This article is mentioned in:

Introduction

Esophageal cancer is the sixth leading cause of cancer-related mortality worldwide, and esophageal squamous cell carcinoma (ESCC) is the main histologic type of esophageal cancer (1). The incidence of ESCC varies widely by nearly 16-fold throughout the world, with the highest rates in Eastern Asia and Southern and Eastern Africa (2). Currently, surgical resection is the main curative therapeutic option. Although recent advances in diagnosis and treatment have improved patient prognosis, the 5-year survival rate is still low at ~30% (3).

Homeobox (HOX) genes encode a group of transcription factors, which can directly drive the transcription of target genes (4). However, further investigation of the detailed molecular mechanisms is needed. HOX genes play crucial roles in embryogenesis and tumorigenesis. There are three aberrations noted in HOX expression in solid tumors: re-expression, new expression, downregulation or deficiency (4). Takahashi et al comprehensively evaluated the expression of all HOX genes in 48 primary ESCC tissues and 7 normal esophageal tissues by RT-qPCR. Their data suggested that disordered expression of HOX genes was significantly associated with the tumorigenesis and development of ESCC (5). In our previous study, homeobox A13 gene (HOXA13) was found to be overexpressed in ESCC tissues when compared to that in normal tissues (3). Colony formation and nude mouse tumorigenicity assays revealed that HOXA13 promoted tumorigenesis in vitro and in vivo. Moreover, the prognosis of HOXA13-positive patients was significantly worse than that of HOXA13-negative patients (6).

Considering that HOXA13 acts as a transcription factor, to identify its potential targets, protein expression changes after HOXA13 knockdown were detected by 2-dimensional electrophoresis (7). Among the proteins downregulated after HOXA13 knockdown, Annexin A2 (ANXA2) and superoxide dismutase 2 (SOD2) were selected for further study. Consistent expression of HOXA13, ANXA2 and SOD2 was validated by western blotting. CHIP-DSL also revealed that SOD2 and ANXA2 were both potential targets of HOXA13. However, little is known concerning the clinical significance of HOXA13/ANXA2 or HOXA13/SOD2 coexpression in ESCC.

ANXA2, a calcium-dependent phospholipid binding protein, is implicated in apoptosis, calcium signaling, tumor invasion, metastasis and angiogenesis. Overexpression of ANXA2, as well as its prognostic value, has been described in colorectal (8), gastric (9) and pancreatic cancer (10,11), hepatocellular carcinoma (12), and lung (13) and breast cancer (14). However, little is known concerning its expression in ESCC.

Superoxide dismutases (SODs) are a family of antioxidant enzymes that neutralize the reactive free superoxide radical (O2−) in mitochondrial reactive oxygen species (ROS) (15,16). SOD2 (also named MnSOD) is highly expressed in cervical carcinoma (17), and in breast (18), gastric and colorectal cancer (19). Studies conducted on SOD2 in cancer focused mainly on its tumor-suppressor role, with a smaller but mounting number of studies suggesting that SOD2 acts as an oncogene (15). It was reported that SOD2 overexpression inhibited POX-induced apoptosis by avoiding oxidative damage to mitochondria (20–22). However, little is known concerning its role in ESCC carcinogenesis.

In the present study, HOXA13, ANXA2 and SOD2 expression was evaluated in normal esophageal mucosa and ESCC tissues, and the correlation between HOXA13 and ANXA2 and SOD2 expression was examined at both the transcriptional and translational levels. In addition, the association of HOXA13, ANXA2 and SOD2 coexpression and prognosis was investigated in ESCC patients.

Materials and methods

Patients and demographic data

The present study cohort consisted of 121 patients from a prospective database of esophageal cancer patients, and all of them underwent surgery at the Department of Thoracic Surgery I, Peking University School of Oncology from July 2000 to November 2009. Our validation cohort consisted of 137 ESCC patients in addition to the above-mentioned database, which were treated between February 1996 and June 2003 at the Department of Thoracic Surgery.

