Metagenomic analysis of formalin-fixed paraffin-embedded tumor and normal mucosa reveals differences in the microbiome of colorectal cancer patients

Abstract

An increased risk of developing colorectal cancer (CRC) and other types of tumor is associated to Lynch syndrome (LS), an inherited condition caused by germline mutations in mismatch repair genes. We selected a cohort of LS patients that had developed CRC and had undergone surgical resection. Formalin-fixed paraffin embedded (FFPE) tissue blocks from matched colorectal and normal mucosa were used for genomic DNA extraction with a commercial kit and sequenced by high-throughput sequencing. A metagenomic approach enabled the taxonomic and functional identification of the microbial community and associated genes detected in the specimens. Slightly lower taxonomic diversity was observed in the tumor compared to the non-tumor tissue. Furthermore, the most remarkable differences between tumors and healthy tissue was the significant increase in the genus Fusobacterium in the former, in particular the species F. nucleatum, as well as Camplylobacter or Bacteroides fragilis, in accordance with previous studies of CRC. However, unlike prior studies, the present work is not based on directed detection by qPCR but instead uses a metagenomic approach to retrieve the whole bacterial community, and addresses the additional difficulty of using long-term stored FFPE samples.

Figure 1

Taxonomic ecological alpha diversity analysis by tissue, gender, gene mutated, tumor tissue, tumor location, grade of differentiation and stage. Analyses are shown for paired-only samples with more than 200 and 400 reads. Diversity within samples is estimated by the Shannon diversity index. In the case of tissue and gender, the p-values of the pairwise comparison is shown, regardless of the significance, whereas in the case of the gene in both normal and tumor tissue, p-values are shown only in significant pairwise comparisons (p-value < 0.05). NT: normal tissue, TT: tumor tissue; F: female; M: male; LSC: left-sided colon (descending + sigmoid); RSC: right-sided colon (ascending + transverse); E: early stage (I + II); A: advanced stage (III + IV).

Figure 2

Taxonomic ecological beta diversity analysis by: tissue, gender, mutated gene, tumor location, stage and grade of differentiation. Analyses are shown for paired-only samples with more than 200 and 400 reads. Diversity among samples is shown through canonical correspondence analysis (CCA) and Adonis tests for significance. NT: normal tissue, TT: tumor tissue, DIFF: grade of differentiation.

Figure 3

Discriminant taxa between tumors and normal mucosa. (Top) Boxplots of Wilcoxon Signed Rank tests for paired samples for Fusobacterium genus (top) showing statistically significant differences between normal mucosa tissue (NT) and tumor tissue (TT), for three sets of reads: all reads without normalization (a), paired-only samples with more than 200 (b) and 400 (c) reads. (Bottom) Linear Discriminant Analysis (LDA) Effect Size (LEfSe) plot of taxonomic biomarkers identified in the FFPE tissue microbiome from five sets of reads: all reads without normalization (d), paired-only samples with more than 200 reads (e), paired plus unpaired samples with more than 200 reads (f), paired-only samples with more than 400 reads (g), and paired plus unpaired samples with more than 400 reads (h). Stars indicate Fusobacterium lineage, circles indicate Bacteroides lineage.

Figure 4

Volcano plots showing the differential abundance of taxa, at genus (top) and species (bottom)level, identified using ANCOMII analysis, between normal mucosa and tumor tissue. Three subsets of samples were used for this analysis: (A) all samples, (B) only samples with more than 200 reads in both samples in the pair, and (C) only samples with more than 400 reads in both samples in the pair. Features are sorted by p value. Significant dots showing features above p-value < 0.05 (dashed line) and log fold change (log2FC) >  = 2, | p-value < 0.01 (solid line) | log2FC ≥ 2, are colored according to the phylum they belong to.

Figure 5

Volcano plots showing the differential abundance of gene functions identified, using ANCOMII analysis, between normal mucosa and tumor tissue. Two subsets of samples were used for this analysis: (A) all samples, (B) only samples with more than 127 reads in both samples in the pair. Features are sorted by p value. Significant dots showing features above p-value < 0.05 (dashed line) |log fold change (log2FC) >  = 2, and p-value < 0.01 (solid line) | log2FC ≥ 2, are colored according to the subcategory they belong to.