The microbiome of kidney stones and urine of patients with nephrolithiasis

Nephrolithiasis is a rising problem globally, with the lifetime incidence having increased from 3% in the 1970s to 10% in Europe, 14% in North America and 20% in Saudi Arabia [1]. In parallel, the rate of stone recurrence has increased with almost 50% of stone patients experiencing an additional stone event within 5 years [2]. In the western world, both sexes are now almost equally affected. However, the prevalence for stone disease (SD) increased with 22.0% in females, while it stayed almost constant in men [3, 4].

For a long time, urinary supersaturation with calcium and oxalate beyond their solubility was considered the main cause of stone formation [5]. Recent studies showed, that recurrent stone formers do not necessarily present with urinary accumulation of calcium oxalate (CaOx) or calcium phosphate (CaPhos). In fact, there is no significant difference in urine chemistry of recurrent stone builders and healthy controls, suggesting that chemical supersaturation alone is not sufficient to explain stone formation [6].

Dysbiosis of intestinal, skin and mucosal microbiomes has been associated with different maladies. Whether urinary dysbiosis is a driver of stone disease, or stone formation rather promotes bacterial imbalance of the urological tract remains to be clarified.

Catheterized urine and kidney stones from 72 male (72%), and 28 female (28%) patients were collected intraoperatively. Voided urine of 20 volunteers (10 males and 10 females) served as healthy reference group. All patient characteristics are shown in Supplementary Fig. 1A. In the study cohort, 18 different combinations of stone types were identified via XRPD. The most prevalent stone type was CaOx monohydrate (COM, Whewellite) present in 71.4% of female and 33.3% of male patients, followed by mixed stones of Whewellite/Apatite (10.7%) in females and Whewellite/CaOx dihydrate and monohydrate (COD, Weddellite) in male patients (18.1%) (Supplementary Fig. 1B).

Due to the low microbial biomass in some samples, 16S rRNA gene amplification was not successful in all collected samples. After quality control and contamination filtering, we only used a sub-cohort of samples for further analysis, where the minimal read number surpassed 500 reads. Urine samples from 33 male (45.8%) and 14 female (50%) patients met the filter criteria, while only 13 stones from male and 12 stones from female patients could be included (18.1 and 42.9%, respectively). The voided urine of all volunteers passed the quality control (Supplementary Fig. 1A). Notably, clinical data in the successfully sequenced samples are similar to the initial cohort (Fig. 1A, B). Finally, for 5 male (6.9%) and 7 female (25%) patients, microbiome data could be retrieved from a full sample set (Supplementary Fig. 1A).

Almost all stone types had a verifiable microbiome. However, all pure uric acid stones (12.7% of all stones) had to be excluded, since none of them displayed > 500 reads (mean = 37.5 reads). The highest read numbers were observed in Apatite and Apatite/Weddellite/Whewellite stones.

We further analyzed the differential relative abundance of microbial taxa, grouped on genus level, between stones, patient urine and urine from healthy volunteers. DeSeq2 analysis revealed 72 significantly differentially abundant genera among the three sample groups (Fig. 3). Comparing the microbiome of stones to corresponding patient urine or healthy urine respectively, we determined that Ureaplasma, Escherichia–Shigella (log2FC: 12.5), Pseudomonas, Proteus, Klebsiella, Enterococcus (log2FC: 10.0), and other Enterobacterales and Enterococcaceae (highlighted in blue) displayed a significantly increased relative abundances in stone samples when compared to both types of urine samples.

In voided urine from volunteers without nephrolithiasis, several genera belonging to Firmicutes, such as Finegoldia, Anaerococcus, Peptoniphilus and Veillonella, were more abundant. Corynebacteriaceae (Lawsonella, Corynebacterium), Bifidobacteriaceae (Alloscardovia, Coriobacteriaceae) and Lactobacillaceae (Lactobacillus, Limosilactobacillus, Facklamia) were also more abundant in healthy mid-stream urine (highlighted in green), compared to patient urine or stone samples.

