Abstract
Objective: HCV infection is frequently asymptomatic, with current diagnosis relying mainly on costly and less accessible HCV RNA testing. While HCV-Ab and biochemical markers exhibit suboptimal diagnostic performance, whether machine learning can enhance their accuracy remains unclear. Methods: This study is a retrospective study, which included data from 179 patients whose HCV-Ab levels were greater than 1.00 S/CO to explore the relationship between HCV-Ab, biochemical indicators, and HCV infection. Univariate logistic regression and restricted cubic splines (RCS) were employed to explore these associations. Machine learning integrated HCV-Ab and biochemical indicators to predict early HCV infection (undiagnosed chronic cases), with validation conducted using receiver operating characteristic curve (ROC) analysis. The machine learning approach randomly divided study participants into training and test sets at a 5:5 ratio, with the training set being used for variable selection and model construction. Results: After full adjustment, TP showed no significant association with HCV infection. Restricted cubic spline (RCS) analysis revealed nonlinear relationships between HCV-Ab, ALT, AST, mAST, GGT, A/G and HCV infection. HCV-Ab exhibited an inflection point at 11.17 (below: OR = 1.04 per unit increase; above: no association). Similar threshold patterns were observed for ALT, AST, mAST and GGT. The integrated HCV-Ab and biochemical marker model achieved excellent predictive performance (AUC = 0.977). Conclusion: TP exhibited a linear association with HCV infection, whereas HCV-Ab, ALT, AST, mAST and GGT showed nonlinear associations with distinct threshold effects. Early prediction of HCV infection using these indicators represents a cost-effective strategy.
Keywords: HCV-RNA, HCV-Ab, biochemical indicators, machine learning, restricted cubic splines
Introduction
Hepatitis C virus (HCV), a single-stranded RNA virus of the Flaviviridae family, primarily targets hepatocytes and leads to both acute and chronic hepatitis [1,2]. Globally, approximately 58 million individuals are chronically infected with HCV, with approximately 399,000 deaths annually attributed to HCV-related complications such as cirrhosis and hepatocellular carcinoma (HCC) [3-6]. The primary modes of transmission include blood exposure (e.g., transfusion, needle sharing, unsafe medical procedures), vertical mother-to-child transmission, and sexual contact. High-risk populations include intravenous drug users, recipients of unscreened blood products, and healthcare workers exposed to blood. Undiagnosed early HCV infections frequently progress to chronic hepatitis (55%-85% of acute cases), subsequently leading to hepatic fibrosis, cirrhosis, and HCC [7-9]. Studies indicate that 15%-30% of chronic HCV patients develop cirrhosis within two decades, with an annual progression rate of 1%-5% to HCC [10-12]. HCV infection is also associated with extrahepatic manifestations, including metabolic abnormalities, cardiovascular diseases, and lymphoma [13-17]. The introduction of direct-acting antivirals (DAAs) has significantly improved treatment outcomes, achieving cure rates exceeding 95% [18,19]. The gold standard for HCV diagnosis is the detection of HCV-RNA [20], which confirms active infection and quantifies viral load. However it exhibits inherent limitations [21]: 1) High intrinsic test cost: The cost of HCV RNA test reagents and specialized equipment (e.g., PCR machines) is substantially higher than that of HCV antibody tests or routine biochemical assays. In resource-limited settings, the cost per test can pose a significant economic barrier. 2) Infrastructure and personnel costs: RNA testing generally requires specialized molecular biology laboratory facilities (e.g., strict containment, specialized equipment, stable power/cold chain) and relies on trained technical personnel. Establishing and maintaining such laboratories is expensive. 3) Accessibility challenges: Molecular diagnostics laboratories are predominantly concentrated in urban centers or large tertiary hospitals, resulting in severely limited accessibility in remote, rural, or resource-poor areas [22]. In contrast, HCV antibody (HCV-Ab) and biochemical indicators assays are cost-effective and widely available in primary care facilities [23]. HCV-Ab refers to antibody produced by activated immune cells in response to HCV infection. This antibody persists in the human body and is typically detected in individuals who have achieved viral clearance (either through treatment or spontaneous resolution) as well as those with chronic active infections. The HCV-Ab test offers significant advantages: low cost, operational simplicity, rapid results, and high accessibility. The hepatitis C virus primarily infects hepatocytes, inducing hepatic injury and subsequent liver dysfunction [24]. Markedly elevated levels of biochemical indicators may indicate active liver injury and, combined with other factors, can heighten suspicion of active infection or disease activity. The combined utilization of serological HCV-Ab and biochemical detection may enhance diagnostic accuracy and improve clinical evaluation of HCV infection status. This study analyzed biochemical parameters and HCV-RNA results from 179 patients with HCV-Ab levels greater than 1.00 S/CO to investigate the dynamic changes in HCV-Ab and biochemical profiles. Utilizing machine learning techniques, we evaluated the predictive performance of combining HCV-Ab with biochemical indicators for early diagnosis of HCV.
