The Application of Immunohistochemical Biomarkers in Urologic Surgical Pathology

Most renal tumors are epithelial neoplasms; clear cell renal cell carcinoma (CRCC), papillary renal cell carcinoma (PRCC), chromophobe renal cell carcinoma (ChRCC), and oncocytoma account for more than 90% of this group of tumors.^1–3  Other uncommon and rare renal epithelial neoplasms include clear cell papillary renal cell carcinoma (CPRCC), mucinous tubular and spindle cell carcinoma, renal medullary carcinoma, collecting duct carcinoma (CDC), tubulocystic renal cell carcinoma, and Xp11 translocation renal cell carcinoma (TRCC).^1–3  Upper urinary tract urothelial carcinoma (UUC) is usually included in the diagnostic consideration. Diagnosis of a renal epithelial neoplasm tends to be straightforward, based on histologic features alone, especially in a nephrectomy specimen. However, histologic subtyping of these renal epithelial tumors and the distinction of an epithelial tumor from a nonepithelial tumor or a metastasis in a limited sample, such as a needle core biopsy or a fine-needle aspiration specimen, can be challenging. In these scenarios, IHC stains may play a crucial role in reaching a definitive diagnosis.

When working on a renal biopsy specimen obtained from a renal mass, in our opinion the following questions should be raised and appropriate diagnostic approaches should be taken: (1) Is this a low-grade renal epithelial neoplasm or a reactive condition, such as reactive tubules in xanthogranulomatous pyelonephritis, interstitial nephritis due to viral infection, malakoplakia, or other benign condition? (2) If this is a renal epithelial tumor, how should one further subtype this tumor? and (3) If this is a neoplastic process, can one exclude a metastasis or a nonepithelial neoplasm? In this portion of the article, we will review and update the most useful immunomarkers and IHC panels, on the basis of the literature and our experience, in attempting to resolve these diagnostic conundrums, with a specific focus on how to classify a renal epithelial neoplasm and how to avoid overutilization or underutilization of the massive number of biomarkers. It should be emphasized here that there are overwhelming numbers of biomarkers reported in the literature, and investigators appear to have their own preference and experience using certain sets of biomarkers.^4–65  As a result, the International Society of Urological Pathology⁴  reached no absolute consensus regarding an individual antibody or panel of antibodies to be used for classifying renal tumors or confirming renal metastases.

The first question to ask when evaluating a biopsy sample from a renal mass is whether it represents a benign or malignant process. Particularly, is this a low-grade renal epithelial neoplasm or a reactive condition? As a general rule, histologic features such as clear cells, papillary structures, and vascular-rich tissues are clues to suggest a neoplastic process. In contrast, mixed acute and chronic inflammation, atrophic tubules and glomeruli, fibrosis, and granulomas are suggestive of a benign/reactive process. Numerous renal epithelial tumor–associated immunomarkers have been reported in the literature in differentiating subtypes of renal epithelial neoplasms.^4–7  Most of these markers, such as cluster of differentiation (CD) 10, renal cell carcinoma marker (RCCma), α-methylacyl-CoA racemase (P504S), epithelial membrane antigen (EMA), paired box gene (PAX) 8, PAX2, S100A1, and kidney-specific cadherin (ksp-cad), are also expressed in proximal and/or distal renal tubules as well. Therefore, most of these markers have little value in the distinction of a benign process from a renal epithelial tumor. However, vimentin, carbonic anhydrase IX (CAIX), CD117, and kidney injury molecule–1 (KIM-1) are not usually expressed in normal renal tubules, with the exception of KIM-1. In the condition of acute tubular injury, KIM-1 is usually expressed in renal proximal tubules. Therefore, if a low-grade CRCC is suspected, staining with CAIX, KIM-1, and vimentin can be performed, and positive staining for these markers will support a diagnosis of malignancy. Table 3 lists the expression of frequently used immunomarkers in 20 cases of benign renal tissues from Geisinger Medical Center, Danville, Pennsylvania.

Based on the histologic features, renal epithelial neoplasms can be classified into 2 main groups: low nuclear grade (Fuhrman nuclear grades 1 and 2) and high nuclear grade (Fuhrman nuclear grades 3 and 4 or sarcomatoid features). The group of low-grade neoplasms can be further divided into (1) tumors with clear cell/granular cell features; (2) tumors with both clear cell and papillary features; and (3) tumors with papillary features. The specific immunoprofile for each entity is summarized in Figure 1. The common differential diagnosis and the frequently used immunomarkers are summarized in Table 4.

