Formal discussant of the RADIANT-3 trial at the European Society for Medical Oncology (ESMO) Congress, Alexandria Phan, MD, of Houston Methodist Cancer Center, agreed that the crossover to open-label everolimus (Afinitor) by placebo patients probably muddied the waters in the survival estimate.
“The survival in the control arm reflects the beneficial effect of the study drug. This is a huge selection bias. Since crossover usually occurs after disease progression, the endpoint affected the most is overall survival,” Dr. Phan told listeners.
She said that the rank-preserving structural failure time analysis used in this study is the best method for adjusting for crossover bias in a randomized controlled trial. “This method assumes a common treatment effect in all patients,” she said.
“The takeaway message for overall survival with everolimus is that without adjustment, it is at least 6.3 months longer than with placebo, and with adjustment, it is at most 23.4 months longer than with placebo,” she said. ■
Disclosure: Dr. Phan reported no potential conflicts of interest.