Human Papillomavirus–Attributable Head and Neck Cancers in India—A Systematic Review and Meta-Analysis

Abstract

PURPOSE

Head and neck cancer accounts for about one third of the global burden in India. Mucosal high-risk human papillomavirus (HPV) has been hypothesized as a contributory risk factor for head and neck cancer (HNC) but its prevalence in Indian patients is not well established. Therefore, this systematic review and meta-analysis aimed to estimate the prevalence of HPV in HNC in India and their attributable fraction by considering the biomarkers of carcinogenesis, p16, and HPV E6/E7 mRNA.

METHODS

A systematic literature search was done in Medline via PubMed, Embase, Scopus, ScienceDirect, ProQuest, and Cochrane to identify studies on HPV and HNC in the Indian population, published between January 1990 and October 2022. Fifty-four eligible studies were identified and relevant clinical information was collected. Meta-analysis was conducted to estimate the pooled prevalence of HPV DNA, p16INK4a, and E6/E7 mRNA percent positivity by random-effect logistic regression model using Metapreg, STATA 18.

RESULTS

Thirty-four high-quality studies were taken for meta-analysis. The pooled prevalence of HPV in HNC was 20% (95% CI, 12 to 32) with a high level of heterogeneity (I² = 90.79%). The proportion of HPV in oropharyngeal cancer (OPC; 22% [95% CI, 13 to 34]) and laryngeal cancer (LC; 29% [95% CI, 17 to 46]) was higher than in oral cancer (OC; 16% [95% CI, 8 to 30]). The HPV-attributable fraction of OPC, considering the E6/E7 mRNA and p16 positivity, was 12.54% and 9.68%, respectively, almost similar to LC (11.6% and 9.57%), while it was much lower in OC (3.36% and 4%).

CONCLUSION

The HPV-attributable fraction is considerably lower for OC, suggesting a negligible causative role of HPV in OC. A significant proportion of OPC and LC are attributed to HPV; however, their exact causative role is unclear because of the presence of other known risk factors.

Meta-analysis reveals a causative role of HPV in a proportion of oropharyngeal and laryngeal cancers in India.

INTRODUCTION

Head and neck cancer (HNC) refers to squamous cell carcinoma arising from oral—lip (C00.0-9), anterior two thirds of the tongue (C02.0-3,9), gum (C03.0-9), floor of the mouth (C04.0-9), hard palate (C05.0), and buccal mucosa and retromolar area (C06.0-9); oropharyngeal—base of the tongue (C01), soft palate (C05.1), tonsil (C02.4, C09.0-9), uvula (C05.2), and other parts of the oropharynx (C10.0-9); laryngeal—glottis (C32.0), the supraglottis (C32.1), subglottis (C32.2), and other parts of the larynx (C32.3-9); and hypopharyngeal—hypopharynx (C13) and pyriform sinus (C12). The global incidence of HNC is about 660,000 cases, with a mortality of 325,000 patients annually.¹ India has the highest HNC incidence (6.1 per 100,000 women and 20.9 per 100,000 men), accounting for about one third of the global burden, primarily because of the habit of tobacco chewing.² However, the recent increase in HNC incidence, particularly oropharyngeal cancers (OPCs) in India, is associated with mucosal high-risk human papillomavirus (HR-HPV), which could be a contributory risk factor.³ The prevalence of HPV-associated HNC varies substantially between countries. A meta-analysis estimated the proportion of HPV-positive OPC to be 45.8%, 24.2% for oral cancer (OC), and 22.1% for laryngeal cancer (LC), on the basis of HPV DNA detection by polymerase chain reaction (PCR).⁴ However, studies from India reported a considerable variation in the prevalence of HPV-related HNC, ranging from 7% to 78.7%, probably because of heterogeneity in these studies.⁵ This variation could be attributed to demographic factors, sample type, anatomic site, and viral detection methods. Most studies have determined the presence of HR-HPV DNA without evaluating the biomarkers of HPV carcinogenesis (ie, E6/E7 mRNA and p16 protein), thus failing to distinguish between transient and persistent HPV infection.⁶ HPV E6/E7 mRNA detection is a gold-standard test for transcriptionally active HPV infection.⁷ The modified eighth American Joint Committee on Cancer/Union for International Cancer Control (AJCC/UICC) classification recommends p16 as an HPV surrogate marker in OPC as it correlates with an improved survival rate.⁸ Nevertheless, the role of p16-IHC for HPV detection in HNC is still questionable, as many studies found discordance because of p16 hypermethylation by prolonged smokeless tobacco exposure.⁹ Hence, p16-positive cases must be confirmed by HPV DNA status to consider causality in OPC.⁶ To our knowledge, there is a lack of meta-analysis on the proportion of HPV-associated HNC in Indian patients with a high burden of HPV-driven malignancies such as cervical cancer. Therefore, this systematic review and meta-analysis aimed to provide information about HPV prevalence and genotype distribution in HNC in India. It also intends to assess the proportion of HPV-attributable HNC by considering the biomarkers of carcinogenesis, p16, and HPV E6/E7 mRNA. The role of sex, risk factors such as tobacco and alcohol, and HPV detection methods in the prevalence of HPV-associated HNC is evaluated. Knowledge of HPV-attributable fraction in HNC in this population will be of considerable interest from a clinical and public health perspective for effective prevention, diagnosis, staging, and management.

