Although immunohistochemistry techniques and applications have made great strides in the past 25 years, diagnostics in the world of dermatopathology is still largely based on H&E stained slides. Despite that fact, there are many ways in which immunohistochemistry (IHC) can assist the pathologist in the diagnosis of problematic cases. Outlined below are a few of the ways in which antibodies can be utilized in dermatopathology.
The differential diagnosis of atypical spindle cell neoplasms on sun-damaged skin includes atypical fibroxanthoma, spindle cell or desmoplastic melanoma, spindle cell squamous cell carcinoma, and leiomyosarcoma. This is a not-infrequent challenging differential in skin tumor pathology. Because of potential overlapping reactivity and rare, anomalous expression, a panel of IHC markers should be selected that will support or refute each of those diagnoses.
Spindle Cell SCC
Squamous cell carcinomas (SCC), in general, stain with cytokeratin and not with vimentin. However, spindle cell SCCs that may lack an obvious origin in the epidermis or evidence of keratinization are often positive for vimentin and negative or only focally positive with routine cytokeratin stains, including AE1 & AE3. Folpe and Cooper1 found high molecular weight (HMW) cytokeratins, such as CK903 (34betaE12) and CK5 & 6 to be sensitive markers in this context; however, some spindle cell SCCs do not stain with either.2-4
Atypical fibroxanthoma (AFX) is a pleomorphic, superficial spindle cell tumor of low-grade malignancy that must be distinguished histologically from other atypical spindle cell tumors of the dermis. Although there are immunohistochemical markers for many of those other tumors, the diagnosis of AFX is generally one of exclusion. A variety of markers have been identified with reactivity in AFX, but those markers alone are not specific. Procollagen 1 (PC1), CD10, and S100A6 are typically reactive in AFX, but those markers often stain a variety of other neoplasms. Procollagen is secreted by fibroblasts and is cleaved to form collagen in the extracellular matrix. Expression of PC1 in most AFXs supports classification of this tumor as fibrohistiocytic. However, PC1 staining should not be used in isolation. Reactivity with keratin, S-100, or desmin should outweigh PC1 positivity because approximately three-fourths of leiomyosarcomas and one-third of desmoplastic melanomas (DMs) and spindle cell SCCs are also PC1 positive.5-7 As with all IHC markers, careful localization of the antibody staining should be established. PC1+ tumor cells must be distinguished from positive background fibroblasts.
S100A6 (calcyclin) is a calcium-binding protein of the S-100 family, which has been isolated from melanocytes, Schwann cells, Langerhans cells, and dermal dendrocytes. Aside from staining nevi and some melanomas, numerous fibrohistiocytic lesions stain positive with S100A6. A small study8 revealed S100A6 in 80% (4 of 5) of the AFX cases. However, other neoplasms in the spindle cell differential have also revealed reactivity with S100A6, including a significant proportion of DMs, leiomyosarcomas, and spindle cell SCCs.9
Similarly, CD10 is sensitive but not specific for AFX. In one small study,10 94% (15 of 16) of the AFXs showed strong expression of CD10, but on the other hand, reactivity was also identified in one-third (3 of 9) of the DMs and one-half (5 of 10) of the spindle cell SCCs.
Focal expression of myogenic markers (calponin, SMA, HHF35), indicative of myofibroblastic differentiation, can be seen in up to one-third of AFXs.1, 11 Therefore, use of actin, smooth muscle (SMA) or calponin alone may be insufficient to diagnosis leiomyosarcoma, and a panel should include another muscle marker, such as desmin or h-caldesmon. Caution is required in interpreting S-100 in AFXs because S-100+ dendritic cells colonize the lesion (possibly Langerhans cells),12,13 but the neoplastic cells are generally S100- essentially excluding DM. Multinucleate giant cells of AFX can also express MART-1/Melan A and HMB-45.14,15
Historically other nonspecific markers with lower reactivity used in AFX have included CD68,13 antichymotrypsin, antitrypsin,16 and CD99.17 Studies have shown that CD163 is more specific for histiocytes than CD68, which is an organelle-specific marker that stains lysosomes. Because CD68 reactivity is seen in some AFXs, it is not surprising that CD163 expression was reported18.
Caution is required in distinguishing AFX with pseudoangiomatous or hemorrhagic features from angiosarcoma. It can be further complicated by occasional expression of D2-40, FLI-1, and CD31 in AFX.19-22 CD34 and ERG may be useful to differentiate those entities.21
Desmoplastic melanoma (DM) often lacks melanin and, sometimes, junctional involvement. When considering its differential diagnosis, it must be remembered that cytokeratin (2%), SMA, and desmin can be anomalously expressed in DMs. CD68+ melanomas may also be mistaken for AFX. HMB-45 is usually negative, and Melan A is positive in only 7% of DMs.23
Microphthalmia transcription factor (MiTF) similarly has low sensitivity in DM. S-100 is a sensitive, albeit not specific, marker for spindle/DM. However, many scars contain S-100+ spindle cells complicating differentiation, particularly of reexcision specimens, but unlike DM, the S-100+ cells are only focal and are predominantly in a horizontal orientation . Nerve growth factor receptor (NGFR) (p75), is a neural crest marker expressed in most desmoplastic and neurotropic melanomas, often more intensely than S-100.24,26
Leiomyosarcomas, AFX, and 81% of spindle cell SCCs are negative for p75, and those spindle cell SCCs that do react show only focal nests of positivity.25 However, many other malignant spindle cell tumors are also positive for p75, including peripheral nerve sheath tumors, dermatofibrosarcoma protuberans (DFSP), rhabdomyosarcoma, synovial sarcoma, and neurotized nevi. Spindle cell melanoma without desmoplasia or neurotropism, many typical epithelioid melanomas, and nevi fail to stain with p75 or do so only focally.26 Scars may reveal p75+ cells, similar to S-100 (possibly myofibroblasts, nerve twigs, or Schwann cells) requiring caution in distinguishing scars from DMs.27
Sex-determing region Y-box 10 (SOX-10) has proven to be as sensitive a marker for DM as S-100 with improved specificity. AFX and spindle cell SCCs fail to show SOX-10 expression, but malignant peripheral nerve sheath tumors show scattered positivity similar to S-100.28 SOX-10 has become an important alternative in identification of DM, especially in reexcision specimens. Ramos-Herberth et al29 reported only rare, weak SOX-10 expression in spindled fibroblasts of scars, in comparison to greater MiTF and S-100 expression in histiocytes and fibroblasts of scars.
