Formal discussant of this trial Ian Tannock, MD, PhD, DSc, of Princess Margaret Cancer Centre and University of Toronto, Canada, took issue with the design of RTOG 0521. He questioned the use of one-sided P values instead of conventional two-sided P values, noting that overall survival would have been P = .08 and not statistically significant if a two-sided P value had been used. He also noted that a hazard ratio of 0.49 is an unlikely and unattainable target, since no phase III trial has shown greater than a 50% reduction in all-cause annual death rate.

“An effect of docetaxel in reducing deaths from other causes or second primary cancers seems likely to be a statistical artifact,” stated Dr. Tannock.

“Based on the one-sided P value, I question whether there is sufficient evidence to recommend docetaxel and androgen-deprivation therapy routinely to men with M0 [nonmetastatic] disease. If there is no effect on survival, chemotherapy delayed is toxicity delayed and is the preferred strategy,” he said. “This might change with longer follow-up,” conceded Dr. Tannock.

In contrast, he recommended docetaxel for men with high-risk metastatic disease at or soon after diagnosis, based on evidence from CHAARTED and STAMPEDE.

Causes of Death

Dr. Tannock called the causes of death “strange” and said that seven more deaths due to prostate cancer only in the control arm were counterbalanced by six deaths due to toxicity or unknown causes in the docetaxel arm.

Dr. Sandler said he would call the causes of death “perplexing” rather than strange. He pointed to the difficulty in accurately obtaining causes of death in a large trial conducted at many centers around the world. “Central review did the best they could, but there was probably some uncertainty in assigning cause of death,” Dr. Sandler commented. ■

Disclosure: Dr. Tannock reported no potential conflicts of interest.