Archives

  • 2019-07
  • 2019-08
  • 2019-09
  • 2019-10
  • 2019-11
  • 2020-03
  • 2020-07
  • 2020-08
  • br Statistical analysis in this

    2020-08-12


    Statistical analysis in this study was performed using R (version 3.5.0). All statistical tests were two-sided, with a P-value < 0.05
    considered significant. All graphs were created using R package “ggplot2” [19].
    2.6. Data availability
    The sequencing data from this study have been deposited in the NCBI Sequence Read Archive under accession numbers SRR8675978 e SRR8676117.
    3. Results
    To reveal the differences in the oral microbiota associated with oral cancer, we collected the salivary samples from 60 OSCC pa-tients (10 females, 50 males) and 80 control subjects (43 females, 37 males). There was no difference in the average age (63.7 and 65.1 years) between OSCC patients and control subjects.
    After sorting, unique representative sequences were classified into 14,057 operational taxonomic units (OTU) at a 97% similarity level, from which 273 genera were detected. After picking 10,000 reads per sample randomly, unique representative sequences were classified into 7647 OTUs, from which 225 genera were detected.
    Different indexes (OTU observed, Chao 1, Shannon, and Simp-son) were employed to estimate the a-diversity of the bacterial community. As revealed by the OTU observed and Chao 1 index, the diversity of the bacterial Y27632 in the OSCC samples was significantly higher compared with that in the control samples (p < 0.05). No difference was observed with other diversity indexes (Shannon, Simpson) (Fig. 1 A e D).
    To compare whether the overall bacterial taxa composition was different between the OSCC samples and controls, we used prin-cipal coordinates analysis (PCoA) on weighted and unweighted Unifrac distances.
    We observed statistically significant differences in both weighted (p ¼ 0.009) and unweighted (p ¼ 0.003) Unifrac distances between the OSCC samples and controls (Fig. 1 E, F).
    3.2. Relative abundance
    A significant level of abundance was observed for Peptos-treptococcus, Fusobacterium, Alloprevotella, and Capnocytophaga (p < 0.05) in OSCC samples relative to the controls (Table 1, Fig. 3 A e D). In contrast, we observed a lower abundance of Rothia and Haemophilus (p < 0.05) in the OSCC samples relative to the controls (Table 1, Fig. 3 E, F). 
    3.2.1. Relationship between bacteria and other factors
    We obtained information regarding age, sex, drinking, and smoking habits as well as denture use from participants. Statistical analysis was performed to compare these items with the six bac-terial genera that were identified to be significantly involved in cancer occurrence (Table 2).
    In the OSCC group, we observed a significantly high abundance of Peptostreptococcus and a low ratio of Haemophilus in female subjects compared to male subjects (Fig. 4 Y27632 A, B). In the OSCC group, Haemophilus was significantly abundant in drinkers (Fig. 4C).
    3.2.2. Logistic regression analysis
    To identify the impact factor for oral cancer occurrence, logistic regression analysis was performed. We conducted an analysis to predict cancer based on age, sex, smoking, drinking habits, denture use, OTU observed index, Chao 1 index, Shannon index, Simpson index, and the OTUs of the six dominant genera (Rothia, Allopre-votella, Capnocytophaga, Peptostreptococcus, Fusobacterium, and Haemophilus). We obtained independent variables by eliminating multicollinearity after a stepwise procedure. The result contained two interaction terms (sex, Chao 1 index). We obtained an odds rate of 10.85 for sex (97.5% Cl) and 1.006 for Chao 1 index (97.5% Cl) (Table 3).
    We performed a stratified analysis to investigate the difference between tumor stage and the relative abundance of dominant taxa using the TNM (tumor-node-metastasis) classification of the UICC (International Union Against Cancer) 7th edition. The number of patients in each stage was T1: 7, T2: 16, T3: 14, T4: 22, TX: 1 and N0: 20, N1: 7, N2: 32, N3: 1. None of the patients showed metastasis.
    We observed that the a-diversity of OTUs decreased with the progression of T stage. There was a negative correlation among the stages, r ¼ - 0.329 (Fig. 5A).
    N stages were divided into two groups: N negative (N0) and N positive (N1 to 3) to investigate the presence or absence of lymph node metastasis.
    We found no significant difference in the a-diversity between N negative and N positive (p ¼ 0.370) (Fig. 5B). We performed a stratified analysis to investigate the relation-ship between the progression of T stage and bacterial community changes. For this purpose we used bacterial genera, which previ-ously indicated a significant difference in cancer occurrence (Pep-tostreptococcus, Fusobacterium, Alloprevotella, Capnocytophaga, Rothia, and Haemophilus). As a result, the abundance of the Rothia genus was relatively decreased with T stage progression, showing a negative correlation among the stages, r ¼ - 0.323 (Fig. 5C).