The surgical specimen the index case revealed three distinct components: pure ATC, pure PTC, and a transitional area where PTC and ATC coexisted (Fig. 1). In the ATC regions, tumor cells displayed epithelioid, squamoid, and spindle cell morphologies. Inflammatory cells were interspersed with tumor cells in varying amounts. The PTC component exhibited a classic subtype. PTC metastasis was identified in a central lymph node. The tumor size measured 5.5 × 4.5 cm, and the American Joint Committee on Cancer stage was determined to be pT4aN1aM1, with evidence of lung metastasis. Immunohistochemical analysis revealed that the ATC cells were negative for thyroglobulin, focally positive for TTF- 1 and PAX8, and positive for BRAF VE1. The Ki- 67 labeling index was 55%.

Next-generation sequencing molecular testing revealed the presence of BRAF p.V600E and TERT promoter C228T mutations, as well as additional mutations in PIK3CA, RAD51D, and CDKN2A, along with EIF1AX amplification. Following surgery, the patient received chemotherapy and external radiation therapy. However, despite these treatments, the patient’s lung metastasis worsened 14 months post-surgery, and she ultimately passed away due to pneumonia related to the progression of her lung metastasis.

Samples of the index case histologically consisted of normal tissue, classic PTC, and five histologic patterns of ATC (Fig. 1). To better understand the spatial and molecular relationships, PTC samples were categorized based on their location: PTC adjacent to normal tissue, which was histologically distant from ATC and PTC adjacent to ATC. Gene expression profiles were analyzed across nine distinct groups: normal thyroid tissue (Fig. 1A), PTC adjacent to normal tissue (Fig. 1B), metastatic PTC in a lymph node (Fig. 1C), PTC adjacent to ATC (Fig. 1D), ATC with an epithelioid pattern (Fig. 1E), ATC with a squamoid pattern (Fig. 1F), ATC with an inflammatory pattern (Fig. 1G), ATC with a spindle cell pattern (Fig. 1H), and ATC with a mixed pattern (Fig. 1I).

Principal Component Analysis of Gene Expression Profiles

To investigate the distribution and transcriptional relationships among the different groups, principal component analysis (PCA) was conducted on gene expression data (Fig. 2). Distinct clustering was observed in the PCA plot, particularly among the classic PTC adjacent to normal tissue, metastatic PTC, PTC adjacent to ATC, and ATC groups. Interestingly, the PTC adjacent to ATC exhibited a transcriptional profile more similar to ATC than to PTC adjacent to normal tissue, highlighting potential molecular overlap with more aggressive thyroid cancer subtypes. As expected, the ATC samples demonstrated a highly heterogeneous pattern, widely dispersed along both PC1 and PC2 axes, underscoring their divergent molecular profiles and potential correlations with unique histopathological characteristics.

Fig. 2

Principal component analysis of gene expression profiles in the index case with anaplastic thyroid carcinoma coexisting with papillary thyroid carcinoma. The first principal component (PC1) accounted for 19.5% of the total variance (Var) and captured the primary transcriptional differences among the groups, while the second principal component (PC2) explained 17.5% of the total variance, representing orthogonal variations that reflected secondary biological patterns. NT, normal thyroid tissue; CPTC, classic papillary thyroid carcinoma adjacent to normal thyroid tissue; PTC, papillary thyroid carcinoma adjacent to anaplastic thyroid carcinoma; PTC_LN, metastatic papillary thyroid carcinoma in a lymph node; ATC, anaplastic thyroid carcinoma

Global Differential Gene Expression Between PTC and ATC

Tissue group-wise differential expression analysis showed differential gene expression across all the ATC histologic patterns (squamoid, spindle, epithelioid, mixed, and inflammatory) compared to PTC. We identified 114 upregulated genes and 261 downregulated genes across all ATC histologic patterns compared to PTC by using Venn diagrams (Fig. 3A).

Fig. 3

Distribution and functional analysis of differentially expressed genes across histologic patterns of anaplastic thyroid carcinoma (ATC) compared to papillary thyroid carcinoma (PTC). A Venn diagrams show 114 upregulated transcripts (left) and 261 downregulated transcripts (right) identified across all histologic patterns of ATC, including epithelioid, squamoid, inflammatory, spindle, and mixed subtypes, in comparison to PTC. B KEGG pathway enrichment analysis highlights the top pathways associated with the significantly upregulated and downregulated genes, illustrating their potential roles in key biological and tumorigenic processes

Tissue group-wise differential expression analysis revealed significant differences in gene expression across all histologic patterns of ATC (squamoid, spindle, epithelioid, mixed, and inflammatory) compared to PTC adjacent to the normal tissue. We identified 114 genes that were upregulated and 261 genes that were downregulated in ATCs (Fig. 3A).

