Diagnostic efficacy and influencing factors of non-invasive liver fibrosis scores in chronic hepatitis B fibrosis: Cross-sectional study

Abstract

BACKGROUND

During the pathological advancement of chronic hepatitis B (CHB) toward cirrhosis and hepatocellular carcinoma, liver fibrosis acts as a pivotal transitional phase. Currently, non-invasive diagnostic techniques have become key substitutes for liver biopsy. However, the diagnostic efficacy of different indicators varies across populations, and the influence of factors such as gender, nutritional status, and liver function reserve remains unclear, necessitating further clarification.

AIM

To evaluate the diagnostic performance of serological indicators, including the aspartate transaminase to platelet ratio index (APRI), the fibrosis-4 index (FIB-4), the gamma-glutamyl transferase to platelet ratio (GPR), and the aspartate aminotransferase to alanine aminotransferase ratio (AST/ALT), with liver stiffness measurement (LSM) via FibroScan as a reference, and to investigate the influence of gender, nutritional status, and Child-Pugh classification on these serum assessment indicators.

METHODS

For this cross-sectional investigation, 627 CHB cases were consecutively enrolled at Chengdu Public Health Clinical Medical Center during the four-year interval from December 2020 to December 2024. Extensive data collection incorporated FibroScan-derived LSM alongside four well-validated serum fibrosis indices: APRI, FIB-4, GPR, and AST/ALT. Associations between these serum-derived indices and different fibrosis stages were examined; the effects of gender, nutritional status, and Child-Pugh class on the diagnostic performance of these non-invasive scoring tools were also assessed.

RESULTS

The median age of CHB participants was 47 years, with male patients comprising 69.86% (438/627). Spearman correlation testing revealed that all non-invasive liver fibrosis assessment indices showed robust associations with FibroScan-measured LSM values and FibroScan-derived liver fibrosis staging (P < 0.001); notably, the correlation between GPR and both LSM values and fibrosis staging was the most pronounced (r = 0.609). Univariate statistical analyses revealed significant disparities across the nutritional risk subgroup, the Child-Pugh staging subgroup, and each of the assessed indices (P < 0.001). Meanwhile, multivariable regression modeling showed that the GPR exhibited the most favorable diagnostic efficacy for two fibrosis categories: “Any fibrosis (F1-F4)” and “significant fibrosis (F3-F4)”. For the GPR model, the area under the curve (AUC) values were 0.816 [95% confidence interval (CI): 0.779-0.853] and 0.871 (95%CI: 0.844-0.899), with both outcomes corresponding to P values < 0.001. Analysis of receiver operating characteristic curves demonstrated that the GPR exhibited strong performance in identifying significant fibrosis: it achieved an AUC of 0.828 (95%CI: 0.796-0.860), alongside 87.6% sensitivity and 64.4% specificity. Moreover, the FIB-4 index demonstrated robust diagnostic performance (AUC = 0.814), surpassing that of both APRI and the AST/ALT ratio.

CONCLUSION

GPR and FIB-4 function as valid serological assessment indicators for identifying liver fibrosis in patients with CHB, especially in instances of advanced fibrosis and cirrhotic stages. In clinical settings, a holistic evaluation of liver fibrosis is recommended by incorporating the patient’s gender, nutritional status, and liver functional reserve to enhance diagnostic precision.

Key Words: Chronic hepatitis B; Gamma-glutamyl transferase to platelet ratio; Fibrosis-4 index; Aspartate aminotransferase to platelet ratio index; Aspartate transaminase to alanine aminotransferase ratio; Child-Pugh classification; Nutritional risk

Core Tip: This investigation was carried out in Southwest China, where the diagnostic efficacy of four non-invasive liver fibrosis scoring tools [aspartate transaminase to platelet ratio index, fibrosis-4 index (FIB-4), gamma-glutamyl transferase to platelet ratio (GPR), and aspartate aminotransferase to alanine aminotransferase ratio] was comprehensively assessed. The results indicated that GPR and FIB-4 achieved optimal diagnostic precision for detecting significant fibrosis and cirrhosis; notably, GPR demonstrated the strongest association with FibroScan-derived liver stiffness measurements. Furthermore, the study revealed that diagnostic performance is influenced by gender, nutritional status, and Child-Pugh grade. These findings indicate that incorporating GPR and FIB-4 into routine clinical assessment, along with patient-specific characteristics, can improve early screening for liver fibrosis in chronic hepatitis B.

INTRODUCTION

Chronic hepatitis B (CHB) endures as a major global public health challenge[1], affecting an estimated 316 million people globally[2]. In its 2024 report, the World Health Organization notes that while hepatitis B incidence has declined, associated mortality rates have increased. Studies indicate that approximately 1.1 million deaths in 2022 were attributed to hepatitis B-related complications. China is identified as an area with a considerable disease burden, with an estimated 79.7 million cases of CHB[3]. This large infected population reflects the severe prevention and control situation in China. In contrast to China’s arduous prevention and control landscape, high-income nations have achieved remarkable progress in curbing new hepatitis B infections. They have reduced the rate of new hepatitis B infections to near zero through comprehensive interventions, including vaccination campaigns, prevention of mother-to-child transmission, and blood safety screening. The primary burden of CHB remains centered in low- and middle-income economies[2,4]. In the natural course of CHB, hepatic fibrosis constitutes a crucial transitional stage. The timely detection of this condition is essential to curbing its progression towards cirrhosis and hepatocellular carcinoma. The essence of liver fibrosis is the abnormal proliferation of fibrous connective tissue within the liver, leading to excessive deposition as part of the injury repair response. When CHB stays persistently active or undergoes repeated disease exacerbations, it can drive the onset of hepatic fibrosis. Such fibrotic changes may gradually progress to decompensated cirrhosis, liver failure, or hepatocellular carcinoma, creating a substantial risk to patients’ survival and overall quality of life[5]. Multiple meta-analyses indicate that the prevalence of hepatic steatosis (an important risk factor for liver fibrosis) among global CHB patients is 29.6% to 34.93%, with approximately 14% to 29% having significant liver fibrosis (F ≥ 2)[2]. Furthermore, it is reported that the yearly progression rate to cirrhosis in untreated CHB individuals varies between 2% and 10%[1]. Consequently, the implementation of early and regular surveillance of liver fibrosis in this population is of substantial clinical importance.

