Abstract CN04-02: Phase I trials: Limits and limitations in proof of concept

Abstract

Abstract We could and maybe should ask the question what is meant by “proof of concept” in early drug development. Do we mean that the drug hits the molecular target it is supposed to hit, which in essence may still be clinically meaningless. Or do we mean that we can proof that the drug has the expected signal of biological anti-tumor activity in a human being? And if so, what level of evidence are we looking for? Obviously the ultimate proof that a drug is beneficial can only be obtained in appropriately designed phase III studies. But these are expensive trials, and it would be worthwhile to be able to estimate earlier on in the development if the drug might be doing what it is supposed to do according to the preclinical information obtained. Is looking for proof of concept in a phase I study stage of development like asking the impossible? Enormous amounts of effort and money have recently been invested in early clinical trials trying to better predict the outcome of later phase studies. Most of the obtained information has not been very supportive for Go/No-Go decisions. Interestingly analyzing the data for some of the recently registered agents it turns out they in retrospect predicted their activity by showing anatomical size changes as classified by RECIST, in a small or larger subset of patients in the phase I studies. For some of these agents, molecular marker changes coincided with the anatomical size changes, but in no instance was a molecular marker change without an anatomical size change predictive of relevant activity. We should bear in mind that when starting phase I studies we frequently still have limited knowledge on the pharmacodynamics of a new agent. Thus the selection of patients is often a limiting factor in trying to assess the desired activity. The more recent focused development of Hedgehoc and ALK-fusion protein inhibitors shows that with a focused patient selection, sometimes based on early observations in the phase I studies, proof of concept prediction seems to come into reach. However for the majority of agents, we still seem to be looking into a crystal ball. Tumor size changes can be considered as biomarkers and should only be replaced by something else once we have certainty that a molecular change coincides with size change. Based on these observations models can be build that might be helpful in clinical development. The fact that more simple assays than measuring anatomical size have not yet been able to provide us with appropriate predictions of efficacy, should not stop us from trying to develop these simple tools. And early hints that we may be on track are available. Citation Information: Mol Cancer Ther 2009;8(12 Suppl):CN04-02.