Oral Microbiota—A New Frontier in the Pathogenesis and Management of Head and Neck Cancers

Simple Summary Head and neck squamous cell carcinoma (HNSCC) is a group of common and aggressive tumors. Recently, oral microbiota has been credited as an important player in carcinogenesis. However, the available knowledge is not always consistent and sometimes conflicting. Therefore, the present comprehensive systematic review of the current clinical reports was conducted to evaluate the role of oral microbiota in HNSCC. Importantly, this study addresses whether oral microbiota targeting could provide diagnostic, prognostic or therapeutic utility in cancer patients. We also discussed the current limitations of this newly emerging field and the potential related strategies for the management of patients with HNSCC and possibly other solid tumors. Abstract Head and neck squamous cell carcinoma (HNSCC) comprises the majority of tumors in head and neck tissues. The prognosis of HNSCC has not significantly improved for decades, signifying the need for new diagnostic and therapeutic targets. Recent evidence suggests that oral microbiota is associated with carcinogenesis. Thus, we conducted a comprehensive systematic review to evaluate the current evidence regarding the role of oral microbiota in HNSCC and whether their targeting may confer diagnostic, prognostic or therapeutic utility. Following the screening of 233 publications retrieved from multiple databases, 34 eligible studies comprising 2469 patients were compiled and critically appraised. Importantly, many oral pathogens, such as Porphyromonas gingivalis and Fusobacterium nucleatum were linked to certain oral potentially malignant lesions and various types of HNSCC. Furthermore, we summarized the association between the expression profiles of different oral bacterial species and their tumorigenic and prognostic effects in cancer patients. We also discussed the current limitations of this newly emerging area and the potential microbiota-related strategies for preventing and treating HNSCC. Whilst many clinical studies are underway to unravel the role of oral microbiota in cancer, the limited available data and experimental approaches reflect the newness of this promising yet challenging field.


Introduction
Head and neck squamous cell carcinoma (HNSCC) arises from the mucosal lining of the oral cavity, pharynx and larynx and it comprises the majority of tumors in the head and neck region [1,2]. Globally, HNSCCs are among the most prevalent cancers with an estimated incidence of 880,000 new cases and 440,000 deaths in 2020 alone [3]. In India, oral squamous cell carcinoma (OSCC) is causing most of the cancer-related deaths among men [3]. The 5-year survival rate of HNSCC remains low and has not significantly

Protocol and Study Registration
The protocol of this study was designed according to the recommendations of the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA). The study was retroactively registered in the International Prospective Register of Systematic Reviews (PROSPERO; registration number CRD42021256877) prior to the initiation of the systematic search [15].

Inclusion and Exclusion Criteria
We included original research studies that assessed the relationship between oral microbiota and the tumorigenesis of HNSCC in human samples. The detailed inclusion and exclusion criteria are listed in the Table S1.

Search Strategy and Study Screening
The literature search was conducted on the 13th of June 2021 without restrictions through four electronic databases: PubMed, Web of Science, Ovid Medline, and Cochrane. The applied search terms included: (("oral microbiota") OR ("oral microbiome") OR ("oral bacteria *") OR ("oral microbial") OR ("oral microorganism") OR ("oral microbe")) AND (("head and neck squamous cell carcinoma") OR ("head and neck neoplasms") OR ("head and neck cancer") OR ("head and neck squamous cell cancer") OR ("oral cancer") OR ("mouth neoplasms") OR ("laryngeal neoplasms") OR ("gingival neoplasms") OR ("lip neoplasms") OR ("palatal neoplasms") OR ("tongue neoplasms") OR ("pharyngeal neoplasms")). Following deduplication, studies were first assessed for eligibility, and then further screened for qualitative assessment. Two reviewers (MM and SH) independently screened and assessed the literatures. Differences in the results, if any, were resolved through discussion with a third reviewer (AS).

