(P 233) IMMUNE CLASSES OF LEIOMYOSARCOMA DRIVEN BY B CELL INFILTRATE AND CTLA4 EXPRESSION

Objective: Leiomyosarcomas (LMS) are aggressive malignancies of smooth-muscle origin. Few targeted molecular options exist, and immunotherapy (IO) agents against T cells have been disappointing. In this study, we examine the immune infiltrate of LMS tumors by immunohistochemistry (IHC) and gene expression. Our objective was to identify actionable targets for immunotherapy in patients with LMS.

Methods: Tumor samples from a series of LMS patients were prospectively collected and banked for DNA and RNA sequencing as part of the MD Anderson Patient MOSAIC project. Associated clinical data for the patients were collected and maintained. IHC was performed on available separate tissue microarrays. Differential gene expression (DGE) analysis was performed using the DESeq2 package for R and the Benjamini-Hochberg procedure used to control for false discovery rate. Immune cell fractions were estimated using the MCPcounter package. Gene set enrichment analysis (GSEA) was performed using clusterprofiler and enrichplot packages.

Results: A total of 72 samples from 53 patients with soft tissue or uterine LMS in the MOSAIC database were included for analysis. Fifty-one samples (70%) were soft tissue LMS and 21 (30%) were uterine LMS. Twenty-four (45%) patients had primary tumors, 23 (43%) had metastatic disease and 6 had local recurrence (11%). Three distinct immune classes were seen based on RNA signature using MCPcounter: a predominantly B-cell group, a monocyte/macrophage group, and a low-infiltrate mixed group (Figure 1a). Analysis by IHC using a tissue microarray demonstrated 3 immune classes similar to the RNA-based subtypes: the presence of a cluster of samples with B-cells and higher expression of immune checkpoints, a CTLA4-driven / immune suppressive cluster, and an immunologically cold cluster (Figure 1b). DGE performed between the subgroups revealed a unique gene expression pattern for tumors in the B cell and monocyte/macrophage subgroups. Interrogation of specific immune markers (CD3, CD8, CTLA4, CD20, CD68 and CD163) revealed low levels of gene expression in the low-infiltrate mixed group, suggesting this may be the most immunologically “cold” subset (Figure 2a). The B cell and monocyte immune classes comprised 42% and 26% of primary tumors, respectively, higher than in the metastatic (27% and 12%) and recurrent (8% and 25%) cohorts (Figure 2b). GSEA of the immune classes revealed enrichment of immune-associated pathways in the B cell group, enrichment of phagosomes/lysosomes and metabolic pathways in the monocyte group, and DNA damage and cell cycle pathways in the low-infiltrate/immune cold group (Figure 3). MCPcounter-based immune infiltrate of LMS was then compared to undifferentiated pleomorphic sarcoma MOSAIC samples, a histologic subtype which is classically associated with dense immune infiltrate. This analysis demonstrated preservation of immune classes in the LMS samples (Figure 4).

Conclusion: Analyses of LMS samples across two large independent datasets demonstrate the existence of 3 distinct immune classes of tumors that persist despite variation in clinical and treatment conditions: B cell, monocytic/macrophage, and low-infiltrate/immune cold group. The B cell and monocytic/macrophage groups have increased expression of immune markers and pathways which may be amenable to targeted therapies.