Pleomorphic Carcinomas of the Lung Show a Selective ... up 4.pdf · Pleomorphic Carcinomas of the Lung Show a Selective Distribution of Gene Products Involved in Cell Differentiation,

ORIGINAL ARTICLE

Pleomorphic Carcinomas of the Lung Show a SelectiveDistribution of Gene Products Involved in Cell Differentiation,Cell Cycle Control, Tumor Growth, and Tumor Cell Motility

A Clinicopathologic and Immunohistochemical Study of 31 Cases

Giuseppe Pelosi, MD, MIAC, Filippo Fraggetta, MD, Oscar Nappi, MD, Ugo Pastorino, MD,Patrick Maisonneuve, Eng, Felice Pasini, MD, Antonio Iannucci, MD, Piergiorgio Solli, MD,

Hossein S. Musavinasab, MD, Giovanni De Manzoni, MD, Alberto Terzi, MD, andGiuseppe Viale, MD, FRCPath

Abstract: We investigated 31 cases of pleomorphic carcinomas ofthe lung, with a double component of neoplastic epithelial cells and ofspindle and/or giant cells. To correlate the morphologic diversity ofthese two cell components with their immunophenotype, we evalu-ated the expression of several gene products involved in cell differ-entiation (cytokeratins, epithelial membrane antigen, carcinoembry-onic antigen, vimentin, S-100 protein, smooth muscle actin, desmin),cell cycle control and apoptosis (p53, p21Waf1, p27Kip1, FHIT), tumorgrowth (proliferative fraction, assessed by Ki-67 antigen, and micro-vascular density, assessed by CD34 immunostaining), and tumor cellmotility (fascin). We found the epithelial component to be signifi-cantly more immunoreactive for cytokeratins, epithelial membraneantigen, carcinoembryonic antigen, cell cycle inhibitors p21Waf1,p27Kip1 and tumor suppressor gene FHIT, whereas the sarcomatoidcomponent, independent of tumor stage and size, was more immuno-reactive for vimentin, fascin, and microvascular density. Accord-ingly, we suggest a model of tumorigenesis whereby the mesenchy-mal phenotype of pleomorphic cells is likely induced by the selectiveactivation and segregation of several molecules involved in cell dif-ferentiation, cell cycle control, and tumor cell growth and motility.Whether pleomorphic carcinomas of the lung are tumors with a dis-mal prognosis still remains an unsettled issue. In our series, however,stage I pleomorphic carcinomas have the same clinical behavior as

ordinary non-small cell lung cancer, and only a high proliferative in-dex (Ki-67 labeling index >35%) is associated with a worse prognosisin these tumors.

Key Words:pleomorphic carcinomas, immunohistochemistry, lung,morphogenesis, prognosis

(Am J Surg Pathol 2003;27:1203–1215)

Among non-small cell lung cancers (NSCLCs), tumors con-taining variable amounts of spindle and/or giant cells, or

entirely made up of these cells, are uncommon but well-documented entities.

Pathologists faced with these tumors have often experi-enced difficulties in rendering a definite diagnosis, especiallybecause of the lack of uniformity in terminology and of stan-dardized diagnostic criteria. Many designations have beenproposed in the past to identify lung tumors containing truesarcomatous or sarcomatoid components, including biphasicand monophasic sarcomatoid carcinoma, pleomorphic carci-noma (PC), spindle cell carcinoma, giant cell carcinoma,pseudosarcoma, pulmonary blastoma, and carcinosar-coma.1,2,5,7,8,13,15,19,20,24,27,28,35–37,45,51,53

In the previous WHO classification, lung tumors show-ing a prevalence of spindle or giant cells were considered vari-ant of squamous cell carcinomas and of large cell carcinomas,respectively.52 The most recently updated WHO classificationstates that there exists “a group of poorly differentiated non-small cell lung cancer that contain a component of sarcoma orsarcoma-like elements.”49 Accordingly, PCs are defined as “apoorly differentiated NSCLC, namely squamous cell carci-noma, adenocarcinoma or large cell carcinoma, containingspindle cells and/or giant cells, or a carcinoma consisting onlyof spindle cells and giant cells. The pleomorphic componentshould comprise at least 10% of the neoplasm.” Thus, the des-ignation spindle cell carcinoma should be restricted to

Received for publication December 12, 2002; accepted January 30, 2003.From the Departments of Pathology and Laboratory Medicine (G.P., H.S.M.,

G.V.), Thoracic Surgery (U.P., P.S.), European Institute of Oncology, Sta-tistics and Epidemiology (P.M.), and University of Milan School of Medi-cine (G.P., H.S.M., G.V.), Milan; Departments of Pathology (F.F.), Can-nizzaro Hospital, Catania; and Cardarelli Hospital (O.N.), Napoli; Depart-ments of Medical Oncology (F.P.,) and General Surgery (G.D.),University of Verona, and Departments of Pathology (A.I.) and ThoracicSurgery (A.T.), City Hospital, Verona, Italy.

Address correspondence and reprint requests to Giuseppe Pelosi, MD, MIAC,Divisione di Anatomia Patologica e Medicina di Laboratorio Istituto Eu-ropeo di Oncologia, Via G. Ripamonti, 435 I-20141, Milano, Italy; e-mail:[email protected]

Copyright © 2003 Lippincott Williams & Wilkins

Am J Surg Pathol • Volume 27, Number 9, September 2003 1203

NSCLCs that are constituted exclusively of spindle cells; giantcell carcinoma to those composed of giant cells, and carcino-sarcoma to those including heterologous elements such as ma-lignant bone, cartilage, vessels or skeletal muscle. Spindle orgiant cell tumors, however, are very uncommon in their pureform,1,51 and most of them fall in the category of PCs owing tothe occurrence of a dual cell component.49 Moreover, the rareoccurrence of a sarcomatoid and a large cell neuroendocrinecomponent in the same tumor18 is now considered a variant oflarge cell neuroendocrine carcinoma.49 The correct identifica-tion of PCs relies exclusively on pure histologic criteria with-out requiring any additional immunohistochemical or ultra-structural evidence.49

Although the actual nature of the pleomorphic cells hasbeen debated at length, an epithelial origin is currently ac-cepted. Several studies have supported the view that epithelialtumor cells of PCs may acquire the expression of vimentin andof other mesenchymal markers,1,8,27,28,37,51,53 and this pheno-type may eventually result in spindling of the cells and reducedcell-to-cell adhesion. Few data, however, are available on thecorrelation between the morphologic diversities of the epithe-lial and sarcomatoid components within individual tumors andtheir corresponding immunophenotypes.