For staging hepatic fibrosis, liver biopsy has long been recognized as the accepted gold standard. Frequently applied in routine clinical settings, the METAVIR scoring system categorizes hepatic fibrosis into five distinct clinical stages: F0 (no fibrosis), F1 (mild fibrosis), F2 (moderate fibrosis), F3 (advanced fibrosis), and F4 (cirrhosis). Even so, this invasive intervention poses risks of adverse events such as localized discomfort, bleeding episodes, and infectious complications, and in rare, severe scenarios, it may even lead to life-threatening consequences. Its low patient acceptability makes it unsuitable for long-term follow-up[6]. Due to these limitations, there is a growing clinical demand for non-invasive diagnostic techniques for liver fibrosis. In response to this need, non-invasive diagnostic technologies for liver fibrosis have rapidly developed, primarily falling into two categories: Serological indicators and imaging techniques. Under serology-focused diagnostic systems, commonly used biomarker indices incorporate the ratio between aspartate aminotransferase (AST) and alanine aminotransferase (ALT) (termed the AST/ALT), the aspartate aminotransferase-platelet ratio index (APRI), the fibrosis-4 scoring index (FIB-4), and the gamma-glutamyl transferase-to-platelet ratio (GPR)[7-10]. These models assess liver fibrosis by measuring blood markers. Studies have shown that various serum biomarkers (e.g., GPR, APRI, and FIB-4) can differentiate advanced from non-advanced hepatic fibrosis[11]. Employing these hepatic fibrosis assessment metrics to accurately determine the disease’s stage can reduce reliance on liver biopsy and may even serve as a viable substitute for this invasive procedure in certain scenarios[12]. This type of blood biochemical examination, which uses formulas to calculate the severity of hepatic fibrosis, is low-cost and easy to perform, making it an ideal alternative to liver biopsy. In terms of imaging technology, transient elastography (TE), represented by FibroScan, has become a mainstream non-invasive method for diagnosing hepatic fibrosis. This technology indirectly measures liver stiffness measurement (LSM) via ultrasonic wave propagation speed to assess the extent of liver fibrosis. Compared with traditional liver biopsy, FibroScan is safer and more convenient to operate, and is widely recognized in clinical practice. Therefore, the combined application of TE with serum biochemical indicators is a promising future direction for the non-invasive diagnosis of liver fibrosis[13]. Currently, TE has secured its position as a frontline non-invasive diagnostic modality in international clinical guidelines; one clinical investigation has proposed[14] that FibroScan ought to be recognized as the preferred non-invasive assessment approach for evaluating overt hepatic fibrosis or cirrhotic lesions in regions with adequate healthcare resources.

Although non-invasive diagnostic technologies for liver fibrosis have made significant progress, there remain many gaps in research. For example, different non-invasive indicators exhibit varying diagnostic efficacy across CHB populations. Especially, data on the population in southwestern China are extremely scarce, leading to the inapplicability of existing research conclusions in the region. In addition, most previous studies did not consider the impact of individual factors such as gender, nutritional status, and liver function on the effectiveness of non-invasive diagnosis. Sex can modulate the progression of hepatic pathology through pathways such as hormonal fluctuations and immune reactivity; malnutrition is not only a consequence of liver disease severity but may also accelerate its progression; the Child-Pugh classification reflects hepatic functional status. Such variables have the potential to disrupt the alignment between serum-based biomarkers and the actual extent of hepatic fibrosis. Thus, it is crucial to analyze how patients’ personal factors influence the diagnostic performance of non-invasive liver fibrosis evaluations to achieve accurate clinical evaluation. This research aims to fill the data gaps in this field in southwest China and provide clinicians with diagnostic references that are more suitable for the characteristics of local populations, so as to improve the accuracy of early screening for liver fibrosis, benefit the majority of CHB patients, and improve the long-term prognosis of CHB patients. In order to make up for the above shortcomings, this investigation aims to systematically examine the association and diagnostic utility between four serum-based biomarkers (APRI, FIB-4, GPR, and AST/ALT) and FibroScan-derived LSM values in CHB patients, with a focus on investigating the influence of gender, nutritional status, and Child-Pugh class on the diagnostic efficacy of these non-invasive tools.

MATERIALS AND METHODS

Study population

This study was designed as a retrospective cross-sectional investigation. The research subjects were selected from inpatients with CHB documented in the inpatient medical record system of Chengdu Public Health Clinical Medical Center. A total of 8675 CHB inpatients were admitted to the center between December 2020 and December 2024. Using a convenience sampling strategy, eligible participants were screened, with the following exclusion criteria applied: 64 cases aged under 18 years, 7626 cases with incomplete demographic or clinical data (including those who did not complete liver transient elastography or relevant laboratory tests), and 358 cases complicated by other chronic liver diseases. Finally, 627 CHB inpatients with complete datasets were enrolled in this study (Figure 1). The present study was conducted in accordance with the principles of the Declaration of Helsinki and has obtained ethical approval from the Institutional Review Board of Chengdu Public Health Clinical Medical Center (approval No. YJ-K2025-87-01).

Figure 1 Participant screening flow diagram. Flowchart of study subject selection for inpatients with chronic hepatitis B in Chengdu Public Health Clinical Center from December 2020 to December 2024. CHB: Chronic hepatitis B.

Sample size calculation

To attain the precision required for assessing the area under the curve (AUC) the receiver operating characteristic (ROC) curve, the study’s sample size was calculated. We utilized the Binormal model-based formula proposed by Hajian-Tilaki[15] for this purpose. The calculation parameters were specified as follows: A two-sided α of 0.05, a statistical power (1-β) of 80%, a target AUC value of 0.801 (based on GPR’s AUC for detecting significant fibrosis)[11], and a margin of error (d) set at 0.07, which corresponds to the maximum allowable half-width of the 95% confidence interval (CI) around the estimated AUC. This calculation yielded a minimum required sample size of 256 participants. The calculation formula is as follows:

.

Inclusion and exclusion criteria

Inclusion criteria: Age of 18 years or older; having undergone both liver transient elastography (FibroScan) and relevant laboratory assessments.

Exclusion criteria were outlined as follows: Individuals with missing demographic or clinical data were excluded from the study. Moreover, patients with a diagnosis of any other chronic liver conditions, such as non-alcoholic fatty liver disease, alcoholic liver disease, autoimmune hepatitis, hepatitis C, and hepatitis D, were excluded as well.

Diagnostic criteria

CHB was defined based on the 2022 guidelines for the prevention and treatment of CHB[1]. This guideline describes the disease as persistent positivity for hepatitis B surface antigen or hepatitis B virus deoxyribonucleic acid for at least 6 months.

Data collection

Clinical, laboratory, and imaging datasets were extracted from 627 CHB individuals admitted to Chengdu Public Health Clinical Medical Center from December 2020 to December 2024. This included demographic information, laboratory test results, and FibroScan findings. Healthcare professionals used the Nutritional Risk Screening 2002 (NRS2002) to assess patients’ nutritional risk and employed the Child-Pugh grading system to evaluate liver function in CHB patients during hospitalization. All research data were exported from the hospitalized electronic medical record system, repeatedly checked, logically verified, and then locked. Only complete and consistent case data were retained for analysis. All detection indicators were completed within 1 week of the patients’ admission to the hospital, and all calculated indicators were computed using internationally recognized formulas. The research team strictly ensured the completeness, accuracy, and authenticity of all data.

Assessment criteria and grouping standards for nutritional evaluation, Child-Pugh grading, serological and imaging indicators in liver fibrosis

LSM: Examinations were performed using FibroScan^® 502 (Echosens, Paris, France) after an overnight fast or at least 2 hours after food intake. Operators with medical backgrounds at our center conducted all FibroScan examinations independently. These operators had received training from Echosens and obtained certificates, each with several years of operational experience. LSM results are reported in kilopascals (kPa). Fibrosis staging was assigned by cross-referencing the measured liver stiffness values with reference criteria for histopathological staging, specifically[16,17]: F0 (no fibrosis): < 7.3 kPa; F1 (mild fibrosis): 7.3-9.7 kPa; F2 (moderate fibrosis): 9.7-12.4 kPa; F3 (advanced fibrosis): 12.4-17.5 kPa; F4 (cirrhosis): > 17.5 kPa.

