Baseline and longitudinal biomarkers predict hepatocellular carcinoma in chronic hepatitis C: A cohort study

Abstract

BACKGROUND

Hepatocellular carcinoma (HCC) remains a major complication in patients with chronic hepatitis C (CHC). Practical, low-cost tools that identify untreated patients at high cancer risk are needed for surveillance and treatment prioritization. Noninvasive biomarkers, including the fibrosis-4 (FIB-4) index and alpha-fetoprotein (AFP), have been identified as potential predictors of HCC development in patients with CHC. While most studies focus on baseline predictors, longitudinal changes in biomarkers has not been fully clarified.

AIM

To evaluate whether baseline and longitudinal trajectories of noninvasive biomarkers, particularly the FIB-4 index, predict subsequent HCC in treatment-naïve CHC.

METHODS

We performed a single-center retrospective cohort study of adults with CHC first evaluated between 1999 and 2015. Patients with other major liver diseases or prior HCC were excluded. Demographics, virology, labs and imaging were collected at baseline and year 3. FIB-4 categories were < 1.45, 1.45-3.25, and > 3.25; 3-year dynamics were assessed by category transition and absolute change (ΔFIB-4). Time-to-event analyses (Kaplan-Meier, Cox models) estimated associations with incident HCC.

RESULTS

Over a median 9.4-year follow-up, 29.9% of patients developed HCC. Baseline FIB-4 > 3.25 significantly predicted HCC [adjusted hazard ratio (aHR) = 2.72; P < 0.001]. At year 3, AFP ≥ 20 ng/mL was independent predictor (aHR = 6.74; P < 0.001). Notably, longitudinal analysis demonstrated that a 3-year absolute increase in FIB-4 (ΔFIB-4 ≥ 1.0) was an independent risk factor (aHR = 2.67; P < 0.001). Stratification by trajectory revealed that the increase group (ΔFIB-4 ≥ +1.0) exhibited the highest 15-year cumulative HCC incidence of 72.7%.

CONCLUSION

High baseline and rising FIB-4 scores strongly predict HCC. Integrating longitudinal FIB-4 monitoring with AFP assessment is essential for risk stratification and prognosis in CHC.

Key Words: Chronic hepatitis C; Hepatocellular carcinoma; Fibrosis-4 index; Alpha-fetoprotein; Longitudinal biomarkers; Risk stratification; Cohort study

Core Tip: This retrospective cohort study evaluated 271 treatment-naïve patients with chronic hepatitis C to determine whether baseline and 3-year trajectories of the fibrosis-4 (FIB-4) index and other biomarkers predict long-term hepatocellular carcinoma (HCC) risk. A baseline FIB-4 value > 3.25 strongly predicted HCC independently, while a 3-year ΔFIB-4 ≥ 1 helped risk stratification. Persistently low FIB-4 values identified a group with negligible HCC incidence. Periodic, low-cost monitoring of FIB-4 and alpha-fetoprotein may offer a practical approach to guide surveillance and treatment prioritization in resource-limited settings.

INTRODUCTION

Hepatitis C virus (HCV) infection remains a significant public health issue globally due to its progression to severe liver complications, including decompensated liver cirrhosis and hepatocellular carcinoma (HCC)[1]. While direct-acting antivirals (DAAs) have revolutionized HCV management with cure rates exceeding 95% across diverse populations[2], the World Health Organization’s goal of eliminating viral hepatitis by 2030 faces substantial hurdles[3]. A significant disparity persists between treatment eligibility and actual initiation, particularly in real-world settings[4]. Global estimates indicate that less than 15% of the 58 million infected individuals have received curative therapy, with pronounced disparities in lowincome settings[5]. In 2020, the cascade of care showed that only 4% of diagnosed patients in lowincome countries accessed treatment[6], leaving a vast reservoir of untreated patients at risk of disease progression.

This gap in the “cascade of care” is multifactorial. System-level barriers include economic constraints, limited healthcare infrastructure, diagnostic challenges, and restrictive eligibility criteria based on fibrosis stages or other parameters[7,8]. At the patient and provider level, common contributors to non-treatment comprise low awareness, concerns about adverse effects, suboptimal adherence, and coexisting comorbidities[9,10]. Consequently, a “treatment-naïve” population persists, representing a unique “natural history” cohort. As this population ages, the cumulative risk of liver-related mortality and HCC is projected to rise through 2030[6], underscoring the urgent need for effective risk stratification to prioritize resource allocation and surveillance.

