Biomarker boom slowed by validation concerns.

By many accounts, the brave new worlds of genomics and proteomics— and their accompanying technologies— have ushered in a new era of cancer biomarker research. Researchers in cancer detection, diagnosis, and treatment all have a vested interest in discovering these proteins and small molecules that are telltale signs of early cancer, characteristics that make a type of cancer unique, or markers that predict whether a patient will respond to a given treatment. As a result, the number of uses for cancer biomarkers has expanded at a rapid pace. “Cancer biomarkers are practically pouring out of research laboratories,” said Bruce Zetter, Ph.D., Charles Nowiszewski Professor of Cancer Biology at Boston’s Children’s Hospital. But as researchers and oncologists use and develop technology to uncover a wealth of markers that may ultimately affect cancer detection and diagnostics, many argue that, at least for now, the biomarker research engine is stuck in neutral. What are missing in many cases, they say, are studies to prove a marker’s reliability and clinical worth. “It’s very rare that a true clinical marker comes along,” said breast cancer specialist Lajos Pusztai, M.D., D.Phil., at the University of Texas M. D. Anderson Cancer Center in Houston. “While a handful of markers appear every year, you’d be hard pressed to find a study that proves its clinical utility.” In the last decade, he said, only one biomarker—HER2 for some cases of breast cancer—has affected treatment. “There are correlations with cancer, but so what? The clinician wants to know, ‘Should I use this [biomarker] or not?’” Pusztai said. For many biomarkers, the answer to that question is no. Many of the tumor marker studies published are poorly done. Pusztai pointed out in a recent editorial that, all too often, biomarker studies are “ancillary” to therapeutic clinical trials and are performed retrospectively. The study “sample size is driven by the availability of tissues” from such trials. The rigorous rules of science often go out the door when it comes to cancer biomarkers, concurred Daniel Hayes, M.D., clinical director of the breast oncology program at the University of Michigan Comprehensive Cancer Center in Ann Arbor. “It is a problem to develop a biomarker with clinical use,” he said. Pusztai and Hayes, who co-chairs the American Society of Clinical Oncology’s tumor marker guideline committee, advocate a clinical development structure for cancer biomarkers that is similar to structure for the clinical development of a drug. According to Pusztai, the step-by-step development of a marker might entail a phase I study, for example, that could demonstrate a potential association between marker and outcome. Phase II could measure the association between the marker and clinical outcome, while phase III then would involve a randomized marker validation trial. “The stepwise approach and widespread application of testing biomarkers in clinical trial–like settings will help push many of these biomarkers through U.S. Food and Drug Administration approval and to clinical use,” said Arul Chinnaiyan, M.D., Ph.D., assistant professor of pathology and urology at the University of Michigan. Making validation studies simpler is one of the ideas behind the National Cancer Institute–backed Early Detection Research Network (EDRN), which got off the ground in 2000. The EDRN established a defined development sequence or “road map” outlining five phases of biomarker development (see Vol. 93, No. 14, p. 1054). EDRN has concentrated on the first three phases: marker discovery, marker validation in terms of distinguishing the presence or absence of cancer, and testing markers for their ability to find preclinical and early-stage cancer. According to Sudhir Srivastava, Ph.D., chief of NCI’s Cancer Biomarkers Research Group, EDRN has several validation studies in progress, including those that focus on microsatellite instability for bladder cancer detection, protein profiling for prostate cancer detection, and several markers for hepatocellular carcinoma (see News, Vol. 95, No. 6, p. 422). Despite these major efforts to streamline the biomarker validation process, potential biomarker discovery well outpaces validation studies. “There’s always been a sort of a bottleneck— there’s this vast research literature about biomarkers for prognosis assessment, biomarkers for treatment selection—yet it never really seems, except in rare occasions, to lead to clinical use of biomarkers,” said biostatistician Richard Simon, D.Sc., chief of the Biometric Research Branch at NCI. “The question is, ‘Why not?’” One reason, he said, is that validating cancer biomarkers can be hard. “Validation is more difficult than discovery,” said Simon, especially when it involves multicenter studies. “I think there is a real misunderstanding of what external validation [of a biomarker] is. I think the notion of ‘phased’ studies is one way of reinforcing that concept.” One stumbling block, he noted, is a lack of multidisciplinary collaboration between clinicians, biologists, and statisticians. In addition, not many scientists really have a good grasp of how to perform validation studies. “People studying biomarkers don’t understand the concept of validation,” Simon said. “They do studies, but they are not hypothesis-based studies.” As a result, he said, “Such studies tend not to get done, and that’s one of the biggest reasons biomarkers don’t get into primetime.” EDRN is doing its share to try to change some of that. The way the network is set up, scientists from its biomarker development laboratories, Biomarker Boom Slowed by Validation Concerns