Gene Expression Profiling of Microsatellite Unstable and Microsatellite Stable Endometrial Cancers Indicates Distinct Pathways of Aberrant Signaling

Research Article

John I. Risinger,1 G. Larry Maxwell,2 Gadisetti V.R. Chandramouli,1 Olga Aprelikova, Tracy Litzi,1 Asad Umar,1 Andrew Berchuck,3 and J. Carl Barrett1

1Laboratory of Biosystems and Cancer, National Cancer Institute, Bethesda, Maryland; 2Walter Reed Army Medical Center, Washington, District of Columbia; and 3Department of Obstetrics and Gynecology/Division of Gynecologic Oncology, Duke University, Durham, North Carolina

Abstract Indeed, substantial correlative evidence exists for this idea, Microsatellite instability (MSI) is a molecular phenotype particularly in colon carcinoma where tumor suppressor or present in f25% of endometrial cancers. We examined the antiapoptotic genes containing hypermutable repeat tracts located global gene expression profiles of early-stage endometrioid within the coding region of the gene are frequently mutated in MSI endometrial cancers with and without the MSI phenotype to cancers. This is best exemplified for the TGFBR2 gene, which is test the hypothesis that MSI phenotype may determine a mutated in a hypermutable (A)10 tract in the majority of colorectal unique molecular signature among otherwise similar cancers. cancers with MSI (1). However, this repeat tract and others are not Unsupervised principal component analysis of the expression universally mutated in other MSI epithelial cancers including data from these cases indicated two distinct groupings of endometrial cancers (2, 3). Cells defective in mismatch repair also cancers based on MSI phenotype. A relatively small number of show defects in cellular signaling mechanisms that normally can array features (392) at high statistical value (P < 0.001) were lead to apoptosis or growth arrest (4). Additionally, data indicate identified that drive the instability signature in these cancers; various survival advantages of mismatch repair–negative cells 109 of these transcripts differed by at least 2-fold. These data under adverse conditions. Specifically, defective mismatch repair in identify distinct gene expression profiles for MSI and micro- yeast can lead to increased survival in the absence of telomerase satellite stable (MSS) cancers, which suggest that cancers with activity, and human mismatch repair–negative cells are more likely MSI develop in part by different mechanisms from their to survive in hypoxic and low-pH environments (5, 6). Cells similar stable counterparts. In particular, we found evidence defective in mismatch repair often fail to transmit proapoptotic that two members of the secreted frizzled related protein signals following various cellular challenges including alkylating family (SFRP1 and SFRP4) were more frequently down- agents, DNA cross-linking agents, as well as oxidative stresses, regulated in MSI cancers as compared with MSS cancers. suggesting the additional possibility of differential responses to Down-regulation was accompanied by promoter hypermethy- chemopreventive or therapeutic agents (4, 7). This effect may be lation for SFRP1. SFRP1 was hypermethylated in 8 of 12 MSI mediated by transcriptional inhibition dependent on MLH1 and cancers whereas only 3 of 16 MSS cancers were methylated. p53 (8). The WNT target fibroblast growth factor 18 was found to be Approximately 20% to 25% of nonfamilial endometrial cancers up-regulated in MSI cancers. These data classify histologically have the MSI phenotype and it occurs almost exclusively in cancers similar endometrioid endometrial cancers into two distinct of endometrioid histology (9). Although many of these cancers are groupings with implications affecting therapy and prevention. generally curable by surgery alone, this disease is the most frequent (Cancer Res 2005; 65(12): 5031-7) gynecologic malignancy in western societies and is associated with high costs in terms of morbidity. Studies suggest that microsatellite Introduction unstable (MSI) and microsatellite stable (MSS) cancers have different clinical outcomes (10–13). In our study of stage I through Microsatellite instability (MSI) is a common feature of many stage IV endometrial cancers, MSI cancers were associated with common epithelial cancers and of those cancers arising in the other favorable prognostic indicators (14). However, when stage I Lynch (hereditary nonpolyposis colorectal cancer) cancer syn- and stage II endometrial cancers are examined separately, MSI drome. The phenotype is diagnostic of an underlying defect in one status is associated with recurrence and with poorly differentiated or more DNA mismatch repair genes either by mutation or by cancers, indicating that these cancers are more likely to require epigenetic silencing. The exact mechanism by which cancer arises additional clinical attention (15). More extensive clinical outcome from this defect remains controversial. One widely accepted studies on endometrial cancer are necessary to resolve these possibility is that an increased mutation rate due to postreplication inconsistencies. DNA mismatch repair deficiency leads to an accumulation of Endometrial cancer is a cancer of which incidence and mortality mutations in genes normally important in the control of cancer. are affected by association with high body mass index, suggesting that the incidence and mortality of this cancer may dramatically increase along with increasing trends of weight gain in western Note: Supplementary data for this article are available at Cancer Research Online (http://cancerres.aacrjournals.org/). societies (16, 17). The clear association with high body mass index Requests for reprints: John I. Risinger, Laboratory of Biosystems and Cancer, NIH/ also suggests that many endometrial cancers may be readily National Cancer Institute, Room 5046, Building 37, 9000 Rockville Pike, MSC 4264, Bethesda, MD. Phone: 301-594-8503; E-mail: [email protected]. preventable. Preventive or therapeutic strategies may depend, in I2005 American Association for Cancer Research. part, on underlying genetic or epigenetic differences in particular www.aacrjournals.org 5031 Cancer Res 2005; 65: (12). June 15, 2005

