Undifferentiated Tumor True Identity by Immunohistochemistry

Armita Bahrami, MD; Luan D. Truong, MD; Jae Y. Ro, MD, PhD

● Context.—‘‘Undifferentiated tumor’’ refers to a hetero- carcinomatous nature of tumor. A diagnostic misinterpre- geneous group of neoplasms with little or no evidence of tation as a consequence of occasional aberrant or unex- differentiation on routine light microscopic morphology. pected antigen expression is best avoided by using a broad Objective.—To identify the true identity of undifferen- panel that includes both antibodies that are anticipated to tiated tumors by immunohistochemical analysis. be positive and those that are expected to be negative. In Data Sources.—Review of the pertinent literature and this treatise, the immunohistochemical dissection of undif- the authors’ experience. ferentiated tumors on the basis of their morphologic fea- Conclusions.—For treatment and prognostic evaluation, tures is outlined, supplemented with algorithmic immu- it is crucial to delineate whether an undifferentiated neo- nohistochemical analysis for each morphologic category of plasm is epithelial, mesenchymal, melanocytic, or hema- small round cell tumors, carcinomatous tumors, sarcoma- topoietic in nature. Application of a screening panel to tous (or sarcoma-like) tumors, and tumors with histologi- demonstrate the expression of markers of major lineages is cally overlapping features, including hematolymphoid ma- fundamental for determination of the broad category of lignancies, melanoma, and sarcomas with epithelioid ap- neoplasia. Because poorly differentiated carcinomas and in pearance. The utility of several organ- or tissue-specific particular sarcomatoid carcinomas are known to be het- markers in the context of undifferentiated tumors is re- erogeneous in their antigen expression, several epithelial viewed. markers in combination may be required to establish the (Arch Pathol Lab Med. 2008;132:326–348)

he term undifferentiated tumor has been used in refer- tastasis of unknown origin are not included in this dis- T ence to a heterogeneous group of tumors with little cussion. or no evidence of differentiation. Some may link this ter- For treatment purposes, it is crucial to determine wheth- minology to morphologically undifferentiated neoplasms er an undifferentiated neoplasm is epithelial, mesenchy- that cannot be otherwise classified, even with the appli- mal, or hematopoietic. In general, the diagnosis of lym- cation of immunohistochemistry. In our view, however, phoma for an undifferentiated tumor predicts a better such tumors are extremely rare and in most instances fur- clinical outcome compared with that of carcinoma.1 The ther sampling or application of ancillary tests should help value of immunohistochemical procedures for identifica- to recognize them as a specific tumor type. For such rea- tion of the true identity of undifferentiated tumors has son, in this review we apply the term undifferentiated to been proved by studies in which approximately 90% of tumors lacking evidence of lineage differentiation on the tumors posing diagnostic difficulties by morphology basis of routine light microscopic morphology alone. could be accurately classified by exploiting immunohis- An undifferentiated malignant tumor represents either tochemistry.1–3 a metastasis of unknown origin or a primary neoplasia Even in undifferentiated tumors, subtle features of ep- without obvious cell line of differentiation. It should be ithelial versus mesenchymal differentiation can often be noted that undifferentiated tumor generally implies a appreciated, which assist the immunohistochemical ap- high-grade malignancy, frequently associated with pleo- proach to these tumors. Hints for epithelial differentiation morphic to anaplastic appearance. Therefore, low-grade include epithelioid cells (round to oval cells) with nesting neoplasms but without an obvious lineage of differentia- arrangement and a desmoplastic stroma with feeding ves- tion (eg, monomorphic spindled cell tumors) or low-grade sels separating tumor cell nests (Figure 1). In contrast, tumors not infrequently encountered in the context of me- mesenchymal differentiation is suggested by a diffuse ar- rangement of spindled cells (Figure 2), without reactive Accepted for publication June 4, 2007. stroma, but with feeding vessels in between tumor cells. From the Departments of Pathology, Baylor College of Medicine (Drs Some tumors, however, may not fit into either of these 2 Bahrami and Truong) and Weill Medical College of Cornell University, categories because of their overlapping histologic features The Methodist Hospital (Drs Truong and Ro), Houston, Tex. (Figure 3), for example, sarcomatoid carcinoma, melano- The authors have no relevant financial interest in the products or ma, lymphoma, neuroendocrine tumors, and sarcoma companies described in this article. Reprints: Jae Y. Ro, MD, PhD, Department of Pathology, Weill Med- with epithelioid cells. ical College of Cornell University, The Methodist Hospital, 6565 Fannin Immunohistochemical dissection of undifferentiated tu- St, Houston, TX 77030 (e-mail: [email protected]). mors is also helped by categorizing them into small round 326 Arch Pathol Lab Med—Vol 132, March 2008 Undifferentiated Tumor, Immunohistochemistry—Bahrami et al blue cell tumors (SRCTs) or large cell tumors. The latter group is further divided into (1) carcinomatous tumors, (2) sarcomatous or sarcoma-like tumors, and (3) tumors with overlapping features. Each category entertains a broad list of entities from epithelial, mesenchymal, he- matopoietic, or melanocytic lineage in the differential di- agnosis. In the following section, the immunohistochemical pro- cedure for a broad lineage determination of undifferenti- ated tumors is discussed, followed by immunohistochem- ical analysis of each individual category of SRCTs, carci- nomatous tumors, sarcomatous (or sarcoma-like) tumors, and tumors with overlapping features, supplemented with diagnostic algorithms. It is emphasized that the out- lined algorithmic immunohistochemical approach is nei- ther meant to be comprehensive nor intended to be an absolute method for immunohistochemical dissection of these tumors. In reality, each tumor requires an ‘‘individ- ually constructed panel’’ composed of carefully selected antibodies that recognize all reasonable diagnostic possi- bilities in the context of the tumor’s morphology, anatomic site, and clinical/radiologic findings. BROAD LINEAGE DETERMINATION The immunohistochemical evaluation of undifferentiat- ed tumors should first aim at determination of the broad category of neoplasia, that is, carcinoma, sarcoma, lym- phoma, or melanoma. A screening panel to demonstrate the expression of markers of major lineages (ie, epithelial, mesenchymal, lymphoid, and melanocytic) often provides the first clue to the nature of an undifferentiated tumor. In certain circumstances, adjuvant immunostains are add- ed; thus, placental alkaline phosphatase (PLAP) and OCT3/4, markers for germ cell tumors (GCTs), may be included for tumors in younger men in view of the high incidence of GCTs in this age group. Based on the result of the screening panel, a more detailed or specific panel is commonly followed to further subclassify the tumor or confirm a particular diagnosis. Screening Markers for Epithelial Lineage Cytokeratins. The low-molecular-weight cytokeratins (LMW CKs), including CK8, CK18, and CK19, recognized by the antibodies CAM 5.2 or 35BH11, and a cocktail of keratins (pankeratin), recognized by the antibody AE1/ AE3, are useful screening markers for the recognition of epithelial differentiation. Because poorly differentiated carcinomas are known to be heterogeneous in their ex- pression of antigens recognized by epithelial markers, when a negative result is obtained with a single antibody and the diagnosis of carcinoma is still suspected, it is pru- dent to use additional antibodies in a backup panel. In a study of 98 poorly differentiated carcinomas, CAM 5.2 and epithelial membrane antigen (EMA) each detected epithe- lial differentiation of 71% of the cases, whereas a com- bined CAM 5.2 and EMA elucidated the epithelial nature Figure 1. Photomicrograph of undifferentiated carcinomatous tumor of 99%.4 composed of epithelioid cells with nesting arrangement and a des- Sarcomatoid carcinomas are in particular known to be moplastic stroma separating tumor cell nests (hematoxylin-eosin, orig- unpredictable in regard to their CK expression; therefore, inal magnification ϫ10). Figure 2. Photomicrograph of undifferentiated sarcomatous tumor composed of spindled cells with a diffuse arrangement with no reactive ← stroma in between tumor cells (hematoxylin-eosin, original magnifi- cation ϫ20). Figure 3. Photomicrograph of undifferentiated tumor with overlap- ping histologic features, displaying epithelioid cells without nesting ar- rangement (hematoxylin-eosin, original magnification ϫ20). Arch Pathol Lab Med—Vol 132, March 2008 Undifferentiated Tumor, Immunohistochemistry—Bahrami et al 327 it is sometimes required to use several markers in com- Table 1. Malignant Neoplasms With Coexpression of bination to detect the epithelial differentiation of these tu- Vimentin and Cytokeratin mors (see ‘‘Sarcomatoid Carcinoma’’). In addition, CK Carcinoma may not be detectable in certain carcinomas; for example, Renal cell carcinoma it is known that CK expression in adrenocortical carcino- Endometrial adenocarcinoma mas is often diminished to levels too low to be recognized Serous ovarian carcinoma following the deleterious effects of fixation.5–7 Thyroid carcinoma, papillary and anaplastic Cytokeratin expression, most frequently with LMW Mesothelioma (biphasic) CKs, has been occasionally described in a variety of sar- Sarcomatoid carcinoma (spindle cell carcinoma) comas8,9 and rarely in hematopoietic malignancies10–14 and Sarcoma melanocytic lesions.9,15,16 Aberrant CK staining in nonepi- Synovial sarcoma thelial malignancies, however, has a weak, focal, and Desmoplastic small round cell tumor patchy staining, which contrasts with the generally diffuse Malignant rhabdoid tumor and strong staining seen in carcinomas or sarcomatoid Epithelioid sarcoma Epithelioid angiosarcoma carcinomas. Conversely, a true strong CK expression, which is frequently associated with a morphologic ap- Carcinosarcoma pearance of epithelial differentiation, is also seen in cer- tain sarcomas8 (see ‘‘Sarcomas With Epithelioid Appear- ance’’). Hence, caution should be taken to avoid overinter- quently demonstrating vimentin/CK coexpression is pro- pretation of CK expression as a specific feature of epithe- vided in Table 1. lial tumors. Epithelial Membrane Antigen. As mentioned previ- Screening Markers for Malignant Melanoma ously, EMA may be used as a supplement to CKs for de- S100 Protein. S100 protein is regarded as a screening tection of epithelial differentiation, especially in sarco- marker for melanoma with more than 95% sensitivity in matoid carcinoma or those undifferentiated carcinomas primary and metastatic sites. A valid positive S100 re- that are negative or only focally positive for CKs. A num- quires both nuclear and cytoplasmic staining. S100 pro- ber of epithelial tumors, including GCTs and some endo- tein, however, is also expressed in various other lesions, crine neoplasms such as medullary carcinoma of thyroid including peripheral nerve sheath, granular cell, cartilag- and adrenocortical carcinomas, lack immunoreactivity for inous and salivary gland tumors, chordomas, Langerhans EMA.17 cell histiocytosis, and occasional adenocarcinomas to vary- Epithelial membrane antigen is not entirely specific for ing degrees.25 Thus, to confirm the melanocytic nature of carcinomas. Epithelial membrane antigen expression has an S100-positive neoplasm, the tumor should be also pos- been seen in some normal and neoplastic hematopoietic itive for one or more melanocyte-specific protein (eg, cells, including reactive and neoplastic plasma cells,12 lym- HMB-45 or MART-1/Melan-A). phocytic and histiocytic (L&H) cells in nodular lympho- HMB-45. HMB-45 is a quite specific marker for mel- cyte predominant Hodgkin lymphoma,18,19 and neoplastic anoma, labeling 90% to 100% of conventional primary cells in some T-cell lymphomas17,19,20; thus, most anaplastic melanomas. The positivity rate declines to 80% in recur- lymphoma kinase (ALK)–positive anaplastic large cell rent or metastatic melanomas and spindle cell melano- lymphomas (ϳ75%) are EMA positive.18,21 mas.26 Desmoplastic melanomas are essentially negative for HMB-45 or other specific melanoma markers.27 HMB- Screening Marker for Mesenchymal Lineage 45 positivity is typically not observed in carcinomas, lym- Vimentin. Vimentin is the sole intermediate filament phomas, or sarcomas. characteristic of mesenchymal cells and present in virtu- ally all sarcomas and melanomas and variably in lympho- Screening Marker for Hematopoietic Malignancies mas.22 The antigenicity of vimentin is best preserved in CD45. Although CD45 (leukocyte common antigen frozen and alcohol-fixed tissues.22,23 Vimentin, however, [LCA]) is known to have high sensitivity (97%) and spec- suffers poor specificity for mesenchymal neoplasms be- ificity (nearly 100%) for lymphoid tumors,28,29 exceptions cause it may be coexpressed with CK in a wide range of have been documented. For example, CD45 is undetect- carcinomas (Table 1).22–24 Therefore, vimentin by itself can- able in most lymphoblastic lymphomas29–32 and is variably not be used to differentiate mesenchymal from nonmesen- expressed in plasma cell neoplasms29,33–35 and anaplastic chymal neoplasms. large T-cell lymphomas.36,37 CD45 immunoreactivity has been considered exquisitely specific for hematopoietic Vimentin/CK Coexpression cells28,29; yet, there are exceptional reports of CD45 ex- Frequent coexpression of vimentin with CK is seen in pression in undifferentiated or neuroendocrine carcino- some carcinomas, for example, renal cell, endometrial, mas.38 papillary and anaplastic thyroid, and ovarian serous (var- iably) carcinomas.22,23 Consistent absence of vimentin, on Aberrant or Unpredicted Antigen Expression the other hand, is observed in colonic, small intestinal, and As stated in several previous examples, an aberrant or prostatic adenocarcinomas or in transitional cell carcino- unexpected antigen expression should be considered as a mas.22 The finding of vimentin/CK coexpression helps fo- source of diagnostic pitfall in the surgical pathology eval- cus on certain types of epithelial tumors as possible pri- uation of undifferentiated tumors. A diagnostic misinter- mary sites in the evaluation of metastatic tumors. On the pretation is best avoided by using a broad panel that in- other hand, a number of sarcomas, in particular those cludes both antibodies that are anticipated to be positive with epithelioid appearance, also express CK in addition and those that are expected to be negative. The occasional to vimentin. A selected list of malignant neoplasms fre- aberrant immunophenotyping must not undermine the 328 Arch Pathol Lab Med—Vol 132, March 2008 Undifferentiated Tumor, Immunohistochemistry—Bahrami et al Table 2. The Immunoprofile of Small Round Cell Tumors* Myogenin/ CK CD45 S100 CD99 Desmin Myo-D1 CD56 WT1 Ewing sarcoma/primitive neuroectodermal tumor ഠ ϪϪϩϪϪϪ† Ϫ Rhabdomyosarcoma ϪϪϪϪ† ϩϩϩϪ‡ Desmoplastic small round cell tumor ϩϪϪഠ ϩϪഠ ϩ Poorly differentiated synovial sarcoma ϩ§ ϪϪ† ϩϪϪϩϪ Mesenchymal chondrosarcoma ϪϪϪ ϩ ഠ Ϫ ഠ Ϫ Wilms tumor ഠ ϪϪ Ϫ† ϩϪ¶ ϩϩ Neuroblastoma Ϫ† ϪϪ# ϪϪϪϩϪ Small cell carcinoma ϩϪϪϪϪϪϩϪ Hematopoietic malignancies Ϫϩ** ϪϪ†† ϪϪϪϪ Melanoma ϪϪϩഠ ϪϪϪ† Ϫ‡ * CK indicates cytokeratin; ഠ, variable; Ϫ, negative; and ϩ, positive. † Often negative but occasionally positive. ‡ Nonspecific cytoplasmic immunoreactivity. § Cytokeratin positive in ϳ50%, epithelial membrane antigen positive in ϳ100%. S100 positive in cartilaginous areas. ¶ Positive only in rhabdomyomatous Wilms tumor. # Positive only in schwannian stroma. ** Frequently negative in lymphoblastic lymphoma. †† Positive in most lymphoblastic lymphomas.

