Vol. 17 •Issue 22 • Page 16
Molecular Profiling in Cancer Diagnostics
A new technique is bringing various disciplines together to offer new standards in oncological prognosis and treatment.
Molecular profiling is an evolving practice, relying on different arrays, including cDNA, genomic, protein and tissue microarrays, or multiple markers to classify tumors and predict outcomes, according to Dave Hoon, MSc, PhD, director of molecular oncology, John Wayne Cancer Institute, Santa Monica, CA.
A New Standard
As array technology and sensitivity is improving, it is becoming easier to distinguish biomarkers and determine which are significant to the analysis, Dr. Hoon reported. The new approach allows pathologists to examine multiple markers using a minimal amount of tissue, providing more information than ever before.
The technology will be used primarily for initial tumor classification and outcome predictions. In the future, health professionals will be capable of characterizing tumors beyond their location into specific tumor types–based on gene expression or genomic DNA, Dr. Hoon said. Additionally, physicians will be able to use the data to predict prognosis and even how the patient will respond to different treatment options. Depending on the availability of surgical specimens, the technique also could be used to monitor a patient’s progress, Dr. Hoon theorized.
Kenneth J. Bloom, MD, FCAP, medical director, Clarient Inc., Irvine, CA, pointed out another use of molecular profiling includes confirming that a second malignancy is a metastasis from a prior tumor (rather than a new primary). Solid data has proven that metastatic lesions look almost identical to the primary tumor when compared with an expression array, he noted.
“At the end of the day, once we know certain characteristics of certain tumors, we’ll be able to look at specific markers and classify them rapidly,” Dr. Hoon added.
Stumbling Blocks
Molecular profiling is still evolving as a discovery tool and has not yet met standard criteria for acceptance (Table), according to Dr. Bloom, although some have attempted to extend its use into the clinical realm. The technology lacks large scale randomized clinical trials and studies assessing sensitivity, specificity and reproducibility. While molecular profiling has tremendous potential, the technology is still in the early stages of inception and faces no shortage of challenges ahead.
Costs
Because the technology is so new, costs are considerable–for the moment. The estimated cost of establishing a laboratory, including personnel, equipment, reagents and validations, could easily exceed $1,000,000, according to Dr. Bloom’s estimates. Even if a facility has an existing molecular laboratory, environmental conditions likely have to be improved because of stringent ozone and oxygen requirements to prepare it for molecular profiling, he said.
The instrumentation is expensive right now, Dr. Hoon agreed. On top of those costs, molecular profiling current standards will require trained personnel dedicated to performing the analysis and also to interpret the results. However, these costs will diminish over time, he stressed. As the technology becomes more widespread, instrumentation costs will drop, and robotics and computers will take some of the burden from specialized employees, he predicted.
Expertise
Until molecular techniques become more automated, implementing the technology will take training costs and time. “All new laboratory techniques require training,” Dr. Bloom explained, but “molecular profiling requires specific expertise that generally is not available in a routine pathology laboratory.”
Traditional methods tend to be standardized and automated, and are much easier to perform, while molecular profiling may require additional training on pre-analytic and analytic variables, and amplification.
It is important that technologists are trained and aware of the goals of molecular profiling and the implications of what could go wrong, Dr. Hoon added, explaining that training also applies to sample preparation. As the array is more sensitive, quality control becomes vital to achieving maximum output.
Samples
Two major issues affect sample collection for molecular profiling– reproducibility and quality. A major challenge is agreeing on what the sample should consist of, Dr. Bloom conveyed. “Some advocate using laser capture microdissection to include only tumor cells in the analysis, while others advocate including tumors as well as surrounding stroma and inflammatory cells,” he said. Also, different areas of a tumor will show differences in expression. Little continuity exists when it comes to sample collection, Dr. Bloom stressed.
Additionally, profiling still requires frozen or fresh tissues, Dr. Hoon reported. Techniques have not yet been developed using paraffin. Also, for nucleic acids, the sample quality has to be very good. Dr. Hoon noted that improving fixation and preparation techniques is necessary, but now healthcare professionals have a better understanding of how tissue preparation affects quality. In the past, tissues may have been left out on the bench and not cared for appropriately, whereas now, protocols are enforced to ensure that samples are immediately prepared or frozen to protect integrity.
Results
The problem of reproducibility extends beyond the analysis to the data results. Though many researchers are experimenting with profiling around the world, each group uses its own set of genes, Dr. Hoon noted.
