Abstract
This study aimed to investigate the association between the atherogenic index of plasma (AIP) and the ultrasound attenuation parameter (UAP) in Chinese adults. A cross-sectional study was conducted among 6141 physical examinees. Demographic, anthropometric, biochemical, and FibroTouch data were collected. AIP was calculated as Log10(triglyceride (TG)/high-density lipoprotein cholesterol (HDL)). Linear and logistic regression, restricted cubic spline (RCS), Receiver operating characteristic (ROC) curves, E-value, IDI, and NRI analyses were performed with multivariable adjustment. Sex-stratified sensitivity analysis was conducted to address gender imbalance. The mean age was 47.9 ± 10.2 years (94.1% male), and NAFLD prevalence (UAP ≥ 244 dB/m) was 71.7%. AIP was linearly and positively associated with UAP (p < 0.001); each 1-unit AIP increase correlated with a 23.03 dB/m UAP elevation (95% CI 18.38–27.68). Elevated AIP was strongly associated with NAFLD (OR = 5.62, 95% CI 3.00–10.52). RCS confirmed linear dose–response relationships (both p for non-linearity > 0.05). The AUC of AIP for NAFLD was 72.22%, superior to TC (70.63%) and HDL (68.45%); IDI/NRI further validated its better predictive capacity. E-value analysis indicated robustness against unmeasured confounding. Among Chinese adults, AIP has a significant linear correlation with NAFLD and UAP. AIP can serve as an independent correlated marker for NAFLD, and its discriminatory efficacy is superior to that of traditional lipid markers.
Keywords: Atherogenic index of plasma (AIP), Nonalcoholic fatty liver disease (NAFLD), Ultrasound attenuation parameter (UAP), Chinese adults
Subject terms: Biomarkers, Diseases, Endocrinology, Gastroenterology, Medical research, Risk factors
Introduction
Nonalcoholic fatty liver disease (NAFLD) has emerged as a global public health burden, being the most prevalent chronic liver disorder worldwide with a steadily increasing prevalence1,2. Recently renamed metabolic-associated fatty liver disease (MAFLD)3, this condition is now explicitly recognized to be etiologically linked to metabolic syndrome, underscoring the critical role of metabolic dysregulation in its pathogenesis. Left undiagnosed and untreated, NAFLD can progress to nonalcoholic steatohepatitis (NASH), liver fibrosis, cirrhosis, and even hepatocellular carcinoma, imposing substantial clinical and economic burdens on healthcare systems2.
Liver biopsy remains the gold standard for the diagnosis and staging of NAFLD due to its ability to provide histopathological details of hepatic steatosis, inflammation, and fibrosis. However, its invasive nature, associated risks (e.g., bleeding, infection), and high cost limit its utility in large-scale population screening and long-term follow-up4. Non-invasive diagnostic tools have therefore gained increasing attention, among these tools, which FibroTouch—a vibration-controlled transient elastography device—has shown promising performance. FibroTouch quantifies liver stiffness measurement (LSM) to assess hepatic fibrosis and the ultrasound attenuation parameter (UAP) to evaluate the degree of hepatic steatosis5,6. Previous studies have demonstrated that FibroTouch exhibits excellent diagnostic accuracy for liver fibrosis and cirrhosis, ranking second only to liver biopsy4, and a UAP threshold of ≥ 244 dB/m has been validated as a reliable indicator for NAFLD diagnosis7. Despite these advantages, FibroTouch remains underutilized in clinical practice for NAFLD diagnosis in China. Meanwhile, conventional lipid markers have limited performance in NAFLD screening, so there is a need to explore novel, convenient lipid-derived indicators.
The atherogenic index of plasma (AIP), proposed by Dobiásová and Frohlich, is a novel lipid-derived marker calculated as the logarithmic transformation of the ratio of triglycerides (TG) to high-density lipoprotein cholesterol (HDL)8. Unlike traditional lipid parameters (e.g., total cholesterol [TC], low-density lipoprotein cholesterol (LDL), AIP integrates the effects of both pro-atherogenic (TG-rich lipoproteins) and anti-atherogenic (HDL) lipids, making it a more comprehensive indicator of lipid metabolism imbalance9. Accumulating evidence has linked AIP to various metabolic disorders, including coronary artery disease, prediabetes, and obesity9,10. A recent study based on the US National Health and Nutrition Examination Survey (NHANES) 2017–2018 data reported a linear dose-response relationship between AIP and MAFLD, with elevated AIP associated with an increased prevalence of MAFLD in US adults11. However, the association between AIP and NAFLD, as well as the correlation between AIP and UAP, remains unclear in the Chinese population.
Given the underutilization of FibroTouch and the lack of data on AIP-NAFLD associations in China, this cross-sectional study aimed to: (1) explore the correlation between AIP and UAP as a quantitative marker of hepatic steatosis; (2) investigate the relationship between AIP and NAFLD (diagnosed by FibroTouch-derived UAP). The findings of this study are expected to provide evidence for identifying a convenient, cost-effective, and non-invasive marker for NAFLD screening in Chinese adults, thereby facilitating early intervention and reducing the burden of NAFLD-related complications.
Methods
Study subjects
This cross-sectional study enrolled participants who underwent routine physical examinations at the Health Management Center of the Affiliated Hospital of North Sichuan Medical College between January 2024 and December 2025. A total of 9067 individuals were initially screened, and 6141 participants were finally included after applying strict inclusion and exclusion criteria. This study follows the STROBE guidelines for cross-sectional studies.
Inclusion criteria: (1) Age ≥ 18 years; (2) Complete clinical, laboratory, and FibroTouch examination data; (3) Voluntary participation and signed informed consent.
Exclusion criteria: (1) History of heavy alcohol consumption (defined as ≥ 30 g/day for men and ≥ 20 g/day for women); (2) Diagnosis of viral hepatitis (hepatitis B or C), autoimmune liver disease, or other chronic liver diseases; (3) Renal dysfunction (estimated glomerular filtration rate < 60 mL/min/1.73 m2); (4) History of malignant tumors; (5) Pregnancy or lactation; (6) Use of lipid-lowering drugs (e.g., statins, fibrates) or medications known to affect liver function within 3 months prior to enrollment.
