March 1, 2010 (Vol. 30, No. 5)
Approaches for Overcoming Obstacles to Successful Implementation in the Clinic
Biomarkers have become some of the most desired clinical and research tools in the life sciences. They have the potential to transform a drug discovery program and save lives in the clinic if they can be discovered, developed, and validated properly.
The hurdles in bringing a biomarker to maturity have, at times, been discouraging, as there have been many early failures and only a few notable successes, such as Her2 and Herceptin. At CHI’s “Biomarker Assay Development” conference held in late January in San Francisco, researchers gathered to discuss novel approaches to discovering biomarkers and taking advantage of what they have to offer. Highlights included smarter discovery methods, tougher validation standards, logical workflows, and a variety of new platform technologies.
Pathway inhibition is a dominant strategy for developing targeted therapies and the biomarkers that go with them. By inhibiting a specific pathway (e.g., in cancer cells), you can theoretically monitor progression of disease and response to therapy based on biomarker changes within the signaling pathway. This has been an active area in cancer therapy, where physicians would like to predict a patient’s response to therapy without wasting weeks or months on a treatment that will not work.
Unfortunately, except in a couple of cases, pathway biomarkers have not produced hoped-for results in predicting patients’ tumor sensitivity to targeted therapy.
Relying on OMICS
Another approach is through omics-based development. Using the tools of proteomics, genomics, and metabolomics, potential biomarkers are sifted out of large quantities of microarray or other high-content data, with an agnostic approach to the mechanism or pathway affected. But omics technologies also have not lived up to expectations.
Research at ArQule comes in somewhere between the two extremes. Neither tightly focused on pathway inhibition alone, nor shotgunning the whole genome or proteome, scientists at ArQule are developing a circuit-based technique to biomarker discovery focusing on cross-talk between pathways.
“If a cell works similarly to an electronic circuit board, if you increase the resistance in one of the circuits, there must be corollary changes of the current, either up or down, in some parallel circuits. You just can’t quietly increase the resistance or take down the electron flow in one part of the circuit and not have something else disturbed in the circuit pathway—hence the idea of cross-talk biomarkers came into being,” explained Thomas Chan, Ph.D., CSO at ArQule.
Dr. Chan’s presentation at CHI’s biomarker conference highlighted ArQule’s recent success in identifying cross-talk biomarkers for several of its kinase inhibitor programs.
As an example, using paired tumor biopsies pre- and post-treatment with their oral c-Met inhibitor (currently in Phase II trials), the researchers found statistically significant inhibition of phosphorylated c-MET and phosphorylated FAK expression. This in-between or “Goldilocks” strategy for biomarker development might be the right-sized strategy for smaller companies that cannot afford a large-scale omics campaign but want to draw upon the overall cell-signaling system rather than isolated signals from one pathway.
Imaging Solutions for Cancer
Sampling circulating tumor cells (CTCs) in blood is a relatively noninvasive method for assessing tumor status or response to treatment. But CTCs present a number of difficulties for practical clinical use. Historically, these have been detected by antibody capture and cytokeratin staining. However, in 40–60% of cases no CTCs are recovered, even though they are likely present. These hurdles have hindered clinical use of CTCs as biomarkers.
Biocept says that OncoCEE™, its clinical biomarker platform, can capture rare CTCs with near 100% efficiency. The technology uses microfluidic channels to flow sample over a streptavidin-coated chip. The entire unit sits on a slide surface.
A cocktail of capture antibodies boosts recovery rates compared to other methods that use a single capture antibody, maintains a company official. Unwanted, non-antibody-bound cells flush through the system, keeping background levels low.
“What’s unique about our device is that there is a glass cover slip that allows users to take the entire unit, right after CTC capture and staining, to a microscope for direct manual analysis, where the person looking through the eyepiece can identify cells based on antibody stain and perform CTC enumeration,” said Farideh Bischoff, senior director, translational research and CLIA development at Biocept.
In early clinical testing, the OncoCEE was comparable to the Veridex (a J&J subsidiary) platform in its ability to enumerate CTCs from blood, according to Bischoff. It also offers an additional advantage of being able to conduct further testing on captured cells directly within channels, such as FISH for aneuploidy detection or gene-amplification studies that could be used to look for genetic biomarkers like Her2, Bischoff noted.
Harold Garner, Ph.D., executive director of the Virginia Bioinformatics Institute at Virginia Tech, introduced the hyperspectral imaging microscope, a new technology developed at the university. It enables researchers to scan each pixel of a microscopic image from 400 to 800 nanometers, quantifying as many as 15 fluorescent biomarker peaks in each scan. A single cell contains hundreds of pixels, so this type of imaging creates a very dense “cube” of data for every cell, permitting a view deep into subcellular compartments for markers of interest, he explained.
