Diagnosis: Histopathology and Differential Diagnosis

The importance of histopathology

The diagnosis of ATC can often be suspected clinically, but the large differential diagnosis, reviewed below, necessitates tissue evaluation to exclude other treatable entities with better prognoses. The diagnosis of thyroid pathology involves correlation of clinical, biochemical, radiographic, and morphological features of the individual case. This is particularly important for ATC (13).

ATCs exhibit wide variations in appearance with several morphologic patterns recognized and many tumors manifesting a mixed morphology (14,15). A common morphologic presentation, and one that is most easily recognized as an anaplastic carcinoma of thyroid, is that of the biphasic spindle and giant cell tumor. Other tumors are dominated by bizarre malignant giant cells, and still others may show a more pure population of spindle and squamoid cells (16). All variations of anaplastic carcinoma of the thyroid are highly proliferative with numerous mitotic figures and atypical mitoses (17). There is usually extensive necrosis, and in some cases the necrosis may be so widespread that the only viable tumor is preserved around blood vessels. Inflammatory infiltrates are frequently seen with the necrosis. Osteoclast-like giant cells may be present and have been shown by immunohistochemical studies to be of the monocytic/histiocytic lineage (18,19). Neoplastic bone and cartilage may also be identified.

Histopathological subtypes

ATC has three main histological growth patterns: spindle cell pattern, pleomorphic giant cell pattern, and squamoid pattern. One of these patterns may predominate in a given tumor, or the tumor may show a mixture of two or three different patterns (14,20–22). Rare histological variants of anaplastic carcinoma include the paucicellular variant and the rhabdoid variant (23–31). The histological subtypes and variants of anaplastic carcinoma have no known prognostic significance, with the possible exception of the paucicellular variant, which in some studies was found to affect younger patients and have a more indolent course (23,25). A variant known in the past as small cell variant of anaplastic carcinoma is practically nonexistent at the present time, since most of these tumors are currently appropriately classified as lymphoma, medullary carcinoma, or poorly differentiated thyroid carcinoma (14,32,33).

Differential diagnosis

Poorly differentiated thyroid cancer. Thyroid carcinomas can exhibit an entire spectrum of differentiation. Poorly differentiated carcinoma is intermediate on the spectrum between well-differentiated and anaplastic carcinoma and may represent a transition form(34,35). The majority of these lesions have an architectural growth pattern that is characterized by large, well-defined solid nests, an appearance that mimics neuroendocrine tumors, hence the terminology ”insular” carcinoma. Other tumors have trabecular growth patterns. The tumor cells are usually small and uniform in size; in contrast to anaplastic carcinomas there is little pleomorphism, and no bizarre, giant, or multinucleated cells are found. There is mitotic activity with three or more mitoses per 10 high power fields (36,37), which is less than that seen in anaplastic carcinomas. Tumor necrosis is usually identified as single cell necrosis or small well-defined necrotic foci in the center of cell nests rather than the large areas of geographic necrosis that are characteristic of anaplastic carcinoma. Preservation of immunohistochemical markers of epithelial and thyroid differentiation, such as thyroglobulin (Tg) and thyroid transcription factor 1 (TTF1), in poorly differentiated carcinoma can help to distinguish it from anaplastic carcinoma.

Squamous cell thyroid cancer. ATC with a predominant squamoid growth pattern may resemble squamous carcinoma morphologically and immunohistochemically (14,15,18,38,39). Primary squamous cell carcinoma of the thyroid is exceedingly rare (40–42), accounting for less than 1% of thyroid malignancies. It can be impossible to distinguish these lesions, and their clinical behavior is not distinct (43). The presence of an adjacent well-differentiated component of papillary or follicular carcinoma or the presence of thyroid-specific molecular events provides evidence that a lesion is ATC (16,44,45). The morphology of primary squamous carcinoma of the thyroid is not distinctive, and its appearance is identical to squamous carcinoma arising elsewhere; therefore, whenever possible an invasive laryngeal or metastatic squamous carcinoma must be excluded prior to making the diagnosis of a primary squamous carcinoma of the thyroid.

