Immunohistochemistry (IHC) has become increasingly important in the evaluation of pathologic conditions in the genitourinary (GU) organs. In addition to careful evaluation of hematoxylin-eosin sections and generation of a differential diagnosis, choosing the optimal panel of IHC markers becomes even more important when the biopsy material is very limited. The following summary of our experience supplemented with relevant literature review exemplifies how to use IHC to facilitate pathologic diagnosis in the GU system.
To describe our experience with the most common immunohistochemical markers used in GU pathology.
Institutional experience and literature search comprise our data sources.
Application of IHC provides enormous benefits to the interpretation of GU pathologic conditions, including benign and malignant lesions. However, both insufficient and excessive types of use of IHC, as well as incorrect interpretations in common and rare GU conditions, could present pitfalls in diagnosis.
Prostatic tissue represents the most common type of specimens in genitourinary pathology and is often examined for the presence and characterization of prostatic adenocarcinoma. The immunohistochemical markers relevant to prostate are summarized in Table 1; these include NK3 Homeobox 1 (NKX3.1), prostate-specific antigen (PSA), prostatic acid phosphatase (PSAP), prostate-specific membrane antigen (PSMA), AMACR (α-methylacyl-CoA racemase, also known as P504S), prostein (also known as P501S), high–molecular weight cytokeratin (HMWCK), and p63.
Usually, prostatic adenocarcinoma displays the phenotype of secretory cells and is positive for PSA, PSAP, and PSMA, and negative for basal cell markers, such as HMWCK and p63. NKX3.1, a marker for both secretory and basal cells, has emerged as the most sensitive and specific marker for prostatic epithelial lesions.1,2
The most common pitfall is that high-grade prostatic adenocarcinoma, particularly treated prostatic adenocarcinoma, often loses expression of some prostatic markers such as PSA, or expresses these markers weakly and focally (Figure 1). NKX3.1, as the most sensitive and most specific prostate marker, is generally the last marker to be lost in high-grade prostatic adenocarcinoma and is therefore highly recommended for determining the prostatic origin. P501S, also known as prostein or Solute Carrier Family 45, Member 3 (SLC45A3), is a protein containing transmembrane domains but showing punctate membranous staining or granular/perinuclear dot in malignant and benign prostatic secretory cells3,4 and is not affected by grade. It has been most useful to demonstrate prostatic origin, with a good sensitivity and excellent specificity by itself5–7 or in combination with other markers, including PSA.4 However, just like the rest of the prostatic markers, p501S is androgen dependent,7 so an antiandrogen-treated case is likely to have weaker expression. Particularly, in our practice we have found that nuclear NKX3.1 is more robust and reliable in decalcified or crushed tissue than cytoplasmic markers.
AMACR is another marker used in confirmation of prostatic adenocarcinoma. It is more strongly expressed in neoplastic secretory cells, including high-grade prostatic intraepithelial neoplasia (HGPIN) and adenocarcinoma, than in benign secretory cells (Figure 1, B). AMACR has been the best marker for confirmation of prostatic adenocarcinoma on the prostate biopsy. However, outside the prostate, AMACR is not prostate specific because other carcinomas, such as colonic adenocarcinoma or papillary renal cell carcinoma (RCC), may express AMACR at high levels.
A nuclear marker ERG (ETS avian erythroblastosis virus E26 oncogene homolog) has recently generated interest. Its expression depends on the TMPRSS2-ERG (transmembrane protease serine 2:v-ETS avian erythroblastosis virus E26 oncogene homolog) fusion status, which ranges from 24.4% to 49% sensitivity depending on population studied.8–10 It is highly specific for prostatic adenocarcinoma and is helpful when positive in a tumor suspected of being metastatic prostatic adenocarcinoma. However, ERG has low sensitivity and is more commonly seen in grade group 1 and 2 cancers,8,9,11–13 so when metastatic prostatic adenocarcinoma is considered, or when the tumor is high grade, ERG is more likely to be negative. Also, an important caveat to remember is that endothelial cells, including vascular tumors, are also positive for ERG. Finally, CK7 and CK20 have no utility for confirmation of prostatic origin in our practice because either, both, or neither may be positive.14
Another common pitfall occurs with overuse of the triple-stain cocktail for prostate biopsies, allowing benign prostatic conditions to be confused in the diagnosis of prostatic adenocarcinoma. A subset of adenosis may display moderate AMACR staining (Figure 2, A and B), and partial atrophy may also have weak AMACR staining (Figure 2, C and D). Both adenosis and partial atrophy will display discontinuous or patchy basal cell staining, which is different from the absence of basal cells in prostatic adenocarcinoma.15,16 It is important to consider that prostatic adenocarcinoma will have cytologic and architectural atypia, whereas adenosis and partial atrophy will lack significant cytologic atypia.