Figure 4. Algorithmic immunohistochemical analysis for small round blue cell tumors. NE mk indicates neuroendocrine markers; SmCC, small cell carcinoma; PDSS, poorly differentiated synovial sarcoma; DSRCT, desmoplastic small round cell tumor; EWS/PNET, Ewing sarcoma/primitive neuroectodermal tumor; MC, mesenchymal chondrosarcoma; NB, neuroblastoma; SCOS, small cell osteosarcoma; MyoG, myogenin; RMS, rhab- domyosarcoma; *, positive cytoplasmic but not nuclear staining in RMS; and WT, Wilms tumor. recognized usefulness of screening markers in character- Ewing Sarcoma/Primitive Neuroectodermal Tumor ization of undifferentiated malignant neoplasms. CD99, the product of the MIC2 gene, is a cell surface SMALL ROUND CELL TUMORS glycoprotein that is expressed in more than 95% of Ewing Small round cell tumors comprise heterogeneous neo- sarcoma/primitive neuroectodermal tumors (EWS/ 39–42 plasms composed of relatively small, round to oval, close- PNETs) with a diffuse membranous staining pattern. ly packed undifferentiated cells with high nuclear-cyto- CD99, however, is also expressed in several other 42–44 plasmic ratio, scant cytoplasm, and round nuclei with SRCTs including lymphoblastic lymphoma/leukemia Ͼ 31,45 46 evenly distributed, slightly coarse chromatin and small or ( 90%), acute myelogenous leukemia (43%), granu- 46 inconspicuous nucleoli. In spite of a similar light micro- locytic sarcoma (55%), small cell variant of poorly dif- scopic morphology, SRCTs include pathologic entities ferentiated synovial sarcoma (PDSS) (nearly all),47–49 mes- from vastly different lineages, including (1) epithelial tu- enchymal chondrosarcoma (80%–100%),50,51 desmoplastic mors, for example, small cell carcinoma (SmCC) (poorly small round cell tumor (DSRCT) (up to 50%),52 and rarely differentiated neuroendocrine carcinoma); (2) mesenchy- in rhabdomyosarcoma (RMS).53 CD99, however, is not ex- mal tumors encompassing malignant solid neoplasms of pressed in neuroblastoma42 and is only exceptionally pres- childhood and other small round cell sarcomas; and (3) ent in blastema-rich nephroblastoma.44,54 In contrast, tumors with overlapping features, such as lymphoma and CD56, a neural cell adhesion molecule (NCAM), which is melanoma. Because of similar routine light microscopic positive in most SRCTs, is expressed in only 10% to 25% features of these tumors, immunohistochemistry is often of EWS/PNETs55–57 and rarely in lymphoblastic lymphoma mandated for a definitive diagnosis. The immunoprofile (Ͻ10%).58,59 Therefore, a CD99ϩ/CD56Ϫ profile supports of SRCTs is provided in the following and in Table 2, and the diagnosis of EWS/PNET or lymphoblastic lymphoma an algorithmic immunohistochemical analysis is outlined against other SRCTs. in Figure 4. The characteristic translocation in EWS/PNET, t(11; Arch Pathol Lab Med—Vol 132, March 2008 Undifferentiated Tumor, Immunohistochemistry—Bahrami et al 329 and desmin is rarely seen in other SRCTs, it is probably the most specific immunohistochemical profile for DSRCT52 (Figure 6, a through d). The characteristic genetic feature of DSRCT is a consistent association with t(11; 22)(p13;q12) that results in the fusion of Wilms tumor (WT1)andEWS genes. The antibody directed against the WT1 part of the fusion protein is positive in 70% to 100% of cases.52,71,72 CD99 expression is variable and seen in one third to one half of DSRCTs.52,70 Poorly Differentiated Synovial Sarcoma Although the classic bimorphic form of synovial sarco- ma is usually easily recognized based on the hematoxylin- eosin (H&E) morphology, its poorly differentiated variants often create diagnostic difficulty. Three types of PDSS have been recognized: a small cell variant, a large cell epithelioid variant, and a high-grade spindle cell variant.47 The small cell variant is often difficult to distinguish from other SRCTs, in particular EWS/PNET. The distinc- Figure 5. Photomicrograph of rhabdomyosarcoma showing a diffuse tion becomes even more complicated as both PDSS and strong nuclear staining for myogenin (original magnification ϫ10). EWS/PNET are reactive for CD99 in most cases.48,49,73 However, PDSS can be differentiated from EWS/PNET by positive staining for EMA (95%–100% of cases), 22)(q24;q12), involving the Ewing sarcoma (EWS) gene on CK (ϳ50%), CD56 (100%), and collagen IV (100%).48,49,73 chromosome 22 and the FLI-1 gene on chromosome 11, These markers are conversely negative in most EWS/ results in overexpression of FLI-1 protein, which can be PNETs.55–57,74 Because synovial sarcoma is nearly always detected immunohistochemically in more than 70% of negative for CD34, a positive stain for CD34 essentially EWS/PNETs.40,60,61 Hence, a nuclear staining for FLI-1 pro- excludes the diagnosis of PDSS. Most synovial sarcomas tein may help distinguish Ewing family of tumors from regardless of histology have the specific translocation t(X; other CD99-positive SRCTs, except for lymphoblastic lym- 18)(SYT-SSX); hence, in questionable cases molecular test- phomas, which also stain for FLI-1 protein in 88% of cas- ing may be required for definitive diagnosis. es.60 Cytokeratin is expressed variably in EWS/PNETs (in 20%–57%); in contrast, desmin expression is extremely un- Mesenchymal Chondrosarcoma common.40,54,62,63 Mesenchymal chondrosarcoma is a biphasic tumor com- posed of highly undifferentiated small round cells, ad- Rhabdomyosarcoma mixed with islands of hyaline-like cartilage. If the histo- Myogenin and Myo-D1, the 2 myogenic transcriptional logic tissue includes only the undifferentiated small cell regulatory proteins, are equally regarded as sensitive and component, the distinction of mesenchymal chondrosar- specific markers of skeletal muscle differentiation.64 These coma from other SRCTs especially EWS/PNET will be dif- markers stain the overwhelming majority (Ͼ95%) of ficult or perhaps impossible. RMSs53,64–66 including the primitive undifferentiated forms, Although the cartilaginous areas are reactive for S100 regardless of morphologic evidence of skeletal muscle dif- protein, the small cell component is typically not.50 The ferentiation.65 Although a nuclear staining is deemed as immunoprofile of mesenchymal chondrosarcoma includes true positive (Figure 5), a cytoplasmic stain, which may variable expression for NSE and desmin but negative be observed in nonmyogenous tumors, should be consid- staining for CKs and myogenin.50 CD99 is demonstrated ered nonspecific. Because of the potential for cytoplasmic in most cases (80%–100%) in the small cell component.50,51 and high background staining with available antibodies against Myo-D1, staining for myogenin is technically pref- Small Cell Osteosarcoma erable.66,67 Because of the lack of a specific immunoprofile for os- Desmin and muscle-specific actin, although highly sen- teosarcoma,75 recognition of the small cell variant of os- sitive markers for RMSs, are not specific for these tumors teosarcoma from other SRCTs essentially relies on the ob- because they also stain smooth muscle neoplasms and servation of osteoid production76 (Figure 7). Although DSRCTs. A strong cytoplasmic staining for WT1, not the monoclonal antibodies for osteocalcin are specific for os- specific nuclear staining, is observed in RMSs (in 100% of teoid-forming tumors,77,78 only a minority of small cell os- cases in 1 report).68 teosarcoma show osteocalcin reactivity.39 Osteosarcomas are immunophenotypically heterogeneous79 and may oc- Desmoplastic Small Round Cell Tumor casionally express desmin, smooth muscle actin, S100 pro- These tumors are characterized by a polyphenotypic tein, and rarely CK,75,79,80 creating a potential diagnostic differentiation, including frequent coexpression of vimen- pitfall in their distinction from other SRCTs. tin, epithelial markers (CAM 5.2, AE1/AE3, EMA), WT1, neuron-specific enolase (NSE), and desmin,52,69,70 the latter Wilms Tumor with a cytoplasmic or characteristic paranuclear dotlike Wilms tumor is classically a triphasic tumor composed pattern. Although most DSRCTs are positive for desmin, of blastemal, epithelial, and stromal elements, but on oc- the absence of reactivity for Myo-D1 and myogenin52,70 dif- casions, particularly in small biopsy materials, it may pre- ferentiates them from RMSs. Because coexpression of CK sent as a monophasic lesion with only blastemal compo- 330 Arch Pathol Lab Med—Vol 132, March 2008 Undifferentiated Tumor, Immunohistochemistry—Bahrami et al Figure 6. Photomicrograph of desmoplastic small round cell tumor (hematoxylin-eosin, original magnifications ϫ10 [a] and ϫ20 [b]), showing coexpression of cytokeratin (original magnification ϫ20 [c]) and desmin (original magnification ϫ20 [d]). Figure 7. Photomicrograph of small cell osteosarcoma recognized by lacelike osteoid closely associated with tumor cells (hematoxylin-eosin, original magnification ϫ20). Figure 8. Photomicrograph of Merkel cell carcinoma showing positive staining for CK20 with a distinct paranuclear dotlike pattern (original magnification ϫ40).

Arch Pathol Lab Med—Vol 132, March 2008 Undifferentiated Tumor, Immunohistochemistry—Bahrami et al 331 nent. Blastema-rich Wilms tumor, mainly composed of un- estrogen receptor (ER) and progesterone receptor (PR), differentiated blastemal cells, may be difficult to distin- which is seen in two thirds of mammary SmCCs accord- guish from other SRCTs from a morphologic perspective, ing to 1 report.96 Although immunoreactivity for steroid especially in the setting of metastatic disease. hormone receptors, especially PR, has been occasionally WT1, the Wilms tumor suppressor gene, located on observed in neuroendocrine tumors from various primary chromosome 11p13, encodes a putative transcription fac- sites,97,98 unlike breast, simultaneous expression of ER and tor implicated in tumorigenesis and in normal urogenital PR is unusual for them. development. WT1 nuclear staining in the blastemal area Thyroid transcription factor 1 (TTF-1) is a sensitive is seen in 70% to 100% of cases.54,68,81 Although the blas- marker for lung SmCC (positive in 80%–100%)99–102;how- temal component is reactive for vimentin and desmin ever, TTF-1 expression has been also observed in various (partially in up to 90%), the absence of staining for my- extrapulmonary SmCCs in several settings (11%–80%).99–101,103 ogenin and Myo-D1 discriminates blastematous Wilms tu- TTF-1, therefore, should not be considered as a sole deter- mor from RMS.82,83 Blastemal foci are either negative or minant in distinguishing extrapulmonary SmCC from its only focally positive for CK. Although rare instances of pulmonary counterpart. TTF-1 expression, however, is CD99 expression have been reported, staining for FLI-1 consistently negative in Merkel cell carcinoma.100,103 has never been observed.54,60 One distinctive example of organ-specific SmCC is that from the ovary. There are 2 distinct types of ovarian Neuroblastoma SmCC. One is the neuroendocrine type, which is morpho- Neuroblastoma is characterized by a neuronal pheno- logically and immunohistochemically indistinguishable type, which is demonstrated by several markers including from the pulmonary SmCC. The second is the hypercal- NSE (positive in 38%–95%), neurofilament (63%), CD56 cemic type, which has a distinctive immunoprofile, dis- (NCAM) (100%), synaptophysin (65%–88%), and chrom- similar to that of traditional SmCC, characterized by pos- ogranin (60%–88%).84–87 Expression of the more specific itive reactivity for CK, calretinin, CD10, N-terminal of neuroendocrine markers, including chromogranin and WT1, p53, and occasionally parathyroid hormone; nega- synaptophysin, may diminish in less differentiated tu- tive staining for chromogranin, CD99, desmin, inhibin, mors.87 S100, and TTF-1; and variable reaction for other neuro- The following markers, which are negative in neuro- endocrine markers.104,105 blastoma, can help differentiate this tumor from EWS/ One should be aware that extrapulmonary SmCCs do PNET (CD99ϩ), SmCC (CKϩ), lymphoma (CD45ϩ), mela- not consistently express the specific antigens of the organs noma (S100ϩ), and RMS (positive myogenous mark- of their origin. For instance, SmCCs of the prostate are ers).42,84,85 Because up to 35% of olfactory neuroblastomas negative for prostate-specific antigen (PSA) and prostate may be CK positive,88,89 a concomitant negative EMA may acid phosphatase (PAP),106 and those of the urinary blad- help differentiate them from SmCC. Although the tumor der are negative for thrombomodulin, uroplakin III, high- cells are negative for S100 protein, the presence of S100- molecular-weight cytokeratin (HMW CK), and CK20.107 A positive dendritic cells around lobules of tumor cells is a positive CDX2, although specific to tumors of intestinal helpful immunohistochemical feature. origin, is seen in only 20% of SmCCs of the colon.108 Sinonasal undifferentiated carcinoma, a rare, highly ag- Small Cell Carcinoma gressive tumor type arising in the nasal cavity and para- Small cell carcinomas (poorly differentiated neuroen- nasal sinuses, should be considered in the differential di- docrine carcinomas) represent poorly differentiated neu- agnosis of both SRCTs and undifferentiated large carci- roendocrine neoplasms with an epithelial lineage. They nomas in the head and neck region. These tumors are gen- may originate from a variety of locations, most commonly erally positive for CKs but exclusively for CKs of simple the lung but also from extrapulmonary sites, including epithelia, such as CK8 (100%), CK19 (50%), and CK7 nasal cavity and paranasal sinuses, breast, uterine cervix, (50%), and as such are negative for CK5/6.109 They stain bladder, prostate, gastrointestinal tract, pancreas, thyroid, variably for EMA (18%), NSE (18%), and CD99 (14%) but adrenal gland, skin (Merkel cell carcinoma), and salivary are typically negative for chromogranin and synaptophy- glands. The histopathologic features of these tumors, re- sin.110 gardless of the site of origin, are nearly identical. The immunodiagnosis of SmCCs rests on the immu- Hematopoietic Malignancies nohistochemical proof of a simultaneous epithelial and A dense infiltrate of relatively small undifferentiated neuroendocrine differentiation. The former is reflected by cells of uncertain origin may be the first presentation of a reactivity for keratin, especially LMW CKs (CK8 and hematolymphoid malignancy. As stated previously, be- CK18),57,90 and the latter by a positive neuroendocrine cause lymphoblastic lymphomas may not express panel, which in order of increasing specificity includes CD45,30,32,111 a negative CD45 (LCA) does not totally ex- NSE, CD56 (NCAM), synaptophysin, and chromogran- clude the possibility of hematopoietic malignancies. Nu- in.91,92 Against a common histologic and immunohisto- clear staining for terminal deoxyribonucleotide transferase chemical background, SmCC from different organs may (TdT), which is a sensitive marker for both T- and B-pre- display additional immunophenotypes that help deter- cursor lymphoid cells, on the other hand, is considered mine the primary site, because not infrequently these tu- the hallmark of lymphoblastic lymphoma. The majority mors present as metastases of unknown origin. For ex- (85%–90%) of lymphoblastic lymphomas derive from pre- ample, among SmCCs from different sites, including gas- cursor T cells, and the rest originate from immature B trointestinal origin, CK20 is positive only in Merkel cell cells.18 CD43 is a sensitive, although not a specific, T-cell carcinomas (Ͼ97%)93,94 and salivary gland SmCCs marker. Cytoplasmic CD3 (not surface CD3) is also fre- (ϳ60%),93,95 often with a distinct paranuclear punctate pat- quently expressed by T-precursor lymphoblasts. CD79a is tern (Figure 8). Another example is positive reaction for a more sensitive marker for detection of immature B cells 332 Arch Pathol Lab Med—Vol 132, March 2008 Undifferentiated Tumor, Immunohistochemistry—Bahrami et al Table 3. Tumor-Specific Markers and Their Staining Pattern* Marker Tumor Staining Pattern TTF-1 Lung, thyroid Nuclear Thyroglobulin Thyroid Cytoplasmic HepPar-1 Hepatocellular Cytoplasmic CDX2 Colorectal/duodenal Nuclear Villin Gastrointestinal (epithelia with brush border) Apical ER/PR Breast, ovary, endometrium Nuclear GCDFP-15 Breast Cytoplasmic Mammaglobin Breast Cytoplasmic RCC marker Renal Membranous PSA Prostate Cytoplasmic PAP Prostate Cytoplasmic Uroplakin III Urothelial Membranous Inhibin Sex cord–stromal, adrenocortical Cytoplasmic Melan-A Adrenocortical, melanoma Cytoplasmic Calretinin Mesothelioma, sex cord–stromal, adrenocortical Nuclear/cytoplasmic WT1 Ovarian serous, mesothelioma, Wilms, desmoplastic small round cell Nuclear Mesothelin Mesothelioma Cytoplasmic/membranous D2-40 Mesothelioma, lymphatic endothelial cell marker Membranous * TTF-1 indicates thyroid transcription factor 1; HepPar-1, hepatocyte paraffin 1; ER/PR, estrogen receptor/progesterone receptor; GCDFP-15, gross cystic disease fluid protein 15; RCC, renal cell carcinoma; PSA, prostate-specific antigen; and PAP, prostate acid phosphatase. compared with CD20. Thus, if lymphoblastic lymphoma cells with a nesting arrangement. These histologic attri- is suspected, the panel should include TdT, CD43, and butes, however, are not specific to carcinomas and may be CD79a, in addition to CD45. mimicked by other tumors, including sarcomas with epi- Myeloid (granulocytic) sarcomas with little or no evi- thelioid appearance, melanoma, and hematopoietic malig- dence of myeloid differentiation may manifest as undif- nancies. Screening panel for major lineage determination ferentiated tumors, sometimes with small round cell mor- (see previous discussion) is essential for clarification of the phology. A high index of suspicion is often required to true identity of these tumors. avoid diagnostic overlook of these malignancies. Myelo- blasts are usually positive for CD45 (in 75% of cases), Tissue- or Organ-Specific Determination CD43 (100%), myeloperoxidase (variably), and lysozyme Once the diagnosis of carcinoma by broad lineage (variably) but are generally negative for B- and T-lineage markers is established, immunohistochemistry may assist markers, including CD79a and CD3, respectively.112 further by delineation of the cell line of differentiation. As mentioned previously, most lymphoblastic lympho- This is achieved by analysis of CK subtypes and other mas45,111 and a large fraction of myeloid sarcomas46 express complementary or tissue- or organ-specific markers (Table CD99. However, application of lymphoid markers usually 3). Although some markers are expressed almost exclu- helps to distinguish hematolymphoid malignancies from sively in a specific tissue, for example, PSA and PAP for Ewing sarcomas, which are negative for these markers. prostate and uroplakin III for urothelial epithelium, others are not strictly organ or tissue specific but may help focus Malignant Melanoma the diagnosis to a specific area. In keeping with various cytomorphologic and architec- It should be mentioned that because most reports on tural manifestations of malignant melanoma (MM), its the sensitivity and specificity of markers in the literature small cell (neuroendocrine-like) variant, which occurs are derived from studies that encompassed primary and more commonly within the nasal cavity and paranasal si- metastatic tumors of various grades, it is difficult to ex- nuses,26 is a differential diagnostic consideration for trapolate those parameters exclusively in respect to un- SRCTs. Because melanoma with small cell morphology of- differentiated or poorly differentiated forms of tumors. ten lacks melanin pigment, immunohistochemistry plays Nevertheless it can be concluded from the literature that an important role in its recognition. for most entities, the sensitivity of tissue-specific markers The immunoprofile of MM is discussed in more detail declines as the tumor grade increases. Also, although the later in the section on tumors with overlapping histologic antigenic expression of some markers appears to be robust features. Herein the potential for expression of aberrant during metastasis, for example, CDX2 for colorectal car- markers in small cell melanoma as a source of diagnostic cinoma, others, such as the renal cell carcinoma (RCC) confusion with other SRCTs is highlighted. These include marker, tend to reduce their sensitivity in metastatic sites. CD99 in rare to up to 60% of cases,113,114 WT1 (cytoplasmic CK Profile. The CK phenotype often provides helpful but not nuclear pattern) in as many as 70%,115 CD10 in clues in determination of the cell line of differentiation in 40% to 50%,116,117 CD68 (a histiocytic marker) in 86%,118 both primary and metastatic carcinomas. This reflects the NSE in approximately 50%,113 and neuroendocrine mark- observation that tumors tend to recapitulate the CK profile ers (CD56, CD57, and synaptophysin) in up to 13%.113 of the normal cells from which they are derived. Although Chromogranin, on the other hand, is consistently absent metastatic tumors may gain or lose antigens compared in melanocytic tumors.119 with primary tumors, their CK profiles usually remain the same, as demonstrated by a number of studies, including UNDIFFERENTIATED ‘‘CARCINOMATOUS’’ TUMORS a comprehensive study of CK7 and CK20 for 384 primary Undifferentiated tumors with a carcinomatous appear- tumors and their metastasis120 and other studies focusing ance are characterized by large round-oval to polygonal on CK20 expression by primary and metastatic colorectal Arch Pathol Lab Med—Vol 132, March 2008 Undifferentiated Tumor, Immunohistochemistry—Bahrami et al 333 Figure 9. Algorithmic immunohistochemical analysis of undifferentiated carcinomas. CA indicates carcinoma; adenoCA, adenocarcinoma; SmCC, small cell carcinoma; SCC, squamous cell carcinoma; RCC, renal cell carcinoma; HCC, hepatocellular carcinoma; ¶, seminoma is keratin negative, OCT3/4 positive; *NE markers, neuroendocrine markers, including synaptophysin, chromogranin, and CD56; ␺, undifferentiated anaplastic thyroid carcinoma is often negative for thyroid transcription factor 1 (TTF-1) and thyroglobulin; and ␰, characteristic canalicular pattern.