Use of a panel of immunohistochemical markers in the setting of atypical dermal spindle cell tumors is required to prevent misdiagnosis; some combination of S-100 and SOX-10 for DM, high–molecular-weight keratin and p63 for spindle cell SCC, desmin and h-caldesmon for leiomyosarcoma, and if hemorrhagic, ERG or CD34 for angiosarcoma.
Differentiation of sebaceous carcinoma and squamous cell carcinoma (SCC) can be challenging. Ansai et al30 found that androgen receptor (AR) and adipophilin expression supports the diagnosis of sebaceous carcinoma, rather than SCC. Sebaceous epitheliomas (sebaceomas) are mostly composed of basaloid cells with a small proportion of mature sebocytes. Thus, there is histologic overlap between the basaloid cells of sebaceomas and those of basal cell carcinoma (BCC). Although IHC is not commonly necessary in differentiation, epithelial membrane antigen (EMA) highlights most mature sebocytes in sebaceoma, whereas its expression is uncommon in BCC.31,32 Reactivity for adipophilin in sebaceous tumors can also distinguish them from BCC.1 The utility of Ber-EP4 in this setting is controversial; although BCCs stain positive for Ber-EP4, there is varying literature on its reactivity in sebaceous lesions.31,33
Adipophilin is a protein on the surface of intracellular lipid droplets and thus shows expression in sebaceous lesions, particularly well-differentiated lesions. Caution is necessary when using adipophilin to distinguish sebaceous neoplasms from other cutaneous lesions with clear cell histology. Metastatic renal cell carcinoma and eccrine-apocrine carcinomas have shown expression.34,35 Xanthomatous lesions, including xanthelasma and xanthogranuloma, reportedly show adipophilin reactivity, but in contrast to the membranous vesicular pattern in sebaceous neoplasms, xanthomatous lesions show a more granular pattern.34 This granular pattern has also been reported in clear cell BCCs and SCCs potentially causing confusion when differentiating those neoplasms from sebaceous carcinoma; however, a granular pattern and decreased intensity of expression favor the former tumors.35,36 The potentially aggressive course of sebaceous carcinoma necessitates accurate differentiation from benign sebaceous neoplasms. Cabral et al37 sought to find an immunohistochemical method that could aid morphologic differentiation, especially in small, partial biopsies. The p53 protein is a tumor suppressor. Mutations lead to an abnormally stable but inactive p53 protein that can be detected immunohistochemically (nuclear staining). Sebaceous carcinomas are associated with increased, presumably defective, p53. Sebaceous carcinomas have significantly increased nuclear staining with p53 and Ki-67 (MIB-1) and reduced levels of BCL2 and p21, in comparison to sebaceous adenomas and sebaceous epitheliomas.
In some cases, IHC can serve as surrogate protein markers for underlying genetic events. Sebaceous gland neoplasms are relatively rare, and that diagnosis should raise the possibility of Muir-Torre syndrome (MTS). The dermatopathologist may be the first to consider the possibility of this syndrome because, in some patients, only a single sebaceous tumor is the first indication. MTS is an autosomal-dominant disease associated with multiple keratoacanthomas and visceral malignancies, especially genitourinary and gastrointestinal. Less common associations include tumors of the breast and lung. MTS is most frequently caused by mutational inactivation of mismatch repair genes MSH2 and MLH1 with resultant microsatellite instability. Other mismatch repair proteins can be involved, such as MSH6 and PMS2. It is reasonable to check this panel (MSH2, MLH1, MSH6, with or without PMS2) in sebaceous neoplasms (other than sebaceous hyperplasia), especially in patients younger than 50 years, when multiple lesions are present, neoplasms involve nonfacial sites, or when cystic or keratoacanthoma-like architecture is present.38,39 Although some authors are conservative and recommend screening only in patients with a personal or family history of colorectal cancer,40 others41,42 recommend screening regardless of age or other clinical characteristics.
Solitary trichilemmomas are typically sporadic but can be associated with Cowden syndrome, which is an autosomal dominant disorder characterized by multisystem hamartomatous growths and carcinomas. This syndrome is linked to a germline mutation in the tumor suppressor phosphatase and tensin homolog (PTEN), located on band 10q23.3. Complete loss of immunoreactivity with PTEN was reported in 5 of 6 patients (83%) with Cowden syndrome–associated trichilemmomas, but in only 1 of 33 patients (3%) with sporadic lesions.43 Shon et al44 reported similar results with 13 Cowden syndrome–associated lesions and 19 sporadic trichilemmomas. This suggests a potential role for immunohistochemical screening similar to its use in MTS. Partial biopsies of desmoplastic tricholemmoma may require distinction from BCC. CD34 immunodetection in the epithelial cells of the tricholemmoma can help in the differentiation.45
Primary Cutaneous Adnexal Neoplasms From Metastatic Adenocarcinoma
Differentiation of metastatic adenocarcinomas from primary cutaneous adnexal neoplasms (PCAN), especially malignant ones, can be difficult. Qureshi et al46 found that most PCANs with sweat gland differentiation were positive for p63 and high–molecular-weight CK5 & 6, whereas expression was rare in metastatic adenocarcinomas. Differential p63 reactivity has also been supported in other studies.47-49
Plumb et al24 similarly found CK 5& 6 expression in more than 95% of PCANs, albeit most tumors studied were benign. In general, metastatic adenocarcinomas expressed CK5/6 in only 33% (9 of 27) of the cases, predominantly with weak intensity. However, metastatic breast carcinoma was reactive for CK 5 & 6 in almost half of their cases.50 Ivan et al51 extended the analysis of p63 to include metastases from adnexal carcinomas and found that 91% of PCANs were strongly marked with p63 and, excluding apocrine and mucinous carcinoma, their metastases labeled similarly.
Podoplanin (D2-40) expression is also a useful adjunct in differentiating adenocarcinomas that have metastasized to the skin from PCAN. Reactivity with D2-40 is seen in primary cutaneous carcinomas and skin adnexal tumors but is absent in metastatic adenocarcinomas to the skin.47,52 A panel of immunohistochemical stains, as suggested by Mahalingam et al,48 may provide the best sensitivity and specificity in distinction of metastatic adenocarcinomas and PCANs. Their recommended panel included p63, D2-40, and CK15. Positive staining with all three favors a PCAN diagnosis, rather than that of metastatic adenocarcinoma.