All of the patients underwent radical esophagectomy, and none of them received adjuvant chemotherapy or radiotherapy prior to surgery. Resected samples were immediately sent for histological examination with hematoxylin and eosin staining. Tumor-node-metastasis (TNM) stage was evaluated according to the criteria of the UICC, 1987. To clarify the survival conditions, life-long follow-up was available by regular review (examination records) after surgery or by direct telephone interview until recurrence, metastasis or death caused by tumor. The demographic information (gender and age) and tumor characteristics (histology, differentiation and TNM stage) were acquired from medical and pathological records. A total of 18 patients with fresh frozen cancerous and non-cancerous samples with complete clinical data were collected from Anyang Cancer Hospital, Henan. Tissues were collected immediately after surgical removal and snap-frozen in liquid nitrogen until further use. All participants provided informed consent, and the study was approved by the Ethics and Academic Committees of Peking University School of Oncology.

Immunohistochemistry (IHC)

Formalin-fixed and paraffin-embedded 4-μm tissue sections were routinely immunostained. After deparaffinization in xylene and rehydration in a graded ethanol series, 3% hydrogen peroxide solution was put on the slide for 10 min, and antigen retrieval was carried out in citrate solution (pH 6.0) by microwave. The sections were blocked with goat serum for 15 min and then incubated with mouse monoclonal antibody to ANXA2 and SOD2 (Abcam) at 4°C overnight. The mouse monoclonal antibody against human ANXA2 and SOD2 was used at 2 μg/ml (1:500) and 0.5 mg/ml (1:400), and the secondary antibody was goat anti-mouse biotin-conjugated IgG. Diaminobenzidine (DAB) chromogenic reaction was used for detection. Two experienced pathologists independently examined the immunohistochemical signals. The scores were evaluated according to the number of stained cells and staining intensity. The percentage of ANXA2 or SOD2-positive tumor cells was evaluated on a scale of 0–3 (0, no staining; 1+, ≤10%; 2+, 11–30%; 3+, 31–50%; 4+, >50%). Thus, the expression level of ANXA2 and SOD2 were divided into two groups in terms of the score: negative (0, 1+, 2+) and positive (3+, 4+). For the evaluation and scoring of HOXA13, the same criteria were used as described in our previous study (6).

Cell culture and RNA isolation

Human esophageal cancer EC109, EC9706, KYSE150, KYSE410 and KYSE510 cells were obtained from ATCC (Manassas, VA, USA). These cell lines were cultured in 1640 medium (HyClone, Logan, UT, USA) supplemented with heat-inactivated fetal bovine serum (Gibco, Carlsbad, CA, USA) and 100 U/ml penicillin sodium in a humidified atmosphere with 5% CO2 at 37°C.

Total RNAs of fresh frozen tumor specimens and sorted cells (EC109, EC9706, KYSE150, KYSE410 and KYSE510) were extracted by TRIzol (Invitrogen, Carlsbad, CA, USA). RNAs were reverse transcripted to single strand cDNAs by two-step RT-PCR (Fermentas Life Sciences).

Real-time RT-qPCR

Quantitative real-time PCR was performed using SYBR-Green Real-Time PCR Master Mix (Applied Biosystems) to detect mRNA expression levels of the target genes. Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) was used as the endogenic control. Special primers were designed using Oligo Primer Analysis Software (version 5.0). The sequence of the primers used are as follows: for HOA13, forward 5′-AGCGCGTGCCTTATACCAAG-3′ and reverse 5′-GCCGCTCAGAGAGATTCGT-3′; for ANXA2, forward 5′-CTCTACACCCCCAAGTGCAT-3′ and reverse 5′-TCAGTGCTGATGCAAGTTCC-3′; for SOD2, forward 5′-AAGGGAGATGTTACAGCCCAGATA-3′ and reverse 5′-TCCAGAAAATGCTATGATTGATATGAC-3′; for GAPDH, forward 5′-GACCCCTTCATTGACCTCAAC-3′ and reverse 5′-CTTCTCCATGGTGGTGAAGA-3′. All assays were carried out in triplicate under the 7500 Real-Time PCR System (Applied Biosystems) and repeated three times according to the manufacturer’s protocol. Evaluation of relative expression was calculated by comparative Ct (threshold cycle) method. 2−ΔCt referred to the fold of the mRNA expression of the target gene compared to GAPDH expression in the same sample.

