Mark Pegram, M.D.
The use of cutting-edge technology and bioinformatics to inform clinical decision-making in oncology is still a ways off, according to Mark Pegram, MD, the Susy Yuan-Huey Hung Professor of Oncology and Director of the Stanford Breast Oncology Program, Stanford University, Palo Alto, California. At the 9th Annual New Orleans Summer Cancer Meeting, Dr. Pegram said the “lofty goal” of targeted therapeutic “cocktails”—which will be needed to address the molecular diversity of tumors—is proving hard to achieve.
Circulating Tumor Cells
Circulating tumor cells as an alternative to serial biopsies of metastatic lesions has great appeal, but the uptake of this technology has been somewhat anemic.
One problem is obtaining a consistent definition of a circulating tumor cell. The cell must be positive for cytokeratin, must have a nucleus, must have a negative control, must be negative for leukocyte cytoplasm (white cell markers), and the nucleus must fit inside the cytoplasm.
“These are the things measured using a huge variety of different approaches for defining and capturing [circulating tumor cells],” he said. The most clinically advanced is the CellSearch System, which uses an antibody/ferrofluid combination to attach specifically to circulating tumor cells, and magnets to draw those cells out of the blood sample to be stained and identified. The test enumerates the number of circulating tumor cells in a patient with metastatic disease, and this is correlated with overall survival.
“The problem with this assay is that it is not sensitive enough to capture [circulating tumor cells] in early stages of disease. While enumeration of [circulating tumor cells] is prognostic, let’s be honest: that’s not what we are interested in,” Dr. Pegram said. “We are interested in predicting response to treatment.”
He has observed that while some clinicians are “enamored” of this technology and do use it, others realize that it holds little value over routine restaging with radiographic studies. The assay also reveals little as to what is happening in these cells, and the small number of circulating tumor cells captured—five or so—is insufficient for fully deciphering the tumor, he said.
Capturing more cells could help, and that is what microfluidics-based cell separation does. This new, simpler technology passes blood through a membrane, separating larger tumor cells from other blood elements and yielding thousands of cells upon which clinically relevant tests can be performed.
“The approaches that have much higher yields will be more useful because they will be informative as to what cells are doing at a molecular level,” he predicted.
Even more sophisticated blood-based technology will someday be better able to capture the genetic heterogeneity of advanced solid tumors at a gene-expression level so they can be compared with the primary tumor. This, however, will present other challenges.
“In one blood sample there are multiple populations of [circulating tumor cells] that are different from another. This will pose a diagnostic challenge and a treatment challenge, as well, if we find unique targets within the same patient at the same time,” he said. “Until we can come to terms with the complexity of solid tumor malignancies, we can’t make informed decisions.”
At this point, guideline committees “have not latched on to [circulating tumor cells] as a ‘must’ in clinical practice,” he indicated, calling circulating tumor cell determination a “consideration,” but one lacking in great value until emerging technologies can interrogate circulating tumor cells at a molecular level.
Genomics and Drug Development
The promise of genomics was to identify mutations within a tumor and thus allow the clinician to concoct a tailored therapeutic cocktail. In reality, however, the scenario is infinitely complex. Within a single MCF-7 human breast cancer cell, for instance, 157 chromosomal break points have been found.
“We have rich genomic information in a tumor cell, but this does not tell the doctor how to treat the patient,” he said.
The Cancer Genome Atlas (TCGA) Network, in its examination of its first 507 breast cancer samples, revealed only four frequently mutated genes out of 50 that were identified: PIK3CA, TP53, MAP3K1, and GATA3.
“This was a stunning observation,” commented Dr. Pegram. “We thought we would discover multiple new therapeutic targets in breast cancer and therefore have home runs in drug development, but we found only four, and all four were already known to be common mutations.”
Drugs are already targeting PI3K, the other three frequent mutations are not druggable, and the rest of the 50 genes are low-frequency mutations (affecting about 2% of breast cancers) for which pharmaceutical companies are unlikely to invest. “This will pose a challenge because our current models of drug development will not survive this reality,” he predicted.
Furthermore, according to Dr. Pegram, deep sequencing identifies even more heterogeneity, revealing individual clones with different mutational profiles within the same tumor. The current next-generation diagnostics are not performing deep sequencing and therefore are not demonstrating the molecular heterogeneity that is critical for selecting the best targeted agent, he said.
Even “more sobering,” he continued, is that this complexity is present at the time of diagnosis, with further alterations piled on due to drug resistance. Cancer and genomes are not static; they are a moving target, he reiterated.
While the situation is clinically frustrating now, there is the potential to tease apart the molecular evolution of cancers with future sequencing technology, and this “extraordinary” achievement could give insights into prevention strategies.
Adding Proteomic Data
Even more complex than genomics is proteomics, the large-scale analysis of protein-expression profiles through mass spectrometry. Proteomic information on post-translational modifications in the tumor (ie, phosphorylation, glycolisation, etc) could be a useful adjunct to genomic information, producing a more “holistic view” of pathway regulation.
“The hope is that mixing proteomic work along with genomic work will facilitate our understanding of what is going on in the dynamic tumor cell,” Dr. Pegram said. “But the problem with proteomics is size: the proteome is much larger than the genome, due to alternative splicing and protein modification.”
The information desired from proteomics includes all protein-to-protein interactions, protein functions and their regulation, protein modifications, subcellular location, and protein concentrations. Current approaches do not provide all this information.
While polymerase chain reaction (PCR) testing determines gene amplification, there is no PCR equivalent for proteomics. Sequencing tools are robust in genomics, but mass spectrometry is still emerging in proteomics. Furthermore, proteomic data is “big data,” and huge servers are needed just to store the data. Novel approaches are currently being pioneered to address these issues, he said.
In summary, Dr. Pegram said, “Mutational events in cancer can yield complex and deranged pathways, but they are still highly functional and they can take the lives of our patients. We need to understand them.”
Disclosure: Dr. Pegram reported no potential conflicts of interest.
The ASCO Post, September 1, 2014, Volume 5, Issue 14