Data Extraction and Study Items
Extracted data were tabulated using the online collaborative tool Google Sheets™ (Google, Menlo Park, CA, US). The following key items were extracted from the eligible studies: The 1st author's name, publication year, title, country, total number and age of patients, sample type, oral microbiota, tumor (type, location, stage, grade), Epstein Barr virus (EBV)/HPV status, methods, antibodies (name, dilution, company, etc.), p-value, confidence interval (CI), hazard ratio (HR), and prognostic data whenever applicable. Data extraction was performed and verified independently by two reviewers (MM and SH).

Assessment of Reporting Quality and the Risk of Bias
The included prognostic studies were evaluated using items adapted from the Reporting Recommendations for Tumor Marker Prognostic Studies (REMARK) guidelines [16]. The assessment criteria are detailed in the Table S2. For analyzing the risk of bias, we used the Meta-Analysis of Statistics Assessment and Review Instrument (MAStARI) tool as we recently described [17]. The assessment was performed independently by two reviewers (MM and SH). Disparities, if any, were resolved through discussion with a third reviewer (AS).

Study Selection
A total of 536 studies were retrieved through the initial searches. After deduplication, 233 articles were identified and screened for eligibility. Of these, 34 studies were deemed relevant and selected for further qualitative analysis and data extraction. The search and selection process is summarized in Figure 1.

Baseline Characteristics of the Studies
The included 34 studies comprised a total of 4432 participants (including 2294 cancer patients, 175 OPMDs and 1963 cancer-free controls). Studies were conducted between 2005-2021 and they were based in the following countries: US (n = 10), Taiwan (n = 5),
The16S ribosomal RNA (rRNA) gene polymerase chain reaction (PCR) was the most commonly used approach for oral bacterial detection and identification. The sampling and characterization methods of the oral microbiota are summarized in Table 2 and Figure 2.

Reporting Quality and the Risk of Bias
To assess the reporting quality of the included studies, we applied six REMARKadapted items. Of note, only one study fulfilled all of the applied criteria while the rest had at least one missing item. The MAStARI tool revealed that the risk of bias was low in 15 studies (44%), moderate in 14 studies (41%), and high in 5 studies (15%). The results for each study using the REMARK and MAStARI assessment tools are detailed in the Table S3.

Oral Microbiota and OPMDs
The relationship between oral microbiota and OPMDs was assessed in 6 studies. Schmidt et al. found that pyhla Firmicutes and Actinobacteria were significantly decreased in the "pre-cancer" lesions including mild, moderate and severe oral epithelial dysplasia compared to the healthy controls [41]. In agreement, Lee et al. found that the epithelial precursor lesions (e.g., dysplasia) had less abundant genera such as Bacillus, Enterococcus, Parvimonas, Peptostreptococcus and Slackia [22]. In contrast to these findings, Gopinath et al. showed that Megaspheara, unclassified Enterobacteriae, Prevotella, and Salmonella were more expressed in oral leukoplakia compared to healthy controls. Additionally, the authors reported a clear overlap between the whole mouth fluid bacteriome of leukoplakia and oral cancer [33]. Similarly, Hashimoto et al. found a significantly higher level of the genus Streptococcus in oral leukoplakia than in OSCC [30]. This was further supported by Ganly et al., who showed that Genera Fusobacterium and Veillonella were significantly increased in OPMDs [29]. Mok et al. demonstrated that phyla Firmicutes and Bacteroidetes had more OPMD related bacteria groups compared to healthy and cancer groups [42].