Even without comparing results to other research, the sheer amount of data generated by the technique can be daunting. It can be difficult to extract “reality from noise,” Dr. Bloom said, and the algorithms used are equally as important as the array technology. Bioinformatics becomes a key stage to interpreting the data and forming meaningful results that can be presented to the physician.
Clearly, molecular profiling has some weaknesses, “but hopefully these will be overcome, as we are still going through a learning curve,” Dr. Hoon said.
To overcome one of these challenges, Dr. Yue (Joseph) Wang, PhD, associate professor of electrical, computer and biomedical engineering, Virginia Polytechnic Institute and State University Advanced Research Institute, Arlington, recommends looking beyond the walls of the lab.
Bioinformatics
As Drs. Bloom and Hoon mentioned, one of the key, if not the biggest, hurdle molecular profiling faces is also one of its biggest strengths–the massive amounts of data that the technique generates.
Dr. Wang, fellow of the American Institute for Medical and Biological Engineering, believes he and his peers have the answer. “The challenge here is that molecular profiling produces a huge volume of raw data that needs to be analyzed to extract useful and meaningful information and knowledge,” he stated. The general solution—quantitative modeling and analysis—requires expertise and tools that engineers can provide, Dr. Wang suggested.
Molecular profiling technologies provide a complete molecular fingerprint of an individual’s health, he explained. “If we apply such technologies to large populations, we also can obtain the molecular background patterns associated with health and disease.”
Dr. Wang broke the process down into three steps:
Analyzers (a joint team of biologists and engineers) identify molecular markers associated with different disease categories and states, i.e., “the players.”
Next, they discover interactions between these molecular markers associated with functional pathways of interest, i.e., how the molecules interact with each other to form certain cellular functions.
Lastly, they predict and confirm the functions of these molecular markers.
“The challenge of cancer treatment has been to target specific therapies to pathogenetically-distinct tumor subtypes, to maximize efficacy and minimize toxicity. However, tumors with similar histopathological appearance can follow significantly different clinical courses and show different responses to therapy,” Dr. Wang stated. “The recent development of molecular profiling provides an opportunity to take a genome-wide approach to predict clinical heterogeneity in cancer treatment.”
Dr. Wang anticipates many roles for engineers to play in the evolving cancer diagnostics field, especially in interpreting the data generated by molecular profiling.
A Turning Point
Today, molecular profiling is being used as a discovery tool, looking for novel pathways or markers that aid in clinical prediction, primarily in universities and research centers. The expression arrays are most successful with tumors of known origin. The technology lacks validation, consistency and accuracy. Studies have yet to prove that molecular profiling can outperform the current methods in predicting disease behavior and responsiveness to therapy. Analyses algorithms need to be standardized and selection criteria need to be developed. Today, molecular profiling is an unrealized dream. But tomorrow, it may be the new gold standard in cancer diagnostics.
Kerri Penno is assistant editor at ADVANCE. She can be reached at [email protected].
Table: Three Criteria for Validated Practices Laboratory testing must fill three criteria:
1. Clinical Validation: the test has been used in at least two randomized trials and has been shown to identify groups with significantly different risks of relapse, survival or response to treatment.
2. Technical Validation: the assay should be sensitive, specific, reproducible, calibrated to outcome and easy to interpret.
3. Usefulness: clinicians will use this test to aid in a specific clinical decision point.
– Kenneth J. Bloom, MD, FCAP
Technology in Practice
Molecular profiling has many uses and likely will impact diagnostics, prognosis, treatment plans and disease monitoring. The concept is based on the knowledge that each patient is unique, and their disease, responsiveness to different treatments, and genetic and environmental history ultimately will affect their prognosis. While the technology to sample a tumor and specifically identify its origin, prognosis and the best treatment course has not yet been validated, the following companies are using a combination of molecular and traditional techniques to profile their patients on a smaller scale.
Accumetrics
As many techniques are used in profiling disease, different reasons predicate their use. One reason for profiling is to ensure efficacy of treatment, Robert S. Hillman, president and CEO, Accumetrics, San Diego, said, but ensuring that treatment levels do not surpass the minimal level needed to achieve the desired results is equally important.
For this reason, Accumetrics offers the VerifyNow™ assay, which determines to what extent aspirin inhibits the COX-1 receptors in individual patients. This technology allows physicians to prescribe a personalized dose for each patient, ensuring that the end goal (inhibiting COX-1 receptors) is met, while keeping the dosage as small as possible to prevent bleeding and other side effects.
