Printer Friendly

Update on selected salivary gland neoplasms.

Neoplasms of the salivary glands account for 2% to 7% of head and neck neoplasms but less than 1% (approximately 0.3%) of malignancies from all body sites. (1) Because of their rarity, the number of histologic subtypes, and the morphologic overlap and heterogeneity among these subtypes, salivary gland tumors often remain diagnostically challenging even for experienced pathologists. Adding to the diagnostic difficulty is the fact that new tumors and improvements in understanding the biologic behavior of previously established subtypes are continually being recognized. This review provides a summary of 4 salivary gland neoplasms that we often see in consultation and that pose diagnostic difficulties. In addition, we highlight some of the relatively recent developments in understanding the biology of these tumors and the significance of reporting critical diagnostic information to surgeons and oncologists to ensure appropriate therapeutic management.


Pleomorphic adenoma (PA), or benign mixed tumor, is the most common salivary gland tumor and accounts for almost two-thirds of all salivary gland tumors. (2) Pleomorphic adenomas are composed of both epithelial cells and myoepithelial cells, which produce a mesenchymal stromal component that is often myxoid or chondromyxoid. Varying amounts and arrangements of the epithelial, myoepithelial, and stromal components give rise to a broad degree of morphologic diversity (ie, "pleomorphic"). The typical PA is composed of roughly equal mixtures of epithelial and stromal elements. Some authors (3) recognize a cellular variant, which is arbitrarily defined as having more than 80% epithelial components. Conversely, an acellular PA has at least 80% stromal components.

At the cellular end of the spectrum, the distinction between a cellular PA, basal cell adenoma, or myoepithelioma can be somewhat arbitrary and subjective. Likewise, an acellular PA may closely resemble a soft tissue chondroma or even myxoma. Generally, the degree of cellularity does not have prognostic significance in PA. The only exception to this statement might concern the acellular myxoid PA, which, on surgical manipulation, can easily rupture through the capsule and contaminate the operative field, thus resulting in implants.

Approximately 6.2% of PAs harbor malignancy. (4) Malignant mixed tumors of the salivary glands can be classified as carcinoma arising in PA (also known as carcinoma ex PA; CaExPA), "benign" metastasizing PA, and carcinosarcoma (true malignant mixed tumor).2 Carcinoma ex PA is defined as a carcinoma that arises in association with a PA. By definition, an antecedent or coexistent PA must be identified to make this diagnosis. According to some authors, (5) those that show prominent zones of hyalinization and at least moderate mitotic activity (mean, 1.5/10 high-power fields [HPFs]) have a 13.8% probability of malignant transformation. Clinical findings that indicate a greater likelihood of malignant transformation include an origin in the submandibular gland, longer duration of tumor, older patient age (mean, 61 years), and larger tumor size (mean, 4 cm). (6)

More than 90% of all malignant mixed tumors are CaExPAs and most are high grade, with salivary duct carcinoma; adenocarcinoma, not otherwise specified; undifferentiated carcinoma; and myoepithelial carcinoma being the most common subtypes. (6,7) Table 1 summarizes the pathologic features that are useful in the diagnosis of CaExPA.

It has become clear that not all CaExPAs behave in similar fashion. Several authors (7-9) have shown that the degree of invasion beyond the tumor capsule predicts the behavior and outcome of CaExPA. The concept of invasion, which is so crucial to the understanding and appropriate classification of tumors in this setting, is not entirely straightforward. Grossly, PAs are usually well circumscribed and encapsulated in the major salivary glands and nonencapsulated in the minor salivary glands. However, careful microscopic examination of most tumors reveals that the degree of encapsulation is actually quite variable from area to area for a given case. Where segments of capsule are attenuated or absent, microscopic "pseudopods" of tumor can extend into peritumoral tissues. Since they are contiguous with the main tumor, pseudopods are not considered to represent true invasion and are consistent with a benign PA. Historically, when PAs were excised with simple nucleation rather than lobectomy, these pseudopods were often left behind and became a nidus for recurrence.

By degree of invasion, CaExPAs can be subdivided into 3 prognostic subtypes: (1) noninvasive, (2) minimally invasive, and (3) invasive carcinomas (Table 2).2 This classification emphasizes that the term carcinoma ex pleomorphic adenoma is imprecise and comprises a spectrum of lesions that is too broad and nonspecific. Clinicians require more detailed information for the management of their patients. Depending on the type of tumor and degree of invasion, patients with CaExPA can have a 5-year survival rate that exceeds 90% or that is less than 20%. (4,7-9) Typically, CaExPAs are widely invasive and associated with advanced stage at presentation with poor 5-year survival rates. In addition to tumor type and differentiation, pathologists must report the presence and degree of invasion when diagnosing CaExPA, as all of these factors impact on prognosis and therapy. Although uncommon, the concept of noninvasive or minimally invasive CaExPA remains relatively unknown to many of our clinical colleagues, who are often unsure how to manage affected patients.

Noninvasive carcinomas arising in PA are sometimes referred to as in situ or intracapsular carcinomas. These tumors are characterized by relatively abrupt transitions between histologically bland PA and areas with obviously malignant cytologic features but are limited to the confines of the PA without extracapsular invasion. The noninvasive carcinoma may invade the matrix of the preexisting PA but is not considered clinically invasive until it penetrates beyond the capsule of the PA. Microscopically, there is obvious contrast between areas replaced with large cells that have hyperchromatic and pleomorphic nuclei with increased mitotic activity and the surrounding benign cells in the PA (Figure 1, A). The difference in cell size between the 2 cell populations is often quite dramatic, as is the amount of mitotic activity; the latter may be highlighted by highly elevated Ki-67 (MIB-1) labeling index (Figure 1, B). HER-2/neu, p53, and androgen receptor may be overexpressed in malignant areas, but it is important to be aware that about 5% to 10% of benign PAs express these markers, such that they are not entirely specific for making a diagnosis. (10)

Minimally invasive CaExPA, in contrast, is defined as a carcinoma that invades no more than 1.5 mm beyond the capsule of the PA (Figure 1, B). Distinguishing between a benign pseudopod and true invasion may, at times, be difficult. A desmoplastic reaction is helpful, if present, since it is not seen with benign pseudopods (Figure 1, C). So-called lateral extension (mushroom-type invasion) into and through the capsule (Figure 1, D), akin to that seen in some thyroid carcinomas, also implies malignancy, as does perineural invasion. In addition, the invasive component should demonstrate atypical cytologic and architectural features.

