The clinical significance of cysteine dioxygenase type 1 methylation in Barrett esophagus adenocarcinoma

SUMMARY

Methylation of cysteine dioxygenase type 1 (CDO1) gene, a tumor suppressor gene, has been studied in various cancers; however, there is no information regarding Barrett esophagus cancer. In this study, the clinical significance of CDO1 methylation in Barrett esophagus adenocarcinoma (BEA) was clarified. CDO1 gene promoter methylation was analyzed for DNA from the patient's specimens using quantitative methylation-specific polymerase chain reaction. Thirty-eight BEA patients who underwent resection were identified between 2000 and 2014. Hypermethylation of CDO1 gene was demonstrated to be frequently recognized even at early stage in BEA by quantitative methylation-specific polymerase chain reaction. In BEA, there is a robust prognostic difference between stage I and stage II/III/IV with regard to 5-year relapse-free survival (P = 0.0016) and 5-year overall survival (P = 0.0024), and the tumor size separated by 7 cm was also a prognostic factor. There was significant difference in CDO1 gene methylation according to the tumor size (P = 0.036). BEA patients with CDO1 gene methylation were shown marginally significantly poorer prognosis (P = 0.054) than otherwise patients. In conclusion, higher CDO1 gene methylation was seen in BEA at earlier stage than in squamous cell carcinoma, and it may account for aggressive phenotype of BEA.

INTRODUCTION

Esophageal cancer is largely composed of squamous cell carcinoma (ESCC) and Barrett esophageal adenocarcinoma (BEA). BEA is a cancer that is increasing at the highest rate in the western world. Since 1995, the frequency of ESCC and BEA was reversed in the United States, and currently, adenocarcinoma has occupied approximately 60% in esophageal cancer.^1–4 On the other hand, BEA remains a minor cancer of the esophagus in Japan, although it has been steadily increasing;^5,6 the proportion of BEA against ESCC was about 1.6% in 2005.^7,8

Norman Barrett for the first time reported Barrett esophagus (BE) in 1950 as gastric epithelium lining the lower esophagus.⁹ It was supposed to be metaplastic change and the carcinogenic rate from BE was reported about 0.3–0.6% per year.^10–12 The carcinogenic high risk factors for BEA are known to be gastroesophageal reflux disease,¹³ the length of BE,^12,14 or the presence of Barrett mucosa.¹⁵ The carcinogenic process has also been reported to be associated with epigenetic abnormalities.¹⁶ Cytosine DNA methylation in the promoter CpG islands of tumor suppressor genes is observed at the early stage of BEA carcinogenesis and is intriguingly seen even in premalignant lesions.^16,17

We recently focused on epigenetic aberration of cysteine dioxygenase type 1 (CDO1) gene in human cancer.¹⁸CDO1 gene methylation was confirmed in colorectal, breast, esophageal, lung, bladder, and gastric cancer.^18–20CDO1 gene is located on chromosome 5q23.2 and is supposed to be a tumor suppressor gene.¹⁸ CDO1 is a nonheme structured and iron-containing metalloenzyme that converts cysteine to cysteine sulfonate.¹⁸ Cysteine biology has gained attention to play a central role in chemosensitivity of cancer cells on downstream of CD44-xCT (cancer stem cell marker-cysteine transporter) axis.^20–22 CDO1 could be implicated in reactive oxygen species (ROS) generation and subsequently could affect cell viability and growth.

In this study, we clarified the clinical significance of CDO1 gene methylation in primary BEA tumors.

PATIENTS AND METHODS

Definition of Barrett esophageal adenocarcinoma

American Gastroenterological Association has defined Barrett's esophagus as follows: Barrett's esophagus is the condition in which any extent of metaplastic columnar epithelium that predisposes to cancer development replaces the stratified squamous epithelium that normally lines the distal esophagus. Presently, intestinal metaplasia is required for the diagnosis of Barrett's esophagus.²³

In this study, we defined it according to the 10th edition of the Japanese Classification Esophageal Cancer (10th JES).²⁴ Barrett's mucosa is the columnar epithelium continuous from the stomach with or without intestinal metaplasia. BE is the esophagus having Barrett mucosa and defined when at least one of the following conditions is satisfied: esophageal glands in the area of columnar epithelium, squamous islands in the columnar epithelium, or double-layered structure of the muscularis mucosa. BE is classified in long-segmental Barret esophagus (LSBE) and short-segmental Barret esophagus (SSBE). LSBE is circular Barrett mucosa extending longitudinally 3 cm or more. SSBE is less than 3-cm Barrett mucosa or noncircular Barrett mucosa. BEA is defined as adenocarcinoma arising in Barrett mucosa.