We collected laboratory test data of patients during hospitalization (within 7 days after admission), including ALT with a reference range of 0-37 U/L; AST with a reference range of 0-37 U/L; γ-glutamyl transferase (GGT) with a reference range of 0-50 U/L; platelet count (PLT) with a reference range of (125-350) × 10⁹/L; and total bilirubin with a reference range of 0-20.5 μmol/L. Using these serum indicators, we calculated APRI, FIB-4, GPR, and the AST/ALT ratio, with the formulas outlined below[8,9]: Formula for calculating APRI: APRI = [AST (U/L)/upper reference limit of AST (U/L)]/[platelet count (× 10⁹/L)] × 100. For the FIB-4 index, the calculation formula is: FIB-4 = [AST (U/L) × patient age (years)]/√[platelet count (× 10⁹/L) × ALT (U/L)]. GPR is computed using the following formula: GPR = [GGT (U/L)/upper reference limit of GGT (U/L)]/[platelet count (× 10⁹/L)] × 100. The AST/ALT is calculated as: AST/ALT = AST (U/L)/ALT (U/L).

To assess nutritional risk among the included patients during their hospital stay, the Nutritional Risk Screening 2002 (NRS2002) was employed. This instrument is specifically designed to detect malnutrition risk in hospitalized populations. The NRS2002 score incorporates three domains: (1) Nutritional status impairment: A score of 1 point is assigned for weight loss > 5% within the preceding 3 months or reduced dietary intake (25%-50%) over one week; 2 points for weight loss > 5% in the past 2 months or 50%-75% reduced intake in one week; and 3 points for weight loss > 5% in the past month, 75%-100% reduced weekly intake, or a body mass index (BMI) < 18.5 kg/m². In scenarios where BMI measurement is precluded (e.g., due to pleural effusion, ascites, or edema), a serum albumin level < 30 g/L can be substituted and also warrants 3 points; (2) Disease severity: Conditions such as hip fracture, chronic disease exacerbations, chronic obstructive pulmonary disease (COPD), hemodialysis, cirrhosis, and common malignancies receive 1 point. For conditions including major abdominal surgical procedures, stroke, severe pneumonia, and hematologic malignancies, a score of 2 points is assigned. Severe conditions, including head trauma, bone marrow transplantation, or intensive care unit admission with an Acute Physiology and Chronic Health Evaluation II score > 10, are designated 3 points; and (3) Age-related scoring adjustment: For patients who are 70 years of age or older, one extra point is assigned. The overall score ranges from 0 to 7, with a score of 3 or higher indicating nutritional risk[18].

During their hospital stay, we conducted Child-Pugh grading assessments for enrolled patients. Used clinically to gauge liver reserve capacity in individuals with hepatic pathology, the Child-Pugh scoring system encompasses five clinical and biochemical parameters: Hepatic encephalopathy (classified by severity): A score of 1 is assigned for the absence of episodes, a score of 2 corresponds to grades 1 through 2, and a score of 3 applies to grades 3 to 4; for ascites: A score of 1 is given when there is no fluid buildup, a score of 2 is designated for mild ascites, and a score of 3 is assigned to moderate or severe ascites; total serum bilirubin (measured in μmol/L): 1 point for < 34, 2 points for 34-51, and 3 points for > 51; serum albumin (g/L): 1 point for > 35, 2 points for 28-35, and 3 points for < 28; prothrombin time prolongation (seconds): 1 point for 1-3 seconds, 2 points for 4-6 seconds, and 3 points for > 6 seconds. These individual scores are summed to categorize patients into three Child-Pugh classes: Class A (total score 5-6) reflects mild hepatic dysfunction; class B (7-9 points) corresponds to moderate impairment; and class C (10-15 points) signals severe liver function compromise[17]. Higher class letters align with greater degrees of hepatic functional decline.

Statistical analysis

All data analyses were conducted using R software (version 4.5.1). Categorical variables were reported as frequencies and percentages. Normality of continuous variables was evaluated using the Shapiro-Wilk and Kolmogorov-Smirnov tests; as these variables failed to exhibit normality, we presented them as median. We conducted a Spearman correlation analysis and visualized the associations between various liver fibrosis indicators and FibroScan (transient elastography) results using heatmaps, enabling intuitive interpretation of the relationships. To compare continuous variables between two groups (e.g., indicators between gender and nutritional risk groups), the Wilcoxon rank-sum test was used. For comparisons across three or more groups, the Kruskal-Wallis test was used; if a significant overall difference was detected, Dunn’s post hoc test was subsequently performed to assess pairwise comparisons (e.g., indicator variations across different Child-Pugh grades or FibroScan stages). Multivariate logistic regression models were constructed to assess the independent effects of individual variables on liver fibrosis-related outcomes; results were presented as adjusted odds ratios with 95%CI. ROC curve analysis was also conducted to assess the diagnostic performance of each index. Statistical significance was determined as a P value less than 0.05.

RESULTS

Baseline characteristics

A cohort of 627 CHB patients with complete case data was included in the present study, with males accounting for 69.86% (438/627). The median age was 47.00 (34.50, 55.00) years. Most patients exhibited significant liver injury, with a median ALT of 114.00 (39.00, 408.50) U/L, AST of 77.00 (36.00, 235.50) U/L, GGT of 57.00 (28.00, 119.00) U/L, PLT of 130.00 (87.00, 173.00) × 10⁹/L, and total bilirubin of 17.20 (11.00, 34.75) μmol/L. Liver fibrosis-related indicators showed: APRI of 1.92 (0.83, 5.42), GPR of 0.99 (0.44, 2.29), FIB-4 of 2.94 (1.49, 6.57), and AST/ALT ratio of 0.72 (0.50, 1.13). Nutritional risk was identified in 138 patients (22.01%). 30% of the enrolled patients were categorized as Child-Pugh grade B or C. FibroScan results showed: F0 (no fibrosis) in 136 patients (21.69%), F1 (mild fibrosis) in 105 patients (16.75%), F2 (moderate fibrosis) in 79 patients (12.6%), F3 (advanced fibrosis) in 117 patients (18.66%), and F4 (cirrhosis) in 190 patients (30.3%) (Table 1).

Table 1 Baseline characteristics (n = 627), n (%)/median (interquartile range).

Variable	Values
Age (years)	47.00 (34.50-55.00)
Sex
Male	438 (69.86)
Female	189 (30.14)
ALT (U/L)	114.00 (39.00-408.50)
AST (U/L)	77.00 (36.00-235.50)
GGT (U/L)	57.00 (28.00-119.00)
PLT (109/L)	130.00 (87.00-173.00)
TBIL (μmol/L)	17.20 (11.00-34.75)
ALB (g/L)	38.20 (34.35-41.85)
TP (g/L)	67.90 (63.70-72.30)
APRI	1.92 (0.83-5.42)
GPR	0.99 (0.44-2.29)
FIB-4	2.94 (1.49-6.57)
AST/ALT	0.72 (0.50-1.13)
NRS2002
NRS2002 < 3	489 (77.9)
NRS2002 ≥ 3	138 (22.01)
Child-Pugh
A	441 (70.33)
B	108 (17.22)
C	78 (12.44)
FibroScan
F0	136 (21.69)
F1	105 (16.75)
F2	79 (12.6)
F3	117 (18.66)
F4	190 (30.3)

APRI: Aspartate aminotransferase to platelet ratio index; GPR: Gamma-glutamyl transferase to platelet ratio index; FIB-4: Fibrosis-4 index; NRS2002: 2002 Nutritional Risk Screening scale; Child-Pugh: Child-Pugh classification; FibroScan: Liver stiffness measurement device (the staging criteria are F0 indicating normal, F4 indicating cirrhosis); AST/ALT: Aspartate aminotransferase to alanine aminotransferase ratio; PLT: Platelet count; TBIL: Total bilirubin; ALB: Albumin; TP: Total protein; GGT: γ-glutamyl transferase.

