Comparison of Guidelines for the Screening, Diagnosis, and Noninvasive Assessment of Nonalcoholic Fatty Liver Disease

We conducted a search of the literature using PubMed (using appropriate MeSH terms), EMBASE (using preferred Emtree terms), and Web of Science. Our screening process and search method is detailed in Figure 1.

We used the following PubMed search protocol: (("Guidelines as Topic"[MeSH Terms] OR "Guideline"[Publication Type] OR "guideline∗"[All Fields] OR "recommendation∗"[All Fields] OR "protocol∗"[All Fields]) AND ("non alcoholic fatty liver disease"[MeSH Terms] OR ("non-alcoholic"[All Fields] AND "fatty"[All Fields] AND "liver"[All Fields]) OR "non alcoholic fatty liver disease"[All Fields] OR "non alcoholic fatty liver disease"[All Fields] OR ("non"[All Fields] AND "alcoholic"[All Fields] AND "fatty"[All Fields] AND "liver"[All Fields] AND "disease"[All Fields]))) AND (2015:2022[pdat]).

We used the following EMBASE search string: “practice guideline” AND (“nafld” OR “nonalcoholic fatty liver”/exp OR “nonalcoholic fatty liver”) AND [2015–2022]/py AND ([adolescent]/lim OR [adult]/lim OR [young adult]/lim OR [middle aged]/lim OR [aged]/lim OR [very elderly]/lim). We used the following Web of Science search string: guideline or protocol AND nafld or non alcoholic fatty liver disease.

Two reviewers (PF and AS) independently reviewed all records and identified 20 relevant articles. PF and AS analyzed and extracted information from the 20 articles about screening recommendations, NITs, risk stratification, fibrosis assessment, and work-up strategies and algorithms.

In our systematic review, no guidelines recommend routine screening for NAFLD in the general population, citing inadequate cost–benefit ratios and uncertainty of treatment options and benefits of screening. We define “routine screening” as population-based detection of disease, and “case-finding” as targeted screening for disease in high-risk groups. Fourteen guidelines^{,  ,  ,  ,  ,  ,  ,  ,  ,  ,  ,  ,  ,}  recommend case-finding for NAFLD in patients with high-risk features. Of those 14 guidelines, diabetes was cited as a high-risk feature in each guideline, and metabolic syndrome and obesity were cited in 13. The KASL does not cite obesity as a high-risk feature, and the Catalan guidelines do not cite metabolic syndrome as a high-risk feature. The KASL “recommends” case-finding for all patients with persistently elevated liver enzymes and/or patients with diabetes and to “consider” case-finding patients with metabolic syndrome. The BSG recommends case-finding for NAFLD in patients with type 2 diabetes, obesity, or metabolic syndrome but notes that the evidence on case-finding strategies in high-risk groups is limited and acknowledges that European and North American guidelines differ in this regard. Similarly, the AAEH recommends case-finding in high-risk groups (diabetes, metabolic syndrome, obesity) but acknowledges that data are inconclusive whether this strategy is sustainable and cost-effective. The EASL recommends case-finding for NAFLD in patients with obesity or metabolic syndrome and stated that case-finding for advanced disease is advisable in high-risk patients (age >50, metabolic syndrome, diabetes). Seven guidelines recommend case-finding in patients with abnormal liver enzymes,^,,,,,, one guideline recommends case-finding in patients with a history of ischemic cardiovascular disease, and one guideline recommended case-finding in patients with a history of arterial hypertension.^{,  ,  ,  ,  ,  ,}  Two guidelines^, recommended against case-finding for NAFLD, even in high-risk groups. The AASLD cites uncertainties in diagnostic tests, treatment options, long-term benefits, and cost-effectiveness as reasons to defer case-finding, while the SBH simply states that there was no policy definition for tracking patients at high risk of developing NAFLD.^, Four guidelines^{,,  ,}  do not mention case-finding strategies (Table 1) . For patients with a diagnosis of NAFLD, 16 guidelines^{,,,,  ,  ,  ,  ,  ,  ,  ,  ,  ,,,} recommend screening for high-risk metabolic comorbidities such as diabetes, metabolic syndrome, dyslipidemia, and cardiovascular disease, noting the strong bidirectional relationship between these metabolic risk factors and NAFLD. Three guidelines (ALEH, CAMFIC, and AFEF) recommend the use of cardiovascular risk scoring systems such as the HeartSCORE for further risk stratification in patients with NAFLD. The BASL guidelines recommend more extensive testing for patients with severe NAFLD, defined as biopsy-proven NASH or ≥F2 fibrosis. For such patients, noninvasive assessment for subclinical atherosclerotic disease (i.e., coronary artery calcium score) is recommended. The INASL guidelines recommend screening for polycystic ovary syndrome in women with NAFLD and the metabolic syndrome.

