Differentiated thyroid carcinoma affects the risk of anaplastic carcinoma by serving as a direct precursor lesion. Indeed, it is currently believed that many, if not most, anaplastic carcinomas develop through the process of dedifferentiation of a well-differentiated or poorly differentiated thyroid carcinoma. All major types of welldifferentiated follicular cell-derived carcinomas can be involved. As anaplastic carcinoma is considered to be a follicular cell-derived neoplasm, cases of “anaplastic transformation” in medullary thyroid carcinoma²⁰ should not be considered as part of the same disease spectrum.

The existence of anaplastic transformation is strongly supported by the common occurrence of anaplastic carcinoma in patients with a history of previously treated well-differentiated thyroid cancer and by frequent finding of a differentiated component adjacent to the anaplastic carcinoma area on pathologic examination (Fig. 13.1). A review of series with >50 cases and adequate adequate histologic characterization demonstrates the presence of preexisting or coexisting well-differentiated carcinoma in 23% to 78% of cases.¹⁴,¹⁷,¹⁸,²⁰,²¹ Papillary carcinoma is the most common coexistent carcinoma and is found in >80% of cases showing both components.²¹ Follicular carcinoma is far less frequent in series published subsequent to the description of the follicular variant of papillary carcinoma, and both conventional type and oncocytic (Hürthle cell) follicular carcinoma can be found.²² Although possibly underdocumented, anaplastic carcinoma is least frequently associated with poorly differentiated carcinomas. Explanations for the underdocumentation may be due to the historical lack of a uniform definition for poorly differentiated carcinoma (see Chapter 12) and low impact of the finding of poorly differentiated carcinoma in those cases where the anaplastic component is detected.

Some cases of anaplastic carcinoma fail to reveal the areas of well-differentiated or poorly differentiated carcinoma. This may be due to the replacement of a less aggressive component by a more malignant tumor, or due to sampling error, or reflect a possibility that some anaplastic carcinomas develop directly, bypassing a stage of more differentiated carcinoma (Fig. 13.2).

A history of long-standing goiter is a well-documented risk factor for anaplastic carcinoma development. In a large cohort of thyroid cancer cases from the United States, ˜25% of patients had a personal history of goiter.¹³ The prevalence of preexisting goiter was even higher in some other series approaching 50%.²⁰,²³ Many patients had a history of goiter present for a very long time, such as for 20 to 60 years.²⁰ It is not entirely clear whether longstanding goiter predisposes to anaplastic carcinoma by increasing the risk of well-differentiated thyroid cancer or independently, by increasing the rate of cell proliferation, or through other and yet unknown mechanisms.

Activating point mutations of RAS are common in follicular adenomas, follicular carcinomas, and follicular variant of papillary carcinomas (see Chapters 8, 10, and 11). In anaplastic carcinomas, the prevalence of RAS mutations ranges from 10% to 60% and is ˜30% in the largest reported series.⁴³,⁴⁴,⁴⁵,⁴⁶

As in well-differentiated tumors, the mutations usually affect NRAS codon 61 and HRAS codon 61, although other hot spots (codons 12/13 of NRAS, KRAS, and HRAS) may be occasionally involved. Mutations in codons other than 12, 13, and 61 have been reported, although their functional significance is unclear.⁴⁷
The differentiated component in these tumors, when present, is typically a follicular carcinoma or follicular variant of papillary carcinoma. An identical RAS mutation is found in both tumor components, consistent with being an early event.

The main oncogenic effects of mutant RAS involve the activation of the mitogen-activated protein kinase (MAPK) and PI3K/AKT signaling cascades. Although RAS mutations occur early in tumorigenesis, their frequent presence in anaplastic carcinomas suggests that they are likely to confer the tumors with propensity to dedifferentiate. This may be due to the effect of mutant RAS on promoting chromosome instability and interfering with DNA damage response, which have been demonstrated in cultured thyroid cells.⁴⁸,⁴⁹ The increasing instability may predispose the tumor cells to acquire additional mutations, which would in turn initiate anaplastic transformation. The ability of RAS activation to induce the loss of multiple thyroid differentiation markers such as thyroglobulin, thyrotropin-stimulating hormone receptor, sodium/iodide symporter (NIS) and thyroid transcription factor 1 (TTF1) in the cultured thyroid cells⁵⁰ is likely to be an in vitro phenomenon. Indeed, the presence of RAS mutations in benign thyroid tumors argues strongly against the direct role of RAS activation on the dedifferentiation of human thyroid cells in vivo.

BRAF mutations are common in papillary carcinomas (see Chapter 11) and are found in ˜25% of anaplastic carcinomas.⁴⁵,⁵¹ Many BRAF-positive anaplastic carcinomas contain a well-differentiated papillary carcinoma component, frequently tall cell variant, whereas some tumors show no morphologic evidence of a differentiated component. Similar to papillary carcinomas, the mutations in anaplastic carcinomas are also V600E and are found in both papillary carcinoma and anaplastic carcinoma areas, indicating that they occur early in carcinogenesis.⁴⁵,⁵²,⁵³

BRAF V600E mutation leads to constitutive activation of the MAPK signaling pathway, and its expression in transgenic mice results in the formation of papillary carcinomas that undergo progression to poorly differentiated carcinomas.⁵⁴ However, the tumors in these animals did not undergo anaplastic transformation, consistent with the requirement for addition mutations to reach a fully undifferentiated phenotype.

