TERT Promoter Mutations Are Enriched in Oral Cavity Cancers and Associated With Locoregional Recurrence

PURPOSE

Telomerase reverse transcriptase (TERT) promoter mutations are prognostic in many cancers and have been observed in human papillomavirus (HPV)–negative head and neck squamous cell carcinomas (HNSCCs). However, the role of TERT promoter mutations in HPV-negative HNSCCs remains poorly understood in these cancers, which have increased risk for locoregional failure (LRF).

PATIENTS AND METHODS

We retrospectively identified patients who were diagnosed with HNSCC between July 1, 2004, and October 12, 2017, at Memorial Sloan Kettering Cancer Center and whose tumors underwent next-generation sequencing using the MSK-IMPACT panel. Patients with HPV-positive oropharyngeal squamous cell carcinoma (SCC) were excluded. Cumulative incidence of LRF, patterns of failure, and overall survival were measured.

RESULTS

We identified 117 patients with SCC of the oral cavity (OSCC), larynx, hypopharynx, or HPV-negative oropharynx whose tumors underwent next-generation sequencing. Sequencing was performed on 95 tumors that were obtained after recurrence and 22 that were obtained before recurrence. TERT promoter mutations were enriched in OSCC compared with laryngopharyngeal cancers (81.1% v 7.0%; P < .001), which was the largest genetic difference between these anatomic disease subsites. TERT promoter mutations were associated with LRF in OSCCs (Gray's test, P < .001) and in the overall cohort (Gray's test, P < .001). On multivariate analysis, TERT promoter mutations were associated with an increased risk for LRF (subdistribution hazard ratio, 2.82; 95% CI, 1.47 to 5.42; P = .0019), independent of oral cavity primary site and TP53 mutation status.

CONCLUSION

TERT promoter status is associated with the cumulative incidence of LRF and patterns of failure. TERT promoter mutations may define a subset of OSCCs with unique pathogenesis that is associated with an increased risk of LRF. Validation in prospective cohorts is warranted.

INTRODUCTION

Squamous cell carcinomas (SCCs) of the oral cavity (OSCC) are a unique subset of head and neck cancers that have distinct management strategies,^1-3 genomic and environmental risk factors,⁴ and prognosis.⁵ Despite management by complete resection followed by radiation or chemoradiation, OSCCs are consistently associated with an increased risk for locoregional failure (LRF) compared with laryngopharyngeal SCCs.^2,6-9 These differences in OSCCs cannot be fully explained by a confounding risk of human papillomavirus (HPV)–positive oropharyngeal cancers.^2,5 Intriguingly, epidemiologic data have shown an increased incidence of OSCCs in young never smokers in the United States. However, no clear reason has been identified to explain these trends.¹⁰

CONTEXT

• Key Objective

• Oral cavity squamous cell carcinomas (OSCCs) have unique clinical behavior among head and neck squamous cell carcinomas (HNSCCs), including resistance to definitive chemoradiotherapy and propensity for locoregional recurrence. Exome sequencing studies to date have not identified clear genetic differences between OSCC and human papillomavirus–negative HNSCC. This study examines the genetic profile of HNSCC using a validated cancer-specific sequencing panel that includes the telomerase reverse transcriptase (TERT) gene promoter. Genetic differences were correlated with clinical outcomes.

• Knowledge Generated

• TERT promoter mutations were strongly associated with OSCC and were enriched among recurrent tumors. Mutations in the TERT promoter represented the greatest genetic difference between OSCC and other human papillomavirus–negative squamous cell carcinomas of the larynx and pharynx. HNSCC-harboring TERT promoter mutations were associated with a locoregional pattern of failure.

• Relevance

• TERT promoter mutations are identified in a subset of OSCCs and are associated with increased risk of locoregional failure.

