Bile Acids Linked to Cirrhosis and HCC Progression

Introduction Chronic hepatitis B virus (HBV) infection, affecting approximately 296 million individuals worldwide, is the primary cause of liver cirrhosis (LC) and hepatocellular carcinoma (HCC), thus constituting a significant global public health challenge.1 In 2019, HBV-related LC and HCC were responsible for an estimated 523,000 deaths, and projections indicate that annual global deaths from HBV will increase by 39% between 2015 to 2030.2,3 Consequently, early detection of LC and HCC is crucial, as it allows for treatments that can either cure the disease or substantially extend patient survival. The current noninvasive methods for assessing the progression of HBV-related liver disease are notably inadequate. Clinical guidelines recommend several serum indicators, such as aspartate aminotransferase to platelet ratio index (APRI), fibrosis index based on four factors (FIB-4), and alpha-fetoprotein (AFP), for diagnosing LC or HCC. Although these methods are simple and cost-effective, they have significant limitations in terms of sensitivity and specificity.4–6 Therefore, there is an urgent need for further research to identify new noninvasive biomarkers that can more accurately assess the progression of HBV-related liver disease. Bile acids (BAs) are the primary metabolites of cholesterol metabolism in the liver and play a pivotal role in regulating lipid, glucose, and energy homeostasis.7,8 Beyond their physiological functions, BAs also act as signaling molecules that bind to nuclear receptors (eg, farnesoid X receptor, FXR) and G protein-coupled receptors (eg, TGR5), thereby modulating hepatic inflammation, fibrosis, and carcinogenesis.9–11 This biological rationale is supported by evidence that HBV infection disrupts hepatocellular BA synthesis, conjugation, and transport, leading to intrahepatic BA accumulation and subsequent activation of profibrotic and pro-oncogenic signaling pathways.12–14 Recent years have witnessed growing attention on the role of BAs in HBV-related LC and HCC, as several clinical studies demonstrating an association between perturbations in BA composition and the development of HBV-related liver disease.15–17 However, existing BA-related studies have primarily focused on individual BA species or small panels, lacking systematic characterization of the entire BA metabolic phenotype across the full disease spectrum from CHB to LC and HCC.7,18–20 Additionally, prior metabolomic studies rarely integrated clinical confounding factors that may alter BA profiles—such as cholestasis (a common complication of advanced liver disease that exacerbates BA dysregulation) and gut microbiota dysbiosis (which modulates secondary BA production via bacterial enzymes like 7α-dehydroxylase)21–23—or directly compared the predictive performance of BA-based models with APRI, FIB-4, and AFP in the same cohort. Understanding the phylogeny of BAs in HBV-related liver diseases is essential for identifying potential biomarkers that can be used to predict and monitor disease progression. Consequently, this retrospective study was conducted to identify BA phenotypes in patients with HBV-related liver diseases, including CHB, LC, and HCC, with the aim of developing a predictive model based on BA metabolism to assess the risk of progression from CHB to LC and HCC. Materials and Methods Study Design and Patients This retrospective analysis utilized data from patients admitted to the Department of Liver Disease at Shuguang Hospital between 2018 and 2023, who were diagnosed with CHB, HBV-related LC, and HCC. Patients were randomly assigned to either the training or validation cohort at a ratio of 7:3 using the R function “Create Data Partition” with a fixed random seed to ensure reproducibility. The training cohort was used to screen variables and construct the prediction model, while the validation cohort was used to internally validate the model performance. Diagnostic criteria were defined as follows: CHB patients were those testing positive for hepatitis B surface antigen (HBsAg) for at least 6 months without evidence of concurrent liver cirrhosis (LC) or hepatocellular carcinoma (HCC).4 HBV-related LC diagnosis was based on a combination of medical history, physical examination, and results of biochemical tests, endoscopic examination, ultrasound, and enhanced computed tomography (CT)/magnetic resonance imaging (MRI) as per clinical practice.24 The diagnosis