Cyclooxygenase-2 expression in human pituitary tumors

Cyclooxygenase-2 Expression in Human PituitaryTumors

Sergio Vidal, Ph.D.1,2

Kalman Kovacs, M.D., Ph.D.1

David Bell, M.D.1

Eva Horvath, Ph.D.1

Bernd W. Scheithauer, M.D.3

Ricardo V. Lloyd, M.D., Ph.D.3

1 Department of Laboratory Medicine and Patho-biology, St. Michaels Hospital, University of To-ronto, Toronto, Ontario, Canada.

2 Department of Anatomy, University of Santiagode Compostela, Lugo, Spain.

3 Department of Laboratory Medicine and Pathol-ogy, Mayo Clinic, Rochester, Minnesota.

Supported in part by a grant from the Ministerio deCiencia y Tecnologı́a Direcciı́n General de Investi-gaciı́n (BFI2001-3336-C02-02) and by a generousdonation from Mr. and Mrs. Jarislowski and theLloyd Carr-Harris Foundation. Dr. Sergio Vidal wassupported by a research grant from University ofSantiago de Compostela, Spain.

The authors thank Mr. Fabio Rotondo for technicalassistance and the staff of St. Michael’s HospitalHealth Sciences Library for their contribution to thestudy.

Address for reprints: Bernd W. Scheithauer,M.D., Department of Laboratory Medicine andPathology, Mayo Clinic, 200 1st Street SW,Rochester, MN, 55905; Fax: (507) 284-1599;E-mail: [email protected]

Received November 20, 2002; revision receivedJanuary 24, 2003; accepted February 11, 2003.

BACKGROUND. Cyclooxygenase-2 (COX-2) plays a role in progression of colon,

breast, pancreas, and lung carcinomas. The authors investigated COX-2 expression

in pituitary tumors.

METHODS. Expression of COX-2 was evaluated in 164 surgically removed human

pituitary tumors. Correlation of COX-2 with MIB-1, a cell proliferation marker, as

well as angiogenesis, patient age, gender, tumor type, size, invasiveness, and

metastatic potential was investigated.

RESULTS. Cyclooxygenase-2 immunoreactivity was confined to the cytoplasm of

tumor cells, whereas the nuclei were unlabeled. Few normal peritumoral adeno-

hypophysial cells showed slight COX-2 cytoplasmic immunoreactivity. The stain-

ing intensity and the percentage of immunopositive cells were higher in tumors.

Most pituitary tumors (96%) were COX-2–immunopositive. Expression was strong

in 60 (44%), moderate in 39 (28%), and weak in 32 (24%). Male gonadotroph

adenomas and null cell adenomas showed a high level of COX-2 expression.

Growth hormone-producing adenomas, prolactin-producing adenomas, thyro-

tropic hormone-producing adenomas, female gonadotroph adenomas, silent ad-

renocorticotropic hormone-producing adenomas, and silent subtype 3 adenomas

had a low level of COX-2 expression. Significant correlation was demonstrated with

patient age, but not with tumor size, invasiveness, and MIB-1 labeling indices.

Expression was medium to high in 76% of macroadenomas and in only 45% of

microadenomas. Strong correlations were noted with angiogenesis markers, such

as microvessel density and surface density.

CONCLUSIONS. Correlation with angiogenesis suggests that COX-2 may be involved

in the regulation of angiogenesis in pituitary tumors. Phamacologic inhibition of

COX-2 activity might suppress angiogenesis in pituitary tumors and may provide a

novel approach for medical therapy. Cancer 2003;97:2814 –21.

© 2003 American Cancer Society.

DOI 10.1002/cncr.11387

KEYWORDS: pituitary tumors, cyclooxygenase-2, immunohistochemistry, angiogen-esis, cell proliferation, hypoxia-inducible factor-1�.

Prostaglandins are small lipid molecules that act locally and areinvolved in various functions, such as the regulation of renal

blood flow, platelet aggregation, neurotransmitter release, immunemodulation, and cytoprotection of the gastric mucosa.1 They mayalso play a role in the development and progression of various humantumors and experimental neoplasms.2 Prostaglandin synthesis beginswith the formation of arachidonic acid, which is converted to PgH2 bythe cyclooxygenase enzymes, COX-1 and COX-2, also known as pros-taglandin endoperoxide synthases, and subsequently is catalyzed tovarious prostaglandins.3 The COX-1 enzyme is expressed constitu-tively in most mammalian tissues where, although often undetected,

