Expression of Cancer Stem Cell-associated DKK1 mRNA Serves as Prognostic Marker for Hepatocellular Carcinoma.

BACKGROUND/AIM
Cancer stem cells (CSCs) are associated with prognosis of hepatocellular carcinoma (HCC). In our previous study, we created cDNA microarray databases on the CSC population of human HuH7 cells. In the present study, we identified genes that might serve as prognostic markers of HCC by employing existing databases.


MATERIALS AND METHODS
Expressions of glutathione S-transferase pi 1 (GSTP1), lysozyme (LYZ), C-X-C motif chemokine ligand 5 (CXCL5), interleukin-8 (IL8) and dickkopf WNT signaling pathway inhibitor 1 (DKK1), the five most highly expressed genes in the CSC cDNA microarray databases, were examined in 99 patients with HCC by real-time polymerase chain reaction (qRT-PCR), and their clinical significance was analyzed.


RESULTS
The Kaplan-Meier analysis showed that both overall and cancer-specific survival were significantly longer in patients with low DKK1 expression than in those with high DKK1 expression. The multivariate analysis revealed that overall survival was negatively associated with albumin and positively associated with alkaline phosphatase (ALP), serosal invasion and stage, and cancer-specific survival was positively associated with ALP, portal vein invasion and DKK1 mRNA.


CONCLUSION
Expression of CSC-associated DKK1 mRNA might be an unfavorable prognostic marker for patients with HCC.


Materials and Methods
Patients and clinical samples. The samples for the gene-expression analysis were obtained from 99 patients with HCC who were admitted to Kyoto University Hospital from 1998 to 2008 and agreed to undergo surgical resection with curative intent under informed consent. Clinicopathological parameters of these patients are summarized in Table I. This study conformed with the principles of the Declaration of Helsinki and was approved by the Institutional Review Board of Tottori University Faculty of Medicine and Kyoto University Hospital (approval number: 1619). HCC clinical specimens were obtained according to the method previously described (9). Total RNA was reverse-transcribed using SuperScript II (Invitrogen, Carlsbad, CA, USA) and oligo (dT) primers. Expression levels of mRNA were measured by Applied Biosystems 7900HT Fast Real Time PCR System using EXPRESS qPCR Supermix with Premixed ROX (Applied Biosystems, Life Technologies, Foster City, CA, USA), universal probes (Roche Applied Science, Basel, Switzerland), and gene-specific primers. Universal probes and gene-specific primers used in this study are summarized in Table II.
Statistical analysis. EXCEL (Microsoft Corporation, Redmond, WA, USA) and PASW statistics (SPSS Inc., Chicago, IL, USA) were used for the statistical calculations in this study. The comparison for statistical differences was performed using Student's t-test, or Mann-Whitney U-test. The Kaplan-Meier analysis was performed for both overall and cancer-specific survival according to CSCrelated genes. The log-rank test was performed to determine the prognostic variables associated with overall survival and cancerspecific survival ratios in patients with HCC. Cox regression model was used for multivariable analysis of variables affecting overall and cancer-specific survival. Diffuses with a value of p<0.05 were considered to be statistically significant.

Results
Differential expression patterns of mRNA from CSCs, non-CSCs, and normal hepatocytes. In our previous study, the genes with at least two-fold up-regulation in CD44-positive compared to CD44-negative HuH7 cells were registered in Gene Expression Omnibus (accession number GSE84226) (9). Of these 604 genes, 216 genes in which the global normalization number was over 20 were divided into three patterns of expression as follows: the first pattern included successive increase of expression in the order of normal hepatocytes, CD44-negative cells, and CD44-positive cells.
The second pattern included successive increase in the order of CD44-negative cells, normal hepatocytes, and CD44positive cells. The third pattern included successive increase in the order of CD44-negative cells, CD44-positive cells, and  Table III. On the assumption that the genes showing a successive increase of expression in the order of normal hepatocytes, CD44-negative cells, and CD44-positive cells are associated with CSCs of HCC, the individual mRNA expressions of the top five genes were examined in normal hepatocytes, CD44-negative cells, and CD44-positive cells by qRT-PCR. C17orf45 was omitted since this gene is a nonprotein coding RNA (10). Expression of GSTP1, LYZ, CXCL5, IL8, and DKK1 was confirmed to be highly expressed in CD44-positive HuH7 cells (Figure 1).