Oral Microbiota and OSCC
As the findings from OSCC studies are extensive, we will only present the statistically significant results. Recently, phyla of Actinobacteria and Cyanobacteria were decreased in OSCC (p = 0.021 and p = 0.013, respectively) compared to paracancerous tissue from the same patients [35]. Yang et al. showed that only Fusobacterium was increased while Streptococcus, Haemophilus, Porphyromonas and Actinomyces were all decreased (p < 0.0001) in OSCC compared to healthy controls [26]. In another recent study, genera Peptostreptococcus, Fusobacterium, Alloprevotella, and Capnocytophaga were all increased in contrast to decreased Rothia and Haemophilus (p < 0.05) in OSCC compared to non-cancer controls [31]. In agreement, OSCC patients had elevated levels of Peptostreptococcus, Fusobacterium, Alloprevotella and Capnocytophaga, while Rothia Streptococcus and Veillonella were all decreased (p < 0.05) [32]. Supporting these findings, genera Prevotella, Fusobacterium and Alloprevotella were increased in OCSCC (p = 0.019, p = 0.016 and p = 0.011, respectively) compared with normal control patients. Interestingly, these genera showed a trend to increase from healthy controls to OPMDs with their highest level seen in OSCC [29]. In contrast, Streptococcus was found decreased in OSCC [29]. Torralba et al. found a higher abundance of Prevotella in saliva from OSCC patients [34]. Hashimoto et al. demonstrated that phylum Bacteroidetes and genus Solobacterium had higher relative abundance in the OSCC group (p < 0.05) than patients with oral leukoplakia [30].
Granato et al. conducted a comparison of oral microbiota in OSCC patients before (L0) and after (L1) surgical excision [36]. Compared with the healthy controls, genera Abiotrophia, Acinetobacter, Alloscordovia, Dialister, Gemella, Granulicatella, Peptostreptococcus, Selenomonas, Staphylococcus, and Stenotrophomonas were considerably higher in both L0 and L1 patients. On the other hand, genera Veillonella, Rothia, Moryella, Kingella, and Centipeda were reduced in both groups. However, L1 patients had higher Alloscordovia and reduced Veillonella levels compared to L0 patients [36]. Very recently, Su et al. identified significant alterations in the bacterial diversity and relative abundance of specific oral microbiota with the most profound finding was the enrichment of Fusobacterium and the loss of Streptococcus in the OSCC [40]. Furthermore, authors suggested that genera Streptococcus, Fusobacterium, Peptostreptococcus, Campylobacter and species Streptococcus pneumoniae and F. Nucleatum (strain CTI-2) could be potential biomarkers for cancer patients [40]. The studied oral microbiota with their relative abundance in OSCC are depicted in Figure 3. with the most profound finding was the enrichment of Fusobacterium and the loss of St tococcus in the OSCC [40]. Furthermore, authors suggested that genera Streptococcus, F bacterium, Peptostreptococcus, Campylobacter and species Streptococcus pneumoniae an Nucleatum (strain CTI-2) could be potential biomarkers for cancer patients [40]. The st ied oral microbiota with their relative abundance in OSCC are depicted in Figure 3.

Oral Microbiota in Other Types of HNSCC
Debelius et al. explored the relationship between NPC and the oral microbiota us 16S rRNA sequencing of 499 NPC patients. They found that the overall microbial diver was lower in NPC patients compared to healthy controls (p < 0.001). They also identi a pair of Granulicatella adiacens amplicon sequence variants were strongly associated w NPC status [46]. Hayes et al. studied mouthwash samples from 129 HNSCC patients cluding cancers of pharynx, larynx and the oral cavity. Greater levels of genera Coryne terium, Kingella, Neisseria, Abiotrophia, Capnocytophaga and species Kingella denitrificans Streptococcus sanguinis were associated with a reduced risk for larynx cancer [48]. L wise, species Actinomyces oris and Veillonella denticariosi were associated with a redu risk of pharynx cancer. However, there were no bacterial genera associated with oral ity or pharynx cancer [48].
As tumor site was strongly associated with the T-stage, Wang et al. stratified t 121 samples of HNSCC based on location: oral cavity/oropharynx versus hypop ynx/larynx, phyla Actinobacteria was increased in low T-stage patients (p = 0.031), w genera Actinomyces and Parvimonas were not significantly changed [47]. In the grou oral cavity/oropharynx, Actinobacteria and Actinomyces approached significance in low In addition, a few species were shown to be increased in some studies and decreased in others (orange circles; orange double-headed arrow).