Other tumors: the role of immunohistochemisty. On fine needle aspiration (FNA) or core biopsy, some thyroid tumors present a variety of differential diagnostic possibilities depending on the predominant morphologic pattern present in the sample. When the tumor presents in the form of a poorly differentiated large B cell carcinoma, the differential diagnosis includes large cell lymphoma (both diffuse large cell lymphoma and anaplastic large cell lymphoma), medullary carcinoma, direct extension of a laryngeal carcinoma,metastatic carcinoma (18), metastatic melanoma, and ATC. When the spindle cell morphology dominates or the tumor appears biphasic with both epithelial and spindle cell components, the differential diagnosis once again includes medullary carcinoma,metastatic carcinoma (in particular a sarcomatoid renal primary), melanoma, and ATC, but now must include sarcomas (both primary and metastatic) as well as involvement of the thyroid by a primary laryngeal sarcomatoid squamous carcinoma. Occasionally infectious and inflammatory lesions can mimic anaplastic carcinoma (46,47). Resolution of these diagnostic possibilities requires careful attention to the morphology, combined with special stains or immunohistochemical studies (48–50) (see Table 4) and clinical information to arrive at the conclusion that the poorly differentiated malignancy is an ATC.


Immunohistochemical marker    DTC  MTC   ATC  SCC Lymphoma
Pankeratin (AE1/AE3) + + +/− + a
High molecular weight keratins + (PTC)

− (FTC)

−/+ +
TTF-1 + +/− −/+
PAX8 + +/− +/− +/−b
Thyroglobulin +
Synaptophysin +
Chromogranin +
Calcitonin +
P53 − (rare +) + +/− +/−
E-cadherin +   +
B-catenin membranous   nuclear or −   +/− (nuclear)
CD45         +

     a Plasmablastic lymphoma, anaplastic large cell lymphoma, and very rarely some diffuse large B cell lymphomas can express cytokeratin that is detected by AE1/AE3 cytokeratin monoclonal antibodies.

     b PAX8 was detected by immunohistochemistry in normal B cells, but was not studied in lymphoma. It is very likely that various lymphomas express PAX8.

     ATC, anaplastic thyroid carcinoma; CD45, protein tyrosine phosphatase receptor type C; CEA, carcinoembryonic antigen; DTC, differentiated thyroid carcinoma; FTC, follicular thyroid carcinoma; MTC, medullary thyroid carcinoma; P53, cellular tumor antigen p53 or tumor protein p53; PAX8, paired box protein Pax 8 or paired box 8; PTC, papillary thyroid carcinoma; SCC, squamous cell carcinoma; TTF-1, thyroid transcription factor 1; +, positive; −, negative; +/−, usually positive, often negative; −/+, usually negative, may be positive.


    Morphologic diagnosis with appropriate immunostaining as relevant is mandatory to exclude other less aggressive and treatable entities that can mimic ATC.

       Strength of Recommendation: Strong

       Quality of Evidence: Moderate

Cytology and pathology procedures

Interobserver variability. How consistently do pathologists make the diagnosis of ATC on the same sample? There are no data on interobserver variation in the diagnosis of ATC. However, this diagnosis is often one of ”exclusion” when there is no evidence of other lesions that can mimic ATC. As with any result of an interpretive nature, accuracy may differ in the hands of experts who have more experience with this disease.

FNA and core biopsy. Morphological diagnosis of FNA biopsy may be diagnostic (51–54), but FNA may not always yield diagnostic material. In cases in which the limited sampling of FNA biopsy yields mainly necrotic or inflamed tissue, there may be a need for core biopsy or open biopsy. There is no cytologic description of the paucicellular variant of ATC, likely due to an inability to obtain diagnostic tissue in this setting (13); core biopsy or open biopsy is usually required for this diagnosis.