Despite differences in the tumor behavior and epidemiology of OSCCs, systematic genomic analyses have failed to identify putative genomic factors that molecularly distinguish OSCCs from laryngopharyngeal SCCs. Studies by The Cancer Genome Atlas and the University of Texas MD Anderson Cancer Center have identified a small subset of oral cavity tumors characterized by mutations in HRAS and CASP8, as well as a few copy number alterations (CNAs).^11-13 However, this subset of genes has not explained the differences in tumor behavior of OSCCs, which have been found to otherwise have similar genomic alterations as HPV-negative laryngopharyngeal SCCs.^11,14,15

Mutations in the telomerase reverse transcriptase (TERT) promoter region have been identified in several types of cancers, including melanoma,^16-18 cutaneous SCC,¹⁹ glioma,²⁰ thyroid carcinoma,^21,22 urothelial carcinoma, renal cell carcinoma,²³ and OSCC.^23,24 Two hotspot mutations, C228T and C250T, constitute the majority of TERT promoter mutations and are located −124 and −146 base pairs upstream from the TERT ATG start site; because of this location in the upstream coding region, hotspot TERT promoter mutations could not be identified in earlier sequencing efforts.^11-15 Additionally, because these mutations are not transcribed, they cannot be identified in RNA sequencing data sets.

In addition to the canonical C228T and C250T mutations, less common variants such as the activating C228A and C250A mutations have also been identified, characterized in vitro, and observed in other cancer types.²⁵ These mutations form de novo E-twenty-six (Ets) transcription factor binding motifs, resulting in increased TERT transcription²⁵ and immortalization through the GA-binding protein-β1L isoform transcription factor.²⁶ In several cancers, these canonical mutations occur early during carcinogenesis²⁷ and are associated with poor prognosis.^28-30 TERT promoter mutations³¹ and telomerase enzymatic activity^32,33 have also been identified in tumor-adjacent tissue, suggesting that telomerase reactivation may occur early during carcinogenesis. We previously observed that patients with metastatic OSCC had poorer overall survival (OS) than those with laryngeal, hypopharyngeal, and oropharyngeal SCCs.⁵ We hypothesized that this effect could be explained by the higher proportion of patients with OSCC who had TERT promoter alterations.

In an earlier analysis, we found that 76% (16 of 21) of recurrent and metastatic OSCCs harbored TERT promoter alterations³⁴; however, this analysis was limited by the sample size. In the present study, we performed targeted capture next-generation sequencing (NGS) on tumors that were obtained before and after recurrence from an expanded cohort of patients to further investigate the differences between oral cavity and laryngopharyngeal SCCs, specifically by focusing on the frequency of TERT promoter mutations. Using data on patterns of failure, we hypothesized that TERT promoter–mutant OSCCs constitute a distinct biologic subgroup with aggressive tumor behavior that drives the poor prognosis observed in patients with OSCC.

PATIENTS AND METHODS

Patient Selection

This retrospective study was approved by the Memorial Sloan Kettering Cancer Center (MSK; New York, NY) Institutional Review Board, and human investigations were also performed in accordance with an assurance filed with and approved by the Department of Health and Human Services. Patients with SCC of the oral cavity, oropharynx, larynx, or hypopharynx, including recurrent or metastatic tumors, were identified. Patients with a prior history of head and neck radiation treatment, cancer predisposition syndromes, or oropharyngeal SCC that was p16-positive by immunohistochemistry or HPV-positive by in situ hybridization were excluded. Data on all patients' clinical history, including the date of initial diagnosis, management strategy, clinicopathologic details, history of leukoplakia, erythroplakia, or lichen planus, and patterns of failure, were abstracted from electronic medical records and reviewed.

