Prognostic value of Dickkopf-1 and ß-catenin expression in advanced gastric cancer

Dickkopf-1 (DKK1) is a Wnt/ß-catenin pathway antagonist related to gastric cancer (GC) carcinogenesis. However, the prognostic role of combined DKK1 and ß-catenin expression in advanced GC (AGC) is not clear. In total, 158 patients with AGC who underwent gastric resection were enrolled in this study. DKK1 and ß-catenin expression was evaluated in whole tumor sections by immunohistochemistry. DKK1 expression was high in 73 (46.2%) patients, while ß-catenin expression was positive in 51 (32.3%) patients. The expression of DKK1 was positively correlated with that of ß-catenin (P < 0.001). The combined expression of DKK1 and ß-catenin was significantly associated with high N stage (N2 and N3) (P = 0.042). In addition, patients with high DKK expression demonstrated poorer overall (OS) (P < 0.001) and disease-free survival (DFS) (P = 0.001). However, there were no differences between high DKK1 expression with ß-catenin positivity and high DKK1 expression with ß-catenin negativity (OS, P = 0.379: DFS, P = 0.255). Multivariate analysis revealed that high DKK1 alone or high DKK1 with ß-catenin positivity were independent prognostic factors for both OS (high DKK1: hazard ratio [HR], 2.130; 95% confidence interval [CI]; 1.370–3.312, P = 0.001; high DKK1 with ß-catenin positivity: HR, 2.140; 95% CI, 1.343–3.409: P = 0.001) and DFS (high DKK1: HR, 2.092; 95% CI, 1.180–3.708; P = 0.012; high DKK1 with ß-catenin positivity: HR, 2.357; 95% CI, 1.291–4.306; P = 0.005). Our results indicate that high DKK1 expression regardless of ß-catenin positivity is a crucial prognostic factor for predicting tumor recurrence and survival in patients with resected AGC.

transport to the nucleus, where ß-catenin forms complexes with T-cell factor (TCF)/lymphoid enhancer factor and activates Wnt-targeted gene expression (cyclin D1, c-MYC, axin-2) [5,8]. Wnt signaling can be repressed by six Wnt antagonist families, including Dickkopf (DKK) proteins, secreted Frizzled-related proteins, WNT-inhibitory factor 1, Wise/SOST, Cerberus, and insulin-like growth-factor binding protein 4 [9]. DKK1, the most well-known Wnt antagonist, is a 35 kDa protein that contains a secreted signal peptide sequence [10]. Antagonism of the Wnt pathway via antagonists such as DKK1 is accomplished through binding to LRP 5/6 [11,12]. Despite being a negative regulator of Wnt signaling, the prognostic role of DKK1 is not well understood. Results from previous studies on DKK1 in diverse tumors have demonstrated conflicting results; some have reported that DKK1 acts as a tumor suppressor, while others have shown that it acts as an oncogene [13]. Lee et al. and Gao et al. reported that overexpression of DKK1 protein and mRNA in tissue and increased levels of DKK1 in serum were significantly associated with unfavorable prognosis in patients with GC [14,15]. In contrast, other recent studies showed low DKK1 expression in GC samples, and that restoration of DKK1 in tumor cells inhibited tumor cell growth and invasion [16,17]. However, the prognostic value of DKK1 in GC is not clearly determined. Furthermore, although DKK1 interacts with ß-catenin, the association between DKK1 and abnormal ß-catenin expression has not been well studied. Therefore, we aimed to confirm the role of DKK1 as a prognostic factor in AGC and to determine the association between DKK1 and ß-catenin in a sizable AGC cohort using immunohistochemistry.

Study population
The clinical and pathological data of 158 patients with AGC who had undergone primary radical resection between 2001 and 2005 at Uijeongbu St. Mary's Hospital of the Catholic University of Korea were reviewed. The inclusion criteria were (i) pathologically confirmed adenocarcinoma, (ii) radical resection without preoperative radiation or chemotherapy, (iii) removal of at least 15 or more lymph nodes, and (iv) available paraffinembedded tumor specimens. Postoperative pathological staging was based on the American Joint Committee on Cancer (AJCC) staging criteria, 7th edition. This study was approved by the Institutional Research Ethics Board of Uijeongbu St. Mary's Hospital of the Catholic University of Korea and adhered to the Declaration of Helsinki. Patient anonymity was preserved.

