Waist-to-Height Ratio, An Optimal Anthropometric Indicator in Prediction of Metabolic Dysfunction Associated Fatty Liver Disease: Results From A Chinese Male Population Survey

Limited available evidence implicated that the anthropometric indicators of adiposity may contribute to the predictive and diagnosis capability of non-alcoholic fatty liver disease (NAFLD) in a facile, low-cost and noninvasive way while NAFLD had been entitled as metabolic-dysfunction associated fatty liver disease (MAFLD) in 2020. This study aimed to validate and compare the predictive and diagnostic capability of eight anthropometric indicators in MAFLD individuals and determined an optimal diagnostic predictor for MAFLD, which may benecial for resource scarce regions. Methods The retrospective cross-sectional population-based study conducted from Fangchenggang Area Male Health and Examination Survey (FAMHES) involved 2428 participants whose comprehensive questionnaire, detailed data, anthropometric parameters and biochemical measurements were collected. Eight anthropometric indicators including body mass index (BMI), waist-to-height ratio (WHtR), waist-hip ratio (WHR), body adiposity index (BAI), cardiometabolic index (CMI), lipid accumulation product (LAP), visceral adiposity index (VAI) and Abdominal volume index(AVI) were enrolled into analyzed. Receiver operating characteristic (ROC) curve analysis and the area under the ROC curves (AUCs) were used to compare the diagnostic ability of each indicator for MAFLD and optimal cut-off points fully determined. Binary logistic regression analysis was used to explore associations of all anthropometric indicators and MAFLD by determined the odds ratios (ORs) and 95% condence interval (CI). physical activity), health status (self-reported and and family history of chronic Anthropometric parameters by using a Body shape measurements including waist, hip, and thigh circumferences which were acquired while the participants in thin clothes without shoes. Weight and by digital scales to the nearest 0.1kg and 0.1cm respectively. hip for for LH, 3.4% for E2, 3.6% for T and 4.4% for SHBG. All assays were conducted to the instructions as previously described [29, 31, 34]. All including size, echogenicity, contour, structure and posterior beam attenuation. The ultrasound diagnostic criteria of fatter liver were based on the following items: diffused liver enhanced near eld echo, with stronger echoes in the hepatic parenchyma than that of the kidney; dirty liver far eld echo decays; intrahepatic duct structure display blurring [31, 34, 37]. by setting optimal cut-off points for each indicator. The AUC analysis results revealed that all eight selected anthropometric indicators had diagnostic value for MAFLD and LAP(AUC:0.868 [95%CI,0.853–0.883])had been conrmed with the largest AUC which reecting the highest diagnostic value, followed by WHtR(0.863 [0.848–0.879]), AVI(0.859 [0.843–0.874]), BMI(0.846 [0.829–0.864]), CMI(0.819 [0.800-0.838]), WHR(0.814[0.796–0.833]), VAI (0.807 [0.787–0.826]),and BAI(0.790[0.769–0.810]). The optimal cut-off points at which the risk of MAFLD increased was 24.0 (sens: sensitivity: 84.08%, spec: specicity: 72.39%) for BMI, 0.49 (sens: 90.68%, spec: ROC curve analysis in our completely AVI, BAI in discrimination of MAFLD. Superior diagnostic values of AVI, CMI and BAI were stressed as AUCs all above 0.79.The diagnostic capability of AVI and CMI (AUCs: 0.859 and 0.819) even better than WHR. As an unexpected harvest, AVI be warranted further study for its outstanding predictive and diagnostic performance in MAFLD as addressed above. Several new indices of adipose accumulation, including ABSI and BRI did not integrate into our analysis due to the limitation of their cumbersome formulas and proven poorly diagnostic power for NAFLD compare to WHtR and WHR [12]. NAFLD the intersection of IR and on NAFLD racial/ethnic groups Existing studies that HOMA-IR, HOMA-β and HOMA-IS used to evaluate evaluation, insulin secretion and insulin sensitivity, respectively In our previous research, the of HOMA-IR and HOMA-β MAFLD, consistent with previous ndings. HOMA-IR HOMA-β as the indicators of predictive capacity for MAFLD, their AUCs 0.70 with cut-off points determined high poor specicity.