NGS

Tumor samples from all patients were evaluated using the MSK-IMPACT (MSK-Integrated Mutation Profiling of Actionable Cancer Targets) NGS panel, an institutional targeted sequencing platform that has previously been described,^31,35 to determine their mutational profiles (Data Supplement). DNA was obtained from either formalin-fixed paraffin-embedded samples or recurrent tumors. The panel identifies single-nucleotide variants, CNA, insertions, deletions, and structural rearrangements within a panel of 410 genes and in selected promoter regions (such as TERT) that are relevant to cancer (Data Supplement). Mutations were functionally annotated using the OncoKB database.³⁶

Study Design and Variant Classification of TERT Promoter Mutations

LRF was defined as any recurrent disease in the primary site or in the neck with a disease-free period of < 10 years. Patients whose cancer recurred in the primary site after a 10-year disease-free period were considered to have second primary tumors. Cumulative incidence of LRF was measured from the date of diagnosis to the date of locoregional progression, with death as a competing risk. OS was measured from the date of diagnosis to the date of death, censoring on the date of the last follow-up visit.

Tumors that harbored C228T, C250T, C228A, and C250A mutations were considered to have functional TERT promoter mutations, whereas other alterations were considered passenger mutations that had a functionally wild-type TERT promoter gene. TERT amplification was not considered to be a canonical TERT promoter mutation for this study because it may represent a distinct mechanism of immortalization.

Statistical Analysis

Survival analysis was performed using the Kaplan-Meier estimator with the log-rank test and the Cox proportional hazards regression model. The risk of LRF and distant metastases over time was assessed using the cumulative incidence method, accounting for the competing risk of death; univariate analyses were performed using Gray's test. To investigate whether the TERT promoter mutation is an independent prognostic factor, rather than a surrogate for OSCCs' anatomic site, we performed a multivariate analysis that controlled for the anatomic subsite (OSCC v other SCCs) and TP53 mutation status; multivariate analyses were performed using Fine and Gray competing risk regression model, characterized by the subdistribution hazard ratio (SHR). Median follow-up time from the date of initial diagnosis was determined using the inverse Kaplan-Meier method.

To control for selection bias, we further investigated the cumulative incidence of LRF in a subset of patients who had tumors that were sequenced before recurrence. Fisher's exact test was used to compare the patterns of failure at the time of tumor sequencing and to compare the proportion of promoter mutations between tumors that were sequenced before and after recurrence.

RESULTS

Patient Characteristics

We identified 117 patients who had SCC of the oral cavity (n = 74), larynx (n = 24), or hypopharynx (n = 5), as well as HPV-negative SCC of the oropharynx (n = 14). Patient characteristics are described in Table 1 (Data Supplement). The initial date of diagnosis ranged from July 1, 2004, to October 12, 2017, and the median follow-up was 5.6 years. Sequencing was performed on 125 tumor samples from 95 patients with recurrent disease and 22 who were newly diagnosed or treated but had no evidence of recurrent disease (Data Supplement). Some tumor samples were sequenced years after treatment to differentiate between metastasis and second primary tumor. Seven patients had two samples that were sequenced, and one patient had three samples that were sequenced.

TABLE 1.

Patient Characteristics at the Time of Tumor Sequencing (N = 117)

TERT Promoter Mutations are Enriched in Oral Cavity SCC

The nonsynonymous mutation and CNA profiles of the tumors, stratified by oral cavity versus hypopharynx, larynx, and oropharynx subsites, are shown in Figure 1. Only mutations that were identified in five or more samples are shown. Like The Cancer Genome Atlas analysis, mutations in TP53, CASP8, and PIK3CA were common, as well as amplifications in 11q13.3, which contains CCND1, FGF19, FGF3, and FGF4, and deletions in CDKN2A (Data Supplement).

FIG 1.

Somatic mutational spectrum and copy number alteration profiles of (A) SCCs of the oral cavity and (B) SCCs of nonoral cavity primary anatomic subsites (including SCCs of the larynx, hypopharynx, and oropharynx and excluding HPV+ or p16+ SCCs of the oropharynx). Only mutations that were identified in ≥ 5% of samples are shown. TERT promoter alterations clustered in oral cavity cancers. amp, amplification; del, deletion; HPV+, HPV-positive; HPX, hypopharynx; intra del, intrachromosomal deletion; LX, larynx; mut, mutation; OC, oral cavity; OPX, oropharynx; p16+, p16-positive; pro mut, promoter mutation; SCC, squamous cell carcinoma; SNV, single-nucleotide variant; TERT, telomerase reverse transcriptase.

