Expression of Class I HistoneDeacetylases Indicates PoorPrognosis in EndometrioidSubtypes of Ovarian andEndometrial Carcinomas1

Wilko Weichert*, Carsten Denkert*, Aurelia Noske*,Silvia Darb-Esfahani*, Manfred Dietel*,Steve E. Kalloger†, David G. Huntsman†

and Martin Köbel†

*Institute of Pathology, Charité Universitätsmedizin, Berlin,Germany; †Genetic Pathology Evaluation Centre of theProstate Research Centre, Department of Pathology,Vancouver General Hospital and British Columbia CancerAgency, Vancouver, BC, Canada

AbstractHistone deacetylase (HDAC) inhibitors are an emerging class of targeted cancer therapeutics, and little is knownabout HDAC expression in gynecologic malignancies. Therefore, we tested the hypothesis whether high-level expres-sion of class 1 HDACs (HDAC1, 2, and 3) is associated with clinically distinct subsets of ovarian and endometrialcarcinomas. Expression was assessed by immunohistochemistry in a population-based cohort of 465 ovarian and149 endometrial carcinomas and correlated with clinicopathologic parameters. Each of the HDACs was expressedat high levels in most ovarian (HDAC1, 61%; HDAC2, 93%; HDAC3, 84%) and endometrial (HDAC1, 61%; HDAC2,95%; HDAC3, 83%) carcinomas. Further, 55% and 56% of ovarian and endometrial carcinomas, respectively, ex-pressed all three HDACs at high levels. Such cases were less common among endometrioid subtypes of ovarianand endometrial carcinomas (36% and 52% positive cases, respectively) compared with high-grade serous sub-types (64 and 69%, respectively, P < .001). High-level expression of all three HDACs is associated with a poorprognosis in ovarian endometrioid carcinomas (hazard ratio, 6.7; 95% confidence interval, 1.9–23.3). The indepen-dent prognostic information and the overall high rate of expression for class I HDACs suggest that these targetsshould be explored as predictive factors in ovarian and endometrial carcinomas prospectively.

Neoplasia (2008) 10, 1021–1027

IntroductionCurrent treatment strategies for ovarian carcinomas with high-riskhistology (e.g., high-grade serous and clear cell subtype) are still in-effective [1]. Even in stage I/II, 30% to 40% of women with theseneoplasms will die of their disease [2]. For endometrial carcinomas ofthe serous and clear cell type, the prognosis is comparably grim [3].Consequently, novel target-based chemotherapeutic strategies areurgently needed. Conversely, endometrioid carcinomas of the ovaryand endometrium, which sometimes present synchronously [4], areless-aggressive tumors compared with their high-risk counterparts.The great majority of women (approximately 80%) will be curedby surgery alone because these tumors are frequently organ confinedand tend to metastasize late [5–7]. Because there is no good strategyto predict the natural course of these tumors, more than 90% ofovarian endometrioid carcinomas are still treated with platinum-based chemotherapy, and a large percentage of endometrial endo-metrioid carcinomas are treated with adjuvant radiotherapy and

chemotherapy [8,9]. Along with the opportunity of subtype-specifictargeted therapy, there is a need to test for novel markers that will helpto classify patient risk independent of currently known risk factors.

Histone modifications are known to be centrally involved in themalignant transformation of cells and histone deacetylase (HDAC)inhibitors have proven to be effective as anticancer agents in a broad

Abbreviations: HDAC, histone deacetylase; TMA, tissue microarray; HG serous, high-grade serousAddress all correspondence to: Martin Köbel, MD, Department of Pathology, Room1259, 1st Floor JPPN, Vancouver General Hospital, 855 West 12th Ave, Vancouver,BC, Canada V5Z 1M9. E-mail: Martin.Koebel@vch.ca1This work was supported by the Cheryl Brown Ovarian Cancer Outcomes Unit ofthe British Columbia Cancer Agency and by an unrestricted educational grant fromsanofiaventis. M.K. received fellowship support from Eli Lilly Canada.Received 10 April 2008; Revised 5 June 2008; Accepted 8 June 2008

Copyright © 2008 Neoplasia Press, Inc. All rights reserved 1522-8002/08/$25.00DOI 10.1593/neo.08474

www.neoplasia.com

Volume 10 Number 9 September 2008 pp. 1021–1027 1021

variety of tumors [10–13]. The family of HDACs to date comprises18 isoenzymes, which, on the basis of structure, could be subclassi-fied into four classes [11], with class I HDACs being the most thor-oughly investigated with respect to function and relevance for tumorformation and progression. Lately, research on these proteins hasgained momentum because small-molecule inhibitors of HDACsare now available and are currently tested as new antitumor drugsin late-phase clinical studies for different types of solid tumors[12,13], including ovarian cancer [14]. In addition, the HDAC in-hibitor vorinostat has recently been approved for therapy in cuta-neous T-cell lymphoma [15].

