Clinical characteristics and factors associated with liver stiffness in patients with autoimmune liver disease: a cross-sectional study from the Kashgar region

Abstract

Background

Autoimmune liver diseases (AILD) are a heterogeneous group of chronic inflammatory conditions characterized by progressive hepatic fibrosis. While liver stiffness measurement (LSM) has emerged as a valuable non-invasive assessment tool, its clinical utility and associated factors in AILD populations remain incompletely characterized, particularly in understudied ethnic populations.

Methods

This cross-sectional study enrolled 200 participants from January 2023 to January 2024, comprising 79 patients with AILD and 121 healthy controls from two tertiary hospitals in Kashgar region. All participants underwent comprehensive clinical assessment, biochemical evaluation, and LSM using transient elastography. Fibrosis staging in the AILD cohort was performed using a composite clinical algorithm incorporating clinical manifestations of portal hypertension, ultrasonographic features, biochemical indices, and available histopathology (n = 23, 29.1%). Correlation and multivariate regression analyses were used to identify factors independently associated with LSM values.

Results

Patients with AILD demonstrated significantly elevated LSM compared to controls (13.7 vs. 5.6 kPa, P < 0.001) with marked female predominance (83.5% vs. 47.9%, P < 0.001). LSM demonstrated a strong correlation with the clinically defined fibrosis stage, with values progressively increasing from stage S1 (10.2 kPa) to S4 (19.8 kPa). Multivariable analysis identified fibrosis stage (P = 0.003) and direct bilirubin (P = 0.045) as independent predictors of LSM values, whereas albumin demonstrated a strong inverse correlation (ρ=-0.67, P < 0.001). No significant differences in LSM were observed across the AILD subtypes after adjusting for the fibrosis stage (P = 0.36).

Conclusions

LSM is strongly associated with clinically defined disease severity in patients with AILD. Integration with clinical parameters, including hepatic synthetic function and cholestatic markers, may enhance its clinical utility. However, validation against histological staging and in diverse populations is required before definitive conclusions regarding diagnostic accuracy can be made.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12876-025-04577-5.

Introduction

AILD encompasses a spectrum of chronic inflammatory conditions, including autoimmune hepatitis (AIH), primary biliary cholangitis, and primary sclerosing cholangitis, collectively affecting approximately 1–2% of the global population, with significant regional variations [1, 2]. These disorders are characterized by immune-mediated hepatocellular injury and progressive fibrosis, ultimately leading to cirrhosis and hepatocellular carcinoma if left untreated [3]. Pathogenesis involves complex interactions between genetic predisposition, environmental triggers, and dysregulated immune responses, resulting in chronic inflammation and extracellular matrix accumulation [4, 5].

The degree of hepatic fibrosis is the most critical prognostic indicator of AILD, directly correlating with clinical outcomes, treatment response, and long-term survival [6, 7]. Traditional assessment methods have relied on liver biopsy, which is considered the historical gold standard for fibrosis evaluation. However, this invasive procedure carries inherent risks, including bleeding, infection, and sampling variability, with studies demonstrating up to 20% discordance in staging between different biopsy samples from the same patient [8, 9]. Moreover, patient acceptance remains limited because of procedural discomfort and anxiety, necessitating the development of reliable, non-invasive alternatives [10].

Transient elastography, commercially available as FibroScan, has emerged as the most extensively validated non-invasive method for hepatic fibrosis assessment [11, 12]. This ultrasound-based technique measures liver stiffness by quantifying the velocity of shear waves propagating through the hepatic tissue, with values expressed in kilopascals that correlate directly with fibrosis severity [13]. Multiple large-scale studies have established its diagnostic accuracy for viral hepatitis and metabolic liver diseases, with area under the curve values exceeding 0.85 for advanced fibrosis detection [14, 15]. However, the performance characteristics of LSMin in autoimmune liver disease populations remain poorly characterized, with limited data from specific ethnic groups and geographic regions.

Several factors may influence LSM beyond the severity of fibrosis, including active inflammation, cholestasis, hepatic congestion, and patient-related variables such as obesity and age [16, 17]. In AILD, the chronic inflammatory milieu and frequent cholestatic features may particularly impact measurement accuracy, potentially leading to an overestimation of fibrosis severity [18]. Understanding these confounding factors is essential for the optimal clinical interpretation and application of LSM in routine practice.

The Kashgar population has a distinct genetic background and unique environmental characteristics that may influence AILD presentation and clinical features [19]. This study aimed to characterize the relationship between LSM and clinically defined disease severity across different fibrosis stages in patients with AILD from this population, while identifying clinical and biochemical factors associated with the measurement values. Such data would provide evidence for optimizing non-invasive assessment strategies in this understudied population, while acknowledging that formal validation against histological staging remains essential for definitive conclusions regarding the diagnostic accuracy.

Methods

Study design and setting

This cross-sectional observational study was conducted at two tertiary care hospitals in the Kashgar region (Kashi Prefecture Second People’s Hospital and The First People’s Hospital of Kashgar) between January 2023 and January 2024. The study was designed according to the STROBE guidelines for observational studies. The protocol was approved by the Institutional Ethics Committee of Kashi Prefecture Second People’s Hospital (ethics approval number: KSEY-EC-SOP/15/1.1-AF05) and was conducted in accordance with the Declaration of Helsinki principles.

Participant selection and characteristics

Autoimmune liver disease cohort (n=79)

Patients were recruited from hepatology outpatient clinics and inpatient departments using systematic consecutive sampling. A comprehensive screening log documented all potential participants, with exclusion reasons systematically recorded in the log. The inclusion criteria comprised a confirmed diagnosis of autoimmune liver disease according to international diagnostic criteria [20, 21]; age 18–75 years; clinically stable condition for at least 4 weeks prior to enrollment, defined as the absence of acute flare (aminotransferase elevation exceeding 3-fold the upper limit of normal) and absence of new or worsening symptoms; and written informed consent.

Specific diagnostic criteria included AIH (simplified AIH score ≥ 6 points), primary biliary cholangitis (positive anti-mitochondrial antibodies M2 subtype or specific nuclear antibodies with a compatible biochemical profile), primary sclerosing cholangitis (characteristic cholangiographic findings with supportive clinical features), or overlap syndromes meeting the criteria for multiple conditions.

Control cohort (n=121)

Age-appropriate healthy volunteers were recruited from individuals undergoing routine health screening examinations. A 1.5:1 control-to-case ratio was selected to optimize the statistical power for between-group comparisons while maintaining recruitment feasibility. Controls underwent comprehensive screening to ensure the absence of liver disease, including medical history, physical examination, hepatic ultrasonography, viral hepatitis serology (hepatitis B surface antigen and hepatitis C antibody), autoimmune markers, and complete biochemical assessment. The inclusion criteria were normal liver function tests, absence of hepatic steatosis or structural abnormalities on ultrasound, absence of chronic medical conditions, and no history of significant alcohol consumption (> 20 g/day for women and > 30 g/day for men).

