Whole genome duplication in oral squamous cell carcinoma in patients younger than 50 years: implications for prognosis and adverse clinicopathological factors.

INTRODUCTION: Oral squamous cell carcinoma (OSCC) in the young (<50 years), without known carcinogenic risk factors, is on the rise globally. Whole genome duplication (WGD) has been shown to occur at higher rates in cancers without an identifiable carcinogenic agent. We aimed to evaluate the prevalence of WGD in a cohort of OSCC patients under the age of 50 years.
METHODS: Whole genome sequencing (WGS) was performed on 28 OSCC patients from the Sydney Head and Neck Cancer Institute (SHNCI) biobank. An additional nine cases were obtained from The Cancer Genome Atlas (TCGA).
RESULTS: WGD was seen in 27 of 37 (73%) cases. Non-synonymous, somatic TP53 mutations occurred in 25 of 27 (93%) cases of WGD and were predicted to precede WGD in 21 (77%). WGD was significantly associated with larger tumor size (p = 0.01) and was frequent in patients with recurrences (87%, p = 0.36). Overall survival was significantly worse in those with WGD (p = 0.05).
CONCLUSIONS: Our data, based on one of the largest WGS datasets of young patients with OSCC, demonstrates a high frequency of WGD and its association with adverse pathologic characteristics and clinical outcomes. TP53 mutations also preceded WGD, as has been described in other tumors without a clear mutagenic driver.

INTRODUCTION

Alterations in ploidy are common occurrences in the evolution of nearly all cancer types. Whole genome duplication (WGD), or tetraploidisation, occurs when the entire complement of diploid chromosomes is doubled. WGD is thought to be a precursor event to aneuploidy and further chromosomal instability with increased somatic copy number alterations. , The tetraploidy that develops following WGD is more tolerant of chromosomal loss as compared with diploidy and thus permits development of aneuploidy more easily. The incidence of WGD is highly variable amongst the various cancer types and ranges from 5% in non‐Hodgkin lymphomas and gastrointestinal neuroendocrine tumors to 58% in germ cell tumors. , WGD can occur at an early stage in tumor development and transitions cells from a pre‐malignant to malignant state, as has been described in lung, oesophageal, and cervical adenocarcinoma. , , In some tumors, it can occur at later stages and may only be seen in some subclonal populations, as documented in renal cell carcinoma, myeloma, and melanoma.

The timing of the gatekeeper Tumor Protein 53 (TP53) mutations and onset of WGD during carcinogenesis have also been studied and TP53 mutations have been found to precede WGD in tumors that lack a clear mutagenic driver such as germ cell tumors. WGD may provide a survival advantage to the cancer cell and is an adverse prognostic factor in several malignancies. Recently, a study of WGD+ and WGD‐ breast cancer cell lines treated with drugs that induce replication stress demonstrated reduced viability in the WGD+ cell lines as compared to their WGD‐ counterparts suggesting that WGD status may guide treatment strategies in the future.

In head and neck squamous cell carcinoma (HNSCC), WGD has been reported in approximately 25% of recurrent and/or metastatic cancers. HNSCC encompasses a heterogeneous range of carcinomas with varying aetiology and biological behavior depending on anatomic location including oral cavity squamous cell carcinoma (OSCC), human papillomavirus (HPV)‐associated oropharyngeal squamous cell carcinoma, and solar damage associated cutaneous squamous cell carcinoma. , OSCC has traditionally affected elderly patients with a history of tobacco use, , , with a SEER database study of OSCC citing a median age of 61 years at diagnosis. However, in the past decade, a disturbing rise in the incidence of OSCC has been observed in a much younger population both in the Asia‐Pacific region as well as the northern hemisphere. , , These patients lack traditional lifestyle risk factors, including smoking and HPV infection. , ,

The prevalence of WGD has not been explored in patients developing OSCC at a younger age, with lower exposure to smoking associated carcinogens. , , The temporal relationship of WGD with TP53 mutation has also not been evaluated in OSCC. This may be due to the lack of samples from young patients with OSCC in international databases such as The Cancer Genome Atlas (TCGA) which predominantly include conventional smoking or HPV‐associated HNSCC .

The primary aim of this study is to evaluate the prevalence of WGD in patients developing HPV independent OSCC at a younger age using whole genome sequencing (WGS), along with validation through orthogonal testing using clinically validated in situ hybridization techniques. The temporal relationship of TP53 mutations to WGD has also been examined. The secondary aim is to evaluate the association of WGD with clinicopathologic prognostic factors and patient outcomes.

MATERIALS AND METHODS

Cohort and tissue selection

The median age at diagnosis of OSCC is 61 years based on SEER data. Schmidt et al. describe a 49.4 pack year smoking history in 67% patients with OSCC. , , Thus, for this study, we defined younger age with low exposure to conventional risk factors as patients with diagnosis of OSCC earlier than 50 years of age. Following institutional human research ethics committee approval (Sydney Local Health District Human Research Ethics Committee X19‐0282), patients treated with curative intent between 2012 and 2018 were identified from the prospectively collected Sydney Head and Neck Cancer Institute (SHNCI) biobank. Inclusion criteria were: (a) patient age under 50 years at the time of diagnosis of HPV independent OSCC; (b) anatomical location in the oral cavity (buccal mucosa, upper and lower alveolar ridges, retromolar trigone, floor of mouth, hard palate mucosa and anterior two‐thirds of the tongue); (c) availability of appropriate samples, including tumor with >30% cellularity on histologic evaluation and matched non‐ tumor tissue, and computational analyses demonstrating tumor purity ranging from 31–90%; (d) clinicopathological parameters, including clinical follow‐up; (e) DNA of sufficient quantity and quality.

Clinical details were obtained, including age, gender, smoking history, and details of adjuvant therapy. Data obtained regarding alcohol consumption was found to be unreliable and was therefore excluded. The histopathology material was reviewed for evaluation of adverse histopathological characteristics, including depth of invasion, perineural invasion, pattern of invasion lymphovascular involvement, nodal involvement (Figure S1) and American Joint Commission on Cancer Staging 8th edition staging categories.

Nucleic acid extraction

DNA was extracted from fresh frozen OSCC tissue (tumor) and oral cavity mucosa (matched normal). DNA extraction was performed using the Qiagen AllPrep extraction kit (Qiagen, Germantown, MD, USA), as per the manufacturer's instructions. Library preparations were processed using the TruSeq Nano preparation kit (Illumina, San Diego, CA, USA).

Quality control

The quality and quantity of DNA were assessed using Qubit V2.0 (Invitrogen, Carlsbad, CA, USA); spectrophotometric methods (Nanodrop, Thermo Scientific, Waltham, MA, USA; Epoch MicroPlate Spectrophotometer, BioTek Instruments, Winooski, VT, USA), Quant‐iT PicoGreen dsDNA assay (ThermoFisher, Waltham, MA, USA), and 0.8% agarose gel electrophoresis. DNA concentration of the SHNCI samples ranged between 35 and 190 ng/μl, with 350 to 1000 ng submitted for whole genome sequencing (WGS).

Nucleic acid sequencing

Tumoral and normal DNA were submitted for WGS at a target coverage of 60× and 30×, respectively. DNA sequencing was undertaken on the Illumina NovaSeq 6000 platform (NovaSeq Xp platform). The sequencing was paired‐end and each read was 150 base pairs in length.

Case selection from The Cancer Genome Atlas (TCGA)

The library of head and neck cancers (referred to as ‘HNSC’ in TCGA nomenclature) was accessed via the Genomic Data Commons (GDC) Portal (https://portal.gdc.cancer.gov/projects/TCGA-HNSC; dbGaP Study Accession #20551, accessed 7 October 2019). Cases below the age of 50 years with HPV independent OSCC were initially accessed from the library. The cases were filtered according to availability of WGS data of coverage at least 30X and the above inclusion criteria. Thus, nine cases from TCGA cohort were included with computed tumor purity of 38–67%. Clinical features including age, smoking status where available, and mortality status, were also obtained from the GDC portal.

