PRKRA expression is upregulated in HBV-related HCC

As shown in Fig. 1A, B, the expression of PRKRA was significantly upregulated in HCC tumor tissues, especially in the HBV-positive group (p < 0.001). Moreover, the expression of PRKRA was higher in the advanced TNM-stage group than the early TNM-stage group (p < 0.001) (Fig. 1C). We then compared the mRNA expression levels of PRKRA in tumor tissues and tumor-adjacent tissues from GSE19665 dataset (n = 5), and these results confirmed the increase expression of PRKRA in HCC (p < 0.001) (Fig. 1D).

Fig. 1

PRKRA is upregulated in HBV-related HCC. A, B PRKRA mRNA expression in HCC tumor tissues (HBV-positive group, n = 145; HBV-negative group, n = 226) and tumor-adjacent tissues (n = 50). Data were derived from TCGA datasets. C The PRKRA mRNA expression levels in controls and at different TNM stages of HBV-related HCC patients (control, n = 50; stage I, n = 44; stage II, n = 41; stage III + IV, n = 53). D PRKRA mRNA expression in matched tumor and tumor-adjacent tissues from 5 patients. Data were derived from GSE19665 dataset

Higher PRKRA expression levels predict a poor prognosis in HBV-related HCC

Next, we analyzed the association between PRKRA expression and prognosis of HBV-related HCC patients. The patients were divided into PRKRA-low and PRKRA-high groups according to the expression levels of PRKRA in tumors. The Kaplan–Meier analysis indicated that patients in the PRKRA-high group had a poor overall survival (Fig. 2A; log-rank p value < 0.001) and disease-free survival (Fig. 2B; log-rank p value < 0.001) than patients in the PRKRA-low group. Then univariate analysis and multivariate Cox regression models including six parameters (PRKRA expression, AFP levels, sex, age, TNM-stage and pathologic grade) were performed. The results showed that PRKRA expression was correlated with patients' clinical outcomes and was an independent risk factor for HBV-related HCC patients (Hazard ratio = 2.208, 95% CI 1.476–3.304, p < 0.001). Collectively, these results indicate that PRKRA could be a prognosis predictor for HBV-related HCC.

Fig. 2

High PRKRA expression levels predict poor prognosis of HBV-related HCC patients. A, B Kaplan–Meier analysis of overall survival (A) and disease-free survival (B) of HBV-related HCC patients based on PRKRA expression (n = 145). The median of PRKRA expression levels was used as the cutoff value for grouping C, D Univariate analysis (C) and multivariate COX regression models (D) including 6 parameters (PRKRA expression, serum AFP level, sex, age, TNM-stage and pathologic grade) were employed to explore risk factors for HBV-related HCC. Symbols indicate Hazard ratio, and bars indicate 95% CIs in forest plots. Uni: Univariate; Mut: multivariate; CI: confidence interval

The diagnostic performance of PRKRA in peripheral blood as a biomarker for HBV-related HCC

The increased expression of PRKRA in HCC, especially in HBV-related HCC was further confirmed in 152 blood samples from HCC patients and healthy controls (Fig. 3A and Additional file 1: Fig. S1). Then the AUROC was plotted to test whether the PRKRA expression in peripheral blood had diagnostic capacity for HBV-related HCC. As shown in Fig. 3B, PRKRA expression in peripheral blood could distinguish HCC patients from healthy controls with AUROC of 0.713 (95% CI 0.621–0.805; p < 0.001). PRKRA expression combining with serum AFP and CEA showed a much higher AUROC (PRKRA + AFP: 0.880, 95% CI 0.806–0.952; p < 0.001; PRKRA + AFP + CEA: 0.908, 95% CI 0.844–0.972; p < 0.001). Notably, the diagnostic sensitivity of PRKRA expression was 54% (cutoff value = 2.341), while PRKRA expression, serum AFP and CEA served as a combined diagnostic indicator for HBV-related HCC could increase sensitivity to 76%. Besides, PRKRA expression values ≥ 2.341 showed a better diagnostic value (PRKRA: AUROC = 0.952, 95% CI 0.914–0.990; p < 0.001; PRKRA + AFP: AUROC = 0.962, 95% CI 0.920–1.00; p < 0.001; PRKRA + AFP + CEA: AUROC = 0.973, 95% CI 0.934–1.00; p < 0.001). These findings revealed that PRKRA expression, serum AFP and CEA could act as a combined diagnostic indicator for HBV-related HCC.