    FNA cytology or core biopsy should play a role in the preoperative diagnosis of ATC. In cases in which the limited sampling of FNA or core biopsy yields material that is nondiagnostic, open biopsy should be performed to obtain diagnostic tissue.

       Strength of Recommendation: Strong

       Quality of Evidence: Low

Intraoperative frozen section and pathology consultation. Intraoperative consultation is used as a method of providing rapid diagnosis to assist in determining the ongoing operative extent and approach. Appropriate utilization of frozen sections should be limited to situations that fulfill three criteria. These are (i) the result of the intraoperative consultation should alter the surgical procedure, (ii) the information provided by the intraoperative consultation should not be available by other preoperative tests, and (iii) the information required must be realistically and practically obtainable by pathologists performing the intraoperative consultation.

These three criteria are not frequently fulfilled in the diagnosis of ATC. Usually, the diagnosis is anticipated prior to surgery for open biopsy, and the procedure is being performed primarily to obtain tissue for diagnosis. Since freezing of tissue compromises morphology and may preclude accurate and definitive diagnosis, it is not recommended as a routine procedure. However, there may be indications for intraoperative evaluation in two scenarios.

If preoperative biopsies yield mainly necrotic tissue, the surgeon’s objective in performing a biopsy is to obtain viable diagnostic material. Confirmation of such may entail intraoperative evaluation of a small piece of tissue to ensure that the remainder of the specimen is appropriate for further studies. This might be done by frozen section, but may be equally as productive when tissue is used to make ”touch preparations” for cytologic assessment, thereby not freezing the tissue and altering its morphology. Involvement of the pathologist at this stage is also helpful to ensure that samples are correctly collected for other nonroutine testing such as microbiological examination to exclude infectious disorders that can mimic anaplastic carcinoma, collection of material for flow cytometry if a hematological malignancy is in the differential diagnosis, or snap freezing material for molecular diagnostics. The involvement of the pathologist for any of these procedures will not yield a diagnosis at the time of intraoperative consultation but should result in a more thorough and accurate final report.

The second situation in which intraoperative pathologist consultation may prove valuable in the diagnosis of ATC arises when the diagnosis is not suspected preoperatively, but is encountered unexpectedly. This rare occurrence may be faced when a patient is thought to have a differentiated thyroid malignancy, even after biopsy has confirmed the diagnosis, but the surgeon encounters unusual features during the operative procedure. In this situation, a frozen section may be used to confirm dedifferentiation of a lesion, and the result may alter the surgical approach.


    Whenever possible, a definitive diagnosis should be obtained prior to surgery. Intraoperative pathology consultation can be used to define the adequacy of the resected tissue for diagnostic evaluation or to identify ATC in a patient when that diagnosis was not anticipated preoperatively. Intraoperative pathology consultation is not usually appropriate for definitive diagnosis.

       Strength of Recommendation: Strong

       Quality of Evidence: Low

Thyroid histopathology. The indications for thyroidectomy are discussed later in this article. Histologic examination of a thyroidectomy specimen provides additional material to examine the extent of disease and to identify additional coexisting differentiated components of the disease. ATC often originates in an abnormal thyroid gland; a history of goiter is reported in >80% of cases (55,56), and the reported association between well-differentiated thyroid carcinoma and ATC ranges from 7% to 89% of cases. The lower figures are likely underestimates, attributable to failure to detect a well-differentiated component due to inadequate sampling (14,15,24,35,55–57). The association of papillary carcinoma, particularly the more aggressive tall cell variant, with anaplastic tumors has also been described (16,35,56).

