Fax: (314) 362-8620
Ribosomal DNA methylation in patients with endometrial carcinoma
An independent prognostic marker
Article first published online: 23 MAY 2002
Copyright © 2002 American Cancer Society
Volume 94, Issue 11, pages 2941–2952, 1 June 2002
How to Cite
Powell, M. A., Mutch, D. G., Rader, J. S., Herzog, T. J., Huang, T. H.-M. and Goodfellow, P. J. (2002), Ribosomal DNA methylation in patients with endometrial carcinoma. Cancer, 94: 2941–2952. doi: 10.1002/cncr.10559
- Issue published online: 23 MAY 2002
- Article first published online: 23 MAY 2002
- Manuscript Accepted: 27 DEC 2001
- Manuscript Revised: 20 DEC 2001
- Manuscript Received: 1 OCT 2001
- National Institutes of Health. Grant Number: CA71754
- ribosomal DNA methylation;
- endometrial carcinoma;
- molecular marker
Surgery is curative for the majority of patients with endometrial carcinoma: Consequently, the use of adjuvant therapy has been controversial. More effective methods to identify patients who are at risk for recurrence and who would benefit from adjuvant therapy are needed. The objective of this study was to investigate the prognostic significance of ribosomal DNA (rDNA) methylation in patients with endometrial carcinoma.
rDNA methylation was assessed in 215 endometrial tumors by Southern blot analysis of HpaII-digested DNA using a probe corresponding to the 5.8 and 28S ribosomal subunits. Tumor rDNA methylation status was correlated with patient outcome.
The majority of tumors demonstrated high levels of rDNA methylation (rDNA-high; 74%). Both disease free survival and overall survival were significantly worse for patients with low-level rDNA methylation (rDNA-low; P < 0.0001). African-American patients were more likely to have rDNA-low tumors than Caucasian patients (P = 0.002). Among the subpopulation of patients with endometrial carcinoma for whom the use of adjuvant therapy is most controversial (148 women with Stage I–II endometrioid tumors), survival was significantly worse for the patients with rDNA-low tumors (P < 0.0001); and, using multivariate analyses, tumor rDNA level was the only significant prognostic factor for both disease free survival and overall survival (hazard ratios, 11.0 and 26.3, respectively; P < 0.01 for both).
rDNA methylation is an independent prognostic indicator for patients with endometrial carcinoma that may serve to identify women with early-stage disease at who are at high risk for recurrence. Racial differences in DNA methylation may explain the historically observed disparity in survival between African-American patients and Caucasian patients. Cancer 2002;94:2941–52. © 2002 American Cancer Society.
Endometrial carcinoma is the most common gynecologic malignancy in the United States. 1 Epidemiologic studies that take into consideration key clinicopathologic and socioeconomic factors have shown that African-American patients are twice as likely as Caucasian patients to die from endometrial carcinoma.2 The molecular and cellular events that drive the evolution of the endometrial cell to a precancerous state and on to a deadly tumor are largely unknown. DNA methylation is an important regulator of gene expression and is likely to play a role in endometrial tumorigenesis. The loss of DNA mismatch repair and the microsatellite instability (MSI) phenotype in endometrial carcinoma is associated with hypermethylation of the MLH1 promoter.3, 4 Abnormalities in DNA methylation appear to be of prognostic significance in patients with breast carcinoma,5 nonsmall cell lung carcinoma,6 and gliomas.7
Given the role that aberrant MLH1 promoter methylation plays in endometrial tumorigenesis, we undertook studies to further investigate DNA methylation in these tumors and to determine whether methylation differences are of prognostic significance. To assess changes in methylation at multiple sites, we investigated the ribosomal DNA (rDNA) genes (present at ≈ 400 copies), which recently demonstrated extensive methylation in patients with breast carcinoma. 8 The discovery of DNA methylation patterns that are associated with specific tumor features may provide insights into the molecular determinants of endometrial tumorigenesis and may serve to define prognostic markers. A molecular prognostic marker could prove important in helping guide adjuvant treatment for patients with early-stage tumors.
