Cancer has a reputation for stealth, a silent submarine of a disease that lurks and grows undetected before revealing itself. But the truth is that, in a biochemical sense, many cancers aren't quiet at all.
Their presence can trigger an alarmed response from the body, which may dispatch tumor-fighting agents and endeavor to choke off the malignancy's blood supply. Cancer cells themselves teem with proteins and other compounds that offer hints of the cells' intentions, hardiness, and vulnerabilities.
For years, doctors and scientists have sought to identify these telltale compounds – technically known as biomarkers – to get an early "read" on whether a tumor is present, how quickly it is likely to grow and spread, and how it will respond to therapy. They've identified dozens, the best known of which may be the prostate specific antigen, or PSA, a substance whose blood level can indicate the presence of prostate cancer.
Today, biomarker tests are used for several forms of cancer. However, most are far from 100 percent reliable, which is a serious shortcoming. Even a 90 percent accuracy rate would mean 100,000 of 1 million people tested would receive false results – and would then need further exams to prove or disprove cancer's presence.
"The term 'biomarker' applies to a range of substances – DNA, RNA, proteins, or other compounds – that can be measured as indicators of health or disease," says Dana-Farber's Charles Fuchs, MD, MPH, who leads research in the field. "In a disease as complex as cancer, there is a role for them at virtually every stage of treatment."
The term 'biomarker' applies to a range of substances – DNA, RNA, proteins, or other compounds – that can be measured as indicators of health or disease.
Though they come in many forms, biomarkers have three main uses in cancer medicine and research: determining whether an individual is at risk for or has developed a malignancy, predicting the course of disease, and deciding on proper treatment. Patients may be most familiar with the first application – diagnosing new cancers and recurrences. But the latter two uses, known as prognostic and predictive, respectively, are becoming increasingly important as cancer medicine moves into the era of personalized therapy, where treatment is tailored to the genetic makeup of each patient's tumor.
Biomarkers are hardly reclusive; they may be found in the bloodstream and, of course, in tumors themselves. They also can reside in the normal, healthy organs of people with an inherited predisposition for developing certain types of cancer.
An ideal blood biomarker would enable physicians to detect cancers when they're small and relatively easy to treat. Even the minutest tumors release proteins and other compounds into the blood that betray cancer's presence. The challenge is to detect these changes amid the normal ebb and flow of substances in the bloodstream. Because some of the changes are very subtle and vary from tumor to tumor, determining which are truly indicative of cancer is daunting. Imagine a modest uptick in the number of blue bicycles on the streets of Manhattan. Then think of the difficulty not only of discovering this increase but of determining what it means – rising gasoline prices, a fitness boom, a growing concern with the environment? Researchers encounter the same type of ambiguities in biomarkers, but with far higher stakes: markers are useful only if they point to cancer, particularly to tumors that are aggressive or malignant.
The search for biomarkers in tumor tissue faces similar hurdles. Technology has made it possible to take the genetic fingerprints of cells, the unique pattern of active genes within them. Researchers hope to use this information to classify tumors not just by the organ where they originate or their appearance under a microscope – the traditional identification method – but by their individual genetic identity, and to devise specific therapies for them. The difficulty lies in the sheer volume of data. A typical study may turn up hundreds of molecular variations between one type of tumor and the next. Culling the meaningful differences – the true biomarkers – from the far more numerous normal ones is an ongoing quest.
Biomarker research at Dana-Farber Cancer Institute runs the gamut from molecular probes of tumor tissue to the development of blood tests for different types of cancer. In many cases, biomarkers are tools of scientific research, enabling investigators to identify new subtypes of cancer and gauge the promise of new therapies.
A particularly ambitious effort is under way in Dana-Farber's colorectal cancer research program. In conjunction with the Nurses' Health Study and the Health Professionals Follow-up Study, which track participants' health over many years, investigators have banked more than 1,500 colon cancer samples from patients. The tumor samples are being scanned for gene expression with a technology platform known as DASL (developed in part by Dana-Farber's Todd Golub, MD) in an approach that Charles Fuchs describes as both comprehensive and "agnostic" – comprehensive, because the equipment reads the activity of all 24,000 genes in cancer cells; agnostic, because the process isn't influenced by preconceived ideas of what the results might be.