Immunohistochemistry
Immunohistochemistry was performed on formalin-fixed paraffin-embedded tissue sections. Whole tissue sections of representative tumor samples were deparaffinized using xylene and graded alcohol and then rehydrated with distilled water. Blocking of endogenous peroxidase activity was achieved by quenching with 3% hydrogen peroxide in methanol for 10 min. Antigen retrieval was then performed by heating the slides for 15 min in 0. 01 M citrate buffer (pH 6.0). The sections were incubated with human specific antibodies against DKK1 (1: 200, Abcam, Cambridge, UK) and ß-catenin (1:100, Cell Signaling, Danvers, MA, USA) at room temperature for 30 min, washed in phosphate-buffered saline, and then incubated with a peroxidase-labeled polymer conjugated to the secondary antibody for 30 min. The immunoreaction was visualized using the Polink-2 plus polymer HRP detection system (Golden Bridge International, Mukilteo, WA, USA). In the negative controls, the primary antibodies were substituted with normal rabbit or mouse IgG at the same concentration as the primary antibody. Immunohistochemical staining was independently examined by two board-certified pathologists (S.A.H. and S.H. Y.) who were blinded to the clinical data. The intensity of DKK1 staining was scored using the following scale: 0, no staining; 1, weakly positive; 2, moderately positive; 3, strongly positive. The proportion of DKK1 staining was calculated as the percentage of positive tumor cells. H scores, ranging from 0 to 300, were determined by multiplying the intensity score by the proportion. The cut-off H score for high DKK1 expression was set at 60 using maximally selected rank statistics (maxstat) calculated in R statistical programming language version 3.2.3 (https://www.r-project.org/). ß-catenin expression was considered positive if more than 10% of tumor cells showed nuclear and cytoplasmic staining [18].

Statistical methods
Chi-square or Fisher's exact tests were conducted to determine the association among DKK1 expression, ßcatenin expression, and clinicopathologic parameters. A P-value of < 0.05 was considered statistically significant. Overall survival (OS) was calculated from the date of diagnosis to the date of death due to any cause or the last follow-up visit. Disease-free survival (DFS) was calculated from the date of diagnosis to the date of the first distant or local disease recurrence or last follow-up. The Kaplan-Meier method was used to analyze 'time-to-event' data, and statistical differences in the cumulative survival curves were evaluated using the log-rank test. Cox proportional hazards regression models were used to investigate the significance of the prognostic factors. All variables with a P-value < 0.05 in the univariate analysis were included in the multivariate analysis. Survival rates and hazard ratios are shown with their respective 95% confidence intervals (CIs). Statistical analyses were performed using R statistical programming language version 3.2.