Serum alpha-fetoprotein (AFP) remains the most commonly used biomarker for HCC screening; however, its specificity is limited, especially in distinguishing HCC from benign liver conditions[11]. The fibrosis-4 (FIB-4) index has emerged as a non-invasive, cost-effective surrogate marker for assessing liver fibrosis severity and stratifying HCC risk in chronic liver disease patients, including those infected with HCV[12]. Specifically, chronic hepatitis C (CHC) patients with a baseline FIB-4 score ≥ 3.25 exhibit a significantly elevated risk of developing HCC, highlighting its potential as a robust risk stratification tool[13]. Furthermore, longitudinal changes in FIB-4 scores have recently gained attention as potential predictors of HCC development in patients who achieved sustained virologic response (SVR) with DAA[14]. However, the prognostic value of dynamic FIB-4 changes in untreated patients remains understudied. In the absence of viral eradication, ongoing inflammation and fibrosis remodeling may manifest as distinct longitudinal biomarker patterns that static scores fail to detect.

Therefore, this study aimed to evaluate the prognostic utility of baseline and longitudinal FIB-4 dynamics in a long-term cohort of treatment-naïve CHC patients. We hypothesized that longitudinal changes in FIB-4 scores (ΔFIB-4) would provide superior risk stratification for HCC compared to static baseline measurements alone. Our findings aim to bridge the knowledge gap in the natural history of HCV and offer evidence for optimizing surveillance strategies in settings where universal DAA coverage is not yet fully realized.

MATERIALS AND METHODS

Study design and populations

A retrospective analysis was conducted on a cohort of treatment-naïve patients with CHC at Chi Mei Hospital, Liouying from 1999 to 2015. These patients underwent longitudinal follow-up at the Department of Hepatology. Patients were excluded if they had coexisting severe liver diseases such as hepatitis B co-infection, autoimmune hepatitis, alcoholic liver disease, prior HCC history, follow-up less than 3 years without development of HCC and incomplete treatment information. This study was conducted and performed in accordance to the ethical principles for medical research involving human subjects of the Declaration of Helsinki, updated in 2013. It was approved by the Institutional Review Board of the Chi Mei Hospital, Liouying (Approval No. 11412-L02) Patient consent was waived due to the retrospective nature of the study and data anonymization.

Baseline characteristics and biochemical variables

The index date was defined as the earliest date of tested positive anti-HCV, and follow-up continued until the first occurrence of HCC, death, loss to follow-up, or the study end date. Data were collected including demographic information (age, sex, comorbidities such as diabetes, hypertension, and hyperlipidemia), hepatitis C-related clinical parameters (HCV genotype and viral load), laboratory values necessary for FIB-4 calculation [aspartate aminotransferase (AST), alanine aminotransferase (ALT), and platelet count], and other liver function markers (albumin, total bilirubin, prothrombin time). Additional biochemical data such as AFP and serum creatinine were also recorded for further analysis. Imaging studies, primarily abdominal ultrasound, were used to assess liver cirrhosis, and computed tomography (CT) or magnetic resonance imaging (MRI) findings were included when available for more accurate diagnostic confirmation. The FIB-4 index was calculated using the following formula: Age (years) × AST (U/L)/[platelet count (10⁹/L) × ALT (U/L)^1/2]. At the three-year follow-up time point, the same FIB-4-related parameters and biochemical markers were reassessed.

The primary end-point was incident HCC, confirmed either histologically or by characteristic arterial-phase enhancement with venous-phase wash-out on multiphasic contrast-enhanced CT or MRI. For every case, the date of diagnosis and the corresponding Barcelona Clinic Liver Cancer (BCLC) stage[15] were documented for time-to-event analysis.

Statistical analysis

Baseline characteristics were presented as median (interquartile range) and n (%). The FIB-4 scores were categorized into three groups: < 1.45, 1.45-3.25, and > 3.25 for analysis[16]. For the variables either at baseline or 3-year time point contributing to HCC development, we firstly use univariable Cox regression models to identify possible correlation factors which were shown in hazard ratio (HR) and 95% confidence intervals (95%CIs). Then, clinically relevant variables (including age, sex, diabetes, hypertension, and hyperlipidemia) and those with P < 0.1 in univariable analysis were included in the multivariable Cox proportional hazards regression models which were shown in adjusted HR (aHR) and 95%CIs. Also, we conducted the Kaplan-Meier method to estimate the cumulative incidence of HCC by baseline and 3-year FIB-4 score along with the Log-Rank test for significance comparison.