Rare cases of prostatic adenocarcinoma expressing p63 (Figure 3, A and B) can present a diagnostic pitfall.17 In such cases the morphologic features are the key for diagnosis, because prostatic adenocarcinoma will have sufficient cytologic atypia and invasive features. Finally, basal cell carcinoma (Figure 3, C), a rare type of prostatic carcinoma, can be challenging for diagnosis.18,19 This tumor is composed of malignant cells with basal cell appearance, forming solid nests or multicell-layered glandular structures that display infiltrating features. This tumor is also positive for HMWCK, p63 (Figure 3, C and D), and NKX3.1, but in haphazard infiltrating patterns, including perineural invasion and/or extraprostatic extension, different from the expected benign basal cell hyperplasia.
Another pitfall happens in patients with a history of prostatic adenocarcinoma who present with new or systemic lesions suspicious for recurrence or metastases. These patients often have a history of therapies, including androgen deprivation therapy or radiation, resulting in morphologic alterations in tumor cells (Figure 4). Such recurrent or metastatic prostatic adenocarcinoma may lose some of the expected prostate markers, or it may express p63 or other markers, such as neuroendocrine differentiation. In combination with ambiguous CK7 and CK20 expression, a pathologist unaware of the history of prostate cancer may not consider prostate as a possible origin. However, it is prudent to do an expanded panel of prostatic markers, including PSA, PSMA, and particularly NKX3.1, because it is generally the last marker to be lost in prostatic adenocarcinoma. Of course, markers of the organ that is biopsied also should be performed, including CDX2 (Caudal-type Homeobox Transcription Factor 2) for colon, GATA3 (Gata-binding protein 3) for urothelial and breast, TTF1 (Transcription Termination Factor, RNA Polymerase I) for lung, and PAX8 (Paired Box Gene 8) for renal origins. In a setting of liver lesions, use of HepPar1 (hepatocyte paraffin 1) by itself is insufficient, because some prostatic adenocarcinoma will express HepPar1.20,21 In this setting, other hepatic markers should be added.22
BLADDER
Urothelial carcinoma (UC), as the most common histologic type of bladder cancer, may display a wide range of histologic patterns, such as glandular, squamous, or micropapillary features. In addition, bladder is the most common site in the genitourinary system for involvement by secondary malignancies, such as colon and prostate. Therefore, neoplasms of the bladder may present diagnostic difficulty and may require immunohistochemical workup. In addition to prostatic markers described above, a usual panel of markers for the bladder may include GATA3, p63, CK20, S100P (S100 Calcium-Binding Protein p), and CK5/623 (Table 2). These markers should be positive in UC, although none of them are urothelium specific and none are 100% sensitive either. We often see high-grade UC showing patchy or even complete loss of CK20 or p63. Variants of UCs, including those with micropapillary component, or glandular differentiation may have morphologic appearance similar to carcinomas from other organs. These cases may also lose expression of CK20 and p63; however, most of these cases should retain the expression of GATA3 (Figure 5)24 although it may be weak and patchy. Finally, less common markers, such as Uroplakin III and thrombomodulin, have shown limited utility because of their poor sensitivity in high-grade tumors.23
Primary bladder adenocarcinoma may present a diagnostic pitfall. Normally, when an adenocarcinoma is encountered in the bladder, metastases or direct extension from other organs, especially from the colon, prostate in a male, or gynecologic tract in a female, must be excluded first. In bladder, these secondary adenocarcinomas are much more common than a primary adenocarcinoma. Typically, prostatic adenocarcinoma will express prostatic markers, such as PSA and NKX3.1, and gynecologic carcinomas will express various combinations of PAX8 (Paired Box Gene 8), WT1 (Wilms Tumor 1), ER/PR (estrogen and progesterone receptors), vimentin, or p16. Colonic adenocarcinoma remains the main challenge in a differential diagnosis. Primary enteric adenocarcinoma of the bladder (Figure 6) displays an immunoprofile similar to that of colonic adenocarcinoma: CDX2+, CK20+, and CK7−. In this situation, we found β-catenin useful: colonic adenocarcinoma should have strong nuclear staining for β-catenin, whereas most primary bladder adenocarcinomas have no β-catenin nuclear staining (Figure 6, D).25–28 Cytoplasmic or membranous expression of β-catenin in bladder adenocarcinomas should not be interpreted as positive. In rare cases of primary bladder adenocarcinomas with nuclear β-catenin expression, intestinal origin must be excluded clinically.