Statistical analysis

Analysis was performed using SPSS 17.0 software. The χ2 test or Fisher’s exact test was used to compare the relationship between HOXA13, ANXA2, SOD2 expression and clinicopathological characteristics of the ESCC patients. HOXA13, ANXA2 and SOD2 mRNA expression levels in cancerous and non-cancerous tissues are presented as the means ± SD and were compared by paired t-test. Wilcoxon test was employed to compare the protein levels of HOXA13, ANXA2 and SOD2 between cancerous and non-cancerous tissues. Pearson’s correlation coefficient analysis was applied to analyze the correlation of HOXA13, ANXA2 and SOD2 mRNA expression. Spearman’s correlation coefficient analysis was applied to analyze the correlation of the protein levels.

The overall survival measured from the day of surgery was estimated by Kaplan-Meier curves, and the differences were analyzed by log-rank test. Cox proportional hazards model was used for multivariate survival analysis. The variables analyzed in the model included age, gender, tumor location, histology, tumor cell differentiation, TNM stage, HOXA13, ANXA2 and SOD2 expression. Hazard ratios and 95% confidence intervals were calculated. p<0.05 was considered to indicate a statistically significant result.

Results

ANXA2 and SOD2 are overexpressed in ESCC tissues

It has been demonstrated that ANXA2 and SOD2 are associated with gastric and colorectal cancer. To investigate whether they are also associated with ESCC, the mRNA levels of ANXA2 and SOD2 in ESCC and matched non-cancerous specimens were analyzed by RT-qPCR. When compared with normal esophageal tissues, ANXA2 and SOD2 showed higher expression in the ESCC tissues (Fig. 1A and B; Mann-Whitney test, ANXA2 p=0.012, SOD2 p=0.016). Furthermore, IHC was employed to assess the protein expression of ANXA2 and SOD2 in 18 ESCC and matched non-cancerous specimens. Positive expression of ANXA2 was observed in 20% of the ESCC tissues and in 0% of the non-cancerous tissues (Fig. 1C; Chi-square test, p=0.023). SOD2 overexpression was detected in 90% of the ESCC tissues and in 25% of the paired cancer margin tissues (Fig. 1D; Chi-square test, p=0.001). In conclusion, expression levels of ANXA2 and SOD2 were significantly higher in the ESCC tissues than levels in the non-cancerous specimens.

Figure 1 ESCC tissues exhibit higher ANXA2 and SOD2 expression when compared with non-cancerous tissues at both the mRNA and protein levels. Comparison of mRNA expression of (A) ANXA2 and (B) SOD2 in cancerous tissues (T) and non-cancerous tissues (N). Representative immunohistochemical staining of (C) ANXA2 and (D) SOD2 in ESCC and non-cancerous (NC) tissues. ESCC, esophageal squamous cell carcinoma; ANXA2, Annexin A2, SOD2, superoxide dismutase 2.

Expression levels of ANXA2 and SOD2 are positively correlated with HOXA13

To investigate the correlation between HOXA13 and its potential target genes, the levels of HOXA13, ANXA2 and SOD2 in 5 ESCC cell lines (EC109, EC9706, KYSE150, KYSE410 and KYSE510) were evaluated by RT-qPCR assay. Both ANXA2 and SOD2 showed a significant positive correlation with HOXA13 in the 5 ESCC cell lines (data not shown; Pearson’s correlation, ANXA2 p=0.005, SOD2 p<0.001), particularly SOD2, with a high R2 score of 0.99.

In the 23 pairs of ESCC tissues, a strong positive correlation was observed between the mRNA expression of ANXA2 and HOXA13 (Pearson’s correlation r=0.878, p<0.001). The same correlation was also found between SOD2 and HOXA13 (Pearson’s correlation r=0.503, p=0.014). To further verify these potential positive correlations, IHC was applied in the study and validation cohorts. In the study cohort, a significant positive correlation was noted between ANXA2 and HOXA13 (Spearman correlation rs=0.200, p=0.028) and the correlation between SOD2 and HOXA13 also approached significance (Spearman correlation rs=0.151, p=0.098). Similar results were observed in the validation cohort. A significant positive correlation was noted between SOD2 and HOXA13 (Spearman correlation rs=0.148, p=0.084) and the correlation between ANXA2 and HOXA13 also approached significance (Spearman correlation rs=0.198, p=0.021).