The concept of noninvasive carcinoma arising in PA was probably first brought to our attention by LiVolsi and Perzin (11) in 1977. They described 47 cases of CaExPA; 6 (13%) of these carcinomas were noninvasive. None of the affected patients had local recurrence or metastases. Review of the literature suggests that metastasis develops in less than 2% of noninvasive carcinomas, with only a single case of locoregional lymph node metastasis reported. (12) It must be emphasized that the diagnosis of noninvasive CaExPA requires careful sampling of the entire tumor periphery to rule out invasive disease.

The significance of minimal extracapsular invasion has changed over time. It was initially addressed in 1984 by Tortoledo et al, (7) who evaluated 40 cases. In their experience, when the extent of invasion was no more than 8 mm from the capsule (17 patients), none of the patients died of the disease. However, all of the 16 patients with follow-up died of disease when invasion exceeded 8 mm. Frequency of local recurrence was also strongly related to distance of invasion and status of resection margins in their series. In 1996, Brandwein et al8 studied 12 patients with noninvasive or minimally invasive CaExPA (minimally invasive defined as those carcinomas with 1.5-mm extracapsular invasion). Among the 8 patients with minimal invasion, there were no recurrences or metastases at mean follow-up at 2.5 years. Finally, in 2001, Lewis et al (6) evaluated their experience with 66 cases of CaExPAs with follow-up and were able to relate prognosis to degree of invasion. No recurrences or metastases were identified for patients with extracapsular invasion of less than 5 mm. In addition, those with extracapsular invasion of at least 15 mm had significantly decreased survival. These authors concluded that patients with extracapsular invasion of less than 5 mm will likely do well.


On the basis of these studies, which correlated degree of invasion in CaExPA with outcome, the World Health Organization has recommended that carcinomas arising in PA be classified as follows: noninvasive (in situ, intracapsular), minimally invasive CaExPA ([less than or equal to] 1.5 mm), and invasive CaExPA (>1.5 mm). (2) Noninvasive CaExPAs with free margins do not require additional treatment since they have a prognosis similar to that of conventional PAs. Minimally invasive carcinomas are capable of metastasis but the incidence is unknown. Treatment of patients with minimally invasive carcinoma is based on margin status and presence of perineural invasion; if either is present, then reexcision and/or radiation therapy may be indicated.


Polymorphous low-grade adenocarcinoma (PLGA) was first described in 1983 almost simultaneously by 2 groups of investigators. Freedman and Lumerman (13) first described 12 cases and referred to the tumor as a lobular carcinoma because they thought it resembled lobular carcinoma of the breast. Batsakis et al, (14) who also reported on 12 cases, referred to the tumor as a terminal duct carcinoma, thus suggesting a proposed histogenetic origin. Shortly thereafter, Evans and Batsakis (15) described 14 new cases and coined the term polymorphous low-grade adenocarcinoma, the currently accepted nomenclature.

Polymorphous low-grade adenocarcinoma occurs almost exclusively in the minor salivary glands of the oral cavity and oropharynx. Because it occurs only exceptionally in the major salivary glands, we generally do not even consider PLGA in the differential diagnosis of major salivary gland neoplasms. In some series, PLGA is the second most common malignant salivary gland tumor in the oral cavity, exceeded in frequency only by mucoepidermoid carcinoma. In our experience, it is probably the most common malignant salivary gland tumor occurring in this site. Most areas within the oral cavity sites are affected (Table 3); however, the palate is the most frequently affected site and accounts for about 60% of all PLGAs. (16,17) Polymorphous low-grade adenocarcinoma is twice as common in women as it is in men and occurs over a broad age range (average, 50-60 years) but may even occur in children. (17) These tumors tend to be nonpainful and slow growing and rarely exceed 6 cm (average, 2 cm).

Histologically, PLGA is characterized by a triad of infiltrative growth, multiple architectural growth patterns (hence the term polymorphous), and cytologic uniformity. (1) Characteristic growth patterns include solid, trabecular, glandular, cribriform, fascicular, cordlike, and papillary. The first 3 patterns are most common (Figure 2, A). Typically, the central portion consists of the more solid growth patterns, with the glandular and cordlike elements seen more often at the infiltrative periphery. Mitoses are rare and necrosis is usually not seen, but perineural invasion is quite common. In fact, perineural invasion is more prominent in PLGA than in almost any other salivary gland carcinoma. The proliferation is usually associated with abundant, paucicellular mucoid or collagenous stroma (Figure 2, B). Cartilage is not seen but occasional tumors have psammoma body-like calcifications. The variable growth patterns and stromal components can complicate the diagnosis, especially in small biopsy specimens. For example, PLGA with a cribriform pattern can be very difficult to separate from adenoid cystic carcinoma. In other cases, PLGA can be indistinguishable from PA and basal cell adenoma, especially the tubulotrabecular variant. As noted above, PLGA usually invades nerves, often in a so-called targetoid arrangement, with cords of cells concentrically arranged around nerves (Figure 2, C). This pattern of perineural invasion is very characteristic of PLGA and can be identified in about 30% of cases. (13)

One of the most characteristic features of PLGA is the nuclear uniformity. The cells are cytologically bland and can be cuboidal, columnar, or spindled with a mixture being quite common. They have scant to moderate amounts of amphophilic or eosinophilic cytoplasm. Occasional tumors have mucus cells, clear cells, or oncocytic cells, but these are typically a minority component in such cases. Even in these different cell types, the nuclei are uniform, round to ovoid, with finely dispersed or ground glass-type nuclear chromatin, with an appearance resembling that of papillary thyroid carcinoma (Figure 2, D). Unfortunately, these characteristic nuclear features are often not identifiable in small biopsy specimens, as these types of specimens tend to be crushed and distorted. Also, the low-grade nuclei can make margins difficult to interpret in tumors with glandular growth patterns, particularly in frozen sections. Recognizing infiltrative growth and perineural invasion thus becomes the most important diagnostic feature in many cases.