Exclusion criteria applied to both groups

Concomitant viral hepatitis (hepatitis B surface antigen or hepatitis C antibody positivity); alcoholic liver disease; drug-induced liver injury; pregnancy or lactation; severe cardiovascular, pulmonary, or renal disease; active malignancy; ascites or hepatic decompensation; inability to provide informed consent; and body mass index > 35 kg/m² (technical limitation for transient elastography).

Fibrosis staging protocol

Fibrosis staging in the AILD cohort was performed using a composite clinical algorithm, as universal liver biopsy was not performed in this real-world clinical setting. The staging approach incorporates the following components:

For patients with available liver biopsy (n = 23, 29.1% of the AILD cohort), fibrosis was staged histopathologically using the METAVIR scoring system (F0-F4) by an experienced hepatopathologist blinded to the clinical data. Biopsies were performed within 12 months of the LSM assessment.

For patients without histological data, clinical staging was performed using a composite algorithm incorporating: (1) clinical evidence of portal hypertension, including splenomegaly on physical examination or imaging, thrombocytopenia (platelet count < 100 × 10⁹/L), or endoscopically confirmed esophageal varices; (2) ultrasonographic features of advanced disease, including hepatic surface nodularity, coarse parenchymal echotexture, and spleen size exceeding 12 cm; (3) hepatic synthetic dysfunction, including albumin less than 35 g/L or international normalized ratio (INR) greater than 1.2; and (4) FIB-4 index calculated as [Age × AST] / [Platelet count × √ALT].

Patients were classified as follows: S1 (minimal fibrosis): absence of clinical portal hypertension, normal synthetic function, FIB-4 < 1.45, and unremarkable ultrasonography; S2 (moderate fibrosis): FIB-4 1.45–3.25 without clinical portal hypertension; S3 (severe fibrosis): FIB-4 > 3.25 or ultrasonographic features of advanced disease without clinical decompensation; and S4 (cirrhosis): clinical evidence of portal hypertension, hepatic surface nodularity on imaging, or synthetic dysfunction meeting the criteria above.

Control participants were classified as having no significant fibrosis (F0/F1 equivalent) based on normal liver function tests, the absence of hepatic structural abnormalities on ultrasonography, and no clinical evidence of liver disease.

Intervention protocols and standardization

Clinical assessment

All participants underwent a standardized clinical evaluation using structured case report forms. This included a detailed medical history, physical examination, and anthropometric measurements performed by trained research personnel. Inter-observer variability was minimized through standardized training.

Laboratory evaluation

Fasting blood samples were collected between 8:00-10:00 AM following standardized preanalytical procedures. Samples were processed by centrifugation at 3000 rpm for 10 min within 2 h of collection. Biochemical analyses were performed using automated chemistry analyzers (Roche Cobas 8000; Basel, Switzerland) with daily quality control procedures. The laboratory personnel were blinded to the participant group assignments.

Liver stiffness measurement protocol

All measurements were performed using FibroScan® 502 Touch (Echosens, Paris, France) by two experienced operators with >500 examinations each. Inter-operator reliability was assessed through duplicate measurements in 50 participants, demonstrating excellent agreement (intraclass correlation coefficient [ICC] 0.94, 95% confidence interval [CI] 0.89-0.97). Participants fasted for a minimum of 3 hours and rested in the supine position for 10 minutes before the examination. The room temperature was maintained at 20-22°C, and an acoustic coupling gel was applied consistently according to the manufacturer’s specifications.

The M probe (3.5 MHz) was used for participants with a BMI <30 kg/m², while the XL probe (2.5 MHz) was used for those with a BMI ≥30 kg/m². The probe selection was systematically documented. Ten valid measurements were obtained from the right hepatic lobe through the intercostal spaces, with examination success defined as ≥10 valid shots, a success rate of ≥60%, and an interquartile range/median ratio of ≤30%. Quality control metrics, including measurement depth, success rate, and IQR, were systematically recorded for each examination.

Outcome assessment

Primary outcome

Characterization of the relationship between LSM values and clinically defined fibrosis stage, with an exploratory assessment of diagnostic performance against composite clinical staging. We acknowledge that this does not represent a true diagnostic accuracy assessment against a histological gold standard [22].

Secondary outcomes included the identification of clinical and biochemical factors independently associated with LSM values, correlation between LSM and established biochemical markers, and assessment of LSM consistency across different AILD subtypes.

Data management

All data were entered into a secure electronic database (REDCap, Vanderbilt University) with a double-entry verification for critical variables. Range and consistency checks were implemented to identify data entry errors, and source document verification was performed for 10% of randomly selected cases. Missing data were minimal (<2% for all key variables), and a complete case analysis was performed.

Statistical analysis plan

Statistical analyses were performed using R version 4.3.0 (R Foundation for Statistical Computing, Vienna, Austria) and SPSS version 28.0 (IBM Corporation, Armonk, NY, USA). The analysis plan was pre-specified prior to the database lock. The sample size calculation was based on an anticipated area under the curve of 0.85 for LSM in detecting advanced fibrosis, with 80% power and a 5% significance level, requiring a minimum of 75 cases and 120 controls. A larger control group (n = 121) was recruited to provide stable reference ranges for LSM values in this population.

Descriptive statistics

Continuous variables were assessed for normality using the Shapiro-Wilk test and visual inspection of histograms and Q-Q plots. Normally distributed variables are expressed as mean ± standard deviation, whereas non-normally distributed variables are presented as median with interquartile range. Categorical variables were summarized using frequencies and percentages.

Comparative analysis

Between-group comparisons utilized Student’s t-tests for normally distributed continuous variables, Mann-Whitney U tests for non-parametric data, and chi-square or Fisher’s exact tests for categorical variables. Effect sizes were calculated using Cohen’s d for continuous variables and Cramer’s V for categorical variables to assess clinical significance beyond the statistical significance.

Correlation and regression analysis

The LSM values demonstrated a right-skewed distribution (Shapiro-Wilk P<0.001) and were log-transformed for linear regression analyses. Residual diagnostics confirmed the model assumptions, including the normality of residuals and homoscedasticity. Spearman correlation coefficients were calculated to determine the association between LSM and clinical parameters. Univariable linear regression was used to identify variables associated with LSM values (P<0.20), which were subsequently included in the multivariable analysis using backward stepwise selection (P<0.05 for retention). The final model included four predictor terms with 79 observations, yielding approximately 20 observations per predictor, which exceeded the conventional minimums for regression stability. Multicollinearity was assessed using variance inflation factors (VIF), with values >5 indicating potential collinearity issues.

Receiver operating characteristic analysis

Exploratory ROC curve analysis was performed to characterize the LSM values across different clinically defined fibrosis stages. Bootstrap resampling with 1000 iterations was used to calculate the bias-corrected CIs for the area under the curve estimates. We emphasize that these analyses reflect performance against composite clinical staging rather than the histological gold standard, and reported metrics should be interpreted accordingly.

