Effectiveness of PIVKA-II in the detection of hepatocellular carcinoma based on real-world clinical data

Protein Induced by Vitamin K Absence or Antagonist-II (PIVKA-II) is an efficient biomarker specific for hepatocellular carcinoma (HCC). Some researchers have proved that levels of PIVKA-II reflect HCC oncogenesis and progression. However, the effectiveness of PIVKA-II based on real-world clnical data has barely been studied. A total of 14,861 samples were tested in Southwest Hospital in over 2 years’ time. Among them, 4073 samples were PIVKA-II positive. Finally, a total of 2070 patients with at least two image examinations were enrolled in this study. Levels of AFP and PIVKA-II were measured by chemiluminescence enzyme immunoassay (CLEIA) and chemiluminescent microparticle Immunoassay (CMIA), respectively. A total of 1016 patients with HCC were detected by PIVKA-II in a real-world application. In all these cases, 88.7% cases primarily occurred and patients with advanced HCC covered 61.3%. Levels of PIVKA-II were significantly higher in advanced group (4650.0 mAU/ml, 667.0–33,438.0 mAU/ml) than early-stage group (104.5 mAU/ml, 61.0–348.8 mAU/ml; P < 0.001). Levels of PIVKA-II elevated significantly in recurrence and residual group than recovery group (P < 0.001). A total of 1054 PIVKA-II positive patients were non-HCC cases. Among them, cirrhosis took the largest part (46.3%), followed by hepatitis (20.6%) and benign nodules (15.3%). High-levels of PIVKA-II in at-risk patients is an indicator of HCC development in two-year time. Our data showed that PIVKA-II effectively increases the detection rate of HCC was a valid complement to AFP and image examination in HCC surveillance.


Background
Recent years have witnessed a huge decrease in cancer mortality rate due to the progression of cancer treatment [1][2][3], especially with the development of next-generation sequencing, immune therapy and targeted drugs [4][5][6]. However, things are different in the area of hepatocellular carcinoma (HCC). Due to the inadequate approaches of early detection, around 50% of HCC cases were diagnosed at late stage when the 5-year overall survival rate is lower than 10% [7]. Chronic hepatitis B virus (HBV) infection ranks the major cause of HCC in Asia and sub-Saharan Africa [8,9]. Researchers have proven that antiviral treatment reduces the risk of HCC [10][11][12]. However, eliminating the risk of HCC in chronic hepatitis B (CHB) patients has a long way to go. Under this circumstance, there is a strong need for a feasible surveillance strategy for at-risk populations to increase the early detection rate of HCC.
Protein Induced by Vitamin K Absence or Antagonist-II (PIVKA-II), also known as Des-γ -carboxy-prothrombin (DCP), is believed to be a suitable serum biomarker specific for HCC since first detected by Libert et al. at 1984 [13]. With the development of accurate measuring methods [14,15], PIVKA-II has been recommended as one of a surveillance method for HCC in at-risk populations and written into the guidelines of the Japan Society of Hepatology (JSH) [16,17].
Clinical researches have revealed that alphafetoprotein(AFP) combined with PIVKA-II elevated the detection rate of about 8-20% with a satisfactory sensitivity and specificity [18][19][20][21]. As for HCC prognosis, treatment response and recurrence monitoring, PIVKA-II could also improve the performance [22][23][24]. However, all these studies were designed in reasonable ways with cases and controls limited to certain groups of people. In real-world settings, different people with different conditions and backgrounds may have great influence on the levels of PIVKA-II. However, the effectiveness of PIVKA-II in detecting HCC based on realworld clinical data has barely been studied. Finally, a total of 2070 patients with at least two image examinations or biopsy were enrolled in this study for cross-sectional analysis, of which 1016 patients (covered 49.1% of all PIVKA-II+ patients) were HCC cases and another 1054 PIVKA-II positive patients were non-HCC cases. For survival analysis, patients with more than 1 years and 3 times of follow-ups were recruited and 252 patients met the criterion and were enrolled.

Study populations
The diagnosis of each case was ascertained by image tests and a few of them were undertaken further pathological examinations. The diagnosis of HCC was determined by at least two enhanced image examinations, enhanced computed tomography (CT)/enhanced magnetic resonance imaging (MRI)/ultrasonography (USG), or by pathological confirmation. For cross-sectional analysis, PIVKA-II levels in HCC group were selected at the time of image diagnosis, while in the non-HCC group, PIVKA-II levels of the last result were selected for analysis. For survival analysis, PIVKA-II levels at baseline point and all follow-up points were analyzed. All clinical data were grabbed from electronic medical record system of Southwest Hospital.

