Liquid Biopsies in Head and Neck Cancers: Recent Developments Across Biofluids, Analytes, and Molecular Features

ABSTRACT

Background

Head and neck cancers (HNCs) are too often diagnosed at advanced stages when outcomes are poor. Additionally, robust tools for the early detection of recurrence remain elusive. These gaps drive interest in so‐called liquid biopsy approaches for HNC detection, prognostication, and surveillance. Molecular heterogeneity presents challenges to liquid biopsy testing, but emerging approaches provide promising avenues toward clinical utility.

Methods

We review the latest developments in HNC liquid biopsies, provide perspectives on viral‐associated and nonviral‐associated cancers, and assess various biofluids, analytes, and molecular profiling approaches.

Results

Liquid biopsy assays targeting viral DNA from peripheral blood plasma have established clinical performance, and utility studies are ongoing, serving as a blueprint for other emerging assays.

Conclusion

The use of multiple biofluid sources and analytes may improve detection sensitivity and clinical applicability. Standardization and harmonization of analysis methods will be critical for enhancing biomarker discovery and enabling reliable clinical implementation.

Abbreviations

CAPP‐seq

Cancer Personalized Profiling by deep sequencing

CD

cluster of differentiation

cfDNA

cell‐free DNA

CTC

circulating tumor cell

ctDNA

circulating tumor DNA

dPCR

digital polymerase chain reaction

EBV

Epstein–Barr virus

EDTA

ethylenediaminetetraacetic acid

EV

extracellular vesicles

HNC

head and neck cancer

HNSCC

head and neck squamous cell carcinoma

HPV

human papillomavirus

MBD‐seq

methyl‐CpG binding proteins capture

MeDIP‐Seq

Methylated DNA Immunoprecipitation Sequencing

MRD

molecular residual disease

MSP

methylation‐specific PCR

NGS

next‐generation sequencing

NPC

nasopharyngeal carcinoma

OPSCC

oropharyngeal squamous cell carcinoma

OSCC

oral squamous cell carcinoma

1. Introduction

Head and neck cancer (HNC), including the most common histological subtype, head and neck squamous cell carcinoma (HNSCC), is a disease with profound functional implications for patients due to its anatomical location and side effects from treatment. Screening and surveillance are limited to clinical examinations and radiographic studies, which can miss signs of low‐volume cancers. Diagnosis is made through tissue biopsy, which is invasive and restricted to a limited number of sites due to patient tolerance. For these reasons, detection is often delayed, leading to a worsened prognosis.

Molecular analysis of blood and other bodily fluids (i.e., “liquid biopsy”) could lead to improvements in HNC diagnosis and optimized patient outcomes. Errors from tissue sampling (i.e., geographic misses) and tumor accessibility can be mitigated. Serial testing during and after treatment provides an opportunity for monitoring treatment response, detecting molecular residual disease (MRD), and identifying mechanisms of emerging treatment resistance. These applications of liquid biopsy could become relevant for making timely treatment decisions as well as supporting the development of new interventional strategies.

In this review, we discuss the pathogenesis of HNCs as it pertains to molecular features that can be leveraged for liquid biopsies, such as oncogenic viruses and specific genetic and epigenetic processes. We next describe various analytes (e.g., circulating tumor cells and cell‐free DNA) and biofluids (e.g., blood and saliva) that serve as sources for putative liquid biopsy biomarkers. We summarize evidence from studies on both viral‐associated and nonviral‐associated HNCs for clinical applications such as detection, monitoring, and prognostication. Finally, we discuss the current challenges and the future direction for unlocking the potential of liquid biopsies in HNC, including the potential of multi‐analyte and multi‐fluid assessments.

2. HNC Subtypes and Molecular Pathogenesis

Worldwide, HNC affects 900 000 people annually, with around 450 000 deaths every year [1]. Major anatomical subtypes of HNC include cancers of the oral cavity, lip, oropharynx, nasopharynx, larynx, and salivary glands (Figure 1). Geographic differences in HNC subtypes include a higher incidence of cancers of the nasopharynx in Southeast Asia and of the oral cavity in the Indian subcontinent [2]. Major causative agents for the development of HNCs include tobacco smoking, alcohol consumption, smokeless tobacco use, and betel nut chewing, along with infectious, environmental, and occupational exposures [3, 4, 5, 6, 7]. Genetic factors [8, 9, 10] and immunological health [11, 12] play important roles in HNC oncogenesis. Common oncogenic viruses include Epstein–Barr virus (EBV) in nasopharyngeal carcinoma (NPC) [13] and human papillomavirus (HPV) in oropharyngeal squamous cell carcinoma (OPSCC) [14]. Emerging liquid biopsy approaches have had to account for this heterogeneity of HNC pathogenesis.

FIGURE 1.

Head and neck cancer liquid biopsies and their composition. Graphical representation of common cancers seen in the head and neck with etiologic origins. Tumors at these sites can shed biomarkers into blood, saliva, and draining lymphatic fluid, with the components of each depicted above.

3. Liquid Biopsy Analytes and Biofluids

Sources of liquid biopsies can be categorized according to the type of analyte and the biofluid under study. This section will review each as it pertains to HNCs.

3.1. Liquid Biopsy Analytes

Major classes of liquid biopsy analytes include intact cells (i.e., circulating tumor cells [CTCs]) and various cell‐free tumor‐derived biomolecules such as cell‐free DNA (cfDNA), extracellular vesicles (EVs), cell‐free RNAs, and proteomic biomarkers. The focus of this section will be on CTCs, cfDNA, and EVs, whereas studies on cell‐free RNA and proteomic biomarkers have recently been reviewed elsewhere [15, 16, 17].

3.1.1. CTCs

The potential for cancer cells to be detectable in the circulation of patients with solid tumors has been recognized for over 150 years [18]. In many disease settings, the relative scarcity of CTCs—often representing less than one cell for every 10 million white blood cells in patients with known metastatic disease [19]—has presented technical challenges to their study. But recent advances in microfluidics, imaging, and high throughput molecular profiling have revealed that CTCs can reflect biological features of primary and metastatic tumors, and that their abundance can be associated with disease burden and risk of metastatic spread. CTCs can be passively shed from primary tumors and metastatic deposits, and they can also represent active invasion as part of the metastatic cascade [20]. As described in more detail below, the presence and abundance of CTCs in HNC patients are correlated with aggressive biology and poor prognosis [20, 21].

CTCs can be identified through selective isolation from the bloodstream or by screening cells for molecular features without selection. CTC isolation techniques rely on physical (e.g., size) and biological (e.g., distinct surface antigens) features that differ between cancer cells and leukocytes [22]. Isolation techniques for CTCs have been reviewed elsewhere [23]. In HNC, most studies on CTCs have employed selective isolation based on epithelial markers and examined associations with prognosis and recurrence [24, 25]. A common drawback to this approach is the inability to detect cancer cells that have lost epithelial markers through epithelial‐mesenchymal transition. For this reason, some studies are expanding the panel of antibodies for CTC selection [26]. To date, CTCs remain most readily detectable in advanced stages of HNC, with few if any cells detectable in early stages.

