Expression quantitative trait loci for PAX8 contributes to the prognosis of hepatocellular carcinoma

Paired-box family member PAX8 encodes a transcription factor that has a role in cell differentiation and cell growth and may participate in the prognosis of hepatocellular carcinoma (HCC). By bioinformatics analysis, we identified several single nucleotide polymorphisms (SNPs) within a newly identified long non-coding RNA (lncRNA) AC016683.6 as expression quantitative trait loci (eQTLs) for PAX8. Hence, we hypothesized that PAX8eQTLs in lncRNA AC016683.6 may influence the HCC prognosis. We then performed a case-only study to assess the association between the two SNPs as well as the prognosis of HCC in 331 HBV-positive HCC patients without surgical treatment. Cox proportional hazard models were used for survival analysis with adjustments for the age, gender, smoking status, drinking status, Barcelona-Clinic Liver Cancer (BCLC) stage, and chemotherapy or TACE (transcatheter hepatic arterial chemoembolization) status. We found that the G allele of rs1110839 and the T allele of rs4848320 in PAX8was significantly associated with a better prognosis compared with the T allele of rs1110839 and the C allele of rs4848320 (adjusted HR = 0.74, 95% CI = 0.61–0.91, P = 0.004 for rs1110839 and adjusted HR = 0.71, 95% CI = 0.54–0.94, P = 0.015 for rs4848320 in the additive model). Furthermore, the combined effect of the variant genotypes for these two SNPs was more prominent in patients with the BCLC-C stage orpatients with chemotherapy or TACE. Although the exact biological function remains to be explored, our findings suggest a possible association of PAX8eQTLs in lncRNA AC016683.6 with the HCC prognosis inthe Chinese population. Further large and functional studies are needed to confirm our findings.


Introduction
Liver cancer is the fifth most common cancer worldwide and the third greatest cause of cancer-related death worldwide [1].Hepatocellular carcinoma (HCC) is the most common type of liver cancer [2].Although surgical resection, liver transplantation, radiotherapy and some other therapies are potentially effective treatments for HCC, studies have shown that HCC has an increasing incidence and a poor 5-year survival rate of approximately 7% despite treatment [3][4][5].The Barcelona Clinic Liver Cancer (BCLC) staging classification has been widely endorsed to predict the prognosis of HCC patients; however, remarkably different survival outcomes among HCC patients at the same stage suggest that the existence of other important factors might affect the prognosis.This current status has inspired researchers to focus on identifying molecular biomarkers to guide individualized treatment and improve the prognosis of cancer patients.
Paired-box gene 8 (PAX8), a member of the PAX gene family, encodes a transcription factor that has a role in cell differentiation and cell growth, and silencing of PAX8 in cell culture results in cell death [6].PAX8 is known to activate the transcription of BCl2, which is an antiapoptotic gene that is also involved in p53suppression, suggesting it plays a role in tumor initiation and progression [7].A recent study also showed that overexpression of PAX8 protein by endometrial cancer is associated with poor disease outcomes [6].Accumulating data have revealed that PAX8is expressed in a high percentage of kidney, thyroid, ovarian, and lung carcinomas [8][9][10][11].However,the potential role of PAX8 in HCC prognosis has been rarely explored.
Long non-coding RNAs (lncRNAs) are defined as non-protein-coding transcripts,which are usuallylonger than 200 nucleotides.Based on the recent studiesonlncRNAs, it is reasonable to believe thatlncRNAsare important for regulating gene expression in the nucleus, exerting their biological functions.Recent studieshave provideda comprehensive generalization on the functions of lncRNAsthatmay modulate transcription or post-transcription via targeting the splicing, stability, or translation of mRNAs [12].LncRNAAC016683.6islocated in the intron region of PAX8, which belongs to chromosome 2q13.Using bioinformatics analyses, we identified two single nucleotide polymorphisms (SNPs) (rs1110839 and rs4848320) in AC016683.6thatmay be the expression quantitative trait loci (eQTLs) for PAX8 (http://www.regulomedb.org)[13].Therefore, it is likely that the two SNPs could influence the interaction between AC016683.6 and PAX8, regulating the expression of PAX8.According to the potential role of PAX8, we hypothesized that PAX8eQTLsmayinfluence the development and progression of HCC.To validate our hypothesis, we examined the associations between the two SNPs of PAX8and the prognosis of 331patients with intermediate or advanced HCC in China.

