America's clinical trial pipeline is drying up, with fewer safe and effective new cancer drugs reaching the market. Some experts contend that FDA's control of the drug approval process hinders the development of new products.

Current research, however, suggests that a "publish or perish" ethos—common among some junior researchers—is at least part of the problem, thus producing an overabundance of dead-end phase II trials that divert resources from more promising new compounds that could go from phase III to FDA approval.

A related question is whether perverse incentives encourage researchers to label their results "positive" without doing the rigorous comparative studies, resulting in trials that produce "encouraging" results, but ultimately go nowhere.

A research team, headed by Ian Tannock, MD, professor of medicine, Princess Margaret Hospital and the University of Toronto, Canada, sought to determine how many promising phase II trials actually lead to phase III testing. They reviewed 200 promising phase II trials presented at ASCO meetings in 1995-1996, and 2006, randomly selecting 20 abstracts with encouraging results for each of five cancer sites (breast, lung, GI, GU, and GYN) for each time period.

"For the trials presented in the 1995-1996 period, we searched for subsequent randomized trials in which one treatment arm was similar to that in the phase II study," Dr. Tannock said. For the studies presented in 2006, the researchers sent a questionnaire to the authors asking whether they recommended testing the drug in phase III, whether a phase III was planned, and whether they had the resources (budget, patients, and drugs) to conduct a randomized controlled trial.

Results question motives

Looking at the data from 1995-1996, the researchers found that 10 years after presenting phase II trials with promising results, only 13 of 100 regimens were evaluated in phase III testing.

Of the 100 investigators who presented a phase II trial in 2006, 42 returned the questionnaire. Of these, 36 confirmed that the results met the efficacy criteria, and 25 felt the regimen should be evaluated in phase III testing. Nevertheless the survey showed that only 10 of the investigators planned to undertake phase III trials, with just 8 stating that they had the necessary resources. The reasons for not initiating a randomized controlled trial included insufficient efficacy, lack of financial support, and limited access to subjects.