Dr. Garner’s team has put together a set of 10 markers for breast cancer and is currently optimizing similar sets for lung and colon cancer. Using data from breast cancer touch preps (where cells are collected from a tumor cell by touching a slide to it), Dr. Garner has shown that hyperspectral imaging microscopy correlates closely to standard 0 to +3 pathology scoring in treatment decision-making, but with a much higher density of information. The data is then entered into an interactive database where the user can view the contributions of each of the biomarkers across an entire image and in individual cells and tissues.
The hyperspectral imaging method addresses the problem of variability between labs in scoring of tumor cells, Dr. Garner pointed out.
“It’s very difficult to compare results from one laboratory to another. By standardizing these hybridization cocktails and the instrumentation and the database and analytical technique, we address those issues so people can use data from all over and be confident that it’s comparable,” Garner said, adding that, at the same time, it provides extremely dense biomarker data that can be used for personalized cancer therapy.
NextGen Sciences has developed a biomarker workflow using multiple reaction monitoring (MRM), a mass-spec method that uses peptides and their fragments to identify and quantitate proteins.
In the first phase, a hybrid LTQ-Orbitrap analyzes biological samples to generate a list of hundreds of potential biomarkers over a period of about six weeks. The next step is to develop assays that can multiplex up to 30 proteins at a time in order to verify the putative markers.
Because the process is iterative and proteins “fall out” of the running after testing samples, all of the proteins in the initial set can be screened, said Michael Pisano, president and CEO of NextGen. The results at this point are relative; in the next phase of the workflow the assay is converted to an absolute assay by spiking in isotope-labeled peptides for quantification.
The final phase is validation, which can comprise several levels depending on the intended use of the assay. The highest validation standards are for GLP and GCP applications that are going to be used in a clinical setting.
Although MRM has been in use for quite some time, the application of the technique to proteins as opposed to small molecules presents some new challenges, explained Pisano.
“A lot of the things you do with small molecule work do not apply to biomolecules. For example when monitoring for levels of drug or metabolites, there is no endogenous level in the patient. We always have to consider that there are components that are potentially present in the patient samples,” Pisano concluded.
Choosing the Right Platform
Biomarker discovery platforms fall into two main categories: pure research and clinical diagnostics. Because of a natural division between manufacturers of instruments for these two types of applications, even people who have worked with biomarkers extensively have generally not used both research and clinical platforms. These two groups of instruments have evolved over time in design and functionality.
However, early-stage biomarker research and clinical biomarker development are uniquely interrelated activities. There is a need for service providers to embrace both sets of technology and access a repertoire of platforms along the entire chain of biomarker development.
“There is no one platform that fits all,” said John Allinson, vp, biomarker laboratory services of Icon Development Solutions. “You need a large range of platforms to be able to give the most appropriate analytical solution to whatever the challenges are in the drug development programs you’re working with.”
Allinson particularly warns against clinically uninformed biomarker investigations. Attempting to characterize biomarkers without a grounding in physiology—such as not understanding the normal physiological range of certain protein markers—can lead to some disastrous mistakes.
“I’ve known some laboratories that have produced biomarker results data that is incompatible with living systems. It’s purely because they don’t have a thorough understanding of the clinical side of the science.”
It can be challenging to take a set of biomarkers identified during a discovery phase and validate them to acceptable clinical standards. To address this task, Rules Based Medicine (RBM) decided to take a multianalyte profile (MAP) approach. The company developed DiscoveryMAP™ technology, based on the Luminex platform, and has come up with a multianalyte profile for schizophrenia. The profile includes 51 biomarkers that provide a pattern to differentiate a healthy state from schizophrenia with diagnostically relevant sensitivity and specificity, explained Ralph L. McDade, Ph.D., strategic development officer at RBM.
The profile also contains biomarker signals that the company believes will help psychiatrists to differentiate schizophrenia from bipolar disorder and major depression. Several of the analytes are treatment-sensitive, and RBM is working with Roche to develop these and others as theranostic candidates.
Although the concept of capturing profiles and patterns for diagnosis of disease is intuitively appealing, the successful execution of a profile-based biomarker is complicated by the precision and reproducibility of results. Matrix interference has been one of the greatest challenges for RBM, which they sought to overcome by developing various cocktails of blocking solutions.
“That probably was the most complicated aspect of getting DiscoveryMAP to reliably validate in a clinical setting. Nobody had ever done that in a multiplexed environment and to my knowledge no one has since,” says Dr. McDade. “Without that you can’t have the precision that is required in the clinical laboratory.”