MATERIALS AND METHODS
Tissue Specimens, DNA Isolation, and Patient Follow-Up
Tumor specimens and normal tissue specimens from women with primary endometrial carcinoma were collected at the time of hysterectomy at Washington University School of Medicine between December, 1991 and November, 1999. Written informed consent was obtained from all study participants (Washington University School of Medicine HSC Protocol 93-0828). Specimens were frozen immediately and stored at −70°C. For the majority of specimens, representative portions of the tumor tissues were formalin fixed and paraffin embedded to assess neoplasia cellularity. A single pathologist reviewed all of the hematoxylin and eosin-stained tissue. The tumor specimens included in this study had > 70% neoplastic cellularity, which was estimated by direct microscopic visualization (80% of samples) or from previous loss of heterozygosity and MSI analyses (20% of samples). Standard methods were used to extract DNA from tumor and corresponding non-neoplastic tissue. Normal DNA was prepared from peripheral blood leukocytes. Of the 258 samples that fulfilled the neoplastic cellularity criteria, 43 samples were inadequate for Southern blot analysis either because the DNA samples were degraded or because the quantity of DNA was insufficient. The presence of MSI was determined and reported previously for this group of tumors. 9 Information on adjuvant therapy, disease progression, recurrence, and patient survival were obtained by a review of the medical records. Adjuvant treatment included external beam radiation, chemotherapy, and radiation implants.
rDNA Methylation Analysis
The methylation status of rDNA was assessed by Southern blot hybridization using a modification of a method previously reported by Brock and Bird. 10 Three micrograms of tumor DNA was digested first with EcoRI (Promega, Madison, WI) and then with an excess of HpaII (Promega, Madison, WI). Digested samples were then size separated in 1.0–1.2% agarose gels and transferred to a nylon membrane (GeneScreenPlus; NEN Life Sciences, Boston, MA). The membranes were hybridized with a 32P-labeled DNA probe directed against the DNA fragment corresponding to the 5.8 and 28S ribosomal subunits. Hybridization patterns were detected using autoradiography and/or phosphorimage analysis (Molecular Dynamics, Sunnyvale, CA).
The methylation pattern for each sample was interpreted independently by each of three observers who were blinded to the clinical and pathologic features of the tumors. Disputed samples (15% or 7%) were reviewed in a group fashion without knowledge of the clinical features, and a consensus was reached. Methylation of rDNA was classified as minimal, intermediate, or extensive. Samples with minimal rRNA methylation had no evidence of EcoRI/HpaII DNA fragments greater than an arbitrarily assigned size of approximately 1.5 kilobases (kb). Tumors were assigned the category of intermediate methylation if no fragments measuring > 4 kb were evident. The category of extensive methylation was assigned to samples in which fragments of ≥ 4.0 kb were evident.
To ensure that the restriction enzyme digestions were complete, the Southern blots were hybridized with a mitochondrial DNA probe containing a HpaII site (207-base-pair polymerase chain reaction product corresponding to nucleotides 11,575–11,781; GenBank accession no. NC-001807). It is known that mitochondrial DNA is unmethylated; therefore, it served as a control for completeness of digestion. In addition, EcoRI/MspI double digests were analyzed for a subset of tumors studied. For the small number of samples that demonstrated partial digestion of the mitochondrial DNA, the digests were repeated, and new Southern blots were prepared. All tumor samples that were scored as exhibiting minimal or intermediate rDNA methylation were analyzed again. Southern hybridization analysis of DNAs digested with EcoRI alone was performed for analysis of the 7.1-kb region to ensure that the DNA was not degraded.
