The study included 59 paired urinary stone and urine samples collected from patients diagnosed with urolithiasis at Suzhou BenQ Medical Center between November 2023 and April 2024. All patients provided informed consent, and the study was approved by the Ethics Committee of Soochow University (Reference 2020CS017). All samples were collected and processed using sterile handling procedures. For downstream comparison, pathogen-dominated groups were defined according to the dominant pathogen profile: Group 1, Mycoplasma/Ureaplasma; Group 2, Pseudomonas; Group 3, E. coli; and Group 4, Staphylococcus aureus.

Fresh urine samples and stone fragments were collected during surgical procedures. Stones were aseptically scraped to obtain biofilm samples, while urine samples were centrifuged to pellet microbial cells [11]. All specimens were stored at −80˚C until DNA extraction.

The urine samples and stone-scraped swabs were processed for microbial analysis. Urine was centrifuged to pellet microbial cells, while swab tips were vortexed in PBS to dislodge adherent bacteria. Genomic DNA was extracted using the MagaBio Soil/Feces Genomic DNA Purification Kit (Bioer Technology, Hangzhou, China), with modifications for low-biomass samples [12][13]. DNA purity and concentration were assessed spectrophotometrically (NanoDrop ND-1000, Thermo Fisher Scientific) based on A260/A280 ratios. All extracts were normalized to 1 ng/µl (to account for lower microbial biomass) and stored at −20˚C.

The hypervariable V3-V4 regions of bacterial 16S rRNA genes were amplified using barcoded primers (Invitrogen). PCR reactions (50 µl total volume) contained: 25 µl Taq polymerase mix (Takara Bio), 1 µl each of forward and reverse primers (10 mM), and 3 µl template DNA. Thermal cycling conditions were: initial denaturation at 94˚C for 30 sec; 30 cycles of 94˚C (30 sec), 52˚C (30 sec), and 72˚C (30 sec); with a final extension at 72˚C for 10 min. PCR products were verified by 1% agarose gel electrophoresis, pooled in equimolar ratios (GeneTools Analysis Software, SynGene), and sequenced on an Illumina MiSeq (PE300, MAGIGENE Genomic Institute).

The raw paired-end sequencing data were processed using BBmerge (BBtools suite v38.94) with the following parameters: “loose = t mininsert = 120 mininsert0 = 100 qtrim2 = t qout = 33 entropy = t maxns = 0 trimq = 10” to merge overlapping reads while maintaining high accuracy. Quality filtering was performed using Vsearch v2.18.0 with a maximum expected error rate of 0.02 (fastq_maxee_rate = 0.02) to remove low-quality sequences [14]. For consistency of downstream analysis, the primary feature unit reported in this study was amplicon sequence variants (ASVs). High-quality reads were denoised with the “cluster_unoise” module, and potential sequencing artifacts were further removed by chimera detection using “uchime3_denovo” to generate the refined ASV table.

For taxonomic classification, ASVs were aligned against the SILVA 138.1 16S rRNA database using BLASTn, retaining only matches with ≥85% sequence identity over ≥50% of the read length. Final taxonomic assignments were determined using a hybrid lowest common ancestor (LCA) approach to ensure robust classification accuracy. Microbial community diversity was assessed using QIIME 1.9.1, which calculated multiple alpha diversity indices including observed richness, Chao1, Shannon, Simpson, ACE, Goods-coverage, and PD_whole_tree [15]. These metrics were visualized through rarefaction curves, rank-abundance plots, and species accumulation curves generated in R v4.1.2.

Statistical comparisons of alpha diversity between sample groups were performed using both parametric (t-test) and non-parametric (Wilcoxon) tests as appropriate based on data distribution characteristics. Beta diversity analysis incorporated principal coordinate analysis (PCoA), principal component analysis (PCA), and non-metric multidimensional scaling (NMDS) using R’s stats, ggplot2, and vegan packages. Differences in beta diversity metrics were evaluated using both parametric and non-parametric statistical approaches. Differential abundance analysis was conducted using STAMP software with default score filtering parameters, supplemented by group-wise t-tests in R for identifying significantly different taxa [16]. All visualizations were generated using R’s advanced plotting capabilities to ensure clear representation of the complex microbial community data.