The sample size was determined by all eligible participants during the study period; no a priori sample size calculation was performed. The study was conducted in accordance with the Declaration of Helsinki and approved by the Medical Ethics Committee of the Affiliated Hospital of North Sichuan Medical College (approval number: 2021ER132-1). All participants provided written informed consent before data collection.
Data collection and measurements
Demographic and anthropometric data
Trained full-time nurses collected demographic information (gender, age) and anthropometric measurements (height, weight, blood pressure). Height and weight were measured with participants wearing light clothing and no shoes, using a calibrated stadiometer and electronic scale, respectively. Body mass index (BMI) was calculated as weight (kg) divided by height squared (m²). Systolic blood pressure (SBP) and diastolic blood pressure (DBP) were measured three times at 5-minute intervals using an Omron upper-arm electronic blood pressure monitor, and the mean of the three readings was recorded.
Clinical biochemical indicators
All participants fasted for 8–10 h overnight, and peripheral venous blood samples were collected the next morning. Serum samples were analyzed in the clinical laboratory of the Affiliated Hospital of North Sichuan Medical College using standard automated biochemical analyzers. The following indicators were measured: TG, TC, HDL, LDL, fasting plasma glucose (FPG), glycosylated hemoglobin (HbA1c), aspartate aminotransferase (AST), alanine aminotransferase (ALT), serum creatinine (Scr), serum urea (Ur), and serum uric acid (UA). AIP was calculated using the formula: AIP = log10(TG / HDL)10.
FibroTouch examination
FibroTouch examinations were performed by qualified sonographers with more than 5 years of experience in liver elastography. Participants were placed in the supine position with the right arm abducted to 90 degrees. The ultrasound probe was placed on the right intercostal space (between the 6th and 8th ribs) to measure LSM and UAP. Each participant underwent 10 valid measurements, and the median value was recorded. NAFLD was diagnosed based on a UAP value of ≥ 244 dB/m7.
Statistical analysis
All statistical analyses were performed using R Statistical Software (Version 4.2.2; The R Foundation for Statistical Computing, Vienna, Austria) and the Free Statistics Analysis Platform (Version 2.2; Beijing, China). A two-tailed P < 0.05 was considered statistically significant.
The normality of continuous variables was verified by the Kolmogorov–Smirnov test. Normally distributed continuous variables were presented as mean ± standard deviation (SD), and non-normally distributed continuous variables as median (interquartile range, IQR, P25–P75). Categorical variables were reported as frequencies (n) and percentages (%). Between‑group comparisons were conducted using one‑way analysis of variance (ANOVA) for normally distributed continuous variables, Kruskal–Wallis test for non‑normally distributed continuous variables, and chi‑square test for categorical variables.
Participants with missing values in any of the demographic, anthropometric, clinical biochemical, or FibroTouch examination indicators were excluded from the final analysis. A complete-case analysis was performed to ensure data completeness, although potential bias cannot be fully excluded, ensuring the integrity and reliability of the analytical dataset.
Candidate covariates were selected based on previously published epidemiological evidence regarding risk factors for NAFLD and metabolic disorders associated with AIP, as well as clinical plausibility. Covariates were not screened based on univariate significance.
Three models were constructed: Model 1: unadjusted; Model 2: adjusted for age and sex; Model 3: adjusted for age, sex, height, weight, BMI, SBP, DBP, HbA1c, FPG, TG, TC, HDL, LDL, Ur, Scr, UA, AST, and ALT. Although AIP is derived from TG and HDL, these variables were retained in the model to account for potential confounding from lipid-related factors.
Linear regression models were established to assess the association between AIP (treated as a continuous variable or tertile categorical variable) and UAP. Restricted cubic spline (RCS) analysis was used to explore the potential non-linear dose-response relationship between AIP and UAP, as well as between AIP and NAFLD prevalence. Univariate and multivariate logistic regression models were applied to investigate the independent correlation between AIP and NAFLD prevalence.
Receiver operating characteristic (ROC) curves were constructed, and the area under the curve (AUC) was calculated to compare the predictive performance of AIP with TC and HDL for NAFLD. To further evaluate the incremental predictive value of AIP beyond traditional risk factors, net reclassification improvement (NRI) and integrated discrimination improvement (IDI) analyses were performed. Categorical NRI was calculated based on predefined clinical risk stratification thresholds to assess reclassification improvement in clinically meaningful risk strata. All NRI and IDI estimates were presented with 95% confidence intervals (CIs).
Sex-stratified subgroup analysis was conducted as a sensitivity analysis to validate the robustness of the core associations. E-value analysis was performed to evaluate the robustness of the AIP–NAFLD association against unmeasured confounding.
Results
Baseline characteristics of included and excluded participants
A total of 9067 participants were initially screened, and 6141 subjects were finally enrolled after exclusion (Fig. 1). Baseline demographic characteristics, anthropometric measurements, clinical biochemical indicators, FibroTouch parameters and NAFLD prevalence were compared between the included and excluded participants (Table 1). Compared with the excluded group, included participants were significantly older and had a higher proportion of males (94.1% vs. 92.5%, p = 0.003). Significant intergroup differences were also detected in height, weight, SBP, UAP, LSM, FPG, AIP, Ur, Scr, UA, AST and ALT (all p < 0.05). No significant differences were observed in BMI, DBP, HbA1c, TG, TC, HDL and LDL between the two groups (all p > 0.05). The prevalence of NAFLD was comparable between included (71.7%) and excluded (72.3%) participants (p = 0.519), suggesting relatively comparable baseline characteristics, although potential selection bias cannot be fully excluded in the study population.
Fig. 1.
Flowchart of participant selection.
Table 1.