2814

© 2003 American Cancer Society

it can be induced rapidly. The COX-2 gene is an early-response gene that is up-regulated by proinflamma-tory stimuli, growth factors, oncogenes, carcinogens,and tumor-promoting phorbol esters.4,5 Previousstudies have suggested that COX-2 plays a role incarcinogenesis. Several epidemiologic investigationshave reported a lower than expected rate of colorectaladenomas and carcinomas in persons taking nonste-roidal antiinflammatory drugs (NSAIDs), inhibitors ofcyclooxygenase activity.6,7 In addition, COX-2 is over-expressed in various tumors, whereas COX-2 inhibi-tors suppress tumor formation in experimental ani-mals.8 –13 COX-2 knockout mice have a lowerincidence of several tumor types, including skin pap-illomas and intestinal polyps.14 –16 It appears thatCOX-2 expression also regulates tumor behavior. Forexample, by activating metalloproteinases, COX-2overexpression confers invasiveness upon tumors.Overexpression of COX-2 also increases production ofproangiogenic factors and decreases E-cadherin, a celladhesion molecule indispensable for tumor growth,invasion, and metastasis.17–21 We evaluated whetherCOX-2 expression occurs in adenohypophysial tumorsand assessed its utility as a marker of tumor behavior.Pituitary tumors are heterogeneous, exhibiting a mul-tifactorial etiology and pathogenesis. Although mostare slow growing and metastases are exceptionallyrare, a significant proportion of adenomas invade ad-jacent structures.22,23 Because the histologic appear-ance of pituitary tumors is a poor predictor of aggres-sive behavior,24 a number of studies have assessed cellproliferation and angiogenesis markers as predictorsof tumor prognosis.25–29 To obtain deeper insight intothe role of COX-2 in pituitary tumor pathobiology, weanalyzed the correlation between COX-2 expressionand immunolabeling for MIB-1, a cell proliferationmarker, as well as microvessel density (MVD), mi-crovessel surface density (MSD), and expression ofhypoxia-inducible factor-1� (HIF-1�).

MATERIALS AND METHODSMaterialsTissue specimens (164 adenohypophysial tumors),which were obtained by transsphenoidal surgery, wereselected from the Mayo Clinic Tissue Registry (Roch-ester, NY) and from the consultation files of 3 of theauthors (B. W. S., K. K., E. H.). Seventy-five patients(45.7%) were men (mean age, 52.9 years; range, 11– 81years) and 89 (54.3%) were women (mean age, 45.7years; range, 17– 80 years). Each tumor was investi-gated and classified according to the criteria of theWorld Health Organization.22 Specimens included 10growth hormone (GH)-producing adenomas, 37 pro-lactin (PRL)-producing adenomas, 19 adrenocortico-

tropic hormone (ACTH)-producing adenomas, 10 thy-rotropic hormone (TSH)-producing adenomas, and 61clinically nonfunctioning adenomas (including 8 si-lent ACTH cell adenomas of Subtype 1, 13 silent Sub-type 3 adenomas, 10 female and 10 male gonadotrophadenomas, 10 nononcocytic adenomas, and 10 onco-cytic null cell adenomas). In addition, we studied 10bromocriptine-treated, PRL-producing adenomas and10 octreotide-treated, GH-producing adenomas. Fi-nally, seven pituitary carcinomas (four produced byPRL and three by ACTH) also were included in thestudy.

Tumor size, invasiveness, proliferative activity,and vascularity were evaluated in each case. Size andinvasiveness were assessed on the basis of preopera-tive magnetic resonance imaging scans and on surgi-cal findings. Tumors were divided into microadeno-mas or macroadenomas (defined as tumors � 1 cmand � 1 cm in diameter, respectively).

Morphologic StudiesEach tumor specimen was stained with hematoxylinand eosin, periodic acid–Schiff, and (in some cases)the Gordon–Sweet silver method for reticulin fibers.For immunohistochemistry, the labeled streptavidin-biotin peroxidase complex method was used as wereantibodies to the full spectrum of adenohypophysialhormones, including GH, PRL, ACTH, luteinizing hor-mone, follicle-stimulating hormone, TSH, and the al-pha subunit of glycoprotein hormones. The sources,dilutions, and clonality of the antibodies used, as wellas the control methods, are described elsewhere.30,31

