Machine learning models are superior to noninvasive tests in identifying clinically significant stages of NAFLD and NAFLD-related cirrhosis.

BACKGROUND AND AIMS: We assessed the performance of machine learning (ML) models in identifying clinically significant NAFLD-associated liver fibrosis and cirrhosis. APPROACH AND RESULTS: We implemented ML models including logistic regression (LR), random forest (RF), and artificial neural network to predict histological stages of fibrosis using 17 demographic/clinical features in 1370 patients with NAFLD who underwent liver biopsy, FibroScan, and labs within a 6-month period at multiple U.S. centers. Histological stages of fibrosis (≥F2, ≥F3, and F4) were predicted using ML, FibroScan liver stiffness measurements, and Fibrosis-4 index (FIB-4). NASH with significant fibrosis (NAS ≥ 4 + ≥F2) was assessed using ML, FibroScan-AST (FAST) score, FIB-4, and NAFLD fibrosis score (NFS). We used 80% of the cohort to train and 20% to test the ML models. For ≥F2, ≥F3, F4, and NASH + NAS ≥ 4 + ≥F2, all ML models, especially RF, had primarily higher accuracy and AUC compared with FibroScan, FIB-4, FAST, and NFS. AUC for RF versus FibroScan and FIB-4 for ≥F2, ≥F3, and F4 were (0.86 vs. 0.81, 0.78), (0.89 vs. 0.83, 0.82), and (0.89 vs. 0.86, 0.85), respectively. AUC for RF versus FAST, FIB-4, and NFS for NASH + NAS ≥ 4 + ≥F2 were (0.80 vs. 0.77, 0.66, 0.63). For NASH + NAS ≥ 4 + ≥F2, all ML models had lower/similar percentages within the indeterminate zone compared with FIB-4 and NFS. Overall, ML models performed better in sensitivity, specificity, positive predictive value, and negative predictive value compared with traditional noninvasive tests. CONCLUSIONS: ML models performed better overall than FibroScan, FIB-4, FAST, and NFS. ML could be an effective tool for identifying clinically significant liver fibrosis and cirrhosis in patients with NAFLD.

MRI-AST (MAST) Score Accurately Predicts Major Adverse Liver Outcome, Hepatocellular Carcinoma, Liver Transplant, and Liver-Related Death.

BACKGROUND & AIMS: The MRI-AST (MAST) score accurately identifies patients with at-risk nonalcoholic steatohepatitis, defined as nonalcoholic steatohepatitis with nonalcoholic fatty liver disease activity score ≥4 and fibrosis stage ≥2 at highest risk for disease progression. It is important to determine the robustness of the MAST score in predicting major adverse liver outcomes (MALO), hepatocellular carcinoma (HCC), liver transplant, and death. METHODS: This retrospective analysis included patients with nonalcoholic fatty liver disease from a tertiary care center who underwent magnetic resonance imaging proton density fat fraction, magnetic resonance elastography, and laboratory testing within 6 months from 2013 to 2022. Other causes of chronic liver disease were excluded. Hazard ratios between logit MAST and MALO (ascites, hepatic encephalopathy, or bleeding esophageal varices), liver transplant, HCC, or liver-related death were computed using a Cox proportional hazards regression model. We computed the hazard ratio of MALO or death associated with MAST scores 0.165-0.242 and 0.242-1.000, using MAST scores 0.000-0.165 as the reference group. RESULTS: Among 346 total patients, average age was 58.8 years with 52.9% females and 34.4% with type 2 diabetes. Average alanine aminotransferase was 50.7 IU/L (24.3-60.0 IU/L), aspartate aminotransferase was 38.05 IU/L (22.00-41.00 IU/L), platelets were 242.9 × 109/L (193.8-290.0 × 109/L), proton density fat fraction was 12.90% (5.90%-18.22%), and liver stiffness on magnetic resonance elastography was 2.75 kPa (2.07-2.90 kPa). Median follow-up was 29.5 months. Fourteen had adverse outcomes, including 10 MALO, 1 HCC, 1 liver transplant, and 2 liver-related deaths. The Cox regression of MAST versus adverse event rate had a hazard ratio of 2.01 (95% confidence interval, 1.59-2.54; P < .0001) for each 1 logit unit increases in MAST. The corresponding Harrell concordance statistic (C statistic) was 0.919 (95% confidence interval, 0.865-0.953). The MAST score ranges of 0.165-0.242 and 0.242-1.0, respectively, had adverse event rate hazard ratio of 7.75 (1.40-42.9; P = .0189) and 22.11 (6.59-74.2; P < .0000) relative to MAST 0-0.165. CONCLUSIONS: The MAST score noninvasively identifies at-risk nonalcoholic steatohepatitis and accurately predicts MALO, HCC, liver transplant, and liver-related death.