Rare benign entities of the bladder can mimic malignancy. Viral infections can have significant cytologic atypia that may mimic malignancy (Figure 7, A and B). Usually, if an immunohistochemical panel containing p53, CK20, CD44s, Ki-67, and/or p16 is performed, these cells may express strong p53 and demonstrate strong Ki-67 proliferative activity (Figure 7, C), mimicking UC in situ. However, a closer look will demonstrate a homogeneous glassy texture to chromatin. Such cells are sometimes seen on cytology and are often called “decoy” cells, but they are rarely seen in biopsy material. In cases of BK virus infection, immunohistochemistry with the antibody specific for SV40 (polyomavirus simian virus 40) will cross-react with BK viral protein and will highlight the atypical infected urothelial cells (Figure 7, D). However, caution should still be exercised because BK virus has been reported to be associated with increased risk of UC.29,30
Evaluation of muscularis propria presence and/or involvement by the tumor in biopsy or transurethral resection specimens is important for clinical management. Sometimes it can be difficult in cases with crush artifact or in patients with altered muscle. Of all smooth muscle markers, smoothelin has shown the most utility. Although markers such as smooth muscle myosin or caldesmon may be useful to demonstrate size of the muscle bundles, smoothelin shows differential staining under the correct conditions: it displays stronger immunoreactivity in muscularis propria than in muscularis mucosae or muscular wall of the arteries. A problem arises when the 2 layers are not present in the same piece or specimen for comparison, or when extensive cautery is present. However, small muscular arteries are nearly always available for evaluation and can be used as a benchmark for strength of smoothelin expression. A muscle bundle expressing stronger smoothelin than background arteriole is usually muscularis propria, whereas a muscle bundle with weaker or equivalent strength of staining is lamina propria (muscularis mucosae; Figure 8, A). This finding in combination with muscle bundle size of 100 to 200 microns can improve accuracy of muscularis propria evaluation (Figure 8, B through E) for determination of UC invading muscularis propria, which is often the indicator for radical cystectomy.
KIDNEY
Renal cell neoplasms may present a diagnostic pitfall when they are discovered at high stage and high grade or present first as a metastatic focus of unknown primary. Careful histologic evaluation of hematoxylin-eosin slides followed by application of reliable markers as well as attention to the overall immunostaining pattern are needed to arrive at the correct diagnosis, providing key diagnostic information for choosing most appropriate treatment modality for the patient.
The 4 most common types of RCC are clear cell RCC, papillary RCC, chromophobe RCC, and recently recognized clear cell papillary RCC. Immunohistochemical workup of renal cell neoplasms has been thoroughly reviewed previously31 ; however, new entities have emerged in recent years.32,33 These new tumors are rare and can be difficult to diagnose without special immunohistochemical and/or molecular studies. Commonly used immunohistochemical markers for working out kidney tumors are summarized in Table 3. Most notable examples include MiTF (Microphthalmia-associated transcription factor) translocation family of carcinomas with TFE3 (Transcription factor for immunoglobulin heavy-chain enhancer 3) or TFEB (T-cell transcription factor EB) translocation.
Briefly, clear cell RCC should be PAX8 positive, CA-IX (carbonic anhydrase IX) positive, and CK7 negative. AMACR can be positive in a subset of clear cell RCC, but its pattern of expression is often weak and focal. In contrast, papillary RCCs are positive for PAX8, AMACR, and CK7, and are negative for CA-IX (Table 3), although high-grade type 2 papillary RCCs often show focal CK7 immunoreactivity. When papillary RCCs grow to large size, they are prone to ischemia, resulting in large areas of necrosis. The cells at the edges of such areas experience partial ischemia, and as a result of normal biologic cell signaling they express CA-IX. This expression should not be interpreted as that of clear cell RCC. Conversely, clear cell RCC may express CK7 or AMACR in a focal or patchy manner in some cases. Correct diagnosis requires consideration of overall immunoprofile and depends on thorough sampling. Utility of CD10 in diagnosis of primary RCCs is limited because it will be positive in most major subtypes of RCC. Likewise, use of CD10 in a metastatic setting should also be limited because of its poor specificity. For example, CD10 is positive in many renal neoplasms in a strong membranous or cytoplasmic pattern that resembles canalicular pattern of hepatocellular carcinoma.