Immunostaining for MIB-1 (Immunotech, West-brook, ME), HIF-1� (Novus Biologicals, Littleton, CO),and CD34 (Dako, Glostrup, Denmark), which specifi-cally detects endothelial cells, was also performed. Foreach case, MIB-1 and HIF-1� labeling indexes weredetermined based on the number of positively stainednuclei divided by the total number of nuclei counted.A mean of 30 fields, each containing approximately100 cells, was assessed in all cases. CD34-immuno-stained sections were evaluated using a computeranalysis system (Microimage, Media Cybernetics, MD)to determine MVD (the percentage of pituitary tissueoccupied by vessels) and MSD (the percentage of thevessel circumference in direct contrast with pituitarytissue). Details of immunohistochemistry, includingthe duration of exposure and control procedures forthese three stains, as well as the morphometric quan-tification method have been described in previouspublications.25,26,32,33 Most tumors (n � 119) werefixed in glutaraldehyde, routinely processed, embed-ded in Epon-Araldite, and studied by transmissionelectron microscopy.

COX-2 in Pituitary Tumors/Vidal et al. 2815

Cyclooxygenase-2 expression was also deter-mined immunohistochemically. Staining was per-formed using a rabbit polyclonal antibody directedagainst a peptide corresponding to amino acids 567–599 of human COX-2 (Cayman Chemical, Ann Arbor,MI). Preliminary titration experiments showed the op-timal working dilution to be 1:200. After routinedeparaffinization, rehydration, and blocking of endog-enous peroxidase activity, 5 �m thick sections weresubject to antigen retrieval by microwave treatment in0.1 mM sodium citrate buffer, pH 6.0, as previouslydescribed.34 Subsequently, the treated slides were in-cubated overnight at 37 °C with the primary antiserumand exposed to the streptavidin-biotin peroxidasecomplex. Diaminobenzidine served as the chromogen.Positive control sections of human colon carcinomatissue were included in each batch. To confirm thespecificity of the primary antibody, control tests in-cluded substitution of the primary antiserum withphosphate-buffered saline, as well as preabsorption ofthe COX-2 antibody with homologous antigen excess.Preabsorption was carried out on paraffin sections.One milliliter of COX-2 antibody (working dilution)was incubated with 10 �g of COX-2 (Cayman Chemi-cal) for a period of 24 hours at 4 °C. Immunolabelingwas abolished totally in both controls.

The COX-2–immunostained sections were evalu-ated by computer image analysis (Microimage, MediaCybernetics, Silver Spring, MD). All quantitative eval-uations were performed blindly by one of the authors(S. V.). The reaction was quantified in 30 randomlyselected fields (6.7 mm2) and expressed as the per-centage of tumor tissue occupied by nucleated COX-2immunoreactive cells. Staining was not interpreted inareas of necrosis, fibrosis, or tissue artefact. The extentof COX-2 reactivity was recorded in four categoriesbased on the percentage of tumor cells stained. Re-sults were immunonegative when no immunopositivecells were detected, low when less than 25% of tumorcells were immunopositive, moderate when 25–75% ofcells were reactive, and high when more than 75% ofcells were stained. Fresh frozen tissue specimens werenot available for Western blotting.

Data were tested for statistical significance usingthe SPSS statistical computer program (SPSS, Chicago,IL). Because assumptions for a parametric test werenot valid (Kolmogorov–Smirnov, P � 0.05), all datawere evaluated by Kruskall–Wallis analysis of varianceand the Mann–Whitney U test as a multiple compar-ison method. The Spearman test was used to assessthe statistical significance of the correlation amongCOX-2 expression, patient age, tumor vascularity,HIF-1� expression, and MIB-1 labeling indices. Only

differences for which P � 0.05 were considered to bestatistically significant.

RESULTSImmunohistochemical examination showed that bothMIB-1 and HIF-1� reactivity were confined to the nu-clei of neoplastic cells, whereas nontumorous adeno-hypophysis showed no staining. In both the nontu-morous adenohypophyses and in tumors, the vascularendothelium was immunoreactive for CD34 alone.The immunopositive cells were identified easily andquantified.

COX-2 immunoreactivity was confined to the cy-toplasm of tumor cells whereas the nuclei were unla-beled (Fig. 1). Although the staining intensity and thepercentage of immunopositive cells were higher intumor, the distribution of COX-2 reactivity was notrestricted to tumor cells. Less than 0.2% of normaladenohypophysial cells showed COX-2 cytoplasmicimmunoreactivity (Fig. 2). Although in most tumors,COX-2 immunopositive cells were distributed ran-domly, expression was restricted to several regions ofthe tumor in some cases. The number of positive cellsvaried among the different tumor types, ranging from0% to 100% (mean, 58.23 � 3.08%; Figs. 3, 4). Insummary, 130 of 135 untreated pituitary tumors (96%)showed COX-2 expression. Among these tumors, ex-pression was strong in 60 (44%), moderate in 39 (28%),and weak in 32 (24%). No COX-2 immunopositivitywas seen in 5 tumors (4%), including 1 GH-producingadenoma, 1 PRL-producing adenoma, and 1 TSH-pro-ducing adenoma.