MRI-based (MAST) score accurately identifies patients with NASH and significant fibrosis.

BACKGROUND & AIMS: Among the large population of patients with non-alcoholic fatty liver disease (NAFLD), identifying those with fibrotic non-alcoholic steatohepatitis (Fibro-NASH) is a clinical priority, as these patients are at the highest risk of disease progression and will benefit most from pharmacologic treatment. MRI-based proton density fat fraction (MRI-PDFF) and MR elastography (MRE) can risk-stratify patients with NAFLD by assessing steatosis and fibrosis, respectively. We developed a highly specific MRI-based score to identify patients with Fibro-NASH. METHODS: This analysis included derivation (n = 103) and validation (n = 244) cohorts of patients who underwent MRI, liver biopsy, transient elastography, and laboratory testing for NAFLD from 2016-2020 in 2 tertiary care centers. To identify Fibro-NASH, a formula was developed based on MRI-PDFF, MRE, and a third variable with highest balanced accuracy per logistic regression. The MRI-aspartate aminotransferase (MAST) score was created and compared to NAFLD fibrosis (NFS), Fibrosis-4 (FIB-4), and FibroScan-aspartate aminotransferase (FAST) scores. RESULTS: The MAST score demonstrated high performance and discrimination in the validation cohort (AUC 0.93; 95% CI 0.88-0.97). In the validation cohorts, the 90% specificity cut-off of 0.242 corresponded to a sensitivity of 75.0%, positive predictive value (PPV) of 50.0% and negative predictive value (NPV) of 96.5%, whereas the 90% sensitivity cut-off of 0.165 corresponded to a specificity of 72.2%, PPV of 29.4%, and NPV of 98.1%. Compared to NFS and FIB-4, MAST resulted in fewer patients having indeterminate scores and an overall higher AUC. Compared to FAST, MAST exhibited a higher AUC and overall better discrimination. CONCLUSION: The MAST score is an accurate, MRI-serum-based score that outperforms previous scores in non-invasively identifying patients at higher risk of Fibro-NASH. LAY SUMMARY: Identifying patients with non-alcoholic steatohepatitis and significant fibrosis - who need treatment and are at risk of clinical liver-related outcomes - is a clinical priority. We developed a more accurate score using MRI-based technologies and a laboratory blood test (aspartate aminotransferase) that outperforms previous non-invasive scores for the identification of patients at higher risk of liver disease progression.

Longitudinal Assessment of Bile Duct Loss in Primary Biliary Cholangitis.

INTRODUCTION: Bile duct involvement is a key finding of primary biliary cholangitis (PBC). The aim of this study was to evaluate baseline ductopenia and disease progression. METHODS: Retrospective longitudinal histological follow-up of treatment-naive patients with PBC. RESULTS: Eighty-three patients were included, with ductopenia correlated to fibrosis stage at baseline. The cumulative incidence of severe ductopenia remained stable after 5 years, whereas fibrosis continually increased over time. Baseline AST-to-Platelet Ratio Index and elevated alkaline phosphatase >2 times the normal with abnormal bilirubin were associated with ductopenia progression. DISCUSSION: Bile duct injury does not seem to follow the same course as fibrosis in PBC.

