Microsatellite Instability, MLH1 Promoter Methylation, and Loss of Mismatch Repair in Endometrial Cancer and Concomitant Atypical Hyperplasia

Gynecologic Oncology 86, 62– 68 (2002) doi:10.1006/gyno.2002.6724

Microsatellite Instability, MLH1 Promoter Methylation, and Loss of Mismatch Repair in Endometrial Cancer and Concomitant Atypical Hyperplasia N. Horowitz,* K. Pinto,† D. G. Mutch,* T. J. Herzog,* J. S. Rader,* R. Gibb,* T. Bocker-Edmonston,‡ and P. J. Goodfellow§ ,1 *Department of Obstetrics and Gynecology, †Department of Pathology and Immunology, and §Department of Surgery, Washington University School of Medicine, St. Louis, Missouri 63110; and ‡Kimmel Cancer Center, Thomas Jefferson University School of Medicine, Philadelphia, Pennsylvania 19107 Received January 7, 2002

tumorigenesis. Microsatellite instability (MSI) is a frequent defect in endometrial cancers. Approximately 25% of endometrial carcinomas have an MSI-high (MSI-H) phenotype because of a defect in DNA mismatch repair [1–3]. Loss of DNA mismatch repair leads to an increased mutation rate, and it is believed that in tumor cells there is strong selection for mutations in genes critical for growth control. Although loss of DNA mismatch repair may contribute to mutation in the PTEN tumor suppressor [4 – 8], the targets for the tumorigenic effects of MSI in endometrial cancers remain obscure [9]. Defective DNA mismatch repair and MSI in endometrial cancers is rarely due to mutation in DNA repair genes. Methylation of the MLH1 promoter accounts for most of the defective DNA mismatch repair that is seen in these malignancies. Detailed investigation of the methylation of the MLH1 promoter region in colorectal cancer cells defined a critical region located ⫺248 to ⫺148 bp upstream of the transcription start site [10]. This region is unmethylated in cell lines that express MLH1, but invariably methylated in MSI-H (high-level MSI defined by aberrant PCR products in ⱖ2 of 5 markers investigated) cell lines that lack immunodetectable MLH1 protein. Approximately 80% of MSI-H endometrial cancers (tumors with a defect in DNA mismatch repair) have methylation in the MLH1 promoter [9, 11, 12]. Immunohistochemistry showed that methylation correlates with loss of immunodetectable MLH1 in endometrial cancers [12]. At present it is unclear whether methylation of the MLH1 promoter triggers gene inactivation or simply participates in the maintenance of gene silencing once it has been established through some as yet unidentified mechanism. Atypical endometrial hyperplasia is believed to be a precursor to type I endometrial cancer [13]. This histologically recognized precancerous lesion provides an opportunity to investigate changes in the genetic makeup of the cell that occur early in endometrial tumorigenesis. Several groups have demonstrated MSI in atypical hyperplasia [7, 14, 15]. Work from our laboratory suggests that loss of mismatch repair may occur

Objective. MLH1 methylation is associated with the microsatellite instability (MSI) phenotype in endometrial cancer and atypical endometrial hyperplasia, a premalignant precursor to carcinoma. The observation that methylation is also seen in atypical endometrial hyperplasia without MSI suggests that methylation is an early event in endometrial tumorigenesis. Our objective was to determine if methylation is always present in MSI-positive atypical hyperplasia concomitant with MSI-positive, methylation-positive carcinoma. Methods. We used laser capture microdissection to study MLH1 methylation and MSI in a large series of endometrial cancer cases that had previously been shown to have methylation and the MSI-high (MSI-H) phenotype. We resampled areas of carcinoma from 27 patients along with 51 foci of concomitant atypical endometrial hyperplasia. Results. Consistent with previous reports, we saw MLH1 methylation in areas of atypical endometrial hyperplasia that did not show MSI. In addition, we noted that 18% of the MSI-H atypical endometrial hyperplasia DNAs lacked methylation of critical cytosines in the MLH1 promoter. Immunohistochemistry studies showed that these MSI-H unmethylated foci of atypical endometrial hyperplasia failed to express MLH1, as did regions of simple hyperplasia. Conclusion. Methylation of the MLH1 promoter is an early event in endometrial tumorigenesis. Given that not all MSI-positive tissues had methylation at cytosines ⴚ229 and ⴚ231, it appears that methylation may not be required for MLH1 silencing and loss of mismatch repair. © 2002 Elsevier Science (USA) Key Words: endometrial carcinoma; atypical endometrial hyperplasia; MLH1 methylation; microsatellite instability.