Short read alignment and short variant calling

FASTQ files for both the SHNCI and TCGA cohort were processed using the same alignment (BWA‐kit version 0.7.17) and variant calling tools (Mutect2) to allow comparisons. BWA‐MEM read aligner was used to align reads to the hg38 reference genome and its alternate contigs. The read alignments were refined using Genome Analysis Tool Kit (GATK) 4.1.2.0 and its BaseRecalibrator tool. Single nucleotide polymorphisms (SNPs) and insertion–deletion (indel) variants were called by implementing GATK's Best Practices Workflow for “Germline short variant discovery (SNPs + Indels)” and “Somatic short variant discovery (somatic SNVs + Indels)”. These workflows use HaplotypeCaller to identify germline variants , and Mutect2 for somatic variants. Annotation of variants was performed via ENSEMBL Variant Effect Predictor version 99.2 (default settings were used, with the exception of pick where one effect per variant was selected using established criteria ). Detailed descriptions of pre‐processing and variant discovery are available on the Sydney Informatics Hub Github repository. , ,

Purity and ploidy

The purity and ploidy of each tumor sample's genome were inferred using the AMBER‐COBALT‐PURPLE pipeline stipulated in HMF Tools. , A summary file was created by PURPLE (PURity and Ploidy Estimator, version 2.4.1) for each tumor sample, in which WGD was described as ‘true’ or ‘false’. If the major allele ploidy surpassed 1.5 for at least half of the bases in at least 11 of 22 autosomes, a value of ‘true’ for WGD was returned. PURPLE determines that if a sufficient number of independent chromosomes is duplicated, the duplication event is likely to have occurred in a single genome‐wide event.

Where the copy number estimate was a non‐integer, the copy number was rounded to the closest integer. A deletion was defined as a copy number of 0 or 1. High‐level amplification was defined as a copy number of 6 or greater.

Mutation timing

PURPLE estimates every somatic SNV's purity‐adjusted variant and minor allele ploidies. If the purity‐adjusted ploidy of a variant (PURPLE_PLOIDY value) is more than 1.5, the variant is considered more likely to occur before WGD. Alternatively, if the purity‐adjusted minor allele ploidy (PURPLE_MAP value) surrounding a variant is less than 1.5 and the copy number of the surrounding region is greater than one, the variant most likely happened after WGD. For other values, the timing is indeterminate.

Fluorescent in situ hybridization (FISH)

Orthogonal testing with FISH was undertaken to confirm the ploidy status reported by the bioinformatics pipeline, using a tissue microarray (TMA) of the SHNCI OSCC cases. In an accredited diagnostic laboratory setting, interphase FISH for centromeric enumeration probes for chromosomes seven (Vysis EGFR/CEP7 Probe Kit, Abbott, Wiesbaden, Germany), 12 (ZytoLight SPEC MDM2/CEN12 Dual Colour Probe, ZytoVision, Bremerhaven, Germany), and 17 (PathVysion HER2 DNA Probe Kit, Abbott, Wiesbaden, Germany), was performed. Briefly, unstained 4 μm formalin‐fixed paraffin‐embedded tissue sections were deparaffinized, followed by heat‐induced epitope retrieval (HIER) pre‐treatment using a pH 7 buffered solution (SPoT‐Light Tissue Pretreatment Solution, Thermo Scientific, Waltham, MA, USA). Proteolytic digestion of tissue sections using Protease 1 (Abbott Molecular, Des Plaines, IL, USA) was then undertaken, after which a saline‐sodium citrate (SSC) buffer rinsing step was performed. Application of the aforementioned probes, with probe denaturation of the target chromosomes and gene for 5 min at 95°C and overnight hybridization at 37°C was completed. After hybridization, unbound probe was removed using 2xSSC/0.3% NP‐40 solution. The FISH slides were dehydrated, and counterstained with 4,6‐diamidino‐phenyldinol (SlowFade™ Gold DAPI, Invitrogen, Waltham, MA, USA). Interphase signals were enumerated in at least 50 tumor cell nuclei for the TMA using an epifluorescence microscope (Zeiss, Dublin, CA, USA).

