DNA-methylation analysis identifies the E-cadherin gene as a potential marker of disease progression in patients with monoclonal gammopathies

Authors


Abstract

BACKGROUND

Silencing of tumor suppressor genes (TSG) by aberrant methylation (referred to as methylation) contributes to the pathogenesis of various human malignancies. However, little is known about the methylation of known and putative TSGs in monoclonal gammopathies. Thus, the authors investigated the methylation frequencies of 10 genes in patients with monoclonal gammopathies.

METHODS

The methylation patterns of the genes p16INK4a (p16), tissue inhibitor of metalloproteinase 3 (TIMP3), p15INK4b (p15), E-cadherin (ECAD), death-associated protein kinase (DAPK), p73, RAS-association domain family 1A (RASSF1A), p14, O6-methylguanine DNA methyltransferase (MGMT), and retinoid acid receptor β2 (RARβ) were determined in patients with monoclonal gammopathy of undetermined significance (MGUS; n = 29), smoldering multiple myeloma (SMM; n = 5), multiple myeloma (MM; n = 113), or plasma cell leukemia (PCL; n = 7) by methylation-specific polymerase chain reaction analysis.

RESULTS

Methylation frequencies for p16, TIMP3, p15, ECAD, DAPK, p73, RASSF1A, p14, MGMT, and RARβ were as follows: 28%, 35%, 10%, 0%, 17%, 21%, 14%, 14%, 7%, and 0%, respectively, in patients with MGUS and 36%, 29%, 27%, 27%, 22%, 15%, 15%, 9%, 4%, and 0%, respectively, in patients with MM. Methylation of at least 1 of these genes was detected in 79% of patients with MGUS and in 80% of patients with MM. Although methylation of ECAD was not detected in patients with MGUS, it was observed frequently in patients with MM and with even greater frequency in patients with PCL. It is noteworthy that an association was found between ECAD methylation and poor prognostic markers in patients with MM.

CONCLUSIONS

Methylation of certain genes can be detected frequently in patients with monoclonal gammopathies. The current data suggest that methylation of ECAD is a marker of disease progression in patients with MM and PCL. Cancer 2004. © 2004 American Cancer Society.

Multiple myeloma (MM) is a B-cell neoplasm characterized by bone marrow (BM) infiltration with malignant plasma cells, which secrete monoclonal immunoglobulin fragments, and by osteolytic lesions. MM ranks as the second most frequently occurring hematologic malignancy after non-Hodgkin lymphoma.1 MM usually is preceded by an age-dependent premalignant disease called monoclonal gammopathy of undetermined significance (MGUS). MGUS cells secrete monoclonal immunoglobulin and progress to malignant MM at a rate of 1% per year.2 According to reported experience over the past decades, MM comprises various entities with a heterogeneous clinical course, ranging from a relatively benign disorder to a highly aggressive disease with rapid progression.3 Cytogenetic studies reveal a wide variety of translocations involving specific subtypes of the disease, and structurally altered genes play important roles in cell proliferation, differentiation, and gene transcription.4, 5 Although several prognostic parameters, including β2-microglobulin, lactate dehydrogenase (LDH), serum creatinine, hemoglobin, and calcium levels; percentage of BM plasma cells; and deletion of chromosome 13q14, have been identified as predictors of the clinical courses of individual patients, there remains a lack of understanding of the molecular features of MM entities with distinct biologic and clinical behavior.6, 7 Molecular mechanisms leading to the transition from MGUS to MM remain poorly defined, and similarly, from a clinical point of view, parameters indicating a high risk of experiencing this transformation are lacking.