Histone deacetylase inhibitors are known to have profound anti-tumor effects in both endometrial [16–20] and ovarian carcinomacell lines [17,21–23] most likely by inducing cell cycle arrest, apopto-sis, and differentiation. These drugs act synergistically with paclitaxel-[24,25] and platinum-based chemotherapeutics [26,27] in vitro.

We have previously found class I HDACs to be highly expressed indifferent types of human cancers [28–30]. In addition, class I HDACexpression was an unfavorable independent prognostic factor in someof these tumor entities. Currently, little is known about HDAC ex-pression in gynecologic malignancies [23,31].

In our study, we focused on this apparent lack of translational re-search in this field and addressed the question whether class IHDACs are expressed in gynecologic malignancies and whetherexpression patterns vary with respect to clinicopathologic and prog-nostic patient subgroups.

Materials and Methods

Study PopulationFrom a population-based cohort of 3501 patients with ovarian

carcinoma in British Columbia diagnosed between 1984 and2000, 518 cases were selected on the basis of optimal surgical treat-ment (i.e., no macroscopic residual tumor after primary surgery) andwere eligible for tissue microarray (TMA) construction after com-plete gynecopathologic review [5]. Two hundred consecutive casesof endometrial carcinoma were retrieved from the archives of theDepartment of Pathology, Vancouver General Hospital, for the pe-riod 1983–1998. All cases underwent gynecopathologic review.Mixed carcinomas were classified according to their highest-gradecomponent and included if the high-grade component was sampledon the TMA. Clinical and pathologic data are shown in Table 1.Both series of cases have been the subject of previous studies[32,33]. A representative area of each tumor was selected and dupli-cate 0.6-mm tissue cores were punched to construct a TMA (BeecherInstruments, Silver Springs, MD).

Adjuvant Therapy and Follow-upNone of the included cases received preoperative radiotherapy or

chemotherapy. Approximately 97% of ovarian carcinoma patientswere treated according to the provincial treatment guidelines of theBritish Columbia Cancer Agency [8]. On the other hand, 3% ofpatients refused the advised chemotherapy and were excluded fromsurvival analysis. All patients with endometrial carcinomas weretreated according the BC Cancer Agency treatment recommenda-tion. All women underwent initial simple hysterectomy and bilateralsalpingo-oophorectomy. For those with documented widespread dis-ease, debulking surgery as in ovarian carcinomas was performed. Sub-sequently, radiation was added if the likelihood of local recurrencewas >10%, and chemotherapy with different regimens was given ifthe predicted risk of death was >25%. Systematic lymph node dissec-tion was not routinely performed. Outcomes were tracked throughthe Cheryl Brown Ovarian Cancer Outcomes Unit at the BritishColumbia Cancer Agency and were available for all patients. Medianfollow-up time was 5.1 and 6.2 years for ovarian and endometrialcarcinomas, respectively. Approval for the study was obtained fromthe Research Ethics Board of the University of British Columbia.

ImmunohistochemistrySerial 4-μm sections were cut for immunohistochemical analysis

and run through an automated system by Ventana, Tuscon, AZ, asper manufacturer’s protocol. Polyclonal rabbit anti-HDAC1 antibodyand monoclonal mouse anti-HDAC2 antibody were obtained fromAbcam, Cambridge, UK. Monoclonal mouse anti-HDAC3 antibodywas obtained from Becton Dickinson, Franklin Lakes, CA. Ampleconfirmation of antibody specificity has been performed in previousstudies [28–30].

Evaluation of ImmunohistochemistryTissue microarrays were scored for HDAC1, HDAC2, and HDAC3

by a pathologist (M.K.; scans available online: http://bliss.gpec.ubc.ca/under OOU ovarian carcinomas, under 02-005 endometrial carcino-mas). Only nuclear staining within tumor cells was considered for eval-uation. A four-tiered scoring system was used: 0 for negative cases, +1for weak intensity, +2 for moderate intensity, and +3 for strong inten-sity (Figure 1). To optimize for disease-specific survival differences, theraw data were binarized as follows: the moderate (+2) and strong (+3)cases were grouped for statistical analysis and assigned the designation 1and considered as high-level expressors, whereas the completely nega-tive (0) and weak (+1) cases were considered HDAC low-level expres-sors and designated as 0. In addition, cases were grouped as beinghigh for all three HDACs versus others. In total, 53 (10%) ovarian car-cinoma and 51 (26%) endometrial carcinomas were not interpretableat least for one marker, and all analyses were restricted to cases withcomplete information.

