from any site in the upper aerodigestive tract, although the oropharynx represents the primary site for the vast majority of cases. Despite this rise in cases, there are no molecular screening tests that currently exist and therefore the majority of cases are detected due to progressive symptoms such as a neck nodal mass. Accordingly, the only means to assess response to treatment are physical exam and cross-sectional imaging. Both have significant limitations, including need for assessment at a medical facility by a specialist and exposure to ionizing radiation.

Biomarkers are used to help guide treatment decisions and judge prognosis throughout the field of oncology; however, head and neck oncology has had a paucity of clinically useful biomarkers, outside of HPV or more recently, PD-L1. Development of a dynamic biomarker has been of significant interest to the field of head and neck oncology as a means of real-time disease assessment. Although biomarkers have been traditionally identified from tumor tissue, they can also be collected non-invasively in other body fluids including blood (e.g., plasma, saliva). Non-invasive collection allows longitudinal collection of larger volumes of samples as tissue biopsies are not needed to obtain tumor tissue. Blood-based biomarkers may be utilized in many stages of the disease course including screening, diagnosis, measurement of therapeutic response, prognostication, and disease surveillance.

The surge of HPV-positive disease presents a unique opportunity to leverage tumor characteristics for improved monitoring and surveillance. HPV-positive tumors shed viral DNA sequences that can be detected in bodily fluid as circulating tumor DNA (HPV ctDNA). Quantifying and monitoring levels of HPV ctDNA has shown promise as a biomarker in patients with HPV-positive HNSCC. Early research examining HPV ctDNA in patients with HNSCC has demonstrated the viability of using this as a biomarker. Current research suggests that HPV ctDNA may aid in monitoring treatment response and surveillance of disease recurrence. Using HPV ctDNA as a marker to predict recurrence prior to radiographic or physical evidence of disease could revolutionize the current standard practice. This review serves to provide an update on the most recent studies examining the use of HPV ctDNA for early detection of recurrence, assessment of treatment response, and profiling of resistance patterns.

Human Papillomavirus (HPV) Associated Head and Neck Squamous Cell Carcinoma

Treatment of HPV-positive HNSCC typically involves multimodal strategies which may include radiation, chemotherapy, and surgery. Although HPV-positive disease confers a more favorable prognosis than HPV-negative disease, 20–25% of patients with HPV-positive cancer will develop recurrent disease within five years of treatment [3–5]. Even in instances of recurrence, there is survival benefit for those with HPV-positive disease after salvage therapy when compared with HPV-negative disease [6]. As for any head and neck cancer, post-treatment surveillance is crucial to identifying early recurrence of disease.

Post-treatment surveillance for HPV-positive HNSCC relies on comprehensive physical examination and serial radiological imaging. The National Comprehensive Cancer Network (NCCN) guidelines for surveillance of disease in patient with HPV-positive HNSCC include clinical examination every 1 to 3 months in the first year following treatment, every 2 to 6 months in the second year, every 4 to 8 months in the third through fifth years, and yearly examinations thereafter. Standard surveillance imaging includes positron emission tomography computed tomography (PET/CT) scan 3 months after definitive treatment [7]. According to NCCN guidelines, additional post-treatment imaging may be considered at 6 months post-treatment and then yearly [7]. These surveillance modalities are unable to detect minimal residual disease or early recurrence.

Using HPV ctDNA as a post-treatment surveillance tool may allow for early detection of disease recurrence and improved survival. HPV-positive disease is associated with a longer post-recurrence survival when compared to HPV-negative disease, but the likelihood of cure is low when disease recurrence is diagnosed at late stage [2, 8, 9]. Early identification may influence treatment course and prompt initiation of salvage therapy sooner than would be indicated by traditional surveillance methods.

Early Detection of Recurrence

The use of ctDNA for surveillance of disease relies on detecting and monitoring tumor-specific genotypes in the circulation. ctDNA has been shown to be useful in evaluating treatment response in a variety of cancers. Using ctDNA biomarkers as a means of surveillance and monitoring of disease is well established for breast, colorectal, and lung cancers [10–13]. Early studies have suggested that monitoring levels of viral DNA sequences in virally-mediated diseases, such as HPV-associated HNSCC, may be useful in assessing treatment resistance, tumor burden, and recurrence of disease [14–18].

