Skip to main content
NCBI home page
Cancer Medicine logoLink to Cancer Medicine
. 2017 Jun 22;6(7):1512–1522. doi: 10.1002/cam4.1067

Genetic variants of cell cycle pathway genes predict disease‐free survival of hepatocellular carcinoma

Shun Liu 1,†, Tian‐Bo Yang 2,†, Yue‐Li Nan 3, An‐Hua Li 4, Dong‐Xiang Pan 4, Yang Xu 1, Shu Li 1, Ting Li 5, Xiao‐Yun Zeng 1,‡,✉, Xiao‐Qiang Qiu 1,‡,✉
PMCID: PMC5504311  PMID: 28639733

Abstract

Disruption of the cell cycle pathway has previously been related to development of human cancers. However, associations between genetic variants of cell cycle pathway genes and prognosis of hepatocellular carcinoma (HCC) remain largely unknown. In this study, we evaluated the associations between 24 potential functional single nucleotide polymorphisms (SNPs) of 16 main cell cycle pathway genes and disease‐free survival (DFS) of 271 HCC patients who had undergone radical surgery resection. We identified two SNPs, i.e., SMAD3 rs11556090 A>G and RBL2 rs3929G>C, that were independently predictive of DFS in an additive genetic model with false‐positive report probability (FPRP) <0.2. The SMAD3 rs11556090G allele was associated with a poorer DFS, compared with the A allele [hazard ratio (HR) = 1.46, 95% confidential interval (95% CI) = 1.13–1.89, P = 0.004]; while the RBL2 rs3929 C allele was associated with a superior DFS, compared with the G allele (HR = 0.74, 95% CI = 0.57–0.96, P = 0.023). Additionally, patients with an increasing number of unfavorable genotypes (NUGs) of these loci had a significant shorter DFS (P trend = 0.0001). Further analysis using receiver operating characteristic (ROC) curves showed that the model including the NUGs and known prognostic clinical variables demonstrated a significant improvement in predicting the 1‐year DFS (P = 0.011). Moreover, the RBL2 rs3929 C allele was significantly associated with increased mRNA expression levels of RBL2 in liver tissue (P = 1.8 × 10−7) and the whole blood (P = 3.9 × 10−14). Our data demonstrated an independent or a joint effect of SMAD3 rs11556090 and RBL2 rs3929 in the cell cycle pathway on DFS of HCC, which need to be validated by large cohort and biological studies.

Keywords: Cell cycle pathway, hepatocellular carcinoma, single‐nucleotide polymorphism, survival

Introduction

Hepatocellular carcinoma (HCC) is one of the most frequently diagnosed cancers and the leading causes of cancer death worldwide. As a high incidence area of HCC, China accounts for about 50% of the total number of cases and deaths worldwide 1, 2. In 2015, it had an estimated of 466,100 and 422,100 patients is diagnosed and die of HCC in China, respectively1, which makes HCC a major public health concern in China. For the past decade, cancer staging systems, such as the Barcelona Clinic Liver Cancer (BCLC) staging system and Cancer of the Liver Italian Program (CLIP) score, have been established for the prognosis assessment by using clinical variables mainly including tumor size, number of primary tumors, portal vein tumor thrombosis, and inflammatory degree 3, 4. But none of these systems has been universally adopted by clinicians. Many novel molecular biomarkers, such as Alpha‐fetoprotein, circulating tumor cells and circulating microRNAs, have been also applied to predict HCC progression 5. However, limitations of low sensitivity and specificity are still associated with these biomarkers when used in disease staging 5. Long‐term outcome prediction for HCC patients remains unfavorable 6, 7, which hampers the development of personalized clinical assessment for HCC patients and calls for identifying additional, more discriminant prognostic indicators.

In recent years, the development of human genome research has been witnessed much success in mining cancer‐related germline genetic variants, especially single nucleotide polymorphisms (SNPs). The current advance in genome‐wide association studies (GWASs) has led to the identification of several SNPs that are associated with HCC risk 8, 9, 10, 11. Additionally, a series of studies also reported that the HCC patients survival was associated with SNPs in genes functioning in pivotal biological pathways or processes, such as cytokine genes 12, 13, telomere maintenance genes 14, DNA repair genes 15, Wnt/β‐catenin pathway genes 16. These pieces of evidence demonstrate that genetic variants act as indicators of different processes of HCC and may provide further information beyond current clinical staging and pathologic prognostic assessments.

The cell cycle is a set of organized and monitored events responsible for proper cell division into two daughter cells, which includes four traditional subdivision phases (i.e., G1, S, G2 and M) and is controlled by the signaling pathway comprising genes encoding cyclins, cyclin‐dependent kinases, and cyclin kinase inhibitors as well as the related regulators (e.g., MYC) 17, 18, 19. Disruption of the cell cycle pathway can result in cell cycle arrest and has previously been related to prognosis of human cancers 20, 21. For example, CDKN2A and RB1 expression levels have been associated with survival of patients with advanced‐stage ovarian cancer 22. In addition, several studies have also demonstrated that aberrant expression of cell cycle pathway genes, including cyclin (A, D1 and E), CDC2, p27 and p21, is associated with prognosis for patients with HCC 23, 24, 25, 26, 27.

We have previously investigated the association of genetic variants in cell cycle pathway genes with susceptibility to HCC and found that SNPs in MCM4, CHEK1 and KAT2B were associated with HCC risk 28. There are additional studies demonstrating the associations between genetic variants of cell cycle pathway genes and survival of cancers of the ovaries, oral cavity and lung 29, 30, 31. However, the associations between genetic variants of cell cycle pathway genes and prognosis of HCC remain largely unknown. In this study, we hypothesized that genetic variants in cell cycle pathway genes are associated with survival of HCC patients. To test this hypothesis, we evaluated the associations of potential functional SNPs in 16 main cell cycle pathway genes (CDC25C, CDC7, CDKN1A, CDKN2A, CHEK1, MCM4, MCM7, MYC, ORC6L, KAT2B, PLK1, RAD21, RBL2, SMAD3, TGFB3, and YWHAB) with clinical outcomes of 271 HCC cases who had undergone radical surgery resection.

