Abstract / Summary
Abstract Background The urogenital microbiome has been implicated in bladder cancer, but the extent to which urine reflects the microbial community in tumour tissue remains unclear. We compared tumour tissue, adjacent normal tissue, and urine from the same patients to identify compartment-specific microbial signatures and infer their functional potential. Methods Paired tumour tissue, adjacent normal tissue, and urine were collected from 20 patients with histologically confirmed urothelial carcinoma. Bacterial DNA was processed alongside sampling, extraction, and no-template amplification controls. Five 16S rRNA gene variable regions were sequenced on an Illumina NovaSeq 6000 platform and integrated using the Short Multiple Regions Framework. Diversity, differential abundance, linear discriminant analysis effect size, exploratory receiver operating characteristic analysis, and PICRUSt2 functional prediction were performed. Results Good's coverage was higher in tumour tissue than in urine (P = 0.012) and in adjacent normal tissue than in urine (P = 0.0021); Chao1 also differed between adjacent normal tissue and urine (P = 0.02). Community composition differed among the three compartments (Adonis, P = 0.01, pseudo-R2 = 0.07). Planctomycetota was more abundant in adjacent normal tissue than in urine. At the genus level, three genera, including Geobacillus, were enriched and 17, including Anaerococcus, were depleted in tumour tissue relative to urine; Coprococcus was more abundant in tumour than in adjacent normal tissue. Exploratory analysis of c_Cytophagia, s_ Hydrogenophaga atypica , and s_ Vibrio diabolicus yielded individual areas under the curve of 0.600, 0.625, and 0.771 and a combined value of 0.897. PICRUSt2 predicted 40 differentially abundant genes or proteins and lower abundance of the superpathway of haem biosynthesis from uroporphyrinogen III in tumour-associated communities. Conclusions Microbial profiles differed mainly between tissue and urine, whereas tumour and adjacent normal tissue were more similar. The three candidate discriminating taxa and the predicted haem-biosynthesis difference are hypothesis-generating findings that require independent validation.