Human Microbiome and Cancer
The microbiome is comprised of microorganisms that live in and on us and contribute to human health and disease.
The microbiome refers to the collection of all of the gene sequences from a community of microbes in the human body. High-throughput DNA amplification and sequencing technologies to characterize the microbial communities provide data for investigations of microbial associations with human disease. Our goal is to create and use established and new cohorts with fecal and oral specimens to prospectively evaluate the association between the human microbiome and cancer risk.
While microbiome research has grown exponentially over the past several years, findings have been difficult to reproduce across studies. The variability induced by sample collection and handling has not been systematically assessed. Therefore, in addition to etiologic studies, DCEG investigators are pursuing a multi-pronged approach to address microbiome methodologic issues to validate optimal collection, storage, and analysis and to understand how to best interpret the complex data.
Etiologic Studies
Prospective study of the oral microbiome and mortality and cancer: Poor oral health and hygiene have been associated with a number of cancers, mortality, and other chronic diseases, which suggests a role for the oral microbiome in the development of these conditions, but few studies have examined direct measures of the bacteria. DCEG investigators are assessing the association between the oral microbiome and incident cancers of the bronchus/lung, colorectum, esophagus, head/neck, hepatobiliary tract, pancreas, small intestine and stomach, and all-cause mortality, using a case-cohort design within the Agricultural Health Study, the NIH-AARP Diet and Health Study, and the Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial (PLCO). This multi-cancer, multi-cohort project allows us to test several hypotheses simultaneously, provide greater insight into the role of the oral microbiome in cancer at several sites, and to investigate cross-sectional associations with various exposures such as diet, body mass index, and smoking.
Lower gastrointestinal cancer studies: The association between the human gut microbiome and colorectal cancer is an important focus of our research. We conduct 16S ribosomal gene sequencing and shotgun sequencing for microbiome analyses of the colorectum using fecal samples as well as tissue samples. We also evaluate microbial metabolites in the blood and fecal samples.
Upper gastrointestinal cancer studies: Gastric cancer is currently the model of bacterially associated cancer; Helicobacter pylori (H. pylori) has been classified as a Group 1 carcinogen in humans by the International Agency for Research on Cancer, indicating agents with the highest level of epidemiologic and mechanistic evidence. Research within DCEG has evaluated how risk factors for cancer, such as tobacco, body mass index, and pepsinogen levels, are associated with microbial diversity using samples collected in China. We have tested multiple methods for collecting tissue microbiome in the upper GI tract and explored the microbiome of tumors and adjacent normal tissue. New studies will explore how H. pylori and other upper GI carcinogens alter the gastric and esophageal tissue microbiome, whether the tumor microbiome can provide diagnostic or prognostic information for upper GI cancers, and will evaluate associations between the oral microbiome and upper gastrointestinal cancer risk in prospective studies.
Breast cancer studies: As the gut microbiome may play a role in the etiology of breast cancer by regulating hormonal, metabolic, and immunologic pathways, we have been investigating the role of the fecal and oral microbiome in relation to breast cancer risk in case-control and prospective studies.
Cross-sectional studies of the microbiome and cancer risk factors: The microbiome may be related to cancer risk through changes related to cancer risk factors. In order to comprehensively study the association between the oral microbiome and extensive cancer risk factors, including tobacco use, alcohol consumption, and periodontal disease, we conducted an oral microbiome analysis of individuals aged 14-69 within the 2009-2012 cycles of NHANES. The alpha and beta diversity data are publicly available on the NHANES website, while the raw sequencing files, amplicon sequence variant tables, and the read count and relative abundance data are available within the Research Data Center. The data for the quality control samples in each sequencing batch are available in SRA. (Vogtmann E et al. Representative oral microbiome data for the US population: the National Health and Nutrition Examination SurveyExit Disclaimer. Lancet Microbe 2022.)
Methods Development
DCEG investigators are evaluating and standardizing fecal and oral sample collection protocols to reliably measure the human microbiome in the context of large-scale population studies. Several issues are being considered, such as, preservation of a microbial signature, stability of samples under field conditions, preservation of samples to maximize multi ‘omics assays, and long-term storage of samples in the freezer.
Standard reference materials: It is crucial to develop appropriate standard reference materials for quality control of samples in large epidemiologic studies. DCEG investigators have developed a large number of aliquots from five individuals (i.e., healthy adult, adult on a low carbohydrate diet, adult with a high body mass index, adult with inflammatory bowel disease, and healthy young person) to optimize the ability to observe microbial differences between these quality control samples. In addition, we developed a community of 45 bacterial strains, which serve as the quantitative “gold standard.” For oral quality control samples, we created a chemostat with an oral sample of saliva, tongue scraping, and plaque, plus oral bacterial communities grown on biofilms. These samples are available to national and international studies. Routine inclusion of quality control samples will allow a laboratory to assess its own performance over time and may be useful for calibration for pooling samples or meta-analyses.
Microbiome Quality Control (MBQC) project: The degree of standardization in microbiome measurement necessary for translation to large-scale studies is early in its development. Sources of variation in microbial profiling must be evaluated and optimized so that independent epidemiologic studies can be pooled. DCEG investigators helped coordinate and conduct the Microbiome Quality Control (MBQC) study, a collaborative effort designed to comprehensively evaluate methods for measuring the human microbiome. The MBQC study evaluated the impact of various DNA extraction, sequencing, and bioinformatics methods for fecal samples.
Feasibility of setting up cohort studies: Large, prospective, population-based cohorts are well suited to address numerous important epidemiologic questions about the pathophysiology of human diseases. Given the growing understanding of the role of human microbiome in diverse health conditions, it is imperative to implement an appropriate sample collection strategy. DCEG investigators are conducting numerous feasibility studies to evaluate whether fecal and oral samples can be collected in existing cohorts, prepaid health plans, or screening populations. These samples are also planned to be collected in the new Connect for Cancer Prevention Study cohort.
Statistical analysis methods: Investigators from the Biostatistics Branch have been evaluating statistical methods for microbiome analysis and creating new methods to analyze and understand the microbiome. For example, they have shown that a simple method of analyzing beta-diversity data had power comparable to more complex procedures. One computes the mean beta-diversity distance from the sample to a set reference, such as stool samples from the Human Microbiome Project. This mean distance is a scalar quantity that can be analyzed by standard statistical methods. In recent work to detect associations of microbiome features with lung cancer risk, these mean distances were studied as time-dependent covariates in a Cox model and were statistically significantly associated with lung cancer incidence.
For more information, contact Rashmi Sinha.