Statistical AnalysisUnivariable survival analysis was performed by the generation of

Kaplan-Meier curves, and differences between the groups were as-sessed using the log rank statistic. Multivariable survival analysiswas performed using the Cox proportional hazards model. Contin-gency tables and the Pearson χ 2 statistic were used to test the changein the distribution of HDAC expression across primary cell types.The Kruskal-Wallis test was used to compare medians for continuousvariables (Ki-67). Spearman rank order correlation was used to deter-mine whether there was a positive or negative correlation between

Table 1. Distribution of Ovarian and Endometrial Carcinomas across Stage.

Cell Type N (%) Stage I/II (%) Stage III (%) Age (Mean ± SE)

Ovarian carcinoma (all) 465 83 17 58.0 ± 13High-grade serous 180 (39%) 66 34 60.9 ± 12Clear cell 122 (27%) 93 7 56.4 ± 13Endometrioid 114 (24%) 96 4 55.8 ± 13Mucinous 24 (5%) 100 0 55.4 ± 13Other 25 (5%) 88 12 60.9 ± 11

Endometrial carcinoma (all) 149 87 13 63.5 ± 13Endometrioid 127 (85%) 93 7 63.0 ± 13Serous 13 (9%) 46 54 70.1 ± 8Clear cell 6 (4%) 83 17 65.5 ± 17Other 3 (2%) 67 33 58.7 ± 12

1022 HDAC Expression Prognosis Endometrioid Carcinoma Weichert et al. Neoplasia Vol. 10, No. 9, 2008

the levels of expression shown by any two types of immunostaining.All analyses were performed using SPSS v15.0 (SPSS Inc, Chicago,IL). P values < .05 were considered significant.

Results

Expression Patterns of Class I HDAC Isoformsin Normal Tissue

HDAC1, HDAC2, and HDAC3 are expressed weakly in nuclei ofovarian surface epithelium of normal ovaries (n = 5). Expression of allclass I isoforms in normal endometrium and in ovarian endometriosiswas variable but did not exceed moderate staining intensity. Normalendometrial stroma cells and inflammatory cells, if present, partlyshowed moderate staining intensity as well.

Expression Patterns of Class I HDAC Isoformsin Ovarian Carcinomas

Most ovarian carcinomas expressed high levels of HDAC1 (61%),HDAC2 (93%), and HDAC3 (84%) in the nuclei of tumor cells(Figure 1 and Table 2). Cytoplasmic tumor cell expression was notobserved. Stromal cells and inflammatory cells were occasionally pos-

itive. Generally, HDAC2 had the highest expression level in alltumor subtypes. Expression of class I HDAC isoforms correlatedwith each other: HDAC1 with HDAC2 (r = 0.150, P = .001),HDAC1 with HDAC3 (r = 0.300, P < .001), and HDAC2 withHDAC3 (r = 0.265, P = .001), suggesting a shared functional regu-lation. High-level expression of all three isoforms was detected in55% of ovarian carcinomas with the mucinous subtype of ovariancarcinoma showing the highest frequency (71%), followed by high-grade serous (64%), clear cell (54%), and endometrioid subtypes(36%). These expression levels are indicative of significant differen-tial expression (P < .001).

Expression Patterns of Class I HDAC Isoformsin Endometrial Carcinomas

In parallel to the results reported for ovarian carcinomas, mostendometrial carcinomas expressed class I HDAC isoforms in the nu-clei of tumor cells (HDAC1, 61%; HDAC2, 95%; HDAC3, 83%;Table 2 and Figure 1). As for ovarian carcinomas, clear cell (83%)and serous subtypes (69%) showed significantly higher expression ratesfor all three HDACs than endometrioid carcinomas (52%, P < .001;Table 2). In all tumor types, the expression of HDAC2 was highest,followed by HDAC3 and HDAC1 (Table 2). Expression of HDAC

Figure 1. HDAC expression: (A to D) HDAC1 expression in endometrial endometrioid carcinomas score 0 (A), score 1 (B), score 2 (C),score 3 (D); (B) insert normal proliferative endometrium. (E) HDAC2 expression in ovarian clear cell carcinoma, score 3. (F) HDAC3expression in ovarian high-grade serous carcinoma, score 3.