For the dynamic change of biochemical parameters during follow up, we focused on the 3-year trajectory of the FIB-4 score compared with baseline. Trajectories were defined based on transitions between these FIB-4 categories. An improved trajectory was defined as a shift from > 3.25 to either 1.45-3.25 or < 1.45, or from 1.45-3.25 to < 1.45. A worsened trajectory was defined as a transition from < 1.45 to either 1.45-3.25 or > 3.25, or from 1.45-3.25 to > 3.25. Patients who remained within the same category over the 3-year period were classified as stable. The heterogeneity between each FIB-4 trajectory group was assessed using the Pearson χ² test.

Absolute change in FIB4 (ΔFIB-4) was defined as the 3-year value minus the baseline value and was analyzed as a continuous variable to preserve information using landmark Cox regression models. To derive an empirical cutpoint for ΔFIB-4, we conducted a grid search across candidate thresholds. Based on this datadriven approach, patients were then further grouped into three categories based on ΔFIB-4: Decrease (ΔFIB-4 ≤ -1.0), stable (-1.0 < ΔFIB-4 < +1.0), and increase (ΔFIB-4 ≥ +1.0) to further evaluate whether improvement in FIB-4 was associated with a lower subsequent HCC risk. The heterogeneity between each FIB-4 trajectory group was assessed using the χ² test and HR was estimated using Cox regression models. Then, the Kaplan-Meier method was conducted to analyze cumulative incidence of HCC by these three categories and compared using the log-rank test. All statistical analyses were performed with a two-tailed significance threshold of P < 0.05.

RESULTS

Baseline characteristics and primary outcome

From 1999 to 2015, a total of 553 CHC patients were identified as treatment-naïve patients at Chi Mei Hospital, Liouying. After excluding those meeting exclusion criteria, 271 patients were included in the analysis with a median age of 66.0 (58.0-73.0) years. The 156 patients (57.6%) were female and 76 patients (28.0%) had baseline liver cirrhosis. Other basic characteristics are shown in Table 1. After a median 9.4 years of follow up, 81 patients (29.9%) developed HCC with an interval of 7.45 years. BCLC stages were distributed as: Stage 0: 8 (9.9%), A: 43 (53.1%), B: 10 (12.3%), C: 12 (14.8%), and D: 8 (9.9%). The 16 patients developed HCC within 3 years of follow up and 4 patients lost to follow up at the 3-year time point; therefore, 251 patients were eligible for 3-year biomarkers analysis (Figure 1).

Figure 1 Flow chart of study design and population. CHC: Chronic hepatitis C; HCC: Hepatocellular carcinoma.

Table 1 Baseline characteristics of the study cohort.

Parameter	Value
Demographics
Age (years, n = 271)	66.0 (58.0-73.0)
Sex (n = 271)
Male	115 (42.4)
Female	156 (57.6)
Follow-up time (years, n = 271)	9.4 (8.7-10.1)
Virological profile
HCV viral load (IU/mL, n = 190)	941202 (64914-2861279)
HCV genotype (n = 136)
Type 1	49 (36.0)
Type 2	48 (35.3)
Type 6	23 (16.9)
Other (mixed)	16 (11.8)
Comorbidities
Hypertension (n = 271)	139 (51.3)
Diabetes mellitus (n = 271)	86 (31.7)
Dyslipidemia (n = 271)	31 (11.4)
Disease severity
Cirrhosis (n = 270)	76 (28.0)
Baseline FIB-4 index (n = 263)
< 1.45	35 (13.3)
1.45-3.25	93 (35.4)
> 3.25	135 (51.3)

Data are presented as median (interquartile range) or n (%). HCV: Hepatitis C virus; FIB-4: Fibrosis-4.

Baseline risk factors and biomarkers regarding HCC development

In baseline univariable Cox models, older age, higher baseline fibrosis burden and tumor marker levels were associated with subsequent HCC. Compared to patients with baseline FIB-4 1.45-3.25, those with FIB-4 > 3.25 had an increased hazard of HCC (HR = 3.13, 95%CI: 1.84-5.31; P < 0.001), whereas FIB-4 < 1.45 was associated with a lower hazard (HR = 0.12, 95%CI: 0.02-0.89; P = 0.039). Per 1-year increase in age (HR = 1.03, 95%CI: 1.00-1.05; P = 0.033), baseline AFP ≥ 20 ng/mL (HR = 3.31, 95%CI: 1.85-5.92; P < 0.001) and total bilirubin ≥ 1 mg/dL (HR = 3.37, 95%CI: 1.85-6.15; P < 0.001) were also significant predictors.