INTRODUCTION Over the past several years significant advances have been made in defining genetic factors that contribute to endometrial 1

To whom correspondence and reprint requests should be addressed at Department of Surgery, Washington University School of Medicine, Campus Box 8109, 660 South Euclid Avenue, St. Louis, MO 63106-1029. Fax: (314) 362-8620. E-mail: [email protected]. 0090-8258/02 $35.00 © 2002 Elsevier Science (USA) All rights reserved.

62

MSI AND MLH1 METHYLATION IN ENDOMETRIAL HYPERPLASIA

prior to mutation in KRAS2 and that in some cases the precursor lesions may be genetically heterogeneous [14]. When methylation occurs in the progression from precancerous to cancer remains to be determined. Esteller and colleagues [16] investigated atypical hyperplasia for MLH1 methylation and MSI. Based on a limited number of samples in which carcinoma and hyperplasia occurred concurrently, these authors concluded that MLH1 promoter methylation is an early event and in some cases may precede a detectable MSI phenotype. We investigated MSI and MLH1 promoter methylation in a large series of endometrial cancer cases with concomitant atypical endometrial hyperplasia in an attempt to better understand the relative timing of MLH1 promoter methylation in endometrial tumorigenesis. MATERIAL AND METHODS Tissue specimens. Tumor and normal tissues were obtained from patients being treated at Washington University School of Medicine for primary endometrial cancer (Human Studies Committee approval 93-0828). Tumor DNA was evaluated for microsatellite instability, and for all MSI-H cases, the methylation status of the MLH1 promoter was determined [12, 17]. For all MSI-H, MLH1 methylation-positive cases the archived pathology specimens were retrieved and evaluated by a qualified gynecologic pathologist (K.P.) for the presence or absence of concomitant hyperplasia. If atypical hyperplasia was evident, the pathology block was retrieved, and slides with serial 5-␮m sections were prepared. One slide was hematoxylin and eosin stained and coverslipped. Three slides were left unstained until the time they would be used for laser capture microdissection. Foci of endometrial carcinoma and hyperplasia were marked on the coverslipped H&E-stained slides. The marked slides were then used to help guide the laser capture microdissection of foci of atypical hyperplasia and carcinoma from adjacent tissue sections (see below). Microsatellite instability analysis. MSI analysis was performed as previously described [2, 14, 18]. DNAs from the tumor tissue collected for research and from the microdissected foci of carcinoma and hyperplasia were genotyped for the BAT-25, BAT-26, D5S346, D2S123, and D17S250 markers. A sample was classified as MSI-H if two or more markers showed instability, MSI-stable (MSI-S) if no instability was noted, and MSI-low (MSI-L) if a single marker revealed novel bands. Laser capture microdissection (LCM). We used an Arcturus PixCell II LCM apparatus in the Tumor Procurement Core at the Siteman Cancer Center, Barnes Jewish Hospital/Washington University School of Medicine, to isolate hyperplastic lesions. The endometrial carcinoma was also resampled by laser capture microdissection. The areas of carcinoma dissected by LCM ranged from 10 to 15 mm 2, while foci of atypical hyperplasia were 1.5–3.0 mm 2. Images of the stained