Downloaded from cancerres.aacrjournals.org on October 1, 2021. © 2005 American Association for Cancer Research. Cancer Research cancers or individuals. It is imperative to understand the principal components accounting for 26% of the total variance (computed underlying biology of endometrial cancers and to develop effective using software Partekpro 5.0, Partek, Inc., St. Charles, MO). preventive and nonsurgical therapeutic strategies for this disease. Differentially expressed genes. Parametric Student’s t tests indicated Like many other cancers, endometrial cancers are thought to 392 gene transcripts were differentially expressed between MSS and MSI at P < 0.001. Multivariate permutation test based on 10,000 permutations develop in part due to accumulation of defects in cellular growth estimated about 3% false discoveries among these transcripts and the control pathways. In addition to the MSI phenotype, many probability of getting at least 392 genes significant by chance if there are no endometrial cancers contain mutations of the PTEN and TP53 real differences between the classes as 0.0001 (BRB Array tools ver. 3.0c, tumor suppressor genes, mutagenic activation of the KRAS2 Richard Simon, Amy Peng, Biometric research branch, National Cancer oncogene, and mutagenic stabilization of h-catenin (CTNNB1; ref. Institute, NIH; http://linus.nci.nih.gov/BRB-ArrayTools.html). 18). None of these events occur in all endometrial cancers and the Methylation-specific PCR. Specific primers for unmethylated and frequency of these alterations differs in various histologic types methylated DNA following sodium bisulfite conversion of DNA were (19). We previously showed that gene expression profiling can also designed within the CpG island of secreted frizzled related protein 1 clearly distinguish between the different histologic types of endo- (SFRP1). Primers for unmethylated PCR were U2F-gtttgtggtaggatgtgtgt and metrial cancer (20). In this study, we sought to determine whether U2R-acaacaacccccctcactaca. Primers for the methylated PCR were M2F- gtttgcggtaggacgcgcgc and M2R-gcgacgacccccctcgctacg. PCR assays were the MSI phenotype in early-stage endometrial cancers is associated verified by analysis of endometrial carcinoma cell lines, HEC-1-A, with characteristic patterns of gene expression that could be containing a methylated promoter and no SFRP1 transcript expression, defined using microarrays. These profiles may be useful in deter- and KLE, which expresses abundant SFRP1 and exhibits no promoter mining future treatment options tailored to individual cancers. methylation. Reverse transcription-PCR primers for SFRP1 were 873F- gagttgaaatctgaggccatca and 1061R-tggtggcagggacagtca. Hierarchical clustering. The set of 109 array features having 2-fold Materials and Methods differential expression (P < 0.001) between MSI and MSS were clustered by Clinical specimens. Flash-frozen endometrial cancers were obtained the similarity of their expression profiles. Hierarchical clustering was done from patients undergoing hysterectomy at Duke University Medical Center. on logarithmic values of expressions using 1 - U as distance metric, where U None of the patients had received preoperative chemotherapy or radiation. is the correlation coefficient of any two gene expressions (22). The In addition, samples of histologically normal endometrium were obtained expression data were shown relative to average gene expression. from patients undergoing hysterectomy for benign gynecologic diseases. Tissues were obtained at Duke with Institutional Review Board–approved Results and Discussion informed consent and this study was approved by the National Cancer Institute Institutional Review Board. Tissue samples were subjected to RNA We first analyzed 24 endometrial cancers of similar stage, grade, isolation using TRIzol and an additional purification using the Rneasy Kit and endometrioid histology (and their matching normal DNAs) for (Qiagen, Valencia, CA) following the recommendation of the manufacturer. the MSI phenotype utilizing the Bethesda panel of simple repeat Following isolation of RNA, the integrity of each RNA sample was verified by markers (23). We found that 9 of 24 