Materials and Methods

Study population

The patients were derived from our previously published case–control study 28. A total of 1127 patients with newly diagnosed and pathologically confirmed HCC were recruited consecutively between June 2007 and December 2013 from the First Affiliated Hospital and Affiliated Tumor Hospital of Guangxi Medical University. Among these patients, 271 undergone radical surgery resection with completed clinical information and long‐term follow‐up data for survival analysis. Clinical information (including age at diagnosis, sex, ethnicity, hepatitis B virus infection, smoking status, drinking status, capsule, portal vein tumor thrombosis, cirrhosis, and BCLC staging) was collected from medical records. Peripheral venous blood was collected in a vacuum EDTA anticoagulant tube from each participant. Genomic DNA was extracted using a standard phenol/chloroform extraction method and stored at −80°C. The study protocol was approved by the ethical committee of Guangxi Medical University, and a written informed consent was obtained from each of the participants.

Follow‐up

All of the patients were followed up to collect data on postoperative recurrence, distant metastasis and death condition through phone calls or the medical records. The endpoint was disease‐free survival (DFS). DFS was defined as the interval between the date of tumor resection and the date of the presence of relapse/metastasis/death or until the last follow‐up. Patients who were lost to follow‐up, who were alive at the last follow‐up and who did not have developed recurrence/metastasis were censored.

SNP selection and genotyping

The methods of SNP selection and genotyping have been described previously 28. In brief, we found 3826 differential expressed genes by analysis of three HCC‐related genome expression microarrays (GSE14520, GSE25097, and GSE12941) from GEO database. After performing the gene ontology classification and pathway enrichment analysis by using blast2GO and DAVID (https://david.ncifcrf.gov/), we identified 40 cell cycle pathway genes involved in the cellular process. Considering the literature on these genes with cancer survival, we selected 16 genes (CDC25C, CDC7, CDKN1A, CDKN2A, CHEK1, MCM4, MCM7, MYC, ORC6L, KAT2B, PLK1, RAD21, RBL2, SMAD3, TGFB3, and YWHAB) for further analysis. The genotype information of the 16 genes were downloaded from Hapmap website (http://hapmap.ncbi.nlm.nih.gov/), and functional SNPs were selected using SNPinfo (https://snpinfo.niehs.nih.gov/) 32. Considering the minor allele frequency (MAF ≥ 0.05), linkage disequilibrium (LD, r 2 > 0.8) and potential functions in transcription factor‐binding, microRNA‐binding and splicing sites, we finally included 24 SNPs with putative functions in this study (Table 1). Then, we performed the genotyping by using a MassARRAY system (Sequenom, San Diego, CA, USA) and a matrix‐assisted laser desorption ionization‐time of flight mass spectrometry method according to the manufacturer's instructions. All the primers for PCR were designed using the Assay Designer software package of the Sequenom system (San Diego, CA, USA).

Table 1.

SNPs selection and prediction function in SNPinfo

Genes SNP CHR Position Allelea MAF Prediction function in SNPinfob
CDC25C rs3734166 5 137693222 A/G 0.42 Nonsynonymous variant
CDC7 rs13447539 1 91758416 A/G 0.05 Splicing site and nonsynonymous variant
CDKN1A rs3176320 6 36754766 A/G 0.24 Transcription factor binding site
CDKN1A rs3176329 6 36755441 G/T 0.09 Transcription factor binding site
CDKN2A rs3088440 9 21958159 G/A 0.12 Transcription factor and microRNA‐binding site
CHEK1 rs3731399 11 125002457 T/C 0.17 Transcription factor binding site
CHEK1 rs515255 11 125002355 C/T 0.40 Transcription factor binding site
MCM4 rs2305952 8 49037162 T/C 0.17 Transcription factor binding site
MCM7 rs2070215 7 99534733 A/G 0.17 Splicing site and nonsynonymous variant
MCM7 rs2261360 7 99530929 C/A 0.40 Transcription factor and microRNA‐binding site
MCM7 rs4928 7 99528626 C/G 0.06 Splicing site
MYC rs4645948 8 128817680 C/T 0.13 Transcription factor binding site
ORC6L rs12596237 16 45284886 C/T 0.05 Transcription factor binding site
ORC6L rs33994299 16 45281099 T/C 0.06 Transcription factor binding site and splicing site
PCAF rs17006625 3 20136100 A/G 0.14 Splicing site and nonsynonymous variant
PCAF rs4858770 3 20169427 C/T 0.47 MicroRNA‐binding site
PLK1 rs2230915 16 23608959 G/A 0.05 Splicing site and microRNA‐binding variant
RAD21 rs16889105 8 117952045 A/G 0.40 Transcription factor and microRNA‐binding site
RAD21 rs6987652 8 117951462 G/A 0.16 Transcription factor binding site
RBL2 rs3929 16 52081809 G/C 0.21 MicroRNA‐binding site
SMAD3 rs11556090 15 65273437 A/G 0.11 MicroRNA‐binding site
SMAD3 rs8025774 15 65270330 C/T 0.50 MicroRNA‐binding site
TGFB3 rs3917148 14 75516274 A/C 0.14 Transcription factor binding site
YWHAB rs2425675 20 42968348 G/A 0.18 Transcription factor binding site
Open in a new tab

SNP, single‐nucleotide polymorphism; CHR, chromosome; MAF, minor allele frequency.

a

Reference allele/minor allele.

Statistical methods

Cox proportional hazards regression analysis was conducted to assess the associations between SNPs and DRS in an additive model with adjustment for age at diagnosis, sex, ethnic, hepatitis B virus infection, smoking status, drinking status, capsule, portal vein tumor thrombosis, cirrhosis and BCLC staging. The false‐positive report probability (FPRP) approach was applied to correct for multiple comparisons 33. We assigned a prior probability of 0.1 to detect an HR of 2.0 for the adverse genotypes and alleles of SNPs with an elevated risk. Only the significant results with FPRP <0.2 were considered noteworthy. The stepwise Cox model including noteworthy SNPs and clinical variables was conducted to choose independent predictive SNPs. Kaplan–Meier estimation of survival functions and Log‐rank tests were used to evaluate the single and combined effects of risk genotypes on DFS. The ROC curve was used to evaluate the performance of different prediction models, and Delong's test was conducted to compare the area under the curves (AUC) across different models 34. Finally, to assess the correlation between genetic variants and mRNA expression of the corresponding genes, the expression quantitative trait loci (eQTL) analysis was also performed using data from the GTEx project (http://www.gtexportal.org/home/) and the HapMap3 Project (Release version 2) by general linear regression model in additive genetic model 35, 36. All analyses were performed using SAS (version 9.1.3; SAS Institute, Cary, NC), unless otherwise specified.

