Back grey_arrow_rt.gif
The Cancer Biomarker Conundrum: Too Many False Discoveries
  The boom in cancer biomarker investments over the past 25 years has not translated into major clinical success. The reasons for biomarker failures include problems with study design and interpretation, as well as statistical deficiencies, according to an article published online August 12 in The Journal of the National Cancer Institute.
The National Institutes of Health defines a biomarker as "a characteristic that is objectively measured and evaluated as an indicator of normal biologic processes, pathogenic processes, or pharmacologic responses to a therapeutic intervention." In the past decade, there have been numerous biomarker discoveries, but most initially promising biomarkers have not been validated for clinical use.
To understand why so-called biomarker "breakthroughs" have not made it to the clinic, Eleftherios P. Diamandis, M.D., Ph.D., professor of pathology and laboratory medicine at Mount Sinai Hospital in Toronto and associate scientist at the Samuel Lunenfeld Research Institute of Mount Sinai Hospital, reviewed some biomarkers initially hailed as breakthroughs and their subsequent failings.
Diamandis first describes the requirements for biomarkers to be approved for clinical use: A biomarker must be released into circulation in easily detectable amounts by a small asymptomatic tumor or its micro-environment; and it should preferably be specific for the tissue of origin. Also, if the biomarker is affected by a non-cancer disease, its utility for cancer detection may be compromised. For example, the prostate-specific antigen (PSA) biomarker, which is used to detect prostate cancer, is also elevated in benign prostatic hyperplasia.
Diamandis looks at seven biomarkers that have emerged in the past 25 years, all of which were considered promising when they were first described. These include nuclear magnetic resonance of serum for cancer diagnosis; lysophosphatidic acid for ovarian cancer; four- and six-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. Problems ranged from inappropriate statistical analysis to biases in case patient and control subject selection. For example, the problems with EPCA-2 included reporting values that were beyond the detection limit of the assay and using inappropriate reagents to test EPCA-2, such as solid surfaces coated with undiluted serum.
Diamandis concludes that "problems with pre-analytical, analytical, and post-analytical study design could lead to the generation of data that could be highly misleading."
Cancer Biomarkers Missing in Action
MedPage Today
Published: August 12, 2010
Biomarkers for cancer -- often reported with great publicity -- routinely fail to be of clinical use, a Canadian researcher said.
Despite large investments and investigation by competent researchers, no new major cancer biomarkers have been approved for clinical use in the past quarter-century, according to Eleftherios Diamandis, MD, PhD, of Mount Sinai Hospital in Toronto.
In a commentary published online in the Journal of the National Cancer Institute, Diamandis said that's because a range of common analytical errors lead to what he called "false discovery," even when reports undergo detailed peer review and appear in high-profile journals.
"Press releases and news conferences that immediately accompany publication of a high-profile biomarker generate high expectations about the new biomarker," he wrote, but such a high level of publicity does not usually accompany the failure of studies attempting to validate the original findings.
Action Points
* Note that this commentary documents the hazards of reports on promising biomarkers for cancer screening.
* The author points out that one of the major problems is finding biomarkers released in significant amounts from asymptomatic (usually small) tumors, but not from normal tissues.
"Consequently," Diamandis argued, "the general public receives skewed information about the biomarker."
To be a useful biomarker for early diagnosis, Diamandis noted, a molecule has to be released in easily detectable amounts by an otherwise asymptomatic tumor. The molecule should also be highly specific for the tissue of origin and a tumor should produce levels that are markedly higher than background.
With a handful of possible exceptions, Diamandis wrote, "very few, if any, molecules have been identified that are expressed only by a cancer tissue but not by the corresponding normal tissue."
Cancer biomarkers are missing in action despite the availability of "highly sophisticated and powerful technologies" to discover them, as well as large investments, Diamandis wrote, adding that the last biomarker approved by the FDA -- in 2009 -- was HE4 protein, indicated for monitoring recurrence but not early detection of ovarian cancer.
Indeed, he argued, most cancer biomarkers in clinical use are not suitable for population screening or for early diagnosis.
"Notable exceptions," he wrote, are human choriogonadotropin for germ cell tumors and gestational trophoblastic disease and β-fetoprotein for hepatocellular and testicular carcinoma.
The lack of such markers is also in contrast with other areas of medicine, where biomarkers such as cardiac troponin for heart attack or serum creatinine for renal function are successfully used in the clinic, Diamandis wrote.
In a series of case studies, he offered examples of highly publicized putative biomarkers that failed.
For example, Diamandis wrote, the 1986 report that proton nuclear magnetic resonance of plasma samples could detect malignant tumors could not be validated, because the original data actually varied with plasma lipid composition rather than cancer.
Plasma lipid composition, in turn, was associated with age, sex, and diet, but the original authors did not control for any of those factors. In a validation study that matched cases and controls by age and sex, there was no ability to distinguish between cancer and no cancer, Diamandis wrote.
The lesson, he wrote, is that patient selection can bias results.
The analysis "makes an incredibly important point," according to Otis Brawley, MD, chief medical officer of the American Cancer Society and a long-time critic of the overhyping of screening tests.
"The biomarkers listed [in the commentary] have not been successful in screening," he told MedPage Today in an e-mail.
Even prostate-specific antigen -- perhaps the most widely used cancer screening test -- remains controversial. It may eventually "prove useful but the jury is still out," Brawley said.
"I am very worried that doctors and others have overpromised the power of screening and the public expects too much," Brawley said. That increases medical costs but also "prevents us from asking the right scientific questions that can ultimately improve health," he said.
The key issue is whether a given test reduces mortality, he added -- something that is often forgotten in the early publicity over a new test.
Another problem, he said, is overdiagnosis. "We cure some people who do not need to be cured and this makes our screening appear more successful than it is."
The commentary was supported by the U.S. National Cancer Institute and the Natural Sciences and Engineering Research Council of Canada. A spokesman for Diamandis said he had no conflicts.
Primary source: Journal of the National Cancer Institute
Source reference:
Diamandis EP "Cancer biomarkers: Can we turn recent failures into success?" J Natl Cancer Inst 2010; 102: 1-6.
  icon paper stack View Older Articles   Back to Top