Abdominal ultrasound (US) is the most frequently used method for the screening and diagnosis of NAFLD. Its advantages include widespread availability, relatively low cost, and overall safety of use. However, its clinical utility is limited due to its variable accuracy and inability to distinguish between NAFL and NASH. Negative findings on US cannot rule out NAFLD, as US requires the presence of steatosis in 12.5–33% of hepatocytes for optimal accuracy.^, Thus, US may miss patients with mild steatosis. Its performance is also less accurate in obese patients.^, 

Despite these limitations, 10 guidelines^{,,  ,  ,  ,  ,  ,  ,,} recommend the use of US as the first-line test for the diagnosis of NAFLD. The AACE and AASLD do not recommend conventional US for the diagnosis of NAFLD.^, The AACE recommends use of transient elastography over US when available, citing its ability to quantify steatosis and assess fibrosis severity in one session.^,  For patients with high pretest probability for NAFLD, the AACE recommends bypassing US and proceeding with fibrosis assessment via the FIB-4 since the probability of having steatosis is high. The AASLD recommends the use of clinical decision aids such as NFS, FIB-4, and VCTE over conventional US, stating that its utility as a screening tool is unproven. Four guidelines^,,, state that US can be helpful but is not required and should be used in conjunction with other NITs. The AFEF stated that steatosis is most often discovered incidentally through imaging such as US but noted that to diagnose NAFLD, these findings must be observed in a metabolic context after other causes of steatosis are ruled out.

Most guidelines agree that US can be useful in detecting moderate to high levels of steatosis and that positive results are highly suggestive of NAFLD. However, US should not be used alone and negative results cannot exclude the presence of NAFLD. Regardless of the results, US should be used in conjunction with other NITs. Once NAFLD is diagnosed, most commonly through US, assessment for hepatic fibrosis should be made using simple blood tests such as the NFS or FIB-4.

The FIB-4 is a simple index developed by Sterling et al. in 2006 to predict liver fibrosis in patients with human immunodeficiency virus and hepatitis C coinfection. The algorithm uses age, AST, INR, and platelet count to generate a numerical score. Scores of <1.45 exclude advanced fibrosis with a negative predictive value of 90% and a sensitivity of 70%, while scores of >3.25 rule in advanced fibrosis with a positive predictive value of 65% and a specificity of 97%. The FIB-4 has since been validated for use in patients with NAFLD, with Shah et al. proposing cutoffs of scores of <1.3 to rule out fibrosis with a negative predictive value of 90%, and scores of >2.67 to rule in fibrosis with a positive predictive value of 80% NAFLD.^{,  ,  ,} 

Eighteen guidelines^{,,  ,  ,  ,  ,  ,  ,  ,  ,  ,  ,  ,  ,  ,,,} recommend the use of FIB-4 for the assessment of NAFLD although guidelines differed on which cutoff values to use. Six guidelines^{,  ,  ,  ,,} recommend the cutoffs proposed by Shah et al. of <1.3 as the lower cutoff to rule out fibrosis and >2.67 as the upper cutoff to rule in fibrosis. The AISF recommends a lower cutoff of <1.3 to rule out fibrosis but does not recommend a higher cutoff to rule in fibrosis. The Catalan guidelines also do not recommend a higher cutoff but instead recommends a lower cutoff of <1.3 in patients aged <65 and a lower cutoff of <2 in patients aged 65 or greater.^, The ALEH and DGVS recommend a lower cutoff of <1.3 for those under the age of 65, and a lower cutoff of <2.0 for patients greater than 65 years old, citing decreased specificity in patients aged 65 or greater, and an upper cutoff of >2.67.^, Of note, both the FIB-4 and NFS have been shown to have poor diagnostic accuracy when used to assess individuals younger than 35 years of age, with area under the receiving operating characteristic curves of 0.52 and 0.6, respectively. In patients aged 65 or greater, the specificity of the FIB-4 and NFS for detecting advanced fibrosis declined to 35% and 25%, respectively. The AASLD recommends the cutoff originally proposed by Sterling et al. of <1.45 as the lower cutoff and >3.25 as the upper cutoff.^,  The JSG recommends a lower cutoff of <1.3 and an upper cutoff of >3.25. Seven guidelines^{,,  ,  ,,,} recommend the FIB-4 but did not specify cutoffs, and two guidelines^, did not mention the FIB-4 (Table 2) .

The NAFLD fibrosis score (NFS) was developed by Angulo et al. in 2007 to assess for advanced fibrosis in patients with NAFLD. It uses six variables: age, presence of hyperglycemia, body mass index (BMI), platelet count, albumin, and the AST/ALT ratio. Advanced fibrosis is ruled out with a score of <−1.45 and ruled in with a score of >0.676. In their retrospective model, use of the NFS score would have avoided liver biopsy in 75% of patients with correct prediction in 90% of patients.