Progression of Nonalcoholic Fatty Liver Disease-Associated Fibrosis in a Large Cohort of Patients with Type 2 Diabetes.

BACKGROUND: Nonalcoholic fatty liver disease (NAFLD) can progress to advanced fibrosis, especially in patients with type 2 diabetes. Small studies have shown that fibrosis can also regress. AIM: We aimed to provide large-scale data on progression and regression of fibrosis in diabetics with NAFLD. METHODS: Adult diabetic patients with the diagnosis of NAFLD based on ICD-9 codes were identified. We used scores from noninvasive tests to identify patients with advanced fibrosis, calculated at first assessment and last follow-up visit. Cutoff values for advanced fibrosis were AST: ALT ratio > 1.4, AST to platelet ratio index > 1.5, FIB-4 score > 2.67, and NAFLD fibrosis score > 0.676. RESULTS: Our cohort included 50,695 diabetics with NAFLD (55.3% female; 71% Caucasian; mean age, 51.2 ± 11.6 y). During median follow-up of 84.4 months, 25.8% transitioned from no advanced fibrosis to advanced fibrosis (progression), 6.4% transitioned from advanced fibrosis to no advanced fibrosis (regression), and the rest remained stable. Factors associated with transition to advanced fibrosis were female sex, older age at first evaluation, African-American race, obesity, chronic kidney disease, or coronary artery disease. Use of insulin increased the risk of progression to advanced fibrosis (odds ratio,1.36; p < .001), whereas use of oral hypoglycemic agents, angiotensin 2 receptor blockers, and fibrates was associated with reduced risk (odds ratios, 0.92, 0.94 and 0.90, respectively; all p < .05). CONCLUSIONS: In a large cohort of patients with type 2 diabetes and NAFLD, more than a quarter progressed to advanced fibrosis. These findings indicate the need for early detection and staging of NAFLD in diabetics.

Comparison of residual shunt rate and complications across 6 different closure devices for patent foramen ovale.

OBJECTIVES: To compare residual shunt rate and complications associated with six different devices used for PFO closure. BACKGROUND: Transcutaneous PFO closure is an effective treatment for preventing recurrent stroke in patients with a history of cryptogenic stroke. The rate of residual shunt is one metric by which the technical success of PFO closure can be measured. METHODS: Patients who underwent PFO closure at a single center between February 2001 and July 2019 were retrospectively enrolled in the study. Right-to-left shunt at baseline and during follow-up was assessed using transcranial Doppler (TCD) or transthoracic echocardiography (TTE). Periprocedural and device-related complications, including atrial fibrillation, were also assessed. RESULTS: Of 467 PFO closures performed during this period, 320 patients received quantitative assessment of right-to-left shunting both before and after percutaneous closure. The highest effective closure was achieved with the Cardioform device (100%, n = 104), followed by the Amplatzer Cribriform (93%, n = 14), Helex (90%, n = 137), Amplatzer ASO (88%, n = 17), CardioSEAL (86%, n = 14), and Amplatzer PFO (85%, n = 33) devices. The most common significant adverse event was atrial fibrillation, which was more common with the Cardioform device (13%) than the Helex (4%) or the Amplatzer PFO (4%) devices. CONCLUSIONS: The Gore Cardioform Septal Occluder provides more robust closure of a PFO when compared to other devices but its effectiveness is offset by the higher prevalence of transient atrial fibrillation.

MR elastography-based liver fibrosis correlates with liver events in nonalcoholic fatty liver patients: A multicenter study.