63

sections were obtained before and after capture. DNA was prepared from LCM tissue as described previously [14]. Sodium bisulfite conversion. Bisulfite conversion of DNA isolated from the tissues prepared by laser capture was performed using The CpGenome modification kit (Intergen Company, Purchase, NY) and the manufacturer’s recommended protocols. The quantity of DNA converted was estimated to be 10 –20 ng. Methylation assay. Two rounds of PCR were used to amplify the bisulfite converted DNAs. A schema for the methylation assay is shown in Fig. 1. The amplification primers used are insensitive to the methylation status of starting DNA. A 115-bp product, located in the critical methylation region “C” [10], was generated. The amplicon contains a potential BstUI (CGCG) site. BstUI restricts the amplification product only if both cytosines in the recognition site were methylated in the starting DNA. Following digestion with BstUI, diagnostic fragments of 115 (unmethylated) or 83 bp (methylated) were resolved on 6 –10% polyacrylamide gels and visualized by autoradiography. For some analyses, the reverse nested primer was end-labeled with 32P and the restriction products were detected directly. To reduce nonspecific background, restriction products separated on 10% gels were transferred to nylon membrane (GeneScreen Plus, NEN Life Science Products, Inc., Boston, MA) and hybridized with a radiolabeled 24-mer probe. The probe used was a pair of oligonucleotides that recognize both methylated and unmethylated amplification products. Sequences for the PCR primers and the 24-mer hybridization probes are as follows: primers for PCR amplification were external forward 5⬘-TTTTTAATTTTGTGGGTTGTTGGG-3⬘, external reverse 3⬘-AAAAACCACAAAAACAAAACCAA-5⬘, nested forward 5⬘-TGTTCGTTATTTAGAAGGATATG-3⬘, and nested reverse 3⬘-TCTACTCCTATTAACTAAATATTTC-5⬘; primers used as hybridization probe were methylated probe 5⬘-TTTATTGGTTGGTTTTTAGGAGTT-3⬘ and unmethylated probe 5⬘-TTTATTGGTTAGTTTTTAGGAGTT-3⬘. Conditions for the PCR assay were as follows: 0.8 ␮M primers, 0.1 mM dNTPs, 1.5 mM MgCl, 0.1 U AmpliTaq polymerase (Perkin-Elmer Cetus, Norwalk, CT). For the nested reaction, the concentration of MgCl was increased to a total of 7.5 mM. Thirty-five cycles of PCR were performed for each round. The external reaction was run at 94, 55, and 72°C for 1 min each while the nested reaction was run at 94, 60, and 72°C again for 1 min each. Immunohistochemistry. MLH1 immunohistochemistry was performed as described previously [12]. Paraffin sections 5 ␮m thick were mounted on charged slides. After routine deparaffinization and rehydration, antigen retrieval was performed by microwaving for 10 min (100 W) in Citra Plus solution (BioGenex, San Ramon, CA). Slides were then incubated overnight with primary antibody [1:200 dilution PharMingen (San Diego, CA) monoclonal antibody G168-728] to hMLH1. This was

64

HOROWITZ ET AL.

followed by a 30-min incubation in biotinylated secondary antibody and a 30-min incubation in ABC complex (LSAB kit, Dako, Carpintera, CA). Each incubation was followed by a phosphate-buffered saline wash. Finally DAB/peroxide (Dako) was applied for 3–7 min, the slides were counterstained by hematoxylin, and the pattern of staining was evaluated. RESULTS We compared the MSI and MLH1 methylation status of endometrial cancers and precancers in a series of patients with concomitant carcinoma and atypical endometrial hyperplasia. The carcinoma DNA was initially investigated using unfixed tumor tissue collected at the time of surgery for research purposes. For these studies we resampled areas of carcinoma and isolated atypical hyperplasia from archived formalin-fixed pathology tissue blocks. Our collection of over 300 prospectively acquired endometrial cancers includes 87 tumors classified as MSI-H. Of these 87 cases, 62 (71%) were shown to have methylation of the MLH1 promoter. Pathology reports from these 62 MSI-H and methylation-positive cancers indicated that 37 arose in a background of atypical endometrial hyperplasia. Following a review of the hematoxylin and eosin-stained slides on file, and the additional sections prepared from the archived paraffin blocks (all reviewed by K.P.), 10 cases were excluded from the study (7 without identifiable atypical hyperplasia and 3 for which paraffin blocks could not be located). A total of 27 MSI-H and methylation-positive cases were available to investigate MSI and MLH1 promoter methylation in atypical hyperplasia concomitant with carcinoma. From these cases we captured 27 areas of carcinoma and 51 foci of hyperplasia (mean, 2 per patient) and prepared DNA from each. Most of the LCM-prepared carcinoma and hyperplasia DNA specimens showed MSI as had been the case for the tumor specimen studied earlier (initial MSI typings performed using DNA prepared from tumor tissues collected for research studies). Twenty-six of 27 carcinomas (96%) and 46 of 51 (90%) atypical hyperplasias were MSI-H (Table 1). Representative examples of MSI analysis for samples 1390 and 1257 are shown in Fig. 2. Although all 27 of the tumor DNAs prepared from tissues collected at the time of surgery were MSI-H and MLH1 methylation positive, a number of the LCM carcinoma and hyperplasia specimens were unmethylated. Twenty-three of 27 (85%) LCM carcinoma and 41 of 51 (80%) LCM hyperplasia specimens had MLH1 methylation (Table 1). A DNA specimen was classified as methylation positive when the 83-bp BstUI (CGCG recognition site) restriction product was evident. Some specimens restricted fully (only the 83-bp product evident) suggesting complete methylation in the carcinoma or atypical hyperplasia DNA, while others showed only the unrestricted 115-bp product (no methylation) or a combination of 115- and 83-bp restriction products. Methylation analyses were