For each sample, researchers can draw on an extensive collection of data – not only the tumor's unique pattern of active and less-active genes, but also information on the patient's course of disease, when it started, how it was treated, and how the patient fared.
Connecting this information may help investigators identify genetic markers for how colon cancers are likely to progress and how they'll respond to certain therapies.
"This work will help us understand the behavior of cancer cells beyond what can be seen under the microscope," says Fuchs. "The goal is to survey the entire genetic landscape to identify different subcategories of tumors and factors that will guide us in the treatment of each."
An example of a discovery to emerge from this research came in a recent study led by Fuchs and Dana-Farber colleague Brian Wolpin, MD, MPH. The investigators found that in people who have undergone surgery for colorectal cancer, the prediagnosis blood levels of two proteins can indicate the individuals' chances of dying from the cancer or other conditions in the near term. Both proteins are related to insulin and rise or fall in response to lifestyle factors such as being overweight, physically inactive, and eating a Western-style diet. It's unclear whether the two proteins, IGFBP-1 and C-peptide, "are part of the actual mechanism that promotes colon cancer recurrence," says Wolpin, "But it is clear that they are markers for the risk of recurrence and death. They may help us identify which patients are most at risk for relapse and who can most benefit from additional treatment."
The goals of biomarker scientists shouldn't obscure the difficulties they face. Despite all that has been learned, only a handful of discoveries have successfully made the passage from the laboratory to clinical tests for bona fide biomarkers.
The sheer complexity of cancer and the genetic variability of tumors, even among those that look similar under a pathologist's microscope, can make it difficult to predict a tumor's clinical course. Studies of genetic activity within a single type of cancer – metastatic melanoma, say – tend to turn up hundreds of points of difference from the activity in normal cells.
"These studies often provide you with a long list of possibilities that include both the 'smoke' and the 'fire,'" the cloud of potential biomarkers shrouding a much smaller group of actual ones, explains Dana-Farber's Lynda Chin, MD.
Such correlative studies are useful in the initial stages of research, providing a "first cut" of candidate markers, she adds. But because cancer is such a diverse disorder, it's necessary to distinguish between substances that are merely associated with the disease and those that are actually driving it.
Chin offers an analogy to buying car insurance: "One of the most predictive factors insurance companies use to set rates is the color of the vehicle: People who drive red cars are more likely to be young males with a risk-taking personality; thus the color red is a correlative marker for a high-risk driver. A more definitive marker would be knowing the gene or genes that dictate a high-risk personality. That is the equivalent of what we're aiming for in our biomarkers research."
In practice, that means gathering all the information possible on a particular type of tumor – not just gene-activity patterns, but also data on how it progresses in patients and how they respond to treatment, much as Fuchs and his colleagues are doing in colon cancer. It means painstakingly examining each potential biomarker gene.
"The key is functionality," Chin says." "Does the gene play a role in the origin or progression of the cancer?" Substances identified through this function-based process can actually do double duty, serving not just as biomarkers, but also – because they help the disease progress – as targets for therapy. In their research in melanoma skin cancer, for example, Chin and her colleagues are searching for genes that enable aspiring cancer cells to migrate, invade tissue, and acquire "anoikis resistance," the ability of cells to survive independently in the blood. Such genes would provide valuable biomarkers for tumors that are likely to metastasize.
For varieties of cancer that are rarely diagnosed before they reach an advanced, aggressive stage, the search for biomarkers can be particularly vexing. Without any physical symptoms to suggest that an individual may be developing a tumor, researchers are hard-pressed to know who should be screened for suspicious blood proteins, or, indeed, which proteins to screen for.
A key step toward a solution has come in pancreatic cancer, the fourth-leading cancer killer and a disease notorious for its ability to appear without warning. Since the initial phases of the disorder are usually veiled, Dana-Farber researchers have engineered a strain of mice that develop aggressive, fatal pancreatic cancer through the same genetic miscues that cause the disease in humans. Knowing the mice are likely to develop the cancer should help researchers find biomarkers in the blood or elsewhere that provide the first whispers of a tumor's presence.
