SEARCH

SEARCH BY CITATION

Keywords:

  • HNPCC;
  • endometrial carcinoma;
  • microsatellite instability;
  • MLH1 methylation

Abstract

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

BACKGROUND

Endometrial carcinoma is a common malignancy in hereditary nonpolyposis colorectal carcinoma (HNPCC). Like colon carcinoma, endometrial carcinoma is diagnosed at an earlier age in women with HNPCC. In contrast to colon carcinoma, the pathologic features of endometrial carcinoma in HNPCC have not been studied in detail. It was the purpose of this study to pathologically characterize a series of HNPCC associated endometrial carcinomas.

METHODS

Fifty women with HNPCC and endometrial carcinoma were analyzed from four different hereditary cancer registries. H&E stained slides and pathology reports were reviewed for clinically important pathologic features of endometrial carcinoma. These results were compared with those for two different groups of sporadic endometrial carcinoma – women younger than age 50 years (n = 42) and women of all ages with tumors demonstrating microsatellite instability (MSI-high) secondary to methylation of MLH1 (n = 26).

RESULTS

Nearly one-fourth of HNPCC patients in this study had endometrial tumors with pathologic features that would require adjuvant therapy after hysterectomy. There was a trend toward the HNPCC patients having more nonendometrioid tumors; all of these patients were carriers of MSH2 mutations. Such nonendometrioid tumors were extremely rare in the MLH1 methylated group. A subset of MLH1 methylated sporadic tumors demonstrated a unique, ‘undifferentiated’ histology that was not observed in HNPCC or the young group.

CONCLUSION

Data suggest a genotype–phenotype relation in which microsatellite instability resulting from MLH1 methylation is almost exclusively associated with classical or ‘undifferentiated’ endometrioid tumors, whereas microsatellite instability secondary to MSH2 mutation can result in a more variable histologic spectrum of endometrial carcinoma. Cancer 2006. © 2005 American Cancer Society.

Traditionally, colon carcinoma has been the most intensely studied malignancy in the familial cancer syndrome hereditary nonpolyposis colorectal carcinoma (HNPCC; recently reviewed1, 2). However, it is knownthat women with HNPCC have an equal or higher likelihood of developing endometrial carcinoma.3, 4 In fact, in more than 50% of women with HNPCC, endometrial carcinoma is the first, or ‘sentinel,’ cancer to develop.5 Similar to colon carcinoma, the age of onset for endometrial carcinoma in HNPCC is significantly younger than the average age of onset of endometrial carcinoma in nonmutation carriers.6 Also, similar to colon carcinoma, high levels of microsatellite instability (MSI-high) in endometrial carcinoma can result from germline mutation of the DNA mismatch repair genes MLH1, MSH2, or MSH6 (HNPCC) or, in the sporadic setting, from methylation of MLH1.7 One reason for the relative paucity of studies regarding endometrial carcinoma in HNPCC is the assumption that endometrial carcinoma is a ‘good prognosis’ tumor in this patient group that only requires hysterectomy for effective treatment. Although there are several epidemiologic studies of endometrial carcinoma in HNPCC,6, 8 detailed pathologic analysis is lacking. It is not known whether HNPCC endometrial carcinomas have unique pathologic features compared with sporadic endometrial carcinomas from the general population. Knowing such pathologic data are critical for the understanding of the natural history of this disease in this unique patient population. Therefore, it was the primary purpose of this study to carefully examine the pathologic features of endometrial carcinomas in women with HNPCC. These results were then compared with sporadic endometrial carcinomas from two different patient groups: 1) nonmutation carriers younger than 50 years old, and 2) women of any age with endometrial carcinoma associated with methylation of MLH1 and MSI-high. The younger than 50 group was chosen because it is known that HNPCC-associated endometrial carcinoma occurs at a younger age than in the general population. The sporadic, MLH1 methylated group provides a comparison group sharing a similar defect in DNA mismatch repair with resulting microsatellite instability.