Changes in the pattern of DNA methylation have been recognized as frequent events in human malignancies; these changes are involved in abnormalities of gene expression, chromosome structure, timing of DNA replication, and chromatin organization.8 Aberrant methylation (referred to as methylation) of promoter-associated CpG islands has been identified as a mechanism of gene silencing and represents the epigenetic equivalent of mutations and deletions in carcinogenesis. The methylation of several malignancy-related genes that affect various molecular pathways has been investigated in different tumor types, and the results suggest that specific tumors may have their own distinct patterns of methylation.9–12 For patients with MGUS and patients with MM, only a limited number of studies addressing the issue of gene methylation, including methylation of p15INK4b (p15), p16INK4a (p16), death-associated protein kinase (DAPK), E-cadherin (ECAD, CDH1), and suppressor of cytokine signaling 1 (SOCS1), have been reported to date,13–17 and most of these studies have involved the analysis of relatively small numbers of samples. Thus, we investigated the frequencies of methylation of the genes p16, tissue inhibitor of metalloproteinase 3 (TIMP3), p15, ECAD, DAPK, p73, RAS-association domain family 1A (RASSF1A), p14ARF (p14), O6-methylguanine DNA methyltransferase (MGMT), and retinoid acid receptor β2 (RARβ) in large numbers of patients with MGUS, smoldering multiple myeloma (SMM), MM, or plasma cell leukemia (PCL). An association between methylation and loss of expression of these genes has been demonstrated previously.18–27 These genes were chosen because methylation of the p73 and MGMT genes can be detected frequently in hematologic malignancies but has not been investigated previously in patients with MM.10, 28, 29 Moreover, little is known about the methylation of RARβ, RASSF1A, TIMP3, and p14 in hematologic malignancies; however, these genes are methylated frequently in certain solid tumor types.11, 30 In addition, we compared our methylation results with clinicopathologic characteristics and data on chromosome 13q14 deletion from the same patients.

MATERIALS AND METHODS

Clinical Specimens

BM aspirates and clinical data were collected from patients with MGUS (n = 29), SMM (n = 5), MM (n = 113), or PCL (n = 7). For control experiments, BM specimens from a healthy BM donor and from patients with localized non-Hodgkin lymphoma without BM infiltration (n = 7) were analyzed. BM aspirates were obtained from the posterior iliac crest and were collected in heparin-coated syringes. For all patients, these procedures were performed for diagnostic purposes, and all individuals provided informed consent according to institutional guidelines, including permission to use an aliquot of the specimen for research purposes. Mononuclear cells were isolated using Histopaque-1077 (Sigma, St. Louis, MO) according to the manufacturer's instructions and were stored at −20 °C.

Clinicopathologic characteristics of all patients are summarized in Table 1. One hundred nine BM samples were collected before the start of chemotherapy, and 45 samples were collected during disease progression. Overall survival was studied in patients with MM after they received standard-dose therapy with either combined melphalan and prednisone (n = 62) or combined vincristine, doxorubicin, and dexamethasone (n = 18) or after they received high-dose therapy (n = 19).

Table 1. Clinical Characteristics of Patients with Monoclonal Gammopathy of Undetermined Significance, Smoldering Multiple Myeloma, Multiple Myeloma, or Plasma Cell Leukemia
CharacteristicNo. of patients
MGUSSMMMMPCL
  • MGUS: monoclonal gammopathy of undetermined significance; SMM: smoldering multiple myeloma; MM: multiple myeloma; PCL: plasma cell leukemia; Ig: immunoglobulin; LDH: lactate dehydrogenase; U: units; SDT: standard-dose therapy; HDT: high-dose therapy.

  • a

    See Zojer et al.7

No. of patients2951137
Gender    
 Female133506
 Male162631
Age (yrs)    
 Median69656660
 Range52–8346–8231–9037–77
Ig subtype    
 IgA81301
 IgG144644
 Others70192
Light chain    
 λ142391
 κ153746
Disease stage    
 I20
 II or III93
Morphology of MM cells    
 Mature or intermediate80
 Plasmablastic7
Serum creatinine (mg/dL)1.9
 Mean range0.7–10.9
β2-Microglobulin (mg/L)6.5
 Mean range1.34–86.3
LDH (U/L)190
 Mean range86–627
Calcium (mmol/L)2.4
 Mean range1.6–4.6
Hemoglobin (g/dL)11.1
 Mean range5.1–17.9
Therapy    
 SDT80
 HDT18
Deletion of 13q14a    
 Yes32
 No44