Adenocarcinomas do metastasize to the skin, and rarely, that can be the first manifestation of the primary diagnosis. Identification of the site of origin has profound prognostic and therapeutic effect. IHC has been useful in a number of cases. Differential cytokeratin staining (CK7 and CK20) may be helpful in this setting but is not specific. Of the most common metastases to skin, breast and lung adenocarcinomas are CK7+/CK20-, and colorectal carcinoma is CK7-/CK20+.53 Other potentially useful immunohistochemical stains include CDX-2 and TTF-1. CDX-2, a nuclear marker, is expressed in almost all colorectal carcinomas but is also expressed in a few other adenocarcinomas, including gastric, biliopancreatic, and mucinous ovarian adenocarcinomas.54,55 TTF-1 is a relatively specific marker for tumors of lung or thyroid origin. Around two thirds of lung adenocarcinomas express TTF-1. However, expression is typically absent in SCC of the lung and lung mucinous adenocarcinomas/bronchioloalveolar carcinomas.54,56 CDX-2 and TTF-1 are reasonable additions to the antibody panel that aims to identify the origin of a metastatic adenocarcinoma of unknown primary. Park et al57 proposed and evaluated a panel of antibodies to improve prediction of the primary site. In addition to CK7, CK20, TTF-1, and CDX-2, the panel included carcinoembryonic antigen (CEA), epithelial mucin genes MUC2 and MUC5AC, ER, and gross cystic disease fluid protein 15 (GCDFP015). GCDFP-15 is a relatively specific marker for breast carcinoma with reactivity ranging from 43% to 77%. The numbers indicate limited sensitivity. Mucin expression varies between carcinomas arising from different organs. MUC2 and MUC5AC are expressed predominantly in colon adenocarcinoma and mucinous ovarian adenocarcinoma, respectively; however, there are overlapping patterns. The combined phenotypes correctly predicted the primary site in 75% of cases.57 Cutaneous metastases occur in up to 11% of patients with renal cell carcinoma (RCC) and may be the presenting sign of disease.58 CD10 is a metallomembrane endopeptidase first identified in acute lymphoblastic leukemia and also expressed in follicular lymphoma, Burkitt lymphoma, hepatocellular carcinoma, urothelial carcinoma, and prostatic carcinoma.59 CD10 may be used to confirm the diagnosis of renal carcinomas, especially clear cell carcinomas and less so papillary or chromophobe RCCs.60 The histologic differential diagnosis of metastatic RCCs includes primary cutaneous adnexal tumors. CD10 was evaluated in adnexal tumors by Bahrami et al58 to determine its utility in the distinction. Six percent of eccrine and apocrine neoplasms and 40% of sebaceous neoplasms demonstrated CD10 reactivity. RCC metastatic to the skin can also simulate other clear cell lesions. Perna et al61 evaluated CD10 in clear cell mimickers, including xanthomas, xanthelasma, and xanthogranulomas. CD10 stained most of those lesions; however, that was predominantly in a membranous pattern compared with the cytoplasmic pattern of RCC. CD10 was also expressed in 25% and 33% of balloon cell nevi and clear cell hidradenomas, respectively, but expression was limited to less than 10% of the clear cells. That overlap of reactivity between metastatic RCC and other cutaneous clear cell lesions is not present with the renal cell carcinoma marker (RCC-Ma). Perna et al62 found RCC-Ma expression in 62.5% of cutaneous RCC metastases and no reactivity in clear cell mimics, including xanthomas, xanthogranulomas, balloon cell nevi, clear cell hidradenoma, and sebaceous neoplasms. PAX-8 is also a renal tumor marker with greater sensitivity than RCC-Ma. PAX-8 also stains thyroid, gynecologic, and neuroendocrine tumors, including Merkel cell carcinoma.63,64
The melanocytic origin of a tumor may not always be apparent. In addition to morphologic clues, IHC can assist in distinguishing melanocytic from nonmelanocytic lesions. Staining with S-100 was one of the first and most enduring markers for melanocytic lesions. Although most melanomas are S-100+, that marker lacks specificity and stains neural tissue, Langerhans cells, Rosai-Dorfman disease, and other tumors. Thus, other antibodies may be necessary to confirm the melanocytic nature of S100þ neoplasms. Both the MART-1 (melanoma antigen recognized by T cells) and Melan A antibodies recognize the same gene product and are expressed by normal melanocytes, nevi, and melanoma, but less frequently by DM.66 HMB-45, also known as anti-gp100, a premelanosome marker, has been shown to mark the intraepidermal and superficial dermal components of melanocytic nevi, with the exception of diffuse dermal staining in blue nevi.67 When compared with MART-1, HMB-45 has been found to have weaker and more-focal staining in both primary and metastatic melanomas. Therefore, HMB-45 is relatively specific, staining only rare other tumors (eg, perivascular epithelioid cells tumors [PEComas]) but is not very sensitive for melanoma.68,69 MiTF is responsible for the normal embryonic development of melanocytes, mast cells, cells of the retinal pigment epithelium, and osteoclasts.70 Melanocytes express nuclear MiTF. Most melanomas retain MiTF reactivity; however, a large proportion of desmoplastic and spindle cell melanomas fail to stain.70,71 Positivity with MiTF has been reported in 88% (235 of 266) of the conventional metastatic melanomas.71 However, MiTF is not specific, and staining has also been reported in neurofibromas, dermatofibromas, AFX, leiomyosarcomas, schwannomas, malignant peripheral nerve sheath tumors, solitary fibrous tumor, giant cell tumors of the tendon sheath, dermal scars, and granular cell tumors.68,72
SOX-10, a nuclear transcription factor expressed in neural crest cells, is crucial for differentiation of Schwann cells and melanocytes. Nuclear staining is noted in normal melanocytes, Schwann cells, secretory cells of the eccrine coil, myoepithelial cells, and predominantly cytoplasmic staining in mast cells.73,74 Expression has been shown in all types of nevi (blue, neurotized, dysplastic, Spitz, capsular) and melanoma (epithelioid, spindled, desmoplastic, metastatic).75,76 The specificity is also high, with expression in only a few other tumors, including granular cell tumor, schwannoma, neurofibroma, myoepithelioma, clear cell sarcoma, and some ductal breast carcinomas. It is notably absent in cellular neurothekeoma, Langerhans cell histiocytosis, perineurioma, MCC, spindle cell SCC, AFX, and leiomyosarcoma.72-75,77,78 SOX-10 is particularly useful in the setting of junctional melanocytic proliferations on sun-damaged skin, DM, and in evaluation of sentinel lymph node biopsies.
Several other melanocytic markers have been studied, including NKI-C3, TRP-1, and TRP-2 (antibodies to tyrosinase), SM5-1, PNL2, and KBA.62.79-84 Multiple myeloma 1 (MUM1), a known hematolymphoid marker, shows positive immunostaining in nevi; in more than 90% of primary melanomas, with the exception of DM; and in 80% of metastatic melanomas.85,86 To date, none of these have shown any significant benefit over those previously discussed.