Among the various tumor types, significant differ-ences were noted in the percentages of cells express-ing COX-2 (Fig. 5). Three types of clinically nonfunc-tioning adenomas (i.e., male gonadotroph adenomasand both nononcocytic and oncocytic null cell adeno-mas) showed the highest level of COX-2 expression. Incontrast, GH-producing adenomas, PRL-producingadenomas, TSH-producing adenomas, female gona-dotroph adenomas, silent ACTH adenomas of Subtype1, and silent Subtype 3 adenomas had the lowestCOX-2 expression levels. In contrast with the previ-ously mentioned tumor types, ACTH-producing ade-nomas and carcinomas showed an intermediate levelof COX-2 expression. In the current study, no signifi-cant differences were observed in the COX-2 expres-sion between PRL and ACTH-producing pituitary car-cinomas compared with their correspondingadenomas.

Considering all types of pituitary tumors, adeno-mas and carcinomas, our statistical study demon-strated a strong significant and positive correlationbetween COX-2 expression and patient age (r � 0.46;

2816 CANCER June 1, 2003 / Volume 97 / Number 11

P � 0.0001). In contrast, no significant differenceswere noted in the percentage of COX-2 immunoposi-tive cells between tumors of male and female patients(P � 0.09), despite the fact that higher counts weredetected in males (64.37 � 4.58%) compared withfemales (54.25 � 4.33%). In addition, we found thatCOX-2 expression was not correlated with pituitary

tumor size (P � 0.35) or with invasiveness (P � 0.91).However, it should be noted that COX-2 expressionwas medium to high in 76% of macroadenomas, butonly in 45% of microadenomas.

With respect to other variables, the Spearman testdid not show a significant correlation between COX-2expression and MIB-1 labeling indices (r � 0.07; P� 0.47) but showed strong correlations between

FIGURE 1. Oncocytic null cell adenoma showing intense immunopositivity for

cyclooxygenase-2 (COX-2). The wall of blood vessels inside the tumor is

immunonegative for COX-2 (star). Scale bar � 20 �m.

FIGURE 2. In the nontumorous area surrounding the tumor, few adenohy-

pophysial cells show cyclooxygenase-2 immunopositivity (arrows). Scale bar

� 20 �m.

FIGURE 3. Male gonadotroph adenoma showing intense immunopositivity for

cyclooxygenase-2. Immunoreactive cells are numerous and are distributed

randomly throughout the tumor. Scale bar � 20 �m.

FIGURE 4. Immunohistochemical staining of an octreotide-treated, growth

hormone–producing adenoma for cyclooxygenase-2. Only a few immunoposi-

tive cells are noted. Scale bar � 20 �m.


COX-2 expression and a number of markers of angio-genesis. The correlation was positive with regard toMVD (r � 0.41; P � 0.0001) and MSD (r � 0.33; P� 0.001), but was negative with respect to HIF-1�

expression (r � � 0.24; P � 0.03).No statistical differences in COX-2 expression

were found between octreotide-treated and untreatedGH adenomas and between bromocriptine-treatedand untreated PRL adenomas. However, in both cases,the percentage of immunonegative tumors increased10 –30% in octreotide-treated GH adenomas and5–20% in bromocriptine-treated PRL-producing ade-nomas (Fig. 6).

DISCUSSIONThe impact of ablating COX-2 activity at the tissuelevel to prevent tumor development or to control tu-mor progression has been reported in relation to dif-ferent neoplasms. There is convincing evidence thatNSAIDs, which are inhibitors of cyclooxygenase activ-ity, lower the risk of developing carcinoma of thecolon,6,35,36 esophagus,9,11,37 and stomach.10,38 Al-though the potential therapeutic efficacy of NSAIDswas related exclusively to tumors located in the diges-tive tract, some studies have suggested that COX-2may be the target for the prevention and treatment ofother carcinomas, including those of the prostate,8,39