Correlation between non-invasive fibrosis assessment indicators and FibroScan-measured LSM

The AST/ALT, APRI, GPR, and FIB-4 all showed good correlations with FibroScan-measured LSM (P < 0.001), with GPR and LSM having the strongest correlation (r = 0.609), and AST/ALT and LSM having the weakest correlation (r = 0.175) (Table 2). When examining correlations between the indicators, APRI and FIB-4 exhibited a strong positive association (r = 0.734), GPR and FIB-4 demonstrated a strong positive correlation (r = 0.639), FIB-4 and LSM exhibited a moderate positive correlation (r = 0.553), and APRI and LSM likewise exhibited a moderate positive correlation (r = 0.518) (Figure 2).

Figure 2 Heatmap of correlations between different non-invasive fibrosis assessment indicators and liver stiffness measurement (n = 627). The heatmap displays Spearman correlation coefficients between non-invasive liver fibrosis assessment indicators (the aspartate transaminase to platelet ratio index, the fibrosis-4 index, the gamma-glutamyl transferase to platelet ratio) and liver stiffness measurement results. Warmer (orange-toned) colors indicate stronger positive correlations, while cooler (blue-toned) colors indicate stronger negative correlations. APRI: Aspartate aminotransferase to platelet ratio index; GPR: Gamma-glutamyl transferase to platelet ratio; FIB-4: Fibrosis-4 index; AST/ALT: Aspartate aminotransferase to alanine aminotransferase ratio.

Table 2 Correlation between serological assessment indicators and liver stiffness measurement (n = 627).

First variable	Second variable	Correlation coefficient	P value
LSM	AST/ALT	0.175	< 0.001
	APRI	0.518	< 0.001
	GPR	0.609	< 0.001
	FIB-4	0.553	< 0.001

Spearman’s correlation analysis was used to assess the association between serological indicators and liver stiffness measurement. LSM: Liver stiffness measurement; AST/ALT: Aspartate aminotransferase to alanine aminotransferase ratio; APRI: Aspartate aminotransferase to platelet ratio index; GPR: Gamma-glutamyl transferase to platelet ratio; FIB-4: Fibrosis-4 index.

Analysis of LSM, AST/ALT, APRI, GPR, and FIB-4 across distinct gender, NRS2002 score, and Child-Pugh grade groups

When grouped by gender and compared across various non-invasive liver fibrosis indicators, the results showed significant differences in LSM, GPR, and the AST/ALT between genders. Regarding LSM staging, the proportion of male patients with cirrhosis (F4) was notably higher than that of females (35.6%; 18.0%), whereas the proportion with no fibrosis (F0) was markedly lower in males than in females (17.1%; 32.3%). GPR was notably higher in males relative to females (1.14; 0.67, P < 0.001), while the AST/ALT was higher in females (0.78; 0.68, P < 0.001). APRI and FIB-4 showed no statistically significant differences between genders (Table 3).

Table 3 Comparison of various indicators by gender (n = 627), n (%)/median (interquartile range).

Indices	Female (n = 189)	Male (n = 438)	P value
LSM
F0	61 (32.3)	75 (17.1)	< 0.001
F1	33 (17.5)	72 (16.4)
F2	23 (12.2)	56 (12.8)
F3	38 (20.1)	79 (18)
F4	34 (18)	156 (35.6)
APRI	1.67 (0.80-4.24)	2.03 (0.85-5.95)	0.208
GPR	0.67 (0.28-1.51)	1.14 (0.53-2.65)	< 0.001
FIB-4	2.88 (1.32-6.61)	2.96 (1.53-6.48)	0.853
AST/ALT	0.78 (0.60-1.28)	0.68 (0.48-1.05)	< 0.001

The χ² test was used to analyze the association between gender and liver stiffness measurement results; the Wilcoxon rank-sum test was used to compare the differences in these continuous variables (aspartate aminotransferase to platelet ratio index, gamma-glutamyl transferase to platelet ratio index, fibrosis-4 index, aspartate aminotransferase to alanine aminotransferase ratio) between the female and male groups. LSM: Liver stiffness measurement; APRI: Aspartate aminotransferase to platelet ratio index; GPR: Gamma-glutamyl transferase to platelet ratio index; FIB-4: Fibrosis-4 index; AST/ALT: Aspartate aminotransferase to alanine aminotransferase ratio.

Using NRS2002 scoring results, the 627 enrolled patients were categorized into a group without nutritional risk (NRS2002 score < 3, n = 489) and a group with nutritional risk (NRS2002 score ≥ 3, n = 138). Differences in various non-invasive liver fibrosis indicators between the two groups were compared. It can be observed that LSM, APRI, GPR, FIB-4, and AST/ALT all showed significant differences between the different nutritional risk groups (Table 4). According to FibroScan liver stiffness staging, the proportion of patients with cirrhotic changes (F4 stage) was markedly higher in the nutritional risk group than in the non-nutritional risk group (62.3% vs 21.3%). In contrast, the shares of patients with no fibrosis (F0) or mild fibrosis (F1) were substantially smaller in the nutritional risk group (F0: 4.3% vs 26.6%; F1: 5.8% vs 19.8%) (Table 4). Meanwhile, serum indices including APRI, GPR, FIB-4, and the AST/ALT also reflected more advanced hepatic fibrosis in the nutritional risk group.

Table 4 Comparison of various indicators by nutritional risk status (n = 627), n (%)/median (interquartile range).

Indices	NRS2002 < 3 (n = 489)	NRS2002 ≥ 3 (n = 138)	P value
LSM
F0	130 (26.6)	6 (4.3)	< 0.001
F1	97 (19.8)	8 (5.8)
F2	67 (13.7)	12 (8.7)
F3	91 (18.6)	26 (18.8)
F4	104 (21.3)	86 (62.3)
APRI	1.50 (0.70-3.60)	4.95 (2.38-10.90)	< 0.001
GPR	0.79 (0.37-1.90)	1.84 (0.97-3.26)	< 0.001
FIB-4	2.38 (1.27-4.80)	6.78 (3.80-10.84)	< 0.001
AST/ALT	0.67 (0.48-1.00)	1.02 (0.61-1.54)	< 0.001

The χ² test was used to compare the categorical variable (liver stiffness measurement stages) across nutritional risk groups; the Wilcoxon rank-sum test was used to compare continuous variables (aspartate aminotransferase to platelet ratio index, gamma-glutamyl transferase to platelet ratio index, fibrosis-4 index, aspartate aminotransferase to alanine aminotransferase ratio) between the two groups. LSM: Liver stiffness measurement; APRI: Aspartate aminotransferase to platelet ratio index; GPR: Gamma-glutamyl transferase to platelet ratio index; FIB-4: Fibrosis-4 index; AST/ALT: Aspartate aminotransferase to alanine aminotransferase ratio; NRS2002: Nutritional Risk Screening 2002.

Using the Child-Pugh classification system, enrolled patients were categorized into three subgroups: Grade A (n = 441), grade B (n = 108), and grade C (n = 78). We compared variations in non-invasive hepatic fibrosis indices across these three subgroups. Our analysis revealed that LSM, APRI, GPR, FIB-4, and AST/ALT exhibit statistically significant disparities across distinct liver function classification subgroups (P value < 0.001) (Table 5). Notably, as LSM-based hepatic fibrosis staging advances from Child-Pugh grade A to grade C, the prevalence of cirrhosis (F4 stage) rises markedly (grade A: 14.5%; grade B: 52.8%; grade C: 88.5%), and all four serological hepatic fibrosis indices also increase significantly as the Child-Pugh classification deteriorates.