Collectively, expression of ANXA2 and SOD2 was positively correlated with HOXA13 in the ESCC cell lines and tissues.

Overexpression of ANXA2 or SOD2 indicates poor prognosis of ESCC patients, respectively

Our previous study revealed that high expression of HOXA13 indicates poor survival; thus, the potential target genes ANXA2 and SOD2 may also have prognostic value in ESCC. To study the correlation of ANXA2 and SOD2 with HOXA13 and their roles in ESCC, we analyzed the expression of HOXA13, ANXA2 and SOD2 in both the study and validation cohorts. In the study cohort, HOXA13, ANXA2 and SOD2 expression was significantly correlated with TNM stage (Table I). In the univariate analysis, tumor invasion (T), lymph node metastasis (N), TNM stage, and expression of HOXA13, ANXA2 and SOD2 were statistically associated with poor prognosis, respectively (Table II). The validation cohort also supported a similar conclusion (data not shown).

Table I Association between HOXA13, ANXA2 and SOD2 expression and clinical characteristics of the patients with ESCC in the study cohort (n=121).

Table I

Association between HOXA13, ANXA2 and SOD2 expression and clinical characteristics of the patients with ESCC in the study cohort (n=121).

	HOXA13			ANXA2			SOD2
Clinicopathological factors	High (n=21)	Low (n=100)	P-value	High (n=22)	Low (n=99)	P-value	High (n=54)	Low (n=67)	P-value
Age (years)			0.219			0.228			0.545
≤50	4	8		4	8		4	8
>50	17	92		18	91		50	59
Gender			0.254			0.780			0.827
Male	18	76		18	76		41	53
Female	3	24		4	23		13	14
Tumor location			0.950			0.090			0.117
Upper	4	48		1	21		14	8
Middle	11	49		15	45		26	34
Lower	6	33		6	33		14	25
Tumor cell differentiation			0.793			0.533			0.363
Well	7	35		10	32		15	27
Moderate	6	35		6	35		20	21
Poor	8	30		6	32		19	19
Tumor invasion (T)			0.075			0.004			0.698
T1	3	25		1	27		18	10
T2	4	30		5	29		19	15
T3	9	39		10	38		24	24
T4	5	6		6	5		6	5
Lymph node metastasis (N)			0.606			0.316			0.120
N0	13	70		13	70		33	50
N1, N2, N3	8	30		9	29		21	17
TNM stage			0.028			0.003			0.039
I, Tis	2	32		1	33		9	25
IIa, IIb	8	42		9	41		25	25
IIIa, IIIb, IIIc	11	26		12	25		20	17

[i] HOXA13, homeobox A13 gene; ANXA2, Annexin A2; SOD2, superoxide dismutase 2; ESCC, esophageal squamous cell carcinoma; TNM, tumor-node-metastasis.

Table II Clinicopathological features, tumor markers and patient survival in the study cohort (n=121, univariate analysis).

Table II

Clinicopathological features, tumor markers and patient survival in the study cohort (n=121, univariate analysis).

Variables	Hazard ratio (95% CI)	P-value
Age, years	1.009 (0.982–1.037)	0.514
Gender	1.229 (0.682–2.217)	0.492
Tumor location (upper/middle vs. lower)		0.561
Middle vs. upper	0.895 (0.307–2.610)	0.337
Lower vs. upper	0.825 (0.474–1.437)	0.497
Tumor cell differentiation (poor/moderate vs. well)		0.534
Moderate vs. poor	1.809 (0.675–4.848)	0.239
Well vs. poor	1.207 (0.519–2.808)	0.662
Tumor invasion (T)		0.021
T1 vs. T4	0.338 (0.142–0.806)	0.014
T2 vs. T4	0.417 (0.184–0.946)	0.036
T3 vs. T4	0.745 (0.354–1.566)	0.437
Lymph node metastasis (N)	0.447 (0.274–0.728)	0.001
TNM stage (I/IIa/IIb vs. III)		0.003
I vs. III	0.194 (0.025–1.537)	0.121
IIa vs. III	0.265 (0.107–0.659)	0.004
IIb vs. III	0.494 (0.210–1.159)	0.105
HOXA13 expression (yes vs. no)	2.020 (1.294–3.145)	0.002
ANXA2 expression (yes vs. no)	2.074 (1.344–3.202)	0.001
SOD2 expression (yes vs. no)	1.764 (1.181–2.634)	0.006

[i] CI, confidence interval; TNM, tumor-node-metastasis; HOXA13, homeobox A13 gene; ANXA2, Annexin A2; SOD2, superoxide dismutase 2; ESCC, esophageal squamous cell carcinoma.