Results

Basic characteristics of study populations

The details of clinical information about age at diagnosis, sex, ethnicity, hepatitis B virus infection, smoking status, drinking status, capsule, portal vein tumor thrombosis, cirrhosis status, BCLC staging and survival outcomes are showed in Table 2. There were 271 cases of staging 0‐C (BCLC staging) who had undergone radical surgery resection with a median follow‐up time of 19.0 months. At the time of analysis, 185 (68.3%) patients reached the DFS endpoint. The median DFS time was 20.3 months and 1, 3 and 5‐year survival rate were 55.4%, 40.2%, and 28.9%, respectively. Univariate analysis indicated that only portal vein tumor thrombosis, cirrhosis, and BCLC staging were significantly associated with HCC DFS (P = 0.007, 0.030 and 0.004, respectively).

Table 2.

Association between characteristic factors and HCC DFS by univariate Cox proportional hazards regression model

Factors Patients Event (%)a HR (95% CI) P
Sex
Female 24 19 (79.2) 1.00
Male 247 166 (67.2) 0.99 (0.62–1.59) 0.964
Age
≤50 164 112 (68.3) 1.00
>50 107 73 (68.2) 0.88 (0.66–1.18) 0.403
Ethnicity
Han 175 125 (71.4) 1.00
Zhuang/others 96 60 (62.5) 0.85 (0.62–1.15) 0.285
HBsAg
Negative 32 22 (68.8) 1.00
Positive 239 163 (68.2) 1.00 (0.64–1.56) 0.987
Smoking
Never 175 120 (68.6) 1.00
Ever 96 65 (67.7) 1.11 (0.82–1.50) 0.503
Drinking
Never 173 118 (68.2) 1.00
Ever 98 67 (68.4) 1.09 (0.81–1.47) 0.587
Intact capsule
No 226 152 (67.3) 1.00
Yes 45 33 (73.3) 1.11 (0.77–1.62) 0.573
Portal vein tumor thrombosis
No 214 139 (65) 1.00
Yes 57 46 (80.7) 1.59 (1.13–2.22) 0.007
Cirrhosis
No 98 75 (76.5) 1.00
Yes 173 110 (63.6) 0.72 (0.54–0.97) 0.030
BCLC staging
0/A 86 50 (58.1) 1.00
B/C 176 128 (72.7) 1.62 (1.17–2.25) 0.004
Missing 9
Open in a new tab

HCC, hepatocellular carcinoma; DFS, disease‐free survival; HRs, hazards ratio; CI, confidence interval; HBsAg, hepatitis B virus surface antigens; BCLC, Barcelona Clinic Liver Cancer.

a

Including relapse, metastasis and death.

Survival analysis of SNPs and DFS

As shown in Table 3, 16 SNPs were genotyped successfully in all HCC patients, while the other eight SNPs had a few missing data. Because ORC6L rs12596237 had only one single genotype, we included 23 SNPs for further analysis. We first performed Cox proportional hazards regression analysis to assess associations of individual SNPs in genes of the cell cycle pathway with DFS. We found that four SNPs were significantly associated with DFS at P < 0.05 under an additive genetic model, of which three SNPs, i.e. rs11556090 (P = 0.004, FPRP = 0.03), rs8025774 (P = 0.020, FPRP = 0.15) and rs3929 (P = 0.023, FPRP = 0.17), had FPRP <0.2. To assess the independent prognostic factor of DFS, we further conducted stepwise Cox regression analysis by including the clinical variables and the significant SNPs. The results demonstrated that BCLC staging, SMAD3 rs11556090, and RBL2 rs3929 were still significantly associated with DFS of HCC patients (Table 4). Taken all together, we selected SMAD3 rs11556090 A>G and RBL2 rs3929G>C as the final independent SNPs for further analyses. The identified SNPs were both predicted to be in the miRNA‐binding sites (Table 1), which may affect the binding capacity of miRNA and the corresponding gene. As shown in Table 5, our results demonstrated that the SMAD3 rs11556090G allele was significantly associated with a shorter DFS (trend test: P = 0.004), while the RBL2 rs3929 C allele was associated with a superior DFS (trend test: P = 0.023). Similarly, the SMAD3 rs11556090 GG+GC genotypes were associated with a shorter DFS, compared with the CC genotype (HR = 1.54, 95% CI = 1.13–2.09, P = 0.006), and similar results were found for the RBL2 rs3929 CC+GC genotypes, compared with the GG genotype (HR = 4.98, 95% CI = 1.57–15.83, P = 0.006). For the visual effect, we used Kaplan–Meier curves to depict associations between the significant SNPs and DFS (Fig. 1A–D).

Table 3.

Association between SNPs of cell cycle pathway genes and HCC DFS by multivariate Cox proportional hazards regression model