Clinical and biochemical characteristics between included and excluded participant.
| Characteristic | Total | Excluded | Included | p |
|---|---|---|---|---|
| n = 9067 | n = 2926 | n = 6141 | ||
| Sex, n (%) | 0.003 | |||
| Female | 580 ( 6.4) | 219 (7.5) | 361 (5.9) | |
| Male | 8487 (93.6) | 2707 (92.5) | 5780 (94.1) | |
| Age (years), Mean ± SD | 47.1 ± 10.8 | 45.3 ± 11.7 | 47.9 ± 10.2 | < 0.001 |
| Height (cm), Mean ± SD | 168.1 ± 7.2 | 168.4 ± 7.7 | 167.9 ± 7.0 | 0.002 |
| Weight (kg), Mean ± SD | 72.5 ± 11.4 | 73.3 ± 12.2 | 72.2 ± 10.9 | < 0.001 |
| BMI (kg/m²), Mean ± SD | 25.6 ± 4.3 | 25.7 ± 3.3 | 25.6 ± 4.6 | 0.220 |
| SBP (mmHg), Mean ± SD | 124.3 ± 16.1 | 125.0 ± 16.2 | 124.1 ± 16.1 | 0.038 |
| DBP (mmHg), Mean ± SD | 80.4 ± 11.6 | 80.4 ± 11.8 | 80.3 ± 11.5 | 0.749 |
| UAP (dB/m), Mean ± SD | 263.1 ± 32.8 | 264.1 ± 33.3 | 262.6 ± 32.5 | 0.044 |
| LSM (kPa), Mean ± SD | 6.0 ± 1.7 | 6.1 ± 1.8 | 6.0 ± 1.7 | 0.044 |
| HbA1C (%), Mean ± SD | 6.1 ± 1.0 | 6.1 ± 0.9 | 6.1 ± 1.0 | 0.180 |
| FPG (mmol/L), Mean ± SD | 5.2 ± 1.6 | 5.2 ± 1.4 | 5.3 ± 1.6 | < 0.001 |
| TG (mmol/L), Median (IQR) | 1.7 (1.2, 2.5) | 1.7 (1.2, 2.5) | 1.7 (1.2, 2.5) | 0.084 |
| TC (mmol/L), Mean ± SD | 5.0 ± 1.0 | 5.0 ± 1.0 | 5.0 ± 1.0 | 0.153 |
| HDL (mmol/L), Mean ± SD | 1.2 ± 0.3 | 1.2 ± 0.3 | 1.2 ± 0.3 | 0.127 |
| LDL (mmol/L), Mean ± SD | 2.9 ± 0.8 | 2.9 ± 0.8 | 2.9 ± 0.8 | 0.929 |
| AIP, Median (IQR) | 0.2 (0.0, 0.4) | 0.2 (0.0, 0.4) | 0.2 (0.0, 0.4) | 0.037 |
| Ur (mmol/L), Mean ± SD | 5.2 ± 1.4 | 5.3 ± 1.6 | 5.2 ± 1.3 | < 0.001 |
| Scr (µmol/L), Mean ± SD | 80.3 ± 16.9 | 86.2 ± 24.7 | 77.6 ± 10.6 | < 0.001 |
| UA (µmol/L), Mean ± SD | 392.6 ± 90.9 | 408.5 ± 95.7 | 385.3 ± 87.7 | < 0.001 |
| AST (U/L), Mean ± SD | 28.3 ± 18.5 | 29.1 ± 28.1 | 28.0 ± 11.5 | 0.008 |
| ALT (U/L), Median (IQR) | 25.0 (18.0, 38.0) | 26.0 (18.0, 40.0) | 25.0 (18.0, 37.0) | 0.001 |
| NAFLD, n (%) | 0.519 | |||
| non-NAFLD | 2550 (28.1) | 810 (27.7) | 1740 (28.3) | |
| NAFLD | 6517 (71.9) | 2116 (72.3) | 4401 (71.7) |
Values are n (%) or mean (standard deviation) or median (P25, P75). BMI, body mass index; SBP, systolic pressure; DBP, diastolic pressure; UAP, ultrasound attenuation parameter; LSM, liver stiffness measurement; HbA1C, glycosylated hemoglobin; FPG, fasting plasma glucose; TG, triglyceride༛TC, total cholesterol༛HDL, high density lipoprotein cholesterol; LDL, low density lipoprotein cholesterol; AIP, atherogenic index of plasma; Ur, urea; Scr, serum creatinine; UA, uric acid; AST, aspartate aminotransferase; ALT, alanine aminotransferase; NAFLD, Nonalcoholic fatty liver disease.
Participant characteristics
A total of 6141 participants were included in the final analysis. The mean age was 47.9 ± 10.2 years, with 94.1% (n = 5780) being male. Participants were divided into three tertiles based on AIP values: Q1 (− 0.95 to 0.04, n = 2047), Q2 (0.04 to 0.31, n = 2046), and Q3 (0.31 to 2.01, n = 2048).
Compared with participants in the lower AIP tertiles, those in higher AIP tertiles were more likely to be male (p < 0.001) and had significantly higher values of weight, BMI, SBP, DBP, UAP, LSM, HbA1c, FPG, TG, TC, UA, AST, and ALT (all p < 0.001). In contrast, HDL levels were significantly lower in higher AIP tertiles (p < 0.001). A total of 4401 participants (71.7%) were diagnosed with NAFLD, and the prevalence of NAFLD increased significantly with increasing AIP tertiles (52.0% in Q1, 75.1% in Q2, and 87.9% in Q3; p < 0.001). Detailed characteristics of the participants are presented in Table 2.
Table 2.
