Visualization of a molecular network integrated with a gene profile.
In recent years, the sequencing of the human genome and advances in supercomputer technology have helped drive hopes that a deeper understanding of the biology of cancer would yield the markers needed to truly usher in the long-awaited era of personalized medicine.
Biomarkers offer great potential for improving management of cancer at every point from screening and detection, to diagnosis, staging, prognosis, and the assessment of treatment response. Striking advances have been made in several fields, particularly breast cancer, and fresh research suggests significant steps forward in lung cancer.
Yet progress has been slow in the vast, ever-changing, and controversial frontier of biomarker research, with fewer than 2 dozen cancer biomarkers approved so far by the FDA among the thousands researchers have explored.1
Experts have advanced a brew of reasons for the gap between potential and performance, in a complex field that requires the coordination of a diverse team of pathologists, molecular biologists, and biostatisticians.
Last year, the AACR-FDA-NCI Cancer Biomarkers Collaborative described a “growing imperative” to modernize the drug development process, and made 27 recommendations in 8 different areas for doing so.2
Meanwhile, the pharmaceutical industry is exploring ways to improve biomarker development, with some industry analysts seeing regulatory issues as a significant sticking point in economically feasible biomarker development (Read More: Industry Testing New Models for Developing Biomarkers
Sudhir Srivastava, PhD, MPH, the founding chief of the Cancer Biomarkers Research Group at the National Cancer Institute, said in an interview with OncLive
that one of the challenges in biomarker research is the unrealistic expectations promoted by many study investigators.
“With almost every paper, even if there is a remote chance of success, you see a press release hyping a discovery, but later on they fail,” Srivastava said. “The public gets so excited about it that they demand we must succeed as soon as possible. But the fact of the matter is, the hype usually does not translate into clinical studies.”
At the same time, Srivastava said, the scientific community is getting organized and gaining momentum. “The infrastructure is in place to move forward in the right direction. More breakthroughs are a matter of time,” he said.
Proposal for classifying Biomarkers
Adapted from Mishra A, Mukesh V. Cancer biomarkers: are we ready for the prime time? Cancers 2010; 2:190-208; doi:10.3390/cancers2010190.
Notable Shortcomings Analyzed
Many researchers who have reviewed the progress made on biomarkers thus far have found much to criticize.
In 2010, Eleftherios P. Diamandis, MD, PhD, professor and head of the Division of Clinical Biochemistry, University of Toronto, Canada, and associate scientist at the Samuel Lunenfeld Research Institute of Mount Sinai Hospital, Toronto, reviewed cancer biomarkers initially hailed as breakthroughs and their subsequent failings.3
These included nuclear magnetic resonance of serum for cancer diagnosis; lysophosphatidic acid for ovarian cancer; 4- and 6-parameter diagnostic panels for ovarian cancer; osteopontin for ovarian cancer; early prostate cancer antigen-2 (EPCA-2) for prostate cancer detection; proteomic profiling of serum by mass spectrometry for ovarian cancer diagnosis; and peptidomic patterns for cancer diagnosis.3
Diamandis found problems ranging from inappropriate statistical analysis to biases in patient and control subject selection. Problems with EPCA- 2, for example, included reporting values that were beyond the detection limit of the assay and using inappropriate agents to test EPCA-2.
Duke University recently gained the national spotlight in one of the most widely publicized biomarker failings. A research team had devised genetic tests to assess tumor cells by looking for gene patterns that would determine which drugs would best attack a particular cancer. The tests turned out to be worthless, though they were once hailed as a breakthrough that was seen as the first fruit of the new genomics.
The Duke research was discovered to be fl awed only because it relied on publicly available data sets and algorithms. Keith A. Baggerly, PhD, and Kevin R. Coombes, PhD, statisticians at MD Anderson Cancer Center in Houston, Texas, spent 2000 hours finding all of the errors in the research and found even simple errors, such as row or column offsets.4