Detection of High-Risk Human Papillomavirus in Oral Cavity Squamous Cell Carcinoma Using Multiple Analytes and Their Role in Patient Survival.

PURPOSE: Accurate detection of human papillomavirus (HPV) in oral cavity squamous cell carcinoma (OSCC) is essential to understanding the role of HPV in disease prognosis and management of patients. We used different analytes and methods to understand the true prevalence of HPV in a cohort of patients with OSCC with different molecular backgrounds, and we correlated HPV data with patient survival.
METHODS: We integrated data from multiple analytes (HPV DNA, HPV RNA, and p16), assays (immunohistochemistry, polymerase chain reaction [PCR], quantitative PCR [qPCR], and digital PCR), and molecular changes (somatic mutations and DNA methylation) from 153 patients with OSCC to correlate p16 expression, HPV DNA, and HPV RNA with HPV incidence and patient survival.
RESULTS: High prevalence (33% to 58%) of HPV16/18 DNA did not correlate with the presence of transcriptionally active viral genomes (15%) in tumors. Eighteen percent of the tumors were p16 positive and only 6% were both HPV DNA and HPV RNA positive. Most tumors with relatively high copy number HPV DNA and/or HPV RNA, but not with HPV DNA alone (irrespective of copy number), were wild-type for TP53 and CASP8 genes. In our study, p16 protein, HPV DNA, and HPV RNA, either alone or in combination, did not correlate with patient survival. Nine HPV-associated genes stratified the virus-positive from the virus-negative tumor group with high confidence ( P < .008) when HPV DNA copy number and/or HPV RNA were considered to define HPV positivity, and not HPV DNA alone, irrespective of copy number ( P < .2).
CONCLUSION: In OSCC, the presence of both HPV RNA and p16 is rare. HPV DNA alone is not an accurate measure of HPV positivity and therefore may not be informative. HPV DNA, HPV RNA, and p16 do not correlate with patients' outcome.

INTRODUCTION

Head and neck squamous cell carcinoma (HNSCC) is the sixth most common cancer worldwide with an incidence of 550,000 cases annually.[1,2] Oral cavity squamous cell carcinoma (OSCC) constitutes a majority of HNSCCs, including tumors of the oral anterior tongue and buccal mucosa.[3] The major known risk factors for OSCC are use of tobacco and alcohol and infection with human papillomavirus (HPV).[4,5] Unlike oropharyngeal tumors, in which HPV incidence is reported to be high (up to 90%),[6,7] the prevalence of HPV in OSCC (although it varies greatly among geographies and choice of analyte and assay[8]) is generally accepted to be low.[9,10] In addition, unlike with oropharyngeal tumors,[11-15] the role of HPV in disease prognosis and response to therapy in patients with OSCC is equivocal. Despite the fact that HPV RNA is shown to function as a better screening and patient management tool,[16,17] the presence of HPV DNA is routinely used as a measure of HPV infection in tumors. HPV DNA results do not always match those for HPV RNA, especially in OSCC.

HPV16 and HPV18 subtypes have been epidemiologically linked with head and neck carcinoma.[18] High-risk HPV16 and HPV18 are the most predominant subtypes in oral cavity tumors from Indian patients, whereas the other subtypes (HPV33, HPV6, and HPV11) are rare.[19,20] HPV E6 interacts with p53 to promote its degradation via the ubiquitin pathway, whereas HPV E7 forms a complex with retinoblastoma (Rb) protein leading to its functional inactivation and dysregulation of the cell cycle.[21,22] In some HPV-related tumors, E6- and E7-mediated inactivation of p53 and Rb result in the accumulation of p16 protein,[23] whereas in others, p16 expression does not directly correlate with HPV positivity.[24] A majority of HPV-negative tumors harbor mutations in TP53 and CASP8, and a significant proportion of HPV-positive tumors harbor mutations in PIK3CA.[25-27] In addition, past studies have identified specific mutations in potential drug targets such as FGFR2/3, lack of EGFR aberrations in HPV-positive patients,[28] and a potential role of CASP8 in HPV-negative cell lines and patients.[26,29] Despite a wealth of information, questions regarding the accuracy of different HPV tests and whether HPV is an important factor in the stratification and treatment of oral cavity tumors remain to be answered.