Another unresolved issue is whether PC has a differentprognosis than conventional NSCLC. Some studies havereported PCs to be associated with more advanced dis-ease stage at presentation and more aggressive clinicalcourse,5,8,19,20,28,35,37,45,51 but other investigations failed todemonstrate any significant poorer prognosis for PCs.1,27

This study was aimed at evaluating in a large series ofPCs the different immunophenotypic features of the epithelialand the pleomorphic components within individual tumors, ac-cording to the expression of several gene products involved incell differentiation (cytokeratins, epithelial membrane antigen[EMA], carcinoembryonic antigen [CEA], chromogranin A,synaptophysin, vimentin, S-100 protein, smooth muscle actin,desmin), cell cycle control and apoptosis (p53, p21Waf1,p27Kip1, FHIT), tumor growth (proliferative fraction assessedby KI-67 labeling and neoangiogenesis inferred by CD34staining of endothelial cells), and tumor cell motility (fascin).Our results indicate that most of these markers are differen-tially expressed in the two tumor components and that theymay be critical in inducing morphologic changes leading to thepleomorphic phenotype.

PATIENTS AND METHODS

PatientsThirty-one patients with pulmonary PCs were identified

in the files of the Departments of Pathology and Thoracic Sur-gery of the City Hospital in Verona (1987–1993) and the Eu-ropean Institute of Oncology in Milan (1998–2001). For eachcase, all paraffin blocks were retrieved, and archival hema-

toxylin and eosin sections were reviewed. All patients hadbeen studied preoperatively with clinical history, physical ex-amination, respiratory tests, chest x-ray, total body CT scan,bone scintigraphy, and routine laboratory profile, and all un-derwent radical surgery. Only patients with a minimum 30-daypostoperative survival were considered in the study. The clini-cal and demographic features of PCs are summarized in Table1. There were 28 males and 3 females, ranging in age from 39to 79 years (mean ± SD, 62.4 ± 10.6 years; median 63 years).All but 5 patients had local or systemic symptoms at the time ofthe initial diagnosis, with Karnowsky performance status rang-ing from 70% to 100%. Complete follow-up information wasavailable for all patients, with a mean duration of 36.5 ± 39.1months (median 13 months; range 2–114 months). Recurrentdisease at different sites was seen in 15 (48.3%) patients; 9(29%) of them died of disease. Three patients died of unrelateddisease (1 of acute gastric hemorrhage, 1 of adult respiratorydistress syndrome, and 1 of acute heart failure). According tothe revised international system for staging for lung cancer,26

there were 16 patients with either p-stage IA (6 cases) or IB (10cases), and 15 with either p-stage IIB (11 cases) or IIIA (4cases). Two patients with disease stage IIB and IIIA underwentneoadjuvant chemotherapy, whereas no patients with diseasestage I underwent any additional therapy.

The diagnosis of PC was rendered according to the cur-rent WHO guidelines.49 The percentages and the types ofthe pleomorphic (spindle, giant cells, or mixed) and epithelial(adenocarcinoma, squamous cell carcinoma, or large cellcarcinoma) components were assessed in every case by scan-ning the whole tumor. Tumor necrosis was evaluated semi-quantitatively on a scale from absent to 2+ (1+, if <50%; 2+, if>50% of the whole tumor). The occurrence of vascular inva-sion within individual tumors was also recorded after CD34immunostaining of the vascular channels, without discriminat-ing the relative impact of arterial, venous and lymphaticinvasion.

ImmunohistochemistryFormalin-fixed, paraffin-embedded surgical tissue

samples were investigated in the study. Tumors were entirelyimmunostained if �3 cm in size, whereas at least two repre-sentative tissue blocks were evaluated in larger neoplasms.The immunohistochemical experiments were performed usingthe primary antibodies listed in Table 2 and a commerciallyavailable detection kit (EnVision Plus-HRP, Dako, Glostrup,Denmark), according to the manufacturer’s suggestions and topreviously refined methods.32 Peroxidase activity was devel-oped with 3-3�-diaminobenzidine-copper sulfate (SigmaChemical Co, St Louis, MO) to obtain a brown-black end prod-uct. The specificity of all immunoreactions was double-checked by substituting the primary antibodies with nonrelatedisotypic mouse immunoglobulins at a comparable dilution, or

Pelosi et al Am J Surg Pathol • Volume 27, Number 9, September 2003

1204 © 2003 Lippincott Williams & Wilkins

with normal serum alone. Appropriate internal and externalpositive controls were also used in all immunostaining proce-dures to ensure specificity of reaction.

In each tumor, the pleomorphic and epithelial compo-nents were assessed separately, recording the percentage (la-beling index) of neoplastic cells immunoreactive for cytoker-atins, EMA, CEA, vimentin, S-100 protein, smooth muscle ac-tin, desmin, fascin, p53, p21Waf1, p27Kip1, Ki-67 antigen, andFHIT. Only immunostaining products confined to the cyto-plasm of tumor cells were taken into account for cytokeratins,

EMA, CEA, chromogranin A, synaptophysin, vimentin, S-100protein, smooth muscle actin, desmin, fascin, and FHIT. Forp53, p21Waf1, and Ki-67 antigen, only nuclear accumulationwas recorded in the corresponding labeling index, whereasp27Kip1 immunoreaction could exhibit either cytoplasmic ornuclear decoration. To decrease the subjectivity of the scoringprocess, tumor cells were counted as positive if moderate tostrong immunolabeling products were recognizable in eitherthe cytoplasm or the nucleus as compared with appropriate in-ternal or external controls. If reactivity was limited to few and

TABLE 1. Clinical Data of 31 Patients With Pleomorphic Carcinoma of the Lung

Case No.Age

(yr)/Sex Symptoms PS Site of RelapseDFS(mo)

OS(mo) Outcome Stage

1 64/M Hemoptysis 100 — 2 2 DOUD* IIIA2 73/M Cough 90 — 2 2 AW IIB3 72/M Dyspnea 90 CNS 12 18 DOD IA4 48/F Hemoptysis 100 Bone, CNS, lung 71 90 DOD IB5 62/M Cough, fever 90 Lung 38 66 DOD IA6 75/M — 90 — 110 110 AW IA7 55/M Fever, cough 80 — 2 2 AW IIB8 72/M Hemoptysis 80 — 13 13 AW IIB9 44/M Thoracic pain 100 Thoracic wall, adrenal gland 12 14 AWD IIB

10 59/M Thoracic pain 100 — 78 78 AW IA11 54/M Thoracic pain 90 Bone 9 10 AWD yIIB12 64/M Cough 80 Bone 24 31 DOD IB13 71/M Hemoptysis 90 — 80 80 AW IB14 69/M Cough, fever 80 — 114 114 AW IB15 72/M — 100 Liver, peritoneum 14 14 AWD IIB16 75/M Hemoptysis, thoracic pain 90 Bone 2 6 DOD IB17 56/M Cough, fever 90 — 18 18 AW IIIA18 64/M Hemoptysis, cough 90 CNS 3 3 AWD yIIIA19 61/M Cough 100 — 79 79 AW IB20 71/M Cough, dyspnea 90 Lung 27 36 DOD IB21 50/M Hemoptysis 90 Thoracic wall 3 3 AWD IIB22 39/M Thoracic pain 100 — 4 4 AW, then lost

to follow-upIIB

23 66/F — 100 — 78 78 AW IA24 58/M — 100 Bone 2 6 DOD IB25 76/M Cough, fever 80 Bone 4 25 DOUD† IB26 63/M — 90 Lung 9 23 DOD IB27 56/M Hemoptysis 100 — 88 97 AWD IA28 43/M Cough, fever, thoracic pain 100 Thoracic wall, liver 5 8 DOD IIIA29 69/M Cough, dyspnea 100 — 2 2 DOUD‡ IIB30 57/F Hemoptysis 90 — 12 12 AW IIB31 79/M Cough 70 — 2 2 AW IIB

PS, performance status; DFS, disease-free survival; OS, overall survival; DOUD, died of unrelated disease; AWD, alive with disease; DOD, died of disease;AW, alive and well; y, neo-adjuvant chemotherapy; CNS, central nervous system.