Contingency tables were analyzed using the chi-square test. Disease free survival and overall survival were calculated from the date of surgery to the time of recurrence, last follow-up, or death. Disease free survival and overall (disease specific) survival curves were generated using the Kaplan–Meier method, 11 and the significance of observed differences were determined using the log-rank test.12 Patients were censored at the time of their last follow-up or at the time of their death in patients without evidence of disease who died of nonendometrial carcinoma-related causes (disease specific survival was determined). Multivariate analyses were performed using SAS software (version 8.1; SAS Institute, Cary, NC) using a Cox proportional hazards regression. A step-down selection using prognostic variables that were identified as significant on univariate analyses (P < 0.1) was performed to determine independent prognostic factors for the entire set of 215 patients. These same prognostic variables were considered in multivariate analysis of the subsets of 148 patients with Stage I and II endometrioid carcinoma and the 87 patients with surgical Stage IB–IC or Stage IIA–IIB (occult) endometrioid carcinomas. All other P values were two-tailed, with P < 0.05 considered significant.
The pathologic and clinical features of the 215 patients who were included in this study are summarized in Table 1. The median age at time of surgery was 66 years (range, 30–93 years). The racial make-up of the patients was as follows: 176 Caucasian patients (82%) with a median age, 66 years, 39 African-American patients (18%) with a median age of 67 years, and 1 Asian patient (< 1%). Eighty-three percent of these patients had their disease completely staged surgically according to the 1988 International Federation of Gynecologists and Obstetricians (FIGO) criteria, 13 which included sampling of the retroperitoneal pelvic and periaortic lymph nodes. When complete surgical staging was not performed, clinical stage was assigned using criteria established by FIGO in 1971.13 Twenty-one of 36 patients (58%) who were clinically staged had Grade 1 tumors confined to the endometrium and negative cytologic washings. Forty-five of 181 patients with endometrioid adenocarcinoma (24.8%) received adjuvant therapy, and 18 of 34 patients who had tumors with more aggressive histologies (53%) had adjuvant therapy. The mean follow-up or the entire group was 35.8 months, and the median follow-up for the entire group was 33 months. Twenty-two patients were lost to follow-up (mean time at loss, 14.7 months; range, 1–44 months), including 18 Caucasian women and 4 African-American women. The pathologic features and outcome of the 43 patients who were excluded from study due to inadequate and/or degraded DNA were evaluated and did not significantly differ from the features and outcome of the study population (data not shown). rDNA methylation was assessed in all 215 endometrial tumors. Whereas most tumors showed extensive methylation, normal noncancerous DNA showed minimal methylation. One hundred fifty-nine of 215 tumors (74%) demonstrated extensive rDNA methylation, 23 of 215 tumors (11%) demonstrated intermediate rDNA methylation, and 33 of 215 tumors (15%) demonstrated minimal rDNA methylation (Fig. 1).
|Feature||No. of patients (%)||White patients (n = 175) and Asian patients (n = 1)||African–American patients (n = 39)|
|Clear cell||15 (7)||12||3|
|Papillary serous||14 (6.5)||9||5|
|Mixed clear cell and papillary serous||5 (2.3)||1||4|
|Tumor grade (FIGO)|
|Lymph-vascular space involvement|
|FIGO surgical stagea|
|FIGO clinical stageb|
|Deaths attributed to endometrial carcinoma||38||29||9|
|Deaths attributed to other causes||9||8||1|
|Adjuvant therapy by histology|
|Mixed clear cell and papillary serous||2||0||2|
Patients who had tumors with either intermediate or minimal rDNA methylation had significantly decreased recurrence free survival rates (P < 0.0001) and overall survival rates (P < 0.0001) compared with patients who had tumors with extensive rDNA methylation (Fig. 2). The outcome of patient who had tumors with intermediate and minimal rDNA methylation was not significantly different (P = 0.47 and P = 0.54, respectively). Thus, the two groups were combined, and the rDNA methylation trait was dichotomized for the remainder of the analyses. Tumors with intermediate or minimal rDNA methylation were categorized as rDNA-low, and extensively methylated rDNA tumors were classified as rDNA-high.