In this study, we addressed the following five questions related to HPV in oral cavity tumors. (1) Does sensitivity of the test matter in the detection of HPV DNA? (2) Does the presence of p16 protein and HPV DNA correlate with HPV E6/E7 RNA? (3) Does the presence of high copy number HPV DNA accurately reflect HPV positivity? (4) Are p16 protein, HPV DNA, and HPV E6/E7 RNA individually or together linked with patient survival? (5) Do somatic mutations and DNA methylation at 5-cytosine residues distinguish the HPV-positive from the HPV-negative tumors?

METHODS

Patients, Cell Culture, and Nucleic Acid–Based Assays

Tumor samples (n = 153) from patients with OSCC (buccal mucosa, bone marrow [including from upper and lower gingivobuccal sulcus and retromolar trigone], and oral tongue) were accumulated consecutively and selected for the assay (Table 1; Data Supplement). For nucleic acid–based assays, we tested five sets of primers published in the literature and two that were newly designed in the amplification reactions (Fig 1; Appendix Table A1). Details of the patients and methodology are provided in the Data Supplement.

Table 1

Characteristics of Patients and Tumor Tissues Used in the Study

Fig 1

Human papillomavirus (HPV) genome organization and locations of different primers (either consensus or type specific) or probes used in the study to detect HPV DNA and RNA. The numbers before and after the arrows represent the corresponding nucleotide number in the HPV genome. ddPCR, droplet digital polymerase chain reaction; qPCR, quantitative PCR; RT-PCR, real-time PCR.

Table A1

Primer and Probe Sequences Used in the Study With Amplicon Size and Conditions for Amplification Reactions

Immunohistochemistry

For p16 immunohistochemistry (IHC), staining was carried out by using formalin-fixed paraffin-embedded tissue blocks and primary antibody from BioGenex (Fremont, CA; catalog No. AM540-5M; Antip16[NK4], Clone G175-405 in the NordiQC list) and using the PolyHRP detection system (catalog No. QD400-60KE, BioGenex) according to the manufacturer’s instructions and a scoring method (Data Supplement). Sections of cervical cancer were used as a positive control.

HPV DNA Copy Number

We deduced the HPV absolute copy number from the quantitative polymerase chain reaction (qPCR) standard curves using cloned HPV16/18. We considered a tumor or cell line to have a relatively high copy number of HPV DNA when the copy number for HPV16 was more than 3.3 × 102 per μg of tumor DNA and that for HPV18 DNA was more than 3.3 × 103 per μg of tumor DNA. To minimize the effect of tumor cellularity, ploidy, and heterogeneity, we expressed the HPV copy number as copies per μg of tumor DNA used in the reaction.

Mutation and Survival Analysis

The mutation data on tumors for TP53, CASP8, and RASA1 were retrieved from previously published data.[29] The χ2 test was used to determine the significance of different clinical parameters of patients. The relationship between tumor HPV status and survival in patients was examined by Kaplan-Meier analysis (Data Supplement). Overall survival (OS) and disease-free survival (DFS) were analyzed, and a log-rank test was used to determine significance (P < .05).

Whole-Genome Methylation and Statistical Data Analyses

Whole-genome methylation data were gathered by using the Illumina Infinium Methylation450 BeadChip kit, chip scanning, and data preprocessing; the process was described previously.[30] Statistical methods used to analyze methylation data are provided in the Data Supplement.

RESULTS

p16 Expression and HPV DNA

In our study, 18% of the tumors were p16 positive (Fig 2A; Table 2). We detected HPV DNA at 0.03 ng or with a larger amount of genomic DNA (Appendix Fig A1) from the cell line UMSCC-47/Hep2 when the following primers were used: GP5+6+, MY09/11, CPI-II, PGMY09/11, or HPV16L1 (Fig 2B). However, the newly designed type-specific primers (HPV16E6 and HPV18L1) could detect HPV16 and HPV18 with as little as 0.0003 ng and 0.003 ng of genomic DNA, respectively (Fig 2B). We also tested the effect of cloned HPV DNA amount on amplification efficiency (Appendix Fig A2). Figure 2C shows the efficiency of the consensus and type-specific primers in a set of representative oral cavity tumors (Appendix Fig A3). Widely used primers from the literature (MY09/11, PGMY09/11, GP5+6+, and HPV16L1) yielded either the least sensitivity or moderate (CPI-II) sensitivity of detection, whereas the newly designed HPV16E6 and HPV18L1 primers showed the optimum sensitivity of detection (Fig 2B). We observed inhibition of the amplification reactions at a high concentration of tumor genomic DNA with a positive cell line spike-in experiment (Appendix Fig A4) and therefore higher concentrations of tumor DNA were avoided in the reactions. In addition, an increase in amplification cycles did not aid in the detection of HPV DNA in PCRs as shown in Appendix Figure A5. Results from qPCRs indicated that 33% of tumors (35 of 106) were positive for HPV DNA (Table 2). Although we found a higher incidence of HPV16 (30% [32 of 106]) than HPV18 (18% [19 of 106]) type, the HPV18-positive tumors had high copy numbers of viral DNA as reflected in their cycle threshold (Ct) values (Fig 3A iii,vi). Quantitative PCR (qPCR) was performed on oral cavity tumors (n = 106), and the tumors were counted as HPV DNA positive if they had Ct values three times the standard deviation for the mean of negative controls (Fig 3A ii,v). Digital PCR has recently been shown to successfully detect HPV DNA in oropharyngeal tumors in a highly specific manner.[31] Digital PCR results indicated that 43% of oral cavity tumors (59 of 136) were positive for HPV16 DNA (Fig 3B iii; Table 2; Appendix Fig A6).

Fig 2

p16 and human papillomavirus (HPV) DNA in oral cavity squamous cell carcinoma (OSCC). (A) Representative images of immunohistochemical staining of p16 in positive and negative OSCC tissue sections; cervical tissue was used as a positive control. (B) Relative polymerase chain reaction (PCR) amplification efficiency and sensitivity of consensus and type-specific primers for detection of HPV using HPV16 or HPV18 individual positive control cell lines (UMSCC-47 and Hep2). (C) Representative HPV DNA PCR in oral cavity tumors with both consensus and type-specific primers. OT, oral tongue.

Table 2

Summary of HPV Assays for Oral Cavity Tumors

Fig A1

Increasing amount of genomic DNA from cell lines used as positive or negative controls for human papillomavirus16 (HPV16) and HPV18 quantitative polymerase chain reaction (qPCR). HPV16 qPCR using negative control (A) Hep2 and (B) UPCI:SCC029B. HPV18 qPCR using negative control (C) UMSCC-47 and (D) UPCI:SCC029B. Error bars are drawn using data from three independent experiments. Ct, cycle threshold.

Fig A2

Amplification efficiency of human papillomavirus16 E6 (HPV16E6) and HPV18L1 primers measured by polymerase chain reaction amplification of serially diluted HPV16/18 cloned plasmid copies. M, marker; N, negative control.

Fig A3

Human papillomavirus (HPV) polymerase chain reaction (PCR) performed by using different sets of consensus or type-specific primers with oral cavity squamous cell carcinoma tumor DNA. PCR for oral cavity tumors with (A) MY09/11, (B) HPV16E6, (C) HPV18L1 using cell lines as positive and negative controls, (D) HPV16E6 for batch 2 (15 tumors), (E) HPV CPI-II using cell lines as positive controls, and (F) HPV16L1 using cell lines for positive controls. BM, buccal mucosa; L, DNA ladder; N1, UPCI:SCC029B DNA (300 ng); N2, no template control (NTC); OT, oral tongue; P1, positive control cervical DNA sample 1; P2, positive control cervical DNA sample 2; P3, UMSCC-47 DNA (HPV16-positive cell line).

Fig A4

Inhibition of amplification reactions for detecting human papillomavirus (HPV) in polymerase chain reactions at high concentrations of tumor genomic DNA (used with PGMY09-11 primer) spiked with HPV-positive UMSCC-47 DNA.

Fig A5

The effect of amplification cycles on polymerase chain reactions. The genomic DNAs used for positive control (UMSCC-47) and negative control cell lines were 63.0 ng and 300 ng, respectively. dNTP, deoxynucleotide triphosphates; NTC, no template control.

Fig 3

Detection of HPV DNA and RNA. (A i-vi) HPV16/18 assays using quantitative polymerase chain reaction (qPCR) and (Bi-iii) droplet digital PCR (DDPCR) in OSCC. (A i,iv, B i) Standard curves were obtained by using cloned HPV16/18 plasmids. (A ii,v, B ii) Data were subsequently obtained by using the positive (UMSCC-47 and Hep2) and negative (UPCI:SCC29B and UPCI:SCC40) cell line DNA to count HPV DNA in (A iii,vi,Biii) oral cavity tumors. (C) HPV16 (top panel) and HPV18 (bottom panel) E6 and E7 mRNA expression in tumors using qPCR. Horizontal dotted lines: threshold lines for negative samples. BM, buccal mucosa; Ct, cycle threshold; OT, oral tongue.