MATERIALS AND METHODS

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

Pathologic Features of Endometrial Carcinoma in HNPCC

After institutional review board (IRB) approval (MDACC protocol LAB 01-047), the hematoxylin and eosin (H&E)-stained slides and pathology reports for a cohort of women with HNPCC and endometrial carcinoma (n = 50) from four different hereditary cancer registries (MDACC, UCSF, Newfoundland, and Creighton) were reviewed. Each HNPCC patient had a confirmed mutation in either MLH1 or MSH2, determined by genetic sequencing. Demographic and germline mutation status were obtained from each registry. The tumor histotype and FIGO grade, depth of myometrial invasion, lymphatic/vascular invasion, cervix involvement, and pathologic stage were analyzed by light microscopy and recorded. These results were compared with those from a cohort of young (younger than 50 years old) women with endometrial carcinoma (n = 42) who had been confirmed as having intact MLH1 and MSH2 via formal genetic testing. This latter group of young endometrial carcinoma patients is from an IRB-approved (ID 01-533), ongoing prospective clinical trial at the M.D. Anderson Cancer Center examining the prevalence of DNA mismatch repair gene mutations in women with endometrial carcinoma younger than 50 years of age. The data from these two groups of young endometrial carcinoma patients were compared with a third group of sporadic endometrial carcinoma patients (n = 26; see below) in which the endometrial carcinomas were associated with loss of MLH1 protein and MSI-high due to methylation of MLH1. Patients in all groups were surgically staged. H&E-stained slides for the three groups were centrally reviewed by a single gynecologic pathologist (R.R.B.) who was blinded as to group. For all the patients in the younger than 50 group, the MLH1 methylated group, and the MDACC subset in the HNPCC group, all H&E-stained slides were microscopically analyzed. For many of the HNPCC cases from other family cancer registries, only representative slides (at least two per case) were reviewed. For the cases in which representative slides were reviewed, if the review tumor grading and histotyping was significantly different from the grade/histotype stated in the pathology report, the case was not used for our study. Such discrepancies were very infrequent in this series. Differences among the pathologic data for the three groups of patients (n = 118) were assessed with Fisher exact test. The frequency of different histotypes in the three specific study groups was compared with that for the total MDACC endometrial carcinoma patient population for the years 2000–2004.

Pathologic Features of Endometrial Carcinoma Associated with Methylation of MLH1

A set of endometrial carcinomas (n = 128; 85 endometrioid, 19 uterine papillary serous carcinoma [UPSC], and 24 malignant mixed mullerian tumor [MMMT]) was examined for methylation of MLH1 by methylation-specific polymerase chain reaction (PCR). Because MLH1 methylation results in loss of MLH1 protein, immunohistochemistry was used to confirm methylation results. MSI analysis was performed for tumors that were methylated at MLH1 and showed loss of MLH1 immunostaining. A total of 26 tumors were identified that had methylation of MLH1 with associated loss of MLH1 immunohistochemical staining and MSI-high. The pathologic features of the MLH1 methylated, MSI-high endometrial carcinomas were then analyzed by light microscopy.

Immunohistochemistry, MSI Analysis, and MLH1 Promoter Methylation

By using standard immunohistochemical techniques, formalin-fixed, paraffin-embedded sections of tumor were analyzed by light microscopy for the immunohistochemical expression of MLH1 (G168-15, 1:30, Pharmingen, San Diego, CA) and MSH2 (FE11, 1:100, Oncogene Science, Boston, MA). The immunoreactive proteins were observed using the EnVision horseradish peroxidase kit (Dako, Carpinteria, CA) for MLH1 and the LSAB2 horseradish peroxidase kit (Dako) for MSH2. Adjacent normal tissue and surrounding tissue lymphocytes served as internal positive controls for each case. Nuclear staining of the tumor was scored as either present or absent compared with the corresponding internal control. Additional immunohistochemistry analyses were performed for estrogen receptor (ER; 6F11, 1:50, Novocastra, Hingham, MA), progesterone receptor (PR; 1A6, 1:30, NeoMarkers/Labvision, Fremont, CA), chromogranin A (1:4000, Chemicon, Temecula, CA), synaptophysin (Snp88, 1:75, Biogenex, San Ramon, CA), and pancytokeratin cocktail (AE1/AE3, 1:500, Dako; CAM5.2, 1:50, Becton Dickinson, San Jose, CA; cytokeratin MNF116, 1:50, Dako; keratin 8&18, 1:25, Zymed, South San Francisco, CA).