SNP Gene Allelea Frequency of genotypes (%) Missing DFS
MhoG HetG RhoG HR (95% CI)b P b FPRP
rs3734166 CDC25C A/G 108 (39.9) 136 (50.2) 27 (10) – 0.92 (0.73–1.16) 0.460 0.81
rs13447539 CDC7 A/G 260 (95.9) 11 (4.1) – – 0.69 (0.28–1.73) 0.433 0.84
rs3176320 CDKN1A A/G 145 (53.5) 112 (41.3) 14 (5.2) – 0.91 (0.70–1.17) 0.445 0.80
rs3176329 CDKN1A G/T 177 (65.3) 73 (26.9) 7 (2.6) 4 (1.5) 1.12 (0.83–1.50) 0.466 0.81
rs3088440 CDKN2A G/A 193 (71.2) 73 (26.9) 5 (1.8) – 0.86 (0.63–1.19) 0.370 0.77
rs3731399 CHEK1 T/C 206 (76.0) 59 (21.8) 5 (1.8) 1 (0.4) 1.22 (0.89–1.68) 0.224 0.67
rs515255 CHEK1 C/T 106 (39.1) 129 (47.6) 31 (11.4) 5 (1.8) 1.01 (0.81–1.26) 0.938 0.89
rs2305952 MCM4 T/C 195 (72.0) 74 (27.3) 2 (0.7) – 1.02 (0.74–1.41) 0.917 0.89
rs2070215 MCM7 A/G 123 (45.4) 110 (40.6) 38 (14) – 1.08 (0.88–1.33) 0.464 0.81
rs2261360 MCM7 C/A 117 (43.2) 111 (41.0) 42 (15.5) 1 (0.4) 0.99 (0.80–1.22) 0.910 0.89
rs4928 MCM7 C/G 243 (89.7) 27 (10.0) 1 (0.4) – 0.93 (0.57–1.54) 0.782 0.88
rs4645948 MYC C/T 192 (70.8) 74 (27.3) 5 (1.8) – 0.95 (0.70–1.30) 0.744 0.87
rs12596237 ORC6L C/T 271 (100) – – – – – –
rs33994299 ORC6L T/C 254 (93.7) 16 (5.9) – 1 (0.4) 0.44 (0.21–0.96) 0.039 0.48
rs17006625 PCAF A/G 153 (56.5) 100 (36.9) 18 (6.6) – 1.03 (0.81–1.30) 0.823 0.88
rs4858770 PCAF C/T 99 (36.5) 130 (48.0) 38 (14.0) 4 (1.5) 0.98 (0.77–1.24) 0.861 0.89
rs2230915 PLK1 G/A 251 (92.6) 19 (7.0) 1 (0.4) – 1.09 (0.65–1.83) 0.742 0.87
rs16889105 RAD21 A/G 114 (42.1) 112 (41.3) 45 (16.6) – 1.17 (0.94–1.45) 0.157 0.59
rs6987652 RAD21 G/A 199 (73.4) 66 (24.4) 6 (2.2) – 0.88 (0.65–1.21) 0.438 0.80
rs3929 RBL2 G/C 160 (59.0) 95 (35.1) 14 (5.2) 2 (0.7) 0.74 (0.57–0.96) 0.023 0.17
rs11556090 SMAD3 A/G 162 (59.8) 98 (36.2) 11 (4.1) – 1.46 (1.13–1.89) 0.004 0.03
rs8025774 SMAD3 C/T 79 (29.2) 141 (52.0) 50 (18.5) 1 (0.4) 0.76 (0.61–0.96) 0.020 0.15
rs3917148 TGFB3 A/C 212 (78.2) 56 (20.7) 2 (0.7) 1 (0.4) 0.97 (0.69–1.38) 0.872 0.89
rs2425675 YWHAB G/A 173 (63.8) 88 (32.5) 10 (3.7) – 1.01 (0.77–1.33) 0.919 0.89
Open in a new tab

SNP, single‐nucleotide polymorphism; HCC, hepatocellular carcinoma; DFS, disease‐free survival; MhoG, major homozygous genotype; HetG, heterozygous genotype; RhoG, rare homozygous genotype; HR, hazards ratio; CI, confidence interval; FPRP, false positive report probability.

a

Reference allele/minor allele.

b

Adjusted by age, sex, ethnicity, HBsAg status, smoking status, drinking status, intact capsule, portal vein tumor thrombosis, cirrhosis and BCLC staging.

Table 4.

Predictors of DFS obtained from stepwise Cox proportional hazards regression analysis

Parameter Parameter estimate Standard error Chi‐square HR (95% CI)a P a
Age −0.11 0.16 0.45 0.90 (0.66–1.22) 0.501
Sex 0.02 0.25 0.01 1.02 (0.63–1.67) 0.930
Ethnicity −0.13 0.15 0.77 0.88 (0.65–1.18) 0.380
BCLC staging 0.45 0.14 9.73 1.57 (1.18–2.08) 0.002
rs3929 −0.27 0.13 4.39 0.76 (0.59–0.98) 0.036
rs11556090 0.30 0.13 5.49 1.36 (1.05–1.75) 0.019
Open in a new tab

DFS, disease‐free survival; HR, hazards ratio; CI, confidence interval; BCLC, Barcelona Clinic Liver Cancer.

a

Stepwise analysis included sex, age, ethnicity, HBsAg status, smoking status, drinking status, intact capsule, portal vein tumor thrombosis, cirrhosis, BCLC staging, RBL2 rs3929, SMAD3 rs8025774, and SMAD3 rs11556090.

Table 5.

Association of SMAD3 rs11556090 and RBL2 rs3929 with DFS of HCC patients

Genotypes Patients Events (%)a Univariate analysis Multivariate analysisb
HR (95%CI) P HR (95%CI) P
SMAD3 rs11556090
AA 162 103 (63.6) 1 1
AG 98 73 (74.5) 1.42 (1.05–1.92) 0.021 1.50 (1.09–2.06) 0.013
GG 11 9 (81.8) 1.45 (0.73–2.86) 0.289 2.00 (0.98–4.05) 0.056
Trend test 0.017 0.004
AA 162 103 (63.6) 1 1
AG+GG 109 82 (75.2) 1.32 (1.04–1.67) 0.023 1.54 (1.13–2.09) 0.006
RBL2 rs3929
GG 160 114 (71.3) 1 1
GC 95 67 (70.5) 0.91 (0.68–1.24) 0.554 0.94 (0.68–1.29) 0.696
CC 14 3 (21.4) 0.18 (0.06–0.57) 0.004 0.20 (0.06–0.63) 0.006
Trend test 0.009 0.023
CC 14 3 (21.4) 1 1
GG+GC 255 181 (71.0) 5.31 (1.7–16.6) 0.004 4.98 (1.57−15.83) 0.006
Combined analysisc
0 NUGs 12 3 (25.0) 1 1
1 NUG 151 99 (65.6) 3.79 (1.20−11.96) 0.023 3.53 (1.10−11.36) 0.034
2 NUGs 106 82 (77.4) 5.42 (1.71−17.17) 0.004 5.46 (1.70−17.58) 0.004
Trend test 0.0003 0.0001
Open in a new tab

DFS, disease‐free survival; HCC, hepatocellular carcinoma; HR, hazards ratio; CI, confidence interval; NUG, number of unfavorable genotypes.

a

Including relapse, metastasis and death.

b

Adjusted by age, sex, ethnicity, HBsAg status, smoking status, drinking status, intact capsule, hepatic vein tumor thrombosis, cirrhosis and BCLC staging.

c

Unfavorable genotypes were SMAD3 rs11556090 AG+GG and RBL2 rs3929 GG+GC.