HPV ctDNA has been shown to be a sensitive and specific biomarker of HPV-positive HNSCC [5, 19–22]. Use of HPV ctDNA assays for surveillance of disease may allow for earlier detection of recurrent disease and earlier intervention. For patients with HPV-positive disease, HPV ctDNA is frequently detectable at the time of diagnosis and HPV ctDNA levels correlate with overall disease burden [19–22]. If accurate, longitudinal monitoring of HPV ctDNA could serve as a powerful tool for assessment of disease status and surveillance for recurrence in patients with HPV associated HNSCC.

While most of the available literature on HPV ctDNA in HNSCC originates from small cohort and pilot studies, prospective clinical trials provide a higher level of evidence in regard to the clinical utility and feasibility of using HPV ctDNA as a surveillance biomarker. A recent prospective clinical trial sought to establish the clinical utility of longitudinal HPV ctDNA monitoring in HPV-positive disease as a method of surveillance after curative intent treatment [20]. Plasma HPV ctDNA levels of patients with nonmetastatic HPV-positive OPSCC treated with curative intent chemoradiotherapy (CRT) were serially analyzed throughout a median follow-up time of 23 months. Plasma samples were analyzed for HPV ctDNA from the following timepoints: pre-treatment, weekly during CRT, and at each post-treatment surveillance visit. HPV ctDNA was undetectable in 76% of patients at all surveillance timepoints and all of these patients remained disease free during the study period. Kinetic data for the patients with abnormal HPV ctDNA levels during post-treatment surveillance showed that all patients who experienced recurrence had at least two consecutive abnormal HPV ctDNA levels. In patients with abnormal HPV ctDNA levels who did not develop recurrent disease, clearance of HPV ctDNA was noted at the subsequent timepoint. Resolution of transient elevation in HPV ctDNA was seen in all patients without recurrent disease. When two subsequent abnormal HPV ctDNA tests was used as criterion for detecting high risk patients, negative predictive value was 100% (95% CI 96% - 100%) and positive predictive value was 94% (95% CI 70%−100%). Sensitivity and specificity were excellent at 100% and 99%, respectively. Study results suggest a high positive predictive value and negative predictive value for identifying recurrent disease in HPV positive OPSCC when HPV ctDNA is detected in two consecutive plasma samples [20].

Those with transient elevation of HPV ctDNA that resolved at a subsequent timepoint did not exhibit recurrent disease [20]. The authors suggest that these patients may have experienced subclinical recurrence, which was immunologically cleared, resulting in a transient spike in HPV ctDNA followed by resolution. Results of this study suggest that two consecutive measures of abnormal HPV ctDNA should serve as criterion for identifying patients at high risk for recurrence. Elevation in HPV ctDNA was identified approximately 4 months prior to biopsy-proven recurrence in one study subject (median lead time 3.9 months). Early identification of recurrence may allow for initiation of salvage therapy and improved outcomes for these individuals.

Early detection of treatment failure prior to radiographic evidence may improve outcomes for patients with HPV positive HNSCC. In a study of patients with HPV positive OPSCC, levels of HPV16 ctDNA measured prior to treatment and at 12 weeks following completion of treatment were compared to radiological assessments. Of the 66 patients included in the study cohort, 65% were found to have complete radiological response and 35% were determined to have incomplete radiological response. Of those with incomplete radiological response, HPV16 ctDNA was detected in 28% of patients, all of which exhibited local or regional treatment failure or distant metastasis. Only 4% of those with complete radiographic response had detectable levels of HPV16 ctDNA at the post-treatment checkpoint [23]. Assessing levels of HPV ctDNA shortly after treatment completion may aid in identification of residual disease not detected by radiographic imaging alone. Interestingly, Rutkowski et al. propose that detection of HPV16 ctDNA may not only indicate residual local disease but may suggest metastatic spread. Two of the patients in their cohort with incomplete radiographic response and elevated HPV ctDNA levels were found to have metastatic disease on PET-CT. Similarly, other studies have shown that detectable HPV16 ctDNA levels following treatment may precede diagnosis of metastatic disease by imaging [19, 24].