carcinoma.94,121,122 Coordinate expression of CK7 and CK20 astatic colorectal carcinoma (CK7Ϫ/CK20ϩ) from primary defines subsets of carcinomas, as shown in Figure 9. nonmucinous ovarian carcinoma or primary endometrial, The CK7/CK20 phenotype can be especially useful in pulmonary, or mammary adenocarcinoma (CK7ϩ/ certain clinical situations, including the differentiation be- CK20Ϫ).122,124 tween prostate (CK7Ϫ/CK20Ϫ) and urothelial (CK7ϩ/ Other CK subtypes may also have a role in defining the CK20 variable) carcinoma123 (Figure 10); poorly differen- lineage of tumor; for example, CK19 combined with CK7 tiated squamous (CK7Ϫ) and poorly differentiated uro- helps to differentiate between hepatocellular carcinoma thelial (CK7ϩ) carcinoma of the urinary bladder; and met- (HCC) (CK7Ϫ/CK19Ϫ) and bile duct carcinoma (CK7ϩ/ 334 Arch Pathol Lab Med—Vol 132, March 2008 Undifferentiated Tumor, Immunohistochemistry—Bahrami et al Figure 10. Photomicrograph of urothelial carcinoma with positive staining for cytokeratin (CK) 7 (a) and CK20 (b) (original magnifications ϫ20). Figure 11. Photomicrograph of lung adenocarcinoma with diffuse nuclear staining for thyroid transcription factor 1 (original magnification ϫ10). Figure 12. Photomicrograph of epithelioid mesothelioma with nuclear and cytoplasmic staining for calretinin (original magnification ϫ20). Figure 13. Photomicrograph of seminoma, showing a diffuse strong nuclear staining for OCT3/4 (original magnification ϫ20). Figure 14. Photomicrograph of sarcomatoid carcinoma with cytokeratin expression in both sarcomatous and carcinomatous areas (original mag- nification ϫ20).

Arch Pathol Lab Med—Vol 132, March 2008 Undifferentiated Tumor, Immunohistochemistry—Bahrami et al 335 Table 4. Carcinoembryonic Antigen (CEA) are nevertheless distinguished based on their different co- Immunostaining in Carcinomas ordinate CK7/CK20. p63 expression in the lung, however, is not specific to SCC; thus, a subset of adenocarcinomas CEA Positive CEA Negative (30%) and large cell carcinomas (37%), 50% of large cell Colonic/rectal adenocarcinoma Endometrial adenocarcinoma neuroendocrine carcinomas (particularly higher-grade tu- Gastric and esophageal adeno- Renal cell carcinoma mors), and a variable proportion of SmCCs (0%–77%) carcinoma Prostate adenocarcinoma 140,142 Pancreatic adenocarcinoma Ovarian serous carcinoma stain for p63. Because p63 is consistently negative in Biliary tract adenocarcinoma Adrenal carcinoma mesothelioma, it can assist in differentiating SCC of the Endocervical adenocarcinoma Mesothelioma lung from epithelioid mesothelioma.127,139 Lung adenocarcinoma Neuroendocrine Markers. Neuroendocrine markers are tissue indicators of neuroendocrine neoplasms, includ- ing those with epithelial lineage (neuroendocrine carci- CK19ϩ).125 CK5/6 serves as a sensitive marker for squa- nomas) and those with neural derivation, such as para- mous cell carcinoma (SCC) (CK5/6ϩ) and epithelioid me- ganglioma/pheochromocytoma and neuroblastoma. sothelioma (CK5/6ϩ) and helps differentiate them from The 2 most reliable markers of neuroendocrine neo- pulmonary adenocarcinoma (CK5/6Ϫ).124,126–129 CK5/6 also plasms, synaptophysin and chromogranin, have a com- labels urothelial carcinoma (in 50%–63% of cases)124,126 and parable sensitivity; thus, they are always used together as the large cell undifferentiated lung carcinoma (vari- complementary reagents for neuroendocrine lineage de- ably).126,130 termination. CD56 is generally regarded as a broad-spec- Figure 9 provides a selected list of specific tumors with trum neuroendocrine marker and perhaps the most sen- their pertinent positive and negative markers, categorized sitive144,145 and is instrumental in confirming the diagnosis according to their CK7/CK20 profile. The choice of tissue- of neuroendocrine tumors in small biopsies.144–146 CD56, specific markers to be included in an investigative panel however, is not specific for neuroendocrine delineation be- is largely determined by the clinical impression, site of cause it may label occasional nonneuroendocrine tissues presentation, and the CK7/CK20 expression of an undif- and their malignant counterpart (eg, follicular and pap- ferentiated carcinoma. illary thyroid, renal cell, and hepatocellular carcinoma), Carcinoembryonic Antigen. Carcinoembryonic anti- natural killer cells, some T-cell lymphomas, and a large gen (CEA) is an oncofetal glycoprotein overexpressed by number of SRCTs.92,147 Neuron-specific enolase suffers a a variety of adenocarcinomas and consistently by gastro- poor specificity for neuroendocrine tumors, and, as such, intestinal adenocarcinomas.131,132 Table 4 summarizes re- it is not used as a sole determinant of neuroendocrine activity for CEA in a variety of carcinomas. Use of CEA differentiation. in a panel may especially assist in certain differential di- An undifferentiated carcinoma may turn out to have an agnostic considerations, for example in the distinction of occult neuroendocrine differentiation, disclosed by im- endocervical (CEA positive) over endometrial (CEA neg- munostain. Although neuroendocrine markers are not ative) adenocarcinoma133 or in the recognition of poorly usually included in the initial diagnostic panel for an un- differentiated colonic adenocarcinoma (CEA positive) differentiated tumor, detection of neuroendocrine differ- from solid areas of ovarian serous adenocarcinoma (CEA entiation is clinically significant, because this may imply negative).132 prognosis and therapy similar to those for SmCC.148,149 Be- Polyclonal CEA (pCEA) antibodies, in contrast to CEA- cause positive staining in a few tumor cells is frequent in specific monoclonal antibody, have cross reactivity with any undifferentiated carcinoma, the diagnosis of neuro- CEA-related antigens, including biliary glycoprotein anti- endocrine carcinoma should only be made if the stain is gen. Hepatocellular carcinoma is negative for monoclonal unequivocal and, in our experience, present in at least 20% CEA but expresses a characteristic bile canalicular staining of the tumor cells. pattern with pCEA134 because of cross reactivity with bil- In the lung, large cell neuroendocrine carcinoma must iary glycoprotein antigen. Carcinoembryonic antigen is be differentiated from basaloid carcinoma. Because of also frequently used as a negative marker in the mesothe- their overlapping H&E morphology, the differential di- lioma panel. agnosis is often difficult but can be facilitated by their MOC-31. MOC-31 is a cell surface glycoprotein large- specific immunoprofiles. Basaloid carcinoma is negative ly found on the epithelial cells. MOC-31 helps differentiate for neuroendocrine markers and TTF-1 and positive for between adenocarcinoma (MOC-31 positive) and meso- HMW CK, whereas large cell neuroendocrine carcinoma thelioma (MOC-31 negative).135,136 MOC-31 is also useful is consistently positive for neuroendocrine markers, posi- in the recognition of cholangiocarcinoma and other met- tive for TTF-1 in about 50% of cases, and often negative astatic adenocarcinomas (MOC-31 positive) from HCC for HMW CK.103,150,151 (MOC-31 negative).137 Thyroid Transcription Factor 1. TTF-1 is a nuclear p63. p63, a p53 homologue, is a marker for basal cells transcription factor that promotes embryogenic pulmo- and helps maintain basal cells in squamous and other ep- nary and thyroid differentiation and is expressed by most, ithelial linings.138 p63 stains the vast majority of SCCs but not all, lung or thyroid neoplasms. Although 80% to (ϳ97%) and, in addition to CK5/6, is recognized as a 100% of SmCCs and 96% to 100% of adenocarcinomas good marker for SCC, especially the poorly differentiated from lung express TTF-1 (Figure 11), pulmonary SCCs are forms.102,139–142 Thus, a combined p63ϩ/CK5/6ϩ immuno- negative in most cases.99–103,150,152,153 profile in a poorly to undifferentiated carcinoma is highly Furthermore, TTF-1 expression tends to decrease in predictive of a squamous origin.139 Because urothelial car- poorly differentiated (50% positive) compared with well- cinoma also expresses p63 (in 70%–95% of cases) and differentiated (100% positive) adenocarcinomas,153 dimin- CK5/6 (50%–63%),126,143 a p63ϩ/CK5/6ϩ immunoprofile ishing the sensitivity of this marker for identification of could also suggest a urothelial carcinoma. The 2 entities the pulmonary origin of a poorly differentiated adenocar- 336 Arch Pathol Lab Med—Vol 132, March 2008 Undifferentiated Tumor, Immunohistochemistry—Bahrami et al cinoma. Among thyroid carcinomas, undifferentiated an- state of tumor differentiation.173 Villin, however, also stains aplastic thyroid carcinomas are unfaithful in terms of TTF- a number of extragastrointestinal adenocarcinomas, in- 1 expression (positive in Ͻ5%).154 TTF-1, as mentioned pre- cluding some adenocarcinomas of ovary, endometrium, viously, may also be expressed in SmCCs arising from a kidney, and bladder.171,173,175 Villin expression in lung ade- variety of locations other than the lung99,101,103,155 (see also nocarcinoma correlates with the presence or absence of ‘‘Small Cell Carcinoma’’). microvilli (positive in 67% and 10% of cases, respective- Thyroglobulin. Thyroglobulin is a very specific mark- ly).174 er for thyroid follicular and papillary carcinomas but only ER and PR. The role of ER and PR in determining the rarely stains cells in undifferentiated anaplastic thyroid primary site in undifferentiated tumors is hampered by a carcinoma.154,156 Anaplastic thyroid carcinomas are gener- wide spectrum of expression among a variety of tumors. ally positive for vimentin and CK157,158 but are negative for Although ER and PR expression is expected to be associ- CEA,157 TTF-1,154 and melanocytic, vascular, myogenous, ated with hormone-responsive organs and their neo- and lymphoid markers. plasms, such as breast, ovary, and endometrium,176 there Hepatocyte Paraffin 1. Hepatocyte paraffin 1 (HepPar- are frequent examples of unexpected ER or PR expression 1) is a monoclonal antibody that stains a cellular antigen reported in a variety of other tumors, including those of unknown function of normal hepatocytes and is known from lung,98,177–179 thyroid,180 stomach,181 and some neuro- to be a highly sensitive (positive in Ͼ90% of cases) and endocrine tumors.97 Furthermore, given that many high- relatively specific marker for HCC.159–164 HepPar-1 expres- grade carcinomas arising from hormone-responsive or- sion, however, decreases in HCC with higher nuclear gans such as breast lack ER expression, we discourage sole grade.164 HepPar-1 is occasionally observed in nonhepatic reliance on ER and PR as determinants of the site of tumor carcinomas, including those of gastrointestinal origin, in origin, except in selected differential diagnoses and in particular gastric signet ring cell carcinoma (47%–83%), combination with other immunomarkers. An example of and rarely in other tumors.159,163–165 the utility of ER in a diagnostic panel is its application The distinction of HCC, especially its poorly differen- combined with CEA133 and vimentin in the distinction of tiated forms, from metastatic adenocarcinoma and cholan- an endometrial (vimentinϩ/ERϩ/CEAϪ) from an endocer- giocarcinoma is commonly a diagnostic challenge. A panel vical (vimentinϪ/ERϪ/CEAϩ) adenocarcinoma. that includes HepPar-1, MOC-31, pCEA, and CD10, in ad- Gross Cystic Disease Fluid Protein. Gross cystic dis- dition to CK7 and CK20, is often helpful in making the ease fluid protein 15 (GCDFP-15), a 15-kd secretory gly- accurate diagnosis. The immunoprofile of HCC includes coprotein of various body fluids, including saliva, milk, CK7Ϫ/CK20Ϫ, HepPar-1 positive, MOC-31 negative, sinu- and seminal fluid, is considered a marker of apocrine dif- soidal cell CD34ϩ, and a canalicular pattern of positive ferentiation182,183 with high specificity for breast carcino- immunostaining with both pCEA and CD10.134,160,161,166 mas.184–186 In a study of 328 metastatic adenocarcinomas, Metastatic adenocarcinoma and cholangiocarcinoma, on Kaufmann et al184 demonstrated that expression of the other hand, are generally MOC-31 and CEA positive, GCDFP-15 and/or ER or PR had a sensitivity of 83%, a the latter with a membrane/cytoplasmic staining pattern. specificity of 93%, and a predictive accuracy of 92% for The canalicular staining pattern with antibodies to CD10, carcinomas of the breast against all other carcinomas ex- positive in 68% of HCC, is distinct from the membrane or cluding ovarian carcinomas. In another study, including cytoplasmic staining seen in RCC.166 105 breast cancers and 585 nonmammary malignancies, CDX2. CDX2, an intestine-specific transcription fac- GCDFP-15 was able to identify breast carcinomas with a tor, is a sensitive and specific marker for colorectal and sensitivity of 74% and a specificity of 95%.185 Other tumor duodenal adenocarcinomas in both primary and meta- types with occasional GCDFP-15 expression include car- static sites.167–169 However, CDX2 expression tends to de- cinomas of the salivary glands, sweat glands, and pros- cline in the higher grade and stage of colorectal tumors tate.185,187 Although a very specific marker, GCDFP-15 is (positive in only 56% of poorly differentiated tumors)169 not a particularly sensitive marker of breast carcino- and is practically absent in an undifferentiated subset of ma.188,189 It is uncertain whether GCDFP-15 expression de- large cell colonic carcinoma, usually associated with DNA clines in poorly differentiated forms of mammary carci- mismatch repair defects.170 CDX2 is variably expressed in noma. other adenocarcinomas of the digestive tract, including Mammaglobin. Mammaglobin, a mammary-specific gastric (55%–70%), esophageal (67%), pancreatic (32%– member of the uteroglobin family, is overexpressed in hu- 60%), and biliary (25%–60%).163,167 man breast carcinoma. Mammaglobin expression has been CDX2 is also observed in a few extraintestinal carcino- observed in 48% to 84% of breast carcinoma and in 8% to mas. For example, 47% to 100% of primary adenocarci- 15% of carcinomas not from breast, including salivary nomas of the bladder168,171 and 10.5% to 100% of ovarian gland tumors (55%) and endometrial carcinoma (13%).188–190 carcinomas, particularly the mucinous type, are positive Although mammaglobin appears to be a more sensitive for CDX2.167–169 Because bladder adenocarcinoma also dis- marker than GCDFP-15, it is not as specific as GCDFP-15 plays a ‘‘urothelial’’ profile (CK7, 65% positive; CK20, 53% for breast carcinoma.188 The predictive value of a com- positive; and thrombomodulin, 59% positive),172 it may be bined mammaglobin and GCDFP-15 panel for mammary differentiated from metastatic colorectal carcinoma by im- carcinomas in tumors of unknown origin needs to be in- munohistochemistry. vestigated. Villin. Villin is a cytoskeletal protein associated with RCC Marker. Renal cell carcinoma marker, a mono- brush border microvilli of the intestine and proximal renal clonal antibody against a normal human proximal tubular tubular epithelium. Villin is a sensitive marker of gastro- brush border glycoprotein known as gp200, decorates con- intestinal adenocarcinomas, staining 82% to 100% of pri- ventional and papillary RCCs with an approximate sen- mary and metastatic colonic adenocarcinomas.168,173,174 Vil- sitivity of 85% and more than 95%, respectively.191,192 lin expression does not seem to be associated with the Among nonrenal carcinomas, a subset of breast (29%) and Arch Pathol Lab Med—Vol 132, March 2008 Undifferentiated Tumor, Immunohistochemistry—Bahrami et al 337 embryonal carcinoma (28%) may stain with RCC mark- can be very challenging. This challenge, which is usually er.191 The sensitivity of RCC marker tends to decline from solved by careful cytomorphologic evaluations (monoto- low-grade (92%) to high-grade (75%) conventional nous tumor cells, prominent nucleolus, and less mitoses RCCs.192 In addition, only 25% of sarcomatoid RCCs stain for prostate cancers), may be facilitated by a panel that for this marker.191 includes a number of discriminatory markers, in addition Although RCC marker remains highly specific (98%) for to CK profile, including PSA and PAP, markers of prostate metastatic RCC, because of a rather low sensitivity in met- adenocarcinoma, and HMW CK, p63, uroplakin III, and astatic sites (positive in 67%), a negative result does not thrombomodulin, markers that are expected to be positive totally rule out metastatic RCC.191 Negative vimentin and in urothelial carcinoma.210,211 CD10, however, enhances the negative predictive value for Inhibin. Inhibin is a peptide hormone produced by conventional RCCs in an investigative panel for metastatic ovarian granulosa cells and testicular Sertoli cells. It serves tumors. CD10, similar to vimentin, is a nonspecific marker as a sensitive and highly specific marker for ovarian and for RCCs. It is, however, a relatively sensitive indicator for testicular sex cord–stromal tumors.21–215 Inhibin is also a RCCs and stains approximately 90% of these tumors with sensitive marker for adrenal cortical neoplasms and reli- a diffuse cytoplasmic or membranous pattern.117 The dis- ably differentiates cortical from medullary adrenal tu- tribution of CD10 appears to be independent of the grade mors.216–218 or histologic type (conventional vs papillary) of the Melan-A. Melan-A, a product of MART-1 gene, is an RCC.192 antigen on melanoma cells that is recognized by the an- PSA and PAP. Immunostain for the prostate-specific tibody A103. Although Melan-A is essentially used as a markers PSA and PAP, 2 glycoproteins produced almost melanoma marker,219 it is also a sensitive marker for ad- exclusively by the prostatic glandular epithelium, is wide- renal cortical neoplasms220 and, similar to inhibin and cal- ly used for determination of the prostatic origin of meta- retinin, a specific marker in differentiating cortical from static tumors. Genega et al193 demonstrated a sensitivity of medullary adrenal tumors.216 Melan-A is additionally ex- 94% for intermediate- to high-grade prostate adenocarci- pressed by perivascular epithelioid cell tumors (PECo- nomas (Gleason score Ն8) with both PSA and PAP. Pros- ma).219,221 tate-specific antigen expression has been rarely reported Calretinin. Calretinin is a