Statistical analysis

Categorical data was analyzed using the Fisher's exact test and the Wilcoxon rank‐sum test for numerical variables. Survival was calculated using the Kaplan Meier method and comparisons were made using the log‐rank test. Overall survival was calculated from the date of surgery to the date of death. Disease‐specific survival and disease‐free survival were not analyzed due to lack of relevant datapoints (including recurrence and cause of death) in the TCGA dataset.

RESULTS

The study includes 37 patients, including 28 from the SHNCI and nine from TCGA. The median genome coverage for the SHNCI tumor samples was 64× (range 54–84×) and 35× (range 27–42×) for the matched normal samples. For the nine cases in the TCGA cohort, the median genome coverage was 75× (range 34–85×) for the tumor tissue and 40× for (range 30–49×) for normal tissue.

The clinicopathologic characteristics are summarized in Table 1. The SHNCI cohort (Table 1A) includes 15 males and 13 females with a median age of 41 years (range 19–50 years) at the time of diagnosis. None of the cohort had a prior history of cancer or previous radiotherapy. The median follow up was 16.5 months (3–175 months). Eight patients developed locoregional recurrence and six died from OSCC during this period.

TABLE 1

Clinicopathological data for (A) SHNCI cohort (n = 28), (B) TCGA cohort (n = 9)

Variable	n	%
Whole genomic duplication (WGD)
WGD b +	19	68
WGD b −	9	32
Age
≥40, ≤50	14	50
<40	14	50
Sex
Male	15	54
Female	13	46
Smoking
Yes	10	36
No	18	64
Anatomical subsite
Tongue	21	75
Floor of mouth	4	13
Buccal	1	4
Alveolar crest	1	4
Hard palate	1	4
Depth of invasion
<5 mm	4	14
5–10 mm	11	39
>10 mm	13	47
Perineural invasion
Yes	16	57
Lymphovascular invasion
Yes	13	46
Pathological T c category
pT1	4	14
pT2	12	43
pT3	7	25
pT4	5	18
Pathological N d category
pN0	15	46
pN1	2	7
pN2	8	29
pN3	3	11
Treatment f
Surgery alone	12	42
Surgery + radiotherapy	4	14
Surgery + radio/chemotherapy	9	32
Recurrence
Locoregional recurrence	8	29
Disease related death	7	25
Whole genomic duplication
WGD b +	8	89
WGD b −	1	11
Age
≥40, ≤50	3	33
<40	6	66
Sex
Male	7	78
Female	2	22
Smoking
Yes	4	44
History not available	5	56
Death g	5	56

Sydney Head and Neck Cancer Institute.

Whole genome duplication.

T = tumor.

N = lymph node involvement.

The Cancer Genome Atlas.

Adjuvant treatment records were unavailable for three patients.

Cause of death not specified in TCGA.

For the cases obtained from TCGA (Table 1B), there were seven men and two women with a median age of 34 years (range 19–49 years). History regarding clinicopathological factors and prior oncologic treatment if any were not available for this cohort. Data regarding recurrence was not available for TCGA cohort. Five of nine patients were deceased at 11.6 months.

WGD is common in OSCC patients <50 years

The FASTQ files of both the SHNCI and TCGA cohorts were processed together and the genomic profiles of all young patients, from both the SHNCI and TCGA cohorts, were examined together (Figure S2). WGD was seen in 27 patients (73%) (Figure 1A). The presence of WGD was further validated in the SHNCI cohort, for whom tissue was available for orthogonal FISH testing. FISH demonstrated concordant results with the computational analyses and a mean ploidy for chromosomes 7, 12, and 17 of 3.55, 2.78, and 2.65, respectively, was observed in those with WGD (Figure 1A–D). For patients without WGD, the mean ploidy for chromosomes 7, 12, and 17 was 1.8, 1.7, and 1.4 (Figure 1E, F, G, H).

