Gemcitabine and nab-paclitaxel with or without the VDR agonist paricalcitol for metastatic pancreatic cancer: A randomized, multi-arm, phase I trial

Vitamin D receptor (VDR) agonists promote quiescence of cancer-associated fibroblasts and improve efficacy of chemotherapy in preclinical models of pancreatic cancer. We conducted a phase I trial (NCT03520790) with primary endpoint of safety when the VDR agonist paricalcitol is given with first-line gemcitabine and albumin-bound paclitaxel (GA) in patients with metastatic pancreatic cancer. Secondary endpoints included pharmacodynamic analyses. Thirty-six patients were randomized to GA plus placebo, GA plus intravenous paricalcitol or GA plus oral paricalcitol with pre-treatment and on-treatment tumor biopsies. Paricalcitol was safely administered with GA, although 5 (42%) patients receiving oral paricalcitol had grade 2-4 hypercalcemia and required dose reduction. Nuclear VDR protein expression was heterogeneous across patients, and VDR was expressed in tumor, immune, and stromal cells. Compared to pre-treatment specimens, on-treatment biopsies had decreased proportion of aSMA+ fibroblasts, altered fibroblast VDR activation signature, and greater density and spatial colocalization of CD8+ T cells with tumor cells in the GA plus paricalcitol arms. VDR expression was predictive of tumor response in the GA plus paricalcitol arms. Paricalcitol can be safely administered with chemotherapy to patients with metastatic pancreatic cancer, and on-treatment biopsies indicated favorable modulation of the tumor microenvironment by paricalcitol as predicted by preclinical models. Overall design: A total of 50 tissue slides (31 pre-treatment, 19 on-treatment) were profiled using the GeoMx Digital Spatial Profiler (DSP) and multiple regions of interest (ROI) were selected within areas containing tumor after hybridization with the Human Whole Transcriptome Atlas (WTA) probe mix. Next, the ROIs were further segmented into areas of illumination (AOIs) using a customized contouring approach, in which cytokeratin is used as a reference to define three non-overlapping areas for ultraviolet light-directed barcode collection: center (tumor epithelial region), ring 1 (stromal region within 1-25 um of the tumor epithelial region), and ring 2 (stromal region within 26-50 um of the tumor epithelial region). The barcodes were then deposited into 96-well plates, subjected to library preparation, and sequencing using S1 flow cells on an Illumina NovaSeq instrument. Samples then underwent demultiplexing. Lastly, raw read counts were analyzed to generate AOI-level expression data. The data was then quality-controlled and third-quartile normalized, resulting in a dataset including counts from 18,412 probes from 676 areas of illumination.