of HBV-related HCC relied on the guidelines of the American Association for the Study of Liver Diseases, including HBsAg positivity, typical imaging features (arterial phase hyperenhancement and venous phase washout) on enhanced CT/MRI, or histological confirmation of malignant hepatocytes via liver biopsy.25 Exclusion criteria were as follows: 1) Comorbidity with other liver diseases, such as chronic hepatitis C, alcoholic liver disease, nonalcoholic steatohepatitis, autoimmune liver disease, or hereditary liver diseases; 2) Presence of severe complications, including severe infections, acute liver failure, or variceal bleeding; 3) Underlying life-threatening medical conditions, such as acute cardiovascular or cerebrovascular accidents, malignant tumors of other systems, or end-stage renal disease; 4) Patients with severely incomplete clinical or laboratory data (missing key variables including bile acid (BA) profiles, APRI, FIB-4, or AFP). Sample size calculation: The sample size was determined based on the methods for logistic regression model development, considering the number of potential predictors and expected event rate. A minimum of 10 events per predictor variable was assumed, with the primary outcome defined as progression from CHB to LC/HCC. Based on preliminary data, the expected event rate (proportion of LC/HCC patients) was approximately 40%.26 We included a series of potential predictors (specially,15 BA species, plus APRI, FIB-4 and AFP), requiring a minimum of 500 events. To account for potential missing data and ensure sufficient statistical power, the final sample size was set at 609 patients, which exceeded the minimum requirement and was consistent with similar metabolomic-based prediction studies in liver diseases. Ethics Statement The study was performed in accordance with the ethical standards of the affiliated institutions and the Declaration of Helsinki (2013 revision). The study protocol was reviewed and approved by the Ethics Committee and Institutional Review Board of Shuguang Hospital (No. 2023-1249-16-01). Written informed consent was obtained from each patient prior to enrollment. The consent form, approved by the Shuguang Hospital Ethics Committee, specifically included permission for access to and use of their medical records for this research and for the publication of deidentified findings. For patients unable to provide written consent due to poor health status, consent was obtained from their legal guardians. Data Collection Basic clinical data, including demographic information, past medical history, and laboratory features, were extracted from electronic medical records. Demographic and clinical variables included age, sex, duration of HBV infection and history of antiviral treatment (before 3 months of enrollment). Laboratory parameters included: 1) BA profiles (16 key species covering primary/secondary, conjugated/unconjugated BAs); 2) routine blood tests (white blood cell count, neutrophil count, lymphocyte count, platelet count (PLT)); 3) biochemical parameters (alanine aminotransferase (ALT), aspartate aminotransferase (AST), total bilirubin(TBIL), direct bilirubin, albumin(ALB), alkaline phosphatase (ALP), gamma-glutamyl transpeptidase (GGT)); 4) virological parameters (Hepatitis B surface Antigen(HBsAg), hepatitis B e antigen (HBeAg) status, HBV DNA load); 5) serum fibrosis indicators (type IV collagen (CIV), laminin (LN), hyaluronic acid (HA), procollagen III N-terminal peptide (PIIIP)); 6) blood coagulation parameters (prothrombin time, international normalized ratio (INR)); 7) tumor biomarkers (alpha-fetoprotein (AFP)). Several clinically relevant ratios were calculated: platelet-to-lymphocyte ratio (PLR) = PLT (109/L) / lymphocyte count (109/L), neutrophil-to-lymphocyte ratio (NLR) = neutrophil count (109/L) / lymphocyte count (109/L), APRI = (AST / upper limit of normal (ULN)) × 100 / PLT (109/L), and FIB-4 = (age × AST) / (PLT × √ALT). Missing data were identified and quantified for each variable. Variables with missing rates < 10% were imputed using the multiple imputation method (with 5 imputed datasets) based on predictive mean matching. Variables with missing rates ≥ 10% were excluded from the final model to avoid bias. Bile Acid Detection and Pre-Analytical Procedures Sample collection and processing: Fasting venous blood samples (5 mL) were collected from all patients within 24 hours of admission, centrifuged at 3000 × g for 10 minutes at 4°C to