Three main entities are included in this group: low-grade CRCC, ChRCC, and oncocytoma. The useful immunomarkers for the distinction of these 3 entities are listed in Table 5. The staining for cytokeratin (CK) 7 tends to be diffuse (>50% of tumor cells stained) in ChRCC, focal and patchy (<10% of tumor cells stained) in oncocytoma, and negative in CRCC. If a CRCC is diffusely positive for CK7, then a CPRCC should be considered. The nuclear and cytoplasmic staining for S100A1 in oncocytoma can be focal, but it is rarely observed in ChRCC, even focally. Caution should be taken, as S100A1 positivity is frequently seen in CRCC. In our experience, CK7, epithelial cell adhesion molecule (Ber-EP4), S100A1, and CD15 can serve as the initial IHC panel to differentiate a ChRCC from an oncocytoma. If this first IHC panel is inconclusive, additional markers, such as CD82, claudin 7, and claudin 8, can be used. In contrast, coexpression of EMA and vimentin and positivity for CAIX, RCCma, and KIM-1 are the diagnostic panel for CRCCs. Epithelioid angiomyolipoma (AML) with clear cytoplasm is a rare entity and sometimes may present as a mimic. Angiomyolipoma is positive for human melanoma black 45 (HMB-45), melanoma-associated antigen recognized by T cells 1 (Mart-1), and smooth muscle actin (SMA) and negative for CK and other renal cell carcinoma–associated markers such as PAX8/PAX2, CAIX, and RCCma.

The main differential diagnosis for a renal cell tumor with both clear cell and papillary features includes CPRCC, CRCC with papillary features, PRCC (type I) with clear cell features, and TRCC. The useful immunomarkers in differentiating these tumors are summarized in Table 6.

Clear cell papillary renal cell carcinoma is a recently recognized, distinct renal epithelial neoplasm that is more frequently seen in patients with end-stage renal disease.^54,55,66  The key histologic features include papillary structures, tubules, solid and cystic components, all lined by tumor cells with clear cytoplasm, and usually low Fuhrman nuclear grade.^54,55,66  The nuclei tend to be located away from the basement membranes and toward the luminal aspect of the tumor cells, creating subnuclear vacuolization similar to secretory endometrium.^54–56,66  These tumors lack the characteristic molecular features of PRCC and CRCC, including trisomies of chromosomes 7 and 17, deletions of 3p25, von Hippel–Lindau gene (VHL) mutation, and VHL promoter hypermethylation.⁵⁴  Immunohistochemically, these tumors are almost always positive for CK7 and CAIX but negative for P504S, CD10, RCCma, and transcription factor E3 (TFE3).⁵⁴  High-molecular-weight CK (34BE12, CK903) was reported to show positivity in 56% of cases⁵⁴  but usually showed negativity in CRCC, PRCC, and TRCC. Interestingly, parafibromin tends to be absent or only weakly expressed in parathyroid carcinoma but showed positivity in 100% of CPRCCs and was also expressed in 7% and 19% of CRCCs and PRCCs, respectively.⁵⁴  To complicate this matter further, Williamson et al⁵³  reported 12 cases of CPRCC-like tumors in patients with VHL disease sharing histologic features of sporadic CPRCC; however, 82% of these cases lacked the characteristic immunohistochemical profile of sporadic CPRCC. These tumors were frequently negative or only focally positive for CK7 and showed diffuse CD10 positivity in 64% of the cases and strong P504S staining in 27% of the cases.⁵³ 

In contrast to the IHC profile for CPRCC, CRCC with papillary features is usually positive for EMA, vimentin, CAIX, CD10, and RCCma, and negative for CK7, parafibromin, and 34BE12. In our experience, P504S expression can be seen in approximately 50% of CRCCs.¹⁹  The staining pattern of CAIX in CRCC is circumferential membrane staining of tumor cells, whereas the staining pattern in CPRCC tends to spare the apical surface of the tumor cells and outline the basolateral membranes.⁴  Papillary renal cell carcinoma with clear cell features, especially type I, is another key differential diagnosis, which typically shows a low nuclear grade and frequent clear cell changes, in comparison to type II PRCC, with high nuclear grade, an abundant amount of eosinophilic cytoplasm, nuclear stratification, and less frequent clear cell changes. Type I PRCC is usually diffusely positive for CK7, P504S, and focally positive for CAIX and CD10. An example of type I PRCC with clear cell features on a small-needle core biopsy specimen is shown in Figure 2, A through D. A reliable IHC panel in the distinction of type I PRCC from type II PRCC is not available yet; however, type II PRCC is frequently negative or only focally positive for CK7, strongly and diffusely positive for P504S, and can be positive for CK20.

