ODAC: More data needed before diagnostic tests affect prescribing restrictions

ODAC panel addressed the influence of diagnostics in the context of colon cancer therapies.

The Oncologic Drugs Advisory Committee met Tuesday to discuss the role of predictive or prognostic diagnostics in providing personalized treatment to patients with cancer.

The discussion began with presentations involving the role of KRAS gene status in the decision to treat patients with colon cancer with cetuximab (Erbitux, Imclone) and panitumumab (Vectibix, Amgen). Researchers from Amgen and Imclone presented data demonstrating that people with the KRAS mutation are not likely to benefit from those two therapies. Among other findings, the data included results presented at the ASCO-NCI-EORTC Annual Meeting on Molecular Markers in Cancer that showed that 13.7% of patients with wild-type KRAS genes responded to treatment with panitumumab, while none of the patients with the KRAS mutation responded to the drug.

Richard Simon, Dsc, chief of the biometric research branch at the NCI and Gary Lyman, MD, MPH, director of health services and outcomes research program of oncology at Duke University Medical Center, were present at the hearing. They agreed that the findings were convincing. “The data were not perfect, but there was a fair amount of consistency, which is what we want to see,” Lyman said.

There was some discussion about whether restriction labeling for the two drugs should change based on these data; however, the panel did not plan to vote on labeling at this time. Rather, the aim of the meeting was to address the development of diagnostics in relation to therapeutics.

Incorporating targeted therapies

“The FDA wanted to get input on how to handle what will likely be a large number of requests for approval of diagnostic, predictive or prognostic tests for patients who are likely to benefit from more specific treatment,” Lyman told HemOnc Today. He said that diagnostics and therapeutics are evolving in parallel, but that their development is not coordinated. “The idea would be to do prospective trials with foreknowledge of who fits into what category based on the assay.” He said that this could prove not only that the therapy works, but that the prognostic or predictive test works, as well.

The panel agreed that conducting prospective trials in every instance would be difficult. One solution might be to conduct high-quality analyses of predictive diagnostics using archived samples from previous clinical trials, according to Simon. To do that there must be a substantial number of archived specimens. “In such cases, if the analysis is really focused on some biomarker that has strong scientific credentials, and if there is an analytically validated assay, then those so-called ‘prospective/retrospective’ analyses can be reliable.” In these cases it might be helpful to require two analyses rather than one, he said.

Lyman suggested that companies conducting a trial to gain approval for a drug be required to collect samples. “We have no idea what is going to come along in terms of an assay or diagnostic. It may have a direct impact on the utilization of these new drugs. In the absence of a coordinated effort to develop assays and therapies simultaneously, it should be strongly encouraged, if not mandatory, for companies to obtain samples in all applicable clinical trials,” Lyman said.

In spite of these suggestions, both researchers agreed that the time it takes to achieve approval may be inhibiting the treatment of patients. “I have some concern that the regulatory environment seems overly restrictive,” Simon said. – by Rob Volansky

This discussion needs to be put into clinical perspective. It is difficult to know when a series of retrospective analyses measures up to data convincing enough to make a change in the treatment paradigm. Ideally, the evidence is black and white. If a test is only 80% predictive, 20% of the patients may not be getting appropriate treatment decisions.

The data regarding the predictive value of the KRAS mutation for EGF-R antibody non-response appears to be black and white. The clinical field has already made that judgment. I suspect that this will lead the FDA to eventually change the label. It is understandable, though, that they are struggling with deciding how much are enough data, but we must remember that these are toxic drugs with very unpleasant side effects. Exposing patients to these drugs if it is certain they will not benefit is unacceptable. Certainly, the companies who market this class of drugs are in favor of the change, so it is hard for me to imagine what other evidence is needed.

– Alan Venook, MD

HemOnc Today Editorial Board Member