MSI analysis was performed using formalin-fixed, paraffin-embedded sections of tumor and corresponding normal tissue. DNA was extracted from microdissected areas of tumor and normal tissue from sections on glass slides. After DNA amplification using fluorescent labeled primers, a panel of five microsatellites recommended by the National Cancer Institute (NCI),9 BAT25, BAT26, D2S123, D5S346, and D173250, were analyzed for allelic shift using a multiplex fluorescence-based PCR assay. The amplified PCR products were analyzed on an ABI Genetic Analyzer (PE/Applied Biosystems, Foster City, CA) using GeneScan Analysis software provided by the manufacturer. Tumors showing allelic shift at two or more markers were classified as MSI-high, while those with allelic shift at one marker were classified as MSI-low. Tumors with no allelic shift at any marker were classified as microsatellite stable.

For the MLH1 methylation assay, DNA was isolated from formalin-fixed, paraffin-embedded tissue sections that were microdissected to provide relatively pure tumor samples. After treatment with bisulfite, the DNA was amplified by PCR using primers that were specific for either the methylated (M) or the unmethylated (U) versions of MLH1 (MLH1-M forward, 5′-gatagcgatttttaacgc-3′ and MLH1-M reverse, 5′-tctataaattactaaatctcttcg-3′; MLH1-U forward, 5′-agagtggatagtgatttttaatgt-3′ and MLH1-U reverse, 5′-actctataaattactaaatctcttca-3′). PCR products were separated on 6% polyacrylamide gels and observed after staining with ethidium bromide. DNA extracted from the RKO colon carcinoma cell line was used as a positive control.

RESULTS

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

Comparison of Pathologic Features among the Three Groups

For this study, the average age for the diagnosis of endometrial carcinoma in women with HNPCC was 46.8 years, similar to previous studies.6 The HNPCC group consisted mainly of MSH2 mutation carriers (n = 47) with a few MLH1 mutation carriers (n = 3). On the basis of previous studies of endometrial carcinoma in the general population, the mean age of diagnosis for UPSC and clear cell carcinoma is 68 years,10–12 while the mean age of diagnosis for uterine MMMT is 65–67 years.13, 14 In the HNPCC group, the mean age of diagnosis for these nonendometrioid tumors was significantly younger, 46.4 years, but was similar to the mean age of the entire HNPCC group. In the HNPCC group, the mean age of diagnosis for women with Stage I tumors was 47.5 years, and the mean age for women with Stages II and III tumors was 46.0 years. Thus, for the HNPCC group age does not distinguish the women with endometrioid tumors from women with nonendometrioid tumors or women with low-stage from women with high-stage tumors.

The pathologic characteristics of HNPCC-associated endometrial carcinoma are summarized in Table 1. For comparison, summaries of the data for endometrial carcinomas from two other patient groups, sporadic endometrial carcinoma in women younger than 50 years of age with intact MLH1 and MSH2 and sporadic endometrial carcinoma associated with methylation of MLH1, are included in Table 1. The mean age at diagnosis of endometrial carcinoma was significantly higher in the sporadic MLH1 methylation group compared with the other two groups. The HNPCC group tended to have more nonendometrioid tumors, but this difference was not statistically significant (Fig. 1). All nonendometrioid tumors were in patients with MSH2 mutations. The nonendometrioid tumors were clear cell carcinoma (n = 3), UPSC with a clear cell carcinoma component (n = 3), and MMMT (n = 1). Nonendometrioid tumors were extremely rare in the sporadic, MSI-high, MLH1 methylated group. Compared with the two groups of young endometrial carcinoma patients (HNPCC and sporadic younger than 50 years of age), the sporadic MLH1 methylated group was associated with higher-grade endometrioid tumors with lymphatic/vascular invasion identified within the myometrium. The prevalence of deep myometrial invasion was not significantly different among the three groups. There was a trend toward advanced stage (Stage III or Stage IV) at diagnosis being more common in the two sporadic groups compared with the HNPCC group, but this was not statistically significant. However, it is important to note that 22% of HNPCC-associated endometrial carcinomas were Stage II or higher at the time of hysterectomy.