The TRCC group of rare renal neoplasms predominately arises in pediatric patients with chromosomal translocation involving the TFE3 transcription factor gene located at 11p11.2 locus or the TFEB gene at 6p21.^6,7,55,58  Based on a study of 443 consecutive nephrectomies from a single institution, the incidence of TRCC in adults has been reported as 1.6%.⁵⁷  The typical morphologic features include papillary and nested structures lined by clear to granular cells with voluminous cytoplasm, hyaline material, and psammoma bodies.⁵⁸  Immunohistochemically, the tumor cells are positive for TFE3, cathepsin K, HMB-45, Mart-1, P504S, RCCma, and PAX8, only focally positive for CAIX, and usually negative for epithelial markers including cytokeratin AE1/3 (AE1/3), EMA, and CK7.⁵⁵ 

There is significant overlapping of the immunostaining profiles of PRCC and mucinous tubular and spindle cell carcinoma.⁶  The distinction between these 2 entities should primarily rely on histologic features. In contrast, metanephric adenoma is usually positive for S100, Wilms tumor 1 (WT1), and CD57 and negative for renal cell carcinoma–associated markers, such as P504S and RCCma.^20,33  The useful immunomarkers for distinguishing these 3 entities are summarized in Table 7.

The main diagnostic considerations for a high-grade renal epithelial neoplasm include high-grade CRCC, high-grade PRCC (type II), sarcomatoid renal cell carcinoma, UUC, CDC, renal medullary carcinoma, TRCC, and metastasis. The useful markers for this differential diagnosis are summarized in Tables 8 and 9. Carvalho and colleagues⁶⁷  suggested that p63, CK7, PAX8, and integrase interactor 1 (INI-1) comprised an optimal IHC panel to differentiate poorly differentiated urothelial carcinoma from other high-grade renal epithelial tumors. Their study demonstrated that (1) all urothelial carcinomas (N = 18) were positive for p63, 15 of 18 (83%) were negative for PAX8, 94% were positive for both CK7 and 34BE12, and all were positive for INI-1; and (2) all CDC cases were negative for p63, and 5 of 6 were positive for PAX8.⁶⁷  The study by Albadine et al⁶⁷  concluded that 100% of CDCs (N = 21) were positive for PAX8, and only 14% were positive for p63; whereas 97% of UUCs (N = 34) were positive for p63, and only 8.8% were positive for PAX8. Loss of INI-1 expression in nearly 100% of renal medullary carcinoma cases has been well documented in many studies.^61,62,67  Interestingly, octamer-binding transcription factor 3/4 (OCT3/4) expression has been demonstrated in 10 of 14 renal medullary carcinoma cases (71%) but was negative in 5 of 5 CDC cases and 10 of 10 UUC cases.⁶⁰  Chang et al⁵⁹  reported PAX8 expression in 69% (31 of 45) of sarcomatoid renal cell carcinomas and only in 4% (2 of 45) of sarcomatoid urothelial carcinomas of the bladder, whereas 18% (2 of 11) of sarcomatoid urothelial carcinomas of the upper urinary tract were positive for PAX8, showing some overlapping staining features with sarcomatoid renal cell carcinoma.⁵⁹ 

Angiomyolipoma is the most common nonepithelial renal neoplasm, and a classic one that can be easily diagnosed by its histologic features of mixed components of spindle cells, adipose tissue, and hyalinized blood vessels. An epithelioid angiomyolipoma with clear cell features may potentially pose a diagnostic challenge; however, a panel of IHC markers such as SMA, HMB-45, Mart-1, and CK is helpful in reaching a definitive diagnosis. Small round cell neoplasms can present as a primary renal tumor, including neuroendocrine carcinoma, Ewing sarcoma/primitive neuroectodermal tumor, neuroblastoma, lymphoma, plasmacytomas, and myeloid sarcoma. We recently encountered a case of primary precursor B-cell lymphoblastic lymphoma with a focal bone metastasis in a 4-year-old boy. The tumor mimicked Ewing sarcoma/primitive neuroectodermal tumor both morphologically and immunohistochemically. Figure 3, A through F, demonstrates the tumor cells to be positive for CD99, Friend leukemia virus integration 1 (Fli-1), PAX5, CD10, and terminal deoxynucleotidyl transferase (TdT) and to have a markedly increased mindbomb homolog 1 (MIB-1/Ki-67) proliferative index (not shown).

