Primary salivary gland oncocytic lesions are traditionally classified as diffuse oncocytosis, nodular oncocytic hyperplasia (nodular oncocytosis), oncocytoma (oncocytic adenoma) and oncocytic carcinoma [3
]. These are uncommon and, according to data from the Armed Forces Institute of Pathology (AFIP), oncocytomas accounted for only 1.4% of 13,749 salivary gland tumors on file [10
]. Among this same group, only 27 cases of oncocytosis were identified. These data are in accordance with those from other groups who found the incidence of oncocytoma to be in the range of 0.1–1.0% [2
Approximately 85–90% occur in the parotid gland [2
] and, accordingly, oncocytic tumors account for a relatively larger percentage of parotid gland tumors. At the AFIP, oncocytomas accounted for about 3.5% of parotid gland tumors [3
]. Similarly, Capone et al. found 21 oncocytic tumors among 561 parotidectomy specimens (3.7%) at The Johns Hopkins University Hospital, which included 13 (2.3%) oncocytomas, 6 (1.1%) cases of oncocytosis and only 2 (0.4%) oncocytic carcinomas [11
While the diagnosis of these lesions is usually straightforward, the histologic distinction between nodular oncocytosis and oncocytoma is admittedly rather arbitrary in certain cases. Many pathologists believe the presence of a single, well-circumscribed and at least partially encapsulated nodule favors the diagnosis of oncocytoma while multiple, unencapsulated nodules distributed in a lobular configuration favors nodular oncocytosis [2
]. Even more confusing is the designation of oncocytoma arising in oncocytosis to describe a dominant often encapsulated nodule in the background of oncocytosis. This distinction, however, is academic and of little to no clinical or prognostic significance.
More important is the distinction of these lesions from their malignant counterparts, oncocytic carcinomas, as well as the oncocytic variants of other salivary gland carcinomas and metastatic lesions with oncocytic morphology. Oncocytic salivary gland carcinoma is uncommon representing only 0.05–0.4% of salivary gland neoplasms and about 5% of oncocytic neoplasms [11
]. Similar to their benign counterparts, nearly 80% occur in the parotid gland. Interestingly, the majority is presumed to arise in a pre-existing oncocytoma but they also may occur de novo [14
]. Diagnostic criteria for salivary gland oncocytic carcinoma include destructive invasion of adjacent salivary or non-salivary tissue, perineural and/or vascular invasion, and metastases.
Renal cell carcinoma metastatic to the major salivary glands is also notoriously difficult to separate histologically from benign and malignant oncocytic salivary gland tumors due in large part to significant morphologic overlap. Furthermore, about 8% of RCC patients initially present with head and neck metastases prior to identification of the primary kidney lesion [15
]. In addition, pathologists are occasionally not given complete historical information regarding prior diagnoses and/or metastatic RCC may not present until many years after nephrectomy. Pathologists increasingly need to maintain a high index of suspicion for the possibility of metastasis when confronted with oncocytic or clear cell salivary gland neoplasms.
Following lung and breast carcinoma, RCC is the most common infraclavicular malignancy to metastasize to the head and neck [16
] with some studies placing it second after lung [10
]. Approximately 15% of patients with RCC will develop non-CNS head and neck metastases [17
] with the thyroid gland being the most common site [18
]. In 1% of patients with RCC, the only documented site of metastasis is to the head and neck [15
Although the incidence varies based on the study evaluated, metastases account for about 3–25% of parotid and submandibular malignancies with the average reported in the literature around 16% [10
]. The vast majority are head and neck squamous cell carcinomas followed in frequency by cutaneous and mucosal head and neck melanomas [19
]. These two account for about 80% of major salivary gland metastases. In contrast, metastases from more distant sites account for only 2–4% of major salivary gland malignancies [20
] and, similar to other head and neck sites, usually are from breast, lung and kidney primaries. In a study of 108 metastases identified from 11,000 salivary gland pathology specimens (9.8%), Seifert et al. found that 6 (5.6%) were metastatic RCC [22
]. Therefore, both locoregional and distant metastatic disease should always be considered in the workup of salivary gland tumors that are not easy to classify histologically. Specifically, metastatic RCC should be considered in tumors morphologically resembling oncocytoma, oncocytosis or oncocytic carcinoma.
