Translation failure and medical reversal: Two sides to the same coin.

Translation failure occurs when the results of preclinical, observational and/or early phase studies fail to predict the results of well done (i.e. appropriately controlled, adequately powered, and properly conducted) phase III or randomised clinical trials. Some failures occur when promising basic science findings fail to replicate in human studies, while others happen when promising uncontrolled trial data show an exaggerated effect that vanishes in the setting of a randomised trial. Medical reversals occur when the results of preclinical, observational and/or early phase studies fail to predict the results of subsequent randomized clinical trials, but the practice has already gained widespread acceptance. Oncologic examples include bevacizumab and the use of autologous stem cell transplant in metastatic breast cancer. In a well-intentioned effort to reduce the rate of translation failure, oncologists must be careful that changes to regulatory processes and clinical trial design do not actually work to increase the approval of ineffective compounds. By trying to cure translation failure, we should be careful to avoid medical reversal. The rise of surrogate end-points and role of hard-wired bias in oncology trials suggest that we may be currently ignoring the simple fact that translation failure and medical reversal are two sides to the same coin.