Metastasis in the kidney is rare; the most common site of origin of a metastatic carcinoma to the kidney is lung, including adenocarcinoma, squamous cell carcinoma, and small cell carcinoma. Other sites of origin may include colon, breast, gastrointestinal tract, pancreatobiliary tract, melanoma, and ovary. Adrenal cortical carcinoma may be a direct invasion or a metastasis. Steroidogenic factor 1 (SF-1) is a recently described, orphan member of the nuclear hormone receptor superfamily, which plays a crucial role in the differentiation of steroidogenic tissue. SF-1 has been reported to show positivity in 100% of adrenal cortical adenomas and 90% of adrenal cortical carcinomas and negativity in PRCC, ChRCC, and oncocytoma.⁶⁴  Only 3 of 20 CRCC cases showed focal positivity, with less than 10% of tumor cells stained.⁶⁴  All adrenal cortical neoplasms were negative for EMA, whereas most renal cell carcinomas were positive.⁶⁴  Another study by Sangoi and colleagues⁶⁵  demonstrated that SF-1 expression was seen in 86% of adrenal cortical neoplasms (N = 54) and none of 185 metastatic CRCCs. Expression of KIM-1, PAX8, and EMA was observed in 83%, 83%, and 78% of CRCCs, respectively.⁶⁵  In contrast, all 54 adrenal cortical neoplasms were negative for these 3 markers.⁶⁵  A small percentage of CRCCs can be positive for calretinin, inhibin-α, and Melan-A; whereas a small percentage of adrenal cortical neoplasms can be positive for CD10, CK, and CAIX.⁶⁵  SF-1, EMA, PAX8/or KIM-1/or pVHL are the optimal initial IHC panel to differentiate these 2 entities.^46,64,65  The useful diagnostic markers for metastasis in the kidney are summarized in Table 9.

In this section, we focus on lesions that may pose a diagnostic dilemma in cystoscopic biopsy material that is often scant, superficial, fragmented, or compromised by cautery artifact. Immunohistochemistry can be a valuable aid; however, it does not replace careful evaluation of morphology, including any normal urothelium that may be present.

Evaluation of urothelial lesions presents a unique problem, as common immunohistochemical markers exhibit staining patterns specific to particular compartments within normal urothelium. Familiarity with staining in normal histology is key to understanding how changes in the normal staining pattern correlate to various conditions in the morphologic spectrum of changes seen in reactive processes, hyperplasia, dysplasia, and carcinoma. Normal urothelium is 4 to 7 cells in thickness and consists of a layer of basal cells, 2 to 5 layers of intermediate cells, and a layer of superficial umbrella cells. Cystoscopic biopsies of a flat urothelial lesion can be problematic and immunohistochemistry can be quite helpful if morphologic evaluation alone is insufficient. We recommend a panel that includes CK20, CD44, p53, and Ki-67. Compartmental staining patterns of normal urothelium and flat lesions are summarized in Table 10^68–83 and illustrated in Figure 4, A through F.

Several lesions can have an exophytic papillary or polypoid cystoscopic appearance, including papillary urothelial neoplasms, polypoid cystitis, and nephrogenic adenoma. Although nephrogenic adenomas occur most commonly in the bladder, they may also develop in the renal pelvis, ureters, and urethra. If the lesion has a polypoid or papillary growth pattern, the differential diagnoses may include papillary urothelial neoplasms and polypoid cystitis; however, nephrogenic adenomas may also exhibit tubular, cystic, and solid histologic pattern, broadening the differential diagnosis to include prostatic adenocarcinoma, ectopic prostatic tissue, and clear cell adenocarcinoma of the bladder. Immunohistochemistry can help differentiate between these various lesions, and the pertinent markers are summarized in Table 11.^82,84–96 

Variants of invasive urothelial carcinoma present a challenge, as many cystoscopic biopsies contain very limited material and may not show urothelial dysplasia or an invasive carcinoma with a more classic appearance. Invasive urothelial carcinoma may show a micropapillary pattern similar to that seen in carcinomas from other primary sites, as well as epithelioid mesothelioma. Table 12 summarizes markers useful in this differential.^50,97–107  Similarly, the plasmacytoid variant of invasive urothelial carcinoma can be confused with plasmacytoma, adenocarcinomas, and other tumors. Table 13 summarizes the markers we recommend for this morphologic pattern.^50,108–114  The nested variant of invasive urothelial carcinoma can be difficult to distinguish from a benign proliferation of von Brunn nests, as well as tumors exhibiting a nested pattern such as paraganglioma, carcinoid, and melanoma. There is great overlap in the immunohistochemical staining pattern of these entities, particularly between paraganglioma and carcinoid tumors where presence or absence of S100-positive sustentacular cells may be the only clue. These stains are not particularly helpful to distinguish between urothelial carcinoma and von Brunn nests, unless the lesion is a carcinoma and diffuse positivity for CK20 and increased Ki-67 expression is present; otherwise, negative or weak CK20 staining and low Ki-67 expression can be indicative of either entity. Markers for tumors with a nested morphology are summarized in Table 14.^115–125 

