CD34 expression by gastrointestinal tract stromal tumors

CD34 Expression by Gastrointestinal Tract Stromal Tumors MATTHIJS MICHAEL

RIJN, MD, PHD, R. HENDRICKSON, MD, AND ROBERT V. ROUSE, MD

VAN DE

Gastrointestinalstromal tumors (GISTS) are neoplasms a&ii in the wall of the gastroiutestinaltract that frequently show evidence of smooth muscle differentiation,eitherby their appearancealone or by immu.uohistology. A significantnumber of these neophuansfail to react with any markersof muscle differentiation,however.A subset of these neoplasms have epithelioidfeatures, and the presence of these features can give rise to confusion with other neoplauns, such as carcinomas and melanomas. Here we show that the CD34 monoclonal antibodyMy10 reacts with 19 of 23 (83%) of these lesions, including both thosewithand withoutepithelioidfeatures.Fiveof 10 epithelioid and one of 13 spindled neoplastuslacked detectable muscle-specific actin (MSA), smooth muscle actin @MA), and desmin, all six were CD34 reactive.Immunoblottingexperimentsshow that the antigenon these stromal neoplasms has a molecular weight identical to that found on hematopoietic cells. The frequency and intensityof the reactivityof GISTS with anti-CD34 antibodies are distinctlyhigher than those reported for smooth muscle neoplastusof soft tissue and myometrium. This reactivitycan be a useful adjunct in the diagnosis of difficult cases, especiallyin those exhibitingepithelioid morphology. HUM PATHOL 25766-771. Copyright0 1994 by W.B. Saunders Company

Gastrointestinal stromal tumor (GIST) is the curneo lasms rent designation for a set of mesenchymal tract.‘. BMany that arise in the wall of the gastrointestinal of these are composed of spindled cells with moderate to abundant amounts of cytoplasm and elongated bluntended nuclei, features characteristic of smooth muscle tumors.3.4 Similarly, a subset of these tumors exhibits immunologic evidence of muscle-type actins or desmin.&’ A subset of GISTS does not show clear evidence of smooth muscle differentiation; some tumors lack muscle type actins and desmin,+’ and some display epithelioid architectural and cytological features that are not reminiscent of normal smooth muscle.2.4.5 To reflect this lack of uniform evidence for muscle differentiation, the members of this group of tumors are described as GISTS rather than leiomyomas or leiomyosarcomas. The antigen designated as CD34 is a 1 IO-kd cell surface glycoprotein. It is defined by a group of monoclonal antibodies and has been well characterized on hematopoietic cells and endothelial cells.“” In addition, CD34 reactivity has been identified in a small sub Stanford UniversitySchool of Medicine, Stanford, CA: and the Veterans Affairs Medical Center, Palo Alto, CA. Accepted for publicationJanuary 25, 1994. Supported in part by the Department of Veterans Affairs. From

the Department

of Pathology,

k$ words: gastrointestinal stromal tumors, immunohistochemistry, CD34, actin, leiomyoblastoma, epithelioid smooth muscle tumor. Address correspondence to Matthijs van de Kjn, MD, PhD, De-

partment of Pathology, Room L203, Stanford University Medical School, Stanford, CA 94305. Copyright 0 1994 by W.B.

Saunders Company 00468177/94/2508-0007$5.00/0

766

set of smooth muscle tumors,‘2.‘” including those with epithelioid features’2.‘4 arising in soft tissue or of unstated site. Here we show that a significant subset of GISTS react with anti-CD34 antibodies in immunoperoxidase studies, including all cases that fail to stain for the muscle markers muscle-specific actin (MSA) , smooth muscle actin (SMA) , and desmin. Because CD34 is only rarely expressed on epithelial tumors,‘2.‘“,‘5 this finding can be useful in the differential diagnosis of GIST, especially for the epithelioid variant. Immunoblotting studies show that the antigen on GIST that is recognized by anti-CD34 antibodies has an identical molecular weight to that found on hematopoietic and endothelial cells.