Table 5 Comparison of various indicators by Child-Pugh classification (n = 627), n (%)/median (interquartile range).

Indices	A grade (n = 441)	B grade (n = 108)	C grade (n = 78)	P value
LSM
F0	131 (29.7)	4 (3.7)	1 (1.3)	< 0.001
F1	98 (22.2)	6 (5.6)	1 (1.3)
F2	65 (14.7)	12 (11.1)	2 (2.6)
F3	83 (18.8)	29 (26.9)	5 (6.4)
F4	64 (14.5)	57 (52.8)	69 (88.5)
APRI	1.33 (0.63-2.82)	3.19 (1.59-10.08)	8.23 (4.29-18.95)	< 0.001
GPR	0.71 (0.32-1.69)	2.19 (0.94-4.56)	1.86 (1.06-3.25)	< 0.001
FIB-4	2.17 (1.23-3.93)	6.03 (3.47-9.22)	8.70 (5.90-16.24)	< 0.001
AST/ALT	0.67 (0.49-0.96)	0.94 (0.56-1.36)	1.10 (0.68-1.61)	< 0.001

The χ² test was used to analyze the association between Child-Pugh classification and the categorical variable (FibroScan stages); the Kruskal-Wallis test was used to compare the differences in continuous variables (aspartate aminotransferase to platelet ratio index, gamma-glutamyl transferase to platelet ratio index, fibrosis-4 index, aspartate aminotransferase to alanine aminotransferase ratio) among different Child-Pugh classification groups. LSM: Liver stiffness measurement; APRI: Aspartate aminotransferase to platelet ratio index; GPR: Gamma-glutamyl transferase to platelet ratio index; FIB-4: Fibrosis-4 index; AST/ALT: Aspartate aminotransferase to alanine aminotransferase ratio.

Comparison and post-hoc analysis of APRI, FIB-4, AST/ALT ratio, and GPR levels based on FibroScan staging

Using FibroScan-assessed fibrosis stages determined by LSM, statistically significant disparities were detected in the AST/ALT, APRI, FIB-4, and GPR across patients with distinct LSM-defined fibrosis stages (P value < 0.001) (Table 6).

Table 6 Aspartate aminotransferase to platelet ratio index, fibrosis-4 index, aspartate aminotransferase to alanine aminotransferase ratio and gamma-glutamyl transferase to platelet ratio based on FibroScan results (n = 627), median (interquartile range).

LSM	APRI	FIB-4	AST/ALT	GPR
F0	0.835 (0.500-1.672)	1.690 (0.992-2.473)	0.670 (0.487-1.008)	0.355 (0.217-0.653)
F1	1.080 (0.500-2.110)	1.600 (0.990-2.950)	0.640 (0.490-0.950)	0.510 (0.250-1.030)
F2	1.500 (0.625-3.980)	2.270 (1.380-3.655)	0.630 (0.465-0.925)	0.830 (0.455-1.710)
F3	2.150 (1.270-5.860)	4.690 (2.500-7.960)	0.830 (0.570-1.160)	1.600 (0.820-2.590)
F4	5.295 (2.212-11.165)	6.320 (3.513-10.688)	0.880 (0.540-1.333)	2.210 (1.132-3.980)
P value	< 0.001	< 0.001	< 0.001	< 0.001

The Kruskal-Wallis test was used to compare the overall differences in indicators across different FibroScan stages. LSM: Liver stiffness measurement; APRI: Aspartate aminotransferase to platelet ratio index; GPR: Gamma-glutamyl transferase to platelet ratio index; FIB-4: Fibrosis-4 index; AST/ALT: Aspartate aminotransferase to alanine aminotransferase ratio.

Findings from post-hoc analyses revealed that the diagnostic discriminatory performance of APRI, FIB-4, AST/ALT ratio, and GPR varied across distinct LSM-based hepatic fibrosis grades (Table 7). Notably, GPR and FIB-4 exhibited the strongest performance, demonstrating statistically significant distinctions not only in differentiating between non-significant/mild fibrosis (stages F0-F2) and moderate/advanced fibrosis (stages F3-F4), but also in differentiating various non-consecutive fibrosis stages (e.g., F0 vs F3, F1 vs F4) with statistical significance. APRI was mainly effective in distinguishing between no fibrosis (F0) and cirrhosis (F4), as well as between mild (F1) or significant (F2) fibrosis and cirrhosis (F4). The AST/ALT had the weakest discriminative ability, showing significance only when comparing early fibrosis stages with extreme cirrhosis stages.

Table 7 Post-hoc analysis of aspartate aminotransferase to platelet ratio index, fibrosis-4 index, aspartate aminotransferase to alanine aminotransferase ratio and gamma-glutamyl transferase to platelet ratio index across different FibroScan-based stages (n = 627).

Analytical index	Comparison group	P value
APRI	F0-F1	1.000
APRI	F0-F2	< 0.01b
APRI	F1-F2	0.264
APRI	F0-F3	< 0.001a
APRI	F1-F3	< 0.001a
APRI	F2-F3	0.137
APRI	F0-F4	< 0.001a
APRI	F1-F4	< 0.001a
APRI	F2-F4	< 0.001a
APRI	F3-F4	< 0.001a
FIB-4	F0-F1	1.000
FIB-4	F0-F2	< 0.05c
FIB-4	F1-F2	0.137
FIB-4	F0-F3	< 0.001a
FIB-4	F1-F3	< 0.001a
FIB-4	F2-F3	< 0.001a
FIB-4	F0-F4	< 0.001a
FIB-4	F1-F4	< 0.001a
FIB-4	F2-F4	< 0.001a
FIB-4	F3-F4	0.082
AST/ALT	F0-F1	1.000
AST/ALT	F0-F2	1.000
AST/ALT	F1-F2	1.000
AST/ALT	F0-F3	0.164
AST/ALT	F1-F3	< 0.05c
AST/ALT	F2-F3	0.090
AST/ALT	F0-F4	< 0.01b
AST/ALT	F1-F4	< 0.01b
AST/ALT	F2-F4	< 0.01b
AST/ALT	F3-F4	1.0000
GPR	F0-F1	0.122
GPR	F0-F2	< 0.001a
GPR	F1-F2	< 0.05c
GPR	F0-F3	< 0.001a
GPR	F1-F3	< 0.001a
GPR	F2-F3	0.009b
GPR	F0-F4	< 0.001a
GPR	F1-F4	< 0.001a
GPR	F2-F4	< 0.001a
GPR	F3-F4	0.115

^aP < 0.001.

^bP < 0.01.

^cP < 0.05.

Post-hoc analysis was used to identify specifically which stages have significant differences. APRI: Aspartate aminotransferase to platelet ratio index; GPR: Gamma-glutamyl transferase to platelet ratio index; FIB-4: Fibrosis-4 index; AST/ALT: Aspartate aminotransferase to alanine aminotransferase ratio.