cancers exhibited the MSI-H denaturing gel electrophoresis. Genomic DNA was isolated from cancer and phenotype wherein at least two of the five markers exhibited noncancer tissue following proteinase K digestion and phenol/chloroform additional alleles as compared with the matched normal sample. extraction. We found no evidence for the MSI-L phenotype in this group of Microsatellite instability analysis. Normal and genomic DNAs from cancers. The remaining 15 cancers did not show evidence of MSI each patient were examined using the National Cancer Institute/Bethesda and were designated MSS. To explore the global gene expression panel of microsatellite markers (BAT25, BAT26, D2S123, D5S346, and D17S250) using a modification of the fluorescent protocol developed by profiles of these samples, we did gene expression array analysis on Berg et al. (21). Briefly, markers were amplified separately but electro- all 24 samples using Affymetrix HU133A and HU133B chips. phoresed simultaneously on an ABI 3700 capillary sequence detection Initially, we examined the possibility of underlying expression system. Peaks were analyzed using genescan software. Cancers were either profiles by examining the data obtained from these 24 endometrial MSI-high (MSI-H) when at least two of the five markers exhibited cancers using unsupervised principal component analysis. Princi- differences in allele size as compared with the patient’s matched normal pal component analysis is a linear transformation of the data to DNA control or microsatellite stable (MSS) when no marker exhibited study variances by minimizing the covariances or correlations. This change. None of the cancers exhibited the MSI-low (MSI-L) phenotype can be used for dimension reduction when most of the variance is where only one of five markers exhibits differences. Cancers in this article accounted by a few principal components. The resulting data will are therefore described as MSS or MSI. separate samples with like expression patterns. Principal compo- Gene expression array. We studied gene expressions using Affymetrix Human Genome-U133A and HG-U133B Gene Chip expression arrays (45,000 nent analysis on the expression data from this set of endometrial gene transcripts covering 28,473 UniGene clusters). Five micrograms of total cancers revealed distinctively separated groups for MSI and MSS RNA from each sample were labeled using the bioarray high yield transcript endometrial cancers, indicating consistent expression changes for kit according to the conditions of the manufacturers (ENZO Life Sciences, each type of sample when plotted in three dimensions (Fig. 1). Inc., Farmingdale, NY). Labeled RNAs were hybridized and washed These groupings were quite distinct, indicating that a consistent according to the recommendations of the manufacturer (Affymetrix, Inc., set of gene expression differences were driving this separation. Santa Clara, CA). Microarray Suite 5.0 software (Affymetrix, 2001) was used To determine the genes responsible for these groupings, we for initial gene expression data analysis. All arrays were normalized to a examined those transcripts that are significantly different between trimmed mean transcript signal level of 500 counts. The genes associated the MSI and MSS groups. We applied stringent statistical analysis with ‘‘absent’’ signal detection calls in 70% of arrays both in MSI and MSS and identified 392 transcripts at P < 0.001. Experimental and samples were excluded from further analysis. The signal values of the remaining 22,378 genes were used for statistical analysis. All the statistical biological validation is an important component of any global calculations were done on the logarithmic values of signals. transcript analysis. Importantly, we identified the hMLH1 gene as Principal component analysis. Principal component analysis minimiz- one of the distinguishing transcripts. Epigenetic silencing of the ing correlations were done on 17,780 gene expressions having present calls hMLH1 gene is the principal defect identified in nonfamilial in at least 50% of the arrays. A projection was made on the largest three endometrial cancers with the MSI phenotype (24–26). In addition,