Fig. 3

PRKRA expression in peripheral blood has the potential diagnostic capability for HBV-related HCC. A PRKRA mRNA expression in the validation cohort including 152 blood samples from 77 healthy controls and 75 HCC patients, 60 of which were infected with HBV. B AUROCs for PRKRA, PRKRA + AFP, PRKRA + AFP + CEA in HBV-related patients and healthy controls. C According to the cut-off value (2.341), patients were divided into PRKRA-high and PRKRA-low groups. AUROCs for PRKRA, PRKRA + AFP, PRKRA + AFP + CEA in PRKRA-high group. D In HBV-related patients with AFP ≤ 200 ng/mL and healthy controls, AUROCs for PRKRA, PRKRA + CEA, PRKRA + AFP + CEA

Next, we paid attention to the diagnostic potential of PRKRA in patients with AFP ≤ 200 ng/mL. PRKRA expression levels in peripheral blood (AUROC = 0.746, 95% CI 0.633–0.858; p < 0.001) indicated a better diagnostic capability than serum AFP (AUROC = 0.626, 95% CI 0.477–0.775; p = 0.095) and CEA (AUROC = 0.682, 95% CI 0.558–0.806; p = 0.010) (Fig. 3C). In addition, a combination of PRKRA expression, AFP and CEA could improve the diagnostic capability when serum AFP was at a low level in HBV-related HCC patients. These data suggested that the PRKRA expression in peripheral blood provided the potential diagnostic capability for HBV-related HCC patients.

PRKRA expression levels are associated with EIF2AK2 and inflammatory cytokine genes

To explore the possible reasons for increased PRKRA expression and poor prognosis in HBV-related HCC, we compared the mRNA expression of EIF2AK2 in tumor tissues and blood samples from the HBV-related HCC patients. Similar to the expression pattern of PRKRA, EIF2AK2 was also upregulated both in tumor tissues and peripheral blood samples (Additional file 1: Fig. S2). Increased EIF2AK2 expression levels were also associated with the poor prognosis of HBV-related HCC patients (Additional file 1: Fig. S3).

Pearson correlation analysis was then used to investigate the link between PRKRA, EIF2AK2 and inflammatory cytokine genes (i.e., IL-2, IL-4, IL-6, IL-10, IL-17, IL-22, TNF-α, TGF-α) through a TCGA database including 145 HBV-positive patients. A significant positive correlation between PRKRA expression and EIF2AK2 expression was observed in tissues (r = 0.658, p < 0.001; Fig. 4A) and peripheral blood samples (r = 0.462, p < 0.001; Fig. 4B). Moreover, PRKRA and EIF2AK2 were both positively correlated to the expression of inflammatory cytokine genes, including IL-2 (r = 0.227, p = 0.01; r = 0.241, p < 0.001), IL-4 (r = 0.247, p < 0.001; r = 0.347, p < 0.001), IL-10 (r = 0.227, p = 0.01; r = 0.351, p < 0.001), TNF-α (r = 0.305, p < 0.001; r = 0.446, p < 0.001) and TGF-α (r = 0.389, p < 0.001; r = 0.441, p < 0.001) (Fig. 4 C, D). These results indicated that PRKRA / EIF2AK2 could lead to poor clinical outcomes through activating inflammatory response (Fig. 4F).

Fig. 4

PRKRA expression levels are associated with EIF2AK2 and inflammatory cytokines. A, B The positive correlation between PRKRA expression levels and EIF2AK2 expression levels in tissues (A) and peripheral blood samples (B). C, D Heatmaps indicating the associations across PRKRA, EIF2AK2 and inflammatory cytokines genes. Pearson r and p values shown in C, D, respectively. Data were derived from TCGA database. E A schematic illustration of PRKRA upregulation was associated with EIF2AK2 and inflammation, which explain the possible reasons for poor prognosis in HBV-related HCC