Screening for Nonalcoholic Fatty Liver Disease in Persons with Type 2 Diabetes in the United States Is Cost-effective: A Comprehensive Cost-Utility Analysis

The US Preventative Services Task Force has no guidelines on the screening for nonalcoholic fatty liver disease (NAFLD), and the American Association for the Study of Liver Diseases (AASLD) guidance does not recommend population screening for NAFLD.1Chalasani N. et al.Hepatology. 2017; 67: 328-357Crossref PubMed Scopus (1795) Google Scholar A reference cited by this guidance concluded that screening is not cost-effective2Corey K.E. et al.Dig Dis Sci. 2016; 61: 2108-2117Crossref PubMed Scopus (30) Google Scholar; however, new data have emerged since then. We hypothesize that screening for NAFLD in patients with type 2 diabetes (T2D), starting with ultrasound (US) and alanine aminotransferase or aspartate aminotransferase and followed by noninvasive testing for fibrosis to detect those most likely to have fibrosis stage ≥F2, is more cost-effective than not screening this population. We developed a Markov model to compare the strategies of screening and treatment versus no screening and treatment of nonalcoholic steatohepatitis (NASH) in hypothetical patients with NAFLD and T2D (Supplementary Methods, Supplementary Figure 1). Our model began with a hypothetical cohort of 55-year-old persons followed across 1-year cycles until their death.3Younossi Z.M. et al.Hepatology. 2016; 64: 1577-1586Crossref PubMed Scopus (437) Google Scholar,4Younossi Z.M. et al.Hepatology. 2019; 69: 564-572Crossref PubMed Scopus (41) Google Scholar We modeled 6 screening approaches, 3 beginning with a patient population receiving US and aspartate aminotransferase and 3 with US and alanine aminotransferase (Table 1). If US + aspartate aminotransferase or US + alanine aminotransferase detected likely NAFLD (or likely NASH phenotype), screening proceeded to the next stage: in screening approaches 1 and 4, liver biopsy; in screening approaches 2 and 5, vibration-controlled transient elastography (VCTE). and if VCTE was suggestive of ≥F2, liver biopsy; and in screening approaches 3 and 6, VCTE.Table 1Results for Each Screening Strategy with ILI Treatment and Patients Aged 55 YearsStrategyTotalIncrementalIncremental Cost-effectiveness RatioCost ($)QALYsLife YearsCost ($)QALYsLife Years($/QALY)($/Life Year)No screening37,20011.590.011400–0.170.61–8345 (dominated)aDominated $/QALY results refer to when the strategy in question is both less effective and more costly than the strategy of no screening.2313Screening strategy 138,60111.430.62No screening37,20011.590.01826–0.080.18–10,948 (dominated)aDominated $/QALY results refer to when the strategy in question is both less effective and more costly than the strategy of no screening.4594Screening strategy 238,02611.520.19No screening37,20011.590.013160.010.1835,2741740Screening strategy 337,51611.600.20No screening37,20011.590.011374–0.160.598385 (dominated)aDominated $/QALY results refer to when the strategy in question is both less effective and more costly than the strategy of no screening.2326Screening strategy 438,57511.430.61No screening37,20011.590.01813–0.070.18–11,032 (dominated)aDominated $/QALY results refer to when the strategy in question is both less effective and more costly than the strategy of no screening.4638Screening strategy 538,01411.520.19No screening37,20011.590.013160.010.1836,7401784Screening strategy 637,51611.600.19NOTE. Screening strategy 1: If US+ALT suspected likely fatty liver, US+ALT was followed by liver biopsy, with 1-year ILI prescribed to diagnosed patients (NASH stage ≥2).Screening strategy 2: If US+ALT suspected likely fatty liver, US+ALT was followed by transient elastography and, if elastography suspected NASH ≥2, liver biopsy, with 1-year ILI prescribed to diagnosed patients (NASH stage ≥2).Screening strategy 3: If US+ALT suspected likely fatty liver, US+ALT was followed by transient elastography, with 1-year ILI prescribed to suspected patients (NASH stage ≥2).Screening strategy 4: If US+AST suspected likely fatty liver, US+AST was followed by liver biopsy, with 1-year ILI prescribed to diagnosed patients (NASH stage ≥2).Screening strategy 5: If US+AST suspected likely fatty liver, US+AST was followed by transient elastography and, if elastography suspected NASH ≥2, liver biopsy, with 1-year ILI prescribed to diagnosed patients (NASH stage ≥2).Screening strategy 6: If US+AST suspected likely fatty liver, US+AST was followed by transient elastography, with 1-year ILI prescribed to suspected patients (NASH stage ≥2).US+ALT, ultrasound + serum alanine aminotransferase; US+AST, ultrasound + aspartate aminotransferase.a Dominated $/QALY results refer to when the strategy in question is both less effective and more costly than the strategy of no screening. Open table in a new tab NOTE. Screening strategy 1: If US+ALT suspected likely fatty liver, US+ALT was followed by liver biopsy, with 1-year ILI prescribed to diagnosed patients (NASH stage ≥2). Screening strategy 2: If US+ALT suspected likely fatty liver, US+ALT was followed by transient elastography and, if elastography suspected NASH ≥2, liver biopsy, with 1-year ILI prescribed to diagnosed patients (NASH stage ≥2). Screening strategy 3: If US+ALT suspected likely fatty liver, US+ALT was followed by transient elastography, with 1-year ILI prescribed to suspected patients (NASH stage ≥2). Screening strategy 4: If US+AST suspected likely fatty liver, US+AST was followed by liver biopsy, with 1-year ILI prescribed to diagnosed patients (NASH stage ≥2). Screening strategy 5: If US+AST suspected likely fatty liver, US+AST was followed by transient elastography and, if elastography suspected NASH ≥2, liver biopsy, with 1-year ILI prescribed to diagnosed patients (NASH stage ≥2). Screening strategy 6: If US+AST suspected likely fatty liver, US+AST was followed by transient elastography, with 1-year ILI prescribed to suspected patients (NASH stage ≥2). US+ALT, ultrasound + serum alanine aminotransferase; US+AST, ultrasound + aspartate aminotransferase. Patients identified as NASH ≥F2 following screening approaches 1–6 underwent a 1-year intensive lifestyle intervention (ILI).5Vilar-Gomez E. et al.Gastroenterology. 2015; 149: 367-378Abstract Full Text Full Text PDF PubMed Scopus (736) Google Scholar In an exploratory analysis, those with NASH ≥F2 patients were alternatively treated with pioglitazone until death or treatment noncompliance. Treatment effects were modeled as rates of improvement in fibrosis compared with rates with placebo, according to 3 outcomes: F3 to F2, F2 to F1, or F1 to NASH with no fibrosis. Initial patient distributions were based on prevalence studies of NAFLD in T2D. Values for quality of life, disutility, and annual costs attached to individual health states are shown in Supplementary Table 1. The main outcome was the median incremental cost-effectiveness ratio. The cost (in US dollars) per quality-adjusted life year (QALY) gained threshold used to determine cost-effectiveness was $50,000/QALY. Deterministic and probabilistic sensitivity analyses were performed to evaluate the effect of parameter uncertainty on the results. An additional scenario analysis made all patients start the model at age 40 years. When screening-detected patients were treated with ILI (Table 1), screening approaches 3 and 6 were cost-effective compared with no screening strategy, with incremental cost-effectiveness ratios of $35,274/QALY and $36,740/QALY, respectively. Of note, we also found that approaches 3 and 6 were cost-effective when they started with US alone (without liver enzymes) followed by VCTE (data not shown). When patients started the model at age 40 years, screening approaches 3 and 6 were cost-effective, with incremental cost-effectiveness ratios of $17,311/QALY and $18,260/QALY, respectively. Screening strategies 3 and 6 were also cost-effective with pioglitazone as the treatment intervention. For screening approaches 3 and 6 when ILI was the intervention, 71.5% and 69.7%, respectively, of the 1000 probabilistic sensitivity analysis iterations were below the $50,000/QALY cost-effectiveness threshold, with 1.0% and 0.7% of all iterations being dominant. The US Preventative Services Task Force has no guidelines for NAFLD screening, and the AASLD has not recommended screening.1Chalasani N. et al.Hepatology. 2017; 67: 328-357Crossref PubMed Scopus (1795) Google Scholar The AASLD reference of Corey et al2Corey K.E. et al.Dig Dis Sci. 2016; 61: 2108-2117Crossref PubMed Scopus (30) Google Scholar compared a no screening strategy with a strategy in which diabetics aged > 50 years were screened with US followed by liver biopsy and then treated with pioglitazone if NASH was present. That approach found screening was not cost-effective, primarily driven by pioglitazone’s side-effect profiles. Corey et al2Corey K.E. et al.Dig Dis Sci. 2016; 61: 2108-2117Crossref PubMed Scopus (30) Google Scholar did not consider the cost of diagnosis of hepatocellular carcinoma or liver transplantation, which would have improved the cost-effectiveness of screening. Finally, Corey et al2Corey K.E. et al.Dig Dis Sci. 2016; 61: 2108-2117Crossref PubMed Scopus (30) Google Scholar used liver biopsy as a screening tool for NAFLD. Given the disease high prevalence and the high cost and invasive nature of liver biopsy, it is plausible that screening with liver biopsy would not be cost-effective. Our analysis also showed the disutility of liver biopsy to result in modeled biopsy-based screening strategies being cost-ineffective. We used VCTE instead of other endorsed noninvasive tests1Chalasani N. et al.Hepatology. 