BACKGROUND & AIMS: Liver fibrosis assessed by liver biopsy is predictive of clinical liver events in patients with nonalcoholic fatty liver disease (NAFLD). Magnetic resonance elastography (MRE) correlates with liver biopsy in assessing liver fibrosis. However, data assessing the relationship between MRE and clinical liver events are lacking. We investigated the association between MRE and clinical liver events/death and identified the cut-off to predict clinical liver events in NAFLD patients. METHODS: We conducted a multicenter retrospective study of NAFLD patients who underwent MRE between 2016 and 2019. Clinical liver events were defined as decompensation events and death. We categorized patients into noncirrhosis, compensated cirrhosis and decompensated cirrhosis. Fisher's exact test was used to test association strength. Receiver operative curve methods were used to determine the optimal cut-off of MRE liver stiffness and to maximize the accuracy for classifying noncirrhosis, compensated cirrhosis and decompensated cirrhosis. Logistic regression modelling was used to predict decompensation. RESULTS: The study included 320 NAFLD patients who underwent MRE. The best threshold for distinguishing cirrhosis from noncirrhosis was 4.39 kPa (AUROC 0.92) and from decompensated cirrhosis was 6.48 kPa (AUROC 0.71). Odds of decompensation increased as liver stiffness increased (OR 3.28) (P < .001). Increased liver stiffness was associated with ascites, hepatic encephalopathy, oesophageal variceal bleeding and mortality (median 7.10, 10.15 and 10.15 kPa respectively). CONCLUSION: In NAFLD patients, liver stiffness measured by MRE with a cut-off of ≥6.48 kPa is associated with decompensation and mortality, and specific MRE cut-offs are predictive of individual clinical liver events.

Frequency of Patent Foramen Ovale and Migraine in Patients With Cryptogenic Stroke.

BACKGROUND AND PURPOSE: Individuals with migraine are at higher risk for stroke, but the mechanism has not been established. On the basis of the association between migraine and intracardiac right-to-left shunt, it has been proposed that stroke in migraineurs could be caused by a paradoxical embolus passing through a patent foramen ovale (PFO) or pulmonary arteriovenous malformation. The aim of this study was to determine the prevalence of PFO with right-to-left shunt in patients who presented with cryptogenic stroke and had a history of migraine. METHODS: Patients between 18 and 60 years old who presented with an ischemic stroke were characterized based on ASCOD phenotyping (atherosclerosis; small-vessel disease; cardiac pathology; other causes; dissection). A migraine diagnosis was identified by reviewing physician notes, and frequent aura was defined if present in at least 50% of attacks. A PFO with right-to-left shunt diagnosis was identified by the presence of a positive bubble contrast study with either transcranial Doppler, transthoracic, or transesophageal echocardiography. RESULTS: Of the 712 patients who presented with ischemic stroke, 127 (18%) were diagnosed as cryptogenic; 68 patients had adequate testing for PFO and a documented migraine history. The prevalence of PFO in patients with cryptogenic stroke without migraine was elevated (59%) compared with the general population (18%). Patients with both cryptogenic stroke and migraine had a higher prevalence of PFO (79%). In patients with cryptogenic stroke who had migraine with frequent aura, the prevalence of PFO was 93%. Only 5 patients (4%) had a history compatible with migrainous infarction. CONCLUSIONS: In patients with cryptogenic stroke who have migraine, there is a high prevalence (79%) of PFO with right-to-left shunt. The timing of the stroke in migraineurs is usually not related to a migraine attack. These observations are consistent with the hypothesis that the mechanism of stroke in migraineurs is most likely because of a paradoxical embolus. Future cryptogenic stroke classification schemes should consider including PFO as a separate etiologic category.

The effect of patent foramen ovale closure in patients with platypnea-orthodeoxia syndrome.