Patients and tissue samples

Between 1 January 2000 and 30 July 2014, 38 patients were diagnosed with BEA in the Kitasato University Hospital. Twenty-four patients had undergone curative esophagectomy, and 14 patients had undergone endoscopic therapy, including endoscopic submucosal dissection (ESD) or endoscopic mucosal resection. The median follow-up period was 42 months (range 3–182 months). Moreover, in order to compare BEA with ESCC, 179 ESCC patients were included between 1996 and 2007.²⁵

Three patients in BEA have received preoperative chemotherapy: one patient was in cStage II and two patients were in cStage IV (M1 lymph). Ten patients have received postoperative chemotherapy: two patients were in cStage I, three patients were in cStage II, one patient was in cStage III, four patients were in cStage IV. In ESCC, 46 patients had received preoperative chemotherapy: 19 cases were in cStage II, 23 cases were in cStage III, and four cases were in cStage IV.

Formalin-fixed, paraffin-embedded tissue samples were collected. Informed consent was obtained from all patients before sample collection. This study was conducted in accordance with the Declaration of Helsinki and was approved by the Research Ethics Committee of Kitasato University School of Medicine.

Clinicopathological factors

Clinicopathological factors were as follows: age, gender, BE type, microscopic Lauren's histological type, tumor diameter, and method of treatment. Clinical and pathological T factor, N factor, M factor, and staging classification were defined according to the 6th edition of American Joint Committee on Cancer/International Union Against Cancer (6th UICC) staging system. Lymphatic invasion, vascular invasion, and infiltrative growth pattern were defined according to the 10th JES. Cancer cells remained in the submucosal layer defined superficial esophageal cancer. And the lesion deeper than the muscle layer was advanced cancer. Details are shown in Table 1, and endoscopic images of the representative cases are shown in Figure 1.

Fig. 1

Representative endoscopic findings and the Kaplan–Meier curves of primary BEA. (a) Endoscopic finding of cT1a BEA before EMR. (b) Endoscopic finding of cT1a BEA after EMR. (c, d) Endoscopic findings of two cases with cT1b BEA before surgical resection. (e, f) Endoscopic findings of two cases with cT2 or cT3 before surgical resection. (g) Five-year RFS of BEA patients according to the 6th UICC cStage using the Kaplan–Meier survival curves. (h) Five-year OS of BEA patients according to the 6th UICC cStage using the Kaplan–Meier survival curves.

Table 1

Patient characters of Barrett esophageal cancer between 2000 and 2014.

Clinicopathological parameters		Categories	Account
Age		Median(range)	69(35–83)
		Average ± S.D.	65.7 ± 12.2
Gender		Male	34
		Female	4
Barrett		LSBE	14
		SSBE	24
Histological type†		Differentiated type adenocarcinoma	29
		Undifferentiated-type adenocarcinoma	9
Tumor diameter (cm)		Median(range)	2.8(0.5–9.3)
		Average ± S.D.	2.5 ± 2.1
Method of treatment		Endoscopic therapy	14
		Surgery	24
Lymphatic invasion		ly0	14
		ly1	12
		ly2	8
		ly3	4
Vascular invasion		v0	22
		v1	3
		v2	9
		v3	4
Infiltrative growth pattern‡		INFa	5
		INFb	21
		INFc	6
Recurrence			9
The 6th UICC	cT factor	T1	23
		T2	4
		T3	11
		T4	0
	cN factor	N0	31
		N1	7
	cM factor	M0	34
		M1a/M1b	3/1
	cStage factor	I	22
		IIA/IIB	8/0
		III	4
		IVA/IVB	3/1
	(y)pT factor	T1	24
		T2	4
		T3	10
		T4	0
	(y)pN factor	N0	30
		N1	8
	(y)pM factor	M0	32
		M1a/M1b	6/0
	(y)pStage factor	I	24
		IIA/IIB	4/0
		III	4
		IVA/IVB	6/0
Esohageal squamous cell carcinoma		cStage 0	1
The 6th UICC		cStage I	57
		cStage IIA/IIB	58/7
		cStage III	40
		cStage IVA/IVB	10/6