Comprehensive Functional Enrichment Analysis of Upregulated and Downregulated Transcriptomics

To explore the biological significance of DEGs, GO and KEGG pathway enrichment analyses were performed using ShinyGO v0.741. The top KEGG pathways enriched for upregulated DEGs in ATCs included protein digestion and absorption, TNF signaling pathway, phagosome, and extracellular matrix (ECM)-receptor interaction, while downregulated DEGs were enriched in metabolic pathways, sphingolipid metabolism, thyroid hormone synthesis, and tight junction (Fig. 3B).

Further analysis of KEGG pathways across ATC histologic patterns (squamoid, spindle, epithelioid, mixed, and inflammatory patterns) revealed histology-specific enrichment patterns (Fig. 4). Spindle, epithelioid, and mixed ATCs showed upregulation of pathways related to tumorigenesis, including microRNAs in cancer, pathways in cancer, focal adhesion, PI3K-Akt signaling, ECM-receptor interaction, and AGE-RAGE signaling in diabetic complications. Cytokine-cytokine receptor interaction and hematopoietic cell lineage pathways were enriched in all histologic patterns except epithelioid. The inflammatory pattern was uniquely enriched in pathways such as neutrophil extracellular trap formation, NF-kappa B signaling pathway, leukocyte transendothelial migration, primary immunodeficiency, and IL- 17 signaling pathway.

Fig. 4

Heatmaps of KEGG pathway enrichment across histologic patterns of anaplastic thyroid carcinoma (ATC). Heatmaps of KEGG pathways enriched for upregulated (A) and downregulated (B) mRNAs across ATC histologic patterns. Color coding represents fold enrichment, with blue indicating negative fold enrichment and red indicating the presence of the corresponding KEGG pathway

For downregulated DEGs in ATCs, pathways such as tight junctions and thyroid hormone synthesis were consistently enriched across all histologic patterns. In ATC with the spindle pattern, oxidative phosphorylation, tricarboxylic acid (TCA) cycle, and glutathione metabolism were uniquely observed. The mixed subtype showed enrichment of ECM-receptor interaction and focal adhesion, while PI3K-Akt signaling was enriched only in squamoid ATCs.

These results suggest that upregulated and downregulated DEGs are involved in key biological functions and KEGG pathways linked to cell cycle regulation, proliferation, thyroid hormone synthesis, and cell migration in ATCs coexisting with PTC.

Selection of Potential mRNA Markers

To identify candidate mRNA markers with biological relevance to tumorigenesis, we focused on the most significantly upregulated and downregulated genes involved in tumor progression pathways, such as epithelial-mesenchymal transition, ECM remodeling, and tumor invasion. Based on their functional roles and relevance, we selected a total of eight genes for further analysis and validation. The expression levels of COL7A1, LAMC2, SPHK1, and SRPX2 were significantly upregulated in ATCs compared to PTCs, highlighting their potential roles in ATC progression. Conversely, the expression levels of CD24, EPHX1, GPX3, and RBM47 were markedly downregulated in ATCs compared to PTCs, suggesting a potential loss of their tumor-suppressive functions during tumor progression towards ATC.

Validation of Selected mRNA Markers in TCGA Data

To validate the expression patterns of the eight selected mRNA markers, we analyzed publicly available TCGA data. While direct comparison of mRNA expression between ATC and PTC was not possible in TCGA, we utilized the PTC dataset to assess mRNA expression levels across different histologic subtypes of PTC.

Using follicular variant PTC as the reference group, we compared the expression levels of the selected markers in classic PTC and tall cell PTC, a subtype known for its aggressive behavior (Fig. 5). All four upregulated markers (COL7A1, LAMC2, SPHK1, and SRPX2) showed significantly higher expression levels in classic PTCs and tall cell PTCs compared to follicular variant PTCs. Similarly, the four downregulated markers (CD24, EPHX1, GPX3, and RBM47) were significantly reduced in classic PTCs and tall cell PTCs relative to follicular variant PTCs. For most mRNA markers, the expression levels in classic PTCs were intermediate, falling between those observed in follicular variant PTC and tall cell PTC. These findings indicate that the expression trends observed for these markers in ATC are mirrored in the aggressive tall cell PTC subtype, further supporting their potential relevance to tumor progression and subtype-specific adverse tumor biology.

