Global Metanalysis of the Urinary Microbiome in Bladder Cancer Patients - Expert Commentary
Data were collected from three published studies conducted in China, Croatia, and Hungary and were then integrated with data from Bukavina et al.’s prospective cohort in the U.S. After adjusting for geographic variability and catheterization contamination; several genera were associated with bladder cancer, including Enterococcus, Acinetobacter, Micrococcus, and Ralstonia. Interestingly, these bacteria all have the ability to metabolize a carcinogen present in cigarette smoke (polycyclic aromatic hydrocarbons). In these studies, most patients had non-muscle invasive bladder cancer (NMIBC). Most patients were male (94%). Urine was collected as a midstream sample in 76% of patients and from a catheter in 24% of patients. The median age was 67 years in the Croatian cohort, 56 years in the Chinese cohort, and was not reported for the cohort in Hungary. The most prevalent orders among bladder cancer patients were Clostridiales, Enterobacterales, and Bacteroides. In healthy controls, these were Bifidobacteriales and Bacillales. Across all cohorts, there was a significant presence of the genera Escherichia/Shigella, Streptococcus, and Gardnerella in bladder cancer patients and a high prevalence of the genera Alloscardovia, Millisia, Lactobacillus, and Finegoldia in healthy controls. A difference in the diversity of microorganisms between healthy controls and bladder cancer patients was detected in the Hungary cohort but not in the China and Croatia cohorts. This was also associated with samples collected via catheterization versus voiding.
These findings were subsequently tested in the prospective U.S. cohort by Bukavina et al. Data was collected from catheterized samples for 29 bladder cancer patients and 31 healthy controls. Similar to the other cohorts, there was a high prevalence of Acinetobacter and Escherichia/Shigella among bladder cancer patients. Contrary to the previous cohorts, there was a significant difference in diversity measurements between healthy controls and bladder cancer patients in the U.S. cohort (p < 0.0001). The researchers then performed a meta-analysis using the four cohorts and reported that 97 out of 548 genera were differentially abundant in bladder cancer patients relative to controls. Samples differed more by country and collection method than by disease. The researchers trained a machine learning model for prediction and found that diagnostic accuracy was higher with catheterized urine.
The findings from this study underscore the challenge of generalizability for urine microbiome studies across countries. Samples were consistently more strongly associated with geographic location and method of urine collection than disease. Accordingly, these variables must be accounted for and reported in detail in studies on urine microbiome. Furthermore, the investigators strongly recommend using catheterized urine in subsequent studies. Establishing urine collection and computational data analysis standards across future studies will be critical for a deeper understanding of the urinary microbiome in bladder cancer.
Written by: Bishoy M. Faltas, MD, Director of Bladder Cancer Research, Englander Institute for Precision Medicine, Weill Cornell Medicine
Reference:
Read the Abstract