METHODS

This systematic review and meta-analysis conform to the Preferred Reporting Items for Systematic Review and Meta-Analyses (PRISMA) 2020 guidelines.¹⁰ This study was registered with the International Prospective Register of Systematic Reviews (PROSPERO, registration No. CRD42022374699).

Search Strategy and Selection Criteria

The CoCoPop framework (condition, context, and population), a recommended approach for the review of prevalence studies,¹¹ is used for the search strategy and inclusion criteria.

Condition: HPV

Context: HNC

Population: India

The inclusion criteria were case-control and cross-sectional studies describing the prevalence of any HPV types in HNC (oral cavity, oropharynx, and larynx/hypopharynx); participants belonging to the Indian population; a clear description of diagnostic methods (PCR-based method); a sample size of more than 20 cases; and histologically classified as squamous cell carcinoma. Hypopharyngeal cases were classified as belonging to the larynx. We have excluded review articles and meta-analyses, articles where the research design is not scientific and reasonable, studies on premalignant lesions and conditions, cancers with nonsquamous origin, and studies restricted to advanced-stage tumors.

A systematic literature search was performed in November 2022 in electronic databases: Medline via PubMed, Embase, Scopus, ScienceDirect, ProQuest, and Cochrane. Search filters were used to restrict retrieval to studies in humans published in English between January 1990 and October 2022 and to journal articles. Search terms such as “Human papillomavirus,” “alpha papillomavirus,” “HPV16,” “HPV18,” “High-risk HPV,” “Head Neck cancer,” “Head Neck squamous cell carcinoma,” “Oral cancer,” “Tongue cancer” “Oropharyngeal cancer” “Laryngeal cancer,” “Prevalence” “Cross-sectional study,” “Case-control study,” “observational study,” “Epidemiology,” “Clinicopathologic,” “India,” and “Indian” were used. Cross-references were reviewed to identify additional sources. The searches were rerun before the final analyses. The search strategy used in different databases is provided in the Data Supplement (Table S1).

The titles and abstracts of all the articles identified through the searches were screened initially for potentially relevant studies, and any irrelevant ones were excluded. The full text of the remaining articles was reviewed to determine the studies that meet the inclusion criteria.