While clinical history and histology are typically adequate in diagnosing metastatic RCC, for reasons mentioned above and below, sometimes distinguishing this from primary oncocytic lesions can be challenging. Metastatic RCC typically demonstrates clear cells with more cellularity and cytologic pleomorphism than benign oncocytic lesions and often has a characteristic prominent vascular stroma that is not usually seen in salivary gland oncocytic lesions [5
]. Metastatic RCC can grow as solid sheets, nests and/or trabeculae and may also contain lumina with or without blood (i.e. blood lakes). However, metastatic RCC can exhibit bland cytologic features among a well-circumscribed nodule with or without the characteristic vascular stroma and may contain cells with oncocytic cytoplasm. Furthermore, primary salivary gland oncocytic lesions, with or without clear cell change, usually demonstrate lumina and, in our experience, occasionally have a prominent vascular stroma reminiscent of that seen in RCC (i.e. RCC-like). Thus, the diagnostic morphologic features of these two tumors can overlap significantly and, especially without a complete clinical history, they may be impossible to separate on routine hematoxylin and eosin-stained sections.
Our findings confirm the significant morphologic overlap frequently noted between salivary gland oncocytic tumors and metastatic RCC. While most RCC are of the conventional (clear cell) type, in the current study, 37% showed a predominance of oncocytic cells. Similarly, while oncocytic tumors typically are predominantly eosinophilic, in this series, 26% of oncocytomas and 33% of oncocytoses were predominantly composed of clear cells while 1 (11%) oncocytic carcinoma showed the presence of only focal clear cells. While Fuhrman nuclear grade was significantly different between the benign and malignant lesions evaluated, there was overlap. Benign oncocytic tumors had significantly lower mitotic rates than metastatic RCC suggesting this feature may be helpful in discriminating these tumors. However, there was no difference benign and malignant oncocytic tumors or between oncocytic carcinoma and metastatic RCC.
Various architectural features also showed overlap among the different tumor types. While an infiltrative growth pattern was unique to oncocytic carcinoma and metastatic RCC, a circumscribed growth pattern (88% vs. 14%) or multinodular growth pattern (12% vs. 50%) could be seen in both oncocytomas and metastatic RCC, respectively, limiting the utility of growth pattern in the differential. The presence of lumina (or pseudolumina) was demonstrated in all primary salivary gland oncocytic tumors and in half of metastatic RCC, however, blood lakes were only identified in cases of metastatic RCC. Therefore when lumina/pseudolumina are accompanied by blood lakes, metastatic RCC should be suspected. Finally, while RCC notoriously demonstrates a unique prominent stromal vascularity, identified in all 16 cases in the current study, a morphologically similar stroma can be identified in salivary gland oncocytic tumors. In fact, we identified this RCC-like vascular stroma in 26, 22, and 56% of oncocytomas, oncocytoses and oncocytic carcinomas, respectively.
For these reasons, immunohistochemical stains are often used to help sort out this differential diagnosis. However, there is only one study comparing the immunohistochemical profile of metastatic RCC to that of salivary gland oncocytomas [9
] and there are no studies comparing them to salivary gland oncocytic carcinomas. Ozolek et al. compared the expression of cytokeratin 7 (CK7), cytokeratin 20 (CK20), epithelial membrane antigen (EMA), vimentin, CD10 and RCCm between ten oncocytomas and ten RCCs metastatic to various head and neck sites (including one parotid gland tumor).
In that study, CD10 was identified as the best single marker to aid in this differential diagnosis, being diffusely and strongly positive in 90% of metastatic RCC and in none of the oncocytomas. In a larger group of tumors, we were unable to confirm this finding. Specifically, 11/15 (73%) metastatic RCCs demonstrated strong diffuse membranous and cytoplasmic positivity for CD10. However, 10/27 (37%) benign oncocytic tumors and 4/8 (50%) oncocytic carcinomas showed strong membranous and cytoplasmic positivity, but the staining was focal in all of these cases. Therefore, in the current study, CD10 yielded a sensitivity of 73% but a specificity of only 60% for RCC when considering positive versus negative tumor staining in salivary gland oncocytic tumors and metastatic RCC.
Vimentin and EMA were not helpful markers in separating these two tumor groups in the Ozolek study. Our experience with vimentin was similar to their findings in that 73% of metastatic RCC (vs. 70% in their study) were positive while 19% of oncocytomas (vs. 40% in their study) were positive. Furthermore, 38% of oncocytic carcinomas were positive limiting the utility of this marker in differentiating it from both oncocytomas and metastatic RCC. In both studies, RCCm was negative in all salivary gland oncocytic lesions, however, only 44% of metastatic RCC were positive in the current study, similar to the 40% found in their study, limiting the utility of this marker based on the low sensitivity.
