Serum protein induced by vitamin K absence or antagonist-II predicts aggressive tumor biology in alpha-fetoprotein-normal hepatocellular carcinoma

Abstract

BACKGROUND

Patients with hepatocellular carcinoma (HCC) beyond the Milan criteria or with portal vein tumor thrombosis are often excluded from the transplant list owing to aggressive biology and recurrence risk. While high alpha-fetoprotein (AFP) signals aggressiveness, the behavior of normal AFP HCC with elevated protein induced by vitamin K absence/antagonist-II (PIVKA-II) is less defined.

AIM

To assess the prognostic value of PIVKA-II in normal AFP HCC.

METHODS

Retrospective cohort of 113 patients with normal AFP and normal or elevated PIVKA-II. “Aggressive” tumors were defined as beyond Milan and/or portal vein tumor thrombosis (n = 63); others were non-aggressive (n = 50). Receiver operating characteristic curve analysis identified PIVKA-II cut-offs.

RESULTS

This study included 78 men and 35 women; mean age 58.4 ± 11.1 years; 62.8% with decompensated cirrhosis. PIVKA-II was higher in aggressive tumors: Median 2785 mAU/mL (interquartile range: 222-8152) vs 239 mAU/mL (interquartile range: 55-727), P < 0.001. Area under receiver operating characteristic curve 0.756 (95% confidence interval: 0.669-0.844). The Youden-optimized cut-off for aggressive HCC was 1609.5 mAU/mL [sensitivity: 0.54, specificity: 0.94; positive predictive value (PPV): 0.919]. A sensitivity-oriented cut-off of 400 mAU/mL gave sensitivity 0.69 and specificity 0.64 (PPV: 0.71). Regression analysis showed that PIVKA-II > 400 mAU/mL was strongly associated with aggressive tumor phenotype (adjusted odds ratio = 5.16, P = 0.001). All the 29 patients with ≥ 4000 mAU/mL were in the aggressive group (PPV: 1.0). All thresholds were dataset-derived.

CONCLUSION

In normal AFP HCC, PIVKA-II discriminates aggressive biology. A cut-off of 1609.5 mAU/mL balances sensitivity and specificity; 400 mAU/mL favors sensitivity; ≥ 4000 mAU/mL delineates an ultra-high-risk subgroup. Findings support the incorporation of PIVKA-II into risk stratification.

Key Words: Alpha fetoprotein; Protein induced by vitamin K absence/antagonist-II; Des-gamma carboxy prothrombin; Hepatocellular carcinoma; Biomarkers; Aggressive tumor

Core Tip: Alpha-fetoprotein (AFP) often underperforms as a biomarker for hepatocellular carcinoma (HCC). Our study shows that protein induced by vitamin K absence/antagonist-II (PIVKA-II) better indicates aggressive disease when AFP is normal. In normal AFP HCC, PIVKA-II ≥ 400 mAU/mL is linked to high-risk features, and ≥ 4000 mAU/mL identifies an especially aggressive group. Using PIVKA-II in surveillance and treatment planning could improve early detection and outcomes, particularly for AFP-negative patients. Transplant choices may change: Low PIVKA-II, even with large tumors, may suggest slower disease, while very high PIVKA-II within size limits signals higher post-transplant recurrence risk. These findings support updating HCC management guidelines and practice.

INTRODUCTION

Hepatocellular carcinoma (HCC) is a leading cause of cancer-related mortality, with prognosis largely determined by tumor burden and vascular invasion[1]. Alpha-fetoprotein (AFP) is widely used to gauge tumor aggressiveness; higher levels generally indicate more aggressive disease and inform prognosis and treatment selection[2]. However, up to 30% of HCC cases present with normal AFP, limiting early risk stratification[3]. These normal AFP cases pose diagnostic and prognostic challenges because aggressive disease can still occur despite normal values[4]. Thus, AFP alone may not adequately reflect tumor behavior in this subgroup.