The primary question when an adenocarcinoma is discovered in the bladder is whether or not it is a primary neoplasm, or has spread to the bladder either by direct extension or metastasis. Table 15 offers a basic immunohistochemical panel to differentiate primary bladder adenocarcinoma from other adenocarcinomas that commonly spread to the bladder, as well as conventional urothelial carcinoma and clear cell renal cell carcinoma.^* Nephrogenic adenoma may also be included in the differential diagnosis and is covered in Table 11.

Spindle cell lesions are uncommon in the bladder; however, they present a difficult diagnostic group in cystoscopic biopsies where the primary consideration is a sarcomatoid urothelial carcinoma versus a mesenchymal tumor. The differential diagnosis of mesenchymal neoplasms is further clouded by lack of a standard classification scheme in the urothelial tract, in part because of their rarity. The markers useful in the differential of spindle cell neoplasms are summarized in Table 16.^{72,82,139–150}  We prefer the term inflammatory myofibroblastic tumor, which encompasses a variety of terms in the literature, including inflammatory pseudotumor, pseudosarcomatous myofibroblastic tumor, inflammatory pseudosarcomatous fibromyxoid tumor, and fibromyxoid pseudotumor; this term is reflective of the low-grade nature of this lesion and does not incite the same alarm and confusion as the terms that include pseudosarcoma.

The diagnosis of prostate carcinoma is largely achieved by histomorphologic assessment on routine hematoxylin-eosin sections of prostate surgical specimens. One of the key diagnostic criteria for malignancy is the absence of basal cells. The identification of basal cells on a hematoxylin-eosin section of core needle biopsies is not always straightforward; some artifacts may make this task challenging.¹⁵¹  Immunohistochemistry using antibodies against basal cells is an objective and reliable tool to demonstrate the presence or absence of basal cells. However, the lack of basal cells in isolation is not diagnostic of malignancy, since benign mimics can show false-negative reactivity. A number of studies have identified biomarkers that are overexpressed in prostate carcinoma and their precursor. The application of cancer overexpressed biomarkers in conjunction with basal cell markers provide additional confidence in making a definitive diagnosis of prostate cancer in challenging cases. In addition, the differential considerations between urothelial and prostatic origin arise occasionally, especially in transurethral resection specimens. Under those circumstances, immunohistochemistry becomes a valuable aid in reaching an accurate diagnosis. The commonly used immunomarkers in prostate surgical pathology include ones aiding in the diagnosis of malignancy and others defining prostate organ specificity.

Two groups of biomarkers are used to aid in the diagnosis of prostate carcinoma on core needle biopsy specimens. The first group is composed of basal cell markers in which a negative immunoreaction indicates malignancy. The second group consists of upregulated biomarkers in which a positive immunoreaction suggests malignancy.

p63, a cloned homologue of the p53 tumor suppressor gene, is a nuclear protein involved in the regulation of growth and development in epithelium of various tissues including prostate.¹⁵²  Anti-p63 labels the p63 protein in the nuclei of basal or progenitor cells in a variety of epithelia, serving as a basal cell marker with a nuclear staining pattern in prostate surgical pathology.¹⁵²  Many studies concluded that the absence of p63 stain indicates prostate carcinoma.^153–157  Thus, the application of p63 immunostain is essential when confronting challenging cases. However, aberrant diffuse p63 expression in prostatic carcinomas had been reported.^158–160  In the study of Osunkoya et al,¹⁵⁸  21 cases of prostatic carcinomas revealed aberrant diffuse expression of p63 (100% of cancer cells in 19 of 21; 75% in 2 of 21); all were negative for high-molecular-weight cytokeratin (HMWCK, or CK903), positive for prostate-specific antigen (PSA) and P504S. A few reports^161,162  also documented false-positive p63 staining in prostate cancers with a basal cell distribution. Although aberrant expression of p63 is rare, caution should be exercised when working on cases with equivocal features.

