Prognostic Role of Porphyromonas Gingivalis in Oral Cavity Squamous Cell Carcinoma Patients

Background: Oral microbial species play direct and/or indirect role in carcinogenesis of oral cavity squamous cell carcinoma (OSCC). Porphyromonas gingivalis (Pg) has been identied a correlation with OSCC. Fimbriae play a vital role for its attribution of initial attachment and adhesion of Pg. Six genotypes (types I-V, Ib) of mA were identied based on sequence variations and the genotype was suggested a relationship to pathogenicity of Pg. Objective: To investigate the abundance of Pg in OSCC as well as the frequency of Pg mA genotypes in OSCC patients. Methods: Ninety-ve OSCC patients and thirty-nine gender- and age-matched non-OSCC subjects were investigated abundance of Pg in saliva. Presence of Pg was compared in OSCC tissue and para-cancerous tissue from patient as well. Clinical data were extracted and patients followed up for a mean period of 13 months. Presence of Pg and mA genotypes were investigated in OSCC tissue and in saliva, then PCR products were sequencing and compared. Results: OSCC patients showed high abundance of Pg in saliva (Chi-square=14.531, P=0.001). OSCC tissue showed strong in situ expression of Pg by in situ hybridization compared with normal tissue adjacent to OSCC. Patients with overabundance of Pg in saliva are associated with systemic disease (Chi-square=10.328, P=0.029), longer disease-free time (Z=-2.988, P=0.003), and lower recurrence rate (Chi-square=5.670, P=0.017). The abundance of Pg was an independent favorable prognostic factor (HR: 0.124, 95%CI: 0.016 to 0.941). There was dominant distribution of Pg with genotype I+Ib (21.1%), II (31.6%) and IV (21.1%) in OSCC patients. The mA genotypes detected in saliva were in accordance with those in OSCC tissue, and there was signicantly correlation in amplied Pg fragments similarity between in saliva and in OSCC tissue. specimens, genotype II in 30 (31.6%) specimens, genotype III in 4 (4.2%) specimens, genotype IV in 20 (21.1%) specimens, genotype V in 2 (2.1%) specimens. We also found two or more genotypes of mA from one sample. FimA genotype I, Ib and II was detected in 1 (1.1%) participant, genotype I, Ib and III in 2 (2.1%) participants, genotype I, Ib and IV in 1 (1.1%) participant, genotype I, Ib and V in 1 (1.1%) participant, genotype II and IV in 3 (3.2%) participants, genotype I, Ib, II and IV in 1 (1.1%) participant. Ten participants showed negative on 2% agarose gel electrophoresis assay. This supported

Results: OSCC patients showed high abundance of Pg in saliva (Chi-square=14.531, P=0.001). OSCC tissue showed strong in situ expression of Pg by in situ hybridization compared with normal tissue adjacent to OSCC. Patients with overabundance of Pg in saliva are associated with systemic disease (Chi-square=10.328, P=0.029), longer disease-free time (Z=-2.988, P=0.003), and lower recurrence rate (Chi-square=5.670, P=0.017). The abundance of Pg was an independent favorable prognostic factor (HR: 0.124, 95%CI: 0.016 to 0.941). There was dominant distribution of Pg with genotype I+Ib (21.1%), II (31.6%) and IV (21.1%) in OSCC patients. The mA genotypes detected in saliva were in accordance with those in OSCC tissue, and there was signi cantly correlation in ampli ed Pg fragments similarity between in saliva and in OSCC tissue.
Conclusions: This study indicated that Pg might involve in the pathogenesis of OSCC, and Pg might consider as a potential prognostic indicator in OSCC. There was a dominant distribution of Pg with genotypes I, Ib, II and IV in OSCC patients. The presence of Pg in tumor might be saliva in provenance.