Obesity substantially contributes to the risk of NAFLD [8]. Obesity and NAFLD have seen rising exponentially both in China and worldwide by parallel [2,5,9]. Mechanistically, hypertrophic obesity leads to dysfunctional and dysregulated subcutaneous adipose tissue and subsequently, the excess accumulation of ectopic fat in many depots including liver [10]. The accumulation of visceral fat has been considered the major risk factor as well as the most important predictor for NAFLD, being highly close related to the severity of the disease, particularly in the setting of dietary overindulgence [11,12].
Recent researches demonstrated that the content and accumulation of adipose tissue can be evaluated using various anthropometric indicators which used to be markers of visceral fat distribution and dysfunction [13][14][15]: Waist circumference (WC), BMI [16], LAP [17], WHR [18],WHtR [18], VAI [9,19], and BAI [20]. Currently, the most sensitive and speci c indicators for discriminating visceral fat have been described in the literature as WHtR and LAP, compared to the classic parameters as WC and BMI [21]. Meanwhile, WHR, WHtR and LAP had been regarded as good predictors of NAFLD [11,12]. Comparatively, body roundness index (BRI) [22] had been requested to be cautiously applied in the assessment of diagnostic capability of NAFLD, and body shape index (ABSI) [23] was identi ed as a poor predictor for NAFLD simultaneously [12]. Limited available evidence implicated that the anthropometric indicators of adiposity may contribute to the predictive and diagnosis capability of NAFLD in a facile, low-cost and noninvasive way despite the controversy regarding the most appropriate anthropometric indicator to predict NAFLD is yet to be elucidated [11].
Additionally, recent evidence explored that the racial/ethnic prevalence of obesity did not completely parallel NAFLD risk, incremental weight gain in Asians appeared more detrimental [24]. NAFLD has signi cant racial/ethnic variability both in the prevalence and characteristics. Of note, the fact that Chinese people have substantially higher risks of NAFLD while at much lower BMI levels compared with the Unite states population, which indicates a strong ethnic heterogeneity in anthropometric indicators and its association with NAFLD [5]. Meanwhile, a meta-analysis revealed that the Asian populations (OR:3.74 [95% CI, 2.51-5.55]) have higher NAFLD risk associated with obesity or increased BMI compared to Caucasian populations (OR: 2.67 [95% CI, 1.58-4.52]) [25]. Hitherto, a large amount of NAFLD patients remain undiagnosed because of the inadequacy of diagnostic tools for convenient, cost-effective clinical assessment and routine screen [4,5]. Thus, it is imperative to validate an effective tool for predict and diagnosis MAFLD while balancing both generalizability and eligibility.
To address this gap, herein we collected eight anthropometric indicators (BMI, WHtR, WHR, BAI, LAP, VAI, CMI [26] and AVI [27]) which re ecting visceral fat and are highly conducive to calculate, aimed to assess predictive capability as well as the optimal cut-off points of these novel indicators in individuals with MAFLD. Meanwhile, we hypothesized that anthropometric indicators of adiposity have strongly racial/ethnic disparities in re ecting liver fat deposition. Moreover, since much of the literature on NAFLD prevalence and anthropometric indicators of adiposity were mainly conducted with white ancestry in the developed areas as the reference group, which constrains generalizability to other populations and less developed regions. Thus, it is essential to further validate the predictive and diagnostic capacity of selected eight anthropometric indicators in MAFLD individuals by a large-scale Chinese male population and determine the accuracy of candidate indicators for optimal prediction and discrimination of MAFLD in Western China.

