Differences in the landscape of colonized microorganisms in different oral potentially malignant disorders and squamous cell carcinoma: a multi-group comparative study

Abstract

Background

The role of microbes in diseases, especially cancer, has garnered significant attention. However, research on the oral microbiota in oral potentially malignant disorders (OPMDs) remains limited. Our study investigates microbial communities in OPMDs.

Materials and methods

Oral biopsies from19 oral leukoplakia (OLK) patients, 19 proliferative verrucous leukoplakia (PVL) patients, 19 oral lichen planus (OLP) patients, and 19 oral lichenoid lesions (OLL) patients were obtained. 15 SCC specimens were also collected from PVL patients. Healthy individuals served as controls, and DNA was extracted from their paraffin-embedded tissues. 2bRAD-M sequencing generated taxonomic profiles. Alpha and beta diversity analyses, along with Linear Discriminant Analysis effect size analysis, were conducted.

Results

Our results showed the microbial richness and diversity were significantly different among groups, with PVL-SCC resembling controls, while OLK exhibited the highest richness. Each disease group displayed unique microbial compositions, with distinct dominant bacterial species. Noteworthy alterations during PVL-SCC progression included a decline in Fusobacterium periodonticum and an elevation in Prevotella oris.

Conclusions

Different disease groups exhibited distinct dominant bacterial species and microbial compositions. These findings offer promise in elucidating the underlying mechanisms of this disease.

Introduction

In recent decades, our understanding of the complex interplay between microorganisms and diseases has markedly advanced. Notably, there is a substantial correlation observed between the occurrence of specific cancers and the presence of particular microorganisms. Helicobacter pylori was the pioneering microorganism linked to malignant tumors, followed by discoveries concerning the association between Epstein-Barr virus and nasopharyngeal carcinoma, as well as Fusobacterium and colorectal cancer [1, 2].

The oral cavity, second only to the gut, harbors the second-largest reservoir of microorganisms. Influenced by diverse lifestyle factors such as smoking, alcohol consumption, and dietary habits, the oral microbiota undergoes dynamic changes [3, 4]. The anatomical structure of the oral cavity provides favorable habitats and essential nutrients for microorganisms. Furthermore, various oral regions, including the hard palate, soft palate, gingiva, posterior tongue, oral floor, and tooth surfaces, exhibit distinct bacterial characteristics [5].

Notably, the oral microbiota has been implicated in a range of diseases. Research has linked it to cancers such as esophageal, hepatic, gastric, and oral cancers [6]. Additionally, the oral microbiota might play a role in diseases like Alzheimer’s disease and cardiovascular diseases [7]. Recent technological advances have shifted our attention to exploring microbial colonization within tissues, whereas earlier research on the oral microbiota primarily focused on saliva [8]. Microbes residing within tissues are more likely to be involved in the pathogenesis of diseases. For instance, the microbial community within tumors influences tumor progression by mechanisms such as DNA damage in gastrointestinal cancers, activation of carcinogenic pathways, and induction of immune suppression in pancreatic ductal adenocarcinoma [9,10,11].

Currently, research on the relationship between the oral microbiota and oral potentially malignant disorders (OPMDs) lags behind that of caries and periodontal diseases. The potential role of microbial dysbiosis in OPMDs remains largely unknown. According to a recent international consensus, OPMDs are defined as any abnormalities of the oral mucosa associated with an increased risk of oral cancer. Leukoplakia, the most prevalent type of OPMD, exhibits an annual malignant transformation rate to oral squamous cell carcinoma (OSCC) ranging from 7 to 10% [12, 13]. Conditions like proliferative verrucous leukoplakia (PVL), previously classified as a type of oral leukoplakia, are now separately categorized as OPMDs by the WHO due to their higher malignant transformation rates (43–66%) and unique clinical and pathological features [14, 15].

Oral lichen planus (OLP), considered a T-cell-mediated inflammatory disease, still lacks clarity regarding the factors triggering T-cell inflammatory responses, despite consideration of various antigens [16]. From a pathological perspective, oral lichenoid lesions (OLL), similar to OLP, are classified as low-malignancy OPMDs [17]. However, they diverge in pathogenesis and treatment approaches, emphasizing the need to examine discrepancies in microbial alterations. Despite distinct pathogenic mechanisms, the connection between diseases and microbial equilibrium remains elusive.