When these variables were entered into a multivariable model, male sex (aHR = 1.74, 95%CI: 1.05-2.86; P = 0.030), baseline FIB-4 > 3.25 (aHR = 2.72 vs 1.45-3.25, 95%CI: 1.51-4.90; P < 0.001), and AFP ≥ 20 ng/mL (aHR = 2.79, 95%CI: 1.50-5.19; P = 0.001) remained independently associated with increased HCC risk. The comparison of FIB-4 < 1.45 vs 1.45-3.25 was not estimable in this multivariable complete-case dataset due to sparse events in the < 1.45 group after exclusion for missing covariates (Table 2).

Table 2 Cox regression analysis of hepatocellular carcinoma in relation to baseline characteristic and biomarkers.

Variable	Univariate		Multivariate
	HR (95%CI)	P value	aHR (95%CI)	P value
Demographics
Sex (male)	1.51 (0.97-2.33)	0.067	1.74 (1.05-2.86)	0.03
Age (per 1-year increase)	1.03 (1.00-1.05)	0.033	1.01 (0.98-1.04)	0.559
Biomarkers
HCV RNA > 6 × 106 (IU/mL)	0.78 (0.28-2.16)	0.634	-	-
AFP ≥ 20 (ng/mL)	3.31 (1.85-5.92)	< 0.001	2.79 (1.50-5.19)	0.001
Cr ≥ 1 (mg/dL)	1.49 (0.87-2.55)	0.142	-	-
T-Bil ≥ 1 (mg/dL)	3.37 (1.85-6.15)	< 0.001	2.06 (0.89-4.75)	0.091
PT ≥ 12 (second)	1.22 (0.56-2.67)	0.621	1.36 (0.48-3.83)	0.564
Alb < 3.5 (g/dL)	1.77 (0.80-3.95)	0.16	1.22 (0.38-3.96)	0.739
Disease severity
FIB-4 score
1.45-3.25	Reference
< 1.45	0.12 (0.02-0.89)	0.039	Not estimable1
> 3.25	3.13 (1.84-5.31)	< 0.001	2.72 (1.51-4.90)	< 0.001
Comorbidities
Diabetes Mellitus	1.53 (0.98-2.39)	0.061	1.43 (0.84-2.43)	0.184
Hypertension	0.90 (0.58-1.40)	0.646	0.62 (0.37-1.06)	0.079
Dyslipidemia	0.77 (0.37-1.59)	0.475	0.71 (0.27-1.87)	0.493

¹Not estimable due to zero events (complete separation) in the fibrosis-4 < 1.45 group.

HR: Hazard ratio; aHR: Adjusted hazard ratio; 95%CI: 95% confidence interval; HCV: Hepatitis C virus; AFP: Alpha-fetoprotein; FIB-4: Fibrosis-4; Cr: Creatinine; T-Bil: Total bilirubin; PT: Prothrombin time; Alb: Albumin.

3-year biomarker associated with HCC development

For the longitudinal 3-year biomarkers, a sub-analysis was conducted in 251 patients who remained under observation at the 3-year time point. In univariable Cox models, third-year AFP ≥ 20 ng/mL was strongly associated with subsequent HCC (HR = 9.27, 95%CI: 4.80-17.87; P < 0.001). Third-year FIB-4 > 3.25 was also associated with higher risk (HR = 2.75, 95%CI: 1.54-4.93; P = 0.001). Third-year total bilirubin ≥ 1 mg/dL (HR = 2.92, 95%CI: 1.38-6.20; P = 0.005) and creatinine ≥ 1 mg/dL (HR = 1.81, 95%CI: 1.00-3.27; P = 0.049) showed a significant association. Sex and age at year 3 were not statistically significant in univariable landmark analyses.

In the multivariable model, third-year AFP ≥ 20 ng/mL remained independently associated with increased HCC risk (aHR = 6.74, 95%CI: 2.48-18.33; P < 0.001). Although the FIB-4 score > 3.25 showed a trend toward increased risk (aHR = 2.30, 95%CI: 0.75-7.03; P = 0.143), it did not reach statistical significance in the adjusted model. And the third-year FIB-4 < 1.45 was not estimable due to zero events in the category within the complete-case multivariable dataset (Table 3).