Neoplasia Vol. 10, No. 9, 2008 HDAC Expression Prognosis Endometrioid Carcinoma Weichert et al. 1023

isoforms showed a moderate degree of correlation with one another:HDAC1 with HDAC2 (r = 0.321, P < .001), HDAC1 with HDAC3(r = 0.512, P < .001), and HDAC2 with HDAC3 (r = 0.339, P <.001). Neither in ovarian nor in endometrial carcinomas a significantassociation of HDAC1, HDAC2, or HDAC3 expression with stagewithin histologic subtypes was observed (data not shown).

Correlation of Class I HDAC Expression with ProliferationBecause data from cell culture studies suggest a tight interlink be-

tween HDACs and cell proliferation [34,35], we aimed to test this po-tential association for ovarian and endometrial carcinomas in vivo.Those tumors showing higher proliferative capacity as determined byKi-67 staining usually expressed higher levels of all three class I HDACisoforms, in both ovarian (P < .001) and endometrial carcinomas (P <.001). This significant association was also observed in most of themajor subtypes, with the exception of ovarian endometrioid carcinomas,which showed a borderline significance of (P = .05; Figure 2).

Correlation of Class I HDAC Expression withPatient Prognosis

High-level expression of all three class I HDACs was significantlyassociated with decreased disease-specific survival in endometrioid

ovarian cancer (Figure 3A; P = .007). In this tumor type, 10-yeardisease-specific survival of women with tumors expressing all threeisoforms at high levels was 67% compared with those patients whosetumors expressed no or only some isoforms with 93%. This effectwas mainly due to a significant association of HDAC1 expressionwith prognosis (HDAC1, P = .012; HDAC2, P = .598; HDAC3,P = .391). In multivariable survival analysis performed under in-clusion of age, FIGO stage, Silverberg grade, and all three class IHDACs, HDAC expression retained its independent statistical sig-nificance (hazard ratio [HR], 6.7, P = .003; Table 3). In serous (Fig-ure 3B; P = .756), mucinous, and clear cell carcinomas (data notshown) of the ovary, none of the HDAC isoforms had significantimpact on patient prognosis in univariate survival analysis.

In endometrial endometrioid carcinomas, 10-year disease-specificsurvival of women with tumors expressing all three isoforms at highlevels was 63% compared with those patients whose tumors ex-pressed no or only some isoforms with 83%. Although this wasnot statistically significant (Figure 3C ; P = .123), the survival differ-ence was similar to that observed in their ovarian counterparts (20%vs 26%). In multivariable analysis under inclusion of age, FIGO stage,grade, and grouped HDAC expression, class I HDAC expressionshowed a trend toward unfavorable prognosis (HR, 2.0, P = .151;Table 4).

Figure 2. Box-percentile plot of Ki-67 labeling index in ovarian carcinoma subtypes. HG serous (indicates high-grade serous), clear cell,and endometrioid ovarian carcinoma and endometrioid subtype of endometrial carcinomas subdivided by low (blue) versus high (red)expression of all three HDACs. P values, Kruskal-Wallis test.

Table 2. HDAC Expression across Ovarian and Endometrial Carcinoma Subtypes.

Cell Type N HDAC1-Positive Cases (%) HDAC2-Positive Cases (%) HDAC3-Positive Cases (%) All Three HDACs-Positive Cases (%)

Ovarian carcinoma 465 61 93 84 55High-grade serous 180 67 98 92 64Clear cell 122 64 86 77 54Endometrioid 114 45 94 78 36Mucinous 24 83 75 83 71Other 25 64 100 92 64

Endometrial carcinoma 149 61 95 83 56Endometrioid 127 58 94 81 52Serous 13 69 100 100 69Clear cell 6 83 100 83 83Other 3 100 100 100 100

Other ovarian carcinomas include nine low-grade serous, six transitional, six undifferentiated, and four adenocarcinoma not otherwise specified.Other endometrial carcinomas include two small cell carcinomas and one large cell carcinoma.

1024 HDAC Expression Prognosis Endometrioid Carcinoma Weichert et al. Neoplasia Vol. 10, No. 9, 2008

DiscussionIn this study, we describe a high-level expression of class I HDAC

isoforms in ovarian and endometrial carcinomas. Histone deacetylaseexpression is especially prominent in high-grade tumors of serous andclear cell subtype. In the endometrioid subtype, we show that ex-pression of the three class I HDACs is an independent prognosticmarker. Reports on in vivo expression of single HDAC isoforms inovarian and endometrioid carcinomas are sparse. In a small set of10 ovarian carcinomas, high expression levels for class I HDACs werereported [23], which is in line with our results. However, in this co-hort, it was not possible to test for prognosis or clinical correlations.

