Analysis of cumulative probabilities shows that the efficacy of 5HT3 antagonist prophylaxis is not maintained.

Abstract

Several investigators have reported that the efficacy of 5HT3 receptor antagonists is maintained over repeated cycles of chemotherapy. These investigators presented conditional probabilities of protection. Because conditional analyses by definition only include patients with protection in previous cycles, the results are flattered. We applied a novel statistical approach to investigate whether the efficacy of the 5HT3 receptor antagonist ICS 205-930 (tropisetron) is maintained over repeated cycles of weekly high-dose cisplatin. Overall protection was determined based on cumulative probabilities with the Kaplan-Meier method. Complete protection was calculated with a three state model for transitional probabilities. Eighty-three patients were studied. Over six consecutive cycles, protection against both acute and delayed emesis decreased significantly. The initial complete and overall protection rates against acute emesis of 71% and 95%, respectively, decreased to 43% and 72% in the sixth cycle of chemotherapy. Similarly, the protection rates of 31% and 68% against delayed emesis decreased to 6% and 40%, respectively. We conclude that overall and complete long-term protection is more accurately measured by cumulative probabilities than with a method that is based on conditional probabilities. Our statistical approach shows that the efficacy of 5HT3 antagonists is not maintained.