HMWCK (CK903, clone 34βE12) is a well-known basal cell marker. Gown and Vogel¹⁶³  investigated the expression of monoclonal antibodies generated against different human intermediate filament proteins in a variety of human normal and neoplastic tissue in 1984; they first described the staining pattern of antibody 34βE12 in prostate as being basal glandular epithelial cells only in distribution. Thereafter, CK903 was studied extensively in normal and neoplastic prostatic tissues, validating its diagnostic utility as a differentiating immunomarker between benign and malignant process, with the lack of basal cell staining being an indicator of malignancy.^163–167  However, lack of basal cell staining is also noted in 5% to 23% of benign prostatic glands.^{151,154,165,168}  Googe et al¹⁶⁶  and Yang et al¹⁶⁷  reported focal patchy immunoreactivity (nonbasal cell distribution) for basal cell markers (especially HMWCK) in high-grade prostatic carcinomas. Thus, Epstein¹⁵¹  advised that negative basal cell staining in small glandular (and other) prostatic lesions is supportive of a diagnosis of prostatic carcinoma only in the appropriate hematoxylin-eosin context. CK5/6, another HMWCK, is expressed in basal cells. A few studies evaluated CK5/6 expression in various prostate normal, benign, and neoplastic tissues, suggesting superior sensitivity and reliability of CK5/6 compared with CK903.^169,170 

P504S, the gene encoding α-methylacyl-coenzyme A racemase (AMACR), was first reported by Xu et al¹⁷¹  in 2000. By using complementary DNA (cDNA) library subtraction in conjunction with high-throughput microarray screening techniques, they found that P504S is overexpressed in prostate carcinomas, while low or undetectable in normal prostatic tissues. In 2001, Jiang et al¹⁷²  found 100% cases of prostate carcinomas expressing P504S regardless of Gleason score, with positive staining defined as continuous, dark cytoplasmic staining or apical granular staining that can be observed easily at low-power magnification. In contrast, 88% of the benign prostatic tissue was negative for P504; only 12% showed focal, weak reactivity. Subsequently, P504S expression was evaluated in a variety of benign and neoplastic prostatic tissues,^173–179  demonstrating that P504S is overexpressed in prostate carcinoma and high-grade prostatic intraepithelial neoplasia (HGPIN), with a reported sensitivity in the range of 62% to 100%; the staining patterns were variable, from diffuse and strong (most reports) to focal and weak.¹⁶²  By comparison, variants of prostate carcinoma showed lower sensitivity, including 62% to 68%¹⁸⁰  and 72%¹⁷⁹  for foamy gland carcinoma; 83%¹⁸¹  and 70% to 77%¹⁸⁰  for pseudohyperplastic carcinoma; and 67% for atrophic carcinoma.¹⁸²  A few studies^181,183,184  investigated P504S expression in postradiation-therapy prostate carcinomas and reported a positive rate of 80% to 100%. The benign atypical glands in postradiation-therapy prostate were P504S negative. Their findings suggested the diagnostic utility of P504S immunostaining in distinction between postradiation prostate carcinoma and radiation-induced atypia in benign prostatic epithelium.

Although benign prostate glands usually lack P504S expression, positive immunoreactivity for P504S was described by several investigators in partial atrophy, atypical adenomatous hyperplasia, and nephrogenic adenoma (NA). Partial atrophy is the most common benign mimicker of prostate carcinoma on core needle biopsy. Wang et al¹⁸⁵  evaluated P504S and basal cell expression in 198 cases of partial atrophy and found P504S expression in 69.2% of cases, and positive basal cells in 68.7% that showed a focal patchy staining pattern. Herawi et al¹⁸⁶  reported patchy basal staining in 87% of partial atrophy foci and P504S positivity in 79% of cases. Atypical adenomatous hyperplasia is another well-known mimicker of prostate carcinoma. Yang et al¹⁸⁷  reported P504S expression in 17.5% cases of atypical adenomatous hyperplasia, while Skinnider et al⁸⁶  reported 14%. The reported expression of P504S in nephrogenic adenoma ranges from 58% to 100%, and most nephrogenic adenomas were also CK903 negative.^86,188,189  The similarity in certain morphologic features and the P504S+/CK903− immunoprofile of nephrogenic adenoma make the distinction between prostate carcinoma from nephrogenic adenoma challenging. Additional immunomarkers, especially PSA, PAX2 (or PAX8), and CK7 can help to differentiate nephrogenic adenoma from prostate carcinoma. Nephrogenic adenoma is usually negative, or only focally and weakly reactive to PSA, and positive for CK7 as well as PAX2,^95,173,190  while prostate carcinoma expresses PSA and is nonreactive to CK7 and PAX2 (see Figure 5, A through H).

