The Prevalence of HPV in Oral Cavity Squamous Cell Carcinoma

Human papillomavirus (HPV) is an important risk factor in a subset of head and neck squamous cell carcinomas (HNSCC), but the association with oral cavity squamous cell carcinomas (OCSCC) remains controversial. This study aimed to identify the prevalence of HPV infection in OCSCC. A systematic search on PubMed and EMBASE was performed, including articles assessing the prevalence of HPV-positive (HPV+) OCSCC published from January 2017 to December 2022. OCSCC was considered HPV+ by the detection of HPV DNA, HPV RNA, and/or p16 overexpression in the tumor mass. A meta-analysis was made determining the overall HPV+ OCSCC prevalence. We included 31 studies comprising 5007 patients from 24 countries. The study size ranged from 17 to 940 patients. The HPV+ OCSCC proportion variated widely and ranged from 0% to 37%. Tumors in the tongue were the predominant sublocation for HPV in the oral cavity. The meta-analysis revealed that the overall HPV+ OCSCC prevalence is 6% (95% CI; 3–10%), and only one study found HPV and OCSCC significantly associated. Thus, HPV may not be a necessary or a strong risk factor in OCSCC oncogenesis, and the possibility of a site misclassification of a mobile tongue with the root of the tongue cannot be excluded.


Introduction
Oral cavity cancer is the most frequent cancer subsite in the head and neck area with more than 377,700 cases worldwide in 2020, placing them as the 16th most common cancer overall, the 11th most prevalent cancer in men, and the 18th most frequent cancer in women [1]. In addition, human papillomavirus (HPV) is widely prevalent worldwide and is the most frequent sexually transmitted infection in the United States [2,3]. It is well established that HPV is an important risk factor in a subset of head and neck squamous cell carcinoma, particularly oropharyngeal squamous cell carcinomas (OPSCC) [4][5][6][7].
Although well-known risk factors of oral cavity cancer are the consumption of tobacco (both smoked and smokeless), alcohol, and poor oral hygiene [8], the association of HPV as a risk factor for oral cavity squamous cell carcinomas (OCSCC) remains controversial since the association was described for the first time in 1983 [9].
There are no universal HPV-testing methods available for OCSCC. The identification of HPV E6/E7 mRNA expression is considered the golden standard for measuring HPV infection by some authors, as this technique detects oncogenic transcriptionally active HPV. However, the method is expensive, technically challenging to perform, and requires
A higher alcohol consumption and a higher number of sexual partners were demonstrated to be associated with HPV+ OCSCC in one study [38]. Two studies found an HPV association with an earlier T-stage [41,45].

The HPV+ OCSCC Prevalence Worldwide
The lowest prevalence of HPV+ OCSCC was 0% found in the Philippines [20], the United Kingdom (UK) [36], India [19,22], the Republic of Korea [25], and France [40]. The highest prevalence was reported to be 37% in Jordan [24]. Additionally, both low and high HPV occurrences were reported across all geographical locations. A study from India was the only study that concluded a statistically significant association of OCSCC to HPV infection by a Chi-Square Test [28], shown in Table 1.
A higher alcohol consumption and a higher number of sexual partners were demonstrated to be associated with HPV+ OCSCC in one study [38]. Two studies found an HPV association with an earlier T-stage [41,45].

The HPV+ OCSCC Prevalence Worldwide
The lowest prevalence of HPV+ OCSCC was 0% found in the Philippines [20], the United Kingdom (UK) [36], India [19,22], the Republic of Korea [25], and France [40]. The highest prevalence was reported to be 37% in Jordan [24]. Additionally, both low and high HPV occurrences were reported across all geographical locations. A study from India was the only study that concluded a statistically significant association of OCSCC to HPV infection by a Chi-Square Test [28], shown in Table 1.
A proportional meta-analysis was conducted, determining the total prevalence of HPV+ OCSCC to 6% (95% CI; 3-10%). There was a great heterogeneity in the prevalence as well, I 2 > 75%, p < 0.01, shown in Figure 2. The HPV status in the tumors located in different anatomical subsites of the oral cavity was specified in 16 studies (n = 284). Eight studies observed the highest proportion of HPV+ OCSCC in the tongue (n = 162) [19,24,27,28,30,35,42,45], and in three studies, the highest prevalence was seen in the floor of mouth (FoM) (n = 89) [37,38,41]. The HPV+ OCSCC proportion in the tongue varied from 0% to 100%, and among patients with tumors in the FoM, it varied from 0% to 57.1%. The specificity of the mobile tongue has been indicated as follows: * specified mobile tongue, ** specification questionable due to not following a classification system or only writing "tongue", *** unspecified due to inclusion of overlapping lesion of tongue and/or lingual tonsil and/or tongue unspecified, shown in Table 1.