breast,13,40,41pancreas,12,42 head and neck,43 and theurinary bladder.44,45 To our knowledge, no epidemio-logic studies have been performed to determine theprotective effects of COX-2 inhibitors on the develop-ment and progression of pituitary tumors. Our obser-vations of higher COX-2 expression in pituitary tumor

specimens compared with normal tissue specimensare consistent with previous findings reporting COX-2overexpression in tumor specimens. Pharmacologicinhibition of COX-2 activity had proven effective inreducing the incidence and growth of pituitary tu-mors. The exact mechanism by which COX-2 overex-pression promotes tumor cell proliferation is un-clear.21 One possibility is that overexpression of COX-2increases tumoral levels of prostaglandins, therebypromoting tumor progression by acting on differenti-ation and growth factors. Consistent with this hypoth-esis, it has been reported that prostaglandins increasetumor cell proliferation and prevent apoptosis bystimulating the expression of Bcl2, an antiapoptoticprotein.5 Conversely, selective inhibition of COX-2 inesophageal carcinoma cells suppresses prostaglandinsynthesis with a resultant increase in apoptotic celldeath and reduced proliferative activity.46 The currentstudy did not support these conclusions. For example,we did not find a significant correlation betweenCOX-2 expression and proliferation of pituitary tumorcells. It is noteworthy, however, that COX-2 expressionwas greater in macroadenomas than in microadeno-mas, a finding that concurs with previous reports oncolorectal carcinoma. There is no ready explanationfor these conflicting results. Yang et al.47 found thatwith familial adenomatous polyposis, elevation of tu-moral prostaglandins was not seen until the lesionreached 6 –7 mm in diameter. It is conceivable that theinnately slow pace of pituitary tumor growth and me-tastasis underlies the noted discrepancies betweenCOX-2 expression and the growth rate of pituitarytumors. Although previous studies have shown a cor-relation between cell proliferation marker expression

FIGURE 5. Cyclooxygenase-2 (COX-2) labeling index in pituitary tumors. The

data are expressed as percentage of COX-2 immunopositive cells and repre-

sent the mean � the standard error of the mean. Values with no letters in

common are significantly different (P � 0.05 according to statistical analysis

with the Kruskall–Wallis analysis of variance and the Mann–Whitney U test).

GH: growth hormone–producing; PRL: prolactin-producing; TSH: thyrotropic

hormone–producing; ACTH: adrenocorticopic hormone–producing.

FIGURE 6. Cyclooxygenase-2 (COX-2) labeling index in treated and untreated

pituitary adenomas. Each bar denotes the percentage of cases that had

immunonegative (0%), low (� 25%), moderate (25–75%), and high (� 75%)

COX-2 expression in each tumor type. GH: growth hormone–producing; PRL:

prolactin-producing. Strong; �, Moderate; Weak; �, Immunonegative.


and tumor aggressiveness, pituitary tumors generallygrow slowly.26,48 The lack of significant correlationbetween COX-2 expression and the presence of distantmetastasis in pituitary tumors is not surprising. Simi-lar results were reported with respect to colorectalcarcinoma.19 Another study indicated that COX-2 ex-pression is associated positively with metastatic po-tential.18 The role of COX-2 in promoting metastasisrelates to its activity as a tumor promoter by its stim-ulation of angiogenesis and metalloproteinase activ-ity.19,20 Although pituitary carcinomas are more vas-cularized than adenomas,25 the occurrence ofcerebrospinal metastases in these rare tumors sug-gests that hematogenous spread is not the exclusivedomain of metastasis. This may explain partially arecent report that concluded that metalloproteinaseactivity in pituitary tumors does not correlate withmetastatic potential.49

In the current study, a strong correlation wasfound between COX-2 expression and vascularizationin pituitary tumors. This result is in agreement withstudies of colorectal carcinomas that show a signifi-cant correlation between COX-2 staining and tumorMVD.19 Overexpression of COX-2 in human colon car-cinoma cells results in angiogenesis by inducing an-giogeneic factors, such as vascular endothelial growthfactor (VEGF) and basic fibroblast growth factor.19,50

The antineoplastic effect of COX-2 inhibitors relatesdirectly to the ability of these agents to suppress bloodvessel formation.51,52 One study showed a correlationbetween the lack of significant angiogenesis in pitu-itary tumors and VEGF expression.53 Collectively,these results suggest that COX-2 that regulates VEGFexpression and angiogenesis in pituitary tumors. Theprecise mechanism by which COX-2 induces tumorcells to produce angiogenic factors may explain whypituitary tumors are so hypovascular. In vitro studiessuggested that the COX-2 gene has a hypoxic respon-sive element and that hypoxia induces the expressionof COX-2 mRNA and its protein in human vascularendothelial cells. Liu et al.54 proposed that inhibitorsof the COX enzymes prevent angiogenesis by stabiliz-ing HIF-1� accumulation under hypoxic conditions.54

