Abstract 2708: Dedifferentiated thyroid cancers have unique expression profiles of stromal cell markers and TIMER predicted immune cell infiltrate

Abstract

Abstract While thyroid cancer is primarily indolent, forming well-differentiated carcinomas (DTCs), a subset of patients develop dedifferentiated tumors that are aggressive. The evolution of dedifferentiated carcinomas is complex, producing two subtypes: poorly differentiated thyroid carcinoma (PDTC) and anaplastic thyroid carcinoma (ATC). ATCs lack markers of differentiated thyroid tissue and have a median survival of 5 months. In contrast, PDTCs retain follicular cell differentiation and have 5-year disease-specific survival of 66%. It's possible that PDTCs and ATCs arise de novo; however, co-occurrences within DTCs have been observed. Consequently, it has been proposed that PDTCs are a transitional stage in the DTC dedifferentiation process, eventually leading to ATC. Dedifferentiation of many cancer types is associated with the accumulation of second hit driver mutations. However, despite large sequencing endeavors, few additional mutations between PDTCs and ATCs have been identified. It is likely that the drivers of dedifferentiation to ATC may be extrinsic to the mutational landscape. Thus, we hypothesize that the tumor microenvironment plays a central role in driving dedifferentiation of DTCs. We performed RNA sequencing on a cohort of 262 thyroid resection specimens encompassing a broad range of thyroid cancers and used DESeq2 to generate fold-change gene expression estimates for poor outcome samples (defined by transformation to PDTC or ATC, recurrence, distant metastasis, or death from disease). Functional enrichment analysis (WebGestalt) identified extracellular matrix (ECM) and regulation of leukocyte activation gene sets as enriched in poor outcome. In the ECM gene set, FAP, PDPN, TGFB1, FN1, COL11A1, LAMB3, POSTN, VCAN, and WNT2, amongst others, were significantly enriched. The neutrophil activating chemokine CXCL5 showed a striking 43-fold increase in poor outcome cases. Interestingly, segregating by histology shows that these stromal markers are upregulated in ATCs but not PDTCs, suggesting that distinct populations of fibroblasts may be present in ATCs. Similarly, the tumor immune cell prediction algorithm TIMER (Tumor Immune Estimation Resource) predicts increased macrophage and neutrophil infiltrate within ATCs relative to PDTCs. Together, these results support a model in which a heterogenous population of fibroblasts drive tumor aggression and granulocyte infiltrate in ATCs but not PDTCs. Single cell resolution will be instrumental to further clarify this model and elucidate the signals that support ATC aggression. Importantly, the distinct differences in stromal markers between PDTCs and ATCs in our cohort suggest that these may be pathologically distinct outcomes rather than existing on a spectrum of dedifferentiation with PDTCs eventually fated to become ATCs. Citation Format: Matthew A. Loberg, George Xu, Thomas P. Stricker, Quanhu Sheng, Barbara A. Murphy, Vivian L. Weiss. Dedifferentiated thyroid cancers have unique expression profiles of stromal cell markers and TIMER predicted immune cell infiltrate [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2021; 2021 Apr 10-15 and May 17-21. Philadelphia (PA): AACR; Cancer Res 2021;81(13_Suppl):Abstract nr 2708.