General characteristics of Kidney stone patients and controls
Urine samples were collected from a total of 43 subjects, and the demographic and clinical data was listed in Table 1. Age, gender, and body mass index showed no significant difference between kidney stone patients and healthy controls. Although comorbidities such as hypertension, diabetes and coronary artery disease were more common in kidney stone formers, they all did not reach statistical significance. The majority of renal stone patients were first onset (20/22, 90.9%) and only two patients were recurrent. All kidney stones were primarily calcium-based and composed of calcium oxalate, calcium phosphate, or a mixture of components. Pure calcium phosphate, uric acid, cystine or struvite stones were not identified. Antibiotics were given immediately after sample collection and no associated postoperative infections were identified in this study.
Sequencing data and biodiversity of the urine microbiome
In total, 5,906,796 clean reads were obtained from the 65 urine samples. The median number of reads in kidney stone patients was 94,966 and in healthy controls was 126,090. The reads were classified into 928 unique operational taxonomic units (OTUs) at 97% similarity level that were used for downstream analysis. We defined three groups according to the kidney stone status and specimen-type: HB represents bladder urine collected from healthy controls, KB represents bladder urine from kidney stone patients, while KP represents renal pelvis urine from kidney stone patients. The HB group showed the largest amount of OTUs, and there was substantial overlap in the OTUs composition among HB, KB and KP groups (Fig. 1). Significant more OTUs were identified in the urine of healthy controls, with an average of 96 OTUs in HB group and 60 OTUs in KB group (P =0.046).
For α−diversity, the values of Good’s coverage index of all libraries were above 99%. The α−diversity indices, including observed species, chao 1 index, ACE index, Shannon diversity index, of the microbiota in HB group were all higher than those of KB group (Fig. 2). Moreover, significant differences were observed in Shannon diversity index and Simpson’s diversity index between HB and KB groups (P<0.001 for both indices). The α−diversity of urinary microbiota between KB and KP group was also evaluated, and all indices showed no significant difference. For β−diversity, we applied unweighted and weighted principal coordinate analysis (PCoA) to display discrepancy among the three groups. It showed that KB and KP samples clustered closer in proximity to each other than HB samples (Fig. 3). We further performed analysis of similarities (ANOSIM), and found the urinary microbiota structure was significantly different between KB and HB groups (ANOSIM, R = 0.11, P < 0.001), while the microbiota structure between KB and KP groups was similar (ANOSIM, R = 0.008, P = 0.251).
Taxonomic analysis of urine microbiota composition
To identify the differentially represented taxa in kidney stone patients and controls, we compared the relative abundance of microbiota between KB and HB group at different taxonomic levels. At phylum level, a statistically significant difference was observed between these two groups in the average abundance of Bacteroidetes, Proteobacteria and Firmicutes. Namely, KB group showed a higher average representation of Proteobacteria (51.8% vs 36.6%, p=0.01) and a lower average representation of Firmicutes (29.3% vs 36.1%, p=0.02) and Bacteroidetes (6.4% vs 19.4%, p<0.001). Significant abundance differences of numerous taxa were also noted between KB and HB groups at other taxonomic levels (table 2). The relative abundance of Faecalibacterium and Lactobacillus was also lower in KP and KB groups compared to HB group, although not statistically significant.
Of interest, we also analyzed the microbiota of paired bladder urine and renal pelvis urine collected from kidney stone patients. At phylum or class level, the overall bacterial compositions of KB and KP groups were quite similar (Fig. 4A-B). However, there were a few taxa differentially represented in these two groups at other taxonomic levels (Fig. 4C-E). A higher average representation of Anoxybacillus (1.2% vs 0.2%, p=0.01) and lower average representation of Fusobacterium (0.6% vs 1.3%, p=0.02) was observed in KP group at genus level.
Specific urinary genera associated with kidney stones
To confirm the differentially abundant taxa in kidney stone patients and controls, we further applied LEfSe, a software using algorithm for high-dimensional biomarker discovery. Only taxa with logarithmic linear discriminant analysis (LDA) score more than 2.0 and P < 0.05 in Wilcoxon test were considered differentially represented. LEfSe identified 31 discriminative features with significant different relative abundance among HB, KB and KP groups (Fig. 5). The taxa at genus level that differentiated the three groups most were Prevotella in HB group, Acinetobacter in KB group and Anoxybacillus in KP group.
Potential functional pathways associated with kidney stone
Having observed a distinct urinary microbiota in kidney stone patients, we further evaluated whether the different bacterial community was associated with specific alterations involved in metabolic processes. The functional pathways of urinary microbiome in HB, KB and KP samples were inferred using PICRUSt tool. Compared to HB group, the significantly enriched KEGG pathways in KB groups included proximal tubule bicarbonate reclamation, ion channels, linoleic acid metabolism and renin−angiotensin system (Supplementary Fig. 1). Meanwhile, the predicted KEGG pathways showed no significant difference between KB and KP groups (Supplementary Fig. 2).