Mutations in the effectors of this signaling pathway occur in 5% to 10% of well-differentiated follicular carcinomas (see Chapter 10). In anaplastic carcinomas, 6% to 23% of tumors demonstrate PIK3CA mutations and 6% to 16% PTEN mutations.⁴⁵,⁴⁶,⁵¹,⁵⁵,⁵⁶ Most PIK3CA mutations are located at exon 20, the kinase domain, and exon 9, the helical domain, and lead to the activation of this pathway, as demonstrated by the increased phosphorylation of the AKT protein.⁴⁵ In addition, an increase in PIK3CA gene copy numbers is found in ˜40% of anaplastic carcinomas.⁵¹,⁵⁷ In the PTEN gene, point mutations and small frameshift deletions most frequently occur in exons 5 and 7, leading to the loss of function of the PTEN protein and the subsequent activation of AKT.⁴⁶

It remains not fully understood whether PIK3CA and PTEN mutations represent an early or late event in thyroid carcinogenesis. They are present with a similar or slightly higher prevalence in anaplastic carcinomas as compared with well-differentiated papillary and follicular carcinomas. However, in contrast to BRAF and RAS mutations, which almost never coexist in the same tumor, PIK3CA and other mutations are frequently seen in anaplastic carcinomas coexisting with BRAF or RAS mutations, suggesting that they may be a late event.⁴⁵,⁵¹ It has also been suggested that gradual increase in AKT stimulation is required for progression from adenoma to well-differentiated follicular carcinoma and then to anaplastic carcinoma.⁴⁶

TP53 mutations are among the most frequent genetic alterations in anaplastic carcinoma and are typically found in 50% to 80% of tumors.⁵⁸,⁵⁹,⁶⁰,⁶¹ The mutational profile is based largely on the screening of exons 5 to 8, where most mutations occur in cancer. The distribution of mutations in thyroid anaplastic carcinomas is similar to that of most other malignancies. The mutation sites are scattered along the evolutionarily conserved domains in these exons. G:C to A:T transitions are the prevalent types of mutation, often in CpG-rich regions, although transversions and small deletions and insertions occur as well.⁶²

TP53 is a nuclear transcription factor that plays an important role in cell cycle regulation, DNA repair and apoptosis, functioning as a tumor suppressor gene. Mutations impair TP53 binding to specific DNA sequences and thus inhibit its transcriptional regulation activity. Mutant TP53 accumulates in the nucleus and manifests immunohistochemically as increased staining (see section Immunohistochemistry). TP53 is commonly altered as a late event in carcinogenesis involving various tumor types. In thyroid cells, TP53 mutations appear to be crucial for the process of tumor dedifferentiation. As such, they are far more common in anaplastic carcinomas than in poorly differentiated carcinomas and are practically absent in well-differentiated carcinomas. In tumors that contain anaplastic and well-differentiated carcinoma components, TP53 mutations are detected only in the anaplastic component.⁴⁰,⁶³ The best illustration of the multistep tumor progression is provided by a case where a well-differentiated component showed a RAS+/TP53– genotype, whereas the adjacent anaplastic carcinoma was RAS+/TP53+.⁶⁴

In mouse models, the loss of TP53 in well-differentiated tumors induces anaplastic transformation.⁶⁵ Reintroduction of wild-type TP53 in anaplastic carcinoma cell lines carrying TP53 mutation results in the reexpression of thyroid specific genes such as TPO and PAX8 and the reacquisition of the ability to respond to thyroid-stimulating hormone stimulation.⁶⁶,⁶⁷ This provides another evidence for the role of TP53 in anaplastic transformation and also points to the restoration of TP53 function as a potential therapeutic approach for these tumors.

Mutations in the gene coding for β-catenin represents another event involved in anaplastic transformation. β-Catenin is a cytoplasmic protein encoded by the CTNNB1 gene and involved in cell adhesion as well as in the wingless (Wnt) signaling pathway.⁶⁸,⁶⁹ In the absence of Wnt signaling, β-catenin is expressed mainly on the cell surface, whereas cytoplasmic protein is rapidly degraded after phosphorylation by a multiprotein complex, which includes the adenomatous polyposis coli (APC) protein. Wnt antagonizes this degradation and allows β-catenin to relocate to the nucleus to stimulate target genes. APC and β-catenin mutations can interfere with degradation, resulting in nuclear localization and gene activation promoting tumorigenesis. Point mutations in exon 3 of β-catenin affect the GSK3β phosphorylation sites required for degradation. These mutations are seen in up to 65% of anaplastic carcinomas and with lower prevalence in poorly differentiated thyroid carcinomas.⁷⁰,⁷¹ They are not found in well-differentiated thyroid carcinomas, with the exception of the cribriform-morular variant of papillary carcinoma.⁷² Unlike this rare variant of papillary carcinoma, β-catenin mutations in anaplastic carcinomas are often multiple, on average 2.4 mutations per case, indicative of the high level of genetic instability.⁷⁰,⁷¹ Mutational status correlates, although not always, with nuclear immunoexpression of β-catenin.

Other components of the Wnt signaling pathway may be altered in anaplastic carcinomas. For instance, APC mutations occur in
about 10% of cases. On the basis of limited evidence, it appears that these mutations are mutually exclusive with β-catenin mutations.⁷³ Neither EGFR amplification nor mutations at exons 18, 19, and 21 are present in anaplastic carcinomas, although there is a high frequency (50%) of polysomy for chromosome 7, where EGFR is located.⁷⁴,⁷⁵,⁷⁶ Although loss of TP53 function promotes anaplasia in mice with papillary carcinomas induced by RET/PTC1,⁶⁵ RET/PTC1 rearrangement is rarely found in human anaplastic carcinoma suggesting that it is less likely to predispose to tumor dedifferentiation than RAS and BRAF mutations.⁴⁷,⁷⁷

Profound alterations in gene expression, microRNA, and protein composition are also found in anaplastic carcinomas,⁷⁸ although the relevance of specific genetic events to pathogenesis of this tumor awaits further analyses.