Study design and patient selection
The current research was performed using a survey from a cross-sectional population-based study, FAMHES, which was completely described previously [28][29][30][31]. Brie y, FAMHES was conducted in Fangchenggang city which located in Guangxi Zhuang Autonomous Region of Western China, namely underdeveloped regions. The study was designed to reveal the effects of environmental and genetic factors in non-hospitalized Chinese male individuals involving their interaction with the development of age-related chronic diseases.
Local men aged ≥ 18 years were convened to participate in the survey upon large-scale health examination at the Medical Center of Fangchenggang First People's Hospital from September 2009 to December 2009. Eventually, a total of 4303 men aged between 18 to 88 year were recruited with their comprehensive demographic and physical examination data exhaustively collected. After exclusions (self-reported cancer, currently diagnosed with chronic disease, acute infectious diseases, use any drug that might affect the endocrine system), 2428 participants were enrolled. Furthermore, after excluded subjects with incomplete or irrational data involved anthropometric and ultrasonography, 2355 men were quali ed for the study, and 515 were diagnosed as MAFLD by the novel expert consensus statement announced in 2020( Fig. 1)  A detailed and standardized structural questionnaire was conducted by trained physicians for all participants. The comprehensive collected data including demographic information (age, education, occupation, nancial status, etc.), lifestyle characteristics (smoking and alcohol consumption history, physical activity), health status (self-reported medical history and current medications) and family history of chronic diseases. Anthropometric parameters were undergone by trained personnel using a standardized protocol. Body shape measurements including waist, hip, and thigh circumferences which were acquired while the participants were dressed in thin clothes without shoes. Weight and height were measured by digital scales to the nearest 0.1kg and 0.1cm accuracy, respectively. The waist and hip circumferences were measured by a non-stretching tape to the nearest 0.1cm. WC was determined to approximate the midpoint between the lowest margin of the last palpable rib and the top of the iliac. Hip circumference (HC) was measured around the widest portion of the buttocks axial plane [28,32,33]. Blood pressure was measured as the mean of the rst 2 measurements taken in a seated position using the right arm by a mercury sphygmomanometer, or an automated device after at least 5-minute rest.
Participants were required to avoid vigorous exercise, drinking, and smoking for at least 30 minutes before the measurement [28].

Biochemical measurements
Blood samples were drawn from the ulnar vein between 7 am to 9 am after overnight fasting, and used to assay triglyceride (TG), total  [28,29]. Meanwhile, follicle stimulating hormone(FSH), luteinizing hormone (LH), total testosterone (TT), Estradiol(E2) and serum sex hormone-binding globulin (SHBG) were measured with electrochemiluminescence immunoassay on COBAS 6000 system E601 (Elecsysmodule) immunoassay analyzers (Roche Diagnostics, IN, Germany). The interassay coe cient of variation was 4.3% for FSH, 3.6% for LH, 3.4% for E2, 3.6% for T and 4.4% for SHBG. All assays were conducted according to the standard manufacturer's instructions as previously described [29,31,34]. HOMA% Sensitivity (HOMA-IS) was calculated as reciprocal of HOMA-IR (1/HOMA-IR) [35] Ultrasonography The abdominal ultrasonography for all participants was performed using a portable ultrasound device (GE, LOGIQ e, 5.0MHz transducer, Fair eld, CT, USA) by two experienced ultrasonographers. All individuals were assessed for liver status including size, echogenicity, contour, structure and posterior beam attenuation. The ultrasound diagnostic criteria of fatter liver were based on the following items: diffused liver enhanced near eld echo, with stronger echoes in the hepatic parenchyma than that of the kidney; dirty liver far eld echo decays; intrahepatic duct structure display blurring [31,34,37].