^†Well- and moderately differentiated tubular adenocarcinomas and papillary adenocarcinomas were classified as differentiated-type adenocarcinoma; poorly differentiated adenocarcinomas, signet-ring cell carcinomas, and mucinous carcinomas were classified as undifferentiated-type adenocarcinoma.

^‡INF was impossible to evaluate in six cases because cancer cells were confined to the mucosa.

LSBE, long-segmental Barret esophagus; SSBE, short-segmental Barret esophagus.

DNA purified from tissue and bisulfite treatment of DNA

Tissue sections from tumor were sharply dissected on hematoxylin and eosinstained slides, and genomic DNA was subsequently extracted using a QIAamp DNA FFPE Kit (Qiagen Sciences, Hilden, Germany). Bisulfite treatment was carried out using an EpiTect bisulfite kit (Qiagen).

Quantitative methylation-specific PCR

TaqMan quantitative methylation-specific polymerase chain reaction (PCR) (Q-MSP) was carried out using iQ Supermix (Bio-Rad, Hercules, California, USA) in triplicate on the iCycler iQ real-time PCR detection system (Bio-Rad). Q-MSP was performed at 95°C for 3 minutes, followed by 40 cycles at 95°C for 20 seconds, 60°C for 30 seconds, and 72°C for 30 seconds in a 25 μL reaction volume containing 1 μL bisulfite-treated genomic DNA, 300 nmol/L of each primer, 200 nmol/L fluorescent probe, and 12.5 μL iQ Supermix. One dense CpG island resides 430 bp upstream of the transcription start site (TSS) in the promoter region of CDO1. Primers for TaqMan-MSP were designed within the region that covered most of the CpG-rich region proximal to the TSS in the CDO1 promoter. Fifteen CGs were analyzed.¹⁸ CDO1 forward primer sequence was 5^΄-CCACAACGACGAAAATAAAACG-3^΄, reverse primer sequence was 5^΄-TCGGCGTTTT-AGGGATCGCG-3^΄, and fluorescent probe sequence was 6FAM 5^΄-TTAACGGCGCGTTTT-AGTCGTTCG-3^΄ TAMRA. β-Actin forward primer sequence was 5^΄-TGGTGATGGAGGAGGTTTAG-TAAGT-3^΄, reverse primer sequence was 5^΄-AA-CCAATAAAACCTACTCCTCCCTTAA-3^΄, and fluorescent probe sequence was 6FAM 5^΄-ACCACCACCCAACACACAATAACAAACACA-3^΄ TAMRA.¹⁸ Serial dilutions of bisulfite-modified DNA from DLD1 of colon carcinoma cell line were used as positive control and HepG2 of hepatocyte carcinoma as negative control, respectively. The methylation value (TaqMeth value) was defined by a ratio of CDO1 divided by β-actin and then multiplied by 100, according to the comparative cycle threshold method.²⁶

Statistical analysis

Analysis of the relationship between the CDO1 TaqMeth value and clinicopathological factors was performed by using Student's t-test, Mann–Whitney's U-test, Tukey's honestly significant difference test and variance, if appropriate. Test of homoscedasticity was performed using the F-test. The Kaplan–Meier method was used to estimate cumulative 5-year relapse-free survival (RFS) and overall survival (OS), and statistical differences were tested by using the log-rank test. RFS and OS were measured from the date of surgery to the date of events or the last follow-up.

Variables suggested to be prognostic factors on univariate analysis were subjected to multivariate analysis using a Cox proportional hazard regression model. A value of P < 0.05 was considered statistically significant. All statistical analyses were performed using the SAS software package JMP, version 11 (SAS Institute Inc., Cary, NC, USA).