Cancer Res 2005; 65: (12). June 15, 2005 5032 www.aacrjournals.org

confirmed this change, indicating a more than 10-fold increase on average as compared with MSI-positive cancers (Fig. 3D). This gene resides on chromosomal location 8q23 and encompasses a region of over 160 kb and contains at least 78 exons. The predicted protein has the features of a cell membrane receptor but its function is unknown (28). Interestingly, the 8q22-24 region is frequently subjected to increased copy number in many endometrial cancers, in particular those endometrial cancers without MSI (29, 30), making PKHD1L1 a possible gene candidate for this region. We also analyzed the expression of PKHD1L1 in laser-captured epithelial, stroma, and myometrial cells from normal endometria and found expression primarily in the epithelial cells (data not shown). We also examined the levels of two negative regulators of WNT signaling, SFRP1 and SFRP4. Both transcripts were down-regulated as determined by the microarray. SFRP4 was previously reported as a stromal specific gene involved in the negative regulation of WNT signals (31). We noted a marked decrease in SFRP4 levels in MSI cancers as compared with normal endometria and with MSS cancers, and confirmed this by real-time PCR analysis (Fig. 3). Furthermore, we examined the tissue-specific nature of SFRP4 in laser-captured normal endometria and found transcripts in epithelial, stromal, and myometrial cell types (data not shown). In addition to SFRP4, the related family member SFRP1 gene was also more down-regulated in MSI cancers. We examined the relationship of down-regulation of this gene to promoter methylation status. SFRP1 and other secreted frizzled related transcripts have been previously identified in an epigenetic screen to identify transcripts silenced by hypermethylation in colon carcinoma (32). First, we examined the expression of SFRP1 in a Figure 1. Unsupervised analysis of early stage endometrioid endometrial small panel of endometrial carcinoma cell lines. The KLE and HEC- cancers. Projection on the first three principal components of highest variance. 59 cells expressed readily detectable transcripts, whereas SFRP1 Enclosing ellipsoids cover a 95% prediction range. transcript was not detectable in HEC-1-A or HOUA cells (data not shown). HEC-1-A cells were treated with the demethylating agent another transcript identified was the KIAA0766 gene, also known as 5-aza-2V-deoxycytidine and examined for reexpression of SFRP1. EPM2AIP1, located and transcribed from the opposite strand from SFRP1 expression was readily detected in 5-aza-2V-deoxycytidine– MLH1. Both these genes share the same CpG island that is treated cells, indicating some level of epigenetic control of this hypermethylated and silenced in many MSI cancers and both were locus either directly or indirectly. We next examined the DNA found down-regulated in MSI cancers (24, 27). These two sequence of the 5V region of the gene and designed primers that transcripts provide an important internal control lending (biolog- could distinguish between methylated and nonmethylated regions ical) validity to our results. In addition to identifying these following sodium bisulfite treatment. Methylation of the promoter important transcripts, we also show that hierarchical clustering was found to be coincident with a lack of expression of SFRP1 in also separated the cancers into two main groupings. A portion of these cell culture models. SFRP1-expressing cells contained these data are depicted in Fig. 2 with clear differences in some 109 unmethylated alleles of SFRP1 whereas nonexpressing cells were array features exhibiting at least 2-fold difference in gene methylated. We next applied this assay to the cancers in our study expression at high statistical significance (P < 0.001), and 537 to determine if promoter hypermethylation might be the features at least 1.5-fold at P < 0.005 (Supplementary data). The mechanism for the observed decreased expression of SFRP1 more expression of these transcripts was also compared with the average frequently noted in MSI cancers. We detected methylated alleles in expression of each in eight normal endometria. 8 of 12 MSI cancers whereas we detected methylation in only 2 of The expression of a subset of these differentially transcripts was 16 MSS cancers (P = 0.0031; Fig. 4). analyzed by real-time PCR to further validate our array results. In this report, we found that FGF18 transcripts were largely Real-time PCR confirmed the differential expression of polycystic undetected in normal endometria and in most MSS cancers but kidney and hepatic disease like 1 (PKHD1L1), heat shock 22 kDa were up-regulated in MSI cancers. The role of FGF18 is largely protein 8 (H11), secreted frizzled related protein 4 (SFRP4), and unexplored in carcinogenesis. Other FGF family members possess a fibroblast growth factor 18 (FGF18) in these cancers including variety of protumorigenic characteristics and a recent report three additional cases of each of MSI and MSS cancers, lending indicates that FGF18 promotes colon tumorigenesis and is also a further confirmation to our data set (Fig. 3). These genes may also direct target of an activated WNT pathway (33). Aberrant WNT contribute to the carcinogenic process in these cancers. signaling has been implicated in endometrial carcinogenesis We first examined the expression of the PKHD1L1 gene to and a subset of endometrial cancers contain activating mutations validate our array findings. This transcript, also known as of h-catenin (CTNNB1) or epigenetic silencing of APC (34–36). DKFZp586C1021, displayed the greatest fold change between MSS However, the downstream transcriptional targets of this pathway and MSI cancers based on our array data (Fig. 2). Real-time PCR have not been explored to any extent in the endometrium. www.aacrjournals.org 5033 Cancer Res 2005; 65: (12). June 15, 2005