3.1.2. Circulating Tumor DNA (ctDNA)

DNA can be released by cells into biological fluids, including the peripheral circulation, in both pathological and physiological conditions [27]. This cfDNA—and ctDNA, the tumor‐derived portion of cfDNA—is a convenient source of potential biomarkers for diagnostic, prognostic, and predictive applications in oncology. cfDNA has a short half‐life of around 6 h and is highly fragmented [28, 29]. ctDNA often only represents a small fraction of the cfDNA that is shed into the circulation, and this fraction is dependent upon the tumor burden and stage, rate of cellular turnover, as well as response to therapy (Figure 2). The molecular makeup of ctDNA reflects the genetic and epigenetic composition of tumor cells, including somatic mutations and aberrant DNA methylation and chromatin structure [30].

FIGURE 2.

Tumor burden detectability utilizing liquid biopsies. Schematic view of ctDNA prior to diagnosis and after treatment. The fraction of plasma cfDNA that consists of ctDNA generally increases as disease progresses, where screening may allow for subclinical disease detection. Definitive treatment is generally associated with a decrease in ctDNA levels, and posttreatment surveillance may allow for the detection of molecular residual disease (MRD). Relapse of disease is generally associated with a rebound in ctDNA levels and may serve to guide therapeutic decision making.

With such a low abundance of ctDNA, sensitive detection methods are required. PCR‐based methods such as digital PCR (dPCR) can have high analytical sensitivity for specific genetic markers (e.g., mutations, methylated cytosines, or viral sequences) but are relatively limited in their ability to simultaneously interrogate multiple targets [31]. In contrast, next‐generation sequencing (NGS) technologies are capable of multiplexing across wide genomic regions [32] and are increasingly capable of interrogating a variety of molecular features within cfDNA.

Current guidelines from the European Society for Medical Oncology recommend that validated ctDNA testing can be used routinely in practice for the genotyping of advanced diseases for lung, breast, gastrointestinal cancer, and others [33]. For HNCs, however, mutation‐based precision medicine and targeted therapies are currently lacking. Nonetheless, as discussed in detail below, ctDNA testing continues to be investigated for other applications in HNCs, including early detection, prognostication, and monitoring.

3.1.3. EVs

EVs are lipid bilayered particles of various sizes that are released by cells into the extracellular environment and act as mediators of intercellular communication [34]. Exosomes represent a subset of EVs that range from 30 to 150 nm in diameter [35, 36, 37] and can transport various biomolecules (e.g., proteins, metabolites miRNA, mRNA, as well as genomic and mitochondrial DNA) out of and between cells [38, 39]. Exosomes have been detected in most biological fluids, including blood and saliva, and hold potential for being utilized as liquid biomarkers in HNSCC [40]. Though they exist in both healthy and diseased states, their presence in the tumor microenvironment has been implicated in tumor invasion, metastasis, and immunologic response to SCC [41, 42, 43]. Additionally, exosomes are able to transfer bioactive molecules between tumor cells, immune cells, and stromal cells, allowing for cancer cells to escape immune surveillance and induce immune tolerance [44].

3.2. Biofluid Sources for Liquid Biopsies

While many different biofluids are being studied as a source for liquid biopsies, blood and saliva are the most relevant and widely examined for HNC (Figure 1) [32, 45]. Recently, surgical drain lymph fluid has also been suggested to have potential utility for MRD detection and will be briefly reviewed.

3.2.1. Blood

Blood is in direct contact with all anatomical sites that give rise to HNC as well as draining nodal basins and distant organs at risk of harboring metastatic deposits. Whole blood (or buffy coat following centrifugation) is used for CTC isolation, whereas serum or plasma is used for cfDNA/ctDNA and EV/exosome analyses. Plasma is generally preferred over serum for cfDNA/ctDNA studies because there is less contamination from leukocytes [46, 47]. Plasma collection tubes also have cation chelators such as EDTA, which block nuclease activity and preserve cfDNA. In recent years, cell stabilizing agents have been added to blood collection tubes to reduce the contamination from leukocytes further and extend the acceptable time from blood draw to processing and storage [48, 49, 50].

3.2.2. Saliva

Saliva is a biological fluid derived from the salivary glands, gingival crevicular fluid, epithelial and inflammatory cell debris, bacteria, nasal and bronchial secretions, and blood [51, 52, 53]. Compared to blood, saliva could provide a more sensitive source for the detection of primary tumors due to the direct shedding of cells and cellular material. However, normal epithelial cells are also shed into saliva and could interfere with the specific detection of tumor cells. The source of cfDNA and EVs in saliva is more complex, as these smaller biomolecules can be secreted into saliva from blood components in addition to direct shedding [54, 55, 56, 57]. There may be less concern for leukocyte contamination in saliva compared to blood [58, 59, 60]. Moreover, whether distinct sizes of cfDNA and EVs in saliva may be relatively enriched for tumor‐derived content is an area of active study [61, 62, 63]. However, the enhanced sensitivity of more proximally collected biomarkers has already been suggested [64].

3.2.3. Lymphatic Fluid

Lymphatic fluid, or lymph, has recently been identified as a source for liquid biopsies in HNCs [65]. Draining lymph has been reported to contain tumor‐derived cfDNA. Surgical drains allow for the collection of lymph‐enriched fluid and are routinely placed after neck dissection in patients with HNCs, allowing for ease of collection and postsurgical MRD detection.

4. Viral‐Associated HNC

4.1. EBV‐Driven NPC

One of the best‐studied liquid biopsy biomarkers in HNC is plasma EBV DNA in patients with NPC. EBV is an oncogenic DNA virus that is found in a variety of cancer types, including NPC as well as certain gastric cancers and lymphomas. EBV‐associated NPC is endemic to Southeast Asia but is also found at lower rates around the world [13].

4.1.1. Detection

In a landmark study, EBV DNA was found to be detectable in the plasma of patients with NPC at the time of diagnosis, with higher levels associated with more advanced stages of disease [66]. Plasma samples were analyzed for the presence of EBV DNA through quantitative PCR (qPCR), targeting the EBNA‐1 and BamHI‐W regions. EBV DNA was detectable in the plasma of 97% of NPC patients prior to oncologic intervention, and copy numbers were 10 times higher in those patients with more advanced disease than those with early‐stage disease. This work suggested that quantitative analysis of plasma EBV DNA may be useful for detecting early‐stage NPC.

Subsequently, plasma EBV DNA detection was investigated as a screening modality for asymptomatic individuals [67]. Among 20 000 participants, persistently positive plasma EBV levels had sensitivity and specificity of 97.1% and 98.6%, respectively. Additionally, there was a notable shift toward earlier stage NPC diagnoses compared to historical controls. With a positive predictive value (PPV) of just 11%, the group later proposed sequencing‐based metrics to boost this metric. They found that integrating fragment size profiles of plasma EBV DNA molecules produced a PPV of 19.6% [68]. Moreover, plasma EBV DNA methylation profiling enabled differentiation of NPC, EBV‐associated lymphoma, and infectious mononucleosis, and integrating all of these metrics improved the PPV to 35.1% [69]. These works together show the great potential of plasma EBV DNA analysis as a screening test for endemic EBV‐driven NPC, with more work required, including comparisons with other promising approaches such as those that incorporate nasopharyngeal brushings [70].

Apart from plasma, urine as a liquid biopsy for EBV DNA detection in NPC has been investigated; however, low transrenal excretion only allows for detection in those patients exhibiting high levels within plasma [71]. While EBV DNA methylation studies of saliva in patients with NPC have been performed, sensitivity was not as high as that in plasma [72]. Therefore, plasma remains the dominant biofluid source for EBV DNA detection in patients with NPC.

