Performance of 16S Metagenomic Profiling in Formalin-Fixed Paraffin-Embedded versus Fresh-Frozen Colorectal Cancer Tissues

SIMPLE SUMMARY: The analysis of colorectal cancer (CRC) gut microbiota can reveal crucial aspects of carcinogenesis and variation of treatment responses. Formalin-fixed, paraffin-embedded (FFPE) tissues represent an invaluable resource for studies in cancer genomics; however, their use in high-throughput metagenomic studies has been questioned due to several limitations in the DNA quality. In this study, we evaluated the impact of sample preservation on CRC-associated microbiota characterization. Using 16S rRNA sequencing and RNA in situ hybridization (RNA-ISH), we found differences in the comparison between paired FFPE and fresh frozen (FF) tissues, mostly derived from contamination issues. A quality index was also outlined to potentially assess the reliability of microbiome profiling obtained from FFPE DNA samples. These results suggest that tissular CRC microbiome studies should preserve internal coherence by using either FFPE or FF samples but not necessarily both. ABSTRACT: Formalin-fixed, paraffin-embedded (FFPE) tissues represent the most widely available clinical material to study colorectal cancer (CRC). However, the accuracy and clinical validity of FFPE microbiome profiling in CRC is uncertain. Here, we compared the microbial composition of 10 paired fresh-frozen (FF) and FFPE CRC tissues using 16S rRNA sequencing and RNA-ISH. Both sample types showed different microbial diversity and composition. FF samples were enriched in archaea and representative CRC-associated bacteria, such as Firmicutes, Bacteroidetes and Fusobacteria. Conversely, FFPE samples were mainly enriched in typical contaminants, such as Sphingomonadales and Rhodobacterales. RNA-ISH in FFPE tissues confirmed the presence of CRC-associated bacteria, such as Fusobacterium and Bacteroides, as well as Propionibacterium allowing discrimination between tumor-associated and contaminant taxa. An internal quality index showed that the degree of similarity within sample pairs inversely correlated with the dominance of contaminant taxa. Given the importance of FFPE specimens for larger studies in human cancer genomics, our findings may provide useful indications on potential confounding factors to consider for accurate and reproducible metagenomics analyses.