Oral Microbiota in Other Types of HNSCC
Debelius et al. explored the relationship between NPC and the oral microbiota using 16S rRNA sequencing of 499 NPC patients. They found that the overall microbial diversity was lower in NPC patients compared to healthy controls (p < 0.001). They also identified a pair of Granulicatella adiacens amplicon sequence variants were strongly associated with NPC status [46]. Hayes et al. studied mouthwash samples from 129 HNSCC patients including cancers of pharynx, larynx and the oral cavity. Greater levels of genera Corynebacterium, Kingella, Neisseria, Abiotrophia, Capnocytophaga and species Kingella denitrificans and Streptococcus sanguinis were associated with a reduced risk for larynx cancer [48]. Likewise, species Actinomyces oris and Veillonella denticariosi were associated with a reduced risk of pharynx cancer. However, there were no bacterial genera associated with oral cavity or pharynx cancer [48].
As tumor site was strongly associated with the T-stage, Wang et al. stratified their 121 samples of HNSCC based on location: oral cavity/oropharynx versus hypopharynx/larynx, phyla Actinobacteria was increased in low T-stage patients (p = 0.031), while genera Actinomyces and Parvimonas were not significantly changed [47]. In the group of oral cavity/oropharynx, Actinobacteria and Actinomyces approached significance in low T-stage patients relative to higher stages (p = 0.100, p = 0.192, respectively), while Parvimonas remained decreased among low T-stage patients (p = 0.006) [47]. In one study on GSCC, Katz et al. performed immunohistochemical staining to assess the presence of p. gingivalis and Streptococcus gordonii in tissue sections from GSCC. They showed a higher level of P. gingivalis (more than 33%, p < 0.05) in the carcinoma samples compared to normal gingiva. The staining intensity was also enhanced for P. gingivalis compared to specimens stained for the S. gordonii [44].

Oral Dysbiosis and Tumor Progression in HNSCC
A possible involvement of oral microbiota in dictating the progression of HNSCC was reported in seven studies [26,27,31,36,37,47,49]. For instance, Fusobacteria was increased during tumor development from stages 1 to 4 in patients with OSCC [26]. In contrast, Actinobacteria and Bacteroidetes were significantly decreased during cancer progression [26,47]. However, higher levels of Bacteroidetes have recently been linked to one of the mutational signature clusters associated with both late stages and larger sizes of OSCC [27]. Higher levels of Fusobacterium [49], Rothia [31], and Actinomyces [47] were associated with a lower T-stage, whereas Parvimonas was decreased in this stage [47] and amount of Peptostreptococcus positively related to a higher stage [49]. Veillonella was inversely correlated with clinical tumor size, lesion, and clinical stage of OSCC patients [36]. Recently, Neuzillet et al. reported that positivity of F. nucleatum was associated with a lower T-stage of OSCC. Furthermore, toll-like receptor 4 and the recruitment of M2-macrophages were both significantly decreased in tumors with high F. nucleatum load suggesting a better clinical outcome [37]. In another recent study on OSCC, Su et al. showed that oral dysbiosis can attenuate the production of anticancer metabolites such as the siderophore group nonribosomal peptides, monoterpenoid biosynthesis and others [40]. Katz et al. reported that tissue abundance of P. gingivalis was associated with the poorly differentiated GSCC, although it was not statistically significant [44].
The Operational Taxonomic Units (OTUs), using mothur's Bayesian classifier, were suggested as predictors of tumor metastasis. Schmidt et al. found that 11 OTUs from Actinomyces, Rothia and Streptococcus were associated with OSCC while only 1 OTU from Fusobacterium was linked to normal samples. This set of OTUs was analyzed and it separated most cancers from other samples with the greatest separation of node metastasis cases [41]. A higher level of Actinobacteria was associated with TP53 mutations, while Firmicutes was associated with recurrent mutations in key driver genes (FAT1, FZR1, and AXIN1) related to the Wnt pathway [27]. However, Ganly et al. found no association between bacterial taxa and tumor stage or metastasis [29].