P504S is upregulated in prostate carcinomas and HGPIN, and can serve as a positive tissue marker; however, by itself, P504S positivity is not specific as it may also be expressed in benign mimickers such as atypical adenomatous hyperplasia and nephrogenic adenoma. Jiang et al¹⁷⁹  suggested using a positive P504S marker along with a negative basal cell–specific marker, such as CK903 and/or p63, to help confirm the diagnosis when small atypical glands are seen. Zhou et al¹⁵⁶  reported that a basal cell cocktail (CK903 plus p63) improves the detection of prostate basal cells. Browne et al¹⁹¹  prospectively stained 123 cases with a basal cell cocktail and P504s; the immunostain results contributed to rendering a final diagnosis in 70% of cases. They suggest that combining P504s and the basal cell cocktail on a single slide would be superior to using either marker separately. Currently, an antibody cocktail composed of P504S, CK903, and p63 is routinely used.

ERG, a transcription factor in the erythroblastosis virus E26 transforming sequence (ETS) family known to be expressed in endothelial cells, is overexpressed in subsets of prostate carcinoma, acute myeloid leukemia, and Ewing sarcoma. A number of studies evaluated ERG expression by immunohistochemistry in benign and neoplastic prostatic tissues, reporting a positive rate of 38% to 45% in prostate carcinoma, 22% to 29% in HGPIN, and rare expression in benign glands.^192–196  In nonprostatic tumors, ERG expression is exclusively identified in vascular tumors, 70% of extramedullary myeloid tumors, 7% of Ewing sarcomas, and rare large cell undifferentiated pulmonary carcinoma, as well as mesothelioma.^192,194  Those data suggest that ERG is highly specific for prostate carcinoma, in addition to vascular tumors, extramedullary myeloid tumors, and rare Ewing sarcoma, but has low sensitivity. Its utility in the differential between benign mimics and prostate carcinoma on core needle biopsies may be limited. However, the high specificity for prostate carcinoma may prove to be of value in the identification of prostate primary cancer when working on tumors of unknown origin.

Prostate-specific antigen and prostate-specific acid phosphatase (PSAP) are 2 well-known prostate tissue–specific immunomarkers, and both are highly sensitive and relatively specific; however, their expression in nonprostatic tissues has been reported, including melanoma, breast carcinoma, and salivary ductal carcinoma.^{162,197–202}  It had been documented that poorly differentiated prostate carcinomas express less PSA and PSAP than benign prostatic tissue and low-grade prostate carcinomas. The reported sensitivities range from 25% to 35% when using monoclonal antibody and approximately 95% when using polyclonal antibody.^{133,162,203,204}  We evaluated PSA expression (using both monoclonal and polyclonal antibodies) in 133 cases of prostate carcinoma (low-intermediate grade: 97; high grade: 36) and 909 cases of nonprostatic tumors from a broad spectrum of human tissues and demonstrated the following: (1) PSA expression is identified in 100% cases of prostate carcinoma with both antibodies; the staining intensity is stronger for polyclonal antibody than monoclonal antibody; (2) No difference can be appreciated when comparing different grades of prostate carcinoma; (3) PSA expression in nonprostatic tumors is 1.3% (12 of 909) when using polyclonal antibody and 1.1% when using monoclonal antibody. The positive cases are 1 of 20 papillary renal cell carcinomas (focal, weak), 1 of 74 breast invasive ductal carcinomas (focal, moderate), and 10 of 93 for polyclonal antibody and 8 of 93 for monoclonal antibody in melanoma (focal, weak). Our data showed a PSA sensitivity of 100%, and specificity of 98.7%.¹⁹⁴ 

P501S, or prostein, is a 553-amino-acid protein identified by cDNA library subtraction followed by high-throughput microarray screening technique.²⁰⁵  Prostein (P501S) messenger RNA and protein expression are prostate specific, restricted to both normal and malignant prostate tissue; all other nonprostatic normal or malignant tissues lack prostein expression.^205,206  Sheridan et al²⁰⁷  immunohistochemically evaluated prostein expression in 20 cases of primary prostate carcinoma, 20 cases of normal prostate tissue, and 69 cases of metastatic prostate carcinoma, finding prostein expression in 100% of primary prostate carcinomas and normal prostate tissue, and 99% of metastatic prostate carcinomas. They described the staining pattern as perinuclear cytoplasmic (Golgi) in distribution. Yin et al²⁰⁸  found that 100% of benign prostate tissue, 94.1% of primary prostate carcinomas, and 87% of metastatic prostate carcinomas were immunoreactive to prostein. Both studies also examined PSA expression, showing similar staining results. The data, although limited, suggest the potential utility of prostein as a prostate-specific marker in the detection of metastatic prostate carcinoma.