good marker for malignant in some nonprostatic tissues and tumors, including sali- mesothelioma with a sensitivity approaching 100%.127,128,208,222 vary gland, breast, and gynecologic tumors.194–197 It is also quite specific for mesothelioma against adeno- Prostate-specific antigen and PAP expression tends to carcinomas, which are positive for it in only 8% to 11% of decrease in higher grade prostate adenocarcinomas.198,199 cases.128,130,208,222 Calretinin decorates both mesothelial cells In a study of 225 prostate adenocarcinomas by Gold- and malignant mesothelioma with both nuclear and cy- stein,200 all tumors with a Gleason score of 6 or 7 were toplasmic staining223 (Figure 12). reactive for PSA, whereas 13% of the adenocarcinomas Calretinin additionally serves as a highly sensitive with a Gleason score of 10 were nonreactive. Although marker for sex cord–stromal tumors. It is, however, not as some investigators found PAP slightly more sensitive than specific as inhibin in this respect because it also labels a PSA in diagnosing poorly differentiated prostatic adeno- fraction of ovarian epithelial neoplasms (22%).215,224,225 Like carcinoma,200,201 others could not confirm these find- inhibin and Melan-A, calretinin also stains adrenal corti- ings.199,202 Nevertheless, a combined immunostain for PAP cal neoplasms and discriminates them from adrenal med- and PSA would be expected to detect most prostatic ad- ullary tumors.216 enocarcinomas in metastatic sites.199,203 Mesothelin. Mesothelin is a cell surface antigen that Parenthetically, ␣-methylacyl-CoA racemase (AMACR) is strongly expressed in normal mesothelial cells, meso- encoded by the gene P504S, which is a highly specific and theliomas, and a number of other carcinomas.226 Although sensitive marker for prostatic carcinoma in prostatic spec- a highly sensitive marker for mesothelioma, mesothelin imens,204 has limited value in detecting metastatic prostate may also be present in approximately half of the lung ad- cancer because it is also expressed in various normal tis- enocarcinomas130,226; as such, it has limited value in dis- sues and nonprostatic tumors.205 criminating between the 2 entities. Carcinomas with con- Thrombomodulin. Thrombomodulin, a cell surface sistently strong exhibition for mesothelin include non- glycoprotein involved in the regulation of intravascular co- mucinous carcinomas of the ovary and adenocarcinomas agulation, is expressed in 70% to 90% of primary or met- of the pancreas and ampulla of Vater.226 astatic urothelial carcinomas.107,206,207 Thrombomodulin, WT1. WT1, in addition to its diagnostic utility for however, is also expressed by a variety of nonurothelial SRCTs (see ‘‘Small Round Cell Tumors’’), is a sensitive tumors, including SCC (majority) and endothelial vascular marker for epithelioid mesothelioma, successfully discrim- tumors.207 Thrombomodulin is additionally known as a inating it from adenocarcinoma.128,227 Another epithelial positive mesothelioma marker, although not as specific or tumor strongly positive for WT1 is ovarian serous carci- sensitive as the other markers discussed later.128–130,208 noma, which is positive in nearly all cases including the Uroplakin III. Uroplakin III, expressed by terminally high-grade forms.228 Because both peritoneal mesothelio- differentiated superficial urothelial cells, is a highly spe- ma and ovarian serous carcinoma metastatic to the peri- cific marker for urothelial origin of tumors.107,209 This toneum express WT1, additional markers are needed for marker, however, is not quite sensitive for urothelial car- their differential diagnosis: A panel including calretinin, cinomas; thus, it is expressed in up to 60% of urothelial Ber-EP4, ER, and PR is helpful in differentiating mesothe- carcinoma in primary sites and slightly less frequently in lioma (calretininϩ/Ber-EP4Ϫ/ERϪ/PRϪ) from primary or metastases (ϳ50%).107,209 metastatic serous ovarian carcinoma (calretininϪ/Ber- The differential diagnosis between a poorly differenti- EP4ϩ/ERϩ/PRϩ).229 ated prostate adenocarcinoma involving the bladder and Other Mesothelioma Markers. Immunohistochemis- a high-grade urothelial carcinoma with prostatic extension try has a crucial role in the differential diagnosis between 338 Arch Pathol Lab Med—Vol 132, March 2008 Undifferentiated Tumor, Immunohistochemistry—Bahrami et al malignant epithelioid mesothelioma and adenocarcinoma could be reclassified as specific types of sarcoma after im- metastatic to the serous membranes. A number of discrim- plementing further studies, including combined immu- inatory markers are available to facilitate this recognition. nohistochemistry and electron microscopy. Coindre et al244 It is generally recommended to include at least 2 positive showed that among 25 tumors initially diagnosed as ret- and 2 negative markers in the differential panel for me- roperitoneal malignant fibrous histiocytoma (MFH), 17 sothelioma. could be reclassified as dedifferentiated liposarcoma by Calretinin, WT1, and CK5/6 (discussed earlier) are es- extensive sampling, combined with immunohistochemis- tablished as highly reliable positive mesothelioma mark- try and comparative genomic hybridization. These dis- ers.128 Other positive markers include mesothelin and tinctions are important above the academic interest in thrombomodulin (discussed earlier). In addition, D2-40,230 view of their variable prognostic implications.245 a lymphatic endothelial marker, and h-caldesmon,231 a spe- It is emphasized that ancillary tests do not replace prop- cific marker for smooth muscle tumors, have recently been er tumor sampling, which often helps detect tumor foci shown to be valuable in discriminating mesothelioma with features suggestive of or diagnostic for a specific line from lung adenocarcinoma. On the other hand, CEA, of differentiation. According to the report of the Interna- MOC-31, Ber-EP4, BG-8, and B72.3128,129 are accepted as tional Chromosomes and Morphology (CHAMP) study reliable negative markers for discriminating epithelioid group on 46 pleomorphic soft tissue sarcomas, it is un- malignant mesothelioma from pulmonary adenocarcino- likely that cytogenetic analysis, mainly because of the kar- ma. yotype complexity of these tumors, can further improve CD5. CD5 is a useful marker of primary thymic car- their differential diagnostic subclassification.246 cinomas, expressed in neoplastic epithelial cells of the thy- With the help of immunohistochemistry in conjunction mic carcinoma but typically not in thymoma or other car- with generous tumor sampling, a high-grade undifferen- cinomas involving the mediastinum.232–234 Of the 2 used tiated sarcoma can often be classified into one specific cat- clones of anti-CD5 antibody, clone CD5/54/B4 identifies egory of pleomorphic sarcomas, including pleomorphic 30% to 67% of thymic carcinomas.235,236 The other clone, leiomyosarcoma, pleomorphic RMS, pleomorphic and de- NCL-CD5-4C7, although more sensitive for thymic carci- differentiated liposarcoma, osteosarcoma, pleomorphic nomas (45%–100% positive), also labels other neoplasms malignant peripheral nerve sheath tumor (MPNST), and occasionally.233–235 PDSS. Markers of GCTs. Germ cell tumors sometimes pre- The diagnosis of pleomorphic leiomyosarcoma requires sent as undifferentiated metastatic carcinomas. Accurate demonstration of a positive reaction for smooth muscle recognition of these tumors is of utmost importance be- markers, including smooth muscle actin, muscle-specific cause of the available effective therapy. Placental alkaline actin, desmin, calponin, and h-caldesmon,246,247 in the pres- phosphatase is consistently present in embryonal carci- ence of at least focal supporting morphologic characteris- noma and seminoma and variably in several other GCTs, tics (eosinophilic spindle cells with vesicular blunt-ended including yolk sac tumor and choriocarcinoma.237,238 Pla- nuclei arranged in a fascicular pattern). The expression of cental alkaline phosphatase immunoreactivity, however, is myoid markers in the pleomorphic areas, compared with not completely specific to GCTs, as it has been seen in the leiomyosarcomatous foci, is usually significantly re- some soft tissue tumors with known myogenic differen- duced.247 Among myoid markers, h-caldesmon appears to tiation.239 be a promising reagent for specific smooth muscle differ- OCT3/4, also known as POU5F1, is a transcription fac- entiation that allows distinction of leiomyosarcoma from tor and a robust diagnostic marker for seminoma (Figure other tumors with smooth muscle–like differentiation, in- 13) and embryonal carcinoma in both primary and met- cluding myofibroblastic tumors.248,249 However, only 40% astatic sites.240–242 Seminoma also stains consistently with of pleomorphic leiomyosarcoma are labeled with this an- D2-40. Yolk sac tumor does not express OCT3/4 but is tibody.247 positive for ␣-fetoprotein, albeit with a patchy staining The immunoprofile of pleomorphic RMS includes ex- pattern. Human chorionic gonadotropin is consistently ex- pression of muscle-related antigens, that is, desmin, myo- pressed by syncytiotrophoblasts of choriocarcinomas. globin, and actin,250 along with a diffuse reaction with at least one skeletal muscle–specific marker including my- UNDIFFERENTIATED ‘‘SARCOMATOUS’’ OR ogenin and Myo-D1.251 The staining pattern for skeletal ‘‘SARCOMA-LIKE’’ TUMORS muscle markers in pleomorphic RMS is typically diffuse, In the approach to undifferentiated tumors with sarco- which should be distinguished from focal staining seen in matous appearance, exclusion of nonmesenchymal neo- a variety of sarcomas with limited areas of rhabdomy- plasms such as sarcomatoid carcinoma and sarcomatoid omatous differentiation. mesothelioma and other mimickers, for example, spindle The diagnosis of liposarcomas, until a specific lipoblast cell melanoma or hematopoietic malignancies with sar- marker is discovered, continues to be made essentially on comatous feature, is essential. The distinction is usually the basis of the H&E light microscopy. Although normal attained by using the screening panel previously dis- fat cells are positive for S100, neoplastic adipocytes stain cussed in the introduction. inconsistently.252 S100 labels lipoblasts in up to half of pleomorphic liposarcomas,253–255 but nonlipogenic areas Undifferentiated High-Grade Sarcomas only rarely stain with S100.253,256 The aP2 gene product Once the true sarcomatous nature of an undifferentiated (aP2 protein) has been suggested as a relatively specific tumor is established, investigation for a specific mode of marker for lipoblasts in all types of liposarcoma.257 This differentiation, that is, neural, myogenous, lipomatous, or marker also stains brown fat cells but reportedly not be- vascular lineage, should be implemented. nign adipocytes or cells of other mesenchymal or epithe- Fletcher243 demonstrated that among 159 neoplastic le- lial tumors.257 Because of the paucity of pertinent studies, sions initially diagnosed as pleomorphic sarcoma, 63% the diagnostic value of this marker remains to be deter- Arch Pathol Lab Med—Vol 132, March 2008 Undifferentiated Tumor, Immunohistochemistry—Bahrami et al 339 mined. Positive immunostainings for MDM2 and CDK4, The tumor cells often coexpress CK (Figure 14) and vi- which correlate with amplification of these genes in de- mentin.266,268–270 However, because of the varied and un- differentiated liposarcomas, may help differentiate dedif- predictable CK immunoreactivity in sarcomatoid carci- ferentiated liposarcomas from poorly differentiated sar- nomas, multiple epithelial markers, including EMA and comas.258 various CK subtypes, such as pancytokeratin, HMW CK The diagnosis of osteosarcoma is essentially rendered (CK34␤E12 or CK5/6), and CK7, may be necessary to at the H&E light microscopic level by demonstration of demonstrate epithelial differentiation.267,269,270 Evaluation of bone or lacelike osteoid associated with malignant cells. a number of alternative immunostains beyond the tradi- Osteocalcin, which seems to specifically highlight the os- tional epithelial markers (CKs and EMA) in the sarco- teoid and the cytoplasm of osteoblasts (90%–100% specif- matoid carcinomas of the head and neck, lung, and uri- ic),77,78 may be potentially helpful in the diagnosis of os- nary bladder showed some values in exposing the epithe- teosarcoma in small biopsy material in the absence of rec- lial identity of tumors in that setting by adding p63 in the ognizable osteoid. Notably, osteosarcomas can be immu- panel.271 noreactive for actin and rarely for epithelial markers. Sarcomatoid Mesothelioma. Immunohistochemistry The diagnosis of pleomorphic MPNST is immunohis- has a limited role in the diagnosis of sarcomatoid meso- tochemically supported by demonstration of nerve sheath thelioma and its distinction from sarcomatoid carcinomas differentiation by a number of neural markers including and sarcomas. Sarcomatoid mesothelioma shows a S100, Leu-7 (CD57), myelin basic protein, and protein marked decline in the expression of epithelial and meso- gene product 9.5.259,260 The coexpression of these antigens thelial markers and has a wide immunophenotypic over- is more reassuring for the diagnosis because none of them lap with sarcomas and sarcomatoid carcinomas.272–274 Only in isolation is specific for nerve sheath tumors.259 A focal 57.8% of sarcomatoid mesotheliomas stain for calretinin, staining for S100 is seen in most MPNSTs (50%–60%),259,261 39% for thrombomodulin, 25% for WT1, 19.6% for CK5/ yet many high-grade MPNSTs display a decreased or neg- 6, and 75% for keratin.272 On the other hand, up to 60% of ative reactivity for S100 and CD57.262 them may stain for actin.273,274 It has been recently reported The diagnosis of high-grade spindle cell variant of PDSS that a combination of calretinin and D2-40 improves the is supported by demonstration of positive reaction for sensitivity for detecting mesothelioma in sarcomatoid ar- EMA, keratin and its subsets (CK7 and CK19), Bcl-2 an- eas to 66%.275 Nevertheless, because of the lack of a specific tigen, and CD9947–49,261 and negative staining for S100 and marker, the diagnosis of sarcomatoid mesothelioma CD34. In questionable cases, molecular testing may be re- should be made after exclusion of alternative diagnostic quired for definitive diagnosis. entities and only with an integration of clinical, radio- Pleomorphic MFH or undifferentiated high-grade pleo- graphic, gross, microscopic, and immunohistochemical morphic sarcoma is a group of pleomorphic sarcomas that findings. do not demonstrate a definitive line of differentiation even Spindle Cell MM. When dealing with a malignant with the application of immunohistochemistry. Malignant spindle cell tumor at any site, the possibility of spindle fibrous histiocytoma is therefore a diagnosis of exclusion, cell melanoma should always be considered. The sensitiv- which is made in the absence of reaction with any lineage- ity of S100 protein in spindle cell MM appears to be more selective markers in a high-grade pleomorphic sarcoma. or less similar to conventional MM,27 but HMB-45 expres- With current advances in diagnostic techniques, it is ex- sion is less frequent (positive in ϳ80% compared with pected that the number of tumors diagnosed as MFH will 90%–100% in the conventional type).26 be progressively smaller. Because a number of nonmesen- Of note, the immunohistologic features of desmoplastic chymal neoplasms may impart an MFH-like morphology, MM, a morphologic subtype of spindle cell MM associ- exclusion of these entities by immunohistochemistry is in- ated with pronounced desmoplasia, is substantially dif- evitable. Although keratin may stain a minor portion (up ferent from conventional melanoma and includes positive to 25%) of pleomorphic MFHs, the staining is often weak staining for S100 (in 94%) but lack of reactivity for HMB- and focal.55,263,264 The tumor cells may be focally positive 45 or other melanoma-specific markers.27,276 for smooth muscle actin or muscle-specific actin, but they are generally negative for desmin and S100. Rare desmin- TUMORS WITH OVERLAPPING HISTOLOGIC FEATURES positive cells in tumors with MFH-like morphology Tumors with nondescript H&E histology that fail to fit should not be regarded as evidence of myogenic differ- into either a sarcoma or a carcinoma category often pose entiation. the greatest diagnostic challenge. The differential diag- nosis is broad and includes melanoma, hematopoietic ma- Sarcoma-like Tumors lignancies, neuroendocrine tumors, poorly differentiated Sarcomatoid Carcinoma. Sarcomatoid (spindle cell) carcinomas, and mesenchymal neoplasms with epithelioid carcinomas are a group of poorly to undifferentiated car- appearance. Ancillary tests, including immunohistochem- cinomas that contain a component of spindle cell differ- istry, play an essential role in the recognition of these tu- entiation with or without differentiated heterologous ele- mors. ments, such as malignant cartilage, bone, or skeletal mus- cle.265,266 Some tumors are histologically biphasic with ob- Sarcomas With Epithelioid Appearance vious sarcomatous and carcinomatous areas, whereas The screening panel can often successfully disclose the others may show no obvious epithelial areas despite gen- broad line of differentiation in tumors with overlapping erous sampling.266,267 In such instances, the sarcomatoid histology (see ‘‘Broad Lineage Determination’’). The epi- foci are hardly distinguishable from true sarcomas at the thelioid morphology of sarcomas, which commonly cor- light microscopic level; therefore, immunohistochemistry relates with their potential CK expression, however, fur- plays a pivotal role in delineation of epithelial differenti- ther complicates the distinction of these tumors from ep- ation. ithelial malignancies. In a proper setting, a vimentinϩ/ 340 Arch Pathol Lab Med—Vol 132, March 2008 Undifferentiated Tumor, Immunohistochemistry—Bahrami et al Figure 15. Algorithmic approach to undifferentiated tumors with overlapping histologic features. CKs indicate cytokeratins; SMA, smooth muscle actin; PDSS, poorly differentiated synovial sarcoma; MPNST, malignant peripheral nerve sheath tumor; and PEComa, perivascular epithelioid cell tumors.