In the study cohort, Kaplan-Meier curve analysis indicated that the median survival time was 22 months for the ANXA2-positive patients, which was significantly shorter than the 64 months for ANXA2-negative patients (Fig. 2A, log-rank p=0.026). As for SOD2, the median survival time was 31 and 84 months for SOD2-positive and SOD2-negative patients, respectively (Fig. 2B, log-rank p=0.039). In conclusion, overexpression of ANXA2 or SOD2 indicated poor prognosis of ESCC patients, respectively. This was similar in the validation cohort. Kaplan-Meier curve analysis indicated that the median survival time was 15 months for ANXA2-positive patients, which was significantly lower than the 26 months for ANXA2-negative patients (Fig. 2C, log-rank p=0.003). As for SOD2, the median survival time was 19 and 28 months for SOD2-positive and SOD2-negative patients, respectively (Fig. 2D, log-rank p=0.003).

Figure 2 Kaplan-Meier survival curves for ESCC patients bearing tumors with or without ANXA2 and SOD2 expression. The overall survival of patients bearing tumors with ANXA2 expression was significantly shorter than that of patients bearing tumors without ANXA2 expression in both the (A) study and (C) validation cohort. The overall survival of patients with SOD2 expression was significantly shorter than that of patients bearing tumors without SOD2 expression in both the (B) study and (D) validation cohort. ESCC, esophageal squamous cell carcinoma; ANXA2, Annexin A2, SOD2, superoxide dismutase 2. 0, negative expression; 1, positive expression.

Combined expression of HOXA13, ANXA2 and SOD2 has increased prognostic value in ESCC

The coexpression of HOXA13 and its potential targets ANXA2 and SOD2 were analyzed. On the basis of HOXA13 and ANXA2 expression, in the study cohort, all of the patients were categorized into three groups: double-positive (HOXA13+/ANXA2+), single-positive (HOXA13+/ANXA2− and HOXA13−/ANXA2+) and double-negative (HOXA13−/ANXA2−). The median survival time of double-positive patients was 10 months, significantly lower than the 64 months for the double-negative patients and 42 months for the single-positive patients (Fig. 3A, log-rank p=0.002). A similar conclusion was found for the validation cohort. The median survival time of the double-positive patients was 13 months, significantly less than the 27 months for the double-negative patients and 22 months for the single-positive patients (Fig. 3B, log-rank p=0.001). For coexpression of HOXA13/SOD2, in the study cohort, the median survival time was 19, 38 and 84 months for the double-positive, single-positive and double-negative patients, respectively (Fig. 3C, log-rank p=0.010). In the validation cohort, the median survival time was 14, 22 and 29 months for the double-positive, single-positive and double-negative patients, respectively (Fig. 3D, log-rank p=0.004). For ANXA2 and SOD2, coexpression of ANXA2 and SOD2 was also predictive of a poor prognosis in the study cohort (Fig. 3E, log-rank p=0.018) and the validation cohort (Fig. 3F, p=0.001).

Figure 3 Kaplan-Meier survival curves for ESCC patients subgrouped according to the combined expression of HOXA13, ANXA2 and SOD2. The survival of patients with HOXA13+/ANXA2+ expression was significantly shorter than that of the patients with HOXA13−/ANXA2− expression in both the (A) study and (B) validation cohort. The survival of patients with HOXA13+/SOD2+ expression was significantly shorter than that of patients with HOXA13−/SOD2− expression in both the (C) study and (D) validation cohort. The survival of patients with ANXA2+/SOD2+ expression was significantly shorter than that of patients with ANXA2−/SOD2− expression in both the (E) study and (F) validation cohort. The survival of patients with HOXA13+, ANXA+ and SOD2+ (triple-positive) expression was significantly shorter than that of the triple-negative patients in both the study cohort (G) and (H) validation cohort. ESCC, esophageal squamous cell carcinoma; ANXA2, Annexin A2, SOD2, superoxide dismutase 2. HA, HOXA13+ANXA2 coexpression; HM, HOXA13+SOD2 coexpression; AM, ANXA2+SOD2 coexpression; HAM, HOXA13+ANXA2+SOD2 coexpression. 0, no markers positive; 1, some markers positive; 2, all markers positive.