Although, in our study, a significant unfavorable prognostic asso-ciation of class I HDACs for endometrioid tumor subtypes was onlyobserved for ovarian but not for endometrial carcinomas, the sametrend and similar differences in 10-year survival rates for class IHDAC expressing groups were evident for tumors arising in bothorgans. The reason for the differences between ovarian and endome-trial carcinomas might be a different assembly of the patient cohorts.The ovarian carcinoma patients were recruited from a population-based cohort from British Columbia and restricted to patients withcompletely resectable disease after primary surgery. Patients withendometrial carcinoma were selected from a cohort treated at theVancouver General Hospital, which is a specialized center where pa-tients with high-risk disease are referred for surgery. This is reflectedby a higher number of grade 3 carcinomas in the endometrial endo-metrioid group (24%) compared to other hospital-based cohorts ofendometrioid endometrial carcinoma (6%) [6] and to the ovarianendometrioid group (7%, P = .002, Pearson χ 2). In multivariableanalysis, grade and age had a stronger influence on prognosis ofendometrial endometrioid than ovarian endometrioid carcinomas.Nevertheless, based on presumably similar biology and given the factthat both tumors occur synchronously in a significant proportionof patients [4], we think that the findings in ovarian endometrioidcarcinomas could be transferable to their endometrial counterpart.Larger numbers might be needed to confirm these finding in endo-metrial endometrioid carcinomas.

Previous immunohistochemical studies of HDAC isoform expres-sion have been published for gastric [28], colorectal [29], prostate[30], and breast [34] cancers by our group and others. In colorectalcancers, class I HDAC expression was associated with poor prognosis[29]. Because colorectal carcinomas show pathogenetic pathways(e.g., microsatellite instability, β-catenin mutation, K-Ras mutation)strikingly similar to endometrioid carcinomas [35,36], similar prog-nostic effects of HDAC expression might be not surprising.

Figure 3. Kaplan-Meier survival analysis: (A) ovarian endometrioidcarcinomas (N = 114); (B) ovarian high-grade serous carcinomas(N = 180); (C) endometrial endometrioid carcinomas (N = 127).P values, log rank test.

Table 3. Univariable and Multivariable Analyses of Disease-Specific Survival for Ovarian Endo-metrioid Carcinomas (N = 114/15 Events).

Patient Characteristics Risk Factor Univariable Analysis Multivariable Analysis

HR 95% CI P HR 95% CI P

Age * 0.423 * 0.246Stage I/II 1.0 0.001 1.0 0.0002

III 9.3 2.5–34.0 16.9 3.8–75.4Silverberg grade 1 1.0 0.264 1.0 0.408

2 1.7 0.5–5.2 1.3 0.4–1.33 3.5 0.7–16.7 3.1 0.6–15.7

All three HDACs 0 1.0 0.007 1.0 0.0031 4.9 1.5–15.7 6.7 1.9–23.3

CI indicates confidence interval; HR, hazard ratio.*Hazard ratio for age is not available because it is a continuous variable.

Table 4. Univariable and Multivariable Analyses of Disease-Specific Survival for Endometrial En-dometrioid Carcinomas (N = 123/20 Events).

Patient Characteristics Risk Factor Univariable Analysis Multivariable Analysis

HR 95% CI P HR 95% CI P

Age * 0.0004 * 0.001Stage I 1.0 0.005 1.0 0.133

II/III 3.6 1.4–8.9 2.1 0.8–5.8WHO Grade 1 1.0 0.042 1.0 0.052

2 1.0 0.2–4.4 0.5 0.1–2.63 3.1 1.2–7.9 2.5 0.9–6.8

All three HDACs 0 1.0 0.123 1.0 0.1511 2.1 0.8–5.4 2.0 0.8–5.4

*Hazard ratio for age is not available because it is a continuous variable.

Neoplasia Vol. 10, No. 9, 2008 HDAC Expression Prognosis Endometrioid Carcinoma Weichert et al. 1025

In the entire cohort and in each subtype, class I HDAC expressionis pronounced in highly proliferating tumors as determined by highKi-67 labeling index. These data fit well with results from functionalin vitro studies showing that HDAC inhibitors induce cell cyclearrest in ovarian and endometrial cancer cell lines [17–23].