separate serum, and stored at −80°C within 1 hour of collection to avoid BA degradation. Samples were thawed only once before analysis. Instrument and reagents: BA profiling was performed using an ultra-high-performance liquid chromatography-tandem mass spectrometry (UHPLC-MS/MS) system (Thermo Scientific Ultimate 3000 UHPLC coupled with TSQ Quantis Triple Quadrupole MS, USA). Analytical columns (ACQUITY UPLC BEH C18, 2.1×100 mm, 1.7 μm; Waters, USA) were used for chromatographic separation. Standards of 24 BA species (purity ≥ 98%) were purchased from Sigma-Aldrich (USA), and chromatographic-grade methanol, acetonitrile, and formic acid were obtained from Fisher Scientific (USA). Chromatographic and mass spectrometric conditions: Mobile phase A consisted of 0.1% formic acid in water, and mobile phase B consisted of 0.1% formic acid in acetonitrile. The gradient elution program was as follows: 0–2 min, 30% B; 2–10 min, 30–90% B; 10–12 min, 90% B; 12–12.1 min, 90–30% B; 12.1–15 min, 30% B. The flow rate was 0.3 mL/min, column temperature was 40°C, and injection volume was 5 μL. MS detection was performed in electrospray ionization (ESI)-negative mode with selected reaction monitoring (SRM). The ion source parameters were: spray voltage = 3500 V, capillary temperature = 320°C, sheath gas pressure = 35 arb, auxiliary gas pressure = 10 arb. Quality control (QC) and calibration: QC samples were prepared by pooling equal volumes of serum from all patients and analyzed at the beginning, middle, and end of each analytical batch (every 10 samples). The relative standard deviation (RSD) of peak areas for each BA species in QC samples was required to be < 15% to ensure analytical stability. Calibration curves were constructed using serial dilutions of standard solutions (0.01–100 μmol/L) with R2 ≥ 0.99. BA concentrations were quantified using external standardization, and results were corrected for matrix effects using the isotope dilution method (stable isotope-labeled cholic acid-d4 and chenodeoxycholic acid-d4 as internal standards). Statistical Analysis Quantitative data are presented as mean ± standard deviation (mean ± SD) for normally distributed data or median with interquartile range (M [Q1, Q3]) for non-normally distributed data, whereas categorical variables are expressed as frequency and proportion (n [%]). Data distribution was assessed using the Kolmogorov–Smirnov normality test. For normally distributed data, differences among the three groups (CHB, LC, HCC) were analyzed using one-way analysis of variance (ANOVA) followed by Tukey’s post-hoc test; for non-normally distributed data, the Kruskal–Wallis H-test followed by Dunn’s post-hoc test was used. Categorical variables were compared using the chi-square test or Fisher’s exact test (when expected frequencies < 5). To establish the prediction model in the training set, univariate logistic regression analysis was first performed to screen potential predictors (p < 0.10). Variables with p < 0.10 in univariate analysis were included in multivariate logistic regression with stepwise forward selection (entry criterion: p = 0.05; removal criterion: p = 0.10). The final model was visualized using a nomogram, with each variable assigned a score based on its regression coefficient. Model evaluation: 1) Discrimination: Assessed by the area under the receiver operating characteristic (ROC) curve (AUC), with 95% confidence intervals (CIs) calculated using the bootstrap method (1000 resamples). AUC values were compared between the BA-based model and traditional indicators (APRI, FIB-4, AFP) using the DeLong test. 2) Calibration: Evaluated by calibration curves, which plot predicted probabilities against observed probabilities, with Hosmer-Lemeshow test (p > 0.05 indicating good calibration) and Brier score (lower values indicate better calibration). A calibration plot with 10 equally sized risk groups was generated, and a slope and intercept correction were applied if necessary. 