Send to printer »

Feature Articles : Oct 1, 2010 (Vol. 30, No. 17)

Profiling Looms Large in Cancer Research

Academics Spearhead Efforts to Enhance Discovery and Validation of Biomarkers
  • Elizabeth Lipp

Protein profiling, the quantitative assessment of protein-expression levels, has been at the vanguard of a brave new drug pipeline for the last four or five years. However, while the technologies that make protein profiling possible continue to evolve, they have not yet fulfilled their potential.

“There are a lot of questions that researchers are trying to answer, but so far, at the early detection level, there hasn’t been much success,” noted Paul Tempst, Ph.D., a researcher at Memorial Sloan-Kettering Cancer Center who spoke at Select Bioscience’s “Cancer Proteomics” conference held recently in Berlin.

“Proteomics hasn’t taken a biomarker to a clinical trial, yet. But we have promising candidates. NCI set up a system to evaluate technologies that make up a classic discovery/verification pipeline, but we now need to apply this pipeline to sample sets.”

Ralph Schiess, Ph.D., researcher at the ETH—Institute of Molecular Systems Biology, agreed. “The identification of biomarkers for diagnosis, prognosis, and treatment selection—or really the lack thereof, at present—has been a barrier to the realization of personalized medicine in the cancer field.” 

Dr. Tempst’s approach to protein profiling is unique to his lab, he said, in that instead of only looking for concentrations of protein they are actually measuring differences in enzymatic activity. “By correlating the proteolytic patterns with disease groups and controls, we have shown that exopeptidase activities contribute to the generation of, not only cancer-specific but also cancer type–specific, serum peptides. So there is a direct link between peptide marker profiles of disease and differential protease activity.” Because of this, Dr. Tempst explained that “the patterns we describe may have value as surrogate markers for detection and classification of cancer.”

Dr. Tempst and his colleagues are focused on the relationship between exopeptidase activities and metastatic disease. “We monitored controlled, de novo peptide breakdown in large numbers of biological samples using mass spec, with relative quantitation of the metabolites, using magnetic, reverse-phase beads for analyte capture and a MALDI-TOF MS read-out,” Dr. Tempst noted.

In a preliminary prostate cancer study, his group found a significant difference in activity levels of a single aminopeptidase in serum from patients with metastatic prostate cancer as compared to primary tumor-bearing individuals and normal healthy controls. However, there were no differences in amounts of the target protein, and this potential biomarker would have been missed if quantitative levels of protein had been the only criterion of selection.

Many studies evaluating correlations between enzyme activity and cancerous states were executed in the 1950s and 1960s, according to Dr. Tempst. Acidic phosphatase activity was used as a blood-based marker for prostate cancer long before the PSA test was put into general use in the 1980s.

“These older observations on enzymes, including aminopeptidases, by a number of different groups caught our attention. This convinced us that we were headed in the right direction,” he added.

Dr. Tempst and his team are currently developing robust biomarker assays. “Our studies on patient samples and mouse models of cancer led to identification of single enzymes (as opposed to complex panels) that can now be individually monitored.”

Serum Protein Profiling

Serum protein profiling has a number of advantages, including convenience for the patient and allowing for frequent screening of patients—especially younger women or those with hereditary predisposition. In addition, protein patterns in serum reflect pathologic changes and specific disease-related changes result in different protein expression, noted Rob Tollenaar, Ph.D., professor of surgical oncology at Leiden University Medical Center. He presented work on proteomic expression done in concert with Dr. Tempst’s lab.

This methodology has not been without its challenges. “Sample handling is problematic,” he said, particularly when there are differences in how samples are handled, whether they are left too long at room temperature, and what sort of tubes are used—“all these things can compromise your data.”

The objective of Dr. Tollenaar’s study was to assess the feasibility of a mass spec approach for the detection of breast cancer. Using a randomized block design, pre-operative serum samples obtained from 78 breast cancer patients and 29 controls were used to generate high-resolution MALDI-TOF protein profiles.

“The spectra generated using C8 magnetic beads assisted mass spec were smoothed, binned, and normalized after baseline correction. Preliminary data suggest that the high sensitivity and specificity indicate the potential usefulness of serum protein profiles for the detection of breast cancer,” said Dr. Tollenaar. “This method and technology are ready for the analysis of a larger patient series.”

Cancer Biomarkers

The themes of discovery and validation resonated throughout the conference. Dr.  Schiess presented a two-stage strategy for discovery and initial validation of serum biomarkers corresponding to specific cancer-causing mutations that involved the use of mass spec assisted discovery, verification, and validation of disease biomarkers. “I take a systems approach and study the mechanisms of the disease by using a mouse model to study tumors with silent progression,” he said. “I used prostate cancer as a model, but I believe applications can be made to other cancers as well.”

Dr. Schiess integrated his study of mouse genetics with proteomics techniques for the diagnosis and stratification of patients with prostate cancer. “We identified a set of biomarkers predictive for the genetic status in human prostate cancer patients, thus identifying potential responders to cancer therapies targeting specific pathways.” 

In the initial discovery phase, Dr. Schiess and his group detected and identified N-linked glycoproteins with distinguishable expression patterns in primary normal and diseased tissue. “The proteins identified in the initial phase will be subjected to targeted MS analysis in plasma samples using highly sensitive and specific selected reaction-monitoring technology. Since glycosylated proteins, such as those secreted or shed from the cell surface, are likely to reside and persist in blood, the two-stage strategy is focused on the quantification of tissue-derived glycoproteins in plasma.”