Fig A6

Positive and negative cell line DNA used for threshold in droplet digital polymerase chain reaction experiment. OTSCC, oral tongue squamous cell carcinoma.

HPV RNA

Compared with the cell lines, tumors showed low levels of expression of E6 or E7 messenger RNA (mRNA; Fig 3C). Only 15% of the tumors showed expression of E6 RNA and/or E7 RNA (unlike HPV DNA), and 6% of the tumors had both HPV DNA (in all three assays) and transcriptionally active HPV genomes (Table 2). In our cohort, younger patients (age 40 years or younger) had significantly more HPV RNA positivity than older patients when χ2 analysis was used (P = .029).

When the results from all of the assays (p16 IHC, HPV DNA, and HPV RNA) were combined, we found that 6% to 48% of the tumors were positive in various assays combined with PCR (Table 2; Appendix Table A2). We found that 22% of the tumors (23 of 106) had relatively high copy numbers of HPV DNA and/or HPV E6 or E7 mRNA.

Table A2

Summary of Tumor HPV Status in Individual Tumors Used in This Study

Linking Tumor Attributes, Somatic Mutations, and HPV With Survival

We performed Kaplan-Meier survival analyses with various tumor attributes that revealed significant association between tumor differentiation (P = .03) and clinical stage (P < .001) with OS (Fig 4A). None of the other tumor attributes showed significant association with survival (Appendix Fig A7A-F). In patients with oral cavity tumors, p16, HPV DNA, and HPV RNA did not correlate with the either OS or DFS (Fig 4A; Appendix Fig A8). HPV DNA status alone measured by any of the DNA-based assays alone or in combination did not correlate with survival (Fig 4A; Appendix Fig A8), except when measured with droplet digital PCR (ddPCR) for OS (P = .03; Appendix Fig A8E). We tested whether tumors with relatively high HPV DNA copy numbers and/or HPV E6 or E7 mRNA were linked with survival. As shown in Appendix Figure A9A-B, we did not find any significant association with this group of tumors for either OS (P = .45) or DFS (P = .68).

Fig 4

(A) Kaplan-Meier survival plots linking tumors with various attributes such as grade, stage, p16 immunohistochemistry (IHC), HPV DNA, and HPV RNA. (B) Clustering of nine methylated genes stratifying the HPV-positive from the HPV-negative group of tumors along with the HPV-associated pathways in HPV-positive tumors. Cyto, cytoplasmic (staining); MDSCC; moderately-differentiated squamous cell carcinoma; Nuc, nuclear staining; PDSCC, poorly-differentiated squamous cell carcinoma; WDSCC, well-differentiated squamous cell carcinoma.

Fig A7

Kaplan-Meier survival analysis with tumors from patients according to habits, age, and nodal status. Overall survival (OS) percentages for patients who were (A) positive v negative for any habit, (B) positive for tobacco chewing v no habit, (C) positive for alcohol consumption and tobacco chewing v no habit, (D) positive for alcohol consumption and smoking v no habit, (E) age older than 40 v age younger than 40 years, and (F) their nodal status.

Fig A8

Kaplan-Meier survival analysis with (A-H) human papillomavirus (HPV) DNA and (I-J) HPV RNA. (A) Overall survival (OS) with DNA polymerase chain reaction (PCR). (B) Disease-free survival (DFS) with DNA PCR. (C) OS with DNA qPCR. (D) DFS with DNA qPCR. (E) OS with DNA ddPCR. (F) DFS with DNA ddPCR. (G) OS with HPV positive in PCR+qPCR+ddPCR vs HPV negative in PCR+qPCR+ddPCR. (H) DFS with HPV positive in PCR+qPCR+ddPCR vs HPV negative in PCR+qPCR+ddPCR. (I) DFS with HPV RNA, and (J) DFS with p16 IHC.

Fig A9

Kaplan-Meier survival analysis of tumors with (A-B) high copy number human papillomavirus (HPV) DNA and/or HPV RNA and (C-D) HPV-negative tumors with mutations in significant genes. DFS, disease-free survival; Mut, mutation; OS, overall survival; WT, wild-type.