These findings indicate significant differences in the aspects of in non-MSI endometrial cancers (34). The APC, FGF18, SFRP1, and SFRP4 WNT pathways between MSI and MSS cancers. Our data is SFRP4 data taken together support the idea that MSI endometrial consistent with the conclusion in a previous report that cancers may exhibit a more active WNT pathway than do MSS APC promoter methylation occurs more frequently in MSI than endometrial cancers. The necessity of the cancer cell to inactivate

Figure 2. Gene expression differences that distinguish MSI-H and MSS endometrial cancers. Hierarchical clustering of 109 transcripts differentially expressed (P < 0.001) with a threshold of at least 2-fold. Red, up-regulation; green, down-regulation.

Cancer Res 2005; 65: (12). June 15, 2005 5034 www.aacrjournals.org

Figure 3. Real time quantitative PCR validation of genes that distinguish MSI-H and MSS endometrial cancers. Mean MSS and MSI-H and normal endometria expressions of SFRP4 (A), FGF18 (B), H11 (C), and (D) PKHD1L1 with reference to h-actin.

multiple negative regulators of the same pathway is curious and estrogens, was reduced in MSI cancers (Figs. 2 and 3C). H11 may reflect our incomplete knowledge of the subtleties of the WNT expression is dependent on the presence of estrogen receptor in signaling pathways or may indicate that the epigenetic down- breast cancers and is a marker for estrogen receptor–positive cases regulations of these genes are less complete than perhaps (38). Although the function of H11 is unknown, it may be a mutational inactivation. An interesting observation is that biomarker for estrogen receptor cancers and its reduced expression inactivation of SFRP family members occurs even in colon cancers in MSI cases may be indicative of reduced estrogenic activity in this with mutation in downstream members of this pathway, such as subset. Relative levels of H11 were reduced in both MSI and MSS APC in colorectal cancer (37). APC mutation has not been cancers as compared with normal endometria; however, this described for endometrial carcinoma; however, some contain observation was more pronounced in MSI cancers. The data activating mutation of CTNNB1. Exploration of the intricacies of support a model in which estrogens are less important in the WNT signaling regarding the various WNT regulators in cancer development of MSI cancers, which may be driven by other factors may be facilitated by our findings in future studies. (perhaps WNT). A large body of literature support a role for unopposed estrogens To further examine the stringency of our data for binary class in the development of endometrioid endometrial carcinomas. comparison on the MSS and MSI groups, we tested the compound Interestingly, the H11 gene, which is under the tight control of covariate class predictor. This class prediction was done using the

Figure 4. Top, microarray expression values for SFRP1 in individual samples. Bottom, agarose gel electrophoresis of SFRP1 methylation–specific PCR for the same samples and controls, normal uterus, and Sss1 in vitro methylated genomic DNA.