Complete local excision is the treatment of choice. Recurrence develops in 9% to 26% of cases, usually months after treatment but possibly after many years. (16-18) About 5% to 10% of tumors metastasize to cervical lymph nodes, but metastases below the clavicles are distinctly unusual. One issue that remains to be settled is the significance of papillary growth and prognosis in PLGA (Figure 2, E). Increasing amounts of papillary growth may be one feature that predicts aggressiveness in terms of lymph node metastases. According to Evans and Luna, (18) the presence of "more than focal" papillary growth pattern is associated with an increased risk of cervical lymph node metastases but not with local recurrence or distant metastases. Unfortunately, what exactly was meant by "more than focal" was not defined and the amount of papillary growth that is associated with this increased risk of lymph node metastasis is currently unclear. We mention it in our reports when 10% or more papillary growth is identified, acknowledging this as an arbitrary cutoff.

The differential diagnosis includes other oral cavity minor salivary gland tumors such as PA, adenoid cystic carcinoma, and sometimes low-grade papillary adenocarcinoma of salivary gland origin (LGPASO). Pleomorphic adenomas of the minor salivary glands are typically not encapsulated, but they are usually well circumscribed and lack a truly infiltrative growth. Unfortunately, when surgeons perform biopsies, they usually obtain the tumor from the center of the lesion, not the periphery, so it is often difficult to assess for invasion. If cartilage is present, PLGA can effectively be ruled out, as this feature is typically only seen in PA. Conversely, if perineural invasion is present, PA can be ruled out. Unlike PLGAs, PAs are characterized by ductal or glandlike structures comprising luminal cells surrounded by a mantle of myoepithelial cells that appear to "melt" into the stroma (ie, chondromyxoid stroma ) that they produce. This biphenotypic cellular composition is usually absent in PLGA; however, rare myoepithelial cells can be identified in PLGA by using immunohistochemical stains. Immunohistochemistry can be useful in this differential diagnosis if a pattern is present in which ductal cells are surrounded by 1 or more layers of myoepithelial cells. Although glial fibrillary acidic protein is a very poor marker for myoepithelial cells, some authors contend that it is usually positive in myoepithelial cells of PA but is negative in PLGA. However, it can be seen in up to 15 % of PLGAs. (17)

Separating PLGA from adenoid cystic carcinoma can also be very challenging in biopsy samples. About 20% of adenoid cystic carcinomas occur in the palate, where it is the third most common salivary gland malignancy (after mucoepidermoid carcinoma and PLGA). Like PLGA, adenoid cystic carcinomas can be architecturally heterogeneous with tubular, cribriform growth patterns occurring in varying quantities. (1) Cribriform growth can be seen in PLGA, but if a cribriform pattern is seen throughout the lesion, this suggests adenoid cystic carcinoma. As mentioned previously, perineural invasion is a prominent feature for both of these salivary gland malignancies, but a targetoid arrangement of perineural invasion is more typical of PLGA. Papillary growth can be seen in PLGA but is not a feature of adenoid cystic carcinoma. Cyst formation and calcific deposits are also more in keeping with PLGA than with adenoid cystic carcinoma. One of the most helpful features is the nuclear characteristics. As opposed to the homogeneous "papillary thyroid carcinoma-like" nuclei characteristic of PLGA, adenoid cystic carcinomas tend to have hyperchromatic and angulated nuclei. The presence of these dark and angulated nuclei can be diffuse or patchy but their presence is quite helpful in separating these 2 lesions on biopsy samples.


Immunohistochemical staining is not extremely helpful in differentiating PLGA from adenoid cystic carcinoma, in our experience. Diffuse and intense c-kit (CD117) staining suggests adenoid cystic carcinoma, although PLGAs show immunoreactivity in most cases. (19,20) Other authors (21) have suggested that the distinction between PLGA and adenoid cystic carcinoma can be made by using S100 protein or BCL-2, but we find these stains to be too unreliable in yielding a definitive diagnosis on small biopsies. Some authors suggest the Ki-67 (MIB-1) proliferation index can be helpful. Usually the proliferative index in PLGA is less than 5%, whereas in adenoid cystic carcinoma it is usually greater than 20%. (22)

Two uncommon lesions that have significant overlap with PLGA may be variants of PLGA and deserve mention. Like PLGA, LGPASO occurs almost exclusively in the oral cavity, with the palate being the most common site. (23) The nuclear features are similar; however, LGPASO consists of diffuse papillary growth with or without intracystic papillary projections and does not have the polymorphic growth patterns so typical of PLGA. In addition, LGPASO appears to be more aggressive, with more frequent local recurrence rates (about 25%) and lymph node metastases in about 15% of cases. Cribriform adenocarcinoma of the tongue (CAT) is likely a lesion closely related to PLGA and occurs most often in the posterior tongue (Figure 2, F).24 The nuclear features are identical to those of PLGA and they have a polymorphic appearance but are dominated by cribriform and solid growth patterns. Like LGPASO, CAT also appears to have a more aggressive pathologic behavior than traditional PLGA, with cervical lymph node metastases (usually multiple nodes) occurring in nearly all patients. Although the number of reported patients with LGPASO and CAT is small, overall survival does not appear to be much worse than for PLGA.


Salivary duct carcinoma (SDC) was first described in 1968 by Kleinsasser and colleagues. (25) It is one of the most aggressive salivary gland carcinomas and histologically resembles high-grade ductal adenocarcinoma of the breast. Salivary duct carcinoma accounts for almost 10% of all salivary gland malignancies. (2,26) Patients present over a broad age range (22-91 years), but most are older than 50 years (mean, 60-79 years). Tumors occurring in individuals younger than age 50 are uncommon. Paradoxically, even though it resembles breast carcinoma, SDC is more common in men by a ratio of 3:1 or 4:1. It can arise de novo or in the setting of a PA (ie, SDCExPA). Approximately 20% of SDCs are reported to arise in a background PA, but the true incidence of this occurrence is unknown, as many SDCs likely "overrun" the PA, thus making its recognition difficult. (26) The parotid gland is by far the most common tumor site (75%-88% of cases), with the submandibular gland and a combination of various minor salivary gland sites affected roughly equally. (1,27) Tumor size can vary between a few millimeters to 7 cm, with a mean of about 3.5 cm. The gross features of SDC closely resemble those of breast carcinoma. Because of an abundant fibrous matrix, they are very firm and often ill-defined, with infiltration of adjacent parenchyma frequently identifiable. Comedotype necrosis, calcification, and cystic change may also be observed macroscopically. Gross features of a background PA can be observed in some examples.