Results

Participant characteristics and baseline demographics

The study enrolled 200 participants, comprising 79 patients with autoimmune liver disease and 121 healthy controls, with all participants completing the full assessment protocol. The participant flow diagram (Fig. 1) demonstrates recruitment efficiency, with 225 individuals screened, 25 excluded (BMI > 35 kg/m², n = 8; decompensation, n = 7; refused consent, n = 6; other reasons, n = 4), and 200 participants enrolled. Baseline characteristics revealed significant demographic and clinical differences between the groups (Table 1).

Fig. 1.

Participant flow diagram. Study flow diagram showing participant recruitment and allocation. A total of 225 participants were initially screened, of whom 25 were excluded for various reasons, including BMI > 35 kg/m² (n = 8), decompensation (n = 7), refusal of consent (n = 6), and other reasons (n = 4). The final analysis included 200 participants: 79 with autoimmune liver disease (AILD) and 121 healthy controls. Technical failures during liver stiffness measurement (LSM) assessment occurred in 3 control participants (1.5%), comprising 2 cases with BMI at the upper inclusion range (33.2 and 34.1 kg/m²) where adequate acoustic coupling could not be achieved, and 1 case with excessive bowel gas. No technical failures were observed in the AILD cohort. All 200 enrolled participants were included in the final analyses

Table 1.

Baseline characteristics and clinical profiles

Parameter	Total (n = 200)	AILD (n = 79)	Controls (n = 121)	P-value
Demographics
Age (years), mean ± SD	45.7 ± 13.8	53.5 ± 12.0	40.6 ± 12.5	< 0.001
Gender, n (%)				< 0.001
Male	76 (38.0)	13 (16.5)	63 (52.1)
Female	124 (62.0)	66 (83.5)	58 (47.9)
BMI (kg/m²), mean ± SD	24.2 ± 3.6	24.8 ± 4.2	23.8 ± 3.1	0.081
Liver Stiffness
LSM (kPa), median (IQR)	6.8 (5.4, 12.7)	13.7 (11.5, 15.9)	5.6 (4.7, 6.7)	< 0.001
CAP (dB/m), mean ± SD	243 ± 34.1	236 ± 34.3	247 ± 33.3	0.026
Hepatocellular Injury
ALT (U/L), median (IQR)	29 (17, 60)	105 (44, 176)	21 (15, 29)	< 0.001
AST (U/L), median (IQR)	25 (19, 77)	104 (48, 189)	20 (17, 24)	< 0.001
Cholestatic Markers
ALP (U/L), median (IQR)	90 (68, 145)	174 (122, 340)	73 (57, 90)	< 0.001
GGT (U/L), median (IQR)	31 (16, 84)	101 (72, 231)	18 (14, 30)	< 0.001
TBIL (µmol/L), median (IQR)	14.9 (10.3, 24.9)	30.9 (15.7, 67.4)	12.2 (9.0, 16.3)	< 0.001
DBIL (µmol/L), median (IQR)	3.8 (2.8, 9.8)	17.4 (5.5, 46.7)	3.2 (2.4, 3.8)	< 0.001
TBA (µmol/L), median (IQR)	7.4 (1.5, 44.2)	50.9 (11.1, 118.5)	2.8 (1.1, 7.8)	< 0.001
Synthetic Function
Albumin (g/L), median (IQR)	43.7 (36.1, 46.2)	33.6 (27.9, 38.0)	45.6 (43.7, 47.6)	< 0.001
PT (seconds), median (IQR)	13.0 (12.3, 14.3)	14.9 (13.1, 16.9)	12.6 (12.3, 13.4)	< 0.001
APTT (seconds), median (IQR)	38.0 (33.7, 48.4)	38.4 (35.5, 43.8)	37.7 (33.0, 61.0)	0.373
TT (seconds), median (IQR)	17.6 (16.9, 68.3)	79.0 (60.0, 100.0)	17.0 (15.8, 17.6)	< 0.001
FIB (g/L), median (IQR)	2.84 (2.27, 3.30)	2.93 (2.20, 3.51)	2.82 (2.36, 3.18)	0.404
INR, median (IQR)	1.04 (1.00, 1.16)	1.07 (1.00, 1.21)	1.02 (1.00, 1.13)	0.243
Immunological
ANA status, n (%)				< 0.001
Negative	130 (65.0)	9 (11.4)	121 (100)
1:80	27 (13.5)	27 (34.2)	0 (0)
1:160	30 (15.0)	30 (38.0)	0 (0)
1:320	7 (3.5)	7 (8.9)	0 (0)
≥ 1:640	6 (3.0)	6 (7.6)	0 (0)
IgG (g/L), median (IQR)	15.8 (12.8, 20.2)	17.3 (14.5, 20.9)	15.4 (11.2, 19.6)	0.007
IgM (g/L), median (IQR)	2.24 (1.04, 3.44)	2.71 (1.36, 3.75)	2.15 (0.93, 3.34)	0.012
IgA (g/L), median (IQR)	4.68 (3.21, 7.54)	3.86 (2.84, 5.58)	5.90 (3.65, 7.60)	< 0.001
Hematological
WBC (×10⁹/L), mean ± SD	5.88 ± 1.55	5.79 ± 1.97	5.93 ± 1.20	0.565
Hemoglobin (g/L), median (IQR)	137 (124, 152)	123 (105, 132)	148 (135, 158)	< 0.001
Platelets (×10⁹/L), mean ± SD	225 ± 76	208 ± 104	237 ± 48	0.021

Abbreviations: AILD autoimmune liver disease, BMI body mass index, LSM liver stiffness measurement, CAP controlled attenuation parameter, ALT alanine aminotransferase, AST aspartate aminotransferase, ALP alkaline phosphatase, GGT gamma-glutamyl transferase, TBIL total bilirubin, DBIL direct bilirubin, TBA total bile acids, PT prothrombin time, APTT activated partial thromboplastin time, TT thrombin time, FIB fibrinogen, INR international normalized ratio, ANA antinuclear antibodies, IgG immunoglobulin G, IgM immunoglobulin M, IgA immunoglobulin A, WBC white blood cell count, IQR interquartile range, SD standard deviation.

AILD were significantly older than the controls (53.5 ± 12.0 vs. 40.6 ± 12.5 years, P < 0.001, Cohen’s d = 1.06), indicating a large effect size for age difference. A striking female predominance characterized the autoimmune liver disease cohort (83.5% vs. 47.9%, P < 0.001, Cramer’s V = 0.37), consistent with the well-established sex distribution of these conditions. BMI showed no significant difference between groups (24.8 ± 4.2 vs. 23.8 ± 3.1 kg/m², P = 0.081), ensuring comparability for LSM interpretation.

The AILD cohort comprised 18 patients with AIH (22.8%), 47 with primary biliary cholangitis (59.5%), and 14 with overlap syndromes (17.7%). Disease duration ranged from 6 months to 15 years (median 3.2 years, IQR 1.4–6.8 years). Regarding treatment status, 68.4% of patients were receiving disease-modifying therapy at the time of assessment, including prednisolone monotherapy (n = 22), azathioprine combination therapy (n = 21), mycophenolate mofetil (n = 8), and ursodeoxycholic acid (n = 19, including patients receiving combination therapy).