Methods
Data collection and processing
This is a retrospective study. The primary data were collected from patients whose HCV-Ab level was greater than 1.00 S/CO, at Shidong Hospital Affiliated to University of Shanghai for Science and Technology between 2019 and 2024, with all data obtained from medical records. The Ethics Committee of Shanghai Shidong Hospital approved the study protocol (Ethics approval report ID: 2025-031-01). For all participants, biochemical indicators data from the first visit were screened and used as features for model construction. Patients with other hepatitis infections or liver/kidney impairment due to other etiologies were excluded. Detection of HCV-RNA is the gold standard for the diagnosis of HCV infection.
Specimen collection and laboratory testing
Venous blood samples were collected using sterile techniques. For serum samples, a standard serum separator tube without anticoagulants was utilized, allowing the blood to clot at room temperature for 30 minutes prior to centrifugation. The samples were then centrifuged at 3,500 RPM for 10 minutes to separate the serum.
The Abbott Alinity i diagnostic kit (America) was employed for quantitative detection of HCV antibody in human serum using the chemiluminescent microparticle immunoassay (CMIA) method according to the manufacturer’s instructions. For HCV-RNA detection, the in vitro nucleic acid amplification kit (Sansure Biotech Inc., Changsha, China) was utilized, with all procedures strictly adhering to the manufacturer’s protocols. Alanine Aminotransferase (ALT), Aspartate Aminotransferase (AST), mitochondrial aspartate aminotransferase isoenzyme (mAST), Alkaline Phosphatase (ALP), Gamma-Glutamyl Transferase (GGT), Total Bilirubin (TB), Conjugated Bilirubin (CB), Total Protein (TP), Albumin (ALB), Albumin-to-Globulin ratio (A/G), Blood Urea Nitrogen (BUN), Creatinine (Cr), Uric Acid (UA), Glomerular Filtration Rate (GFR), and Glucose (GLU) were quantified using a clinical chemistry analyzer (Beckman5800 Chemistry System). All the reagents were provided by the respective manufacturers as part of pre-packaged kits, ensuring consistency and reliability across tests. All tests were performed in accordance with rigorous quality control protocols.
Study methods
First, trend analysis of multivariate logistic regression was used to assess the association between HCV-Ab, biochemical indicators and HCV infection. TP was introduced into the logistic regression model as a continuous variable, and the results were expressed as odds ratio (OR) and 95% confidence interval (95% CI). Three models were constructed by adjusting for different confounding variables. In model 1 was unadjusted for variables. In model 2, confounders including gender and age were adjusted. Model 3 was further adjusted for HCV-Ab, ALT, AST, mAST, GGT, A/G. To further explore the potential nonlinear relationship between HCV-Ab, biochemical indicators and HCV infection, restricted cubic spline (RCS) regression analyses were performed. Likelihood ratio tests were used to detect nonlinearity. The threshold effect of HCV-Ab and biochemical indicators on HCV infection risk was further analyzed by a two-stage linear regression model.
Machine learning algorithms to construct HCV infection prediction model
In this study, multiple machine learning methods were used to investigate the application and predictive value of HCV-Ab and biochemical indicators in the diagnosis of HCV infection. The study participants were randomly divided into a train set and a test set in a ratio of 5:5. The train set was used to screen the variables and construct the model. The test set was used to evaluate the performance of the final model. The diagnostic performance of each model was evaluated by calculating the area under the curve (AUC) value. The optimal model was selected based on the AUC value to plot the receiver operating characteristic curve (ROC) [25].
Statistical analysis
Continuous variables with non-normal distributions were assessed using the Mann-Whitney U test and expressed as medians (interquartile range [Q1, Q3]). Categorical variables were compared using the chi-square test and reported as counts (percentages). All statistical analyses were performed using R software (version 4.4.0), and statistical significance was defined as P < 0.05.
Results
Baseline clinical characteristics of participants
This study included a total of 179 participants, with 130 cases (72.63%) in the HCV control group and 49 cases (27.37%) in the HCV disease group. Statistically significant differences (P < 0.05) were observed between the two groups in Age, HCV-Ab, ALT, AST, mAST, GGT, TP, A/G ratio, and Gender; whereas no significant differences (P > 0.05) were found in ALP, TB, CB, ALB, BUN, Cr, UA, GFR, or GLU (Table 1).
Table 1.