Microscopically, SDCs are characterized by both infiltrative and intraductal carcinoma resembling breast ductal carcinoma (Figure 3, A). (1,27) The most characteristic histologic feature is variably sized, rounded nodules with a cribriform, solid, cystic, or papillary architecture. These so-called ductal lesions frequently demonstrate comedonecrosis, which may undergo calcification akin to intraductal breast carcinoma. Often associated with these large nodules are small infiltrative nests, ducts, or cords of cells with similar cytologic features. Salivary duct carcinomas tend to be very high grade cytologically, with enlarged pleomorphic nuclei, coarse chromatin, prominent nucleoli, and numerous mitoses including many atypical forms (Figure 3, B). The cells have moderate to abundant amounts of eosinophilic to amphophilic cytoplasm, frequently imparting an apocrine-like phenotype. Some tumors may focally have a squamoid appearance (Figure 3, C); however, goblet cells are not found. Whereas the ductal lesions are truly infiltrative (ie, they lack a peripheral basal/myoepithelial layer), intraductal foci can often be discerned by using immunohistochemistry to identify peripheral layer of residual nonneoplastic myoepithelial cells in the involved ducts. (28) Large zones of dense fibrosis can be seen in many examples and may be indirect evidence that the SDC arose from a background PA. Perineural invasion can be seen in 60% of cases and about one-third show lymphovascular invasion. (27)

Several histologic SDC variants have been described in recent years in addition to those that arise in a background PA (SDCExPA). Some SDCs have an associated malignant spindle cell or pleomorphic growth pattern of varying proportion. This sarcomatoid variant (29) differs from true carcinosarcomas, as the sarcomatoid component is immunoreactive with epithelial markers. Like those seen in the breast, lung, and genitourinary tract, a micropapillary variant has been described and is similarly reported to have a more aggressive behavior. (30) Rare examples are composed of foci of traditional SDCs associated with areas of tumor cells embedded in lakes of extracellular mucin (mucin-rich variant) similar to mucinous (colloid) carcinomas of the breast. (31) An extremely rare osteoclast-like giant cell carcinoma variant has also been described.32 Finally, some SDCs demonstrate pure intraductal growth (intraductal variant) as confirmed by immunohistochemistry and are not associated with the adverse outcome of typical infiltrative SDCs. (28) In our opinion, before a diagnosis of intraductal variant of SDC is made, the entire lesion should be submitted for histologic evaluation and judicious use of immunohistochemistry should be made to confirm the intraductal nature. A lesion reported initially as low-grade salivary duct carcinoma and now known as low-grade cribriform cystadenocarcinoma is, in our opinion and that of others, unrelated to SDC. (33)


Salivary duct carcinoma is typically positive for both low-molecular-weight and high-molecular-weight cytokeratins as well as epithelial membrane antigen. Myoepithelial markers are negative in the tumor cells but they may highlight a peripheral myoepithelial layer in some foci. S100 protein can stain positively or negatively in SDCs but is not typically immunoreactive in a significant portion of the tumor. Like breast carcinomas, SDCs are often positive for gross cystic disease fluid protein (BRST2) and carcinoembryonic antigen by immunostaining, but they are almost uniformly negative for estrogen and progesterone receptors. Most tumors, approximately 85% to 95%, are positive for androgen receptor and SDC is now virtually defined by androgen receptor expression (Figure 3, D; Table 4). (26,34-41) More than two-thirds of SDCs are positive for HER-2/ neu overexpression (3+) as well, and two-thirds of these demonstrate HER-2/neu gene amplification by fluorescence and chromogenic in situ hybridization (Table 5). (42-45) Prostate specific antigen and prostatic acid phosphatase expression can be seen in a few tumors as well. (46)


Complete surgical excision with lymph node dissection and postoperative radiation with or without chemotherapy is the preferred treatment for this aggressive tumor. Local recurrence is seen for about one-third of patients with SDC and most have advanced stage at presentation. (26) Regional lymph node metastasis occurs in 60% of cases and is often present at initial diagnosis. (27) Unfortunately, nearly half of the patients have distant metastases, particularly to the lungs, liver, and bone. Overall, prognosis is dismal. Of 104 cases reported by Barnes et al, (27) 65% described patients who died of the disease, usually within 4 years of diagnosis. Because SDCs strongly express androgen receptor and overexpress HER-2/neu, some studies are now starting to evaluate treatment with Herceptin and anti-androgen therapy. The verdict about efficacy of these treatments is still pending; some tumors have responded to treatment, but this has not been a universal finding. (44,47)

The differential diagnosis of SDC includes metastases, especially breast carcinoma. Gnepp (48) looked at 400 cases of metastatic tumors to the parotid and only 8 (2%) were from the breast. Metastatic prostatic carcinomas can be a problem, especially since some SDCs secrete prostate specific antigen and may be associated with elevated serum levels of this analyte. In the analysis by Gnepp, only 0.5% of parotid metastases were from prostatic primary cancers. A thorough history and physical examination with serologic and immunohistochemical studies will resolve this differential in most cases. Because of the eosinophilic or apocrine epithelium in SDC, oncocytic adenocarcinoma may enter the differential diagnosis. Oncocytic adenocarcinoma typically consists of solid tumor nodules without prominent comedonecrosis or papillary growth. In contrast to SDC, oncocytic adenocarcinoma is associated with numerous mitochondria and results for mitochondria will be positive with stains such as phosphotungstic acid hematoxylin or with immunohistochemical stains for cytochrome c oxidase. The absence of epidermoid and mucus cells rules out a high grade mucoepidermoid carcinoma. The strong resemblance to breast carcinoma effectively rules out a diagnosis of high-grade adenocarcinoma, not otherwise specified, which is a tumor demonstrating ductal differentiation but lacking features diagnostic of another histologic subtype of salivary gland malignancy. The demonstration of strong positive nuclear staining for androgen receptor can be quite helpful in problematic cases, as oncocytic adenocarcinoma, mucoepidermoid carcinoma, and adenocarcinoma, not otherwise specified, typically yield negative results for this test.