Figure 1.

Figure 1

Open in a new tab

SMAD3 rs11556090 and RBL2 rs3929 associated with disease‐free survival of hepatocellular carcinoma patients. (A–D) Kaplan–Meier survival curves of single nucleotide polymorphisms in different genetic models: rs11556090 in an additive genetic model (A) and dominant model (B); rs3929 in an additive genetic model (C) and recessive model (D). (E) Kaplan–Meier survival curves of combined effects of the unfavorable genotypes. (F) Receiver operating characteristic (ROC) curve and area under the curve (AUC) estimation for prediction of 1‐year DFS.

Combined effects and ROC curves for DFS prediction

To evaluate the joint effect of SMAD3 rs11556090 and RBL2 rs3929, we combined risk genotypes of rs11556090 CC and rs3929GG+GC into a single variable as the number of unfavorable genotypes (NUGs) (Table 5). The trend test indicated that increased NUG was associated with a shorter DFS regardless of univariate (trend test: P = 0.0003) or multivariate analysis (trend test: P = 0.0001). The Kaplan–Meier curves for depicting associations between NUG and DFS is also shown in Figure 1E. Using the ROC curves, we further evaluated predictive value of the unfavorable genotypes. The results demonstrated that, as classification of 1‐year DFS, the AUC was significantly increased from 68.4% to 74.1% (P = 0.011), when adding NUG to the model including clinical variables (age, sex, ethnic, hepatitis B virus infection, smoking status, drinking status, capsule, portal vein tumor thrombosis, cirrhosis and BCLC staging) as classifiers (Fig. 1F).

In silico functional validation

As mapped on the UCSC website (https://genome.ucsc.edu/), rs11556090 is located in exon 9, 3ʹUTR of SMAD3, while rs3929 is located in exon 22, 3ʹUTR of RBL2 (Fig. 2A and B). Both of them were predicted to be miRNA binding sites of the corresponding genes by the SNPinfo. We further conducted the expression quantitative trait loci (eQTL) analysis by searching on the GTEx Portal (http://www.gtexportal.org/home/) and using data from the 1000 Genomes Project to examine the effects of SNPs on the mRNA expression levels of corresponding genes. We only found information about RBL2 on GTEx Portal that the rs3929 C allele was significantly associated with increased expression levels of RBL2 in liver tissues (P = 1.8 × 10−7, Fig. 2C) and the whole blood (P = 3.9 × 10−14, Fig. 2D) by general linear regression analysis. Data of Asian population (the Chinese and Japanese together) from the HapMap 3 Project showed no significant association for neither rs11556090 nor rs3929 with their corresponding gene expression levels (P = 0.486 and 0.686, respectively, Fig. 2E and F).

Figure 2.

Figure 2

Open in a new tab

In silico functional validation of SMAD3 rs11556090 and RBL2 rs3929. Location‐map of rs11556090 (A) and rs3929 (B) in UCSC website (https://genome.ucsc.edu/); The expression quantitative trait loci analysis (eQTL) from the GTEx Portal (http://www.gtexportal.org/home/) for RBL2 rs3929 in liver tissue (C) and the whole blood (D); eQTL analysis by using Asian population of HapMap3 data for rs11556090 (E) and rs3929 (F).

Discussion

In this hypothesis‐driven study, we identified SMAD3 rs11556090 A>G and RBL2 rs3929 G>C in the cell cycle pathway may independently or jointly modulate DFS of HCC patients. We also observed that combining with unfavorable genotypes of these loci significantly improved prediction performance of HCC survival in the model that included the known clinical prognostic variables. Furthermore, the results from the GTEx Portal demonstrated that the C allele was significantly associated with increasing expression levels of RBL2, suggesting that RBL2 rs3929 may modulate DFS of HCC patients possibly through a mechanism of modulating gene expression.

Cell cycle control is an important process for DNA damage repair. When DNA damage occurs, cell cycle checkpoints are activated and cell cycle progression is paused, allowing cells sufficient time to repair the damage and determining whether cells will survive or die, which is controlled by the cell cycle pathway players 21. Previous studies have shown that the identified genes, SMAD3 and RBL2, are important players in the cell cycle control. For example, SMAD3 plays a critical part in the regulation of transforming growth factor‐β (TGF‐β), which is a potent inhibitor of cell cycle progression at the G1 phase 37. Likewise, RBL2, encoding one of the retinoblastoma family proteins (RBL2, also known as p130 and pRb2), is a key switch at the restriction point (R) and inhibits the S phase entry by physical combination with E2F transcription factor and actively represses gene transcription that regulates DNA synthesis in the S phase 38. Other studies reported that SMAD3 functioned as a tumor suppressor in cancers of the stomach, 39, breast 40 and prostate 41, while RBL2 functions as a tumor suppressor in cancers of the breasts, ovaries and endometrium 42. A similar function in HCC has also been illustrated for both of SMAD3 and RBL2 in the previous studies 43, 44, suggesting the significant function of SMAD3 and RBL2 as a tumor suppressor in HCC.