Table 3 Cox regression analysis of hepatocellular carcinoma relation to 3-years characteristic and biomarkers.

Variable	Univariate		Multivariate
	HR (95%CI)	P value	aHR (95%CI)	P value
Demographics
Sex (male)	1.32 (0.81-2.15)	0.27	1.72 (0.64-4.66)	0.282
Age (per 1-year increase)	1.02 (1.00-1.05)	0.098	1.01 (0.96-1.06)	0.78
Biomarkers
AFP ≥ 20 (ng/mL)	9.27 (4.80-17.87)	< 0.001	6.74 (2.48-18.33)	< 0.001
Cr ≥ 1 (mg/dL)	1.81 (1.00-3.27)	0.049	0.92 (0.35-2.39)	0.859
T-Bil ≥ 1 (mg/dL)	2.92 (1.38-6.20)	0.005	2.57 (0.96-6.88)	0.06
PT ≥ 12 (second)	1.76 (0.69-4.47)	0.233	-	-
Alb < 3.5 (g/dL)	1.70 (0.70-4.16)	0.243	-	-
Disease severity
FIB-4 score
1.45-3.25	Reference
< 1.45	0.14 (0.02-1.06)	0.057	Not estimable1
> 3.25	2.75 (1.54-4.93)	0.001	2.30 (0.75-7.03)	0.143

¹Not estimable due to zero events (complete separation) in the fibrosis-4 < 1.45 group.

HR: Hazard ratio; aHR: Adjusted hazard ratio; 95%CI: 95% confidence interval; HCV: Hepatitis C virus; AFP: Alpha-fetoprotein; FIB-4: Fibrosis-4; Cr: Creatinine; T-Bil: Total bilirubin; PT: Prothrombin time; Alb: Albumin.

Cumulative incidence of HCC by different FIB-4 groups

The cumulative incidence of HCC stratified according to different FIB-4 scores either at baseline or 3-year time point was illustrated in Figure 2. At baseline, the 5-, 10-, and 15-year cumulative risk of HCC in patients with FIB-4 index < 1.45 was 0%, 4.3%, and 4.3%, respectively, while they were 3.2%, 12.2%, and 31.9% in patients with FIB-4 index 1.45-3.25, and 14.2%, 40.2%, and 64.2% in those with FIB-4 index > 3.25, respectively. Using the 3-year FIB-4 index stratification, the corresponding 5-, 10-, and 15-year cumulative incidence was 3.0%, 3.0%, and 3.0% for FIB-4 index < 1.45; 7.0%, 23.4%, and 33.9% for FIB-4 index 1.45-3.25; and 19.6%, 48.8%, and 78.1% for FIB-4 > 3.25. Both log-rank tests showed clinical significance with P < 0.001.

Figure 2 The cumulative incidence of hepatocellular carcinoma stratified according to different fibrosis-4 scores either at baseline or 3-year time point. A: Cumulative incidence of hepatocellular carcinoma (HCC) by baseline fibrosis-4 (FIB-4) score. The 5-, 10-, and 15-year cumulative risk of HCC was 0%, 4.3%, and 4.3% for FIB-4 index < 1.45; 3.2%, 12.2%, and 31.9% in patients with FIB-4 index 1.45-3.25; and 14.2%, 40.2%, and 64.2% in those with FIB-4 index > 3.25, respectively; B: Cumulative incidence of HCC by third-year FIB-4 score. The 5-, 10-, and 15-year cumulative incidence of HCC was 3.0%, 3.0%, and 3.0% for FIB-4 index < 1.45; 7.0%, 23.4%, and 33.9% for FIB-4 index 1.45-3.25; and 19.6%, 48.8%, and 78.1% for FIB-4 > 3.25, respectively. FIB-4: Fibrosis-4.