3) Clinical utility: Assessed by decision curve analysis (DCA) and clinical impact curves (CIC). DCA was used to compare the net benefit of the BA-based model with traditional indicators and the “treat all” or “treat none” strategies across a range of threshold probabilities (0.1–0.9). CIC was used to estimate the number of true positives and false positives at a clinically relevant threshold (preset at 0.3 based on clinical practice and preliminary data). All statistical tests were two-tailed, with a significance level of p ≤ 0.05. Statistical significance was defined as follows: *P < 0.05, **P < 0.01, ***P < 0.001. Statistical analyses were performed using SPSS (V28.0, IBM, USA), R software (V4.3.1, R Foundation for Statistical Computing), and GraphPad Prism (V8.0, GraphPad Software, USA). The nomogram was constructed using the “rms” package in R, and DCA was performed using the “rmda” package. Results Baseline Characteristics A total of 609 eligible patients diagnosed with CHB, HBV-related LC, or HCC between 2018 and 2023 were recruited for this study. The demographic and clinical characteristics of patients are shown in Table 1. The mean ages of patients with CHB, HBV-related LC, and HCC were 44, 59, and 59 years, respectively. The proportions of male patients in the CHB, HBV-related LC, and HCC groups were 78.53%, 53.29%, and 73.28%, respectively. Enrolled patients were randomly divided into two cohorts at a ratio of 7:3, resulting in a training cohort of 431 patients and a validation cohort of 178 patients. A detailed flow diagram illustrating the patient selection process is shown in Figure 1. Importantly, there were no significant differences in the clinical variables between the training and validation cohorts (Supplementary Table 1). The Phenotype of Bile Acid Metabolism in CHB, LC and HCC Patients We investigated BA metabolism and identified 15 BA-related metabolites in patients with CHB, LC, and HCC using LC-MS on plasma samples from a training cohort comprising 260CHB, 101 LC, and 70 HCC patients (detailed in Supplementary Tables 2 and 4). Notably, significant differences in key clinical metabolic markers were observed between any two of the three groups, except between LC and HCC. As shown in Figure 2A, analysis of BA metabolites across the three groups revealed elevated levels of multiple primary bile acids (PBA) in the LC and HCC groups higher than in the CHB group, whereas levels of multiple secondary bile acids (SBA) were lower in the LC and HCC groups than in the CHB group. Specifically, cholic acid (CA) and chenodeoxycholic acid (CDCA) species were significantly increased in LC and HCC relative to CHB, whereas lithocholic acid (LCA) and deoxycholic acid (DCA) were significantly decreased. Although no statistically significant differences in BA levels were found between LC and HCC, patients with decompensated cirrhosis exhibited significantly higher levels of total BA (TBA), glycocholic acid (GCA), glycochenodeoxycholic acid (GCDCA), taurocholic acid (TCA), taurolithocholic acid (TLCA), taurochenodeoxycholic acid (TCDCA), and tauroursodeoxycholic acid (TUDCA) than those in the compensated stage (P<0.05, Supplementary Figures 1A, 2 and 3). To validate our initial findings, we conducted targeted metabolomic profiling of 15 BAs in plasma samples from a validation cohort of 178 participants comprising 84 CHB, 48 LC, and 46 HCC patients. The validation cohort exhibited BA metabolite alterations similar to those observed in the training cohort, with significant upregulation of TBA, GCA, GCDCA, TCDCA, LCA, GUDCA, and TUDCA in both the LC and HCC groups compared with the CHB group (Figure 2B). To further explore BA metabolism in the gut, we examined the ratio of SBA metabolites to their PBA precursors, as SBA production involves multiple enzymatic reactions facilitated by gut microbiota.11 Targeted metabolomics revealed a significant decrease in the ratio of DCA to CA in the LC and HCC groups compared with that in the CHB group (Figure 2C). A comparable decrease was noted in the ratio of LCA and UDCA metabolites to CDCA metabolites in the LC and HCC groups compared to that in the CHB group (Figure 2D). These findings suggest that variations in the enzyme activity of the intestinal microbiota contribute to changes in secondary BAs.27 Additionally, we observed differences in the ratios of circulating conjugated to unconjugated PBAs or SBAs, such as glycine(G)/taurine(T)-DCA/DCA and G/T-LCA/LCA (Figure 2E–H), indicating that imbalances in SBAs are related to enzymatic reactions involving their conjugation with T and G.27 Furthermore, an increase in the ratio of CA metabolites to CDCA metabolites was observed in the LC and HCC groups compared to the CHB group (Figure 2I), suggesting a shift in BA synthesis in the liver from the classical pathway to the alternative pathway.27 ROC analysis was conducted to evaluate the predictive power of the BAs for the risk of disease progression. The