Clinical and biochemical characteristics by AIP tertile grouping.
| Characteristic | AIP group | p | |||
|---|---|---|---|---|---|
| Total (n = 6141) | Q1 (n = 2047) | Q2 (n = 2046) | Q3 (n = 2048) | ||
| Sex, n (%) | < 0.001 | ||||
| Female | 361 ( 5.9) | 231 (11.3) | 89 (4.3) | 41 (2) | |
| Male | 5780 (94.1) | 1816 (88.7) | 1957 (95.7) | 2007 (98) | |
| Age (years), Mean ± SD | 47.9 ± 10.2 | 49.3 ± 10.7 | 48.3 ± 10.3 | 46.2 ± 9.4 | < 0.001 |
| Height (cm), Mean ± SD | 167.9 ± 7.0 | 166.9 ± 7.2 | 168.0 ± 6.6 | 168.8 ± 7.0 | < 0.001 |
| Weight (kg), Mean ± SD | 72.2 ± 10.9 | 67.0 ± 9.9 | 72.9 ± 10.1 | 76.7 ± 10.6 | < 0.001 |
| BMI (kg/m²), Mean ± SD | 25.6 ± 4.3 | 24.0 ± 2.9 | 25.8 ± 2.9 | 27.0 ± 5.8 | < 0.001 |
| SBP (mmHg), Mean ± SD | 124.1 ± 16.1 | 121.6 ± 16.2 | 124.2 ± 16.0 | 126.6 ± 15.7 | < 0.001 |
| DBP (mmHg), Mean ± SD | 80.3 ± 11.5 | 77.7 ± 11.4 | 80.5 ± 11.3 | 82.9 ± 11.2 | < 0.001 |
| UAP (dB/m), Mean ± SD | 262.6 ± 32.5 | 245.4 ± 31.0 | 264.5 ± 30.0 | 277.8 ± 28.1 | < 0.001 |
| LSM (kPa), Mean ± SD | 6.0 ± 1.7 | 5.7 ± 1.5 | 6.0 ± 1.6 | 6.4 ± 1.9 | < 0.001 |
| HbA1C (%), Mean ± SD | 6.1 ± 1.0 | 6.0 ± 0.7 | 6.1 ± 0.9 | 6.4 ± 1.2 | < 0.001 |
| FPG (mmol/L), Mean ± SD | 5.3 ± 1.7 | 5.0 ± 1.2 | 5.2 ± 1.4 | 5.7 ± 2.1 | < 0.001 |
| TG (mmol/L), Median (IQR) | 1.7 (1.2, 2.5) | 1.0 (0.8, 1.2) | 1.7 (1.4, 1.9) | 3.0 (2.4, 4.2) | < 0.001 |
| TC (mmol/L), Mean ± SD | 5.0 ± 1.0 | 4.8 ± 0.9 | 5.0 ± 0.9 | 5.2 ± 1.0 | < 0.001 |
| HDL (mmol/L), Mean ± SD | 1.2 ± 0.3 | 1.4 ± 0.3 | 1.1 ± 0.2 | 0.9 ± 0.2 | < 0.001 |
| LDL (mmol/L), Mean ± SD | 2.9 ± 0.8 | 2.7 ± 0.7 | 3.1 ± 0.7 | 2.9 ± 0.8 | < 0.001 |
| AIP, Median (IQR) | 0.2 (0.0, 0.4) | -0.1 (-0.2, 0.0) | 0.2 (0.1, 0.2) | 0.5 (0.4, 0.7) | < 0.001 |
| Ur (mmol/L), Mean ± SD | 5.2 ± 1.3 | 5.4 ± 1.3 | 5.1 ± 1.2 | 5.1 ± 1.3 | < 0.001 |
| Scr (µmol/L), Mean ± SD | 77.6 ± 10.6 | 76.2 ± 11.0 | 78.4 ± 10.4 | 78.2 ± 10.2 | < 0.001 |
| UA (µmol/L), Mean ± SD | 385.3 ± 87.7 | 349.4 ± 77.4 | 387.0 ± 80.9 | 419.4 ± 89.8 | < 0.001 |
| AST (U/L), Mean ± SD | 28.0 ± 11.5 | 26.5 ± 11.0 | 27.2 ± 10.2 | 30.2 ± 12.9 | < 0.001 |
| ALT (U/L), Median (IQR) | 25.0 (18.0, 37.0) | 21.0 (15.0, 29.0) | 25.0 (18.0, 36.0) | 31.0 (22.0, 46.0) | < 0.001 |
| NAFLD, n (%) | < 0.001 | ||||
| non-NAFLD | 1740 (28.3) | 983 (48) | 510 (24.9) | 247 (12.1) | |
| NAFLD | 4401 (71.7) | 1064 (52) | 1536 (75.1) | 1801 (87.9) |
Values are n (%) or mean (standard deviation) or median (P25, P75). BMI, body mass index; SBP, systolic pressure; DBP, diastolic pressure; UAP, ultrasound attenuation parameter; LSM, liver stiffness measurement; HbA1C, glycosylated hemoglobin; FPG, fasting plasma glucose; TG, triglyceride༛TC, total cholesterol༛HDL, high density lipoprotein cholesterol; LDL, low density lipoprotein cholesterol; AIP, atherogenic index of plasma; Ur, urea; Scr, serum creatinine; UA, uric acid; AST, aspartate aminotransferase; ALT, alanine aminotransferase; NAFLD, Nonalcoholic fatty liver disease.
Association between AIP and UAP
Linear regression analysis revealed a significant positive association between AIP and UAP across all three models (all p < 0.001). In the fully adjusted model (Model 3), each 1-unit increase in AIP was associated with a 23.03 dB/m increase in UAP (95% CI 18.38–27.68).
When AIP was analyzed as tertiles (with Q1 as the reference group), a significant upward trend in UAP was observed with increasing AIP tertiles (p for trend < 0.001). In the fully adjusted model, compared with Q1, Q2 was associated with a 5.84 dB/m increase in UAP (95% CI 4.21–7.47, p < 0.001), and Q3 was associated with an 8.61 dB/m increase (95% CI 6.46–10.76, p < 0.001) (Table 3).
Table 3.