Moreover, when combining the expression of HOXA13, ANXA2 and SOD2, a better prognostic model was obtained in the two cohorts. We found that coexpression of HOXA13, ANXA2 and SOD2 was significantly associated with overall survival in the study cohort (Fig. 3G, log-rank p<0.001) as well as in the validation cohort (Fig. 3H, log-rank p=0.001). In the study cohort, TNM stage (p=0.006) and HOXA13/ANXA2/SOD2 (p=0.002) coexpression are both independent poor predictors of overall survival time in the multivariate analysis (Table III). In the validation cohort, consistent with the above, multivariate analysis showed that TNM stage (p=0.002) and HOXA13/ANXA2/SOD2 (p=0.017) coexpression are both independent predictors of poor overall survival (Table IV).

Table III Independent predictors of the overall survival time in the study cohort (multivariate analysis, n=121).

Table III

Independent predictors of the overall survival time in the study cohort (multivariate analysis, n=121).

Variables	Hazard ratio (95% CI)	P-value
TNM stage (I/IIa/IIb vs. III)		0.006
I vs. III	0.404 (0.208–0.784)	0.007
IIa vs. III	0.358 (0.152–0.840)	0.018
IIb vs. III	0.382 (0.206–0.711)	0.002
HOXA13/ANXA2/SOD2		0.002
None positive vs. all positive	0.171 (0.068–0.433)	<0.001
Partly positive vs. all positive	0.294 (0.124–0.699)	0.294

[i] CI, confidence interval; TNM, tumor-node-metastasis; HOXA13, homeobox A13 gene; ANXA2, Annexin A2; SOD2, superoxide dismutase 2; ESCC, esophageal squamous cell carcinoma.

Table IV Independent predictors of the overall survival time in the validation cohort (multivariate analysis, n=137).

Table IV

Independent predictors of the overall survival time in the validation cohort (multivariate analysis, n=137).

Variables	Hazard ratio (95% CI)	P-value
TNM stage (I/IIa/IIb vs. III)		0.002
I vs. III	0.197 (0.048–0.813)	0.025
IIa vs. III	0.592 (0.390–0.899)	0.014
IIb vs. III	0.363 (0.164–0.806)	0.013
HOXA13/ANXA2/SOD2		0.017
None positive vs. all positive	0.332 (0.129–0.858)	0.023
Partly positive vs. all positive	0.535 (0.208–1.374)	0.194

[i] CI, confidence interval; TNM, tumor-node-metastasis; HOXA13, homeobox A13 gene; ANXA2, Annexin A2; SOD2, superoxide dismutase 2; ESCC, esophageal squamous cell carcinoma.

Discussion

Homeobox (HOX) genes function as primary regulators in embryogenesis and tumorgenesis. Transcriptional factors encoded by HOX, which have been detected as deregulated in various types of tumors, regulate cell proliferation and differentiation (4). Previously, we performed the first comprehensive investigation on the 39 HOX genes in ESCC; 8 of the 39 HOX genes were detected in cancerous tissues rather than non-cancerous tissues. The upregulation of HOXA13 was observed in ESCC cell lines and cancerous tissues. Colony formation and nude mouse tumorigenicity assays revealed that HOXA13 promotes tumor cell proliferation in vitro and in vivo, and HOXA13 expression is significantly associated with disease-free survival. Subsequently, a proteomics study and CHIP-DSL revealed that ANXA2 and SOD2 are potential targets of HOXA13.

In the present study, we revealed that both ANXA2 and SOD2 were overexpressed in ESCC tissues when compared to the levels in the normal esophageal tissues. Further analysis of ANXA2 and SOD2 expression combined with HOXA13 expression in the same series of ESCC tissues indicated a significantly positive correlation between them at both the protein and mRNA levels. Collectively, ANXA2 and SOD2 may participate in ESCC tumorigenesis as well as HOXA13.