Fig. 5

Identification and validation of key mRNA markers associated with tumorigenesis in ATC using the TCGA papillary thyroid carcinoma (PTC) dataset. The top four upregulated mRNA markers (COL7A1, LAMC2, SPHK1, and SRPX2) demonstrated significantly higher expression levels in tall cell PTC, a subtype with aggressive behavior, compared to the follicular variant PTC (reference group). Conversely, the four downregulated mRNA markers (CD24, EPHX1, GPX3, and RBM47) showed significantly reduced expression in tall cell PTC relative to follicular variant PTC

Validation of Protein Expression by Immunohistochemistry

To validate the protein expression of the eight selected markers, we performed immunohistochemistry using TMAs constructed from a cohort of thyroid tumors, including follicular adenoma, PTC, poorly differentiated thyroid carcinoma, medullary thyroid carcinoma, and ATC (Figs. 6 and 7). Protein expression levels were quantified using H-scores, allowing for comparative analysis across tumor types (Figs. 8 and 9).

Fig. 6

Representative immunohistochemistry images of four upregulated protein markers (COL7A1, LAMC2, SPHK1, and SRPX2) in anaplastic thyroid carcinoma (ATC) compared to other thyroid tumors. While each tumor type may show varying expression levels ranging from low to high, the images included were specifically chosen to highlight the statistically significant differences in expression observed among the different tumor types. FA, follicular adenoma; PTC, papillary thyroid carcinoma; PDTC, poorly differentiated thyroid carcinoma; ATC, anaplastic thyroid carcinoma; MTC, medullary thyroid carcinoma

Fig. 7

Representative immunohistochemistry images of four downregulated protein markers (CD24, EPHX1, GPX3, and RBM47) in anaplastic thyroid carcinoma (ATC) compared to other thyroid tumors. While each tumor type may show varying expression levels ranging from low to high, the images included were specifically chosen to highlight the statistically significant differences in expression observed among the different tumor types. FA, follicular adenoma; PTC, papillary thyroid carcinoma; PDTC, poorly differentiated thyroid carcinoma; ATC, anaplastic thyroid carcinoma; MTC, medullary thyroid carcinoma

Fig. 8

Comparison of H-scores for the four upregulated markers in anaplastic thyroid carcinoma across thyroid tumor types using immunohistochemistry. H-scores representing protein expression levels of the four upregulated protein markers (COL7A1, LAMC2, SPHK1, and SRPX2) are compared across follicular adenoma (FA), papillary thyroid carcinoma (PTC), poorly differentiated thyroid carcinoma (PDTC), medullary thyroid carcinoma (MTC), and anaplastic thyroid carcinoma (ATC)

Fig. 9

Comparison of H-scores for the four downregulated markers in anaplastic thyroid carcinoma across thyroid tumor types using immunohistochemistry. H-scores representing protein expression levels of the four downregulated protein markers (CD24, EPHX1, GPX3, and RBM47) are compared across follicular adenoma (FA), papillary thyroid carcinoma (PTC), poorly differentiated thyroid carcinoma (PDTC), medullary thyroid carcinoma (MTC), and anaplastic thyroid carcinoma (ATC)

Among the upregulated markers (Fig. 8), COL7A1 and SPHK1 showed the highest protein expression levels in ATC compared to all other thyroid tumor types. LAMC2 expression was significantly higher in ATC compared to follicular adenoma and PTC, but there was no significant difference when compared to poorly differentiated thyroid carcinoma and medullary thyroid carcinoma. SRPX2 expression was significantly higher in ATC compared to follicular adenoma, PTC, and medullary thyroid carcinoma; however, its expression levels were generally elevated across all types of thyroid tumors.

For the downregulated markers (Fig. 9), CD24 exhibited the lowest protein expression levels in ATC, significantly lower than in other thyroid tumors. The expression of EPHX1 and RBM47 was significantly lower in ATC compared to follicular adenoma, PTC, and poorly differentiated thyroid carcinoma, while no significant difference was observed between ATC and medullary thyroid carcinoma. GPX3 expression was significantly lower in ATC compared to follicular adenoma, PTC, and medullary thyroid carcinoma, while no significant difference was observed between ATC and poorly differentiated thyroid carcinoma. Therefore, all mRNA-selected markers were successfully validated at the protein level through immunohistochemistry.

Correlation of ATC Biomarkers with BRAF p.V600E Mutation Status

To investigate the relationship between BRAF p.V600E mutation status and the expression of the eight identified biomarkers, we analyzed 40 ATC samples with available BRAF p.V600E mutation data. Among these, the BRAF p.V600E mutation was detected in 14 cases (35%). Expression levels of all eight markers—both at the mRNA and protein levels—showed no statistically significant differences between ATC cases with and without the BRAF p.V600E mutation.

These findings suggest that the expression of COL7A1, LAMC2, SPHK1, SRPX2, CD24, RBM47, EPHX1, and GPX3 is independent of BRAF p.V600E mutation status. This independence underscores the potential utility of these biomarkers as diagnostic predictors of ATC, regardless of the molecular heterogeneity associated with BRAF p.V600E mutations.