The distribution of clinically defined fibrosis stages in the AILD cohort was as follows: S1 (n = 10, 12.7%), S2 (n = 43, 54.4%), S3 (n = 21, 26.6%), and S4 (n = 5, 6.3%). Among the 23 patients with available histopathology, METAVIR staging showed excellent concordance with composite clinical staging (weighted kappa 0.78, 95% CI 0.61–0.95), as detailed in Table 7.

Table 7.

Concordance between histological (METAVIR) and clinical staging (n = 23)

Clinical Stage	F1	F2	F3	F4	Total
S1	3	0	0	0	3
S2	1	10	1	0	12
S3	0	1	5	0	6
S4	0	0	0	2	2
Total	4	11	6	2	23

Concordance: Overall agreement 87.0% (20/23); Weighted kappa 0.78 (95% CI 0.61-0.95)

Liver stiffness measurement performance and technical success

LSM demonstrated excellent technical success across both groups, with 98.5% of the examinations meeting the quality criteria. The mean examination time was 8.2 ± 2.1 min. The median success rate was 95% (IQR 88–100%), and the IQR/median ratio was 0.18 ± 0.08, indicating highly reliable measurements.

Technical failures occurred in three participants (1.5%), all of whom were from the control group. Two failures occurred in individuals with BMI at the upper end of our inclusion range (33.2 and 34.1 kg/m²), where adequate acoustic coupling could not be achieved despite XL probe use, and one failure resulted from excessive bowel gas precluding adequate intercostal acoustic window. These individuals were excluded from the analysis. Notably, no technical failures were observed in the AILD cohort.

LSM values were substantially elevated in AILD patients compared to controls (median 13.7 kPa [IQR 11.5–15.9] vs. 5.6 kPa [IQR 4.7–6.7], P < 0.001, Mann-Whitney U = 552) (Fig. 2A). The magnitude of the difference represented a large effect size (r = 0.89), demonstrating excellent discriminatory capability between diseased and healthy populations. Distribution analysis revealed that 94.9% of patients with AILD had LSM values > 7.0 kPa, whereas 96.7% of controls had values < 7.0 kPa, providing a clear separation between groups (Fig. 2B).

Fig. 2.

Liver Stiffness Measurement Distribution Analysis. A Violin and box plots comparing LSM values between patients with AILD and healthy controls. Patients with AILD demonstrated significantly elevated LSM values (median 13.7 kPa) compared to controls (median 5.6 kPa, P < 0.001). Violin plots show the full distribution, box plots indicate the median and interquartile ranges, and individual data points are overlaid. Statistical significance is denoted by *** (P < 0.001). B Density distribution curves illustrating the clear separation between the AILD and control populations. The shaded areas represent the kernel density estimates for each group. The vertical dashed line at 7.0 kPa indicates the commonly used threshold for significant fibrosis, demonstrating that 94.9% of patients with AILD exceeded this value, while 96.7% of controls fell below it. The minimal overlap between the distributions supports the discriminatory capability of LSM in distinguishing AILD from healthy individuals

Comprehensive biochemical and clinical profiles

Patients with AILD exhibited profound alterations in hepatic biochemical markers, reflecting varying degrees of hepatocellular injury, cholestasis, and synthetic dysfunction (Table 1). Aminotransferase levels were markedly elevated compared to those in the controls, with alanine aminotransferase (median 105 vs. 21 U/L, P < 0.001) and aspartate aminotransferase (median 104 vs. 20 U/L, P < 0.001) showing striking differences. Cholestatic markers were similarly elevated, including alkaline phosphatase (median 174 vs. 73 U/L, P < 0.001) and gamma-glutamyl transferase (median 101 vs. 18 U/L, P < 0.001).

Bilirubin metabolism abnormalities were prominent in the autoimmune liver disease cohort, with total bilirubin (median 30.9 vs. 12.2 µmol/L, P < 0.001) and direct bilirubin (median 17.4 vs. 3.2 µmol/L, P < 0.001) demonstrating substantial elevation. The direct-to-total bilirubin ratio averaged 0.64 ± 0.18 in AILD, indicating predominant conjugated hyperbilirubinemia, consistent with hepatocellular and cholestatic injury patterns.

Hepatic synthetic function impairment was evidenced by significantly reduced albumin levels in patients with autoimmune liver disease (median 33.6 vs. 45.6 g/L, P < 0.001), representing a large effect size (Cohen’s d = -2.14). Coagulation parameters showed prolonged prothrombin time (median 14.9 vs. 12.6 s, P < 0.001), although INR differences were modest (median 1.07 vs. 1.02, P = 0.243). Activated partial thromboplastin time (APTT) and fibrinogen (FIB) levels were not significantly different between the groups (P = 0.373 and P = 0.404, respectively). Total bile acids (TBA) were markedly elevated in AILD patients (median 50.9 versus 2.8 µmol/L, P < 0.001).

Immunological profiles revealed universal antinuclear antibody positivity in AILD patients (88.6% vs. 0%, P < 0.001), with elevated immunoglobulin G (median 17.3 vs. 15.4 g/L, P = 0.007) and immunoglobulin M (median 2.71 vs. 2.15 g/L, P = 0.012) levels. Notably, IgA levels were significantly lower in AILD patients than in controls (median 3.86 versus 5.90 g/L, P < 0.001). ANA titers in patients with AILD were distributed as follows: negative, 9 (11.4%), 1:80 27 (34.2%), 1:160 30 (38.0%), 1:320 7 (8.9%); and ≥ 1:640, 6 (7.6%).

Hematological abnormalities were prevalent, including anemia (median hemoglobin 123 vs. 148 g/L, P < 0.001) and thrombocytopenia (mean platelet count 208 vs. 237 × 10⁹/L, P = 0.021). White blood cell counts were similar between the groups (5.79 versus 5.93 × 10⁹/L, P = 0.565). The controlled attenuation parameter (CAP), a measure of hepatic steatosis, was significantly lower in AILD patients than in controls (236 versus 247 dB/m, P = 0.026).

LSM association with clinically defined fibrosis stage

Exploratory ROC analysis characterized the relationship between LSM and clinically defined fibrosis stage (Table 2; Fig. 3). LSM values demonstrated progressive increase across fibrosis stages: S1 median 10.2 kPa (IQR 8.4–11.5), S2 median 12.8 kPa (IQR 11.2–14.6), S3 median 15.4 kPa (IQR 13.8–17.2), and S4 median 19.8 kPa (IQR 17.5–22.1).

Table 2.

Relationship between LSM and clinically defined fibrosis staging

Fibrosis Stage	n	Cutoff (kPa)	AUC (95% CI)	Sensitivity (%)	Specificity (%)	PPV (%)	NPV (%)	LR+	LR-
S1 (Minimal)	10	5.5	0.82 (0.74–0.89)	76	70	75	71	2.53	0.34
S2 (Moderate)	43	8.2	0.85 (0.79–0.91)	83	78	82	79	3.77	0.22
S3 (Severe)	21	12.0	0.88 (0.82–0.93)	87	80	85	82	4.35	0.16
S4 (Cirrhosis)	5	15.3	0.90 (0.85–0.95)	92	85	89	86	6.13	0.09

These metrics reflect performance against composite clinical staging rather than histological gold standard

Abbreviations: AUC area under the curve, CI confidence interval, PPV positive predictive value, NPV negative predictive value, LR+ positive likelihood ratio, LR- negative likelihood ratio

Fig. 3.