This proposed mechanism is not supported by ourfindings, which showed a negative statistical correla-tion between COX-2 expression and HIF-1�. In nor-moxia, the HIF-1� protein is maintained at low andoften undetectable levels and is overexpressed underhypoxic conditions.55 Our results suggest that adeno-hypophysial tumor cells submitted to hypoxia have adiminished capacity to express COX-2. The lack of anassociation between HIF-1� expression and the vari-ous factors involved in the regulation of angiogenesisin pituitary tumors is not surprising. A correlation has

not been demonstrated between HIF-1� and MVD inpituitary tumors.33 Given the expression of COX-2 inpituitary tumors, pharmacologic inhibition of COX-2may not only decrease the risk of pituitary tumorformation, but may be an effective treatment. Consis-tent with this notion, our findings showed an increasein the percentage of COX-2 immunonegative tumorswhen comparing octreotide-treated, GH-producingadenomas or bromocriptine-treated, PRL-producingadenomas with untreated tumors. This may be of par-ticular importance in clinically nonfunctioning pitu-itary tumors, precisely the group showing the highestlevels of COX-2 expression and for which no medicaltherapy is currently available.

REFERENCES1. Crofford LJ. Prostaglandin biology. Gastroenterol Clin North

Am. 2001;30:863– 876.2. Bishop-Bailey D, Calatayud S, Warner TD, Hla T, Mitchell JA.

Prostaglandins and the regulation of tumor growth. J Envi-ron Pathol Toxicol Oncol. 2002;21:93–101.

3. Subbaramaiah K, Zakim D, Weksler BB, Dannenberg AJ. In-hibition of cyclooxygenase: a novel approach to cancer pre-vention. Proc Soc Exp Biol Med. 1997;216:201–210.

4. Kelley DJ, Mestre JR, Subbaramaiah K, et al. Benzo[a]pyreneup-regulates cyclooxygenase-2 gene expression in oral epi-thelial cells. Carcinogenesis. 1997;18:795–799.

5. Fosslien E. Molecular pathology of cyclooxygenase-2 in neo-plasia. Ann Clin Lab Sci. 2000;30:3–21.

6. Xu XC. COX-2 inhibitors in cancer treatment and preven-tion, a recent development. Anticancer Drugs. 2002;13:127–137.

7. DuBois RN. Review article: cyclooxygenase—a target for co-lon cancer prevention. Aliment Pharmacol Ther. 2000;14(Suppl 1):64 – 67.

8. Fujita H, Koshida K, Keller ET, et al. Cyclooxygenase-2 pro-motes prostate cancer progression. Prostate. 2002;53:232–240.

9. Buskens CJ, Van Rees BP, Sivula A, et al. Prognostic signifi-cance of elevated cyclooxygenase 2 expression in patientswith adenocarcinoma of the esophagus. Gastroenterology.2002;122:1800 –1807.

10. Van Rees BP, Saukkonen K, Ristimaki A, et al. Cyclooxygen-ase-2 expression during carcinogenesis in the human stom-ach. J Pathol. 2002;196:171–179.

11. Shamma A, Yamamoto H, Doki Y, et al. Up-regulation ofcyclooxygenase-2 in squamous carcinogenesis of the esoph-agus. Clin Cancer Res. 2000;6:1229 –1238.

12. Kokawa A, Kondo H, Gotoda T, et al. Increased expression ofcyclooxygenase-2 in human pancreatic neoplasms and po-tential for chemoprevention by cyclooxygenase inhibitors.Cancer. 2001;91:333–338.

13. Ristimaki A, Sivula A, Lundin J, et al. Prognostic significanceof elevated cyclooxygenase-2 expression in breast cancer.Cancer Res. 2002;62:632– 635.

14. Oshima M, Dinchuk JE, Kargman SL, et al. Suppression ofintestinal polyposis in Apc delta716 knockout mice by inhi-bition of cyclooxygenase 2 (COX-2). Cell. 1996;87:803– 809.


15. Tiano H, Chulada P, Spalding J, et al. Effects of cyclooxy-genase deficiency on inflammation and papilloma develop-ment in mouse skin. Proc Am Assoc Cancer Res. 1997;38:1727.