TABLE 1 Patterns of MSI and MLH1 Promoter Methylation in LCM Prepared Endometrial Carcinoma and Concomitant Atypical Hyperplasia Atypical hyperplasia specimen Case identifier

Carcinoma

Focus 1

0037 0064 1070 1089 1094 1102 1128 1135 1141 1144 1150 1155 1158 1174 1185 1231 1267 1318 1330 1334 1363 1390 1411

H/⫹ H/⫹ H/⫹ H/⫹ H/⫹ H/⫹ H/⫹ H/⫹ H/⫹ H/⫹ H/⫹ H/⫹ H/⫹ H/⫹ H/⫹ H/⫹ H/⫹ H/⫹ H/⫹ H/⫹ H/⫹ H/⫹ H/⫹

H/⫹ H/⫹ H/⫹ H/⫹ H/⫺* H/⫹ H/⫹ H/⫹ H/⫹ H/⫹ H/⫹ H/⫹ H/⫹ H/⫹ H/⫹ H/⫹ H/⫹ H/⫹ H/⫹ H/⫹ H/⫹ H/⫹ H/⫺*

1061 1257 1368

H/⫺* H/⫺* H/⫺*

1235

S/⫺*

Focus 2

Focus 3

H/⫹

L/⫹* H/⫺* H/⫺* H/⫹ S/⫹* H/⫹ H/⫹ H/⫹ H/⫹ H/⫹ H/⫹

H/⫹ H/⫺* H/⫹ H/⫹

H/⫺* H/⫺*

H/⫹

H/⫹* H/⫺ H/⫺

H/⫺

S/⫹*

H/⫹*

S/⫹*

H/⫹* L/⫹*

Note. MSI state was reported for 0037, 0064, 1070, 1094, 1102, 1128, 1135, 1141, 1144, 1150, 1158, 1174, 1231, 1061, and 1235 specimens in Cohn et al. [14]. H/⫹ ⫽ MSI-high and MLH1 promoter methylation positive. H/⫺ ⫽ MSI-high and MLH1 promoter methylation negative. S/⫹ ⫽ MSI-stable and MLH1 promoter methylation positive. S/⫺ ⫽ MSI-stable and MLH1 promoter methylation negative. L/⫹ ⫽ MSI-low (instability at a single marker) and MLH1 promoter methylation positive. *Discrepancy with respect to MSI or methylation status.

performed twice, starting with the first-round PCR amplification. For those cases in which the methylation was negative, reactions were repeated beginning at the bisulfite conversion stage. Representative examples of the MLH1 promoter methylation analyses are shown in Fig. 3. There were two major differences in methylation and MSI status in the LCM-prepared and previously studied DNAs. Some of the DNAs prepared from the pathology tissue blocks were shown to be MSI-H but lacked MLH1 methylation. Thirteen of 72 (18%) LCM-prepared MSI-H specimens were unmethylated. The 13 unmethylated LCM-prepared DNAs came from 11 cases. Three were carcinoma and 10 were hyperplasia specimens. The second major difference was that 5 LCM-prepared DNAs (one each from cases 1061, 1094, and

MSI AND MLH1 METHYLATION IN ENDOMETRIAL HYPERPLASIA

65

FIG. 1. Schematic representation of the PCR amplification/restriction digestion assay to assess methylation in the MLH1 promoter. The location of the first-round primers (external forward and reverse) is approximate. Following the second round of amplification, restriction with BstUI gives rise to either 32- and 83-bp fragments (template DNA methylated) or the uncut 115-bp product (either CpG in the BstUI site unmethylated). Second-round (nested) PCR products were visualized by autoradiography. FIG. 2. Representative MSI analysis. MSI typing DNAs from case 1390 and case 1257. (A) D17S250 genotypes. (B) BAT-26 genotypes. Arrowheads indicate novel PCR products and arrows point to the normal allelic fragments. FIG. 3. Representative methylation analyses. Restriction products were detected by hybridization with an internal oligonucleotide probe. Lane 1, case 1150 atypical endometrial hyperplasia, focus 2; lane 2, case 1158 atypical endometrial hyperplasia, focus 3; lane 3, case 64 atypical endometrial hyperplasia, focus 1; lane 4, case 1257 atypical endometrial hyperplasia. Lanes 3 and 4, examples with complete methylation and no methylation at the BstUI site, respectively.