Protein induced by vitamin K absence/antagonist-II (PIVKA-II), also known as des-gamma-carboxy prothrombin, has emerged as a complementary biomarker, particularly in AFP-negative cohorts[5,6]. It is produced by defective vitamin K-dependent carboxylation in malignant hepatocytes, and its serum levels correlate with tumor size, vascular invasion, and poor differentiation in HCC[7]. Despite its diagnostic and prognostic utility in Asian guidelines, PIVKA-II remains underutilized in Western protocols, where AFP and imaging predominate[8]. We therefore evaluated the predictive value of PIVKA-II in HCC patients with normal AFP, focusing on its ability to identify aggressive tumor biology[9]. To our knowledge, this is among the first studies dedicated specifically to the normal AFP HCC population.

MATERIALS AND METHODS

Study design and population

This retrospective study analyzed 113 HCC patients with normal AFP levels (≤ 20 ng/mL) and available PIVKA-II measurements, seen over the past three years at our center. Clinical data, including tumor characteristics and treatment outcomes, were collected. Patients were categorized into two groups based on tumor behavior: Aggressive tumors that exceeded Milan criteria, or there was portal vein tumor thrombosis (PVTT), and non-aggressive tumors (within Milan criteria and no PVTT). Clinical parameters included etiology, liver function, and tumor characteristics. PIVKA-II levels were measured via the ARCHITECT PIVKA-II chemiluminescence immunoassay 3C10 with a cutoff value of 40 mAU/mL. The AFP concentration was analyzed using an ARCHITECT AFP 3P36 (Abbott Laboratories, IL, United States) kit, which uses a two-step immunoassay for quantitative measurement. AFP was considered normal if the value was less than 20 ng/mL.

Inclusion criteria

Patients were eligible for inclusion if they had a confirmed diagnosis of HCC, either through histopathological examination or non-invasive imaging criteria, specifically demonstrating arterial hyperenhancement with washout on contrast-enhanced computed tomography/magnetic resonance imaging. Additionally, participants were required to have normal AFP levels, defined as ≤ 20 ng/mL at diagnosis. Furthermore, pre-treatment serum PIVKA-II measurements needed to be available in the medical records. To ensure real-world heterogeneity in the study population, patients with various etiologies of liver disease were included, encompassing hepatitis B and C viral infections, alcohol-related liver disease, non-alcoholic fatty liver disease, and autoimmune liver disease.

Exclusion criteria

Patients were excluded from the study if they presented with mixed histology, specifically intrahepatic cholangiocarcinoma or mixed HCC-cholangiocarcinoma. Furthermore, those with concurrent malignancies, including extrahepatic cancers or metastatic HCC, were not eligible for participation. Additionally, patients who had received any prior anti-tumor therapy, such as chemotherapy, transarterial chemoembolization, microwave ablation, or resection before PIVKA-II measurement, were excluded from the study. Moreover, the study excluded patients who were on vitamin K antagonists, specifically those taking warfarin or who had received recent vitamin K supplementation, as these medications could potentially alter PIVKA-II levels.

Sample size justification

While the study did not explicitly calculate a sample size, the cohort of 113 patients aligns with similar retrospective studies evaluating PIVKA-II in AFP-negative HCC. With 113 patients, our study had > 80% power to detect the observed difference in mean PIVKA-II levels at α = 0.05, based on descriptive sensitivity check. This sample size is in line with prior studies of PIVKA-II in HCC (e.g., Saeki et al[10], n = 90).

Statistical analysis

For baseline comparisons, continuous variables, such as PIVKA-II levels, were compared using the Mann-Whitney U test due to their non-normal distribution as evident from the Shapiro-Wilk test. Categorical variables, including etiology and decompensation, were analyzed using either Pearson’s χ² test or Fisher’s exact test, where appropriate. A two-tailed P-value < 0.05 was considered statistically significant. Receiver operating characteristic (ROC) curve analysis was performed to determine the optimal PIVKA-II cutoff for predicting aggressive behavior using Youden’s index, which maximizes the sum of sensitivity and specificity. The area under the ROC curve was calculated to evaluate diagnostic accuracy. Significant variables for aggressive tumors with univariable P < 0.10 were entered into a multivariable logistic regression. Adjusted odds ratios and P values are reported. The statistical analysis was performed by IBM SPSS Statistics 30.0 (IBM, Armonk, NY, United States).