*Acute gastric hemorrhage.†Acute heart failure.‡Adult respiratory distress syndrome.

Am J Surg Pathol • Volume 27, Number 9, September 2003 Pleomorphic Carcinoma of the Lung

© 2003 Lippincott Williams & Wilkins 1205

minute particles of faint immunostaining, it was considered atrace reaction and not computed in the corresponding index.

One observer (G.P.) evaluated all results of immuno-staining blindly, without knowledge of the patients’ identity orclinical outcome. For labeling indexes, the percentages of im-munoreactive cells were obtained scanning at least 1,000 tu-mor cells in randomly selected fields: three different countswere obtained for each marker from either epithelial or sarco-matoid areas, and the highest score for each marker was chosenas the corresponding labeling index. For tumor angiogenesis,inferred by microvessel density (MVD) after CD34 immuno-staining of endothelial cells, the mean number of vessels de-tected in three ×200 microscopic fields (each field correspond-ing to an area of 0.785 mm2) in the hot spot areas was insteadrecorded.

Statistical AnalysisAssociations of categorical variables were evaluated by

Fisher exact test t test of �2 test. The comparison of continuousvariables was performed using Wilcoxon’s signed rank test forpairs. The intraobserver reproducibility was evaluated byanalysis of variance. Overall survival was defined as the timebetween surgery and the last follow-up or cancer death. If apatient died without cancer recurrences, the patient’s survivaltime was censored at the time of death. Only lung cancer re-lated deaths or recurrences were considered as events. Dis-ease-free survival was calculated from the date of surgery tothe date of progression or the date of the last follow-up. Sur-

vival estimates were calculated with Kaplan-Meier’s methodand compared by the log-rank test. All analyses were carriedout using the SAS statistical software (SAS Institute, Inc.,Cary, NC). All P values were based on two-sided testing, andconfidence intervals were set at a 95% level.

RESULTSThe pathologic findings are summarized in Table 3.

Gross FindingsGrossly, the neoplasms ranged from 1.9 cm to 17 cm in

size. Typically, the tumors showed areas of necrosis and hem-orrhage, but a double component of sarcomatoid and carcino-matous areas could not be macroscopically suspected in all butone case (case no. 28) (Fig. 1).

Microscopic FindingsRepresentative features of PCs are depicted in Figure 2.

The epithelial components ranged from 10% to 85% of thewhole tumor area and consisted of squamous cell carcinoma in3 cases (Fig. 2A), adenocarcinoma in 14 cases (Fig. 2B), andlarge cell carcinoma in 13 cases. One tumor (case no. 28)showed a triple epithelial component of bronchioloalveolarcarcinoma, adenocarcinoma, and squamous cell carcinoma.The epithelial component was poorly differentiated in 24 tu-mors (13 large cell carcinomas, 9 adenocarcinomas, and 2squamous cell carcinomas), whereas 6 cases showed an inter-mediate differentiation (5 adenocarcinomas and 1 squamous

TABLE 2. Antibody Panel Used in the Current Study

Antibody Clonality Clone Source Dilution Pretreatment

Cytokeratin pool m AE1-AE3 Novocastra Laboratories, Newcastle upon Tyne, UK 1:50 MWO-CBVimentin m V9 DAKO, Glostrup, Denmark 1:50 MWO-CBEMA m E29 DAKO, Glostrup, Denmark 1:20 MWO-CBCEA m 11-7 DAKO, Glostrup, Denmark 1:50 NoneSmooth muscle actin m 1A4 DAKO, Glostrup, Denmark 1:200 NoneDesmin m D33 DAKO, Glostrup, Denmark 1:150 MWO-CBS-100 protein p — DAKO, Glostrup, Denmark 1:1600 Nonep53 m DO7 DAKO, Glostrup, Denmark 1:1000 MWO-CBp21Waf1 m EA10 Oncogene Science, Cambridge, MA 1:100 MWO-CBp27Kip1 m 56 Transduction Laboratories, Lexington, KY 1:600 MWO-CBFHIT p — Immunotech, Marseille, France 1:400 MWO-EDTAKi-67 antigen m MIB-1 Immunotech, Marseille, France 1:400 MWO-EDTAFascin m IM20 Novocastra Laboratories, Newcastle upon Tyne, UK 1:400 MWO-EDTACD34 m QBEnd/10 Novocastra Laboratories, Newcastle upon Tyne, UK 1:400 MWO-CBChromogranin A m LK2H10 Signet Laboratories, Dedham, MA 1:40 NoneSynaptophysin m SY38 DAKO, Glostrup, Denmark 1:20 MWO-CB

MWO-CB, microwave oven at 750 W for 20 minutes in citrate buffer, pH 6; MWO-EDTA, microwave oven at 750 W for 12 minutes in EDTA buffer, pH 8;M, monoclonal; p, polyclonal.



cell carcinoma). The tumor with triple epithelial componentshowed a well-differentiated bronchioloalveolar carcinomaand poorly differentiated adenocarcinoma and squamous cellcarcinoma. The sarcomatoid component accounted for 15% to90% of the tumor and was spindle in 5 cases (Fig. 2A, C, D, E),giant in 7 (Fig. 2B, F) or mixed in the remaining 19 cases.There was no preferential association between the differenttypes of either epithelial and sarcomatoid components. Thespindle cells component featured either relatively uniformspindle cells resembling fibrosarcoma or malignant nervesheath tumor (Fig. 2C) or highly pleomorphic cells resemblingmalignant fibrous histiocytoma. The latter were arranged ei-

ther haphazardly or in a fascicular or vaguely storiform con-figuration. The giant cell component consisted predominantlyof bizarre giant cells with multilobated nuclei and abundanteosinophilic cytoplasm, which sometimes engulfed leukocytes(Fig. 2F). The mixed type resulted from a variable proportionof highly atypical giant and spindle tumor cells, intimately ad-mixed or segregated in distinct tumor areas. In some cases,especially those composed of spindle cells, the tumor matrixwas at least partially myxoid (Fig. 2D) or contained a variableamount of leukocytes, plasma cells, and collagen fibers. Mi-totic figures could be numerous, especially in spindle cell areaswhere prominent vascularization was also seen (Fig. 2E).