CKϪ or a coordinate vimentinϩ/CKϩ in a tumor with Miscellaneous Mesenchymal Tumors With overlapping histology should raise the possibility of sar- Epithelioid Appearance comas with epithelioid features and prompt performance Some mesenchymal neoplasms of borderline or uncer- of a battery of immunohistochemical stains for further tain malignant behavior may impart epithelioid appear- characterization of these tumors (Figure 15). Sarcomas ance and be considered in the differential diagnosis of with epithelioid cells include epithelioid sarcoma, epithe- undifferentiated carcinomas; a few examples of which are lioid angiosarcoma, clear cell sarcoma, epithelioid herein given. Perivascular epithelioid cell tumors are mes- MPNST, and other rare sarcomas, such as sclerosing epi- enchymal neoplasms of perivascular epithelioid cells,289 thelioid fibrosarcoma, alveolar soft part sarcoma, and ep- with a wide anatomic distribution, sometimes presenting ithelioid leiomyosarcoma. as unusual clear cell tumors in various locations.290–292 Im- The immunoprofile of epithelioid sarcoma includes pos- munohistochemically, these tumors are characterized by itive staining for vimentin (Figure 16, a), LMW CKs, and coexpression of melanocytic, including HMB-45 and Me- EMA and variable staining for CD34 (positive in 50%).277–280 lan-A, and smooth muscle markers.219,221,290 Sex cord–stro- The latter finding, if present (Figure 16, b), helps to dis- mal tumors are a heterogeneous group of neoplasms of tinguish epithelioid sarcoma from carcinomas, which are the ovary and testis with various histologic patterns, not rarely CD34 positive. uncommonly confused as an undifferentiated carcinoma. Most epithelioid angiosarcomas are positive for both vi- Epithelial membrane antigen, which is nearly always neg- mentin and CKs.281–284 These tumors are recognized by ex- ative, is a more reliable marker in distinction of these tu- pression of at least one of the endothelial markers, includ- mors from epithelial neoplasms because sex cord–stromal ing CD31, CD34, and factor VIII.281–283 FLI-1 protein has tumors, in particular granulosa cell tumor, may variably been recently shown to have equal or superior sensitivity stain for CKs (0%–64%).214,225,293 Both inhibin and calretinin and specificity compared with the traditional markers for are useful markers for confirming the diagno- vascular tumors60,61 (Figure 17). sis.212,214,215,224,225 Gastrointestinal stromal tumor, including Clear cell sarcoma has a similar immunohistochemical the epithelioid subtype, is immunohistochemically char- constituent to melanoma that includes expression of S100, acterized by expression of CD117 (c-Kit) and CD34 in HMB-45, and microphthalmia transcription factor in the most cases, rare expression of smooth muscle actin and majority of cases.285,286 It is, however, genetically different desmin, and negative staining for S100.294 from melanoma by its specific t(12;22) translocation.287 Clear cell sarcomas are typically negative for CK and Malignant Melanoma EMA. Epithelioid MPNST, a rare variant of MPNST, may be S100 is a sensitive, albeit not a specific, melanoma mark- recognized by positive staining for S100 (in 80%) and er, decorating more than 95% of MMs of primary and met- NSE.288 The epithelioid variant of pleomorphic liposarco- astatic sites. The diagnosis of MM requires confirmation mas in up to half of the cases may show reaction with with a melanocytic-specific marker, including HMB-45, S100 or epithelial markers, an important finding to be con- Melan-A, microphthalmia transcription factor, and tyros- sidered in the differential diagnosis of these tumors from inase. solid carcinomas with clear cell feature.253,255 The diagnosis HMB-45 is a specific monoclonal antibody against an of a number of rare sarcomas with epithelioid appearance, antigen present in the cytoplasm of neoplastic melanocytic for example, sclerosing epithelioid fibrosarcoma and al- cells. It shows positive cytoplasmic staining in the major- veolar soft part sarcoma, is essentially made on the basis ity of MMs (Ͼ95% in ethanol-fixed tissues and slightly of their distinct light microscopic features. less in formalin-fixed tissues), with the exception of des- Arch Pathol Lab Med—Vol 132, March 2008 Undifferentiated Tumor, Immunohistochemistry—Bahrami et al 341 moplastic MM, which rarely demonstrates HMB-45 ex- pression.27,295 Melan-A/MART-1 is a newer melanocytic marker with sensitivity comparable to that of HMB-45 for conventional melanoma. Like HMB-45, it is negative in most desmoplastic MMs.295 Although both Melan-A and HMB-45 are positive in more than 90% of conventional forms, their sensitivity declines in recurrent, spindle cell, or metastatic melanomas to 80%.15,26 For metastatic mela- nomas, however, Melan-A seems to be more sensitive (positive in 82%) than HMB-45 (positive in 76%).219,221 Both HMB-45 and Melan-A stain the PEComa family of neo- plasms,219,221 yet Melan-A also stains adrenal cortical neo- plasms.220 Malignant melanomas stain intensely for vimentin but are typically negative for epithelial markers, glial fibrillary acidic protein, CD45, and desmin. However, aberrant ex- pression for a number of immunostains, including CK, EMA, pCEA, or glial fibrillary acidic protein, has been rarely reported (also see ‘‘Malignant Melanoma’’ under ‘‘Small Round Cell Tumors’’).15,16,26,296 Hematopoietic Malignancies The spectrum of hematolymphoid malignancies pre- senting as a mass of large atypical cells of uncertain origin is broad and includes various types of lymphoma, myeloid sarcoma, and plasma cell neoplasm, each with a diverse immunoprofile, the complexity of which is far beyond the scope of this manuscript. Some examples of the unex- pected immunohistochemical expression in selected groups of hematopoietic tumors are herein provided. As mentioned previously, CD45 (LCA) is a valuable tool in the diagnostic separation of hematopoietic neoplasms, particularly of the lymphoid type, from poorly differen- tiated epithelial and mesenchymal neoplasms. Non-Hodg- kin lymphomas of the B- and T-cell types are immunore- active for LCA in 93% to 100% .28,29 A negative LCA, how- ever, does not absolutely preclude the diagnosis of lym- phoma in a large cell undifferentiated tumor because LCA is rarely absent in some lymphomas of the T-cell type and some plasma cell neoplasms. The immunoprofile of anaplastic large cell lymphoma is highly variable and includes a positive reaction for LCA (60%–70% of cases), CD30 (nearly all), T-cell markers (54%–70%), ALK protein (60%–85%), and EMA (ϳ75%).18,297–299 Hence, a potential CD45Ϫ/EMAϩ immu- noprofile in anaplastic large cell lymphoma may pose di- agnostic confusion with an undifferentiated carcinoma. On the other hand, the typical CKϪ/EMAϩ immunoprofile in anaplastic large cell lymphoma may help to differenti- ate it from embryonal carcinoma (CKϩ/EMAϪ), which is also reactive for CD30. Most but not all granulocytic sarcomas (ϳ75%) express LCA.112 CD43 is a more sensitive marker for myeloid sar- comas, staining up to 100% of cases with an intense wide- spread staining pattern.112 The diagnosis of myeloid tu- mor, however, requires demonstration of more specific markers, including myeloperoxidase and lysozyme. The immunoprofile complexity of neoplastic plasma cells (plasmacytoma or plasma cell myeloma), including Figure 16. Photomicrograph of epithelioid sarcoma showing positive variable expression for CD45, EMA, and CK, combined ϫ staining for vimentin (a) and CD34 (b) (original magnifications 20). with commonly negative reaction for B-cell markers Figure 17. Photomicrograph of epithelioid angiosarcoma showing (CD19, CD20), may pose potential diagnostic pitfalls in positive nuclear staining for FLI-1 protein (original magnification ϫ20). their distinction from epithelial neoplasms.10,33,34 A high index of suspicion should direct to a panel that includes CD38, CD138, and ␬ and ␭ light chains. Of note, although 342 Arch Pathol Lab Med—Vol 132, March 2008 Undifferentiated Tumor, Immunohistochemistry—Bahrami et al CD138 is an excellent marker of plasmacytic differentia- mas using monoclonal and polyclonal antibodies effective in routinely fixed wax embedded tissues. Histopathology. 1986;10:1243–1260. tion within the hematopoietic system (positive in nearly 21. ten Berge RL, Snijdewint FG, Mensdorff-Pouilly S, et al. MUC1 (EMA) is all), it is also expressed in various nonhematopoietic neo- preferentially expressed by ALK positive anaplastic large cell lymphoma, in the plasms; hence, it is not considered a definitive marker for normally glycosylated or only partly hypoglycosylated form. J Clin Pathol. 2001; 54:933–939. plasmacytic derivation, unless the hematolymphoid origin 22. Azumi N, Battifora H. The distribution of vimentin and keratin in epithelial of tumor is established.300 and nonepithelial neoplasms. A comprehensive immunohistochemical study on In summary, classification of a tumor beyond its appar- form. Am J Clin Pathol. 1987;88:286–296. 23. McNutt MA, Bolen JW, Gown AM, Hammar SP, Vogel AM. Coexpression ent undifferentiated morphology is possible in the major- of intermediate filaments in human epithelial neoplasms. Ultrastruct Pathol. 1985; ity of cases. This not only is important for prognostication 9:31–43. and therapy but should also provide further insights into 24. Dabbs DJ. Diagnostic Immunohistochemistry. 2nd ed. Philadelphia, Pa: the pathobiology of these tumors. The diagnostic pathway Churchill Livingstone Elsevier; 2006:194. 25. Herrera GA, Turbat-Herrera EA, Lott RL. S-100 protein expression by pri- is started with recognizing a tumor as SRCT, carcinoma- mary and metastatic adenocarcinomas. Am J Clin Pathol. 1988;89:168–176. tous tumor, sarcomatous (or sarcoma-like) tumor, or tu- 26. Banerjee SS, Harris M. Morphological and immunophenotypic variations mor with histologically overlapping features, followed by in malignant melanoma. Histopathology. 2000;36:387–402. 27. Longacre TA, Egbert BM, Rouse RV. Desmoplastic and spindle-cell malig- an immunohistochemical panel, which is judiciously con- nant melanoma: an immunohistochemical study. Am J Surg Pathol. 1996;20: structed for each tumor. Awareness of the diagnostic util- 1489–1500. ity of several tissue- or organ-specific immunomarkers, as 28. Michels S, Swanson PE, Frizzera G, Wick MR. Immunostaining for leu- kocyte common antigen using an amplified avidin-biotin-peroxidase complex summarized in this review, should help meet one of the method and paraffin sections: a study of 735 hematopoietic and nonhemato- most significant challenges in diagnostic pathology. poietic human neoplasms. Arch Pathol Lab Med. 1987;111:1035–1039. 29. Kurtin PJ, Pinkus GS. Leukocyte common antigen—a diagnostic discrimi- References nant between hematopoietic and nonhematopoietic neoplasms in paraffin sec- 1. Coindre JM, Tanguy F, Merlio JP, De Mascarel I, De Mascarel A, Trojani M. tions using monoclonal antibodies: correlation with immunologic studies and The value of immunohistological techniques in undifferentiated cancers. Tumori. ultrastructural localization. Hum Pathol. 1985;16:353–365. 1986;72:539–544. 30. Van Eyken P, Wolf-Peeters C, Van den OJ, Tricot G, Desmet V. Expression 2. Gatter KC, Alcock C, Heryet A, Mason DY. Clinical importance of analysing of leukocyte common antigen in lymphoblastic lymphoma and small noncleaved malignant tumours of uncertain origin with immunohistological techniques. Lan- undifferentiated non-Burkitt’s lymphoma: an immunohistochemical study. J cet. 1985;1:1302–1305. Pathol. 1987;151:257–261. 3. Vege DS, Soman CS, Joshi UA, Ganesh B, Yadav JN. Undifferentiated tu- 31. Ozdemirli M, Fanburg-Smith JC, Hartmann DP, et al. Precursor B-Lympho- mors: an immunohistochemical analysis on biopsies. J Surg Oncol. 1994;57:273– blastic lymphoma presenting as a solitary bone tumor and mimicking Ewing’s 276. sarcoma: a report of four cases and review of the literature. Am J Surg Pathol. 4. Kamel OW, Rouse RV, Warnke RA. Heterogeneity of epithelial marker ex- 1998;22:795–804. pression in routinely processed, poorly differentiated carcinomas. Arch Pathol Lab 32. Kahwash SB, Qualman SJ. Cutaneous lymphoblastic lymphoma in chil- Med. 1991;115:566–570. dren: report of six cases with precursor B-cell lineage. Pediatr Dev Pathol. 2002; 5. Gaffey MJ, Traweek ST, Mills SE, et al. Cytokeratin expression in adreno- 5:45–53. cortical neoplasia: an immunohistochemical and biochemical study with impli- 33. Kumar S, Rajkumar SV, Kimlinger T, Greipp PR, Witzig TE. CD45 expres- cations for the differential diagnosis of adrenocortical, hepatocellular, and renal sion by bone marrow plasma cells in multiple myeloma: clinical and biological cell carcinoma. Hum Pathol. 1992;23:144–153. correlations. Leukemia. 2005;19:1466–1470. 6. Wick MR, Cherwitz DL, McGlennen RC, Dehner LP. Adrenocortical carci- 34. Wotherspoon AC, Norton AJ, Isaacson PG. Immunoreactive cytokeratins noma: an immunohistochemical comparison with renal cell carcinoma. Am J in plasmacytomas. Histopathology. 1989;14:141–150. Pathol. 1986;122:343–352. 35. Ferry JA, Young RH, Scully RE. Testicular and epididymal plasmacytoma: 7. Cote RJ, Cordon-Cardo C, Reuter VE, Rosen PP. Immunopathology of ad- a report of 7 cases, including three that were the initial manifestation of plasma renal and renal cortical tumors: coordinated change in antigen expression is as- cell myeloma. Am J Surg Pathol. 1997;21:590–598. sociated with neoplastic conversion in the adrenal cortex. Am J Pathol. 1990; 36. Clavio M, Rossi E, Truini M, et al. Anaplastic large cell lymphoma: a clin- 136:1077–1084. icopathologic study of 53 patients. Leuk Lymphoma. 1996;22:319–327. 8. Miettinen M. Keratin subsets in spindle cell sarcomas: keratins are wide- 37. Perkins PL, Ross CW, Schnitzer B. CD30-positive, anaplastic large-cell spread but synovial sarcoma contains a distinctive keratin polypeptide pattern lymphomas that express CD15 but lack CD45: a possible diagnostic pitfall. Arch and desmoplakins. Am J Pathol. 1991;138:505–513. Pathol Lab Med. 1992;116:1192–1196. 9. Banks ER, Jansen JF, Oberle E, Davey DD. Cytokeratin positivity in fine- 38. Nandedkar MA, Palazzo J, Abbondanzo SL, Lasota J, Miettinen M. CD45 needle aspirates of melanomas and sarcomas. Diagn Cytopathol. 1995;12:230– (leukocyte common antigen) immunoreactivity in metastatic undifferentiated and 233. neuroendocrine carcinoma: a potential diagnostic pitfall. Mod Pathol. 1998;11: 10. Shin JS, Stopyra GA, Warhol MJ, Multhaupt HA. Plasmacytoma with ab- 1204–1210. errant expression of myeloid markers, T-cell markers, and cytokeratin. J Histo- 39. Devoe K, Weidner N. Immunohistochemistry of small round-cell tumors. chem Cytochem. 2001;49:791–792. Semin Diagn Pathol. 2000;17:216–224. 11. Menestrina F, Lestani M, Scarpa A, et al. Common acute lymphoblastic 40. Folpe AL, Goldblum JR, Rubin BP, et al. Morphologic and immunophe- leukaemia-lymphoma expressing cytokeratin: a case report. Virchows Arch. 1994; notypic diversity in Ewing family tumors: a study of 66 genetically confirmed 425:83–87. cases. Am J Surg Pathol. 2005;29:1025–1033. 12. Beschorner R, Horny HP, Petruch UR, Kaiserling E. Frequent expression of 41. Weidner N, Tjoe J. Immunohistochemical profile of monoclonal antibody haemopoietic and non-haemopoietic antigens by reactive plasma cells: an im- O13: antibody that recognizes glycoprotein p30/32MIC2 and is useful in diag- munohistochemical study using formalin-fixed, paraffin-embedded tissue. Histol nosing Ewing’s sarcoma and peripheral neuroepithelioma. Am J Surg Pathol. Histopathol. 1999;14:805–812. 1994;18:486–494. 13. Listrom MB, Dalton LW. Comparison of keratin monoclonal antibodies 42. Hess E, Cohen C, DeRose PB, Yost BA, Costa MJ. Nonspecificity of p30/ MAK-6, AE1:AE3, and CAM-5.2. Am J Clin Pathol. 1987;88:297–301. 32MIC2 immunolocalization with the O13 monoclonal antibody in the diagnosis 14. Gustmann C, Altmannsberger M, Osborn M, Griesser H, Feller AC. Cyto- of Ewing’s sarcoma: application of an algorithmic immunohistochemical analysis. keratin expression and vimentin content in large cell anaplastic lymphomas and Appl Immunohistochem. 1997;5:94–103. other non-Hodgkin’s lymphomas. Am J Pathol. 1991;138:1413–1422. 43. Scotlandi K, Serra M, Manara MC, et al. Immunostaining of the p30/ 15. Bishop PW, Menasce LP, Yates AJ, Win NA, Banerjee SS. An immunophe- 32MIC2 antigen and molecular detection of EWS rearrangements for the diag- notypic survey of malignant melanomas. Histopathology. 1993;23:159–166. nosis of Ewing’s sarcoma and peripheral neuroectodermal tumor. Hum Pathol. 16. Selby WL, Nance KV, Park HK. CEA immunoreactivity in metastatic ma- 1996;27:408–416. lignant melanoma. Mod Pathol. 1992;5:415–419. 44. Ramani P, Rampling D, Link M. Immunocytochemical study of 12E7 in 17. Pinkus GS, Kurtin PJ. Epithelial membrane antigen—a diagnostic discrim- small round-cell tumours of childhood: an assessment of its sensitivity and spec- inant in surgical pathology: immunohistochemical profile in epithelial, mesen- ificity. Histopathology. 1993;23:557–561. chymal, and hematopoietic neoplasms using paraffin sections and monoclonal 45. Riopel M, Dickman PS, Link MP, Perlman EJ. MIC2 analysis in pediatric antibodies. Hum Pathol. 1985;16:929–940. lymphomas and leukemias. Hum Pathol. 1994;25:396–399. 18. Ioachim HL, Ratech H. Ioachim’s Lymph Node Pathology. 3rd ed. Phila- 46. Zhang PJ, Barcos M, Stewart CC, Block AW, Sait S, Brooks JJ. Immunore- delphia, Pa: Lippincott Williams & Wilkins; 2002:341–476. activity of MIC2 (CD99) in acute myelogenous leukemia and related diseases. 19. Delsol G, Gatter KC, Stein H, et al. Human lymphoid cells express epi- Mod Pathol. 2000;13:452–458. thelial membrane antigen. Implications for diagnosis of human neoplasms. Lan- 47. van de Rijn M, Barr FG, Xiong QB, Hedges M, Shipley J, Fisher C. Poorly cet. 1984;2:1124–1129. differentiated synovial sarcoma: an analysis of clinical, pathologic, and molecular 20. Norton AJ, Isaacson PG. An immunocytochemical study of T-cell lympho- genetic features. Am J Surg Pathol. 1999;23:106–112.