Kaplan-Meier analysis of association of CSC-related genes
with overall and cancer-specific survival. The association of expression of LYZ, CXCR5, DKK1, IL8 and GSTP1 with overall survival was examined. Both overall survival (p=0.016, Figure 2A) and cancer specific-survival (p=0.002, Figure 2B) were significantly associated with DKK1 mRNA, but not with that for other genes.

Univariate and multivariate analyses of clinical factors and CSC-related genes for overall and cancer-specific survival.
Overall survival was significantly associated with AFP; albumin; ALP; tumor size; invasion of serosa, portal vein and bile duct; tumor stage; and DKK1 mRNA by univariate analysis, and was associated with albumin, ALP, serosal invasion and stage by multivariate analysis (Table IV). Cancer-specific survival was significantly associated with hepatitis B surface antigen (HBs-Ag), hepatitis C virus antibody (HCV-Ab); AFP; ALP; tumor number; tumor size; invasion of capsule, serosa, portal vein, bile duct and hepatic vein; stage; and DKK1 mRNA by univariate analysis, and was associated with ALP, portal vein invasion and DKK1 mRNA by multivariate analysis.  Association of DKK1 mRNA with clinicopathological variables. Expression of DKK1 mRNA was negatively associated with albumin, but was positively associated with AFP, GSTP1 mRNA, CXCL5 mRNA, and IL8 mRNA ( Figure 3A). Expression of DKK1 mRNA was higher in patients with HBsAg-positive, stage III/IV, or capsular invasion-positive tumor than HBsAg-negative, stage I/II or capsular invasion-negative, respectively ( Figure 3B).

Discussion
In the present study, we showed that DKK1 mRNA is significantly associated with cancer-specific survival in patients with HCC. A meta-analysis of prognostic significance in solid tumors reported that DKK1 overexpression predicted poor overall survival in HCC, ovarian cancer, and other cancer types (11)(12)(13)(14)(15). On the other hand, serum DKK1 was reported to be a biomarker useful for diagnosis of HCC by a large-scale and multicenter study (16). These data suggest that DKK1 plays an important role in HCC from viewpoints of biology and clinical settings. DKK gene family comprise an evolutionary conserved small genes (17). DKK genes encode secreted proteins that antagonize WNT/β-catenin signaling by binding to the WNT co-receptors LRP5 and -6. The human DKK family consists of five members, DKK1, DKK2, DKK3, DKK4, and a unique DKK3-related gene, DKKL1 (18). Dysregulation of WNT signal activation is thought to play a causative role in several types of cancer and is involved in the acquisition of stem cell-like properties of CSCs. DKK1 is an antagonist of the WNT signaling pathway, and plays crucial roles in tumor growth and progression. DKK1 levels are elevated in a wide variety of cancer types including HCC (11)(12)(13)(14)(15), and breast (19), colorectal (20), and pancreatic (21) cancer. Although how DKK1 regulates WNT signaling remains to be solved, one important fact is that DKK1 is a downstream target of WNT signaling, allowing for a negative feedback loop (18). Indeed, activation of canonical WNT signaling causes an increase in DKK1 mRNA and protein (22). These data suggest that DKK1 overexpression is a result of WNT/β-catenin pathway expression.
In the present study, by employing the previously developed databases of cDNA microarray in CSCs population, we identified DKK1 mRNA expression as a prognostic marker. Since liver CSCs have been reported to be associated with increased chemo/radioresistance, earlier recurrence after surgical or locoregional treatment, increased invasiveness, metastasis, and poor prognosis (23), it is clinically useful to identify novel prognostic genes which are highly expressed in liver CSCs.
In conclusion, by employing our previously developed databases of cDNA microarray in CSCs, we identified DKK1 mRNA expression in HCC tissues as representing cancerspecific survival of patients with HCC. Therefore it is expected to serve as a novel prognostic marker in HCC.