FIGURE 1

CIRCOS plots (PURity and Ploidy Estimator, version 2.4.1) demonstrating (A) presence of whole genome duplication. The inner red ring of the CIRCOS plot shows chromosomal loss, while the adjacent green ring shows chromosomal gain. (B) to (D) FISH CEP of chromosomes 7, 12, and 17 showing increased ploidy; (E) absence of whole genome duplication, with the majority of autosomes showing no notable ploidy changes, confirmed by (F) to (H) FISH CEP of chromosomes 7, 12, and 17

Gender did not differ significantly between those with WGD (17 males, 10 females) and those without WGD (four males, six females) (p = 0.27). Smoking status also was not significantly different between those with WGD (eight smokers of 27 patients), and those without WGD (four smokers of 10 patients) (p = 0.45). Mutation signature analysis was performed (Figure S3). No evidence of Catalogue of Somatic Mutations in Cancer (COSMIC) smoking signatures (single base substitution signature [SBS] 4 and SBS 92) was seen in any of the cohort, regardless of WGD status.

The majority of WGD patients had pathogenic non‐synonymous mutation in TP53 in its DNA binding domain preceding the WGD

Somatic single nucleotide variants (SNVs) in TP53 were seen in 32 patients (84%) overall, including 17 missense, three frameshift, three splicesite, three nonsense variants, and six cases with multiple TP53 SNVs (Figures 2 and 3A). There were 27 patients with loss of heterozygosity (LOH) encompassing TP53, 19 of which showed biallelic somatic mutation of TP53. The remaining eight samples either showed copy number neutral LOH (CN‐LOH) with monoallelic TP53 variant, CN‐LOH with no somatic SNVs, or deletion overlapping the TP53 region (chromosome 17p whole arm or partial deletion). Of the 32 TP53‐mutant samples, 25 were located in the DNA binding domain of the p53 protein (Figure 2).

FIGURE 2

Lollipop plots demonstrating location of TP53 mutations in relation to p53 protein in those with WGD and without WGD

FIGURE 3

Oncoplot of patients with WGD and without WGD, demonstrating most common (A) somatic SNVs, and (B) CNVs. Figure 3A shows cases which lack TP53 variants but show TP53 LOH, due to deletion events overlapping the TP53 region

Of the 27 patients who had WGD, 25 also had a non‐synonymous mutation in TP53 (93%). Amongst the 25 patients with WGD and non‐synonymous TP53 SNV, 21 (84%) were inferred to have preceded WGD. One patient also had a second non‐synonymous mutation, predicted to have stop‐gain effect, and inferred to have occurred after WGD.

Amongst the ten patients without WGD, TP53 somatic SNVs were seen in seven patients (Figures 2 and 3A). Of these, three TP53 somatic variants occurred in the DNA binding domain, one of which was a frameshift variant, and the other two of which were missense variants also found in patients with WGD (Figure 2). While not reaching statistical significance, largely due to the small number of patients without WGD, TP53 somatic variants in the DNA binding domain were more frequently observed in patients with WGD (22 [82%]) vs. three (30%) non‐WGD patients, p = 0.07) Figure 2).

Most frequent genomic alterations observed in OSCC patients with and without WGD

Tumor mutation burden did not differ significantly between those with WGD (3.34 mutations/mb) and those without WGD (3.24 mutations/mb) (Figure S4). None of the patients showed microsatellite instability.

Figure 3A,B summarize the most frequent genomic changes observed in patients with and without WGD. Somatic SNVs in CSMD2, a gene that may act as a tumor suppressor in gastrointestinal cancers, were seen in four patients with WGD (15%), but were not identified in those without WGD (Figure 3A). Similarly, EGFR amplification was seen exclusively in patients with WGD (8/27, 30%), and not seen in patients without WGD (p = 0.07, Figure 3B).

WGD is significantly associated with presence of adverse clinicopathological features

Information regarding histopathological prognostic features was available for evaluation only for the 28 SHNCI patients. WGD was significantly more frequent in tumors with a larger tumor size (median size of 30 mm in WGD patients [range 12–95 mm]), versus median size of 15 mm in non‐WGD patients (range 5–40 mm), (p = 0.01, Figure 4A). There was a trend toward more advanced pT category (p = 0.08) and a higher rate (68%) of perineural invasion in patients with WGD (p = 0.11).

FIGURE 4

(A) Violin plot demonstrating association of WGD with tumor size; (B) Kaplan‐ Meier curve of overall survival in those with and without WGD

There was no association between WGD and lymphovascular invasion (p = 0.43), depth of invasion (p = 0.24, Figure S5), infiltrative pattern of invasion (p = 1.0), the presence of nodal metastases (p = 1.0), and the presence of extranodal extension (p = 1.0).