Study Selection and Data Extraction

Data extraction from the full text of the articles was performed by a single reviewer (N.V.V.) and verified by a second reviewer (R.M.) before inclusion. Any disagreements were resolved through discussion with a third researcher (R.S.). Authors were contacted electronically when there was incomplete or inadequate information. Data focusing on the first author, year of publication, region, cancer site (oral, oropharyngeal, and laryngeal [larynx and hypopharyngeal cases were combined]), sample size, sample type, age, sex, tobacco and alcohol habits, HPV detection method, HPV DNA positivity, genotypes detected, HPV DNA-positive cases tested for HPV E6/E7 mRNA and p16, and the number of HPV DNA-positive cases additionally positive for E6/E7 mRNA and p16 were extracted. HPV prevalence is calculated by dividing the number of HPV DNA-positive cases by the total number of patients tested. In one study, the HPV prevalence was determined by three different PCR techniques, and the data from a highly sensitive method with a larger sample size were included in the analysis. Likewise, studies that contain various HNC sites were classified as separate studies for the analysis. The data reported in one study were deemed unclassifiable, mainly because it was impossible to separate the HNC studied into constituent cancer types. HPV E6/E7 mRNA and p16 percent positivities refer to the percentage of HPV DNA-positive samples that were also positive for these biomarkers. The HPV-attributable fraction is calculated as the product of HPV prevalence and the percent positivities of the biomarkers (HPV E6/E7 mRNA and p16) in HPV DNA-positive cases.

Quality Assessment

The articles were assessed for quality using eight quality criteria as described by Ndiaye et al¹²—broad-spectrum PCR, human gene marker, histologic confirmation of cases, control for contamination, details about age and sex, statement of the recruitment period, sample size more than 50 cases, and random or consecutive or incident cases. The quality score was calculated as the sum of the result (0 = no/1 = yes) for all quality control items. This grading tool was deemed most suitable and would account for the robustness of the studies in assessing the prevalence of HPV.

Statistical Analysis

The pooled prevalence of HPV, p16INK4a, and E6/E7 mRNA percent positivity was estimated by random-effect logistic regression (logit) model using Metapreg, a statistical procedure in STATA (version 18; StataCorp LLC, College Station, TX). The percentage of total variation across studies because of heterogeneity was assessed by Pearson's Chi-square test using the I² measure. An I² value above 75% indicates high heterogeneity. The effect of covariates was explored by stratified analysis with meta-regression to estimate the effect size. The regression-based Egger test was done to examine the small-study effect, and the funnel plot was used to test for publication bias. The values of P ≤ .05 were considered statistically significant.

RESULTS

The systematic search of the electronic databases identified 300 publications whose titles and abstracts were reviewed for eligibility. Around 232 articles were excluded as they were duplicates and nonrelevant to the topic. An additional 21 studies were identified through cross-reference. The full text of 92 articles was reviewed for inclusion, of which 54 articles that met the inclusion criteria were assessed for quality. Three studies have used the same data and samples, of which the most recent study was included. Around four studies conducted exclusively on tobacco users were excluded from the analysis, as they were considered a high-risk group for cancer and might bias the results. Studies with an assigned quality score of <5 were excluded. The authors of 12 studies were contacted for additional information, of which only one author provided the required clinical details. Meta-analysis was performed for 34 high-quality studies with a quality score of ≥5. The PRISMA flowchart summarizes the study identification process (Fig 1).

FIG 1.

PRISMA flowchart of the selection process of studies on HPV-associated HNC in India. HNC, head and neck cancer; HPV, human papillomavirus; PRISMA, Preferred Reporting Items for Systematic Review and Meta-Analyses.

The characteristics and variables collected from the studies and their quality score are shown in the Data Supplement (Table S2). The references of all the articles included for analysis are listed in the Data Supplement (Material S3). These studies were observational, with a cross-sectional design and estimated point prevalence. All 34 studies used the presence of HPV DNA as positivity, whereas three studies analyzed HPV mRNA, and eight studies tested p16 biomarkers in HPV DNA-positive samples.

The random-effect model was used for the analysis because of the substantial heterogeneity among the included studies. With this approach, the overall pooled prevalence of HPV (determined by HPV DNA positivity) in HNC analyzed in 4,492 patients from 34 eligible studies was 20% (95% CI, 12 to 32) with a high level of heterogeneity (I² = 90.79%). When analyzed by geographical region, the highest HPV prevalence was noted in the Eastern part (four studies, 805 cases) with 58% (95% CI, 44 to 70). It is lower in the West (seven studies, 1,305 cases) with a 12% prevalence. Considerable heterogeneity was identified among HPV prevalence and geographic regions and anatomic sites.