Presently, predominant methodologies for investigating OPMDs encompass 16 S rRNA gene sequencing and metagenomic sequencing to elucidate the microbial community composition within tissues. Nevertheless, challenges persist in conditions necessitating retrospective diagnosis, such as PVL, due to sampling complexities. Furthermore, the scarcity of certain OPMDs subtypes poses constraints on scrutinizing the interplay between microbial communities and diseases owing to limited sample accessibility.

Meanwhile, hurdles endure in detecting microbes in formalin-fixed paraffin-embedded (FFEP) tissues, underscoring the necessity of introducing innovative microbial exploration methods, such as 2bRAD sequencing for microbiome (2bRAD-M). This technique excels in handling severely degraded DNA tissues, including FFEP samples, and proficiently delineates low-biomass microbial communities at the species level while upholding precision and cost-effectiveness [18].

In this study, we curated four subcategories of OPMDs: OLK, OLP, OLL, and PVL, alongside a collection of tumor samples originating from PVL progression, labeled as PVL-SCC, for our research cohort. Our objective was to juxtapose the microbial community architectures of diverse OPMDs and scrutinize PVL samples along with their associated tumors. To our knowledge, our study represents the first use of FFEP tissues and includes various types of OPMDs for microbial analysis.

Methods

Recruitment of patients and volunteers

This study was conducted in accordance with the Declaration of Helsinki and received approval from the Ethics Committee of Peking University Hospital of Stomatology (PKUSSIRB-2023-02-85-02). According to the WHO 2020 consensus standards for OPMD, we selected four groups of 19 patients each (OLK, OLP, OLL, PVL) based on clinical-pathological diagnostic criteria from 2014 to 2020 at Peking University Hospital of Stomatology [19]. The diagnostic criteria for PVL are based on the standards proposed by Irene et al. in 2022, which include: (1) the presence of multiple leukoplakias at more than two different sites of the oral mucosa, (2) clinically homogeneous and/or nonhomogeneous leukoplakias (verrucous, erythroleukoplakia, speckled, nodular), (3) leukoplakias that proliferate, grow, and spread during their evolution, and (4) recurrent leukoplakias in previously treated areas [14]. All participants in our study had not received antibiotic treatment in the six months prior to sampling. Pathological diagnoses were performed by two experienced pathologists.

Age distribution was consistent across all groups (ranging from 50 to 76 years old). Each group included one smoker, one drinker, and one patient who both smoked and drank. The PVL-SCC samples were obtained from PVL patients who had undergone malignant transformation. We included 15 SCC biopsies (PVL-SCC) from these patients in our study. Detailed patient information was provided in Supplemental Table 1. Each patient underwent precise diagnosis and presented complete clinical records along with FFEP samples. For OLK, OLL, OLP, and PVL-SCC groups, one biopsy was collected per patient. In the case of PVL, two distinct regions were sampled from each patient, except for one instance (P63), where only one lesion was available. FFEP samples were utilized for this portion of the study.

For the control group, we recruited individuals who had undergone wisdom tooth extraction without any prior history of OPMDs or SCC. We collected ordinary mucosal tissue surrounding the extraction site of the third molar, embedding it in paraffin to serve as the disease control, totaling 11 healthy donors. Fresh FFEP samples were used for the control group.

DNA extraction, Library preparation, and sequencing

Paraffin tissues from 120 samples were utilized, with each paraffin block cut into 6–8 sections approximately 8 micrometers thick. These sections were subsequently placed in 1.5 ml Eppendorf tubes for DNA extraction using the Upure FFPE Tissue DNA Kit (Qiagen). The methodology for establishing the 2bRAD-M library was previously described [18]. Briefly, genomic DNA (200 ng) was digested with 4 U of BcgI enzyme (NEB) at 37 °C for 3 h, followed by overnight ligation to adaptors (Ada1, Ada2) using T4 DNA ligase (NEB, America) at 4 °C. The ligated products underwent PCR amplification with Phusion High Fidelity DNA polymerase (NEB, America). The resulting PCR products were separated on an 8% polyacrylamide gel, and a 100-base pair band was excised. DNA was allowed to elute into nuclease-free water at 4 °C over 12 h. PCR products were purified using the QIQuick PCR purification kit (Qiagen) and subsequently sequenced on the Illumina Nova PE150 platform by Qingdao OE Biotechnology Co., LTD.