Association between AIP and UAP.
| Characteristic | Model 1 | Model 2 | Model 3 | |||
|---|---|---|---|---|---|---|
| B (95% CI) | p | B (95% CI) | p | B (95% CI) | p | |
| AIP | 39.15 (37 ~ 41.31) | < 0.001 | 38.14 (35.94 ~ 40.35) | < 0.001 | 23.03 (18.38 ~ 27.68) | < 0.001 |
| Q1 | Ref | Ref | Ref | |||
| Q2 | 19.08 (17.26 ~ 20.9) | < 0.001 | 18.6 (16.77 ~ 20.43) | < 0.001 | 5.84 (4.21 ~ 7.47) | < 0.001 |
| Q3 | 32.32 (30.5 ~ 34.14) | < 0.001 | 31.45 (29.59 ~ 33.31) | < 0.001 | 8.61 (6.46 ~ 10.76) | < 0.001 |
| p for trend | < 0.001 | < 0.001 | < 0.001 | |||
Linear regression models: Model 1: no covariates were adjusted. Model 2 was adjusted for age and gender. Model 3 was adjusted for age, gender, Height, Weight, BMI, SBP, DBP, HbA1C, FPG, TG, TC, HDL, LDL, Ur, Scr, UA, AST, ALT.
Sex‑stratified subgroup analysis further confirmed the stability of the AIP‑UAP association (Table 4). In male participants, AIP was significantly positively associated with UAP in all models. In female participants, similar significant positive associations were observed. The P for interaction was 0.014 in Model 3, a trend toward stronger effect magnitude in females was observed, but the small female sample (n = 361, 5.9%) limits the robustness of this finding.
Table 4.
Subgroup association between AIP and UAP.
| Subgroup | model 1 | model 3 | ||||
|---|---|---|---|---|---|---|
| Sex | n | B (95% CI) | p | B (95% CI) | p | p for interaction |
| Male | 5780 | 38.36 (36.13 ~ 40.59) | < 0.001 | 21.92 (17.17 ~ 26.67) | < 0.001 | 0.014 |
| Female | 361 | 42.5 (31.96 ~ 53.05) | < 0.001 | 35.42 (12.58 ~ 58.26) | 0.003 | |
Model 1: no covariates were adjusted. Model 3 was adjusted for age, gender, Height, Weight, BMI, SBP, DBP, HbA1C, FPG, TG, TC, HDL, LDL, Ur, Scr, UA, AST, ALT.
Restricted cubic spline analysis further confirmed a robust linear positive relationship between AIP and UAP (p for overall association < 0.001, p for non-linearity = 0.647), which remained significant after adjusting for multiple confounders (Fig. 2).
Fig. 2.
Association of AIP with UAP. Solid and dashed yellow lines represent the predicted value and 95% CI, respectively. Adjusted for age, gender, Height, Weight, BMI, SBP, DBP, HbA1C, FPG, TG, TC, HDL, LDL, Ur, Scr, UA, AST, ALT.
Association between AIP and the Risk of NAFLD
Multivariate logistic regression analysis showed a significant positive association between AIP and NAFLD prevalence in all three models (all p < 0.001). In the fully adjusted model (Model 3), each 1-unit increase in AIP was associated with a 5.62-fold higher odds of NAFLD (95% CI 3.00–10.52).
When AIP was categorized into tertiles (Q1 as the reference group), a significant upward trend in NAFLD prevalence was observed with increasing AIP tertiles (p for trend < 0.001). In the fully adjusted model, compared with Q1, Q2 was associated with a 1.34-fold higher prevalence of NAFLD (95% CI 1.11 ~ 1.62, p = 0.002), and Q3 was associated with a 1.62-fold higher prevalence (95% CI 1.2 ~ 2.19, p = 0.001) (Table 5).
Table 5.
Logistic regression analysis of the association between AIP and NAFLD.
| Characteristic | Model 1 | Model 2 | Model 3 | |||
|---|---|---|---|---|---|---|
| OR (95% CI) | p | OR (95% CI) | p | OR (95% CI) | p | |
| AIP group | 15.4 (12.41 ~ 19.11) | < 0.001 | 14.9 (11.96 ~ 18.56) | < 0.001 | 5.62 (3 ~ 10.52) | < 0.001 |
| Q1 | 1 (Ref) | 1 (Ref) | 1 (Ref) | |||
| Q2 | 2.78 (2.44 ~ 3.18) | < 0.001 | 2.72 (2.38 ~ 3.11) | < 0.001 | 1.34 (1.11 ~ 1.62) | 0.002 |
| Q3 | 6.74 (5.75 ~ 7.9) | < 0.001 | 6.53 (5.56 ~ 7.67) | < 0.001 | 1.62 (1.2 ~ 2.19) | 0.001 |
| p for trend | < 0.001 | < 0.001 | < 0.001 | |||
Model 1: no covariates were adjusted. Model 2 was adjusted for age and gender. Model 3 was adjusted for age, gender, Height, Weight, BMI, SBP, DBP, HbA1C, FPG, TG, TC, HDL, LDL, Ur, Scr, UA, AST, ALT.
Sex‑stratified subgroup analysis was performed to assess the robustness of the AIP‑NAFLD association (Table 6). In male participants with NAFLD, AIP was strongly associated with NAFLD prevalence in all models. In female participants with NAFLD, significant positive associations were detected in Model 1 while a consistent positive trend was observed in Model 3. The p for interaction was 0.209 in Model 3, indicating no significant effect modification by sex on the AIP‑NAFLD association.
Table 6.
Subgroup logistic regression analysis of the association between AIP and NAFLD.
| Subgroup | Model 1 | Model 3 | ||||
|---|---|---|---|---|---|---|
| Sex | n (%) | OR (95% CI) | p | OR (95% CI) | p | p for interaction |
| Male | 4200 (72.7) | 14.49 (11.58 ~ 18.14) | < 0.001 | 5.01 (2.62 ~ 9.58) | < 0.001 | 0.209 |
| Female | 201 (55.7) | 23.43 (9.19 ~ 59.72) | < 0.001 | 11.75 (0.83 ~ 165.42) | 0.068 | |
Model 1: no covariates were adjusted. Model 3 was adjusted for age, gender, Height, Weight, BMI, SBP, DBP, HbA1C, FPG, TG, TC, HDL, LDL, Ur, Scr, UA, AST, ALT.