analysis revealed that several BAs, including TBA, CA, CDCA, GCA, GCDCA, TCA, TCDCA, and TUDCA, demonstrated strong predictive capacities in distinguishing HCC and LC from CHB compared to differentiating between HCC and LC. These findings indicate that BAs serve as effective biomarkers for liver disease progression (Supplementary Figure 1B–I). Identification of Risk Factors We began our analysis with a collinearity test, which led to the exclusion of TUDCA from the logistic regression analysis because of collinearity. Subsequently, we conducted univariate logistic regression analysis incorporating clinical baseline characteristics and BAs as independent variables to identify potential confounders. Univariate analysis revealed several significant variables, including male sex, age, AFP, PLR, NLR, TBIL, ALB, ALP, CIV, LN, HA, PIIIP, FIB-4, TBA, CA, DCA, CDCA, GCA, GCDCA, GUDCA, and TCA (Table 2). These significant variables were then subjected to multivariate logistic regression analysis, which identified TCA, ALB, and age as the independent predictors of disease progression (Table 3 and Supplementary Tables 5 and 6). Establishment and Validation of the Nomogram Prediction Models A nomogram prediction model was established based on the risk factors identified through multivariate logistic regression analysis in the training cohort, enabling visual representation of the predictive framework (Figure 2J). The model incorporated three key variables: age, ALB, and TCA. The probability of disease progression in individual patients can be determined by calculating the cumulative points across these variables. In the training set, the model demonstrated excellent discriminative ability, with an AUC value of 0.9151 (95% CI: 0.888–0.9421), achieving a sensitivity of 0.876 and specificity of 0.814 (Figure 3A). The model’s robust performance was further confirmed in the validation set, where it achieved an even higher AUC of 0.9413 (95% CI: 0.9096 −0.973), with an enhanced sensitivity (0.895) and specificity (0.86 (Figure 3B). The calibration curves for both sets showed strong alignment with the standard curve, indicating reliable concordance between the predicted probabilities and the observed outcomes (Figure 3C and D). The clinical utility of the model was substantiated by DCA and CIC assessments (Figure 3E–H). To comprehensively evaluate the predictive advantage of the nomogram, we compared its performance with that of three conventional clinical indicators (APRI, FIB-4, and AFP) in the training cohort. Detailed comparative data are presented in Supplementary Table 3. The nomogram showed the highest AUC (0.915, 95% CI: 0.888–0.942) among all tested indicators, which was significantly superior to APRI (0.610, 95% CI: 0.541–0.678), FIB-4 (0.764, 95% CI: 0.708–0.819), and AFP (0.798, 95% CI: 0.739–0.857). In addition, combined with the optimal cut-off value, sensitivity and specificity, these confirms that the Nomogram has better discriminant power in predicting CHB progression to LC/HCC compared with traditional non-invasive indicators and classical tumor markers AFP. Discussion HBV infection is the primary cause of cirrhosis and HCC, with recent statistics indicating incidence rates of 4.91 and 2.57 per 100,000 persons for HBV-related liver cirrhosis and liver cancer, respectively.1,28 Given that LC and HCC progression develops over time, early recognition and prediction have substantial clinical value. Although noninvasive assessments, including serum biomarkers through laboratory testing, are increasingly being proposed for predicting LC and HCC development, their performance remains controversial. Therefore, it is crucial to identify new indicators of HBV-related LC and HCC progression. Studies have shown that HBV infection disrupts BAs metabolism.29,30 As the end products of cholesterol catabolism, BAs act as signaling molecules and metabolic regulators that maintain hepatic lipid, glucose, and energy homeostasis.8 CA and CDCA PBAs are synthesized in human livers, conjugated with taurine or glycine, and secreted into the intestine.31 Subsequently, the gut microbiota converts these PBAs into SBAs, including LCA, DCA, and UDCA.32,33 Excess hepatic BAs can contribute to liver fibrosis, LC, and HCC.34,35 Given this, we conducted a preliminary validation using serum samples from 609 patients with chronic liver disease (including CHB, HBV-related LC, and HCC) to investigate whether BA indicators can predict HBV-related liver