Patients with tumors that were classified as rDNA-high had a very favorable prognosis. The rDNA-low tumors were associated significantly with other known poor prognostic factors, including nonendometrioid histology, Grade 3 differentiation, presence of lymph-vascular space involvement, and advanced stage (P < 0.05 for all). There was also a significant association between rDNA methylation and patient race. Tumors from African-American women were significantly more likely to be rDNA-low tumors compared with tumors from Caucasian women. This difference in tumor rDNA methylation level also persisted when only endometrioid histologic subtype tumors were considered (Table 2).
|Histology||rDNA methylation (%)||P value|
|All||18 (46)||21 (54)||38 (22)||138 (78)||0.003|
|Endometriod only||11 (41)||16 (59)||25 (16)||128 (84)||0.012|
Univariate analysis of recurrence free and overall survival for all 215 patients determined that tumor stage, lymph-vascular space involvement, histology, grade, and rDNA levels all were significant prognostic factors. Nonsignificant trends in disease free survival and overall survival were observed with patient age (P = 0.48 and P = 0.32, respectively), race (P = 0.47 and P = 0.46, respectively), and presence of MSI (P = 0.67 and P = 0.66, respectively). Multivariate analysis using the univariate variables with P < 0.1 determined that rDNA methylation was an independent predictor of both recurrence free survival and overall survival for patients with endometrial carcinoma (hazard ratios of 3.56 and 3.57, respectively). Both tumor stage and lymph-vascular space involvement also were statistically significant predictors of outcome (Table 3).
|Variable||Disease free survival||Overall survival|
|HR (95% CI)||P value||HR (95% CI)||P value|
|rDNA methylation (low vs. high)||3.56 (1.72–7.41)||0.006||3.57 (1.50–8.47)||0.0041|
|LVSI (absent vs. present)||3.12 (1.38–7.05)||0.012||3.75 (1.43–9.89)||0.0074|
|Stage (I–IV)||2.22 (1.57–3.13)||< 0.0001||2.46 (1.53–2.88)||< 0.0001|
|Histologic subtype (endometrioid vs. nonendometrioid)||1.05 (0.48–2.78)||0.910||1.12 (0.46–2.70)||0.80|
|Tumor grade (1–3)||1.14 (0.52–2.07)||0.649||1.07 (0.54–2.12)||0.84|
|Race (Caucasian/Asian vs. African American)||1.14 (0.54–2.49)||0.740||1.12 (0.50–2.93)||0.67|
Because the use and benefits of adjuvant radiotherapy is unclear for patients with early-stage endometrioid tumors, we assessed the prognostic significance of rDNA methylation in this subpopulation. One hundred forty-eight of the patients in our series had Stage I or II tumors of the endometrioid histologic subtype. In this group, there were 8 recurrences among 23 women with rDNA-low tumors compared with 5 recurrences among 125 women with rDNA-high tumors (P < 0.0001). This resulted in a relative risk of 8.7 (95% confidence interval, 3.1–24.2) for recurrence in patients with rDNA-low tumors. Disease free survival and overall survival also were significantly worse in the patients who had rDNA-low tumors compared with patients who had rDNA-high tumors (P = 0.0002 and P = 0.002, respectively) among these 148 patients (Fig. 3). A lack of complete surgical staging did not appear to account for the worst outcome in the group of patients with rDNA-low tumors. Thirty-four of 148 patients had clinically staged tumors (lymph nodes not sampled), and nearly all of them (31 of 34 patients) were in the rDNA-high group. Using multivariate analyses, tumor rDNA level was the only significant prognostic factor for both disease free survival and overall survival (Table 4), with hazard ratios of 11.0 and 26.3, respectively (P = 0.0012 and P = 0.008, respectively). Of the 148 patients with early-stage endometrioid carcinoma, 87 patients would have been classified as intermediate risk and would meet the criteria for enrollment in a recently completed, large, randomized, National Cancer Institute-sponsored Gynecologic Oncology Group trial assessing the role of adjuvant radiation therapy. 14 In this group of 87 patients, all of whom were staged surgically (with lymph nodes sampled) with Stage IB–IC or Stage IIA–IIB disease, ribosomal DNA methylation was the only statistically significant predictor of both disease free survival and overall survival in both univariate (Fig. 3) and multivariate analyses, with hazard ratios of 18.5 and 43.0, respectively (P = 0.002 and P = 0.01, respectively).