De nition of MAFLD
The novel positive diagnostic criteria of MAFLD was particularly noteworthy in the following content: (1) Regardless of alcohol consumption or other concomitant liver diseases; (2) Based on hepatic histological (biopsy), imaging or blood biomarker evidence of fat accumulation in the liver (hepatic steatosis); (3) In addition to one of the major three situations, namely overweight/obesity, presence of T2DM, or evidence of metabolic dysregulation; (4) The latter must de ned by the presence of at least two metabolic risk abnormalities which have been described in more detail [1].

Statistical analysis
The demographic and clinical data of the study subjects were presented using mean ± standard deviation for normality distribution and median (interquartile range) for skewed distribution as continuous variables. Comparisons of Clinical characteristics and adipose tissue accumulation indicators between groups (MAFLD and non-MAFLD) were made using Students t-test and Mann-Whitney rank sum tests appropriately. The normal percentile method was adopted to formulate the estimated reference range for all adipose tissue accumulation indicators, which were represented by P2.5-P97.5. ROC curve analysis was used to compare the diagnostic ability of each indicator for MAFLD according to the AUC. The indicator with the largest AUC was accepted as the most valuable indicator. The optimal cut-off value was determined with the maximal "Youden index" (de ned as [Sensitivity + Speci city-1]) [38]. Binary logistic regression models were used to explore associations between anthropometric indicators and MAFLD. Potential confounding variables including age (continuous), blood glucose, blood pressure, plasma uric acid, lipid parameters and sex hormone parameters were entered into models in a stepwise manner. Statistical analysis was conducted with SPSS version 25.0(IBM). Statistical tests were performed as 2-sided, and P < 0.05 was assumed statistically signi cant.

Subject Characteristic
The total 2355 enrolled participants were divided into MAFLD group (n = 515) and non-MAFLD group (n = 1840) according to the novel consensus of diagnostic criteria [1]. The prevalence of MAFLD among all participants was 21.87%. Compared to the individuals without MAFLD, age, weight, WC, HC, systolic blood pressure (SBP), diastolic blood pressure (DBP), ALT, TC, TG, LDL, UA, FPG, Insulin, HOMA-IR and HOMA-β were signi cantly higher (p < 0.001; Table 1), while HDL, HOMA-IS, LH, E2, TT and SHBG were substantially lower among individuals with MAFLD (p < 0.001; Table 1). Meanwhile, subject height and FSH concentration present no difference between the two groups (Table 1). The Mean ± Standard Deviation and recommended reference range of anthropometric indicators between individuals with and without MAFLD for men Table 2 showed the mean ± standard deviation for the anthropometric indicators in individuals with and without MAFLD for men.
According to our results, the dramatically higher values of anthropometric indicators were related to MAFLD sufferers as compared to others (p < 0.001; Table 2). We established recommended reference range of anthropometric indicators by the 2.5th and 97.5th percentiles ( Table 3). The reference of each selected anthropometric indicator in subjects with MAFLD is signi cantly higher than those without MAFLD.  The cut-off points and area under the receiver operating characteristic curves (95% con dence interval)of anthropometric indicators and HOMA-index in predicting MAFLD.
We assessed the diagnostic abilities of selected indicators in the diagnosis of MAFLD.ROC curves were analyzed by setting optimal cut-off points for each indicator.  Fig. 2). When cut-off point determined, WHtR was noticed had the highest sensitivity (90.68%), whereas LAP had the best speci city (74.73%), separately.  Table S1). The results were generally consistent with the ndings of previous studies [8,31,37,39,40]. In model II, HDL was a protective factor and diabetes was a risk factor, but after adjusting for sex hormones and SHBG (model III), HDL was no more a statistically signi cant protective factor for MAFLD in CMI, LAP and VAI models, and hypertriglyceridemia was no longer a statistically signi cant risk factor in CMI and LAP models. Plasma glucose also was no longer a signi cant factor for MAFLD in all anthropometric models except BAI. Details of multivariate binary logistic regression analysis are presented in Supplementary Table S1.

