Large scale multiomic analysis suggests mechanisms of resistance to immunotherapy in leiomyosarcoma.

11512 Background: Leiomyosarcomas (LMS) have been reported to have immunohistochemical (IHC) and gene expression signatures suggestive of an immune-responsive tumor microenvironment. Despite this, immune checkpoint inhibitors have demonstrated minimal activity in LMS. We examined molecular profiles of LMS specimens from multiple institutions to explore mechanisms of immunotherapy (IO) resistance. Methods: LMS specimens (n = 1115), including 701 uterine (uLMS) and 414 soft tissue site (stLMS) samples, underwent next-generation sequencing (NGS) of DNA (592-gene panel or whole exome) and RNA (whole transcriptome, n = 537) at Caris Life Sciences (Phoenix, AZ). A threshold of 10 mut/Mb was used to identify high tumor mutational burden (TMB-H). IHC was performed for PD-L1 (SP142; 2+|5% positive). Deficient mismatch repair (dMMR)/high microsatellite instability (MSI-H) was tested by IHC and NGS, respectively. RNA expression was analyzed using Gene Set Enrichment Analysis and Microenvironment Cell Populations-counter, with results compared to melanoma (n = 1255) as a representative immunogenic tumor type. P-values were adjusted for multiple hypothesis testing. Results: TMB-H was observed in 3.8% (n = 41) of LMS specimens, with a median of 5 mut/Mb (IQR 3.3-6.7). dMMR/MSI-H was rarely detected (1.5%, n = 17), whereas 8.2% (n = 88) were positive for PD-L1 expression. uLMS and stLMS did not differ in TMB-H (3.4 vs 4.5%, p = 0.277), PD-L1 expression (8.6 vs 7.4%, p = 0.322), or dMMR/MSI-H (2.0 vs 0.7% p = 0.207). stLMS demonstrated upregulation of immune-related gene sets, including interferon γ (p = 0.035) and α (p = 0.033) response, inflammatory response (p = 0.038), interleukin-6/STAT3 signaling (p = 0.030), and TNFα signaling (p = 0.026) compared to uLMS. Immune cell infiltration was increased in stLMS over uLMS, most notably for CD8 T-cell and B-cell abundance ( > 2-fold increase, p < 0.0001). Compared to melanoma, all LMS had lower abundance of CD8 T cells, cytotoxic lymphocytes, and B-cells ( > 2-fold decrease, p < 0.0001). Fibroblasts were more prevalent in LMS relative to melanoma (3.2-fold increase, p < 0.0001). Interestingly, while higher CD8 T-cell infiltration was positively associated with dMMR/MSI-H among LMS specimens (p = 0.032), TMB-H and PD-L1 expression were associated with lower CD8 T-cell infiltration (p < 0.01). Conclusions: Only a small proportion of LMS are TMB-H or MSI-H, suggesting that the neoantigen burden in LMS may be insufficient to promote a robust anti-tumor response, even in the presence of PD-L1 positive tumor cells. Traditional predictive biomarkers of response to IO are unlikely to be useful in LMS. Furthermore, both uLMS and stLMS have an immune microenvironment characterized by a high fibroblast and low T cell abundance relative to melanoma. Future IO trials in LMS should focus on combination therapies that may reverse the observed T-cell exclusion/desmoplastic phenotype.

Duke Scholars

Author Richard Francis Riedel Medicine, Medical Oncology