MATERIALS

AND METHODS

lmmunophenotypic

Studies

Twenty-three cases of GIST were selected from the files of the Department of Surgical Pathology at Stanford University Medical Center, Stanford, CA; several were consultation cases on which blocks or unstained slides were obtained. A panel of antibodies was used to stain formalin-fixed, paraffinembedded sections: antidesmin (clone D33, Dako, Carpentiera, CA), anti-MSA (HHF35, Dako), anti&MA (clone lA4, Sigma, St Louis, MO), and anti-CD34 (anti-HPCA-1, clone MylO, Becton Dickinson, Mountain View, CA). Immunohistochemical studies were performed on deparaffinized sections using an indirect biotin-avidin method on a Ventana 320 automated immunohistochemistry system (Ventana Medical Systems, Inc., Tucson, AZ). After inhibition of endogenous peroxidase the automated method uses primary antibody incubation at a temperature of 37°C for 30 minutes. The Ventana DAB (diaminobenzidine) Readout System (Ventana Medical Systems, Inc.) was used for detection of antibody reactivity. The sections were counterstained with hematoxylineosin, dehydrated, and then coverslipped. The intensity of the staining for the above antigens was recorded as +, + +, or +++ and was designated as focal (f) if less than 10% of the neoplastic cells stained. Care was taken to exclude vascular staining from the evaluation.

Immunoblotting Cell lysates were prepared from KGla cells and samples of a GIST and a CD34negative metastatic uterine leiomyosarcoma. KGla is the human myeloid leukemia cell line from which the CD34 molecule and gene were originally identified.‘” The lysates were analyzed on 7.5% acrylamide sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS PAGE) followed by Coomassie staining to determine the amount of protein present in the preparations. Equal amounts from the three samples were next analyzed on SDSPAGE and transferred to nitrocellulose by electrophoresis. Prestained markers (Sigma) were used to determine molecular weight.

CD34

IN GASTROINTESTINAL

TRACT

STROMAL

After overnight blocking of residual binding sites by incubation in n-is-buffered saline (TBS), 0.1% Triton X-100, and 5% milk powder, the membranes were incubated with primary antibodies against CD34 (anti-HPGA-1, clone MylO, and clone QBENDlO, Harlan, Indianapolis, IN) and anti-vimentin (DAKO). Subsequently the membranes were incubated with goat anti-mouse immunoglobulin (Ig) coupled to biotin (Sigma) and avidin coupled to alkaline phosphatase (Jackson Immunoresearch, West Grove, PA). Each incubation step was followed by washing in TBS. The membranes were developed with the nitro blue tetrazolium (NBT) staining method.”

lmmunophenotypic Studies Seventeen of the 23 cases of GIST reacted with one or more of the three muscle markers, desmin, MSA, and SMA. Desmin was positive in six cases, whereas MSA and SMA were detectable in 10 and 14 cases, respectively. When reactive they were generally strong. Five of the 10 epithelioid tumors lacked detectable muscle markers, whereas only one of 13 spindled neoplasms lacked all three. Vascular and enteric smooth muscle served as a positive internal control for the actin and desmin antibodies in all cases. CD34, using antibody MylO, was detected in 19 of 23 cases (83%). All six cases that lacked all three muscle markers reacted with CD34 (Fig 2). In most CD34reactive cases the majority of the cells were intensely stained and clearly included the predominant cell type in all of the positive tumors. Endothelial staining served as a positive control for the CD34 antibody in all cases. The immunohistological staining results were compared with the site of origin, clinical outcome, histological grade, and presence or absence of epithelioid fea-

Clinical Features The clinical and morphological features of the GIST are listed in Table 1. Ten cases arose in the stomach, four cases in the small intestine, six cases in the large intestine, and three cases in the anorectal area. The age of the patients ranged from 8 to 85 years. The size of the lesions ranged from 0.6 to 30 cm. The clinical features of the 23 cases of GIST did not differ significantly from those of the larger series reported in the literature.“.lg

Histology The cases of GIST were graded histologically as benign, of uncertain malignant potential, and malignant, according to the criteria proposed by Lewin et al.’ Many showed histological features of smooth muscle differ-

Case No.