Sixteen studies with 2,706 HNC cases presented HPV prevalence stratified by sex (1,976 males and 730 females). The pooled prevalence was slightly higher for females (29% [95% CI, 16 to 46]) compared with males (27% [95% CI, 14 to 47]) with high heterogeneity. The studies included in the analysis used fresh-frozen or formalin-fixed paraffin-embedded (FFPE) tissue specimen types for HPV detection. The pooled prevalence of HPV was higher in studies that used FFPE samples (35% [95% CI, 24 to 48]) than in fresh-frozen samples (15% [95% CI, 7 to 29]). Twenty-six studies have detected HPV by single-round PCR (23% [95% CI, 12 to 38]) and had a higher prevalence than the nested PCR approach (13% [95% CI, 2 to 52]). The presence of HPV was lower in studies using type-specific primers amplifying the E6/E7 region (14% [95% CI, 3 to 42]) than L1 consensus primers (22% [95% CI, 12 to 36]). Most HPV type–specific molecular tests detect the two most common oncogenic types, HPV16 and HPV18. Tobacco smokers (34% [95% CI, 23 to 49]) had lower HPV-associated HNC than chewing tobacco users (41% [95% CI, 25 to 60]) and combined tobacco and alcohol users (51% [95% CI, 3 to 97]). Analysis by the number of cases showed that studies with ≥100 patients with HNC had a higher HPV prevalence of 23% (95% CI, 12 to 38) compared with those with <100 cases (17% [95% CI, 7 to 37]). Stratified by tumor grade, poorly differentiated tumors (grade 3) showed an increased HPV prevalence (60% [95% CI, 28 to 86]) than well-differentiated tumors.

HPV16 was the most prevalent genotype detected in all the qualified studies, except four studies where the HPV18 genotype was predominant. HPV16 accounted for 86% (95% CI, 73 to 94) and existed as coinfection with HPV18 (HPV16/18) in 12% (6 to 26) of all HPV-positive cases. HPV18 was the second most prevalent type in HNC (27% [95% CI, 11 to 52]). Apart from the common oncogenic types HPV16 and HPV18, the prevalence of other HPV types (6, 8, 9, 11, 22, 31, 32, 33, 35, 38, 39, 40, 52, 53, 56, 58, 68, and 73) was 20% (9 to 38). The effect of covariates on HPV prevalence in HNC is shown in Table 1.

TABLE 1.

Effect of Covariates on the Pooled Prevalence of HPV in HNC

Variable	Studies, No.	Cases, No.	HPV+, No.	Prevalence, % (95% CI)	Heterogeneity Test I2, %
Overall	34	4,492	1,304	20 (12 to 32)	90.79
Region
North	3	381	105	28 (23 to 32)	0.00
South	13	1,152	306	14 (3 to 44)	62.14
East	4	805	484	58 (44 to 70)	89.79
West	7	1,305	221	12 (3 to 33)	90.98
Central	4	543	107	16 (8 to 30)	76.98
Northeast	3	306	81	23 (11 to 41)	88.67
Sex
Male	16	1,976	724	27 (14 to 47)	94.29
Female	16	730	238	29 (16 to 46)	84.91
Sample type
Fresh-frozen	17	2,046	555	15 (7 to 29)	90.09
FFPE	10	1,443	594	35 (24 to 48)	87.47
Detection method
Single-round PCR	26	3,090	1,087	23 (12 to 38)	91.31
Nested PCR	3	427	31	13 (2 to 52)	83.72
Primers type
Consensus	25	3,079	968	22 (12 to 36)	90.82
Type-specific	5	852	179	14 (3 to 42)	92.92
Risk factors
Smoking	12	1,058	431	34 (23 to 49)	66.72
Chewing tobacco	10	764	295	41 (25 to 60)	87.10
Alcohol	6	455	137	31 (19 to 45)	81.46
Tobacco + alcohol	5	135	61	51 (3 to 97)	7.09
Sample size
<100	17	944	280	17 (7 to 37)	81.00
≥100	17	3,548	1,024	23 (12 to 38)	96.57
Tumor grade
Well-differentiated	9	712	329	50 (33 to 66)	88.67
Moderately differentiated	10	559	261	47 (27 to 67)	83.97
Poorly differentiated	10	143	88	60 (28 to 86)	58.99
HPV types
HPV 16	21	956	793	86 (73 to 94)	45.59
HPV 18	13	688	163	27 (11 to 52)	80.21
HPV 16/18	6	429	41	12 (6 to 26)	64.59
Other	7	331	79	20 (9 to 38)	82.75