Study subjects
This study was approved by the institutional review board of Nanjing Medical University.All participants provided written informed consent to participate in this study and the ethics committees approved of this consent procedure.Theenrollment of participants was previously described [14,15].In consideration for constructing a relatively homogenous population, our current study was restrained to HCC patients who did not undergo surgery in the intermediate stage (B) or advanced stage (C) according to the Barcelona Clinic Liver Cancer (BCLC) staging system [16,17].Briefly, 414 intermediate or advanced HCC patients were consecutively recruited from Nantong Tumor Hospital and the First Affiliated Hospital of Nanjing Medical University, Jiangsu, China.All participants were newly diagnosed and histopathologically confirmed HCC cases.Then, we prospectively conducted a follow-up study every 3 months from the time of enrollment by personal or family contacts until death or final follow-up.As a result, 331 intermediate or advanced HCC patients who had completed follow-ups and whose clinical information was available were enrolled in our study with the response rate of 80.0%.The maximum follow-up time (MFT) for the 331 patients was 60.7 months (final follow-up in January 2013), and the median survival time (MST) was 14.5 months.

Genotyping
Followingthe traditional method [18], genomic DNA was extracted from a leukocyte pellet by a series of treatments.Then, all SNPs were genotyped using the Sequenom Mass ARRAY iPLEX platform (SequenomInc).To reduce the false positive and error rates, three blank (water) controls were detected in each384-well plate during the sample testingevery time, and more than 10% samples were randomly repeated for quality control, checking whether the latter results coincided with the former.The success rates of genotyping for the two SNPs were all above 98%.

Statistical analysis
The survival time was calculated from the date of HCC diagnosis to the date of patient death or the last follow-up.The associations between the median survival time(MST) and demographic characteristics, clinical features and genotypes were estimated using the Kaplan-Meier method and log-rank test.Univariate and multivariable Cox proportional hazard regression analyses were conducted to estimate the crude or adjusted hazard ratio (HR) and their 95% confidence intervals (CI) with adjustment for the age, gender, smoking status, drinking status, BCLC stage, and chemotherapy or TACE (transcatheter hepatic arterial chemoembolization) status.The Cox stepwise regression model was also performed to determine the predictive factors of HCC prognosis with a significance level of 0.050 for entering and 0.051 for removing the respective explanatory variables.The heterogeneity between subgroups was assessed with the chi-square-based Q-test, and the heterogeneity was considered significant for P<0.10.Analysiswas performed using Stata software (version 11.0;Stataconference, Chicago).All tests were two-sided, and the criterion of statistical significance was set at P<0.05.

Patients' characteristics and clinical features
The demographic characteristics and clinical information of the 331 HCC patients in stage B or C included in this study were previously described [15].In brief, 258 of the 331 patients died from HCC, and 2 died from other causes during a period of up to 60.7 months of follow-up.For the disease-specific survival analysis, the latter were considered censored data in the analyses.Chemotherapy or TACE and the drinking status were significantly associated with the patient survival time (log-rank P 0.001 and 0.006 for the drinking status and chemotherapy or TACE status, respectively).Compared to those who received neither chemotherapy nor TACE therapy (MST = 3.4 months), patients with chemotherapy or TACE therapy (MST = 16.8 months) had a significantly decreased risk of death (61%, HR = 0.39; 95% CI = 0.29-0.51).

Effects of PAX8 polymorphisms on HCC survival
The associations of the two SNPs with HCC survival were examined in an additive model by the Kaplan-Meier method.As shown in Table 1, patients carrying rs1110839 GT/GG genotypes and rs4848320 CT/TT genotypes had a longer survival time(MST:14.3months for rs1110839 GT/GG and 15.4 months for rs4848320 CT/TT) than those carrying the rs1110839 TT and rs4848320 CC genotypes (MST:13.4monthsfor rs1110839 TT and 13.0 months for rs4848320 CC).Furthermore, multivariable Cox regression analysis showed that rs1110839 and rs4848320 could be considered significant prognostic markers for HCC (Table 1).After adjusting for the age, gender, smoking status, drinking status, BCLC stage, and chemotherapy or TACE status, variant genotypes of rs1110839 and rs4848320 were significantly associated with a favorable HCC prognosis (adjusted HR = 0.74, 95% CI = 0.61-0.91,P = 0.004 for rs1110839 andadjusted HR = 0.71, 95% CI = 0.54-0.94,P = 0.015 for rs4848320).

Stratified analysis and interaction effects
We then examined the combined effect of these two variants on the HCC survival and observed a significant locus-dose effect between favorable genotypes and the risk of death (P for trend <0.001).Compared to patients without favorable genotypes (MST = 12.6 months), those patients with 1 or 2-4 favorable genotypes had a significantly longer MST (13.3 and 14.9 months, respectively).After adjusting for the age, gender, smoking status, drinking status, BCLC stage, and chemotherapy or TACE status, patients with 1 or 2-4 favorable genotypes had 0.22-and 0.43-fold decreased risks of HCC-specific deaths, respectively (Table 2).
Theassociations between PAX8SNPs and HCC survival were further investigated by stratified analysis of the age, gender, smoking status, drinking status, BCLC stage, and chemotherapy or TACE status.As shown in Table 3, we found that the protective effect of combined variant genotypes was more prominent in B stage patients and patients without chemotherapy and TACE (adjusted HR = 0.73, 95% CI = 0.62-0.86;adjusted HR = 0.66, 95% CI = 0.49-0.87,respectively) than in C stage patients or patients with chemotherapy or TACE therapy (adjusted HR = 1.30, 95% CI = 0.77-2.21and adjusted HR = 0.96, 95% CI = 0.80-1.15;P = 0.040 and 0.026 for the heterogeneity test, respectively).Therefore, the gene-BCLC stage and genechemotherapy or TACE status interaction analysis were performed, and statistically significant multiplicative interactionswere observed, as shown in Tables 4 and 5 (P for multiplicative interaction = 0.029 and <0.001, respectively).