Association between 3-Year FIB-4 dynamic trajectory and HCC development

For the impact of longitudinal fibrosis progression on HCC risk, patients were stratified into three groups based on their 3-year FIB-4 score trajectory: Improved (n = 39), stable (n = 159), and worsened (n = 45). The respective incidences of HCC were 20.5%, 28.9%, and 20.0%. The χ² test for trend across the three groups was P = 0.376, indicating HCC across the three trajectory groups having no difference. When further subdividing the stable group into three FIB-4 strata: Stable FIB-4 < 1.45, stable FIB-4 1.45-3.25, and stable FIB-4 > 3.25, resulting in five comparison groups overall. The incidences of HCC were 0% for stable FIB4 < 1.45, 15.2% for stable FIB4 1.45-3.25, and 42.4% for stable FIB4 > 3.25. The P value of five-level Pearson χ² test across these groups was P < 0.001. Using stable FIB-4 1.45-3.25 as the reference, patients with FIB-4 remaining > 3.25 had 4.1-fold increased risk of HCC (95%CI: 1.66-10.13, P = 0.002). Whereas, FIB-4 staying < 1.45 appeared protective but this did not reach statistical significance (HR = 0.12, 95%CI: 0.01-2.25, P = 0.157). Improved and worsened trajectories did not differ significantly from the reference group (Table 4).

Table 4 Pair-wise comparisons between five groups of 3-year fibrosis-4 trajectories and risk of hepatocellular carcinoma.

	HCC cases (n)	Incidence (%)	HR (95%CI)	P value
Stable 1.45-3.25	7/46	15.2	Reference	-
Stable < 1.45	0/21	0	0.12 (0.01-2.25)	0.157
Stable > 3.25	39/92	42.4	4.10 (1.66-10.13)	0.002
Improved	8/39	20.5	1.44 (0.47-4.40)	0.525
Worsened	9/45	20	1.39 (0.47-4.13)	0.55

Improved trajectory: Fibrosis-4 from > 3.25 to either 1.45-3.25 or < 1.45, or from 1.45-3.25 to < 1.45; Worsened trajectory: From < 1.45 to either 1.45-3.25 or > 3.25, or from 1.45-3.25 to > 3.25. HR: Hazard ratio; 95%CI: 95% confidence interval; HCC: Hepatocellular carcinoma.

In the absolute change of value in FIB-4 score, using landmark Cox proportional hazards model for events occurring after year 3, each 1-point increase in ΔFIB-4 was associated with a significantly higher risk of subsequent HCC development (HR = 1.19, 95%CI: 1.04-1.36; P = 0.0107). Then, using a ΔFIB-4 rise of ≥ 1 over the three year as the cut-off point, the incidence of HCC was 37.7% (26/69) among patients with ΔFIB-4 ≥ 1 compared with 21.3% (37/174) among those with ΔFIB-4 < 1, corresponding to a HR of 2.56 (95%CI: 1.53-4.28, P < 0.001). In sensitivity analyses adjusting for age and sex, the association between ΔFIB-4 ≥ 1.0 and subsequent HCC was consistent and slightly strengthened (aHR = 2.67, 95%CI: 1.58-4.50; P < 0.001). In a categorical analysis (decrease: ΔFIB-4 ≤ -1.0, stable: -1.0 < ΔFIB-4 < +1.0, and increase: ΔFIB-4 ≥ +1.0), the risk of HCC development was 26.5% (13/49), 19.2% (24/125), and 37.7% (26/69), respectively. The three-level χ² test for trend across the three groups was P = 0.019. Relative to the stable group, patients in the increase group exhibited a significantly higher risk of developing HCC (HR = 2.54, 95%CI: 1.32-4.92, P = 0.005), whereas the decrease group did not differ statistically (HR = 1.52, 95%CI: 0.70-3.30, P = 0.291). The cumulative incidence of HCC stratified according to different ΔFIB-4 group was illustrated in Figure 3. The 5-, 10-, and 15-year cumulative incidence of HCC was 4.4%, 29.9%, and 72.7% for increase group; 9.2%, 30.7%, and 35.7% for decrease group; and 0%, 12.1%, and 27.1% for stable group, respectively. The log-rank tests showed clinical significance with P < 0.001.

Figure 3 Cumulative incidence of hepatocellular carcinoma by change in fibrosis-4 group. Increase [change in fibrosis-4 (FIB-4) ≥ +1.0], decrease (ΔFIB-4 ≤ -1.0), stable (-1.0 < ΔFIB-4 < +1.0). The 5-, 10-, and 15-year cumulative incidence of hepatocellular carcinoma was 4.4%, 29.9%, and 72.7% for increase group; 9.2%, 30.7%, and 35.7% for decrease group; and 0%, 12.1%, and 27.1% for stable group, respectively.