ROC Curves for Hepatic Fibrosis Staging. Receiver operating characteristic (ROC) curves demonstrating the relationship between LSM and clinically defined fibrosis stages. Each curve represents detection of specific fibrosis stages: S1 (minimal fibrosis, AUC = 0.82), S2 (moderate fibrosis, AUC = 0.85), S3 (severe fibrosis, AUC = 0.88), and S4 (cirrhosis, AUC = 0.90). The diagonal dashed line represents the line of no discrimination (AUC = 0.50). Progressive improvement in association strength is evident with advancing fibrosis stages. Note: These analyses reflect performance against composite clinical staging rather than histological gold standard. Fibrosis staging was performed using a composite clinical algorithm incorporating clinical manifestations of portal hypertension, ultrasonographic features, biochemical indices, and available histopathology (n = 23, 29.1% of AILD cohort). Validation against histopathological staging is required before definitive conclusions regarding diagnostic accuracy can be drawn

For the detection of clinically defined S1, the optimal cutoff of 5.5 kPa yielded an AUC of 0.82 (95% CI 0.74–0.89) with a sensitivity of 76% and specificity of 70%. We emphasize that these metrics reflect performance against composite clinical staging rather than the histological gold standard. Moderate fibrosis (S2) detection showed an AUC of 0.85 (95% CI 0.79–0.91) with an 8.2 kPa cutoff, sensitivity of 83%, and specificity of 78%.

Advanced fibrosis stages showed stronger associations. Severe fibrosis (S3) detection using a 12.0 kPa cutoff achieved an AUC of 0.88 (95% CI 0.82–0.93) with a sensitivity of 87% and specificity of 80%. Cirrhosis (S4) detection with a 15.3 kPa cutoff yielded an AUC of 0.90 (95% CI 0.85–0.95), sensitivity of 92%, and specificity 85%. Figure 4A illustrates the comparative sensitivity and specificity values across all fibrosis stages, demonstrating a progressive improvement in diagnostic metrics with increasing disease severity. Figure 4B shows the systematic increase in AUC values from S1 (0.82) to S4 (0.90), visually confirming the superior performance for advanced disease stages. These cutoff values are consistent with the ranges reported in the meta-analysis by Chen et al. examining transient elastography in AILD populations (F3:9.6–12.5 kPa; F4:14.4–16.9 kPa for primary biliary cholangitis), providing external support for their potential clinical utility [23].

Fig. 4.

Association Metrics by Fibrosis Stage. A Bar chart comparing sensitivity (green) and specificity (purple) for the detection of clinically defined fibrosis stages S1-S4. Both metrics improved progressively with increasing fibrosis severity, reaching optimal values for S4 (cirrhosis) with 92% sensitivity and 85% specificity. Error bars represent the 95% confidence intervals. B Line plot showing the progressive increase in area under the curve (AUC) values from S1 (0.82) to S4 (0.90), demonstrating enhanced association between LSM and clinically defined disease severity for advanced fibrosis stages. The shaded area represents 95% confidence bands. Note: These metrics reflect performance against composite clinical staging rather than histological gold standard. The reported sensitivity and specificity values characterize the relationship between LSM and clinically defined fibrosis stages. Validation against histopathological staging in prospective studies is essential for definitive conclusions regarding diagnostic accuracy and before these thresholds can be recommended for clinical implementation. The cutoff values (S1: 5.5 kPa, S2: 8.2 kPa, S3: 12.0 kPa, S4: 15.3 kPa) demonstrated reasonable concordance with ranges reported in meta-analyses of transient elastography in AILD populations

Correlation analysis and biomarker relationships

Comprehensive correlation analysis revealed strong associations between the LSM and key clinical parameters (Table 3; Fig. 5). The most prominent correlation was observed with albumin (Spearman’s ρ = -0.67, P < 0.001) (Fig. 5A), indicating that progressive liver stiffness closely parallels the decline in hepatic synthetic function. This relationship was consistent across all autoimmune liver disease subtypes, suggesting a fundamental biological connection between fibrosis progression and protein synthesis capacity.

Table 3.

Correlation analysis between LSM and clinical parameters

Parameter	Correlation Coefficient (ρ)	95% CI	P-value	Interpretation
Albumin	-0.67	-0.74 to -0.58	< 0.001	Strong negative
Total bilirubin	0.72	0.64 to 0.79	< 0.001	Strong positive
Direct bilirubin	0.55	0.44 to 0.64	< 0.001	Moderate positive
Prothrombin time	0.43	0.31 to 0.54	< 0.001	Moderate positive
Alkaline phosphatase	0.43	0.31 to 0.54	< 0.001	Moderate positive
AST	0.38	0.26 to 0.49	< 0.001	Moderate positive
ALT	0.35	0.22 to 0.47	< 0.001	Moderate positive
Age	0.24	0.11 to 0.37	0.001	Weak positive

Abbreviations: LSM liver stiffness measurement, CI confidence interval, AST aspartate aminotransferase, ALT alanine aminotransferase

Fig. 5.

Correlation Analysis Between LSM and Clinical Parameters. A Scatter plot showing a strong inverse correlation between albumin levels and LSM (ρ=-0.67, P < 0.001). Patients are distinguished by disease subtype: red circles for autoimmune hepatitis (AIH, n = 18), green triangles for primary biliary cholangitis (PBC, n = 47), and purple squares for overlap syndromes (n = 14). The control participants are shown as blue circles (n = 121). Patients with AILD clustered in the lower-right quadrant with low albumin and high LSM values, whereas the controls showed the opposite pattern. The regression line with a 95% confidence band is shown. B Positive correlation between total bilirubin and LSM (ρ = 0.72, P < 0.001), indicating cholestatic involvement. Point shapes and colors distinguish AILD subtypes and controls, as described in panel A. C Moderate positive correlation between direct bilirubin and LSM (ρ = 0.55, P < 0.001). Point shapes and colors distinguish AILD subtypes and controls, as described in panel A. D Comprehensive correlation matrix heatmap displaying relationships among all measured parameters. Color intensity indicates the correlation strength: red for positive correlations and blue for negative correlations. The correlation coefficients are displayed within each cell. LSM showed the strongest correlations with albumin (negative) and bilirubin (positive) levels

Cholestatic markers demonstrated substantial correlations with liver stiffness, including total bilirubin (ρ = 0.72, P < 0.001) (Fig. 5B) and direct bilirubin (ρ = 0.55, P < 0.001) (Fig. 5C). The stronger correlation with TBIL suggests that overall bilirubin metabolism dysfunction, rather than isolated conjugation defects, reflects disease severity more closely. ALP levels showed a moderate correlation (ρ = 0.43, P < 0.001), indicating that bile duct injury patterns contribute to LSM.