We also investigated whether somatic mutations in significantly mutated genes in OSCC play a role in survival in patients with HPV DNA–positive tumors. We analyzed three genes (TP53, CASP8, and RASA1) shown to be significantly mutated in oral cavity tumors.[26,29,32] Ninety-five percent of the HPV-positive tumors in the group were wild-type for TP53 and CASP1 genes, and 85% of the HPV-positive tumors were wild-type for RASA1 gene (Appendix Fig A10). We tested whether the mutations in any of the genes, alone or in combination in the HPV-negative tumor group, were linked with survival. We did not find any significant association for this group of tumors with survival (Appendix Fig A9C-D).

Fig A10

Mutational frequency in tumors with mutations in three commonly mutated (Mut) genes. WT, wild-type.

Linking Methylation With HPV

Supervised clustering of the first group of patients (a group defined as having high copy number HPV DNA and/or E6 or E7 RNA) resulted in a list of 60 genes of which nine (FERMT3, GIT2, HK3, PRKCZ, ZCCHC8, IRF5, IFFO1, ARID3A, HOXA2) were mapped to the HPV pathway (Fig 4B). Methylation of those genes is involved in the downstream control of the expression of different target genes. For example, ZCCHC8 methylation is linked with the expression of RB1, PRKCZ methylation controls state change of DLG1, methylation in ARID3A, IRF5, IFFO1, and HOXA2 are connected with the expression of TP53, and FERMT3, HK3, and GIT2 genes control the expression of AP1 (JUN) (Fig 4B). All of the genes except HOXA2 were significantly hypermethylated in the HPV-positive group of tumors compared with the HPV-negative group (Fig 4B). The four linked genes obtained from the nine significantly methylated genes were mapped to the pathways involving HPV E6 and E7 proteins (Fig 4B). To test significance, we performed unpaired t tests between the two groups: group 1 had relatively high copy numbers of HPV DNA and/or HPV RNA, and group 2 was negative for both HPV DNA and HPV RNA. All of the eight hypermethylated genes and one hypomethylated gene (HOXA2) showed high significance (P < .001 and P < .007, respectively; Appendix Table A3). However, when the patients were grouped on the basis of HPV DNA positivity alone (irrespective of copy number), most of these nine HPV-linked genes did not show a significant association.

Table A3

P Values From Unpaired t Tests Measuring Significance in Differences Between Differential Methylation in Nine HPV-Associated Genes Between HPV-Positive and HPV-Negative Groups

DISCUSSION

HPV plays a vital role in the prognosis of patients with oropharyngeal tumors.[33,34] Unlike with disease in the oropharynx, the incidence of HPV and its role in disease prognosis in oral cavity tumors are not well established. Past results regarding HPV DNA incidence in oral cavity tumors varied widely (from low to high; Appendix Table A4) depending on the assay sensitivity, analyte, and patient cohort were used.[34,35] Questions regarding the accuracy of the HPV tests and HPV positivity need to be answered to make confident treatment decisions for treating patients with head and neck tumors.[36] There are only a few studies that used multiple analytes (protein, DNA, and RNA) and various molecular tests (IHC, PCR, qPCR, and digital PCR) to establish HPV positivity in oral cavity tumors and that correlated HPV with tumor attributes (including somatic mutations and methylation) and survival. In this study, we attempted to assess correlations between HPV DNA, RNA, and p16 protein and survival in 153 patients with oral cavity tumors.

Table A4

Literature Survey of HPV Studies in Oral Cavity Tumors

Although p16 expression (as measured by IHC) is a commonly used proxy for HPV in HNSCC, its expression is not specific in HPV-associated tumors.[34] Several past studies have correlated p16 expression with HPV,[37-39] but p16 IHC has shortcomings, especially when relating the expression of p16 to patient survival. Limitations, such as variations in staining intensities,[14,40] nonspecific binding of antibodies, and the lack of scoring and interpretive criteria for p16 staining make the method less reliable.[41] Associating p16 status with survival of patients with OSCC has been inconclusive, and some previous reports[42] have suggested additional study to derive any conclusive evidence in this regard. In our study, although we found an unusually high percentage (51%) of tumor cells that showed immunopositive staining, only a small percentage (18%) had both cytoplasmic and nuclear staining, an accurate reflection of HPV positivity as described earlier.[43-45] Unlike the authors of some of the previous studies,[38] we could not find any correlation (either positive or negative) between p16 expression and survival (Fig 4). A weak correlation between p16 IHC and HPV in situ hybridization was reported earlier.[37] As in previous reports,[46] we found that p16 expression was not a useful surrogate marker for HPV in oral cavity tumors.