Some melanomas exhibit aberrant immunohistochemical expression and may express desmin, SMA, CD68, CEA, and EMA.87 Korabiowska et al88 found cytokeratin expression in 6% of malignant melanomas, with staining tending to be focal, and only 3% or more of the cells staining. Use of a panel, including the previously mentioned melanocytic markers, should prevent misdiagnosis.
PEComas have immunophenotypic features of smooth muscle and melanocytes. Although many are retroperitoneal or visceral, a subset arises in the skin. Those lesions can mimic melanocytic neoplasms on routine sections and immunohistochemically. HMB-45 is the most-sensitive marker, but Melan A and MiTF are also expressed in many cases. PEComas differ from melanocytic lesions in absence of S-100 expression.89
Heavily pigmented melanocytic neoplasms are difficult to assess on routine hematoxylin-eosin (H&E)–stained slides because pigmented melanocytes can be confused with pigmented keratinocytes and melanophages. Immunostaining using diaminobenzidine, which forms a brown product, as the chromogen is challenging to distinguish from melanin. Alternatives include use of a red chromogen or use of melanin bleach; however, the bleaching may result in loss of antigenicity, incomplete melanin removal, or loss of cytologic detail.
Junctional Melanocytic Proliferations
It can be difficult to reliably identify intraepidermal melanocytes in H&E–stained sections, especially on sun-damaged skin. Even close inspection may not unequivocally discriminate pigmented keratinocytes from melanocytes or clearly delineate the melanocyte density. Thus, many pathologists employ IHC to differentiate melanoma in situ from its mimics. S-100 also highlights intraepidermal Langerhans cells making it less favored. MART-1/Melan A has historically been valued in this setting, but studies90-92 have shown that those markers may artificially increase the perceived number of melanocytes because of the labeling of cytoplasmic dendrites encircling neighboring keratinocytes. Available nuclear melanocytic markers are sensitive, specific, and circumvent the problem of cytoplasmic dendrites.91 SOX-10 is such a nuclear marker, with the same or better sensitivity as S-100, and it does not stain additional distracting cells, such as Langerhans cells of the epidermis.91 Nuclear expression of MiTF has similar benefits to SOX-10 but is much less sensitive in identifying subtle underlying DM. Caution is required with initial use of nuclear melanocytic markers and should be performed in parallel with cytoplasmic markers until one is comfortable with the altered staining pattern. Confluence, as seen with cytoplasmic markers, may not be as obvious at the outset because nuclear markers lack reactivity in the juxtaposed melanocytic cytoplasm.
Intraepidermal lesions with a pagetoid distribution can be diagnostically difficult. CK7 positivity reportedly differentiates Paget disease of the breast and extramammary Paget disease (EMPD) from pagetoid Bowen disease and melanoma in situ; however, the marker is not 100% sensitive, and occasional cases of CK7+ pagetoid Bowen disease have been described.93,94 Similarly, CAM 5.2 expression favors Paget disease or EMPD, rather than SCC in situ, but reactivity has been reported in both.94,95 Sellheyer and Krahl96 proposed the addition of Ber-EP4 to the panel used to evaluate pagetoid cutaneous tumors. In their study,30 Ber-EP4 labeled all cases of EMPD but failed to label Bowen disease and melanoma in situ. Other authors97 have identified strong nuclear p63 in pagetoid SCC in situ, in contrast to no reactivity in EMPD. Several studies have attempted to use IHC to differentiate primary from secondary EMPD, but the immunoprofiles tend to overlap, including occasional CK20 and prostate specific antigen (PSA) expression in primary EMPD.98 Thus far, CDX-2 reactivity has only been reported in EMPD secondary to anorectal adenocarcinoma.
Sclerosing Epithelial Neoplasms
Differentiation of sclerosing epithelial neoplasms is not only of academic interest but also paramount to clinical management, and diagnosis is often requested on small, superficial biopsies. Various immunohistochemical markers have been studied to help distinguish desmoplastic trichoepitheliomas (DTE), infiltrating or morpheaform BCCs, and microcystic adnexal carcinomas, including CD23, CD5, CD10, CD34, CK20, CK15, stromelysin-3, BCL2, AR, pleckstrin homology-like domain, family A, member 1 protein (PHLDA1), p75 neurotrophin receptor (p75NTR), fibroblast activation protein (FAP), Ber-EP4, p63, and others.99-103 Much of the available data are based on a few cases reported by a single group of authors and require additional validation. Conflicting results were found with the same marker in other studies. A few deserve further mention, but the histopathologic criteria and clinical data remain the current gold standard.
The panel of AR and CK20 has proven useful in distinguishing DTE and morpheaform BCC.100, 104 CK20+ Merkel cells are identified as colonizing DTEs but not BCCs and microcystic adnexal carcinomas.105 However, the Merkel cells identified in DTE may be few, requiring serial sections, and care is needed to avoid misinterpretation of Merkel cells in preexisting vellus follicles. Nuclear AR expression is focally present in most morpheaform BCCs and is absent in most DTEs.100, 104 AR reactivity has also been noted in classic or conventional trichoepitheliomas.106
Several studies have attempted to identify immunomarkers to differentiate conventional trichoepithelioma or trichoblastoma from nonaggressive forms of BCC. Although there are supporting data for use of CD34 and BCL2 in this setting, minimal research has been conducted evaluating those markers in the setting of sclerosing epithelial neoplasms. As with conventional trichoepitheliomas, Kirchmann et al107 found the stroma of DTEs was positive for CD34, differentiating it from the negative staining stroma of morpheaform BCC and microcystic adnexal carcinoma; however, Costache et al104 failed to identify stromal staining in any of their 19 DTEs. BCL2 tends to show expression only at the periphery of the tumor islands in conventional trichoepitheliomas but is present throughout the tumor island in BCC, yet the small islands of DTE and morpheaform BCC make that distinction nebulous, as seen in the study by Costache et al.104
Recent studies have evaluated the utility of p75NTR (CD271) and PHLDA1. Morpheaform BCCs tend to lack expression of p75 in the tumor cells or only show focal or weak expression, whereas DTEs are strongly and diffusely positive.108, 109 p75 is typically expressed in the outer root sheath of follicles; therefore, reactivity in DTEs supports their classification as a follicular hamartoma mimicking that portion of the follicle. Microcystic adnexal carcinomas cannot be reliably distinguished with p75 because nearly one-half are strongly positive.108 PHLDA1 is a hair-follicle stem-cell marker in the bulge area of the follicle. Reactivity for that marker in more than 50% of tumor cells characterizes DTEs versus the lack or minimal expression seen in infiltrating or morpheaform BCCs.102, 103, 110 Colonizing melanocytes and tumor cells in proximity to an ulcer can be positive requiring caution in interpretation.102 As noted above with CD34, some studies have concentrated on tumoral stroma when attempting to differentiate these tumors.