Abbreviations: FFPE, formalin-fixed paraffin-embedded; HNC, head and neck cancer; HPV, human papillomavirus; PCR, polymerase chain reaction.

Role of HPV in OC, OPC, and LC

The pooled prevalence of HPV in OC (30 studies, 3,264 cases) was 16% (95% CI, 8 to 30), while it was significantly higher for OPC (10 studies, 455 cases; 22% 95% CI, 13 to 34) and LC (12 studies, 390 patients; 29% 95% CI, 17 to 46). The forest plot illustrating the individual and pooled prevalence estimates of HPV in HNC cases stratified by the anatomic site is shown in Figures 2A-2C.

FIG 2.

(A-C) Forest plots of meta-analysis on HPV prevalence in HNC by anatomic site: (A) oral, (B) oropharyngeal, and (C) laryngeal cancers. HNC, head and neck cancer; HPV, human papillomavirus.

Only three studies have tested for E6/E7 mRNA in HPV DNA-positive OC (67 cases), yielding a pooled percent positivity of 21% (95% CI, 9 to 43). However, it was higher for OPC (57% [95% CI, 36 to 75]), followed by LC (40% [95% CI, 19 to 65]). The HPV-attributable fraction of OPC, considering the E6/E7 mRNA positivity in addition to HPV DNA, was 12.54%, almost similar to LC (11.6%), while it is much lower in OC (3.36%). The pooled p16 percent positivity in OPC was 44% (95% CI, 33 to 56), higher than LC (33% [95% CI, 21 to 49]) and OC (25% [95% CI, 18 to 34]). The attributable fraction of HPV, considering both p16 and HPV DNA positivity, is similar for OPC (9.68%) and LC (9.57%), while it is lower in OC (4%). The estimates of HPV-attributable fraction in HNC stratified by the anatomic site are summarized in Table 2.

TABLE 2.

HPV-Attributable Fractions in HNC on the Basis of E6/E7 mRNA and p16 Positivity

Variable	Oral Cavity	Oropharynx	Larynx
Samples tested for HPV DNA, No.	3,264	455	390
HPV DNA+ cases, No.	809	130	121
HPV prevalence, % (95% CI)	16 (8 to 30)	22 (13 to 34)	29 (17 to 46)
HPV DNA+ cases tested for E6/E7 mRNA, No.	67	23	15
HPV DNA+ and E6/E7 mRNA + cases, No.	14	13	6
Percent of E6/E7 mRNA+ cases in HPV DNA+ cases, % (95% CI)	21 (9 to 43)	57 (36 to 75)	40 (19 to 65)
Attributable fraction of E6/E7 mRNA and HPV DNA+ cases, %	3.36	12.54	11.60
HPV DNA+ cases tested for p16, No.	218	68	42
HPV DNA+ and p16+ cases, No.	52	30	14
Percent of p16+ cases in HPV DNA+ case, % (95% CI)	25 (18 to 34)	44 (33 to 56)	33 (21 to 49)
Attributable fraction of p16 and HPV DNA+ cases, %	4	9.68	9.57

Abbreviations: HNC, head and neck cancer; HPV, human papillomavirus.

Egger's regression test for small study effects showed a statistically significant association (P < .001), and the displayed funnel plot also demonstrates that there is evidence of publication bias (Fig 3).

FIG 3.

Funnel plot displaying publication bias. SE, standard error.