*Se OJrs)

1E 44 2E 59 3E 62 4E 68 5E 62 6E 62 7E 49 SE 68 9 85 10 67 11 45 12 75 13 60 14 74 15E 59 16E 8 17 48 18 24 19 56 20 55 21 54 22 48 23 34 Total epithelioid Total spindled Total

Case Descriptions

Sex

Site

Size (cm)

status

F M M M M M M F F F F F M F ? M F F M M M F F

Stomach Stomach Stomach Stomach Stomach Stomach Stomach Stomach Stomach Stomach Sm int Sm int Sm int Sm int Lg int Lg int Lg int Lg int Lg int Lg int Anorectal Anorectal Anorectal

4.5 Unknown 16 7.5 19 13 14 25 2.5 4.8 7 5 10 0.6 >15 4 1 30 6 3 3 3.2 3

ANED ANED AWD ANED AWD ANED ANED NA NA ANED AWD NA LFU ANED AWD AWD ANED ANED DOD ANED NA NA ANED

(van de Rijn et al)

entiation with spindled cells with abundant eosinophilic cytoplasm growing in fascicles (Fig 1A). Ten cases had an epithelioid appearance of the cells with loss of spindle cell appearance and rounded appearance of the cells (Figs 1B and 1C). It was this epithelioid variant that was found to be the most difficult to distinguish from other lesions, such as carcinomas, melanomas, and anaplastic lymphomas, in the differential diagnosis.

RESULTS

TABLE 1.

TUMORS

and Immunologic Follow-Up (mos) 3 82 30 29 15 3 3 <2 <2 10 23 <2 2 11 1 12 14 204 6 <2 <2 11

Findings

Grade

CD34

MSA

UMP UMP Malig Malig Malig UMP UMP Malig Malig Malig Malig UMP UMP Benign Malig Malig UMP Malig Malig Malig Malig Malig Malig

++ +++

_ _ _ +++ +++ +++ _ ++ _ _ +++ +++ _ _ +++ +++ _

++f ++f +++ +++ +++ _ +++ +++ +++ +++ +++ _ +++ +++ ++ f _ _ +++ +++ s+/tlo 10/13 19/23

+++f _ ++ _ 3/10 7/13 lo/23

SM.4

Desmin

_ _ _

_ _ + _ _ +

+++f +++f +++ +++ ++f -

+++f _ _ -

+f +++ +++ +++ _ _

+++ _ _

+++ +++ +++ ++f _ _ ++ 4/10 10/13 14/23

+++ _ _ -

-

+++ 3/10 3/13 6/23

NOTE. Immunologic stains scored for intensity as + , + + , and + + +, and indicated as f for focal if < 10% of cells stained. Abbreviations: E, epithelioid; F, female; M, male; Sm int, small intestine: Lg int, large intestine; ANED, alive with no evidence of disease; AWD, alive with disease; LFU, lost to follow-up; DOD, dead of disease; NA, not assessable because follow-up was <2 months; UMP, uncertain malignant potential; Malig, malignant.

767

HUMAN

PATHOLOGY

Volume 25, No. 8 (August 1994)

FIGURE 1. Histological appearance of GIST. (A) Typical appearance of spindled GIST (case no. 20). (B) epithelioid GIST (case no. 151, and (C) epithelioid GIST (case no. 1).