4.1.2. Monitoring and Surveillance

In addition to its use in NPC screening, plasma EBV DNA has been assessed for its role in monitoring and surveillance. Since the first reports by Lo et al. over 25 years ago [73, 74], numerous studies have evaluated the association between plasma EBV DNA levels at baseline and throughout the treatment course with clinical events. A meta‐analysis of 14 studies involving more than 7836 cases found that patients with plasma EBV DNA levels that were high prior to initiation of treatment, patients with detectable levels mid‐ and posttreatment, and patients with slow clearance after treatment all experienced poor clinical outcomes with overall survival hazard ratios of 2.81, 3.29, 4.26, and 3.58, respectively [75]. In general, the prognostic value increased during and after treatment as compared to pretreatment. Moreover, the associations with distant metastasis were stronger than local or regional failure [75]. These results confirmed that plasma EBV DNA levels, especially posttreatment levels, have clinical validity as a prognostic biomarker and could help guide clinical decision making.

Based on these reproducible findings, recent investigations have focused on utilizing post‐chemoradiotherapy plasma EBV levels to guide adjuvant systemic treatment decisions. As the value of adjuvant chemotherapy remains controversial, a prospective randomized controlled trial sought to evaluate the benefit of adjuvant cisplatin and gemcitabine vs. observation in patients with residual detectable plasma EBV DNA 6–8 weeks after concurrent chemoradiotherapy [76]. With a median of 6.6 years follow‐up, there was no significant difference in 5‐year relapse‐free survival between the arms. This unexpected result could indicate the difficulty in sterilizing residual treatment‐refractory cancer cells after chemoradiotherapy. As cisplatin was included in both the concurrent and adjuvant phases of the trial, cross‐resistance may have developed that hindered the efficacy of the adjuvant regimen. Other studies such as NRG‐HN001 are evaluating regimens with less potential for cross‐resistance in this setting.

In contrast to the adjuvant setting, there has been greater progress in demonstrating potential clinical utility for plasma EBV DNA in the neoadjuvant setting. In a previously completed multi‐center randomized trial that established neoadjuvant cisplatin and gemcitabine as a standard treatment for locally advanced NPC [77], ad hoc analysis suggested that the benefit of neoadjuvant treatment was restricted to patients with pretreatment plasma EBV DNA ≥ 4000 copies/mL (5‐year overall survival 85.9% vs. 70.8%). Strikingly, there was no apparent benefit among patients with pretreatment EBV DNA < 4000 copies/mL (5‐year overall survival 90.6% vs. 91.4%). Future trials will need to replicate these findings in a prospective manner to confirm the specific thresholds and strength of the biomarker interaction.

Compared to the standard qPCR assay used in the vast majority of studies, more sensitive detection of plasma EBV DNA could provide greater prognostic value in the posttreatment setting. Emerging sequencing‐based methods could drive down technical detection limits by over 100‐fold. Indeed, sequencing‐based analysis of post‐chemoradiotherapy plasma EBV DNA in NPC patients from the aforementioned randomized adjuvant study [76] demonstrates higher sensitivity for both local (85.5%) and distant recurrences (97.1%) [78]. The added prognostic value of targeted sequencing over qPCR could lead to more informed clinical treatment decisions, but practical trade‐offs such as cost and turnaround time, which are both generally higher with sequencing‐based assays, also need to be taken into consideration when implementing these tests into clinical trials and standard practice.

4.1.3. EBV DNA Test Standardization

As EBV DNA testing becomes more broadly utilized, there has been a need for standardization of assay methods [79]. Le et al. 2013 assessed tests across different laboratories to harmonize international protocols, reagents, as well as cut‐off criteria for ctDNA detection. Further, in 2015, the National Cancer Institute convened a workshop on the harmonization of EBV testing for NPC to create standards for its appropriate clinical validation [80]. Although there continues to be a need to improve standardization efforts, the current progress with EBV DNA may serve as a blueprint for other assay types.

4.2. HPV‐Driven Oropharyngeal Carcinoma

The discovery that HPV infection can be a driver of oropharyngeal carcinogenesis, among other cancers, has provided the opportunity for major advancement in both detection and treatment. Compared to HPV‐negative OPSCC, HPV‐associated OPSCC is characterized by significantly better treatment responses and clinical outcomes [81]. While there are well over 150 different strains of HPV, 90% of HPV‐associated OPSCCs are attributed to the high‐risk HPV genotype 16 [14]. HPV16 and other high‐risk subtypes drive carcinogenesis in part by integrating their double‐stranded DNA into the host genome [14]. This integration event can lead to dysregulated expression of viral E6 and E7 oncogenes, which promote cellular proliferation, genetic instability, and remove cell cycle checkpoints, thereby promoting malignant transformation [82, 83].

4.2.1. Detection

HPV can be detected in tumor tissues by PCR, in situ hybridization, and immunohistochemistry targeting the surrogate marker p16 [84]. Unfortunately, most cases of OPSCC are initially detected through the presentation of enlarged lymph nodes in the neck, indicating lymphatic spread has already occurred, and diagnosed through tissue biopsy of the neck node and/or primary lesion. Currently, liquid biopsy approaches are being investigated for less invasive detection and diagnosis.

Plasma HPV DNA is being extensively studied for a range of liquid biopsy applications in OPSCC. An early study using qPCR detected HPV16 DNA in peripheral blood serum from only 9% of patients with HPV‐associated OPSCC [85]. Subsequent studies utilizing plasma instead of serum have shown higher sensitivities in a manner dependent on disease burden as determined by clinical stage [86, 87, 88, 89, 90, 91, 92, 93]. Incorporating HPV types other than HPV16, including HPV 18, 31, 33, and 35, can also boost sensitivity. One multi‐institutional prospective study found sensitivity of 89% and specificity of 97% among patients treated with chemoradiotherapy using dPCR [86]. Other notable cross‐sectional studies have found sensitivities of 89%–96% with similarly high specificity [87, 88]. Importantly, patients with primary‐only disease display lower detection rates compared to those who have nodal metastases, and N category is a strong driver of plasma HPV DNA levels.

The lower sensitivity for early‐stage disease has motivated the development of newer techniques for plasma HPV DNA detection. NGS provides a means for simultaneous detection of HPV DNA fragments derived from the entire ~8 kb genome, potentially boosting sensitivity by up to ~50‐fold compared to PCR‐based techniques. Indeed, studies using cell lines and patient plasma samples have confirmed higher analytical and clinical sensitivity when targeting the full HPV genome using hybrid capture NGS [94, 95]. In mostly early‐stage cases of HPV+ OPSCC, HPV genome NGS produced sensitivity of 99% and specificity of 98%, which compared favorably to dPCR [95]. Recognizing the additional cost and complexity of NGS assays compared to PCR, future studies are needed to define the clinical scenarios that require this additional boost in sensitivity to provide meaningful benefits in clinical utility.

Despite the reported association with clinical stage, plasma HPV DNA levels vary considerably among patients even after accounting for the known burden of disease. This variability may be explained in part by different rates of shedding and clearance from the circulation [96, 97] as well as the number of viral genome copies within malignant cells [98]. Moreover, viral DNA integration into the host genome is negatively associated with plasma HPV DNA levels and has also been linked with worse outcomes in OPSCC [99, 100]. Thus, the relationship between pretreatment plasma HPV DNA levels, disease burden, and prognosis is complex and multifactorial.