Restricted cubic spline analysis confirmed an approximately linear positive dose-response relationship between AIP and NAFLD prevalence (p for overall association < 0.001, p for non-linearity = 0.506). The odds ratio (OR) for NAFLD increased continuously with increasing AIP levels, indicating statistical stability of the association (Fig. 3).
Fig. 3.
Dose-response curve of AIP and NAFLD. Solid and dashed yellow lines represent the predicted value and 95% CI. Adjusted for age, gender, Height, Weight, BMI, SBP, DBP, HbA1C, FPG, TG, TC, HDL, LDL, Ur, Scr, UA, AST, ALT.
ROC curve analysis showed that the AUC for AIP in predicting NAFLD was 72.22% (95% CI: 70.83%–73.61%), which was significantly higher than that of TC (AUC = 70.63%, 95% CI 69.22–72.04%) and HDL (AUC = 68.45%, 95% CI 67.00–69.91%), suggesting that AIP is a better predictor of NAFLD (Fig. 4).
Fig. 4.
ROC curve analysis of AIP and NAFLD. AIP AUC (95% CI): 72.22% (70.83% ~ 73.61%), TG AUC (95% CI): 70.63% (69.22% ~ 72.04%), HDL AUC (95% CI): 68.45% (67.00% ~ 69.91%).
Sensitivity analysis for unmeasured confounding
We also performed an E-value analysis to assess the potential influence of unmeasured confounding. The relatively large E-value suggests that a confounder would need to have a strong association with both AIP and NAFLD to fully explain the observed results. While this does not eliminate the possibility of residual confounding, it provides some support for the robustness of the findings(Fig. 5).
Fig. 5.
E-value plots for the associations of AIP with NAFLD. The x-axis represents the exposure-confounder risk ratio, and the y-axis represents the confounder-outcome risk ratio. The green solid line is the critical threshold, and the red triangle indicates the symmetric critical E-value. The results indicate that extremely strong unmeasured confounding is required to reverse the original associations, suggesting the result is relatively robust to potential unmeasured confounding.
Incremental predictive value of AIP evaluated by NRI and IDI
To further verify the incremental predictive value of AIP for NAFLD on the basis of TG and HDL, NRI and IDI analyses were performed (Table 7). Compared with the base model (TG + HDL), adding AIP to construct the new model (AIP + TG+HDL) resulted in a significant categorical NRI of 0.0341 (95% CI 0.0212–0.0471, P < 0.001), a continuous NRI of 0.256 (95% CI 0.2016–0.3105, P < 0.001), and an IDI of 0.015 (95% CI 0.0121–0.0178, P < 0.001). All three indicators were positive with 95% CIs not crossing 0, indicating that the addition of AIP significantly improved the model’s reclassification ability and overall discrimination performance for NAFLD prevalence.
Table 7.
NRI and IDI for adding AIP to the base model (TG + HDL).
| Indicator | 95% CI | p |
|---|---|---|
| Categorical NRI | 0.0341 (0.0212–0.0471) | < 0.001 |
| Continuous NRI | 0.2560 (0.2016–0.3105) | < 0.001 |
| IDI | 0.0150 (0.0121–0.0178) | < 0.001 |
Base model = TG+HDL; New model = AIP+ TG + HDL.
Discussion
This study examined the relationship between AIP and NAFLD, and further evaluated its association with UAP as a quantitative indicator of hepatic steatosis. We found that AIP was positively associated with both UAP and NAFLD prevalence. Individuals with higher AIP levels tended to have higher UAP values and a greater likelihood of NAFLD. In addition, AIP showed slightly better discrimination than traditional lipid indicators such as TC and HDL. These findings provide novel clinical evidence for the screening and risk assessment of NAFLD in the Chinese adult population. Importantly, this study extends previous findings by using UAP as a quantitative measure of hepatic steatosis, providing more precise evidence for the association between AIP and liver fat accumulation.
NAFLD is one of the most prevalent chronic liver diseases globally, affecting approximately 30 to 40% of adults7. The incidence of this disease is still on the rise. It has been well established that NAFLD is closely linked to metabolic syndrome, and the recent renaming to MAFLD further emphasizes the central role of metabolic disturbances in its pathogenesis3. Although liver biopsy remains the gold standard for NAFLD diagnosis, its invasive nature severely limits its application in large-scale population screening4. FibroTouch, a non-invasive diagnostic tool, quantifies hepatic steatosis and fibrosis by measuring UAP and LSM. Previous studies have validated that a UAP value ≥ 244 dB/m is an effective diagnostic criterion for NAFLD12; meanwhile, studies have shown that UAP increases as the degree of fatty degeneration increases13,14, which supports its use in NAFLD assessment for the diagnosis of NAFLD in this study.
Insulin resistance is widely considered a key factor in the development of NAFLD15,16. Previous studies have shown that AIP is associated with insulin resistance, suggesting that higher AIP levels may reflect an impaired metabolic state16–18. Under insulin resistance, more free fatty acids are released from adipose tissue and transported to the liver, which may contribute to hepatic fat accumulation15. Lipid overload in the liver is often described as lipotoxicity. Studies have reported that excess lipid intermediates, such as diacylglycerols and ceramides, may interfere with insulin signaling pathways and worsen metabolic imbalance19–21. In addition, cholesterol accumulation in hepatocytes has been linked to the progression of steatohepatitis and liver injury20,22. Excess free fatty acids may also increase oxidative stress. Experimental studies have shown that high levels of free fatty acids can impair mitochondrial function and promote the production of reactive oxygen species19,21. Increased oxidative stress has been associated with hepatocellular damage in NAFLD19. Inflammation is another important process in NAFLD. Previous studies have reported that lipid accumulation and oxidative stress may activate inflammatory pathways in the liver, including the activation of Kupffer cells and the release of cytokines such as TNF-α and IL-619,20,22. Overall, AIP may reflect a combination of insulin resistance, lipid imbalance, oxidative stress, and inflammation. These processes are widely reported to be associated with NAFLD, but the causal relationships cannot be confirmed in this cross-sectional study15,19,20.