RESULTS

A total of 113 patients were included, with a mean age of 58.4 ± 11.1 years; 69% were male. Table 1 summarizes the baseline characteristics, including etiologies of liver disease and liver function status. Seventy-one (62.8%) patients had decompensated liver disease due to ascites or encephalopathy. Out of 63 patients with aggressive tumors, 30 patients fulfilling the Milan criteria also had portal PVTT. In only 4 patients, PVTT was present with HCC of less than the Milan criteria. Patients with aggressive tumors had markedly higher PIVKA-II levels than those with non-aggressive tumors (n = 50): Median with interquartile range values were 2785 mAU/mL (222-8152) vs 239 mAU/mL (55-727), P < 0.001 (Figure 1). There was no significant difference in the PIVKA-II levels of patients with (n = 34) and without PVTT (n = 29) within the aggressive groups (P = 0.202 by Mann-Whitney U test).

Figure 1 Comparison of protein induced by vitamin K absence levels in patients with aggressive and non-aggressive hepatocellular carcinoma. PIVKA-II: Protein induced by vitamin K absence/antagonist-II; HCC: Hepatocellular carcinoma; CI: Confidence interval.

Table 1 Patient demographics and characteristics, n (%).

Variables	Values
Age (years), median (IQR)	58 (52-66)
Gender (male/female)
Male	78 (69)
Female	5 (31)
Etiology
HC	49 (43.4)
HB	27 (23.9)
HB + HD	3 (2.6)
HC + HB	2 (1.7)
MASLD	6 (14.1)
Alcohol	6 (5.3)
Autoimmune liver disease	6 (5.3)
Cryptogenic	4 (3.5)
Lab parameters, median (IQR)
Hemoglobin (g/dL)	11.0 (9.3-13.0)
White cell count (× 109/L)	6.8 (5.0-9.0)
Platelets (× 109/L)	116.5 (84-168)
Total bilirubin (mg/dL)	1.3 (0.7-2.3)
Alanine aminotransferase (IU/L)	47.0 (30.0-73.0)
Aspartate aminotransferase (IU/L)	58 (37.5-87.5)
Alkaline phosphatase (IU/L)	96.5 (140.5-219.0)
International normalized ratio	1.27 (1.10-1.44)
Albumin (g/dL)	2.76 (3.10-3.61)
Creatinine (mg/dL)	0.89 (0.70-1.20)
PIVKA-II (mAU/mL)	492.8 (106.6-4284.8)
Alpha-fetoprotein (ng/mL)	6.0 (3.6-10.0)
Elevated PIVKA-II levels	97 (85.8)
PIVKA-II levels ≥ 400	62 (54.8)
PIVKA-II levels ≥ 4000	16 (14.2)
Child Class
A	38 (33.6)
B	56 (49.6)
C	19 (16.8)
Decompensated liver disease (ascites or encephalopathy)	71 (62.8)

IQR: Interquartile range; HC: Hepatitis C; HB: Hepatitis B; HD: Hepatitis D; MASLD: Metabolically-dysfunction-associated steatotic liver disease; PIVKA-II: Protein induced by vitamin K absence/antagonist-II.

ROC curve analysis for PIVKA-II value for aggressive tumors revealed an area under the ROC curve of 0.756 (SE: 0.045; 95% confidence interval: 0.669-0.844; P < 0.001) (Figure 2). The Youden-optimized cut-off was 1609.5 mAU/mL, yielding sensitivity 0.54 and specificity 0.94; corresponding precision [positive predictive value (PPV)] on the precision-recall curve was 0.919. Given the aim to prioritize sensitivity while maintaining reasonable specificity, we selected a pragmatic threshold of 400 mAU/mL. with a sensitivity 0.69 and specificity 0.64 (PPV: 0.71). At a high-stringency threshold of 4000 mAU/mL (n = 29), precision was 1.00 with recall 0.24-0.25 and specificity 1.00, indicating a very specific but insensitive rule (Figure 3). Two-way contingency table analysis for aggressive tumors with PIVKA-II of 4000 or more showed a P value < 0.001 and relative risk with 95% confidence interval of 2.471 (1.929-2.471).