We further investigated the correlation of dysbiosis in SD patients with clinical data.

Patients were clustered to the cohort “features of metabolic syndrome” (FMS), when at least two of the following morbidities were exhibited: BMI > 30, Type 2 diabetes, dyslipidemia and arterial hypertension. Interestingly, PCoA analysis revealed a significant difference in the stone microbiome in patients with FSM compared to those without FMS (p = 0.015) (Fig. 4A). In kidney stones derived from patients with FSM, the genera Escherichia–Shigella, Enterococcus, Sphingomonas and Klebsiella were significantly more relatively abundant. Stones retrieved from patients without FMS were characterized by increased relative abundance of different Staphylococcus species and Ureaplasma instead, which were almost absent in stones from patients with FMS (Fig. 4B).

Additionally, patients with FMS displayed a significantly increased relative abundance of classical gastrointestinal tract-associated microorganisms, such as E. coli, and Enterococcus faecalis in pre-surgical urine cultures (> 10³ colony-forming units/ml) compared to SD patients without FMS (p = 0.0362) (Fig. 4C).

No significant difference was found regarding recurrence of SD, and other clinical parameters, when compared to the cohort without the respective condition (data not shown). Given the small number of patients with the same stone type fulfilling the quality criteria, we also found no significant similarities in the microbiome within a certain stone type (Supplementary Fig. 2).

We performed the so far largest prospective study investigating the composition of kidney stones and urine microbiome in patients with nephrolithiasis. Our analyses lead to several important findings. First, the highest amounts of confidently detectable bacterial colonization were found in Apatite and Apatite/CaOx/CaPhos stones, although these stones are classified as “non-infection stones” in the EAU guidelines [19]. Second, we observed that the microbiome of stones is well reflected in the corresponding patients’ urine sample. Thus, the analysis of the microbial community composition of catheterized urine samples may help to guide therapy decisions for tailored antibiosis, to avoid complications after stone removal but also antimicrobial overtreatment. Moreover, our analysis revealed, that stones display an increased relative abundance of distinct microorganisms, classically associated with the gastrointestinal microbiome, but also known to be pathogenic [20].

The systemic change of public health, with increasing cases of severe obesity, raises enormous medical expenditure and imposes new challenges in patient care. Nephrolithiasis is a well-known co-morbidity of obesity, and metabolic syndrome and diabetes have been considered as risk factors for kidney stone formation for more than 20 years [21]. For a long time, hyper-nutrition and unilateral diet, resulting in a supersaturation of carbonates, considered mainly as the underlying reasons. Moreover, obese persons are characterized by hyperinsulinemia, which is associated with increased intestinal absorption and renal excretion of calcium [22]. However, besides resulting in chemical supersaturation, nutrition has a strong effect on the gut microbiome itself, and consequently also on urogenital microbial populations [23]. Consistent with previous reports, we found significant dysbiosis in patients with SD, characterized by an enrichment of classical gastrointestinal microorganisms in urine and kidney stones [23]. While others focused on dysbiosis of the gut microbiome in SD, we were able to confirm this observation also in stone and urine of patients with nephrolithiasis [24]. We showed that the bacterial composition in stones of patients with features of metabolic syndrome is characterized by increased abundance of gastrointestinal microorganisms, such as E. coli, Shigella, Klebsiella, Enterococcaceae, Proteus and Sphingomonas, while SD patients without FMS display a higher relative abundance of Staphylococcaceae and Ureaplasma. This supports the study from Chen et al. where an increased abundance of Escherichia–Shigella, Klebsiella, Enterococcus and Aerococcus, but a reduction of Prevotella and Lactobacillus in patients with type II diabetes and lower urinary tract symptoms (LUTS) was described [25]. The origin of these classical fecal opportunistic pathogens and commensals is not easily explained. It is possible that poor diet and the enrichment of carbohydrates promote the colonization of certain bacteria, while overgrowing others.