Sequencing processing and quantitative analysis

The 2bRAD-M library database encompasses a total of 173,165 microbial genes, encompassing bacteria, fungi, and archaea, each annotated with a distinctive 2bRAD tag label under a specific taxon. Firstly, mapping of all sequenced 2bRAD tags after quality control (QC) to the constructed 2bRAD marker database. In order to control false positives, the G score was calculated for each species using the formula shown below: $\:{G\:score}_{species\:i}$= $\:\sqrt{{S}_{i}\times\:{t}_{i}}$ (where S stands for the number of reads assigned to all 2bRAD markers attributed to $\:species\:i$ within a sample, and t represents the count of all 2bRAD markers of $\:species\:i$ that have been sequenced within that sample).To identify false positives in the discovery of microbial species, the threshold for the G score was established at 5. Subsequently, the relative abundance of each species in the sample was calculated using the following formula: $\:{Relative\:abundance}_{species\:i}$=$\:\frac{{S}_{i}/{T}_{i}}{{\sum\:}_{i=1}^{n}{S}_{i}/{T}_{i}}$ (where S denotes the number of reads assigned to all 2bRAD markers of $\:species\:i$ within a sample, and T represents the total count of theoretical 2bRAD markers associated with $\:pecies\:i$). Using FEAST (version: 0.1.0), decontam (version: 1.10.0), and microDecon (version: 1.0.2) to identify and remove contaminating species from the microbial community. FEAST employs a maximum likelihood algorithm to predict the proportion of each species in experimental samples originating from user-defined “Source” environmental samples. If decontam or microDecon identifies a species as a contaminant and FEAST calculates that its probability of originating from each “Source” source is greater than 0, removal is conducted based on the probability ratio [20,21,22]. Following this process, we recalculated the relative abundance of species composition based on the post-removal outcomes. (Supplemental Table 2)

Microbial diversity analysis and statistical analysis

The statistical analysis was performed using IBM SPSS Statistics 25.0 (IBM Corp., Armonk, NY, USA) and R 4.1.1. Visualization tools such as box plots, Venn plots, bar plots, and PCoA plots were generated using R 4.1.1 with the ggplot2 package. Alpha diversity within groups was assessed using the Kruskal-Wallis test for chao1 (species richness) and Shannon index (species diversity) with the “vegan” package (version 2.6.4). Between-group comparisons were conducted using the Wilcoxon rank-sum test. Beta diversity analysis utilized Bray-Curtis distance, Binary Jaccard distance, and Euclidean distance algorithms also implemented with the “vegan” package. These results were visualized using principal coordinate analysis (PCoA). To identify taxa with differential representation between groups, the Linear Discriminant Analysis Effect Size (LEfSe) technique was employed (LEFSe software version 1.0). Paired analysis of PVL and PVL-SCC groups was conducted using the paired T-test. Statistical significance was defined as p < 0.05.

Result

Biodiversity of OPMDs Microbiota

The quality control information for sequencing was presented in Supplemental Table 3, showing raw reads of 10.55 million and clean reads of 8.70 million per sample across 120 samples (Supplemental Table 3). To investigate the diversity and richness of microbial communities in these diseases, we initially analyzed six groups. Regarding alpha diversity, as indicated by the chao1 (p = 0.049) and shannon (p = 0.003) indices, discernible differences in species richness and diversity were observed among the six groups (Fig. 1. a-b). Specifically, the PVL-SCC group exhibited the lowest richness and diversity among the disease groups, similar to that of the control group. Conversely, OLK exhibited significantly higher richness and diversity compared to the control group. Noteworthy differences were also observed among various OPMDs. We have identified within-group comparisons with statistically significant differences in the graph. However, even if there are no statistical differences between certain groups, we can observe a comparative trend in microbial species richness and diversity. The control group had lower levels compared to of OPMDs groups, while compared carcinoma group with PVL or OLK groups, there was a decreasing trend in microbial species and quantities, consistent with the trend results obtained by Hu et al. [23]. OLK is higher in biodiversity than OLL (p = 0.008) and higher in richness than PVL (p < 0.001). Additionally, OLP has a higher abundance than PVL (p = 0.004).