PET-CT is broadly utilized to assess treatment response to radiotherapy in HNSCC; however, ability to predict treatment failure via imaging alone is inadequate. In a recent study by Tanaka et al., the authors prospectively correlated HPV16 ctDNA levels and metabolic response, as indicated by PET-CT, with treatment failure [25]. They quantified HPV16 ctDNA in patients with HPV-positive HNSCC undergoing treatment with radiation or CRT. The study found that the negative predictive values were similar between HPV16 ctDNA and PET-CT. Notably, HPV16 ct6DNA positive predictive value was 100% whereas PET-CT positive predictive value was 50% after a median follow-up of 21 months. When examining treatment failure, risk was high in those with detectable post-treatment HPV16 ctDNA levels and incomplete metabolic response. All individuals who had detectable post-treatment HPV16 ctDNA experienced treatment failure [25]. Risk of treatment failure was deemed intermediate in individuals who experienced discordant results between post-treatment HPV16 ctDNA levels and metabolic response. Analysis of post-treatment HPV16 ctDNA may complement PET-CT in assessing treatment response and may inform clinical decisions in patients with HPV-positive HNSCC.

Using HPV ctDNA as a complementary biomarker may serve as an effective point of entry for introduction of this biomarker into clinical practice. Combining unique HPV ctDNA data for a patient with current standard of care imaging results may enhance the clinician’s ability to assess treatment response. With advances in both surgical salvage and novel immunotherapy, early detection of treatment failure is crucial to the care of patients with recurrent disease.

Assessment of Treatment Response

Several studies have illustrated the role of HPV ctDNA as a biomarker to closely monitor therapeutic response of patients undergoing CRT for HNSCC, correlating HPV ctDNA levels with treatment response in patients with localized or metastatic disease [19, 26].

A pilot study by Hilke et al. evaluated ctDNA in patients with locally advanced HNSCC receiving treatment with combined CRT. ctDNA levels were shown to be tightly correlated with tumor volume. There was also a correlation between HPV ctDNA dynamics and response to treatment [27]. In this study, tumor samples of patients with locally advanced HNSCC were sequenced to identify driver mutations and ctDNA dynamics were analyzed throughout treatment course. There was a negative correlation of ctDNA and treatment dose in the majority of patients. Analysis of HPV ctDNA kinetics during CRT demonstrated a time and dosage dependency. There was a decline of ctDNA levels with successful treatment via CRT, suggesting that monitoring dynamic changes of ctDNA may inform assessment of treatment response [27]. When observing ctDNA kinetics during treatment, it was noted that several patients had a spike in ctDNA levels at the first timepoint following treatment initiation [27]. Later timepoints showed a rapid decline in ctDNA levels. One possible explanation for such a spike is increase in tumor cell apoptosis after the initial initiation of treatment, similar to a phenomenon described in metastatic melanoma [28]. However, this transient elevation in ctDNA was not seen in all patients. These findings suggest that ctDNA may be utilized as a surrogate marker for disease burden and treatment response.

Retrospective studies have shown that HPV16 ctDNA levels become essentially undetectable after CRT in the majority of patients and conversely increase with disease recurrence [19, 22, 29–31]. In a multi-institutional prospective biomarker trial, Chera et al. examined patients with p16-positive OPSCC who received definitive CRT as treatment. Plasma was analyzed at baseline, each week during CRT, and at all follow-up visits. Using optimized multianalyte digital PCR assays, ctDNA of HPV types 16, 18, 31, 33, and 35 were quantified. Results showed that baseline plasma HPV ctDNA has high sensitivity and specificity for identifying HPV positive OPSCC. Pretreatment HPV16 ctDNA copy number correlated with disease burden, tumor HPV copy number, and HPV integration status [32]. A focus of this trial was analysis of the clearance profile of individuals in relation to disease status. In order to assess clearance kinetics, patients were examined every 2 to 4 months for years 1 and 2 following treatment and then every 6 months thereafter. The assay used in this study was designed to amplify and quantify a region of the HPV16 E7 gene. Other HPV subtypes were not cross detected. Results demonstrated that rapid clearance of HPV ctDNA during CRT therapy correlates with disease control in HPV-positive OPSCC. A favorable clearance profile was defined as patients with a high baseline copy number (>200 copies/mL) and clearance of >95% of HPV16 ctDNA by day 28 of CRT [32]. Those with favorable clearance profiles during CRT exhibited excellent disease control while patients with unfavorable profiles demonstrated significantly worse control [32]. Ability to predict likelihood of disease control may be advantageous when selecting patients for treatment deintensification. Individuals with a rapid clearance profile may be candidates for deintesified therapy. Furthermore, if clearance profiles can be routinely monitored during treatment, levels of HPV16 ctDNA may be able to predict local and regional disease control.