RESULTS

The prognosis of BEA according to stage

Representative images of BEA by endoscopy are shown in Figure 1. BEA patients with cT1 are shown in Figure 1a–d, while those with cT2 or beyond are shown in Figure 1e,f. Figure 1a shows the BEA patient with cT1a, and Figure 1b shows endoscopic image after ESD for the case of Figure 1a. Figure 1c,d represents BEA patients with cT1b, who were treated by esophagectomy with right thoracotomy and hiatal approaches, respectively. Cases as shown in Figure 1a,c,d are cStage I, while those in Figure 1e,f are diagnosed as cStage II. RFS and OS of the patients with cStage I and cStage II/III/IV were compared using the Kaplan–Meier survival curves (Fig. 1g,h). There were robust differences with regard to RFS (P = 0.0016) and OS (P = 0.0024) according to cStage. The 5-year RFS rate was 95.2% in cStage I, whereas that was 47.4% in cStage II/III/IV. The 5-year OS rate was 94.4% in cStage I, whereas that was 36.4% in cStage II/III/IV.

Univariate prognostic analysis of clinicopathological factors in BEA

We next performed univariate prognostic analysis for RFS and OS by using the log-rank test for all prognostic factor candidates of clinicopathological factors. Significant prognostic factors representing poor RFS were female sex (P = 0.038), undifferentiated type (P = 0.041), tumor diameter > 7.0 cm (P = 0.021), lymphatic invasion (P = 0.016), venous invasion (P = 0.0010), advanced carcinoma (P = 0.0033), N factor (P < 0.0001), and cStage (P < 0.0001). For OS, significant negative prognostic factors were undifferentiated type (P = 0.0073), tumor diameter > 7.0 cm (P < 0.0001), lymphatic invasion (P = 0.030), venous invasion (P = 0.0002), advanced carcinoma (P = 0.0006), N factor (P < 0.0002), and cStage (P < 0.0001).

Multivariate prognostic analysis in BEA

Clinical staging factors such as cT factor and cN factor were not included in the multivariate prognostic analysis because they are thought to be critical confounding factors for stage determination. The multivariate Cox proportional hazard model identified the tumor diameter, which was determined as a cutoff value of 7 cm, as a poor prognostic factor in both RFS (HR; 48.25, 95% CI; 3.26–1394.58, P = 0.0058) and OS (HR; 51.00, 95% CI; 3.44–1469.71, P = 0.0052) independently of cStage.

Relationship between the CDO1 TaqMeth value and clinicopathological factors

We then investigated the relationships between the CDO1 TaqMeth value and various clinicopathological factors. The result summary is shown in Table 2. The median of the TaqMeth value of BEA was 40.28 (3.88–135.37). There was a significant correlation between the clinicopathological factors and CDO1 TaqMeth value with regard to the tumor diameter; BEA patients with length beyond 7 cm showed a significantly higher CDO1 TaqMeth value than those with length below 7 cm (P = 0.036). In addition, it was marginally significant in terms of lymphatic invasion (P = 0.053). The results of comparison of the CDO1 TaqMeth value in clinicopathological factors including the tumor diameter and lymphatic invasion are shown in Figure 2. No significant correlation with other clinicopathological factors was obtained according to the CDO1 TaqMeth value.

Fig. 2

The relationship between the clinicopathological factors and CDO1 TaqMeth value in primary BEA. (a) CDO1 methylation level is significantly correlated with the tumor diameter (P = 0.036). (b) CDO1 methylation level is marginally correlated with degree of lymphatic invasion (P = 0.053). There was no significant correlation of the CDO1 methylation level with (c) age, (d) length of Barrett, (e) infiltrative growth pattern, and (f) cStage. LSBE, long-segmental Barret esophagus; SSBE, short-segmental Barret esophagus.

Table 2

Analysis of the relationship between the CDO1 TaqMeth value and clinicopathological factors.