Myoepithelial salivary gland neoplasms are tumors composed exclusively or almost exclusively of myoepithelial cells. (1,2) Whereas some authors believe that myoepithelial neoplasms should be purely myoepithelial, some allow for a minor ductal component to be present, typically less than 5%. Myoepithelial neoplasms can exist in both benign (myoepithelioma) and malignant (myoepithelial carcinoma or malignant myoepithelioma) forms. Most are benign but around 10% to 20% are malignant. (49) Myoepithelial neoplasms are being recognized more frequently and currently represent around 1% to 2% of all salivary gland neoplasms. The most common sites of origin are the parotid gland and palate, with these 2 sites accounting for approximately 75% of all myoepithelial neoplasms. (1,49) They can also be seen in other major and minor salivary glands, as well as in the seromucous glands within the sinonasal cavity and larynx.

The age distribution of persons affected with myoepithelioma is similar to that associated with PA, with most cases occurring in the fifth or sixth decade of life (range, 14-81 years). (50) It usually presents as a slowly enlarging mass, rarely more than 5 cm in greatest dimension. Grossly, it can vary from tan to white to yellow, and is characteristically quite firm. Like PA, it is encapsulated when it involves the major salivary glands but nonencapsulated when it involves the minor salivary glands.

Myoepithelial tumors can be difficult to recognize histologically because of variety in their architectural and cytologic appearances. The cellular morphology is often spindled (Figure 4, A) or plasmacytoid-hyaline (Figure 4, B), but clear (Figure 4, C), epithelioid (Figure 4, D), and oncocytic cell types can also be seen. A mixture of 2 or more of these cell types is not infrequent. (1) In addition to these 5 characteristic cell types, there are also 6 characteristic growth patterns for myoepithelial tumors: they can be solid, nodular, mucinous, trabecular, pseudoglandular, or reticulated (Figure 4, E). The amount of extracellular matrix in these tumors varies but is typically absent in the spindle cell-rich tumors and is often mucinous or fibrous in the other types. On ultrastructural examination, myoepithelial cells have basal lamina, desmosomes, pinocytotic vesicles, and cytoplasmic filaments with focal dense bodies. (49)


A few general statements about the histologic subtypes of myoepithelioma can be made. First, the plasmacytoid and reticulated lesions usually have a mucoid and/or myxoid matrix. Second, the solid spindle cell type typically occurs in the parotid gland and usually has a well-formed capsule. Third, clear cell myoepitheliomas contain abundant cytoplasmic glycogen. Although it looks histologically and cytologically bland, the clear cell variant recurs frequently and occasionally may metastasize. Therefore, we often make a comment about this when diagnosing clear cell myoepitheliomas. Fourth, it can be easy to mistake the solid or pseudoglandular patterns for an adenocarcinoma. These variants will be cytologically bland and show no evidence of destructive invasion of adjacent parenchyma. Fifth, the plasmacytoid variant is the most common type seen in the palate. Finally, some myoepitheliomas may resemble schwannomas or they may show focal squamous differentiation. This is not surprising, as we know that the source of squamous epithelium in PA is the myoepithelial cell. It is interesting that one never sees cartilage in myoepitheliomas, as the source of cartilage in PA is also the myoepithelial cell.

Immunohistochemical reactivity for cytokeratin and at least 1 other marker of myoepithelial differentiation is required to make this diagnosis. (2) Commonly used myoepithelial markers include smooth muscle actin, calponin, p63, CD10, vimentin, glial fibrillary acidic protein, smooth muscle myosin heavy chain, and S100 protein. Savera et al51 demonstrated that there is really no best single specific immunohistochemical marker for a myoepithelial cell. According to these authors, the markers that are more likely to be positive in a myoepithelial lesion are cytokeratin AE1/AE3, vimentin, S100 protein, calponin, p63, 34[beta]E12, and CD10. The take home message from this study, as is also our experience, is that a combination of myoepithelial markers should be used to confirm a suspected diagnosis of myoepithelial neoplasm. For cases we suspect are myoepithelial, we like to use a combination of about 3 to 5 antibodies including cytokeratin (AE1/AE3 or CAM 5.2), p63 (Figure 4, F), S100 Protein (Figure 4, G), smooth muscle actin (Figure 4, H), and calponin. We find p63 particularly helpful because of its high sensitivity and specificity, as well as the fact that it has crisp nuclear staining and serves as a nice internal control (normal myoepithelial cells in salivary acini and ducts; Figure 4, F). S100 protein is sensitive but demonstrates much less specificity (Figure 4, G). One of the reasons we like to use S100 protein is for the added benefit that it allows the identification of more subtle perineural invasion. In our experience, glial fibrillary acidic protein is a poor marker for myoepithelial cells and is characterized by low sensitivity. In cases we strongly suspect of being myoepithelial, for which these markers are negative, we will use some of the other markers.

Myoepithelial carcinoma is an uncommon salivary gland tumor accounting for about 0.1% to 0.5% of all salivary gland neoplasms. The World Health Organization has endorsed the term myoepithelial carcinoma rather than malignant myoepithelioma but both are appropriate in our opinion. (2) Myoepithelial carcinomas are seen in patients who are about a decade older than those with myoepitheliomas (mean, 50-60 years) but occur over a broad age range (24-81 years). (51,52) They may arise de novo, from a preexisting myoepithelioma, or as the carcinomatous component of a CaExPA. Approximately 50% to 70% of myoepithelial carcinomas are associated with a preexisting benign tumor. (51,52) Two-thirds occur in the parotid gland, but they can occur in other major salivary glands and various minor salivary gland sites as well. (51) Grossly, they share features with myoepitheliomas but typically also have infiltrative growth into adjacent parenchyma or soft tissues and may demonstrate hemorrhage and necrosis. Most are between 3.5 and 5 cm, but some can be quite large.

Myoepithelial carcinomas are composed of the same cell types and growth patterns as seen for myoepitheliomas. In contrast to myoepithelioma, demonstration of destructive invasive growth is the essential diagnostic feature of myoepithelial carcinoma, as some tumors can be exceptionally bland (Figure 5, A). (2) Other histologic features suggesting the diagnosis of malignancy in myoepithelial lesions include cellular pleomorphism, more than 7 mitoses per 10 high-power fields, perineural or lymphovascular invasion, and necrosis (Figure 5, B and C). (49) They also frequently exhibit Ki-67 (MlB-1) labeling indexes greater than 10%, aneuploidy, and p53 overexpression. (52) Some are very high-grade and myoepithelial differentiation may not be readily apparent in these cases. A characteristic feature of some myoepithelial carcinomas is neoplastic lobules that demonstrate central coagulation-type necrosis with a periphery of viable myoepithelial carcinoma (Figure 5, D). (51) As with myoepithelioma, immunohistochemical reactivity for cytokeratin and at least 1 other marker of myoepithelial differentiation is required to make this diagnosis.