Prognostic indicators of melanoma on H&E–stained sections include ulceration and tumor depth, but there has been increasing study into the use of IHC-assisted prognostication. Identification of lymphatic invasion can be increased with the use of D2-40, a lymphatic endothelial marker, or CD34, and recognition is associated with decreased disease-free and overall survival.111-113 The presence of dermal mitoses in melanoma is a critical variable in management and prognostic considerations. Identification of mitoses has become increasingly important in staging melanoma in the 2010 American Joint Committee on Cancer (AJCC) guidelines, as identification of one mitotic figure in a melanoma less than 1 mm thick will alter the stage.114 However, it can be time consuming to search for mitoses, and distinguishing apoptotic, pyknotic, and hyperchromatic nuclei from mitotic figures can be difficult and subjective. Histone H3 is phosphorylated during mitotic chromatin condensation in the late G2 and M phases of the cell cycle. Immunohistochemical staining of phosphohistone H3 (PHH3) facilitates mitotic counting, even at 200X magnification, in contrast to the 400X magnification used when counting on H&E. PHH3 is specific to the M phase of the cell cycle, unlike MIB-1, and is not expressed in apoptotic cells. PHH3 aids identification of the mitotic “hot spot”, shortens the time required to determine the mitotic count, and improves reproducibility beyond counts on standard H&E–stained sections.115 It is not surprising that PHH3-determined counts are higher than those from H&E–stained sections because the prophase that comprises a large fraction of the entire mitotic phase is essentially undetectable on standard sections but is well highlighted with PHH3. The PHH3 count may also be falsely elevated by misinterpretation of nonmelanocytic mitotic figures. PHH3- determined mitotic counts have proven to be a stronger prognostic indicator than mitotic count determined from H&E–stained sections.116 Although use of PHH3 improves accurate identification of mitoses and is known to show an association with survival,117 the current melanoma staging criteria are based on a mitotic rate determined from H&E–stained slides. PHH3-based cutoff levels for (stage) pT1b will need to be determined.
SOX-10 is a highly sensitive marker for metastatic melanoma in sentinel lymph nodes but, unlike S-100, is not expressed in follicular dendritic cells or sinus histiocytes, which improves the identification of small micrometastases and isolated cells . HMB-45 can differentiate metastatic melanoma (when HMB-45 +) from nodal nevi ( HMB-45-). Unlike HMB-45, SOX-10 does not differentiate nodal nevi from metastases, and distinction requires attention to the nodal location, cytology, and comparison to the primary.
The small blue cell appearance of Merkel cell carcinoma (MCC) elicits a differential diagnosis including Ewing sarcoma/primitive neuroectodermal tumor (EWS/PNET), metastatic small cell carcinoma of the lung, neuroblastoma, lymphoma, and melanoma. Small cell melanoma can be distinguished immunohistochemically because MCC does not typically express S-100 but does express neuroendocrine markers, such as chromogranin, neuron-specific enolase (NSE), and synaptophysin, and displays paranuclear dot staining with neurofilament protein and CK20. Although metastatic small cell carcinoma of the lung is typically negative for CK20 and neurofilament protein, it is classically positive for CK7 and thyroid transcription factor 1 (TTF-1), a finding not usually seen in MCC.118, 119 CK20 is present in most MCCs (87%), whereas CK7 is rarely identified.119 Despite the classic CK20+/CK7- pattern, CK20-/CK7+ MCCs have been described.119, 120 It is important to be aware of the staining pattern of the primary tumor when evaluating sentinel lymph nodes if IHC is needed. The absence of CD45/LCA in MCC is typically sufficient to exclude lymphoma, but some markers used in evaluation of lymphoma can be detected in MCC, leading to erroneous diagnoses, including ALK1 (depending on the clone used) as in anaplastic large cell lymphoma (ALCL), BCL2 as in B-cell lymphomas, TdT and PAX-5 as in B-lymphoblastic lymphoma, and CD56 as in natural killer T-cell lymphoma.119,121-124
As in MCC, EWS/PNET also has the potential to express NSE, chromogranin, and synaptophysin. CD99 is a marker used for EWS/PNET, but it is not specific and has been seen in lymphoblastic lymphoma, selected rhabdomyosarcomas, small cell carcinomas, carcinoid tumors, melanomas, and even a few MCCs.125-128 The FLI-1 antibody is also a nuclear marker for EWS/PNET, as well as for vascular tumors. However, it is also expressed in subsets of lymphomas, MCCs, and melanomas.127, 129-132 Pankeratin reactivity is usually absent in EWS/PNET, and CK20 expression has not been reported, unlike MCCs. The surest way to distinguish EWS/PNET from MCC is cytogenetic analysis revealing t(11;22), usually involving EWS and FLI-1 genes, in EWS/PNET.128, 133
The c-kit proto-oncogene codes for a transmembrane receptor tyrosine kinase (KIT receptor/CD117). CD117 is expressed in a variety of tumors, including acute myeloid leukemia, mast cell disease, melanoma, small cell lung cancer, and gastrointestinal stromal tumors. CD117 is expressed by most MCCs.134, 135 Although KIT expression is present in MCC, response to imatinib mesylate has been poor and anecdotal.135-137 Rarely, MCC can mimic BCC, and both can and often do express Ber-EP4, BCL2, and NCAM/CD56. In addition, BCC, on occasion, can express chromogranin and synaptophysin.
Nerve sheath myxomas, believed to be of peripheral nerve sheath origin, were redesignated as neurothekeoma by Gallager and Helwig138 in 1980. In 1986, the more-cellular form was designated cellular neurothekeoma (CNT) by Rosati et al.139 Currently, there are 3 recognized subtypes of neurothekeoma: myxoid, cellular, and mixed. Controversy exists about the lineage of CNT, with arguments for myofibroblastic, nerve sheath, melanocytic, and leiomyomatous differentiation. Immunoreactive markers span a spectrum of cell types, including focal staining with SMA, CD68, and CD10.140, 141 Protein gene product 9.5 (PGP 9.5, also known as ubiquitin carboxyl-terminal hydroxylase-1) reactivity can be helpful in the diagnosis of neurothekeoma but also suffers from low specificity.142 PGP 9.5 is positive in nerve sheath tumors, including granular cell tumors; fibroblastic tumors, including dermatofibromas, vascular tumors, and other tumors, such as leiomyomas.143, 144 D2-40, as discussed previously, is expressed in lymphatics and PCANs, but reactivity has also been reported in the 15 CNTs thus far studied.146 Cellular neurothekeomas and fibrous histiocytomas can share morphologic features, and both have shown expression of D2-40, further complicating differentiation.