|Variable||Disease free survival||Overall survival|
|HR (95% CI)||P value||HR (95% CI)||P value|
|rDNA methylation (low vs. high)||11.0 (2.5–45.4)||0.0012||26.32 (2.3–3.33)||0.0079|
|LVSI (absent vs. present)||2.6 (0.6–10.5)||0.19||1.9 (0.3–13.0)||0.51|
|Stage (I–II)||1.1 (0.04–3.0)||0.36||1.6 (0.14–17.0)||0.71|
|Tumor grade (1–3)||1.8 (0.66–4.8)||0.25||0.8 (0.2–3.2)||0.68|
|Race (Caucasian vs. African-American)||0.6 (0.13–2.8)||0.55||0.5 (0.06–3.3)||0.44|
The results of this study suggest that rDNA methylation level is an important prognostic indicator for recurrent disease and patient survival in patients with endometrial carcinoma. Although this observation provides further evidence that DNA methylation is important in endometrial tumorigenesis, the role that rDNA methylation plays is unclear. It is possible that rDNA methylation is simply a marker for aberrant methylation involving key growth-regulatory genes. Hypermethylation of the MLH1 promoter appears to be the dominant mechanism of inactivation of the mismatch-repair system in endometrial malignancies 3 that results in MSI. Recently published studies by our group and others have examined MSI as a prognostic factor in patients with endometrial tumors. Maxwell and colleagues found that the presence of MSI was be a favorable prognostic factor,15 whereas Fiumicino et al. came to the opposite conclusion in a smaller series of patients.16 Like MacDonald and coworkers,17 we found no correlation between MSI status (data not shown) and outcome, as reported previously.9 Thus, it seems unlikely that rDNA methylation is a surrogate for MLH1 promoter methylation associated with MSI. The extensive tumor rDNA methylation that is correlated with favorable outcome actually may influence cell growth properties and tumor aggressiveness. Further studies are necessary to elucidate the cellular mechanisms whereby aberrant rDNA methylation influences tumor growth.
The vast majority of women with endometrial carcinoma in the United States present with Stage I or II tumors of the endometrioid histologic subtype. 18 It is in this patient population that the use of adjuvant therapy is particularly controversial19 and for which the identification of additional prognostic markers has the greatest potential clinical benefit. Adjuvant therapy for patients with early-stage endometrial carcinoma usually consists of radiation therapy. The reported rates of toxicity or serious toxicity associated with radiation therapy range from 7% to 20%.14, 20–22 A marker to identify those patients who are at a very low risk of recurrence could be used to avoid unnecessary adjuvant therapy and lessen the morbidity and expense associated with treating endometrial carcinoma. High-level tumor rDNA methylation appears to identify such a group of patients. In this series of 148 patients with early-stage endometrioid carcinoma, only 5 of 125 women (4%) with high levels of rDNA methylation developed recurrent disease (3 women died of disease; 2.4%). However, among the 23 patients with rDNA-low tumors, 8 patients developed recurrent disease (35%), and 6 patients died of disease (26%). Of the 148 patients with Stage I–II disease, 28 patients received adjuvant radiotherapy. Twenty-one of 125 women with rDNA-high tumors received adjuvant therapy (only 1 of 5 women who developed recurrent disease received adjuvant therapy). Seven of these 23 women with rDNA-low tumors were treated with adjuvant radiotherapy, including only 3 of 8 women who developed recurrent disease. Because adjuvant therapy is expected to influence outcome, we assessed the power of the rDNA tumor marker to predict recurrence in the 120 patients with Stage I–II disease who did not receive adjuvant therapy. Five of 16 women with rDNA-low tumors developed recurrent disease compared with 4 of 104 patients with rDNA-high rDNA tumors (P = 0.002). The specificity and sensitivity of rDNA methylation for predicting recurrence in these 120 patients were 90% and 56%, respectively.