However, to date, the molecular pathway linking HOXA13 and its potential targets is not yet clear. As a transcriptional factor, the core binding motif of HOXA13 has been identified: a core sequence of TAA, and TAA-containing sequences were TAAA (50%), TAAC (30%) and TAAT (20%) (23), which were also found in the promotor region of both ANXA2 and SOD2 (data not shown).

ANXA2 is a member of the calcium and phospholipid-dependent proteins. Binding of t-PA and ANXA2 on the membrane of pancreatic cancer cells was found to activate tumor cell invasion (10). ANXA2 was found to facilitate cell cycle and proliferation in non-small cell lung cancer by inhibiting p53, while the silencing of ANXA2 increased p53 expression, which led to p53-dependent and -independent G2 arrest (13). The present study suggests that overexpression of ANXA2 is indicative of the poor prognosis of ESCC patients, which corroborates the role of oncogenic ANXA2 revealed by such mechanistic studies.

SOD2 is a member of the manganese superoxide dismutase family, which encodes a mitochondrial protein. Studies suggest that SOD2 overexpression is associated with tumor invasion and metastasis. NF-κB was found to reduce tumor progression through binding to intronic enhancer element to activate the expression of SOD2 (18). Our results also indicated that overexpression of SOD2 is predictive of poor prognosis of ESCC patients.

Considering the oncogenic role of ANXA2 and SOD2, and our previous result of their coexpression in ESCC, we speculate that HOXA13 may act as an oncogene in ESCC by regulating ANXA2 and SOD2 expression, which still needs further investigation. Revealing the specific mechanism of the above association may further our understanding of ESCC carcinogenesis, with the potential to develop new drug targets of ESCC and possibly, to establish a more personalized prognosis for each patient.

Since ANXA2 and SOD2 were found to be involved in ESCC and were associated with HOXA13, elucidation of their clinical significance was of great concern. We revealed that not only ANXA2 or SOD2 expression alone but also their coexpression with HOXA13 was significantly correlated with the overall survival of ESCC patients. Kaplan-Meier survival curve analysis showed that coexpression of HOXA13/ANXA2/SOD2 was indicative of a poor prognosis of ESCC patients, while Cox proportional hazards regression model indicated that coexpression of HOXA13/ANXA2/SOD2, as well as TNM stage, are both independent prognosis factors of ESCC. To strengthen our conclusion, all of the results were validated in two independent cohorts. Collectively, both ANXA2 and SOD2 had a significant prognostic value for ESCC patient, and their coexpression with HOXA13 may have added prognostic value, as a complement to the TNM staging system.

In conclusion, HOXA13 as well as its target genes ANXA2 and SOD2 are potential negative predictors of overall survival time of ESCC patients. Thus, combination of their expression profile and the TNM stage classification may provide a more accurate prediction of the postoperative outcome of ESCC patients.

Acknowledgements

This work was supported by the National Natural Science Foundation for Distinguished Young Scholars (grant no. 81301748), Science Fund for Creative Research Groups of the National Natural Science Foundation of China (grant no. IRT13003). We are grateful to the patients who participated in this study. For providing patient care and the data base establishment, we would like to thank Hong-Chao Xiong, MD, Zhen Liang, MD, Qin Bin, MD, Shao-Hua Ma, MD, Xiao-Zheng Kang, Msc, Yong-Bo Yang, MD, Liang Dai, MD, Wan-Pu Yan, Msc, He-Li Yang, MD (Thoracic Surgery I). We would like to thank Bin Dong, MD, Zhong-Wu Li, MD (Pathology Department) and Yang Ke, MD, PhD, Hong Cai, MD, PhD, Jing-Jing Li, PhD (Laboratory of Genetics). In addition, we would like to thank the Ethics and Academic Committees of Peking University School of Oncology, for approving this study and Zhen-Dong Gu, MD, Meng-Meng Song, MD, Wen Wang, PhD, Hao Fu, Msc, Ya-Bing Du, Msc, Yun-Fan Ma, Msc, Hui Wang, Msc, Chuan Huang, Msc (Thoracic Surgery I).

References