The Prognostic Value of Oral Microbiota in HNSCC
A total of five studies reported prognostic data associated with oral microbiota (Table 3). Recently, Granato et al. demonstrated that a higher relative abundance of Stenophotromonas, Staphylococcus, Centipeda, Selenomonas, Alloscordovia, and Acinetobacter in saliva was associated with poor overall survival of OSCC patients [36]. In contrary, Veillonella relative abundance inversely correlates with clinical tumor size and clinical stage, suggesting a better prognosis for OSCC patients [36]. Neuzillet et al. concluded that OSCC patients with F. nucleatum-positive samples had longer relapse-free survival (median: 7.06 vs. 2.11 months, p = 0.0091) and metastasis-free survival (9.71 vs. 3.54 months, p = 0.0016) compared to F. nucleatum-negative tumors [37]. Consistently, a higher relative abundance of F. nucleatum in tumor tissue was correlated with a better 3-year diseasespecific survival and disease-free survival. This elevated status of F. nucleatum was also associated with non-smokers, lower tumor stage, lower rate of recurrence. Further, depletions of Neisseria, Haemophilus, and Rothia in HNSCC cases were associated with worse cancer-specific survival [49]. Robayo et al. analyzed the survival data of HPV-positive OSCC patients with those co-infected with HPV and Streptococcus anginosus. A tendency towards a poorer survival outcome was recognized for patients co-infected with both microorganisms, although it was not statistically significant [43]. As a novel finding, Capnocytophaga, a genus of gram-negative bacteria, was significantly increased in patients with recurrent OSCC after tumor resection, with a median abundance of 5.62-fold higher than in normal control patients [29].
To our knowledge, this is the first systematic review to compile and analyze an extensive amount of clinical data regarding the association between oral microbiota and different aspects of HNSCC. Although it was challenging to perform this study, yet it exhibits several strengths. Given the newness of this field and its limited experimental approaches, our comprehensive search, via multiple databases, yielded studies with a relatively large number of patients. Secondly, the study design was based on the PRISMA guidelines with a PROSPERO-registered protocol. Finally, the included studies were qualitatively appraised using both quality control and bias analysis tools. However, we acknowledge certain limitations. More than half of the eligible studies showed a tendency towards a reporting bias (i.e., moderate and high). This could be in part explained by the failure to perform a sufficient analysis of the patients' data. Additionally, prognostic studies were limited and the REMARK criteria were not always applied. Unfortunately, variations among the studies did not permit us to conduct a meta-analysis. Finally, our analysis was restricted only to data reported in English language; thus, this study may not reveal all the available knowledge in the field.
The analyzed microbiota was isolated from saliva, oral swabs, and/or tissue samples ( Figure 2). Although saliva was the most commonly used sample, there was a variation in the collection techniques including oral washes, stimulated and unstimulated saliva. These variations could subsequently affect the bacterial composition and hence the results' compatibility [6,7]. Furthermore, results inconsistency in onco-microbiome studies could arise from focusing on the compositional analysis alone, which does not account for functional redundancy in oral microbiota [25]. Tissue samples were utilized only in a few studies with a smaller cohort size as compared to studies based on salivary analysis. Noteworthy, the controls were not always healthy or cancer-free individuals in these studies. In a recent OSCC study, for instance, the negative controls comprised patients with thyroid nodules (benign and malignant). However, these patients were considered representative of normal population as they showed no evidence of oral cavity pathology following close clinical examinations [29].
