Biomarkers, by whatever name, have been a cornerstone of medicine at least since the days of Hippocrates and the four humors. In more modern times we’ve come to see an excess of blood glucose as a marker for insulin deficiency (and therefore, for diabetes), while Her2 levels in a biopsy may indicate the treatment course for a breast cancer patient.
According to the generally accepted definition of the NIH’s biomarker definition working group, a biomarker is an objectively measured and evaluated indicator of a normal or pathogenic process, or of the response to a therapeutic intervention. In other words, it’s used to tell whether things are normal and whether treatments are working.
The search for such indicators, their validation, how to turn them into viable assays, and what they may mean for pharmaceutical development and patient diagnosis and care were among the many topics discussed and debated at CHI’s recent “Biomarker Assay Development” conference.
There is a wealth of data in biorepositories awaiting mining. Hospitals routinely store biopsies and the like as FFPE samples. This tissue—going back ten years or more—associated with clinical histories that provide a trove of collateral information about the samples and the disease process, can be analyzed and used for biomarker discovery.
Immunohistochemistry (IHC) remains the tool of choice to examine protein expression in tissue, said Marina Guvakova, Ph.D., senior research investigator and adjunct assistant professor at the University of Pennsylvania. As long as there exists an appropriate antibody to a candidate marker, the technique is inexpensive, works on archived tissue, requires only relatively small amounts to sample, and can be used with a microscope to see the morphology and localize the protein of interest. But IHC has its drawbacks.
Dr. Guvakova is interested in determining the likelihood that breast cancer remains contained where it is (in situ), as opposed to becoming invasive. To do that, she looked for markers on FFPE tissue representing the breast cancer progression series: normal tissue, benign lesions, carcinoma-in-situ (both with and without associated invasion), and two groups of invasive breast cancer tissue. The problem is that IHC is generally scored by a pathologist on either a +/- or a four-point scale (negative, weak, moderate, or strong), not the “truly continuous measurement of the protein expression which is needed for biomarker studies,” she said.
Dr. Guvakova and her colleagues have developed image-based analytical tools and algorithms to measure the protein expression in a tissue without relying on very sophisticated equipment that quite often is not found in an academic lab setting, and without using the very expensive software being developed for big pharma.
They start with a single antibody with a single chromogen, utilize a highly sensitive digital camera to acquire a monochrome image, and use the NIH’s free ImageJ software to evaluate area, size, and pixel-value statistics. To compensate for nonspecific staining, they always measure relative rather than absolute intensity.
After finding concordance on IGF-1R expression between their results and the published literature, the group went on to examine expression of a new potential biomarker, Ras-related protein 1 (Rap1). They “found that not only were normal and benign tissue different from invasive tissue, but measurements of candidate biomarkers on continuous scales allowed us to identify the difference between normal benign tissue and carcinoma in situ with an invasive component,” Dr. Guvakova explained.
When More than One Is Too Much
Often a cadre of molecules, rather than just a single species, is used as a biomolecular marker.
Vibha Jawa, Ph.D., senior scientist, clinical immunology, and her team at Amgen are concerned about aggregation of recombinant proteins like monoclonal antibodies in pharmaceutical formulations. Proteins can aggregate when present in high concentration or due to unknown factors associated with storage. Such aggregates can cause an adaptive immune reaction—a concern “that the FDA has commented on.”
As a first step, they investigated whether peripheral blood mononuclear cells (PBMCs) from normal humans would respond to artificially aggregated monoclonals in culture and found evidence of a response to the challenge after 20–24 hours. A multiplexed panel assayed 47 different cytokines produced by the culture, of which “there were seven or eight different cytokines that we regard as the monocyte-derived cytokine signature that came up consistently for the 22 donors that were looked at.”
Yet an early innate inflammatory response is not the same as a mature adaptive immune response—which can take about 14 days to manifest and may involve different cell types—and the cytokine signatures for the two are different. So to try and correlate the two, the Amgen researchers used a mAb known to be clinically immunogenic and compared it to those that are not.
Aggregates of monoclonals that were already immunogenic as monomers turned out to be significantly more potent activators of monocytes (precursor to the macrophage) in the one-day culture than were aggregates of nonimmunogenic antibodies. In a related set of experiments, the group found that particles of about 2–10 micrometers tend to be the most inflammatory in the innate-phase assays.
“The high-level question we are asking is: ‘What particle size, and how much, is going to induce an inflammatory response?’” Dr. Jawa said. These experiments are just the first project in the process of developing a model that Amgen can use in quality-control assays. “Once we have the whole story of inflammation markers completed, we can focus more on the translatability” of the initial results to the adaptive phase of immune response.