DISCUSSION

This systematic review and meta-analysis critically analyzed the available literature on the role of HPV and HNC in the Indian population. There is a considerable variation in the prevalence of HPV-positive HNCs in India, ranging from 0% to 73%. A 20% pooled prevalence of HPV in HNC was reported with clear heterogeneity insights. Some possible reasons for the variability in the effect size include sample types, number of cases analyzed, different HPV detection methods, and interlaboratory variability. Our results were similar to the Asian prevalence of 21.5%,¹³ but lesser than the global prevalence of 34.8% reported in a meta-analysis on PCR-based studies.¹⁴ Our data indicate that the proportion of mucosal HPV associated with HNC in India differs from those in developed parts of the world, probably because of differences in tobacco usage and sexual behavior.

The association between HPV infection and HNC in specific sites suggests that the strongest association is with OPC and LC. However, the pooled prevalence of HPV in OPC (22%) in our study was lower compared with 33% globally.¹⁵ The prevalence of HNC in various subsites is similar to the one reported by Saulle et al¹⁶ on the basis of a meta-analysis of studies from 1988 to 2011. An increasing incidence of OPC has been observed in the United Kingdom, the United States, Europe, New Zealand, and parts of Asia. HPV accounts for 71% and 51.8% of all OPC in the United States and the United Kingdom, respectively, surpassing cervical cancer in women.¹⁷ Changes in sexual behavior and a decline in tonsillectomy rates could explain this increase in oral HPV exposure.^6,18 However, unlike the Western literature, our data are inconclusive in defining a time trend for an increase in incidence. Despite bearing most of the global burden of HPV-associated cervical cancer, data about HPV-related OPC are scarce in lower- to middle-income countries in South Asia and sub-Saharan Africa. Hence, it becomes difficult to understand whether such increasing time trends are absent or unreported in these regions.¹⁹ Although the incidence of HPV-associated OPC appears to be highest in high-income countries, more epidemiologic data are needed from low- and middle-income countries. Most of the global data on HPV-associated HNCs are reported for OPC; nevertheless, Indian studies focused predominantly on the prevalence of HPV in OC. The literature focused exclusively on HPV in OPC in India is limited, with a single study showing a 22.8% prevalence rate.²⁰ Among the two studies on LC, only a single study reporting a prevalence of 13.3% was taken for analysis as it fulfills the eligibility criteria.²¹ The differences in the proportion between the sites indicate that HPV has a varying degree of pathogenicity in these anatomic sites. However, other risk factors such as tobacco and alcohol increase the overall risk of oncogenesis in these regions.

The vast and ethnically different populations in India affect the prevalence of HPV in HNC, accounting for the variation in the geographical area. The eastern region had a higher prevalence of HPV-positive HNC, probably because of the high incidence of OC in the age group of 40-69 years in West Bengal.²² The north and northeast (NE) regions were underrepresented as there were only three studies with smaller sample sizes from this region. The NE region lacks the required infrastructure for specialized treatment facilities, and many patients with cancer undergo treatment outside the region.²³ The local culture and lifestyle factors could explain the heterogeneity among the geographic regions. Although most studies have shown an increased HPV-associated HNC in men, four studies have reported a higher incidence in women.^24-27 Information about age was insufficient in many studies; hence, it was difficult to assess its association with HPV prevalence. However, one study reported that the mean age of HPV-positive OC was significantly lower than its HPV-negative counterpart.²⁸ The study on OPC reported that the HPV-positive patients were 8 years younger than the HPV-negative patients.²⁰ The absence of information about sexual behavior in most of these studies made it impossible to determine their role. Bahl et al²⁰ reported that HPV-positive cancers were more common in patients from an urban background than rural (66% v 40%) and had practices of oral sex and a higher number of lifetime sexual partners. These sexual practices were significantly more common in HPV-positive patients with cancer.²⁹

