Protein induced by vitamin K absence or antagonist II as a prognostic marker in hepatocellular carcinoma patients with normal serum alpha-fetoprotein levels.

1. INTRODUCTION
In 2021, hepatocellular carcinoma (HCC) was responsible for an estimated 483 900 deaths worldwide, making it the third leading cause of cancer mortality.1,2 Recent Taiwanese data also highlight environmental and metabolic factors such as air pollution as contributing to HCC risk even after adjusting for viral hepatitis.3 Despite advances in surveillance and management, the 5-year overall survival (OS) rate for patients with HCC remains approximately 20%.2 The diagnosis of HCC primarily relies on dynamic imaging modalities, such as contrast-enhanced computed tomography (CT) or magnetic resonance imaging (MRI), along with histopathological confirmation when necessary.4,5 To improve early detection among high-risk populations, particularly individuals with liver cirrhosis or chronic viral hepatitis, current clinical practice guidelines recommend incorporating serum levels of tumor biomarkers alongside imaging for effective HCC surveillance.4–6
Alpha-fetoprotein (AFP) is a circulating protein that is secreted by dedifferentiated HCC cells and has long been used as a serum biomarker for the screening and diagnosis of HCC.4,5,7–9 AFP secretion is associated with the degree of tumor-cell differentiation, which varies widely among patients.10 However, the sensitivity and specificity of serum AFP for HCC detection remain suboptimal, and its utility as a prognostic marker is limited.11 Moreover, a substantial proportion of patients have normal serum levels of AFP at the time of HCC diagnosis.12 Patients with normal AFP exhibit heterogeneous clinical characteristics, including variations in tumor burden, liver functional reserve, treatment strategies, and outcomes. Therefore, there is an unmet need to identify alternative prognostic biomarkers specifically for this subgroup.
Prothrombin induced by vitamin K absence or antagonist II (PIVKA-II) is an abnormal form of prothrombin that HCC cells produce and has emerged as a promising serum biomarker for HCC detection.7,13–15 Also known as des-gamma-carboxy prothrombin (DCP), elevated serum levels of PIVKA-II have been linked to adverse tumor features such as microvascular invasion.16,17 However, evidence regarding its sensitivity for large-scale HCC surveillance has been limited and inconsistent, so current clinical guidelines do not recommend using PIVKA-II alone for HCC screening.5
Comparative studies have extensively evaluated the diagnostic performance of AFP and PIVKA-II in HCC. Several reports have demonstrated that the sensitivity and specificity of HCC detection can be significantly improved by combining serum AFP and PIVKA-II levels with other methods such as the “gender, age, AFP-L3, AFP, and des-carboxy-prothrombin” (GALAD) score.14,18–20 In addition, PIVKA-II-based prognostic models have shown clinical utility in predicting survival outcomes, which further supports its role as a valuable biomarker across various stages of HCC management.21–23
However, most studies have primarily focused on the diagnostic or surveillance value of these biomarkers and often rely on models that integrate both AFP and PIVKA-II. These approaches may be less applicable to patients with normal AFP levels, for whom AFP provides limited prognostic information.24 To address this gap, this study evaluated the prognostic value of serum PIVKA-II specifically in patients with HCC and normal AFP levels.

2. METHODS

2.1. Patients and follow-up
This study enrolled 330 consecutive treatment-naïve patients with HCC with serum AFP levels <20 ng/mL, which were defined as “normal” based on both the 2018 AASLD practice guidelines and prior Taiwanese studies.12,25,26 The patients were diagnosed between 2020 and 2023 at Taipei Veterans General Hospital (VGHTPE). Serum biochemistry and imaging data from the time of diagnosis were available for all patients (Fig. 1). Clinical information was obtained from the VGHTPE HCC registration system, a prospectively maintained database in which demographic characteristics, baseline laboratory data, tumor characteristics, treatments, and clinical outcomes are recorded. This database has been utilized in several studies.27–30 Patients were followed every 3 months until death, loss to follow-up, or the endpoint of the study (January 30, 2024).
The study protocol was conducted in accordance with the Declaration of Helsinki and relevant ethical guidelines and was approved by the Institutional Review Board (IRB) of VGHTPE, Taiwan (VGHIRB No. 2025-07-033CC). Due to the retrospective cohort study design, the requirement for informed consent was waived by the IRB. All patient data were de-identified before the initiation of this study to ensure confidentiality.