Arch Pathol Lab Med—Vol 132, March 2008 Undifferentiated Tumor, Immunohistochemistry—Bahrami et al 343 48. Folpe AL, Schmidt RA, Chapman D, Gown AM. Poorly differentiated sy- tin immunoreactivity in extraskeletal osteosarcoma: a study of 28 cases. Hum novial sarcoma: immunohistochemical distinction from primitive neuroectoder- Pathol. 1999;30:32–38. mal tumors and high-grade malignant peripheral nerve sheath tumors. Am J Surg 78. Fanburg JC, Rosenberg AE, Weaver DL, et al. Osteocalcin and osteonectin Pathol. 1998;22:673–682. immunoreactivity in the diagnosis of osteosarcoma. Am J Clin Pathol. 1997;108: 49. Pelmus M, Guillou L, Hostein I, Sierankowski G, Lussan C, Coindre JM. 464–473. Monophasic fibrous and poorly differentiated synovial sarcoma: immunohisto- 79. Hasegawa T, Hirose T, Kudo E, Hizawa K, Usui M, Ishii S. Immunophe- chemical reassessment of 60 t(X;18)(SYT-SSX)-positive cases. Am J Surg Pathol. notypic heterogeneity in osteosarcomas. Hum Pathol. 1991;22:583–590. 2002;26:1434–1440. 80. Okada K, Hasegawa T, Yokoyama R, Beppu Y, Itoi E. Osteosarcoma with 50. Hoang MP, Suarez PA, Donner LR, et al. Mesenchymal chondrosarcoma: cytokeratin expression: a clinicopathological study of six cases with an emphasis a small cell neoplasm with polyphenotypic differentiation. Int J Surg Pathol. 2000; on differential diagnosis from metastatic cancer. J Clin Pathol. 2003;56:742–746. 8:291–301. 81. Muir TE, Cheville JC, Lager DJ. Metanephric adenoma, nephrogenic rests, 51. Granter SR, Renshaw AA, Fletcher CD, Bhan AK, Rosenberg AE. CD99 and Wilms’ tumor: a histologic and immunophenotypic comparison. Am J Surg reactivity in mesenchymal chondrosarcoma. Hum Pathol. 1996;27:1273–1276. Pathol. 2001;25:1290–1296. 52. Zhang PJ, Goldblum JR, Pawel BR, Fisher C, Pasha TL, Barr FG. Immu- 82. Domagala W, Chosia M, Bedner E, Kubicka C, Weber K, Osborn M. Im- nophenotype of desmoplastic small round cell tumors as detected in cases with munocytochemistry in fine needle aspirates of small cell-, round-, blue-cell ma- EWS-WT1 gene fusion product. Mod Pathol. 2003;16:229–235. lignant tumors of childhood (neuroblastoma, nephroblastoma, lymphoma, Ew- 53. Hicks J, Flaitz C. Rhabdomyosarcoma of the head and neck in children. ing’s sarcoma, rhabdomyosarcoma). Patol Pol. 1991;42:79–82. Oral Oncol. 2002;38:450–459. 83. Folpe AL, Patterson K, Gown AM. Antibodies to desmin identify the blas- 54. Jimenez RE, Folpe AL, Lapham RL, et al. Primary Ewing’s sarcoma/primitive temal component of nephroblastoma. Mod Pathol. 1997;10:895–900. neuroectodermal tumor of the kidney: a clinicopathologic and immunohisto- 84. Hasegawa T, Hirose T, Ayala AG, et al. Adult neuroblastoma of the retro- chemical analysis of 11 cases. Am J Surg Pathol. 2002;26:320–327. peritoneum and abdomen: clinicopathologic distinction from primitive neuroec- 55. Weiss SW, Goldblum JR. Enzinger and Weiss’s Soft Tissue Tumors. 4th ed. todermal tumor. Am J Surg Pathol. 2001;25:918–924. St Louis, Mo: Mosby; 2001:235, 549. 85. Franquemont DW, Mills SE, Lack EE. Immunohistochemical detection of 56. Garin-Chesa P, Fellinger EJ, Huvos AG, et al. Immunohistochemical anal- neuroblastomatous foci in composite adrenal pheochromocytoma-neuroblasto- ysis of neural cell adhesion molecules: differential expression in small round cell ma. Am J Clin Pathol. 1994;102:163–170. tumors of childhood and adolescence. Am J Pathol. 1991;139:275–286. 86. Wirnsberger GH, Becker H, Ziervogel K, Hofler H. Diagnostic immuno- 57. Miettinen M, Cupo W. Neural cell adhesion molecule distribution in soft histochemistry of neuroblastic tumors. Am J Surg Pathol. 1992;16:49–57. tissue tumors. Hum Pathol. 1993;24:62–66. 87. Hachitanda Y, Tsuneyoshi M, Enjoji M. An ultrastructural and immunohis- 58. Ravandi F, Cortes J, Estrov Z, et al. CD56 expression predicts occurrence tochemical evaluation of cytodifferentiation in neuroblastic tumors. Mod Pathol. of CNS disease in acute lymphoblastic leukemia. Leuk Res. 2002;26:643–649. 1989;2:13–19. 59. Khanlari B, Buser A, Lugli A, Tichelli A, Dirnhofer S. The expression pattern 88. Frierson HF Jr, Ross GW, Mills SE, Frankfurter A. Olfactory neuroblastoma: of CD56 (N-CAM) in human bone marrow biopsies infiltrated by acute leukemia. additional immunohistochemical characterization. Am J Clin Pathol. 1990;94: Leuk Lymphoma. 2003;44:2055–2059. 547–553. 60. Folpe AL, Hill CE, Parham DM, O’Shea PA, Weiss SW. Immunohistochem- 89. Banerjee AK, Sharma BS, Vashista RK, Kak VK. Intracranial olfactory neu- ical detection of FLI-1 protein expression: a study of 132 round cell tumors with roblastoma: evidence for olfactory epithelial origin. J Clin Pathol. 1992;45:299– emphasis on CD99-positive mimics of Ewing’s sarcoma/primitive neuroectoder- 302. mal tumor. Am J Surg Pathol. 2000;24:1657–1662. 90. Moll R, Franke WW. Cytoskeletal differences between human neuroen- 61. Rossi S, Orvieto E, Furlanetto A, Laurino L, Ninfo V, Dei Tos AP. Utility of docrine tumors: a cytoskeletal protein of molecular weight 46,000 distinguishes the immunohistochemical detection of FLI-1 expression in round cell and vas- cutaneous from pulmonary neuroendocrine neoplasms. Differentiation. 1985;30: cular neoplasm using a monoclonal antibody. Mod Pathol. 2004;17:547–552. 165–175. 62. Collini P, Sampietro G, Bertulli R, et al. Cytokeratin immunoreactivity in 91. Mertz H, Vyberg M, Paulsen, SM, Teglbjærg, PS. Immunohistochemical 41 cases of ES/PNET confirmed by molecular diagnostic studies. Am J Surg Pathol. detection of neuroendocrine markers in tumors of the lungs and gastrointestinal 2001;25:273–274. tract. Appl Immunohistochem. 1998;6:175–180. 63. Gu M, Antonescu CR, Guiter G, Huvos AG, Ladanyi M, Zakowski MF. 92. Shipley WR, Hammer RD, Lennington WJ, Macon WR. Paraffin immuno- Cytokeratin immunoreactivity in Ewing’s sarcoma: prevalence in 50 cases con- histochemical detection of CD56, a useful marker for neural cell adhesion mol- firmed by molecular diagnostic studies. Am J Surg Pathol. 2000;24:410–416. ecule (NCAM), in normal and neoplastic fixed tissues. Appl Immunohistochem. 64. Kumar S, Perlman E, Harris CA, Raffeld M, Tsokos M. Myogenin is a spe- cific marker for rhabdomyosarcoma: an immunohistochemical study in paraffin- 1997;5:87–93. embedded tissues. Mod Pathol. 2000;13:988–993. 93. Chan JK, Suster S, Wenig BM, Tsang WY, Chan JB, Lau AL. Cytokeratin 20 65. Sebire NJ, Malone M. Myogenin and MyoD1 expression in paediatric immunoreactivity distinguishes Merkel cell (primary cutaneous neuroendocrine) rhabdomyosarcomas. J Clin Pathol. 2003;56:412–416. carcinomas and salivary gland small cell carcinomas from small cell carcinomas 66. Wang NP, Marx J, McNutt MA, Rutledge JC, Gown AM. Expression of of various sites. Am J Surg Pathol. 1997;21:226–234. myogenic regulatory proteins (myogenin and MyoD1) in small blue round cell 94. Miettinen M. Keratin 20: immunohistochemical marker for gastrointestinal, tumors of childhood. Am J Pathol. 1995;147:1799–1810. urothelial, and Merkel cell carcinomas. Mod Pathol. 1995;8:384–388. 67. Cessna MH, Zhou H, Perkins SL, et al. Are myogenin and myoD1 expres- 95. Nagao T, Gaffey TA, Olsen KD, Serizawa H, Lewis JE. Small cell carcinoma sion specific for rhabdomyosarcoma? A study of 150 cases, with emphasis on of the major salivary glands: clinicopathologic study with emphasis on cytokeratin spindle cell mimics. Am J Surg Pathol. 2001;25:1150–1157. 20 immunoreactivity and clinical outcome. Am J Surg Pathol. 2004;28:762–770. 68. Carpentieri DF, Nichols K, Chou PM, Matthews M, Pawel B, Huff D. The 96. Shin SJ, DeLellis RA, Ying L, Rosen PP. Small cell carcinoma of the breast: expression of WT1 in the differentiation of rhabdomyosarcoma from other pedi- a clinicopathologic and immunohistochemical study of nine patients. Am J Surg atric small round blue cell tumors. Mod Pathol. 2002;15:1080–1086. Pathol. 2000;24:1231–1238. 69. Gerald WL, Ladanyi M, de Alava E, et al. Clinical, pathologic, and molec- 97. Bacchi CE, Garcia RL, Gown AM. Immunolocalization of estrogen and ular spectrum of tumors associated with t(11;22)(p13;q12): desmoplastic small progesterone receptors in neuroendocrine tumors of the lung, skin, gastrointes- round-cell tumor and its variants. J Clin Oncol. 1998;16:3028–3036. tinal and female genital tracts. Appl Immunohistochem. 1997;5:17–22. 70. Ordonez NG. Desmoplastic small round cell tumor, II: an ultrastructural 98. Viale G, Doglioni C, Gambacorta M, Zamboni G, Coggi G, Bordi C. Pro- and immunohistochemical study with emphasis on new immunohistochemical gesterone receptor immunoreactivity in pancreatic endocrine tumors: an immu- markers. Am J Surg Pathol. 1998;22:1314–1327. nocytochemical study of 156 neuroendocrine tumors of the pancreas, gastroin- 71. Barnoud R, Sabourin JC, Pasquier D, et al. Immunohistochemical expres- testinal and respiratory tracts, and skin. Cancer. 1992;70:2268–2277. sion of WT1 by desmoplastic small round cell tumor: a comparative study with 99. Johansson L. Histopathologic classification of lung cancer: relevance of other small round cell tumors. Am J Surg Pathol. 2000;24:830–836. cytokeratin and TTF-1 immunophenotyping. Ann Diagn Pathol. 2004;8:259–267. 72. Hill DA, Pfeifer JD, Marley EF, et al. WT1 staining reliably differentiates 100. Cheuk W, Kwan MY, Suster S, Chan JK. Immunostaining for thyroid tran- desmoplastic small round cell tumor from Ewing sarcoma/primitive neuroecto- scription factor 1 and cytokeratin 20 aids the distinction of small cell carcinoma dermal tumor: an immunohistochemical and molecular diagnostic study. Am J from Merkel cell carcinoma, but not pulmonary from extrapulmonary small cell Clin Pathol. 2000;114:345–353. carcinomas. Arch Pathol Lab Med. 2001;125:228–231. 73. Dei Tos AP, Wadden C, Calonje E, et al. Immunohistochemical demon- 101. Ordonez NG. Value of thyroid transcription factor-1 immunostaining in stration of glycoprotein p30/32 MIC2 (CD99) in synovial sarcoma: a potential distinguishing small cell lung carcinomas from other small cell carcinomas. Am cause of diagnostic confusion. Appl Immunohistochem. 1995;31:68–73. J Surg Pathol. 2000;24:1217–1223. 74. Mechtersheimer G, Staudter M, Moller P. Expression of the natural killer 102. Zhang H, Liu J, Cagle PT, Allen TC, Laga AC, Zander DS. Distinction of cell-associated antigens CD56 and CD57 in human neural and striated muscle pulmonary small cell carcinoma from poorly differentiated squamous cell carci- cells and in their tumors. Cancer Res. 1991;51:1300–1307. noma: an immunohistochemical approach. Mod Pathol. 2005;18:111–118. 75. Devaney K, Vinh TN, Sweet DE. Small cell osteosarcoma of bone: an im- 103. Kaufmann O, Dietel M. Expression of thyroid transcription factor-1 in munohistochemical study with differential diagnostic considerations. Hum Pathol. pulmonary and extrapulmonary small cell carcinomas and other neuroendocrine 1993;24:1211–1225. carcinomas of various primary sites. Histopathology. 2000;36:415–420. 76. Nakajima H, Sim FH, Bond JR, Unni KK. Small cell osteosarcoma of bone: 104. McCluggage WG, Oliva E, Connolly LE, McBride HA, Young RH. An review of 72 cases. Cancer. 1997;79:2095–2106. immunohistochemical analysis of ovarian small cell carcinoma of hypercalcemic 77. Fanburg-Smith JC, Bratthauer GL, Miettinen M. Osteocalcin and osteonec- type. Int J Gynecol Pathol. 2004;23:330–336.

344 Arch Pathol Lab Med—Vol 132, March 2008 Undifferentiated Tumor, Immunohistochemistry—Bahrami et al 105. McCluggage WG. Ovarian neoplasms composed of small round cells: a chemical study of primary and metastatic carcinomas of the liver. Am J Clin review. Adv Anat Pathol. 2004;11:288–296. Pathol. 1993;99:551–557. 106. Oesterling JE, Hauzeur CG, Farrow GM. Small cell anaplastic carcinoma 135. Oates J, Edwards C. HBME-1, MOC-31, WT1 and calretinin: an assess- of the prostate: a clinical, pathological and immunohistological study of 27 pa- ment of recently described markers for mesothelioma and adenocarcinoma. His- tients. J Urol. 1992;147:804–807. topathology. 2000;36:341–347. 107. Parker DC, Folpe AL, Bell J, et al. Potential utility of uroplakin III, throm- 136. Sosolik RC, McGaughy VR, De Young BR. Anti-MOC-31: a potential ad- bomodulin, high molecular weight cytokeratin, and cytokeratin 20 in noninva- dition to the pulmonary adenocarcinoma versus mesothelioma immunohisto- sive, invasive, and metastatic urothelial (transitional cell) carcinomas. Am J Surg chemistry panel. Mod Pathol. 1997;10:716–719. Pathol. 2003;27:1–10. 137. Proca DM, Niemann TH, Porcell AI, DeYoung BR. MOC31 immunore- 108. Erickson LA, Papouchado B, Dimashkieh H, Zhang S, Nakamura N, Lloyd activity in primary and metastatic carcinoma of the liver: report of findings and RV. Cdx2 as a marker for neuroendocrine tumors of unknown primary sites. En- review of other utilized markers. Appl Immunohistochem Mol Morphol. 2000;8: docr Pathol. 2004;15:247–252. 120–125. 109. Franchi A, Pasquinelli G, Cenacchi G, et al. Immunohistochemical and 138. Yang A, McKeon F. P63 and P73: P53 mimics, menaces and more. Nat ultrastructural investigation of neural differentiation in Ewing sarcoma/PNET of Rev Mol Cell Biol. 2000;1:199–207. bone and soft tissues. Ultrastruct Pathol. 2001;25:219–225. 139. Kaufmann O, Fietze E, Mengs J, Dietel M. Value of p63 and cytokeratin 110. Cerilli LA, Holst VA, Brandwein MS, Stoler MH, Mills SE. Sinonasal un- 5/6 as immunohistochemical markers for the differential diagnosis of poorly dif- differentiated carcinoma: immunohistochemical profile and lack of EBV associ- ferentiated and undifferentiated carcinomas. Am J Clin Pathol. 2001;116:823– ation. Am J Surg Pathol. 2001;25:156–163. 830. 111. Ozdemirli M, Fanburg-Smith JC, Hartmann DP, Azumi N, Miettinen M. 140. Wang BY, Gil J, Kaufman D, Gan L, Kohtz DS, Burstein DE. P63 in pul- Differentiating lymphoblastic lymphoma and Ewing’s sarcoma: lymphocyte mark- monary epithelium, pulmonary squamous neoplasms, and other pulmonary tu- ers and gene rearrangement. Mod Pathol. 2001;14:1175–1182. mors. Hum Pathol. 2002;33:921–926. 112. Traweek ST, Arber DA, Rappaport H, Brynes RK. Extramedullary myeloid 141. Wu M, Wang B, Gil J, et al. p63 and TTF-1 immunostaining: a useful cell tumors: an immunohistochemical and morphologic study of 28 cases. Am J marker panel for distinguishing small cell carcinoma of lung from poorly differ- Surg Pathol. 1993;17:1011–1019. entiated squamous cell carcinoma of lung. Am J Clin Pathol. 2003;119:696–702. 113. Thompson LD, Wieneke JA, Miettinen M. Sinonasal tract and nasopha- 142. Au NH, Gown AM, Cheang M, et al. P63 expression in lung carcinoma: ryngeal melanomas: a clinicopathologic study of 115 cases with a proposed stag- a tissue microarray study of 408 cases. Appl Immunohistochem Mol Morphol. ing system. Am J Surg Pathol. 2003;27:594–611. 2004;12:240–247. 114. Wilkerson AE, Glasgow MA, Hiatt KM. Immunoreactivity of CD99 in 143. Chu PG, Weiss LM. Expression of cytokeratin 5/6 in epithelial neoplasms: invasive malignant melanoma. J Cutan Pathol. 2006;33:663–666. an immunohistochemical study of 509 cases. Mod Pathol. 2002;15:6–10. 115. Nakatsuka S, Oji Y, Horiuchi T, et al. Immunohistochemical detection of 144. Kaufmann O, Georgi T, Dietel M. Utility of 123C3 monoclonal antibody WT1 protein in a variety of cancer cells. Mod Pathol. 2006;19:804–814. against CD56 (NCAM) for the diagnosis of small cell carcinomas on paraffin 116. Bilalovic N, Sandstad B, Golouh R, Nesland JM, Selak I, Torlakovic EE. sections. Hum Pathol. 1997;28:1373–1378. CD10 protein expression in tumor and stromal cells of malignant melanoma is 145. Lantuejoul S, Moro D, Michalides RJ, Brambilla C, Brambilla E. Neural associated with tumor progression. Mod Pathol. 2004;17:1251–1258. cell adhesion molecules (NCAM) and NCAM-PSA expression in neuroendocrine 117. Chu P, Arber DA. Paraffin-section detection of CD10 in 505 nonhema- lung tumors. Am J Surg Pathol. 1998;22:1267–1276. topoietic neoplasms: frequent expression in renal cell carcinoma and endometrial 146. Rossi G, Marchioni A, Milani M, et al. TTF-1, cytokeratin 7, 34␤E12, and stromal sarcoma. Am J Clin Pathol. 2000;113:374–382. CD56/NCAM immunostaining in the subclassification of large cell carcinomas of 118. Pernick NL, DaSilva M, Gangi MD, Crissman J, Adsay V. ‘‘Histiocytic the lung. Am J Clin Pathol. 2004;122:884–893. markers’’ in melanoma. Mod Pathol. 1999;12:1072–1077. 147. Wong KF, Chan JK, Ng CS. CD56 (NCAM)-positive malignant lymphoma. 119. Wick MR, Stanley SJ, Swanson PE. Immunohistochemical diagnosis of Leuk Lymphoma. 1994;14:29–36. sinonasal melanoma, carcinoma, and neuroblastoma with monoclonal antibodies 148. Wick MR. Immunohistology of neuroendocrine and neuroectodermal tu- HMB-45 and anti-synaptophysin. Arch Pathol Lab Med. 1988;112:616–620. mors. Semin Diagn Pathol. 