Background Bladder cancer is the fifth most common cancer in the US, and the ninth most common cancer worldwide. Treatment of bladder cancer has evolved over time to encompass traditional modalities of chemotherapy and surgery, but has been particularly impacted by the recent use of immunotherapy. Modern immunotherapy has focused on checkpoint protein inhibitors that impede immune function. The inhibitors for several checkpoint targets (programmed death-ligand 1 [PD-L1], programmed cell death protein1 [PD-1], and cytotoxic T-lymphocyte-associated protein 4 [CTLA4]) were either approved or in late-stage development. In this study we examined the effect of PD-1 inhibitor pembrolizumab and cisplatin in a panel of bladder patient-derived xenografts (PDX) with distinct patterns of PD-L1 expression in CD34+ stem cell humanized NCG (HuCD34NCG) mice.

Methods Three bladder PDX models PNX0428, PNX0434 and PNX1028 have been established under informed consent from the patients at the Fox Chase Cancer Center, Philadelphia. These models have been profiled for the levels of PD-L1 protein using immunohistochemical staining with SP263 antibody (Ventana) and used to establish the growth kinetics and sensitivity to the PD-1 check point inhibitor pembrolizumab and standard of care chemotherapeutic agent cisplatin in female HuCD34NCG and standard NCG mice from Charles River Laboratories.

Results We have established the ability of three bladder PDX models to grow in both the HuCD34NCG and standard NCG mice. The tumor growth kinetics of these models was slightly delayed in HuCD34NCG animals compared to NCG. We observed variable responses to cisplatin and pembrolizumab treatments among the PDX models that did not correlate with the level of PD-L1 expression in these tumors. Despite the presence of ~70% PD-L1 positive cells in the PNX0428 model, these tumors produced minor responses to pembrolizumab in HuCD34NCG mice that correspond to progressive disease in patients. Interestingly, pembrolizumab treatment in the PNX1028 model and even more significantly in the PNX0434 model in HuCD34NCG mice produced strong statistically significant tumor growth inhibition that correlates with stable disease in patients despite negative staining for PD-L1 protein in these tumors. The standard of care treatment cisplatin produced significant tumor growth inhibition in all three PDX models in both HuCD34NCG and standard NCG mice.

Conclusions Our data indicates that abundant expression of PD-L1 protein in tumors should not be used as the only biomarker for patient stratification for the treatment with PD-1/PD-L1 check point inhibitors. The HuCD34NCG mouse model is an effective tool for supporting tumor growth and evaluating immunotherapies.

Ethics Approval Animal studies were approved by Nexus Pharma, IACUC number 08-22.Three bladder PDX models PNX0428, PNX0434 and PNX1028 have been established under informed consent from the patients at the Fox Chase Cancer Center, Philadelphia, IRB protocol 11-866.