Beyond plasma cfDNA, EVs have also been assessed for their capacity to serve as a diagnostic biomarker of HPV‐associated OPSCC. In one study, DNA from EVs and total plasma cfDNA were both assessed using dPCR [101]. Sensitivity for the detection of locally advanced HPV‐associated OPSCC favored total plasma cfDNA (91%) over EVs (42%), indicating that EVs may not be as effective for assessing HPV DNA. EVs also contain RNA and other molecules that could be cancer‐specific, so additional work is needed to systematically compare these different analytes.

Saliva has also been evaluated for its utility in the detection of HPV‐associated OPSCC. A number of groups have employed PCR to assess for salivary HPV DNA in patients with HPV‐positive OPSCC. A cross‐sectional investigation assessed copy numbers of E6 and E7 together in salivary rinses with matched tumor samples and found a high specificity of 98.3% but only 30.4% sensitivity [102]. In a separate study with slightly different methods, sensitivity was higher at 81.4% with 96.4% specificity, indicating that the methodology of saliva collection is an important determinant of test performance [103].

The utility of salivary HPV16 DNA as a screening modality was recently evaluated in a prospective study of otherwise healthy individuals [104]. Participants with persistently positive results after 30 months were assessed by an otolaryngologist. Of the 650 participants recruited, 12 had persistent HPV16 DNA, and 1 of these was found to have a small primary lesion that was positive on pathology for HPV‐associated OPSCC. Longer follow‐up would be required to determine how many of these individuals with persistently detectable salivary HPV16 DNA eventually go on to develop invasive cancer.

Taken together, these studies indicate the promise of HPV DNA detection for liquid biopsy applications. Future studies are needed to define the clinical utility as a diagnostic or screening biomarker.

4.2.2. Monitoring and Surveillance

As the optimal strategy for posttreatment surveillance remains controversial, plasma HPV DNA detection has been proposed due to its noninvasive and quantitative properties. One of the first studies to examine dPCR for this purpose enrolled 22 patients with advanced HPV‐associated OPSCC [105]. Plasma HPV DNA levels correlated positively with the total tumor burden prior to and during treatment, where all patients who experienced clinical evidence of treatment response exhibited drops in plasma HPV copy numbers. The decrease in plasma HPV cfDNA was also shown to precede clinical response by 15 days. Conversely, patients with disease progression all showed either persistent or increasing levels of HPV DNA.

In the setting of nonmetastatic HPV‐associated OPSCC, plasma HPV DNA could be a useful tool for surveillance following chemoradiotherapy. An early demonstration of this strategy was shown in a cohort of 115 patients, 28 of whom developed a positive plasma HPV DNA test [106]. The median lead time between a positive plasma test and biopsy‐proven recurrence was 3.9 months, and the PPV of consecutive positive results was 94%. Subsequently, a multi‐center study of 1076 patients with HPV‐associated OPSCC demonstrated that only 80 (7.43%) had plasma HPV DNA‐positivity during surveillance after definitive therapy, with the majority (74%) of these positive results occurring in patients with no clinical evidence of disease [107]. Estimates of patient‐level positive and negative predictive values in this surveillance setting were 95% and 98%, respectively, albeit with limited follow‐up duration [107, 108].

With the promising results seen with plasma HPV DNA in the surveillance setting, there has been interest in using on‐treatment kinetics to adapt treatment intensity during chemoradiotherapy. Cao et al. observed a rapid decline in plasma HPV DNA levels in 14 patients during the course of chemoradiotherapy [93]. Chera et al. examined plasma HPV DNA kinetics weekly among 103 patients treated with chemoradiotherapy and found that rapid clearance 4 weeks into treatment correlated with favorable prognosis [86]. These intriguing results remain to be replicated, and the clinical utility for guiding treatment intensity will need to be evaluated in prospective studies.

Beyond plasma, HPV DNA in saliva and oral rinses has also been evaluated for monitoring of disease recurrence. In a two‐institution prospective study, qPCR was performed for the detection of high‐risk HPV subtypes in oral rinses after treatment for HPV‐associated HNSCC. Two‐year overall survival was only 65% in those patients with persistent high‐risk HPV DNA in oral rinses vs. 95% in those patients without [109]. Another study focusing on only HPV‐associated OPSCC patients found that persistent HPV16 DNA in oral rinses was strongly associated with recurrence, with specificity of 100% but modest sensitivity of 26% [110].

In an effort to boost sensitivity for posttreatment surveillance, a number of studies have evaluated dual plasma‐saliva liquid biopsies [56, 111, 112]. One study found that pretreatment sensitivity of HPV qPCR rose to 76% when analyzing both biofluids from 52.8% and 67.3% when analyzing just saliva or plasma, respectively [56]. Moreover, overall survival was reduced in patients who exhibited posttreatment HPV DNA in saliva, whereas survival was not affected if HPV DNA was detected in plasma alone. In terms of recurrence, combined plasma and saliva posttreatment samples were 90.7% specific and 69.5% sensitive in predicting recurrent disease in 3 years. These results were replicated in a subsequent prospective clinical trial with 87% specificity and 65% sensitivity for the combined HPV DNA test [111]. The mean lead time for a positive test when compared to clinically detected recurrence was 122 days, indicating that clinical implementation of dual liquid biopsies for disease detection holds promise.

To further address the limited sensitivity of HPV DNA detection in the posttreatment setting, another liquid biopsy source, lymph, has been introduced [65]. Plasma and paired lymphatic drainage fluid were assessed postsurgical resection, and cell‐free HPV DNA was found to be enriched in lymph compared to plasma and correlated with advanced nodal stage. While further work is needed to validate these findings, this proof‐of‐concept study demonstrates that lymph analysis could enable more sensitive HPV DNA detection for MRD applications. In particular, lymph may provide a window into regional recurrence risk while saliva and plasma reflect local and distant recurrence risk, respectively.

5. Nonviral‐Associated HNC

Nonviral HNCs present different challenges with respect to biomarker elucidation. While viral DNA offers convenient molecular targets for detection that are often present in multiple copies per cell, the heterogeneous nature of molecular aberrations driving nonviral HNCs necessitates more complex assays for liquid biopsy applications.

5.1. Mutational Profiling as a Means of Cancer Detection and Monitoring

Carcinogenesis is characterized by the accumulation of mutations that promote oncogene expression and/or inhibit the function of tumor suppressors. Mutations may be in the form of single‐nucleotide variants, insertions, deletions, and large chromosomal structural variants, and may be both inherited and acquired. Landmark advancements in genomic sequencing technologies have revealed hundreds of oncologic drivers [113, 114, 115]. Genes important for cell survival and proliferation (TP53, HRAS, EGFR, PTEN, and PIK3CA), cell‐cycle control (CDKN2A and CCND1), cellular differentiation (NOTCH1), reactive oxygen species scavenging (NFE2L2 and KEAP1), and adhesions and invasion signaling (FAT1) have all been implicated in nonviral‐associated HNSCC. Table 1 summarizes liquid biopsy studies investigating mutational targets in plasma, saliva, and oral rinses.