2.2. Biochemical, serological markers, and image features
Baseline demographic data and serum biochemistry parameters were collected at the time of diagnosis for all enrolled patients. The data collected included age, sex, serum albumin, creatinine, alanine aminotransferase (ALT), aspartate aminotransferase (AST), alkaline phosphatase (ALK-P), total bilirubin, AFP, PIVKA-II, platelet count, and the prothrombin time international normalized ratio (PT INR). Serum concentrations of AFP were measured using an electrochemiluminescence immunoassay (ECLIA) on a Cobas e801 module (Roche Diagnostics, Basel, Switzerland), for which the manufacturer’s reference cutoff is 7 ng/mL. To ensure robustness, we performed an exploratory analysis comparing OS between patients with AFP >7 ng/mL and those with AFP ≤7 ng/mL. Serum PIVKA-II levels were quantified using a chemiluminescent immunoenzymatic assay on a LUMIPULSE® G1200 system analyzer (Fujirebio Europe N.V., Gent, Belgium), which has an analytical range of 5 to 75 000 mAU/mL and a reference cutoff value of 48 mAU/mL. Imaging data were obtained from contrast-enhanced CT or MRI at diagnosis and included the tumor number, tumor size, presence of vascular invasion, and distant metastasis.

2.3. Statistical analysis
The primary outcome was OS, which was defined as the time from the date of HCC diagnosis to death, the last visit, or loss to follow-up. Data were censored as of June 30, 2024. Categorical and ordinal variables are presented as frequencies and percentages, while continuous variables are expressed as medians with interquartile ranges (IQRs). The optimal cutoff value for serum PIVKA-II was determined using the Youden index for OS. According to a receiver operating characteristic (ROC) curve analysis, the optimal threshold was 100 mAU/mL, which provided the best discriminatory performance in our cohort (Supplementary Fig. S1, https://links.lww.com/JCMA/A356, and Supplementary Table S1, https://links.lww.com/JCMA/A357). Cutoff values for other clinical parameters were based on standard laboratory reference values.
Comparisons between categorical variables were performed using Fisher’s exact test or the chi-squared test with Yates’ correction as appropriate. The Mann-Whitney U test was used for comparisons of continuous variables. OS was analyzed using Kaplan-Meier survival curves, and differences were assessed using the log-rank test. Risk factors associated with OS were evaluated using Cox proportional-hazards regression models. All statistical analyses were conducted using SPSS Statistics for Windows version 24.0 (IBM Corp., Armonk, NY) and R software version 3.6.3. A two-tailed p value <0.05 was considered statistically significant.

3. RESULTS

3.1. Comparison of baseline characteristics based on serum PIVKA-II levels
The analysis included a total of 330 patients with HCC who had normal serum AFP levels and met the inclusion criteria. The participants’ baseline clinical characteristics stratified by serum PIVKA-II levels are summarized in Table 1. The majority of patients were male, and more than half had viral hepatitis-related HCC. Overall, 60.3% of patients were diagnosed at an early stage (Barcelona Clinic Liver Cancer [BCLC] stage 0 or A), and 83.0% had preserved liver function classified as Child-Pugh class A.
Among the 224 patients who underwent curative treatment, 159 received surgical resection, 56 underwent local ablation therapy, and 9 underwent liver transplantation. Among the 106 patients who received non-curative treatment, 36 underwent transarterial chemoembolization, 39 received systemic therapy, and 31 received best supportive care. Compared to patients with low serum PIVKA-II levels, those with high levels exhibited more aggressive tumor features, including more advanced BCLC stages, higher frequency of multinodular tumors, larger tumor sizes, greater incidence of extrahepatic metastases and macrovascular invasion, and lower likelihood of receiving curative treatment modalities. However, there were no significant differences between the groups in terms of age, sex, underlying etiology, or liver-functional reserve.