Discussion
We conducted a systematic review and meta-analysis that examined the HPV+ OCSCC prevalence worldwide. Thirty-one studies were included with a total of 5007 patients. There was a wide variety in the prevalence of HPV+ OCSCC reported, with the lowest prevalence being 0% in the Philippines, the UK, India, the Republic of Korea, and France and the highest being 37% in Jordan [20,22,24,25,33,36,38,40]. Studies with a higher HPV+ proportion had greater statistical uncertainty as demonstrated in our meta-analysis, shown in Figure 2. The Jordanian paper consisted of a small study size comprising 27 patients, and they only detected HPV DNA, which should be noted. Furthermore, they found no HPV association with OCSCC or other clinical factors like tobacco, alcohol, age, or gender [24]. The studies that reported 0% HPV+ OCSCC proportion had a sample size ranging from 31 to 166 individuals and defined HPV+ as double positivity with p16+/HPV DNA or solely HPV DNA. The study from India observed the HPV+ OCSCC fraction to be 13% and was the only study to find a statistically significant association between the HPV status and OCSCC [28]. Moreover, with the HPV−OCSCC being more frequent than HPV+ OCSCC in all the studies and with our meta-analysis determining the total prevalence of HPV+ OCSCC globally being 6% (95% CI; 3-10%), it may indicate that HPV infection is not a mandatory nor a strong risk factor and does not constitute a high proportion of the OCSCC worldwide. Additionally, a study found no difference in survival outcomes between patients with HPV+ OCSCC and HPV−OCSCC when stratified on p16 overexpression status [15]. Regarding survival, our enrolled studies support that neither p16-status nor HPV-status have an impact on OCSCC patients [32,35,37,41]. Only one study reported a trend towards increased survival in HPV-positive individuals, although they did not find it statistically significant [42]. HPV+ OPSCC is more prevalent in the Western world [52,53], but in our enrolled studies there was no correlation, both high and low HPV+ OCSCC prevalence were observed regardless of the geographical area. Additionally, studies have shown HPV+ OPSCC to be more apparent among the younger patients that generally consuming less alcohol or tobacco [54,55]. Compared to HPV+ OCSCC in our study, only two enrolled studies reported such significant association with younger individuals [38,41]. Solely, one study observed HPV+ OCSCC association with less tobacco consumption [37]. A higher alcohol consumption and number of sexual partners were demonstrated to be associated with HPV+ OCSCC in one study as well [38]. Interestingly, the same study demonstrated a significant HPV+ OCSCC association to an earlier T-stage and concluded that it may indicate that both alcohol intake and oral HPV infection act synergistically, explaining earlier tumor onset. Overall, this could emphasize that the HPV+ OCSCC is most frequently seen in elderly patients, and the association with less alcohol and tobacco consumption is weak in contrast to the HPV+ OPSCC. Presumably, the oncogenicity of HPV-infection is of less magnitude in the genesis of OCSCC compared to OPSCC. This is supported by the decrease in incidences of OCSCC (i.e., in the USA while HPV-related OPSCC incidents have been increasing) [56].
We found that HPV16 was the predominant genotype and HPV18 was the next common genotype in the oral cavity. Other HPV HR-subtypes (HPV 31,33,45,52 and 59) were less frequent, but as a shortcoming, some studies only examined the presents of HPV16 and/or HPV 18. The 9-valent HPV vaccine (Gardasil9) may in theory reduce the possible HPV-caused carcinogenesis in the oral cavity since it covers all these subtypes except for HPV 59, which was only present in one case [49].
The anatomical subsites of HPV+ OCSCC was specified in 14 of the studies. Discrepancies were observed regarding the most prevalent sublocation. Ten studies did not investigate the subsites of the tumors, which complicates the interpretation of the results and is a limitation for determining the true representation of HPV+ OCSCC sublocations [23,29,31,37,39,43,44,[46][47][48]. However, most of the included studies reported that HPV+ OCSCC were more apparent in the tongue followed by the FoM. This raises an interesting issue as to whether some of the tumors in the tongue might have arisen from the base of the tongue, which is classified as a part of the oropharynx instead of the mobile tongue, which is considered a subsite of the oral cavity.
