Using integromics to better understand and treat colorectal cancer

MD Anderson’s Colorectal Cancer Moon Shot™ is committed to the prevention and early detection of the third most common cancer diagnosed in men and women in the United States, as well as improving therapeutic approaches to the disease, which is expected to cause more than 50,000 deaths this year. The moon shot is part of the Moon Shots Program™, a collaborative effort to accelerate the development of scientific discoveries into clinical advances that save patients’ lives.

Jeffrey Morris, Ph.D., professor of Biostatistics, and his colleagues are addressing a challenge posed by the complexity of colon and rectal cancer, which leads to major differences in responses to treatment and patient survival.

“Some cancers have two or three very well-defined biomarkers that, if found in a patient’s tumor, can be reliably used to guide treatment,” says Morris. “For many cancers, it’s not that simple, and colorectal cancer is one of those.”

Morris is leading a project to tackle these complexities by more clearly defining the molecular characteristics of colorectal cancer (CRC) subtypes, which will allow for more personalized and effective treatments. This requires a deep analysis of patient samples and integration of multiple data sets into a complete picture of the disease using a method known as “integromics.” This integromic approach provides an integrated analysis of genetic variations, expression of genes and regulation of genes by nongenetic factors.

He spoke with Cancer Frontline about the moon shot’s efforts to leverage integromics into clinical assays and novel therapies to improve outcomes for CRC patients.

Can you briefly explain the discovery and significance of the consensus molecular subtypes of colorectal cancer? How do these change the traditional view of colorectal cancers?

One of the key problems in any cancer, including CRC, is the heterogeneity of the cancer across patients. We know that the molecular characteristics of each patient’s tumor is different, and these differences may be the primary reasons why patients don’t respond equally to a particular treatment or why some patients’ cancers are more aggressive than others.

Together with several moon shot colleagues, including Scott Kopetz, M.D., Ph.D., David Menter, Ph.D., and Bradley Broom, Ph.D., we initiated an international Colorectal Cancer Subtyping Consortium to try to establish consensus subtypes of CRC that would help explain this heterogeneity. This was an incredible cooperative effort that combined data from more than 4,500 patients in 22 studies with information from existing subtyping systems. Through that collaboration, we were able to identify four consensus molecular subtypes (CMS), which we published in 2015 in Nature Medicine.

The patients within each CMS are much more alike than patients from different CMS, so the identification of these subtypes allows us to redefine the disease and study it in a more precise manner. In fact, the molecular characteristics of each subtype suggest it may be possible to personalize therapy for each CMS, and we are now working to better understand each subtype in order to treat them more effectively.

What is “integromics”? And what is the value of this approach in understanding these subtypes?

Integromics is a term we use to describe the act of integrating together information across many molecular types and resolution levels to uncover deeper knowledge into the characteristics of the cancer. 

Modern biomedical research generates a vast array of molecular data at different resolution levels, ranging from DNA to RNA to protein. At each level, we also obtain epigenetic data including methylation, histone modifications and microRNA, along with environmental factors like the microbiome. Each of these data sources provide particular information on the underlying biology, but are merely a part of the whole picture. Only by integrating it together can we hope to gain a complete picture of the key mechanisms underlying the cancer and its subtypes. 

We are hopeful this deep molecular characterization will enable the identification of subsets of patients responding to existing therapies and enhance the discovery and development of novel CMS-specific therapeutic targets.

How is the Colorectal Cancer Moon Shot using integromics to improve patient care?

We work primarily with two sources of multi-platform genomics data – publicly available data from The Cancer Genome Atlas (TCGA) data and internal data from several hundred of our own CRC patients. At MD Anderson, we have the unique opportunity to access and analyze both of these large data sets, allowing us to better understand CRC biology.

Our team has overcome significant data management challenges and developed innovative statistical methods to pursue this goal. A few years ago, there were practically no established statistical methods for these types of integrative analyses, and the MD Anderson biostatistics and bioinformatics research groups have taken leadership in the development of such tools. 

Using what we learn through integromics, we are now working with internal collaborators, led by Dipen Maru, M.D., and Raja Luthra, Ph.D., to deploy a clinical assay to better classify a patient’s tumor into a specific CMS. With that, our hope from the integromics approach is to learn fundamental information about the tumors that can help guide the therapeutic discovery process.

Can you describe the clinical assay that your team is working on?

In order to utilize CMS in the clinic, it is necessary to have a clinical assay for classifying patients’ tumors into one of the four CMS. The original Nature Medicine paper reported a CMS classifier, but it required fresh-frozen tissue and measurements on nearly 6,000 genes, which is not feasible for clinical use.

Our moon shot team has worked extensively to develop an improved CMS classifier and adapt it for clinical use.  We first performed an extensive survey of potential classification methods to build a more effective CMS classifier than that previously reported. We also wanted our assay to be useful on formalin-fixed paraffin-embedded samples commonly found in the clinic. We developed an assay that could be used with either sample type and required profiling of just 100 genes.

We are currently going through CLIA certification for this classifier so it can be used in the clinic.  We are also attempting to refine this classifier for use with metastatic samples and to build more relevant patient-derived xenograft (PDX) models.

How is your team working to evaluate new and existing therapies in the context of these subtypes in order to identify and optimize the best treatments for each subgroup of patients?

As I mentioned, the four CMS have distinct molecular characteristics, and we hope to identify existing or novel therapeutics to target these characteristics. This would allow us to perform CMS-based precision therapy for CRC, with the potential to greatly improve patient outcomes.

We have taken a multi-modal approach and have engaged several Moon Shots platforms, including the Institute for Applied Cancer Science (IACS) and Center for Co-Clinical Trials, to work toward developing and testing therapeutic targets with relevance in specific CMS.

First, we are attempting to classify patients in existing clinical trials by CMS to assess whether their subtype is predictive of response to therapy. We also have preclinical studies underway to investigate some novel therapeutics under development by IACS.

Second, we are using lentivirus and CRISPR screens to identify potential CMS-specific therapeutic targets, which also involves the development of CMS-specific preclinical models, including PDXs that can be used in these screens. 

Finally, we have initiated studies making use of the clinical classifier to enrich for patients with specific CMS, and we are evaluating other completed and ongoing randomized studies to understand how CMS predicts activity to novel therapies.

We are incorporating information from our integromic analysis of patient samples to help guide this process, and we are hopeful this strategy can lead to target discovery with the greatest potential for success when taken into the clinic. Moon Shot funding enabled our highly-integrated collaborations and accelerated progress, making these groundbreaking discoveries possible when they otherwise would not have been.