TABLE 1.

Mutational DNA liquid biopsy targets for HNSCCs.

Study	Country	Patients	Tstage (early, late)	HPV status	Detection method	Tumor‐informed	Genes	Personalized panel	Sample type	Detection	Clinical context
Wang et al., 2015 [116]	USA	93 HNSCC	20, 73	40% HPV+	Multiplex PCR	Yes	TP53, PIK3CA, CDKN2A, FBXW7, HRAS, NRAS	No	Plasma	87%	Diagnosis, recurrence
								No	Saliva	76%
Perdomo et al., 2017 [117]	France	36 HNSCC	14, 22	HPV−	Sequencing	Yes	TP53, NOTCH1, CDKN2A, CASP8, PTEN	No	Plasma	42%	Detection, diagnosis
		37 HNSCC	0, 37				TP53	No	Oral rinse	70%
Schwaederle et al., 2017 [118]	USA	25 HNSCC	—	—	NGS	No	70‐gene panel	No	Plasma	88%	Detection, diagnosis
Egyud et al., 2019 [119]	USA	7 HNSCC	1, 6	—	NGS	Yes	TP53, ARID1B, ATM, CDK8, FANCA, RASA1, SMARCA4, XRCC2, BCL10	No	Plasma	43%	Recurrence
Mes et al., 2020 [120]	Netherlands	40 HNSCC	6, 34	25%	NGS	Yes	TP53, NOTCH1, HRAS, CDKN2A, PTEN, PIK3CA, AJUBA, CASP8, FAT1, FBXW7, KMT2D, NSD1	No	Plasma	67%	Detection, diagnosis
Bai et al., 2020 [121]	China	7 HNSCC	0, 7	—	NGS	Yes	TP53, PIK3CA	No	Plasma	71%	Monitoring
Galot et al., 2020 [122]	Belgium	39 HNSCC	0, 39	12.8%	NGS	Yes	604‐gene panel	No	Plasma	51%	Recurrence, metastasis
Hilke et al., 2020 [123]	Germany	20 HNSCC	0, 20	15%	NGS	Yes	327‐gene panel	No	Plasma	85%	Monitoring
Khandelwal et al., 2020 [124]	USA	22 OPSCC	0, 22	50%	Sequencing	Yes	56‐gene panel	No	Plasma	—	Prognosis, progression
Porter et al. 2020 [125]	USA	48 HNSCC	8, 29	25%	NGS	Yes	73‐gene panel	No	Plasma	83%	Recurrence, metastasis
Wilson et al., 2021 [126]	USA	75 HNSCC	28, 47	26.7%	NGS	Yes	73‐gene panel	No	Plasma	65.3%	Prognosis
Wu et al., 2021 [127]	China	27 HNSCC	9, 18	HPV—	NGS	Yes	1021‐gene panel	No	Plasma	70.4%	Recurrence
								No	Saliva	63%
Cui et al., 2021 [128]	Korea	11 OSCC	4, 7	—	NGS	Yes	493‐gene panel	No	Plasma	40%	Recurrence
								No	Saliva	90.9%
Lee et al., 2021 [129]	Korea	7 HNSCC	3, 4	—	Multiplex PCR	Yes	275‐gene panel	No	Saliva	85.7%	Diagnosis, recurrence
D'Cruz et al., 2021 [130]	India	15 OSCC	—	—	PCR	Yes	TP53	No	Salivary rinse	66.7%	Diagnosis
Shanmugam et al., 2021 [131]	USA	121 OSCC	53, 68	—	NGS	Yes	TP53, PIK3CA, FAT1, CDKN2A, NOTCH1, CASP8, HRAS	No	Salivary rinse	95.9%	Diagnosis, recurrence
Burgener et al., 2021 [132]	Canada	30 HNSCC	11, 19	HPV—	CAPP‐seq	No	389‐gene panel	No	Plasma	66.7%	Recurrence, prognosis
Rapado‐González et al., 2023 [133]	Spain	3 HNSCC	0, 3	HPV—	NGS	Yes	170‐gene panel	No	Plasma	100%	Detection
					dPCR		EP300, NOTCH1, TP53		Saliva	33.3%
Flach et al., 2022 [134]	Germany	17 HNSCC	0, 17	HPV—	WES and Multiplex PCR	Yes	48‐gene panel	Yes	Plasma	100%	Detection, recurrence
Honoré et al., 2023 [135]	Belgium	53 HNSCC	12, 41	19%	NGS	No	26‐gene panel + E6, E7	No	Plasma	77.4%	Detection, recurrence, prognosis
Ahmed et al., 2024 [136]	England	11 OSCC	1, 10	—	PanelSeq	Yes	TP53, CDKN2A, KDM6B, NSD1, DNAH7, PIK3CA	No	Saliva	82%	Detection
Sanz‐Garcia, et al., 2024 [137]	Canada	32 HNSCC	32	53%	NGS	Yes	RaDaR CAPP‐seq HPV‐seq	Yes	Plasma	RaDaR 88% CAPP‐seq 51.7%	Detection, recurrence
Hanna et al., 2024 [138]	USA	116 HNSCC	34, 82	HPV—	WES	Yes	16‐gene panel	Yes	Plasma	91%	Recurrence, prognosis
Sim et al., 2025 [139]	USA	24 HNSCC	4, 20	4% (42% n/a)	WGS	Yes	MAESTRO	No	Plasma	92.9%	Recurrence, prognosis

Abbreviations: CAPP‐seq = Cancer Personalized Profiling by deep sequencing; MAESTRO = minor allele‐enriched sequencing through recognition oligonucleotides; NGS = next‐generation sequencing; WES = whole‐exome sequencing.

A notable study by Wang et al. assessed multi‐source liquid biopsies in the detection of mutations in HNSCC patients [116]. This observational study utilized both cfDNA from plasma and total salivary DNA for the detection of HPV DNA and somatic mutations using multiplex PCR targeting potential oncogenes. Plasma alone had a sensitivity of 87%, while saliva alone had a sensitivity of 76%; when combined, sensitivity improved to 96%. Interestingly, salivary DNA provided 100% sensitivity in oral squamous cell carcinoma (OSCC) patients, highlighting the value of interrogating both sources of liquid biopsy material.

These findings have been corroborated by another small study of 11 OSCC patients in which both plasma and saliva were interrogated using an NGS‐based technique [128]. Sensitivity was 27% in plasma and 91% in saliva when analyzed in isolation, but 100% when combined. Recurrent disease was detected by the combined liquid biopsy with 100% sensitivity. It is noted that in all cases of recurrent disease, detection by liquid biopsy was made prior to the emergence of clinical and radiographic findings, with up to 6 months lead time. While larger studies are needed, these findings indicate that mutational profiling within multiple biofluids holds promise for detection and monitoring of nonviral HNC.

Apart from multiple liquid biopsy sources, tumor tissue analysis has been investigated as a means for designing personalized, bespoke ctDNA assays. Flach et al. interrogated plasma cfDNA of patients with HPV‐negative HNSCC following tumor whole‐exome sequencing for personalized variant selection; multiplex amplicon NGS of tumor‐specific variants achieved high sensitivity [134]. In all patients that recurred, ctDNA was detected prior to clinically observed progression, with lead times of 108–253 days. Similarly, Sanz‐Garcia et al. found higher sensitivity with a tumor‐informed bespoke ctDNA assay as compared to a tissue‐agnostic panel‐based NGS such as CAPP‐seq [137], and Honore et al. observed lower sensitivity using a tumor‐agnostic 26‐gene NGS panel.