Myoepithelial carcinomas can be remembered as a tumor of thirds: one-third of patients will remain disease free after treatment, about one-third will develop local recurrence, about one-third will have metastases (mainly hematogenous, although they can spread to lymph nodes), and about one-third of these patients will die of their disease, on average, somewhere around three years from diagnosis. (51) Because of the low rate of lymph node metastases, lymphadenectomy is not indicated in the absence of enlarged nodes. Surgical resection is the treatment of choice and chemoradiation therapy remains largely untested.

The differential diagnosis for the spindle cell myoepithelial neoplasms includes benign and malignant soft tissue tumors. Immunohistochemistry is necessary to sort this differential diagnosis out. Clear cell types may be confused with other clear cell salivary gland neoplasms such as epithelial-myoepithelial carcinoma; clear cell carcinoma, not otherwise specified; and clear cell variants of acinic cell carcinoma and mucoepidermoid carcinoma. Demonstration of myoepithelial cells will rule out these last 3 possibilities. Epithelial-myoepithelial carcinoma can overlap significantly with myoepithelial carcinoma but the demonstration of a ductal component suggests the former diagnosis. Finally, metastatic renal cell carcinoma may also enter the differential diagnosis. A thorough history and p63 immunohistochemistry can easily rule out metastatic renal cell carcinoma, as it is always negative for p63. (53)


(1.) Ellis GL, Auclair RL. Tumors of the Salivary Glands. Washington, DC: Armed Forces Institute of Pathology; 1996. Atlast of Tumor Pathology; 3rd series.

(2.) Gnepp DR, Brandwein-Gensler MS, El-Nagger AK, Nagao T. Carcinoma ex pleomorphic adenoma. Barnes EL, Eveson JW, Reichart P, Sidransky D, eds. Pathology and Genetics of Head and Neck Tumors. Lyon, France: IARC Press; 2005:242-243. World Health Organization Classification of Tumours; vol 9.

(3.) Seifert G, Langrock I, Donath K. A pathological classification of pleomorphic adenoma of the salivary glands [author's transl; in German]. HNO. 1976;24(12):4 1 5-426.

(4.) Gnepp DR. Malignant mixed tumors of the salivary glands: a review. Pathol Annu. 1993;28(pt 1):279-328.

(5.) Auclair PL, Ellis GL. Atypical features in salivary gland mixed tumors: their relationship to malignant transformation. Mod Pathol. 1996;9(6):652-657.

(6.) Lewis JE, Olsen KD, Sebo TJ. Carcinoma ex pleomorphic adenoma: pathologic analysis of 73 cases. Hum Pathol. 2001;32(6):596-604.

(7.) Tortoledo ME, Luna MA, Batsakis JG. Carcinomas ex pleomorphic adenoma and malignant mixed tumors: histomorphologic indexes. Arch Otolaryngol. 1984;110(3):172-176.

(8.) Brandwein M, Huvos AG, Dardick I, Thomas MJ, Theise ND. Noninvasive and minimally invasive carcinoma ex mixed tumor: a clinicopathologic and ploidy study of 12 patients with major salivary tumors of low (or no?) malignant potential. Oral Surg Oral Med Oral Pathol Oral Radiol Endod. 1996;81(6):655-664.

(9.) Olsen KD, Lewis JE. Carcinoma ex pleomorphic adenoma: a clinicopathologic review. Head Neck. 2001;23(9):705-712.

(10.) DeRoche TC, Hoschar AP, Hunt JL. Immunohistochemical evaluation of androgen receptor, HER-2/neu, and p53 in benign pleomorphic adenomas. Arch Pathol Lab Med. 2008;132(12):1907-1911.

(11.) LiVolsi VA, Perzin KH. Malignant mixed tumors arising in salivary glands--I, carcinomas arising in benign mixed tumors: a clinicopathologic study. Cancer. 1977;39(5):2209-2230.

(12.) Felix A, Rosa-Santos J, Mendonca ME, Torrinha F, Soares J. Intracapsular carcinoma ex pleomorphic adenoma: report of a case with unusual metastatic behaviour. Oral Oncol. 2002;38(1):107-110.

(13.) Freedman PD, Lumerman H. Lobular carcinoma ofintraoral minor salivary gland origin: report of twelve cases. Oral Surg Oral Med Oral Pathol. 1983;56(2): 157-166.

(14.) Batsakis JG, Pinkston GR, Luna MA, Byers RM, Sciubba JJ, Tillery GW. Adenocarcinomas of the oral cavity: a clinicopathologic study of terminal duct carcinomas. J Laryngol Otol. 1983;97(9):825-835.

(15.) Evans HL, Batsakis JG. Polymorphous low-grade adenocarcinoma of minor salivary glands: a study of 14 cases of a distinctive neoplasm. Cancer. 1984;53(4):935-942.

(16.) Vincent SD, Hammond HL, Finkelstein MW. Clinical and therapeutic features of polymorphous low-grade adenocarcinoma. Oral Surg Oral MedOral Pathol. 1994;77(3):41-47.

(17.) Castle JT, Thompson LD, Frommelt RA, Wenig BM, Kessler HP. Polymorphous low grade adenocarcinoma: a clinicopathologic study of 164 cases. Cancer. 1999;86(2):207-219.

(18.) Evans HL, Luna MA. Polymorphous low-grade adenocarcinoma: a study of 40 cases with long-term follow up and an evaluation of the importance of papillary areas. Am J Surg Pathol. 2000;24(10):1319-1328.

(19.) Penner CR, Folpe AL, Budnick SD. C-kit expression distinguishes salivary gland adenoid cystic carcinoma from polymorphous low-grade adenocarcinoma. Mod Pathol. 2002;15(7):687-691.

(20.) Edwards PC, Bhuiya T, Kelsch RD. C-kit expression in the salivary gland neoplasms adenoid cystic carcinoma, polymorphous low-grade adenocarcinoma, and monomorphic adenoma. Oral Surg Oral Med Oral Pathol Oral Radiol Endod. 2003;95(5):586-593.