Background
There were approximately 354,864 new cases of lip and oral cavity cancer and 177,384 deaths from these tumors worldwide in 2018 [1] . Oral cavity squamous cell carcinoma (OSCC) is the most common malignant disease in head and neck besides non melanoma skin cancer. Traditionally, risk factors associated with OSCC include tobacco and alcohol consumption, betel quid chewing, HPV or HIV infection, dietary factor, vitamin and mineral de ciencies, occupational exposures and heritable conditions. But overwhelming epidemiologic, clinicopathologic and molecular studies have proved oral microbial species play direct and/or indirect role in carcinogenesis of OSCC [2] . chemoradiotherapy (Docetaxel, Cisplatin, 5-Fluorouracil) respectively. Pathological diagnosis was established by one pathologist and con rmed by another experienced pathologist.
Patients were followed up from discharge by telephone or clinical assessment. Pathologic con rmation of recurrence was obtained in patients with clinical signs or symptoms. Disease-free survival (DFS) is de ned as the time (in months) from the date of discharge to April 2020 or until the date recurrence was diagnosed.
A total of 134 saliva samples were collected between 6 a.m. and 8 a.m. following an overnight fast and refrainment of tooth brushing. Subjects were asked to swish vigorously with 40 mL sterilised double distilled water (bacteria negative in PCR assay) for 1 minute, and then to expectorate in to another specimen tube [16] . The saliva samples were centrifuged at 14,000 rpm for 15 minutes, and then the cell pellet was suspended in 1 mL of sterile TE buffer. Saliva samples stored at -80 ºC until testing.
Bacterial DNA was extracted from saliva samples using a commercial DNA extraction kit (DP302, Tiangen, China) according to the manufacturer's protocol, except adding an enzymatic lysis step with lysozyme (20 mg/ml, 37 ºC, 60 minutes). The resultant DNA was stored at -20 ºC until in PCR.
A total of 15 out of 95 OSCC patients fresh-frozen OSCC tissue samples were obtained during the operation. DNA was extracted using the Total DNA/RNA/Protein Kit (R6734, Omega Bio-tek, USA) according to the procedure recommended by the manufacturer. Quanti cation of Pg in saliva samples and detection of mA genotypes were measured by Real time quantitative PCR. Ampli cations were performed in duplicated on Bio-Rad CFX96 thermal cycler (Bio-Rad Laboratories, USA). The primers, synthesized by Sangon Biotech (Shanghai, China), used in Real time quantitative PCR with the annealing temperature is listed in Table 1. Ampli ed PCR products of mA genotype from Real time quantitative PCR were checked on 2% agarose gel (ST004L, Beyotime, China). This was done using 1X Tris Acetate-EDTA buffer (TAE) from 50X TAE (ST716, Beyotime, China). Gels were stained with 4S GelRed (A616697, Sangon Biotech, China). Image results were captured with the digital imaging system (NuGenius, SYNGENE, UK). One paire of ampli ed Pg fragments from OSCC tissue and saliva were con rmed following nucleotide sequencing by Sangon Biotech (Shanghai, China) and the correlation of two sequence by aligning two sequences with BLAST (http://www.ncbi.nlm.nih.gov/BLAST) [20] .
Among 15 OSCC tissue patients, remaining OSCC tissue and normal tissue adjacent to OSCC from one patient were xed in 4% paraformaldehyde, para n-embedded, and cut into 4 m sections, which was stained with haematoxylin and eosin, gram and subjected to in situ hybridization (ISH) using Enhanced Sensitive ISH Detection kit I (POD) (MK1030, Boster, China) according to the manufacturer's instructions. The probe is listed in Table 1. Omission of the probe was obtained as the negative controls.