Our observation that HDAC expression levels vary considerablyand have impact on patient survival in ovarian endometrioid carci-noma implicates that expression of these proteins might be usefulas a prognostic marker. The absence of or only partial expressionof class I HDACs indicates a very good prognosis (a 93% 10-yeardisease-specific survival). Because 94% of these patients received ad-juvant platinum-based chemotherapy, we are not able to compare theoutcome with a nontreated group. Anecdotally from seven patients,where adjuvant chemotherapy was not advised, only one woman diedof disease, and this tumor expressed a high level of all three HDACs.Further, if we estimate the cure rate of women in the chemotherapygroup by using the effectiveness of chemotherapy (33% of patientsare cured by adjuvant therapy) described by the authors of the Actiontrial [37], we calculate that 7% died despite chemotherapy and only4% of our patients were actually cured by chemotherapy; hence,89% were treated without benefit. Therefore, we would hypothesizethat HDACs’ expression assessment can identify candidates that canbe spared from first-line chemotherapy to minimize overtreatment.Future larger studies might clarify whether it is necessary to assessan antibody panel against all three HDACs or whether this prognosticeffect is largely based on the expression of one isoform, e.g., HDAC1.

An even more interesting aspect of our results is that response ofHDAC inhibitors might vary with different expression levels of class IHDACs. Treatment of tumors expressing low levels of HDACsmight be ineffective because it has been shown that HDAC-negativetumors caused by the mutation of HDAC genes are resistant toHDAC inhibitors [38]. Hence, we hypothesize that class I HDACexpression in ovarian and endometrial carcinomas might be associ-ated with a positive response to HDAC inhibitors. This would beespecially desirable in endometrioid subtypes, because those patientsthat highly express class I HDACs have a compromised prognosis.Because determination of class I HDAC expression status in tumortissue is easy, straightforward, and possible on very small tissue sam-ples such as the 0.6-mm cores on a tissue microarray, such analysis isfeasible in clinical trials. As we have shown subtype-specific differ-ences in the prognostic value, it is possible that predictive effects willalso be restricted to specific pathologic subtypes.

References[1] Winter WE III, Maxwell GL, Tian C, Carlson JW, Ozols RF, Rose PG,

Markman M, Armstrong DK, Muggia F, McGuire WP, et al. (2007). Prognosticfactors for stage III epithelial ovarian cancer: a Gynecologic Oncology GroupStudy. J Clin Oncol 25, 3621–3627.

[2] Kristensen GB, Kildal W, Abeler VM, Kaern J, Vergote I, Tropé CG, andDanielsen HE (2003). Large-scale genomic instability predicts long-term outcomefor women with invasive stage I ovarian cancer. Ann Oncol 14, 1494–1500.

[3] Soslow RA, Bissonnette JP, Wilton A, Ferguson SE, Alektiar KM, Duska LR,and Oliva E (2007). Clinicopathologic analysis of 187 high-grade endometrialcarcinomas of different histologic subtypes: similar outcomes belie distinctivebiologic differences. Am J Surg Pathol 31, 979–987.

[4] Fujii H, Matsumoto T, Yoshida M, Furugen Y, Takagaki T, Iwabuchi K, NakataY, Takagi Y, Moriya T, Ohtsuji N, et al. (2002). Genetics of synchronous uterineand ovarian endometrioid carcinoma: combined analyses of loss of heterozygos-ity, PTEN mutation, and microsatellite instability. Hum Pathol 33, 421–428.

[5] Gilks CB, Ionescu DN, Kalloger SE, Köbel M, and Swenerton K (2008). Tumorcell type can be reproducibly diagnosed and is of independent prognostic sig-

nificance in patients with maximally debulked ovarian carcinoma. Hum Pathol39, 1239–1251.

[6] Köbel M, Langhammer T, Hüttelmaier S, Schmitt WD, Kriese K, Dittmer J, StraussHG, Thomssen C, and Hauptmann S (2006). Ezrin expression is related to poorprognosis in FIGO stage I endometrioid carcinomas. Mod Pathol 19, 581–587.

[7] Creutzberg CL, van Putten WL, Koper PC, Lybeert ML, Jobsen JJ, Wárlám-Rodenhuis CC, De Winter KA, Lutgens LC, van den Bergh AC, van de Steen-Banasik E, et al. (2000). Surgery and postoperative radiotherapy versus surgeryalone for patients with stage-1 endometrial carcinoma: multicentre randomisedtrial. PORTEC Study Group. Post Operative Radiation Therapy in EndometrialCarcinoma. Lancet 355, 1404–1411.

[8] Swenerton KD (1992). Prognostic indices in ovarian cancer. Their significancein treatment planning. Acta Obstet Gynecol Scand Suppl 155, 67–74, and up-dated ovarian carcinoma treatment guidelines in British Columbia (http://www.bccancer.bc.ca/PPI/TypesofCancer/Ovary/default.htm).