DNA Extraction

Genomic DNA was isolated from mononuclear cells using proteinase K (Gibco, Lofer, Austria) digestion followed by phenol-chloroform extraction and ethanol precipitation.31

Methylation Analysis

Genomic DNA was treated with sodium bisulfite (Sigma) as described previously.32 Using methylation-specific polymerase chain reaction (MSP) analysis, the methylation status of the genes p16, TIMP3, p15, ECAD, DAPK, p73, RASSF1A, p14, MGMT, and RARβ was analyzed. Primer sequences of all genes, annealing temperatures, and the expected product sizes have been reported previously.9, 12, 26, 33 For each PCR assay, 0.1 μg DNA was used. To confirm successful sodium bisulfite treatment and the intact nature of the genomic DNA, MSP also was performed with the primer pair specific for the unmethylated form of p16. For additional control experiments, DNA prepared from BM aspirates obtained from one healthy BM donor and from patients with localized non-Hodgkin lymphoma were analyzed. DNA from peripheral blood lymphocytes treated with SssI methylase (New England BioLabs Inc., Beverly, MA) were used as positive controls for methylated alleles. PCR products were analyzed on 2% agarose gels and were visualized under ultraviolet illumination. To confirm our methylation results, sequencing of several PCR products in MGUS and MM samples were performed.

Statistical Analysis

The methylation status of the genes p16, TIMP3, p15, ECAD, DAPK, p73, RASSF1A, p14, MGMT, and RARβ was compared with clinicopathologic characteristics, including age; gender; β2-microglobulin, LDH, serum creatinine, hemoglobin, and calcium levels; type of paraprotein; type of light chain; tumor stage (Durie-Salmon Staging System); tumor grade (morphology of MM cells); and chromosome 13q14 deletion in patients with MM. Statistical analyses were performed using the chi-square test and the Student t test for correlations between groups. Overall survival curves were plotted according to the method of Kaplan and Meier and were compared using the log-rank test and the Breslow test.

RESULTS

Frequency of Methylation in MGUS, SMM, MM, and PCL Samples

We analyzed the methylation frequencies of the p16, TIMP3, p15, ECAD, DAPK, p73, RASSF1A, p14, MGMT, and RARβ genes in BM aspirates from patients with MGUS, SMM, MM, or PCL using MSP. The methylation frequencies of these genes are summarized in Figure 1. The unmethylated form of p16 was found in all samples, confirming the presence and integrity of DNA in all specimens examined. Figure 2 summarizes methylation results in detail for the different types of samples. Examples of MSP are shown in Figure 3. The sequencing of several PCR products confirmed methylation as detected by MSP. Representative parts of the sequencing of p16 in methylated and unmethylated MGUS and MM samples are shown in Figure 3.

Figure 1.

The percentage of methylation of 10 selected genes in patients with monoclonal gammopathy of undetermined significance (MGUS), smoldering multiple myeloma (SMM), multiple myeloma (MM), or plasma cell leukemia (PCL).

Figure 2.

Summary of methylation of 10 selected genes in patients with (A) multiple myeloma (MM), (B) monoclonal gammopathy of undetermined significance (MGUS), (C) smoldering multiple myeloma (SMM), or (D) plasma cell leukemia (PCL) compared with (E) a control group. Gray boxes represent unmethylated samples, and black boxes represent methylated samples.

Figure 3.