disease progression. Our analysis revealed significant elevations in CA, GCA, TCA, CDCA, and GCDCA levels in patients with LC and HCC compared with those in the CHB group (P<0.05). Conversely, free secondary BAs, DCA, and LCA levels were lower in patients with LC and HCC than in those with CHB group (P<0.05). This result was in accordance with that of a previous study, which showed the alterations of some BAs among different stages of hepatitis B, with a decreased conversion of primary to secondary BAs.36 Therefore, we suspected that primary BAs dehydroxylation might be impaired in HBV-related disease progression because secondary BAs are formed through dehydroxylation of primary BAs catalyzed by bacteria in the intestine.8 Studies have shown that bacteria involved in BAs metabolism are significantly decreased in CHB- and HBV-induced LC.36,37 It has been reported that the gut microbiome and HBV-induced chronic liver diseases are linked, and that abnormal BAs circulation caused by the intestinal microbiota may be the reason for CHB progression.38 The elevation in serum BA levels observed in our study can be partially attributed to cholestasis—a common complication of HBV-related LC and HCC that impairs bile flow and leads to intrahepatic BA accumulation.39,40 Even though there was no significant difference in HBV-related LC and HCC in the present study, we found that the levels of BAs, including TBA, GCA, GCDCA, TCA, TLCA, TCDCA, and TUDCA, were significantly increased in patients with LC decompensation compared to the compensated stage, which indicated that BAs were evidently elevated in end-stage liver disease. When hepatitis and cirrhosis progress to HCC, serum levels of GCDCA, TCA, and GCA are elevated.41 Regarding the indistinct BA trend between LC and HCC, as well as the decreased ROC performance of TCA in distinguishing these two diseases, the core reasons are as follows: ① Pathophysiological overlap between advanced LC and early-to-mid stage HCC, where cirrhosis-driven cholestasis and metabolic disorders mask HCC-specific BA alterations, with TCA elevation primarily driven by cirrhotic dysfunction rather than tumorigenesis itself;42 ② Biological heterogeneity of HCC (variations in differentiation degree, size/location) leading to overlapping BA profiles with LC;43,44 ③ Prior clinical interventions (eg, antiviral therapy, hepatoprotective agents) may homogenize BA levels between the two groups, reducing discriminatory efficacy. Collectively, BAs can serve as potential biomarkers for HBV-related liver diseases, such as HBV-related LC and HCC, particularly end-stage liver disease. The above analysis suggests that BAs (especially TCA) are more suitable for assessing liver injury severity and predicting CHB progression to LC/HCC, rather than distinguishing LC from HCC alone. Future studies may combine BA indicators with tumor-specific biomarkers or imaging features to improve discriminatory power. We confirmed that age, ALB, and TCA were independent factors for the development of HBV-related LC and HCC via univariate and multivariate logistic regression analyses, all of which were significantly correlated with the progression of HBV-related liver disease (all P<0.001). Specifically, ALB was associated with a decreased risk, while age and TCA were associated with increased risks of HBV-related LC and HCC. As widely recognized, age and ALB level have been well established as predictors of liver disease progression.45–47 Notably, our study further identifies elevated serum TCA as an independent risk factor for HBV-related LC or HCC, consistent with the findings of Tan et al,48 who confirmed TCA as a potential marker for distinguishing HCC from HBV-related chronic liver diseases. Mechanistically, previous studies have demonstrated that high TCA levels promote cirrhotic progression by acting on hepatic cells (eg, hepatic stellate cells and hepatocytes).49,50 This finding supports the biological rationality of TCA as a predictive biomarker for HBV-related LC and HCC. Based on these three independent factors, we constructed a predictive nomogram, which exhibited high discrimination and clinical applicability: the area under the curve (AUC) was 0.9151 (95% CI: 0.888–0.9421) in the training set and 0.9413 (95% CI: 0.9096–0.973) in the validation set. Calibration curves, decision curve analysis (DCA), and clinical impact curves (CIC) further verified the reliable predictive accuracy of this model in both cohorts. The