Measurement of PIVKA-II and AFP
Serum levels of PIVKA-II were determined by chemiluminescence enzyme immunoassay (CLEIA) (LUMIPULSE® G1200, FUJIREBIO INC., Japan). The cut-off value was 40 mAU/ml. Serum levels of AFP were measured by AFP Reagent kit via chemiluminescent microparticle immunoassay (CMIA) ARCHITECT i2000, Abbott Laboratories, America). The cut-off value was set at 20 ng/ml. were applied for all statistical analysis and the graphs were constructed on the Prism version 6.00 (GraphPad Software Inc., USA). Each variable was represented as median with interquartile range. For cross-sectional analysis, normality and homogeneity of all data were evaluated by Kolmogorov-Smirnov test. Student T-test or Mann-Whitney test was applied to compare the differences between two categorical variables and for multi-categorical variables, Fig. 1 Flow diagram of the selection procedure. A cross-sectional study was conducted in PIVKA-II (+) patients with pathological or imaging confirmation. Survival analysis was conducted based on confirmed populations with follow-up one-way ANOVA or Kruskal-Wallis test was used. Sensitivity, specificity, Kappa value and diagnostic accuracy were calculated by 2 × 2 table in SPSS. Pearson Chi-square test was employed to evaluate statistical differences of diagnostic performance at different cut-off values. Receiver Operating Characteristics (ROC) Curves and area under ROC (AUROC) were calculated to evaluate the detecting efficiency of PIVKA-II, and DeLong test was applied to compare the different AUROC. For survival analysis, the cumulative incidence of HCC by patient groups with different levels of PIVKA-II was assessed with Kaplan-Meier analyses, and crude differences were calculated by log-rank test. Cox proportional hazard models were used to calculate hazard ratios and 95% confidence intervals of HCC. Covariates with a P value less than 0.1 in univariate analysis were included in multivariate analysis. Two-tailed P value less than 0.05 was defined to be statistically significant.

Effectiveness of PIVKA-II in diagnosing HCC
In about two and a half years' time, a total of 1016 patients with HCC (covered 49.1% of all PIVKA-II+ patients) were detected by PIVKA-II in the clinical application at Southwest Hospital, Chongqing, China. Among these diagnosed HCC patients, serum AFP (cutoff: 20 ng/ml) levels in 230 cases (22.6%) were negative at the time of diagnosis. Besides, 241 cancer cases (23.7%) of PIVKA-II positive presented no signs of tumor in image examination the first time but were diagnosed as HCC later. The average gap between the elevation of PIVKA-II level and positive results in image examination was 402.5 ± 192.3 days. 3) mAU/ml, respectively. Although levels of PIVKA-II elevated in other diseases, they were significantly higher in HCC group than any other groups (Mann-Whitney P < 0.001). However, there was no significant difference among other groups (Fig. 2b). The influence of different etiology on the level of PIVKA-II was also considered.

Cumulative incidence of HCC by PIVKA-II
Levels of PIVKA-II of all CHB patients were tested in two and a half years' time, and among them, 252 patients with more than 1 years and 3 times of followups were enrolled for analysis. Based on the outcome, they were divided into HCC group and non-HCC group. Table 1 shows the baseline characteristics and Cox survival analysis of all enrolled patients. Among all the most at-risk follow-up patients, 86 cases developed into HCC during the 2 years' follow-up. Male, age per year, ALT < 40 IU/L, TBA < 10 μmol/L, APRI < 0.5, HBV-DNA < 5*10 2 IU/ml, HBsAg negative, HBeAg negative, PIVKA-II < 200 mAU/ml, AFP < 20 ng/ml were selected as a reference. In univariate analysis of this study, female/low level of TBA/low level of PIVKA-II/median level of AFP were protective factors. After adjustment, TBA and PIVKA-II were two variables that significantly influence the incidence of HCC for the most at-risk population and the hazard ratios were 1.918 (95% CI: 1.111-3.310, P = 0.019) and 0.433(95% CI: 0.277-0.678, P < 0.001). This was consistent with our previous study that constant high level of TBA increased the risk of HCC [25]. Figure 5 shows the Kaplan-Meier curve for the cumulative incidence of HCC. At-risk patients were divided into two groups: low-level group (baseline PIVKA-II < 200 mAU/ml) and high-level group (baseline PIVKA-II ≥ 200 mAU/ml), and cumulative incidence were analyzed in all at-risk patients and a sub-cohort group of cirrhotic patients. Figure 5a suggested that in all at-risk patients, the cumulative incidence was 82.0% for the low-level group and reduced significantly to 46.2% for the high-level group (log-rank P < 0.001) at the end of follow-up. Likewise, the cumulative incidence was 82.0% for the low-level group and reduced significantly to 54.1% for the high-level group (log-rank P < 0.001) at the end of follow-up, as shown in Fig. 5b.