Fourteen guidelines^{,,,  ,  ,  ,  ,  ,  ,  ,  ,,,} recommend the use of NFS. Eight guidelines^{,,,  ,  ,,,} recommend the cutoff values proposed by Angulo et al. AISF recommends a single cutoff of <−1.45 to rule out fibrosis. Patients with scores <−1.45 are recommended to repeat NITs in 1–3 years, and those with scores >−1.45 are recommended to pursue further workup for fibrosis. The Catalan guidelines and DGVS recommend a lower cutoff of <−1.4 in patients younger than 65 and a lower cutoff of <0.12 in patients 65 and older. The Catalan guidelines do not specify an upper cutoff and consider all patients with scores >−1.4 to be medium-to-high risk for fibrosis and recommend further testing with elastography.^, The DGVS recommends an upper cutoff of >0.67 to rule in advanced fibrosis.^,  Six other guidelines^{,,  ,  ,,} recommend NFS but do not specify cutoff values. The AACE recommends against using NFS in primary care settings as it overestimates fibrosis in patients with obesity and diabetes. Similarly, AFEF recommends avoiding the NFS in patients with diabetes, stating that approximately 80% of patients with hyperglycemia or diabetes have high or indeterminate NFS scores, which significantly overestimates the prevalence of advanced fibrosis in this population. The AGA clinical care pathway prefers FIB-4 over NFS, citing superior diagnostic accuracy in predicting advanced fibrosis. The NICE-UK guidelines do not comment on NFS (Table 2).

The APRI was originally developed to predict significant fibrosis and cirrhosis in patients with hepatitis C and has since been shown to reliably predict fibrosis in patients with NAFLD.^,, The formula is [[AST/upper limit of normal AST] × 100]/platelet count. Five guidelines (Egyptian guidelines, APASL, AGA, SBH, and EASL) recommended the use of APRI in NAFLD assessment. The Egyptian guidelines and APASL recommend a lower cutoff of <0.5 to rule out advanced fibrosis and an upper cutoff of >1.5 to rule in advanced fibrosis.^, The AGA recommends cutoffs of <0.48 and >1.34. Three guidelines^,, recommend the use of APRI but do not specify cutoffs. Five guidelines^,,,, do not recommend the APRI over other tests such as the NFS and FIB-4. The ALEH questions the accuracy of the APRI in NAFLD patients, citing a meta-analysis that showed superior performance of NFS and FIB-4 in detecting advanced fibrosis in NAFLD.^{,  ,  ,  ,}  Similarly, the AASLD, BASL, CSG, and DGVS recommend the NFS and FIB-4 over APRI in predicting advanced fibrosis in NAFLD (Table 2).^,,,

The ELF test is a panel that measures the levels of three serum markers of liver fibrosis: hyaluronic acid (HA), type III procollagen peptide (PIIINP), and tissue inhibitor of matrix metalloproteinase 1 (TIMP-1). This model estimates the rate of liver extracellular matrix metabolism which correlates with the severity of liver fibrosis. It was first proposed by as the “Original European Liver Fibrosis Test” in 2004 using the three aforementioned biomarkers along with patient age to estimate fibrosis in patients with chronic liver disease. In 2008, age was removed from the model and the algorithm was updated in a validation study that focused on patients with NAFLD. This new model was deemed the “Enhanced Liver Fibrosis Test” and was demonstrated to reliably detect advanced fibrosis in patients with NAFLD with an AUC of 0.9. The Siemens Healthineers medical device company later revised the algorithm using a different scale. Subsequent studies reported varying results using different manufacturer recommendations for cutoffs. A recent meta-analysis showed that the ELF test had a sensitivity of 93% and specificity of 34% at the lower cutoff of 7.7, and a sensitivity of 65% and specificity of 86% at the upper cutoff of 9.8. In low-prevalence populations, authors reported high NPV and lower PPV.

Eight guidelines^,,,,,,, recommended the use of ELF as a second-line tool for fibrosis assessment. The AACE recommends cutoffs of <7.7 to rule-in fibrosis and >9.8 to rule-out fibrosis. The Egyptian guidelines recommend a single cutoff of <10.35.^{,  ,  ,  ,  ,  ,  ,}  The KASL recommends a single cutoff of <0.3576 citing the study by Guha et al., which is equivalent to <10.17 on the Siemens scale.^,, The NICE-UK guidelines recommend a single cutoff of <10.51.^{,  ,  ,  ,  ,  ,  ,}  The ELF is recommended by APASL, EASL, BSG, and AFEF, but no cutoffs were specified.^,,, The BASL acknowledges the utility of the ELF but notes that given its use of nonroutine laboratory tests and lack of reimbursement, its use was limited in daily practice. The AASLD notes that the ELF had been approved for commercial use in Europe but states that it was not available for clinical use in the United States. Of note, 3 years after the AASLD guidelines were published, the United States Food and Drug Administration (FDA) approved the ELF for commercial use in August 2021. Similarly, the DGVS acknowledges that the ELF has a high NPV in ruling out advanced fibrosis, especially in populations with low NAFLD prevalence but states that further studies are needed to determine the use of this marker in primary care before it could be routinely recommended. The ELF is not mentioned in AISF, AGA, ALEH, Catalan guidelines, CSG, SBH, JSG, AAEH, and INASL (Table 3).^,,,,,,,,