16. Boolbol SK, Dannenberg AJ, Chadburn A, et al. Cyclooxy-genase-2 overexpression and tumor formation are blockedby sulindac in a murine model of familial adenomatouspolyposis. Cancer Res. 1996;56:2556 –2560.

17. Tsujii M, DuBois RN. Alterations in cellular adhesion andapoptosis in epithelial cells overexpressing prostaglandinendoperoxide synthase 2. Cell. 1995;83:493–501.

18. Tsujii M, Kawano S, DuBois RN. Cyclooxygenase-2 expres-sion in human colon cancer cells increases metastatic po-tential. Proc Natl Acad Sci U S A. 1997;94:3336 –3340.

19. Cianchi F, Cortesini C, Bechi P, et al. Up-regulation of cy-clooxygenase 2 gene expression correlates with tumor an-giogenesis in human colorectal cancer. Gastroenterology.2001;121:1339 –1347.

20. Costa C, Soares R, Reis-Filho JS, Leitao D, Amendoeira I,Schmitt FC. Cyclo-oxygenase 2 expression is associated withangiogenesis and lymph node metastasis in human breastcancer. J Clin Pathol. 2002;55:429 – 434.

21. Cao Y, Prescott SM. Many actions of cyclooxygenase-2 incellular dynamics and in cancer. J Cell Physiol. 2002;190:279 –286.

22. Kovacs K, Scheithauer BW, Horvath E, Lloyd RV. The WorldHealth Organization classification of adenohypophysialneoplasms. A proposed five-tier scheme. Cancer. 1996;78:502–510.

23. Kovacs K, Horvath E, Vidal S. Classification of pituitaryadenomas. J Neurooncol. 2001;54:121–127.

24. Pernicone PJ, Scheithauer BW, Sebo TJ, et al. Pituitary car-cinoma: a clinicopathologic study of 15 cases. Cancer. 1997;79:804 – 812.

25. Vidal S, Kovacs K, Horvath E, Scheithauer BW, Kuroki T,Lloyd RV. Microvessel density in pituitary adenomas andcarcinomas. Virchows Arch. 2001;438:595– 602.

26. Thapar K, Kovacs K, Scheithauer BW, et al. Proliferativeactivity and invasiveness among pituitary adenomas andcarcinomas: an analysis using the MIB-1 antibody. Neuro-surgery. 1996;38:99 –106.

27. Turner HE, Nagy Z, Gatter KC, Esiri MM, Harris AL, Wass JA.Angiogenesis in pituitary adeno. J Endocrinol. 2000;165:475–481.

28. Turner HE, Nagy Z, Gatter KC, Esiri MM, Wass JA, Harris AL.Proliferation, bcl-2 expression and angiogenesis in pituitaryadenomas: relationship to tumour behaviour. Br J Cancer.2000;82:1441–1445.

29. Vidal S, Kovacs K, Horvath E, et al. Topoisomerase II alphaexpression in pituitary adenomas and carcinomas: relation-ship to tumor behavior. Mod Pathol. 2002;15:1205–1212.

30. Kovacs K, Lloyd R, Horvath E, et al. Silent somatotrophadenomas of the human pituitary. A morphologic study ofthree cases including immunocytochemistry, electron mi-croscopy, in vitro examination, and in situ hybridization.Am J Pathol. 1989;134:345–353.

31. Kovacs K, Stefaneanu L, Horvath E, et al. Effect of dopamineagonist medication on prolactin producing pituitary adeno-mas. A morphological study including immunocytochemis-try, electron microscopy and in situ hybridization. VirchowsArch A Pathol Anat Histopathol. 1991;418:439 – 446.

32. Vidal S, Scheithauer BW, Kovacs K. Vascularity in nontu-morous human pituitaries and incidental microadenomas: amorphometric study. Endocr Pathol. 2000;11:215–227.

33. Vidal S, Horvath E, Kovacs K, Kuroki T, Lloyd RV,Scheithauer BW. Expression of hypoxia inducible factor-1�

(HIF-1�) in pituitary tumors. Histol Histopathol. In press.34. Vidal S, Roman A, Oliveira MC, De La Cruz LF, Moya L.

Simultaneous localization of Pit-1 protein and gonadotro-pins on the same cell type in the anterior pituitary glands ofthe rat. Histochem Cell Biol. 1998;110:183–188.

35. Saha D, Roman C, Beauchamp RD. New strategies for colo-rectal cancer prevention and treatment. World J Surg. 2002;26:762–766.