Fig. 1

Comparison of microbial diversity and structure among disease groups and the control group. (a, b) Evaluation of alpha diversity (Chao1, Shannon index) among the six groups (Control, OLK, OLL, OLP, PVL, PVL-SCC). (c) Assessment of beta diversity among the six groups using PCoA. (d) Beta diversity between different groupings

Subsequently, we conducted beta diversity analysis of the six groups, revealing significant differences between disease and control groups (p = 0.001). However, distinguishing between different disease groups within the same plot was challenging (Fig. 1.c). Therefore, we conducted a more detailed analysis (Fig. 1.d). Initially, dimensionality reduction analysis on the four OPMDs showed a significant distinction between the OLL group and the other disease groups (p = 0.001). In exploring diseases that are challenging to differentiate pathologically in clinical practice, we analyzed various groups and found a significant distinction between OLL and OLP (p = 0.032), while no significant difference was observed between PVL and OLK (p = 0.085). Additionally, we analyzed microbial samples from PVL patients and those post-cancer transformation, revealing a significant distinction between the two (p = 0.003).

Microbial Community Composition in different groups

Initially, substantial variations in genus composition were observed among the five disease groups and the control group (Fig. 2.a). Notably, the dominant genera in the control group were Porphyromonas (21.24%), Actinomyces (16.56%), and Tannerella (9.53%). However, the predominant genera showed both similarities and differences across different OPMDs. Among the four OPMDs, Pseudomonas consistently emerged as the dominant genus, with abundances exceeding 20%. Streptococcus showed relatively high abundance in OLK, OLP, and PVL, accounting for 8.41%, 12.32%, and 6.31%, respectively. Paracoccus also demonstrated relatively high abundance in OLK, OLP, and OLL, accounting for 5.65%, 8.98%, and 11.24%, respectively. Noteworthy differences were observed, particularly in OLK, where Meiothermus (4.31%), Neisseria (4.22%), and Actinomyces (4.19%) were notable. In OLP, Rothia exhibited relatively high abundance (7.05%). OLL showed significant differences from the other OPMDs, with high abundance of Cupriavidus (5.44%), Candida (5.14%), and Roseomonas (4.57%). Additionally, in PVL, other genera with relatively high abundance included Haemophilus (4.70%), Campylobacter (3.91%), and Rothia (3.78%). In PVL-SCC, the top five genera were Pseudomonas (13.96%), Prevotella (9.54%), Capnocytophaga (7.07%), Neisseria (6.19%), and Streptococcus (5.15%) (Supplemental Table 4).

Fig. 2

Bacterial abundance and distribution in groups. (a) Relative abundance of the top 30 most abundant genera in six groups. (b) Relative abundance of the top 30 most abundant species in six groups. (c) Venn diagram showing shared and unique species in OLK, OLL, OLP, and PVL. (d) Heatmap displaying relative abundance of the top 15 most abundant species in OLK, OLL, OLP, and PVL. (e) Venn diagram showing shared and unique species between PVL and PVL-SCC groups. (f) Heatmap displaying relative abundance of the top 15 most abundant species between the PVL and PVL-SCC groups.

When examining the top 30 microbial species, distinct representatives emerged within each group (Fig. 2.b). In the control cohort, Porphyromonas gingivalis (20.20%), Tannerella forsythia (9.01%), and Corynebacterium matruchitii (5.52%) were the most prevalent. Fusobacterium nucleatum (2.5%) and Prevotella intermedia (0.45%) were also notable. Conversely, these species showed relatively lower proportions within the disease groups. Comparative analysis between the control and various disease cohorts revealed notable elevations in specific bacterial levels within the latter. In OLK, Streptococcus pneumoniae (6.3%) and Pseudomonas fluorescens (5.4%) dominated, whereas their presence was minimal in the control group (P. fluorescens at 0.03% and S. pneumoniae undetected). In OLL, both Cupriavidus metallicurans and Candida albicans constituted substantial proportions, surpassing 5%, whereas only a minute amount of C. albicans (0.002%) was detected in the control group. S. pneumoniae also prevailed in OLP, closely trailed by Pseudomonas extremaustralis (6.8%) and Rothia mucilaginosa (4.6%). Notably, predominant bacteria in the PVL group (P. extremaustralis and Pseudomonas veronii) were absent in the control group. Conversely, the tumor group exhibited elevated levels of P. extremaustralis (5.33%), absent in the control group, while P. intermedia (3.4%) and F. nucleatum (3.2%) were also detected in the control group, albeit at lower proportions of 0.04% and 2.5%, respectively (Supplemental Table 5).