Stepwise Cox regression analyses
We finally performed stepwise Cox proportional hazard analysis to evaluate the effects of the age, gender, smoking status, drinking status, BCLC stage, and chemotherapy or TACE status and PAX8eQTLs on HCC survival.Four variables (Chemotherapy or TACE status, age, drinking status and PAX8eQTLs) were selectedfor use in the final regression model.Furthermore, when the gender, drinking status, chemotherapy or TACE status, and PAX8eQTLswere included in the final model, the PAX8eQTLs remained an independent protective factor for HCC survival (HR = 0.78, 95% CI = 0.67-0.91,P = 0.001)(Table 6).

Discussion
In the present case-cohort study, we demonstrated that PAX8eQTLs (rs1110839 and rs4848320) may be an independent candidate biomarker for predicting HCC survival in Chinese with a prospective study design.Furthermore, the combined effect of these two SNPs was more prominent in patients with BCLC-B stage who did not receivechemotherapy or TACE.Since the ENCODE Project and RNA-seq analysis have identified thousands of new lncRNAs, the genetic variants and biological function of lncRNAs are becoming a focus in studies of complex diseases.We previously reported that ZNRD1(Zinc ribbon domain containing 1)eQTLsincreased the risk of HCCLncRNA AC016683.6,located in the intron region of PAX8, may influence the expression of PAX8.A PAX family member, PAX8, is implicated in the development of the kidney, thyroid gland, Mullerian and Wolffian ducts, and others [19,20].Tumors derived from these organs, as mentioned above, also typically express PAX8 [21].One study showed that PAX8 activates the transcription of the NCAM gene through binding sequences resembling paired domain binding sites in the NCAM promoter [22].NCA-Mis expressed on the surface of immune cells [23], and it has been shown to mediate adhesion between immune cells, which is particularly important in inflammation [24].Previous studies have shown that abnormal cell growth and proliferation are often associated with high expression levels of PAX genes.In fact, overexpression of PAX proteins does not appear to be an initiating or transforming molecular event in tumor pathogenesis, but it facilitates malignant development through the effects of PAX genes on apoptosis resistance, tumor cell proliferation and migration, and repression of terminal differentiation [25].Our results showed that the variant genotypes (rs1110839 GT/GG and rs4848320 CT/TT genotypes) of the two PAX8-eQTLs in lncRNAAC016683.6promotedbetterprognosis in HCC patients with intermediate or advanced disease.
There are several limitations ofthe study that need to be addressed by further research.First, further validation needs to be conducted.Large-scale studies are required to validate the associations between the two eQTLs in AC016683.6 and the HCC prognosis.Second, the biological function of the two eQTLsin AC016683.6 was not further investigated in this study.Additionally, our previous study found that ZNRD1eQTLs rs3757328 in ZNRD1-AS1 (ZNRD1 antisense RNA1) was associated with an increased risk for HCC, and further eQTL analysis indicated the significant association between the genotypes of rs3757328 and the expression of ZNRD1 and ZNRD1-AS1.In vitro experiments have also demonstrated that ZNRD1 knockdown inhibits the expression of HBV mRNA and promotes the proliferation of HepG2.2.15 cells.Given the findings from our previous study [26], we hypothesized that lncRNAAC016683.6might regulate the expression of a related protein (PAX8) based on its variation, influencing the prognosis of hepatic tumors.
To the best of ourknowledge, this is the first study investigating the association between PAX8eQTLs in the lncRNAAC016683.6andHCCprognosis.This study showed that the variant rs1110839 GT/GG and rs4848320 CT/TT genotypes in lncRNA AC016683.6influenced the prognosis of hepatic tumor patients.These results suggested that AC016683.6rs1110839 and rs4848320 might serve as susceptibility markers for HCC survival.Further studies incorporating diverse populations and functional assays are warranted to validate and extend our findings.

Table 4 . Interactive effect of combined effect of two SNPs genotypes and bclc stage associated with HCC patients' survival. BCLCstage No. of favorable genotype Patients Deaths Adjusted HR a (95%CI) P a
Adjusted for age, gender,smoking, drink,Chemotherapy/TACE. a

Table 6 . Multivariate Cox regression analysis on HCC patients' survival. Variables β a SE b HR (95% CI) P
β is the estimated parameter of the regression model.SE is the standard error of the regression model.https://doi.org/10.1371/journal.pone.0173700.t006 a b