In this study, both of the two loci we identified were computationally predicted to be located in the microRNA‐binding sites by the SNPinfo. SMAD3 rs11556090 is located in the binding site of hsa‐miR‐132, hsa‐miR‐188, hsa‐miR‐212, hsa‐miR‐30b, hsa‐miR‐337, hsa‐miR‐431, hsa‐miR‐532 and hsa‐miR‐9, while RBL2 rs3929 was located only in has‐miR‐134. Surprisingly, the above‐mentioned miRNAs, except for has‐miR‐337 and has‐miR‐532, have been associated with development or survival of HCC 45, 46, 47, 48, 49, 50, 51. Therefore, it is likely that the underlying mechanisms of the two SNPs in modulating HCC survival are to regulate the transcription of proteins by affecting activities of miRNA‐binding. SNPs in the potential miRNA‐binding sites are suggested to be functional and always good candidates for causal variants of human disease occurrence or progress 52. Specifically, we found that the RBL2 rs3929 C allele associated with a longer DFS was correlated with increased mRNA expression levels of RBL2 in normal liver tissues and peripheral blood lymphocytes by using data from the GTEx Portal. These results are consistent with the characteristics of a tumor suppressor, although the majority of the donors for the GTEx project are of the white population. However, somatic mutations in RBL2 were observed previously only in primary nasopharyngeal carcinomas, lung tumors, and Burkitt's lymphomas but not in HCC. Taken together, we can infer that RBL2 rs3929 may be a functional and important variant for HCC progression.

The results also demonstrated a significant locus‐dosage effect in patients with a larger number of unfavorable genotypes, who had a shorter DFS, suggesting an interaction of SMAD3 rs11556090 and RBL2 rs3929 in the cell cycle pathway on HCC progress. The potential mechanism under the observed association is probably that both of SMAD3 and RBL2 are the key substrates of CDK4, which can control cell cycle progression from the G1 to S phase by phosphorylation of these proteins 37, 53. Importantly, The combination of the two loci significantly improved prediction performance of the model that included the known prognosis factors of HCC patients.

It should be noted that there were limitations of this study. The first limitation included restriction of the recruit population. We conducted the study only in the population from Guangxi Zhuang Autonomous Region, where the main cause of HCC is HBV infection 28. The results may not apply to other ethnic groups or disease etiologies, such as hepatitis C virus infection, nonalcoholic fatty liver disease and alcoholic liver disease. Secondly, the significance of RBL2 rs3929 and SMAD3 rs11556090 in the prognosis of HCC has been demonstrated in this study, however, it still lack epidemiology validation and biological mechanism confirmation, besides the significant eQTL results for RBL3 rs3929 from the GTEx project. Finally, because of lack of more detailed clinical information, we did not efficiently evaluated the potential effects of different therapies after radical surgery resection on the outcomes of HCC patients, or their potential associations with the identified SNPs. Therefore, larger scale, more comprehensive and multi‐institutional epidemiology investigations and biological studies are warranted to further validate these results.

In conclusion, this study identified a prognostic role of SMAD3 rs11556090 and RBL2 rs3929 of the cell cycle pathway in HCC patients. Although both of SMAD3 and RBL2 are important in the control of cell cycle progression from the G1 to S phase, the interpretation of our findings should be cautious, until validated by mort patient cohorts and HCC cell lines.

Conflict of Interest

The authors declare that they have no conflict of interest.

Acknowledgements

We thank the individuals who participated in this project. We would like to thank the staff members of the First Affiliated Hospital and Affiliated Tumor Hospital of Guangxi Medical University for their support in recruiting the study participants as well as Da‐Fang Chen and his research team at the Peking University Health Science Center for technical help. This work was funded in part by National Natural Science Foundation of China, No. 81360448, No. 81402756, No. 81660563; Natural Science Foundation of Guangxi, No. 2014GXNSFAA118139; University Science and Technology Research Project of Guangxi, No. YB2014067. Shun Liu was supported by the Innovation Project of Guangxi Graduate Education (2015).

Cancer Medicine 2017; 6(7):1512–1522

Contributor Information

Xiao‐Yun Zeng, Email: zxyxjw@21cn.com.

Xiao‐Qiang Qiu, Email: xqqiu9999@sina.com.