Survival is significantly poorer in patients with WGD

For those whom treatment and outcome data were available, the rate of post‐operative radiotherapy (p = 0.41) and chemotherapy (p = 1.0) was similar amongst those with and without WGD.

Eight patients developed locoregional recurrence. Of these, seven harbored WGD (87%). Overall survival was significantly worse in those with WGD (p = 0.05) (Figure 4B). The 24‐month overall survival of those with WGD was 67% compared to 100% for those without WGD.

DISCUSSION

Our study presents WGS data from one of the largest cohorts of OSCC in patients younger than 50 years with low cumulative burden of traditional risk factors, such as smoking. A high prevalence (73%) of WGD in patients younger than 50 years with OSCC is demonstrated. The computational analyses have been confirmed through orthogonal testing using clinically validated in situ hybridization techniques. In the vast majority (84%) of these patients, mutations in the gatekeeper TP53 DNA binding domain were found to precede the WGD event. Our data demonstrate that patients with WGD were more likely to adverse pathological features such as larger tumor size. Most notably, patients with WGD were more likely to recur and had poor overall survival.

Morris et al. previously reported WGD in 25% of recurrent/metastatic head and neck cancers using a cohort of HPV‐positive and ‐negative mucosal HNSCC (all sites), salivary gland cancers, and cutaneous carcinomas using targeted sequencing. The median age in their study was 59 years with a male preponderance (2.8:1). Of the 30 HPV‐negative OSCC cases older than 50 years, five cases (17%) had WGD. This rate of WGD reported in the study by Morris et al. for all metastatic and recurrent head and neck cancers irrespective of site and histologic type is significantly lower than that observed in the WGS data of a homogenous cohort of HPV independent OSCC patients younger than 50 years of age in our data. This study also found that WGD occurred more frequently in HPV‐negative versus primary HPV‐positive disease (odds ratio 4.8), than in HPV‐negative versus recurrent/metastatic HPV‐positive disease (odds ratio 0.71). (A relationship between TP53‐mutant status and WGD was not examined in this study.) The differences in the sequencing methods (targeted vs WGS) may account for some of this difference. However, other factors such as differences in age and cumulative risk factor exposure, such as smoking, HPV, and ultraviolet damage are more likely to account for the significant disparity. Indeed, Morris et al. report that WGD occurred more frequently in HPV‐negative versus primary HPV‐positive disease (odds ratio 4.8) in their cohort.

OSCC in those under the age of 50 is a rare cancer. However, an increasing incidence is being documented globally. , , Despite the rising incidence, the pathways to carcinogenesis have not been explored, and a consistent mutagenic risk factor for OSCC in young patients has not been identified. , While our cohort includes 14 (37%) smokers, they are younger than 50 years of age at diagnosis and unlikely to have the 49.4 pack year smoking history observed by Schmidt et al. in older patients with OSCC31.This is further strengthened by the absence of COSMIC mutational signature 4 associated with smoking in this cohort. While it is difficult to precisely delineate the relationship between carcinogenic risk factors and WGD, WGD has been described as an early event, preceding subclonal proliferation in tumors that lack a mutagenic risk factor such as germ cell tumors and lung adenocarcinoma in non‐smokers. , In contrast, WGD has been described as a late event in malignancies with known mutagenic risk factors or driver point mutations. , , For example, WGD was seen to occur after the accumulation of a number of ultraviolet light exposure‐related mutations including BRAF and KRAS alterations, and prior to the occurrence of copy number alterations in melanoma, a UV damage driven malignancy. Similarly, WGD is also a late event in myeloma, where point mutations often occur early. Interestingly, nearly all of the patients ith germ cell tumors and lung adenocarcinoma in non‐smokers tend to be younger at the time of diagnosis , ; a demographic feature shared with patients in this cohort. Thus, WGD may play an important role in OSCC carcinogenesis in young patients with limited cumulative exposure to risk factors or in those where a causative agent has not been identified.