[9] Maggi R, Lissoni A, Spina F, Melpignano M, Zola P, Favalli G, Colombo A, andFossati R (2006). Adjuvant chemotherapy vs radiotherapy in high-risk endo-metrial carcinoma: results of a randomised trial. Br J Cancer 95, 266–271.

[10] Yoo CB and Jones PA (2006). Epigenetic therapy of cancer: past, present andfuture. Nat Rev Drug Discov 5, 37–50.

[11] Minucci S and Pelicci PG (2006). Histone deacetylase inhibitors and the prom-ise of epigenetic (and more) treatments for cancer. Nat Rev Cancer 6, 8–51.

[12] Xu WS, Parmigiani RB, and Marks PA (2007). Histone deacetylase inhibitors:molecular mechanisms of action. Oncogene 26, 5541–5552.

[13] Bolden JE, Peart MJ, and Johnstone RW (2006). Anticancer activities of histonedeacetylase inhibitors. Nat Rev Drug Discov 5, 769–784.

[14] Candelaria M, Gallardo-Rincón D, Arce C, Cetina L, Aguilar-Ponce JL, ArrietaO, González-Fierro A, Chávez-Blanco A, de la Cruz-Hernández E, CamargoMF, et al. (2007). A phase II study of epigenetic therapy with hydralazineand magnesium valproate to overcome chemotherapy resistance in refractorysolid tumors. Ann Oncol 18, 1529–1538.

[15] Marks PA and Breslow R (2007). Dimethyl sulfoxide to vorinostat: develop-ment of this histone deacetylase inhibitor as an anticancer drug. Nat Biotechnol25, 84–90.

[16] Zhou XC, Dowdy SC, Podratz KC, and Jiang SW (2007). Epigenetic considera-tions for endometrial cancer prevention, diagnosis and treatment. Gynecol Oncol107, 143–153.

[17] Takai N and Narahara H (2007). Human endometrial and ovarian cancer cells:histone deacetylase inhibitors exhibit antiproliferative activity, potently inducecell cycle arrest, and stimulate apoptosis. Curr Med Chem 14, 2548–2553.

[18] Ahn MY, Jung JH, Na YJ, and Kim HS (2008). A natural histone deacetylaseinhibitor, Psammaplin A, induces cell cycle arrest and apoptosis in human en-dometrial cancer cells. Gynecol Oncol 108, 27–33.

[19] Takai N, Desmond JC, Kumagai T, Gui D, Said JW, Whittaker S, Miyakawa I,and Koeffler HP (2004). Histone deacetylase inhibitors have a profound anti-growth activity in endometrial cancer cells. Clin Cancer Res 10, 1141–1149.

[20] Jiang S, Dowdy SC, Meng XW, Wang Z, Jones MB, Podratz KC, and Jiang SW(2007). Histone deacetylase inhibitors induce apoptosis in both type I and typeII endometrial cancer cells. Gynecol Oncol 105, 493–500.

[21] Takai N, Kawamata N, Gui D, Said JW, Miyakawa I, and Koeffler HP (2004).Human ovarian carcinoma cells: histone deacetylase inhibitors exhibit anti-proliferative activity and potently induce apoptosis. Cancer 101, 2760–2770.

[22] Strait KA, Dabbas B, Hammond EH, Warnick CT, Iistrup SJ, and Ford CD(2002). Cell cycle blockade and differentiation of ovarian cancer cells by thehistone deacetylase inhibitor trichostatin A are associated with changes inp21, Rb, and Id proteins. Mol Cancer Ther 1, 1181–1190.

[23] Khabele D, Son DS, Parl AK, Goldberg GL, Augenlicht LH, Mariadason JM,and Rice VM (2007). Drug-induced inactivation or gene silencing of class Ihistone deacetylases suppresses ovarian cancer cell growth: implications for ther-apy. Cancer Biol Ther 6, 795–801.

[24] Cooper AL, Greenberg VL, Lancaster PS, van Nagell JR Jr, Zimmer SG, andModesitt SC (2007). In vitro and in vivo histone deacetylase inhibitor therapywith suberoylanilide hydroxamic acid (SAHA) and paclitaxel in ovarian cancer.Gynecol Oncol 104, 596–601.

[25] Sonnemann J, Gänge J, Pilz S, Stötzer C, Ohlinger R, Belau A, Lorenz G, andBeck JF (2006). Comparative evaluation of the treatment efficacy of suberoyla-nilide hydroxamic acid (SAHA) and paclitaxel in ovarian cancer cell lines andprimary ovarian cancer cells from patients. BMC Cancer 6, 183.