References

  • 1. Chen, W. , Zheng R., Baade P. D., Zhang S., Zeng H., Bray F., et al. 2016. Cancer statistics in China, 2015. CA Cancer J. Clin. 66:115–132. [DOI] [PubMed] [Google Scholar]
  • 2. Torre, L. A. , Bray F., Siegel R. L., Ferlay J., Lortet‐Tieulent J., and Jemal A.. 2015. Global cancer statistics, 2012. CA Cancer J. Clin. 65:87–108. [DOI] [PubMed] [Google Scholar]
  • 3. Farinati, F. , Rinaldi M., Gianni S., and Naccarato R.. 2000. How should patients with hepatocellular carcinoma be staged? Validation of a new prognostic system. Cancer 89:2266–2273. [PubMed] [Google Scholar]
  • 4. Llovet, J. M. , Bru C., and Bruix J.. 1999. Prognosis of hepatocellular carcinoma: the BCLC staging classification. Semin. Liver Dis. 19:329–338. [DOI] [PubMed] [Google Scholar]
  • 5. Chatterjee, R. , and Mitra A.. 2015. An overview of effective therapies and recent advances in biomarkers for chronic liver diseases and associated liver cancer. Int. Immunopharmacol. 24:335–345. [DOI] [PubMed] [Google Scholar]
  • 6. Pons, F. , Varela M., and Llovet J. M.. 2005. Staging systems in hepatocellular carcinoma. HPB (Oxford) 7:35–41. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 7. Farinati, F. , Vitale A., Spolverato G., Pawlik T. M., Huo T. L., Lee Y. H., et al. 2016. Development and validation of a new prognostic system for patients with hepatocellular carcinoma. PLoS Med. 13:e1002006. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 8. Jiang, D. K. , Sun J., Cao G., Liu Y., Lin D., Gao Y. Z., et al. 2013. Genetic variants in STAT4 and HLA‐DQ genes confer risk of hepatitis B virus‐related hepatocellular carcinoma. Nat. Genet. 45:72–75. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 9. Zhang, H. , Zhai Y., Hu Z., Wu C., Qian J., Jia W., et al. 2010. Genome‐wide association study identifies 1p36.22 as a new susceptibility locus for hepatocellular carcinoma in chronic hepatitis B virus carriers. Nat. Genet. 42:755–758. [DOI] [PubMed] [Google Scholar]
  • 10. Qu, L. S. , Jin F., Guo Y. M., Liu T. T., Xue R. Y., Huang X. W., et al. 2016. Nine susceptibility loci for hepatitis B virus‐related hepatocellular carcinoma identified by a pilot two‐stage genome‐wide association study. Oncol. Lett. 11:624–632. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 11. Li, S. , Qian J., Yang Y., Zhao W., Dai J., Bei J. X., et al. 2012. GWAS identifies novel susceptibility loci on 6p21.32 and 21q21.3 for hepatocellular carcinoma in chronic hepatitis B virus carriers. PLoS Genet. 8:e1002791. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 12. Tarhuni, A. , Guyot E., Rufat P., Sutton A., Bourcier V., Grando V., et al. 2014. Impact of cytokine gene variants on the prediction and prognosis of hepatocellular carcinoma in patients with cirrhosis. J. Hepatol. 61:342–350. [DOI] [PubMed] [Google Scholar]
  • 13. Pan, D. , Zeng X., Yu H., Yang Y., Bei C., Bai H., et al. 2014. Role of cytokine gene polymorphisms on prognosis in hepatocellular carcinoma after radical surgery resection. Gene 544:32–40. [DOI] [PubMed] [Google Scholar]
  • 14. Jung, S. W. , Park N. H., Shin J. W., Park B. R., Kim C. J., Lee J. E., et al. 2014. Prognostic impact of telomere maintenance gene polymorphisms on hepatocellular carcinoma patients with chronic hepatitis B. Hepatology 59:1912–1920. [DOI] [PubMed] [Google Scholar]
  • 15. Long, X. D. , Yao J. G., Zeng Z., Ma Y., Huang X. Y., Wei Z. H., et al. 2013. Polymorphisms in the coding region of X‐ray repair complementing group 4 and aflatoxin B1‐related hepatocellular carcinoma. Hepatology 58:171–181. [DOI] [PubMed] [Google Scholar]
  • 16. Kim, S. S. , Cho H. J., Lee H. Y., Park J. H., Noh C. K., Shin S. J., et al. 2016. Genetic polymorphisms in the Wnt/beta‐catenin pathway genes as predictors of tumor development and survival in patients with hepatitis B virus‐associated hepatocellular carcinoma. Clin. Biochem. 49:792–801. [DOI] [PubMed] [Google Scholar]
  • 17. Vermeulen, K. , Van Bockstaele D. R., and Berneman Z. N.. 2003. The cell cycle: a review of regulation, deregulation and therapeutic targets in cancer. Cell Prolif. 36:131–149. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 18. Maddika, S. , Ande S. R., Panigrahi S., Paranjothy T., Weglarczyk K., Zuse A., et al. 2007. Cell survival, cell death and cell cycle pathways are interconnected: implications for cancer therapy. Drug Resist. Update 10:13–29. [DOI] [PubMed] [Google Scholar]
  • 19. Bretones, G. , Delgado M. D., and Leon J.. 2015. Myc and cell cycle control. Biochim. Biophys. Acta 1849:506–516. [DOI] [PubMed] [Google Scholar]
  • 20. Kohrman, A. Q. , and Matus D. Q.. 2017. Divide or conquer: cell cycle regulation of invasive behavior. Trends Cell Biol. 27:12–25. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 21. Park, M. T. , and Lee S. J.. 2003. Cell cycle and cancer. J. Biochem. Mol. Biol. 36:60–65. [DOI] [PubMed] [Google Scholar]
  • 22. Kommoss, S. , du Bois A., Ridder R., Trunk M. J., Schmidt D., Pfisterer J., et al. 2007. Independent prognostic significance of cell cycle regulator proteins p16(INK4a) and pRb in advanced‐stage ovarian carcinoma including optimally debulked patients: a translational research subprotocol of a randomised study of the Arbeitsgemeinschaft Gynaekologische Onkologie Ovarian Cancer Study Group. Br. J. Cancer 96:306–313. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 23. Mineta, H. , Miura K., Suzuki I., Takebayashi S., Amano H., Araki K., et al. 1999. Low p27 expression correlates with poor prognosis for patients with oral tongue squamous cell carcinoma. Cancer 85:1011–1017. [DOI] [PubMed] [Google Scholar]
  • 24. Chao, Y. , Shih Y. L., Chiu J. H., Chau G. Y., Lui W. Y., Yang W. K., et al. 1998. Overexpression of cyclin A but not Skp 2 correlates with the tumor relapse of human hepatocellular carcinoma. Cancer Res. 58:985–990. [PubMed] [Google Scholar]
  • 25. Ito, Y. , Takeda T., Sakon M., Monden M., Tsujimoto M., and Matsuura N.. 2000. Expression and prognostic role of cyclin‐dependent kinase 1 (cdc2) in hepatocellular carcinoma. Oncology 59:68–74. [DOI] [PubMed] [Google Scholar]
  • 26. Qin, L. F. , and Ng I. O.. 2001. Expression of p27(KIP1) and p21(WAF1/CIP1) in primary hepatocellular carcinoma: clinicopathologic correlation and survival analysis. Hum. Pathol. 32:778–784. [DOI] [PubMed] [Google Scholar]
  • 27. Peng, S. Y. , Chou S. P., and Hsu H. C.. 1998. Association of downregulation of cyclin D1 and of overexpression of cyclin E with p53 mutation, high tumor grade and poor prognosis in hepatocellular carcinoma. J. Hepatol. 29:281–289. [DOI] [PubMed] [Google Scholar]
