November 15, 2014 (Vol. 34, No. 20)
When targeting genes, drug developers may opt for knockdown via CRISPR or silencing via RNAi. These weapons, however, may fit different battle plans.
Tailoring healthcare decisions based on biological information represents the promise of personalized medicine.
A key hurdle slowing progress toward tailored therapeutics is developing appropriate biomarkers for diagnostics, prognostics, and prediction.
Despite much research, only a trickle of biomarkers has made it into clinical practice. Researchers, however, are pursuing novel avenues and making headway by dissecting the subtle architecture of normal versus cancerous cells. New approaches include scrutinizing aberrant glycans, microRNAs shed from tumors, and epigenetic modifications, as well as mining and merging the vast stores of multi-omics data.
There are a number of key issues that must be considered prior to the successful development of a biomarker. “One of the most fundamental aspects that is too often overlooked is the specimen procurement process,” suggests Chad R. Borges, Ph.D, assistant professor, chemistry and biochemistry, The Biodesign Center, Arizona State University. “Many sites store biospecimens, at least temporarily, at −20°C. Unfortunately, some samples such as serum and plasma do not freeze completely until the temperature is −30°C or lower.
“We have found that this causes certain proteins in the samples to oxidize—and this can happen quite rapidly. Also, while −20°C frost-free freezers do not result in sample oxidation faster than nonfrost-free units, they do lead to sample dehydration. Many clinical study collection sites, such as some doctor’s offices, do not have the capability to store plasma or serum appropriately (−80°C).”
Another issue relates to trial design itself. “It’s important to precisely define the clinical context for which the biomarker will be used,” Dr. Borges continues. “Will it be for diagnostics, prognostics, or for prediction? The Institute of Medicine has provided a three-part framework for biomarker evaluation that includes analytical validation, qualification (based on an association between the biomarker and the clinical endpoint of concern), and utilization—which defines the specific context of the proposed biomarker use.”
Mindful of these issues, Dr. Borges’ group is developing oncology biomarkers using glycan “node” analysis for detecting and monitoring cancer.
“Our approach pursues the idea that monosaccharide and linkage-specific glycan polymer chain links and branch points (“glycan nodes”) differ in control patients versus those with various types of cancers,” explains Dr. Borges. “The construction and display of aberrant glycans is a hallmark of nearly every known type of cancer and likely facilitates metastasis. Since glycotransferases (GTs) mediate glycan structure, aberrant GT expression or activity leads to irregular glycan production.”
Using gas chromatography-mass spectrometry (GC-MS), Dr. Borges quantifies unique glycan features, such as core fucosylation, in a single analytical signal.
“This is a new type of bottom-up approach in which we fragment the glycans prior to analyzing the sample,” notes Dr. Borges. “We are able to derive a complete, targeted analysis in one readout that includes dozens of analytical signals, instead of hundreds or thousands of signals. This methodology can be applied to whole body fluids (without needing prepurification), is inexpensive, and employs a clinically accepted and practiced analytical modality (GC-MS).”
According to Dr. Borges, this approach can detect some early-stage cancer-induced glycan changes: “We are also encouraged by new findings suggesting that some glycan nodes are correlated with malignancy progression in several types of cancer.”
Colorectal Cancer Biomarkers
Colorectal cancer is the third most common cancer and cancer death in the United States. Each year, more than 150,000 people are diagnosed with this disease, and more than 50,000 succumb to it. However, it is also one of the most preventable cancers, and if diagnosed early, it is readily cured.
“The problem is that the current method for screening for colon polyps, the asymptomatic precursors of early-stage colon cancer, is primarily colonoscopy, which is expensive and inconvenient,” says William M. Grady, M.D., professor and director of translational research, Division of Gastroenterology, University of Washington School of Medicine.
In response to the need for a more convenient, accurate, and noninvasive screening test for colorectal cancers, a new test that is based on advances made in our understanding of the molecular pathogenesis of colorectal cancer over the last 30 years has been developed and recently approved by the FDA.
“Colorectal cancer is a complex disease with substantial molecular heterogeneity with every cancer having a wide array of gene mutations and epigenetic changes,” observes Dr. Grady. “The new test, called Cologuard, uses these alterations as biomarkers for the basis of a noninvasive stool DNA-based screening assay. It detects both fecal blood and three molecular alterations, methylated NDRG4, methylated BMP3, and mutant KRAS. It has been shown to detect >40% of polyps and >80% of cancers with >90% specificity.”
Dr. Grady is developing additional early-detection biomarkers in his lab as well as predictive biomarkers to accurately determine which colon cancers will respond to specific forms of chemotherapy. His research team recently assessed whether a molecular subclass of colon cancers that have the CpG island methylator phenotype (CIMP), characterized by a high frequency of genome-wide aberrant DNA methylation, can predict response to chemotherapy.
In a study published in the journal Gastroenterology, CIMP cancers (compared to other colon cancers) were more sensitive to the chemotherapeutic drugs 5-fluorouracil and irinotecan.
Although the number of clinically useful biomarkers for directing the prevention and treatment of colon cancer is still limited, Dr. Grady feels the next 3–5 years should produce many more biomarkers that will be used to direct the management of colorectal cancer: “The promise of precision medicine is now a clinical reality. Advances in identifying genetic alterations are driving new biomarker development for use as early detection markers, prognostic markers, and markers that more accurately predict treatment responses.”
