Two Different Shades of Anaplastic Thyroid Cancer

Anaplastic thyroid cancer (ATC) remains a lethal disease and one of the most aggressive tumors overall. The recent implementation of neoadjuvant targeted therapies and of immune checkpoint blockade (ICB) treatments has extended the survival of patients with BRAF^V600E-mutant ATCs, but resistance invariably arises, and prognoses for most ATC patients remain dismal. Biomarkers that could inform optimal treatments in this fast-growing cancer are thus needed. ¹ On this issue of Thyroid, Byun, Lee, and colleagues ² conduct a transcriptomic study of ATCs and their precursors, integrating data from both their own series ³ and other published datasets, ^4,5 and propose a subclassification of ATCs based on their distinct tumor microenvironment (TME), which might ultimately help inform refined management strategies for these patients.

Here the authors primarily rely on deconvolution analyses of bulk RNAseq data from ATCs, their differentiated thyroid cancer (DTC) counterparts, including both “BRAF^V600E-like” and “RAS-like” specimens, as well as normal thyroid samples and benign adenomas. As expected, compared to DTCs, ATCs were enriched in dysfunctional immune cells and in fibroblasts associated with aggressive cancer ecotypes. The authors go further and propose a subclassification of ATCs in two entities with specific transcriptomic features and distinctive TMEs: ATC-epithelial-endothelial (ATC-E) and ATC-immune-fibroblast (ATC-IF). No differences in ATC cytomorphology, in the distribution of genetic alterations, or in the BRAF^V600E-RAS score were found between ATC-E and ATC-IF subtypes. The former suggests that, once an ATC is generated, its biology largely relies on the interactions between tumor cells and their TME, rather than on DTC-initiating genetic events (typically BRAF or RAS mutations) or on markers of thyroid cancer progression (e.g., p53 loss). However, the TME of ATC-IF was characterized by the highest levels of M2 macrophages, resting NK and CD4⁺ T-cells, and promigratory fibroblasts, as well as of immune checkpoint markers such as Programmed cell death ligand 1 (PD-L1). Compared to ATC-E, ATC-IF was enriched in markers of epithelial-mesenchymal transition and activated pathways related to cell adhesion and extracellular matrix organization (suggestive of a continuous crosstalk with the TME). ATC-E, on their part, retained certain epithelial identity and were comparatively enriched in endothelial cells. Remarkably, in the absence of any differences in treatment modalities, survival of patients with ATC-IF was even shorter than that of those with ATC-E.

In addition, this work represents a notable effort to integrate ATC profiling results from multiple groups and to harmonize observations across cohorts toward generalizable tumor classifications, highlighting the value of open and collaborative science. In this regard, tumor signatures defined in other relevant bulk RNA sequencing efforts were evaluated here. For instance, compared to ATC-E, ATC-IF correlated with higher ERK signaling ⁶ and MAP (Molecular Aggression and Prediction) scores. ⁷ Data from single-cell transcriptomics on ATC were also incorporated here: ATC-IF correlated with a mesenchymal ATC signature (defined by Lu et al. ⁸ ) and with higher expression of CREB3L1 transcription factor (from Luo et al. ⁹ ). Overall, this work represents a good compromise between the greater granularity into biology from single-cell approaches versus the likely easier applicability of more broadly defined categories used here.

Moving forward, it will be interesting to see if the ATC classifications proposed here and elsewhere are useful tools to select, monitor, and predict responses to currently used therapies to treat this fatal disease or to identify subsets of patients likely to benefit from novel approaches. Among those, patient-tailored combinatorial therapies should help improve outcomes. Preclinical efforts employing MAPK inhibitors that are effective in non-BRAF^V600E tumors, immunotherapies other than ICB, epigenetic inhibitors and/or treatments exploiting the metabolic preferences of these cancers might help reduce mortality in ATC. Of note, ATC characterization was performed here on pretreated tumors, but it would be logical to assess some of these signatures as potential predictors of response to current ATC treatments, e.g., ERK scores in those treated with dabrafenib plus trametinib and/or the outcomes of ATC-E versus ATC-IF subtypes in the context of ICB therapy. I look forward to these novel classifications furthering scientific discourse in ATC, with the potential for inclusion in future ATC management guidelines. ¹⁰