Taken together, this work highlights that tumor‐informed mutation‐based ctDNA assays—while more complex and costly than tissue‐agnostic assays—more accurately detect disease recurrence in nonviral HNSCC. In the absence of tissue, analysis of multiple biofluid sources could provide a means for driving greater sensitivity.

5.2. Methylation Profiling as a Means of Cancer Detection and Monitoring

In addition to mutations, DNA methylation has emerged as a promising class of liquid biopsy biomarkers for HNC. Cancer cells harbor hypermethylation of CpG‐dense regions (i.e., CpG islands) and hypomethylation of CpG‐sparse regions across the genome. Of these, hypermethylated CpG islands located in the vicinity of gene promoters have been the most commonly investigated regions for ctDNA detection due to their potential for cancer specificity with a high signal‐to‐noise ratio in cfDNA [140, 141]. Many cancers show more instances of conserved CpG island hypermethylation than mutated sequences, which could drive greater sensitivity for ctDNA detection in HNC [142]. Table 2 summarizes liquid biopsy studies investigating methylation changes in DNA in plasma, saliva, and oral rinses. Until recently, methylation‐specific PCR (MSP) has been the most commonly utilized technique. In a large observational study utilizing multi‐source liquid biopsies, Carvalho et al. paired serum and oral rinse samples from patients with HNSCC, and healthy participants were analyzed using quantitative MSP. Detection of MINT31, MGMT, CCNA1, and CDKN2A in serum and oral rinse resulted in 84.5% and 85.0% sensitivity, respectively, and 50% and 30% specificity [151]. This group made note that differential patterns of methylation in normal subjects between the serum and salivary liquid biopsies existed, which influenced the utility of single markers.

TABLE 2.

Methylated DNA liquid biopsy targets for HNSCCs.

Study	Country	Patients	T category (T1–2, T3–4, unknown)	Detection method	Tumor‐informed	Methylated targets	Personalized panel	Sample detection	Clinical context
Sanchez‐Cespedes et al., 2000 [143]	USA	23 HNSCC	4, 19	MSP	Yes	P16, MGMT, GSTp1DAPK	No	Tumor: 42% Serum: p16 (Sens 31%), MGMT (Sens 48%), DAPK (Sens 18%)	Detection
Rosas et al., 2001 [144]	USA	30 HNSCC	7, 23	MSP	Yes	p16, MGMT, DAPK	No	Tumor: 56% Oral rinse: Sens 36.7%	Detection
Wong et al., 2003 [145]	China	20 HNSCC	—	MSP	Yes	p15, p16	No	Plasma: p15 (Sens 60%, Spec 50%), p16 (Sens 65%, Spec 80%)	Detection
Nakahara et al., 2005 [146]	Japan	17 OSCC	9, 8	MSP	Yes	p16	No	Tumor: 64% Serum: Sens 54.5%	Detection
Righini et al., 2007 [147]	France	90 HNSCC	29, 61	MSP	Yes	p16, MGMT, DAPK, RASSF1A, TIMP3, ECAD	No	Tumor: Sens 77% Oral Rinse (n = 60): Sens 82%	Detection, recurrence
Franzmann et al., 2007 [148]	USA	102 HNSCC	40, 62	MSP	No	CD44	No	Oral rinse: Sens 62%–70%	Detection, recurrence
Viet et al., 2007 [149]	USA	14 OSCC/dysplasia	—	Methylight	Yes	p16, MGMT, p15, APC, ECAD	No	Saliva: Sens 7%–35%	Detection
Viet et al., 2008 [150]	USA	13 OSCC	7, 6	GoldenGate Methylation Array	Yes	GARB3, IL11, INSR, NOTCH3, NTRK3, PXN	No	Saliva: Sens 77% Spec 87%	Detection, monitoring
Carvalho et al., 2008 [151]	USA	176 HNSCC	—	qMSP	Yes	MINT31, MGMT, CCNA1, p16	No	Serum: Sens 84.5%, Spec 50%	Detection
							No	Oral rinse: Sens 85%, Spec 30%
Pattani et al., 2010 [152]	USA	161 OSCC/dysplasia	—	qMSP	No	EDNRB	No	Oral rinse: Sens 71%, Spec 58%	Detection
Demokan et al., 2010 [153]	USA	71 HNSCC	23, 48	qMSP	Yes	EDNRB, KIF1A	No	Oral rinse: Sens 77%, Spec 93%	Detection, prognosis
Guerrero‐Preston et al., 2011 [154]	USA and Spain	16 OSCC	—	Human Methylation27 BeadChip	Yes	HOXA9, NID2	No	Oral rinse: Sens 94%, Spec 97%	Detection
Nagata et al., 2012 [155]	Japan	34 OSCC	24, 10	MSP	No	ECAD, TMEFF2, RARb, MGMT	No	Oral rinse: Sens 100%, Spec 87.5%	Detection
Liu et al., 2012 [156]	China	32 OSCC	18, 14	qMSP	Yes	DAPK	No	Blood: Sen 52.2% Spec 86.6%	Detection
							No	Oral rinse: Sens 3.4%
Kusumoto et al., 2012 [157]	Japan	10 OSCC	—	MSP	No	CDKN2A	No	Oral rinse: Sens 40%, Spec 100%	Detection
Ovchinnikov et al., 2012 [158]	Australia	143 HNSCC	—	Nested MSP	No	RASSF1A, DAPK1, p16	No	Saliva: Sens 80%, Spec 87%	Detection
Rettori et al., 2013 [159]	Brazil	146 HNSCC	58, 88	qMSP	Yes	CCNA1, TIMP3	No	Oral rinse: Sens 14%, Spec 93%	Detection
Puttipanyalears et al., 2013 [160]	Thailand	43 OSCC	14, 29	Combined‐bisulfite restriction analysis (hypomethylation)	No	Alu	No	Oral rinse: Sens 87%, Spec 57%	Detection
Schussel et al., 2013 [161]	USA	191 HNSCC	—	qMSP	No	EDNRB, DCC	No	Oral rinse: Sens 75%, Spec 48%	Detection
Ovchinnikov et al., 2014 [162]	Australia	46 HNSCC	—	MSP	No	MED15, PCQAP5′	No	Saliva: Sens 70%, Spec 63%	Detection
Gaykalova et al., 2015 [163]	USA	59 HNSCC	15, 41	HumanMethylation27 BeadChips	Yes	ZNF160	No	Oral rinse: Sens 17%, Spec 100%	Detection
Langevin et al., 2015 [164]	USA	OSCC/OPSCC	36, 65	HumanMethylation450 BeadArray	No	22 CpG island classifier	No	Oral rinse: Sens 64.7%, Spec 96.0%	Detection
Lim et al., 2016 [165]	Australia	88 HNSCC	32, 36, 20	MSP	No	RASSF1a, p16 INK4a , TIMP3, PCQAP5′, PCQAP3′	No	Saliva: Sens 71%, Spec 80%	Detection
Mydlarz et al., 2016 [166]	USA	100 HNSCC	20, 80	qMSP	No	EDNRB	No	Serum: Sens 10%, Spec 100%	Detection
Ferlazzo et al., 2017 [167]	Italy	58 OSCC	—	MSP	No	p16, MGMT	No	Saliva: Sens 21%	Detection
De Vos et al., 2017 [168]	Germany	278 HNSCC	140, 131, 19	dMSP	No	SEPT9, SHOX2	No	Plasma: Sens 64%–65%, Spec 90%–91%	Detection
Schröck et al., 2017 [169]	Germany	141 HNSCC	—	dMSP	No	SEPT9, SHOX2	No	Plasma: Sens 52%, Spec 95%	Detection, prognosis, recurrence
Cheng et al., 2018 [170]	Taiwan	94 OSCC	—	qMSP	No	ZNF582	No	Oral rinse: Sens 65%, Spec 75%	Detection
Puttipanyalears et al., 2018 [171]	Thailand	42 OSCC/24 OPSCC	17, 25 (OSCC)/7, 17 (OPSCC)	qMSP	No	TRH	No	Oral rinse: OSCC Sens 88%, Spec 93% OPSCC Sens 83%, Spec 93%	Detection
Liyanage et al., 2019 [172]	Sri Lanka	54 OSCC/34 OPSCC	8, 23, 23 (OSCC)/2, 13, 19 (OPSCC)	MSP	No	p16, RASSF1a, TIMP3, PCQAP/MED15	No	Saliva: OSCC Sens 92%, Spec 92% OPSCC Sens 100%, Spec 92%	Detection
De Jesus et al., 2020 [173]	Brazil	15 OPSCC	—	dPCR	Yes	CCNA1, DAPK, CDH8, TIMP3	No	Tumor: 71% Plasma: Sens 73.3%, Spec 100%	Detection
Srisuttee et al., 2020 [174]	Thailand	43 OSCC	—	qMSP	No	NID2	No	Oral rinse: Sens 79%, Spec 100%	Detection
Shen et al., 2020 [175]	USA	21 OPSCC	2, 19	qMSP	No	PAX5, EDNRB	No	Oral rinse: Sens 95%, Spec 90%	Detection, recurrence
Gonzalez‐Perez et al., 2020 [176]	Columbia	43 OSCC	24, 19	MSP	No	p16, RASSF1a	No	Saliva: Sens 54%, Spec 88%	Detection, prognosis
Burgener et al., 2021 [132]	Canada	30 HNSCC	11, 19	MeDIPSeq	No	DMR: 941 hypermethylated regions, 56 hypomethylated regions	No	Plasma	Recurrence, prognosis
Patel et al., 2023 [177]	USA	8 OSCC	3, 5	MBD‐seq	No	Detection: PENK, NXPH1, ZIK1, TBXT, CDO1 Prognosis: ZIK1, IRF4, PCDH17, PENK	No	Plasma	Detection, prognosis
Rapado‐González et al., 2024 [178]	Spain	4 OSCC	4, 0	MethylationEPIC BeadChip	No	A2BP1, ANK1, ALDH1A2, GFRA1, TTYH1, PDE4B	No	Saliva: Sens 75%–100%, Spec 85%–100%	Detection
Liu et al., 2025 [179]	Canada	325 HNSCC	81, 68	MeDIPSeq	No	DMR panel	No	Plasma: Sens 91%, Spec 88%	Recurrence