www.aacrjournals.org 5035 Cancer Res 2005; 65: (12). June 15, 2005

392 genes that distinguished MSS and MSI by two-sample t test identified in both studies and seemed to be the only consistently (P < 0.001; ref. 39). The prediction was done by leaving one sample observed feature. out at a time for cross-validation and using all other samples for It is possible that the strong separation of MSI and MSS cancers classification. Based on 10,000 random permutations, the compound using microarrays is based on another underlying defect that covariate predictor was correct for 92% of the arrays with a P value cosegregates with the MSI phenotype or MLH1 defect. Subsequent of 0.001. The validity of this set of genes was further verified using analysis of the MSI-positive cancers in our panel indicated that three other classifiers including support vector machine, linear they all exhibited hypermethylation of the MLH1 gene promoter discriminant analysis, and the nearest centroid classifier (BRB Array whereas this gene was not methylated in any of the MSS group tools ver. 3.0c, Richard Simon, Amy Peng, Biometric research (data not shown). Perhaps the profile is indicative of an branch, National Cancer Institute, NIH; http://linus.nci.nih.gov/BRB- undescribed phenotype related to global methylation phenotype? ArrayTools.html), which also correctly classified 88% to 92% of the However, promoter methylation analysis of other genes on this MSI and MSS arrays, further validating the data. same group of cancers does not support the general contention It is perhaps somewhat surprising that such distinct groupings that MSI cancers strictly fall into a ‘‘methylator phenotype group’’. were determined from our unsupervised analysis because the Many cancers in the MSS group also contain methylated promoters underlying defect is in the DNA mismatch repair system and not in and MSI cancers have unmethylated ones.4 Additionally, it is not a transcription factor(s), which might be anticipated to give such clear whether an endometrial cancer with a mutation (rather than expression profiles. For example, estrogen receptor status defines epigenetic silencing) of MLH1 or of another mismatch repair gene distinct gene expression patterns in breast cancers (40). However, (i.e., MSH2 or MSH6) would exhibit a gene expression profile this phenomenon is not without precedent. During the preparation similar to the MSI cases described here or of a separate category. of our manuscript, microarray data that discriminate MSI-H, MSS, Our data suggest that the MSI and MSS early-stage endometrioid and MSI-L groups have been identified in colorectal cancers as endometrial cancers have distinct patterns of gene expression. The distinct global molecular phenotypes (41). Similar to our data gene expression data related to this phenotype can also be obtained in the endometrium, colon cancers designated MSI-H interpreted to support the idea that the carcinogenic program in and MSS were separated into distinct groupings by principal MSI cancers likely extends beyond the simple hypothesis that these component analysis. In addition, this report was able to further DNA repair–deficient cancers are solely the result of an identify the MSI-L group as a valid entity in colorectal carcinomas. accumulation of mutations in growth control genes. Importantly, We have found no evidence for the MSI-L group in this panel of we identified new molecular targets which may be useful in endometrial cancers and our principal component analysis understanding the etiology of this cancer type. A better under- supports this in that cases were either MSI-H or MSS. However, standing of the molecular pathogenesis of MSI and MSS it is possible that a MSI-L group is rare or may exist for endometrial endometrial cancers has the potential to facilitate improvements carcinomas in more advanced cases or in cases of nonendome- in the prevention and treatment of these cancers. trioid histology. We addressed the issue of whether the same set of genes were driving the molecular expression profiles of colon and endome- Acknowledgments trial MSI-H cancers. Mori et al. reported 20 genes in addition to Received 3/9/2004; revised 2/2/2005; accepted 3/7/2005. MLH1 as transcripts that most strongly drove the MSI-H The costs of publication of this article were defrayed in part by the payment of page phenotype in colorectal cancer. We therefore examined those 20 charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. genes (and also MLH1) which were present on the Affymetrix array chips and found that none were present on our list of differentially expressed genes even when we examined our data at a reduced statistical stringency (P < 0.005). However, MLH1 was 4 J.I. Risinger, unpublished data.

