Demographic data of enrolled patients
All the clinical data, including age, gender, leukocyte counts, differential counts, absolute eosinophil count, platelets, erythrocyte sedimentation rate (mm/hour), C-reactive protein (mg/L), rheumatoid factor (IU/mL), AST (mg/dL), ALT (mg/dL), creatinine (mg/dL), uric acid (mg/dL), total IgE, and total cholesterol collected by chart review did not differ significantly between the PSA and NO PSA subgroups (all p>0.05) (Table 1).
Alpha diversity and beta diversity
Microbial diversity was assessed either within a subgroup (alpha diversity) or between the sample subgroups (beta diversity). We calculated several different values to evaluate alpha diversity: “Rarefaction Curve” to calculate species richness for a given number of individual samples (Figure 1), “Chao Index” to estimate species abundance (Figure 2a), and “observed species” to estimate the amount of unique OTUs found in each sample (Figure 2b). Gut microbiota communities from the PSA patients demonstrated a Chao index (p = 0.014) and observed species (p=0.048) that were significantly increased compared to the NO PSA arthritis controls (Figure 2a, b). The Rarefaction Curve demonstrates that both the PSA and the NO PSA subgroups were adequately sampled and the species richness of the microbial DNA reach plateau (Figure 1). Beta diversity was assessed using a partial least squares discriminated analysis (PLS-DA) plot to compare between NO PSA arthritis controls and PSA patients. The results demonstrate that the microbial diversity differed significantly between the PSA and NO PSA groups (Figure 3a). The unweighted unifrac and weighted unifrac PCoA plots indicated that the gut microbiomes were different between the NO PSA arthritis controls and the PSA patients (Figure 3b). Overall, these results indicated that the gut microbiota communities differed between NO PSA arthritis controls and PSA patients.
Gut microbiome profiling
The gut microbiome dataset from NO PSA arthritis controls and PSA patients revealed a total of ten phyla, four of which accounted for more than 99% of all the bacteria. The NO PSA arthritis controls demonstrated a dominance of Bacteroidetes (75.95%), followed by Firmicutes (14.13%), Proteobacteria (9.24%), and Fusobacteria (0.34%). The bacteria composition at the phylum level of the PSA group closely resembled that of the NO PSA arthritis controls. The bacteria composition at the phylum level of the PSA group consisted of Bacteroidetes (68.06%), followed by Firmicutes (24.11%), Proteobacteria (6.1%), and Actinobacteria (1.10%). The Actinobacteria was significantly higher in the PSA patients than in the NO PSA arthritis controls (p<0.01, see following Figure 4b).
Comparison microbiology differences between PSA and NO PSA subgroups with Welch's t-test, metagenome sequencing, and LEfSe analysis
We used the Welch's t-test (Figure 4a), Metagenome Sequencing (Figure 4b), and LEfSe analysis (Figure 4c) to identify the specific bacteria phylotypes that were differentially altered between the NO PSA arthritis controls and the PSA patients. The Welch’s test, named for creator Bernard Lewis Welch, is an adaptation of the Student's t-test (19) for analysis of significantly different families of species between NO PSA arthritis controls and PSA patients. The result shows that Family: XIII_AD3011 is significantly higher in the stool samples of NO PSA patients than those of PSA patients (p=0.039, Figure 4a). Metagenome Sequencing is used to demonstrate the significant difference of kingdom, phylum, class, order, and family between the NO PSA arthritis controls and the PSA patients. In Figure 4b, only those that differed significantly between NO PSA arthritis controls and PSA patients are shown (Figure 4b, all p<0.05). These differences include Actinobacteria Phylum, Erysipelotrichia Class, Erysipelotrichales Order, Fusobacteriales Order, Acidaminococcaceae Family, Erysipelotrichaceae Family, Peptostreptococcaceae Family, Clostridium innocuum Genus, Catenibacterium Genus, Family XIII_AD3011 Genus, Intestinimonas Genus, Morganella Genus, Turicibacter Genus, and Tyzzerella Genus (Figure 4b, all p<0.05).
The LEfSe plot as a result of comparison Genus, and Species between NO PSA arthritis controls and PSA patients is shown in Figure 4c. The relative abundance of Megasphaera elsdenii in the PSA group was significantly higher than 10000 times in the NO PSA arthritis controls group.