  • 28. Nan, Y. L. , Hu Y. L., Liu Z. K., Duan F. F., Xu Y., Li S., et al. 2016. Relationships between cell cycle pathway gene polymorphisms and risk of hepatocellular carcinoma. World J. Gastroenterol. 22:5558–5567. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 29. Murali, A. , Varghese B. T., Kumar R. R., and Kannan S.. 2016. Combination of genetic variants in cyclin D1 and retinoblastoma genes predict clinical outcome in oral cancer patients. Tumour Biol. 37:3609–3617. [DOI] [PubMed] [Google Scholar]
  • 30. Yin, J. , Lu C., Gu J., Lippman S. M., Hildebrandt M. A., Lin J., et al. 2011. Common genetic variants in cell cycle pathway are associated with survival in stage III‐IV non‐small‐cell lung cancer. Carcinogenesis 32:1867–1871. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 31. Song, H. , Hogdall E., Ramus S. J., Dicioccio R. A., Hogdall C., Quaye L., et al. 2008. Effects of common germ‐line genetic variation in cell cycle genes on ovarian cancer survival. Clin. Cancer Res. 14:1090–1095. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 32. Xu, Z. , and Taylor J.A.. 2009. SNPinfo: integrating GWAS and candidate gene information into functional SNP selection for genetic association studies. Nucleic Acids Res. 37(Web Server Issue):W600–W605. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 33. Wacholder, S. , Chanock S., Garcia‐Closas M., El Ghormli L., and Rothman N.. 2004. Assessing the probability that a positive report is false: an approach for molecular epidemiology studies. J. Natl Cancer Inst. 96:434–442. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 34. DeLong, E. R. , DeLong D. M., and Clarke‐Pearson D. L.. 1988. Comparing the areas under two or more correlated receiver operating characteristic curves: a nonparametric approach. Biometrics 44:837–845. [PubMed] [Google Scholar]
  • 35. Consortium GT . 2013. The genotype‐tissue expression (GTEx) project. Nat. Genet. 45:580–585. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 36. Stranger, B. E. , Montgomery S. B., Dimas A. S., Parts L., Stegle O., Ingle C. E., et al. 2012. Patterns of cis regulatory variation in diverse human populations. PLoS Genet. 8:e1002639. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 37. Matsuura, I. , Denissova N. G., Wang G., He D., Long J., and Liu F.. 2004. Cyclin‐dependent kinases regulate the antiproliferative function of Smads. Nature 430:226–231. [DOI] [PubMed] [Google Scholar]
  • 38. Nevins, J. R. 1992. E2F: a link between the Rb tumor suppressor protein and viral oncoproteins. Science 258:424–429. [DOI] [PubMed] [Google Scholar]
  • 39. Han, S. U. , Kim H. T., Seong D. H., Kim Y. S., Park Y. S., Bang Y. J., et al. 2004. Loss of the Smad3 expression increases susceptibility to tumorigenicity in human gastric cancer. Oncogene 23:1333–1341. [DOI] [PubMed] [Google Scholar]
  • 40. Walker, L. C. , Fredericksen Z. S., Wang X., Tarrell R., Pankratz V. S., Lindor N. M., et al. 2010. Evidence for SMAD3 as a modifier of breast cancer risk in BRCA2 mutation carriers. Breast Cancer Res. 12:R102. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 41. Lu, S. , Lee J., Revelo M., Wang X., Lu S., and Dong Z.. 2007. Smad3 is overexpressed in advanced human prostate cancer and necessary for progressive growth of prostate cancer cells in nude mice. Clin. Cancer Res. 13:5692–5702. [DOI] [PubMed] [Google Scholar]
  • 42. Scambia, G. , Lovergine S., and Masciullo V.. 2006. RB family members as predictive and prognostic factors in human cancer. Oncogene 25:5302–5308. [DOI] [PubMed] [Google Scholar]
  • 43. Huynh, H. 2004. Overexpression of tumour suppressor retinoblastoma 2 protein (pRb2/p130) in hepatocellular carcinoma. Carcinogenesis 25:1485–1494. [DOI] [PubMed] [Google Scholar]
  • 44. Yang, Y. A. , Zhang G. M., Feigenbaum L., and Zhang Y. E.. 2006. Smad3 reduces susceptibility to hepatocarcinoma by sensitizing hepatocytes to apoptosis through downregulation of Bcl‐2. Cancer Cell 9:445–457. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 45. Liu, K. , Li X., Cao Y., Ge Y., Wang J., and Shi B.. 2015. MiR‐132 inhibits cell proliferation, invasion and migration of hepatocellular carcinoma by targeting PIK3R3. Int. J. Oncol. 47:1585–1593. [DOI] [PubMed] [Google Scholar]
  • 46. Fang, F. , Chang R. M., Yu L., Lei X., Xiao S., Yang H., et al. 2015. MicroRNA‐188‐5p suppresses tumor cell proliferation and metastasis by directly targeting FGF5 in hepatocellular carcinoma. J. Hepatol. 63:874–885. [DOI] [PubMed] [Google Scholar]
  • 47. Dou, C. , Wang Y., Li C., Liu Z., Jia Y., Li Q., et al. 2015. MicroRNA‐212 suppresses tumor growth of human hepatocellular carcinoma by targeting FOXA1. Oncotarget 6:13216–13228. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 48. Liu, Z. , Wei X., Zhang A., Li C., Bai J., and Dong J.. 2016. Long non‐coding RNA HNF1A‐AS1 functioned as an oncogene and autophagy promoter in hepatocellular carcinoma through sponging hsa‐miR‐30b‐5p. Biochem. Biophys. Res. Commun. 473:1268–1275. [DOI] [PubMed] [Google Scholar]
  • 49. Pan, L. , Ren F., Rong M., Dang Y., Luo Y., Luo D., et al. 2015. Correlation between down‐expression of miR‐431 and clinicopathological significance in HCC tissues. Clin. Transl. Oncol. 17:557–563. [DOI] [PubMed] [Google Scholar]
  • 50. Zhang, J. , Cheng J., Zeng Z., Wang Y., Li X., Xie Q., et al. 2015. Comprehensive profiling of novel microRNA‐9 targets and a tumor suppressor role of microRNA‐9 via targeting IGF2BP1 in hepatocellular carcinoma. Oncotarget 6:42040–42052. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 51. Zha, R. , Guo W., Zhang Z., Qiu Z., Wang Q., Ding J., et al. 2014. Genome‐wide screening identified that miR‐134 acts as a metastasis suppressor by targeting integrin beta1 in hepatocellular carcinoma. PLoS ONE 9:e87665. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 52. Zhang, W. , Liu H., Yin J., Wu W., Zhu D., Amos C. I., et al. 2016. Genetic variants in the PIWI‐piRNA pathway gene DCP1A predict melanoma disease‐specific survival. Int. J. Cancer 139:2730–2737. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 53. Sherr, C. J. , and Roberts J. M.. 1999. CDK inhibitors: positive and negative regulators of G1‐phase progression. Genes Dev. 13:1501–1512. [DOI] [PubMed] [Google Scholar]

Articles from Cancer Medicine are provided here courtesy of Wiley

RESOURCES