Statistical analysis
Shapiro-Wilk test was used to assess whether or not data were normally distributed. Normally distributed data were analysed by Student's t test and presented as Mean ± Standard Deviation. The data without normal distribution presented as median and inter-quartile range (M, Q) and analysed by the Mann-Whitney U test. Categorical variables were analysed by Pearson Chi-square test or Fisher's exact test. The cutoff point to convert the number of Pg 16S rRNA gene copies into categorical data (weak, <2.04 and strong, >2.04) was performed using X-tile software [21] . The Kaplan-Meier method was used to estimate the recurrence rate. Univariate and multivariate Cox regression models were used to evaluate the association between DFS and clinicopathological variables. Parameters considered statistically signi cant (P<0.20) in the univariate model were analysed in the multivariate models with backward selection. All two-tailed P values < 0.05 were considered as signi cant. All analyses were carried out using IBM SPSS Statistics software (IBM SPSS Statistics V.25.0, USA).

Results
Overabundance of Pg was associated with the incidence of OSCC As showed in table 2, compared with controls matched for gender and age (P < 0.05), OSCC patients showed overabundance of Pg in saliva (P < 0.05). To exclude contamination of samples, Pg was also detected in tissues by ISH from one patient. As showed in gure 1, compared with normal tissue which is adjacent to OSCC, OSCC tissue showed strong in situ expression of Pg.   The presence of Pg in tumor were saliva in provenance To clarify the homogeneity of Pg among saliva and OSCC tissue, the frequency of mA genotypes was also detected in fteen OSCC tissues. Among fteen patients, the mA genotypes detected in saliva were in accordance with those in OSCC tissue (Table 5). Besides, ampli ed Pg fragments from OSCC tissue and saliva were examined in one patient, we found signi cantly correlation in nucleotide similarity ( Figure  3 and Figure 4). Collectively, these results provided evidence supporting the presence of Pg in tumor was saliva in provenance, and the pro le of Pg in saliva was in accordance with those in tissue.