Patterns of Treatment Response in Recurrent or Metastatic Disease

Survival for patients with unresectable locoregional recurrent or metastatic (R/M) HNSCC is poor, with few patients surviving greater than one year [34]. The minority of patients respond to commonly used systemic therapies including platinum-based therapy, epidermal growth factor receptor (EGFR) inhibitors and immunotherapeutics [34–36]. Low response rates likely relate to tumor resistance patterns as well as patient functional status and comorbidities. We lack a clear understanding of which patients will respond to each therapy. To date, the only established predictive biomarker is programmed cell death ligand 1 (PD-L1) combined positive score (CPS), specifically using the 22C3 PD-L1 antibody. Patients with tumors expressing PD-L1 (CPS ≥ 1%) are more likely to have an improvement in survival with first line immunotherapy versus chemotherapy [36]. There is significant interest in interest in identifying additional biomarkers to predict treatment response.

HPV ctDNA analyses allow for non-invasive longitudinal assessments of tumor genomes which may assist in identifying predictive biomarkers. Hanna et al. assessed HPV plasma ctDNA levels in patients with HPV associated R/M HNSCC receiving systemic therapy. CtDNA levels were correlated with disease burden and treatment outcomes. Results demonstrated that tumor burden strongly correlated with HPV ctDNA viral load. Patients with lower overall tumor burden and lower HPV ctDNA viral load had improved survival. Additionally, results showed that a decrease in plasma HPV ctDNA viral load over time predicted a decline in tumor burden (defined as treatment response), whereas an increase in plasma HPV ctDNA viral load over time predicted progressive disease. These results suggest that viral ctDNA kinetics reflect dynamic changes in disease burden which may elucidate patterns of treatment response and resistance.

Saliva Analysis

While the majority of extant literature focuses on plasma levels of HPV ctDNA, Hanna et al. has explored the clinical utility of analyzing salivary HPV ctDNA in patients with HPV-positive HNSCC. Plasma HPV ctDNA levels have been shown to correlate with tumor stage and size [26, 37]. HPV ctDNA in saliva could serve as an additional, non-invasive surrogate for disease monitoring. In an observational study of patients with advanced HPV-positive OPSCC, Hanna et al. evaluated the predictive and prognostic potential of paired plasma and saliva HPV ctDNA [26]. In this study, total tumor burden in those with locoregional disease was correlated with HPV ctDNA levels in saliva [26]. HPV ctDNA levels in saliva were noted to increase and decrease in concordance with clinical and radiographic locoregional disease progression or response [26]. Higher plasma HPV ctDNA levels were associated with poorer patient outcomes, whereas higher saliva HPV ctDNA levels were not. Results of this observational study are limited and require prospective validation in order to garner clinical significance.

In another study, Tang et al. assayed HPV-16 in saliva of OPC patients at different stages and reported that high HPV-16 DNA viral loads in the saliva showed significant concordance with advanced disease stages [38]. HPV-16 DNA was detected in the saliva of 80% p16 positive OPC patients (n = 89) and only 3% of the p16 negative OPC patients (n = 32) indicating an 80% sensitivity and 91% specificity with a positive predictive value of 91% and negative predictive value of 62% [38] Remarkably, their HPV-16 detection assay using oral rinse samples led to the detection of OPC in an asymptomatic individual, who had enrolled in the study for surveillance as a high-risk, but cancer free, healthy subject [39]. For this individual, a marked increase in HPV-16 levels was observed in oral rinse samples peaking at 36 months, presumably indicative of viral sequences shed from a growing tumor. After surgical resection of the tumor via tonsillectomy, a steep drop in HPV-16 levels was observed [39]. This demonstrates the potential for saliva-based HPV assay in clinical application of early disease detection and treatment response monitoring.