We were not able to demonstrate that the accumulation of a specific bacterial genus or combination of bacteria promotes the formation of a particular type of kidney stone. However, we showed that the incidence of pathogenic Enterobacteriaceae was high in all stone types, proving that besides struvite, CaOx and CaPhos stones are also associated with these bacteria. Barr-Baere et al. also identified E. coli and other bacteria of the family Enterobacteriaceae as the most prevalent uro-pathogen in pediatric urinary tract infections accompanied by CaOx stones [26]. Taking advantage of a murine model for CaOx nephropathy, they showed increased CaOx deposition in mice after transurethral inoculation with E. coli compared to CaOx nephropathy alone. The underlying mechanism for enhanced stone formation was both the adhesion of bacterial metabolites to crystals and increased CaOx crystal aggregation [26].

Furthermore, we hypnotized that patients with recurrent SD display an aberrant microbiome compared to patients with primary SD. Reasons therefore might be previous interventions, transporting bacteria from skin, genitals, urethra and bladder up to the renal pelvis. Moreover, antibiosis and other therapeutics can also sustainably change the bacterial composition of the urogenital tract, resulting in dysbiosis and therefore recurrent stone formation. Surprisingly, we were not able to detect significant divergence in stone and urine microbiomes of recurrent SD patients compared to primary SD patients. However, the five patients with the highest read sequence yield per stone (> 100.000 reads/stone) had multiple stone removals within the study period, with increasing read numbers in stones but not in urine. Notably, all of them had Apatite or Apatite mix stones.

Interestingly, patients with verifiable bacteria in kidney stones displayed a prolonged LOS due to different post-operative complications. Similarly, Wagenius et al. demonstrated, that patients with positive stone cultures harbor a greater risk for post-operative urosepsis, than those with just positive urine culture [27]. So the evidence of pathogens in the kidney is a tremendous risk for infectious complications.

The major limitation of our study was the limited number of patients for which sufficient microbial biomass was retrievable, impairing statistic correlations.

We also want to address the problem of a suitable control group in urological microbiome research. Catheterized urine from healthy individuals would be the most suitable control to reveal disease-related dysbiosis. Since this cohort was not available due to ethical reasons, we investigated the voided mid-stream urine of volunteers without prior stone disease. Hence, we do not consider the data obtained from voided urine as a control, but rather as a benchmark and reference, to detect bacteria unique to patients with stone disease.

As previously described, we observed a sex-specific voided urine microbiome in healthy men and women (p-value 0.002, Supplementary Fig. 3A), but not in catheterized urine of patients with SD (Supplementary Fig. 3B) [28, 29]. Moreover, we observed no difference in stone microbiome between man and women. This result is certainly influenced by the approach of urine collection, since voided urine represents a combination of bladder, urethral, and perimeatal microbiota. Therefore, with the available data, we are not able to conclude if the difference of sexes between healthy urine and urine form patients with stone disease is due to dysbiosis.

Additional studies including whole metagenome sequencing are required to improve our understanding of the involvement and role of bacteria in kidney stone disease. Moreover, concordant analyses of patients` voided and catheterized urine, stones and gut microbiome are necessary to illuminate the extent of dysbiosis and interactions of the microbes to prevent stone formation, tailor antibiotic therapy and circumvent postsurgical complications.

The current study showed that patients presenting with features of metabolic syndrome displayed a distinct stone microbiome compared to metabolically fit patients. They displayed a significant enrichment of classical gastrointestinal bacteria, such as E. coli, Shigella, Klebsiella, Enterococcaceae, Proteus and Sphingomonas. Stones of patients without features of metabolic syndrome were characterized by Ureaplasma and Staphylococcaceae.

Generally, patients with bacteria in their kidney stones exhibit a longer length of stay, presumably due to more complex care.