Hence, a shortcoming can be attributed to the different definitions of the oral cavity among the studies regarding the included subsites. The Surveillance, Epidemiology, and End Results (SEER) database and Systematized Nomenclature of Medicine-Clinical Terms (SNOMED-CT) do not distinguish between the base of the tongue and the mobile tongue when describing the oral sites but labels them both as lingual or tongue, which can cause some anatomical site misclassification and contribute to the confounding of the reported HPV+ OCSCC proportion [57,58]. Furthermore, SNOMED-CT consider the palate as a part of the oral cavity without distinguishing between the hard palate and the soft palate, with the latter being a part of the oropharynx [58]. Our study demonstrates some outlying studies that report a higher prevalence of HPV+ OCSCC than the pooled prevalence, and a high proportion of the tumors are found in the tongue [19,24,28,30,38]. In the paper with the highest reported HPV+ OCSCC proportion, we observed that all HPV+ cases were seen in the tongue, and 50% (10/20) of all the tongue tumors were HPV+, shown in Table 1. Nonetheless, they distinguished between the base of tongue and the mobile part, but considering the ICD-10 codes, overlapping lesions might have been involved (C.02.8 "overlapping lesion of tongue") as they do not report the exact included ICD-10 codes [59]. Another example to notice is that the only study that found a significant association between HPV and OCSCC included the soft palate as a part of the oral cavity, as shown in Table 1. Moreover, we excluded some relevant studies due to the unclear definition of the term "oral", which might solely refer to the oral cavity or the combination of the oral cavity and oropharynx. On the contrary, these mistakes should easily be avoided as the anatomical sites are clearly defined [59,60]. A previous study has investigated the prevalence of HPV in palatine tonsillar squamous cell carcinoma, subdivided, according to the certainty of tonsillar tumor origin into specified tonsillar squamous cell carcinoma (STSCC) and nonspecified tonsillar squamous cell carcinoma (NSTSCC). The study observed the proportion of HPV+/p16+ to be 72% for STSCC and 21% for NSTSCC [61]. Hence, it would be interesting to explore if there was a big difference in the prevalence of HPV+ tumor according to the certainty of mobile tongue origin as well. Eventually, the HPV+ OCSCC fraction globally might be smaller than the 6% determined in this study.
The inconsistency of HPV detection methods among the studies was remarkable and a limitation to this review. PCR, ISH, hybrid capture 2 high-risk HPV (HC2-HPV DNA), and p16 IHC were among the different methods applied in the studies. We prioritized results from the methods that have defined HPV presence by double positivity since studies based on HPV DNA and p16 combined are more reliable, as the method has demonstrated a high sensitivity and specificity when considering HPV16 E6/E7 mRNA detection as the golden standard [13,16,17]. Many of the enrolled studies only detected HPV DNA with PCR, which is not the equivalent to the virus being transcriptionally active and could lead to false-positive test results, making this method suboptimal [17,62]. A few studies were based on p16 IHC positivity, and they all reported a higher HPV+ OCSCC prevalence than the pooled result determined by our meta-analysis. The overexpression of p16 might not reveal the true result, as it is only identified as valid for OPSCC, where it furthermore yields false-positive results in up to 20% when compared to the detection of HPV DNA [50,62]. It has also been recommended not to use p16 IHC as a tool for OCSCC in study trials since it has shown poor performance as a prognostic marker for overall survival in OCSCC [15].
To compare the studies and determine the HPV+ OCSCC fraction worldwide, it is important to find a homogenous HPV detection method that is both highly sensitive and specific for biologically active HPV and considered beneficial in the cost-effective context.
Moreover, there is no consistency in the definition of p16-positivity. Most of the studies defined p16 IHC overexpression as positive if staining ≥70% of the tumor cells was observed. Other studies set the limit of the staining for p16 to be positive at >75%, >50%, and >10%, and one study used a 5-tiered point system to determine p16+. The wide variety of p16-positive definition can also result in a wrong interpretation of the HPV+ OCSCC proportion.
Lastly, a meta-analysis was conducted, revealing significant heterogeneity I 2 > 75%, p < 0.01, and the global burden of HPV+ OCSCC to 6% (95% CI; 3-10%). (Figure 2). We did not stratify for the various detection methods because some of the detection methods were too few for a valid meta-analysis, which is a limitation that should be noted. We excluded from our review studies and meta-analyses that assessed the prevalence of HPV in the head and neck area if data from the oral cavity could not be extracted due to pooled data from multiple anatomical locations. Hence, it entails some missing data that could be used in our study.
To the best of our knowledge, with over 5000 patients and 31 studies, this is the largest systematic review and meta-analysis conducted evaluating the association of HPV infection in solely OCSCC worldwide within the last five years.

Conclusions
In conclusion, there was a significant heterogeneity in the HPV+ OCSCC prevalence worldwide, varying from 0% to 37%. Studies with higher HPV+ fraction had greater statistical uncertainty. The most prevalent HPV genotype was HPV 16, and the second most prevalent was HPV 18. Tumors in the tongue were the predominant sublocation for HPV in the oral cavity. The meta-analysis revealed the pooled HPV+ OCSCC prevalence worldwide to be 6% (95% CI; 3-10%) from the included studies. HPV−OCSCC were more frequent than HPV+ OCSCC in all the studies and only one study found HPV significantly associated with OCSCC. Thus, taking the heterogeneity and limitations into account, HPV