Discussion
The total NAFLD population has been estimated to increase to 314.58 million cases in China by 2030, the greatest increase in NAFLD prevalence globally [4]. Given the rapid geographic expansion and abrupt increase in the NAFLD population of China, tremendous economic and clinical burden is inevitable. The development of economical tools for widespread screening, prediction and diagnosis of MAFLD in the general population deserves further concern.
The prevalence of MAFLD in FAMHES conducted in Guangxi area turned out to be 21.87%, signi cantly lower than the overall NAFLD prevalence of China nationwide over the past two decades[29.6%(95%CI,28.2%-31.0%)] [4]. Furthermore, the prevalence of NAFLD throughout the country had reached 32.9% (95%CI, 28.9%-36.8%) in 2018 [4]. The counterintuitive outcome re ecting complex yet potentially modi able interaction between NAFLD prevalence and the area urbanization along with industrialization. According to the latest meta-analysis, the national gross domestic product (GDP) rank was inversely associated with the prevalence of NAFLD in the past two decades in China. Meanwhile, the emerging available studies regard to the prevalence of NAFLD were mainly within the region with higher GDP per capita in recent decades. The few studies performed in less developed regions may attribute to larger bias and should warrant more attention [5]. Clinical vigilance for MAFLD should be maintained without exclusion in less developed areas.
In the present study, eight selected anthropometric indicators (BMI, WHtR, WHR, BAI, CMI, LAP, VAI, AVI) were calculated, and statistical results illustrated that all anthropometric indicators were substantially associated with MAFLD. The connection powerfully emerged and persisted with MAFLD independent of potential confounding factors after logistic analysis. WHtR exhibited the persistently largest OR by stepwise binary logistic regression analysis regardless of potential confounders, suggesting WHtR may be de ned as a critical anthropometric indicator that had optimal advantages in the prediction of MAFLD over the remainders. hypertrophy, and hyperuricemia [12]. A latest systematic review and dose-response meta-analysis stressed the indices of central fatness (WC, WHR, WHtR, BAI, ABSI and waist-to-thigh ratio) were positively and signi cantly associated with higher all-cause mortality risk [15]. Yet few studies adequately evaluate the diagnostic abilities of anthropometric indicators in predictive of MAFLD.
BMI used to be a surrogate index of visceral adiposity [41], whereas its validity as an appropriate indicator of obesity has been criticized for it cannot differentiate lean body mass and fat mass, fails to characterize regional fat distribution, racial/ethnic heterogeneity and varies by gender despite comparable body fat proportion [15,24]. Nevertheless, BMI is still widely used in the evaluation of NAFLD individuals [11]. The prevalence of NAFLD is proportional to the increase in BMI [8]. 24.0kg/m 2 (sens: 84.08%, spec: 72.39%), separately. Although BMI remained that with reliable predictive value and satisfactory sensitivity and speci city for MAFLD, yet absented complete dominance compare to WHtR, as aforementioned in the present results.
Varying research ndings existed regarding the prediction e cacy of WHR and WHtR for NAFLD, nonetheless, those results are too limited to be conclusive. Although the limited number of available studies were found that addressed the prediction e cacy and diagnostic competency of WHR and WHtR in NAFLD, the WHR had been determined the indicator with the highest diagnostic value prediction performance and optimal diagnostic capability for MAFLD is WHtR, as we mentioned above. There are diverse results had been reported. Motamed et al. used a cross-sectional study with 4872 Iranian adults, reported an AUC above 0.70 in WHR as well as above 0.80 in WHtR with both sexes to detect NAFLD, emphasized that the diagnostic power of WHtR was higher than WHR [12], which is consistent with our results. The optimal cut-off points for WHtR were 0.53(sens: 82.7%, spec: 70.8%) for men and 0.58(sens: 83.3%, spec: 71.7%) for women while for WHR were not presented [12]. Yoo et al. analyzed a observational cohort including 456 subjects, identi ed AUC of WHtR in NAFLD above 0.70 in total sample and the cut-off points presented 0.52 (sens:71%, spec:65%) for men and 0.53 (sens:90%, spec: 63%) for women [44]. Despite the inconsistent ndings presented among the above studies may be partially attributed to the racial/ethnic discrepancy along with economic and geographic diversi cation, WHtR and WHR can be regarded as outstanding predictors for MAFLD undoubtedly. Notably, the ndings of Zhang et al. [43] were conducted on a population with high proportion of males and our study was performed completely by a male survey, the cut-off points of WHR and WHtR we shared consistently may suggest more useful in the discrimination of MAFLD in males.