Examples of the methylation-specific polymerase chain reaction (MSP) assay and sequencing in patients with monoclonal gammopathy of undetermined significance (MGUS) and patients with multiple myeloma (MM). (A) Bands represent the methylated form of the investigated gene. PCR product sizes are shown on the left. (B) Bands represent the unmethylated form of the p16 gene. PCR product size is shown on the left. (C) Representative sections of the sequencing analysis of the p16 promoter region in MGUS and MM samples. In samples that were identified as being methylated on methylation-specific PCR analysis (p16/M), cytosines in CpG dinucleotides remained cytosines (indicated by arrows). In samples that were identified as being unmethylated on methylation-specific PCR analysis (p16/U), all cytosines in CpG dinucleotides were changed to thymines (indicated by asterisks). bp: base pairs; ECAD/M: methylated E-cadherin gene; MGMT/M: methylated O6-methylguanine DNA methyltransferase gene; DAPK/M: methylated death-associated protein kinase gene.

In 79% of MGUS samples, methylation of at least 1 gene was detected: Forty-five percent of MGUS samples exhibited methylation of 1 gene, 14% of MGUS samples exhibited methylation of 2 genes, 14% of MGUS samples exhibited methylation of 3 genes, 3% of MGUS samples exhibited methylation of 4 genes, and 3% of MGUS samples exhibited methylation of 5 genes. Methylation of at least 1 gene was detected in 80% of MM samples, including methylation of 1 gene in 27% of MM samples, methylation of 2 genes in 25% of MM samples, methylation of 3 genes in 14% of MM samples, methylation of 4 genes in 9% of MM samples, methylation of 5 genes in 3% of MM samples, methylation of 6 genes in 2% of MM samples, and methylation of 8 genes in 1% of MM samples. When more than one gene was methylated, no correlation was found between the methylation of a given gene and the methylation of any other given gene.

For each of eight patients with MM, we investigated the methylation status of two BM samples at different time points during the course of their disease. Methylation results together with data on time differences, chemotherapy regimens, and disease stage are shown in Figure 4. Comparing the methylation patterns in these samples, we observed methylation in the second sample in 10 cases in which the first sample was not methylated. In six cases, methylation was detected in both the first and second samples. In one patient, methylation of p73 was seen in the first sample but not in the second sample. One patient exhibited methylation of five genes (p16, p15, ECAD, DAPK, and p73) in the first sample; in the second sample, four of those genes (p16, ECAD, DAPK, and p73) were found to be unmethylated, with p15 remaining methylated and TIMP3 methylation being newly observed.

Figure 4.

Comparison of methylation results for samples obtained from each patient at two different time points. SDT: standard-dose therapy; 0: no chemotherapy.

We also compared the methylation patterns of p16, TIMP3, p15, ECAD, DAPK, p73, RASSF1A, p14, MGMT, and RARβ between MGUS samples and MM samples. The methylation frequencies of p16, TIMP3, DAPK, p73, RASSF1A, p14, MGMT, and RARβ were similar in both MGUS and MM samples, whereas methylation of p15 was detected in a greater proportion of MM samples (P = 0.054). SMM samples exhibited methylation frequencies that were similar to those observed in MM samples. The methylation frequencies noted in PCL samples were similar to those observed in MM samples for the TIMP3, p15, DAPK, p73, RASSF1A, p14, MGMT, and RARβ genes and greater than what was observed in MM samples for the p16 (57%) and ECAD (57%) genes.

It is noteworthy that methylation of ECAD was not detected in MGUS samples but was noted in 20% of SMM samples, 27% of MM samples, and 57% of PCL samples. The difference in ECAD methylation between MGUS samples and MM samples was statistically significant (P = 0.002). Because the numbers of SMM and PCL samples were limited, comparison of ECAD methylation in these samples with ECAD methylation in MGUS or MM samples was not possible.