2000;17:194–203. 120. Wang NP, Zee S, Zarbo RJ. Coordinate expression of cytokeratins 7 and 149. Wick MR, Berg LC, Hertz MI. Large cell carcinoma of the lung with 20 defines unique subsets of carcinomas. Appl Immunohistochem. 1995;3:99– neuroendocrine differentiation: a comparison with large cell ‘‘undifferentiated’’ 107. pulmonary tumors. Am J Clin Pathol. 1992;97:796–805. 121. Shah KD, Tabibzadeh SS, Gerber MA. Comparison of cytokeratin expres- 150. Sturm N, Lantuejoul S, Laverriere MH, et al. Thyroid transcription factor sion in primary and metastatic carcinomas: diagnostic application in surgical 1 and cytokeratins 1, 5, 10, 14 (34betaE12) expression in basaloid and large-cell pathology. Am J Clin Pathol. 1987;87:708–715. neuroendocrine carcinomas of the lung. Hum Pathol. 2001;32:918–925. 122. Loy TS, Calaluce RD, Keeney GL. Cytokeratin immunostaining in differ- 151. Lyda MH, Weiss LM. Immunoreactivity for epithelial and neuroendocrine entiating primary ovarian carcinoma from metastatic colonic adenocarcinoma. antibodies are useful in the differential diagnosis of lung carcinomas. Hum Pathol. Mod Pathol. 1996;9:1040–1044. 2000;31:980–987. 123. Bassily NH, Vallorosi CJ, Akdas G, Montie JE, Rubin MA. Coordinate 152. Folpe AL, Gown AM, Lamps LW, et al. Thyroid transcription factor-1: expression of cytokeratins 7 and 20 in prostate adenocarcinoma and bladder immunohistochemical evaluation in pulmonary neuroendocrine tumors. Mod urothelial carcinoma. Am J Clin Pathol. 2000;113:383–388. Pathol. 1999;12:5–8. 124. Chu P, Wu E, Weiss LM. Cytokeratin 7 and cytokeratin 20 expression in 153. Nakamura N, Miyagi E, Murata S, Kawaoi A, Katoh R. Expression of epithelial neoplasms: a survey of 435 cases. Mod Pathol. 2000;13:962–972. thyroid transcription factor-1 in normal and neoplastic lung tissues. Mod Pathol. 125. Maeda T, Kajiyama K, Adachi E, Takenaka K, Sugimachi K, Tsuneyoshi 2002;15:1058–1067. M. The expression of cytokeratins 7, 19, and 20 in primary and metastatic car- 154. Miettinen M, Franssila KO. Variable expression of keratins and nearly cinomas of the liver. Mod Pathol. 1996;9:901–909. uniform lack of thyroid transcription factor 1 in thyroid anaplastic carcinoma. 126. Ordonez NG. Value of cytokeratin 5/6 immunostaining in distinguishing Hum Pathol. 2000;31:1139–1145. epithelial mesothelioma of the pleura from lung adenocarcinoma. Am J Surg 155. Agoff SN, Lamps LW, Philip AT, et al. Thyroid transcription factor-1 is Pathol. 1998;22:1215–1221. expressed in extrapulmonary small cell carcinomas but not in other extrapul- 127. Ordonez NG. The diagnostic utility of immunohistochemistry in distin- monary neuroendocrine tumors. Mod Pathol. 2000;13:238–242. guishing between epithelioid mesotheliomas and squamous carcinomas of the 156. Carcangiu ML, Steeper T, Zampi G, Rosai J. Anaplastic thyroid carcinoma: lung: a comparative study. Mod Pathol. 2006;19:417–428. a study of 70 cases. Am J Clin Pathol. 1985;83:135–158. 128. Ordonez NG. The immunohistochemical diagnosis of mesothelioma: a 157. Ordonez NG, El Naggar AK, Hickey RC, Samaan NA. Anaplastic thyroid comparative study of epithelioid mesothelioma and lung adenocarcinoma. Am J carcinoma: immunocytochemical study of 32 cases. Am J Clin Pathol. 1991;96: Surg Pathol. 2003;27:1031–1051. 15–24. 129. Carella R, Deleonardi G, D’Errico A, et al. Immunohistochemical panels 158. Venkatesh YS, Ordonez NG, Schultz PN, Hickey RC, Goepfert H, Samaan for differentiating epithelial malignant mesothelioma from lung adenocarcinoma: NA. Anaplastic carcinoma of the thyroid: a clinicopathologic study of 121 cases. a study with logistic regression analysis. Am J Surg Pathol. 2001;25:43–50. Cancer. 1990;66:321–330. 130. Miettinen M, Sarlomo-Rikala M. Expression of calretinin, thrombomo- 159. Wennerberg AE, Nalesnik MA, Coleman WB. Hepatocyte paraffin 1: a dulin, keratin 5, and mesothelin in lung carcinomas of different types: an im- monoclonal antibody that reacts with hepatocytes and can be used for differential munohistochemical analysis of 596 tumors in comparison with epithelioid me- diagnosis of hepatic tumors. Am J Pathol. 1993;143:1050–1054. sotheliomas of the pleura. Am J Surg Pathol. 2003;27:150–158. 160. Lau SK, Prakash S, Geller SA, Alsabeh R. Comparative immunohisto- 131. Sheahan K, O’Brien MJ, Burke B, et al. Differential reactivities of carci- chemical profile of hepatocellular carcinoma, cholangiocarcinoma, and meta- noembryonic antigen (CEA) and CEA-related monoclonal and polyclonal anti- static adenocarcinoma. Hum Pathol. 2002;33:1175–1181. bodies in common epithelial malignancies. Am J Clin Pathol. 1990;94:157–164. 161. Morrison C, Marsh W Jr, Frankel WL. A comparison of CD10 to pCEA, 132. Berezowski K, Stastny JF, Kornstein MJ. Cytokeratins 7 and 20 and car- MOC-31, and hepatocyte for the distinction of malignant tumors in the liver. Mod cinoembryonic antigen in ovarian and colonic carcinoma. Mod Pathol. 1996;9: Pathol. 2002;15:1279–1287. 426–429. 162. Siddiqui MT, Saboorian MH, Gokaslan ST, Ashfaq R. Diagnostic utility of 133. Alkushi A, Irving J, Hsu F, et al. Immunoprofile of cervical and endo- the HepPar1 antibody to differentiate hepatocellular carcinoma from metastatic metrial adenocarcinomas using a tissue microarray. Virchows Arch. 2003;442: carcinoma in fine-needle aspiration samples. Cancer. 2002;96:49–52. 271–277. 163. Chu PG, Ishizawa S, Wu E, Weiss LM. Hepatocyte antigen as a marker 134. Ma CK, Zarbo RJ, Frierson HF Jr, Lee MW. Comparative immunohisto- of hepatocellular carcinoma: an immunohistochemical comparison to carci-

Arch Pathol Lab Med—Vol 132, March 2008 Undifferentiated Tumor, Immunohistochemistry—Bahrami et al 345 noembryonic antigen, CD10, and alpha-fetoprotein. Am J Surg Pathol. 2002;26: 190. Sasaki E, Tsunoda N, Hatanaka Y, Mori N, Iwata H, Yatabe Y. Breast- 978–988. specific expression of MGB1/mammaglobin: an examination of 480 tumors from 164. Fan Z, van de RM, Montgomery K, Rouse RV. Hep par 1 antibody stain various organs and clinicopathological analysis of MGB1-positive breast cancers. for the differential diagnosis of hepatocellular carcinoma: 676 tumors tested using Mod Pathol. 2007;20:208–214. tissue microarrays and conventional tissue sections. Mod Pathol. 2003;16:137– 191. McGregor DK, Khurana KK, Cao C, et al. Diagnosing primary and met- 144. astatic renal cell carcinoma: the use of the monoclonal antibody ‘renal cell car- 165. Lugli A, Tornillo L, Mirlacher M, Bundi M, Sauter G, Terracciano LM. cinoma marker.’ Am J Surg Pathol. 2001;25:1485–1492. Hepatocyte paraffin 1 expression in human normal and neoplastic tissues: tissue 192. Avery AK, Beckstead J, Renshaw AA, Corless CL. Use of antibodies to microarray analysis on 3,940 tissue samples. Am J Clin Pathol. 2004;122:721– RCC and CD10 in the differential diagnosis of renal neoplasms. Am J Surg Pathol. 727. 2000;24:203–210. 166. Borscheri N, Roessner A, Rocken C. Canalicular immunostaining of ne- 193. Genega EM, Hutchinson B, Reuter VE, Gaudin PB. Immunophenotype of prilysin (CD10) as a diagnostic marker for hepatocellular carcinomas. Am J Surg high-grade prostatic adenocarcinoma and urothelial carcinoma. Mod Pathol. Pathol. 2001;251297–1303. 2000;13:1186–1191. 167. Barbareschi M, Murer B, Colby TV, et al. CDX-2 homeobox gene ex- 194. Levesque M, Hu H, D’Costa M, Diamandis EP. Prostate-specific antigen pression is a reliable marker of colorectal adenocarcinoma metastases to the expression by various tumors. J Clin Lab Anal. 1995;9:123–128. lungs. Am J Surg Pathol. 2003;27:141–149. 195. Bostwick DG. Prostate-specific antigen: current role in diagnostic pa- 168. Werling RW, Yaziji H, Bacchi CE, Gown AM. CDX2, a highly sensitive thology of prostate cancer. Am J Clin Pathol. 1994;102:S31–S37. and specific marker of adenocarcinomas of intestinal origin: an immunohisto- 196. Alanen KA, Kuopio T, Collan YU, Kronqvist P, Juntti L, Nevalainen TJ. chemical survey of 476 primary and metastatic carcinomas. Am J Surg Pathol. Immunohistochemical labelling for prostate-specific antigen in breast carcino- 2003;27:303–310. mas. Breast Cancer Res Treat. 1999;56:169–176. 169. Kaimaktchiev V, Terracciano L, Tornillo L, et al. The homeobox intestinal 197. Dennis JL, Hvidsten TR, Wit EC, et al. Markers of adenocarcinoma char- differentiation factor CDX2 is selectively expressed in gastrointestinal adenocar- acteristic of the site of origin: development of a diagnostic algorithm. Clin Cancer cinomas. Mod Pathol. 2004;17:1392–1399. Res. 2005;11:3766–3772. 170. Hinoi T, Tani M, Lucas PC, et al. Loss of CDX2 expression and micro- 198. Varma M, Berney DM, Jasani B, Rhodes A. Technical variations in pros- satellite instability are prominent features of large cell minimally differentiated tatic immunohistochemistry: need for standardisation and stringent quality assur- carcinomas of the colon. Am J Pathol. 2001;159:2239–2248. ance in PSA and PSAP immunostaining. J Clin Pathol. 2004;57:687–690. 171. Suh N, Yang XJ, Tretiakova MS, Humphrey PA, Wang HL. Value of CDX2, 199. Steffens J, Friedmann W, Lobeck H. Immunohistochemical diagnosis of villin, and alpha-methylacyl coenzyme A racemase immunostains in the distinc- the metastasizing prostatic carcinoma. Eur Urol. 1985;11:91–94. tion between primary adenocarcinoma of the bladder and secondary colorectal 200. Goldstein NS. Immunophenotypic characterization of 225 prostate ade- adenocarcinoma. Mod Pathol. 2005;18:1217–1222. nocarcinomas with intermediate or high Gleason scores. Am J Clin Pathol. 2002; 172. Wang HL, Lu DW, Yerian LM, et al. Immunohistochemical distinction 117:471–477. between primary adenocarcinoma of the bladder and secondary colorectal ad- 201. Keillor JS, Aterman K. The response of poorly differentiated prostatic tu- enocarcinoma. Am J Surg Pathol. 2001;25:1380–1387. mors to staining for prostate specific antigen and prostatic acid phosphatase: a 173. Bacchi CE, Gown AM. Distribution and pattern of expression of villin, a comparative study. J Urol. 1987;137:894–896. gastrointestinal-associated cytoskeletal protein, in human carcinomas: a study 202. Abrahamsson PA, Lilja H. Three predominant prostatic proteins. Andro- employing paraffin-embedded tissue. Lab Invest. 1991;64:418–424. logia. 1990;22(suppl 1):122–131. 174. Tan J, Sidhu G, Greco MA, Ballard H, Wieczorek R. Villin, cytokeratin 7, 203. Stein BS, Vangore S, Petersen RO. Immunoperoxidase localization of and cytokeratin 20 expression in pulmonary adenocarcinoma with ultrastructural prostatic antigens: comparison of primary and metastatic sites. Urology. 1984;24: evidence of microvilli with rootlets. Hum Pathol. 1998;29:390–396. 146–152. 175. Tamboli P, Mohsin SK, Hailemariam S, Amin MB. Colonic adenocarci- 204. Jiang Z, Wu CL, Woda BA, et al. Alpha-methylacyl-CoA racemase: a mul- noma metastatic to the urinary tract versus primary tumors of the urinary tract ti-institutional study of a new prostate cancer marker. Histopathology. 2004;45: with glandular differentiation: a report of 7 cases and investigation using a limited 218–225. immunohistochemical panel. Arch Pathol Lab Med. 2002;126:1057–1063. 205. Jiang Z, Fanger GR, Woda BA, et al. Expression of alpha-methylacyl-CoA 176. Carcangiu ML, Chambers JT. Sex steroid receptors in gynecologic neo- racemase (P504s) in various malignant neoplasms and normal tissues: a study of plasms. Pathol Annu. 1992;27(pt 2):121–151. 761 cases. Hum Pathol. 2003;34:792–796. 177. Cagle PT, Mody DR, Schwartz MR. Estrogen and progesterone receptors 206. Ordonez NG. Value of thrombomodulin immunostaining in the diagnosis in bronchogenic carcinoma. Cancer Res. 1990;50:6632–6635. of transitional cell carcinoma: a comparative study with carcinoembryonic anti- 178. Ollayos CW, Riordan GP, Rushin JM. Estrogen receptor detection in par- gen. Histopathology. 1997;31:517–524. affin sections of adenocarcinoma of the colon, pancreas, and lung. Arch Pathol 207. Ordonez NG. Thrombomodulin expression in transitional cell carcinoma. Lab Med. 1994;118:630–632. Am J Clin Pathol. 1998;110:385–390. 179. Vargas SO, Leslie KO, Vacek PM, Socinski MA, Weaver DL. Estrogen- 208. Comin CE, Novelli L, Boddi V, Paglierani M, Dini S. Calretinin, throm- receptor-related protein p29 in primary nonsmall cell lung carcinoma: pathologic bomodulin, CEA, and CD15: a useful combination of immunohistochemical and prognostic correlations. Cancer. 1998;82:1495–1500. markers for differentiating pleural epithelial mesothelioma from peripheral pul- 180. Diaz NM, Mazoujian G, Wick MR. Estrogen-receptor protein in thyroid monary adenocarcinoma. Hum Pathol. 2001;32:529–536. neoplasms: an immunohistochemical analysis of papillary carcinoma, follicular 209. Kaufmann O, Volmerig J, Dietel M. Uroplakin III is a highly specific and carcinoma, and follicular adenoma. Arch Pathol Lab Med. 1991;115:1203–1207. moderately sensitive immunohistochemical marker for primary and metastatic 181. Wu CW, Chi CW, Chang TJ, Lui WY, P’eng FK. Sex hormone receptors urothelial carcinomas. Am J Clin Pathol. 2000;113:683–687. in gastric cancer. Cancer. 1990;65:1396–1400. 210. Kunju LP, Mehra R, Snyder M, Shah RB. Prostate-specific antigen, high- 182. Mazoujian G, Pinkus GS, Davis S, Haagensen DE Jr. Immunohistochem- molecular-weight cytokeratin (clone 34␤E12), and/or p63: an optimal immuno- istry of a gross cystic disease fluid protein (GCDFP-15) of the breast: a marker of histochemical panel to distinguish poorly differentiated prostate adenocarcinoma apocrine epithelium and breast carcinomas with apocrine features. Am J Pathol. from urothelial carcinoma. Am J Clin Pathol. 2006;125:675–681. 1983;110:105–112. 211. Mhawech P, Uchida T, Pelte MF. Immunohistochemical profile of high- 183. Viacava P, Naccarato AG, Bevilacqua G. Spectrum of GCDFP-15 ex- grade urothelial bladder carcinoma and prostate adenocarcinoma. Hum Pathol. pression in human fetal and adult normal tissues. Virchows Arch. 1998;432:255– 2002;33:1136–1140. 260. 212. Stewart CJ, Jeffers MD, Kennedy A. Diagnostic value of inhibin immu- 184. Kaufmann O, Deidesheimer T, Muehlenberg M, Deicke P, Dietel M. Im- noreactivity in ovarian gonadal stromal tumours and their histological mimics. munohistochemical differentiation of metastatic breast carcinomas from meta- Histopathology. 1997;31:67–74. static adenocarcinomas of other common primary sites. Histopathology. 1996; 213. McCluggage WG, Shanks JH, Whiteside C, Maxwell P, Banerjee SS, Big- 29:233–240. gart JD. Immunohistochemical study of testicular sex cord-stromal tumors, in- 185. Wick MR, Lillemoe TJ, Copland GT, Swanson PE, Manivel JC, Kiang DT. cluding staining with anti-inhibin antibody. Am J Surg Pathol. 1998;22:615–619. Gross cystic disease fluid protein-15 as a marker for breast cancer: immunohis- 214. Iczkowski KA, Bostwick DG, Roche PC, Cheville JC. Inhibin A is a sen- tochemical analysis of 690 human neoplasms and comparison with alpha-lact- sitive and specific marker for testicular sex cord-stromal tumors. Mod Pathol. albumin. Hum Pathol. 1989;20:281–287. 1998;11:774–779. 186. Raab SS, Berg LC, Swanson PE, Wick MR. Adenocarcinoma in the lung 215. Movahedi-Lankarani S, Kurman RJ. Calretinin, a more sensitive but less in patients with breast cancer: a prospective analysis of the discriminatory value specific marker than alpha-inhibin for ovarian sex cord-stromal neoplasms: an of immunohistology. Am J Clin Pathol. 1993;100:27–35. immunohistochemical study of 215 cases. Am J Surg Pathol. 2002;26:1477–1483. 187. Tian W, Osawa M, Horiuchi H, Tomita Y. Expression of the prolactin- 216. Zhang PJ, Genega EM, Tomaszewski JE, Pasha TL, Livolsi VA. The role of inducible protein (PIP/GCDFP15) gene in benign epithelium and adenocarcinoma calretinin, inhibin, melan-A, BCL-2, and C-kit in differentiating adrenal cortical of the prostate. Cancer Sci. 2004;95:491–495. and medullary tumors: an immunohistochemical study. Mod Pathol. 2003;16: 188. Bhargava R, Beriwal S, Dabbs DJ. Mammaglobin vs GCDFP-15: an im- 591–597. munohistologic validation survey for sensitivity and specificity. Am J Clin Pathol. 217. Pelkey TJ, Frierson HF Jr, Mills SE, Stoler MH. The alpha subunit of inhibin 2007;127:1–11. in adrenal cortical neoplasia. Mod Pathol. 1998;11:516–524. 189. Han JH, Kang Y, Shin HC, et al. Mammaglobin expression in lymph nodes 218. Renshaw AA, Granter SR. A comparison of A103 and inhibin reactivity is an important marker of metastatic breast carcinoma. Arch Pathol Lab Med. in adrenal cortical tumors: distinction from hepatocellular carcinoma and renal 2003;127:1330–1334. tumors. Mod Pathol. 1998;11:1160–1164.