TABLE 3. Pathologic Characteristics of 31 Patients with Pleomorphic Carcinoma

Case no.Size(cm) pT pN

Epithelialcomponent

Grading ofepithelial component

Pleomorphiccomponent % PLC

Vascularinvasion

Tumornecrosis*

LN metastasismorphology

1 4.5 3 2 LCC G3 Mixed 75 Yes + LCC + mixed2 3.6 2 1 LCC G3 Giant 40 Yes ++ AC + giant3 1.9 1 0 LCC G3 Mixed 40 No ++ —4 6.5 2 0 LCC G3 Mixed 90 No ++ —5 2.5 1 0 SCC G2 Spindle 15 No − —6 1.8 1 0 AC G3 Mixed 60 Yes − —7 5.5 2 1 AC G2 Mixed 80 Yes ++ AC + mixed8 5.5 2 1 AC G3 Spindle 80 No ++ AC9 3.5 3 0 AC G3 Mixed 70 No ++ —

10 1.5 1 0 AC G3 Mixed 50 Yes − —11 4.5 y3 y0 LCC G3 Mixed 75 No + —12 5 2 0 LCC G3 Giant 25 No + —13 9 2 0 AC G3 Giant 80 No ++ —14 8 2 0 LCC G3 Spindle 50 No ++ —15 5.5 3 0 LCC G3 Mixed 50 No ++ —16 4 2 0 LCC G3 Giant 40 No − —17 7.5 2 2 AC G2 Mixed 90 Yes + Diffuse necrosis18 3.5 y2 y2 AC G2 Mixed 70 Yes + AC + mixed19 6 2 0 SCC G3 Mixed 40 Yes + —20 4 2 0 LCC G3 Giant 40 Yes + —21 9 3 0 AC G3 Mixed 60 Yes ++ —22 10.5 3 0 LCC G3 Giant 50 No + —23 2.2 1 0 AC G3 Giant 25 Yes ++ —24 6 2 0 AC G3 Mixed 30 Yes ++ —25 11 2 0 AC G2 Mixed 65 Yes ++ —26 3.5 2 0 LCC G3 Mixed 40 Yes + —27 3 1 0 LCC G3 Mixed 15 Yes + —28 17 3 2 BAC/AC/SCC G1/G3 Spindle 70 Yes ++ AC29 10.5 2 1 SCC G3 Mixed 45 Yes ++ SCC30 3.2 2 1 AC G3 Spindle 90 No + AC + spindle31 6 3 0 AC G2 Mixed 95 Yes ++ —

% PLC, percentage of the pleomorphic component within nonnecrotic parts of individual tumors; AC, adenocarcinoma; LCC, large cell carcinoma; SCC,squamous cell carcinoma; BAC, bronchioloalveolar carcinoma; Mixed, spindle and giant cell components; y, patients undergoing preoperative chemotherapy.

* Necrosis: −, absent; +, <50%; ++, >50%.



Tumor necrosis was encountered in all but 4 cases, rang-ing from focal (<50% of the whole tumor) in 11 cases to diffuse(>50%) in the remaining 16 tumors. Vascular invasion wasrecognizable in 18 (58%) cases, either within tumors or at theedge of the neoplastic masses, sometimes with evidence of dis-ruption of the vascular elastic wall and luminal effacement bytumor cells. However, no systematic discrimination was madeas for arterial, venous, or lymphatic invasion because of therelatively small number of examined tumors. There was nopreferential distribution of tumor necrosis or vascular invasionin the two components, either epithelial or sarcomatoid.

Nine cases had lymph node metastases: in 5 cases, themetastases showed the same biphasic growth pattern as seen inthe primary tumors, in 3 cases, only the epithelial component(ie, adenocarcinoma or squamous cell carcinoma) was identi-fied; and in the last case, the metastasis was extensively ne-crotic and not further assessable.

No correlations were found between the percentage ofpleomorphic cells within individual tumors and age, sex, per-formance status, vascular invasion, tumor necrosis, or tumorsize. Only tumor stage was statistically related to the percent-age of pleomorphic cells: stage I tumors were usually made up�50% of pleomorphic component, whereas the opposite oc-curred in stages II and III (Table 4).

Immunohistochemical FindingsThe results of the immunohistochemical study are de-

tailed in Table 5. Evaluation of intraobserver variation showed

no statistically significant differences in immunostaining forall markers under evaluation in the whole series of tumors(F < 0.01, P > 0.90).

Cell Differentiation

The epithelial components, independent of the histo-logic type, tumor size, and stage were more consistently andmore intensely immunoreactive for cytokeratins and EMA,whereas the opposite occurred for vimentin (Fig. 3A–D).CEA immunostaining was observed in only 16 tumors butwas significantly associated with the epithelial components.Occasional cells (<5%) were immunoreactive for smoothmuscle actin, desmin, or S-100 protein, but they were con-fined to the sarcomatoid component, especially when spindleshaped. No tumor cells, either epithelial or sarcomatoid, dis-played chromogranin A or synaptophysin immunoreactivity.

Cell Cycle Control and Apoptosis

Immunohistochemical expression of p21Waf1 andp27Kip1 proteins (the latter seen both in the nucleus and in thecytoplasm) was confined to the epithelial component. Thiswas independent of tumor growth patterns, size, and stage ofdisease. FHIT gene product was more commonly upregu-lated in the epithelial components, although the differenceswere statistically significant only for tumors with stage >I.No differences were observed in the distribution of p53-immunoreactive cells between the two cell components (Fig.3E, G, H).

Tumor Cell Motility

Fascin immunoreactivity appeared as a fine, granularto diffuse cytoplasmic staining of both normal and neo-plastic cells (Fig. 3F). In the normal lung of either non-carcinoma or carcinoma patients’ group, fascin im-munoreactivity decorated invariably endothelial cellsof bronchial and alveolar wall microvessels and dendriticcells of the mucosa-associated lymphoid tissue. Fascin im-munoreactivity was most common in the pleomorphic areas,independent of tumor growth patterns, size, and stage ofdisease.

Tumor Growth

No statistically significant difference was observed inthe distribution of Ki-67-immunoreactive cells between thetwo cell components, even after adjusting for tumor growthpatterns, size, and stage of disease. MVD was more promi-nent in the pleomorphic component, independent of tumorgrowth patterns, size, and stage of disease.

Survival Data AnalysisStage I tumors behaved more aggressively than ordi-

nary NSCLCs, but differences in disease-free or overall sur-vival were not statistically significant. About 50% of these

FIGURE 1. Gross pathology of a large PC (case no. 28) occu-pying almost the entire lung and with invasion of the thoracicwall and the diaphragm. A, The latter component was pre-dominantly composed of sarcomatoid cells and appearedsofter and gelatinous. B, Pulmonary scintigraphy showedblood perfusion only in the residual portion of the upper pul-monary lobe.



patients died of disease or showed tumor recurrence after 5years, a value strictly comparable with the clinical outcome ofordinary stage I NSCLCs (Fig. 4). No survival analysis wasperformed in higher stage tumors because of the lack of ad-equately long follow-up time.

Moreover, we investigated whether the expression ofdifferent gene products involved in cell differentiation, cellcycle control, apoptosis regulation, tumor growth, and cell mo-tility could have a prognostic relevance in PCs of the lung. Inour study, however, only a Ki-67 labeling index >35% corre-lated negatively with both overall and disease-free survival(Table 6).

