Supplementary Material for: Transient Elastography and Serum-based Tests for Diagnosis of Fatty Liver and Advanced Fibrosis in a Community Cohort- a Cross Sectional Analysis

Background: Non-invasive tests (NITs) are necessary for knowing the true prevalence of fatty liver (FL) and advanced fibrosis (AF). Noninvasive tests (NITs) for diagnosis of FL and fibrosis were compared. Methods: Data were obtained from the National Health and Examination Survey (NHANES; 2017-2018). Participants were excluded with other liver diseases, missing data for NIT calculation and/or excessive alcohol use. Area under the receiver operating characteristic (AUROC) compared the accuracy of 4 FL NITs (CAP, HIS, FLI, USFLI) among themselves and to CAP value of 285 dB/m and 5 fibrosis NITs (transient elastography, APRI, NFS, FIB-4, HEPAmet) among themselves and to LSM ≥ 8.7 kPa. Results: Among 2051 participants (average age 47 (±17.7), 48% males, 62% white, 73% overweight/obese, 39% metabolic syndrome), demographics were similar among NIT groups (CAP=812; HSI=1,234; FLI=935; USFLI-824). FL prevalence by NIT: 39% CAP, 58% HSI, 47% FLI, 37% USFLI. AF prevalence by test- LSM (≥ 8.7 kPa) 10%-14%; FIB-4 (≥2.67) and APRI (≥0.7) 1.3%- 2.7%; HEPAmet (>0.47) 14%-21%. Compared to CAP ≥285, FLI (AUROC= 0.823) and USFLI (AUROC=0.833) performed better than HSI (AUROC: 0.798). Compared to LSM ≥8.7kPa, only NFS (AUROC= 0.722) performed well (Fib-4 AUROC=0.606; APRI=0.647; HEPAmet=0.629). Among the CAP cohort, the strongest FL predictor was obesity (OR 15.2, 95%CI 7.97-28.9, P<0.001); the only fibrosis predictor was elevated AST (OR: 1.06, 95%CI 1.00-1.12, P=0.04). The addition of CAP or LSM as a second NIT reduced the number of indeterminate patients especially for FL. Conclusions: Regardless of diagnostic method in 2017-2018, the prevalence of NAFLD was >35%. NITs for FL performed well but not for AF. CAP and LSM as a second NIT reduced those considered indeterminate.

背景：明确脂肪肝（Fatty Liver, FL）与进展期纤维化（Advanced Fibrosis, AF）的真实患病率，需依赖非侵入性检测（Non-invasive Tests, NITs）。本研究对用于脂肪肝与纤维化诊断的非侵入性检测进行了对比分析。
方法：数据来源于2017-2018年国家健康与营养检查调查（National Health and Examination Survey, NHANES）。排除合并其他肝脏疾病、存在非侵入性检测计算所需数据缺失或过量饮酒的受试者。采用受试者工作特征曲线下面积（Area Under the Receiver Operating Characteristic Curve, AUROC）对比4种脂肪肝非侵入性检测：受控衰减参数（Controlled Attenuation Parameter, CAP）、肝脂肪变指数（Hepatic Steatosis Index, HSI）、脂肪变指数（Fatty Liver Index, FLI）、超声脂肪肝指数（Ultrasound-based Fatty Liver Index, USFLI）的诊断效能，并比较各检测与CAP界值285 dB/m的差异；同时对比5种纤维化非侵入性检测：瞬时弹性成像、天冬氨酸转氨酶与血小板比值指数（Aspartate Aminotransferase to Platelet Ratio Index, APRI）、非酒精性脂肪性肝病纤维化评分（NAFLD Fibrosis Score, NFS）、FIB-4指数、HEPAmet评分的诊断效能，并比较各检测与肝脏硬度测量（Liver Stiffness Measurement, LSM）界值≥8.7 kPa的差异。
结果：本研究共纳入2051名受试者，平均年龄47岁（±17.7），其中48%为男性，62%为白人，73%存在超重/肥胖，39%合并代谢综合征。各非侵入性检测组（CAP组n=812；HSI组n=1234；FLI组n=935；USFLI组n=824）的人口学特征均衡可比。不同非侵入性检测对应的脂肪肝患病率分别为：CAP组39%，HSI组58%，FLI组47%，USFLI组37%。不同检测对应的进展期纤维化患病率分别为：LSM≥8.7kPa组10%~14%；FIB-4≥2.67与APRI≥0.7组1.3%~2.7%；HEPAmet>0.47组14%~21%。以CAP≥285 dB/m为金标准时，FLI（AUROC=0.823）与USFLI（AUROC=0.833）的诊断效能优于HSI（AUROC=0.798）。以LSM≥8.7kPa为金标准时，仅NFS（AUROC=0.722）具有良好的诊断效能（FIB-4的AUROC=0.606；APRI=0.647；HEPAmet=0.629）。在CAP队列中，肥胖是脂肪肝最强的预测因素（比值比（Odds Ratio, OR）=15.2，95%置信区间（95% Confidence Interval, 95%CI）：7.97~28.9，P<0.001）；而升高的天冬氨酸转氨酶（Aspartate Aminotransferase, AST）是进展期纤维化的唯一预测因素（OR=1.06，95%CI：1.00~1.12，P=0.04）。将CAP或LSM作为第二轮非侵入性检测可减少无法明确诊断的受试者数量，在脂肪肝诊断中尤为显著。
结论：2017-2018年，无论采用何种诊断方法，非酒精性脂肪性肝病（Non-Alcoholic Fatty Liver Disease, NAFLD）的患病率均超过35%。用于脂肪肝诊断的非侵入性检测效能良好，但用于进展期纤维化诊断的效能欠佳。将CAP与LSM作为第二轮非侵入性检测可有效减少无法明确诊断的受试者数量。