Skip to Main Content
Principal Investigator
Name
Jennifer Davis
Degrees
Ph.D.
Institution
UT MD Anderson Cancer Center
Position Title
Instructor
Email
About this Project
Study
PLCO (Learn more about this study)
Project ID
2018-0009
Title
Colorectal cancer risk factors, risk prediction and blood-based biomarker by tumor consensus molecular subtype
Summary

Despite the availability of effective screening techniques, only 40% of colorectal cancer (CRC) cases are diagnosed at a localized stage of disease. This is likely due to a combination of factors, including screening non- compliance, limitations in the screening sensitivity and specificity, and heterogeneity of CRC biology. Specifically the existence of more aggressive tumor subtypes, which may have a shorter natural history, combined with screening inadequacies hinder our ability to detect more early stage disease and further reduce CRC morbidity and mortality. The recently described consensus molecular subtypes (CMS) of CRC include a more aggressive, mesenchymal subtype and provide a framework for stratified risk assessment, screening recommendations and prevention interventions. The four subtypes identified have distinct biology and clinical outcomes, suggesting the possibility of unique risk factors, prevention, and screening strategies. Specifically, CMS1 (Immune, 14% of cases) is associated with high micro-satellite instability (MSI), BRAF mutations and immune infiltration, CMS2 (Canonical, 37%) accounts for the largest percent of tumors and is characterized by activation of WNT and MYC, CMS3 (Metabolic, 13%) is characterized by low somatic copy number alterations, KRAS mutations and tumor metabolic dysregulation, CMS4 (Mesenchymal, 23%) is characterized by stromal infiltration, TGF-β activation, angiogenesis and worse overall and relapse-free survival (Nat Med. 2015, 21:1350). The studies in this proposal utilize the CMS framework to develop and test a risk-prediction tool and test the CMS-specific performance of a validated blood-based three-marker panel. Building on our preliminary data and using the high-quality, longitudinal data and tumor RNA from the Prostate, Lung, Colorectal and Ovarian Cancer Screening Trial, we plan to test the associations of CMS with age, smoking status, and tumor stage at diagnosis (Aim 1). Using the associations from aim 1, we plan to build and test a CMS-specific CRC risk prediction tool to facilitate screening and prevention efforts (Aim 2). We also plan to further test the performance of our validated blood-based three-marker panel across CMS and in the years prior to diagnosis (Aim 3). The combination of a CMS-specific risk prediction tool and a validated blood-based biomarker has the potential to greatly improve CRC screening compliance and early detection, leading to a reduction in morbidity and mortality from CRC. PUBLIC HEALTH RELEVANCE: PLCO As much as one-third of the United States population is not up-to-date on screening for colorectal cancer, which contributes to later stages of disease at diagnosis. Improved non-invasive screening methods and risk prediction tools are needed to reduce the colorectal cancer burden. The work proposed here seeks to improve screening efficacy and compliance by building and testing a tool to predict an individual's risk of developing four different subtypes of colorectal cancer, including an aggressive subtype, and further characterize a validated blood-based biomarker panel for its ability to detect each subtype prior to diagnosis.

NOTE: Aim 3 was not funded

Aims

Aim 1. Evaluate CMS distribution and association with race/ethnicity, age at diagnosis, smoking status, diabetes, aspirin/Non-Steroidal Anti-Inflammatory Drug (NSAID) use, diet quality, tumor stage at diagnosis and overall survival.
Hypothesis: Overall distribution of CMS will mirror that published in Guinney, et al6 and MD Anderson analyses, race/ethnicity, age at diagnosis, smoking status, diabetes, aspirin/NSAID use, diet quality, tumor stage at diagnosis and overall survival will vary by CMS.
Aim 1a. Determine CMS from formalin-fixed, paraffin embedded (FFPE) tumor tissue using a parsimonious Nanostring geneset.
Aim 1b. Compare race/ethnicity, age at diagnosis, smoking status, diabetes, aspirin/NSAID use, diet quality, tumor stage at diagnosis and overall survival across CMS.
Aim 2. Develop a CMS-specific CRC risk prediction tool to facilitate enhanced screening for subjects at increased risk for aggressive subtypes, such as CMS4 (Mesenchymal).
Hypothesis: Patient profiles can be used to predict a person’s likelihood of developing a particular CRC subtype and this prediction can facilitate risk-adjusted screening and prevention recommendations.
Aim 2a. Develop a CMS specific risk prediction tool in the Intervention Arm using both cases and controls, matched 4:1 for sex, length of follow-up and study center.
Aim 2b. Test CMS specific risk prediction tool in the Control Arm data, using cases and controls as defined in Aim 2a.
Aim 2c. Compare the CMS specific risk prediction tool with an existing tumor-site specific risk prediction tool9, 10.

Collaborators

Jennifer Davis (UT MD Anderson Cancer Center)
Robert Bresalier (UT MD Anderson Cancer Center)