Clinical characteristics
| Variables | Total (n = 179) | Control (n = 130) | Disease (n = 49) | Statistic | P |
|---|---|---|---|---|---|
| Age, M (Q1, Q3) | 63.00 (55.50, 69.00) | 64.50 (56.25, 69.75) | 60.00 (54.00, 64.00) | Z = -2.22 | 0.026 |
| HCV-Ab, M (Q1, Q3) | 5.56 (1.54, 14.18) | 2.27 (1.31, 8.27) | 13.84 (11.85, 15.45) | Z = -7.03 | < .001 |
| ALT, M (Q1, Q3) | 21.00 (14.50, 50.00) | 18.00 (13.00, 25.75) | 70.00 (48.00, 135.00) | Z = -7.75 | < .001 |
| AST, M (Q1, Q3) | 26.00 (19.00, 47.00) | 22.00 (18.00, 28.00) | 68.00 (37.00, 99.00) | Z = -7.55 | < .001 |
| mAST, M (Q1, Q3) | 3.00 (2.70, 5.25) | 3.00 (2.40, 4.00) | 5.00 (3.00, 8.20) | Z = -3.84 | < .001 |
| ALP, M (Q1, Q3) | 82.00 (68.50, 100.00) | 82.00 (68.50, 102.75) | 82.00 (69.00, 98.00) | Z = -0.26 | 0.797 |
| GGT, M (Q1, Q3) | 30.50 (21.00, 63.00) | 30.00 (18.00, 46.75) | 44.00 (30.50, 103.00) | Z = -3.80 | < .001 |
| TB, M (Q1, Q3) | 14.00 (10.90, 17.30) | 14.00 (11.20, 16.45) | 14.00 (10.20, 19.00) | Z = -0.75 | 0.454 |
| CB, M (Q1, Q3) | 2.80 (2.30, 3.90) | 2.80 (2.23, 3.50) | 2.80 (2.60, 4.90) | Z = -1.53 | 0.127 |
| TP, M (Q1, Q3) | 69.00 (64.55, 73.40) | 69.00 (63.62, 72.72) | 70.40 (65.90, 75.00) | Z = -2.32 | 0.020 |
| ALB, M (Q1, Q3) | 39.30 (35.75, 41.95) | 39.30 (35.70, 42.18) | 39.20 (36.10, 40.70) | Z = -0.51 | 0.611 |
| A/G, M (Q1, Q3) | 1.33 (1.18, 1.50) | 1.33 (1.19, 1.55) | 1.26 (1.16, 1.33) | Z = -2.44 | 0.015 |
| BUN, M (Q1, Q3) | 5.35 (4.60, 6.95) | 5.35 (4.48, 6.88) | 5.35 (4.80, 7.30) | Z = -0.50 | 0.619 |
| Cr, M (Q1, Q3) | 68.60 (58.10, 89.45) | 68.60 (58.05, 88.20) | 68.60 (58.30, 91.00) | Z = -0.02 | 0.987 |
| UA, M (Q1, Q3) | 339.30 (270.00, 399.65) | 339.30 (269.50, 395.70) | 339.30 (287.60, 415.50) | Z = -0.43 | 0.669 |
| GFR, M (Q1, Q3) | 97.80 (74.25, 107.30) | 97.80 (73.77, 107.12) | 97.80 (78.70, 111.40) | Z = -0.63 | 0.528 |
| GLU, M (Q1, Q3) | 5.36 (4.99, 6.08) | 5.36 (4.97, 5.96) | 5.36 (5.16, 6.26) | Z = -1.20 | 0.230 |
| Gender, n (%) | Χ2 = 5.65 | 0.017 | |||
| Female | 88 (49.16) | 71 (54.62) | 17 (34.69) | ||
| Male | 91 (50.84) | 59 (45.38) | 32 (65.31) |
Z, Mann-Whitney test, χ2, Chi-square test; M (Q1, Q3), Median (1st Quartile, 3st Quartile); n (%), numbers (percentages); HCV-Ab, hepatitis C virus antibody; ALT, alanine aminotransferase; AST, aspartate aminotransferase; mAST, mitochondrial aspartate aminotransferase isoenzyme; ALP, alkaline phosphatase; GGT, gamma-glutamyl transferase; TB, total bilirubin; CB, conjugated bilirubin; TP, total protein; ALB, albumin; A/G, albumin-to-globulin ratio; BUN, blood urea nitrogen; Cr, creatinine; UA, uric acid; GFR, glomerular filtration rate; GLU, glucose.
A linear relationship between TP and the risk of HCV infection
Three models were constructed by adjusting for different confounding variables to evaluate the associations between TP with the risk of HCV infection. After adjusting for all confounding variables, in the final model, the relationship between TP (OR = 1.05, 95% CI: 0.98-1.13, P = 0.149) and HCV infection was not significant. However, the p-value showed a trend toward significance (Table 2).
Table 2.