(21.) Vargas h SD, Kaplan MJ, et al. Mixed tumor, polymorphous low grade adenocarcinoma and adenoid cystic carcinoma of the salivary gland: pathogenic implication and differential diagnosis by Ki-67 (MIB 1), BLC 2 and S-100 immunohistochemistry. Applied Immunohistochemistry. 1997;8:8-16.

(22.) Skalova A, Simpson RH, Lehtonen H, Leivo I. Assessment of proliferative activity using the MIB1 antibody helps to distinguish polymorphous low grade adenocarcinoma from adenoid cystic carcinoma of salivary glands. Pathol Res Pract. 1997;193(10):695-703.

(23.) Mills SE, Garland TA, Allen MS Jr. Low-grade papillary adenocarcinoma of palatal salivary gland origin. Am J Surg Pathol. 1984;8(5):367-374.

(24.) Michal M, Skalova A, Simpson RH, et al. Cribriform adenocarcinoma of the tongue: a hitherto unrecognized type of adenocarcinoma characteristically occurring in the tongue. Histopathology. 1999;35(6):495-501.

(25.) Kleinsasser O, Klein HJ, Hubner G. Speichelgangcarcinoma: Ein den Milchgangcarcinoma der Brustdruse analoge Gruppe von Speildrusen tumoren [in German]. Arch Kin Exp Ohre Nasen Kehlkopfheilkd. 1968;192(1):100-105.

(26.) Lewis JE, McKinney BC, Weiland LH, Ferreiro JA, Olsen KD. Salivary duct carcinoma: clinicopathologic and immunohistochemical review of 26 cases. Cancer. 1996;77(2):223-230.

(27.) Barnes L, Rao U, Krause J, Contis L, Schwartz A, Scalamogna P. Salivary duct carcinoma--part I: a clinicopathologic evaluation and DNA image analysis of13 cases with review of the literature. Oral Surg Oral Med Oral Pathol. 1994; 78(1):64-73.

(28.) Simpson RH, Desai S, Di Palma S. Salivary duct carcinoma in situ of the parotid gland. Histopathology. 2008;53(4):416-425.

(29.) Henley JD, Seo IS, Dayan D, et al. Sarcomatoid salivary duct carcinoma of the parotid gland. Hum Pathol. 2000;31(2):208-213.

(30.) Nagao T, Gaffey TA, Visscher DW, et al. Invasive micropapillary salivary duct carcinoma: a distinct histologic variant with biologic significance. Am J Surg Pathol. 2004;28(3):319-326.

(31.) Simpson RH, Prasad AR, Lewis JE, Skalova A, David L. Mucin-rich variant of salivary duct carcinoma: a clinicopathologic and immunohistochemical study of four cases. Am JSurgPathol. 2003;27(8):1070-1079.

(32.) Tse LL, Finkelstein SD, Siegler RW, Barnes L. Osteoclast-type giant cell neoplasm of salivary gland: a microdissection-based comparative genotyping assay and literature review--extraskeletal "giant cell tumor of bone" or osteoclast-type giant cell "carcinoma"? Am J Surg Pathol. 2004;28(7): 953-961.

(33.) Delgado R, Klimstra D, Albores-Saavedra J. Low grade salivary duct carcinoma: a distinctive variant with a low grade histology and a predominant intraductal growth pattern. Cancer. 1996;78(5):958-967.

(34.) Dimery IW, Jones LA, Verjan RP, Raymond AK, Goepfert H, Hong WK. Estrogen receptors in normal salivary gland and salivary gland carcinoma. Arch Otolaryngol Head Neck Surg. 1987;113(10):1082-1085.

(35.) Delgado R, Vuitch F, Albores-Saavedra J. Salivary ductcarcinoma. Cancer. 1993;72(5):1503-1512.

(36.) Barnes L, Rao U, Contis L, Krause J, Schwartz A, Scalamogna P. Salivary duct carcinoma--part II: immunohistochemical evaluation of 13 cases for estrogen and progesterone receptors, cathepsin D, and c-erbB-2 protein. Oral Surg Oral Med Oral Pathol. 1994;78(1):74-80.

(37.) Kay PA, Roche PC, Olsen KD. Salivary duct carcinoma: immunohistochemical analysis of androgen receptor, prostatic markers and HER-2/Neu oncoprotein in 40 cases. Lab Invest. 2001;81:140A.

(38.) Ockner DM, Mills SE, Swanson PE, et al. Salivary duct carcinoma: an immunohistologic comparison with invasive ducal carcinoma of the breast. Lab Invest. 1994;70:100A.

(39.) Nasser SM, Faquin WC, Dayal Y. Expression of androgen, estrogen, and progesterone receptors in salivary gland tumors: frequent expression of androgen receptor in a subset of malignant salivary gland tumors. Am J Clin Pathol. 2003; 119(6):801-806.

(40.) Hoang MP, Callender DL, Sola Gallego JJ, et al. Molecular and biomarker analyses of salivary duct carcinomas: comparison with mammary duct carcinoma. Int J Oncol. 2001;19(4):865-871.

(41.) Kapadia SB, Barnes L. Expression of androgen receptor, gross cystic disease fluid protein, and CD44 in salivary duct carcinoma. Mod Pathol. 1998; 11(11):1033-1038.

(42.) Skalova A, Starek I, Vanecek T, et al. Expression of HER-2/neu gene and protein in salivary duct carcinomas of parotid gland as revealed by fluorescence in-situ hybridization and immunohistochemistry. Histopathology. 2003;42(4): 348-356.

(43.) Cornolti G, Ungari M, Morassi ML, etal. Amplification and overexpression of HER2/neu gene and HER2/neu protein in salivary duct carcinoma of the parotid gland. Arch Otolaryngol Head Neck Surg. 2007;133(10):1031-1036.

(44.) Nabili V, Tan JW, Bhuta S, Sercarz JA, Head CS. Salivary duct carcinoma: a clinical and histologic review with implications for trastuzumab therapy. Head Neck. 2007;29(10):907-912.

(45.) Johnson CJ, Barry MB, Vasef MA, Deyoung BR. Her-2/neu expression in salivary duct carcinoma: an immunohistochemical and chromogenic in situ hybridization study. Appl Immunohistochem Mol Morphol. 2008;16(1):54 58.