Coagulation parameters revealed significant associations, with prothrombin time demonstrating a correlation coefficient of 0.43 (P < 0.001). The comprehensive correlation matrix (Fig. 5D) displays the relationships among all measured parameters.

Multivariable predictive modeling

Univariable regression analysis identified multiple variables significantly associated with LSM values, including fibrosis stage, inflammatory grade, albumin level, direct bilirubin level, and antinuclear antibody status (all P < 0.20). The univariable analysis revealed that the fibrosis stage explained 52% of the liver stiffness variance (R² = 0.52), while albumin accounted for 44% (R² = 0.44), highlighting these as the most influential single predictors.

Multivariable linear regression modeling using backward stepwise selection retained fibrosis stage and direct bilirubin as three independent predictors in the final model (Table 4; Fig. 6). Fibrosis stage emerged as the strongest predictor (P = 0.003), with progressive increases in liver stiffness across stages: F2 versus F1 (β = 3.2, 95% CI 0.38-6.0), F3 versus F1 (β = 4.1, 95% CI 0.81–7.5), and F4 versus F1 (β = 8.7, 95% CI 4.3–13.0). The large coefficient for S4 indicates that cirrhosis produces a profound increase in liver stiffness beyond that explained by earlier fibrosis stages.

Table 4.

Multivariable linear regression analysis of LSM predictors

Variable	β Coefficient	SE	95% CI	P-value	VIF
Fibrosis Stage (S)				0.003
S1 (reference)	—	—	—	—	—
S2	3.2	1.4	0.38, 6.0	0.024	1.8
S3	4.1	1.7	0.81, 7.5	0.015	2.1
S4	8.7	2.2	4.3, 13.0	< 0.001	2.4
Direct Bilirubin (µmol/L)	0.01	0.007	0.00, 0.03	0.045	1.6
Constant	7.8	1.1	5.6, 10.0	< 0.001	—

Model statistics: R² = 0.73, Adjusted R² = 0.71, F(4,74) = 15.8, P<0.001

Abbreviations: SE standard error, CI confidence interval, VIF variance inflation factor

Fig. 6.

Multivariable Linear Regression Analysis. Forest plot illustrating the independent predictors of LSM values from the multivariable linear regression analysis. The points represent β coefficients, with horizontal lines indicating 95% confidence intervals. Fibrosis stage comparisons (S2 vs. S1, S3 vs. S1, S4 vs. S1) show progressive increases in effect size, with S4 demonstrating the largest coefficient (β = 8.7, 95% CI 4.3–13.0). Direct bilirubin emerged as an independent predictor (β = 0.01, 95% CI 0.00-0.03). The vertical dashed line at zero represents no effect. P-values and model fit statistics (R²=0.73, adjusted R²=0.71, F(4,74) = 15.8, P < 0.001) are indicated. Variance inflation factor (VIF) values ranged from 1.6 to 2.4, indicating acceptable multicollinearity

Direct bilirubin maintained independent significance (β = 0.01, 95% CI 0.00-0.03, P = 0.045), suggesting that cholestatic components contribute to liver stiffness beyond fibrosis-related changes.

The final model demonstrated a good fit (R² = 0.73, adjusted R² = 0.71, F = 15.8, P < 0.001), explaining 73% of the LSM variance. Residual analysis confirmed model assumptions, with normally distributed residuals (Shapiro-Wilk P = 0.31 for log-transformed model) and absence of influential outliers (Cook’s distance < 0.1 for all observations). VIF values ranged from 1.6 to 2.4, indicating acceptable multicollinearity.

Albumin, despite a strong univariable correlation, was not retained in the multivariable model due to collinearity with the fibrosis stage (variance inflation factor 4.2), suggesting that synthetic dysfunction and architectural changes represent overlapping rather than independent contributors to liver stiffness. Sensitivity analysis using ridge regression confirmed this interpretation.

Subgroup analysis by disease type

A detailed subgroup analysis by AILD subtype is presented in Table 5. Comparative analysis across AILD subtypes revealed no significant differences in LSM values after adjusting for the fibrosis stage (F = 1.2, P = 0.36). AIH patients had mean LSM 13.5 ± 2.7 kPa, primary biliary cholangitis 13.2 ± 4.7 kPa, and overlap syndromes 16.0 ± 2.8 kPa. The numerically higher values in overlap syndromes (n = 14) may reflect the combined hepatocellular and cholestatic injury patterns characteristic of these conditions, although this observation requires confirmation in larger cohorts of patients. The inflammatory grade distribution across the AILD cohort was as follows: G1, 15 (19.0%), G2 51 (64.6%); and G3, 13 (16.5%) patients, with no significant differences across disease subtypes (P = 0.17).

Table 5.