Abbreviations: dPCR = digital polymerase chain reaction; DMR = differentially methylated regions; MBD‐seq = methyl‐CpG binding proteins capture; MeDIP‐Seq = Methylated DNA Immunoprecipitation sequencing; MSP = methylation‐specific PCR; sens = sensitivity; spec = specificity.

As MSP is limited in the number of targets that can be simultaneously interrogated, recent studies have applied NGS for broader profiling of HNC‐specific DNA methylation biomarkers. Burgener et al. compared cell‐free Methylated DNA ImmunoPrecipitation and high throughput sequencing (cfMeDIP‐seq) with mutation‐based NGS from plasma cfDNA [132]. In a cohort of HPV‐negative HNSCC, pretreatment ctDNA levels determined by cfMeDIP‐seq and mutation‐based NGS were highly correlated, and posttreatment persistence of methylated ctDNA was strongly correlated with recurrence. Interestingly, DNA methylation biomarkers associated with clinical outcomes when measured in tumor tissue were also strongly prognostic in plasma cfDNA in a manner that was independent of ctDNA levels. These results suggest that the value of epigenetic profiles of cfDNA can extend beyond detection and quantification to more phenotypic characterization of ctDNA, which could contribute to clinical utility by boosting prognostic performance. Importantly, many of the results from this study were replicated in a larger cohort of patients consisting of both HPV‐associated and HPV‐negative HNSCC [179], indicating that this approach could be generalizable across subtypes of HNC.

5.3. CTCs in Nonviral HNSCC

CTCs are relatively scarce in HNSCC, with only late‐stage disease demonstrating consistent detection [180]. In one study of patients with Stages II–IV HNSCC, E48 RNA, an epithelial marker, was detected by RT‐PCR in 35% of bone marrow samples and 10% of blood samples, with only late‐stage patients demonstrating detectability. In another study, peripheral pretreatment blood from HNSCC patients was subjected to CTC enrichment using EpCAM antibodies [181]. Of the 18 HNSCC patients studied, only 8 (44%) had detectable CTCs, 7 of whom had late‐stage disease. The CellSearch technique also targets EpCAM‐positive cells, along with cytokeratins 8, 18, and 19, and is coupled with flow cytometry [182]. In a multi‐center prospective study of patients with advanced‐stage HNSCC, 40% of pretreatment blood samples had detectable CTCs.

While CTCs have low sensitivity in early‐stage disease, multiple studies have shown potential prognostic value. CTC detectability was prognostic among patients with recurrent or metastatic HNSCC [183] as well as locally advanced OSCC [184, 185]. Other studies have evaluated dynamic changes in CTC levels in patients with HNSCC [186]. A reduction in CTC levels 2–4 weeks after treatment initiation correlated with better progression‐free survival and overall survival. Taken together, detectable CTCs appear to reflect a group of patients with highly aggressive disease behavior with poor prognosis.

5.4. EVs in Nonviral HNSCC

Apart from cfDNA and CTCs, EVs have also been assessed in liquid biopsies of HNC patients. Large EVs that are 100–1000 nm in size, termed microvesicles, have been quantified within saliva from newly diagnosed OSCC patients [187]. Purified salivary microvesicles were quantified by flow cytometry with higher counts in patients with OSCC than healthy controls or patients with benign oral disease. Salivary microvesicle counts were also correlated with disease burden as measured by lymph node status and clinical stage. While these findings are of interest, further work is needed on salivary EVs as a liquid biopsy.

EVs can also be quantified from peripheral blood plasma. In one study, plasma exosomes were measured pre‐, mid‐, and posttreatment in HNSCC patients receiving cetuximab and ipilimumab with radiotherapy [188]. Tumor‐derived exosomes were isolated by size‐exclusion chromatography, differentiating them from T‐cell‐derived exosomes via immunocapture Cluster of Differentiation (CD)3‐positivity. Levels of tumor‐derived exosomes increased from pretreatment levels among those patients with disease recurrence. In another study, exosomal levels of the immunosuppressive molecule, programmed death‐ligand 1 (PD‐L1), were correlated with clinical stage [189]. These data highlight the potential of evaluating specific targetable molecules within exosomes that may inform prognosis and/or therapeutic targeting.