Dr. Schiess noted that the focus on N-glycoproteome not only reduces the complexity of the analytes, but also targets an information-rich subproteome—relevant for remote sensing of diseases in the plasma. “The discovery and validation workflow allows for the robust identification of protein candidate panels that can be selectively monitored in blood plasma at high sensitivity in a reliable, noninvasive, and quantitative fashion.

“We further discovered serum biomarkers for the prognosis of localized prostate cancer providing additional noninvasive prognostic markers for cancer aggressiveness, and thereby, supporting the decision of active surveillance or immediate surgical intervention.”

Dynamic Response Networks

“There are scientific and practical reasons for pursuing individualized treatment of cancer,” said Serhiy Souchelnytskyi, Ph.D., associate professor at the Karolinska Institute. “My primary activity involves proteome profiling of human breast cancer to allow generation of networks suitable for systemic analysis. Initially, it was not my intent to focus on breast cancer, as this could very well be applicable to other cancers, but breast cancer proved to be a good model.”

Dr. Souchelnytskyi’s methodology of choice is Dynamic Response Network (DRNet). “DRNet is a tool that can be used in the clinic to manage genetic and proteomic information to predict disease development and response to treatments,” he explained. “Among other things, DRNet can discriminate malignant vs. benign neoplasia—up to 70 percent of breast neoplasia are not malignant. This provides possibilities for clinical applications such as the individualization of treatments.”

Dr. Souchelnytskyi presented two case studies demonstrating the clinical applications of DRNet that focused on improved diagnostics and selection of treatment in one case and selection of treatment of resistant breast cancer in the other.

“The case-studies indicated that individual differences are so significant that selection of the most efficient treatment requires assessment of each patient individually,” said Dr. Souchelnytskyi. “Tumorigenesis studies indicated specific network features of the main steps of tumorigenesis. Proteomics is the most reliable in reflecting the functional status of cells and tumors, which is required for DRNet building.”

When a treatment is only 80% effective, noted Dr. Souchelnytskyi, you don’t want to be in that other 20%. However, with first-line treatment, there are strict guidelines to treat. “With second-line, there are more options, but less certainty, and this is where we come into the picture. Cancer can be aggressive or indolent. If it grows, go after it; with malignancy you have to be careful. I’ve seen cases where tumors have been activated to metastasis. We step up and suggest ways for treatment and let the clinician decide.”

Omic Screens

Researchers have a range of protein-profiling methodologies at their disposal. In his presentation, William M. Gallagher, Ph.D., associate professor of cancer biology at University of California Davis, talked about how tissue microarrays and digital slide scanning technologies can greatly speed up biomarker development.

“We have actually stopped most of our omics-based discovery programs and shifted more of our efforts toward validation of our findings via high-throughput antibody-based assays on tissue microarrays—which, in a sense, can also be included within the proteomics sphere,” said Dr. Gallagher.

A core activity within Dr. Gallagher’s lab centers on the creation and use of tissue microarrays to provide a means for screening large numbers of clinical samples on a simultaneous basis. “This assay affords its own problems, namely surrounding downstream analysis via pathologist-based interpretation.” 

“Accordingly, we have developed an automated approach to assist investigators to quantify expression of biomarkers that have been assessed via immunohistochemistry.” This approach, termed IHC-MARK, can discriminate tumor-specific expression of biomarkers at different subcellular levels, e.g., nuclear, cytoplasmic, and membranous. “We are currently in the process of commercializing this technology via our spin-out company, OncoMark.

“IHC-MARK has been road-tested so far on multiple tumor and marker types,” explained Dr. Gallagher, who is also CSO of OncoMark.

One of the key bottlenecks that Dr. Gallagher’s group initially faced was the difficulty in performing downstream validation of identified biomarkers of interest. “This required us to refocus our efforts on this crucial phase and develop new solutions such as the automated image-analysis approach to overcome these problems.

“Our findings suggest that such immunohistochemistry-based surrogates may provide more clinically applicable assays than complex gene-expression signatures. The result is a comprehensive biomarker development pathway that extends from discovery through validation on tissue microarrays, which is yielding clinically relevant biomarker panels for predicting outcome in breast cancer.”

Sidebar: Improving Reproducibility

Proteomics starts with the discovery effort—customers look to identify protein biomarkers, noted Dominic Gostick, Ph.D., director of biomarker MS business at AB Sciex. “We provide the tools and technologies they need to do comparison biology. We developed the AB Sciex TripleTOF 5600 with that in mind. It gives us the ability to see deep and find more peptides than ever before within a given proteome.”

This product integrates comprehensive qualitative exploration, rapid profiling, and high-resolution quantitation workflows on a single platform, Dr. Gostick reported. It combines the highest sensitivity detection, high-resolution with at least five times better acquisition speed, and stable ~1 ppm mass accuracy over days of acquisition.

“Reproducibility used to be a challenge. The key to peptide quantification is not just seeing a peptide, but doing cross validation in multiple labs across multiple days, and the robustness we now have in our platforms really enables us to move to the next level. When I talk to pioneering customers, I am encouraged and excited by the increase in reproducibility.”