Although myxoid neurothekeomas are S-100+, CNTs are not, arguing strongly against a peripheral nerve sheath or melanocytic histogenesis. However, CNTs are positive for S100A6 protein (calcyclin).142, 146 Reactivity with that protein is not equivalent to a diagnosis of CNT because it is also expressed in melanocytes and dermal dendrocytes.142 Cellular neurothekeomas also can be confused with melanocytic lesions, histologically. When considering this differential diagnosis, S100A6 reactivity, with negative S-100 staining, essentially eliminates a melanocytic lesion. S-100 expression of colonizing antigen-presenting cells can be seen and should not be confused with lesional cell reactivity, causing an erroneous diagnosis. Although the typical melanocytic markers, S-100, HMB-45, and Melan A are absent in CNT, other less-specific melanocytic markers, such as S100A6 have been identified.141, 142 Many CNTs express NKI-C3 (CD63), but that marker lacks specificity and has been identified in nevi, melanomas, granular cell tumors, and some fibrohistiocytic lesions.142, 147 Expression of KBA.62, an antimelanoma monoclonal antibody, has recently been reported148 in 18 of 18 CNTs (100%). MiTF is also a melanocytic marker that highlights most CNTs.
Dermatofibroma and Dermatofibrosarcoma Protuberans
Most dermatofibromas (DFs) can be easily distinguished from dermatofibrosarcoma protuberans (DFSPs); however, morphologic features alone do not always allow reliable distinction between deep or cellular DF and DFSP. Commonly, CD34 and factor XIIIa have been used in differentiation. Typically, DFSP is CD34+ and factor XIIIa-, and DF is CD34- and factor XIIIa+; however, that is not absolute.149-151 Some DFs exhibit focal staining with CD34, especially at the periphery of cellular and deep DFs, and occasional DFSPs are negative for CD34.152 Caution is required in interpretation of DFSP margins with CD34 because CD34 disappears from scars but proliferates in pericicatricial tissue.153 CD34 positivity is not exclusive to DFSP. It is also an endothelial marker and stains solitary fibrous tumor, spindle cell lipoma, superficial acral fibromyxoma, sclerotic fibroma, Kaposi sarcoma, neurofibroma, tricholemmoma, scleromyxedema, and nephrogenic systemic fibrosis, among others.154-156 Factor XIIIa is typically weak and only identified at the periphery of DF or more diffusely in spindle cell–dominant cellular DFs.150 Many alternative markers have been explored but mostly in isolated studies with few cases. Those markers include HMGA1 and 2, CD163, ApoD, tenascin, S100A6, MMP-2 and -11, IGFBP7, cathepsin K, and D2-40.
Nestin, a neuroectodermal and mesenchymal stem cell marker, was evaluated in a larger cohort of nearly 200 cases of DFSP and DF in 4 different studies,157-160 which showed similar results. Strong nestin expression was noted in DFSP, and no or only rare focal expression was seen in DFs. Unlike CD34, which can be absent or decreased in fibrosarcomatous areas of DFSP, nestin remains unaltered.160 Immunohistochemistry for stromelysin 3, a member of the matrix metalloproteinase family, MMP-11, has also been studied in at least 5 separate studies151, 161-164 with similar outcomes. Stromelysin 3 has a role in tissue remodeling during wound healing and tumor invasion, and there is diffuse cytoplasmic staining of the spindle cells in most DFs, whereas it has rarely been noted in DFSPs.
Several endothelial markers are available including: von Willebrand factor (factor VIII–related antigen), CD34, CD31 (platelet-endothelial cell adhesion molecule type 1), FLI-1, and ERG. Factor VIII–related antigen is specific for endothelial cells but has a low sensitivity, and because it circulates in the serum, it can be seen in zones of necrosis and hemorrhage. CD34 is sensitive but is also expressed in several nonvascular tumors, including DFSP and epithelioid sarcoma. CD34 has been considered a sensitive marker for Kaposi sarcoma, but Kaposi sarcoma–associated herpes virus 8 latent nuclear antigen (HHV-8) has supplanted use of CD34 in that context.1, 165 CD31 has been considered the most-sensitive marker for vascular endothelium and vascular tumors. It is expressed in endothelial cells but is not specific and has been reported in a subset of carcinomas, lymphomas, AFXs, epithelioid sarcoma, EWS/PNET, macrophages, platelets, and plasma cells.21, 166 More recently, FLI-1 and ERG were reported as nuclear markers of endothelial differentiation. Folpe et al167 found FLI-1 expression in 94% of benign and malignant vascular tumors, but FLI-1 is also a marker for EWS/PNET, with expression in approximately 90% of cases, and it shows expression in some melanomas, leiomyosarcomas, SCCs, and AFXs.91 Thus far, ERG has proven a sensitive and specific marker for vascular lesions, showing expression only in a subset of prostate carcinomas, EWS/PNET, epithelioid sarcomas, and myeloid sarcoma.168-170 Rao et al166 evaluated 34 angiosarcomas in a tissue microarray and found ERG and FLI-1 to be the most sensitive immunomarkers. Angiosarcomas often show loss of expression of one or more endothelial markers, especially epithelioid angiosarcomas, which can also show keratin expression, suggesting that a panel should be used in diagnosis.