The relationship between TP and the risk of HCV infection
| Variables | Model 1 | Model 2 | Model 3 | |||
|---|---|---|---|---|---|---|
|
|
|
|
||||
| OR (95% CI) | P | OR (95% CI) | P | OR (95% CI) | P | |
| TP | 1.06 (1.01-1.11) | 0.030 | 1.06 (1.01-1.11) | 0.032 | 1.05 (0.98-1.13) | 0.149 |
| TP (median) | ||||||
| 1 | 1.00 (Reference) | 1.00 (Reference) | 1.00 (Reference) | |||
| 2 | 2.07 (1.03-4.15) | 0.042 | 2.23 (1.06-4.69) | 0.035 | 1.79 (0.58-5.52) | 0.314 |
| P for trend | 0.042 | 0.035 | 0.314 | |||
Model 1: Crude, Model 2: Adjust: gender, age, Model 3: Adjust: gender, age, HCV-Ab, ALT, AST, mAST, GGT, AG. OR, Odds Ratio; CI, Confidence Interval; TP, total protein; HCV-Ab, hepatitis C virus antibody; ALT, alanine aminotransferase; AST, aspartate aminotransferase; mAST, mitochondrial aspartate aminotransferase isoenzyme; GGT, gamma-glutamyl transferase; A/G, albumin-to-globulin ratio.
A nonlinear relationship and threshold effect were observed between HCV-Ab, partial biochemical indicators, and the risk of HCV infection
Comparative analysis was performed of HCV-Ab and biochemical indicators between the HCV Control group and Disease group (Figure 1A-F). To further ensure the robustness of the results, the potential nonlinear relationship between HCV-Ab, partial biochemical indicators and the risk of HCV infection was examined. In the RCS regression model, after adjusting for all confounding factors, significant nonlinear associations were observed between HCV-Ab, ALT, AST, mAST, GGT, A/G and HCV infection (nonlinear P < 0.05) (Figure 2A-F).
Figure 1.
Comparative analysis of HCV-Ab and biochemical indicators between HCV Control group and Disease group. *P < 0.05; ***P < 0.001. A. HCV-Ab; B. ALT; C. AST; D. mAST; E. GGT; F. A/G. HCV-Ab, hepatitis C antibody; ALT, alanine aminotransferase; AST, aspartate aminotransferase; mAST, mitochondrial aspartate aminotransferase isoenzyme; GGT, gamma-glutamyl transferase; A/G, albumin-to-globulin ratio.
Figure 2.
Restricted cubic spline analyses the association of HCV-Ab and biochemical indicators (A. HCV-Ab; B. ALT; C. AST; D. mAST; E. GGT; F. A/G) with HCV infection. HCV-Ab, hepatitis C antibody; ALT, alanine aminotransferase; AST, aspartate aminotransferase; mAST, mitochondrial aspartate aminotransferase isoenzyme; GGT, gamma-glutamyl transferase; A/G, albumin-to-globulin ratio.
In Table 3, further analysis revealed a threshold effect in the association between HCV-Ab and the risk of HCV infection (P for likelihood ratio test < 0.001), with an inflection point at 11.17. When HCV-Ab were below 11.17, a positive correlation exists between HCV-Ab and the risk of HCV infection (OR = 2.04, 95% CI: 1.34-3.10, P < 0.001). Each unit increase was associated with a 1.04-fold increase in the risk of HCV infection. However, when HCV-Ab exceeded 11.17, no significant association with the risk of HCV infection was observed (OR = 0.77, 95% CI: 0.59-1.01, P = 0.057). Additionally, a threshold effect was identified for ALT (P for likelihood ratio test = 0.017) with an inflection point at 54.00, and for AST (P for likelihood ratio test = 0.007) with an inflection point at 39.00, and for mAST (P for likelihood ratio test = 0.005) with an inflection point at 7.00. Overall, a positive correlation was observed between GGT and HCV infection (OR = 1.00, 95% CI: 1.00-1.01, P = 0.026). A threshold effect was identified for GGT (P for likelihood ratio test = 0.017). However, no significant association was found when GGT was below or above 28.00.
Table 3.