Figure 2 Receiver operating characteristic curve of protein induced by vitamin K absence/antagonist-II values for predicting aggressive hepatocellular carcinoma. ROC: Receiver operating characteristic; PIVKA: Protein induced by vitamin K absence.

Figure 3 Proportion of aggressive vs non-aggressive tumors at two proteins induced by vitamin K absence/antagonist-II thresholds (> 400 mAU/mL and > 4000 mAU/mL). All tumors with protein induced by vitamin K absence/antagonist-II > 4000 mAU/mL were aggressive, whereas at > 400 mAU/mL the cohort splits between aggressive and non-aggressive tumors, as shown. PIVKA: Protein induced by vitamin K absence; HCC: Hepatocellular carcinoma.

Demographic and etiologic variables (age, sex, hepatitis B virus/hepatitis C virus status), liver function metrics (albumin, international normalized ratio, bilirubin, Child class) and tumor variables were compared for aggressive and non-aggressive tumors. Parameters for aggressive tumors with P-values < 0.1 were hepatitis B etiology (0.080), hepatitis C etiology (0.039), white blood cell count (0.094), platelets count (0.015), serum sodium (0.025), bilirubin (0.039), aspartate transaminase (0.010), alanine transaminase (0.018), alkaline phosphatase (0.004), and PIVKA-II levels > 400 mAU/mL (P < 0.001). In the logistic regression analysis, PIVKA-II > 400 mAU/mL was strongly associated with aggressive tumor phenotype (adjusted odds ratio = 5.16, P = 0.001). Platelet count and alkaline phosphatase showed inverse associations (Table 2). Model calibration was assessed using the Hosmer-Lemeshow goodness-of-fit test, which showed no evidence of lack of fit (χ² = 2.786, df = 8, P = 0.947).

Table 2 Logistic regression analysis of significant variables identified by univariate analysis.

Variable	Sig	Exp(B)
Hepatitis B etiology	0.083	2.977
Hepatitis C etiology	0.581	0.744
White blood cell count (× 109/L)	0.130	1.097
Platelets count (× 109/L)	0.014	0.988
Total bilirubin (mg/dL)	0.264	0.929
Aspartate transaminase (IU/L)	0.311	0.994
Alanine transaminase (IU/L)	0.962	1.000
Alkaline phosphatase (IU/L)	0.009	0.994
PIVKA-II 400 or more	0.001	5.155

PIVKA-II: Protein induced by vitamin K absence/antagonist-II.

DISCUSSION

Our study emphasizes the need for improved assessment tools for patients with HCC. By evaluating PIVKA-II in patients with normal AFP levels, healthcare providers can gain a better understanding of tumor behavior, which is crucial for treatment planning and patient management[6,11]. This study demonstrates that elevated PIVKA-II levels were a significant predictor of aggressive HCC in normal-AFP patients, aligning with prior evidence linking PIVKA-II to vascular invasion and tumor progression[12]. In a study by Bhatti et al[13] involving 57 patients with microvascular invasion, the combination of AFP ≥ 40 ng/mL plus PIVKA-II ≥ 350 mAU/mL yielded a sensitivity of 69.6%. In another study by the same group, a combination of AFP > 20 ng/mL and PIVKA-II > 1000 mAU/mL predicted 47.1% of HCC recurrences after liver transplant, whereas HCC recurred in 6.1% of patients not meeting this threshold[14]. Our study differed from the above studies as our patients had normal AFP levels. By ROC analysis, the Youden-optimized PIVKA-II cut-off was 1609.5 mAU/mL (sensitivity: 0.54, specificity: 0.94). To prioritize sensitivity while maintaining reasonable specificity, we also considered a pragmatic threshold of 400 mAU/mL (sensitivity: 0.69, specificity: 0.64). Notably, all patients with PIVKA-II ≥ 4000 mAU/mL (n = 29) were in the aggressive group, suggesting this value delineates an ultra-high-risk stratum. However, this finding is exploratory and limited by sample size and prevalence.