768

CD34 IN GASTROINTESTINAL TRACT STROMAL TUMORS (van de Rijn et al)

FIQURE 2. Staining pattern in GIST with CD34 (case no. 15). Strong staining of GIST (lower part figure) with CD34 (B) in contrast to negative staining with MSA (A). Note reactivity of muscularis propria and smooth muscle in vessel walls with MSA.

tures. There was no significant association with site or histological grade and any of the markers. CD34 stained all categories equally. Only MSA and desmin reactivity correlated with clinical outcome; of cases with no evidence of disease, seven of 11 stained for MSA and five of 11 stained for desmin, whereas of those with recurrent disease, none of six reacted for either. Only MSA and SMA correlated with epithelioid morphology; of 10 epithelioid cases, three were MSA reactive and four were SMA reactive, whereas seven of 13 and 10 of 13 spindled cases were MSA and SMA positive.

lmmunoblotting Immunoblotting experiments showed that antiCD34 antibody My10 recognizes a protein with a molecular weight of 110 kd in GIST (Fig 3, lane a). This protein is identical in molecular weight to that found on KGla cells (Fig 3, lane b), the cell line from which the CD34 antigen was cloned.16 A control case of metastatic leiomyosarcoma that was negative for CD34 by immunohistology failed to show a band in this area (Fig 3, lane c). In all three lanes minor background bands with molecular weights of 135 and 60 kd were observed. Identical results were obtained with a second anti-CD34 antibody, QBENDlO (data not shown). The antibodies QBENDlO and MYIO are known to react with different carbohydrate residues on the CD34 molecule.” Parallel blots were developed with anti-vimentin antibody as a primary antibody to confirm that equal amounts of lysate from the GIST sample and the metastatic leiomyosarcoma sample were loaded (data not shown).

DISCUSSION Our immunohistological findings indicate that CD34 is an effective marker of GIST. In agreement with

769

previously reported studies4’ we found the conventional markers of muscle differentiation (MSA, SMA, and desmin) to react with only a subset of GISTS. We found reactivity for muscle-type actins to correlate strongly with morphology. Actin expression was present in 10 of 13 spindled tumors but in only four of 10 epithelioid tumors. Although some reports5>i found no difference in actin expression between different histological types of GIST, another study mentioned similar differential expssion of actin in spindled versus epithelioid GIST. In this respect it is important to note that CD34 stained epithelioid and spindled GISTS equally well. Its reactivity in 83% of all cases of GIST compared with MSA reactivity in 43% and SMA reactivity in 64% makes it a more sensitive marker for GIST in our series. Significantly, in the five epithelioid and one spindled cases that failed to express either muscle-type actins or desmin, CD34 was present. This common lack of actin and desmin expression and frequent CD34 expression in epithelioid GIST may have important ramifications in the application of immunologic stains to differential diagnosis. It is this sub set of GISTS that is more frequently confused with other neoplasms, most significantly carcinoma, melanoma, and large-cell lymphoma. An a&n-or desmin-positive neoplasm would be separable from all of these. However, as we demonstrate, these markers will not be helpful in a significant number of such cases. Keratin7~‘2~23 and S1005.6.25 also may be expressed by some GISTS, whereas some carcinomas, melanomas, and lymphomas fail to express the appropriate keratin,‘4 S100,25 leukocyte common antigen, or B- and T-lineage markers,‘6 complicating the distinction from these neoplasms. CD34 expression by an epithelioid neoplasm would be very helpful in such cases. Combining the three large series published to date, only three of 280 carcinomas

HUMAN

PATHOLOGY

Volume 25, No. 8 (August 1994)