6. Future Opportunities for Clinical Impact

With a growing catalog of methodologies available for HNC liquid biopsy applications (Table 3), major gaps remain in establishing clinical utility and optimal implementation of specific tests. Here, we propose a number of topics for future investigation.

TABLE 3.

Feature table of common detection platforms for HNC liquid biopsies.

Detection methods and liquid biopsy source		Sensitivity	Advantages	Disadvantages	References
qPCR	Plasma cfDNA	9%–97%	Low cost Accessible Quantitative	Limited discovery Limited multiplexing capabilities	[73, 85, 91, 102, 103, 130, 190]
	Saliva cfDNA	30%–81%
Digital PCR	Plasma cfDNA	12%–100%	Accessible Quantitative Reproducible	Moderate cost Limited discovery Limited multiplexing capabilities	[88, 89, 90, 91, 92, 133, 168, 169, 173, 191, 192]
	Saliva cfDNA	33%–73%
Antibody enrichment	Blood CTCs	40%–44%	Fast turnaround time Direct detection of malignant cells	Moderate cost Limited discovery Low sensitivity	[181, 182]
Methylation‐specific PCR and qPCR	Plasma cfDNA	18%–71%	Low/moderate cost Accessible Reproducible	Limited discovery Limited multiplexing capabilities	[143, 144, 145, 147, 148, 151, 152, 153, 155, 156, 159, 161, 162, 165, 167, 170, 171, 172, 174, 175, 176, 193, 194]
	Saliva cfDNA	3%–100%
Next‐generation sequencing	Plasma cfDNA	40%–100%	High sensitivity Increased discovery Multiplexing capabilities Quantitative	High cost Increased turnaround time	[65, 95, 118, 119, 120, 121, 122, 123, 125, 126, 127, 128, 129, 130, 133, 134, 135, 136, 137, 138]
	Saliva cfDNA	63%–96%

6.1. Multimodal Liquid Biopsy Profiling

In this review, we have summarized the diversity of molecular analytes and biofluid sources in HNCs. Evaluating multiple analytes and/or biofluids may boost test performance and expand clinical utility. A number of studies have already begun to conduct multimodal liquid biopsy assessments. For instance, Burgener et al. [132] and Sanz‐Garcia et al. [137] both profiled multiple molecular targets (i.e., mutations, viral sequences, and/or aberrant DNA methylation) within ctDNA, while Wang et al. [116] and Earland et al. [65] both evaluated multiple biofluids. These are just a few examples of multimodal profiling approaches that could allow for a high‐resolution snapshot of disease burden, biology, and behavior. One tempting application where such multimodal profiling could have a major clinical impact is in the setting of post‐operative MRD detection, where complementary tests of saliva, lymph, and peripheral blood could be used to address the risk of local, regional, and distant relapse, respectively.

6.2. Protocol Standardization for Saliva

Techniques for peripheral blood handling for the purposes of liquid biopsy are now well established [195]. A similar degree of standardization would be of great benefit for saliva and lymph fluids. For saliva, different groups have used various combinations of whole saliva, saliva rinses, and oral swabs [196, 197, 198]. Typically, saliva is collected in a receptacle that can be subjected to centrifugation to isolate or remove the cellular components. DNA should be isolated within hours of collection to avoid degradation, or else preserved with fixative solutions to allow for batch processing. However, most fixative solutions contain lysis buffers that lead to a release of genomic DNA, which can dilute cfDNA. Unfortunately, the lack of standardization of saliva collection and analyte isolation procedures makes cross‐study comparisons difficult.

The ideal saliva receptacle should allow for the isolation of different liquid biopsy analytes. Cell lysing agents should be avoided so that cellular components and cfDNA can be analyzed separately. A DNA stabilizing agent would allow for delayed processing and improve clinical workflows. Finally, an antimicrobial agent would decrease bacterial outgrowth. Saliva has a unique microbiome that is altered in patients with HNC [199, 200, 201], and an accurate representation of salivary bacteria could lead to additional promising research directions [202].

6.3. Clinical Use of HNC Liquid Biopsies

Moving liquid biopsies into clinical use for HNC requires a number of considerations. The distribution of disease in local, regional, and distant sites will impact the potential utility of different tests. Due to its proximity to primary tumors, saliva may be best utilized for screening and local disease assessment. Many HNC patients experience treatment‐related xerostomia, which can impede salivary collection. Peripheral blood may be better suited for posttreatment surveillance in these patients and has demonstrated higher sensitivity for detecting metastatic spread. While sampling of peripheral blood can be easily repeated, lymph fluid has thus far only been studied in the immediate post‐operative period while surgical drains are in situ.

6.4. Implementation of HNC Liquid Biopsies Into Clinical Practice

To drive broad adoption of new tests, clinical utility studies are needed to demonstrate improvements in disease‐related outcomes, toxicity, and/or health system costs. To date, strong evidence of clinical utility for liquid biopsy testing in HNC has remained elusive. This gap is reflected in the absence of such tests in NCCN guidelines (v.4.2005). Prospective clinical trials are currently testing the potential for plasma EBV DNA and HPV DNA to lead to improved outcomes in NPC and OPSCC, respectively, but caution should be exercised until definitive results are available. The failure of adjuvant chemotherapy to improve relapse‐free survival in NPC patients with detectable plasma EBV DNA after chemoradiotherapy [76] serves as an important reminder that rigorously conducted interventional studies are needed.

Early investigations have pointed to the potential for cost savings from implementing liquid biopsy tests in certain clinical settings. For instance, detecting recurrence of HPV‐driven OPSCC in the surveillance setting could result in reduced costs by decreasing the need for other diagnostic tests [203]. Future investigations could follow the blueprint of ctDNA in Stage II colon cancer and Oncotype DX in breast cancer, which provide prognostic information that leads to cost savings related to reductions in unnecessary therapy in favorable risk patients [204, 205]. As the cost of sequencing modalities decreases, accessibility and cost‐effectiveness of emerging tests may continue to improve.

For patients to have access to liquid biopsy tests, clinical labs must be capable of providing specialized services. The success of implementation depends on reimbursement, assay cost, complexity, turnaround time requirements, testing volume, and regulatory scrutiny. Centralized lab testing often addresses several of these considerations through increased scale and lower variability, but can lead to limited access. Test standardization and harmonization between multiple clinical labs could increase accessibility; recent efforts to harmonize plasma EBV DNA testing illustrate the potential for this approach.

The future is bright for liquid biopsy research in HNC. Existing technologies are already making their way into clinical practice, with much work remaining to establish their optimal clinical settings and utility. Emerging technologies that target new analytes and biofluids have the potential to revolutionize how HNC patients are diagnosed and managed.

Conflicts of Interest

Scott V. Bratman declares ownership and leadership of Adela and patents/licensing with Roche and Adela. The other author declares no conflicts of interest.

Brooks P. J. and Bratman S. V., “Liquid Biopsies in Head and Neck Cancers: Recent Developments Across Biofluids, Analytes, and Molecular Features,” Head & Neck 47, no. 11 (2025): 3150–3171, 10.1002/hed.70009.

Data Availability Statement

Data sharing is not applicable to this article as no new data were created or analyzed in this study.