2017; 67: 328-357Crossref PubMed Scopus (1795) Google Scholar because it more accurately detects fibrosis at a lower stage (F2), which allows early therapeutic intervention. AASLD-endorsed tests (eg, FIB-4) are precise in detecting fibrosis of ≥F3, which might be too late for therapy. We project that if the FIB-4 is used, cost-effectiveness will be maintained because this score relies on routine blood tests. Our analysis refrained from modeling potential new NASH treatments, because without known costs conclusions may not be applicable. However, newer, more-expensive therapies might not be cost-effective for screening unless they target more advanced fibrosis stages. Finally, our analysis focused on administering ILI to those with NASH ≥F2. However, ILI will likely be beneficial in many clinical aspects in diabetics, even if they had NAFLD 10%77.08F2 to F1 if ILI induces weight loss > 10%77.0F1 to F0 if ILI induces weight loss > 10%100.0F3 to F2 if pioglitazone-treated12.61,5,11F2 to F1 if pioglitazone-treated12.6F1 to F0 if pioglitazone-treated11.4Intervention cost (US dollars)ILI (1 year)2864.6013Elastography39.6514Liver biopsy922.404US127.56Alanine aminotransferase test6.5415Aspartate aminotransferase test5.1815Pioglitazone (monthly)126.8416Associated health states cost (US dollars)Year 1(NASH detected) NAFL no fibrosis431.006(NASH detected) NASH no fibrosis431.00(NASH detected) F1431.00(NASH detected) F21631.00(NASH detected) F31731.00(NASH detected) F420 103.00(NASH detected) DCC36 868.00(NASH detected) HCC detected94 428.00(NASH detected) HCC undetected34 796.73(NASH detected) liver transplant356 200.00(NASH detected) after liver transplant50 036.00Year 2(NASH detected) NAFL no fibrosis431.00(NASH detected) NASH no fibrosis431.00(NASH undetected) NAFL no fibrosis431.00(NASH undetected) NASH no fibrosis431.00(NASH detected) post-liver transplant14 036.00Health states utilities defaultNAFL no fibrosis0.7301, 6NASH no fibrosis0.730F10.730F20.730F30.730F40.710DCC0.570Liver transplant0.567Post liver transplant0.576HCC (detected and undetected)0.496Intervention disutilities1-year duration ILI (treatment)0.01Author assumptionLiver biopsy F20.2017Liver biopsy F30.20Liver biopsy F40.20Liver biopsy complications F20.43Liver biopsy complications F30.43Liver biopsy complications F40.43Pioglitazone0.015DetectionProbability of chance detection0.2104Probability of US + ALT detecting fatty liver0.94018, 19Probability of US + AST detecting fatty liver0.91718, 20Probability of liver biopsy detecting ≥NASH F21.0003, 4Sensitivity and specificity of elastography≥ NASH F20.71,aArea under the receiver operating characteristic curve (95% confidence interval) = 0.77 (0.72–0.82). 0.70bYouden index, cutoff (kPa) = 8.2.4≥ NASH F30.71,cArea under the receiver operating characteristic curve (95% confidence interval) = 0.80 (0.75–0.84). 0.75dYouden index, cutoff (kPa) = 9.7.NASH F40.85,eArea under the receiver operating characteristic curve (95% confidence interval) = 0.89 (0.84–0.93). 0.79fYouden index, cutoff (kPa) = 13.6.Treatment noncompliance (%)Pioglitazone (annual)27.24NOTE. DCC, decompensated cirrhosis; HCC, hepatocellular carcinoma; NAFL, Non-alcoholic fatty liver; US + ALT, ultrasound + serum alanine aminotransferase; US + AST, ultrasound + aspartate aminotransferase.a Area under the receiver operating characteristic curve (95% confidence interval) = 0.77 (0.72–0.82).b Youden index, cutoff (kPa) = 8.2.c Area under the receiver operating characteristic curve (95% confidence interval) = 0.80 (0.75–0.84).d Youden index, cutoff (kPa) = 9.7.e Area under the receiver operating characteristic curve (95% confidence interval) = 0.89 (0.84–0.93).f Youden index, cutoff (kPa) = 13.6. Open table in a new tab NOTE. DCC, decompensated cirrhosis; HCC, hepatocellular carcinoma; NAFL, Non-alcoholic fatty liver; US + ALT, ultrasound + serum alanine aminotransferase; US + AST, ultrasound + aspartate aminotransferase. CorrectionGastroenterologyVol. 160Issue 6PreviewNoureddin M, Jones C, Alkhouri N, et al. Screening for nonalcoholic fatty liver disease in persons with type 2 diabetes in the United States is cost-effective: a comprehensive cost-utility analysis. Gastroenterology 2020;159:1985-1987.e4. Full-Text PDF