The threshold effect of HCV-Ab and biochemical indicators on HCV infection was analyzed using a two-stage phased regression model
| Variables | Models | Adjusted OR (95% CI) | P |
|---|---|---|---|
| HCV-Ab (S/CO) | Model 1 Fitting model by standard linear regression | 1.30 (1.20-1.42) | < .001 |
| Model 2 Fitting model by two-piecewise linear regression | |||
| Inflection point | 11.17 | ||
| < 11.17 | 2.04 (1.34-3.10) | < .001 | |
| ≥ 11.17 | 0.77 (0.59-1.01) | 0.057 | |
| P for likelihood test | < .001 | ||
| ALT (U/L) | Model 1 Fitting model by standard linear regression | 1.04 (1.03-1.06) | < .001 |
| Model 2 Fitting model by two-piecewise linear regression | |||
| Inflection point | 54.00 | ||
| < 54.00 | 1.09 (1.04-1.14) | < .001 | |
| ≥ 54.00 | 1.02 (1.00-1.04) | 0.112 | |
| P for likelihood test | 0.017 | ||
| AST (U/L) | Model 1 Fitting model by standard linear regression | 1.05 (1.03-1.07) | < .001 |
| Model 2 Fitting model by two-piecewise linear regression | |||
| Inflection point | 39.00 | ||
| < 39.00 | 1.16 (1.05-1.28) | 0.004 | |
| ≥ 39.00 | 1.02 (1.00-1.04) | 0.086 | |
| P for likelihood test | 0.007 | ||
| mAST (U/L) | Model 1 Fitting model by standard linear regression | 1.06 (1.01-1.11) | 0.016 |
| Model 2 Fitting model by two-piecewise linear regression | |||
| Inflection point | 7.00 | ||
| < 7.00 | 1.56 (1.14-2.13) | 0.006 | |
| ≥ 7.00 | 1.00 (0.93-1.07) | 0.983 | |
| P for likelihood test | 0.005 | ||
| GGT (U/L) | Model 1 Fitting model by standard linear regression | 1.00 (1.00-1.01) | 0.026 |
| Model 2 Fitting model by two-piecewise linear regression | |||
| Inflection point | 28.00 | ||
| < 28.00 | 1.14 (0.99-1.30) | 0.071 | |
| ≥ 28.00 | 1.00 (1.00-1.01) | 0.290 | |
| P for likelihood test | 0.017 |
OR, odds ratio; CI, confidence interval; HCV-Ab, hepatitis C antibody; ALT, alanine aminotransferase; AST, aspartate aminotransferase; mAST, mitochondrial aspartate aminotransferase isoenzyme; GGT, gamma-glutamyl transferase.
Prediction of HCV infection risk by ALT, AST and HCV-Ab changes
A total of 179 samples were randomly divided into train and test sets in a 5:5 ratio. No statistically significant differences were observed between the two groups. The model developed with Gradient Boosting Machine (GBM) demonstrated the best performance in predicting the risk of HCV infection, achieving an AUC of 0.997 in the train set, 0.953 in the test set and 0.977 for the total samples (Figure 3A-C). This performance significantly outperformed that of individual indicators, suggesting that the scoring model can effectively identify the risk of developing HCV infection during early infection.
Figure 3.
Machine learning algorithms to construct HCV infection prediction models. A. The ROC curves of HCV-Ab, ALT and AST (GBM) model and the HCV infection prediction model in the train group. B. The ROC curves of HCV-Ab, ALT and AST (GBM) model and the HCV infection prediction model in the test group. C. The ROC curves of HCV-Ab, ALT and AST (GBM) model and the HCV infection prediction model in the total group.
Discussion
This study established a Gradient Boosting Machine (GBM)-based predictive model integrating serological and biochemical indicators using clinical data from 179 patients with high HCV-Ab titers (S/CO > 1.00). The multi-parameter model demonstrated superior diagnostic accuracy compared to single-marker approaches (AUC = 0.977), offering new insights for optimizing early HCV screening strategies.
As the cornerstone of HCV screening, HCV-Ab testing provides rapid and cost-effective population-level surveillance. However, its limitations are notable: 1) The prolonged seroconversion window (2-6 months) may delay early diagnosis; 2) 15%-30% of virologically cured patients maintain persistent antibodies, complicating differentiation between active and resolved infections [22,26]; 3) F false-positive results may occur due to rheumatoid factor interference or immunosuppressive conditions [27,28]. Our threshold effect analysis revealed a nonlinear relationship between HCV-Ab titers and infection probability. At HCV-Ab < 11.17 S/CO, each unit increase correlated with 30% elevated infection risk (OR = 1.3, 95% CI: 1.20-1.42, P < 0.001), whereas no significant correlation was observed above this threshold. This phenomenon aligns with antigen-antibody complex dynamics: subthreshold titers (< 11.17 S/CO) potentially indicate insufficient neutralizing capacity, allowing viral replication, whereas suprathreshold levels (> 11.17 S/CO) may represent either effective immune containment or chronic infection states - a phenomenon analogous to “antigen trap” mechanism observed in HIV affinity maturation [29,30].
Chronic HCV infection induces progressive hepatic injury through inflammation-fibrosis cascades, as evidenced by perturbations in serum biomarkers including TP, ALB, ALT, AST, mAST, ALP, GGT, TB, and CB [31-34]. Our threshold analysis identified nonlinear associations for liver enzymes: ALT < 54 U/L (OR = 1.04, P < 0.001), AST < 39 U/L (OR = 1.05, P < 0.001), and mAST < 7 U/L (OR = 1.06, P = 0.016) showed positive correlations with infection risk, with diminishing effects beyond these cutoffs. This may reflect progression from early to late liver injury [35-37]. While GGT showed statistical association (P = 0.026), its nonspecific elevation in alcoholic liver disease and cholestatic conditions limits diagnostic specificity. TP showed linear association in unadjusted models (P = 0.030), but significance attenuated after multivariable adjustment (Model 3: P = 0.149), likely confounded by compensatory hepatic synthesis mechanisms.
