FDA to Consider Lower Standards for Amgen and ImClone in Cancer Studies

In its review of applications for ImClone’s Erbitux and Amgen’s Vectibix, the FDA has released a statement simultaneously frowning upon retrospective analyses of drug studies and suggesting a path for their use in new indication approvals. The statement says in part:

There are multiple examples of “retrospective” or post-hoc genomic biomarker assessment or clinical disease characteristics. In the worst examples, this involves a retrospective re-analysis of a “failed” clinical trial in which efficacy is purported to be established in a subset defined by a genomic biomarker/patient characteristics without consideration of multiplicity (i.e., data dredging), substantial missing data, and poorly characterized assays. FDA discourages such practices…

Both companies have submitted retrospective analyses of their cancer drugs. Early research has indicated that patients with the normal, so-called “wild type” of a gene known as KRAS respond far better to treatment with ImClone’s Erbitux and Amgen’s Vectibix than those with a mutated form of the gene, according to Reuters.

The retrospective studies have put the FDA in a bind, the statement says:

In discussions with these manufacturers, FDA stated that the optimal approach would be to conduct an adequate and well-controlled trial, prospectively designed to assess efficacy in subgroups based on KRAS testing by a validated assay. However, the widespread publication and presentation of the retrospective KRAS analyses have resulted in practice changes in the community, thus, a prospectively designed trial may no longer be feasible.

The FDA is therefore going to debate on Tuesday whether retrospective analyses may be acceptable as long as they are well controlled. BNET has weighed in before on the dangers of retrospectively combing failed drug trial data in hopes in finding hidden gems.

In response, the FDA says, Amgen and ImClone:

have proposed retrospective testing of KRAS status from large randomized trials that have completed accrual (Amgen) or are actively accruing patients (Imclone). The ongoing studies have been modified to enroll only patients with KRAS wild type (WT) tumors through the completion of the studies.

The development is an interesting one because it suggests that if a company floods the market with enough off-label study data, then a prospective study becomes virtually impossible, and the FDA will then compromise its standards for a new approval.

However, the compromise may be worth it because use of the drugs could be targeted only at patients most likely to benefit, and not wasted on those who won’t respond. The ironic result: That the FDA could lower its standards a little if it thinks the end result makes it more likely that sales of the drugs will be reduced because they are being administered in more effective ways.

Amgen says its data is due in 2009. ImClone says:

There may be limitations of such retrospective analyses in terms of (i) the studies not having been originally designed to test this hypothesis, (ii) analyses were conducted on a subset of the study population, and (iii) though performed in a blinded manner, the data were analyzed subsequent to the analysis of efficacy results. Nevertheless, the consistency of results across studies strongly suggests that K-ras status is a predictive biomarker.