Usually, renal angiomyolipoma can be easily diagnosed based on typical morphology of a variable proportion of adipose tissue, spindle or epithelioid smooth muscle–like cells, and abnormal thick-walled blood vessels. Sometimes, immunohistochemical staining may be necessary to confirm the diagnosis by its positivity for one or more melanocytic markers: HMB-45, Melan-A, MiTF, tyrosinase, and one or more actins (smooth muscle actin or muscle-specific actin). However, a pitfall may arise when epithelioid variant of angiomyolipoma is encountered. Epithelioid angiomyolipoma had 2 major architectural patterns: carcinoma-like growth pattern, which is characterized by large cells arranged as broad alveoli, nests, and sheets with thin vascular-rich septae, and diffuse growth pattern, composed of epithelioid to plump spindled cells arranged in diffuse sheets. Multinucleated giant cells and melanin pigment can be present. Necrosis and microabscesses are occasionally seen. Five clinicopathologic features were associated with aggressiveness of the pure epithelioid angiomyolipoma: (1) the presence of tuberous sclerosis complex; (2) tumor size greater than 7 cm; (3) carcinoma-like growth; (4) involvement of perinephric soft tissue and/or kidney; and (5) presence of necrosis (Figure 9). Therefore, it is very important to provide these key pieces of diagnostic information to clinicians for choosing the most successful treatment modality for the patient. Cathepsin K is a cysteine proteinase that shows strong cytoplasmic staining in renal angiomyolipomas, including epithelioid, oncocytoma-like, and epithelial cystic variants, which may be variably negative for melanocytic or muscle markers.34 This marker also demonstrated good expression in tumors of perivascular epithelial cells (PEComas) from other locations of the body.35 Caution should be exercised when a differential diagnosis includes a translocation-associated RCC because it would express cathepsin K.36 In such cases, morphologic features and an expanded muscle and renal marker panel could be useful.
Renal medullary carcinoma is a rare aggressive tumor affecting young patients (second and third decades of life) with sickle cell trait and rarely with sickle cell disease. Although not associated with other hemoglobinopathies,37 a very rare provisional entity called renal cell carcinoma, unclassified, with medullary phenotype (RCCU-MP) has recently been reported to show immunohistochemical loss of SMARCB1 (Swi/SNF-Related, Matrix-Associated, Actin-Dependent Regulator of Chromatin, Subfamily B, Member 1).38
Morphologically, renal medullary carcinoma presents as an infiltrative tumor with pleomorphic nuclei, eosinophilic cytoplasm, and a variety of growth patterns with desmoplastic stroma. This appearance is similar to that of a collecting duct carcinoma, and for a time these 2 entities were considered by some authors to be related. Sickling renal cells can often be observed inside renal medullary carcinoma. Immunohistochemical profile of RMS is not specific, but it is usually positive for PAX8, CK7, and OCT4 (Octamer-Binding Transcription Factor 4)39 and may express CK20 and vimentin.40 The key to diagnosis is immunohistochemical loss of SMARCB1, more commonly known as INI1 (Integrase Interactor 1), which is a chromatin remodeling regulator. It is important to remember that in addition to renal medullary carcinomas, other renal tumors, such as pediatric rhabdoid tumors,41 and epithelioid sarcomas,37 even in collecting duct carcinoma, may show immunohistochemical loss of SMARCB1.42,43 The genetic alterations underlying the loss of expression are varied.44 Most of the other RCCs are reported to retain SMARCB1, including those with rhabdoid change.45
Another pitfall may arise when MiTF family translocation RCCs are encountered. Accurate diagnosis of such tumors is important because these tumors in adults usually have aggressive behavior and do not respond to usual chemotherapy designed for clear cell RCC. The first cases of such carcinomas were found in young patients. Now, more and more translocation RCCs are found in adults, even in older patients of 70 or 80 years old. Different from its pediatric counterpart, the adult translocation RCCs are often high grade and high stage at presentation (Figure 10, A). In contrast to the usual RCCs, translocation RCCs have multiple relatively unique morphologic patterns within the same tumor, including clear cytoplasm in papillary configuration, high nuclear grade, psammoma bodies, and sometimes rosette formation. This type of RCC is usually positive for PAX8 but often negative for CA-IX and/or AMACR, different from conventional clear cell RCC or papillary RCC. Furthermore, a subset of these translocation carcinomas is positive for Melan-A or HMB-45, which is not typical of other types of RCC,33 although a rare clear cell RCC has been reported to express weak Melan-A.46 At this point, immunohistochemistry for TFE3 is warranted (Figure 10, B through E). It is important to point out, however, the TFE3 antibody is technically challenging to work with and may demonstrate high background staining and a poor specificity, which is evident when benign renal tubules and glomeruli are falsely positive for TFE3 staining. Fluorescence in situ hybridization or molecular analysis for translocations may be the most appropriate next step in questionable cases.