3.2. Prognostic ability of serum PIVKA-II levels
After a median follow-up of 17 months (IQR: 8.0-27.0 months), 50 patients had died, and the 3-year OS rate was 77% for the entire cohort. The Kaplan-Meier survival analysis demonstrated a significant difference in OS between the high and low-PIVKA-II groups, which had 3-year OS rates of 54.9% and 87.1%, respectively (p < 0.001) (Fig. 2). As shown in Table 2, the multivariate Cox regression analysis identified high serum levels of PIVKA-II as an independent predictor of poor OS (hazard ratio [HR]: 2.422, 95% CI, 1.128-5.198, p = 0.023). Other independent predictors included the presence of vascular invasion (HR: 1.996, 95% CI, 1.026–3.882, p < 0.001), elevated serum creatinine (HR: 4.108, 95% CI, 2.166-7.793, p < 0.001), elevated ALK-P (HR: 2.346, 95% CI, 1.225-4.492, p = 0.001), prolonged PT INR (HR: 2.964, 95% CI, 1.576-5.574, p = 0.001), and the absence of curative treatment modalities (HR: 3.630, 95% CI, 1.775-7.421, p < 0.001).

3.3. Subgroup analysis according to liver function, tumor stage, and treatment modality
Subgroup analyses were conducted based on liver function (determined by albumin-bilirubin [ALBI] grade), tumor stage, and treatment modality. As shown in Fig. 3, high serum PIVKA-II levels were significantly associated with poorer OS among patients with impaired liver function (ALBI grade 2 or 3), those with more advanced tumor stages (BCLC stages B-D), and those who did not receive curative treatment. In contrast, among patients who had well-preserved liver function, had early-stage disease, or underwent curative treatment, OS was comparable between high and low-PIVKA-II groups. Furthermore, even when stratifying patients according to their serum AFP concentrations within the normal range (high-normal vs low-normal with a cutoff value of 7 ng/mL), serum PIVKA-II levels effectively discriminated survival outcomes.

