Study of CBP501/Cisplatin/Nivolumab Combinations in Advanced Pancreatic Cancer

Multicenter, randomized, open-label, parallel group phase 2 study to assess the efficacy and tolerance of four combinations of CBP501, cisplatin, and nivolumab administered once every 21 days to patients with stage IV exocrine pancreatic cancer and WBC \< 10,000/mm3 at screening. Patients will be randomized 1:1:1:1 to the following four treatment groups, with randomization stratified by ECOG PS (0 vs 1) and liver metastasis (present vs absent): 1. CBP501 25 mg/m2 + cisplatin 60 mg/m2 + nivolumab 240 mg 2. CBP501 16 mg/m2 + cisplatin 60 mg/m2 + nivolumab 240 mg 3. CBP501 25 mg/m2 + cisplatin 60 mg/m2 4. Cisplatin 60 mg/m2 + nivolumab 240 mg No more than 4 cycles of combination therapy may be administered but patients who remain progression-free after 4 cycles may receive up to 6 additional cycle of single-agent nivolumab. A Fleming two-stage design will be used. For each study arm, the null hypothesis that the true percentage of patients progression-free at 3 months is 10% will be tested against a one-sided alternative. In the first stage, 9 patients will be accrued to each study arm. In the first stage, if there are 1 or fewer patient progression-free at 3 months the study will be stopped for futility and if there are 4 or more patients progression-free at 3 months the study will stopped and the null hypothesis rejected. Otherwise, 14 additional patients will be accrued to the study arm for a total of 23. The null hypothesis will be rejected if 6 or more of 23 patients are progression-free at 3 months. This design yields a type I error rate of 2.5% and power of 80% when the true percentage of patients progression-free at 3 months is 35%.