References 6. Kondo A, Safaei R, Mishima M, Niedner H, Lin X, of microsatellite instability in sporadic colorectal cancer. Howell SB. Hypoxia-induced enrichment and mutagen- Int J Colorectal Dis 2004;19:533–7. 1. Parsons R, Myeroff LL, Liu B, et al. Microsatellite esis of cells that have lost DNA mismatch repair. Cancer 12. Klump B, Nehls O, Okech T, et al. Molecular lesions in instability and mutations of the transforming growth Res 2001;61:7603–7. colorectal cancer: impact on prognosis? Original data h factor type II receptor gene in colorectal cancer. 7. Vaisman A, Varchenko M, Umar A, et al. The role of and review of the literature. Int J Colorectal Dis Cancer Res 1995;55:5548–50. hMLH1, hMSH3, and hMSH6 defects in cisplatin and 2004;19:23–42. 2. Myeroff LL, Parsons R, Kim SJ, et al. A transforming oxaliplatin resistance: correlation with replicative by- 13. Gervaz P, Buhler L, Scheiwiller A, Morel P. A tale of growth factor h receptor type II gene mutation common pass of platinum-DNA adducts. Cancer Res 1998;58: two colons and two cancers. Distinct carcinogenesis and in colon and gastric but rare in endometrial cancers with 3579–85. clinical outcome according to location proximal or microsatellite instability. Cancer Res 1995;55:5545–7. 8. Yanamadala S, Ljungman M. Potential role of MLH1 in distal to the splenic flexure. Swiss Surg 2003;9:3–7. 3. Ohwada M, Suzuki M, Saga Y, et al. Mutational the induction of p53 and apoptosis by blocking 14. Maxwell GL, Risinger JI, Alvarez AA, Barrett JC, analysis of transforming growth factor h receptor type II transcription on damaged DNA templates. Mol Cancer Berchuck A. Favorable survival associated with micro- and DNA mismatch repair genes in sporadic endome- Res 2003;1:747–54. satellite instability in endometrioid endometrial can- trial carcinomas with microsatellite instability. Oncol 9. Risinger JI, Berchuck A, Kohler MF, Watson P, Lynch cers. Obstet Gynecol 2001;97:417–22. Rep 2000;7:789–92. HT, Boyd J. Genetic instability of microsatellites in 15. Fiumicino S, Ercoli A, Ferrandina G, et al. Micro- 4. Hardman RA, Afshari CA, Barrett JC. Involvement of endometrial carcinoma. Cancer Res 1993;53:5100–3. satellite instability is an independent indicator of mammalian MLH1 in the apoptotic response to 10. Gleich LL, Wang J, Gluckman JL, Fenoglio-Preiser recurrence in sporadic stage I-II endometrial adenocar- peroxide-induced oxidative stress. Cancer Res 2001;61: CM. Microsatellite instability as a predictor of survival cinoma. J Clin Oncol 2001;19:1008–14. 1392–7. in head and neck cancer—is there a link with colon 16. Calle EE, Rodriguez C, Walker-Thurmond K, Thun 5. Rizki A, Lundblad V. Defects in mismatch repair cancer? ORL J Otorhinolaryngol Relat Spec 2003;65: MJ. Overweight, obesity, and mortality from cancer in a promote telomerase-independent proliferation. Nature 193–8. prospectively studied cohort of U.S. adults. N Engl J Med 2001;411:713–6. 11. Lim SB, Jeong SY, Lee MR, et al. Prognostic significance 2003;348:1625–38.