346 Arch Pathol Lab Med—Vol 132, March 2008 Undifferentiated Tumor, Immunohistochemistry—Bahrami et al 219. Busam KJ, Jungbluth AA. Melan-A, a new melanocytic differentiation clinicopathologic and immunohistochemical study with special emphasis on its marker. Adv Anat Pathol. 1999;6:12–18. distinction from ordinary leiomyosarcoma and malignant fibrous histiocytoma. 220. Ghorab Z, Jorda M, Ganjei P, Nadji M. Melan A (A103) is expressed in Am J Surg Pathol. 2001;25:1030–1038. adrenocortical neoplasms but not in renal cell and hepatocellular carcinomas. 248. Ceballos KM, Nielsen GP, Selig MK, O’Connell JX. Is anti-h-caldesmon Appl Immunohistochem Mol Morphol. 2003;11:330–333. useful for distinguishing smooth muscle and myofibroblastic tumors? An immu- 221. Jungbluth AA, Busam KJ, Gerald WL, et al. A103: an anti-melan-a mono- nohistochemical study. Am J Clin Pathol. 2000;114:746–753. clonal antibody for the detection of malignant melanoma in paraffin-embedded 249. Watanabe K, Kusakabe T, Hoshi N, Saito A, Suzuki T. h-Caldesmon in tissues. Am J Surg Pathol. 1998;22:595–602. leiomyosarcoma and tumors with smooth muscle cell-like differentiation: its spe- 222. Doglioni C, Tos AP, Laurino L, et al. Calretinin: a novel immunocyto- cific expression in the smooth muscle cell tumor. Hum Pathol. 1999;30:392–396. chemical marker for mesothelioma. Am J Surg Pathol. 1996;20:1037–1046. 250. Gaffney EF, Dervan PA, Fletcher CD. Pleomorphic rhabdomyosarcoma in 223. Chhieng DC, Yee H, Schaefer D, et al. Calretinin staining pattern aids in adulthood: analysis of 11 cases with definition of diagnostic criteria. Am J Surg the differentiation of mesothelioma from adenocarcinoma in serous effusions. Pathol. 1993;17:601–609. Cancer. 2000;90:194–200. 251. Furlong MA, Mentzel T, Fanburg-Smith JC. Pleomorphic rhabdomyosar- 224. McCluggage WG, Maxwell P. Immunohistochemical staining for calretin- coma in adults: a clinicopathologic study of 38 cases with emphasis on morpho- in is useful in the diagnosis of ovarian sex cord-stromal tumours. Histopathology. logic variants and recent skeletal muscle-specific markers. Mod Pathol. 2001;14: 2001;38:403–408. 595–603. 225. Deavers MT, Malpica A, Liu J, Broaddus R, Silva EG. Ovarian sex cord- 252. Dei Tos AP, Wadden C, Fletcher CDM. S-100 protein staining in liposar- stromal tumors: an immunohistochemical study including a comparison of cal- coma: its diagnostic utility in the high-grade myxoid (round cell) variant. Appl retinin and inhibin. Mod Pathol. 2003;16:584–590. Immunohistochem. 1996;4:95–101. 226. Ordonez NG. Application of mesothelin immunostaining in tumor diag- 253. Gebhard S, Coindre JM, Michels JJ, et al. Pleomorphic liposarcoma: clin- nosis. Am J Surg Pathol. 2003;27:1418–1428. icopathologic, immunohistochemical, and follow-up analysis of 63 cases: a study 227. Foster MR, Johnson JE, Olson SJ, Allred DC. Immunohistochemical anal- from the French Federation of Cancer Centers Sarcoma Group. Am J Surg Pathol. ysis of nuclear versus cytoplasmic staining of WT1 in malignant mesotheliomas 2002;26:601–616. and primary pulmonary adenocarcinomas. Arch Pathol Lab Med. 2001;125: 254. Hornick JL, Bosenberg MW, Mentzel T, McMenamin ME, Oliveira AM, 1316–1320. Fletcher CD. Pleomorphic liposarcoma: clinicopathologic analysis of 57 cases. 228. O’Neill CJ, Deavers MT, Malpica A, Foster H, McCluggage WG. An im- Am J Surg Pathol. 2004;28:1257–1267. munohistochemical comparison between low-grade and high-grade ovarian se- 255. Miettinen M, Enzinger FM. Epithelioid variant of pleomorphic liposar- rous carcinomas: significantly higher expression of p53, MIB1, BCL2, HER-2/neu, coma: a study of 12 cases of a distinctive variant of high-grade liposarcoma. Mod and C-KIT in high-grade neoplasms. Am J Surg Pathol. 2005;29:1034–1041. Pathol. 1999;12:722–728. 229. Barnetson RJ, Burnett RA, Downie I, Harper CM, Roberts F. Immunohis- 256. Hasegawa T, Seki K, Hasegawa F, et al. Dedifferentiated liposarcoma of tochemical analysis of peritoneal mesothelioma and primary and secondary se- retroperitoneum and mesentery: varied growth patterns and histological grades— rous carcinoma of the peritoneum: antibodies to estrogen and progesterone re- a clinicopathologic study of 32 cases. Hum Pathol. 2000;31:717–727. ceptors are useful. Am J Clin Pathol. 2006;125:67–76. 257. Joyner CJ, Triffitt J, Puddle B, Athanasou NA. Development of a mono- 230. Chu AY, Litzky LA, Pasha TL, Acs G, Zhang PJ. Utility of D2-40, a novel clonal antibody to the aP2 protein to identify adipocyte precursors in tumours of mesothelial marker, in the diagnosis of malignant mesothelioma. Mod Pathol. adipose differentiation. Pathol Res Pract. 1999;195:461–466. 2005;18:105–110. 258. Binh MB, Sastre-Garau X, Guillou L, et al. MDM2 and CDK4 immuno- 231. Comin CE, Dini S, Novelli L, Santi R, Asirelli G, Messerini L. h-Caldes- stainings are useful adjuncts in diagnosing well-differentiated and dedifferentiated mon, a useful positive marker in the diagnosis of pleural malignant mesothelioma, liposarcoma subtypes: a comparative analysis of 559 soft tissue neoplasms with epithelioid type. Am J Surg Pathol. 2006;30:463–469. genetic data. Am J Surg Pathol. 2005;29:1340–1347. 232. Hishima T, Fukayama M, Fujisawa M, et al. CD5 expression in thymic 259. Meis JM, Enzinger FM, Martz KL, Neal JA. Malignant peripheral nerve carcinoma. Am J Pathol. 1994;145:268–275. sheath tumors (malignant schwannomas) in children. Am J Surg Pathol. 1992;16: 233. Berezowski K, Grimes MM, Gal A, Kornstein MJ. CD5 immunoreactivity 694–707. of epithelial cells in thymic carcinoma and CASTLE using paraffin-embedded 260. Hoang MP, Sinkre P, Albores-Saavedra J. Expression of protein gene prod- tissue. Am J Clin Pathol. 1996;106:483–486. uct 9.5 in epithelioid and conventional malignant peripheral nerve sheath tumors. 234. Tateyama H, Eimoto T, Tada T, Hattori H, Murase T, Takino H. Immuno- Arch Pathol Lab Med. 2001;125:1321–1325. reactivity of a new CD5 antibody with normal epithelium and malignant tumors 261. Smith TA, Machen SK, Fisher C, Goldblum JR. Usefulness of cytokeratin including thymic carcinoma. Am J Clin Pathol. 1999;111:235–240. subsets for distinguishing monophasic synovial sarcoma from malignant periph- 235. Kornstein MJ, Rosai J. CD5 labeling of thymic carcinomas and other non- eral nerve sheath tumor. Am J Clin Pathol. 1999;112:641–648. lymphoid neoplasms. Am J Clin Pathol. 1998;109:722–726. 262. Zhou H, Coffin CM, Perkins SL, Tripp SR, Liew M, Viskochil DH. Malig- 236. Dorfman DM, Shahsafaei A, Chan JK. Thymic carcinomas, but not thy- nant peripheral nerve sheath tumor: a comparison of grade, immunophenotype, momas and carcinomas of other sites, show CD5 immunoreactivity. Am J Surg and cell cycle/growth activation marker expression in sporadic and neurofibro- Pathol. 1997;21:936–940. matosis 1-related lesions. Am J Surg Pathol. 2003;27:1337–1345. 237. Bailey D, Marks A, Stratis M, Baumal R. Immunohistochemical staining 263. Litzky LA, Brooks JJ. Cytokeratin immunoreactivity in malignant fibrous of germ cell tumors and intratubular malignant germ cells of the testis using histiocytoma and spindle cell tumors: comparison between frozen and paraffin- antibody to placental alkaline phosphatase and a monoclonal anti-seminoma an- embedded tissues. Mod Pathol. 1992;5:30–34. tibody. Mod Pathol. 1991;4:167–171. 264. Rosenberg AE, O’Connell JX, Dickersin GR, Bhan AK. Expression of ep- 238. Burke AP, Mostofi FK. Placental alkaline phosphatase immunohistochem- ithelial markers in malignant fibrous histiocytoma of the musculoskeletal system: istry of intratubular malignant germ cells and associated testicular germ cell tu- an immunohistochemical and electron microscopic study. Hum Pathol. 1993;24: mors. Hum Pathol. 1988;19:663–670. 284–293. 239. Goldsmith JD, Pawel B, Goldblum JR, et al. Detection and diagnostic 265. Wick MR, Swanson PE. Carcinosarcomas: current perspectives and an utilization of placental alkaline phosphatase in muscular tissue and tumors with historical review of nosological concepts. Semin Diagn Pathol. 1993;10:118–127. myogenic differentiation. Am J Surg Pathol. 2002;26:1627–1633. 266. Nappi O, Glasner SD, Swanson PE, Wick MR. Biphasic and monophasic 240. Looijenga LH, Stoop H, de Leeuw HP, et al. POU5F1 (OCT3/4) identifies sarcomatoid carcinomas of the lung: a reappraisal of ‘carcinosarcomas’ and ‘spin- cells with pluripotent potential in human germ cell tumors. Cancer Res. 2003; dle-cell carcinomas.’ Am J Clin Pathol. 1994;102:331–340. 63:2244–2250. 267. Carter MR, Hornick JL, Lester S, Fletcher CD. Spindle cell (sarcomatoid) 241. Jones TD, Ulbright TM, Eble JN, Baldridge LA, Cheng L. OCT4 staining carcinoma of the breast: a clinicopathologic and immunohistochemical analysis in testicular tumors: a sensitive and specific marker for seminoma and embryonal of 29 cases. Am J Surg Pathol. 2006;30:300–309. carcinoma. Am J Surg Pathol. 2004;28:935–940. 268. Ikegami H, Iwasaki H, Ohjimi Y, Takeuchi T, Ariyoshi A, Kikuchi M. Sar- 242. Cheng L. Establishing a germ cell origin for metastatic tumors using OCT4 comatoid carcinoma of the urinary bladder: a clinicopathologic and immunohis- immunohistochemistry. Cancer. 2004;101:2006–2010. tochemical analysis of 14 patients. Hum Pathol. 2000;31:332–340. 243. Fletcher CD. Pleomorphic malignant fibrous histiocytoma: fact or fiction? 269. Lewis JE, Olsen KD, Sebo TJ. Spindle cell carcinoma of the larynx: review A critical reappraisal based on 159 tumors diagnosed as pleomorphic sarcoma. of 26 cases including DNA content and immunohistochemistry. Hum Pathol. Am J Surg Pathol. 1992;16:213–228. 1997;28:664–673. 244. Coindre JM, Mariani O, Chibon F, et al. Most malignant fibrous histio- 270. Thompson LD, Wieneke JA, Miettinen M, Heffner DK. Spindle cell (sar- cytomas developed in the retroperitoneum are dedifferentiated liposarcomas: a comatoid) carcinomas of the larynx: a clinicopathologic study of 187 cases. Am review of 25 cases initially diagnosed as malignant fibrous histiocytoma. Mod J Surg Pathol. 2002;26:153–170. Pathol. 2003;16:256–262. 271. Lewis JS, Ritter JH, El Mofty S. Alternative epithelial markers in sarco- 245. Fletcher CD, Gustafson P, Rydholm A, Willen H, Akerman M. Clinico- matoid carcinomas of the head and neck, lung, and bladder-p63, MOC-31, and pathologic re-evaluation of 100 malignant fibrous histiocytomas: prognostic rel- TTF-1. Mod Pathol. 2005;18:1471–1481. evance of subclassification. J Clin Oncol. 2001;19:3045–3050. 272. Suster S, Moran CA. Applications and limitations of immunohistochem- 246. Mertens F, Fletcher CD, Dal Cin P, et al. Cytogenetic analysis of 46 pleo- istry in the diagnosis of malignant mesothelioma. Adv Anat Pathol. 2006;13:316– morphic soft tissue sarcomas and correlation with morphologic and clinical fea- 329. tures: a report of the CHAMP Study Group. Chromosomes and MorPhology. 273. Kung IT, Thallas V, Spencer EJ, Wilson SM. Expression of muscle actins Genes Chromosomes Cancer. 1998;22:16–25. in diffuse mesotheliomas. Hum Pathol. 1995;26:565–570. 247. Oda Y, Miyajima K, Kawaguchi K, et al. Pleomorphic leiomyosarcoma: 274. Lucas DR, Pass HI, Madan SK, et al. Sarcomatoid mesothelioma and its

Arch Pathol Lab Med—Vol 132, March 2008 Undifferentiated Tumor, Immunohistochemistry—Bahrami et al 347 histological mimics: a comparative immunohistochemical study. Histopathology. mosome translocation, encodes a constitutive transcriptional activator. Onco- 2003;42:270–279. gene. 1996;12:159–167. 275. Hinterberger M, Reineke T, Storz M, Weder W, Vogt P, Holger M. D2-40 288. Laskin WB, Weiss SW, Bratthauer GL. Epithelioid variant of malignant and calretinin-a tissue microarray analysis of 341 malignant mesotheliomas with peripheral nerve sheath tumor (malignant epithelioid schwannoma). Am J Surg emphasis on sarcomatoid differentiation. Mod Pathol. 2007;20:248–255. Pathol. 1991;15:1136–1145. 276. Skelton HG, Smith KJ, Laskin WB, et al. Desmoplastic malignant mela- 289. Bonetti F, Pea M, Martignoni G, et al. Clear cell (‘‘sugar’’) tumor of the lung is a lesion strictly related to angiomyolipoma—the concept of a family of noma. J Am Acad Dermatol. 1995;32:717–725. lesions characterized by the presence of the perivascular epithelioid cells (PEC). 277. Miettinen M, Fanburg-Smith JC, Virolainen M, Shmookler BM, Fetsch JF. Pathology. 1994;26:230–236. Epithelioid sarcoma: an immunohistochemical analysis of 112 classical and var- 290. Folpe AL, Mentzel T, Lehr HA, Fisher C, Balzer BL, Weiss SW. Perivascular iant cases and a discussion of the differential diagnosis. Hum Pathol. 1999;30: epithelioid cell neoplasms of soft tissue and gynecologic origin: a clinicopatho- 934–942. logic study of 26 cases and review of the literature. Am J Surg Pathol. 2005;29: 278. Manivel JC, Wick MR, Dehner LP, Sibley RK. Epithelioid sarcoma: an 1558–1575. immunohistochemical study. Am J Clin Pathol. 1987;87:319–326. 291. Zamboni G, Pea M, Martignoni G, et al. Clear cell ‘‘sugar’’ tumor of the 279. Hasegawa T, Matsuno Y, Shimoda T, Umeda T, Yokoyama R, Hirohashi S. pancreas: a novel member of the family of lesions characterized by the presence Proximal-type epithelioid sarcoma: a clinicopathologic study of 20 cases. Mod of perivascular epithelioid cells. Am J Surg Pathol. 1996;20:722–730. Pathol. 2001;14:655–663. 292. Martignoni G, Pea M, Reghellin D, Zamboni G, Bonetti F. Perivascular 280. Arber DA, Kandalaft PL, Mehta P, Battifora H. Vimentin-negative epithe- epithelioid cell tumor (PEComa) in the genitourinary tract. Adv Anat Pathol. 2007; lioid sarcoma: the value of an immunohistochemical panel that includes CD34. 14:36–41. Am J Surg Pathol. 1993;17:302–307. 293. Costa MJ, DeRose PB, Roth LM, Brescia RJ, Zaloudek CJ, Cohen C. Im- munohistochemical phenotype of ovarian granulosa cell tumors: absence of ep- 281. Fletcher CD, Beham A, Bekir S, Clarke AM, Marley NJ. Epithelioid an- ithelial membrane antigen has diagnostic value. Hum Pathol. 1994;25:60–66. giosarcoma of deep soft tissue: a distinctive tumor readily mistaken for an epi- 294. Miettinen M, Sobin LH, Lasota J. Gastrointestinal stromal tumors of the thelial neoplasm. Am J Surg Pathol. 1991;15:915–924. stomach: a clinicopathologic, immunohistochemical, and molecular genetic 282. Deshpande V, Rosenberg AE, O’Connell JX, Nielsen GP. Epithelioid an- study of 1765 cases with long-term follow-up. Am J Surg Pathol. 2005;29:52–68. giosarcoma of the bone: a series of 10 cases. Am J Surg Pathol. 2003;27:709– 295. Bacchi CE, Bonetti F, Pea M, Martignoni G, Gown AM. HMB-45: a re- 716. view. Appl Immunohistochem. 1996;4:73–85. 283. Zhang PJ, Livolsi VA, Brooks JJ. Malignant epithelioid vascular tumors of 296. Ben Izhak O, Stark P, Levy R, Bergman R, Lichtig C. Epithelial markers the pleura: report of a series and literature review. Hum Pathol. 2000;31:29–34. in malignant melanoma: a study of primary lesions and their metastases. Am J 284. Meis-Kindblom JM, Kindblom LG. Angiosarcoma of soft tissue: a study Dermatopathol. 1994;16:241–246. of 80 cases. Am J Surg Pathol. 1998;22:683–697. 297. Jaffe ES, Harris NL, Stein H, Vardiman JW. Pathology and Genetics of 285. Granter SR, Weilbaecher KN, Quigley C, Fletcher CD, Fisher DE. Clear Tumours of Haematopoietic and Lymphoid Tissues. Lyon, France: IARC Press; cell sarcoma shows immunoreactivity for microphthalmia transcription factor: fur- 2001:230–235. World Health Organization Classification of Tumours. 298. Chott A, Kaserer K, Augustin I, et al. Ki-1-positive large cell lymphoma: ther evidence for melanocytic differentiation. Mod Pathol. 2001;14:6–9. a clinicopathologic study of 41 cases. Am J Surg Pathol. 1990;14:439–448. 286. Swanson PE, Wick MR. Clear cell sarcoma: an immunohistochemical 299. Agnarsson BA, Kadin ME. Ki-1 positive large cell lymphoma: a morpho- analysis of six cases and comparison with other epithelioid neoplasms of soft logic and immunologic study of 19 cases. Am J Surg Pathol. 1988;12:264–274. tissue. Arch Pathol Lab Med. 1989;113:55–60. 300. O’Connell FP, Pinkus JL, Pinkus GS. CD138 (syndecan-1), a plasma cell 287. Fujimura Y, Ohno T, Siddique H, Lee L, Rao VN, Reddy ES. The EWS- marker immunohistochemical profile in hematopoietic and nonhematopoietic ATF-1 gene involved in malignant melanoma of soft parts with t(12;22) chro- neoplasms. Am J Clin Pathol. 2004;121:254–263.

348 Arch Pathol Lab Med—Vol 132, March 2008 Undifferentiated Tumor, Immunohistochemistry—Bahrami et al