Sixty-three patients had TERT promoter alterations. TERT promoter mutations were enriched in OSCC compared with other anatomic subsites (81.1% [60 of 74] v 7.0% [3 of 43], respectively; odds ratio [OR], 54.2; 95% CI, 14.4 to 310.8; P < .001; Fig 2A). The spectrum of TERT promoter alterations is shown in Figure 2B. The most common TERT promoter mutations were located at two specific loci on hg19 chromosome 5 position 1,295,228 (228) and hg19 chromosome 5 position 1,295,250 (250). Seven patients with C250T alterations harbored synchronous C254T alterations on the same strand. TERT promoter C228T and C250T mutations were mutually exclusive. One tumor harbored a CC434TT variant, which has unknown function.

FIG 2.

(A) Distribution of TERT promoter mutations in head and neck squamous cell carcinomas. TERT promoter mutations were more prevalent among oral cavity squamous cell carcinomas than in the oropharynx, larynx, and hypopharynx squamous cell carcinomas (OR, 54.2; 95% CI, 14.4 to 310.8; P < .001). (B) TERT promoter mutation spectrum and distribution. HPV, human papillomavirus; HPX, hypopharynx; OP, oropharynx; OR, odds ratio; SCC, squamous cell carcinoma; TERT, telomerase reverse transcriptase.

Among seven patients from whom multiple tumor samples were sequenced, TERT promoter mutations were not observed in any sample from three patients and four patients had at least one sample that harbored a TERT promoter mutation. In all four cases, the TERT promoter mutation was preserved across all samples within a given patient. Across all seven patients, no patient had discordant TERT promoter alterations between sequenced samples.

Patients with alterations at the −228 and −250 loci were considered to have functional TERT promoter alterations, as these have been shown to correspond to increased telomerase activity in vivo.²⁵ In total, 98.4% (62 of 63) of mutations in the TERT promoter were functional. For the remaining analysis, TERT promoter mutant tumors consisted of those that harbored functional TERT promoter mutations.

TERT Promoter Mutations Are Associated with a Higher Cumulative Incidence of LRF

In the overall study cohort, TERT promoter alterations were associated with a higher cumulative incidence of LRF (Gray's test, P < .001; Fig 3A) and the same trend was observed in the subset of patients with OSCC (Gray's test, P < .001; Fig 3B). Multivariate analyses showed that TERT promoter mutation status was associated with LRF (SHR, 2.82; 95% CI, 1.47 to 5.40; P = .0019), independent of oral cavity primary site (P = .34) and TP53 mutation status (P = .89). In a separate multivariable model, the association between TERT promoter mutations risk of LRF (SHR, 3.28; 95% CI, 140 to 7.71, P = .0063) was independent of margin status (positive v negative, P = .93) and extracapsular extension (P = .69).

FIG 3.

Cumulative incidence of locoregional failure from the time of diagnosis, stratified by the presence or absence of TERT promoter mutations. The cumulative incidence of locoregional failure in (A) the entire study cohort of patients with HPV-negative squamous cell carcinoma of the head and neck, (B) the subset of patients with primary squamous cell carcinoma of the oral cavity, (C) the entire study cohort of patients whose tumors were stratified by tumor type, and (D) the subset of patients whose tumors were sequenced before recurrence. HNSCC, head and neck squamous cell carcinomas; HPV, human papillomavirus; LRF, locoregional failure; mut, mutation; SCC, squamous cell carcinoma; TERT p, telomerase reverse transcriptase promoter; wt, wild-type.