Discussions
With recent breakthroughs in high-throughput genetic-based tools, there has been a hot issue concerning the relationship between the oral microbiome and neoplasms, especially OSCC. Recently accumulating evidence indicated the relationship between Pg and OSCC. Immortalized oral keratinocytes stimulated with Pg led to a more aggressive malignant pro le phenotype and contributing to enhanced tumor features [22] . The serum immunoglobulin G antibody against Pg was higher in OSCC patients compared with non-OSCC patients [4] . Pg increased the size and the multiplicity of carcinoma to promote the development of oral cancer [5] . Previous studies have con rmed the association between Pg and OSCC by examining the abundance of Pg in saliva of patients, and unveiled that patients with medium and poor differentiation, overall clinical stage III and stage IV as well as lymph node metastasis suggested association with Pg involvement [3] . Our study also veri ed the overabundance of Pg from OSCC patients compared with non-OSCC subjects in saliva. To exclude contamination of samples, we examined the presence of Pg in OSCC tissue by ISH. There was high enrichment of Pg in OSCC tissue compared with normal tissue adjacent to OSCC, as previous study [3] . After an average follow-up period of 13 months, the disease recurred in 18.1% (17/94) of our patients. As the same as our ndings, it was reported early stage patients have a 90-95% survival rate for one year or more, and advanced stage patients have a 65-70% survival rate [23] .
We found that overabundance of Pg in saliva was more likely to correspond with systemic disease. Systemic disease included hypertension, diabetes, coronary arteriesarteriosclerosis and chronic hepatitis B. It is may because diabetes mellitus and hypertension were correlation with increasing abundance of Pg [24,25] .
Compared with weak group, patients with the overabundance of Pg in saliva had longer disease-free time, and patients with the overabundance of Pg in saliva had a lower recurrence rate than those with low abundance of Pg. Those suggested that Pg may have an effect on prognosis of oral carcinoma. Unexpectedly, we found the overabundance of Pg is a prognostic factor for longer DFS. Contrary to popular belief, they found that Pg was associated with higher risk of pancreatic cancer [26,27] , esophageal squamous cell carcinoma [28,29] and oral squamous cell carcinoma [3] . Meanwhile, Pg was associated with overall survival rate in esophageal squamous cell carcinoma [30] . Furthermore, patients with high level of Pg had the worst prognosis in esophageal squamous cell carcinoma [29] . Besides population and followup period contributed to this prognostic incongruity, the inherent mechanistic also needs to elucidate.
One of the most vital virulence of Pg has been supposed to the presence of mbriae, which plays an important role in adhesion, colonization and invasion to tissues [31] . Most of studies focused on distribution of mA genotypes in periodontitis. However, the frequency of mA genotype of Pg in OSCC was not clear. FimA genotypes I and Ib could be discriminated by Rsa I enzyme digest, however, clearly discrimination of genotypes I and Ib seem to be improbable [11] . Besides, there were no differences in immunological analysis between mA I and Ib mbriae [11] . So we consider mA genotypes I and Ib as a whole.
In this study, the association of mA genotypes and clinicopathological parameters was not statistically signi cant. However, the predominant detected mA genotypes in OSCC were genotypes I, Ib, II and IV. Several studies concluded that nucleotide genetic variation was likely associated with virulence. Some reported mA genotypes Ib, II and IV are the most virulent mbriae in periodontitis and assist in adhesion and invasion [11,32] . It was reported that mA genotype Ib, II and IV led to more severe infections and in ammations [33,34] . Clinical isolation of Pg from chronic periodontitis patients also supported the virulence of mA genotypes Ib, II and IV [35] . Different Pg mA genotypes was injected subcutaneously and Nakano et al found that the weakest in ammatory response was induced by genotype III [34] . FimA genotype V was the least amount of genotypes in this study. The reason might be the low prevalence (0%-29%) of this genotype reported in other studies [36] . Single Pg mA genotype was determined more than 70% of OSCC patients, and two or more genotypes were also detected in a subset of the subjects. Approximately 10% of the samples were multiple genotypes in this population, which was less than the results of other studies [11,14] . Researchers attributed to the limitations of PCR in discrimination of mA genotypes and the possibility of classifying new genotypes [37] .
Due to conservative properties of DNA, the bacterial 16S ribosomal DNA allows identi cation of the genus and species. Analysis of Pg nucleotide sequences in the OSCC tissue and in the saliva showed a homology of 100%. Moreover, the distribution of mA genotypes in OSCC tissue is according to those in saliva. Those results support the provenance of Pg in OSCC tissues is those in saliva [38] .
The limitation of this study is the short follow-up period. Further research will need to be done to elucidate the prognostic role of Pg in the long run. Except mbriae of Pg, other important virulence such as: encapsulation (K1-K6), gingipain (types A, B, C) as well as lysine-speci c types I and II, may also play a vital role in the association between Pg and OSCC. Those remain to be uncovered in the future study.

Conclusions
This study found the overabundance of Pg was associated with OSCC. However, patients with systemic disease, longer disease-free time and the lower recurrence rate were related to the overabundance of Pg. Meanwhile, the abundance of Pg was an independent favorable prognostic factor. Furthermore, there was a dominant distribution of Pg with genotype I, Ib, II and IV from OSCC patients and the presence of Pg in tumor was saliva in provenance.

Abbreviations
Oral cavity squamous cell carcinoma: OSCC Porphyromonas gingivalis: Pg In situ hybridization: ISH

Declarations
Ethics approval and consent to participate Approval from the institutional review board was obtained at the Hospital of Stomatology Wuhan University before starting the study (2016-60). Informed consent was obtained from each patient. Animal Ethics clearance is not applicable, as this study does not involve any animals.

Consent for publication
Not applicable.

Availability of data and materials
All data generated or analysed during this study are included in this published article.

Competing interests
The authors declare that they have no competing interests.  Cumulative recurrence curve of 94 patients.

Figure 3
The PCR product examined of Porphyromonas gingivalis in saliva (A) and in oral cavity squamous cell carcinoma tissue (B).

Figure 4
The homology analysis of Porphyromonas gingivalis detected in saliva and oral cavity squamous cell carcinoma tissue.