DISCUSSION

In this retrospective cohort study of 271 treatment-naïve CHC patients, both baseline FIB-4 burden and AFP level were strongly associated with subsequent HCC, and these signals remained robust when reassessed at year 3 using a landmark design. Although year3 FIB4 categories alone did not significantly predict incident HCC, analyses of FIB4 trajectories demonstrated that patients with worsening fibrosis over three years had a substantially higher risk of HCC compared with those whose FIB4 scores remained stable. These findings underscore the importance of not only initial biomarker and baseline characteristic assessment but also continuous monitoring for identifying high risk patients. Accordingly, patients presented with a high baseline FIB-4 index or evidence of worsening FIB-4 scores during follow-up warrant intensified surveillance for HCC.

In the modern era, DAAs have been approved worldwide for the treatment of HCV to improve clinical outcomes and reduce transmission[17]. A notable characteristic of our cohort is the treatment-naïve patients managed under long-term follow-up. This observation must be interpreted within the historical context of the study data collection period (1999-2015), which predates the era of universal DAA reimbursement in Taiwan[18]. Many patients in our cohort declined treatment due to advanced age or fear of severe side effects, opting instead for a “watchful waiting” strategy. These patient-level factors align with well-documented barriers to HCV therapy[9,10]. Consequently, this study captures a unique “natural history” cohort unconfounded by antiviral therapy, offering valuable insights into fibrosis progression and HCC risk that are increasingly difficult to observe in the modern DAA era.

Consistent with established literature identifying older age, male sex, metabolic comorbidities, and coinfections as key risks of hepatocarcinogenesis[19,20], our multivariate analysis confirmed that male sex acts as an independent predictor of HCC. However, the most potent driver of risk in our cohort was liver fibrosis burden. The strong association between a baseline FIB-4 > 3.25 and HCC development likely reflects a dual mechanism: It serves not only as a surrogate for advanced fibrosis/cirrhosis but also indicates a biologically permissive hepatic microenvironment that favors carcinogenesis[21]. Our findings align with prior studies linking elevated FIB-4 to HCC risk and liver-related mortality, extending these observations to a treatment-naïve natural history cohort[13,22-24]. Our results demonstrate a clear, stepwise gradient in long-term outcomes, with 5-, 10-, and 15-year cumulative HCC incidence progressively increasing across the < 1.45, 1.45-3.25, and > 3.25 tiers. Notably, patients with a baseline FIB-4 > 3.25 exhibited a 15-year cumulative incidence that was 15-fold higher than those with FIB-4 < 1.45 (which remained low at approximately 4%). Complementing the fibrosis-driven risk, an AFP threshold of 20 ng/mL is widely used in surveillance algorithms and has been shown to be associated with HCC risk[25]; this threshold remained a significant predictor of HCC in our analysis. This distinct stratification supports a risk-adapted surveillance strategy: Patients with baseline FIB-4 below 1.45 may be candidates for less-intensive monitoring, whereas those with FIB-4 > 3.25 or elevated AFP warrant rigorous follow-up.

Longitudinal monitoring of albumin, AFP, FIB-4 index, and liver stiffness provides better risk prediction than baseline-only measurements in patients with HCV who achieved SVR with DAA[14,26]. These investigations have employed landmark intervals ranging from one to three years for biomarkers follow-up and HCC risk assessment[14,27]. In our 3-year landmark analysis, the vast majority of participants had complete data at the threeyear timepoint, preserving statistical power; moreover, the median interval from enrolment to HCC diagnosis was 7.45 years, making year 3 a clinically relevant intermediate point for pre-HCC biomarker review. We excluded patients with less than 3 years of follow-up who remained HCC-free; this exclusion criterion was methodologically necessary to ensure valid data points for calculating longitudinal dynamic changes (ΔFIB-4), aligning with established methodologies in previous study focusing on HCC risk stratification in treatment-naïve CHC patients[13]. Indeed, on univariate analysis a three-year FIB-4 > 3.25 was associated with a significantly higher HCC risk. After multivariable adjustment, only AFP ≥ 20 ng/mL retained independent significance. Even so, the cumulative incidence gradient across FIB-4 strata persisted to 15 years (78% in > 3.25 vs 3% in < 1.45), aligning with the concept that the duration of advanced fibrosis and cirrhosis are dominant driver and predictors of HCC[14]. Thus, reassessing the AFP level at year 3 offers valuable prognostic utility for HCC prediction in addition to FIB-4 score[28].