4. DISCUSSION
This study yielded several important findings. First, we demonstrated that elevated serum PIVKA-II levels were an independent risk factor for poor OS in patients with HCC and normal AFP levels. Subgroup analyses revealed that its prognostic value was particularly evident among patients who had impaired liver function reserve, had more advanced tumor stages, and did not receive curative treatment. These findings address the unmet need for a reliable prognostic serum biomarker in this distinct subset of patients with HCC, for which AFP provides limited predictive utility.
Moreover, our findings confirm that elevated serum PIVKA-II levels are associated with more aggressive tumor characteristics in patients with HCC with normal AFP levels, including macrovascular invasion, which is consistent with previous reports.14 We also observed a 3-year OS rate of 77%, which is higher than that reported in unselected HCC cohorts and aligns with earlier studies showing a relatively favorable prognosis for patients with normal AFP.2,12 AFP has long been regarded as the primary serum biomarker for patients with HCC. Experimental data suggest that AFP may promote hepatocarcinogenesis by modulating apoptotic pathways and cytoplasmic signaling cascades.31,32
Overexpression of AFP is influenced by multiple factors, including the degree of cellular differentiation and molecular tumor subtypes.33–35 Nevertheless, a substantial proportion of patients with HCC present with normal AFP levels, and research has described distinct clinical characteristics and molecular profiles within this subgroup.12,36,37 Moreover, our previous study demonstrated that serum AFP levels (>7 vs ≤7 ng/mL) were not an independent prognostic factor in patients with HCC with normal AFP values.12 This highlights the unmet need for alternative prognostic biomarkers in this subgroup.
PIVKA-II is an abnormal form of prothrombin that results from an acquired defect in the post-translational carboxylation of its precursor in malignant hepatocytes.37,38 Immunohistological studies indicate that the expression and secretion pathways of AFP and PIVKA-II are distinct, which suggests that they may reflect different biological aspects of tumor behavior.8 Several studies have demonstrated the utility of PIVKA-II in the early diagnosis of HCC. Poté et al14 reported that PIVKA-II is a reliable biomarker for detecting early and very-early-stage HCC and is strongly correlated with the presence of vascular invasion. Xu et al showed that PIVKA-II expression is associated with epithelial-mesenchymal transition (EMT), a critical process in tumor invasion and metastasis in HCC.17,39 These findings support the role of PIVKA-II as a marker of tumor aggressiveness and adverse prognosis.
Despite the well-established roles of AFP and PIVKA-II in the diagnosis and surveillance of HCC, their prognostic significance has received relatively little attention. Norman et al22 demonstrated that serum AFP-L3 and PIVKA-II levels could predict early HCC recurrence following liver transplantation. Similarly, Kang et al reported that patients with concurrently low serum AFP and PIVKA-II levels exhibited more favorable clinical characteristics and outcomes.40 However, no prior study has specifically evaluated the prognostic role of PIVKA-II in all patients with HCC who have normal AFP levels, nor conducted detailed subgroup analyses in this setting.
To date, only one recent study has suggested that on-treatment changes in PIVKA-II were associated with survival outcomes in patients with unresectable HCC and normal serum AFP levels who received systemic therapy.24 To our knowledge, our study is the first to comprehensively evaluate baseline PIVKA-II as a prognostic biomarker in patients with HCC and normal serum AFP levels across different disease stages and treatment modalities. Our study provides novel and robust evidence that serum PIVKA-II can be used to independently predict survival outcomes and stratify risk in this clinically important population.
Interestingly, the prognostic value of PIVKA-II was less apparent for patients who had early-stage disease, had preserved liver function, or underwent curative treatments. This likely reflects the excellent survival outcomes in these subgroups, which limited the ability to detect differences based on biomarker levels. Moreover, as PIVKA-II testing was only implemented at our center in 2020, the median follow-up duration was relatively short (17 months), which potentially led to underestimation of its long-term prognostic impact. Prospective studies with longer follow-up are needed to clarify the role of PIVKA-II in these lower-risk populations.
Several limitations of this study should be acknowledged. First, it was a single-center retrospective cohort study, and the relatively small number of patients in some subgroups may have limited the generalizability of the findings. Second, 47.9% of the patients in our cohort had infections with the hepatitis B virus, and 13.9% had infections with the hepatitis C virus, indicating that the majority of cases were viral hepatitis-related HCC. This is consistent with the epidemiological distribution of HCC in Eastern countries.41–43 We acknowledge that non-viral etiologies, such as metabolic dysfunction-associated steatotic liver disease and alcohol-related liver disease, were underrepresented.
Because viral and non-viral HCC may differ in hepatocarcinogenesis pathways and tumor biology, the prognostic value of PIVKA-II could potentially differ between these groups.44 However, the limited number of non-viral patients with HCC in our study precluded meaningful subgroup analyses. Prospective multicenter studies with more diverse etiologies are warranted to further clarify this important issue.
PIVKA-II measurements were obtained only at the time of diagnosis, so the potential prognostic significance of dynamic changes in PIVKA-II could not be evaluated. The median follow-up period was only 17 months in our cohort, which might underestimate the long-term survival differences by these biomarkers. The multivariate model may not have been optimal due to the limited cohort size and potential multicollinearity among covariates. Machine-learning approaches such as Least Absolute Shrinkage and Selection Operator (LASSO) regression may help to improve variable selection and model robustness in future studies.29 Finally, the optimal cutoff values for both AFP and PIVKA-II in specific subgroups and prognostic purposes remain debated and may vary between different patient populations. Therefore, larger multicenter studies with more diverse etiologies and standardized biomarker thresholds are warranted to validate our findings.
In conclusion, serum PIVKA-II was identified as an independent prognostic factor for patients with HCC and normal AFP levels and effectively stratified patients into different risk groups. Incorporating PIVKA-II measurement into clinical practice may improve risk assessment and guide treatment decision-making for this distinct subgroup of patients with HCC.

APPENDIX A. SUPPLEMENTARY DATA
Supplementary data related to this article can be found at https://links.lww.com/JCMA/A356 and https://links.lww.com/JCMA/A357.