Validation of the Association Between the Abundance of Three Bacterial Genera in Stool Samples and the Risk of Gallbladder Cancer in South American Patients With Gallstones

Gallbladder cancer (GBC) is a highly lethal disease with a low incidence worldwide, but it is relatively frequent in indigenous populations of the Americas. The absence of specific symptoms and the lack of early detection tests often lead to delayed diagnosis and a very poor prognosis. Biomarkers identification for early detection are currently unavailable. The gut microbiome plays an important role in the modulation of individual metabolic and immunologic functions. Although alterations in the gut microbiome have been reported in gallbladder cancer with several genera proposed as potential risk biomarkers, these findings remain largely exploratory and lack consistent validation across independent cohorts, particularly in populations with a high GBC incidence. In our previous research, we carried out metagenomic sequencing of DNA extracted from faecal samples from 178 South American patients (n=48 with GBC and n=130 with gallstones) from the EULAT Eradicate GBC study. We found that the abundance of the bacterial genus Veillonella was higher, whilst that of Adlercreutzia and Gordonibacter was lower, in the faecal samples from GBC patients. The aim of the present study is to validate the three identified associations using an independent set of South American stool samples (n=40 from GBC and n=80 from gallstone disease patients). DNA extracted from the faecal samples will be pre-processed using Illumina DNA prep 1/4 for library preparation and sequenced on a NovaSeq X 25B 300c lane machine. Raw data will be pre-processed following a cleaning protocol with bbduk (bbmap-version 38.93). Cleaned reads will be analysed using the PathSeq pipeline in GATK (v4.1.6.0). Reads aligning to the human reference genome (GRCh38) will be removed, and remaining reads will be mapped to microbial reference databases obtained from the European Molecular Biology Laboratory (EMBL-Broad Institute). Analyses will be performed using established reference sets and default parameters, with a minimum clipped read length of 50. Microbiome data will be analyzed using the R phyloseq framework. Batch effects will be assessed through Bray-Curtis distances and PERMANOVA, low-prevalence taxa (\<10% of samples) will be filtered, and outliers identified via principal component analysis and Mahalanobis distance removed. Rarefaction will be applied to calculate Alpha diversity, while beta diversity will be assessed using Aitchison distance on CLR-transformed abundances. To validate the three associations identified, the newly generated bacterial abundances will be analysed using robust logistic regression (glmrob, robustbase R package), adjusting for potential confounding factors and correcting the probability values for multiple comparisons using the Bonferroni correction.