1144, and 2 from 1235) were found to be MSI-S or MSI-L but had MLH1 methylation (Table 1). Ten of the 51 atypical hyperplasia DNAs (20%) examined were not methylated at the BstUI site in the MLH1 promoter (Table 1). There were also three (10%) LCM carcinoma specimens (1061, 1257, and 1368) that were MSI-H but did not have MLH1 promoter methylation (as noted earlier, the DNA prepared from tumor collected for research was methylation positive). We were able to investigate six foci of hyperplasia associated with these three cases in which the carcinoma DNA lacked methylation: three from 1061, one from 1257, and two from 1368. For all three of these cases, at least one focus of atypical hyperplasia was like the LCM carcinoma in that it was

MSI-H but lacked MLH1 methylation. For two of the three cases (1061 and 1368), one focus of hyperplasia was found to be both MSI-H and methylation positive (like the tumor DNA initially investigated) and molecularly distinct (MLH1 methylation state) from the LCM-prepared carcinoma from the same patient. Seven additional MSI-H and methylation-negative foci of hyperplasia were identified (Table 1). In each case the corresponding research tumor specimen (fresh-frozen tissue) and LCM carcinoma were MSI-H and methylation positive. These foci of hyperplasia again represent molecularly distinguishable populations. Three cases (1094, 1235, and 1411) were associated with atypical hyperplasia in which molecular features (MSI and

66

HOROWITZ ET AL.

FIG. 4. MLH1 immunohistochemistry in case 1363 endometrium. (A) Atypical endometrial hyperplasia fails to stain. Note: faint brown stain in the stroma. (B) Simple hyperplasia showing a mixture of MLH1-positive (brown staining, black arrowhead) and MLH1-negative glands (white arrowheads). Bar, 100 ␮m.

MLH1 methylation) were distinct from both the LCM carcinoma or the research tumor DNA. However, for all other cancer cases, at least one focus of hyperplasia was found with a molecular signature identical to the concomitant LCM carcinoma or research tumor DNA. Expression of MLH1 was assessed by immunohistochemistry (IHC) in two cases in which MLH1 methylation patterns in atypical hyperplasia differed from the tumor DNA initially studied (case 1363 and case 1390). The carcinoma and three foci of hyperplasia used to prepare DNA did not stain with the MLH1 antibody. A representative area in which atypical hyperplasia did not stain for MLH1 is shown in Fig. 4A. In addition, regions of simple hyperplasia also failed to stain for MLH1 (Fig. 4B). DISCUSSION Over the past several years there has been an increasing awareness that epigenetic changes contribute to a variety of malignancies [19]. Methylation of the MLH1 gene is an important epigenetic alteration in colon, endometrial, and a number of other tumors [11, 12, 20, 21]. Loss of DNA mismatch repair (through methylation of the MLH1 promoter or mutation in MLH1 or another DNA mismatch repair family member) may be a critical early step in some tumors [22]. Esteller and colleagues [10] assessed MLH1 promoter methylation in normal endometrium and four different histologically defined proliferative lesions: simple endometrial hyperplasia, complex endometrial hyperplasia, atypical endometrial hyperplasia, and endometrial carcinoma. In their study, 91% of MSI-H carcinomas showed methylation of the MLH1 promoter, while the rates of methylation for the normal endometrium, simple endometrial hyperplasia, complex hyperplasia, and atypical hy-

