Abstract
Current methodologies for developing PDX in humanized mice in preclinical trials with immune-based therapies are limited by GVHD. Here, we compared two approaches for establishing PDX tumors in humanized mice: (1) PDX are first established in immune-deficient mice; or (2) PDX are initially established in humanized mice; then established PDX are transplanted to a larger cohort of humanized mice for preclinical trials. With the first approach, there was rapid wasting of PDX-bearing humanized mice with high levels of activated T cells in the circulation and organs, indicating immune-mediated toxicity. In contrast, with the second approach, toxicity was less of an issue and long-term human melanoma tumor growth and maintenance of human chimerism was achieved. Preclinical trials from the second approach revealed that rigosertib, but not anti-PD-1, increased CD8/CD4 T cell ratios in spleen and blood and inhibited PDX tumor growth. Resistance to anti-PD-1 was associated with PDX tumors established from tumors with limited CD8+ T cell content. Our findings suggest that it is essential to carefully manage immune editing by first establishing PDX tumors in humanized mice before expanding PDX tumors into a larger cohort of humanized mice to evaluate therapy response.