To facilitate comprehensive comparisons and visualizations of diverse diseases, we conducted Venn plot and heatmap analyses, delineating bacterial dominance across four OPMDs. At the species level, we identified 712 shared species among the four OPMDs, with numerous unique species specific to each disease (Fig. 2.c). Heatmaps were employed to compare the proportions of microbial communities in different groups (Fig. 2.d). Notably, in OLK, Meiothermus silvanus accounted for 4.11%, exhibiting the highest proportion among the four groups. In OLP, S. pneumoniae and R. mucilaginosa were dominant species. C. albicans showed a significantly higher proportion in OLL compared to the other groups. In PVL, several species of Pseudomonas (P. extremaustralis, P. fluorescens, P. veronii) were predominant, alongside Haemophilus parainfluenzae and Campylobacter concisus. Similar analyses were conducted for the PVL group and PVL-SCC group (Fig. 2.e-f). The PVL group exhibited the highest number of detected species, with 907 species shared with the PVL-SCC group and 1314 unique species. The top 15 dominant bacteria in both groups are shown in the heatmap.

Differential abundances of bacterial taxa in different groups

To identify statistically significant biomarkers among diverse groups, we employed the LEfSe technique. Initially, we applied the Kruskal-Wallis rank-sum test to multi-group samples to identify species abundances significantly distinct across various groups. Subsequently, using the species identified as significant from the preceding step, we conducted inter-group difference analysis employing a paired Wilcoxon rank-sum test. Ultimately, Linear Discriminant Analysis (LDA) was employed with a threshold greater than 3.5 for data dimensionality reduction to evaluate the influence of these notably distinct species.

In the differential analysis across the six groups, we observed that the majority of distinct bacteria in different groups, especially in the control group, were representative species of the respective diseases, such as P. gingivalis, T. forsythia, and Corynebacterium matruchotii (Fig. 3.a). In multi-group comparisons, each group often needed to be compared with several others to identify species distinguishing that disease. Therefore, we showcased the most unique bacteria in the comparison across the six groups and depicted their true distribution. For instance, in the comparison across the six groups, the top two species in the PVL-SCC group were Capnocytophaga sputigena and Alloprevotella tannerae, while in the PVL group were Pseudomonas koreensis and Actinomyces israelii, in the OLP group S. pneumoniae, in the OLL group Cupriavidus metallidurans and C. albicans dominated, and in the OLK group, M. silvanus was predominant (Fig. 3.a-b and Supplemental Table 4).

Fig. 3

Differential abundances of bacterial taxa among different groups. (a) Histogram illustrating the taxonomic hierarchical structure of discriminative biomarkers across six groups. (b) Expression profiles of ten distinct species. (c) Histogram illustrating the taxonomic hierarchical structure of discriminative biomarkers between OLP and OLL groups. (d) Histogram illustrating the taxonomic hierarchical structure of discriminative biomarkers between OLK and PVL groups. (e) Histogram illustrating the taxonomic hierarchical structure of discriminative biomarkers between PVL and PVL-SCC groups.

To better explore differences in microbiota between clinically related diseases, we compared two pathologies, OLL and OLP, which cannot be distinguished based solely on histology but rather require differentiation based on medical history. Our analysis revealed a high detection rate of C. albicans in OLL, whereas S. pneumoniae was notably prevalent in OLP (Fig. 3.c). Further investigation included PAS staining on samples from OLL patients positive for C. albicans, revealing conspicuous hyphae under the microscope in all three patients (P20, P24, P33) with high detection rates (Supplemental Fig. 1).

PVL, characterized by higher malignancy risk compared to OLK, exhibited significant differences. M. silvanus and Micrococcus luteus were more prevalent in OLK, while A. israelii dominated in PVL (Fig. 3.d). Lastly, comparing PVL with its subsequent cancerous transformation, PVL-SCC, we identified significant differences, with distinct bacterial profiles remaining predominant in each condition (Fig. 3.e).