Clinicopathological parameters	Compare items	n = 38	Average of TaqMeth value	P value
Age	<69	19	42.0 ± 28.8	0.28
	≤69	19	52.6 ± 31.0
Gender	Male	34	44.9 ± 28.6	0.15
	Female	4	67.9 ± 38.4
Barrett	LSBE	14	53.1 ± 35.2	0.37
	SSBE	24	44.0 ± 26.7
Histological type†	Differentiated type	29	45.9 ± 28.7	0.61
	Undifferentiated type	9	51.8 ± 35.2
Tumor diameter (cm)	≤7 cm	34	55.8 ± 27.3	0.036
	<7 cm	4	76.8 ± 40.2
Method of treatment	Endoscopic therapy	14	54.4 ± 29.0	0.27
	Surgery	24	43.2 ± 30.3
Lymphatic invasion	ly0	14	48.2 ± 26.3	0.053
	ly1	12	37.8 ± 27.7
	ly2	8	41.9 ± 22.7
	ly3	4	83.7 ± 43.1
Vascular invasion	v0	22	45.0 ± 28.8	0.82
	v1	3	59.1 ± 32.7
	v2	9	45.4 ± 21.8
	v3	4	55.7 ± 55.0
Infiltrative growth pattern	INFa	5	34.2 ± 12.1	0.26
	INFb	21	44.5 ± 28.8
	INFc	6	62.6 ± 38.1
cT (6th UICC)	cT1	23	46.1 ± 28.5	0.76
	cT2-cT3	15	49.2 ± 33.1
(y)pT (6th UICC)	pT1	24	47.7 ± 26.8	0.91
	pT2-pT3	14	46.6 ± 35.8
cN (6th UICC)	Negative‡	31	45.3 ± 29.1	0.38
	Positive§	7	56.4 ± 34.5
(y)pN (6th UICC)	Negative‡	30	43.6 ± 26.9	0.14
	Positive§	8	61.4 ± 38.2
cStage factor (6th UICC)	Stage I	22	43.6 ± 26.4	0.20
	Stage II	8	38.7 ± 17.9
	Stage III	4	58.9 ± 53.6
	Stage IV	4	73.6 ± 34.4
(y)pStage factor (6th UICC)	Stage I	24	45.3 ± 25.6	0.22
	Stage II	4	26.2 ± 28.1
	Stage III	4	51.4 ± 15.2
	Stage IV	6	55.8 ± 33.2

^†Well- and moderately differentiated tubular adenocarcinomas and papillary adenocarcinomas were classified as differentiated-type adenocarcinoma; poorly differentiated adenocarcinomas, signet-ring cell carcinomas, and mucinous carcinomas were classified as undifferentiated-type adenocarcinoma; ^‡Even one lymph node metastasis is not allowed; ^§One or more of the lymph node metastases are found.

LSBE, long-segmental Barret esophagus; SSBE, short-segmental Barret esophagus.

Prognosis of BEA patients with a high CDO1 TaqMeth value

We calculated the cutoff value of the CDO1 TaqMeth value using the log-rank plot method to determine the most optimal cutoff value of the CDO1 TaqMeth value with regard to prognosis. Using the cutoff value of 88.4, the hypermethylation group (n = 4) tended to poorer prognosis for RFS compared with the hypomethylation group (n = 34, P = 0.054), and no other cutoff points could stratify the differential prognosis of BEA.

DISCUSSION

BEA is reported to be rapidly increasing in Europe and the United States.⁴ However, BEA still remains a rare disease in Japan, while its mortality admits an increasing trend.^7,8 In Asia including Japan, ESCC accounts for the majority of esophageal cancer, and standard treatment for BEA has therefore not yet established due to minor disease, and its prognosis has not been clarified. In our institute, 409 patients with esophageal cancer were treated for esophagectomy during the latest decades, and BEA was identified only in 23 cases (5.6%). Its frequency was slightly higher than that reported as nationwide registry in Japan in 2012 (3.4%).⁸

Our current study is one of the biggest prognostic data of BEA in Japan from a single institution. In this study, excellent prognosis was confirmed in cStage I BEA, but dismal prognosis was recapitulated in cStage II–IV BEA. Prognosis of BEA is known to be poor in the western country, and the 5-year survival rate has been reported to be dismally 15–20%.²⁷ Prognosis of advanced BEA in the western country may be thus further poorer than our advanced cases. Periodic surveillance for BE would be required in order to achieve early detection of BEA at curable stage, and the optimal selection of better treatment plays key roles in improving the prognosis. Such early detection system and establishment of suitable treatment strategies are being highly anticipated in BEA clinics, and biomarker with regard to BEA should be urgently identified.