[26] Strait KA, Warnick CT, Ford CD, Dabbas B, Hammond EH, and Ilstrup SJ(2005). Histone deacetylase inhibitors induce G2-checkpoint arrest and apoptosis

1026 HDAC Expression Prognosis Endometrioid Carcinoma Weichert et al. Neoplasia Vol. 10, No. 9, 2008

in cisplatinum-resistant ovarian cancer cells associated with overexpression of theBcl-2–related protein Bad. Mol Cancer Ther 4, 603–611.

[27] Ozaki KI, Kishikawa F, Tanaka M, Sakamoto T, Tanimura S, and Kohno M(2008). Histone deacetylase inhibitors enhance the chemosensitivity of tumorcells with cross-resistance to a wide range of DNA-damaging drugs. Cancer Sci99, 376–384.

[28] Weichert W, Röske A, Gekeler V, Beckers T, Ebert MP, Pross M, Dietel M,Denkert C, and Röcken C (2008). Association of patterns of class I histonedeacetylase expression with patient prognosis in gastric cancer: a retrospectiveanalysis. Lancet Oncol 9, 139–148.

[29] Weichert W, Roeske A, Niesporek S, Noske A, Buckendahl AC, Dietel M,Gekeler V, Boehm M, Beckers T, and Denkert C (2008). Class I histone de-acetylase expression has independent prognostic impact in human colorectalcancer—specific role of class I HDACs in vitro and in vivo. Clin Cancer Res14, 1669–1677.

[30] Weichert W, Röske A, Gekeler V, Beckers T, Stephan C, Jung K, Fritzsche FR,Niesporek S, Denkert C, Dietel M, et al. (2008). Histone deacetylases 1, 2and 3 are highly expressed in prostate cancer and HDAC2 expression is asso-ciated with shorter PSA relapse time after radical prostatectomy. Br J Cancer 98,604–610.

[31] Krusche CA, Vloet AJ, Classen-Linke I, von Rango U, Beier HM, and Alfer J(2007). Class I histone deacetylase expression in the human cyclic endometriumand endometrial adenocarcinomas. Hum Reprod 22, 2956–2966.

[32] Prentice LM, Klausen C, Kalloger S, Köbel M, McKinney S, Santos JL, KenneyC, Mehl E, Gilks CB, Leung P, et al. (2007). Kisspeptin and GPR54 immuno-

reactivity in a cohort of 518 patients defines favourable prognosis and clear cellsubtype in ovarian carcinoma. BMC Med 5, 33.

[33] Alkushi A, Clarke BA, Akbari M, Makretsov N, Lim P, Miller D, Magliocco A,Coldman A, van de Rijn M, Huntsman D, et al. (2007). Identification of prog-nostically relevant and reproducible subsets of endometrial adenocarcinoma basedon clustering analysis of immunostaining data. Mod Pathol 20, 1156–1165.

[34] Krusche CA, Wülfing P, Kersting C, Vloet A, Böcker W, Kiesel L, Beier HM,and Alfer J (2005). Histone deacetylase-1 and -3 protein expression in humanbreast cancer: a tissue microarray analysis. Breast Cancer Res Treat 90, 15–23.

[35] Matias-Guiu X, Catasus L, Bussaglia E, Lagarda H, Garcia A, Pons C, Munoz J,Arguelles R, Machin P, and Prat J (2001). Molecular pathology of endometrialhyperplasia and carcinoma. Hum Pathol 32, 569–577.

[36] Shen L, Toyota M, Kondo Y, Lin E, Zhang L, Guo Y, Hernandez NS, Chen X,Ahmed S, Konishi K, et al. (2007). Integrated genetic and epigenetic analysisidentifies three different subclasses of colon cancer. Proc Natl Acad Sci USA 104,18654–18659.

[37] Trimbos JB, Vergote I, Bolis G, Vermorken JB, Mangioni C, Madronal C,Franchi M, Tateo S, Zanetta G, Scarfone G, et al. (2003). Impact of adjuvantchemotherapy and surgical staging in early-stage ovarian carcinoma: EuropeanOrganisation for Research and Treatment of Cancer-Adjuvant ChemoTherapyin Ovarian Neoplasm trial. J Natl Cancer Inst 95, 113–125.

[38] Ropero S, Fraga MF, Ballestar E, Hamelin R, Yamamoto H, Boix-Chornet M,Caballero R, Alaminos M, Setien F, Paz MF, et al. (2006). A truncating muta-tion of HDAC2 in human cancers confers resistance to histone deacetylaseinhibition. Nat Genet 38, 566–569.

Neoplasia Vol. 10, No. 9, 2008 HDAC Expression Prognosis Endometrioid Carcinoma Weichert et al. 1027