Recent studies have shown that nested PCR with two sets of primers PGMY/GP+ is more efficient for HPV detection in OPC, which may have a very low viral load.³⁰ Only three studies in our analysis have used a nested PCR approach,^31-33 while most have adopted single-round PCR with either consensus or type-specific primer. The HPV positivity was twofold greater when FFPE samples were analyzed. In a meta-analysis by Mehanna et al,⁴ there were no significant differences in the prevalence estimate when fresh-frozen or FFPE tissue samples were used. However, a study on cervical cancer has shown higher HPV positivity in samples from paraffin blocks, although statistically insignificant.³⁴ Generally, amplification of shorter fragments of the HPV genome is preferred for FFPE tissues.³⁵ HPV detection rates are significantly higher when targeting shorter PCR amplicons than larger ones because the shorter segments are likely to remain intact and successfully amplified.³⁶ This indicates a need for validated PCR assays similar to cervical cancer–related HPV testing to assess HPV in HNC. The lack of specificity in the number of PCR assays with the different types of samples has probably contributed to the heterogeneity. Studies using FFPE specimens for HPV detection have shown more varied results owing to heavy crosslinking in the samples and associated nonspecific amplification contributing to the heterogeneity.

The attributable fraction of HPV adjusted for tobacco and alcohol habits cannot be fully assessed because of the lack of sufficient data in most studies. Analysis revealed that the prevalence of HPV-related HNCs in combined tobacco and alcohol users was higher (51% [95% CI, 3 to 97]) than in nonusers. The prevalence was also higher in tobacco chewers (41% [95% CI, 25 to 60]) compared with smokers (34% [95% CI, 23 to 49]) and alcohol users (31% [95% CI, 19 to 45]). Tobacco use is known to increase the chance of HPV infection and has an additive effect on carcinogenesis.³⁷ It also decreases the survival benefit of HPV-positive status in patients with OPC.³⁸ The carcinogenic role of the virus cointeracting with other risk factors such as tobacco/alcohol needs further research to elucidate the exact role of the virus in such conditions. This necessitates extensive research to determine whether HPV prevalence is overestimated than their actual contribution to the development of these cancers.

The association of sample size on HPV prevalence estimates has shown decreased prevalence with smaller sample sizes. This is in contrast to the reduced HPV prevalence noted with increased sample sizes for OC and LC by Ndiaye et al.¹² It could be attributed to differences in detection technique, sample types, and publication bias. Besides, sample size calculation needs to be included in most analyzed studies. Regarding tumor differentiation, HPV-positive HNC in India is significantly associated with a poorly differentiated tumor grade. Generally, most HNCs are moderately differentiated, whereas HPV-associated cancers are predominantly nonkeratinizing or poorly differentiated/basaloid carcinomas.³⁹ HPV16 and HPV18 subtypes have been etiologically linked with HNC. HPV16 was the most common oncogenic HPV type, accounting for 86% of all HPV-positive HNCs. The second most common high-risk HPV type, HPV18, was detected in 27% of all HNCs. Therefore, the types targeted in the current HPV vaccines, HPV16 and HPV18, were seen in most HPV-positive cases. It has been reported that high-risk HPV16 and HPV18 are the most predominant subtypes in oral tumors from Indian patients, whereas the other subtypes (HPV33, HPV6, and HPV11) are rare.⁴⁰

In addition to HPV prevalence, biomarkers of HPV oncogenic activity E6/E7 mRNA and p16 positivity were considered to estimate the HPV-attributable fractions in HNC. HPV-attributable fractions with E6/E7 mRNA and p16INK4a were much lower in OC (3.36% and 4%, respectively) compared with OPC (12.5% and 9.7%, respectively) and LC (11.6% and 9.6%, respectively). Nevertheless, very few studies tested for these biomarkers show the need to explore the attributable fraction of HPV in HNC further with standardized procedures. The HPV biomarkers E6/E7 mRNA and p16INK4a also have their limitations. Although HPV E6/E7 mRNA is accepted as the gold-standard test to elucidate the oncogenic role of HPV in the tumor, nucleic acid degradation in FFPE tissues and the presence of residual DNA in cDNA samples can affect the sensitivity and specificity.¹² Previous studies have shown that most HPV-positive OCs were negative for p16 overexpression. At the same time, it was overexpressed in some HPV-negative tumors, suggesting that the p16 levels might not indicate active HPV status.⁴¹ Another study has shown that p16 expression in OC was associated with a higher tumor grade with 21.4% overexpression; however, none of the p16-positive tumors were associated with HPV.⁴² Studies with only p16INK4a as an HPV detection method in HNC were not included, as many studies have found discordance between p16 and HPV positivity. The reported attributable fraction concerning p16INK4a positivity reflects maximum estimates because some positive tumors for p16INK4a might not have been caused by HPV infection. The widely used test for HPV detection in clinical practice for OPC is p16 (INK4A) immunohistochemistry (IHC). In the modified eighth AJCC/UICC classification, p16 was recommended as an HPV surrogate marker for OPC if there was diffuse (≥75%) expression and at least moderate (+2/3) staining intensity in the tumor⁴³; however, this has not been validated in OC.⁴⁴ Therefore, overexpression of p16 may probably not be representative of HPV in OC. This is quite apparent in Indian patients because of p16 hypermethylation driven by prolonged smokeless tobacco exposure.³⁹