The company recently received FDA clearance for another assay, the VerifyNow P2Y12, which measures ADP receptor inhibition and Plavix® (clopidogrel) effectiveness.
Beyond the role that COX-1 receptiveness plays in treatment and management of disease, decreased COX-1 inhibition has been clinically correlated with a higher probability of stroke, myocardial infarction, cerebrovascular events and death, Hillman noted. Therefore, the assays can be used to determine the degree of this risk.
Clarient
While developers at Clarient Inc., Irvine, CA are waiting for validation of molecular profiling technology, Kenneth J. Bloom, MD, FCAP, medical director, told ADVANCE that all Clarient’s current techniques—even the pathologist looking at a slide under a microscope—are types of classification.
Currently, Clarient offers PCR-based molecular profiling and a suite including immunohistochemistry (IHC) for key regulatory molecules such as ER, PR, HER-2, EGFR, CD20 and CD117, as well as flourescence in-situ hybridization looking for genes that may be amplified, such as HER-2, BCL-1 or EGFR, common translocations, trisomies and deletions.
Clarient is focusing on developing low cost, simple IHC methods to identify antibodies to key proteins identified by molecular profiling techniques, according to Dr. Bloom—for now. Looking forward, Clarient is investigating the possibility of using molecular profiling to determine if a current malignancy is a metastasis from a previous malignancy or a new primary.
GE Healthcare
Discovery Systems, a segment of GE Healthcare, Piscataway, NJ, offers multiple products and solutions that contribute to molecular profiling, according to Andy Bertera, head of marketing, and gene and protein discovery–such as the IN Cell Analyzer (used for cellular analysis), Ettan™ 2-D DIGE (a technique for proteomics research to compare protein samples from normal and diseased tissues), MegaBACE (used for DNA sequencing and genetic variation analysis), as well as the software necessary to analyze data generated by molecular profiling studies.
GE’s CodeLink™ microarray enables analysis of the entire genome on a single slide, allowing users to study changes in, and levels of, gene expression associated with disease. Bertera described the results as “key to interpreting the predisposition to disease, the treatment protocol to be applied and the potential acceptance of a drug, personalized for the patient.”
CodeLink has been used to research the effects of myocardial infarction on gene expression, and how dietary fat composition affects the initiation and promotional stages of tumor development in the colon.
Key Players
Kenneth J. Bloom, MD
Dr. Bloom has been medical director at Clarient Inc., headquartered in San Juan Capistrano, CA, since August 2004. Previously, Dr. Bloom served as senior medical director at Irvine, CA-based US LABS. Prior to that, Dr. Bloom spent more than 20 years serving in senior academic, consultative and clinical roles with leading hospitals and healthcare enterprises principally in the specialization of cancer care.
Dr. Bloom’s academic posts have included 10 years as assistant professor of pathology at Chicago-based Rush Medical College, as well as one year as a visiting professor in the Department of Computer Science at DePaul University.
Dave S.B. Hoon, MSc, PhD
Dr. Hoon is the director of the Department of Molecular Oncology, at the John Wayne Cancer Institute in Santa Monica, CA, where he is investigating novel molecular approaches to diagnose minimal tumor metastasis.
Dr. Hoon holds an MSc. degree in pathology from the University of Manitoba (Canada), and a PhD in tumor metastasis from the University of Saskatchewan (Canada). He has served on review study sections for the National Cancer Institute, the Department of Defense Army Breast Cancer Program, the American Cancer Society, and the National Institute of Allergy and Infectious Diseases.
Yue (Joseph) Wang, MS, PhD
Dr. Wang is an associate professor at the Virginia Polytechnic Institute and State University (Virginia Tech) in Blacksburg, VA. He received his PhD in electrical engineering from the University of Maryland. His teaching interests encompass machine learning, bioinformatics, stochastic signals and systems, biomedical imaging, signal detection and estimation, and image informatics.
He is actively involved in researching intelligent computing, machine learning, pattern recognition, statistical visualization, and advanced imaging and image analysis, with applications to computational bioinformatics and bioimaging. He has served on review committees for the National Cancer Institute, the Department of Defense Army Breast Cancer Program, National Science Foundation, and the National Institute of Biomedical Imaging and Bioengineering.