Currently, PIVKA-II is not widely included in most HCC management guidelines. While American Association for the Study of Liver Diseases guidelines omit PIVKA-II for screening, Japanese protocols endorse its combined use with AFP[15]. In light of our findings, PIVKA-II could be adopted alongside AFP in risk stratification protocols for HCC, particularly in patients with normal AFP, where it fills a prognostic gap. Integrating PIVKA-II into practice could improve risk assessment and guide management decisions, potentially leading to better patient outcomes. Understanding the role of PIVKA-II in predicting tumor aggressiveness allows for more tailored treatment strategies. For instance, patients with elevated PIVKA-II levels, despite normal AFP, may require more aggressive treatment approaches or closer monitoring to manage the risk of recurrence and improve survival rates[16]. Post-treatment PIVKA-II dynamics (e.g., decline after resection) may predict recurrence, as shown in longitudinal studies[17-20].

The strength of this study is that it focuses on AFP-negative HCC. Unlike many earlier studies that included all HCC patients, our normal AFP cohort allowed us to highlight the subset of patients for whom AFP fails to flag risk. This population is underrepresented in guidelines. Our patients had diverse liver disease etiologies and both early and advanced tumor stages, which improves the applicability of our findings to general HCC populations, albeit still needing external validation. Patients with HCC exceeding Milan criteria and those with PVTT are often excluded from transplant lists due to their aggressive nature. Biomarker-based assessments (including PIVKA-II) might one day be used to refine transplant eligibility; for instance, distinguishing patients who, despite large tumors, have indolent tumor biology and could benefit from transplant, as opposed to those with aggressive biology where transplant would be futile.

The limitations of this study are the retrospective design, relatively small sample size, single-center cohort, and lack of longitudinal PIVKA-II measurements to assess dynamic changes. Because we only included patients who had PIVKA-II measured, there may be an inherent selection bias; clinicians might have ordered PIVKA-II in cases they suspected to be high-risk, potentially enriching our cohort with aggressive cases. Our definition of ‘aggressive tumor’ is partly based on tumor size and number (Milan criteria) or vascular invasion, which is inherently correlated with PIVKA-II levels. This relationship might inflate the predictive performance of PIVKA-II for our endpoint. However, it’s noteworthy that PIVKA-II identified this aggressive phenotype even in normal-AFP patients, supporting its utility.

Future directions

Our findings open avenues for further research into the role of PIVKA-II in HCC. While AFP is a traditional marker for assessing HCC aggressiveness, its limitations in patients with normal levels necessitate the exploration of alternative biomarkers. Prospective trials could test whether incorporating PIVKA-II into HCC surveillance or treatment allocation improves outcomes. By incorporating PIVKA-II into evaluation, clinicians might identify certain patients with traditionally ‘aggressive’ tumors who could still benefit from transplant, for example, those beyond Milan criteria but with low biomarker levels suggesting less aggressive biology. Monitoring PIVKA-II after curative treatment (resection/ablation) in AFP-negative patients might help predict recurrence. This could provide additional insights into the behavior of HCC.

CONCLUSION

In HCC patients with normal AFP, elevated PIVKA-II levels (especially ≥ 400 mAU/mL) are strongly associated with aggressive tumor characteristics. Incorporating PIVKA-II into risk assessments could improve prognostication and guide management, such as transplant eligibility and adjuvant therapy decisions. Our findings support including PIVKA-II alongside AFP in HCC workups, but prospective multi-center studies are warranted to validate the cutoff values and refine their clinical use.