The microbial pairwise analysis of PVL and cancerous transformation PVL-SCC group

Taking into account the unique diagnostic characteristics of PVL patients [24], we initially conducted beta analysis on two sets of lesion samples from 19 PVL patients to assess their heterogeneity (Supplemental Fig. 2). Our analysis revealed significant differences among different patients, as well as substantial variations within the same patient. Subsequently, to delve deeper into microbial community changes before and after PVL carcinogenesis, we focused on 12 patients for paired analysis of microbial changes at identical anatomical sites before and after the onset of cancer (Fig. 4.a). Following the paired analysis, we observed significant trends at the genus level, with Capnocytophaga and Porphyromonas exhibiting a noteworthy downward trend, while no obvious genera showed an upward trend (Fig. 4. b, d).

Fig. 4

Microbial variation in paired PVL patients. (a) Relative abundance of the top 30 most abundant species in 12 paired PVL patients (24 cases in total). (b) Significant genera in paired samples. (c) Significant species in paired samples. (d) Variation in two genera and six species during disease progression.

At the species level, transitioning from PVL to cancer was accompanied by a significant decrease in Deinococcus sp S9, Fusobacterium periodonticum, and Roseomonas rosea levels. In contrast, there was a notable increase in Capnocytophaga leadbetteri, Capnocytophaga sp. oral taxon 329 str. F0087, and Prevotella oris levels (Fig. 4.c, d). Although we cannot definitively ascertain whether the disease influenced microbial composition or vice versa, our pairwise comparison suggests that these identified taxa may indeed play crucial roles in disease progression, with significant implications. Particularly striking is the decrease in F. periodonticum and the rise in P. oris, highlighting their potential significance in elucidating the dynamics of malignancy progression.

Discussion

The correlation between microorganisms and OPMDs has gained increasing attention in recent years. However, our understanding of the role of bacteria in the progression of these prevalent oral diseases remains limited. This study aims to elucidate the complexity of microbial communities in OPMDs, highlighting correlations between changes in microbial enrichment and disease progression. Previous reports indicate that smoking consumption can significantly influence the composition of oral microbial communities. Smoking, in particular, disrupts oral biofilm structure, potentially leading to unstable colonization compared to non-smokers [25]. Furthermore, smokers may experience dysregulation of innate and adaptive immune responses, thereby increasing susceptibility to bacterial infections [26]. One study found that a slightly reduced presence of an unclassified species within the Capnocytophaga genus was observed in the oral microbiomes of frequent drinkers. However, despite changes in microbial composition, alterations in microbial metabolism were not associated with drinking status [3]. To mitigate these variables, we employed a paired approach to ensure consistent proportions of tobacco and alcohol use among patients in each disease group.

Recent research has increasingly focused on understanding the colonization of microbial communities associated with diseases. Unlike transient microorganisms found in saliva, those residing within tissues are more likely to play a direct role in pathogenicity or disease progression [8]. Microbes within tissues can influence disease development by inducing inflammatory responses, activating carcinogenic pathways, and modulating cytokine production to facilitate tumor progression through immune suppression. Notably, the role of Fusobacterium nucleatum in colorectal cancer is well-documented. Studies have shown that F. nucleatum activates the TLR4 signaling pathway, influencing microRNA levels, regulating autophagy, and promoting chemotherapy resistance [27]. Additionally, recent research by Kong et al. demonstrated that F. nucleatum promotes invasion and metastasis of colorectal cancer by activating the TLR4/Keap1/NRF2 signaling pathway, enhancing levels of CYP2J2 and 12,13-EpOME [28]. Moreover, microorganisms within tumors can metabolize anti-tumor drugs, alter anti-tumor immunity, and affect cancer treatment efficacy, thereby influencing both tumor occurrence and therapeutic outcomes [29]. These microorganisms may also impact checkpoint proteins and the immune microenvironment, with studies showing that F. nucleatum directly binds to the checkpoint protein TIGIT, inhibiting the anti-tumor activity of NK and T cells [30].