Table 1. Pathologic Features of HNPCC-Associated Endometrial Carcinoma and Two Different Groups of Sporadic Endometrial Carcinoma
 HNPCCSporadic younger than 50 yrsSporadic with MLH1 methylationP valuea
  • a

    Statistical analysis was performed using analysis of variance (ANOVA) for age and Fisher exact test for the remaining parameters. Differences were considered significant if P < 0.05.

n (total = 118)504226N/A
Mean age at diagnosis, yrs46.839.961.10.001
Endometrioid histology86.0%97.6%96.2%0.115
 Endometrioid Grade 144.2%39.0%8.0% 
 Endometrioid Grade 239.5%51.2%56.0%0.005
 Endometrioid Grade 316.3%9.8%32.0% 
Myometrial invasion > 50%26.0%23.8%32.0%0.798
Lymphatic/vascular invasion present24.0%40.5%52.0%0.041
 Stage I78.0%66.7%60.0% 
 Stage II10.0%7.1%4.0%0.099
 Stage III/Stage IV12.0%26.2%36.0%
thumbnail image

Figure 1. Representative nonendometrioid tumors associated with HNPCC. (A) Papillary serous carcinoma. H&E. (B) Clear cell carcinoma with characteristic clear tumor cell cytoplasm and eosinophilic, extracellular, hyaline globules. H&E staining; original magnification ×200 (both panels).

Download figure to PowerPoint

For the overall endometrial carcinoma patient population at MDACC, 78% had endometrioid tumors and 22% had nonendometrioid tumors. This percentage of nonendometrioid tumors was significantly greater than the percentages observed in the sporadic younger than 50 group and the sporadic MLH1 methylated group (P = 0.006). The HNPCC group had fewer nonendometrioid tumors (14%) than the general population, but this difference was not statistically significant. Analysis of the overall endometrial carcinoma patient population at MDACC demonstrated the following FIGO grade distribution: Grade 1, 28.5%; Grade 2, 58.8%; and Grade 3, 12.7%. The HNPCC and sporadic younger than 50 groups had similar distributions of endometrioid tumor grades compared with the overall population. However, the sporadic MLH1 methylated group had significantly fewer Grade 1 tumors and more Grade 3 tumors (P = 0.009).

Sporadic Endometrial Carcinoma Associated with Methylation of MLH1

Approximately 18% (26 of 138) of the sporadic endometrial carcinomas were methylated at the MLH1 locus, showed loss of MLH1 protein by immunohistochemistry, and were MSI-high (Figs. 2, 3). Nearly all of these tumors with methylation of MLH1 (96.2%; 25 of 26) were the endometrioid subtype; one tumor was an MMMT (Table 1). This contrasts with the endometrial carcinomas in HNPCC, in which 14% were nonendometrioid (Table 1, Fig. 1). A small subset (5 of 26, 19%) of the MLH1 methylated tumors had distinctive histologic features not usually associated with endometrioid tumors. These tumors had been previously classified as endometrioid adenocarcinoma, FIGO Grade 3. This distinctive histology was characterized as ‘undifferentiated’ in that the tumor cells were monotonous and small-to-medium sized, larger than histiocytes, but smaller than usual endometrioid tumor cells (Fig. 4A,B). The ‘undifferentiated’ cells grew in solid, discohesive sheets without a pattern or gland formation. By comparison, Grade 3 endometrioid adenocarcinomas characteristically consist of larger tumor cells with at least focal gland formation (Fig. 4A,B). By immunohistochemistry, the ‘undifferentiated’ tumors were positive for pancytokeratin, but the staining was weaker and more patchy compared with that seen in Grade 3 endometrioid tumors (Fig. 4C,D). The ‘undifferentiated’ tumors were entirely negative for ER and PR, whereas most Grade 3 endometrioid tumors have at least focal staining for these hormone receptors (Fig. 4E,F). No staining for the neuroendocrine markers synaptophysin and chromogranin was detected (data not shown). The age of diagnosis for these patients was quite variable (average age, 56 yrs; range, 40–76 yrs), as was the clinical stage (Stage IA–IIIC). Interestingly, despite the common molecular defect in DNA mismatch repair, the ‘undifferentiated’ histotype was not observed in the endometrial carcinomas from women with HNPCC, including those with germline defects in MLH1.

thumbnail image

Figure 2. MLH1 methylation in sporadic endometrial carcinoma. (A) Analysis of MLH1 methylation by methylation-specific PCR followed by polyacrylamide gel electrophoresis. The presence of a band in the M lane indicates methylation of MLH1 (Tumors 1 and 2). Tumor 3 shows no evidence of MLH1 methylation. (B,C) Immunohistochemistry performed on a representative endometrial tumor with MLH1 methylation shows (B) positive nuclear expression for another DNA mismatch repair protein, MSH2, and (C) lack of MLH1 nuclear staining.