DISCUSSIONThe occurrence of a sarcomatoid or pleomorphic com-

ponent, with giant or spindle cells, is an uncommon but well-documented feature in some NSCLCs, and it is usually consid-ered to confer the neoplasms a more aggressive clinical coursethan ordinary lung cancer.5,8,19,20,28,35,37,45,51 Although the na-ture of these pleomorphic cells has been debated at length, anepithelial derivation with divergent mesenchymal differentia-tion is now accepted.28,47

In the current study, we sought to correlate the morpho-logic diversity of the cell components of PCs with their immu-nophenotype for different molecules involved in cell differen-

FIGURE 2. Histologic features of biphasic PCs of the lung composed of spindle and squamous cells (A) or glandular and giant cells(B). Spindle tumor cells could grow up in fascicles resembling fibrosarcoma or malignant nerve sheath tumor (C), sometimes withaccumulation of myxoid matrix in the tumor stroma (D), numerous mitotic figures, or prominent vascularization (E). A few PCsshowed a giant cell component with numerous leukocytes resembling malignant fibrous histiocytoma (F).



tiation, cell cycle control and apoptosis, tumor growth, and cellmotility. From the results, we could derive a model of tumormorphogenesis, whereby the pleomorphic features of the neo-plastic cells closely parallel the expression of the differentgene products (Fig. 5). Although we cannot exclude that someof these molecules studied are actually downstream covariantsand that other genes are involved in the causative developmentof these tumors, we first indicate that epithelial and sarcoma-toid cells in PCs of the lung are differentially regulated at thecellular level as far as the semiquantitative distribution of sev-eral molecules is concerned. This model is useful to summa-rize the statistically significant relationships between mor-phology and immunohistochemical results, without implyingthat the expression of the targets studied is able alone to induceepithelial–mesenchymal transition in these tumors.

The epithelial malignant component retains the expres-sion of epithelial differentiation markers (ie, cytokeratins,CEA, and EMA), and consistently shows upregulation of thecell cycle inhibitors (ie, p21Waf1 and p27Kip1) and of FHIT, amolecule related to apoptosis. However, the pleomorphic com-ponent predominantly expresses cytoskeletal proteins of the

mesenchymal type (vimentin, but occasionally also desminand smooth muscle actin) and fascin, a cell motility-relatedprotein, as well as shows increased levels of MVD, eventhough the pleomorphic cells may retain residual expression ofcytokeratins and other epithelial markers. Altogether, thesefindings are consistent with a more severe dysregulation ofseveral aspects of cellular life (cell cycle control, apoptosis,tumor growth, and cell motility) in the pleomorphic tumorcells, thus supporting the view that they may be an especiallyaggressive component of these tumors, albeit derived from ep-ithelial cells. Interestingly, no tumor cells, either epithelial orsarcomatoid, displayed neuroendocrine differentiation as as-sessed by chromogranin A or synaptophysin immunoreactivi-ty. This indicates that PCs can not be considered in the list ofpulmonary tumors showing neuroendocrine differentiation.Although neuroendocrine properties may be unexpectedly de-tected in 10% to 20% of ordinary NSCLCs, especially adeno-carcinomas and large cell carcinomas and less commonlysquamous cell carcinomas,5,49 our results in PCs support theview that the engagement of neuroendocrine properties doesnot play an important role in the development and morphogen-esis of these tumors.

It has been suggested that carcinoma cells may undergoa spectrum of progressive phenotypic changes that lead to theappearance of sarcomatoid features, via the activation of mes-enchymally conditioned pathways resulting in the accumula-tion of vimentin and the loss of cytokeratin filaments.6,47 Thecytoplasm reorganization confers higher plasticity to the tumorcells, which show reduced intercellular cohesion, due to thelack of keratin-based cell adhesion systems, progressive cellelongation with spindling, and reduced production of base-ment membranes with extracellular matrix invasion.47 Manyof these changes derive from cytoskeletal rearrangement, andthe bundling of actin microfilaments at the cell periphery mayplay an important role in this change. Human fascin, a 55-kDaactin-bundling protein, is known to induce membrane protru-sions at the leading edges of the cells and to increase cell mo-tility of normal and transformed epithelial cells.9,23,44,46,48,54 Itis therefore not surprising that fascin immunoreactivity isclosely related to the pleomorphic phenotype, independent oftumor growth pattern, size, and stage. This suggests that fascinupregulation is an early event in the morphogenesis of PCs,and not a feature of more advanced tumors.

Down-regulation of p21Waf1 and p27Kip1 is also a con-sistent feature of the pleomorphic cell component of PCs, andan aberrant regulation of these cyclin-dependent kinase inhibi-tors has been commonly seen in the development of manytypes of human malignancies.4,14,17,21,22,29 These alterationsare not accompanied by any corresponding abnormal accumu-lation of p53, in keeping with a previous study reporting nodifferential immunoreactivity for p53 in the epithelial and sar-comatoid components of pulmonary carcinosarcoma.13

TABLE 4. Correlation between the Percentage ofPleomorphic Cells and Clinicopathologic Data

% Pleomorphic

Exact P value<50% >50%

Age0–59 yr 4 860+ yr 12 7 NS

SexMales 15 13Females 1 2 NS

InvasionNo 7 6Yes 9 9 NS

StageI 12 4>I 4 11 0.012

NecrosisNo 9 4Yes 7 11 NS

PS�80 3 390 6 7100 7 5 NS

Diameter<5 cm 7 85 cm+ 9 7 NS

PS, performance status; NS, not significant.



The FHIT gene product, a tumor suppressor gene locatedat the 3p14.2 fragile site that is frequently and early deleted inlung cancerogenesis,38,40,43 also shows a different prevalencein the two cell components of PCs, but only in higher stagecarcinomas.

Finally, in the pleomorphic component an increasedMVD may be documented, in keeping with previous data onthe higher vascularization of sarcomatous than carcinomatousmetastases to the lung.50 This suggests that the pleomorphiccomponent may acquire some phenotypic similarities to truesarcoma, even though it remains basically a carcinoma in itsnature. A consequence of this finding, however, could be thepossibility of using antiangiogenetic drugs in the managementof these patients, although further studies are needed to con-firm this assumption.

The selective distribution of most of the molecules underinvestigation strongly suggests that PC represents a distinctpathologic entity in the scenario of NSCLC. The differential

regulation at the cellular level of several markers suggests thetwo cell components, either epithelial and sarcomatoid, are notsimply the result of intratumoral heterogeneity of NSCLC. Infact, a similar differential distribution of these markers was notdemonstrable within individual NSCLCs of conventional typewhen the latter were assessed in different areas of the sametumor (data not shown). As the sarcomatoid cells in our PCsretained variable phenotypic expressions of epithelial markers,we speculate that PCs are metaplastic carcinomas derived fromtrans-differentiation of malignant epithelial cells into sarcoma-like cells. A metaplastic origin of biphasic PCs is also sup-ported by a recent study on carcinosarcomas, a classic proto-type of biphasic tumors of the lung.13 In these neoplasms, amultistep progression of the sarcomatous component fromcommon totipotent cells through an intermediate stage of car-cinoma has been demonstrated, showing a greater amount ofgenetic damages in the mesenchymal than in the epithelialcomponent.