Relation between Methylation and Clinicopathologic Parameters

Methylation results were analyzed for potential correlations with clinicopathologic characteristics, including age; gender; hemoglobin, LDH, β2-microglobulin, calcium, and serum creatinine levels; type of paraprotein; type of light chain; tumor stage; tumor grade (morphology of MM cells); and chromosome 13q14 deletion in all patients with MM. Statistically significant correlations were found between ECAD methylation and plasmablastic morphology of myeloma cells (P = 0.012); between ECAD methylation and prognostic factors, including decreased hemoglobin levels and increased LDH levels; and between DAPK methylation and increased β2-microglobulin levels. In addition, patients who had BM samples with four or more methylated genes exhibited higher β2-microglobulin and serum creatinine levels. These data are summarized in Table 2. However, we did not find methylation status to be correlated with age, gender, disease stage, type of paraprotein, type of light chain, calcium levels, or chromosome 13q14 deletion. Methylation results also were analyzed for correlations with clinicopathologic characteristics in only those patients who were treated with standard-dose melphalan; however, no statistically significant correlations were found in these patients. No association was found between methylation of any of the other investigated genes and overall survival in patients with MM. With regard to ECAD, we observed a trend toward poorer prognosis for patients who had MM with ECAD methylation (Fig. 5).

Table 2. Statistically Significant Correlations between Methylation and Certain Clinicopathologic Characteristics
Methylation parameterNo. of patientsMean ± SD of correlated parameterP value
  1. SD: standard deviation; LDH: lactate dehydrogenase; ECAD: E-cadherin gene; U: units; DAPK: death-associated protein kinase gene.

ECAD Hemoglobin (g/dL)0.024
 Methylated29 10.3 ± 2.2 
 Unmethylated82 11.4 ± 2.3 
ECAD LDH (U/L)0.035
 Methylated30 219 ± 120 
 Unmethylated79 179 ± 73 
DAPK β2-microglobulin (mg/L)0.015
 Methylated22 11.0 ± 18.8 
 Unmethylated82 5.2 ± 5.4 
No. of methylated genes β2-microglobulin (mg/L)0.005
 < 489 5.3 ± 5.4 
 ≥ 415 13.2 ± 22.4 
No. of methylated genes Serum creatinine (mg/dL)0.01
 < 493 1.7 ± 1.8 
 ≥ 416 3.1 ± 2.8 
Figure 5.

Overall survival and E-cadherin gene methylation in 108 patients with multiple myeloma (MM) who exhibited (n = 30; median survival, 40.5 months) or did not exhibit (n = 78; median survival, 59.5 months) E-cadherin gene methylation (P = 0.13).

DISCUSSION

To date, only limited information has been published on methylation in monoclonal gammopathies. Thus, the objective of the current study was to determine the methylation patterns of 10 different genes in patients with MGUS, SMM, MM, or PCL and to investigate possible correlations between methylation and the clinicopathologic characteristics of these patients. Study genes were chosen based on their frequent methylation in other hematologic malignancies and/or solid tumors.

MSP analysis was used for methylation analysis, because it is a very sensitive and fast method and is much better suited to the analysis of large numbers of samples compared with bisulfite genomic sequencing.32, 34–36 MSP has been used to detect methylation in both unsorted mononuclear cells and sorted CD138-positive cells.13–15, 35, 37 Although several authors have studied the methylation of p16, p57, SOCS1, and DAPK in unsorted mononuclear cells in patients with MM and various B-cell malignancies,15, 35, 37–39 Gonzalez et al.13 and Guillerm et al.14 used sorted CD138-positive cells to investigate the methylation of p15 and p16. However, the reported methylation frequencies of p16 were comparable in both types of studies. In addition, the methylation frequencies of the genes p15 and p16 in the current study were similar to the frequencies reported by Gonzalez et al.13 and Guillerm et al.14 These findings indicate that it is not mandatory to sort plasma cells from BM samples for MSP analysis.

The current results demonstrate that methylation of the investigated genes is already detectable in MGUS and is a frequent event in MM. These results are similar to the results reported by other authors, who demonstrated frequent methylation of p16 and p15 in MGUS and MM.13, 14, 40, 41 The higher frequency of DAPK methylation reported by Ng et al. may have been attributable to the limited number of patients in that study.15 Another gene that reportedly is methylated frequently in MM is SOCS1.16 This gene, which is involved in the Jak/STAT pathway, was found to be methylated in 23 of 35 MM samples. However, confirmation of these results in a larger study, along with the analysis of possible correlations of clinicopathologic characteristics with SOCS1 methylation, is required.

