Oral Presentation Advances in Neuroblastoma Research Congress 2016

Background: Tumor-associated macrophages (TAM) have prognostic significance in tumor microenvironment (TME) of neuroblastomas lacking MYCN amplification (MYCN-NA). However, little is known about the regulation of immune checkpoint expression or polarization state of macrophages in neuroblastomas.

Methods: Neuroblastomas microarrays (n=160) were analyzed using CIBERSORT to estimate relative proportion of immune cells in TME. Multiplex-immunofluorescence of CD163 (macrophages) and PDL1 (immune-checkpoint) was performed on tumor tissues (n=45). Polarized human macrophages (M1, M2a, or M2c) were characterized alone and in co-cultures with neuroblastoma cell-lines. PDL1 and B7H3 were evaluated at baseline for each cell type, post 24-hour incubation with IFNɣ, and after 24-hour of co-culture.

Results: A serum-free in vitro macrophage polarization platform was used to create and characterize distinct macrophage phenotypes: M1 (CD163^dim/CD206^dim/B7H3^mod/PD1^high), M2a (CD163^dim/CD206^high/B7H3^high/PDL1^mod), and M2c (CD163^high/ CD206^low/B7H3^low/ PDL1^low). IFNɣ exposure caused upregulation of PDL1 in M2a and M2c macrophages without altering polarization markers. Neuroblastoma cell-lines, which at baseline have PDL1^low/B7H3^mod phenotype, also upregulated their PDL1 expression in response to IFNɣ exposure. In co-culture experiments, neuroblastoma cell-lines upregulated PDL1 in presence of the IFNɣ-induced M1 macrophages. Microarray CIBERSORT analyses showed high proportion of M2 macrophages, but not M0 or M1, in MYCN-NA tumors; this correlated highly with CD163 expression (r=0.59), but not PDL1 expression (r=0.2). Multiplex-immunofluorescence analyses of human tissues confirmed presence of CD163⁺ TAM, but no evidence of PDL1/CD163 co-localization was identified.

Conclusions: While PDL1 positivity was observed in subset of neuroblastomas, TAM did not express PDL1. The high CD163, low PDL1 pattern suggests TAM have M2c phenotype. M2c macrophages and neuroblastoma cells, however, can robustly upregulate PDL1 in face of anti-tumor immune response, as mimicked by IFNɣ exposure. Our results suggest that lack of PDL1 expression on tumors may not be a reliable estimate of response to anti-PD1 therapies and alternate biomarkers should be developed.