Liquid miRNA Biopsy
MicroRNAs (miRNAs or miRs) are small, noncoding RNAs that regulate the expression of multiple genes. They play significant roles in a diverse set of physiologic and pathologic processes. “There has been considerable work, especially in the last four years, demonstrating that miRNAs are shed into the circulation from distant tissues and that their patterns of expression differ in the types and states of diseases,” notes Anton Wellstein, Ph.D., professor of oncology, pharmacology, and medicine, Georgetown University Medical School.
Dr. Wellstein says that miRNAs in the circulation are remarkably stable and can provide signatures for distinguishing different types of diseases (such as cancer) and disease prognoses as well as revealing new targets including altered signaling pathways.
“In cases such as colorectal cancers, stages are usually determined by tumor size for a predictive analysis, but utilizing biomarkers in blood could distinguish stages and different types of such cancers,” explains Dr. Wellstein. “For example, in a recent Phase II clinical study, we found that miR-296 is progressively lost during tumor progression and correlates with metastatic disease in colorectal cancer. Further, it provides a biomarker of treatment as circulating miR-296 is associated with shorter survival and poor response to treatment.”
One of the challenges remaining before miRNAs can be more commonly utilized as biomarkers is population diversity. “Every patient and cancer is different, even if it is the same type such as colon or breast cancer,” comments Dr. Wellstein. “Studies are now focusing on developing and analyzing about a dozen or so markers to obtain an miRNA signature as well as seeking the best miRNA to evaluate. Other challenges in monitoring treatment are determining when to evaluate since drugs differ in time of action and how soon they translate into outcome.”
According to Dr. Wellstein, the concept of liquid biopsies is making headway: “Ultimately, being able to analyze a blood sample—that is, a liquid biopsy—to diagnose and evaluate disease is faster and a less invasive tool for patients. We continue looking for new disease-relevant targets and are seeking companies to help fund our research.”
Mining large databases of genomics, proteomics, metabolomics, etc. is creating new hope for identifying biomarkers because “omics” data is a growing goldmine of information. But where does one start? “A standard way employs statistical approaches that rely only on the information contained in the omics dataset itself,” says Melinda Baker, Ph.D., solution specialist, GeneGo, Thomson Reuters. But there is a problem with this approach.
“While this method can produce accurate biomarkers, the discovery cohort must be sufficiently large to derive an accurate analysis,” points out Dr. Baker. “An improved approach focuses on pathway analysis that provides the biological context for identified genes. No gene is an island. Genes communicate with each other. It is important to have a good grasp of the underlying biology and network of biological interactions.”
Thomson Reuters developed and markets software suites for functional analysis of next-generation sequencing (NGS) called MetaCore™ and MetaBase™. These combine literature evidence of transcriptional, signaling, and metabolic pathways to define molecular classes such as transcription factors, ligands, receptors, and much more.
“Once investigators derive a data set, these tools can simultaneously analyze multi-omics data to find or predict functionally significant gene variants or proteins,” explains Dr. Baker. “Pathway-based approaches move away from sample numbers only and rather enable the analysis of small patient cohorts or multiple experiments that used different platforms.
“For example, if your analysis uncovers 10 nonoverlapping proteins in a small number of patients, you could determine they are related through a similar pathway. In this case you use biological knowledge to fill in the blanks.”
Another example Dr. Baker recalls is an internal analysis the company performed on a published article studying colon cancer patients and featuring multiple types of data from NGS to protein expression: “Pooling all the data and analyzing with a single platform allowed us to identify several relevant target genes that not only correlated with relapse but also were frequently seen in inflammatory bowel disease and related disorders.”
Dr. Baker says that new goals for biomarker research are less to identify blockbuster drugs and more to understand and improve efficacy in clinical trials for the right patients: “It’s best to know the underlying biology. People are becoming excited because NGS data can be integrated with multiple other omics data, and this provides the missing link for biomarker discovery and validation.”
Delivering on the promise of personalized medicine has proven to be very challenging. “Considerable resources have been devoted to drug-gene relationships, that is, pharmacogenomics, hoping that genetic variants would lead the way,” notes Lin Li, Ph.D., associate director, biostatistics, BioStat Solutions. “This has failed to deliver, despite new breakthroughs in high-density SNP genotyping and DNA sequencing.
“While the FDA’s listing of Pharmacogenomic Biomarkers in Drug Labeling includes 158 total entries, little progress has been made on genes that correlate with drug efficacy. According to a recent study, only 12% of drugs licensed between 1998 and 2012 had pharmacogenomic biomarker information in their label.”
Dr. Li says one solution is to redefine the genomic feature of interest and utilize a region-based testing strategy: “We recommend broadening investigations and not limiting analysis to a single variant but instead including other genomic regions in the analysis. There is an underlying structure to genetic data that has biologic relevance. In particular, linkage disequilibrium structure and blocks are natural groupings that are not random. Association studies using linkage disequilibrium can help identify genes that contribute to phenotypic variation.”
There are many advantages to employing a region-based strategy. “It has the potential to improve power over examining one variant at a time,” explains Dr. Li. “This alleviates the multiplicity burden by orders of magnitude and better leverages underlying linkage disequilibrium structure. In addition, it can improve the chances of replication. A gene is a consistent unit across populations and clinical trials. Finally, it may enable the detection of complex genetic effects such as the joint effect of multiple SNPs across the same gene or more than one gene.”
Dr. Li’s take-home message is that “whether evaluating the impact of genomic variation on treatment response or informing development of similar compounds, genomic region-based testing that employs a strategic combination of gene-based methods and traditional single-SNP testing may improve chances for success in tailoring specific biomarker development.”