Download figure to PowerPoint

thumbnail image

Figure 3. Representative MSI chromatograms for an MSI-high, MLH1 methylated endometrial carcinoma. The endometrial tumor exhibited allelic shift (additional, abnormal peaks in tumor DNA compared with normal) in all five microsatellites examined (results not shown for D2S123 and D5S346).

Download figure to PowerPoint

thumbnail image

Figure 4. Histologic features of (A,C,E) ‘undifferentiated’ carcinoma associated with MLH1 methylation compared with (B,D,F) typical FIGO Grade 3 endometrioid adenocarcinoma. (A,B) H&E. (C,D) Pancytokeratin immunohistochemistry. (E,F) Estrogen receptor (ER) immunohistochemistry. Compared with Grade 3 endometrioid adenocarcinomas, the ‘undifferentiated’ carcinomas had smaller tumor cells with no evidence of glandular differentiation. By immunohistochemistry, the Grade 3 endometrioid tumors were strongly positive for the epithelial marker pancytokeratin and ER, but the ‘undifferentiated’ tumors were weakly positive for pancytokeratin and entirely negative for ER. The ‘undifferentiated’ tumors were also entirely negative for progesterone receptor (data not shown). Original magnification ×100 (C–F); ×200 (A,B).

Download figure to PowerPoint

DISCUSSION

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

In this study, 22% of the HNPCC endometrial carcinomas were Stage II or Stage III, and would therefore require adjuvant radiation therapy or chemotherapy. In addition, 14% of the HNPCC-associated endometrial tumors were classified as a histotype (UPSC, clear cell carcinoma, or MMMT) associated with a more aggressive clinical course.15, 16 All of the tumors with nonendometrioid histology occurred in women with MSH2 mutations. As the HNPCC cases that we studied were collected from several different large registries, clinical follow-up information is limited for most of these patients. Previous epidemiologic studies have reported that the overall 5-year survival for women with HNPCC and endometrial carcinoma is similar to that for women with sporadic endometrial carcinoma.8 Classically, in the general population approximately 80% of sporadic endometrial carcinomas are of the endometrioid type, while approximately 20% are nonendometrioid.17, 18 These reported percentages are quite similar to the percentages at MDACC (78% endometrioid and 22% nonendometrioid). The HNPCC group had slightly fewer nonendometrioid tumors (14%) than the general population. In contrast, the younger than 50 group and MLH1 methylated group in the present study had significantly fewer nonendometrioid tumors than the general population (2.4% and 3.8%, respectively). Our study highlights that HNPCC-associated endometrial carcinoma includes a spectrum of disease, including tumors with nonendometrioid histology, very similar to the spectrum of tumor histologies seen in the general population.

In contrast to the broader spectrum of endometrial carcinoma in HNPCC, loss of MLH1 function via methylation is associated primarily with endometrioid tumors (Table 1). Similar results for MLH1 methylation have been obtained by previous investigators.19, 20 In our study, the sporadic MLH1 methylated group has a statistically significant association with higher-grade endometrioid tumors. Both sporadic groups have more tumors with advanced pathologic stage (Stage III or Stage IV) compared with the HNPCC group. Previously, one small study also showed that sporadic, MSI-high endometrial carcinomas are significantly associated with higher FIGO grade and higher FIGO stage.21 These differences between the HNPCC group and the sporadic, MLH1 methylated group suggest a genotype–phenotype relationship in which microsatellite instability resulting from MLH1 methylation is almost exclusively associated with higher-grade endometrioid tumors, while microsatellite instability secondary to MSH2 mutation can result in a more varied spectrum of endometrial carcinomas. Because these associations are not absolute, it is likely that the actions of additional genes also contribute to determining the histotype of an endometrial tumor.