Discussion
In this study, high DKK1 expression was significantly associated with high N stage, and patients with high DKK1 expression demonstrated shorter OS and DFS. High DKK1 expression was also correlated with β-catenin positivity. On multivariate analysis, high DKK1 expression and β-catenin positivity were found to be independent adverse prognostic factors for both OS and DFS. Interestingly, patients with high DKK1 expression and concomitant β-catenin negativity also had shorter OS and DFS. Thus, the prognostic value of DKK1 in GC patients may be independent of β-catenin expression.
Various tumors have shown conflicting results regarding the role of DKK1 as a tumor suppressor or oncogene. DKK1 is a secreted protein that plays a crucial role as a negative regulator of the Wnt signaling pathway, and downregulation of DKK1 in colon cancer, breast cancer, hepatocellular carcinoma, and renal cell carcinoma suggests that DKK1 is an antagonist of the Wnt signaling pathway. However, overexpression of DKK1 has been observed in various malignant tumors and is also correlated with adverse prognosis in patients with multiple myeloma, hepatoblastoma, Wilm's tumor, lung cancer, and breast cancer [19][20][21][22]. Although DKK1 has been investigated extensively in various tumors, including GC, no previous studies have evaluated the association between DKK1 and β-catenin expression in AGC. Previous studies have shown that overexpression of DKK1 is related to adverse prognosis in GC [14,15]. This is consistent with the results we obtained in the high DKK1 expression group, which showed that regardless of β-catenin expression, high DKK1 expression is related to unfavorable prognosis. In previous studies, high DKK1 expression was found to be associated with intestinal GC, advanced T stage, vascular and lymphatic invasion, and distant metastasis. However, we only found a significant association between N stage and high DKK1 expression. This discrepancy might stem from different cut-off values and patient cohorts among studies. Our cut-off value was meticulously determined and optimized for patient survival using maximally selected rank statistics. Therefore, we suggest that our cut-off value better reflects the prognostic role of DKK1 in GC, independently of other adverse clinicopathologic parameters. Previous clinical studies of DKK1 expression in GC included both early GC and AGC, with a number of distant metastasis cases [14,15]. Survival of early gastric carcinoma patients is over 90%, but survival of patients with both GC and distant metastasis is less   than 10% [23]. Therefore, clarification of the prognostic impact of DKK1 could be confounded by the heterogenous patient cohorts used in previous studies.
Wan et al. reported that adenovirus-mediated DKK1 overexpression in CD44 + GC cells inhibits tumorigenicity through attenuating Wnt signaling [16]. The results Activation of Wnt/β-catenin is found in 30% to 50% of GC tissue samples and cell lines [6,24]. Furthermore, recent studies that used high-throughput sequencing methods showed that CTNNB1, which encodes β-catenin, is a driver gene for GC [25][26][27]. However, CTNNB1 gene mutations were only detected in 4% to 9% of sporadic GC tumors [28,29]. These findings suggest that aberrant activation of Wnt signaling pathway is modulated mainly by Wnt ligands and negative regulators, not through mutation of the CTTNB1 gene. In this study, high DKK1 expression was significantly related to positive β-catenin expression in AGC samples. Given that DKK1 inhibits β-catenin, this finding may seem contradictory. However, several tumor studies revealed that high DKK1 expression is correlated with activation of the Wnt/β-catenin pathway in both hepatocellular carcinoma and hilar cholangiocarcinoma [20,30]. These findings could be explained by a disruption in the negative feedback loop between DKK1 and the Wnt/ β-catenin pathway, which could result from a high level of secreted DKK1 [31]. In addition, DKK1 is also a downstream target gene of β-catenin/TCF, which is a direct target of activated β-catenin [32].
The OS and DFS of patients with high DKK1 and negative β-catenin expression were not different from those with high DKK1 and positive β-catenin expression. These results suggest that the effect of high DKK1 expression in AGC could be an independent of β-catenin status. Conversely, high DKK1 expression, which does not affect canonical Wnt/β-catenin signaling, is still a prognostic factor for patients with AGC.
Our results suggest that high DKK1 expression affects prognosis regardless of β-catenin activation. Several previous studies showed that DKK1 promotes malignancy via non-canonical Wnt pathway mechanisms. In hepatocellular carcinoma, high DKK1 mRNA and protein expression was correlated with poor OS and DFS. In addition, a positive relationship among DKK1 expression, JNK phosphorylation, and RhoA levels was identified [33]. These results indicate that the malignant potential can be increased by the interaction between DKK1 and the non-canonical Wnt pathway, which consists of the Wnt/Ca 2+ and Wnt/PCP pathways and does not involve activation of β-catenin [34]. Moreover, Kimura et al. reported that cytoskeleton-associated protein 4, a receptor for DKK1, mediates DKK1 signaling to promote cancer cell proliferation via the PI3K/ AKT pathway and was associated with an unfavorable prognosis in pancreatic and lung cancer patients [35]. Together, these results suggest that high DKK1 expression acts through β-catenin-independent mechanisms to increase the malignant potential and decrease survival in patients with AGC. Unfortunately, molecular targeting therapies for AGC are limited to trastuzumab and ramucirumab [36,37]. Due to the shortage of promising target agents for GC, new targets molecules with potential agents are urgently needed. The efficacy of an anti-DKK1 antibody has been investigated in multiple myeloma and prostate cancers that were associated with bone resorption [38,39]. However, further preclinical studies to determine the effectiveness of anti-DKK1 antibody in GC are required.

Conclusions
We found that high DKK1 expression was correlated with a positive β-catenin status. In addition, patients with high DKK1 expression who were positive for βcatenin had a poor prognosis. However, patients with high DKK1 expression who were negative for β-catenin also demonstrated a poor prognosis. In the multivariate analysis, high DKK1 expression only or high DKK1 expression with β-catenin positivity were an independent prognostic factor for OS and DFS in patients with AGC. Together, these results suggest that DKK1 may act as a biomarker and therapeutic target in AGC.

Additional file
Additional file 1: Table S1. The funding bodies had no role in the design of the study and collection, analysis, and interpretation of data and in the writing of the manuscript.

Availability of data and materials
Additional data and materials may be obtained from the corresponding author on reasonable request.