A prior study conducted in the U.S. population reported a linear dose-response relationship between AIP and MAFLD, showing that increased AIP levels were associated with a higher prevalence of MAFLD11. This study, focusing on Chinese adults, not only confirms this association but also provides additional critical evidence. After adjusting for multiple confounding factors including age, gender, BMI, blood pressure, blood glucose, liver and kidney function, and traditional lipid markers, each 1-unit increase in AIP was associated with a significant 23.03 dB/m elevation in UAP and a 5.62-fold increase in the prevalence of NAFLD. We calculated the E-value to assess the robustness of the observed association against potential unmeasured confounding. Such a high E-value suggests that the findings are reasonably robust to unmeasured confounding. This indicates that an unmeasured confounder would need to be associated with both AIP and NAFLD by an odds ratio of 10.72 above and beyond the adjusted covariates to fully explain away the observed association. Furthermore, the tertile analysis of AIP revealed that as AIP levels increased from Q1 to Q3, the prevalence of NAFLD rose from 52.0% to 87.9%, accompanied by gradual increases in UAP and LSM. These results showed a trend of increasing LSM with elevated AIP, suggesting a possible association with liver fibrosis that requires further validation. Previous studies have indicated this viewpoint21.
Sex‑stratified subgroup analyses further verified the stability of the AIP‑UAP and AIP‑NAFLD associations. For the AIP‑UAP association, a significant positive correlation was present in both males and females, and the effect magnitude of AIP on UAP was significantly stronger in females (p for interaction = 0.014). This result should be interpreted with caution due to the extremely small female sample size (5.9%). For the AIP‑NAFLD association, no significant sex‑dependent effect modification was observed (p for interaction = 0.209), suggesting that the predictive value of AIP for NAFLD prevalence is consistent across genders. This sex‑specific difference in the AIP‑UAP association may be attributed to disparities in lipid metabolism, body fat distribution, and hormonal status between males and females, which warrants further mechanistic exploration.
Restricted cubic spline analysis further confirmed a stable linear positive correlation between AIP and both UAP and NAFLD prevalence (p for non-linearity were 0.647 and 0.506, respectively). This suggests that within the observed AIP range in this study population (− 0.95 to 2.01), a progressive increase in AIP is consistently associated with worsening hepatic steatosis and an elevated prevalence of NAFLD. These findings are consistent with the pathophysiological mechanisms underlying metabolic disorders: elevated AIP reflects an imbalance in lipid metabolism, and the liver, as the central organ regulating lipid homeostasis, is prone to abnormal lipid accumulation—a key pathological process in the development of NAFLD2. Additionally, the predictive value of AIP for NAFLD (AUC = 72.22%) was superior to that of TC (AUC = 70.63%) and HDL (AUC = 68.45%). This suggests that AIP, as a composite indicator integrating TG and HDL, can more comprehensively reflect the state of lipid metabolism disorders and thus has higher clinical utility in NAFLD screening, consistent with previous studies22.
Notably, NRI and IDI analyses further confirmed the incremental predictive value of AIP beyond TG and HDL. The significant positive categorical NRI, continuous NRI, and IDI all demonstrated that adding AIP to the basic model composed of TG and HDL significantly improved the model’s ability to reclassify NAFLD risk and discriminate between patients with and without NAFLD. Specifically, the continuous NRI (0.256) was substantially higher than the categorical NRI (0.0341), indicating that the improvement in predictive performance was not limited to predefined risk stratification thresholds but was evident across the entire risk spectrum, further validating the robustness of AIP’s predictive value. These findings provide supporting evidence supporting the clinical utility of AIP as a valuable supplementary biomarker for NAFLD risk assessment.
Our study also had a few limitations. First, as a cross-sectional design, this study can only identify associations rather than establish causal relationships. Future prospective cohort studies or Mendelian randomization analyses are warranted to verify the causal direction between AIP and NAFLD. Secondly, the study population was mainly from the northern Sichuan region, with a high proportion of males (94.1%), which may introduce regional and gender biases. The generalizability of the results needs to be validated in more diverse regions and populations with a balanced gender distribution. Thirdly, detailed information on lifestyle (diet, exercise, smoking, drinking) and family history was not collected. These factors may indirectly affect the association between AIP and NAFLD by influencing lipid metabolism, and future studies should incorporate these variables for more in-depth subgroup analyses. Forth, given the highly imbalanced sex distribution, propensity score-based methods such as IPTW could further reduce selection bias. However, due to the extremely small proportion of female participants, such methods may not achieve stable matching. Therefore, sex-stratified analysis was performed as a sensitivity approach.
In conclusion, AIP was positively associated with both NAFLD and UAP in this study population. Higher AIP levels were consistently linked to greater hepatic steatosis. Although the findings are limited by the cross-sectional design and sample structure, they suggest that AIP may serve as a simple and accessible marker in clinical practice. Further studies, especially prospective designs, are needed to clarify its role in disease progression.
Author contributions
All author contributions as follows: H.Z. (First Author): Conceptualization, data curation, formal analysis, funding acquisition, investigation, methodology, writing-original draft, writing-review and editing; D.Y. (First Author): Methodology, writing-review and editing, data curation, supervision; Y.L. (Corresponding Author): Supervision, resources; R.L.: Supervision, validation; X.Z.: Data curation; Y.J.: Supervision;
Funding
Research Project of North Sichuan Medical College (CBY23-QNB11). The Primary Health Development Research Center of Sichuan Province Program (SWFZ25-Q-96).
Data availability
The datasets generated and analysed during the current study are not publicly available due to the need for further research but are available from the corresponding author on reasonable request.
Declarations
Ethics approval and consent to participate
This study was reviewed and approved by the Medical Ethics Committee of Affiliated Hospital of North Sichuan Medical College (2021ER132-1), and all subjects gave informed consent.