It is well known that promoter DNA cytosine methylation for tumor suppressor genes can be useful as cancer biomarkers.^20,28 Recently, the methylation of CpG islands in the promoter region of CDO1 gene has been demonstrated in various cancers, and its outstanding high frequency in the tumor tissues should be considered as a unique character of this gene.^18–20 In this study, we for the first time reported extraordinary high level of promoter DNA methylation in CDO1 gene in primary BEA. Using the TaqMeth value as an indicator of CDO1 gene promoter methylation, promoter methylation was actually demonstrated in all patients.

In addition, CDO1 gene promoter methylation suppresses its expression at the mRNA level, and the loss of function of CDO1 in the cancer cells would contribute to avoid oxidative damage and cell death by enhancing the detoxification of ROS.²⁸ Such tumor-suppressive ability may be actively involved in cancer progression in primary BEA. Clinicopathological analysis showed a close correlation between the promoter methylation of CDO1 with advanced tumor features such as large tumor diameter, aggressive lymphatic invasion, or even phenotypes representing poor prognosis in this study.

In our study, prognostic significance of CDO1 is marginal putatively due to a small number of the disease. Hypermethylation of CDO1 has been reproducibly identified as a negative prognostic factor in various human cancers (ESCC, breast cancer, and renal cell carcinoma).^25,29,30 As in other human cancers, CDO1 may have a potential to represent poor prognosis of BEA. It would be required analysis of accumulated cases to confirm the results in the future or from the western countries, where BEA was no longer minor diseases.

Cancer-specific DNA methylation can also be used as cancer diagnosis by liquid biopsy as a screening method for early cancer detection, and such surveillance is highly expected in every oncological field.^20,31 In regard to esophageal cancer, although histopathological features of BEA and ESCC have been clarified, the investigation of the DNA methylation difference between BEA and ESCC has never been assessed. We also compared CDO1 methylation in BEA with that in ESCC (Supporting Information Fig. S1). Analyzing the degree of promoter DNA methylation of CDO1 gene in 179 cases of esophageal ESCC in our hospital,²⁵ the median of the CDO1 TaqMeth value of ESCC was 9.38 (0.01–279.51). Comparing it with that of BEA in the total cases, the TaqMeth value was significantly higher in BEA (P < 0.0001) and higher in cStage I in BEA (P = 0.030). These findings may explain why BEA harbors aggressive phenotype as compared to ESCC. With the proviso that many patients actually received preoperative chemotherapy in the ESCC patients, it is necessary that the possibility of modification of the CDO1 methylation level by chemotherapy should be considered.

CDO1 is involved in ROS generation, which leads to apoptosis of epithelium-derived cells in the inflammatory tissues,²⁸ suggesting that tumor-suppressive function of CDO1 is defective in primary BEA and the methylation of CDO1 gene plays a role in tumor progression in BEA. We have to pay attention that CDO1 methylation could be detected in early phase of BEA, such as cStage IA BEA, because esophagectomy or endoscopic resection can rescue the patient in this stage in our study. CDO1 methylation utility as the early detection marker would be beneficial in order to improve BEA.

Recently, there have been several reports of CDO1 methylation and chemosensitivity.^25,30,32 Hypermethylation of CDO1 has been reported to be associated with less chemotherapeutic effect representing poor prognosis in breast cancer with anthracycline treatment or less histological grades in ESCC with neoadjuvant chemotherapy.²⁵ However, the latter data only included postoperative histological data of the surgical specimens, and preoperative histological information was not included. Moreover, in BEA, preoperative treatment has seldom been experienced in Japan, and there were no data for relation of CDO1 hypermethyaltion to chemotherapeutic effect by using preoperative biopsy samples.

In conclusion, our study revealed that high methylation of CDO1 gene was seen in BEA at early stage than in ESCC, and it may account for aggressive phenotype of BEA. Such early abnormalities in the tumor tissues could have a promising feature to be developed as early detection tools such as liquid biopsy or prognostic biomarkers.

Acknowledgments

None.

SUPPORTING INFORMATION

Supplementary data are available at DOTESO online.

Fig. S1 Comparison of CDO1 methylation in primary BEA and ESCC. (a) Whole cStage and (b) cStage I had significant difference, while (c) cStage II, III and (d) cStage IV showed no significant difference.