Baseline characteristics and LSM values by AILD subtype

Parameter	Total (n = 79)	AIH (n = 18)	PBC (n = 47)	Overlap (n = 14)	P-value
Demographics
Age (years), mean ± SD	53.5 ± 12.0	54.8 ± 12.7	52.8 ± 11.2	54.1 ± 14.6	0.82
Gender, n (%)					0.91
Male	13 (16.5)	2 (11.1)	9 (19.1)	2 (14.3)
Female	66 (83.5)	16 (88.9)	38 (80.9)	12 (85.7)
BMI (kg/m²), mean ± SD	24.8 ± 4.2	25.0 ± 3.4	24.8 ± 4.8	24.7 ± 2.9	0.98
Fibrosis Stage (S), n (%)					0.51
S1	10 (12.7)	2 (11.1)	7 (14.9)	1 (7.1)
S2	43 (54.4)	10 (55.6)	27 (57.4)	6 (42.9)
S3	21 (26.6)	5 (27.8)	9 (19.1)	7 (50.0)
S4	5 (6.3)	1 (5.6)	4 (8.5)	0 (0)
Inflammatory Grade (G), n (%)					0.17
G1	15 (19.0)	2 (11.1)	12 (25.5)	1 (7.1)
G2	51 (64.6)	13 (72.2)	30 (63.8)	8 (57.1)
G3	13 (16.5)	3 (16.7)	5 (10.6)	5 (35.7)
Liver Stiffness
LSM (kPa), mean ± SD	13.8 ± 4.1	13.5 ± 2.7	13.2 ± 4.7	16.0 ± 2.8	0.083
LSM adjusted for Sᵃ	—	13.4 ± 2.6	13.3 ± 4.5	15.8 ± 2.9	0.36
CAP (dB/m), mean ± SD	235.9 ± 34.3	241.4 ± 29.9	237.0 ± 36.9	225.3 ± 30.3	0.40
Biochemical Profile
ALT (U/L), median (IQR)	105 (44, 176)	132 (46, 293)	105 (40, 166)	93 (52, 165)	0.65
AST (U/L), median (IQR)	104 (48, 189)	149 (41, 273)	84 (46, 184)	106 (63, 161)	0.69
ALP (U/L), median (IQR)	174 (122, 340)	127 (93, 182)	174 (131, 359)	273 (219, 563)	0.009
GGT (U/L), median (IQR)	101 (72, 231)	92 (59, 129)	106 (64, 230)	163 (83, 437)	0.42
TBIL (µmol/L), median (IQR)	30.9 (15.7, 67.4)	20.4 (17.0, 42.4)	30.0 (13.2, 61.5)	51.5 (20.6, 184)	0.17
DBIL (µmol/L), median (IQR)	17.4 (5.5, 46.7)	10.3 (7.2, 42.3)	16.4 (4.6, 43.5)	36.7 (12.5, 124)	0.10
Albumin (g/L), mean ± SD	33.3 ± 7.2	34.9 ± 8.0	33.8 ± 6.9	29.8 ± 6.7	0.12
Coagulation
PT (seconds), mean ± SD	15.1 ± 2.5	14.7 ± 2.1	15.2 ± 2.4	15.5 ± 3.1	0.64
APTT (seconds), mean ± SD	39.2 ± 7.5	38.7 ± 7.2	38.6 ± 7.9	41.8 ± 6.5	0.37
INR, median (IQR)	1.07 (1.00, 1.21)	1.07 (1.02, 1.22)	1.09 (0.98, 1.17)	1.03 (1.00, 1.33)	0.71
Immunological
IgG (g/L), median (IQR)	17.3 (14.5, 20.9)	18.8 (16.8, 26.4)	15.8 (13.6, 19.0)	20.3 (18.4, 24.6)	0.004
IgM (g/L), median (IQR)	2.71 (1.36, 3.75)	1.39 (0.94, 2.43)	3.10 (1.53, 4.01)	3.24 (2.25, 3.68)	0.051
IgA (g/L), mean ± SD	4.19 ± 2.03	4.78 ± 2.24	3.56 ± 1.71	5.53 ± 1.99	0.002
Hematological
WBC (×10⁹/L), mean ± SD	5.79 ± 1.97	6.36 ± 1.96	5.50 ± 1.93	6.04 ± 2.08	0.26
Hemoglobin (g/L), mean ± SD	117.6 ± 23.3	128.9 ± 14.1	115.9 ± 25.7	108.7 ± 19.8	0.035
Platelets (×10⁹/L), mean ± SD	207.6 ± 103.6	195.9 ± 85.0	214.7 ± 117.7	198.6 ± 73.5	0.76
Treatment, n (%)
Any treatment	54 (68.4)	14 (77.8)	30 (63.8)	10 (71.4)	0.56
Corticosteroids	43 (54.4)	16 (88.9)	18 (38.3)	9 (64.3)	< 0.001
UDCA	19 (24.1)	2 (11.1)	14 (29.8)	3 (21.4)	0.38

ᵃAdjusted for fibrosis stage using ANCOVA

Abbreviations: AILD autoimmune liver disease, AIH autoimmune hepatitis, PBC primary biliary cholangitis, BMI body mass index, LSM liver stiffness measurement, CAP controlled attenuation parameter, ALT alanine aminotransferase, AST aspartate aminotransferase, ALP alkaline phosphatase, GGT gamma-glutamyl transferase, TBIL total bilirubin, DBIL direct bilirubin, PT prothrombin time, APTT activated partial thromboplastin time, INR international normalized ratio, IgG immunoglobulin G, IgM immunoglobulin M, IgA immunoglobulin A, WBC white blood cell count, UDCA ursodeoxycholic acid, IQR interquartile range, SD standard deviation

Biochemical profiles differed across subtypes in expected patterns: ALP levels were significantly higher in primary biliary cholangitis and overlap syndromes than in AIH (P = 0.009), consistent with the more prominent cholestatic phenotype. IgG levels were highest in AIH (median 18.8 g/L) compared to primary biliary cholangitis (median 15.8 g/L, P = 0.004), reflecting the characteristic hypergammaglobulinemia associated with AIH. IgM levels were higher in primary biliary cholangitis and overlap syndromes (P = 0.051). IgA levels differed significantly across subtypes (P = 0.002), being highest in overlap syndromes (5.53 g/L) and lowest in primary biliary cholangitis (3.56 g/L) patients. Hemoglobin levels were significantly lower in patients with overlap syndromes (108.7 g/L) than in those with AIH (128.9 g/L, P = 0.035).

Treatment status analysis revealed no significant difference in LSM values between treated and untreated patients within the same fibrosis stage (13.6 ± 4.1 versus 14.2 ± 3.9 kPa, P = 0.42). Corticosteroid use was significantly more common in AIH (88.9%) than in primary biliary cholangitis (38.3%) and overlap syndromes (64.3%, P < 0.001), whereas ursodeoxycholic acid use was most common in primary biliary cholangitis (29.8%).

Stratified analysis by inflammatory activity

To address potential confounding by inflammatory activity, we performed stratified analyses by ALT level (dichotomized at 2× the upper limit of normal) (Table 6). Among patients with ALT ≤ 2× upper limit of normal (n = 31), the median LSM was 12.1 kPa (IQR 10.4–14.2), while those with ALT > 2× upper limit of normal (n = 48) had a median LSM of 14.8 kPa (IQR 12.2–17.1), P = 0.02. However, fibrosis stage remained significantly associated with LSM within both strata, and the inclusion of ALT as a covariate in multivariable models did not materially alter the association between fibrosis stage and LSM (β coefficients changed < 10%). These findings suggest that while inflammatory activity may modestly elevate LSM values, the relationship with the clinically defined fibrosis stage remains robust.

Table 6.

LSM values stratified by inflammatory activity (ALT Level)

Parameter	ALT ≤ 2× ULN (n = 31)	ALT > 2× ULN (n = 48)	P-value
Overall LSM (kPa), median (IQR)	12.1 (10.4, 14.2)	14.8 (12.2, 17.1)	0.02
LSM by Fibrosis Stage
S1 (n = 10)	9.8 (8.2, 10.9)	11.4 (9.6, 12.8)	0.18
S2 (n = 43)	11.8 (10.2, 13.4)	13.9 (11.8, 15.2)	0.04
S3 (n = 21)	14.2 (12.4, 15.8)	16.1 (14.2, 18.0)	0.08
S4 (n = 5)	18.2 (16.8, 20.1)	21.4 (18.9, 23.8)	0.12

Fibrosis stage remained significantly associated with LSM within both strata (P<0.01)

Abbreviations: LSM liver stiffness measurement, ALT alanine aminotransferase, ULN upper limit of normal, IQR interquartile range

Discussion

AILD encompasses a complex group of chronic inflammatory conditions, and accurate assessment of the severity of hepatic fibrosis remains fundamental to clinical management and prognostic evaluation [3, 4]. This cross-sectional study provides evidence supporting the clinical utility of LSM as a non-invasive assessment tool for characterizing disease severity in patients with AILD while identifying key clinical factors associated with measurement values. We observed strong correlations between LSM and clinically defined fibrosis stages, with values progressively increasing from minimal fibrosis to cirrhosis [14, 15].