TABLE 5. Individual Numeric Results of the Immunohistochemical Study

Marker

All stages Stage I State > I

Mean ± SD Median P Value‡ Mean ± SD Median P Value‡ Mean ± SD Median P Value‡

Ck E* 75.7 ± 13.8 80 <0.001 71.7 ± 18.6 80 0.002 80.0 ± 0 80 <0.001Ck Pl* 29.3 ± 25.2 18 39.1 ± 27.6 31.5 18.8 ± 17.7 18Vim E* 25.5 ± 26.4 18 <0.001 29.3 ± 26.6 18 0.002 21.5 ± 26.5 18 <0.001Vim Pl* 70.1 ± 21.8 80 63.0 ± 27.7 80 77.6 ± 9.0 80Ema E* 69.5 ± 19.6 80 <0.001 69.0 ± 16.7 80 0.004 70.0 ± 22.9 80 0.001Ema Pl* 29.9 ± 29.2 18 37.3 ± 29.4 31.5 22.1 ± 27.8 18CEA E* 14.7 ± 21.3 2.5 <0.001 13.3 ± 19.7 10.2 0.008 16.3 ± 23.5 2.5 0.008CEA Pl* 3.2 ± 9.4 0 6.3 ± 12.5 0 0.0 ± 0.0 0p53 E* 27.9 ± 31.9 12 0.521 26.1 ± 33.7 6 0.695 29.8 ± 30.9 35 0.203p53 Pl* 31.2 ± 33.8 15 22.9 ± 27.5 10.5 40.2 ± 38.4 52Ki67 E* 40.0 ± 16.6 36 0.101 39.2 ± 16.9 33 0.224 40.9 ± 16.8 40 0.002Ki67 Pl* 47.2 ± 20.9 50 35.0 ± 18.6 30.5 60.2 ± 14.6 65p21 E* 28.8 ± 19.9 30 <0.001 32.0 ± 22.8 32.5 <0.001 25.4 ± 16.4 30 0.002p21 Pl* 13.5 ± 12.7 14 14.7 ± 12.8 16.5 12.3 ± 12.9 9p27nE* 40.5 ± 29.1 40 <0.001 33.5 ± 28.9 34.5 0.003 48 ± 28.3 50 0.019p27nPl* 24.2 ± 21.0 15 17.5 ± 21.7 10.5 31.4 ± 18.3 32p27cE* 10.8 ± 15.0 4 0.001 5.6 ± 9.3 1 0.047 16.3 ± 18.1 10 0.017p27cPl* 2.3 ± 4.5 0 1.5 ± 3.3 0 3.2 ± 5.5 0FHIT E* 24.1 ± 26.8 20 0.008 27.5 ± 27.2 20 0.195 20.6 ± 26.9 20 0.031FHIT Pl* 14.8 ± 22.6 5 21.5 ± 28.1 10 7.6 ± 11.9 0Fascin E* 52.6 ± 29.5 45 <0.001 47.3 ± 33.9 45 0.031 58.3 ± 23.8 45 0.008Fascin Pl* 70.3 ± 22.2 80 63.4 ± 28.3 80 80.0 ± 0.0 80MVD E† 71.9 ± 52.0 57 <0.001 80.4 ± 63.3 66.3 0.015 62.7 ± 36.4 51 0.001MVD Pl† 05.1 ± 67.3 90 110.1 ± 81.7 89 99.7 ± 49.9 92.6

E, epithelial; Pl, pleomorphic; p27n, nuclear expression of p27; p27c, cytoplasmic expression of p27.* Numeric results expressed as percentages of immunoreactive tumor cells.† Numeric results expressed as microvessel counts.‡ Exact paired t test (Wilcoxon signed rank test).



FIGURE 3. Immunohistochemical evaluation of PCs for a variety of antigens involved in different aspects of tumor cell differen-tiation and growth. Cytokeratins were variably expressed in the sarcomatoid component, ranging from quite diffuse labeling (A)to occasional (B) or completely negative (C) immunoreactivity, whereas the epithelial component reacted strongly for cytokeratins(C). The opposite occurred for vimentin, with the sarcomatoid tumor cells being intensely reactive and the epithelial cellscompletely negative (D). FHIT was immunolocalized in epithelial tumor cells (E), whereas fascin reacted predominantly with thesarcomatoid elements (F). Similar results were achieved for both p27Kip1 (with either nuclear or cytoplasmic labeling) (G) andp21Waf1 (H) immunoreactivities: the epithelial components (on the left in G and at the bottom in H) were immunoreactive for bothcell cycle inhibitors, whereas the sarcomatoid cells appeared to be consistently negative. A few p27Kip1-immunoreactive lympho-cytes served as a helpful internal positive control (G).



Although PCs have been reported to run a more aggres-sive clinical course,5,8,19,20,28,35,37,45,51 stage I tumors in ourseries did not show a significantly worse prognosis than ordi-nary NSCLCs; about 50% of the patients died of disease after5 years, a value not different from that expected for patientswith stage I ordinary NSCLC.33 Our results are in keeping withthose of Nakajima et al, who failed to demonstrate any signifi-cant clinical difference of sarcomatoid carcinomas as com-

pared with conventional NSCLCs.27 In our study, however,tumors with >50% pleomorphic cell component were morelikely to belong to higher pathologic stages, suggesting that thepercentage of sarcomatoid cells may be relevant to prognosis.

TABLE 6. Prognostic Relevance of Different Gene ProductsInvolved in Cell Differentiation, Cell Cycle Control, ApoptosisRegulation, Tumor Growth, and Cell Motility

Log-rankTest

Overall survival Disease-free Survival

Epithelial Pleomorphic Epithelial Pleomorphic

CK 0.137 0.750 0.323 0.600VIM 0.848 0.788 0.659 0.390EMA 0.398 0.434 0.892 0.085CEA 0.529 0.339 0.827 0.408p53 0.375 0.623 0.947 0.657p21Waf1 0.977 0.981 0.862 0.441p27Kip1n 0.919 0.970 0.704 0.943p27Kip1c 0.500 0.684 0.295 0.363Ki67 0.040 0.077 0.011 0.034FHIT 0.777 0.749 0.475 0.304Fascin 0.887 0.143 0.992 0.348CD34 0.874 0.199 0.458 0.545

p27Kip1 n, nuclear expression of p27Kip1; p27Kip1 c, cytoplasmic expres-sion of p27Kip1.

FIGURE 4. Overall (A) and disease-free (B) survival curves ac-cording to the different histology (adenocarcinoma, epider-moid carcinoma, and PC) in stage I NSCLC. Although therewas a trend for stage I PC to behave more aggressively thanordinary NSCLC, differences were not statistically significant inboth overall (A) and disease-free (B) survival curves.