Multivariable logistic regression analysis: Assessing the independent impacts of predictor variables on fibrosis-related outcomes

Results from multivariable logistic regression analyses revealed that, in both the assessment of any fibrosis (comparing stages F1-F4 to F0) and significant fibrosis (comparing stages F3-F4 to F0-F2), the GPR model exhibited the strongest performance. It achieved corresponding AUC values of 0.816 (95%CI: 0.779-0.853) and 0.871 (95%CI: 0.844-0.899); both outcomes were statistically significant (P value < 0.001). The FIB-4 model showed the next best diagnostic efficacy, with AUCs of 0.791 (0.750-0.831) and 0.825 (0.793-0.857), with P values of (P < 0.001, 0.002), respectively. The APRI model and AST/ALT model showed relatively weaker diagnostic performance (some models were not statistically significant). Child-Pugh classification (Grade B/C vs Grade A) and male gender were significant predictors, while nutritional risk showed no statistical significance in all models (Tables 8 and 9).

Table 8 Multivariate logistic regression results for any fibrosis (F1-F4 vs F0) (n = 627).

Model	Predictor variable	AOR (95%CI)	P value	Model AUC
APRI	The model itself	1.072 (1.018-1.145)	0.021	0.777 (0.736-0.818)
	Age Q2 (vs Q1)	1.071 (0.630-1.819)	0.8
	Age Q3 (vs Q1)	1.495 (0.841-2.686)	0.174
	Age Q4 (vs Q1)	1.783 (0.971-3.343)	0.066
	Male (vs female)	2.096 (1.373-3.199)	< 0.001
	Nutritional risk (yes vs none)	2.049 (0.835-5.910)	0.144
	Child-Pugh B (vs A)	6.455 (2.507-22.031)	< 0.001
	Child-Pugh C (vs A)	9.821 (1.691-187.753)	0.036
GPR	The model itself	1.58 (1.282-2.017)	< 0.001	0.816 (0.779-0.853)
	Age Q2 (vs Q1)	0.981 (0.574-1.676)	0.945
	Age Q3 (vs Q1)	1.324 (0.741-2.388)	0.207
	Age Q4 (vs Q1)	1.492 (0.807-2.811)	0.346
	Male (vs female)	1.728 (1.120-2.663)	0.013
	Nutritional risk (yes vs none)	2.087 (0.852-6.006)	0.133
	Child-Pugh B (vs A)	5.596 (2.186-19.007)	0.001
	Child-Pugh C (vs A)	12.127 (2.125-230.362)	0.024
FIB-4	The model itself	1.193 (1.086-1.329)	< 0.001	0.791 (0.750-0.831)
	Age Q2 (vs Q1)	0.899 (0.526-1.534)	0.697
	Age Q3 (vs Q1)	0.99 (0.543-1.818)	0.975
	Age Q4 (vs Q1)	0.957 (0.485-1.907)	0.899
	Male (vs female)	2.218 (1.446-3.408)	< 0.001
	Nutritional risk (yes vs none)	1.971 (0.799-5.706)	0.169
	Child-Pugh B (vs A)	5.212 (2.011-17.850)	0.002
	Child-Pugh C (vs A)	7.041 (1.192-135.262)	0.074
AST/ALT	The model itself	0.972 (0.623-1.593)	0.906	0.745 (0.703-0.787)
	Age Q2 (vs Q1)	1.071 (0.631-1.818)	0.799
	Age Q3 (vs Q1)	1.393 (0.770-2.546)	0.275
	Age Q4 (vs Q1)	1.77 (0.934-3.419)	0.084
	Male (vs female)	2.106 (1.378-3.221)	< 0.001
	Nutritional risk (yes vs none)	2.186 (0.902-6.253)	0.108
	Child-Pugh B (vs A)	8.232 (3.248-27.872)	< 0.0001
	Child-Pugh C (vs A)	15.884 (2.800-301.739)	0.011

Age was categorized into quartiles: Q1 (≤ 34 years, reference group), Q2 (35-47 years), Q3 (48-55 years), and Q4 (≥ 56 years). This table presents four independent multivariate logistic regression models, each evaluating the independent predictive ability of a different serum non-invasive indicator for any liver fibrosis (F1-F4) vs no fibrosis (F0). The “model AUC” represents the area under the receiver operating characteristic curve for each regression model, indicating its overall diagnostic performance in distinguishing patients with any fibrosis (F1-F4) from those without fibrosis (F0). The area under the curve (AUC) was calculated by deriving a predicted probability for each subject based on the model’s predictor variables, plotting an receiver operating characteristic curve using this probability against the actual fibrosis status (F1-F4 vs F0), and then determining the area under this curve. An AUC value closer to 1 indicates superior discriminatory ability of the model. AOR: Adjusted odds ratio; CI: Confidence interval; AUC: Area under the curve; APRI: Aspartate aminotransferase to platelet ratio index; GPR: Gamma-glutamyl transferase to platelet ratio index; FIB-4: Fibrosis-4 index; AST/ALT: Aspartate aminotransferase to alanine aminotransferase ratio.

Table 9 Multivariate logistic regression analysis results for significant fibrosis (F3-F4 vs F0-F2) (n = 627).

Model	Predictor variable	AOR (95%CI)	P value	Model AUC
APRI	Model itself	1.011 (0.995-1.036)	0.26	0.800 (0.766-0.834)
	Age Q2 (vs Q1)	1.291 (0.780-2.143)	0.322
	Age Q3 (vs Q1)	1.526 (0.903-2.587)	0.115
	Age Q4 (vs Q1)	2.464 (1.445-4.241)	0.001
	Male (vs female)	1.746 (1.178-2.666)	0.006
	Nutritional risk (yes vs none)	1.363 (0.737-2.517)	0.295
	Child-Pugh B (vs A)	6.419 (3.768-11.310)	< 0.001
	Child-Pugh C (vs A)	23.951 (8.361-87.708)	< 0.001
GPR	Model itself	1.551 (1.350-1.806)	< 0.001	0.871 (0.844-0.899)
	Age Q2 (vs Q1)	1.16 (0.688-1.959)	0.578
	Age Q3 (vs Q1)	1.356 (0.786-2.345)	0.274
	Age Q4 (vs Q1)	2.028 (1.158-3.573)	0.014
	Male (vs female)	1.379 (0.901-2.124)	0.141
	Nutritional risk (yes vs none)	1.29 (0.673-2.454)	0.438
	Child-Pugh B (vs A)	4.728 (2.718-8.467)	< 0.001
	Child-Pugh C (vs A)	22.635 (7.834-83.517)	< 0.001
FIB-4	Model itself	1.085 (1.035-1.146)	0.002	0.825 (0.793-0.857)
	Age Q2 (vs Q1)	1.142 (0.686-1.904)	0.61
	Age Q3 (vs Q1)	1.215 (0.706-2.091)	0.481
	Age Q4 (vs Q1)	1.737 (0.976-3.100)	0.061
	Male (vs female)	1.847 (1.224-2.813)	0.004
	Nutritional risk (yes vs none)	1.367 (0.736-2.534)	0.32
	Child-Pugh B (vs A)	5.188 (2.998-9.264)	< 0.001
	Child-Pugh C (vs A)	16.712 (5.779-62.87)	< 0.001
AST/ALT	The model itself	1.344 (0.915-2.021)	0.144	0.787 (0.751-0.822)
	Age Q2 (vs Q1)	1.241 (0.748-2.067)	0.404
	Age Q3 (vs Q1)	1.343 (0.778-2.319)	0.290
	Age Q4 (vs Q1)	2.162 (1.233-3.817)	0.007
	Male (vs female)	1.865 (1.238-2.838)	0.003
	Nutritional risk (yes vs none)	1.329 (0.715-2.464)	0.366
	Child-Pugh B (vs A)	6.35 (3.743-11.160)	< 0.001
	Child-Pugh C (vs A)	24.416 (8.606-88.915)	< 0.001