Markers that distinguish blood vessel endothelial cells from lymphatic endothelial cells have also been investigated. Prox1, a transcription factor important in the regulation and maintenance of the lymphatic endothelial phenotype, is expressed in lymphangiomas, Kaposi sarcoma, tufted angiomas, and Kaposiform hemangioendothelioma.171 The monoclonal antibody D2-40 binds to podoplanin, a glycoprotein highly expressed in lymphatic endothelium. D2-40 has been identified as a sensitive marker for lymphatic endothelium in normal tissue and in a subset of vascular lesions, including Kaposi sarcoma, Dabska tumor, lymphangioma, hobnail hemangioma (targetoid hemosiderotic hemangioma) and in a subset of angiosarcomas, especially those with epithelioid endothelial cells. In general, classic hemangiomas do not express D2-40.172, 173 D2-40 is variably expressed in a wide variety of lesions, including sebaceous neoplasm, epithelioid sarcoma, AFX, SCCs, CNTs, DFs, and PCANs. Expression of lymphatic markers, including Prox1 and D2-40, in angiosarcoma suggests that it may show a mixed endothelial phenotype. 22, 166, 171 Infantile hemangiomas become apparent postnatally, enlarge rapidly during the first year of life, and spontaneously involute in the first decade. The characteristic histopathologic features diminish during that evolution, and differentiating them from vascular malformations can be difficult. Endothelial immunoreactivity for the erythrocyte- type glucose transporter protein (GLUT1), normally restricted to endothelia with blood-tissue barrier function, as in the brain and the placenta, is identified in all phases of infantile hemangiomas, whereas reactivity is absent in vascular malformations, pyogenic granulomas, and granulation tissue.174-176 Normal reactivity in erythrocytes must be discounted when interpreting the expression. Expression of GLUT1 has also been demonstrated in normal perineural cells and perineuriomas.177 These S-100-, benign peripheral nerve sheath tumors are also positive with EMA but often in a focal and weak pattern because of their very thin and widely separated cytoplasmic processes. Claudin-1, a component of tight junctions, and GLUT1 are stronger and stain more diffusely in perineurioma than does EMA.178 Infantile hemangiomas and vascular malformations have distinct clinical courses, and the distinction is important for management considerations.
Wilms tumor 1 (WT1) cytoplasmic endothelial expression has been reported in vascular tumors, including infantile hemangiomas, noninvoluting congenital hemangiomas, rapidly involuting congenital hemangiomas, tufted angiomas, pyogenic granulomas, microvenular hemangiomas, and cherry hemangiomas, but is lacking in lymphatic and venous vascular malformations.179-181 WT1 positivity was, however, reported in the 5 stage 2 arteriovenous malformations studied,180 possibly because of active proliferation and clinical enlargement during that stage. Current data are limited, but WT1 results are conflicting for verrucous hemangioma and hobnail hemangioma, further drawing into question the appropriate classification of those lesions as neoplasms or malformations.181-183 Another important distinction includes distinguishing rapidly involuting congenital hemangioma and noninvoluting congenital hemangioma from infantile hemangioma. Those lesions behave differently, as the names suggest, but have histologic features that overlap with infantile hemangioma. However, neither rapidly involuting congenital hemangioma nor noninvoluting congenital hemangioma expresses GLUT1.176, 184
Cutaneous B-Cell Lymphomas
CD20 is a B-cell–specific marker expressed in 98% of B-cell lymphomas, but it may be lost in the lymphomas treated with the anti-CD20 antibody (rituximab). CD79a is expressed in the precursor B-cell stage and disappears later than CD20 in B-lymphocyte differentiation, explaining why plasma cells are positive for CD79a but not for CD20. CD79a, therefore, stains most B-cell lesions, even rituximab- treated B-cell lymphomas. PAX-5 is a nuclear marker expressed in early B-cell development and is, therefore, negative in plasma cells, but it can stain recurrent B-cell lymphoma following rituximab therapy.185, 186 CD38 and CD138 (syndecan-1) are expressed in plasma cells.
Aside from lymphocytoma cutis, a significant B-cell infiltrate in the skin is rarely reactive and should suggest possible B-cell lymphoma. Architecturally, the infiltrate may be nodular or diffuse. The diffuse pattern includes follicle center cell lymphoma, diffuse large B-cell lymphoma (DLBCL) in both the leg type and the otherwise-not-specified type. The nodular pattern is seen in lymphocytoma cutis, marginal zone lymphoma, and a small proportion of follicle center cell lymphomas. Normal germinal centers are BCL6+, BCL2-, and CD10+ with a CD21+ and/or CD23+ follicular dendritic cell network and high proliferative index. Benign lymphoid follicles can be seen in lymphocytoma cutis and marginal zone lymphoma. The interfollicular component of marginal zone lymphoma is neoplastic, and immunohistochemically is BCL2+, whereas BCL6 and CD10 are negative. When there is plasmacytoid differentiation, light-chain restriction can be evaluated immunohistochemically; 2:1 is the normal K:λ ratio, but above 5:1 or a 3:1 λ:K ratio supports clonality. However, reproducibility for this ratio is low.187
Cutaneous follicle center cell lymphoma is of germinal center cell origin and is thus BCL6+ and BCL2-. If BCL2 expression is present, it suggests spread of the disease to the skin from the lymph nodes.185 T cells are BCL2+ and comprise a proportion of the lymphoid cells in a follicle center cell lymphoma. BCL2 expression must be interpreted in the neoplastic cells alone and must not be confused with those colonizing T cells. The leg-type DLBCL is also CD20+ and BCL6+, but unlike follicle center cell lymphoma, it expresses BCL2, MUM1, and FOXP1.188 MUM1 is also expressed in melanoma and anaplastic large cell lymphoma (ALCL).