The comparative analysis of clinical and urinary parameters across the four pathogen-defined groups revealed distinct microbial-associated profiles. Demographic analysis showed significant gender disparities, with Group 1 (Mycoplasma/Ureaplasma-dominated infections) exhibiting a striking female predominance (85.7%) compared to Group 2 (Pseudomonas-dominated, 81.5% male; p < 0.001). This gender distribution pattern was similarly significant between Group 1 and Group 4 (Staphylococcus aureus-dominated, p = 0.015), suggesting potential pathogen-specific tropisms for different host sexes.

Urinary inflammatory markers demonstrated clear group-specific patterns. Group 1 showed markedly elevated leukocyte counts (median 535.6, IQR 911.7) compared to all other groups (all p ≤ 0.012), indicating a particularly robust inflammatory response to Mycoplasma/Ureaplasma infections. Biochemical analysis revealed that Group 2 maintained significantly lower urinary pH (median 6.19) than Group 1 (median 6.79, p = 0.037), whereas the crystal-examination result was retained as a descriptive pattern rather than a statistically significant pairwise difference. Group 2 showed the highest median crystal count (57.8, IQR 1.1) and Group 4 showed the lowest median value (0.2, IQR 0.2), but this contrast should be interpreted cautiously in light of the non-significant pairwise results shown in Table 1. The group-level clinical and laboratory characteristics are summarized in Table 2.

Microbiological and cellular findings provided further differentiation between groups. While bacterial and fungal loads showed no significant differences, epithelial cell patterns varied meaningfully. Group 4 displayed the highest squamous epithelial cell counts (median 9.3, IQR 1.2), whereas Group 3 (E. coli-dominated) showed elevated small round epithelial cells (median 17.4, IQR 11), potentially indicating different sites or mechanisms of epithelial damage. Group 1 demonstrated numerically higher red blood cell counts (median 17,996, IQR 8069), though this did not reach statistical significance compared to other groups.

Table 1.Statistical comparison of clinical and urinary parameters across pathogen-defined groups. The table presents p-values from pairwise comparisons (Wilcoxon/t-tests) of demographic and urinary parameters among four pathogen-dominated groups: Group 1: Mycoplasma/Ureaplasma-dominant infections (female-predominant, high inflammation). Group 2: Pseudomonas-dominant infections (male-predominant, lower urinary pH and higher median crystal-examination value in descriptive analysis). Group 3: E. coli-dominant infections. Group 4: Staphylococcus aureus-dominant infections. Significant differences (*p* < 0.05, marked with ※) highlight group-specific patterns, such as gender disparity (Group 1 vs. 2, *p* < 0.001), urinary pH (Group 1 vs. 2, *p* = 0.037), and leukocyte counts (Group 1 vs. others, *p* ≤ 0.012).

Table 2.Clinical and laboratory characteristics of pathogen-defined groups. Detailed median values (IQR) for each group: Group 1 (n = 10): Elevated leukocytes (median 535.6, IQR 911.7), female predominance (85.7%). Group 2 (n = 27): Low urinary pH (median 6.19), higher median crystal-examination value (median 57.8 counts), male predominance (81.5%). Group 3 (n = 10): Intermediate profiles with elevated small round epithelial cells (median 17.4). Group 4 (n = 12): lower median crystal-examination value (median 0.2 counts). Consistent parameters (e.g., conductivity, osmolality) suggest microbial identity drives phenotypic variation.

Notably, several parameters showed remarkable consistency across groups. Electrolyte markers (conductivity, osmolality), specific gravity, and cast counts maintained stable values regardless of the dominant pathogen, suggesting these factors may be less influenced by specific microbial presence. These findings collectively demonstrate that different uropathogens associate with distinct clinical and laboratory profiles. Mycoplasma/Ureaplasma infections present with pronounced inflammation and female predominance, whereas Group 2 is better described as showing lower urinary pH and a higher median crystal value without firm statistical support for crystalluria differences, while E. coli and S. aureus show more intermediate patterns with specific epithelial cell signatures.

The heatmap analysis revealed distinct microbial community structures across the samples (labeled as JS and NY series). Pseudomonas and Staphylococcus were the most dominant genera, showing high relative abundance in multiple samples (e.g., JS33, NY10, NY23). Other notable genera included Enterobacteriaceae, Proteus, and Mycoplasma, which exhibited variable distribution patterns. Samples such as JS20 and NY18 displayed a higher abundance of Ureaplasma, while Clostridia and Dialister were more prevalent in NY4 and JS11, respectively. The “Others” category, representing low-abundance taxa, was consistently present across all samples, indicating a diverse but unevenly distributed microbial ecosystem. Figure 1 shows the distribution of the dominant genera across the paired stone and urine samples.