Discussion
The main associations for the study of liver in the world simultaneously suggest that tumor biomarkers should not be regarded as a diagnostic criterion but strongly calls for biomarkers in HCC surveillance. In the lately released American Association for the Study of Liver Diseases (AASLD) guidelines for the treatment of HCC, US with or without AFP every 6 months is the recommended strategy for HCC surveillance [26]. It should be noticed that in this guideline biomarkers are conditionally recommended for the first time, though the quality of evidence is low. European Association for the Study of the Liver (EASL) still suggests US every 6 months for HCC surveillance but emphasizes on developing accurate tumor biomarkers [27]. Asian-Pacific Association for the Study of the Liver (APASL) and JSH explicitly recommends US with tumor biomarkers as an efficient strategy for HCC [28]. Therefore, biomarkers are still critical in helping HCC surveillance and diagnosis.
Real-world researches often enrolled an abundant number of participants and a relatively less limited and strict inclusion criterion provide an actual and satisfactory external validity and could be easily spread for widely application [29,30]. Our study for the first time analyzed the efficacy of PIVKA-II in the detection of HCC based on real-world clinical data. We hope to provide some clinical evidence for the use of PIVKA-II.
Between 2014 and 2016, 1016 patients with HCC were revealed by PIVKA-II in our hospital and among them, 230 cases would be neglected if using AFP alone. These results showed that PIVKA-II is a necessary complement to AFP and image examination in HCC surveillance. A total of 241 cases were detected in advance compared with image examination. Importantly, levels of PIVKA-II elevated over 1 year before image discovery in HCC  It has been suggested that levels of PIVKA-II would rise with the progression of HCC and our results gave the same conclusion [31]. But interestingly, levels of PIVKA-II in recurrence group and the residual group were significantly higher than recovery group and there was no difference between recurrence group and residual group. This phenomenon suggested that PIVKA-II could help to predict prognosis of HCC after surgery and our study gave a cut-off value of 282.5 mAU/ml. Some clinical researches have proven that PIVKA-II is a predictive factor of HCC prognosis after ablation or resection [32,33]. Some researches go even further. Atsushi Hiraoka et al. used the number of tumor markers (including PIVKA-II) to predict the response to TACE [34]. Seok-Hwan Kim et al. found that PIVKA-II could be used for expansion of selection criteria of liver transplantation for HCC [35]. However, further largesample multicentered studies are needed to evaluate its prognostic value and determine the cut-off.
Among all the factors that increased the levels of PIVKA-II, cirrhosis and hepatitis are familiar to us. As a result, cirrhosis and hepatitis groups are regarded as the controlled group in many studies. But as a matter of fact, any factors that damage liver cells or trigger liver cell regeneration may increase the serous level of PIVKA-II, although the mechanisms are still unclear [36]. However, clinical researches seldom pay attention to other hepatic diseases that increase levels of PIVKA-II. In our analysis, there was a large part of patients of hepatic adipose infiltration, liver cyst, liver abscess, pregnancy and so on that have elevated levels of PIVKA-II. However, compared with other groups, levels of PIVKA-II were significantly higher in HCC group, suggesting that PIVKA-II is still a biomarker proper for HCC. In addition, a high level of PIVKA-II also warns these participants of the risk of vitamin K deficiency, especially for those who were normal in image examinations. In clinical practice, further examinations may be necessary for this group of people.
Cirrhosis, HBV/HCV infection, aflatoxin B1, alcohol assumption are proven risk factors for HCC and HBeAg seropositive, high viral load, and genotype C are independent predictors of the development of HBV-related HCC. In our analysis, we provide another evidence for PIVKA-II in predicting HCC tumorigenesis. In 1 years' time, many enrolled patients developed into HCC, because all these enrolled participants were the most atrisk population of HCC. But separately, high-level group (PIVKA-II level > 200 mAU/ml) took more risk than low-level group (P < 0.001) with about 80% of patients developing into HCC. This clue strongly indicated that even if PIVKA-II was not a diagnostic marker, but a high-level of PIVKA-II was still an indicator for HCC. However, although a great number of participants were enrolled in our research, the observation time was short. Further research should extend observation time and get more detailed information.

Conclusions
This study was the first real-world research on the effectiveness of PIVKA-II in the detection of HCC. By detecting PIVKA-II, 230 AFP(−) and 241 US(−) patients were diagnosed as HCC in advance in 2 years' time. Levels of PIVKA-II elevated more than 1 year before image diagnosis. High levels of PIVKA-II in at-risk populations were a potent indicator of developing into HCC in 2 years. Our real-world data suggested that the use of PIVKA-II improved the detection rate of PIVKA-II and was a proper complement to AFP and US.