In many cases (20%–90%), histological examination of ATC identifies a coexisting well-differentiated carcinoma, or patients have a history of previously resected well-differentiated or poorly differentiated thyroid carcinoma (14,15,24,26, 55,58,59). This supports the notion that many ATCs develop by dedifferentiation of a preexisting well-differentiated thyroid carcinoma. Well-differentiated papillary carcinoma, often the tall cell variant, is the most common coexistent carcinoma, followed in frequency by follicular carcinoma of conventional type or oncocytic (Hürthle cell) type (16, 22,35,60) An adjacent component of poorly differentiated thyroid carcinoma may also be seen. In some anaplastic carcinomas, no well-differentiated or poorly differentiated cancer component is found on histological examination. The presence of coexisting well-differentiated or poorly differentiated carcinoma may be substantiated by finding preserved immunohistochemical staining for Tg, TTF1, and paired box protein Pax 8 (PAX8) in better differentiated areas (22). The finding of coexisting or preexisting well-differentiated or poorly differentiated carcinoma is helpful in establishing a definitive diagnosis of ATC by ruling out metastatic carcinoma and nonepithelial malignancy. The proportion of anaplastic carcinoma in a given tumor may influence prognosis. Although data are limited, survival is likely to be more prolonged in cases in which anaplastic carcinoma comprises only a small component of an otherwise well-differentiated papillary or follicular thyroid carcinoma (55,61,62)


    Pathological evaluation should provide information on the proportion of tumor that comprises ATC and coexistent well-differentiated or poorly differentiated thyroid carcinoma, which may affect prognosis and guide management.

       Strength of Recommendation: Strong

       Quality of Evidence: Low

Molecular techniques. ATCs are typically aneuploid and have a complex karyotype with multiple numerical and structural chromosomal abnormalities (63–69). Loss of heterozygosity at multiple chromosomal regions is very common (70–73). A progressive accumulation of chromosomal alterations can be observed when comparing well-differentiated carcinomas with poorly differentiated carcinomas and anaplastic carcinomas, which supports the multistep dedifferentiation process (69,74). Most common somatic mutations found in ATCs are in the TP53 and β-catenin (CTNNB1) genes (75–80). These mutations rarely occur in well-differentiated thyroid cancer. Other mutations, such as BRAF and RAS, are common in both well-differentiated thyroid cancer and ATCs and are likely to be early events in thyroid carcinogenesis that may predispose to tumor dedifferentiation (60,81). PIK3CAand PTEN gene mutations also occur in both well-differentiated thyroid cancer and ATCs. Papillary thyroid carcinomas carrying RET/PTC3 rearrangement may also be prone to dedifferentiation (82,83). Other rearrangements found in well-differentiated papillary and follicular carcinomas, such as RET/PTC1 and PAX8/PPARγ, have not been detected in anaplastic carcinomas, but this may be due to technical limitations since the studies have been based on RNA expression that may be lost with dedifferentiation. Profound alterations in gene expression, microRNA, and protein composition are also found in ATCs (84). These molecular alterations provide potential therapeutic targets for novel therapies that may be clinically relevant in the future. Molecular markers have also been studied as candidate prognostic factors; it appears that higher urokinase-type plasminogen activator receptor (u-PAR) expression and protein formation may be associated with increased mortality (85). Currently, molecular techniques do not play a significant role in the diagnosis of ATC or provide guidance for patient management, although they may occasionally be used to aid the diagnosis. Finding a BRAF or RAS mutation in a small biopsy sample would generally favor thyroid origin of an undifferentiated tumor, but these mutations are not specific for thyroid carcinomas and may occur in other malignant tumors. Detection of RET/PTC and PAX8/PPARγ expression is fairly specific for thyroid cell origin; however, it is rarely found in ATC (79). Although TP53 and β-catenin mutations may occasionally be seen in poorly differentiated carcinomas, their presence would suggest tumor dedifferentiation.

  • RECOMMENDATION 5 Molecular studies based on DNA/RNA analysis are not currently required for diagnosis and management of patients with ATC.

       Strength of Recommendation: Strong

       Quality of Evidence: Moderate