(46.) Fan CY, Wang J, Barnes EL. Expression of androgen receptor and prostatic specific markers in salivary duct carcinoma: an immunohistochemical analysis of 13 cases and review of the literature. Am J Surg Pathol. 2000;24(4):579-586.

(47.) Williams MD, Roberts D, Blumenschein GR Jr, et al. Differential expression of hormonal and growth factor receptors in salivary duct carcinomas: biologic significance and potential role in therapeutic stratification of patients. Am J Surg Pathol. 2007;31(11):1645-1652.

(48.) Gnepp DR. Metastatic disease to the major salivary gland. In: Ellis, GL, Auclair PL, Gnepp DR, eds. Surgical Pathology of the Salivary Glands. Philadelphia, PA: WB Saunders; 1991:560-569.

(49.) Alos L, Cardesa A, Bombi JA, et al. Myoepithelial tumors of salivary glands: a clinicopathologic, immunohistochemical, ultrastructural, and flow cytometric study. Semin Diagn Pathol. 1996;13(2):138-147.

(50.) Barnes L, Appel BN, Perez H, El-Attar AM. Myoepithelioma of the head and neck: case report and review. J Surg Oncol. 1985;28(1):21-28.

(51.) Savera AT, Sloman A, Huvos AG, Klimstra DS. Myoepithelial caricnomaof the salivary glands: a clinicopathologic study of 25 patients. Am J Surg Pathol. 2000;24(6):761-774.

(52.) Nagao T, Sugano I, Ishida Y, et al. Salivary gland malignant myoepithelioma: a clinicopathologic and immunohistochemical study often cases. Cancer. 1998;83(7):1292-1299.

(53.) McHugh JB, Hoschar AP, Dvorakova M, Parwani AV, Barnes EL, Seethala RR. p63 immunohistochemistry differentiates salivary glands oncocytoma and ococytic carcinoma from metastatic renal cell carcinoma. Head Neck Pathol. 2007;1(2):123-131.

Jonathan B. McHugh, MD; Daniel W. Visscher, MD; E. Leon Barnes, MD

Accepted for publication June 3, 2009.

From the Department of Pathology, University of Michigan, Ann Arbor (Drs McHugh and Visscher);and the Department of Pathology, University of Pittsburgh, Pittsburgh, Pennsylvania (Dr Barnes).

The authors have no relevant financial interest in the products or companies described in this article.

Presented at New Frontiers in Pathology: An Update for Practicing Pathologists, University of Michigan, Ann Arbor, September 20, 2008.

Reprints: Jonathan B. McHugh, MD, Department of Pathology, University of Michigan, 1500 E Medical Center Dr, Room 2G332, Ann Arbor, MI 48109 (e-mail:
Table 1. Histologic Features Helpful in the Diagnosis
of Carcinoma Ex Pleomorphic Adenoma

1. Diffuse and marked nuclear pleomorphism and
2. Increased mitotic activity, especially atypical mitoses
3. Necrosis
4. Hyalinization of matrix (especially if calcified)
5. Capsular invasion
6. Perineural invasion
7. Lymphovascular invasion
8. Androgen receptor, p53, and HER-2/neu immunoreactivity (if
   strong and diffuse)

Table 2. Classification of Malignant Mixed Tumors

Carcinoma ex pleomorphic adenoma
   Noninvasive (in situ, intracapsular)
   Minimally invasive ([less than or equal to] 1.5 mm)
   Invasive (>1.5 mm)
Carcinosarcoma (true malignant mixed tumor)
"Benign" metastasizing pleomorphic adenoma

Table 3. Anatomic Distribution of 337 Polymorphous
Low-Grade Adenocarcinomas (a)

Site                         %

Palate                      57
Buccal mucosa               16
Lip                         13
Alveolar mucosa              4
Retromolar trigone           3
Tongue                       2
Floor of mouth               1
Parotid gland                1
Nasal cavity                 0.8
Nasopharynx                  0.3
Not otherwise specified      7

(a) Data derived from Vincent et al (16) and Castle et al. (17)

Table 4. Hormone-Staining Status in Salivary
Duct Carcinoma

Source, y                      Cases        ER         PR        AR

Dimery et al, (34) 1987           1         1          ND        ND
Delgado et al, (35) 1993         11         0          ND        ND
Barnes et al, (36) 1994          12         1           0        ND
Kay et al, (37) 2001             40         0           9        33
Ockner et al, (38) 1994          17         0           3        ND
Nasser et al, (39) 2003           6         0           0         6
Hoang et al, (40) 2001           28        ND          ND        20
Kapadia & Barnes, (41) 1998      12         0           0        11
Lewis et al, (26) 1996           26         0           1        ND
Williams et al, (47) 2007        84        57 (a)                56
Total, No. (%)                  237       2/125      13/113    126/170
                               (100%)    (2%) (a)    (12%)      (74%)

Abbreviations: AR, androgen receptor; ER, estrogen receptor;
ND, stain not performed in study; PR, progesterone receptor.

(a) The study by Williams et al47 used antibodies to the ER[beta]
receptor and these data are not included in the calculation for
total number with ER expression.

Table 5. HER-2/neu Expression and HER-2/neu Amplification in
Salivary Duct Carcinoma

                            Total        IHC
Source, y                   Cases   Expression (a)   Amplification (b)

Skalova et al, (42) 2003     11        8                 4/8
Cornolti et al, (43) 2007    13       10                 8/10
Nabili et al, (44) 2007       7        7                 3/7
Johnson et al, (45) 2008     12        4                 4/4
Total, No. (%)               43       29/43 (67)        19/29 (66)

Abbreviation: IHC, immunohistochemical.

(a) IHC expression with 3+ membranous staining.

(b) Amplification detected by fluorescence in situ hybridization
(ISH) or chromogenic ISH (Johnson et al 45).
COPYRIGHT 2009 College of American Pathologists
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2009 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:McHugh, Jonathan B.; Visscher, Daniel W.; Barnes, E. Leon
Publication:Archives of Pathology & Laboratory Medicine
Article Type:Report
Date:Nov 1, 2009
Previous Article:New frontiers in pathology.
Next Article:Ovarian cancer update: lessons from morphology, molecules, and mice.

Terms of use | Privacy policy | Copyright © 2019 Farlex, Inc. | Feedback | For webmasters