Distribution of patient characteristics and prognosis analyses
All patient characteristics and clinical data are summarized in Table 1. The patients included 334 men (90.8%) and 34 women (9.2%) with a median age of 50 years (range: 22–82 years). Among this cohort, 62.5% (230) were non-drinkers and 58.4% (215) were non-smokers, while 84.0% (309) were hepatitis B surface antigen (HBsAg) positive. The patient cohort comprised 238 (64.7%) patients with a tumor size of ≥ 5 cm, 77 (20.9%) with multiple tumor nodules (≥ 2), 105 (28.5%) with cancer embolus, and 151 (41.0%) with significantly increased serum alpha-fetoprotein (AFP) levels (≥ 400 ng/mL). The percentage of patients with Barcelona clinic liver cancer (BCLC) stage A/B and stage C was 55.4% (204) and 44.6% (164), respectively.
To determine whether there were any confounding factors influencing patient deaths or survival time, we performed a Cox proportional hazards regression analysis to assess whether there were any associations between OS and clinicopathological characteristics. Multivariate analyses showed that large tumor size (≥ 5 cm) (adjusted HR = 1.83, 95% CI = 1.32–2.53; P < 0.001), advanced BCLC stage C (adjusted HR = 2.17, 95% CI = 1.59–2.96; P < 0.001), and cancer embolus (adjusted HR = 1.54, 95% CI = 1.13–2.11; P = 0.006) predicted poor overall survival of patients with HCC. There was no association between HCC outcome and age (P = 0.470), gender (P = 0.671), smoking status (P = 0.702), drinking status (P = 0.226), HBV infection status (P = 0.801), number of tumor nodules (P = 0.125), or AFP levels (P = 0.378) (Table 1).
Association between RAD51 rs12593359 and clinical outcome in HCC patients
The effect of RAD51 rs12593359 on HCC prognosis is shown in Table 2. We observed that the dominant genetic model of the SNP rs12593359 had a significant correlation with HCC prognosis. Compared to the rs12593359 GG genotype, the GT/TT genotype conferred a significant increase in the risk of death in multivariate Cox proportional hazards model adjusted for all variables (adjusted HR = 1.34, 95% CI = 1.02–1.76; P = 0.035). Kaplan-Meier curve analysis showed a markedly shorter survival time in HCC patients with GT/TT genotypes of SNP rs12593359 compared with that of patients carrying a GG genotype (19.0 months vs. 36.0 months; P log-rank = 0.012) (Figure 1a).
Stepwise regression analysis of HCC prognosis
To determine the independent prognostic factors associated with HCC OS, we performed a stepwise multivariate Cox proportional hazards regression analysis using covariates listed in Table 1 and the RAD51 SNP rs12593359. The results demonstrated that the survival of patients with HCC had a relationship with BCLC stage (A/B vs. C), tumor size (< 5 cm vs. ≥ 5 cm), cancer embolus (no vs. yes), and rs12593359 genotypes (GG vs. TT/TG) (Table 3).
Stratified analysis of association between SNP rs12593359 with OS by HCC patient characteristics
We further analyzed the effect of the SNP rs12593359 on OS in HCC patients stratified by clinical characteristics. As shown in Table 4, the significant increase in risk of death conferred by the SNP rs12593359 was observed in males (adjusted HR = 1.40, 95% CI = 1.05–1.87; P = 0.021), but not in females (P = 0.707); in HBV-positive patients (adjusted HR = 1.37, 95% CI = 1.01–1.85; P = 0.042), but not in HBV-negative patients (P = 0.901); in those with a single tumor nodule (adjusted HR = 1.51, 95% CI = 1.11–2.06; P = 0.009), but not in those with multiple tumor nodules (P = 0.940); in those with AFP levels < 400ng/mL (adjusted HR = 1.58, 95% CI = 1.10–2.29; P = 0.014), but not in those with levels ≥ 400ng/mL (P = 0.451); and in patients who were cancer embolus-free (adjusted HR = 1.55, 95% CI = 1.10–2.20; P = 0.014), but not in those with a cancer embolus (P = 0.461). No significant association between SNP rs12593359 and HCC OS existed in other subgroups (P > 0.05).
As indicated in Figure 1b-f, log-rank tests indicated that there was a significant difference in overall survival between patients with TT/TG genotypes and those with the GG genotype of SNP rs12593359 in the above-mentioned subgroups.
Correlation between rs12593359 genotypes and the mRNA expression level of RAD51
We performed the eQTL analysis to correlate the rs12593359 genotypes and mRNA expression levels of RAD51 using data of the 483 cell-cultured fibroblasts from the GTEx database. The rs12593359 GG genotype was significantly associated with lower expression levels of RAD51 mRNA than the TT genotype, but not TG genotype (P = 1.1×10-4, Figure 2).
Association of RAD51 mRNA expression levels with survival of patients with HCC
By using the GEPIA database, we evaluated mRNA expressions levels of RAD51 in 371 HCC and 160 normal tissue samples. HCC tissues exhibited a significantly higher expression of RAD51 as compared to that of the normal tissues (P < 0.001, Figure 3), and patients with higher RAD51 mRNA expression in tumor tissues showed a poorer overall survival (HR=1.6, Plog-rank= 0.0054, Figure 4). Furthermore, we also found that a higher expression level of RAD51 was associated with poorer survival in 364 liver cancer patients from the online tool Kaplan-Meier plotter (P = 0.00097, Figure 2S).