36. Gupta RA, DuBois RN. Colorectal cancer prevention andtreatment by inhibition of cyclooxygenase-2. Nat Rev Can-cer. 2001;1:11–21.

37. Buttar NS, Wang KK, Leontovich O, et al. Chemopreventionof esophageal adenocarcinoma by COX-2 inhibitors in ananimal model of Barrett’s esophagus. Gastroenterology.2002;122:1101–1112.

38. Lim HY, Joo HJ, Choi JH, et al. Increased expression ofcyclooxygenase-2 protein in human gastric carcinoma. ClinCancer Res. 2000;6:519 –525.

39. Kirschenbaum A, Liu X, Yao S, Levine AC. The role ofcyclooxygenase-2 in prostate cancer. Urology. 2001;58:127–131.

40. Badawi AF, Badr MZ. Chemoprevention of breast cancer bytargeting cyclooxygenase-2 and peroxisome proliferator-ac-tivated receptor-gamma. Int J Oncol. 2002;20:1109 –1122.

41. Moran EM. Epidemiological and clinical aspects of nonste-roidal anti-inflammatory drugs and cancer risks. J EnvironPathol Toxicol Oncol. 2002;21:193–201.

42. Ding XZ, Tong WG, Adrian TE. Cyclooxygenases and lipoxy-genases as potential targets for treatment of pancreatic can-cer. Pancreatology. 2001;1:291–299.

43. Dannenberg AJ, Altorki NK, Boyle JO, Lin DT, SubbaramaiahK. Inhibition of cyclooxygenase-2: an approach to prevent-ing cancer of the upper aerodigestive tract. Ann N Y AcadSci. 2001;952:109 –115.

44. Shirahama T, Arima J, Akiba S, Sakakura C. Relation be-tween cyclooxygenase-2 expression and tumor invasivenessand patient survival in transitional cell carcinoma of theurinary bladder. Cancer. 2001;92:188 –193.

45. Ristimaki A, Nieminen O, Saukkonen K, Hotakainen K, Nord-ling S, Haglund C. Expression of cyclooxygenase-2 in humantransitional cell carcinoma of the urinary bladder. Am JPathol. 2001;158:849 – 853.

46. Souza RF, Shewmake K, Beer DG, Cryer B, Spechler SJ.Selective inhibition of cyclooxygenase-2 suppresses growthand induces apoptosis in human esophageal adenocarci-noma cells. Cancer Res. 2000;60:5767–5772.

47. Yang VW, Shields JM, Hamilton SR, et al. Size-dependentincrease in prostanoid levels in adenomas of patients withfamilial adenomatous polyposis. Cancer Res. 1998;58:1750 –1753.

48. Turner HE, Wass JA. Are markers of proliferation valuable inthe histological assessment of pituitary tumours? Pituitary.1999;1:147–151.

49. Turner HE, Nagy Z, Esiri MM, Harris AL, Wass JA. Role ofmatrix metalloproteinase 9 in pituitary tumor behavior.J Clin Endocrinol Metab. 2000;85:2931–2935.

50. Gallo O, Franchi A, Magnelli L, et al. Cyclooxygenase-2 path-way correlates with VEGF expression in head and neckcancer. Implications for tumor angiogenesis and metastasis.Neoplasia. 2001;3:53– 61.


51. Dermond O, Ruegg C. Inhibition of tumor angiogenesis bynon-steroidal anti-inflammatory drugs: emerging mecha-nisms and therapeutic perspectives. Drug Resist Updat.2001;4:314 –321.

52. Masferrer JL, Leahy KM, Koki AT, et al. Antiangiogenic andantitumor activities of cyclooxygenase-2 inhibitors. CancerRes. 2000;60:1306 –1311.

53. Lloyd RV, Scheithauer BW, Kuroki T, Vidal S, Kovacs K,Stefaneanu L. Vascular endothelial growth factor (VEGF)

expression in human pituitary adenomas and carcinomas.Endocr Pathol. 1999;10:229 –235.

54. Liu XH, Kirschenbaum A, Lu M, et al. Prostaglandin E2induces hypoxia-inducible factor-1� stabilization and nu-clear localization in a prostate cancer cell line. J Biol Chem.55. Wang GL, Jiang BH, Rue EA, Semenza GL. Hypoxia-inducible factor 1 is a basic-helix-loop-helix-PAS het-erodimer regulated by cellular O2 tension. Proc Natl Acad SciUSA. 1995;92:5510 –5514.


Documents

Cyclooxygenase-2 expression in human pituitary tumors