BACKGROUND: Platypnea-orthodeoxia syndrome is a rare condition characterized by hypoxemia in the upright position that is improved in the supine position. Although several etiologies of platypnea-orthodeoxia exist, it is frequently associated with right-to-left shunting of blood at the cardiac or pulmonary level, usually via a patent foramen ovale (PFO). The aim of this study was to evaluate the incidence of platypnea-orthodeoxia syndrome in a select patient population with right-to-left shunting and to describe the outcomes after PFO closure. METHODS: Patients with platypnea-orthodeoxia were prospectively identified from a population of patients who had a PFO and were referred to UCLA from 2001 to 2012. Those patients who elected to have their PFO closed were assessed for the severity of their symptoms and interval SaO2 changes. The changes in SaO2 before and after closure were compared in the supine and upright position. Patients were classified depending on the result of PFO closure as having "improved SaO2 " or "no change." RESULTS: Of 683 patients with PFO-associated conditions, 17 (2.5%) had platypnea-orthodeoxia and elected to close their PFO. The results in 11 of 17 patients (64.8%) were classified as having "improved SaO2 "; they experienced improvement or complete resolution of their dyspnea and hypoxemia (improved SaO2 from baseline 5.2 ± 4.7% when recumbent and 15.6 ± 3.0% when upright, P = 0.03 and P < 0.0001, respectively). Patients with no change after PFO closure predominantly had a pulmonary etiology for their hypoxia, with elevated mean pulmonary pressures measured before closure (51.4 ±16.8 mmHg, P = 0.06). CONCLUSION: PFO closure may resolve symptomatic postural dyspnea and hypoxemia and is an effective method for treating platypnea-orthodeoxia, but is not effective when the primary etiology of the hypoxemia is due to a pulmonary cause.

Monitoring disease progression in metabolic dysfunction-associated steatotic liver disease.

BACKGROUND: Metabolic dysfunction-associated steatotic liver disease (MASLD) is the most common cause of chronic liver disease. Its prevalence is increasing with the epidemic of obesity and metabolic syndrome. MASLD progression into metabolic dysfunction-associated steatohepatitis (MASH) and advanced fibrosis may lead to decompensated cirrhosis and development of liver-related events, hepatocellular carcinoma and death. Monitoring disease progression is critical in decreasing morbidity, mortality, need for transplant and economic burden. Assessing for treatment response once FDA-approved medications are available is still an unmet clinical need. AIMS: To explore the most up-to-date literature on testing used for monitoring disease progression and treatment response METHODS: We searched PubMed from inception to 15 August 2023, using the following MeSH terms: 'MASLD', 'Metabolic dysfunction-associated steatotic liver disease', 'MASH', 'metabolic dysfunction-associated steatohepatitis', 'Non-Alcoholic Fatty Liver Disease', 'NAFLD', 'non-alcoholic steatohepatitis', 'NASH', 'Biomarkers', 'clinical trial'. Articles were also identified through searches of the authors' files. The final reference list was generated based on originality and relevance to this review's broad scope, considering only papers published in English. RESULTS: We have cited 101 references in this review detailing methods to monitor MASLD disease progression and treatment response. CONCLUSION: Various biomarkers can be used in different care settings to monitor disease progression. Further research is needed to validate noninvasive tests more effectively.

Natural history of clinical outcomes and hepatic decompensation in metabolic dysfunction-associated steatotic liver disease.

BACKGROUND & AIMS: The natural progression of hepatic decompensation in metabolic dysfunction-associated steatotic liver disease (MASLD) is not well-characterised. We aimed to describe it by conducting a retrospective analysis. METHODS: This longitudinal, retrospective analysis of well-characterised MASLD cohorts followed for hepatic decompensation and death. The sequence of liver-related events was evaluated, and the median time between hepatic decompensation episodes and death versus. transplantation was measured. RESULTS: Of the 2016 patients identified, 220 (11%) developed at least one episode of hepatic decompensation during a median follow-up of 3.2 years. Ascites was the most common first liver-related event [153 (69.5%)], followed by hepatic encephalopathy (HE) [55 (25%)] and variceal haemorrhage (VH) [30 (13.6%)]. Eighteen out of the 220 (8.1%) patients had more than one liver-related event as their first hepatic decompensation. Among the patients who had the first episode, 87 (39.5%) had a second episode [44 (50.5%) HE, 31 (35.6%) ascites, and 12 (13.7%) VH]. Eighteen out of 220 (8.1%) had a third episode [10 (55.5%) HE, 6 (33.3%) VH, and 2 (11.1%) ascites]. Seventy-three out of 220 (33.1%) died, and 31 (14%) received liver transplantation. The median time from the first episode to the second was 0.7 years and 1.3 years from the second episode to the third. The median survival time from the first episode to death or transplantation was 2.0 years. CONCLUSION: The most common first liver-related event in MASLD patients is ascites. The median survival from the first hepatic decompensation to either death or transplantation is 2 years.