This analysis must be interpreted considering the heterogeneity of the studies, such as geographic region, tumor site, sample size, sample type, sensitivity of HPV primers, and different PCR protocols. Hence, the random-effects model was used to combine data, and meta-regression was performed to identify possible associations between the presence of HPV and predefined covariates. Large-scale collaborative research with standardized HPV testing and study designs exploring major confounding factors are recommended to identify the exact role of HPV in HNC etiology. Although HPV molecular testing by p16 IHC is recommended for staging OPC in the modified eighth AJCC/UICC classification, the Centers for Disease Control and Prevention does not recommend routine HPV screening for detecting persistent oral HPV infection. Validated HPV DNA-based screening for early detection of persistent high-risk HPV infection and gender-neutral HPV vaccination can potentially prevent a substantial fraction of HNC, particularly OPCs. The US Advisory Committee on Immunization Practices recommends HPV vaccination for females and males age 9-26 years to prevent HPV infection and associated cancers.⁴⁵ In India, the HPV vaccine was rolled out for girls age 9-14 years and is planned to be introduced in three phases as part of the cervical cancer elimination program. This will become part of their routine immunization programs once successfully introduced.⁴⁶ However, recommending the vaccination to boys requires adequate scientific evidence that establishes the attributable fraction of HPV to HNC, particularly oropharyngeal, independent of other known risk factors. HPV status is a recommended biomarker for OPC in patient stratification toward de-escalation treatment regimens to minimize treatment-associated toxicity without affecting the outcome. However, the ongoing randomized clinical trials evaluate the treatment de-escalation strategies with well-defined selection criteria.

In conclusion, this study presents the prevalence of HPV and their attributable fraction in HNC in Indian patients, the region where tobacco-related HNC is a major burden. Although a significant proportion of HNCs are associated with a transient HPV infection (HPV DNA positivity), the HPV-attributable fraction based on a clinically (transcriptionally) active infection is considerably lower for OC, suggesting a negligible causative role of HPV in OC. Despite a substantial fraction of OPCs and LCs being attributed to HPV, the exact causative role of HPV is unclear because of the presence of other known risk factors. Further studies exclusively on OPCs and LCs in nonsmoking and nonalcoholic individuals are recommended using standardized molecular techniques on larger, well-defined Indian patients to elucidate the etiologic role of HPV in these cancers. If established, they can be prevented by screening for persistent HPV infection and prophylactic vaccination.

AUTHOR CONTRIBUTIONS

Conception and design: Nandimandalam Venkata Vani, Rajendran Madhanagopal, Rajaraman Swaminathan

Collection and assembly of data: Nandimandalam Venkata Vani, Ranganathan Rama, Rajendran Madhanagopal, Vijayalakshmi Ramshankar, Rajaraman Swaminathan

Data analysis and interpretation: Nandimandalam Venkata Vani, Ranganathan Rama, Rajendran Madhanagopal, Vijayalakshmi Ramshankar, Rajaraman Swaminathan

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 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/go/authors/author-center.

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

No potential conflicts of interest were reported.