In the current study, immunohistochemical staining for p63 proved to be the most reliable marker for differentiating salivary gland oncocytic tumors from metastatic RCC. The p63
gene is a recently recognized member of the p53
tumor suppressor gene family [23
]. It has been shown that p63
plays a major role in epithelial development [24
] and is an integral gene whose protein is expressed in basal and myoepithelial cells in various tissues including salivary glands [26
]. The gene gives rise to six different major transcripts that segregate into two functional protein classes. Three isoforms (referred to as TAp63
) function like p53 as inducers of apoptosis, a function mediated by a transactivating (N)-terminal domain. The other three (referred to as ΔNp63
) lack the (N)-terminal transactivation domain and therefore inhibit p53 activity. It is this latter group of isoforms that is expressed in basal cells of multilayered epithelia and in myoepithelial cells [28
]. The major theory for its function in these epithelia is to maintain the proliferative capacity of these important progenitor cells.
Few studies have evaluated the expression of p63 in salivary gland oncocytic tumors. Using TMA methodology, Weber et al. found rare positive cells in five of five oncocytomas with the positive cells being distributed in a basal cell-type distribution [29
]. Similarly, Bilal et al. demonstrated scattered positive cells in four oncocytomas with the positive cells being present at the tumor-stromal interface in a basal cell-type distribution [30
]. Finally, Foschini et al. demonstrated positive p63 staining in one oncocytoma studied [31
] while Seethala et al. found no p63 staining in the evaluation of one oncocytoma [32
All benign oncocytic tumors and oncocytic carcinomas in the current study demonstrated diffuse p63 nuclear-positive cells distributed in a basal cell-type pattern with positive cells being located predominantly towards the periphery of the tumor cell nests. Some tumor nests showed extension of p63 reactivity towards the center of tumor nests, but with decreased intensity. This staining pattern was in distinct contrast to metastatic RCC in which no tumor stained positive for p63. Therefore, p63 immunohistochemical staining, in our experience, demonstrates 100% sensitivity and 100% specificity in distinguishing both primary benign and malignant salivary gland oncocytic tumors from metastatic RCC. Furthermore, the expression of p63 in oncocytic carcinomas was maintained in the invasive component in tumors arising in an underlying oncocytoma.
The exact nature of these p63-positive cells is not clear but they may represent basal cells/tumor stem cells with oncocytic cytoplasm. They do not appear to be myoepithelial cells as they are typically negative on immunohistochemical staining for other myoepithelial markers in our experience (data not shown) as well as that of other investigators [33
]. While the majority of p63 positive cells were present at the periphery of tumor nests, some foci did show a diminishing gradient of staining towards the center of the nests. This supports a maturation of phenotype with gradual loss of p63 in primary oncocytic lesions, rather than a truly biphenotypic differentiation profile seen in tumors such as pleomorphic adenomas, or epithelial-myoepithelial carcinomas.
To our knowledge, p63 expression in salivary gland oncocytic carcinomas has not been previously reported. Our findings are the first to demonstrate the frequency and pattern of staining of this marker in this rare malignant salivary gland tumor. The utility of immunohistochemical p63 staining in differentiating between oncocytic carcinoma and metastatic RCC has been demonstrated herein, however, its use in separating it from other malignant oncocytic salivary gland tumors, such as oncocytic mucoepidermoid carcinoma, remains to be proven. Similarly, p63 expression in primary renal tumors has not been studied extensively. In accordance with our findings, Langner et al., using a TMA, found this marker to be negative in all 188 RCC, which included 133 conventional (clear cell) subtypes [34
]. Two smaller studies evaluating p63 expression found no staining in four conventional (clear cell) RCC [35
] and in one of thirteen RCC [36
], however, the histologic types were not provided in the later study. We additionally demonstrate that p63 is negative in all 60 renal oncocytomas and 12 chromophobe renal cell carcinomas tested on our TMA. While this finding is not of diagnostic importance, since essentially all RCCs that have metastasized above the clavicle are of the conventional type, it is of biologic significance in that it demonstrates that p63 reactivity is more closely linked to the tissue of origin rather than the oncocytic phenotype.
In summary, primary oncocytic salivary gland lesions and metastatic RCC show considerable morphologic overlap. In most cases, a careful consideration of morphologic features along with a clinical history is sufficient to separate these tumors. We have shown that immunohistochemical p63 positivity effectively distinguishes primary salivary oncocytic lesions from metastatic RCC in problematic cases with 100% sensitivity and specificity.