perplasia were much lower: 0, 0, 3, and 33%, respectively. Of over 100 hyperplasias that were investigated, only 6 were associated with endometrial cancer in the same patient. Interestingly, despite the fact that all 6 of these hyperplasias had MLH1 methylation, only 4 were MSI-H (2 were MSI-S). Based on these findings, the authors concluded that methylation of MLH1 is an early event in endometrial tumorigenesis and that methylation of MLH1 may precede the MSI phenotype in a subset of hyperplasias. Our study of 27 endometrial cancer cases and 51 foci of concomitant atypical hyperplasia further supports the findings of Esteller et al. [16] and strengthens the argument that MLH1 methylation is an early event in the progression from normal endometrium to carcinoma. Ten percent (5 of 51) of the atypical hyperplasias we studied were classified as MSI-S (N ⫽ 3) or MSI-L (N ⫽ 2) but had MLH1 methylation. The methylation-positive, MSI-S, or MSI-L status of these five DNAs is consistent with a model in which methylation precedes MSI. Furthermore, we demonstrated by IHC that MLH1 protein is absent in simple endometrial hyperplasia concomitant with carcinoma, further supporting the notion that alterations in DNA mismatch repair occur early in endometrial tumorigenesis. Eighty-five percent of the carcinoma and 71% of the atypical hyperplasia DNAs we prepared using tissue isolated by LCM were MSI-H and exhibited methylation of the MLH1 promoter. Although MLH1 methylation may be correlated with loss of mismatch repair (MSI) in endometrial cancer, our study suggests that methylation of MLH1 may not be necessary for gene silencing. From the 27 endometrial cancer cases we examined, 13 MSI-H tissues were identified that were not methylated. The discordance between MSI and methylation states occurred in 3

MSI AND MLH1 METHYLATION IN ENDOMETRIAL HYPERPLASIA

carcinomas and 10 atypical hyperplasias. There are several explanations for why these 13 MSI-H DNAs might not have MLH1 methylation. First, loss of DNA repair could be attributable to a mechanism other than MLH1 methylation (i.e., genetic mutation). Alternatively, a defect in a different mismatch repair gene could account for the MSI. None of the cases in our series had features suggestive of inherited disease which would predispose to multiple precancers, and in the case of a germline MLH1 defect could lead to different “second” events in different lesions. It is also unlikely that another DNA repair gene accounts for MSI in the tissues we studied. The foci of atypical hyperplasia in case 1363 and 1390 that were unmethylated both failed to express MLH1 based on immunohistochemical staining. One possible explanation for the discordant MSI-H/methylation-negative state is that methylation is heterogenous. Methylation of the two cytosines (position ⫺229 and ⫺231) we studied may not be required for MLH1 inactivation. Other methylation events could be sufficient for gene silencing. Further methylation analysis (sequence-based) could determine if there is methylation of other CpG pairs in these 13 specimens that are MSI-H but unmethylated for the BstUI site we examined. A sequenced-based approach to assessing methylation, while technically possible, would be difficult given the limited tissues available. Regardless, our studies suggest that within the same uterus there are multiple, molecularly distinct clones (different methylation patterns) of hyperplasia. As discussed previously, five atypical hyperplasia specimens in our series had methylation of the MLH1 promoter yet did not exhibit the MSI-H phenotype. These specimens suggest that methylation may precede the loss of mismatch repair. Alternately, the number of mutations attributable to defective DNA mismatch may be few in a precancer (for which there would have been fewer cell divisions than for a frank carcinoma) and microsatellite instability may have gone undetected. We used the five reference panel markers established for the evaluation of MSI in colorectal cancers [5] to test these foci of hyperplasia. Perhaps with the use of additional markers we might reclassify these foci of hyperplasia as MSI-H. Immunohistochemistry might also reveal that such methylated MSI-S lesions do not express MLH1, as was demonstrated for the atypical hyperplasia foci lacking methylation, but manifesting MSI. Early loss of mismatch repair gives rise to a mutator phenotype as described by Loeb [23], thus enabling the multistep process of carcinogenesis to proceed. Our observation that hyperplastic glands lacking atypia (simple hyperplasia) do not stain for MLH1 is indicative of genetically abnormal precancerous cell populations. Evaluation of the MLH1 methylation status or IHC may thus provide a more sensitive test for the early identification of these genetically aberrant but cytologically bland cells that are likely to progress to cancer. Although we selected the cases in our series based on the fact that the research tumor specimen was MSI-H and MLH1 methylated, there was one case in our series (1235) in which the LCM carcinoma was MSI-S and methylation negative.