Oral microbiota were shown to be associated with cancers other than HNSCC including lung, colorectal and pancreatic cancers [9,11]. Importantly, carcinogenesis has recently been linked to periodontitis-a chronic inflammation largely mediated by oral dysbiosis [10]. In a recent meta-analysis, periodontitis and periodontal bacteria were associated with an increased incidence of cancer and poor survival rates. Interestingly, authors found that a higher cancer risk was associated with P. gingivalis and P. intermedia but not with F. nucleatum, Tannerella forsythia, Treponema denticola or Aggregatibacter actinomycetemcomitans [10]. Fusobacterium, specifically F. nucleatum, has a strong association to the tumorigenesis of colorectal cancer [52][53][54]. In agreement with this evidence, our review demonstrated that genus Fusobacterium [26,29,31,32,40] and F. nucleatum [24,32,37,40] are connected to HNSCC. However, it was also proposed that abundance of Fusobacterium could have a favorable effect on HNSCC progression and survival [37,49]. Another wellstudied species is P. gingivalis, an anaerobic bacteria that has been connected among others to pancreatic cancer [55,56] and OSCC [57][58][59]. Noteworthy, only one study showed a statistically significant evidence of the association between P. gingivalis and HNSCC using immunostaining on tissue samples [44]. This finding raises the question whether stimulated/unstimulated saliva, swabs or tissue samples would represent the most reliable method for analyzing oral microbiota in cancer patients. Data are, however, conflicting in this regard. While unstimulated saliva was considered inferior to stimulated saliva [60,61], another study showed that there are no major differences in their reliability [62].
The utility of oral microbiota as biomarkers in HNSCC has also been elucidated. Lee et al. proposed that differences in the abundance of genera Bacillus, Parvimonas, Peptostreptococcus and Slackia could be used as a marker for the prediction, detection, and prognosis of patients with OSCC [22]. Su et al. had results supporting that the profusion of F. nucleatum (strain CTI-2) and a decreased abundance of S. pneumoniae could distinguish cancers from healthy controls [40]. Indeed, a prediction tool for metastasis would be crucial since locoregional metastasis can drastically worsen the prognosis of HNSCC patients [63]. In this context, it was suggested that certain OTUs can separate node-positive cases from the negative ones in patients with OSCC [41]. Additionally, studies indicated that oral microbiota can substantially change through the course from potentially malignant lesions to cancer [22,29,33]. This finding could have a clinical utility in future. Currently, the follow-up of OPMDs is based on the clinical appearance of the lesion and, if needed, a surgical biopsy. Thus, obtaining saliva to follow up any changes in the microbiota, together with the clinical inspection, could provide clinicians with a simple, non-invasive, approach to early diagnosing malignant changes of OSCC.
The potential role of oral microbiota in modifying the immune response in tumor microenvironment (TME) has been studied. For instance, it has been shown that infection with F. nucleatum enhanced M2 polarization of macrophages through TLR-4 activation, which increased tumor growth in colorectal cancer [64]. In this regard, higher levels of M2 macrophages showed poorer prognosis and clinical outcomes in HNSCC patients [65,66]. Controversially, one study reported that high levels of F. nucleatum was associated with decreased recruitment of M2-macrophages and low TLR4 signaling and lower T-stage [37]. Patients with HNSCC are often given antibiotics during the course of treatment. In the included studies, the oral microbiota samples were collected either prior to the treatment or without reporting which antibiotics were given. Thus, a detailed follow-up regarding the change in oral microbiota due to the administration of antibiotics and their possible effect on cancer progression is needed in the future studies.

Conclusions
This systematic review asserts the association of oral microbiota to the tumorigenesis of HNSCC. Microbial dysbiosis is evident in HNSCC patients and several bacterial genera and species seem to affect HNSCC progression, metastasis, recurrence, and/or survival. However, it remains uncertain exactly which genera, species or bacteria combinations are truly significant. Therefore, we encourage further research in this newly emerging area, which could lead to the development of effective diagnostic and prognostic targets and even therapeutic measures for patients with HNSCC.
Supplementary Materials: The following are available online at https://www.mdpi.com/article/ 10.3390/cancers14010046/s1, Table S1: The inclusion and exclusion criteria applied in this study, Table S2: Evaluation criteria used to assess the reporting quality of the included studies, Table S3: The risk of bias and reporting quality of the included studies.

Conflicts of Interest:
The authors declare no conflict of interest.