FIGURE 5. A model of tumor morphogenesis of PCs is pro-posed, whereby the pleomorphic features of tumor cellsclosely paralleled the differential expression of several geneproducts, revealing a more severe dysregulation of the cellcycle control, apoptosis, cell motility, and tumor growth inpleomorphic areas. In the transition from epithelial to pleo-morphic phenotype, tumor cells lost the expression of epithe-lial markers (cytokeratins EMA and CEA) and cell cycle inhibi-tors (p27Kip1, p21Waf1, and FHIT) and acquired the expressionof cytoskeletal proteins of mesenchymal type, such as vimentin(but occasionally also desmin and smooth muscle actin, notshown) and fascin, a cell motility-related protein. Also, micro-vascular density (grey circles) was significantly increased in thepleomorphic areas.



From a clinical viewpoint, the formal distinction of PCs fromcarcinosarcomas, according to the presence of true chondro-,osteo-, angio-, or rhabdomyosarcomatous components, mayalso be unwarranted because even carcinosarcomas do notshow a significantly different prognosis.19 Moreover, thethreshold of 10% for the sarcomatoid component, as proposedby WHO for separating PCs from ordinary NSCLCs, is instru-mental in excluding tumors with occasional sarcomatoid cellsbut is probably “per se” insufficient to identify lesions withdistinct clinicopathologic characteristics that may result in dif-ferent biologic behavior.

We also investigated whether the expression of differentgene products involved in cell differentiation, cell cycle con-trol, apoptosis regulation, tumor growth, and cell motilitycould have a prognostic significance in PCs of the lung. It iswell known that downregulation of cell cycle inhibitors and ofpromoters of the apoptosis, and an increased tumor vascular-ization are associated with the progression of several tumortypes.10,11,16,17,22,29,34 Moreover, the subcellular localizationof p27kip1, which has been reported to lead to protein inactiva-tion and degradation,42 and fascin overexpression could sig-nificantly contribute to the progression of PCs. We have re-cently found that fascin overexpression could be involved inthe metastatic spread of stage I NSCLC via either direct vas-cular permeation or the development of newly formed mi-crovessels (Pelosi et al., manuscript in preparation). However,in the current study, only a Ki-67 labeling index >35% wasfound to impact adversely on overall and disease-free survival,thus confirming the well-established prognostic role of Ki-67immunostaining in several pulmonary and non-pulmonaryneoplasms.3,12,25,30,31,33,39,42

In summary, this study supports the view that PCs, com-posed of a dual cell population of epithelial and sarcomatoidcells, represent a well-defined pathologic entity in the spec-trum of NSCLC. We could construct a model of tumor mor-phogenesis whereby the sarcomatoid phenotype is closely cor-related to specific changes in the expression of markers of celldifferentiation, cell cycle control and apoptosis, tumor growth,and cell motility. Whereas stage I PCs do not prognosticallydiffer from ordinary NSCLCs, the percentage of pleomorphiccells may have clinical relevance in more advanced-stage tu-mors.

AcknowledgmentThe authors thank Mr. Steven Lawson for the English

revision of the manuscript.

REFERENCES1. Attanoos RL, Papagiannis A, Suttinont P, et al. Pulmonary giant cell car-

cinoma: pathological entity or morphological phenotype? Histopathol-ogy. 1998;32:225–31.

2. Berho M, Moran CA, Suster S. Malignant mixed epithelial/mesenchymalneoplasms of the lung. Semin Diagn Pathol. 1995;12:123–39.

3. Brown DC, Gatter KC. Ki67: the immaculate deception? Histopathology.2002;40:2–11.

4. Chetty R. Cyclin E and p27/Kip1: two proteins on a see-saw or anotherfalse dawn? Histopathology. 2002;40:393–4.

5. Colby TV, Koss MN, Travis WD. Tumors of the Lower Respiratory Tract.Washington, DC: Armed Forces Institute of Pathology, 1995.

6. Dandachi N, Hauser-Kronberger C, More E, et al. Co-expression of tena-scin-C and vimentin in human breast cancer cells indicates phenotypictransdifferentiation during tumor progression: correlation with histopath-ological parameters, hormone receptors and oncoproteins. J Pathol. 2001;193:181–9.

7. Davis MP, Eagan RT, Weiland LH, et al. Carcinosarcoma of the lung:Mayo Clinic experience and response to chemotherapy. Mayo Clin Proc1984;59:598–603.

8. Fishback NF, Travis WD, Moran CA, et al. Pleomorphic (spindle/giantcell) carcinoma of the lung: a clinicopathologic correlation of 78 cases.Cancer. 1994;73:2936–45.

9. Fischer D, Tucker RP, Chiquet-Ehrismann R, et al. Cell-adhesive re-sponses to tenascin-C splice variants involve formation of fascin mi-crospikes. Mol Biol Cell. 1997;8:2055–75.

10. Fontanini G, Faviana P, Lucchi M, et al. A high vascular count and over-expression of vascular endothelial growth factor are associated with un-favourable prognosis in operated small cell lung carcinoma. Br J Cancer.2002;86:558–63.

11. Hasan J, Byers R, Jayson GC. Intra-tumoural microvessel density in hu-man solid tumours. Br J Cancer. 2002;86:1566–77.

12. Heslin MJ, Cordon-Cardo C, Lewis JJ, et al. Ki-67 detected by MIB-1predicts distant metastasis and tumor mortality in primary, high gradeextremity soft tissue sarcoma. Cancer. 1998;83:490–7.

13. Holst VA, Finkelstein S, Colby TV, et al. p53 and K-ras mutational geno-typing in pulmonary carcinosarcoma, spindle cell carcinoma, and pulmo-nary blastoma: implications for histogenesis. Am J Surg Pathol. 1997;21:801–11.

14. Hommura F, Dosaka-Akita H, Mishina T, et al. Prognostic significance ofp27KIP1 protein and ki-67 growth fraction in non-small cell lung cancers.Clin Cancer Res. 2000;6:4073–81.

15. Humphrey PA, Scroggs MW, Roggli VL, et al. Pulmonary carcinomaswith a sarcomatoid element: an immunocytochemical and ultrastructuralanalysis. Hum Pathol. 1988;19:155–65.

16. Inoue K, Kamada M, Slaton JW, et al. The prognostic value of angiogen-esis and metastasis-related genes for progression of transitional cell car-cinoma of the renal pelvis and ureter. Clin Cancer Res. 2002;8:1863–70.

17. Kawana H, Tamaru J, Tanaka T, et al. Role of p27 Kip1 and cyclin-dependent kinase 2 in the proliferation of non-small cell lung cancer. AmJ Pathol. 1998;153:505–13.

18. Khalifa M, Hruby G, Ehrlich L, et al. Combined large cell neuroendocrinecarcinoma and spindle cell carcinoma of the lung. Ann Diagn Pathol.2001;5:240–5.

19. Koss MN, Hochholzer L, Frommelt RA. Carcinosarcomas of the lung: aclinicopathologic study of 66 patients. Am J Surg Pathol. 1999;23:1514–26.

20. Krefting IP, Nunez LA, Sherer P, et al. Pleomorphic carcinoma (spindleand giant cell) of the lung. Md Med J. 1994;43:787–90.

21. Li X, Hui AM, Shi YZ, et al. Reduced p21 (WAF1/CIP1) expression is anearly event in gallbladder carcinogenesis and is of prognostic significancefor patients with carcinomas of the gallbladder. Hum Pathol. 2001;32:771–7.