67

Evaluating the three foci of concomitant atypical hyperplasia isolated from this patient’s hysterectomy specimen we found one focus to be MSI-H and methylation positive, one to be MSI-S but methylation positive, and one MSI-L and methylation positive. Molecularly, the MSI-H and methylation-positive hyperplasia appears to be the precursor lesion of the research tumor specimen. With additional cell divisions and accumulation of microsatellite alterations, the other two foci of hyperplasia might have become molecularly indistinguishable from the research tumor specimen. The MSI-S and methylation-negative LCM carcinomas appear to represent genetically distinct clones of cancer that developed via a different pathway than the research tumor. Isolation of additional foci of atypical hyperplasia by microdissection might reveal an MSI-S, methylation-negative precursor to the LCM carcinoma from case 1235. In general, resampling would likely provide further evidence of tumor heterogeneity. It is possible that the foci of LCM carcinoma and atypical hyperplasia that were not classified as MSI-H or methylation positive do not represent molecularly distinct clones, but rather were the result of isolating normal tissues rather than foci of carcinoma or atypical hyperplasia. Using laser capture microdissection and relying upon a gynecologic pathologist to select areas of cancer and hyperplasia helped ensure that appropriate tissues were captured and that they represented pure cell populations. When samples were found to be methylation negative, assays were repeated from the bisulfite conversion stage to confirm our findings. Furthermore, immunohistochemistry was performed to show absence of MLH1 staining in two of the methylation-negative foci. We recognize that our methylation assay did not evaluate all CpG pairs in the region critical for inactivation of MLH1. The assay, based on the work by Deng et al. [10], corresponds to the CpG island located ⫺248 to ⫺178 bp proximal to the transcription start site and we assessed only the BstUI site, much as was done by Esteller et al. [16]. In conclusion, our study confirms that MSI and methylation of the MLH1 promoter are early events in the tumorigenesis of sporadic endometrial carcinoma, and for some atypical hyperplasias methylation may precede MSI. Furthermore, our data suggest that methylation of the BstUI site in the region at ⫺229 and ⫺231 bp upstream of the MLH1 transcription start site may not be required for gene silencing. The observed patterns of MSI and MLH1 methylation are consistent with a model in which carcinoma and atypical hyperplasia are polyclonal with multiple genetically distinct populations within the same uterus. Whether these genetically distinct populations are biologically different remains to be determined. ACKNOWLEDGMENTS This work was supported by CA71754 (P.J.G.) and P30 CA91842 to the Alvin J. Siteman Cancer Center at Washington University. We thank Dr. Mark Watson, Siteman Cancer Center Tumor Procurement Core, for his assistance with laser capture microdissection, and Drs. Richard Fishel and Juan P. Palazzo, Kimmel Cancer Center, Thomas Jefferson University, for their help

68

HOROWITZ ET AL.

with immunohistochemistry. We also thank Tonia Thompson for typing this manuscript and Christina Menke for helping with the MSI analyses.

REFERENCES 1. Burks RT, Kessis TD, Cho KR, Hedrick L. Microsatellite instability in endometrial carcinoma. Oncogene 1994;9:1163– 6. 2. Kowalski LD, Mutch DG, Herzog TJ, Rader JS, Goodfellow PJ. Mutational analysis of MLH1 and MSH2 in 25 prospectively-acquired RER ⫹ endometrial cancers. Genes Chromosomes Cancer 1997;18:219 –27. 3. Risinger JI, Berchuck A, Kohler MF, Watson P, Lynch HT, Boyd J. Genetic instability of microsatellites in endometrial carcinoma. Cancer Res 1993;53:5100 –3. 4. Ali IU. Gatekeeper for endometrium: the PTEN tumor suppressor gene. J Natl Cancer Inst 2000;92:861–3. 5. Boland CR, Thibodeau SN, Hamilton SR, Sidransky D, Eshleman JR, Burt RW, Meltzer SJ, Rodriguez-Bigas MA, Fodde R, Ranzani N, Srivastava S. A National Cancer Institute workshop on microsatellite instability for cancer detection and familial predisposition: development of international criteria for the determination of microsatellite instability in colorectal cancer. Cancer Res 1998;58:5248 –57. 6. Kong D, Suzuki A, Zou TT, Sakurada A, Kemp LW, Wakatsuki S, Yokoyama T, Yamakawa H, Furukawa T, Sato M, Ohuchi N, Sato S, Yin J, Wang S, Abraham JM, Souza RF, Smolinski KN, Meltzer SJ, Horii A. PTEN1 is frequently mutated in primary endometrial carcinomas. Nature Genet 1997;17:143– 4. 7. Levine RL, Cargile CB, Blazes MS, van Rees B, Kurman RJ, Ellenson LH. PTEN mutations and microsatellite instability in complex atypical hyperplasia, a precursor to lesion to uterine endometroid carcinoma. Cancer Res 1998;58:3254 – 8. 8. Tashiro H, Blazes MS, Wu R, Cho KR, Bose S, Wang SI, Li J, Parsons R, Ellenson LH. Mutations in PTEN are frequent in endometrial carcinoma but rare in other common gynecological malignancies. Cancer Res 1997; 57:3935– 40. 9. Gurin CC, Federici MG, Kang L, Boyd J. Causes and consequences of microsatellite instability in endometrial carcinoma. Cancer Res 1999;59: 462– 6. 10. Deng G, Chen A, Hong J, Suk Chae H, and Kim YS. Methylation of CpG in a small region of the hMLH1 promoter invariably correlates with the absence of gene expression. Cancer Res 1999;59:2029 –33. 11. Esteller M, Levine R, Baylin SB, Ellenson LH, Herman JG. MLH1