nomogram was established with a clinically relevant threshold (sensitivity = 0.663, specificity = 0.84; optimal cutoff value for the nomogram total score: 2.152) to facilitate practical application. Its primary application scenario lies in the early risk stratification of CHB patients during routine clinical surveillance—especially for those with unclear disease progression trends or limited access to advanced imaging examinations—offering a convenient, noninvasive tool to identify high-risk individuals who may require intensified monitoring or intervention. When compared with classical noninvasive indices (APRI, FIB-4) and the gold-standard HCC marker AFP, our nomogram exhibited superior discriminatory power (AUC of nomogram vs APRI: 0.915 vs 0.610; vs FIB-4: 0.915 vs 0.764; vs AFP: 0.915 vs 0.798; all P < 0.05), attributed to the integration of TCA, a metabolite directly linked to liver pathophysiology and BAs metabolism dysfunction induced by HBV infection. We also acknowledge that potential confounding factors may influence the interpretation of our findings: for instance, cholestasis (a common complication in advanced HBV-related liver disease) can independently elevate serum BAs levels, which may partially overlap with the pathogenic effects of TCA on liver fibrosis and carcinogenesis. However, subgroup analysis (data not shown) revealed that the predictive value of TCA remained significant after adjusting for cholestasis-related indicators (eg, total bilirubin, gamma-glutamyl transpeptidase), supporting its independent role beyond mere cholestatic manifestation. The abnormal BAs profile in our study serves as a predictive biomarker for HBV-related liver disease and identifies potential therapeutic targets to improve BAs metabolism and patient outcomes. First, UDCA, a well-established bile acid replacement therapy for cholestatic liver diseases,51 inhibits toxic primary BAs synthesis via hepatic CYP7A1 negative feedback.52 Preclinical and clinical studies confirm UDCA alleviates HBV-induced liver fibrosis,53 supporting its potential as adjuvant therapy for high-risk individuals. Second, gut microbiota modulation (eg, probiotics) corrects HBV-related BAs imbalance. Due to HBV reducing BAs-dehydroxylating bacteria and impairing primary-to-secondary BAs conversion,54 Probiotics could restore microbial diversity and BAs dehydroxylation, reversing the imbalance.55,56 It is worth noting that numerous prediction models for HCC risk have been reported previously,57 and as a single-center retrospective study without external validation, the current model may not be immediately translatable to broad clinical practice. However, the integration of TCA (a metabolite with clear biological relevance to liver pathophysiology) with age and ALB provides a novel, concise, and biologically coherent predictive tool that complements existing models—especially given the accessibility of serum TCA detection in clinical settings. The limitations of this study should be acknowledged: first, this was a single-center retrospective cohort study, which may introduce selection bias; second, we did not perform dynamic monitoring of serum bile acid (BAs) levels in patients, and the temporal relationship between TCA changes and disease progression remains to be clarified; third, no prognostic analysis of BAs in HBV-related liver disease was conducted. Therefore, future prospective, large-cohort, multicenter studies are warranted to further validate the value of BAs metabolism (especially TCA) in the prognosis of HBV-related liver disease and to perform external validation of the proposed nomogram. Additionally, in vitro cellular and in vivo animal experiments are needed to further elucidate the underlying molecular mechanisms by which BAs metabolism regulates HBV-related liver disease progression. Conclusion Our results provided preliminary evidences for BAs to be used as biomarkers for the HBV-related liver disease, and developed a Nomogram enabling precise prediction of LC and HCC in CHB patients. This nomogram provides an accurate and accessible method for early screening and prevention, potentially improving routine surveillance and prediction of HBV-related LC or HCC in clinical practice. However, prospective, multicenter, large-cohort studies are needed to validate the model further, which will assist clinicians in timely and accurate identification of high-risk CHB patients for targeted prevention and personalized management.