Cancer Res 2005; 65: (12). June 15, 2005 5036 www.aacrjournals.org

17. Calle EE, Kaaks R. Overweight, obesity and cancer: 25. Gurin CC, Federici MG, Kang L, Boyd J. Causes and 33. Shimokawa T, Furukawa Y, Sakai M, et al. Involve- epidemiological evidence and proposed mechanisms. consequences of microsatellite instability in endome- ment of the FGF18 gene in colorectal carcinogenesis, as Nat Rev Cancer 2004;4:579–91. trial carcinoma. Cancer Res 1999;59:462–6. a novel downstream target of the h-catenin/T-cell factor 18. Lax SF. Molecular genetic pathways in various types 26. Simpkins SB, Bocker T, Swisher EM, et al. MLH1 complex. Cancer Res 2003;63:6116–20. of endometrial carcinoma: from a phenotypical to a promoter methylation and gene silencing is the primary 34. Zysman M, Saka A, Millar A, Knight J, Chapman W, molecular-based classification. Virchows Arch 2004;444: cause of microsatellite instability in sporadic endome- Bapat B. Methylation of adenomatous polyposis coli in 213–23. trial cancers. Hum Mol Genet 1999;8:661–6. endometrial cancer occurs more frequently in tumors 19. Risinger JI, Maxwell GL, Berchuck A, Barrett JC. 27. Herman JG, Umar A, Polyak K, et al. Incidence and with microsatellite instability phenotype. Cancer Res Promoter hypermethylation as an epigenetic compo- functional consequences of hMLH1 promoter hyper- 2002;62:3663–6. nent in Type I and Type II endometrial cancers. Ann N Y methylation in colorectal carcinoma. Proc Natl Acad Sci 35. Fukuchi T, Sakamoto M, Tsuda H, Maruyama K, Acad Sci 2003;983:208–12. U S A 1998;95:6870–5. Nozawa S, Hirohashi S. h-Catenin mutation in carcino- 20. Risinger JI, Maxwell GL, Chandramouli GV, et al. 28. Hogan MC, Griffin MD, Rossetti S, Torres VE, Ward ma of the uterine endometrium. Cancer Res 1998;58: Microarray analysis reveals distinct gene expression CJ, Harris PC. PKHDL1, a homolog of the autosomal 3526–8. profiles among different histologic types of endometrial recessive polycystic kidney disease gene, encodes a 36. Kobayashi K, Sagae S, Nishioka Y, Tokino T, Kudo R. cancer. Cancer Res 2003;63:6–11. receptor with inducible T lymphocyte expression. Hum Mutations of the h-catenin gene in endometrial 21. Berg KD, Glaser CL, Thompson RE, Hamilton SR, Mol Genet 2003;12:685–98. carcinomas. Jpn J Cancer Res 1999;90:55–9. Griffin CA, Eshleman JR. Detection of microsatellite 29. Suzuki A, Fukushige S, Nagase S, Ohuchi N, Satomi S, 37. Suzuki H, Watkins DN, Jair KW, et al. Epigenetic instability by fluorescence multiplex polymerase chain Horii A. Frequent gains on chromosome arms 1q and/or inactivation of SFRP genes allows constitutive WNT reaction. J Mol Diagn 2000;2:20–8. 8q in human endometrial cancer. Hum Genet 1997;100: signaling in colorectal cancer. Nat Genet 2004;36: 22. Eisen MB, Spellman PT, Brown PO, Botstein D. 629–36. 417–22. Cluster analysis and display of genome-wide expression 30. Hirasawa A, Aoki D, Inoue J, et al. Unfavorable 38. Charpentier AH, Bednarek AK, Daniel RL, et al. patterns. Proc Natl Acad Sci U S A 1998;95:14863–8. prognostic factors associated with high frequency of Effects of estrogen on global gene expression: identifi- 23. Boland CR, Thibodeau SN, Hamilton SR, et al. A microsatellite instability and comparative genomic cation of novel targets of estrogen action. Cancer Res national cancer institute workshop on microsatellite hybridization analysis in endometrial cancer. Clin 2000;60:5977–83. instability for cancer detection and familial predisposi- Cancer Res 2003;9:5675–82. 39. Radmacher MD, McShane LM, Simon R. A paradigm tion: development of international criteria for the 31. Abu-Jawdeh G, Comella N, Tomita Y, et al. Differential for class prediction using gene expression profiles. determination of microsatellite instability in colorectal expression of frpHE: a novel human stromal protein of the J Comput Biol 2002;9:505–11. cancer. Cancer Res 1998;58:5248–57. secreted frizzled gene family, during the endometrial cycle 40. Perou CM, Sorlie T, Eisen MB, et al. Molecular 24. Esteller M, Levine R, Baylin SB, Ellenson LH, and malignancy. Lab Invest 1999;79:439–47. portraits of human breast tumours. Nature 2000;406: Herman JG. MLH1 promoter hypermethylation is 32. Suzuki H, Gabrielson E, Chen W, et al. A genomic 747–52. associated with the microsatellite instability pheno- screen for genes up-regulated by demethylation and 41. Mori Y, Selaru FM, Sato F, et al. The impact of type in sporadic endometrial carcinomas. Oncogene histone deacetylase inhibition in human colorectal microsatellite instability on the molecular phenotype of 1998;17:2413–7. cancer. Nat Genet 2002;31:141–9. colorectal tumors. Cancer Res 2003;63:4577–82.

www.aacrjournals.org 5037 Cancer Res 2005; 65: (12). June 15, 2005

Downloaded from cancerres.aacrjournals.org on October 1, 2021. © 2005 American Association for Cancer Research. Gene Expression Profiling of Microsatellite Unstable and Microsatellite Stable Endometrial Cancers Indicates Distinct Pathways of Aberrant Signaling

John I. Risinger, G. Larry Maxwell, Gadisetti V.R. Chandramouli, et al.

Cancer Res 2005;65:5031-5037.

Updated version Access the most recent version of this article at: http://cancerres.aacrjournals.org/content/65/12/5031

Cited articles This article cites 41 articles, 19 of which you can access for free at: http://cancerres.aacrjournals.org/content/65/12/5031.full#ref-list-1

Citing articles This article has been cited by 6 HighWire-hosted articles. Access the articles at: http://cancerres.aacrjournals.org/content/65/12/5031.full#related-urls

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://cancerres.aacrjournals.org/content/65/12/5031. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.