12.

13.

14.

15.

16.

17.

18.

19. 20.

21.

22. 23.

promoter hypermethylation is associated with the microsatellite instability phenotype in sporadic endometrial carcinomas. Oncogene 1998;17: 2413–7. Simpkins SB, Bocker T, Swisher EM, Mutch DG, Gersell DJ, Kovatich AJ, Palazzo JP, Fishel R, Goodfellow PJ. MLH1 promoter methylation and gene silencing is the primary cause of microsatellite instability in sporadic endometrial cancers. Hum Mol Genet 1999;8:661– 6. Kurman R, Kaminiski P, Norris H. The behavior of endometrial hyperplasia a long-term study of “untreated” hyperplasia in 170 patients. Cancer 1985;56:403–12. Cohn DE, Mutch DG, Herzog TJ, Rader JS, Dintzis SM, Gersell DJ, Todd CR, Goodfellow PJ. Absence of PTEN repeat tract mutation in endometrial cancers with microsatellite instability. Gynecol Oncol 2000;79: 101– 6. Mutter GL, Boynton KA, Faquin WC, Ruiz RE, Jovanovic AS. Uteri of women with endometrial carcinoma contain a histopathological spectrum of monoclonal putative precancers, some with microsatellite instability. Cancer Res 1996;56:4483– 6. Esteller M, Catasus L, Matias-Guiu X, Mutter GL, Prat J, Baylin SB, Herman JG. hMLH1 promoter hypermethylation is an early event in human endometrial tumorigenesis. Am J Pathol 1999;155:1767–72. Whelan AJ, Babb S, Mutch DG, Rader JS, Herzog, TJ, Todd C, Ivanovich JL, Goodfellow PJ. MSI in endometrial carcinoma: absence of MLH1 promoter methylation is associated with increased familial risk for cancer. Int J Cancer 2002;99:697–704. Peiffer SL, Herzog TJ, Tribune DJ, Mutch DG, Gersell DJ, Goodfellow PJ. Allelic loss of sequences from the long arm of chromosome 10 and replication errors in endometrial cancers. Cancer Res 1995;55:1922– 6. Jones PA. DNA methylation errors and cancer. Cancer Res 1996;56: 2463–7. Kane MF, Loda M, Gaida GM, Lipman J, Mishra R, Goldman H, Jessup JM, Kolodner R. Methylation of the hMLH1 promoter correlates with lack of expression of hMLH1 in sporadic colon tumors and mismatch repairdefective human tumor cell lines. Cancer Res 1997;57:808 –11. Cunningham JM, Christensen ER, Tester DJ, Kim CY, Roche PC, Burgart LJ, Thibodeau SN. Hypermethylation of the hMLH1 promoter in colon cancer with microsatellite instability. Cancer Res 1998;58:3455– 60. Lengauer C, Kinzler KW, Vogelstein B. Genetic instabilities in human cancers. Nature 1998;396:643– 49. Loeb LA. Mutator phenotype may be required for multistage carcinogenesis. Cancer Res 1991;51:3075–9.