Cutaneous T-Cell Lymphomas
Cutaneous lymphoproliferative disorders can be one of the most vexing problems in dermatopathology. Mycosis fungoides-type cutaneous T-cell lymphoma (MF) must be differentiated from reactive T-cell infiltrates. Polymerase chain reaction analysis of T-cell receptors for clonality has become helpful in this context; however, many cases of early MF are polyclonal by PCR, and monoclonality has been identified in several cases of benign dermatoses.189 Like benign inflammatory infiltrates, most cases of MF have a CD4+ T helper cell phenotype but, in contrast, may show loss of expression of other T-cell markers in the epidermal component. The most frequent down-regulated antigen is CD7, closely followed by CD5.187 However, that down regulation is not absolute, and some benign inflammatory infiltrates, especially acute ones, show loss of CD7.190 Therefore, interpretation of CD7 should be made in the context of the clinical and histologic features. The relative absence of a pan–T-cell marker CD3, CD5, or CD43, or the concordant expression or loss of both CD4 and CD8, are more likely to be seen in MF than in benign conditions. Some studies have suggested that a CD4:CD8 epidermal ratio greater than 2 supports MF, rather than an inflammatory process.191 However, there is significant overlap, with inflammatory conditions showing a ratio up to 6.192 Ortonne and colleagues193 attempted to improve diagnosis of early MF and found an epidermal CD8:CD3 ratio of less than 25% to be suggestive, but not specific, for MF. Use of CD3, rather than CD4, obviates concern about misinterpretation of CD4+ Langerhans cells. In practice, application of these ratios is not reproducibly measured for precise diagnosis. In contrast, there is a subset of CD8+ MF, which is most common in children and hypopigmented cases.187 The International Society for Cutaneous Lymphoma proposed an algorithm for diagnosis of early MF, based on a scoring system involving clinical, histopathologic, biomolecular, and immunopathologic criteria. The immunohistologic features included in the score are CD2, CD3, or CD5 positivity in less than 50% of T cells, CD7 positivity in less than 10% of T cells, and epidermal/dermal discordance in expression of CD2, CD3, CD5, or CD7 expression.194
Subcutaneous panniculitis-like T-cell lymphoma is a CD8+ cytotoxic T-cell neoplasm, by definition, of the alpha-beta receptor class and is, therefore, β-F1+, distinguishing it from cutaneous gamma-delta lymphoma. Subcutaneous panniculitis-like T-cell lymphoma is typically positive for TIA-1, granzyme B, and perforin. The adipocyte rimming by T cells, particularly when they are Ki-67+, is a useful feature in diagnosis.187
IHC currently provides a rapid morphologic diagnosis of infections in tissue samples from patients, thus facilitating clinical decisions in patient care. In infectious disease diagnosis, IHC has been shown to be useful in identifying microorganisms that are (1) difficult to detect by routine or special stains, (2) stain poorly, (3) present in low numbers, or (4) noncultivable. Although several antibodies have been developed to aid in the immunohistochemical diagnosis of many infectious diseases, only a small subset of these antibodies is useful in dermatopathology for multiple reasons.195-197 The use of IHC in the diagnosis of cutaneous infections has been applied to confirm the diagnosis of specific viral and bacterial infections that may be difficult to diagnose with certainty using routine microscopy alone.
IHC and Viral Pathogens
IHC can be quiet useful in herpes simplex virus (HSV) detection and identification, especially in the case of atypical cutaneous manifestations. For example, IHC was able to identify HSV infection in 5 bedridden geriatric patients (type I in 3 and type II in 2) with genital ulcers, when histology was suggestive of HSV infection in only 2 of the 5 patients.197, 198 One study showed that the sensitivity and specificity of IHC was comparable with ISH in diagnosing HSV.199 Despite the high sensitivity of the polyclonal antibodies used, they do not differentiate between the antigenically similar HSV I and HSV II.7 Similarly, IHC has also been shown to have higher specificity and sensitivity than standard microscopic assessments in diagnosing varicella zoster virus (VZV) infection (varicella and herpes zoster) through detection of VZVORF63 encoded protein (IE63) and VZV late protein gE on both smears and formalin-fixed paraffin-embedded skin sections.200 This can be of special significance in allowing early diagnosis of VZV infection in immunocompromised patients and thus early treatment.
In addition to its presence in various benign (inflammatory pseudotumor) and malignant (non-Hodgkin lymphoma, Hodgkin disease, nasopharyngeal carcinoma, gastric carcinoma) lesions, Epstein-Barr virus (EBV) causes oral hairy leukoplakia and infectious mononucleosis.201 IHC has long been used in the diagnosis of EBV infections through identification of EBV-expressed proteins, which also allow distinction of latent from replicative infection based on expression profiles. Actually, LMP1 immunostain (cytoplasmic and membranous localization) has been shown to be nearly as effective as EBV-encoded RNA ISH in identifying EBV in lymph nodes of patients with infectious mononucleosis. Detection of other EBV proteins such as BZLF1, EBNA1, EBNA2, and LMP2A can be achieved by IHC.201
Cytomegalovirus (CMV) infection is especially common in immune-compromised patients (such as HIV-infected patients), and the cutaneous manifestations can be quiet variable including ulcers, vesicles, papules, purpuric macules, verrucous lesions, prurigo nodularis–like lesions, and digital infarcts. Although the sensitivity of light microscopy is high in the detection of CMV inclusions mostly within endothelial cells and, occasionally, within keratinocytes, sweat gland epithelial cells, macrophages, and fibroblasts, immunohistochemical analysis may be needed to confirm the diagnosis in certain cases such as the presence of intense inflammation or in cases where only 1 or 2 cells are infected.202 IHC allows for rapid diagnosis and its sensitivity is better than light microscopy and is comparable with culture and ISH.
Human herpesvirus 8 (HHV-8) is a major cause of Kaposi sarcoma (KS). Through positive nuclear immunostaining for HHV-8 latent nuclear antigen-1 in spindle cells and cells lining the primitive and thin-walled vascular channels, immunohistochemical detection of HHV-8 in paraffin-embedded sections has been shown to be highly sensitive and specific for the diagnosis of KS and thus allowing its differentiation from its mimickers including angiosarcoma, kaposiform hemangioendothelioma, spindle-cell hemangioma, among others.203, 204 Immunolocalization in the nuclei was considered positive.
IHC and Bacterial Pathogens
During routine evaluation, the use of Steiner or Warthin-Starry silver stains to confirm the diagnosis of suspected syphilis is of low sensitivity (range from 33% to 71%), as it is usually hampered by significant background staining.204-207 Several studies have shown that immunohistochemical staining provides better sensitivity than silver stains for the detection of Treponema pallidum (T pallidum).205-207 Using a monoclonal antibody, Hoang et al205 showed positive staining in 12 of 17 (71%) biopsy specimens taken from patients with secondary syphilis compared with 41% detection using silver stain. In another study in which Buffet et al206 used a polyclonal antibody to T pallidum, the spirochetes were detected in 11 of 12 (91%) skin biopsy specimens taken from patients with secondary syphilis. Recently, Mart´ın-Ezquerra et al207 studied the sensitivity of silver staining and immunohistochemical T pallidum polyclonal antibody expression in 34 biopsy specimens of 8 patients with primary syphilis and 26 with secondary syphilis. Immunohistochemical analysis showed greater sensitivity than Warthin-Starry staining, as spirochetes were identified in 29 of 34 (85%) cases (8 of 8 primary and 21 of 26 secondary syphilis) using IHC, although Warthin-Starry staining disclosed spirochetes in 17 of 34 (50%) biopsy specimens (4 of 8 primary and 13 of 26 secondary syphilis). In this study, the authors, using IHC, were also able to identify different distribution patterns of T pallidum in primary and secondary syphilis. Although primary syphilis showed both an epitheliotropic (abundant spirochetes in an intercellular distribution in the lower mucosa/epidermis) and vasculotropic (treponemes surrounding the vascular walls) patterns, secondary syphilis only showed an epitheliotropic pattern. The authors concluded that the immunohistochemical distribution pattern of T pallidum may permit diagnostic differentiation of primary from secondary syphilis.207