The majority of patients with WGD in our study demonstrated somatic non‐synonymous TP53 alterations. This is in keeping with previous pan‐cancer analyses of WGD, where TP53‐mutant tumors are twice as likely to undergo WGD as compared with TP53‐wildtype tumors. , Indeed, inactivation of the p53 protein has been shown to increase tolerance of WGD. Most TP53 mutations in our cohort occurred in the DNA binding domain of the p53 protein, a region essential for recognizing transcription errors and maintaining genome integrity during cell cycle. Most somatic TP53 mutations were inferred to have occurred before the WGD event. TP53 mutations have been shown to precede WGD, which would be consistent with the cell cycle regulator function of the intact p53 protein, preventing a tetraploid cancer cell from re‐entering the cell cycle. , Interestingly, not all samples with WGD had TP53 LOH in our cohort. Thus, there is likely to be another cause of WGD yet to be discovered. Zack et al. describe amplification of CCNE1, or cyclin E1, a gene that encodes for a cyclin protein involved in cell cycle transition, in all patients with wild‐type p53. Interestingly, CCNE1 amplification was not observed in this cohort which shows a predominance of early TP53 mutations.

WGD has been shown to be associated with poor prognosis in a diverse range of cancers. , , In a pan‐cancer study of advanced cancer patients, WGD was an independent predictor of adverse outcome after adjusting for well‐established adverse clinicopathologic features and genetic alterations. For example, WGD was found to be an independent predictor of death in KRAS‐mutant colorectal cancer and estrogen receptor positive breast cancer. Correlation of WGD with survival has not been previously performed in OSCC. Similar to Bielski and colleagues in other cancer types, our data demonstrate for the first time in OSCC, that overall survival is significantly reduced in OSCC patients with WGD.

Association of WGD and adverse histopathological features, representing in vivo characteristics of the tumor have not been previously examined in OSCC, or in other cancer types. The patients in our cohort with WGD showed a significantly larger tumor size, with a trend toward higher pT category at clinical presentation, and the presence of perineural invasion. Wangsa et al. describe similar findings in their study of the effect of WGD and polyploidy in single‐cell derived tetraploid clones, and demonstrate increased invasive and migratory ability of cancer cells ex vivo . Tumor recurrence was also frequent in those OSCC patients with WGD. Thus, our data indicate that WGD is an adverse prognostic indicator in OSCC. If verified, this information could be harnessed to guide prognostic stratification of OSCC in light of the increasing availability, accuracy, and decreasing cost of next‐generation sequencing.

One of the main limitations of the study is its small cohort size. HPV‐independent OSCC is a rare cancer in young people and the clinicopathologic data of the SHNCI cohort has been carefully curated and the DNA selected for sequencing has undergone stringent quality controls. Of note, search of publicly available databases such as TCGA yielded only 11 cases of HPV‐independent OSCC in patients younger than 50 years. Of these, nine had sufficient read depth and tumor purity that could be utilized for further analyses with very limited clinicopathologic and follow up data. In addition, another limitation of using TCGA data was the lack of detailed carcinogenic exposure data, particularly with regards to smoking history (Table 1B). The need for well‐annotated clinical data in cancer genomic datasets, as obtained for the SHNCI cohort, is highlighted by our study. The paucity of WGS data for HPV‐independent OSCC in young patients in one of the largest genomic consortia highlights the unique nature of the current study.

The current study is the first to demonstrate the high prevalence of WGD among young patients who develop HPV‐independent OSCC, in the absence of a conventionally recognized mutagen. The majority of cases with WGD demonstrated TP53 mutations and these predominantly occurred prior to WGD. Of clinical relevance, the presence of WGD may represent a potential prognostic biomarker of adverse outcome in young HPV‐independent OSCC patients. The frequency of adverse prognostic histopathological features and poor outcome in OSCC patients with WGD highlight the need for targeted therapies. Novel agents targeting the mitotic spindle‐assembly checkpoints in WGD‐driven cancers are under development, and may represent a new therapeutic horizon in the treatment of OSCC13.

Figure S1

Click here for additional data file.

Figure S2

Click here for additional data file.

Figure S3

Click here for additional data file.

Figure S4

Click here for additional data file.

Figure S5

Click here for additional data file.