LAP expressed a continuous marker of lipid over-accumulation, which had been proposed to be a good predictor for cardiovascular disease and diabetes [26]. VAI expressed the visceral fat function associated with cardiometabolic risk and a predictor of complications related to visceral obesity [19]. Both LAP and VAI were de ned as novel sex-speci c indices which had high accuracy in visceral obesity discrimination and were effective in cardiovascular risk assessment, especially tended to elderly men [21]. In literature, up to now, only two studies have reported the prediction of NAFLD concerning LAP. Cuthbertson et al. demonstrated an AUC of 0.78 without calculated cut-off points for NAFLD by evaluated 4 cohorts from Germany and England [45]. A large cross-sectional study utilized 40459 Chinese in Changsha had already highlighted, which LAP showed high accuracy for discriminating NAFLD in both males and females (AUCs 0.843 and 0.887, respectively), with cut-off points of 30.5 (sens:77%, spec: 75%) and 23.0 (sens:82%, spec: 79%), respectively [46]. Similarly, according to the present study, LAP had the largest AUC [0.868(95%CI 0.853-0.883)] which indicating the highest diagnostic value of MAFLD compare to the remainders. As for the cut-off points substantially lower than Dai et al. [46] (24.29 vs.30.5 in males) may be due to regional economic gap and high disparity in area utilization. The diagnostic effectiveness of VAI in NAFLD individuals remains controversial. Vongsuvanh et al. [47], Díez-Rodríguez et al. [48] and Ercin et al. [49] claimed that VAI was no more powerful than WC in discriminating hepatic steatosis from steatohepatitis and not associated with liver histology either.
While Patta et al. reported VAI was a marker of IR, both qualitative and quantitative of adipose tissue dysfunction, which signi cantly correlated with brosis in NAFLD patients [50]. Our present study demonstrated the AUC of VAI was 0.807(95% CI 0.787-0.826) in MAFLD subjects, yet VAI still ranked as posterior position in all indicators regardless of predictive capability or diagnostic value.
Of the remainders, AVI, CMI and BAI have not been applied as predictors for NAFLD in the existing literature to date. AVI was suggested to estimated overall abdominal volume, which is strongly related to impaired glucose tolerance and diabetes [27]. Yet the study about the effectiveness of AVI has not been reported so far. CMI was proposed as a good predictor and discriminator of a central component of metabolic syndrome and diabetes, re ecting both adiposity and blood lipids [26]. Cantero et al. reported that the substantial decline of CMI could achieve the largest decrease in TG, LDL-C, total and visceral fat mass [51]. Verma [14]. The ROC curve analysis in our study completely empowered AVI, CMI and BAI with new vision in discrimination of MAFLD. Superior diagnostic values of AVI, CMI and BAI were stressed as AUCs all above 0.79.The diagnostic capability of AVI and CMI (AUCs: 0.859 and 0.819) is even better than WHR. As an unexpected harvest, AVI should be warranted further study for its outstanding predictive and diagnostic performance in MAFLD as addressed above. Several new indices of adipose accumulation, including ABSI and BRI did not integrate into our analysis due to the limitation of their cumbersome formulas and proven poorly diagnostic power for NAFLD compare to WHtR and WHR [12].
Insulin resistance (IR) occurs as the cardinal feature of both excess adiposity and NAFLD [24]. The lipotoxicity produced by the fatty acids from dysfunctional and insulin-resistant adipocytes derived from the accumulation of triglyceride-derived toxic metabolites in ectopic tissues (skeletal muscle, pancreas, liver, etc.) [9]. Ultimately, IR triggers compensatory hyperinsulinemia, which intimately activated de novo hepatic lipogenesis, further exacerbating NAFLD [24]. Noteworthy, across racial/ethnic groups, IR presented differential manifestations. Report from the Dallas Heart Study showing a decreased prevalence of NAFLD among non-Hispanic black population despite high IR, which suggests that the intersection of IR and metabolic abnormalities on NAFLD has variable impacts among the different racial/ethnic groups [24]. Existing studies established that HOMA-IR, HOMA-β and HOMA-IS were used to evaluate IR evaluation, insulin secretion and insulin sensitivity, respectively [52]. In our previous research, the extent of HOMA-IR and HOMA-β was signi cantly correlated with the severity of NAFLD. The moderate-severe NAFLD group exhibited higher HOMA-IR (4. 16  Ideally, liver histological (biopsy) remains the gold standard for diagnosis of NAFLD, but its various defects like invasive nature,expenditure, sampling errors and potential complications, clearly indicating the impractical for routinely screen [2,5]. Consequently, non-invasive detection mothed for NAFLD relies on blood biomarkers and imaging techniques. Despite its limited sensitivity, ultrasonography contains the most widely used methodology as the rst-line diagnostic modality for NAFLD. Computed tomography and magnetic resonance imaging are available and more precisely, yet still not widely used in large-scale populations or clinical settings for the reasons including unequivocally higher cost and the ionizing radiation. Blood biomarkers appear only to be appropriate for extensive epidemiological studies, remain inconvenient for public healthy prediction and evaluation [1]. Comparably, our study contributing that the clinical anthropometric indicators of visceral obesity, especially WHtR, of easy applicability and costeffective, have been very promising in the prediction and discrimination of MAFLD, particularly in the less economically developed districts of China.

Study strengths and limitations
This is the rst large cross-sectional study that uses a Chinese male population to rigorously highlight and compare the predictive capability of eight anthropometric indicators of visceral adiposity in MAFLD individuals. And we revealed that all selected anthropometric indicators do bolster optimism in valuable diagnostic capacity for MAFLD. Furthermore, we located the optimal diagnostic predictor and its successors for MAFLD.
The study has several limitations that should be acknowledged. Since the starting point of FAMHES was focusing on male health and investigating the effects of genetic and environmental factors of age-related chronic diseases in men, we lacked female subjects enrolled in this study and absented gender comparisons. Secondly, the study was a retrospective cross-sectional design and could not draw out the identi cation of causal relationships. Thirdly, the limited available literature evaluated the association between anthropometric indicators and liver fat deposition were fully conducted with NAFLD populations as the original reference group. Correspondingly, the discrepancy of the population between NAFLD and MAFLD might lead to some extent bias and confusion when citing relevant results.

Conclusion
To summarize, our study detailed the signi cant impact of MAFLD prevalence in Western China and simultaneously underscored the effect of race/ethnic heterogeneity and regional variations in the anthropometric indicators and their associations with MAFLD. All eight anthropometric indicators illustrated the superior diagnostic value of MAFLD with AUCs fully above 0.79. Principally, we stressed WHtR as the most powerful diagnostic predictor for MAFLD, while LAP and AVI presented highly potential only after WHtR. The present ndings representing promising applied tools with widespread availability and high reproducibility may bene cial for the early prediction and discrimination of MAFLD, especially true for underdeveloped, resource scarce regions.

Availability of data and materials
All data used in this study are available from the corresponding author.

Declarations
Ethics approval and consent to participate  Flowchart of the study participants and exclusions.