By Turna Ray
The conditions under which the US Food and Drug Administration accepted retrospective analyses from two pharmaceutical companies in updating labeling for a class of colorectal cancer drugs with pharmacogenetic information may serve as a model for other sponsors attempting to personalize treatments with genomic-biomarker assessments, according to an FDA official.
When FDA relabeled two colorectal cancer drugs — Amgen's Vectibix and Bristol-Myers Squibb/ImClone's Erbitux — informing doctors and patients that those with KRAS mutations in codon 12 or 13 do not respond to EGFR inhibitors, it marked the first time the agency used retrospective clinical trial data to update the label of a drug with PGx information [see PGx Reporter 07-22-2009].
However, just because the companies performed genomic analyses on samples collected during previously completed trials doesn't mean "they went on a fishing expedition," Lawrence Lesko, director of the Office of Clinical Pharmacology at FDA's Center for Drug Evaluation and Research, said last week at a conference hosted by Ohio State University Center for Personalized Healthcare.
Lesko listed "critical evidence" submitted by the sponsors, and noted that "if one wants to utilize retrospective data … then this would be a good way to do it."
BMS and Amgen's submissions hit a number of FDA's key criteria for accepting retrospective data, including presentation of a biologically plausible hypothesis, replication of findings, reduced bias, a pre-specified statistical plan, and the availability of an analytically valid test.
According to Lesko, the agency accepted the companies' retrospective analysis because the data "plausibly" showed that EGFR-inhibiting drugs were not efficacious in patients with mutant KRAS. The companies then replicated this finding, showing "consistent differences" in objective response between wild-type and mutated KRAS in six pooled single-arm studies. Furthermore, the companies had a 90 percent ascertainment rate of tumor samples from previously conducted studies, and presented a statistical plan ahead of conducting genomic analysis for data collection and retrospective analysis.
Furthermore, around the same time FDA was reviewing this data, diagnostics developer DxS began submitting materials to the agency for pre-market approval of its KRAS mutation test kit [see PGx Reporter 03-04-2009]. The test is commercially available in Europe and for research use in the US until the FDA approves it.
"If you think about these elements, they kind of represent a framework of policy," Lesko said at the meeting, referring to the BMS/ImClone and Amgen's retrospective genomic data submissions.
'Uniform Standards'
The agency has historically discouraged sponsors from conducting retrospective genomic-biomarker assessments, and had initially expressed reservations about using such data as the evidence base for recommending KRAS testing prior to treatment with EGFR-inhibiting colorectal cancer drugs.
The FDA's decision to update the labels for Vectibix and Erbitux came more than a year after European regulators narrowed the indication for the two drugs based on retrospective data, and several months after US professional societies, such as the American Society of Clinical Oncology and the National Comprehensive Cancer Network, recommended that doctors test all metastatic colorectal cancer patients for KRAS variants prior to treatment with an anti-EGFR antibody therapy.
In an effort to understand the circumstances under which the FDA would accept retrospective data to narrow the indication of a drug for a particular subpopulation, the agency convened an advisory committee meeting last fall to discuss Amgen and BMS/ImClone's submissions for the labeling changes [see PGx Reporter 12-17-2008].
[ pagebreak ]
This protracted process to update Vectibix and Erbitux's labels, however, is part of the FDA's overall attempt at establishing uniform regulatory guidelines by which to review pharmacogenomically guided products.
"Our ways of developing consensus on evidence to support both new drug approval and re-labeling are pretty tedious. It comes about through a large series of meetings, and a large number of interactions with companies," Lesko said.
"Right now the evidence is on a drug-by-drug basis," he added. "So, developing some uniform standards around, for example, pharmacogenomics and metabolizing enzymes, and what the framework of those decisions should be, would be a big help."
Challenges and Priorities
Another priority at the agency is to develop "more unambiguous drug product labels" that can guide treatment decisions. For example, although the FDA updated the labeling for the anticoagulant warfarin with genetic-risk information two years ago, the agency plans to update labeling again with PGx-guided dosing information once several large studies are completed [see PGx Reporter 09-05-2007].
The warfarin example illustrates FDA's challenge of having to update drug labeling absent completion of critical clinical trials. "Imperfect data," as Lesko called it, was also a factor when the agency relabeled Plavix in May.
In re-labeling the popular anti-platelet drug marketed by BMS and Sanofi-Aventis, FDA informed physicians and patients that CYP2C19 poor-metabolizers will have diminished response and higher risk of heart attack when treated with the drug, but did not provide PGx-guided dosing information [see PGx Reporter 06-17-2009].
"We did omit the recommendation for optimal doses [of Plavix] for poor metabolizers [of CYP2C19] because again the data wasn't sufficient to make a recommendation," Lesko said.
Another challenge at the agency in advancing genomically guided personalized medicine — a critical component of which is the development of drug/diagnostic combination products — is that therapeutics and tests are monitored and reviewed by two separate entities within the FDA.
"We work in silos at FDA," Lesko said, referring to CDER and the Center for Devices and Radiological Health. "It's a challenging to have adequate communication between the centers."
Finally, Lesko noted that the agency is increasingly focusing on product safety in the post-marketing setting, and pharmacogenomics has a role to play in this effort. Specifically, the FDA is considering incorporating genomic data into two of its post-marketing drug safety initiatives, the Sentinel System and the Safety First database, Lesko said.
In conducting post-marketing safety assessments, the agency has said it may use the large gene-risk databases of companies or groups conducting consumer genomics studies [see PGx Reporter 11-19-2008].