To investigate heterogeneity between the different TERT promoter mutations, we subdivided the canonical TERT promoter mutations into C250T and C228A/T groups. The cumulative incidence of LRF was similar between the two types of TERT promoter mutations (P = .61), and both were associated with a higher cumulative incidence of LRF compared with TERT promoter wild-type tumors (C228A/T v wild-type, P = .0015; C250T v P = .0035; Fig 3C).

To control for selection bias in the overall cohort, we investigated whether TERT promoter status was prognostic among a limited subset of 22 patients whose tumors were sequenced before recurrence (Fig 3D). In this subset, all eight subsequent locoregional recurrences occurred among patients with TERT promoter alterations (Gray's test, P < .001).

Among 74 OSCC tumors, 17 were sequenced before recurrence and 57 were sequenced after recurrence. Ten of 17 (58.8%) OSCC tumors that were sequenced before recurrence harbored TERT promoter mutations; by contrast, 49 of 57 (86.0%) recurrent OSCC tumors harbored TERT promoter mutations (OR, 4.2; 95% CI, 1.04 to 17.03; P = .034).

Patterns of Failure in TERT Promoter–Mutant Versus TERT Promoter Wild-Type Tumors

Among 95 patients whose tumors were sequenced after recurrence, the patterns of failure at the time of sequencing were correlated with TERT promoter mutation status (Table 2). TERT promoter mutations were associated with LRF at the time of tumor sequencing (80.8% [42 of 52] v 51.2% [22 of 43] with TERT promoter wild-type tumors, respectively; OR, 3.95; 95% CI, 1.5 to 11.2; P = .004). Patients with TERT promoter–mutant tumors had a lower frequency of distant metastases at the time of sequencing than those with TERT promoter wild-type tumors (59.6% [31 of 52] v 76.7% [33 of 43], respectively; OR, 0.45; 95% CI, 0.16 to 1.2; P = .08). The distribution of locoregional, distant, or both locoregional and distant failures was significantly different between TERT promoter–mutant and wild-type tumors (P = .018).

TABLE 2.

Patterns of Failure at the Time of Tumor Sequencing Performed After Recurrence (n = 95)

Ninety-two percent (47 of 51) of patients with OSCC that developed locoregional recurrence at any time in the follow-up period harbored TERT promoter mutations. By comparison, 52% (12 of 23) of patients with OSCC who did not develop LRF harbored TERT promoter mutations.

TERT promoter mutations were associated with shorter disease-free survival (median 0.8 v 1.1 years, P = .003). Despite differences in disease-free survival and the cumulative incidence of LRF, no differences in distant metastasis-free survival (P = .91; Data Supplement) or OS (P = .22; Data Supplement) were identified between TERT promoter–mutant and wild-type tumors. This may reflect a greater proportion of isolated LRFs treated with definitive locoregional salvage therapy among patients with TERT promoter–mutant tumors (73% [35 of 48] v 46% [13 of 28], Fisher's P = .028).

DISCUSSION

Our data demonstrate that TERT promoter mutations are enriched in OSCCs as compared with head and neck squamous cell carcinomas (HNSCCs) originating from other anatomic subsites and are further enriched in recurrent OSCCs (86%). TERT promoter mutations represent the greatest genetic difference between OSCCs and HPV-negative HNSCCs, which otherwise shared recurrent mutations and CNAs in known drivers of HNSCC, including in PIK3CA, TP53, CDKN2A, NOTCH1, and FAT1. Furthermore, we found TERT promoter mutations to be associated with a high risk for LRF. This increased risk for LRF is independent of the oral cavity primary site, TP53 mutation status, extracapsular extension, and positive surgical margins, suggesting that TERT promoter mutations are an independent biomarker of LRF, rather than a surrogate for OSCC or other known prognostic markers. The subset of TERT promoter–mutant OSCCs may explain the high risk of LRF that has historically been associated with OSCCs.^2,5,6,8,9

TERT promoter mutations were significantly enriched in recurrent vs nonrecurrent OSCC (86.0% v 58.8%). The proportion of tumors harboring TERT promoter mutations is also enriched compared with the baseline risk of 32%-65% in published cohorts of newly diagnosed OSCC.^37-39 The reasons for this variability are unclear. TERT promoter mutations have been associated with exposure to betel nut³⁷ and alcohol,³⁹ which may vary between studies and populations.