Conventional linear models inadequately capture complex biomarker interactions. Our GBM-based machine learning model demonstrated robust performance across training, testing, and pooled datasets, outperforming conventional methods through automated feature engineering and nonlinear relationship modeling.
Several limitations should be acknowledged. First, sample size constraints necessitate multicenter validation for generalizability. Second, exclusion of emerging markers like HCV core antigen (HCV-cAg), detectable within 7-10 days post-infection, could address HCV-Ab’s window period limitations [38,39]. Future studies should explore “HCV-Ab + ALT + HCV-cAg” triage protocols integrated with deep learning algorithms for dynamic prediction. Third, the inclusion criterion of selecting patients with HCV-Ab > 1.00 S/CO may introduce potential selection bias. Although this threshold facilitates the identification of suspected infections with definitive serological evidence, it may skew the results toward reflecting clinical characteristics of patients with elevated HCV-Ab levels, potentially underrepresenting populations with borderline antibody values or those in early seroconversion phases. This limitation could compromise the applicability of our conclusions to broader hepatitis C virus-infected populations, particularly in clinical scenarios involving equivocal serological status. Lastly, all sample data were derived from a single-center cohort. Despite rigorous standardization of data collection protocols, the generalizability of findings may be constrained by regional homogeneity in healthcare practices, demographic profiles, and diagnostic-therapeutic expertise. To enhance external validity and elucidate the mechanistic associations between HCV-Ab dynamics and clinical outcomes, future investigations should employ multi-center, large-scale designs incorporating patients with diverse antibody levels and geographical distributions. Such efforts will enable systematic validation of our observations while advancing understanding of HCV serological evolution in relation to disease progression.
This study proposes two optimization pathways for resource-limited settings: 1) A stepwise “HCV-Ab screening → liver function retesting” cascade to reduce unnecessary HCV-RNA testing; 2) Portable GBM model deployment for real-time risk assessment in primary care. Furthermore, patients with mild ALT elevation (40-80 U/L) and HCV-Ab positivity should prioritize antiviral therapy to mitigate fibrosis progression.
Acknowledgements
This work was supported by the Key Disciplines in Yangpu District (No. 22YPZB03).
Disclosure of conflict of interest
None.
References
- 1.Di Stasio D, Guida A, Romano A, Petruzzi M, Marrone A, Fiori F, Lucchese A. Hepatitis C virus (HCV) infection: pathogenesis, oral manifestations, and the role of direct-acting antiviral therapy: a narrative review. J Clin Med. 2024;13:4012. doi: 10.3390/jcm13144012. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Alzahrani N. Hepatitis C virus, insulin resistance, and diabetes: a review. Microbiol Immunol. 2022;66:453–459. doi: 10.1111/1348-0421.13023. [DOI] [PubMed] [Google Scholar]
- 3.Magri A, Manfredi GF, Smirne C, Pigni S, Burlone ME, Bellan M, Vercellino N, Minisini R, Pirisi M. Impact of age and sex on viral load in hepatitis C virus infection. Viruses. 2024;17:21. doi: 10.3390/v17010021. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.El-Emshaty HM, Saad EA, Gouida MS, Elshahawy ZR. Associations between CD133, CK19 and G2/M in cirrhotic HCV (genotype-4) patients with or without accompanying tumor. Biocell. 2018;42:55–60. [Google Scholar]
- 5.Burki T. WHO’s 2024 global hepatitis report. Lancet Infect Dis. 2024;24:e362–e363. doi: 10.1016/S1473-3099(24)00307-4. [DOI] [PubMed] [Google Scholar]
- 6.Wu W, Wang X, Ma R, Huang S, Li H, Lyu X. Deciphering the roles of neddylation modification in hepatocellular carcinoma: molecular mechanisms and targeted therapeutics. Genes Dis. 2024;12:101483. doi: 10.1016/j.gendis.2024.101483. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Adugna A. Therapeutic strategies and promising vaccine for hepatitis C virus infection. Immun Inflamm Dis. 2023;11:e977. doi: 10.1002/iid3.977. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Gajos-Michniewicz A, Czyz M. WNT/β-catenin signaling in hepatocellular carcinoma: the aberrant activation, pathogenic roles, and therapeutic opportunities. Genes Dis. 2023;11:727–746. doi: 10.1016/j.gendis.2023.02.050. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Quan W, Bello KE, Shueb Shomiad RH, Mustaffa N. Circular RNAs in hepatitis B virus-induced hepatocellular carcinoma: a comprehensive review and recent advances. Genes Dis. 2025;12:101605. doi: 10.1016/j.gendis.2025.101605. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Westbrook RH, Dusheiko G. Natural history of hepatitis C. J Hepatol. 2014;61:S58–68. doi: 10.1016/j.jhep.2014.07.012. [DOI] [PubMed] [Google Scholar]
