PLoS One. 2016 May 12;11(5):e0155362. doi: 10.1371/journal.pone.0155362. eCollection 2016.

Colorectal Cancer and the Human Gut Microbiome: Reproducibility with Whole-Genome Shotgun Sequencing.

Vogtmann E1,2, Hua X1, Zeller G3, Sunagawa S3, Voigt AY3,4,5,6, Hercog R7, Goedert JJ1, Shi J1, Bork P3,6,8,9, Sinha R1.

1Division of Cancer Epidemiology & Genetics, National Cancer Institute, Bethesda, Maryland, United States of America.

2Division of Cancer Prevention, National Cancer Institute, Bethesda, Maryland, United States of America.

3Structural and Computational Biology Unit, European Molecular Biology Laboratory, Heidelberg, Germany.

4Department of Applied Tumor Biology, Institute of Pathology, University Hospital Heidelberg, Heidelberg, Germany.

5Clinical Cooperation Unit Applied Tumor Biology, German Cancer Research Center (DKFZ), Heidelberg, Germany.

6Molecular Medicine Partnership Unit (MMPU), University Hospital Heidelberg and European Molecular Biology Laboratory, Heidelberg, Germany.

7Genomics Core Facility, European Molecular Biology Laboratory, Heidelberg, Germany.

8Max Delbrück Centre for Molecular Medicine, Berlin, Germany.

9Department of Bioinformatics Biocenter, University of Würzburg, Würzburg, Germany.

Abstract

Accumulating evidence indicates that the gut microbiota affects colorectal cancer development, but previous studies have varied in population, technical methods, and associations with cancer. Understanding these variations is needed for comparisons and for potential pooling across studies. Therefore, we performed whole-genome shotgun sequencing on fecal samples from 52 pre-treatment colorectal cancer cases and 52 matched controls from Washington, DC.

We compared findings from a previously published 16S rRNA study to the metagenomics-derived taxonomy within the same population. In addition, metagenome-predicted genes, modules, and pathways in the Washington, DC cases and controls were compared to cases and controls recruited in France whose specimens were processed using the same platform.

Associations between the presence of fecal Fusobacteria, Fusobacterium, and Porphyromonas with colorectal cancer detected by 16S rRNA were reproduced by metagenomics, whereas higher relative abundance of Clostridia in cancer cases based on 16S rRNA was merely borderline based on metagenomics.

This demonstrated that within the same sample set, most, but not all taxonomic associations were seen with both methods. Considering significant cancer associations with the relative abundance of genes, modules, and pathways in a recently published French metagenomics dataset, statistically significant associations in the Washington, DC population were detected for four out of 10 genes, three out of nine modules, and seven out of 17 pathways.

In total, colorectal cancer status in the Washington, DC study was associated with 39% of the metagenome-predicted genes, modules, and pathways identified in the French study. More within and between population comparisons are needed to identify sources of variation and disease associations that can be reproduced despite these variations. Future studies should have larger sample sizes or pool data across studies to have sufficient power to detect associations that are reproducible and significant after correction for multiple testing.

PMID:

27171425

PMCID:

PMC4865240

DOI:

10.1371/journal.pone.0155362