To date, three studies have evaluated the prognostic value of TERT promoter mutations in HNSCC with discordant results. Chang et al used Sanger sequencing to identify TERT promoter mutations in the tumors from 201 patients with newly diagnosed oral cavity cancers, including 94 tumors with C228T mutations and 26 with C250T mutations. This study did not identify an association between TERT promoter mutations and disease-free survival, disease-specific survival, or OS. Of note, TERT promoter mutations were analyzed on a per locus basis, comparing TERT promoter C228T versus wild-type tumors and TERT promoter C250T versus wild-type tumors. Because TERT promoter mutations are mutually exclusive,^24,37 the TERT promoter C228 wild-type group comprised a mixture of C228 wild-type and either C250 wild-type or mutant (C250T) tumors and the TERT promoter C228 wild-type group comprised a mixture of C228 wild-type and either C250 wild-type or mutant (250T) tumors and the TERT promoter C250 wild-type control group comprised a mixture of C250 wild-type and either C228 wild-type or mutant (C228T) tumors.³⁷ By contrast, Arantes et al³⁹ evaluated the prognostic significance of TERT promoter mutations in a cohort of 88 patients with HNSCC treated with primary surgery, including 69 patients with OSCC and 24 patients with TERT promoter mutations. The authors identified an association between C228T—but not C250T mutations—and decreased disease-free survival, suggesting potential functional heterogeneity between the different mutations. In a third study, Boscolo-Rizzo et al³⁸ evaluated the prognostic impact of TERT promoter mutations in HNSCC, but this study included just 27 patients with OSCC, 10 of which harbored TERT promoter mutations. Although this study did not reveal an association between TERT promoter mutations and OS, this was potentially limited by the small number of TERT promoter–mutant tumors.

In contrast to the published data, this study specifically examined the distribution of TERT promoter mutation patterns of failure on the basis of TERT promoter mutation status; in our patient population, TERT promoter mutations were associated with an increased risk of LRF, but not distant failure or OS. Of note, both C228A/T and C250T mutations were associated with increased risk of LRF in our data set and we did not observe a difference in the cumulative incidence of LRF between these different TERT promoter mutations.

The correlation between LRF and TERT promoter mutations identified in this study is consistent with a previous body of work on telomerase enzymatic activity in head and neck cancers, much of which predated the discovery of TERT promoter mutations in 2013.^17,33 Patel et al demonstrated telomerase activity in 78.2% of invasive cancers, 53.1% of adjacent normal tissue, and 85% of precancerous lesions and found disease-free survival to be correlated with both telomerase activity in the adjacent normal tissue and telomere length in malignant tissues. These data are also recapitulated in the study by Boscolo-Rizzo et al,³⁸ which identified an association between short telomeres within the tumor and clinical outcomes.

Several mechanisms have been proposed for the prognostic effect of TERT promoter mutations. In thyroid cancers, TERT promoter mutations alone only weakly predict disease-free survival and OS, but tumors that harbor both TERT promoter and BRAF^V600E mutations account for nearly all the recurrences and deaths that have been observed in differentiated thyroid cancers. TERT promoter mutations are known to generate de novo binding sites for the family of Ets transcription factors, which are downstream of the BRAF pathway. Thus, comutation in both BRAF and the TERT promoter may synergistically drive proliferation, genomic instability, and immortalization.²² TERT promoter might also be coupled with other upstream regulators; for instance, hypoxia has been shown to induce Ets-1 transcription factors via HIF-1, potentially driving reactivation of telomerase in TERT promoter–mutant tumors.