NKX3.1 is a prostate-specific homeobox gene located on chromosome band 8p21. In human normal and neoplastic tissues, NKX3.1 expression is highly specific, restricted to prostate, testis, and breast with a nuclear staining pattern as illustrated in Figure 6, A and B.^209−211 Gelmann et al²¹⁰  evaluated the expression of NKX3.1 in 4061 samples encompassing a broad spectrum of human cancers and normal tissues and found that NKX3.1 is expressed primarily in benign and malignant prostatic epithelial cells, but also normal testis, 5% to 9% of breast invasive ductal carcinomas, and 26% to 27% of breast invasive lobular carcinomas. Chuang et al¹³³  evaluated the sensitivity of prostate-specific immunomarkers, including PSA, prostein (P501S), prostate-specific membrane antigen (PSMA), NKX3.1, and proPSA (pPSA) on 38 poorly differentiated prostate carcinomas and 35 high-grade invasive urothelial carcinomas. They found the sensitivities for labeling prostate carcinoma were 97.4% for PSA, 100% for P501S, 92.1% for PSMA, 94.7% for NKX3.1, and 94.7% for pPSA. Thus, they suggested using PSA as the first screening marker in differentiating high-grade prostate carcinoma from high-grade urothelial carcinoma; other markers are useful when PSA shows negative or equivocal staining. Gurel et al²⁰⁹  evaluated the expression of NKX3.1, PAS, and PSAP in 69 cases of metastatic prostate carcinomas and 349 cases representing a variety of nonprostatic tumors. The sensitivity for identifying metastatic prostatic adenocarcinomas overall was 98.6% for NKX3.1, 94.2% for PSA, and 98.6% for PSAP. The specificity of NKX3.1 was 99.7% (1 of 349 nonprostatic tumors positive). The sole positive nonprostatic cancer case was an invasive lobular carcinoma of the breast. The available limited data suggest that NKX3.1 is a highly sensitive and relatively specific tissue marker for prostate carcinoma and may have utility in identification of prostate origin when working on tumors of unknown primary.

Testicular tumors account for only 1% of all tumors, so they are relatively rare for many pathologists and may present a significant diagnostic challenge. Most testicular tumors occurring at any age are germ cell tumors (GCTs). Approximately 50% of GCTs in adults are pure classic seminomas, but approximately 30% will be mixed GCTs. Each component in mixed GCTs may exhibit multiple morphologic patterns, creating additional challenges. The histologic patterns of GCTs often overlap with those of sex cord–stromal tumors (SCSTs), and some tumors may have both germ cell and sex cord–stromal components to further complicate matters. Identifying all components of a testicular tumor is critical for patient management, as the various components may or may not respond to radiation or chemotherapy.^1,212  The newer stem cell markers, such as SALL4, OCT4, and Nanog, are particularly useful in identification of GCTs. Immunomarkers useful in the differential identification of GCTs are presented in Table 17^213–255 and illustrated in Figure 7, A through F. Markers that are helpful in distinguishing between GCTs and SCSTs are presented in Table 18.^{212,232,252,256}  We recommend an initial panel consisting of sal-like 4 (SALL4), placental-like alkaline phosphatase (PLAP), inhibin α, and calretinin to differentiate between GCTs and SCSTs; if all 4 markers show negativity, then other neoplasms such as adenocarcinomas, melanoma, or lymphoma should be considered. Markers useful in the identification of SCSTs are included in Table 19^{217,256–280} and a typical pattern of SF-1 expression is illustrated in Figure 8.

Tumors of the paratesticular structures are rare and it can be difficult to distinguish the primary site of origin and whether or not the tumor is primary to the paratesticular structures, or metastatic from the abdomen or pelvis. The distinction between epithelial or mesothelial origin is also problematic in these tumors. Markers that may help in the distinction of primary epithelial neoplasms of the rete testis and epididymis, as well as mesothelial neoplasms, are presented in Table 20.^{217,256,281–302}  If the tumor in this region appears to be metastatic, Table 21 lists markers that may help in the identification of adenocarcinomas occurring in the male pelvis.^†

Immunohistochemisty has broad diagnostic applications in genitourinary pathology and is a useful adjunct to morphology in difficult cases. It can be applied to benign and malignant neoplasms, but this must be done in the context of comparing to expression of these markers in normal tissues.