Age grouping follows Table 8. This table presents four independent multivariate logistic regression models. Each model aims to evaluate the independent predictive ability of different serum non-invasive indicators for significant liver fibrosis (F3-F4) vs no or mild fibrosis (F0-F2). “Model AUC” refers to the area under the receiver operating characteristic (ROC) curve of the regression model, indicating its overall diagnostic performance in distinguishing patients with “significant fibrosis (F3-F4)” from those with “no or mild fibrosis (F0-F2)”. The calculation method is as follows: A predicted probability is computed for each subject based on the model's predictor variables; an ROC curve is then plotted using this predicted probability as the test variable against the actual grouping (F3-F4 vs F0-F2) as the state variable; finally, the area under this curve is calculated. An area under the curve value closer to 1 indicates a stronger discriminatory ability of the model. AOR: Adjusted odds ratio; CI: Confidence interval; AUC: Area under the curve; APRI: Aspartate aminotransferase to platelet ratio index; GPR: Gamma-glutamyl transferase to platelet ratio index; FIB-4: Fibrosis-4 index; AST/ALT: Aspartate aminotransferase to alanine aminotransferase ratio.

Evaluating diagnostic performance through ROC curve analysis: Four non-invasive indices (APRI, FIB-4, GPR, AST/ALT) for distinguishing different hepatic fibrosis stages (subgroup comparisons: F0 vs F1-F4; F0-F2 vs F3-F4)

The diagnostic performance of APRI, GPR, FIB-4, and the AST/ALT in distinguishing liver fibrosis stages was evaluated using receiver operating characteristic curves (Figure 3). To identify cases of any fibrosis (comparing F1-F4 to F0), GPR exhibited the highest area under the ROC curve (AUROC) of 0.814 (95%CI: 0.774-0.853; P value < 0.001). FIB-4 and APRI exhibited moderate AUROC values of 0.743 and 0.736, respectively, whereas the AST/ALT ratio had the lowest AUROC (0.553). With a cutoff value set at 0.715, GPR attained a sensitivity of 72.3%, specificity of 79.4%, positive predictive value (PPV) of 92.7%, negative predictive value (NPV) of 44.3%, and diagnostic accuracy of 73.8% thus outperforming the remaining indices.

Figure 3 Receiver operating characteristic curves of different fibrosis assessment indicators. A: Receiver operating characteristic (ROC) curves for comparing any fibrosis (F1-F4) vs no fibrosis (F0); B: ROC curves for comparing significant fibrosis (F3-F4) vs no or mild fibrosis (F0-F2). The curves correspond to different liver fibrosis assessment indicators: Aspartate transaminase to platelet ratio index, gamma-glutamyl transferase to platelet ratio, fibrosis-4 index, and the aspartate aminotransferase to alanine aminotransferase ratio. The values in the figure represent the area under the curve for each non-invasive liver fibrosis indicator. APRI: Aspartate aminotransferase to platelet ratio index; GPR: Gamma-glutamyl transferase to platelet ratio; FIB-4: Fibrosis-4 index; AST/ALT: Aspartate aminotransferase to alanine aminotransferase ratio; ROC: Receiver operating characteristic.

In distinguishing significant fibrosis (F3-F4) from milder stages (F0-F2), GPR maintained its leading performance with an AUROC of 0.828 (95%CI: 0.796-0.860; P < 0.001). FIB-4 and APRI followed with AUROCs of 0.814 and 0.776, and the AST/ALT again performed inadequately (AUROC = 0.615). At the 0.715 cutoff, GPR’s sensitivity rose to 87.6%, with corresponding specificity, PPV, and NPV of 64.4%, 70.2%, and 84.4%. Its diagnostic accuracy (75.8%) was similar to that of FIB-4 (75.9%) and higher than that of APRI (70.5%) and the AST/ALT (60.9%) (Table 10).

Table 10 Diagnostic performance of various indicators in differentiating F0 vs F1-F4 and F0-F2 vs F3-F4.

Comparison	Indices	AUC (95%CI)	P value	Cut-off	Sensitivity (%)	Specificity (%)	PPV (%)	NPV (%)	DA (%)
F0 and F1-F4	APRI	0.736 (0.69-0.781)	< 0.001	1.45	66.6	70.6	89.1	36.9	67.5
	GPR	0.814 (0.774-0.853)	< 0.001	0.715	72.3	79.4	92.7	44.3	73.8
	FIB-4	0.743 (0.699-0.786)	< 0.001	2.395	67.4	74.3	90.4	38.7	68.9
	AST/ALT	0.553 (0.5-0.606)	0.028	0.825	43.8	68.4	83.3	25.2	49.1
F0-F2 and F3-F4	APRI	0.776 (0.741-0.812)	< 0.001	1.76	72.6	68.4	68.8	72.3	70.5
	GPR	0.828 (0.796-0.86)	< 0.001	0.715	87.6	64.4	70.2	84.4	75.8
	FIB-4	0.814 (0.781-0.848)	< 0.001	3.88	66.4	85	81	72.5	75.9
	AST/ALT	0.615 (0.571-0.659)	< 0.001	0.825	52.1	69.4	62	60.2	60.9

This table shows the diagnostic performance of different indicators in differentiating F0 vs F1-F4 and F0-F2 vs F3-F4. APRI: Aspartate aminotransferase to platelet ratio index; GPR: Gamma-glutamyl transferase to platelet ratio index; FIB-4: Fibrosis-4 index; AST/ALT: Aspartate aminotransferase to alanine aminotransferase ratio; AUC: Area under the curve; CI: Confidence interval; PPV: Positive predictive value; NPV: Negative predictive value; DA: Diagnostic accuracy.

DISCUSSION

Among patients diagnosed with chronic hepatitis B, the present study carried out a thorough evaluation of the diagnostic performance of four non-invasive serological indices (APRI, FIB-4, GPR, and the AST/ALT) alongside FibroScan-measured LSM in the diagnosis of liver fibrosis in patients with chronic hepatitis B. Study findings revealed that GPR exhibited the strongest diagnostic performance, achieving the highest AUC values in differentiating any fibrosis (F0 vs F1-F4) and significant fibrosis (F0-F2 vs F3-F4): 0.814 (95%CI: 0.774-0.853) and 0.828 (95%CI: 0.796-0.86), respectively. FIB-4, APRI, and the AST/ALT ranked sequentially afterward in terms of diagnostic performance. In the multivariate logistic regression results for any fibrosis (F0, F1-F4) and moderate-to-severe fibrosis (F3-F4, F0-F2), the GPR model also performed best, with corresponding AUCs of 0.816 and 0.871, while the FIB-4 model showed the second-highest diagnostic performance with AUCs of 0.791 and 0.825, and the APRI model and AST/ALT model demonstrated relatively weaker diagnostic performance. In the comparison and post-hoc pairwise analysis of APRI, FIB-4, AST/ALT, and GPR indicator levels based on LSM staging, GPR performed outstandingly in multiple comparisons, followed by FIB-4, APRI, and AST/ALT. GPR and FIB-4 were more aligned with LSM in identifying moderate-to-severe liver fibrosis and cirrhosis.