Unlike antibody-based methods, nucleic acid–based methods detect HPV with high sensitivity and are therefore widely used.[47] Meta-analysis of 5,478 oral cavity tumors suggested that overall prevalence of HPV DNA was 24.2% with 11% of the tumors being positive for both HPV DNA and E6 or E7 RNA.[48] India has one of the highest incidence rates of oral cavity cancers, and there is a significant difference in the incidence trend between oropharyngeal and oral cavity cancer.[49] Previously, PCR coupled with mass array was shown to provide highly sensitive detection with a small amount of genomic DNA input.[50] Our results showed that 38% of tumors were positive and 13% were negative in all three DNA-based assays (PCR, qPCR, and ddPCR). Overall, the prevalence of HPV DNA (33% to 58%) was dependent on the type of test used; PCR yielded the highest incidence over the more sensitive methods such as qPCR and ddPCR assays (Table 2). This was possibly due to the result of consensus primers used in PCR (but not in qPCR and ddPCR) in addition to the type-specific primers that resulted in the detection of non-HPV16/18 subtypes. As expected, digital PCR, which was the most sensitive of the three DNA-based assays, showed more tumors being HPV16 DNA positive, which resulted in the detection of low copy number viral genomes in tumor samples. On the basis of several levels of evidence, we conclude that the presence of low copy numbers of HPV DNA alone may not be a reflection of functionally active HPV. First, we found that only a fraction of the tumors (15%) had HPV E6 or E7 RNA. Second, only 6% of the tumors were positive for the presence of both the HPV genome and E6 or E7 RNA. Third, almost all of the tumors with relatively high copy numbers of the HPV genome and/or HPV RNA had wild-type TP53 and CASP8 genes, which was not the case for tumors with low copy numbers of HPV DNA. Both TP53 and CASP8 are known to be wild-type primarily in HPV-positive tumors. In our study, we found that this corresponds to tumors with high copy numbers of the HPV genome and/or a transcriptionally active genome only (Table 2). High prevalence of HPV DNA, as demonstrated in some assays, might suggest the presence of passenger HPV genomes coming from adjacent normal cells (as reported earlier[51-53]), or it could be a reflection of inactive or passenger viruses in oral cavity tumors. Although the numbers are low (n = 3), we cannot explain why some tumors in our study with HPV E6 or E7 RNA did not show the presence of HPV DNA. It is possible that the genomic DNA for those tumors was degraded and therefore could not serve as an ideal template for DNA-based assays. An additional factor that might have added to this is the presence of inhibitors for DNA-based assays in those tumors.

The fact that there were only two tumors that were p16 positive and HPV RNA negative means that a definitive conclusion on the lack of correlation between p16 and HPV RNA cannot be made from our study. Similarly, there were two tumors that were positive for HPV RNA and negative for p16. In HNSCC, p16 is often mutated or silenced, which results in its loss of expression. This could have led to the lack of p16 expression in those two tumors. We did not find any significant correlation between p16, HPV-DNA, and/or HPV RNA and disease outcome (Fig 4A; Appendix Fig A8). Even the tumors with relatively high copy number of HPV genomes and/or E6 or E7 RNA did not support the role of HPV in patient survival (Appendix Fig A9A-B). Our study has highlighted that understanding HPV prevalence in OSCC is complicated. In fact, in oropharyngeal tumors in which p16 has been a definite prognostic marker, a recent study[54] recommended additional HPV DNA testing to accurately predict prognosis.[55] These aspects need further study and analysis.

Although more research is needed to determine how HPV gets to the mouth cavity, it is believed that oral sex and/or bad oral hygiene are two responsible factors. However, a causal role between bad oral hygiene and HPV infection is unclear.[56] In addition, recent data show the role of the oral microbiome in HPV-positive and HPV-negative oral tumors.[57] Future studies linking oral sex and bad oral hygiene with HPV in the mouth cavity among patients belonging to different sociogeographic strata might shed additional light on this problem. Although our study is comprehensive, it has several limitations. Not all the tumors were assayed with all of the analytes, which makes the sample number different for different methods. We could not perform additional survival analyses for the tumors that were HPV RNA positive, given the small sample size. In our study, we did not perform in situ hybridization, which could have provided additional information on p16 positivity and HPV prevalence. It is possible that the presence of high copy numbers of the HPV genome in the tumors studied does not correlate with the presence of the biologically active virus. Further studies may help answer this question.