- 11.Lee HW, Lee H, Kim BK, Chang Y, Jang JY, Kim DY. Cost-effectiveness of chronic hepatitis C screening and treatment. Clin Mol Hepatol. 2022;28:164–173. doi: 10.3350/cmh.2021.0193. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Liu T, Diao H. A double-edged sword: the HBV-induced non-coding RNAs alterations in hepatocellular carcinoma. Biocell. 2023;47:27–32. [Google Scholar]
- 13.Butt ZA, Wong S, Rossi C, Binka M, Wong J, Yu A, Darvishian M, Alvarez M, Chapinal N, McKee G, Gilbert M, Tyndall MW, Krajden M, Janjua NZ. Concurrent hepatitis C and B virus and human immunodeficiency virus infections are associated with higher mortality risk illustrating the impact of syndemics on health outcomes. Open Forum Infect Dis. 2020;7:ofaa347. doi: 10.1093/ofid/ofaa347. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Choi GH, Jang ES, Kim YS, Lee YJ, Kim IH, Cho SB, Lee HC, Jang JW, Ki M, Choi HY, Baik D, Jeong SH. Hepatocellular carcinoma, decompensation, and mortality based on hepatitis C treatment: a prospective cohort study. World J Gastroenterol. 2022;28:4182–4200. doi: 10.3748/wjg.v28.i30.4182. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Fujiyama S, Akuta N, Sezaki H, Kobayashi M, Kawamura Y, Hosaka T, Kobayashi M, Saitoh S, Suzuki F, Suzuki Y, Arase Y, Ikeda K, Kumada H. Mortality rates and risk factors in 1412 Japanese patients with decompensated hepatitis C virus-related cirrhosis: a retrospective long-term cohort study. BMC Gastroenterol. 2021;21:189. doi: 10.1186/s12876-021-01770-0. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Roguljic H, Nincevic V, Bojanic K, Kuna L, Smolic R, Vcev A, Primorac D, Vceva A, Wu GY, Smolic M. Impact of DAA treatment on cardiovascular disease risk in chronic HCV infection: an update. Front Pharmacol. 2021;12:678546. doi: 10.3389/fphar.2021.678546. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Song AT, Sobesky R, Vinaixa C, Dumortier J, Radenne S, Durand F, Calmus Y, Rousseau G, Latournerie M, Feray C, Delvart V, Roche B, Haim-Boukobza S, Roque-Afonso AM, Castaing D, Abdala E, D’Albuquerque LA, Duclos-Vallée JC, Berenguer M, Samuel D. Predictive factors for survival and score application in liver retransplantation for hepatitis C recurrence. World J Gastroenterol. 2016;22:4547–4558. doi: 10.3748/wjg.v22.i18.4547. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Gobran ST, Pagliuzza A, Khedr O, Fert A, Chomont N, Bruneau J, Klein MB, Ancuta P, Shoukry NH. DAA-mediated HCV cure reduces HIV DNA levels in HCV/HIV coinfected people. J Virol. 2023;97:e0110523. doi: 10.1128/jvi.01105-23. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Moretti S, Mancini F, Borsetti A. Long term immunological perturbations post DAA therapy in chronic HCV/HIV co-infected patients. Biocell. 2022;46:2695–2699. [Google Scholar]
- 20.Sepúlveda-Crespo D, Treviño-Nakoura A, Bellon JM, Ardizone Jiménez B, Jiménez-Sousa MA, Fernández-Rodríguez A, Martínez I, Resino S. Meta-analysis: diagnostic accuracy of hepatitis C core antigen detection during therapy with direct-acting antivirals. Aliment Pharmacol Ther. 2022;56:1224–1234. doi: 10.1111/apt.17198. [DOI] [PubMed] [Google Scholar]
- 21.Hansoongnern P, Pratedrat P, Nilyanimit P, Wasitthankasem R, Posuwan N, Wanlapakorn N, Kodchakorn K, Kongtawelert P, Pimsing N, Poovorawan Y. An amino acid substitution in HCV core antigen limits its use as a reliable measure of HCV infection compared with HCV RNA. PLoS One. 2023;18:e0287694. doi: 10.1371/journal.pone.0287694. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Wang Y, Jie W, Ling J, Yuanshuai H. HCV core antigen plays an important role in the fight against HCV as an alternative to HCV-RNA detection. J Clin Lab Anal. 2021;35:e23755. doi: 10.1002/jcla.23755. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Woo J, Choi Y. Biomarkers in detection of hepatitis C virus infection. Pathogens. 2024;13:331. doi: 10.3390/pathogens13040331. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Baber AS, Suganthan B, Ramasamy RP. Current advances in Hepatitis C diagnostics. J Biol Eng. 2024;18:48. doi: 10.1186/s13036-024-00443-2. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Liang WEI, Zhang Z, Yang K, Hu H, Luo Q, Yang A, Chang LI, Zeng Y. Genetic algorithm-optimized backpropagation neural network establishes a diagnostic prediction model for diabetic nephropathy: combined machine learning and experimental validation in mice. Biocell. 2023;47:1253–1263. [Google Scholar]