22. Lloyd RV, Erickson LA, Jin L, et al. p27kip1: a multifunctional cyclin-dependent kinase inhibitor with prognostic significance in human can-cers. Am J Pathol. 1999;154:313–23.

23. Maekawa S, Ohta K, Sakai H. A novel 53 kDa actin binding protein fromporcine brain: further biochemical and immunological characterization.Cell Struct Funct. 1988;13:373–85.

24. Matsui K, Kitagawa M. Spindle cell carcinoma of the lung: a clinicopath-ologic study of three cases. Cancer. 1991;67:2361–7.

25. Molino A, Micciolo R, Turazza M, et al. Ki-67 immunostaining in 322primary breast cancers: associations with clinical and pathological vari-ables and prognosis. Int J Cancer. 1997;74:433–7.

26. Mountain CF. Revisions in the International System for Staging LungCancer. Chest. 1997;111:1710–7.



27. Nakajima M, Kasai T, Hashimoto H, et al. Sarcomatoid carcinoma of thelung: a clinicopathologic study of 37 cases. Cancer. 1999;86:608–16.

28. Nappi O, Glasner SD, Swanson PE, et al. Biphasic and monophasic sar-comatoid carcinomas of the lung: a reappraisal of ‘carcinosarcomas’ and‘spindle-cell carcinomas.’ Am J Clin Pathol. 1994;102:331–40.

29. Palazzo JP. Cyclin-dependent kinase inhibitors: a novel class of prognos-tic indicators. Hum Pathol. 2001;32:769–70.

30. Pence JC, Kerns BJ, Dodge RK, et al. Prognostic significance of the pro-liferation index in surgically resected non-small-cell lung cancer. ArchSurg. 1993;128:1382–90.

31. Pelosi G, Bresaola E, Bogina G, et al. Endocrine tumors of the pancreas.Ki-67 immunoreactivity on paraffin sections is an independent predictorfor malignancy: a comparative study with proliferating-cell nuclear anti-gen and progesterone receptor protein immunostaining, mitotic index, andother clinicopathologic variables. Hum Pathol. 1996;27:1124–34.

32. Pelosi G, Fraggetta F, Sonzogni A, et al. CD99 immunoreactivity in gas-trointestinal and pulmonary neuroendocrine tumors. Virchows Arch.2000;437:270–4.

33. Pelosi G, Fraggetta F, Pasini F, et al. Immunoreactivity for thyroid tran-scription factor-1 in stage I non-small cell carcinomas of the lung. Am JSurg Pathol. 2001;25:363–72.

34. Pezzella F, Harris AL, Gatter KC. Ways of escape: are all tumours angio-genic? Histopathology. 2001;39:551–3.

35. Przygodzki RM, Koss MN, Moran CA, et al. Pleomorphic (giant andspindle cell) carcinoma is genetically distinct from adenocarcinoma andsquamous cell carcinoma by K-ras-2 and p53 analysis. Am J Clin Pathol.1996;106:487–92.

36. Przygodzki RM, Koss MN, O’Leary TJ. Pleomorphic (giant and/orspindle cell) carcinoma of lung shows a high percentage of variantCYP1A12. Mol Diagn. 2001;6:109–15.

37. Ro JY, Chen JL, Lee JS, et al. Sarcomatoid carcinoma of the lung: immu-nohistochemical and ultrastructural studies of 14 cases. Cancer. 1992;69:376–86.

38. Roz L, Gramegna M, Ishii H, et al. Restoration of fragile histidine triad(FHIT) expression induces apoptosis and suppresses tumorigenicity inlung and cervical cancer cell lines. Proc Natl Acad Sci USA. 2002;99:3615–20.

39. Rudolph P, Olsson H, Bonatz G, et al. Correlation between p53, c-erbB-2,and topoisomerase II alpha expression, DNA ploidy, hormonal receptorstatus and proliferation in 356 node-negative breast carcinomas: prognos-tic implications. J Pathol. 1999;187:207–16.

40. Sard L, Accornero P, Tornielli S, et al. The tumor-suppressor gene FHITis involved in the regulation of apoptosis and in cell cycle control. ProcNatl Acad Sci USA. 1999;96:8489–92.

41. Sgambato A, Ratto C, Faraglia B, et al. Reduced expression and alteredsubcellular localization of the cyclin-dependent kinase inhibitorp27(Kip1) in human colon cancer. Mol Carcinog. 1999;26:172–9.

42. Shiba M, Kohno H, Kakizawa K, et al. Ki-67 immunostaining and otherprognostic factors including tobacco smoking in patients with resectednonsmall cell lung carcinoma. Cancer. 2000;89:1457–65.

43. Sozzi G, Veronese ML, Negrini M, et al. The FHIT gene 3p14.2 is abnor-mal in lung cancer. Cell. 1996;85:17–26.

44. Tao YS, Edwards RA, Tubb B, et al. �-Catenin associates with the actin-bundling protein fascin in a noncadherin complex. J Cell Biol. 1996;134:1271–81.

45. Terzi A, Gorla A, Piubello Q, et al. Biphasic sarcomatoid carcinoma of thelung: report of 5 cases and review of the literature. Eur J Surg Oncol.1997;23:457.

46. Tilney LG, Connelly PS, Vranich KA, et al. Why are two different cross-linkers necessary for actin bundle formation in vivo and what does eachcross-link contribute? J Cell Biol. 1998;143:121–33.

47. Thompson LD, Wieneke JA, Miettinen M, et al. Spindle cell (sarcoma-toid) carcinomas of the larynx: a clinicopathologic study of 187 cases. AmJ Surg Pathol. 2002;26:153–70.

48. Tseng Y, Fedorov E, McCaffery JM, et al. Micromechanics and ultra-structure of actin filament networks crosslinked by human fascin: a com-parison with alpha-actinin. J Mol Biol. 2001;310:351–66.

49. Travis WD, Colby TV, Corrin B, et al. Histological Typing of Lung andPleural Tumours. New York: Springer Verlag, 1999.

50. Veronesi G, Landoni C, Pelosi G, et al. Fluoro-deoxi-glucose uptake andangiogenesis are independent biological features in lung metastases. Br JCancer. 2002;86:1391–5.

51. Wick MR, Ritter JH, Humphrey PA. Sarcomatoid carcinomas of the lung:a clinicopathologic review. Am J Clin Pathol. 1997;108:40–53.

52. World Health Organization. The World Health Organization histologicaltyping of lung tumors. Am J Clin Pathol. 1982;77:123–36.

53. Xu X, Yin H, Meng K, et al. Pathological study of the pleomorphic car-cinoma of the lung. Zhonghua Bing Li Xue Za Zhi. 2000;29:424–7.

54. Yamashiro S, Yamakita Y, Ono S, et al. Fascin, an actin-bundling protein,induces membrane protrusions and increases cell motility of epithelialcells. Mol Biol Cell. 1998;9:993–1006.



Documents

Pleomorphic Carcinomas of the Lung Show a Selective ... up 4.pdf · Pleomorphic Carcinomas of the Lung Show a Selective Distribution of Gene Products Involved in Cell Differentiation,