The distance metric (ranging from 0 to 0.88) highlighted varying degrees of microbial similarity between samples. For instance, the samples JS33 and NY10 (distance: 0.22) shared similar microbial profiles, dominated by Pseudomonas and Staphylococcus. In contrast, JS18 and NY19 (distance: 0.86) showed markedly divergent communities, with the former enriched in Mycoplasma and the latter in

Figure 1.Microbial community composition in urinary stones and paired urine samples. Boxplots showing the relative abundance of dominant bacterial genera in stone (JS series) and urine (NY series) samples. Stone samples (JS) are enriched in environmental taxa (e.g., Christensenella), whereas urine samples (NY) dominate in uropathogens (e.g., Klebsiella).

Enterobacteriaceae. The clustering pattern suggested that samples from the same series (JS or NY) did not always group together, implying that microbial composition was influenced by factors beyond simple sample origin. Gram-negative bacteria (Pseudomonas, Proteus, Enterobacteriaceae) were widespread, particularly in samples with higher distance values (e.g., NY21, JS28), indicating potential polymicrobial infections. Gram-positive bacteria (Staphylococcus, Enterococcus, Streptococcus) were prominent in samples like NY17 and JS7, often co-occurring with other taxa. Anaerobic genera (Prevotella, Dialister, Clostridia) were more abundant in samples with intermediate distances (e.g., NY15, JS9), suggesting niche-specific colonization.

Permutational multivariate analysis (PERMANOVA) of the combined dataset indicated that bacterial communities were significantly stratified by sample type (urine vs. stone surface; p = 0.00067, R² = 0.16) and by patient group (JS vs. ENY; p = 0.00001, R² = 0.50), whereas no significant difference was driven by group identity alone (p = 0.29). The constrained PCoA (PCoA1 = 32.85 % of total variance) displayed a clear separation of stone-surface swabs toward the negative end of axis 1 and urine samples toward the positive end, with a modest overlap zone at the origin. Notably, JS samples exhibited a tighter cluster along the negative–neutral range, whereas NY samples were more dispersed across the ordination space, suggesting that inter-individual heterogeneity is more pronounced in urine than on stone surfaces. Collectively, these data support niche differentiation between urine and stone-surface communities, but the magnitude of this effect should be interpreted cautiously because a full patient-blocked reanalysis was not available in the current revision. As shown in Figure 2, alpha-diversity patterns and constrained ordination support partial but significant separation between the two niches.

At the genus level, 27 bacterial taxa exhibited significant shifts in mean relative abundance (95% CI) between stone matrices (JS) and matched urine (NY). Klebsiella was markedly enriched in urine (NY; Δ = 3.96%, p = 0.036), whereas Christensenella was almost exclusive to stones (JS; Δ = 4.72%, p = 0.047). Conversely, stone surfaces carried higher proportions of environmental- or soil-associated genera such as Chromobacterium (Δ = 0.14%, p = 0.0014), Micropepsaceae (Δ = 0.32%, p = 0.0032) and Haliangium (Δ = 0.68%, p = 0.0068), while urine showed over-representation of Acinetobacter (Δ = 4.97%, p < 0.05) and Cyanobacteria. These data indicate that, despite patient-level overlap, stone biofilms selectively accumulate specific environmental taxa absent from contemporaneous urine, underscoring niche-specific microbial partitioning within the same urological ecosystem. The taxa showing significant niche-specific enrichment are illustrated in Figure 3.

(A)

(B)

(C)

Figure 2.Diversity analysis of stone (JS) and urine (NY) microbiomes. (A) Boxplots showing the relative abundance of dominant bacterial genera in stone (JS series) and urine (NY series) samples. (B) Boxplots of alpha diversity indices (Shannon and Chao1) for stone and urine samples (p < 0.05, Wilcoxon test). (C) Constrained PCoA (Principal Coordinate Analysis) based on Bray-Curtis distances, showing significant separation between stone-surface (JS, blue) and urine (NY, red) microbiomes.

Figure 3.Differential abundance of bacterial taxa between stone (JS) and urine (NY) niches. Bar plot of significantly enriched genera in stones (e.g., Chromobacterium, Haliangium) versus urine (e.g., Acinetobacter, Cyanobacteria). Error bars represent 95% confidence intervals (*p* < 0.05, STAMP analysis).