It is essential to contextualize our findings within the limitations of our study. Fibrosis staging in our cohort was performed using a composite clinical algorithm rather than universal histological confirmation, as liver biopsy was not routinely performed in this real-world clinical setting. Although the 23 patients (29.1%) with available histopathology showed excellent concordance between histological and clinical staging (weighted kappa 0.78, Table 7), we could not definitively establish diagnostic accuracy in the absence of systematic biopsy correlation. Therefore, our reported performance metrics should be interpreted as characterizing the relationship between LSM and clinically defined disease severity rather than true diagnostic accuracy against a histological gold standard.

The association between LSM and clinical fibrosis stage was strongest for advanced disease, with AUC values of 0.88 for severe fibrosis and 0.90 for cirrhosis against clinical staging. This pattern reflects the fundamental relationship between liver stiffness and architectural changes, where advanced fibrosis produces more substantial alterations in tissue mechanical properties [11, 13]. If validated against histology, the high sensitivity (92%) for cirrhosis detection could provide clinical confidence in identifying patients requiring intensive monitoring or consideration for liver transplantation evaluation, while the adequate specificity (85%) would minimize false-positive diagnoses [7]. However, prospective validation against histological staging is required before these thresholds can be recommended for clinical use.

Multivariate analysis revealed that fibrosis stage and direct bilirubin were independent predictors of liver stiffness values, providing important insights into the factors influencing measurement interpretation. The strong association with the fibrosis stage validates the fundamental premise underlying LSM, demonstrating that architectural changes remain the primary determinant of tissue mechanical properties [22]. The independent contribution of direct bilirubin suggests that cholestatic components, particularly those prominent in primary biliary cholangitis and primary sclerosing cholangitis, may influence liver stiffness beyond fibrosis-related changes [17]. This finding aligns with previous observations that extrahepatic cholestasis can increase liver stiffness values, irrespective of fibrosis severity, and has practical implications for clinicians interpreting LSM in patients with significant cholestasis, where values might overestimate fibrosis severity due to bile duct inflammation and edema [17].

Our cutoff values (S3:12.0 kPa, S4:15.3 kPa) demonstrate reasonable concordance with those reported in the systematic review by Chen et al. examining transient elastography specifically in AILD populations (F3:9.6–12.5 kPa, F4:14.4–16.9 kPa for primary biliary cholangitis) [23]. This concordance provides external support for the potential clinical utility of the observed thresholds, although disease- and population-specific validation remains essential.

The profound correlation between liver stiffness and albumin levels (r=-0.67) highlights the close relationship between architectural changes and hepatic synthetic function. This association strengthens the evidence supporting LSM as a comprehensive marker of liver disease severity, encompassing both structural and functional aspects of hepatic impairment [6]. The consistency of this relationship across different autoimmune liver disease subtypes suggests universal biological mechanisms linking fibrosis progression with protein synthesis capacity, independent of the specific disease pathophysiology [4, 5].

The demographic characteristics of our autoimmune liver disease cohort, particularly the marked female predominance (83.5%) and older age at presentation, align with the established epidemiological patterns reported globally for AIH (female: male ratio 3.6:1) and primary biliary cholangitis (female: male ratio 9:1) [1, 3]. The absence of significant differences in LSM across disease subtypes after fibrosis adjustment supports the potential universal applicability across the AILD spectrum, although the sample sizes within individual subgroups (AIH, n = 18; PBC, n = 47; overlap syndromes, n = 14) limit confidence in disease-specific conclusions [4].

Several methodological strengths enhance the reliability and clinical relevance of our findings. Rigorous standardization of LSM procedures, including operator training and quality control metrics, ensures measurement reliability and reproducibility consistent with established guidelines [22]. Comprehensive biochemical evaluation provides a detailed characterization of liver dysfunction patterns. The excellent inter-operator reliability (ICC 0.94) and high technical success rate (98.5%) support the measurement validity. The inclusion of healthy controls establishes clear baseline thresholds and validates the ability to discriminate between diseased and normal populations [12].

This study has several limitations. First, the absence of a universal histological confirmation is fundamental. Our composite clinical staging, while reflecting real-world practice, requires validation through histopathological analysis. Second, the cross-sectional design precludes conclusions regarding the utility of LSM in monitoring disease progression or treatment response. Third, inflammatory activity may confound LSM interpretation, potentially causing an overestimation of fibrosis severity in patients with active flares. Fourth, the single-region recruitment from the Kashgar population limits the generalizability of the findings to other ethnic groups. Fifth, modest subgroup sample sizes (AIH, n = 18; overlap syndromes, n = 14) limit disease-specific conclusions. Sixth, the exclusion of decompensated patients and those with severe obesity restricts the applicability to clinically stable populations.

The clinical implementation of LSM values should consider cholestatic markers, particularly direct bilirubin levels. The strong correlation with albumin suggests that LSM provides complementary information to traditional biochemical markers.

Future research should prioritize longitudinal studies assessing LSM changes during treatment response and disease course, providing insights into the utility of serial measurements for monitoring therapeutic efficacy [18]. A systematic correlation with histological staging across diverse AILD populations would establish definitive diagnostic accuracy and optimal cutoff values. Investigation of LSM performance in different ethnic populations would enhance the understanding of population-specific considerations for clinical implementation [19]. The development of AILD-specific diagnostic algorithms incorporating LSM with other non-invasive biomarkers, such as the enhanced liver fibrosis score or FibroTest, could further optimize clinical utility [20, 22].

Conclusions

This cross-sectional study of 79 patients with AILD and 121 controls from the Kashgar region demonstrated that LSM was strongly associated with clinically defined disease severity, with values progressively increasing across fibrosis stages. Fibrosis stage and direct bilirubin levels emerged as independent predictors of LSM, whereas albumin levels showed a strong inverse correlation (ρ=-0.67). LSM performance was consistent across the AILD subtypes after adjustment for the fibrosis stage.

These findings support the use of LSM as a promising non-invasive tool for characterizing the severity of hepatic fibrosis in patients with AILD. However, the lack of universal histological confirmation represents a critical limitation. Before clinical implementation, validation against histopathological staging in multicenter studies across diverse populations is essential. The integration of LSM with clinical and biochemical parameters may optimize the assessment of disease severity in this patient population.

Authors’ contributions

PP conceived and designed the study, performed data collection, conducted the statistical analysis, and drafted the manuscript. MT participated in the study design and data collection. AA contributed to patient recruitment and clinical assessment. MT performed the LSM and quality control. MT contributed to the data interpretation and manuscript revision. AA participated in the study coordination and manuscript review. All the authors have read and approved the final manuscript.

Funding

This research received a grant from the Natural Science Foundation of the Xinjiang Uygur Autonomous Region (Project No.:2023D01F35).

Data availability

The datasets used and analyzed in the current study are available from the corresponding author upon reasonable request, subject to institutional data-sharing policies and participant privacy protection requirements.

Declarations

Ethics approval and consent to participate

This study was approved by the Institutional Ethics Committee of Kashi Prefecture Second People’s Hospital (ethics approval number: KSEY-EC-SOP/15/1.1-AF05) and was conducted in accordance with the Declaration of Helsinki principles. Written informed consent was obtained from all participants before enrollment.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.