Interpretation of LRF-free survival, distant metastasis-free survival, and OS analyses can be confounded by patient selection and immortal time biases. At our institution, NGS was preferentially performed after recurrence or development of distant metastases. Thus, our population was enriched with patients who recurred and cumulative incidence estimates are therefore inflated. Despite this limitation, the differences in cumulative incidence between patients with and without TERT promoter mutations are more robust to this bias. To address this limitation, we repeated our analysis in the subset of patients whose tumors were sequenced before recurrence and found that all cases with later locoregional recurrence harbored TERT promoter mutations. We then characterized the patterns of failure at the time of tumor sequencing, which controlled for immortal time bias, and similarly identified a higher likelihood of locoregionally recurrent disease among TERT promoter–mutant tumors.

Our data mirror the prognostic effect of TERT promoter mutations that has been observed in several other cancer types^{21,22,29,30,40-42} and confirmed the observations from our previous study.³⁴ Additional validation in a group of patients with newly diagnosed high-risk HNSCC is needed to confirm these findings for OSCCs.

TERT promoter mutations occur early during oncogenesis and are likely preserved between primary and recurrent tumors. In the evolutionary analysis from the Pan-Cancer Analysis of Whole Genomes, TERT promoter mutations were most often early clonal alterations that were projected to occur during the oncogenesis of head and neck SCCs.^27,43 Among seven patients in our cohort from whom multiple tumor samples were sequenced, TERT promoter mutations were preserved within patients. This is also supported by the detection of TERT promoter mutations in tumor-adjacent normal mucosa in some³⁷ but not all studies.³⁸ Although paired sequencing of primary and recurrent tumors would be needed to definitively answer this question, our study suggests that TERT promoter mutations are likely preserved across primary tumors and recurrences and that their use as a biomarker is less likely to be complicated by differences in sequencing technique or by heterogeneity within the same tumor.

In conclusion, our study demonstrates that TERT promoter mutations are enriched in OSCCs and represent the most significant genomic difference between oral cavity and laryngopharyngeal SCCs. TERT promoter mutations are also enriched in recurrent OSCCs compared with newly diagnosed OSCCs and are associated with a high risk for LRF, independent of the oral cavity primary anatomic subsite. The subset of TERT promoter–mutant tumors may identify a distinct subtype of OSCCs that drives the high risk of locoregional recurrence in tumors arising from the oral cavity. Validation in prospective cohorts may facilitate strategies to personalize the management of patients with OSCC and confirm a potential therapeutic target.

AUTHOR CONTRIBUTIONS

Conception and design: Yao Yu, Linda Chen, Nora Katabi, Jennifer Cracchiolo, Jingming Wang, Nadeem Riaz, Nancy Lee

Financial support: Nancy Lee

Administrative support: Luc Morris, Nancy Lee

Provision of study materials or patients: Jung Kang, Jay O. Boyle, Luc Morris

Collection and assembly of data: Yao Yu, Dan Fan, Xinmao Song, Jung Kang, Snjezana Dogan, Cristina Valero, Jingming Wang, Nancy Lee

Data analysis and interpretation: Yao Yu, Kaveh Zakeri, Linda Chen, Jung Kang, Sean McBride, C. Jillian Tsai, Lara Dunn, Eric Sherman, Jennifer Cracchiolo, Marc Cohen, Jay O. Boyle, Mark Lee, Jingming Wang, Richard Wong, Luc Morris, Nadeem Riaz, Nancy Lee

Manuscript writing: All authors

Final approval of manuscript: All authors

Accountable for all aspects of the work: All authors

AUTHORS' DISCLOSURES OF POTENTIAL CONFLICTS OF INTEREST

The following represents disclosure information provided by the authors of this manuscript. All relationships are considered compensated unless otherwise noted. Relationships are self-held unless noted. I = Immediate Family Member, Inst = My Institution. Relationships may not relate to the subject matter of this manuscript. For more information about ASCO's conflict of interest policy, please refer to www.asco.org/rwc or ascopubs.org/po/author-center.

Open Payments is a public database containing information reported by companies about payments made to US-licensed physicians (Open Payments).