This study’s outcomes align with the results of multiple previous studies. Yan et al[9] found that GPR outperformed APRI and FIB-4 when diagnosing moderate-to-severe fibrosis among the CHB population. Similarly, a study conducted in West Africa reported that GPR was a more accurate, non-invasive indicator of liver fibrosis in CHB patients than APRI and FIB-4, with a markedly stronger predictive value for F3 and F4 stages than APRI and FIB-4. GPR's predictive performance was closest to that of FibroScan-measured LSM[7]. Compared with other non-invasive serum indices, the AST/ALT showed lower diagnostic performance, a finding consistent with previous reports[8]. This may reflect population heterogeneity or different patterns of liver injury. Studies have demonstrated that the AST/ALT performs well in assessing liver fibrosis among patients with hepatitis C[19].

GPR is composed of GGT and platelet count, of which GGT mainly reflects biliary damage and oxidative stress[20], while platelet count is related to the progression of liver fibrosis and portal hypertension[21]. Pairing GGT with platelet count can more comprehensively evaluate changes in liver fibrosis, which may explain why GPR performs more robustly in the diagnosis of liver fibrosis.

Some important clinical phenomena were also identified in this study. The FibroScan results indicated that approximately 30% of patients had developed cirrhosis (stage F4), which was highly consistent with the share of Child-Pugh class B and C patients. This finding suggests that FibroScan and Child-Pugh gradings are consistent in the evaluation of advanced liver disease, thereby providing a practical basis for the promotion of non-invasive, low-cost screening tools in primary care settings. Additionally, we observed that the percentage of male patients with cirrhosis (stage F4) was notably higher than that of female patients (35.6% vs 18.0%), and GPR and other indicators were also higher in males. These patterns may be attributed to factors including differences in sex hormones, immunomodulation, and lifestyle. Studies suggest that androgens can suppress immune function, reducing the body’s ability to clear the virus, prolonging infection, and contributing to chronic liver disease. This persistent injury to hepatocytes and stimulation of hepatic stellate cells may worsen liver fibrosis[22]. Additionally, research indicates that male-predominant habits such as smoking and alcohol consumption can accelerate the advancement of liver fibrosis[23]. These observations underscore the necessity for more proactive surveillance and intervention targeting liver fibrosis in male CHB patients within clinical care.

Beyond this, we examined how gender, nutritional status, and Child-Pugh classification influence the performance of non-invasive diagnostic markers for liver fibrosis. Findings from univariate analysis revealed that male patients, those with nutritional risk (as defined by NRS2002 scores ≥ 3), and individuals with elevated Child-Pugh classification were closely associated with the severity of liver fibrosis. Notably, though, when we applied a multivariate logistic regression model, nutritional risk did not emerge as an independent predictor of liver fibrosis severity. This difference may be because nutritional risk does not directly cause fibrosis but rather does so indirectly through poor liver function. When nutritional risk is included in the model alongside other predictive factors, such as Child-Pugh grading, its role is not obvious. The single-factor analysis shows a higher rate of cirrhosis in the nutritional risk group (62.3% vs 21.3%). This obvious association has important clinical significance, indicating that malnutrition may affect the progression of liver fibrosis, which is consistent with the latest consensus of the American Journal of Gastroenterology[24]. We can also think in reverse: Patients with cirrhosis are more prone to malnutrition. Studies have indicated that the reduction of serum albumin concentration is one of the long-term recognized characteristics of patients with cirrhosis[25]. Cirrhosis can cause extensive loss of functional liver tissue, induce persistent inflammation, and impair albumin synthesis capacity[26]. In addition, cirrhosis is often complicated by portal hypertension, which may lead to complications such as gastrointestinal stasis and edema. This, in turn, impairs the barrier function of the gastrointestinal mucosa and reduces digestive enzyme secretion, thereby affecting the digestion and absorption of nutrients, including carbohydrates, fats, and proteins. Furthermore, ascites can compress the gastrointestinal tract in some patients, making it more difficult for them to eat. Patients with cirrhosis are in a state of chronic inflammation for a long time. Studies have shown that cirrhosis can increase levels of inflammatory factors tumor necrosis factor alpha and interleukin-6, accelerate the body’s energy metabolism, and cause the body to consume more[27,28]. The above factors together cause patients with cirrhosis to fall into the vicious circle of “reduced synthesis, impaired absorption, and increased consumption”, finally leading to malnutrition among CHB individuals. Against this backdrop, in clinical practice, individuals with cirrhosis should not only conduct routine monitoring of liver fibrosis progression, but also regularly assess nutritional status using tools such as the NRS2002. Alongside this, healthcare providers should promptly offer these patients enteral nutritional support or targeted dietary interventions. These measures are intended to break the aforementioned vicious cycle and enhance the nutritional status of affected individuals.

The advantage of this study is that it systematically investigates the impacts of gender, nutritional status, and Child-Pugh grading on the diagnostic efficacy of multiple non-invasive liver fibrosis indicators (including APRI, FIB-4, GPR, and AST/ALT), based on clinical data from Southwest China. This, in turn, lays a new foundation for personalized clinical evaluation. Additionally, through integrating multivariate regression and ROC curve analysis, this study thoroughly assesses how each marker performs diagnostically across distinct liver fibrosis stages. Thereby providing a reference for selecting efficient, practical, non-invasive screening tools in primary care settings. Nevertheless, this study has notable limitations: As a single-center cross-sectional investigation, it cannot monitor longitudinal changes in patients’ liver fibrosis status, assess fibrosis progression or regression, or easily determine the causal relationships among variables. Moreover, the study used FibroScan rather than liver biopsy to assess liver fibrosis, lacking the histopathological “gold standard” as a reference. Although liver biopsy is associated with risks such as bleeding, pain, sampling error, and low patient acceptance, some studies have indicated that FibroScan and liver biopsy yield relatively consistent results in fibrosis evaluation[29]. Other studies have shown that liver biopsy samples represent only 1/50000 of the total liver tissue, resulting in significant sampling error, whereas FibroScan has almost no sampling error[30]. Samples for this study were sourced from a clinical medical center in the southwest of China; as such, they do not capture the characteristics of the wider population across the entire country. Future research should conduct multicenter, prospective, larger-scale, and more diverse cohort studies to validate the diagnostic efficacy of GPR and FIB-4 across different regions and populations in China; it is also necessary to implement longitudinal studies to assess the predictive ability of these non-invasive hepatic fibrosis indices for the progression of fibrosis. Finally, this study did not incorporate liver biopsy as the gold standard for direct comparative analysis with non-invasive liver fibrosis indicators such as FibroScan, GPR, and FIB-4, making it impossible to further compare the consistency between non-invasive indicators and the “gold standard” diagnostic results, and thus difficult to more precisely validate the diagnostic efficacy of non-invasive indicators.

CONCLUSION

Our study results indicate that, among the CHB population in Southwestern China, the non-invasive diagnostic tools GPR and FIB-4 exhibit superior performance in evaluating liver fibrosis, particularly for initial screening of advanced fibrosis and cirrhosis. In primary care settings, GPR and FIB-4 can serve as first-line non-invasive screening tools and be integrated into routine liver disease management. Additionally, this study revealed that factors such as gender, nutritional risk, and Child-Pugh grade were strongly linked to the extent of liver fibrosis. These findings imply that clinical evaluations should comprehensively consider individual patient characteristics, particularly nutritional status and liver function. Future multicenter prospective studies should be undertaken to further validate the effectiveness and applicability of these indicators in real-world practice.

ACKNOWLEDGEMENTS

Thanks to Dr. Zhu Chen, Lin Wang, Chuang-Jie Mao and Li Wang (the First Ward, the Second Ward and the Third Ward of liver disease, Public Health Clinical Center of Chengdu).