were CD34 positive.“.‘“.” CD34 is not detectable on melanomas (none of 33 tested)‘s and is not reported to react with nonlym hoblastic lymphomas (none of 63 in r three series). “.“.‘. Other published markers of carcinoma and melanoma have not been tested on GIST; their utility must be considered limited. Of greatest importance in all of these distinctions, CD34 provides a positive marker for those GISTS that lack muscle markers, removing the need to depend entirely on the negative finding of a lack of other lineage markers. Their frequent and strong CD34 reactivity supports the concept that GISTS are distinct from smooth muscle neoplasms arising in other sites. Investigators have used two different anti-CD34 antibodies, My10 and QBENDIO, for the study of smooth muscle neoplasms. With My10 reactivity has been identified in three of 22 leiomyosarcomas of unstated site.‘” The first report of QBENDIO staining found none of four leiomyosarcomas to be CD34 reactive,i5 but a more recent series found QBENDlO to stain three of seven leiomyosarcomas, two of three epithelioid leiomyosarcomas, and three of three epithelioid leiomyomas (all of unstated site).“Another recent publication describes My10 reactivity with five of seven epithelioid leiomyosarcomas of soft tissue (spindled tumors were not tested) .I4 We have recently examined a series of uterine and soft tissue smooth muscle neoplasms using antibody MylO, confirming that strong CD34 expression is infrequent.” Twenty-six percent of 58 uterine smooth muscle tumors were CD34 positive, only one of which was strongly reactive. Furthermore, of the 17 uterine tumors exhibiting classical epithelioid features, only one was positive. Of 13 soft tissue leiomyosarcomas, we found six to be CD34 positive but only one was strongly so. More epithelioid soft tissue smooth muscle neoplasms will have to be tested, but our present findings with CD34 staining indicate that GISTS as a group are strikingly different from spindled and epithelioid uterine smooth muscle tumors and spindled smooth muscle tumors of soft tissue. This confirms the importance of separating GISTS from smooth muscle neoplasms of other sites and we strongly urge that all reports on the immunophenotype of smooth muscle-type neoplasms specify the sites of origin. In addition to showing immunoperoxidase staining for MylO, our Western blot results strongly indicate that the molecule being detected in GIST is indeed the CD34 molecule. The band detected by My10 is identical in migratory characteristics to that detected by My10 in cell line KGla, a granulocytic leukemia cell line used to define the CD34 molecule.‘” Identical results were ob tained when another CD34 monoclonal antibody, QBENDlO, was used. This finding virtually excludes cross-reactivity between epitopes on unrelated molecules as an explanation for an anti-CD34 reaction with KGla cells and GIST. Although Western blots of fresh human neoplasms certainly contain CD34 of vascular origin, several points argue against this as an explanation for our findings. The control leiomyosarcoma, which was CD34 negative by immunohistology, contained many CD34positive vessels; this contaminating endothelial cell CD34 is not sufficient to produce a de770

abc FIGURE 3. lmmunoblots on GIST (case no. 11,lane a>, KGla cell line (lane b), and metastatic leiomyosarcoma (lane c). Markers are indicated on the left side of the figure and represent 125. 94,715, and 51 kd.

tectable band on Western blotting. The intensity of the band detected also indicates an abundant molecule in the GIST case examined. The CD34 antigen has a perplexing distribution pattern on human neoplastic and normal tissues. In addition to GIST and smooth muscle neoplasms (described previously), it also has been found to stain significant numbers of leukemias, vascular tumors, hemangiopericytomas, nerve sheath tumors, epithelioid sarcomas,13 dermatofibrosarcomas,‘5 and solitary fibrous tumors of pleura and other sites.‘a Its pattern of reactivity with normal human cells is no less confusing. The CD34 antigen is present on approximately 1% of normal marrow hematopoietic cells.” This population includes progenitors of B, T, and myeloid cells because studies in nonhuman primates indicate that CD34pos itive bone marrow cells are both necessary and sufficient for completely reconstituting irradiated animals.H Endothelial cells also have been shown to express the CD34 molecule and messenger RNA (mKNA).’ The only other normal human cells that have been described as reactive with CD34 antibodies are precursors of bone marrow stromal cells” and dendritic and spindled fibroblastic cells in the dermis3i and soft tissue.3’ Molecular and genetic characterization of the basis of these last reactivities has not been reported. The function of the CD34 molecule in the above described cell types is unknown. Its molecular configuration suggests a role in cell signaling in hematopoietic cells with a rigid extracellular domain and multiple intracytoplasmic phosphorylation sites.‘8.33.34On endothelial cells its localization to the lumenal surface” and its regulation with other adhesion moleculess5 suggest a role in cell adhesion. It is not clear what role such a molecule might play in neoplastic smooth muscle cells. However, a recent publication shows downregulation of CD34 in cultured endothelial cells at the same time that endothelial leukocyte adhesion molecule 1 (ELAM-I) and intracellular adhesion molecule 1 (ICAM-1) are upregulated. This suggests a negative role in the mod-