A smaller subgroup of endometrial carcinomas with MLH1 methylation demonstrated a distinctive, ‘undifferentiated’ histology (Fig. 4). In contrast to more conventional endometrioid tumors, the ‘undifferentiated’ tumors were entirely negative for ER and PR and only weakly positive for pancytokeratin, a cocktail of antibodies to a variety of epithelial markers (Fig. 4). This ‘undifferentiated’ histology was not observed in the HNPCC endometrial carcinomas, including the tumors from individuals with a germline MLH1 mutation. It must be stressed that most of the endometrial carcinomas with microsatellite instability and MLH1 methylation were endometrioid tumors with no microscopic characteristics that could reliably distinguish them from microsatellite stable endometrial carcinomas. This is similar to the situation with MSI-high colon carcinomas. Some histologic features, such as mucinous histology, poor differentiation, tumor infiltrating lymphocytes, and a Crohn-like inflammatory reaction at the tumor periphery can be associated with some MSI-high colon carcinomas.22 However, the presence of microsatellite instability cannot be reliably predicted based on light microscopic examination of H&E-stained slides.

The variety of endometrial carcinoma histotypes observed in HNPCC is similar to the limited experience with ovarian carcinoma in HNPCC,23 in which a wide variety of ovarian tumors were documented. It should be noted, however, that this HNPCC ovarian carcinoma study relied on clinical records for documentation of tumor types; a formal pathologic review was not conducted. HNPCC-associated gynecologic malignancies contrast with ovarian carcinoma from women with BRCA1 or BRCA2 mutations. In these women, nearly all of the ovarian tumors are restricted to one histotype—high-grade papillary serous carcinoma.24–26

In the general population, many low-grade endometrial endometrioid adenocarcinomas are derived from the precursor lesion complex atypical hyperplasia (CAH). It is known that approximately 29% of women with CAH detected on endometrial biopsy will progress to endometrial carcinoma.27 For HNPCC-associated colon carcinoma, it has been hypothesized that colon adenomas, especially proximal ones, are more likely to progress to colonic adenocarcinoma, and progress more rapidly, than adenomas in the general population.28–30 A similar hypothesis could be posed for CAH in HNPCC-associated endometrial carcinoma, but at this time we do not have sufficient data to address this issue. During the collection of cancer cases for the HNPCC group, we did find two cases of CAH in hysterectomy specimens. For the NCI-sponsored endometrial carcinoma chemoprevention trial that we are conducting at MDACC, we have encountered two women with HNPCC who had complex hyperplasia with atypia at their baseline endometrial biopsies. At hysterectomy, both of these women had endometrial endometrioid adenocarcinoma, Grade 1, associated with complex hyperplasia. Neither tumor was invasive. Therefore, our limited information at this time suggests that CAH is indeed a part of the pathogenesis of endometrial endometrioid tumors in HNPCC.

In conclusion, we have found that HNPCC-associated endometrial carcinoma includes a spectrum of disease, including endometrioid adenocarcinoma, clear cell carcinoma, UPSC, and MMMT. While the majority of these tumors are early stage, 22% are Stage II or higher, which may require therapy beyond a simple hysterectomy. Microsatellite instability resulting from MLH1 methylation is almost exclusively associated with classical or ‘undifferentiated’ endometrioid tumors, while microsatellite instability due to MSH2 mutation can result in endometrial carcinoma with a variety of histotypes. The HNPCC group in our study was almost entirely composed of MSH2 mutation carriers. As more endometrial carcinomas are identified for study in women with germline MLH1 mutations, it will be interesting to determine whether mutations in these two different DNA mismatch repair genes result in tumors with different pathologic features.

Acknowledgements

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

The authors thank Jane S. Green, Ph.D., Discipline of Genetics, Faculty of Medicine, Memorial University of Newfoundland, St. John's, Newfoundland, Canada for assistance and the MDACC gynecological pathologists (Elvio Silva, M.D., Michael Deavers, M.D., Anais Malpica, M.D., and Jinsong Liu, M.D., Ph.D.) for their expertise in determining microscopic tumor grade and histotype.

REFERENCES

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES