Associations of TyG-related indices with NAFLD-related outcomes: A systematic review and meta-analysis Authors: Jinyang Li, Wantao Liu Published: 2026 ABSTRACT Background:The triglyceride-glucose (TyG) index and related composite indices have been increasingly studied as accessible metabolic markers for fatty liver disease. However, evidence remains difficult to interpret because different TyG-related indices have been evaluated across heterogeneous populations and nonalcoholic fatty liver disease (NAFLD)-related outcomes. We conducted a systematic review and meta-analysis to clarify the pooled associations that could be formally synthesized and to summarize the broader descriptive evidence.Methods:We systematically searched PubMed, Embase, the Cochrane Library, China National Knowledge Infrastructure, Wanfang Data, VIP Database, and China Biology Medicine Database from inception to December 2025. Observational studies evaluating the TyG index or related composite indices, including TyG-body mass index, TyG-waist circumference, and TyG-waist-to-height ratio, in relation to NAFLD-related outcomes were eligible. Random-effects meta-analyses were performed in R. Only studies with sufficient comparability in exposure operationalization, outcome definition, and effect-size reporting were quantitatively pooled; the remaining studies were retained in the descriptive synthesis.Results:Nineteen studies involving 212,618 participants were included in the systematic review, including 16 cross-sectional studies and 3 cohort studies. Of these, 8 studies contributed to at least 1 quantitative synthesis. In continuous-scale analyses, higher TyG levels were associated with increased odds of NAFLD/steatosis (5 studies; pooled odds ratio [OR], 2.72; 95% confidence interval [CI], 2.08–3.55; I2 = 85.7%). In highest-versus-lowest analyses, the association remained significant (4 studies; pooled OR, 4.36; 95% CI, 3.39–5.61; I2 = 92.8%). TyG-body mass index (2 studies; pooled OR, 1.04; 95% CI, 1.03–1.05; I2 = 0.0%) and TyG-waist circumference (2 studies; pooled OR, 1.01; 95% CI, 1.01–1.01; I2 = 15.1%) were also positively associated with NAFLD/steatosis. TyG-waist-to-height ratio showed a positive association in supportive synthesis, but evidence for liver fibrosis and nonalcoholic steatohepatitis remained limited and was mainly descriptive.Conclusion:TyG-related indices were consistently associated with NAFLD/steatosis across the available literature. The basic TyG index had the broadest evidence base, whereas composite indices suggested possible additional value in selected settings. These markers may be useful for metabolic fatty liver risk stratification, but further studies with standardized outcome definitions and clearer exposure parameterization are needed. 1. Introduction Nonalcoholic fatty liver disease (NAFLD) has become one of the most common chronic liver diseases worldwide and is closely linked to obesity, insulin resistance, type 2 diabetes, and other manifestations of metabolic dysfunction.[1] Its clinical importance extends beyond simple steatosis because a subset of patients may progress to steatohepatitis, fibrosis, cirrhosis, and liver-related complications.[1] Early identification of individuals at increased risk is therefore of practical value, particularly in settings where imaging- or biopsy-based assessment is not readily available. Because insulin resistance plays a central role in the development of NAFLD, simple metabolic surrogate markers have attracted increasing interest. The triglyceride-glucose (TyG) index has been widely explored as an accessible marker of insulin resistance-related metabolic risk, and several recent studies have suggested that TyG-related composite indices may further improve the identification of fatty liver disease by incorporating anthropometric information.[2–4] In population-based or screening-oriented settings, composite indices such as TyG-body mass index (BMI), TyG-waist circumference index (TyG-WC), and TyG-waist-to-height ratio (WHtR) have shown stronger discriminatory performance than the basic TyG index in some cohorts.[2–4] However, the current evidence remains difficult to interpret as a whole. Existing studies have evaluated different TyG-related indices across heterogeneous populations, including general adults, nonobese individuals, adolescents, chronic kidney disease populations, and biopsy-characterized clinical cohorts.[5–7] The reported outcomes have also varied considerably, ranging from NAFLD/steatosis to liver fibrosis, nonalcoholic steatohepatitis (NASH), and at-risk NASH.[5,8–10] In addition, some studies have suggested that the performance of TyG-related indices may differ according to disease stage or clinical context, particularly when moving from simple steatosis toward fibrosis- or inflammation-related phenotypes.[8,11,12] These differences make it difficult to judge the relative evidential strength of the basic TyG index and its composite derivatives from individual studies alone. Therefore, we conducted a systematic review and meta-analysis of observational studies evaluating TyG-related indices in relation to NAFLD-related outcomes. Our aims were to clarify the pooled associations that could be formally synthesized, to summarize the broader descriptive evidence that could not be quantitatively pooled, and to compare the overall evidence profiles of TyG and its related composite indices across NAFLD/steatosis, liver fibrosis, and NASH-related domains. 2. Methods 2.1. Protocol and reporting standard This systematic review and meta-analysis was conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses 2020 statement. The review protocol was prospectively registered in the International Prospective Register of Systematic Reviews (CRD420251153116). Because this study was a systematic review and meta-analysis based exclusively on previously published aggregate data, it did not involve any new recruitment of human participants, collection of individual-level patient data, or animal experiments. Therefore, approval by an ethics committee or institutional review board was not required, and informed consent was not applicable. 2.2. Literature search strategy A systematic literature search was conducted in 3 English-language databases (PubMed, Embase, and the Cochrane Library) and 4 Chinese-language databases (China National Knowledge Infrastructure, Wanfang Data, VIP Database, and China Biology Medicine Database). The search covered the period from database inception to December 31, 2025. The search strategy combined controlled vocabulary terms and free-text keywords related to the TyG index and its derivative indices, including TyG, TyG-BMI, TyG-WC, and TyG-WHtR, together with terms related to NAFLD, metabolic dysfunction-associated fatty liver disease (MAFLD), steatosis, liver fibrosis, and NASH. Reference lists of relevant articles were also manually screened to identify additional eligible studies. The full database-specific search strategies are provided in the Supplementary Search Appendix. 2.3. Eligibility criteria Studies were considered eligible if they met the following criteria: observational study design, including cross-sectional or cohort studies; evaluation of the TyG index or related composite indices, including TyG-BMI, TyG-WC, or TyG-WHtR, as the exposure of interest; reporting of NAFLD-related outcomes, including NAFLD/steatosis, liver fibrosis, or NASH; and provision of sufficient effect-size data for qualitative synthesis and, where applicable, quantitative pooling. Studies were excluded if they were reviews, meta-analyses, case reports, conference abstracts without extractable data, animal studies, or studies conducted in populations not relevant to the review question. Studies were also excluded if the exposure of interest was not a TyG-related index, if the outcome definition was incompatible with the review scope, or if no usable association estimates or sufficient raw data were available. 2.4. Study selection After removal of duplicates, all records were screened independently by 2 reviewers in 2 stages: title/abstract screening followed by full-text assessment. Discrepancies were resolved through discussion until consensus was reached. Studies meeting the predefined eligibility criteria were included in the systematic review. The study selection process is summarized in Figure 1. PRISMA 2020 flow diagram of study selection. Flow diagram of study identification, screening, eligibility assessment, and final inclusion. n = number of studies, PRISMA = Preferred Reporting Items for Systematic Reviews and Meta-Analyses. 2.5. Data extraction and evidence structuring Data were extracted independently by 2 reviewers using a standardized form. Extracted items included study characteristics, country, study design, population characteristics, sample size, exposure definition, outcome definition, diagnostic method, covariate adjustment, and effect-size estimates with corresponding 95% confidence intervals (CIs). Because the included literature was methodologically heterogeneous, evidence synthesis was structured at 2 levels. First, all eligible studies were retained in the systematic review and contributed to the descriptive synthesis, evidence map, and synthesis framework. Second, only studies with sufficient comparability in exposure operationalization, outcome definition, and effect-size reporting were entered into quantitative pooling. For synthesis purposes, reported outcomes were grouped into 3 prespecified domains: NAFLD/steatosis, liver fibrosis, and NASH. Original outcome definitions as reported in the source articles were preserved during data extraction and were used to determine whether a study was sufficiently comparable for formal meta-analysis. When a study reported multiple estimates for the same exposure-outcome combination, only one estimate was selected for the primary pooled analysis within each analysis stratum. Priority was given to the most fully adjusted model that was methodologically appropriate and did not introduce obvious overadjustment related to intrinsic components of the composite index. Studies not meeting these criteria remained in the descriptive synthesis. To further assess robustness, we applied 2 pooling rules. A lenient eligibility rule was used for the primary quantitative synthesis to retain clinically interpretable and methodologically acceptable studies. A stricter eligibility rule was then applied as a supplementary robustness analysis. Because most exposure-outcome pathways became too sparse under the strict rule, only the continuous-scale TyG-NAFLD/steatosis analysis remained eligible for additional strict-rule pooling. 2.6. Quality assessment Methodological quality was evaluated using the Newcastle–Ottawa Scale (NOS) for cohort studies and the Agency for Healthcare Research and Quality (AHRQ) checklist for cross-sectional studies. Quality assessments were performed independently by 2 reviewers, with disagreements resolved by discussion. Study-level quality profiles are presented in the Supplementary Figures. 2.7. Statistical analysis All statistical analyses were performed in R (version 4.5.2; R Foundation for Statistical Computing), primarily using the metafor, readxl, and dplyr packages. Random-effects models were used for all quantitative syntheses because clinical and methodological heterogeneity was anticipated across studies. Odds ratios (ORs) and corresponding 95% CIs were used as the common effect measure for pooled analyses. Continuous-scale and highest-versus-lowest analyses were synthesized separately and were not combined within the same meta-analysis. Between-study heterogeneity was quantified using the I2 statistic. Leave-one-out sensitivity analyses were performed for pooled outcomes to evaluate the influence of individual studies on the overall estimates. Formal assessment of publication bias was not performed because the number of studies in each pooled analysis was too small to support meaningful funnel plot interpretation or statistical testing. Studies that could not be pooled quantitatively were retained in the descriptive synthesis and incorporated into the evidence map and synthesis framework. Additional methodological details and supplementary results are provided in the Supplementary Methods and results file. 3. Results 3.1. Study selection and study characteristics A total of 19 studies involving 212,618 participants were included in the systematic review (Fig. 1). Of these, 16 were cross-sectional studies and 3 were cohort studies. The included studies were conducted across Asia, North America, Europe, and South America, with China and the United States contributing the largest number of studies. The study populations were heterogeneous and included both general-population and clinically enriched cohorts, such as nonobese adults, older adults, individuals with metabolic syndrome or obesity, adolescents, and biopsy-based NAFLD populations. The evaluated exposures comprised the TyG index and related composite indices, including the TyG-BMI, TyG-WC, and TyG-WHtR. NAFLD/steatosis was the most frequently reported outcome, whereas liver fibrosis and NASH were reported less often and were mainly synthesized descriptively. Outcome ascertainment varied across studies and included ultrasonography, vibration-controlled transient elastography/FibroScan (Echosens)-based measures, and liver biopsy, with a few studies using surrogate scores for steatosis- or fibrosis-related phenotypes. Detailed study characteristics are presented in Table 1. Table 1Baseline characteristics table for TyG-related indices and NAFLD-related outcomes.StudyFirst authorYearCountry/ regionStudy designStudy periodPopulation/ settingSample sizeAgeSexBMITyG-related index/ exposureExposure grouping/ comparisonOutcome definitionDiagnostic methodFollow-upFully adjusted covariatesNotesZhang et al, 2024Fan Zhang2024China (Changzhou; eastern China)Single-center cross-sectional studyApr 2020-May 2024Adults with biopsy-proven NAFLD at Changzhou Third People’s Hospital393 total (training 320; validation 73)Training cohort median 50 (38–57) yearsTraining cohort male 129/320 (40.31%)Training cohort median 25.12 (23.44–27.45) kg/m2TyG-BMIQuartiles in training cohort; also analyzed as continuous variableNASH; NAS ≥ 4; at-risk NASH; significant fibrosis; advanced fibrosis; cirrhosisLiver biopsy (gold standard); CAP and LSM also measured for combined modelsNo follow-upAge, gender, CRP, PLT, TBIL, TP, TC, eGFR, AST, hypertensionHospital-based biopsy cohort; disease severity/staging study among existing NAFLD patientsShi et al, 2024Jiejun Shi2024United States (NHANES 2017–2018)Retrospective population-based cross-sectional studyNHANES 2017–2018General adult population3095 total; 1368 with NAFLDNon-NAFLD: 47.3 (45.5–49.0); NAFLD: 53.4 (51.9–54.9) yearsMale: 51.5% in non-NAFLD; 58.1% in NAFLDNon-NAFLD: 26.4 (25.6–27.1); NAFLD: 33.0 (32.3–33.8) kg/m2TyG included among 6 indices (TyG, HOMA-IR, METS-IR, LAP, VAI, WTI)Continuous and quartile analyses of 6 metabolic indicesNAFLD prevalence; steatosis severity; fibrosis severityVCTE/ FibroScan; NAFLD defined as CAP ≥ 274 dB/m after excluding secondary causes; fibrosis staged by LSMNo follow-upAge, gender, race (model 2)Multi-index comparison study rather than TyG-only analysisWu et al, 2024Zhouxiao Wu2024United States/ American adults (NHANES 2017–2018)Population-based cross-sectional studyNHANES 2017–2018General adult population4634 adultsBy TyG-WC quintiles: Q1 37.8 ± 16.6 to Q5 50.8 ± 15.8 yearsBy TyG-WC quintiles: male 35.3% (Q1) to 57.9% (Q5)By TyG-WC quintiles: 22.1 ± 2.8 (Q1) to 38.4 ± 6.7 (Q5) kg/m2TyG-WCQuintiles plus continuous-variable analysesMAFLD prevalence; steatosis severity (S1/S2/S3); significant fibrosis; advanced fibrosis; cirrhosisTransient elastography; CAP for steatosis, LSM for fibrosis; MAFLD defined by steatosis plus metabolic criteriaNo follow-upGender, age, race, hypertension, diabetes, smoking, drinking, activity level, lipid-lowering medication, antidiabetic medication, hyperlipidemia, HbA1cMAFLD terminology study; stronger association reported in nonobese and Non-Hispanic White groupsKuo et al, 2024Tzu-chia Kuo2024United States (NHANES 2017–2018)Cross-sectional studyNHANES 2017–2018Adults with metabolic syndrome (MetS)876 total; 587 with hepatic steatosis; 151 with hepatic fibrosis52.37 (SE 1.18) yearsMale 407/876 (50.07%); female 469/876 (49.93%)NR in currently extracted total baseline rowTyG, TyG-WHtR, HOMA-IR, METS-IR (TyG-WHtR as key TyG-related index)Indicators categorized into tertilesHepatic steatosis; hepatic fibrosis; severe steatosis subgroupVCTE/ FibroScan 502 V2 Touch; CAP ≥ 274 dB/m for steatosis; LSM ≥ 8.2 kPa for fibrosisNo follow-upAdjusted for extensive demographic, lifestyle, clinical, medication, and nutritional covariatesMetS-specific study; TyG linked mainly with steatosis, while TyG-WHtR showed stronger performance for steatosis/fibrosisLi et al, 2023Shuying Li2023United States (NHANES 2017–2018; nonobese adults)Population-based cross-sectional studyNHANES 2017–2018nonobese adults aged ≥20 years (BMI < 30 kg/m2)1776 adultsTotal: 48.6 (46.9–50.3) years; non-NAFLD 44.1 (42.3–46.0); NAFLD 55.8 (53.9–57.7)Male 48.8% overall; 42.8% in non-NAFLD; 58.3% in NAFLD25.0 (24.7–25.2) kg/m2 overall; 24.0 (23.7–24.3) non-NAFLD; 26.5 (26.1–26.9) NAFLDTyG, TyG-BMI, TyG-WC, TG/HDL-cContinuous-variable analyses; TyG-WC also examined across 5 quantile groupsNAFLD; liver stiffnessVCTE/ FibroScan; NAFLD defined as CAP ≥248 dB/m; liver stiffness defined as LSM ≥7.0 kPaNo follow-upAge, BMI, gender, ethnicity, education, alcohol use, diabetes, hypertension, physical activity, TC (BMI not adjusted in TyG-BMI model)Focused specifically on nonobese NAFLD screening; TyG-WC showed the best AUC for NAFLDXue et al, 2022Yan Xue2022United States (NHANES 2017–2018)Population-based cross-sectional analysisNHANES 2017–2018General adult population with VCTE data1727 participants; 737 with NAFLD; 718 with MAFLD53.00 (36.00–65.00) years overall; 47.50 (31.75–63.00) non-NAFLD; 57.00 (43.50–67.00) NAFLDMale 47.8% overall; 44.6% in non-NAFLD; 52.0% in NAFLD28.00 (24.40–32.80) kg/m2 overall; 25.80 (23.00–29.40) non-NAFLD; 31.30 (27.80–35.90) NAFLDTyG, TyG-WC, TyG-WHtR, TyG-BMI (with HOMA-IR comparator)Quartile-stratified and continuous ROC analyses across total population and subgroupsNAFLD; MAFLD; liver fibrosis; moderate-to-advanced fibrosisVCTE/ FibroScan; NAFLD defined as CAP ≥274 dB/m; liver fibrosis LSM ≥7.0 kPa; moderate-to-advanced fibrosis LSM ≥8.2 kPaNo follow-upAge, gender, PIR, education, smoking, diabetes, hypertension, and overweight/obesity status (multivariable models)Comprehensive multiparameter comparison study; TyG-WC, TyG-WHtR, and TyG-BMI ranked highest for NAFLD/MAFLD screeningLee et al, 2022Jun-Hyuk Lee2022South Korea (KoGES-Ansan and Ansung cohort)Longitudinal prospective cohort studyBaseline 2001–2002; follow-up through 2017–2018Adults aged 40–69 years with NAFLD at baseline from a community-based Korean cohort2218 with NAFLD at baseline; 1368 without ALF at baseline and followed longitudinallyNAFLD baseline cohort mean 54.1 years; analytic NAFLD without ALF cohort mean 53.6 ± 8.4 years42.4% male in NAFLD baseline cohort; 42.8% male in analytic cohortNAFLD baseline cohort mean BMI 27.0 ± 2.9 kg/m2; analytic cohort mean 26.7 ± 2.9 kg/m2TyG included with METS-IR and HOMA-IR comparatorsTertiles plus continuous ROC/Cox analysesIncident advanced liver fibrosis (ALF) among patients with baseline NAFLDNAFLD-liver fat score > -0.640 for NAFLD; FIB-4 ≥2.67 for ALFMean follow-up 15.8 yearsSex, age, BMI, physical activity, smoking, drinking, total energy intake, vitamin E, MBP, TC, CRP, ALTCommunity-based longitudinal NAFLD cohort; evaluates progression to ALF rather than cross-sectional liver fat detectionAmzolini et al, 2022Anca Maria Amzolini2022Romania (Craiova)Cross-sectional non-interventional study2017–2019Adults with metabolic syndrome who underwent liver biopsy113 patientsMedian 53 years [IQR 20]41 males, 72 femalesMedian BMI 29 kg/m2 [IQR 6]TyGCompared across BMI categories; continuous analysis versus histopathologic lesionsSteatosis, inflammation, hepatocyte ballooning, NASH category, fibrosis, significant fibrosisLiver biopsy; Kleiner and Brunt classification; NAS used for NASH categorizationNo follow-upMultiple logistic regression adjusting for covariates (not fully detailed in abstract/snippet)Metabolic syndrome biopsy cohort; TyG associated more strongly with definite NASH and significant fibrosis than simple steatosisSong et al, 2022Kyungchul Song2022Korea (Seoul; Gangnam Severance Hospital Health Promotion Center)Retrospective cross-sectional studyJan 2007-Apr 2021Children and adolescents aged 10–19 years undergoing health checkup program; none had diabetes or alcohol use258 total (136 males; 122 females)18.02 ± 1.41 yearsMale 136/258 (52.71%)22.03 ± 4.12 kg/m2TyG, TyG-BMI, TyG-BMI SDS, TyG-WC, TyG-WHR, TyG-WHtRTertiles for each marker; comparison by NAFLD grade (0–3)NAFLD presence and NAFLD grade in youthAbdominal ultrasonography; grades 1–3 considered NAFLD, grade 0 normalNo follow-upNo multivariable adjustment reported; primarily group comparison, tertile-based ORs, and ROC analysesYouth cohort; modified TyG indices outperformed TyG alone. Methods/results report n = 258 although abstract states 225.Smiderle et al, 2021Carla Alessandra Smiderle2021Brazil (Southern Brazil; Santa Casa de Misericordia de Porto Alegre)Retrospective cross-sectional study2007–2016Adults with obesity undergoing bariatric-surgery with intraoperative liver biopsy423 total37.1 ± 9.7 yearsFemale 333/423 (78.7%); male 90/423 (21.3%)43.8 ± 5.6 kg/m2TyGContinuous TyG index with ROC-derived cutoffs for histologic outcomesSignificant steatosis (grade 2–3), significant NASH (grade 2–3), advanced fibrosis (F3–4)Intraoperative liver biopsy; steatosis, NASH, and fibrosis staged histologically by Kleiner criteriaNo follow-upNo multivariable adjustment reported; diagnostic performance study using AUROC and cutoff analysisBiopsy-based obese bariatric-surgery cohort; TyG showed modest performance for steatosis/NASH but useful NPV for excluding advanced fibrosisKhamseh et al, 2021Mohammad E. Khamseh2021Iran (Tehran)Cross-sectional studyNROverweight/obese adults without diabetes, aged 30–65 years184 total (96 with NAFLD; 88 without NAFLD)NAFLD group 44.4 ± 9.3 years; control group 45.0 ± 8.7 years93 men and 91 women overallNAFLD group median 30.9 (29.0–32.7) kg/m2; control group median 30.1 (28.0–31.6) kg/m2TyG, TyG-BMI, TyG-WCCompared as continuous indices; ROC analyses for NAFLD and liver fibrosisNAFLD and liver fibrosisTransient elastography (FibroScan); CAP ≥ 288 dB/m for NAFLD and LSM ≥ 5.9 kPa for liver fibrosisNo follow-upAge, gender, waist-to-hip ratio, SBP, DBP, serum cholesterol, ALT, AST, HOMA-IR, statin medication, smoking, physical activityObesity-focused screening study; TyG-WC best for NAFLD prediction, TyG-BMI relatively better for fibrosisGuo et al, 2020Wen Guo2020China (Nanjing)Retrospective cross-sectional studyMay 2017-Nov 2019Adults undergoing health checkups at the First Affiliated Hospital of Nanjing Medical University4784 total (2902 NAFLD; 1882 non-NAFLD)NAFLD group 49.15 ± 10.23 years; non-NAFLD group 48.77 ± 10.65 years3231 males and 1553 females overallNAFLD group 26.96 ± 2.94 kg/m2; non-NAFLD group 22.85 ± 2.21 kg/m2TyG indexContinuous TyG index; quartiles also analyzed for NAFLD prevalence and fibrosis riskNAFLD; hepatic steatosis severity; liver fibrosis in NAFLDUltrasound for NAFLD; transient elastography/FibroTouch with FAP for steatosis severity and LSM > 7.3 kPa for liver fibrosisNo follow-upBMI, sex, age, SBP, DBP, uric acid, HbA1c, ALT, AST, GGT, HDL-C; some models also included TC and LDL-CLarge Chinese health-check population; TyG predicted NAFLD well but had limited AUROC for fibrosisFedchuk et al, 2014L. Fedchuk2014France (Paris)Retrospective biopsy-based cross-sectional study2000–2010Consecutive adults with clinical and/or ultrasonographic suspicion of NAFLD undergoing liver biopsy324 totalMedian 54 (45–60) yearsMale 206/324 (64%)Median 29 (26–33) kg/m2TyG included among 5 steatosis biomarkers (FLI, NAFLD-LFS, HSI, VAI, TyG)Biomarkers analyzed continuously and across histologic steatosis gradesAny steatosis (>5%); moderate/severe steatosis (>33%); NASH; advanced NASH; advanced fibrosisLiver biopsy (gold standard); ultrasound data also retrievedNo follow-upNot a conventional multivariable epidemiologic model; analyses tested confounding by fibrosis and necroinflammationBiopsy-validation study focused on diagnostic performance and limitations of steatosis biomarkers in suspected NAFLDChoe et al, 2020A Reum Choe2020Republic of Korea (single tertiary center, Seoul)Retrospective cross-sectional study with temporal validation2011–2016Adults with chronic kidney disease attending nephrology clinic and undergoing baseline abdominal ultrasound819 total (derivation 567; validation 252)Derivation 64.7 ± 15.3 years; validation 64.6 ± 14.8 yearsMale 317/567 (55.9%) in derivation; 165/252 (65.5%) in validationDerivation 24.6 ± 3.8; validation 25.4 ± 3.8 kg/m2TyG index (HSI also calculated; TyG used in final NAFLD prediction model)Continuous-variable logistic regression and derived prediction model with temporal validationNAFLD presence; liver fibrosis severity assessed secondarily among NAFLD patientsAbdominal ultrasound for NAFLD after excluding other liver diseases and excess alcoholNo follow-upFinal multivariable model: BMI, hemoglobin, ALT, eGFR, TyG indexCKD-specific study; disease-detection model rather than general-population TyG screening studyWang et al, 2025Xiuli Wang2025Japan (Murakami Memorial Hospital; NAGALA database)Cross-sectional study (secondary analysis of public health-check data)Health examinations 2004–2015Japanese nonobese adults undergoing health examinations11,987 participantsOverall mean 43.45 ± 8.98 yearsFemale 6227/11,987 (51.95%); male 5760/11,987 (48.05%)nonobese population by design (BMI < 25 kg/m2); by outcome group: 20.85 ± 2.09 vs 23.08 ± 1.40 kg/m2TyG, TyG-BMI, TyG-WC, TyG-WHtRQuartiles and continuous-variable analysesNAFLD presence/risk in nonobese populationAbdominal ultrasound for NAFLD; alcohol threshold exclusions appliedNo follow-upAge, gender, BMI, WC, WHtR, AST, ALT, GGT, HDL-c, TC, SBP, DBP, smoking status, alcohol consumptionnonobese Japanese screening cohort; compares multiple TyG-related parameters and reports nonlinear associationsYang et al, 2025Xin Yang2025China (Sichuan Province; Affiliated Hospital of Southwest Medical University)Retrospective cross-sectional health-check studyJul 2021-Jun 2024Adults undergoing health examinations with abdominal ultrasonography71,299 participantsMAFLD: median 49 (38,56) years; non-MAFLD: 41 (32,52) yearsMale: 72.01% in MAFLD vs 42.21% in non-MAFLDMAFLD: 26.59 (24.92,28.55) vs non-MAFLD: 22.36 (20.62,24.17) kg/m2TyG, TyG-BMI, TyG-WC, TyG-WHRContinuous-variable analyses with ROC and subgroup analysesMAFLD presenceAbdominal ultrasound plus MAFLD metabolic criteriaNo follow-upModel 3: sex, age, BMI, hypertension, diabetes mellitus, AST/ALT, AST, ALT, TC, GGT, HDL-C, LDL-CVery large Chinese screening cohort; TyG-BMI showed highest overall predictive valueMa et al, 2025Hanchen Ma2025United States (NHANES 2017-March 2020 pre-pandemic)Population-based cross-sectional studyNHANES 2017-March 2020 (pre-pandemic)U.S. adults with NAFLD diagnosed by VCTE/ FibroScan1679 NAFLD participants; 170 with at-risk NASHMedian 53.0 (37.0–64.0) yearsMale 844/1679 (52.1%)NR in overall baseline tableTyG, TyG-WC, TyG-WHR, TyG-WtHR, TyG-BMITertiles plus continuous-variable analysesAt-risk NASH among adults with NAFLDVCTE/ FibroScan; NAFLD defined by CAP ≥ 248 dB/m; at-risk NASH defined by FAST score ≥ 0.35No follow-upAge, gender, race, education level, hypertension, T2DM, PLT, ALT, AST, GGTNationally representative NAFLD cohort; focuses on at-risk NASH rather than general-population NAFLD detectionChen et al, 2025Yuwei Chen2025China (Quanzhou, Fujian Province)Hospital-based cohort study of older adults2022–2024Adults aged ≥60 years undergoing health examinations at Nan’an General Hospital Nanqiao Branch3954 total; 2067 with NAFLDMean 72.01 +/- 5.46 yearsFemale 2103/3954 (53.19%); male 1851/3954 (46.81%)Mean 23.00 +/- 3.32 kg/m2TyG, TyG-WC, TyG-BMI, TyG-WHtRQuartiles; also analyzed continuously and by change in TyG over follow-upNAFLD status; hepatic steatosis severity (mild/moderate/severe); NAFLD progressionB-mode ultrasonography after excluding other liver diseases and excess alcohol2-year follow-upAge, sex, marital status, alcohol consumption, smoking status, exercise, hypertension, diabetes, coronary heart disease, uric acid, HDL, ALT, ASTOlder-adult-specific study; also evaluated delta TyG as a predictor of NAFLD progressionCao et al, 2025Tengrui Cao2025China (Beijing) and TaiwanLongitudinal cohort study with external validation cohortPrimary cohort 2016–2022; external validation cohort 1996–2006Adults free of NAFLD at baseline from the Beijing Health Management Cohort; external validation in the Taiwan MJ health-screening cohort27,052 in primary cohort; 75,023 in external validation cohortPrimary cohort median 36.0 (30.0–44.0) yearsPrimary cohort male 10,625/27,052 (39.3%)Primary cohort median 22.49 (20.61–24.47) kg/m2TyG indexContinuous and quartile analyses; causal inference using EBCT and TMLEIncident NAFLD during follow-upLiver ultrasound during follow-up based on standard NAFLD diagnostic criteria after excluding secondary causes and excess alcoholPrimary cohort median 2.49 years (IQR 1.35–3.98); longer follow-up in external validation cohortAge, sex, BMI, WC, smoking status, exercise, SBP, DBP, AST, ALT, LDL-C, HDL-C, SUA, examination yearKey longitudinal TyG study; evaluates incident NAFLD and strengthens temporal/causal inference with EBCT and TMLE plus external validationAHRQ = Agency for Healthcare Research and Quality, ALF = advanced liver fibrosis, ALT = alanine aminotransferase, AST = aspartate aminotransferase, AUC = area under the curve, BMI = body mass index, CAP = controlled attenuation parameter, CRP = C-reactive protein, DBP = diastolic blood pressure, EBCT = ensemble bootstrap causal tree, eGFR = estimated glomerular filtration rate, FAST = FibroScan-aspartate aminotransferase score, FIB-4 = fibrosis-4 index, GGT = gamma-glutamyl transferase, HbA1c = glycated hemoglobin, HDL-c = high-density lipoprotein cholesterol, HOMA-IR = homeostatic model assessment of insulin resistance, IQR = interquartile range, LAP = lipid accumulation product, LDL-C = low-density lipoprotein cholesterol, LSM = liver stiffness measurement, MAFLD = metabolic dysfunction-associated fatty liver disease, MBP = mean blood pressure, MetS = metabolic syndrome, METS-IR = metabolic score for insulin resistance, NAFLD = nonalcoholic fatty liver disease, NAS = NAFLD activity score, NASH = nonalcoholic steatohepatitis, NHANES = National Health and Nutrition Examination Survey, NR = not reported, PLT = platelet count, ROC = receiver operating characteristic, SBP = systolic blood pressure, SUA = serum uric acid; T2DM, type 2 diabetes mellitus, TBIL = total bilirubin, TC = total cholesterol, TMLE = targeted maximum likelihood estimation, TP = total protein, TyG = triglyceride-glucose index, VAI = visceral adiposity index, VCTE = vibration-controlled transient elastography, WC = waist circumference, WHR = waist-to-hip ratio, WHtR = waist-to-height ratio, WTI = waist triglyceride index. Baseline characteristics table for TyG-related indices and NAFLD-related outcomes. AHRQ = Agency for Healthcare Research and Quality, ALF = advanced liver fibrosis, ALT = alanine aminotransferase, AST = aspartate aminotransferase, AUC = area under the curve, BMI = body mass index, CAP = controlled attenuation parameter, CRP = C-reactive protein, DBP = diastolic blood pressure, EBCT = ensemble bootstrap causal tree, eGFR = estimated glomerular filtration rate, FAST = FibroScan-aspartate aminotransferase score, FIB-4 = fibrosis-4 index, GGT = gamma-glutamyl transferase, HbA1c = glycated hemoglobin, HDL-c = high-density lipoprotein cholesterol, HOMA-IR = homeostatic model assessment of insulin resistance, IQR = interquartile range, LAP = lipid accumulation product, LDL-C = low-density lipoprotein cholesterol, LSM = liver stiffness measurement, MAFLD = metabolic dysfunction-associated fatty liver disease, MBP = mean blood pressure, MetS = metabolic syndrome, METS-IR = metabolic score for insulin resistance, NAFLD = nonalcoholic fatty liver disease, NAS = NAFLD activity score, NASH = nonalcoholic steatohepatitis, NHANES = National Health and Nutrition Examination Survey, NR = not reported, PLT = platelet count, ROC = receiver operating characteristic, SBP = systolic blood pressure, SUA = serum uric acid; T2DM, type 2 diabetes mellitus, TBIL = total bilirubin, TC = total cholesterol, TMLE = targeted maximum likelihood estimation, TP = total protein, TyG = triglyceride-glucose index, VAI = visceral adiposity index, VCTE = vibration-controlled transient elastography, WC = waist circumference, WHR = waist-to-hip ratio, WHtR = waist-to-height ratio, WTI = waist triglyceride index. Of the 19 included studies, 8 contributed to at least 1 quantitative synthesis and were incorporated into the main forest plots. The remaining studies were retained in the review and integrated into the evidence map and synthesis framework as descriptive evidence because they did not meet the prespecified criteria for pooling, most commonly owing to differences in exposure operationalization, outcome definition, or limited comparability across studies. 3.2. Quality assessment Methodological quality was assessed using the NOS for cohort studies and the AHRQ checklist for cross-sectional studies. Study-level quality profiles are presented in Figure S5 and Figure S6. Overall, the cohort studies were of moderate-to-good methodological quality on the NOS, with generally acceptable performance in participant selection and outcome assessment. However, as expected for observational studies, comparability across studies was influenced by differences in adjustment strategies and covariate selection. Most cross-sectional studies showed acceptable methodological quality on the AHRQ checklist, although several items were variably reported across studies. In particular, limitations were more commonly observed in the reporting of handling of missing data, assessment and control of potential confounding, and response-related methodological details. These methodological patterns were taken into account when interpreting the pooled associations, especially in analyses with substantial between-study heterogeneity. 3.3. Synthesis framework of TyG-related indices and NAFLD-related outcomes A synthesis framework of TyG-related indices and NAFLD-related outcomes is shown in Figure 2, and the broader evidence landscape is summarized in the evidence map (Fig. S1). Across the 19 included studies, TyG was the most frequently evaluated exposure and contributed the broadest evidence base across NAFLD-related outcomes. By contrast, evidence for composite indices, including TyG-BMI, TyG-WC, and TyG-WHtR, was more limited and was concentrated primarily in analyses of NAFLD/steatosis. Synthesis framework of TyG-related indices and NAFLD-related outcomes. Framework of the included studies across TyG-related exposures and NAFLD-related outcome domains, showing studies entered into quantitative synthesis and those retained as descriptive evidence. BMI = body mass index, n = number of participants, NAFLD = nonalcoholic fatty liver disease, NASH = nonalcoholic steatohepatitis, TyG = triglyceride-glucose, WC = waist circumference, WHtR = waist-to-height ratio. NAFLD/steatosis represented the dominant outcome domain and accounted for most studies eligible for quantitative synthesis. In contrast, evidence for liver fibrosis and NASH was comparatively sparse and was incorporated mainly into the descriptive synthesis. Under the predefined pooling strategy, only exposure-outcome combinations with sufficient comparability in exposure operationalization and outcome definition were entered into quantitative meta-analysis, whereas the remaining evidence was retained in the framework and evidence map as supportive descriptive data. This structure allowed us to distinguish the overall evidence landscape from the subset of studies contributing to pooled analyses, thereby clarifying why only a proportion of the included studies were incorporated into the main forest plots. The main pooled and supportive associations are summarized in Table 2. Table 2Summary of the main pooled and supportive associations.FigureExposureOutcomeAnalysis typeEvidence roleStudies (k)Participants (n)Pooled OR95% CIHeterogeneity(I2)Interpretation noteFigure 3TyGNAFLD/steatosisContinuous-scale analysisMain pooled analysis5123,7172.722.08 to 3.5585.7%Positive association; substantial heterogeneityFigure 4TyGNAFLD/steatosisHighest-versus-lowest analysisMain pooled analysis4110,0004.363.39 to 5.6192.8%Positive association; substantial heterogeneityFigure 5TyG-BMINAFLD/steatosisContinuous-scale analysisMain pooled analysis213,7631.041.03 to 1.050.0%Positive association; limited pooled evidenceFigure 6TyG-WCNAFLD/steatosisContinuous-scale analysisMain pooled analysis219601.011.01 to 1.0115.1%Positive association; limited pooled evidence Figure S2 TyG-WHtRNAFLD/steatosisHighest-versus-lowest analysisSupportive pooled analysis248309.624.09 to 22.6689.5%Supportive signal only; sparse evidence base Figure S7 TyGNAFLD/steatosisContinuous-scale analysis (strict rule)Strict-rule sensitivity analysis4118,9332.691.99 to 3.6389.9%Directionally consistent under strict ruleBMI = body mass index, CI = confidence interval, NAFLD = nonalcoholic fatty liver disease, OR = odds ratio, TyG = triglyceride-glucose index, WC = waist circumference, WHtR = waist-to-height ratio. Summary of the main pooled and supportive associations. BMI = body mass index, CI = confidence interval, NAFLD = nonalcoholic fatty liver disease, OR = odds ratio, TyG = triglyceride-glucose index, WC = waist circumference, WHtR = waist-to-height ratio. 3.4. Association between the TyG index and NAFLD/steatosis 3.4.1. Continuous-scale analyses In continuous-scale analyses, higher TyG levels were significantly associated with an increased likelihood of NAFLD/steatosis (Fig. 3).[13–17] Based on 5 studies involving 123,717 participants, the pooled OR was 2.72 (95% CI 2.08–3.55), although between-study heterogeneity was substantial (I2 = 85.7%). The direction of effect was generally consistent across studies despite variation in effect magnitude, suggesting that the heterogeneity mainly reflected differences in study populations, exposure scaling, diagnostic approaches, and adjustment strategies rather than reversal of the overall association. Association between the TyG index and NAFLD/steatosis in continuous-scale analyses. Forest plot of the pooled association between TyG and NAFLD/steatosis in continuous-scale analyses under the lenient eligibility rule. CI = confidence interval, NAFLD = nonalcoholic fatty liver disease, OR = odds ratio, TyG = triglyceride-glucose. Leave-one-out sensitivity analysis showed that the direction of the pooled association remained stable after sequential exclusion of individual studies (Fig. S3). Under the strict eligibility rule, the continuous-scale TyG analysis remained poolable and showed a directionally similar association based on 4 studies involving 118,933 participants (pooled OR 2.69, 95% CI 1.99–3.63; I2 = 89.9%) (Fig. S7 and Fig. S8).[13,14,16,17] Taken together, these findings support the robustness of the positive association between TyG and NAFLD/steatosis, while also indicating that the magnitude of association should be interpreted with caution in light of the marked between-study variability. 3.4.2. Highest-versus-lowest analyses A similar pattern was observed in highest-versus-lowest analyses (Fig. 4).[13,17–19] Based on 4 studies involving 110,000 participants, individuals in the highest TyG category had a significantly higher likelihood of NAFLD/steatosis than those in the lowest category (pooled OR 4.36, 95% CI 3.39–5.61), although heterogeneity remained considerable (I2 = 92.8%). Association between the TyG index and NAFLD/steatosis in highest-versus-lowest analyses. Forest plot of the pooled association between TyG and NAFLD/steatosis in highest-versus-lowest analyses under the lenient eligibility rule. CI = confidence interval, NAFLD = nonalcoholic fatty liver disease, OR = odds ratio, TyG = triglyceride-glucose. Sensitivity analysis further supported the stability of this finding (Fig. S4). Sequential omission of individual studies did not materially alter the direction of the pooled effect, suggesting that the observed association was not driven by any single study alone. Overall, the categorical and continuous analyses were concordant in showing that higher TyG was associated with a greater burden of NAFLD/steatosis. 3.5. Association between the TyG BMI and NAFLD/steatosis In continuous-scale analyses, TyG-BMI was also positively associated with NAFLD/steatosis (Fig. 5).[14,16] Based on 2 studies involving 13,763 participants, the pooled OR was 1.04 (95% CI 1.03–1.05), with no observed heterogeneity (I2 = 0.0%). The numerical effect size was not directly comparable with those of other indices because of differences in exposure parameterization; however, the association remained directionally positive and statistically stable. Association between the TyG-BMI and NAFLD/steatosis in continuous-scale analyses. Forest plot of the pooled association between TyG-BMI and NAFLD/steatosis in continuous-scale analyses under the lenient eligibility rule. CI = confidence interval, NAFLD = nonalcoholic fatty liver disease, OR = odds ratio, TyG-BMI = triglyceride-glucose body mass index. Because TyG-BMI incorporates BMI as an intrinsic component, interpretation of this finding should take into account differences in study populations, adiposity profiles, and multivariable adjustment strategies. Nonetheless, the absence of observed heterogeneity and the consistency of direction across the included studies support a stable positive association between TyG-BMI and NAFLD/steatosis. 3.6. Association between the TyG-WC and NAFLD/steatosis TyG-WC was likewise associated with NAFLD/steatosis in continuous-scale analyses (Fig. 6).[14,20] Based on 2 studies involving 1960 participants, the pooled OR was 1.01 (95% CI 1.01–1.01), with low heterogeneity (I2 = 15.1%). Although the magnitude of the pooled estimate should not be directly compared with that of other indices because of differences in exposure scaling, the direction of association was consistent. Association between the TyG-WC and NAFLD/steatosis in continuous-scale analyses. Forest plot of the pooled association between TyG-WC and NAFLD/steatosis in continuous-scale analyses under the lenient eligibility rule. CI = confidence interval, NAFLD = nonalcoholic fatty liver disease, OR = odds ratio, TyG-WC = triglyceride-glucose waist circumference index. Despite the relatively coherent pooled signal, interpretation of the TyG-WC findings still requires caution. The number of eligible studies was limited, and study-level differences in participant characteristics, NAFLD ascertainment, and covariate adjustment may still have influenced the pooled estimate. Thus, TyG-WC appears to be positively associated with NAFLD/steatosis, but the supporting evidence base remains modest. 3.7. Supplementary and descriptive findings Additional evidence beyond the main pooled analyses is summarized in the evidence map (Fig. S1) and the synthesis framework (Fig. 2). These visual summaries show that TyG was the most extensively studied exposure, whereas evidence for TyG-BMI, TyG-WC, and TyG-WHtR was comparatively more limited and unevenly distributed across outcome domains. For TyG-WHtR, only a small number of studies were sufficiently comparable for quantitative synthesis. In the highest-versus-lowest analysis, based on 2 studies involving 4830 participants, TyG-WHtR showed a positive association with NAFLD/steatosis (pooled OR 9.62, 95% CI 4.09–22.66), but the limited number of eligible studies and substantial between-study heterogeneity (I2 = 89.5%) precluded firm inference (Fig. S2).[18,19] Accordingly, this result was treated as supportive rather than definitive evidence. Evidence for liver fibrosis and NASH was relatively sparse overall and did not support broad quantitative synthesis under the prespecified framework. These outcomes were therefore incorporated primarily into the descriptive synthesis and evidence map. This approach allowed retention of clinically relevant but methodologically heterogeneous studies without overextending the scope of formal pooling. As an additional robustness analysis, the continuous-scale association between TyG and NAFLD/steatosis was reexamined under the strict eligibility rule (Fig. S7). Although the number of eligible studies decreased, the direction of association remained consistent with the main analysis. Leave-one-out sensitivity analysis under the strict rule likewise showed no reversal of the overall signal (Fig. S8), supporting the stability of the primary finding while underscoring the reduced evidence base under stricter pooling criteria. 4. Discussion In this revised systematic review and meta-analysis, higher TyG-related indices were consistently associated with NAFLD/steatosis, although the amount and stability of evidence differed across indices. Among the pooled analyses, the basic TyG index provided the broadest quantitative evidence base, whereas TyG-BMI and TyG-WC were also positively associated with NAFLD/steatosis but were supported by fewer directly comparable studies. Accordingly, our findings support a positive overall association between TyG-related metabolic markers and NAFLD/steatosis, while suggesting that the evidential maturity of individual indices is not uniform. A key clinical implication of the present analysis is that TyG-related indices should be viewed as accessible tools for risk stratification rather than as substitutes for imaging, elastography, or histology. The basic TyG index appears to be the most broadly applicable marker because it was evaluated most often and remained directionally stable across analytic approaches. Composite indices that incorporate anthropometric information may provide incremental value in selected settings, particularly where adiposity is closely linked to hepatic metabolic risk. A similar pattern was reported in a metabolically high-risk population, in which TyG-related composite indices, particularly TyG-WHtR, showed stronger discriminatory relevance for steatosis and fibrosis than TyG alone.[18] However, the current evidence does not support declaring any single index universally optimal, especially for fibrosis- or NASH-related outcomes, where the evidence base remains more limited. The non-pooled studies retained in the descriptive synthesis help place the pooled findings into context. Several studies suggest that TyG-related indices may extend beyond simple steatosis detection toward stratification of disease severity. This interpretation is also supported by Guo et al,[15] who reported that TyG was associated not only with NAFLD but also with steatosis severity and the presence of fibrosis, while its discrimination appeared stronger for steatosis than for fibrosis. In biopsy-based NAFLD cohorts, TyG-BMI was associated with histologic NASH activity, at-risk NASH, and fibrosis progression.[5] In metabolically high-risk biopsy cohorts, the basic TyG index was also associated with definite NASH and significant fibrosis.[9] In addition, composite TyG-related indices, particularly TyG-WHtR, showed possible value for identifying at-risk NASH in a population-based setting.[10] These data support extension of TyG-related markers toward more advanced disease phenotypes, but the current evidence remains too sparse and heterogeneous for broad formal pooling. Other descriptive studies support the practical screening relevance of composite TyG-related markers in broader populations. In a nationally representative United States cohort, TyG-WC was strongly associated with MAFLD and with greater steatosis and fibrosis severity.[2] Another NHANES-based study reported that TyG-WC, TyG-WHtR, and TyG-BMI had stronger discriminatory performance than TyG alone for NAFLD/MAFLD detection and fibrosis-related stratification.[3] A large Chinese screening cohort also showed positive dose-response relationships between TyG-related indices and MAFLD, with composite indices outperforming TyG alone.[4] These studies do not carry the same interpretive weight as pooled evidence, but they support the view that anthropometric augmentation may improve the practical screening value of TyG-based markers. The descriptive literature also suggests that the performance of TyG-related indices varies across populations and disease settings. In nonobese adults, Li et al[14] further suggested that adding WC to TyG may enhance screening performance, supporting the view that TyG-related composite indices can capture metabolically unfavorable phenotypes even outside overt obesity. In children and adolescents, modified TyG indices performed better than the basic TyG index for identifying NAFLD.[6] In patients with chronic kidney disease, TyG-related metabolic information also remained clinically informative for ultrasound-defined NAFLD.[7] By contrast, in bariatric-surgery candidates, the basic TyG index showed limited value for identifying significant steatosis or NASH, although it may still help exclude advanced fibrosis.[11] Similarly, biopsy-based comparative work suggested that TyG may detect the presence of steatosis but is less reliable for grading disease severity and may be influenced by fibrosis and necroinflammation.[12] Taken together, these studies indicate that TyG-related indices are not interchangeable across all clinical scenarios. An important strength of the present revision is the clearer separation between overall evidence and pooled evidence. Although 19 studies were included in the systematic review, only 8 contributed to at least 1 quantitative synthesis. This was not due to arbitrary exclusion, but to prespecified requirements for comparability in exposure operationalization, outcome definition, and effect-size reporting. The remaining studies were retained as descriptive evidence and were incorporated into the synthesis framework and evidence map rather than being discarded. This approach addresses the analytical clarity concerns raised during peer review and provides a more transparent account of why the quantitative evidence base was necessarily narrower than the full review dataset. Heterogeneity was substantial in several pooled analyses and should be interpreted carefully. In our view, this heterogeneity mainly reflects structural and clinical differences across studies rather than simple statistical instability. The included studies varied markedly in population composition, including general adults, nonobese individuals, adolescents, elderly populations, obesity-enriched cohorts, and biopsy-based clinical samples. Outcome assessment also differed across ultrasonography, FibroScan-based measures, liver biopsy, and surrogate scoring systems. These approaches do not capture exactly the same disease construct. In addition, exposure parameterization varied across studies, with some analyses using continuous-scale models and others highest-versus-lowest comparisons. Taken together, these differences provide a coherent explanation for why substantial between-study heterogeneity persisted despite an overall stable direction of association. The TyG-BMI analysis merits particular attention because it was specifically questioned by the reviewer. TyG-BMI is mathematically derived from both TyG and BMI. If a multivariable model further adjusts for BMI itself, WC, or closely related adiposity covariates, part of the biological signal carried by the composite exposure may be statistically absorbed by the adjustment set. This is not merely a matter of including many confounders; rather, it reflects adjustment for variables that overlap structurally or biologically with the index itself. Under such conditions, the apparent independent association of TyG-BMI may be attenuated and may become less stable across studies. This provides a plausible explanation for why the TyG-BMI results were more sensitive to study-level modeling differences, precisely the issue highlighted in peer review. Within this context, the influence of Xiuli Wang 2025 should not be framed as a post hoc exclusion undertaken to obtain a preferred result. Rather, high heterogeneity was identified first, and leave-one-out analysis was then used, as planned, to explore potential contributors. Xiuli Wang 2025 focused on a nonobese population, in which the biological meaning of a BMI-containing index may differ from that in general or obesity-enriched populations. In such a setting, the same increment in TyG-BMI may not reflect the same adiposity-related metabolic gradient. If BMI-related or central adiposity covariates are also included in the adjustment set, the incremental information carried by TyG-BMI may be further reduced. From this perspective, the change in heterogeneity after exclusion of Xiuli Wang 2025 is analytically interpretable and supports model sensitivity and effect modification rather than arbitrary data trimming. This framing is also directly aligned with the reviewer’s request that the exclusion be presented as an investigation of heterogeneity rather than a data-driven attempt to strengthen the conclusion. The fibrosis-focused literature further supports cautious interpretation beyond simple steatosis. In a Korean longitudinal NAFLD cohort, TyG-related markers were not uniformly positively associated with incident advanced fibrosis.[8] This may reflect stage-specific metabolic remodeling in more advanced disease, where lipid handling and insulin resistance patterns become more complex. Such evidence helps explain why fibrosis-related findings were less stable and why we deliberately avoided overextending formal pooling in that domain. At present, TyG-related indices appear most mature as markers of fatty liver risk, whereas their extension to fibrosis-or NASH-focused stratification remains promising but less settled. Several limitations should be acknowledged. First, most included studies were cross-sectional, which limits causal inference. Second, although 19 studies were included in the systematic review, only 8 contributed to at least 1 quantitative synthesis. We addressed this explicitly by separating pooled evidence from descriptive evidence, but several exposure-outcome pathways remained too sparse for formal meta-analysis. Third, evidence for liver fibrosis and NASH was limited and should be regarded as supportive rather than definitive. Fourth, heterogeneity remained substantial in several pooled analyses even after sensitivity analyses. Finally, subgroup-like signals derived from limited descriptive evidence, including findings in adolescents, nonobese populations, or fibrosis-focused settings, should be considered hypothesis-generating and require external validation, as specifically recommended during peer review. Overall, the present evidence supports a consistent positive association between TyG-related indices and NAFLD/steatosis, while also showing that the evidential depth differs across indices and outcome domains. The basic TyG index currently has the broadest evidence base, whereas composite indices may provide additional information in selected settings. However, further studies with more standardized designs and clearer exposure-outcome definitions are still needed to strengthen cross-study comparability. 5. Conclusion In conclusion, TyG-related indices were consistently associated with NAFLD/steatosis across the available literature. The basic TyG index had the broadest evidence base, whereas composite indices suggested possible additional value in selected settings. At present, these markers are best regarded as accessible tools for metabolic fatty liver risk stratification rather than as substitutes for established diagnostic methods. Future studies should use more standardized outcome definitions, clearer exposure parameterization, and prospective validation across diverse clinical populations to improve comparability and reduce heterogeneity. Acknowledgments The authors acknowledge the use of ChatGPT for English-language polishing during manuscript revision. All AI-assisted content was reviewed and validated by the authors, who take full responsibility for the final manuscript. Author contributions: Conceptualization: Jinyang Li. Data curation: Jinyang Li. Formal analysis: Jinyang Li. Investigation: Jinyang Li. Methodology: Jinyang Li. Project administration: Jinyang Li. Resources: Jinyang Li. Software: Jinyang Li. Supervision: Wantao Liu. Validation: Jinyang Li. Visualization: Jinyang Li. Writing – original draft: Jinyang Li. Writing – review & editing: Jinyang Li.