Introduction:

Resistance to immune checkpoint blockade is an important challenge for patients with bladder cancer. Approximately 25-35% of patients have objective responses in the 2^nd-line metastatic setting and 40% have pathologic complete response during neoadjuvant treatment.  Studies in other solid tumor types have identified immune suppressor cells that are associated with resistance to immune checkpoint blockade, and sometimes targeting these immune suppressor cells can increase therapeutic responses.  The purpose of his study was to determine cellular and bulk gene expression correlates of resistance to anti-PD-L1 therapy in immune competent murine bladder cancer models.

Methods:

C57BL6 female mice were subcutaneously inoculated with 1 million cells from the BBN963 and UPPL1541 cell lines.  Once tumors reached 3-5mm in size, subjects were treated three times weekly with anti-PD-L1 monoclonal antibody (BioXCell, 5mg/kg, intraperitoneal administration). In combination therapy studies, subjects were also treated with the tyrosine kinase inhibitor ibrutinib (25mg/kg/day oral).  Tumor volumes were monitored over time based on 2-dimensional caliper measurements and subjects were classified as responders (complete or partial) or non-responders (stable disease or progression) based on RECIST criteria.  At end of study, tumors were excised and flash frozen for subsequent immune gene expression profiling and/or digested into single-cell suspensions for subsequent surface antibody staining and flow cytometry.  Spleens were processed into single-cell suspensions that were also subject to surface antibody staining for lineage immune markers followed by flow cytometry. Proportions of cell populations were compared via two-tailed t-testing, and immune gene expression differences after quantile normalization were considered significant between treatment groups if at least 2-fold differences in expression were detected at a p<0.01. 

Results:

In the BBN936 (basal) model, the response rate (partial or complete) in control subjects was 0% (0/10) and among anti-PD-L1 treated subjects, 73% (16/22).  Comparing control (IgG-treated) subjects to non-responder (anti-PD-L1 treated) subjects, non-responders had a greater proportion of monocytic myeloid derived suppressor cells (M-MDSC) in the tumor microenvironment (TME) and less M-MDSC in the spleen, suggesting a shift into tumors of macrophage-like MDSC are a biomarker of resistance to anti-PD-L1 treatment. In the UPPL1541 (luminal) model, the opposite effect was observed:  non-responders had lower levels of M-MDSC in the TME.  When bulk tumor was processed into RNA and analyzed on the Nanostring Immune Panel gene expression platform, basal bladder tumors treated with anti-PD-L1 in combination with the MDSC-inhibitor ibrutinib demonstrated lower levels of macrophage markers Ptgs2 and Serpinb2 compared to control, suggesting that targeted therapy has the potential to remodel the TME in the setting of immune checkpoint blockade.

Conclusion:

In a basal model of bladder cancer, macrophage like immune suppressor cells (M-MDSC) appear to be a biomarker of resistance to immune checkpoint inhibitor (anti-PD-L1) treatment. This does not appear to be the case in a luminal bladder cancer model.  The macrophage content of non-responder tumors can be potentially decreased by combination treatment incorporating ibrutinib, which inhibits MDSC signaling.

Funding: Urology Care Foundation