Publication outcomes of phase II oncology clinical trials

Objectives: Translation of evidence-based medicine into oncology practice depends on timely and full publication of clinical trials. We investigated publication outcomes of Phase II trial abstracts from the annual meetings of the American Society of Clinical Oncology (ASCO). Methods: We searched the 1997, 1999, and 2001 ASCO annual meeting proceedings and identified all Phase II trials, excluding those that reported preliminary results. Literature search was performed using PubMed, EMBASE, and Google for corresponding publications in peer-reviewed journals. We attempted to contact authors of all unpublished trials. Results: Only 60.8% of t he 559 trials identified were published, with a median time to publication of 41 months. At 5 years, 65.9%, 62.7%, and 57.0% of studies from 1997, 1999, and 2001, respectively, were published. Studies with larger samples were associated with a shorter time to publication, as were oral and poster presentations versus print only (P < 0.001). Common reasons for not publishing were uninteresting results, lack of time, and relocation of authors. Among abstracts reporting response rates, 37.7% showed different results in subsequent publications. Though not statistically significant, over the 5-year period, abstracts presented at later years had a lower rate of publication, longer time to publication, and a higher likelihood of showing a better tumor response. Conclusions: Almost half of Phase II trials presented at ASCO annual meetings within the last 10 years remain unpublished. Over one-third of published trials reported results different from those presented in abstracts. Like Phase I and III trials, Phase II trials often are unpublished.