- 26.Long X, Chen Y, Yin M, Liu W. Improving HCV diagnosis following a false-negative anti-HCV result. Clin Chem Lab Med. 2024;62:e229–e231. doi: 10.1515/cclm-2024-0161. [DOI] [PubMed] [Google Scholar]
- 27.Kim MH, Kang SY, Lee WI, Lee MY. Evaluation of HCV RNA by PCR and signal-to-cutoff ratios of HCV antibody assays for diagnosis of HCV infection. Lab Med. 2021;52:240–244. doi: 10.1093/labmed/lmaa074. [DOI] [PubMed] [Google Scholar]
- 28.Xu B, Chen B, Ma ZH, Ren YN, Ma JQ, Pei LJ, Xing WW. Application of anti-HCV and HCV RNA detection in intravenous drug users. Zhonghua Gan Zang Bing Za Zhi. 2021;29:415–420. doi: 10.3760/cma.j.cn501113-20200724-00423. [DOI] [PubMed] [Google Scholar]
- 29.Curtis KA, Rudolph DL, Pan Y, Delaney K, Anastos K, DeHovitz J, Kassaye SG, Hanson CV, French AL, Golub E, Adimora AA, Ofotokun I, Bolivar H, Kempf MC, Peters PJ, Switzer WM. Evaluation of the Abbott ARCHITECT HIV Ag/Ab combo assay for determining recent HIV-1 infection. PLoS One. 2021;16:e0242641. doi: 10.1371/journal.pone.0242641. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Liu HY, Lin YH, Lin PJ, Tsai PC, Liu SF, Huang YC, Tsai JJ, Huang CI, Yeh ML, Liang PC, Lin ZY, Dai CY, Huang JF, Chuang WL, Huang CF, Yu ML. Anti-HCV antibody titer highly predicts HCV viremia in patients with hepatitis B virus dual-infection. PLoS One. 2021;16:e0254028. doi: 10.1371/journal.pone.0254028. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Lu JL, Yu CX, Song LJ. Programmed cell death in hepatic fibrosis: current and perspectives. Cell Death Discov. 2023;9:449. doi: 10.1038/s41420-023-01749-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Roudot-Thoraval F. Epidemiology of hepatitis C virus infection. Clin Res Hepatol Gastroenterol. 2021;45:101596. doi: 10.1016/j.clinre.2020.101596. [DOI] [PubMed] [Google Scholar]
- 33.Li H, Liu T, Yang Y, Cho WC, Flynn RJ, Harandi MF, Song H, Luo X, Zheng Y. Interplays of liver fibrosis-associated microRNAs: molecular mechanisms and implications in diagnosis and therapy. Genes Dis. 2023;10:1457–1469. doi: 10.1016/j.gendis.2022.08.013. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Yang J, Zhang J, Zhang L, Yang Z. Mitochondrial oxidative stress-associated mechanisms in the development of metabolic dysfunction-associated steatotic liver disease. Biocell. 2025;49:399–417. [Google Scholar]
- 35.Ferreira J, Bicho M, Serejo F. Effects of HCV clearance with direct-acting antivirals (DAAs) on liver stiffness, liver fibrosis stage and metabolic/cellular parameters. Viruses. 2024;16:371. doi: 10.3390/v16030371. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Tarancon-Diez L, Carrasco I, Jiménez de Ory S, Berzosa Sánchez A, Hernanz-Lobo A, Montero-Alonso M, Laguno M, Bernardino JI, López-Cortés L, Aldamiz-Echevarría T, Collado P, Bisbal O, Samperiz G, Gavilán C, Ríos MJ, Ibarra S, Navarro ML, Muñoz-Fernández M. Long-term evolution in liver disease markers and immune and lipid profiles in vertically HIV/HCV-coinfected youths with sustained viral response after direct-acting antivirals therapy. Biomed Pharmacother. 2023;162:114587. doi: 10.1016/j.biopha.2023.114587. [DOI] [PubMed] [Google Scholar]
- 37.Li H, Yu K, Zhang X, Li J, Hu H, Deng X, Zeng S, Dong X, Zhao J, Zhang Y. YTHDF1 shapes immune-mediated hepatitis via regulating inflammatory cell recruitment and response. Genes Dis. 2024;12:101327. doi: 10.1016/j.gendis.2024.101327. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Al Alawi AM, Al Shuaili HH, Al-Naamani K, Al Naamani Z, Al-Busafi SA. A machine learning-based mortality prediction model for patients with chronic hepatitis C infection: an exploratory study. J Clin Med. 2024;13:2939. doi: 10.3390/jcm13102939. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.Feld JJ. What is needed to move toward single-step diagnosis of current HCV infection? J Infect Dis. 2024;229:S316–S321. doi: 10.1093/infdis/jiad453. [DOI] [PMC free article] [PubMed] [Google Scholar]