Pathogenic bacteria commonly associated with periodontal disease have also been implicated in oral cancer. In our study of the PVL-SCC sample group, Prevotella intermedia, Fusobacterium nucleatum, and Porphyromonas gingivalis were equally prevalent. Experimental evidence has demonstrated synergistic pathogenicity between P. gingivalis and F. nucleatum in an in vivo oral cancer model [31]. Furthermore, recent findings have shown enrichment of F. nucleatum and P. intermedia in patients with nasopharyngeal carcinoma, suggesting the potential roles of these bacteria in tumorigenesis across different anatomical sites [32]. These findings underscore the importance of exploring the contributions of these bacteria to tumor development.

Research on microbial community changes in various OPMDs remains relatively scarce, with many studies focusing on phylum or genus-level analyses rather than species-specific investigations, leading to disparate findings [33]. For instance, Amer et al. noted enrichment of Fusobacterium, Leptotrichia, Campylobacter, and Rothia species in the oral mucosa of patients with OLK [34]. Similarly, our study also observed enrichment of Fusobacterium and Rothia in OLK. Interestingly, Pseudomonas, Neisseria, and Streptococcus were notably abundant, consistent with findings by Hu et al. in saliva samples from OLK patients [23]. Additionally, both our study and Amer et al. found increased abundance of R. mucilaginosa in OLK, a bacterium known to convert ethanol into acetaldehyde, potentially inducing oxidative stress [35]. The specific role of R. mucilaginosa in OPMDs patients warrants further investigation, especially since its concentration does not correlate with alcohol consumption.

In our investigation of OLP, our microbial analysis results align with previous studies. A study utilizing 16 S rRNA sequencing on mucosal samples from OLP patients also identified Streptococcus, Haemophilus, and Neisseria as dominant genera, with Streptococcus predominant in our findings, alongside Paracoccus and Rothia [36]. Bornstein et al. conducted a study comparing the colonization of 74 bacterial species on mucosa with asymptomatic OLP lesions and healthy sites of OLP patients, as well as healthy subjects, using the checkerboard DNA-DNA hybridization method [37]. However, our study detected these bacteria variably across samples, highlighting methodological influences on detection rates. Notably, Streptococcus pneumoniae was highly prevalent in both OLP and OLK in our study, a phenomenon requiring further investigation to validate its significance. In the latest review, a systematic review of the microbiome composition of OLP and OLK was conducted. The conclusion drawn was that for OLP, the bacteria with the highest abundance were Fusobacterium, Capnocytophaga, Gemella, Granulicatella, Porphyromonas, and Rothia, while for OLK, Prevotella was predominant [38]. These insights underscore the complex microbial dynamics within OPMDs and emphasize the need for standardized methodologies to elucidate their roles in disease progression.

Currently, there is limited literature on microbial communities associated with OLL. Despite sharing pathological similarities with OLP, OLL exhibits a higher transformation rate to cancer compared to OLP [17]. Common features include hyperkeratosis, incomplete keratinization, microabscesses, varying degrees of mucosal inflammation ranging from mild erythema to severe ulceration, and a mononuclear cell infiltrate composed of activated T lymphocytes, macrophages, and dendritic cells, suggesting a link between OLP and OLL. In our study, we identified an overlap of 850 microbial species between OLP and OLL. However, significant differences in species composition exist between these two conditions. Notably, OLL shows a higher incidence of Candida albicans infection, consistent with its association with white plaque diseases. Risk factors for oral candidiasis include frequent use of corticosteroids (especially inhalation aerosols for asthma), antibiotic use, and oral mucosal disorders like OLP [39]. Sarkate et al. tested for C. albicans infection in OLP and OLL patients using a germ tube test and found positivity in 40% of patients [40]. The involvement of C. albicans-induced inflammation in white plaque diseases likely stems from its filamentous development, a critical virulence factor causing host cell damage. Epithelial cells act as the primary barrier against C. albicans invasion, triggering the production of cytokines such as interleukins (IL)-1α, IL-1β, IL-6, and IL-17 through activation of proteins like Activator Protein-1 (AP-1), c-Fos, and Mitogen-activated Protein Kinase Phosphatase-1 (MKP1), subsequently recruiting immune cells like macrophages [41]. Further microbial studies on OLL are crucial for enhancing our understanding of its pathogenesis and potential role in carcinogenesis, emphasizing the importance of maintaining oral microbial balance in OLL patients.

