There is an Erratum (March 2004) associated with this News.
Cancer trials get set for biomarkers
Aaron Bouchie
New York
The use of biomarkers in clinical trials is slowly becoming a reality that the US Food and Drug Administration (FDA; Rockville, MD, USA) has begun to recognize. But standards need to be agreed upon for what determines the validity of a biomarker before any can be used in the drug regulatory process. Even though the drug industry voices approval for what it hears from the FDA, the major questions that remain are: "who will validate the biomarkers, and when?"
Herceptin, the first marketed personalized medicine, was approved using a coordinated drug/diagnostic approval process that will become more common.
Samuel K. Collins
Personalized medicine, or pharmacogenomics, is the prescribing of drugs based on a patient's individual biological profile and relies on valid biomarkers that pinpoint what drugs a patient should receive. Biomarkers that researchers currently look for are typically genes, proteins or metabolites that are associated with certain diseases. The breast cancer drug Herceptin broke new ground in 1998 when it was approved along with a molecular diagnostic that could determine, based on a genetic biomarker, whether a person with breast cancer was part of the 30% of the population that would benefit from the drug.
But many in the healthcare industry worry about biomarkers not being properly validated. For example, a cancer drug may be given accelerated approval (after phase 2 trials) based on the lowering of certain proteins in participants, but then not increase survival time—the golden standard for primary endpoints in cancer trials. Alternatively, a drug's development may be halted because it does not affect known protein or metabolite levels, although it could actually prove efficacious if phase 3 clinical trials were conducted.
The FDA announced a system for classifying biomarkers in its November guidance "Pharmacogenomics data submissions." The agency put forth that a 'valid biomarker' is widely accepted as having physiological, toxicological, pharmacological or clinical significance. Less well defined was the 'probable biomarker,' which has not yet reached the status of a valid biomarker because it has been used by only a single company or because its significance has not yet been replicated from the scientific literature.
The introduction of such classification has raised the hopes of drug firms that plan to some day use biomarkers for designing clinical trials. If trial sponsors can accurately select people that have a greater chance of being successfully treated based on their biological profiles, then they can reduce trial costs and time to approval a great deal.
Large pharmaceutical firms typically look for pharmacogenomic data to determine drug dosing schedules and not to develop treatments for genetically defined populations (Nat. Biotechnol.21, 590, 2003). But Susan Arbuck, global head of the oncology therapeutic area at Aventis Pharmaceuticals (Bridgewater, NJ, USA), says that although incorporating molecular profiling into clinical trials will segment the market of a single indication, it has the potential to increase market penetration by easing the identification of successful patients across indications.
Arbuck also said at a conference 'Molecular targets and cancer therapeutics' on November 20 that valid biomarkers would probably have enabled the firm to perform fewer and smaller trials for Iressa, its nonsmall cell lung cancer therapeutic, than the 2,000-person phase 3 trials, which cost hundreds of millions of dollars. "We know Iressa works in some patients [specifically females and nonsmokers] better than others, but we need help understanding the pathway," says Arbuck.
Still, an easy strategy for validating biomarkers has yet to be established. After-the-fact demographic analyses are currently the most common for identifying biomarkers, but this strategy is crude and inefficient and there is often no science to back up the resulting correlation. Preclinical validation of biomarkers would be ideal, but this is often unrealistic considering the infancy of pharmacogenomics.
Although there are various biomarker identification projects taking place in academia and with the support of the FDA and the National Cancer Institute (Bethesda, MD, USA; Nat. Biotechnol.21, 718, 2003), skeptics say the only way to truly validate a biomarker is during large clinical trials. For example, Genentech's (S. San Francisco, CA, USA) Avastin showed promise in phase 2 trials for at least four types of cancer, but the drug failed to reach its primary endpoint in at least one phase 3 trial for breast cancer.
If determining the proper indication for Avastin, which has a known target in a validated pathway (see page 15), is problematic until phase 3 clinical trials, then how reliable can other cancer biomarkers be? Spiros Rombotis, CEO for Cyclacel (Dundee, UK), believes the key is tailoring techniques that measure cell death to match the drug's mechanism and molecular effects. For example, inducing apoptosis is different from directly killing cells via radiotherapies or starving a cell with angiogenesis inhibitors. One of the methods used by Cyclacel is an ELISA-based biomarker to measure caspase 3−dependent apoptosis induction in solid tumors, and this allows the firm to confirm that tumor cell death is due to apoptosis induced by its drugs.
Rombotis also says pharmacodynamic biomarker validation in phase 1 will help clinical researchers design phase 2 trials by allowing more confidence in giving appropriate doses to ensure the death of tumor cells without causing unnecessary toxicity. "Modern mechanism-based cancer drugs often fail because of toxicity in high doses, whereas lower doses might be sufficient for them to be effective in certain populations," explains Rombotis.
The FDA seems ready and willing to work with the increasing number of firms that generate pharmacogenomic data, based on its recent pharmacogenomics guidance and decisions concerning molecular diagnostics (Nat. Biotechnol.21, 1423, 2003). As Richard Pazdur, director of oncology drug products at the FDA, says: "It's not about drug approval; it's about drug development."