MGUS plasma cells share many abnormal features with MM plasma cells. It has been recognized that cytogenetic abnormalities, including 14q32 translocations, deletions of 13q, and numeric gains of multiple chromosomes, occur at the level of MGUS, suggesting early onset of karyotypic instability in monoclonal gammopathies.42 Using gene expression profiling, large numbers of abnormally expressed genes have been observed in MGUS, leading to the notion that based on gene expression analysis, MGUS plasma cells and MM plasma cells are largely indistinguishable.43 Results of the current investigation also indicate that methylation patterns are characteristic of both MGUS and MM. Although MGUS plasma cells exhibit many genetic and epigenetic abnormalities, it remains unclear which critical events may contribute to the transformation from MGUS to MM. In this respect, it is interesting to note that ECAD was unmethylated in all MGUS samples in the current series. This finding is in contrast to the increasing percentages of ECAD methylation observed in SMM, MM, and PCL samples, suggesting that ECAD methylation may be a marker for disease progression in monoclonal gammopathies. The transmembrane glycoprotein ECAD mediates Ca2+-dependent intercellular adhesion, which is essential for normal tissue homeostasis. In a variety of epithelial tumors, loss of ECAD has been associated with invasion and metastasis,44, 45 and methylation has been identified as the main reason for loss of expression.46, 47 In agreement with the hypothesis that ECAD methylation is an event that indicates disease progression, in two patients with follow-up samples, ECAD was found to be unmethylated at the time of diagnosis but methylated at the time of disease progression (Fig. 4; Patients 1 and 4).

In most cases, by comparing methylation patterns in samples obtained during follow-up, we observed either methylation in both the first sample and second sample or methylation in the second sample without methylation in first sample. An obvious exception was Patient 8 (Fig. 4), who had discordant results in the first and second samples, possibly indicating the outgrowth of a different myeloma cell clone during disease progression after a period of > 5 years without specific antimyelotic therapy.

Another objective of the current study was to investigate whether methylation of the 10 study genes was associated with clinicopathologic parameters, and particularly overall survival, in these patients. It is noteworthy that in the current study, methylation of ECAD and DAPK in patients with MM was associated with poor prognostic markers, and ECAD methylation also was correlated with plasmablastic morphology of MM cells. Although the difference in overall survival between patients with and without ECAD methylation was not statistically significant, patients who had ECAD methylation appeared to have a poorer prognosis compared with patients who did not (Fig. 5). Methylation of any of the other investigated genes or of a subset of these genes was not associated with overall survival. This finding is in agreement with results of Guillerm et al.,14 who reported that methylation of p15 and p16 in patients with MGUS and patients with MM was not associated with shorter overall survival. However, this finding is in contrast to the results of Mateos et al.,41 who observed a statistically significant survival disadvantage for patients with MM who had p16 methylation compared with patients who did not have p16 methylation.

It is worth noting that the demethylating agent 5-aza-2′-deoxycytidine has shown activity in patients with myelodysplasia48 and acute myelocytic leukemia.49 Demethylation of p15 has been found to be correlated with clinical response in patients with chronic myelocytic leukemia.50, 51 Recently, it has been suggested that there is synergy between methylation and histone deacetylase activity.52 Thus, investigation of whether demethylating agents and/or histone deacetylase inhibitors may be of benefit in the treatment of patients with MM, as suggested by the results of studies involving MM cell lines, is warranted.53–55

In conclusion, the current study demonstrated that methylation is an important mechanism for silencing known and putative TSGs in the pathogenesis of monoclonal gammopathies. Although methylation is already detectable in of most of the genes investigated in MGUS, the results of the current study suggest that ECAD methylation may be a marker of disease progression.

Acknowledgements

The authors thank Alexandra Budinsky for her assistance with statistical analysis and Dr. Albert Wölfler for his assistance in collecting clinicopathologic data.

Ancillary