Fig. 2. Characteristic microbial communities appeared in the nasopharynx of NPC patients.
a The PCoA plot based on Bray-Curtis dissimilarity distance showing the compositional differences between NPC (N = 70) and control (N = 86) groups with adjusting for the confounding variables (R2 and P value was obtained from PERMANOVA, two-sided). The dot plots showing the relative abundances of NPC-enriched (b) and NPC-delepted (c) species (ANCOM-BC, two-sided, FDR q < 0.05) with adjusting for the confounding variables. Data represented mean ± SEM (N = 70 for NPC, and N = 86 for control). The co-occurrence network of nasopharyngeal microbiota in NPC patients (d, N = 70) and controls (e, N = 86). Only significant correlations were shown in the networks (SparCC, |r| > 0.25 and P < 0.05, two-sided). Each node represented a microbial species, NPC-enriched and control-enriched species were shown in pink and green, respectively. Each edge represented the correlation between paired species, and its width reflected the absolute value of the correlation coefficient. Co-inclusion associations were colored in red, whereas co-exclusion correlations were colored in blue. f The venn plot showing 13 species were both NPC-enriched and high-translocation-enriched which were defined as “NPCOtoNP” species. g The heatmaps show the relative abundance and prevalence of “NPCOtoNP” species in oral and nasopharyngeal samples from NPC patients (N = 70) and controls (N = 86). h The bar plots showing the R2 and P values obtained from PERMANOVA (two-sided) of NPC (N = 70)/NPC_H (N = 16)/NPC_L (N = 54) vs control (N = 86) groups with 1001 permutations. i The ROC curves of all NPC vs control, NPC_H vs control, and NPC_L vs control groups, corresponding AUCs (95%CI) were shown. Data were based on the nasopharyngeal microbiota of Cohort 1. The confounding variables contain age, sex, cigarette smoking status, alcohol drinking status, whether have caries and whether have oral or nasal diseases. PCoA principal coordinate analysis, PERMANOVA Permutational multivariate analysis of variance, ANCOM-BC analysis of compositions of microbiomes with bias correction, FDR false discovery rate, SparCC sparse correlations for compositional data algorithm, NPC_H NPC patients with high-translocation, NPC_L NPC patients with low-translocation, ROC Receiver operating characteristic, AUC Area Under the Curve, 95% CI 95% confidence intervals. Source data are provided as a Source Data file.