"This model shows great promise as a platform for rapid and efficient testing of novel therapeutic agents and for the discovery of markers for each stage of tumor growth," says Dana-Farber's Ronald DePinho, MD, who has led the research with Nabeel Bardeesy, PhD, and graduate student Andrew Aguirre.
By comparing thousands of proteins in the blood of pancreatic cancer patients with proteins in the blood of mice with the disease, investigators zeroed in on a handful that could help identify patients more than a year before symptoms prompted a visit to their doctor. These proteins are now undergoing stringent clinical testing for potential use in people at high risk for the disease. Cross-species comparisons are now being used for many types of cancer.
While much research involves biomarkers for diagnosing and classifying tumors, other work focuses on predictive markers, which reveal whether a particular therapy is working.
In many cases, the discovery of predictive markers is an outgrowth of research into the inner circuitry of cells. Consider a recent study by Dana-Farber's Sam Sidi, PhD, and Thomas Look, MD. By working with zebrafish and, later, human cells, Sidi, Look, and their associates identified several proteins in a "pathway" that signals potentially cancerous cells to self-destruct for the good of the body. As investigators around the world begin testing drugs that target this particular pathway – known as "Chk1-suppressed apoptosis" – they can check whether one of the newly identified proteins, caspase-2, has been cleaved, which is a sign of a drug's effectiveness in certain patients.
In today's era of targeted therapies, it's critical to have a way of quickly determining whether a drug is hitting the desired gene or protein and causing tumor cells to die.
Those enrolled in studies of Chk1-targeting drugs could have a small section of their tumors removed, or biopsied, shortly after beginning therapy and analyzed for caspase-2 cleavage, which would indicate whether cancer cells are self-destructing. Doctors would know, long before an MRI or X-ray could show evidence of tumor remission, whether the drug is working, or if another treatment needs to be tried.
"In this new era of targeted therapies, it's critical to have a way of determining, quickly, whether a drug is hitting the desired gene or protein and causing tumor cells to die," says Sidi. "For the Chk1 inhibitors now coming on line, the proteins we've identified can serve precisely that function."
The high-tech nature of Dana-Farber's biomarker research is reflected in the fact that much of it is conducted in centralized, "core" facilities that serve a range of scientists. Two such facilities have a particularly large role in this area.
One is the Center for Molecular Oncologic Pathology at Dana-Farber/Brigham and Women's Cancer Center. The center houses large silver tanks for storing tissue samples at low temperatures, as well as equipment for extracting genetic material and proteins from tissues and body fluids. Technicians analyze tissue to obtain its genetic signature and explore how it is altered by experimental drugs.
To determine if discoveries made in animal tissue are applicable to human tumors, pathologists identify, bank, and diagnose these tumors, as well as calculate the relative abundance of tumor to normal tissue and the percentage of tumor cells with a particular mutation. Specialists grow human tumors in mice to learn how the growths respond to specific treatments.
"We act as a hub where researchers can interact, collaborate, and use novel technologies to translate basic science discoveries into patient-care applications through the use of human tumor specimens," says Massimo Loda, MD, director of the center. Established in 2007, the center has already been involved in more than 30 biomarker-related studies.
A second resource is the Oncomap program in Dana-Farber's Center for Cancer Genome Discovery, which scans cancer tissue for gene mutations that might be targeted by drugs. The program originated when Dana-Farber researchers showed that technology designed to uncover inherited mutations in groups of people at risk for cancer could be adapted to search for specific mutations in tumor cells.
"The presence of a particular mutation or group of mutations can be predictive of a tumor's response or lack of response to targeted agents," says Dana-Farber's Levi Garraway, MD, PhD, who directs the program. The Oncomap initiative can help researchers learn how prevalent a certain pattern of mutations is in a particular type of cancer, or enable them to discover which of nearly 1,000 known mutations are in a set of tissue samples.
Though still used solely for research purposes, some version of this technology is likely to one day be put to therapeutic use, helping doctors classify and treat individual tumors, Garraway says. In this way, equipment that paves the way for research to move from the lab to the clinic may itself make that same journey.
Spring/Summer 2009 Table of Contents
Dana-Farber Cancer Institute, 450 Brookline Avenue, Boston, MA 02215 | Call us toll-free: