Inactivation of p16, RUNX3, and HPP1 occurs early in Barrett's-associated neoplastic progression and predicts progression risk

Article metrics

Abstract

Patients with Barrett's esophagus (BE) are at increased risk of developing esophageal adenocarcinoma (EAC). Clinical neoplastic progression risk factors, such as age and the length of the esophageal BE segment, have been identified. However, improved molecular biomarkers predicting increased progression risk are needed for improved risk assessment and stratification. Using real-time quantitative methylation-specific PCR, we screened 10 genes (HPP1, RUNX3, RIZ1, CRBP1, 3-OST-2, APC, TIMP3, p16, MGMT, p14) for promoter hypermethylation in 77 EAC, 93 BE, and 64 normal esophagus (NE) specimens. A subset of genes manifesting significant differences in methylation frequencies between BE and EAC was then analysed in 20 dysplastic specimens. All 10 genes except p14 were frequently methylated in EACs, with RUNX3, HPP1, CRBP1, RIZ1, and OST-2 representing novel methylation targets in EAC and/or BE. p16, RUNX3, and HPP1 displayed increasing methylation frequencies in BE vs EAC. Furthermore, these increases in methylation occurred early, at the interface between BE and low-grade dysplasia (LGD). To demonstrate the silencing effect of hypermethylation, we selected the EAC cells BIC1, in which the HPP1 promoter is natively methylated, and subjected them to 5-aza-2′-deoxycytidine (Aza-C) treatment. Real-time RT–PCR indicated increased HPP1 mRNA levels after 3 days of Aza-C treatment, as well as decreased levels of methylated HPP1 DNA. Hypermethylation of a subset of six genes (APC, TIMP3, CRBP1, p16, RUNX3, and HPP1) was then tested in a retrospective longitudinal study of 99 BE and nine LGD specimens obtained from 53 BE patients undergoing surveillance endoscopy. Only high-grade dysplasia (HGD) or EAC were defined as progression end points. Two patient groups were compared: eight progressors (P) and 45 nonprogressors (NP), using Cox proportional hazards models to determine the relative progression risks of age, BE segment length, and methylation events. Multivariate analyses revealed that only hypermethylation of p16 (odds ratio (OR) 1.74, 95% confidence interval (CI) 1.33–2.20), RUNX3 (OR 1.80, 95% CI 1.08–2.81), and HPP1 (OR 1.77, 95% CI 1.06–2.81) were independently associated with an increased risk of progression, whereas age, BE segment length, and hypermethylation of TIMP3, APC, or CRBP1 were not independent risk factors. In combined analyses, risk was detectable up to, but not earlier than, 2 years preceding neoplastic progression. Hypermethylation of p16, RUNX3, and HPP1 in BE or LGD may represent independent risk factors for the progression of BE to HGD or EAC. These findings have implications regarding risk stratification, early EAC detection, and the appropriate endoscopic surveillance interval for patients with BE.

Introduction

Esophageal adenocarcinoma (EAC) arises in the setting of Barrett's esophagus (BE), a premalignant condition in which the normal squamous epithelium of the lower esophagus is replaced by a metaplastic columnar lining. The incidence of EAC has risen steadily over the past several decades (Powell and McConkey, 1992; Blot and McLaughlin, 1999; Brown et al., 2002). Patients with BE have a 30–125-fold increased risk of developing EAC relative to the general population (Hameeteman et al., 1989). Therefore, regular endoscopic surveillance has been recommended for patients with BE (Sampliner, 2002; Spechler, 2002), and it has been shown that cancers detected in surveillance programs occur at an earlier stage and have a better prognosis (Streitz et al., 1993; Peters et al., 1994; Schnell et al., 2001; Corley et al., 2002). However, endoscopic surveillance has several drawbacks, including its high cost, low yield, and procedure-related risks. As only a small percentage of BE patients ultimately develop EAC, the merits and appropriate interval of endoscopic surveillance for all BE patients have been a source of some debate. Moreover, the accuracy of clinical risk factors to predict early progression in BE is limited. For example, the length of the BE segment has been considered as a major risk factor for progression to EAC (Hameeteman et al., 1989; Miros et al., 1991; Williamson et al., 1991; Iftikhar et al., 1992; Menke-Pluymers et al., 1993; van der Burgh et al., 1996; Avidan et al., 2002) and dysplasia (Gopal et al., 2003). However, several studies have also shown that EAC occurs in short-segment BE (Hamilton et al., 1988; Schnell et al., 1992; Cameron et al., 1995; Sharma et al., 2000).

Similarly, the presence and grade of dysplasia currently constitute the basis for risk stratification in patients with BE, but significant problems have emerged in studies of dysplasia. These problems have included poor reproducibility of dysplasia interpretation (except for high-grade dysplasia (HGD)), poor predictive value for negative, indefinite, and low-grade dysplasia (LGD), and sampling error (Alikhan et al., 1999; Montgomery et al., 2001a, 2001b). Moreover, multiple studies have shown that the majority of patients do not progress to dysplasia or carcinoma, with an annual rate of neoplastic transformation of BE to EAC of only 0.4–0.5% (Drewitz et al., 1997; Shaheen and Ransohoff, 2002).

More precise markers to stratify the risk of progression of BE to EAC would permit more effective targeting of surveillance endoscopy to those patients at the highest risk of progression. In addition, identifying the molecular events accompanying the development of LGD, HGD, and EAC would aid in identifying patients at greatest risk of progression, while simultaneously providing clues to the biology underlying neoplastic transformation in BE.

However, only a limited number of molecular events have been evaluated as prospective markers in EAC (McManus et al., 2004). For example, loss of heterozygosity (LOH) at chromosome 17p (17p-LOH) and ploidy status define the risk of developing EAC in prospective trials (Teodori et al., 1998; Barrett et al., 1999; Rabinovitch et al., 2001; Reid et al., 2001). Aneuploid subclones were detectable prior to the p53 inactivation (Neshat et al., 1994; Galipeau et al., 1996). In addition, cell cycle-regulatory genes including p16 (CDKN2, INK4a) and cyclin D1 have been implicated in the early stages of EAC pathogenesis (Arber et al., 1996; Bani-Hani et al., 2000; Wong et al., 2001). p16 has been shown to be inactivated through mutation or deletion in EAC (Suzuki et al., 1995); however, both point mutation and homozygous deletion are rare in EAC (Barrett et al., 1996), suggesting an alternative mode of p16 gene inactivation. Recently, promoter hypermethylation has emerged as a fundamental epigenetic mechanism of tumor suppressor gene silencing in carcinogenesis (Esteller, 2003; Herman and Baylin, 2003; Jones, 2003). Certain methylation events can occur with aging or inflammation (which itself underlies BE), conditions associated with increased neoplastic risk (Issa et al., 1994, 1996, 2001; Issa, 1999). Hypermethylation of several genes has been observed in primary BE and EAC tissues and is the most frequent mechanism of APC and p16 inactivation in EAC (Wong et al., 1997, 2001; Eads et al., 2000, 2001; Corn et al., 2001; Bian et al., 2002). However, to our knowledge, prospective studies of the predictive value of epigenetic alterations in BE have not been reported. In the current study, we first sought to identify novel hypermethylation targets in EAC and its precursor lesions, BE and dysplasia (LGD and HGD). We therefore performed a cross-sectional analysis of promoter hypermethylation of 10 genes in 77 EACs, 93 BEs, and 64 normal esophagus (NE) tissue specimens to identify genes that were methylated early in neoplastic progression, but not in NE mucosa. In addition, in 14 LGD and six HGD specimens, we restricted our study to genes showing different methylation levels or frequencies in BE vs EAC. Next, we performed a retrospective longitudinal analysis of promoter methylation of a subset of six genes in 53 BE patients with known long-term outcome in a surveillance program to validate that methylation events may predict progression of BE or BE-associated LGD to HGD or EAC.

Results

Cross-sectional analysis of promoter hypermethylation in NE, BE, dysplasia, and EACs

We analysed promoter hypermethylation of 10 genes in 77 EACs, 93 BEs, and 64 NEs (17 NEs from EAC patients and 47 NEs from BE patients). p14 was not methylated in EAC, while MGMT was unable to discriminate NE from EAC. All eight remaining genes showed a highly discriminative receiver–operator characteristic (ROC) curve profile of promoter hypermethylation, which clearly distinguished EAC from NE. ROC curves with corresponding areas under the ROC curves (AUROCs) for p16, RUNX3, and HPP1 are shown in Figure 1a–c.

Figure 1
figure1

ROC curve analysis of NMV of NE vs EAC for hypermethylation of (a) p16, (b) RUNX3, and (c) HPP1. The AUROC for each gene is shown and conveys each gene's accuracy in distinguishing NE from EAC in terms of its sensitivity and specificity

Normalized methylation values (NMVs) were significantly higher in BE than in NE for all genes studied except p14. NMVs were significantly higher in EAC than in BE for p16 (P<0.0001, Student's t-test) and RUNX3 (P=0.01, Student's t-test).

We selected cutoff values according to the ROC curve with maximum sensitivity and specificity and calculated the frequency of hypermethylation in 64 NE, 93 BE, and 77 EAC (Table 1; AUROC is shown as a measurement of discriminative value). All genes, including the novel methylation targets RUNX3, CRBP1, RIZ1, 3-OST-2, and HPP1, were methylated in at least 20% of EACs. Based on a dichotomized categorization, HPP1, RUNX3, and p16 were methylated significantly more frequently in EAC than in BE, whereas 3-OST-2 was more frequently methylated in BE than in EAC.

Table 1 Methylation frequencies of 10 genes in NE, BE and EAC

Based on the finding that p16, RUNX3, and HPP1 were differentially methylated between BE and EAC, we hypothesized that hypermethylation of these three genes was involved in the Barrett's metaplasia–dysplasia–adenocarcinoma sequence. To find further evidence supporting this hypothesis, we analysed 14 LGD and six HGD specimens for methylation levels and frequencies of these three genes. Among HGD specimens, hypermethylation rates of all three genes were 67% (4/6 samples). Among 14 LGD samples, hypermethylation rates varied somewhat for these three genes: methylation of HPP1 occurred in 11 (79%), of RUNX3 in eight (57%), and of p16 in seven (50%). Differences in methylation levels (NMVs) between LGDs and HGDs were not significant. NMVs of p16, RUNX3, and HPP1 in NE, BE, LGD, HGD, and EAC are shown in Figure 2. Significant differences in NMVs between BE and dysplasia were observed for all three genes. Comparing BE to LGD alone, RUNX3 (P=0.005, Student's t-test) and p16 (P=0.002, Student's t-test) retained significant, whereas the difference in mean NMVs between BE and LGD for HPP1 just failed to reach significance (P=0.059, Student's t-test).

Figure 2
figure2

Quantitative display of HPP1, p16, and RUNX3 hypermethylation in NE (group 1, n=64), BE (group 2, n=93), dysplasia (D, group 3, n=20) and EAC (group 4, n=77). Progressive increases in methylation (NMV) from NE to BE to D are seen for all the three genes, with a slight decline from D to EAC. Methylation values are displayed as means with 95% CIs. *P<0.05; **P<0.01; $P<0.005; #P<0.001; ##P< 0.0001

5-Aza-dC treatment of EAC cell line BIC-1

To demonstrate the gene-silencing effect of HPP1 hypermethylation in Barrett's neoplastic progression, we selected BIC1 esophageal carcinoma cells, in which the HPP1 promoter is natively methylated, and subjected them to 5-aza-2′-deoxycytidine (Aza-C) treatment. Real-time RT–PCR analysis revealed increased levels of HPP1 mRNA levels as well as decreased levels of methylated HPP1 after Aza-C treatment (Figure 3).

Figure 3
figure3

Correlation of HPP1 methylation and mRNA expression in 5-Aza-dC-treated BIC cells. On the left side MSP values of HPP1 are presented, on the right side HPP1 mRNA expression levels are shown. The RT–PCR value was defined by the ratio of the expression value on a certain and the untreated expression value (the expression values for each timepoint itself are normalized to ACTB). This figure shows a trend toward increasing HPP1 expression as HPP1 methylation decreases after Aza-C treatment

Correlation of RUNX3 promoter hypermethylation and RUNX3 mRNA levels in 21 EACs

To validate the functional effect RUNX3 hypermethylation exerted on gene function in primary esophageal tumors, we correlated promoter hypermethylation and mRNA levels of RUNX3 in 21 EACs. In all, 13 EACs did not exhibit promoter methylation, while eight EACs did manifest RUNX3 hypermethylation (Figure 4). All eight cases showing RUNX3 hypermethylation exhibited low RUNX3 RNA levels. Conversely, all cases showing high RUNX3 expression levels were unmethylated. A subset of unmethylated samples exhibited low levels of RUNX3 RNA as well, suggesting alternative modes of RUNX3 silencing in these samples (correlation coefficient R=−0.192498).

Figure 4
figure4

Scatter plot display of RUNX3 hypermethylation and expression levels in EACs. RUNX3 hypermethylation and expression levels in 21 EACs are shown in scatter plot format (X-axis, RUNX3 mRNA expression level; Y-axis, RUNX3 methylation level). All cases with RUNX3 hypermethylation exhibited low RUNX3 RNA levels. Conversely, all cases with high expression levels exhibited low methylation levels. A subset of unmethylated samples also exhibited low levels of RUNX3 RNA, suggesting alternative modes of RUNX3 silencing in these samples

Retrospective longitudinal validation study

Clinical characteristics of the patients in the retrospective longitudinal study are shown in Table 2. Age and length of Barrett's segment were determined at the date of the first endoscopic biopsy in the study. Ethnic origin was self-reported. Gender, age, and race did not differ significantly among groups. Barrett's segment length ranged from 1 to 15 cm and was significantly lower in patients with BE alone vs patients with dysplasia or cancer (BE vs LGD, HGD or EAC, P=0.0001, t-test). Follow-up was significantly longer in patients having BE or LGD as their most advanced lesions vs patients who developed HGD or EAC (P<0.0005, t-test). Number of samples per patient did not differ between these two patient groups.

Table 2 Clinical characteristics of patients from the retrospective longitudinal study

We analysed a subset of six genes (p16, RUNX3, HPP1, APC, TIMP3, CRBP1) showing high methylation frequencies in EAC but rare methylation in NE in 53 patients under surveillance for BE. We classified patients as NP if their most advanced lesion was LGD during follow-up (n=45 patients) or as P if they developed HGD or EAC during follow-up (n=8 patients). We investigated 87 pre-progression tissue samples from the NP group (84 BE, three LGD) and 21 samples from the P group (15 BE, six LGD). Thus, none of the tissues analysed in this study were HGD or EAC; only the pre-progression metaplastic BE or LGD were considered in this analysis. We calculated time-to-progression distributions using the Kaplan–Meier method, and employed a Cox proportional hazards model to correlate progression-free interval with methylation levels as continuous variables. In this analysis, we also included age and length of BE segment at the timepoint each sample was obtained (Table 3). In univariate analyses of samples, hypermethylation of p16 (odds ratio (OR) 1.54, 95% confidence interval (CI) 1.24–1.83), RUNX3 (OR 1.75, 95% CI 1.33–2.21), HPP1 (OR 1.84, 95% CI 1.29–2.54), CRBP1 (OR 1.64, 95% CI 1.21–2.13), TIMP3 (OR 1.68, 95% CI 1.14–2.38), and BE segment length (OR 1.51, 1.01–2.27) was associated with a significantly increased risk of progression to HGD or EAC, whereas age and hypermethylation of APC were not significant risk factors.

Table 3 Cox proportional hazards model of risk factors for progression to HGD or EAC in patients with BE (n=53 patients, n=108 samples)

We also performed multivariate analyses. These multivariate analyses revealed that only hypermethylation of p16 (OR 1.74, 95% CI 1.33–2.20, P=0.0005), RUNX3 (OR 1.80, 95% CI 1.08–2.81), and HPP1 (OR 1.77, 95% CI 1.06–2.81) was independently associated with a significantly increased risk of progression to HGD or EAC. BE segment length, age, and hypermethylation of TIMP3, APC, and CRBP1 were not independent significant risk factors (Table 3).

A combined hazard ratio (HR) index was calculated from the Cox proportional hazards model for each sample, including age and BE segment length, as well as the weighted influences of all six continuous methylation values (Figure 5). At 2 years or less preceding progression to HGD or EAC, the HR index was >5 for samples from progressors, whereas samples from nonprogressors had an HR index of <5. Thus, when the HR index was >5, the likelihood of progression within 2 years was increased. However, HR index was not predictive of progression for samples derived more than 2 years prior to progression. This finding suggests that the HR index begins to predict progression at 2 years prior to its occurrence.

Figure 5
figure5

Correlation of combined HR and time to progression to HGD or EAC. A combined Cox's HR index was calculated from the multivariate Cox model, taking into account NMVs of all six genes and age and BE segment length at the timepoint each sample was obtained. ▪, specimens from progressors (P); , specimens from nonprogressors (NP). At 2 years or sooner preceding progression (i.e., the end point), the HR indices of all P specimens were greater than 5, while the HR indices of most NP specimens remained below 5. This figure suggests that the HR index begins to predict progression approximately 2 years in advance of its occurrence

Discussion

The most established clinical marker for the risk of developing EAC in BE is dysplasia. However, the natural history of LGD is not well characterized, with rates of progression to HGD or EAC ranging from 10 to 30%, as well as frequent regression to BE (Hameeteman et al., 1989; Weston et al., 1997; O'Connor et al., 1999; Weston et al., 1999, 2001; Skacel et al., 2000; Conio et al., 2003). Moreover, histological classification of dysplasia, especially LGD, has high interobserver variability (Alikhan et al., 1999; Montgomery et al., 2001a, 2001b).

For these reasons, molecular biomarkers are being sought for improved risk classification of BE patients. Recently, phases of biomarker development for the early detection of cancer have been defined by the Early Detection Research Network (EDRN) by the National Cancer Institute (Sullivan Pepe et al., 2001).

In previous studies, aneuploidy and LOH at chromosome 17p defined patients at high risk for developing EAC in prospective trials, respresenting Phase 4 biomarker development studies (Teodori et al., 1998; Barrett et al., 1999; Rabinovitch et al., 2001; Reid et al., 2001). In addition, inactivation of p16 was shown to be an early and frequent event in EAC (Suzuki et al., 1995; Barrett et al., 1996, 1999; Maley et al., 2004).

In addition to genetic alterations such as LOH or point mutation, hypermethylation, the epigenetic addition of methyl groups to DNA at CpG islands has been established as a common mechanism of gene inactivation in human carcinogenesis (Esteller, 2003; Herman and Baylin, 2003; Jones, 2003). Certain methylation events can occur with aging or inflammation (which itself underlies BE), conditions associated with increased neoplastic risk (Issa et al., 1994, 1996, 2001). Hypermethylation of several genes, including p16, APC, HPP1, and TIMP3 has been observed in primary BE and EAC tissues (Wong et al., 1997, 2001; Eads et al., 2000, 2001; Corn et al., 2001; Bian et al., 2002; Geddert et al., 2004; Sarbia et al., 2004).

By performing a cross-sectional analysis of candidate methylation targets in EAC, BE, and NE, we sought to identify novel epigenetic gene inactivation sites in EAC representing a Phase 1–Phase 2 study of biomarker development, as defined by the EDRN (Sullivan Pepe et al., 2001). In the current study, promoter hypermethylation of RUNX3, RIZ1, 3-OST-2, and CRBP1 occurred frequently in EAC and BE specimens. CRBP1 is known to be frequently hypermethylated in cancers of the stomach, colon, liver, lymphatic, and hematologic systems (Esteller et al., 2002), while hypermethylation of RUNX3 has been reported in gastric cancer (Li et al., 2002; Waki et al., 2003; Oshimo et al., 2004b) and other tumors, including those of the liver, larynx, lung, breast, and prostate (Kang et al., 2004; Kim et al., 2004; Li et al., 2004; Xiao and Liu, 2004). Hypermethylation of RIZ1 has been reported in breast, liver, gastric, nasopharyngeal, and parathyroid cancers (Du et al., 2001; Carling et al., 2003; Chang et al., 2003; Tokumaru et al., 2003; Oshimo et al., 2004a). 3-OST-2 hypermethylation in human cancer has been identified in only one study thus far, involving breast, colon, lung, and gastric cancers (Miyamoto et al., 2003). Since the mechanistic role of 3-OST-2 in carcinogenesis is unclear, we investigated its expression in normal squamous esophagus by quantitative RT–PCR and found it expressed only at low levels (data not shown). Therefore, we concluded that although 3-OST-2 hypermethylation may represent a biomarker of BE per se, there is little evidence that 3-OST-2 functions as a tumor suppressor gene in the esophagus.

In addition, the current study confirmed a recent finding of frequent HPP1 hypermethylation in EAC (Geddert et al., 2004). However, the previous study did not examine any BE specimens (Geddert et al., 2004). We found HPP1 to be very frequently methylated in BE specimens. Thus, although Geddert et al. proposed HPP1 methylation as an esophageal cancer-specific event, our findings suggest that HPP1 methylation occurs at the metaplastic step, rather than being cancer-specific.

Hypermethylation of APC, TIMP3, and p16 was also observed in BE in the current study, in agreement with previously published work (Eads et al., 2001; Wong et al., 2001; Bian et al., 2002). In this context, it should be noted that p16 inactivation, either by homozygous or heterozygous deletion (LOH), methylation, point mutation, or a combination of these events, is one of the earliest events in the pathogenesis of EAC (Galipeau et al., 1999; Wong et al., 2001). In contrast, however, Klump et al. (1998) detected frequent p16 methylation in LGD, HGD, and indefinite dysplasia, but found it in only 3% of BE. Our cross-sectional prevalence findings agree with several previously published reports on p16 methylation (Eads et al., 2000, 2001; Wong et al., 2001; Bian et al., 2002).

Notably, we observed significant differences between BE and EAC in the prevalence and levels of p16, RUNX3, and HPP1 methylation. We therefore sought to determine whether these epigenetic events occurred early or late in esophageal neoplastic transformation. When we investigated BE-associated dysplastic lesions, we found significant differences in methylation levels between BE and dysplasia for all three genes. In fact, when dysplastic lesions were compared to EACs, there were no significant differences in methylation levels between these two groups. This finding suggested that inactivation of p16 RUNX3, and HPP1 occurs early in BE-associated neoplastic progression, at the interface between BE and LGD.

While evaluating the gene silencing mechanism of promoter hypermethylation of HPP1, we demonstrated an increased level of HPP1 mRNA levels after 3 days of Aza-C treatment, as well as decreased levels of methylated HPP1 DNA in the EAC cell line BIC1, which has native methylation of the HPP1 promoter. This finding emphasizes the functional role HPP1 hypermethylation appears to play in esophageal neoplastic transformation.

In our process of validating RUNX3 hypermethylation as a functional mechanism of mRNA expression silencing in primary tumors, we found that all cases with RUNX3 hypermethylation displayed low mRNA expression levels, that all cases with high RUNX3 expression were unmethylated, whereas a subset of samples without hypermethylation also showed low mRNA expression levels, suggesting alternative mechanisms of RUNX3 silencing (Bae and Choi, 2004).

Based on the findings from our cross-sectional study and our in vitro and in vivo findings on methylation-induced gene silencing, we performed a retrospective longitudinal Phase 3 biomarker development study to ascertain whether hypermethylation of a panel of genes could identify BE patients at increased risk of progression to HGD or EAC. We included HGD as an end point because it has a significantly higher incidence of progression to EAC (Montgomery et al., 2001b) and a high prevalence of concomitant EAC (Tseng et al., 2003) relative to BE or LGD.

Most importantly, we found that hypermethylation of p16, RUNX3 and HPP1 constituted independent risk factors for the progression to HGD or EAC, whereas BE segment length, which is generally considered a risk factor for EAC (Hameeteman et al., 1989; Miros et al., 1991; Williamson et al., 1991; Iftikhar et al., 1992; Menke-Pluymers et al., 1993; van der Burgh et al., 1996; Avidan et al., 2002), failed to achieve statistical significance in our multivariate model, suggesting that the chosen molecular biomarkers are more powerful in predicting the risk of progression to HGD or EAC.

We calculated a combined HR index for each sample from the Cox proportional hazard model, which showed that specimens from progressors had an HR index >5 at 2 years or less preceding neoplastic progression. This finding suggests that the HR index begins to detect progression at 2 years prior to its occurrence. Stated alternatively, this finding implies that patients with an increased HR index may require more frequent endoscopic surveillance or interventions, such as chemopreventive agents (Heath et al., 2003; Jankowski and Sharma, 2004). Conversely, these findings suggest that even when using this hypermethylation panel, it is difficult to predict progression more than two years in advance.

Several caveats are in order regarding this retrospective longitudinal Phase 3 biomarker development study. Firstly, the number of patients who progressed to HGD or EAC was relatively low, due to the low annual rate of neoplastic transformation of BE to EAC, which is only 0.4–0.5% (Drewitz et al., 1997; Shaheen and Ransohoff, 2002). Secondly, the study was retrospective, due primarily to the low rate of progression in this disease. Thirdly, BE segment length and age, the only clinical factors available to us, were not the only clinical factors impacting on neoplastic progression; for example, duration and severity of gastroesophageal reflux, smoking, the presence or absence of esophageal ulcers, weight, and other parameters have all been implicated to some degree. However, we included these two factors because both are readily ascertainable at the time of endoscopy, making them highly practical in the clinical setting. Fourthly, the combined end point of HGD and EAC is not universal, since regression of HGD has also been reported (Schnell et al., 2001). Even for the combined end point of HGD and EAC, the estimate of the annual progression rate is only 3.3% (Sharma et al., 2001). However, HGD is incorporated as an end point in most studies of the neoplastic progession of EAC. All of these caveats can only be overcome in a concerted Phase 4 prospective multicenter trial involving many large clinical centers (Sullivan Pepe et al., 2001).

In conclusion, in the current study, we sought evidence that promoter hypermethylation influences the neoplastic progression of BE to HGD and EAC. We found evidence that hypermethylation of p16, RUNX3, and HPP1 may indeed be associated with a significantly increased risk of neoplastic progression, and that these events themselves have the potential to serve as clinically useful biomarkers of the risk of progression. Therefore, these three genes appear to constitute promising biomarker candidates worthy of validation in future large prospective multicenter trials. Moreover, this analysis and other methylation analyses like it can be extended to other premalignant conditions in need of better biomarkers of increased neoplastic progression risk, such as ulcerative colitis, where endoscopic surveillance faces similar dilemmas.

Materials and methods

Patients

All patients provided written informed consent under a protocol approved by the Institutional Review Boards of the University of Maryland and Baltimore VA Hospitals. Patients underwent endoscopy (esophagogastroduodenoscopy (EGD)) or surgery at the Baltimore VA and University of Maryland Hospitals. Biopsies were taken using a standardized biopsy protocol. At each EGD, four-quadrant biopsies were obtained at 2-cm intervals throughout the grossly apparent BE segment (1 cm intervals on follow-up after an EGD with LGD or HGD). Research tissues were obtained from aliquots of grossly apparent Barrett's epithelium or mass lesions in patients manifesting these changes at endoscopic examination or after surgical removal. Simultaneously obtained parallel aliquots were sent for histological examination at the time of the endoscopy or surgery. The diagnosis and grading of dysplasia was made by between at least two pathologists at the two participating institutions.

For the cross-sectional part, 77 EACs, 93 BEs, 20 dysplasias (14 LGDs, six HGDs), and 64 NEs (17 NEs from EAC patients; 47 NEs from BE patients) were included. NE specimens were taken at least 5 cm away from macroscopic BE or mass lesions. Clinical characteristics (age, sex, and ethnic origin) of included patients did not differ between groups (NE, BE, LGD, HGD, and EAC).

For the retrospective-longitudinal study, patients were enrolled sequentially during endoscopy at the Baltimore VA and University of Maryland Hospitals since 1992. We performed a retrospective study of BE tissues of 53 patients from a surveillance program at the University of Maryland and VA Hospitals. The detailed endoscopic surveillance protocol is provided as Supplementary File 1. Research tissues were obtained during endoscopy from aliquots of grossly apparent BE or grossly abnormal lesions at endoscopic examination. Simultaneously obtained parallel tissue aliquots were sent for histological examination at the initial endoscopy. Histologic confirmation of gross endoscopic impressions was obtained in all cases. The diagnosis and grading of dysplasia was made by members of an expert team of gastrointestinal pathologists at the two participating institutions. There was concordance between at least two pathologists for all diagnoses of dysplasia. All patients provided written informed consent under a protocol approved by the Institutional Review Boards of the University of Maryland and Baltimore VA Hospitals.

We defined the development of HGD or EAC as the primary end point for progression, resulting in NP and P groups. Nonprogressors were censored at the date of their last follow-up endoscopy. All samples were obtained prior to the date of the end point (HGD or EAC) or of censoring.

Of the 53 patients in this analysis, 38 patients did not develop LGD, HGD, or EAC during the follow-up period, whereas seven patients had LGD at least at one timepoint as their highest-ranked pathology, four patients developed HGD, and four patients developed EAC. The four patients who ultimately developed EAC had early-stage lesions (two CIS, two T1N0). All four patients with EAC and three of four patients with HGD underwent esophagectomy, whereas one HGD patient was managed by endoscopic mucosal resection. There were 108 samples evaluated in this study: 87 specimens from different timepoints during surveillance in the NP group, plus 21 samples obtained at different timepoints preceding the development of HGD or EAC in the P group.

DNA and RNA extraction

Tissue specimens were snap-frozen immediately after biopsy or surgical removal and stored in liquid nitrogen until further processing. Genomic DNA from clinical samples and cell lines was extracted as described (Meltzer et al., 1990). RNA was extracted from clinical samples and cell lines using Qiagen RNeasy Mini Kit (Qiagen, Hilden, Germany).

Bisulfite treatment and real-time quantitative methylation-specific PCR (MSP) protocol

DNA was treated with bisulfite to convert unmethylated cytosines to uracils prior to MSP as described previously (Sato et al., 2002a, 2002b; Mori et al., 2004). DNA methylation status and levels of 10 genes were determined with real-time quantitative MSP using the ABI 7700 Sequence Detection (Taqman) System, as described previously (Sato et al., 2002b; Shibata et al., 2002; Mori et al., 2004). Primers and probes for quantitative MSP were as described for p16, TIMP-3, APC, MGMT, RIZ1, HPP1, ACTB, and p14 (Eads et al., 2001; Sato et al., 2002a, 2002b; Tokumaru et al., 2003) or designed anew for RUNX3, CRBP1, and 3-OST-2; primers and probes are available on request. An NMV reflecting the percentage of DNA methylated for the gene of interest (GoI) was defined as follows: NMV=(GoI-S/GoI-FM)/(ACTB-S/ACTB-FM) × 100, where GoI-S and GoI-FM represent GoI methylation levels in the sample and fully methylated DNAs, respectively, while ACTB-S and ACTB-FM correspond to β-actin in the sample and fully methylated (FM) DNAs, respectively.

Real-time quantitative RT–PCR

The expression level of target genes was measured using the Taqman quantitative RT–PCR system. Probes were designed over exon boundaries of the respective mRNA sequence to avoid amplification of potentially contaminated genomic DNA (primers and probes are available on request). We performed one-step, real-time, quantitative RT–PCR of HPP1 (in EAC cell line BIC1), RUNX3 (in 21 EACs), and 3-OST-2 (in 11 NEs) using the Qiagen QuantiTect Probe RT–PCR Kit (Qiagen, Hilden, Germany). In all, 40 ng of total RNA template was used per reaction. ACTB was used for normalization of data. The standard curve was generated using serial dilutions of human fetal brain poly A mRNA (BD Biosciences, San Jose, CA, USA) for HPP1 and 3-OST-2, and the colon cancer cell line HCT 116 for RUNX3. The expression index was calculated according to the following formula for the relative expression of target mRNA: Expression index=(TarS/TarC)/(ACTB-S/ACTB-C), where TarS and TarC represent levels of mRNA expression for the target gene in the sample and control mRNA, respectively, whereas ACTB-S and ACTB-C correspond to the amplified ACTB levels in the sample and control mRNA, respectively.

5-Aza-dC treatment of EAC cell lines BIC-1

The EAC cell line BIC-1 (Soldes et al., 1999) expresses low levels of HPP1 as measured by quantitative RT–PCR, and exhibits high methylation levels as determined by quantitative MSP. Demethylating 5-Aza-dC treatment was carried out as previously described (Bender et al., 1999). Briefly, 1 × 105 cells were seeded in a 100 mm dish, grown for 24 h (day 0), and were treated with 5 × 10−6 M of 5-Aza-dC for three days (days 1, 2 and 3). The demethylating agent was replaced with medium at the end of the treatment, and once in every 2 days. DNA and RNA were obtained on days 0, 1, 2, 4 and 6.

Data analysis and statistics

For the cross-sectional part of the study, receiver-operator characteristic (ROC) curve analysis was performed using CpG island methylation data for the 77 EAC and 64 NE tissues (Hanley and McNeil, 1982). Using this approach, the AUROC yielded optimal sensitivity and specificity, and corresponding NMV thresholds were calculated for each gene. These cutoff values were applied to the 77 EAC and 64 NE samples to determine the frequency of methylation of each gene in each tissue type.

Based on methylation frequency (>30% methylation in EAC) and specificity (<15% methylation in NE), we selected genes for further analysis of BE specimens. A subset of these genes, namely those with a significantly higher frequency and/or level of hypermethylation in EAC compared to BE, was studied for promoter hypermethylation in LGD and HGD specimens.

For the retrospective-longitudinal part of the study, each specimen in group P, progression-free interval was defined as the time between the date the specimen was obtained and the date at which HGD or EAC was first detected. All specimens in group NP were indicated as censored, and progression-free interval was defined as the number of months between the date the specimen was obtained and the date of the patient's last EGD. Estimates of time-to-progression distributions were calculated using the Kaplan–Meier method (Kaplan and Meier, 1958).

We used a Cox proportional hazards model (Cox, 1972; Cox and Oakes, 1984) to correlate progression-free interval with methylation levels as continuous variables, as well as including age and length of BE segment at the time each sample was obtained. The Cox proportional hazards model was calculated as follows:

where t denotes the progression-free time interval, h0(t) indicates the baseline hazard, and h(t, …) denotes the resultant hazard, given by the values (v1, … ,vn), of the n covariates for each case, at the respective progression-free time interval. N coefficients (b1, …, bn) were determined by maximum likelihood regression methods. The univariate and multivariate OR and 95% CI were calculated for each gene, age, and BE segment length. This HR index was correlated with progression-free intervals for the P vs the NP specimen groups.

P-values less than 0.05 were considered statistically significant. All tests were two-sided. ROC curve analysis was performed using Analyse-it© software (Version 1.71, Analyse-it Software, Leeds, UK). For all other tests, the software package Statistica (version 6.1; StatSoft, Inc., Tulsa, OK) was used.

References

  1. Alikhan M, Rex D, Khan A, Rahmani E, Cummings O and Ulbright TM . (1999). Gastrointest. Endosc., 50, 23–26.

  2. Arber N, Lightdale C, Rotterdam H, Han KH, Sgambato A, Yap E, Ahsan H, Finegold J, Stevens PD, Green PH, Hibshoosh H, Neugut AI, Holt PR and Weinstein IB . (1996). Cancer Epidemiol. Biomarkers Prev., 5, 457–459.

  3. Avidan B, Sonnenberg A, Schnell TG, Chejfec G, Metz A and Sontag SJ . (2002). Am. J. Gastroenterol., 97, 1930–1936.

  4. Bae SC and Choi JK . (2004). Oncogene, 23, 4336–4340.

  5. Bani-Hani K, Martin IG, Hardie LJ, Mapstone N, Briggs JA, Forman D and Wild CP . (2000). J. Natl. Cancer Inst., 92, 1316–1321.

  6. Barrett MT, Sanchez CA, Galipeau PC, Neshat K, Emond M and Reid BJ . (1996). Oncogene, 13, 1867–1873.

  7. Barrett MT, Sanchez CA, Prevo LJ, Wong DJ, Galipeau PC, Paulson TG, Rabinovitch PS and Reid BJ . (1999). Nat. Genet., 22, 106–109.

  8. Bender CM, Gonzalgo ML, Gonzales FA, Nguyen CT, Robertson KD and Jones PA . (1999). Mol. Cell. Biol., 19, 6690–6698.

  9. Bian YS, Osterheld MC, Fontolliet C, Bosman FT and Benhattar J . (2002). Gastroenterology, 122, 1113–1121.

  10. Blot WJ and McLaughlin JK . (1999). Semin. Oncol., 26, 2–8.

  11. Brown LM, Devesa SS and Fraumeni Jr JF . (2002). Atlas of Oncology: Cancer of the Upper Gastrointestinal Tract, Posner MC, Vokes EE and Weichselbaum RR (eds). BC Decker Inc.: London, pp. 1–22.

  12. Cameron AJ, Lomboy CT, Pera M and Carpenter HA . (1995). Gastroenterology, 109, 1541–1546.

  13. Carling T, Du Y, Fang W, Correa P and Huang S . (2003). Surgery, 134, 932–939; discussion 939–940.

  14. Chang HW, Chan A, Kwong DL, Wei WI, Sham JS and Yuen AP . (2003). Clin. Cancer Res., 9, 1033–1038.

  15. Conio M, Blanchi S, Lapertosa G, Ferraris R, Sablich R, Marchi S, D'Onofrio V, Lacchin T, Iaquinto G, Missale G, Ravelli P, Cestari R, Benedetti G, Macri G, Fiocca R, Munizzi F and Filiberti R . (2003). Am. J. Gastroenterol., 98, 1931–1939.

  16. Corley DA, Levin TR, Habel LA, Weiss NS and Buffler PA . (2002). Gastroenterology, 122, 633–640.

  17. Corn PG, Heath E, Heitmiller R, Fogt F, Forastiere AA, Herman JG and Wu TT . (2001). Clin. Cancer Res., 7, 2765–2769.

  18. Cox DR . (1972). J. Roy Statist. Soc. B, 34, 187–220.

  19. Cox DR, Oakes D and Oakes DO . (1984). Analysis of Survival Data, 1st edn. CRC Press: Boca Raton, FL. ISBN: 041224490X.

  20. Drewitz DJ, Sampliner RE and Garewal HS . (1997). Am. J. Gastroenterol., 92, 212–215.

  21. Du Y, Carling T, Fang W, Piao Z, Sheu JC and Huang S . (2001). Cancer Res., 61, 8094–8099.

  22. Eads CA, Lord RV, Kurumboor SK, Wickramasinghe K, Skinner ML, Long TI, Peters JH, DeMeester TR, Danenberg KD, Danenberg PV, Laird PW and Skinner KA . (2000). Cancer Res., 60, 5021–5026.

  23. Eads CA, Lord RV, Wickramasinghe K, Long TI, Kurumboor SK, Bernstein L, Peters JH, DeMeester SR, DeMeester TR, Skinner KA and Laird PW . (2001). Cancer Res., 61, 3410–3418.

  24. Esteller M . (2003). Lancet Oncol., 4, 351–358.

  25. Esteller M, Guo M, Moreno V, Peinado MA, Capella G, Galm O, Baylin SB and Herman JG . (2002). Cancer Res., 62, 5902–5905.

  26. Galipeau PC, Cowan DS, Sanchez CA, Barrett MT, Emond MJ, Levine DS, Rabinovitch PS and Reid BJ . (1996). Proc. Natl. Acad. Sci. USA, 93, 7081–7084.

  27. Galipeau PC, Prevo LJ, Sanchez CA, Longton GM and Reid BJ . (1999). J. Natl. Cancer Inst., 91, 2087–2095.

  28. Geddert H, Kiel S, Iskender E, Florl AR, Krieg T, Vossen S, Gabbert HE and Sarbia M . (2004). Int. J. Cancer, 110, 208–211.

  29. Gopal DV, Lieberman DA, Magaret N, Fennerty MB, Sampliner RE, Garewal HS, Falk GW and Faigel DO . (2003). Digest. Dis. Sci., 48, 1537–1541.

  30. Hameeteman W, Tytgat GN, Houthoff HJ and van den Tweel JG . (1989). Gastroenterology, 96, 1249–1256.

  31. Hamilton SR, Smith RR and Cameron JL . (1988). Hum. Pathol., 19, 942–948.

  32. Hanley JA and McNeil BJ . (1982). Radiology, 143, 29–36.

  33. Heath EI, Canto MI, Wu TT, Piantadosi S, Hawk E, Unalp A, Gordon G and Forastiere AA . (2003). Dis. Esophagus, 16, 177–186.

  34. Herman JG and Baylin SB . (2003). N. Engl. J. Med., 349, 2042–2054.

  35. Iftikhar SY, James PD, Steele RJ, Hardcastle JD and Atkinson M . (1992). Gut, 33, 1155–1158.

  36. Issa JP . (1999). Crit. Rev. Oncol. Hematol., 32, 31–43.

  37. Issa JP, Ahuja N, Toyota M, Bronner MP and Brentnall TA . (2001). Cancer Res., 61, 3573–3577.

  38. Issa JP, Ottaviano YL, Celano P, Hamilton SR, Davidson NE and Baylin SB . (1994). Nat. Genet., 7, 536–540.

  39. Issa JP, Vertino PM, Boehm CD, Newsham IF and Baylin SB . (1996). Proc. Natl. Acad. Sci. USA, 93, 11757–11762.

  40. Jankowski J and Sharma P . (2004). Aliment. Pharmacol. Ther., 19 (Suppl 1), 54–59.

  41. Jones PA . (2003). Ann. N.Y. Acad. Sci., 983, 213–219.

  42. Kang GH, Lee S, Lee HJ and Hwang KS . (2004). J. Pathol., 202, 233–240.

  43. Kaplan EL and Meier P . (1958). J. Am. Stat. Assoc., 53, 457–481.

  44. Kim TY, Lee HJ, Hwang KS, Lee M, Kim JW, Bang YJ and Kang GH . (2004). Lab. Invest., 84, 479–484.

  45. Klump B, Hsieh CJ, Holzmann K, Gregor M and Porschen R . (1998). Gastroenterology, 115, 1381–1386.

  46. Li QL, Ito K, Sakakura C, Fukamachi H, Inoue K, Chi XZ, Lee KY, Nomura S, Lee CW, Han SB, Kim HM, Kim WJ, Yamamoto H, Yamashita N, Yano T, Ikeda T, Itohara S, Inazawa J, Abe T, Hagiwara A, Yamagishi H, Ooe A, Kaneda A, Sugimura T, Ushijima T, Bae SC and Ito Y . (2002). Cell, 109, 113–124.

  47. Li QL, Kim HR, Kim WJ, Choi JK, Lee YH, Kim HM, Li LS, Kim H, Chang J, Ito Y, Youl Lee K and Bae SC . (2004). Biochem. Biophys. Res. Commun., 314, 223–228.

  48. Maley CC, Galipeau PC, Li X, Sanchez CA, Paulson TG and Reid BJ . (2004). Cancer Res., 64, 3414–3427.

  49. McManus D, Olaru A and Meltzer S . (2004). Cancer Res., 64, 1561–1569.

  50. Meltzer SJ, Mane SM, Wood PK, Resau JH, Newkirk C, Terzakis JA, Korelitz BI, Weinstein WM and Needleman SW . (1990). Cancer Res., 50, 3627–3630.

  51. Menke-Pluymers MB, Hop WC, Dees J, van Blankenstein M and Tilanus HW . (1993). Cancer, 72, 1155–1158.

  52. Miros M, Kerlin P and Walker N . (1991). Gut, 32, 1441–1446.

  53. Miyamoto K, Asada K, Fukutomi T, Okochi E, Yagi Y, Hasegawa T, Asahara T, Sugimura T and Ushijima T . (2003). Oncogene, 22, 274–280.

  54. Montgomery E, Bronner MP, Goldblum JR, Greenson JK, Haber MM, Hart J, Lamps LW, Lauwers GY, Lazenby AJ, Lewin DN, Robert ME, Toledano AY, Shyr Y and Washington K . (2001a). Hum. Pathol., 32, 368–378.

  55. Montgomery E, Goldblum JR, Greenson JK, Haber MM, Lamps LW, Lauwers GY, Lazenby AJ, Lewin DN, Robert ME, Washington K, Zahurak ML and Hart J . (2001b). Hum. Pathol., 32, 379–388.

  56. Mori Y, Yin J, Sato F, Sterian A, Simms LA, Selaru FM, Schulmann K, Xu Y, Olaru A, Wang S, Deacu E, Abraham JM, Young J, Leggett BA and Meltzer SJ . (2004). Cancer Res., 64, 2434–2438.

  57. Neshat K, Sanchez CA, Galipeau PC, Blount PL, Levine DS, Joslyn G and Reid BJ . (1994). Gastroenterology, 106, 1589–1595.

  58. O'Connor JB, Falk GW and Richter JE . (1999). Am. J. Gastroenterol., 94, 2037–2042.

  59. Oshimo Y, Oue N, Mitani Y, Nakayama H, Kitadai Y, Yoshida K, Chayama K and Yasui W . (2004a). Int. J. Cancer, 110, 212–218.

  60. Oshimo Y, Oue N, Mitani Y, Nakayama H, Kitadai Y, Yoshida K, Ito Y, Chayama K and Yasui W . (2004b). Pathobiology, 71, 137–143.

  61. Peters JH, Clark GW, Ireland AP, Chandrasoma P, Smyrk TC and DeMeester TR . (1994). J. Thorac. Cardiovasc. Surg., 108, 813–821.

  62. Powell J and McConkey CC . (1992). Eur. J. Cancer Prev., 1, 265–269.

  63. Rabinovitch PS, Longton G, Blount PL, Levine DS and Reid BJ . (2001). Am. J. Gastroenterol., 96, 3071–3083.

  64. Reid BJ, Prevo LJ, Galipeau PC, Sanchez CA, Longton G, Levine DS, Blount PL and Rabinovitch PS . (2001). Am. J. Gastroenterol., 96, 2839–2848.

  65. Sampliner RE . (2002). Am. J. Gastroenterol., 97, 1888–1895.

  66. Sarbia M, Geddert H, Klump B, Kiel S, Iskender E and Gabbert HE . (2004). Int. J. Cancer, 111, 224–228.

  67. Sato F, Harpaz N, Shibata D, Xu Y, Yin J, Mori Y, Zou TT, Wang S, Desai K, Leytin A, Selaru FM, Abraham JM and Meltzer SJ . (2002a). Cancer Res., 62, 1148–1151.

  68. Sato F, Shibata D, Harpaz N, Xu Y, Yin J, Mori Y, Wang S, Olaru A, Deacu E, Selaru FM, Kimos MC, Hytiroglou P, Young J, Leggett B, Gazdar AF, Toyooka S, Abraham JM and Meltzer SJ . (2002b). Cancer Res., 62, 6820–6822.

  69. Schnell TG, Sontag SJ and Chejfec G . (1992). Digest. Dis. Sci., 37, 137–143.

  70. Schnell TG, Sontag SJ, Chejfec G, Aranha G, Metz A, O'Connell S, Seidel UJ and Sonnenberg A . (2001). Gastroenterology, 120, 1607–1619.

  71. Shaheen N and Ransohoff DF . (2002). JAMA, 287, 1972–1981.

  72. Sharma P, Reker D, Falk GW, Johnston M, Weston AP and Sampliner RE . (2001). Gastroenterology, 120, A16.

  73. Sharma P, Weston AP, Morales T, Topalovski M, Mayo MS and Sampliner RE . (2000). Gut, 46, 9–13.

  74. Shibata DM, Sato F, Mori Y, Perry K, Yin J, Wang S, Xu Y, Olaru A, Selaru F, Spring K, Young J, Abraham JM and Meltzer SJ . (2002). Cancer Res., 62, 5637–5640.

  75. Skacel M, Petras RE, Gramlich TL, Sigel JE, Richter JE and Goldblum JR . (2000). Am. J. Gastroenterol., 95, 3383–3387.

  76. Soldes OS, Kuick RD, Thompson II IA, Hughes SJ, Orringer MB, Iannettoni MD, Hanash SM and Beer DG . (1999). Br. J. Cancer, 79, 595–603.

  77. Spechler SJ . (2002). N. Engl. J. Med., 346, 836–842.

  78. Streitz Jr JM, Andrews Jr CW and Ellis Jr FH . (1993). J. Thorac. Cardiovasc. Surg., 105, 383–387.

  79. Sullivan Pepe M, Etzioni R, Feng Z, Potter JD, Thompson ML, Thornquist M, Winget M and Yasui Y . (2001). J. Natl. Cancer Inst., 93, 1054–1061.

  80. Suzuki H, Zhou X, Yin J, Lei J, Jiang HY, Suzuki Y, Chan T, Hannon GJ, Mergner WJ and Abraham JM . (1995). Hum. Mol. Genet., 4, 1883–1887.

  81. Teodori L, Gohde W, Persiani M, Ferrario F, Tirindelli Danesi D, Scarpignato C, Di Tondo U, Alo P and Capurso L . (1998). Cytometry, 34, 257–263.

  82. Tokumaru Y, Nomoto S, Jeronimo C, Henrique R, Harden S, Trink B and Sidransky D . (2003). Oncogene, 22, 6954–6958.

  83. Tseng EE, Wu TT, Yeo CJ and Heitmiller RF . (2003). J. Gastrointest. Surg., 7, 164–170; discussion 170–171.

  84. van der Burgh A, Dees J, Hop WC and van Blankenstein M . (1996). Gut, 39, 5–8.

  85. Waki T, Tamura G, Sato M, Terashima M, Nishizuka S and Motoyama T . (2003). Cancer Sci., 94, 360–364.

  86. Weston AP, Badr AS and Hassanein RS . (1999). Am. J. Gastroenterol., 94, 3413–3419.

  87. Weston AP, Banerjee SK, Sharma P, Tran TM, Richards R and Cherian R . (2001). Am. J. Gastroenterol., 96, 1355–1362.

  88. Weston AP, Krmpotich PT, Cherian R, Dixon A and Topalosvki M . (1997). Am. J. Gastroenterol., 92, 407–413.

  89. Williamson WA, Ellis Jr FH, Gibb SP, Shahian DM, Aretz HT, Heatley GJ and Watkins Jr E . (1991). Arch. Intern. Med., 151, 2212–2216.

  90. Wong DJ, Barrett MT, Stoger R, Emond MJ and Reid BJ . (1997). Cancer Res., 57, 2619–2622.

  91. Wong DJ, Paulson TG, Prevo LJ, Galipeau PC, Longton G, Blount PL and Reid BJ . (2001). Cancer Res., 61, 8284–8289.

  92. Xiao WH and Liu WW . (2004). World J. Gastroenterol., 10, 376–380.

Download references

Acknowledgements

This work was supported in part by a grant of the Mildred Scheel Foundation of German Cancer Aid (Deutsche Krebshilfe) (KS) and NIH grants CA85069, CA01808, CA95323, CA098450, CA77057 and The Medical Research Service of the Department of Veterans Affairs (SJM).

Author information

Correspondence to Stephen J Meltzer.

Additional information

Supplementary Information accompanies the paper on Oncogene website (http://www.nature.com/onc)

Supplementary information

Rights and permissions

Reprints and Permissions

About this article

Keywords

  • esophageal cancer
  • Barrett's esophagus
  • hypermethylation
  • early cancer detection
  • neoplastic progression
  • cancer biomarkers

Further reading