The existing body of literature on salivary analysis is small and consists primarily of observational pilot studies. Potential benefits of a saliva-based HPV ctDNA assay include the non-invasive nature of the test and ability to collect serial samples. While saliva analysis is non-invasive, obtaining the quantity of saliva necessary for adequate analysis is a limitation of this approach. Although groups have attempted to standardize saliva collection protocols, they note variability in the volume and content of saliva collected for each specimen [26]. Additionally, studies have shown that the sensitivity of HPV ctDNA detection in saliva varies based on tumor location [40]. Saliva-based HPV ctDNA data may serve to compliment plasma HPV ctDNA data, but larger scale trials are necessary to elucidate the potential benefits of this approach.

Challenges

While all of this data regarding the detection of HPV ctDNA and association with treatment outcomes is exciting, the field faces significant challenges which must be tackled prior to moving forward into the clinic. Although the studies have shown varied outcomes based on HPV ctDNA detection at various points, we do not understand the normal kinetics of HPV ctDNA in large cohorts of homogenous patients undergoing standard of care therapies. For example, at which point during chemoradiation are the kinetics discriminatory for outcomes? Is it the presence of a peak during the first week, the absolute level at a given week, or clearance at a specified time point during therapy? Are there differential response kinetics in patients treated with chemotherapy versus immunotherapy? Although seemingly mundane questions, it is imperative we know this data before designing clinical trials or moreover changing clinical care. Without a full understanding of the normal kinetics, we may be missing actionable time points earlier during patients’ treatment course. This is even more challenging should we move to use HPV ctDNA as a test for early detection of HPV-positive HNSCC or surveillance of recurrence after therapy. Large cohorts will be necessary for each of these populations to establish the implication of detection of HPV ctDNA on the development of disease or recurrence. For example, a small number of patients post-chemoradiation transiently had HPV ctDNA detectable then cleared which was hypothesized to be due to immune clearance [32]. Without a comprehensive population analysis, we will not be able to characterize the implications of these events given the low rate of disease recurrence. Furthermore, the field needs to harmonize the technical characteristics of each assay, including procedures for specimen processing, control of batch effects and defining analytic sensitivity. There are numerous HPV ctDNA assays in development and given that each has separate molecular characteristics, one cannot assume a uniform sensitivity/specificity. Even small differences in sensitivity between assays will impact the ability to interpret and compare results. These aims will require a high degree of multi-institutional collaboration and meeting the challenge of enrolling patients receiving standard of care therapy. Until a HPV ctDNA assay is fully analytically validated and the normal kinetics are established against ground truth controls, it should not be employed to guide decisions in clinical trials.

Work from other tumors has shown that biomarker directed therapy does not improve patient outcomes across all settings [41]. Although a circulating biomarker offers a new tool for the management of HPV-positive HNSCC, the field must be cautious and examine how and if it improves survival and quality of life. For example, if employed for primary surveillance across an at-risk population, will a false positive rate (albeit minimal) lead to an increased rate of tonsil biopsies and radiographic imaging in patients with benign mucosa? Similarly, if used for early detection of recurrence, will early detection of surveillance improve survival or simply subject patients to more interventions and impaired quality of life? Perhaps more provocative, how will clinicians manage molecular recurrence in the absence of radiographic findings? These are difficult questions that will challenge the field to incorporate novel measures and endpoints.

Conclusion

HPV ctDNA offers a potentially promising biomarker for monitoring and surveillance of patients with HPV-positive HNSCC. Recent studies illustrate how analysis of ctDNA may aid in early detection of recurrence, assessment of treatment response, and profiling of resistance patterns for HPV-positive disease. Incorporating HPV ctDNA monitoring into clinical care of patients with HPV-positive disease may revolutionize disease treatment and monitoring. Numerous challenges surrounding clinical validation, understanding of the normal kinetics, and appropriately designed trials remain. Until clinical utility and efficacy are demonstrated in appropriately designed and executed clinical trials, HPV ctDNA testing should not be incorporated into routine management of patients with HPV-positive HNSCC.