A large, multicenter, randomized trial of adjuvant therapy involving patients with Stage IB–IC and Stage IIA–IIB endometrioid endometrial carcinoma was completed recently (Gynecologic Oncology Group protocol 99). Adjuvant therapy appeared to decrease recurrences but did not change overall survival. 14 The tumor rDNA methylation levels in our series of 87 patients, all of whom would have been eligible for the trial, identified the patients who developed recurrent disease and died with a striking accuracy (Fig. 3). In fact, none of the patients with rDNA-high tumors died from endometrial carcinoma. This suggests that rDNA may be a clinically useful prognostic marker.
Referral and sampling biases complicate the interpretation of molecular correlative studies, such as the one we describe here. We recognize that this series of tumors may not be entirely representative of endometrial carcinoma in general. Our population came from a tertiary referral center; thus, the patients we studied were more likely to have either higher grade tumors with nonendometrioid histology or more advanced tumors. We generally collect > 0.5 grams of tumor for study; thus, patients with small tumors (thought to have a better prognosis 23) often do not have adequate tumor tissue to be included. Both factors biased our series toward more aggressive tumors. Thus, rDNA methylation may be an even more powerful prognostic factor in a population-based series. Additional studies in a separate set of patients are necessary to validate the finding that rDNA methylation is a robust prognostic marker. The hazard ratios for disease free survival (11) and overall survival (26) have broad confidence intervals, and a larger study should define more precisely the risks associated with low rDNA tumor methylation.
Many studies previously reported a racial disparity in outcomes in women with endometrial carcinoma. 2, 24 It has been suggested that African-American women present with more advanced stage disease and higher grade tumors, which are more often of the nonendometrioid subtype. A population-based study from the California Cancer Registry, however, revealed that the incidence of high-risk tumors was identical in Caucasian women and African-American women and that African-American women had three-fold fewer low-risk tumors than Caucasian women.25 Even when controlling for clinicopathologic and socioeconomic factors, African-American race remained a significant poor prognostic factor.2 In the current study, we found striking racial differences in rDNA methylation levels. African-American patients were significantly more likely than Caucasian patients to have rDNA-low tumors. Differences in DNA methylation may reflect underlying factors that contribute to the observed racial differences in outcome. Racially based biologic differences suggest that racial genetic variations may have a role in endometrial carcinoma tumorigenesis.25, 26 This study provides evidence that altered methylation is an important factor in endometrial tumorigenesis. Understanding the role that rDNA and other methylation changes play in tumors should facilitate more effective and targeted therapies for the treatment of patients with endometrial carcinoma.
The authors thank Dr. Adrian Bird for providing the ribosomal DNA plasmid construct used in this study; Washington University School of Medicine Division of Biostatistics and Emily S. Brouwer, M.P.H., for their assistance with statistical analyses; and Dr. Steven Scholnick for his helpful comments on the interpretation of data.
- 13International Federation of Gynecologists and Obstetricians. Annual report on the results of treatment in gynecological cancer. Int J Gynecol Obstet. 1989; 28: 189.
- 14A Phase III randomized study of surgery vs. surgery plus adjunctive radiation therapy in intermediate risk endometrial adenocarcinoma. Abstract of the 29th annual meeting of the Society of Gynecologic Oncologists. 1998; 35: 70., , , et al.