The variation of FIB-4 score from baseline is also informative and can serve as valuable indicators for stratifying HCC risk in patients with CHC receiving treatment of DAAs[29]. A prospective study conducted in South Korea reported that CHC patients, regardless of treatment status, who exhibited a ΔFIB-4 index/year ≥ 0.5, were at an increased risk of cirrhosis, liver decompensation, HCC, and mortality[30]. As for treatment-failure cases, ΔFIB-4 index/year using a cut-off value of 0.4 correlates cumulative incidence of fibrosis progression to cirrhosis[31]. In our cohort, a three-year rise of FIB-4 index ≥ 1 identified a subgroup with 2.67-fold higher hazard of the 3-year HCC risk compared with those with smaller rises or declines. This threshold is comparable with previous studies and simple to operationalize from routine laboratory data for high risk screening. In our categorical analysis, patients with a ≥ 1.0 increase in FIB-4 demonstrated the highest risk burden, with cumulative HCC incidences reaching 72.7% at 15 years. These results highlight the importance of examining time-to-event outcomes rather than relying solely on crude incidence at a single follow-up point[32].

The longitudinal fibrosis dynamics by category transitions dividing into five groups confirmed that persistently high FIB-4, consistent with advanced fibrosis, dominates subsequent cancer risk. In contrast, those who remained FIB-4 < 1.45 over three years experienced no HCC events. These findings indicate that serial assessments of the FIB4 index may serve as a prognostic marker for identifying a highrisk cohort HCC, who could benefit from more intensive surveillance strategies[33]. Whereas the FIB-4 index remaining < 1.45 over three years identifies a low-risk subset suitable for less intensive surveillance intervals. Taken together, the magnitude of change of the FIB-4 index and between group change are both informative for risk assessment of HCC. This correlation is explained by fibrosis progression, which, even in the absence of overt clinical cirrhosis, involves excessive accumulation of extracellular matrix proteins causing chronic liver injury, thereby inducing oncogenic changes in the hepatic microenvironment[21].

This study demonstrates that serial changes in FIB-4, along with longitudinal biomarker follow-up, provide prognostic information independent of baseline fibrosis stage. The long follow-up in a treatment-naïve cohort captured incident HCC events and therefore reflects the natural history of CHC before exposure to antiviral therapy. From a clinical standpoint, our data support a risk-adapted approach: (1) Patients with FIB-4 persistently > 3.25 and/or ΔFIB-4 ≥ 1.0 over three years should be prioritized for antiviral therapy (where applicable) and intensified HCC surveillance; (2) Those persistently < 1.45 appear to have very low risk and may be suitable for less-intensive surveillance intervals; and (3) Individuals in the mid-tier (1.45-3.25) warrant continued monitoring given their non-trivial long-term risk. Because FIB-4 and AFP are low-cost and widely available[34], this strategy is particularly attractive in settings with limited DAA uptake.

There are limitations that should be noted. First, we acknowledge the potential for selection bias and immortal time bias inherent in excluding patients with less than 3 years of follow-up who remained HCC-free. However, as noted, this was necessary to construct the dynamic model. Therefore, our findings should be interpreted as reflecting risk profiles specifically among patients who survive the initial 3-year natural history phase. Second, the patient cohort was derived from a single center, which may introduce regional selection bias. Third, the optimal cut-off values for longitudinal ΔFIB-4 were identified using a data-driven approach within the same study cohort. Given the exploratory nature of this study and the rarity of long-term untreated cohorts, our primary aim was to demonstrate the concept that dynamic changes matter, rather than to establish a definitive clinical threshold. Therefore, the specific ΔFIB-4 cut-offs proposed here should be regarded as preliminary. Further validation in independent, larger, and multi-center cohorts is essential to confirm their generalizability and clinical applicability before routine implementation.

CONCLUSION

In conclusion, our findings demonstrate that the longitudinal assessment of non-invasive fibrosis scores, specifically the dynamic worsening of FIB-4, combined with AFP offers additional prognostic value to baseline measures alone. This accessible and cost-effective strategy is particularly valuable in resource-limited settings or regions where universal DAA coverage remains restricted. By identifying treatment-naïve patients with worsening fibrosis trajectories, clinicians can better prioritize scarce resources for intensified surveillance and therapeutic intervention, ultimately optimizing long-term outcomes in this vulnerable population.