CD34 IN GASTROINTESTINAL

TRACT

ulation of adhesion of endothelial cells for the CD34 molecule. It is intriguing to speculate that expression of CD34 in GIST might be associated with a loss of normal adhesion capability in the neoplastic cells. Acknowledgment. The authors thank Katherine Marlowe and Lourdes Villanueva for assistance with the immunohistochemistry. We also thank our colleagues who took the effort to send us tissue blocks and provided us with clinical information.

REFERENCES 1. Appelman HD, Helwig EB: Gastric epithelioid leiomyoma and leiomyosarcoma (leiomyoblastoma). Cancer 38:708-728, 1976 2. Lewin KJ, Riddell RH, Weinstein WM: Mesenchymal tumors, in Gastrointestinal Pathology and Its Clinical Implications. New York, NY, IgakuShoin, 1992, pp 284342 3. Enzinger FM, Weiss SW Soft Tissue Tumors. St Louis, MO, Mosby, 1983 4. Saul SH, Rast ML, Brooks JJ: The immunocytochemistry of gastrointestinal stromal tumors. Evidence supporting an origin from smooth muscle. Am J Surg Pathol 11:464-473, 1987 5. Pike AM, Lloyd RV, Appelman HD: Cell markers in gastrointestinal stromal tumors. HUM PATHOL 19:83@834, 1988 6. Miettinen M: Gastrointestinal stromal tumors. An immune histochemical study of cellular differentiation. Am J Clin Pathol 89:601-610,1988 7. Franquemont DW, Frierson KF: Muscle differentiation and clinicopathologic features of gastrointestinal stromal tumors. Am J Surg Pathol 16:947-954, 1992 8. Berenson RJ, Andrews RG, Bensinger WI, et al: Antigen CD34+ marrow cells engraft lethally irradiated baboons. J Clin Invest 81:951-955, 1988 9. Fina L, Molgaard HV, Robertson D, et al: Expression of the CD34 gene in vascular endothelial cells. Blood 75:2417-2426, 1990 10. Hanson CA, Ross CW, Schnitzer B: Anti-CD34 immunoperoxidase staining in paraffin sections of acute leukemia: Comparison with flow cytometric immunophenotyping. HUM PATHOI.23:2&32, 1992 11. Soligo D, Delia D, Oriani A, et al: Identification of CD34+ cells in normal and pathological bone marrow biopsies by QBENDlO monoclonal antibody. Leukemia 5:1026-1030,199l 12. Aziza J, Mazerolles C, Selves J, et al: Comparison of the reactivities of monoclonal antibodies QBENDlO (CD34) and BNH9 in vascular tumors. Appl Immunohistochem 1:51-57, 1993 13. Traweek ST, Kandalaft PL, Mehta P, et al: The human hematopoietic progenitor cell antigen (CD34) in vascular neoplasia. Am J Clin Pathol96:25-31, 1991 14. Sirgi KE, Wick MR, Swanson PE: B72.3 and CD34 immunoreactivity in malignant epithelioid soft tissue tumors. Adjuncts in the recognition of endothelial neoplasms. Am J Surg Pathol 17:179-l 85, 1993 15. Ramani P, Bradley NJ, Fletcher CD: QBEND/lO, a new monoclonal antibody to endothelium: Assessment of its diagnostic utility in paraffin sections. Histopathology 17:237-242, 1990 16. Satterthwaite AB, Borson R, Tenen DC: Regulation of the gene for CD34, a human hematopoietic stem cell antigen, in KG1 cells. Blood 75:2299-2304, 1990

STROMAL

TUMORS

(van de Rijn et al)