Research on the microbial aspects of PVL is more limited compared to that of OLL, with only a few studies focusing on PVL. This scarcity is mainly due to challenges in understanding the risk of OSCC progression from PVL and difficulties in sample collection that have hindered research progress. A study on PVL identified specific microbial enrichments in PVL tissues, including Oribacterium sp. oral taxon 108, Campylobacter jejuni, uncultured Eubacterium sp., Campylobacter, Tannerella, Porphyromonas, and Peptoniphilus [42]. Another study comparing microbial differences among leukoplakia, PVL, and SCC arising from PVL highlighted Prevotella salivae and C. concisus as predominant species in PVL [43]. In our findings, we also observed enrichment of Campylobacter and Porphyromonas in PVL, with C. concisus showing significantly increased relative abundance. Reports indicate that heightened C. concisus levels are notably associated with marked hyperplasia in oral leukoplakia patients. A case control study involving 25 cases of OSCC and 27 cases of fibrous epithelial polyps concluded that the abundance of C. concisus in OSCC was higher than in fibrous epithelial polyps [44]. This bacterium was also enriched in the tumor group in our study, suggesting a potential significant association with tumorigenesis. Pseudomonas presence and its association with oral cancer, along with its impact on oxygen consumption index, have also been documented. This suggests that Pseudomonas might induce a hypoxic environment post-oxygen consumption, facilitating anaerobic bacterial proliferation and accumulation of virulence factors that could promote carcinogenesis. Upon pairing PVL with SCC among patients, we observed a decrease in F. periodonticum and an increase in P.oris, which deserves our attention. F. periodonticum is associated with various diseases and exhibits varying degrees of aggregation in oral mucosal diseases such as leukoplakia and OLP, as well as in periodontitis. While its association with other tumors has been reported, its connection to oral cancer is less established. We speculate that its decrease may be more related to the disease itself rather than the disease progression. Meanwhile, the significant increase in P.oris, potentially related to biofilm production, suggests its possible association with carcinogenesis.

Different sampling methods in oral microbiome studies often yield disparate results, highlighting the absence of a standardized approach [45]. However, regarding oral mucosal sampling, information obtained from swabs, punctures, and biopsies typically depicts similar microbial communities. Comparing different sampling and detection techniques poses a challenge in various studies. For example, membrane sampling reveals an increased proportion of genus such as Fusobacterium, Leptotrichia, Lautropia. Conversely, saliva samples from OLP patients show an increase in the proportion of Prevotella melaninogenica [46].

In our selected control group, we detected a relatively higher abundance of bacteria associated with periodontal disease, including F. nucleatum. Recent reports highlight Treponema, Porphyromonas, Parvimonas, and Fusobacterium as core microbial communities in periodontitis [47]. This may be attributed to sampling mucosal tissue around third molars, often associated with deep periodontal pockets and a microbial composition akin to inflammation. Nonetheless, this finding underscores the robustness of our novel sequencing method, 2bRAD-M, in our study.

Detecting microbes in FFPE tissues presents challenges such as formalin-induced DNA damage and degradation over time. The 2bRAD-M technique mitigates these issues using BcgI enzyme digestion to amplify short DNA fragments, effectively managing DNA degradation in paraffin samples [18]. While innovative, this approach has limitations—it struggles with predicting microbial functional pathways compared to metagenomic sequencing and lacks resolution in strain-level identification [48]. Moreover, it does not detect or analyze viruses.

There are several limitations in our study. Our study did not encompass all types of OPMDs and had a limited sample size. Given the challenge of non-invasively collecting mucosal tissues from buccal or gingival areas of healthy individuals, we opted for mucosal tissues surrounding wisdom teeth extractions as the control group. Hence, potential inflammation and age difference may influence results. Despite these challenges, our findings highlight varying microbial community compositions across different OPMDs, indicating microbial differences among diseases sharing similar histological features. This prompts speculation regarding microbial changes in PVL carcinogenesis.

Conclusions

Investigations revealed distinct predominant bacterial species across various groups, with F.periodonticum exhibiting a decline and P.oris demonstrating an elevation, suggesting potential significance in the progression from PVL to carcinoma. Further exploration into the oral microbiota is warranted to elucidate these findings.