Abstract 6244: Comparative diversity analysis of microbial composition across tissue and fecal samples in multi-cohort, multi-study cancer analyses

Abstract The microbial ecosystems in tumors, surrounding tissues, and the gut (which influences processes throughout the body) offer new opportunities for oncology diagnostics and treatment, but inconsistencies in metadata quality and parameters across abundant public data hinder generalizable conclusions. To address these issues, we analyzed alpha and beta diversity metrics across multiple studies, focusing on how variables like collection site or cancer type influence microbiome profiles. We compiled a list of over 180 publicly available studies from the European Nucleotide Archive that sequenced human stool and/or tissue. Studies without publications or sufficient metadata were removed, and remaining studies were narrowed to those using 16S rRNA gene amplicon sequencing, leaving 38 studies with 10, 796 samples. Using Qiita, an open-source multi-omics platform, samples were split into fecal and tissue. All fecal samples were analyzed as their own category and came from patients with various cancer types (FEC). Each tissue sample was placed into at least one of the following categories: all cancer tissue samples (CAN), tissue from patients with both healthy adjacent biopsies and cancerous biopsies (ADJ), and tissue from healthy donors and matched cancer patients (HEA). Some samples fell into multiple categories, but each category was analyzed independently. After default QC and rarefying, we used QIIME2, a microbiome bioinformatics platform, to apply diversity metrics. We explored microbial differences across cancer types using Bray Curtis (BC) and Jaccard diversity metrics for tissue (CAN) and fecal (FEC) categories. In the CAN category, PERMANOVA results for both beta diversity metrics showed a statistically significant difference across the six cancer types (p < 0.00001). The same result was replicated in the FEC category. In the CAN category, Kruskal-Wallis results for Shannon diversity (SH) showed a significant difference across cancer tumor types (p = 1.5e-28). The same result was observed in the FEC category (p = 4.6e-48), indicating that cancer type may be related to microbial diversity differences in both contexts. In the ADJ category, significant differences were observed between adjacent and cancerous tissue for both BC (p < 0.00001) and SH (p = 0.017). In the HEA category, significant differences were found between matched healthy donors and cancer patients for both BC (p < 0.00001) and SH (p = 0.00004), highlighting variations driven by tissue context and disease state. Finally, our findings reveal substantial and statistically significant variability using both BC and SH diversity when comparing across the unique study ID in both CAN (BC p<0.00001, SH p=3.9e-38) and FEC (BC p<0.00001, SH p=4.1e-48) categories. This underscores the need for standardized guidelines in microbiota sample collection and processing to enhance comparability in future analyses. Citation Format: Tiffany Zhang, Kylie Chan, Caitlin Guccione, Bertalan Takács, Antonio González, Gail Ackermann, Kathleen Curtius, Rob Knight. Comparative diversity analysis of microbial composition across tissue and fecal samples in multi-cohort, multi-study cancer analyses [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2025; Part 1 (Regular Abstracts); 2025 Apr 25-30; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2025;85(8_Suppl_1):Abstract nr 6244.