Competing interests
The authors declare no competing interests.
Footnotes
Publisher’s note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
HongYuan Zhao and Dan Yang contributed equally to this work.
References
- 1.Nassir, F. NAFLD: Mechanisms, Treatments, and Biomarkers. Biomolecules12 (6), 824. 10.3390/biom12060824 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Younossi, Z. M. et al. The economic and clinical burden of nonalcoholic fatty liver disease in the United States and Europe. Hepatology64 (6), 1577–1586. 10.1002/hep.28641 (2016). [DOI] [PubMed] [Google Scholar]
- 3.Rinella, M. E. et al. A multisociety Delphi consensus statement on new fatty liver disease nomenclature. Hepatology78 (6), 1966–1986. 10.1097/HEP.0000000000000520 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Peng, X. et al. Diagnostic value of FibroTouch and non-invasive fibrosis indexes in hepatic fibrosis with different aetiologies. Dig. Dis. Sci.67 (6), 2627–2636. 10.1007/s10620-021-07049-4 (2022). [DOI] [PubMed] [Google Scholar]
- 5.Serra, J. T. et al. Prospective comparison of transient elastography using two different devices: Performance of FibroScan and FibroTouch. Hepatol. Med. Policy. 12, 41–48. 10.1159/000509585 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Singh, S. et al. American Gastroenterological Association Institute technical review on the role of elastography in chronic liver diseases. Gastroenterology152 (7), 1544–1577. 10.1053/j.gastro.2017.02.027 (2017). [DOI] [PubMed] [Google Scholar]
- 7.TILG, H. et al. Metabolic dysfunction-associated steatotic liver disease in adults: A review. JAMA335 (2), 163–174. 10.1001/jama.2025.19615 (2026). [DOI] [PubMed] [Google Scholar]
- 8.Dobiásová, M. & Frohlich, J. The plasma parameter log (TG/HDL-C) as an atherogenic index: correlation with lipoprotein particle size and esterification rate in apoB-lipoprotein-depleted plasma (FER(HDL)). Clin. Biochem.34 (7), 583–588. 10.1016/S0009-9120(01)00263-6 (2001). [DOI] [PubMed] [Google Scholar]
- 9.Shin, H. R. et al. Atherogenic index of plasma and its association with risk factors of coronary artery disease and nutrient intake in Korean adult men: the 2013–2014 KNHANES. Nutrients14 (5), 1071. 10.3390/nu14051071 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Heng, X. et al. Nonlinear relationship between atherogenic index of plasma and the risk of prediabetes: a retrospective study based on Chinese adults. Cardiovasc. Diabetol.22 (1), 205. 10.1186/s12933-023-01773-8 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Chen, Y. et al. Elevated AIP is associated with the prevalence of MAFLD in the US adults: evidence from NHANES 2017–2018. Front. Endocrinol.15, 1405828. 10.3389/fendo.2024.1405828 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Qu, Y. et al. Diagnostic performance of FibroTouch ultrasound attenuation parameter and liver stiffness measurement in assessing hepatic steatosis and fibrosis in patients with nonalcoholic fatty liver disease. Clin. Transl. Gastroenterol.12 (4), e00323. 10.14309/ctg.0000000000000323 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Yu, H. et al. Accuracy of FibroTouch in assessing liver steatosis and fibrosis in patients with metabolic-associated fatty liver disease combined with type 2 diabetes mellitus. Ann. Palliat. Med.10 (9), 9702–9714. 10.21037/apm-21-2339 (2021). [DOI] [PubMed] [Google Scholar]
- 14.Zhu, S. H. et al. Optimal thresholds for ultrasound attenuation parameter in the evaluation of hepatic steatosis severity: evidence from a cohort of patients with biopsy-proven fatty liver disease. Eur. J. Gastroenterol. Hepatol.33 (3), 430–435. 10.1097/MEG.0000000000001746 (2021). [DOI] [PubMed] [Google Scholar]
- 15.Loomba, R., Friedman, S. L. & Shulman, G. I. Mechanisms and disease consequences of nonalcoholic fatty liver disease . Cell184 (10), 2537–2564. 10.1016/j.cell.2021.04.015 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Yin, B. et al. Non-linear association of atherogenic index of plasma with insulin resistance and type 2 diabetes: A cross-sectional study. Cardiovasc. Diabetol.22 (1), 157. 10.1186/s12933-023-01886-5 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Liu, J. et al. Fractional esterification rate of cholesterol in high-density lipoprotein associates with risk of coronary heart disease. Lipids Health Dis.16 (1), 162. 10.1186/s12944-017-0545-z (2017). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Dong, J. et al. A simple and precise method for direct measurement of fractional esterification rate of high density lipoprotein cholesterol by high performance liquid chromatography. Clin. Chem. Lab. Med.52 (4), 557–564. 10.1515/cclm-2013-0525 (2014). [DOI] [PubMed] [Google Scholar]
- 19.Svegliati-Baroni, G. et al. Lipidomic biomarkers and mechanisms of lipotoxicity in non-alcoholic fatty liver disease. Free Radic. Biol. Med.144, 293–309. 10.1016/j.freeradbiomed.2019.05.029 (2019). [DOI] [PubMed] [Google Scholar]
- 20.Horn, C. L. et al. Role of cholesterol-associated steatohepatitis in the development of NASH. Hepatol. Commun.6 (1), 12–35. 10.1002/hep4.1801 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Neuschwander-Tetri, B. A. Non-alcoholic fatty liver disease. BMC Med.15 (1), 45. 10.1186/s12916-017-0806-8 (2017). Published 2017 Feb 28. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Buzzetti, E., Pinzani, M. & Tsochatzis, E. A. The multiple-hit pathogenesis of non-alcoholic fatty liver disease (NAFLD). Metabolism65 (8), 1038–1048. 10.1016/j.metabol.2015.12.012 (2016). [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Data Availability Statement
The datasets generated and analysed during the current study are not publicly available due to the need for further research but are available from the corresponding author on reasonable request.