Background: The application of next-generation sequencing techniques has enabled characterization of urinary tract microbiome. Although many studies have demonstrated associations between the human microbiome and bladder cancer (BC), these have not always reported consistent results, thereby necessitating cross-study comparisons. Thus, the fundamental questions remain how we can utilize this knowledge.
Objective: The aim of our study was to examine the disease-associated changes in urine microbiome communities globally utilizing a machine learning algorithm.
Design, setting, and participants: Raw FASTQ files were downloaded for the three published studies in urinary microbiome in BC patients, in addition to our own prospectively collected cohort.
Outcome measurements and statistical analysis: Demultiplexing and classification were performed using the QIIME 2020.8 platform. De novo operational taxonomic units were clustered using the uCLUST algorithm and defined by 97% sequence similarity and classified at the phylum level against the Silva RNA sequence database. The metadata available from the three studies included were used to evaluate the differential abundance between BC patients and controls via a random-effect meta-analysis using the metagen R function. A machine learning analysis was performed using the SIAMCAT R package.
Results and limitations: Our study includes 129 BC urine and 60 healthy control samples across four different countries. We identified a total of 97/548 genera to be differentially abundant in the BC urine microbiome compared with that of healthy patients. Overall, while the differences in diversity metrics were clustered around the country of origin (Kruskal-Wallis, p < 0.001), collection methodology was a driver of microbiome composition. When assessing dataset from China, Hungary, and Croatia, data demonstrated no discrimination capacity to distinguish between BC patients and healthy adults (area under the curve [AUC] 0.577). However, inclusion of samples with catheterized urine improved the diagnostic accuracy of prediction for BC to AUC 0.995, with precision-recall AUC = 0.994. Through elimination of contaminants associated with the collection methodology among all cohorts, our study identified increased abundance of polycyclic aromatic hydrocarbon (PAH)-degrading bacteria Sphingomonas, Acinetobacter, Micrococcus, Pseudomonas, and Ralstonia to be consistently present in BC patients.
Conclusions: The microbiota of the BC population may be a reflection of PAH exposure from smoking, environmental pollutants, and ingestion. Presence of PAHs in the urine of BC patients may allow for a unique metabolic niche and provide necessary metabolic resources where other bacteria are not able to flourish. Furthermore, we found that while compositional differences are associated with geography more than with disease, many are driven by the collection methodology.
Patient summary: The goal of our study was to compare the urine microbiome of bladder cancer patients with that of healthy controls and evaluate any potential bacteria that may be more likely to be found in patients with bladder cancer. Our study is unique as it evaluates this across multiple countries, to find a common pattern. After we removed some of the contamination, we were able to localize several key bacteria that are more likely to be found in the urine of bladder cancer patients. These bacteria all share their ability to break down tobacco carcinogens.
Keywords: 16S rRNA sequencing; Bladder cancer; Microbiome; Urinary microbiome; Urothelial carcinoma.
Copyright © 2023 European Association of Urology. Published by Elsevier B.V. All rights reserved.