When a renal neoplasm is centrally located, UC comes into the differential diagnosis. Urothelial carcinoma can have extensive clear cell change, whereas RCCs often have papillary architecture, mimicking UC. Commonly, PAX8 is used to demonstrate renal origin. However, a diagnostic pitfall is that some upper tract UCs express weak to moderate PAX8, and PAX8 may also be expressed in sarcomatoid variants of both renal and urothelial tumors.47 This may cause diagnostic difficulty when limited biopsy material is available, or when a metastatic focus is biopsied. In such circumstances, it is crucial to do an expanded panel of PAX8, p63, GATA3, and S100P, as well as a full RCC panel to definitively identify the nature of the neoplasm, because surgical and medical treatments for these 2 tumors are significantly different.
Finally, more and more newly defined subtypes of RCCs are emerging. Some have relatively specific morphologic characteristics and defined immunohistochemical profiles, such as tubulocystic RCC, acquired cystic disease–associated RCC, succinate dehydrogenase–deficient RCC, hereditary leiomyomatosis–associated RCC, and RCC syndrome–associated RCC. It is important to be aware of these tumors and confirm with appropriate immunohistochemistry, fluorescence in situ hybridization, or molecular studies.
Another recently described subtype is eosinophilic solid cystic RCC, which is characterized by tumor cells with prominent eosinophilic cytoplasm, and prominent and characteristic cytoplasmic stippling (Figure 11, A and B). The tumor cells form solid mixed with cystic patterns. The tumor cells are positive for PAX8 and CK20, focally positive for AMACR (Figure 11, C and D), and negative for CD117 and CK7, which is significantly different from an eosinophilic papillary RCC or a renal oncocytoma.
TESTIS
Testicular neoplasms that present the most diagnostic concern are germ cell tumors. In combination with morphology and serologies, immunohistochemistry has become important in delineating the nature and relative amounts of various components of these neoplasms.
Most of the markers used in the evaluation of germ cell neoplasms of the testis have some degree of overlap (Table 4). Seminoma and germ cell neoplasia in situ are generally positive for CD117, and most germ cell neoplasms will express OCT4. Embryonal carcinomas are positive for CD30 and OCT4. Both seminoma and embryonal carcinoma can be positive for D2-40, but they show different patterns: seminoma exhibits diffuse membranous D2-40 staining,48 whereas embryonal carcinomas display focal luminal D2-40 staining (Figure 12). Yolk sac component is positive for AFP and glypican 3,49 although glypican 3 is also expressed in trophoblasts50,51 and choriocarcinoma, which is also positive for β-hCG (beta subunit of human chorionic gonadotropin; Figure 13, A and B). It is important to remember that a small subset of seminoma may either secrete β-hCG directly or contain scattered syncytiotrophoblasts secreting β-hCG, which should not be interpreted as choriocarcinoma component.
SALL4 (SAL-Like 4) is another marker for germ cell tumors (Figure 13, C and D). It has been reported to mark seminomas, most embryonal carcinomas, yolk sac tumor, and teratomas52,53 ; it is rarely expressed in non–germ cell neoplasms.54 In our practice, we have found that SALL4 is a good marker to determine the germ cell origin, but it lacks specificity in classification of germ cell subtypes.
Finally, it is always important to remember that metastatic carcinoma, such as prostatic adenocarcinoma, or hematologic neoplasms become more common in older men. If a testicular neoplasm is found in an older man which is not showing an expected immunophenotype, a panel of hematopoietic markers, such as CD45, CD20, and CD3, is warranted. A pitfall may arise if a single marker, such as CD117 in a case of myeloid sarcoma or α-keratin in the case of embryonal carcinoma, is used—the tumor cells will be positive; however, without other hematologic markers, such a hematopoietic tumor may be misdiagnosed as a seminoma or embryonal carcinoma.
CONCLUSIONS
Immunohistochemistry has become an indispensable tool in pathologic diagnosis and classification of pathologic conditions in genitourinary organs. Choosing the appropriate panel of markers and correct interpretation of immunostaining pattern and results are the key to a more accurate diagnosis.
References
Author notes
The authors have no relevant financial interest in the products or companies described in this article.
Presented in part at the 5th Princeton Integrated Pathology Symposium; April 15, 2018; Plainsboro, New Jersey.