Analysis Of Features Of The Microbiota Associated With Gastric Cancer Progression

BackgroundThe gastric microbiome is mainly composed of Proteobacteria,Firmicutes,Bacteroidetes,Fusobacteria,and Actinobacteria.During the development of gastric cancer,the profile and composition of gastric microbiome change continuously,suggesting that gastric microbiome is associated with gastric cancer progression.To date,few studies have been conducted to explore features of the microbial community in the progression of gastric cancer.ObjectiveWe aimed to analysis the profile and composition of gastric microbiome in patients with advanced gastric cancer(AC)and patients with early gastric cancer(EC),and to further identify microbial signatures capable of classifying gastric cancer progression by random forest model screening.MethodsA total of 60 patients with gastric cancer were enrolled in the study,the gastric microbiome was assessed in AC(n = 30),and EC(n = 30).The biopsies was taken at least 5 cm away from the cancerous lesion.To analyze the microbial communities of the gastric mucosa,genomic DNA was extracted from gastric mucosa samples.The16 S r RNA gene was PCR amplified,to conduct high-throughput sequencing analyses gastric mucosa microbial.Linear discriminant analysis(LDA)effect size(LEf Se)was used to analyze the difference of bacteria between AC and EC.A classifier based on a random forest(RF)machine learning algorithm identified a microbial signature that distinguished AC from EC.The importance of bacteria characteristics was determined by Gini coefficient(the sum of reductions in a variable).Out of bag error rate(OOB)was used to evaluate the accuracy of RF classification model.A 20-times repeated10-fold cross-validation of the RF model was performed.The performance of the RF model was demonstrated by the receiver operating characteristic(ROC)curve.Results1.The general clinical data of AC and EC were analyzed respectively.The average age of the AC was 58.5±14.5 years,and the gender ratio was male to female was 2.3:1.The average age of the EC was 60.1±11.5 years,and the gender ratio was male to female was 3.3:1.There was no significant difference in the age or sex between AC and EC(P>0.05;P>0.05).2.The microbial communities of gastric mucosa from 60 patients were analyzed using high-throughput sequencing of 16 S r RNA amplicons.A total of 9,828,400 pairedends reads were obtained from 60 samples.After quality control and filtration,8,166,894 reads were produced with an average of 68,058 reads per sample.The coverage index of both groups of samples was > 0.99.It can be considered that at least more than 99%of the species were detected in the samples,and the sequencing depth was reasonable.3.We analyzed the diversity of microbiome in the two groups.The results showed that the Shannon index of AC was 2.673,which was significantly higher in AC than EC(2.454)(P=0.013).In contrast,there was no significant difference in the Chao 1 index between EC and AC(P=0.538).The comparison of Chao 1 index showed that the gastric flora in AC group was slightly lower than that in EC group,there was no significant difference in the Chao 1 index between EC and AC(P=0.538).The β-diversity of the two groups was analyzed.The principal coordinate analysis(PCo A) chart showed that the gastric microbiome of patients in the two groups could be clearly distinguished from each other,while the microbiome in each group was significantly clustered,indicating that the structure of the microbiome in the two groups was different(P=0.04).This indicated that members of microbiota in gastric cancer were more distantly related,suggesting a diversified microbiota profile during the progression of gastric cancer4.We analyzed the species composition of microbiome of the two groups.The histogram of species composition showed that there was no significant change in the microbiome composition at phylum level.The Mann-Whitney U test performed that the relative abundance of Deinococcus-Thermus in AC(2.840)was significantly higher than that EC(0.868)(P=0.000398).The bacteria with potential cancer-promoting activities were enriched in AC.5.Analysis of the significance of the difference between the two groups of patients,linear discriminant analysis(LDA)effectsize(LEf Se)analyses identified 12 bacterial genera that were associated with AC and EC,had an LDA score higher than 2.0.A majority of the genera which were(8/12,Thermus,Burkholderia,Pseudomonas,Brevibacillus,Stenotrophomonas,Salinivibrio,Uruburuella and Schlegelella)showed significantly increased abundances in AC,the highest LDA value was Thermus(LDA=4.06).Compared with AC,Bradyrhizobium,Nocardioides,Bacillus and Ochrobactrum were increased relative abundance in AC.The relative abundances of the 12 genera were further compared that the relative abundances of the enrichment genera changed significantly in the AC group.The most significant change was Thermus(P < 0.05),which was 3.86(2.7/0.7)times of relative abundance in EC group。6.A RF classification model was built with the RF algorithm that the results showed a minimal set of 24 bacterial genera,15 of which were Proteobacteria: Burkholderia,Uruburuella,Enhydrobacter,Ochrobactrum,Stenotrophomonas,Salinivibrio,Serratia,Bradyrhizobium,Helicobacter,Schlegelella,Pseudomonas,Rhizobium,Cellvibrio,Comamonas,Brevundimonas,6 bacteria belonged to Firmicutes: Bacillus,Thermus,Brevibacillus,Streptococcus,Anoxybacillus,Paenibacillus.3 bacteria belonged to Actinobacteria: Nocardioides,Pseudonocardia,Tsukamurella,that maximally differentiated AC from EC(AUC= 0.916,95% CI: 0.841-0.990,sensitivity:0.800,specificity: 0.767).ConclusionIn the process of gastric cancer progression,the composition,structure and diversity of gastric microbiome of AC were significantly different from those of EC,and Thermus may be related to the progression of gastric cancer.The study also identified a set of bacteria(a minimal set of 24 bacterial genera,15 of which were Proteobacteria: Burkholderia,Uruburuella,Enhydrobacter,Ochrobactrum,Stenotrophomonas,Salinivibrio,Serratia,Bradyrhizobium,Helicobacter,Schlegelella,Pseudomonas,Rhizobium,Cellvibrio,Comamonas,Brevundimonas,6 bacteria belonged to Firmicutes: Bacillus,Thermus,Brevibacillus,Streptococcus,Anoxybacillus,Paenibacillus.3 bacteria belonged to Actinobacteria: Nocardioides,Pseudonocardia,Tsukamurella)by RF model that could accurately distinguish AC from EC.The signature of bacteria can be used as a biomarker to judge the progression of gastric cancer.