17. Blake MS, Johnston KH, Russell JG, et al: A rapid, sensitive method for detection of alkaline phosphatase-conjugated anti-antibody on Western blots. Anal Biochem 136:175-179, 1984 18. Ranched M, Kempson RL: Smooth muscle tumors of the gastrointestinal tract and retroperitoneum. A pathologic analysis of 100 cases. Cancer 39:255-262, 1977 19. Evans HL: Smooth muscle tumors of the gastrointestinal tract. A study of 56 cases followed for a minimum of 10 years. Cancer 56:2242-2250, 1985 20. Sutherland DR, Marsh JC, Davidson J, et al: Differential sensitivity of CD34 epitopes to cleavage by Pasteurella haemolytica glycoprotease: Implications for purification of CD34positive progenitor cells. Exp Hematol20:590-599, 1992 21. Ma CK, Amin MB, Kintanar E, et al: Immunohistologic characterization of gastrointestinal stromal tumors: A study of 82 cases compared with 11 cases of leiomyomas. Mod Pathol6:139-144, 1993 22. Norton AJ, Thomas JA, Isaacson PG: Cytokeratin-specific monoclonal antibodies are reactive with tumours of smooth muscle derivation. An immunocytochemical and biochemical study using antibodies to intermediate filament cytoskeletal proteins. Histopathology 11:487499,1987 23. Hurlimann J, Gardiol D: Gastrointestinal stromal turnours: An immunohistochemical study of 165 cases. Histopathology 19:311320, 1991 24. Spagnolo DV, Michie SA, Crabtree GS, et al: Monoclonal anti-keratin (AEl) reactivity in routinely processed tissue from 166 human neoplasms. Am J Clin Pathol84:697-704, 1985 25. Ordonez NG, Xiaolong J, Hickey RC: Comparison of HMB 45 monoclonal antibody and S-100 protein in the immunohistochemical diagnosis of melanoma. Am J Clin Pathol90:385-390, 1988 26. Falini B, Pileri S, Stein H, et al: Variable expression of leucocyte-common (CD45) antigen in CD30 (Kil)positive anaplastic largecell lymphoma: Implications for the differential diagnosis between lymphoid and nonlymphoid malignancies. HUM PATHOL.

21:624629,1990 27. Rizeq M, van de Rijn M, Hendrickson MR, et al: A comparative immunohistochemical study of uterine smooth muscle neoplasms with emphasis on the epithelioid variant. HUM PATH~L (in press) 28. Greaves MF, Brown J, Molgaard HV, et al: Molecular features of CD34: A hemopoietic progenitor cell-associated molecule. Leukemia 1:31-36, 1992 29. van de Rijn M, Lombard CM, Rouse RV: Expression of CD34 by solitary fibrous tumors of the pleura, mediastinum and lung. Am J Surg Pathol (in press) 30. Simmons PJ, Torok SB: CD34 expression by stromal precursors in normal human adult bone marrow. Blood 78:2848-2853,199l 31. Nickoloff BJ: The human progenitor cell antigen (CD34) is localized on endothelial cells, dermal dendritic cells, and perifollicular cells in formalin-fixed normal skin, and on proliferating endotheha1 cells and stromal spindle-shaped cells in Kaposi’s sarcoma. Arch Dermatol 127:523529, 1991 32. Hanau CH, Machera H, Miettinen M: Immunohistochemical evaluation of vascular invasion in carcinomas with five different markers. Appl Immunohistochem 1:46-50, 1993 33. He XY, Antao VP, Basila D, et al: Isolation and molecular characterization of the human CD34 gene. Blood 79:22962302, 1992 34. Simmons DL, Satterthwaite AB, Tenen DG, et al: Molecular cloning of a cDNA encoding CD34, a sialomucin of human hematopoietic stem cells. J Immunol 148:267-271, 1992 35. Delia D, Lampugnani MG, Resnati M, et al: CD34 expression is regulated reciprocally with adhesion molecules in vascular endothelial cells in vitro. Blood 81:1001-1008. 1993