Low liver fat in non-alcoholic steatohepatitis-related significant fibrosis and cirrhosis is associated with hepatocellular carcinoma, decompensation and mortality.

BACKGROUND: Progression to cirrhosis in non-alcoholic steatohepatitis (NASH) is associated with a decrease in liver fat. However, the prognostic significance of liver fat content in NASH-related significant fibrosis and cirrhosis is unclear. AIM: To investigate the risk of decompensation, hepatocellular carcinoma (HCC) and mortality stratified by liver fat content in NASH-related significant fibrosis and cirrhosis. METHODS: In this meta-analysis of individual participant data, 456 patients with both magnetic resonance elastography (MRE) and MRI-derived protein density fat fraction (MRI-PDFF) were enrolled, and 296 patients with longitudinal follow-up were analysed. MRE combined with fibrosis-4 (MEFIB-index), and MRI-PDFF were used to measure liver fibrosis and fat, respectively. MEFIB-negative, MEFIB-positive+ MRI-PDFF ≥5% and MEFIB-positive+ MRI-PDFF <5% were defined as no significant liver fibrosis, NASH with significant fibrosis and higher liver fat content, and NASH with significant fibrosis and low liver fat content groups, respectively. The primary outcome was hepatic decompensation, HCC and death. RESULTS: The rates of decompensation, HCC and mortality were highest in the NASH with significant fibrosis and low liver fat group (33%, 17% and 17%, respectively), followed by the NASH with significant fibrosis and higher liver fat group (18%, 13% and 13% respectively), and lowest in the no significant fibrosis (MEFIB-negative) group (0%, 1% and 2% respectively). In multivariable-adjusted analysis, low liver fat content was strongly associated (HR = 42.2 [95% CI: 7.5-235.5, p < 0.0001]) with HCC, decompensation and death. Sensitivity analyses for patients with cirrhosis (MRE ≥5 kPa) determined consistent findings. CONCLUSIONS: Low liver fat content in patients with burnt-out NASH-related significant fibrosis and cirrhosis is associated with an increase in hepatic decompensation, HCC and mortality.

Noninvasive tests in the evaluation of nonalcoholic fatty liver disease: A review.

1_kpnc0f20Kaltura.

MEFIB-Index and MAST-Score in the assessment of hepatic decompensation in metabolic dysfunction-associated steatosis liver disease-Individual participant data meta-analyses.

BACKGROUND: There are limited data regarding the longitudinal association between MEFIB-Index (MRE combined with FIB-4) versus MAST-Score (MRI-aspartate aminotransferase) and hepatic decompensation. AIM: To examine the longitudinal association between MEFIB-Index versus MAST-Score in predicting hepatic decompensation in patients with metabolic dysfunction-associated steatotic liver disease (MASLD). METHODS: This was a longitudinal, retrospective analysis of subjects from United States, Japan, and Turkey who underwent a baseline MRE and MRI-PDFF and were followed for hepatic decompensation. Cox-proportional hazard analyses were used to assess the association between MEFIB-Index versus MAST-Score with a composite primary outcome (hepatic decompensation) defined as ascites, hepatic encephalopathy, and varices needing treatment. RESULTS: This meta-analysis of individual participants (IPDMA) included 454 patients (58% women) with a mean (±SD) age of 56.0 (±13.5) years. The MEFIB-Index (MRE ≥3.3 kPa + FIB 4 ≥1.6) and MAST-Score (>0.242) were positive for 34% and 9% of the sample, respectively. At baseline, 23 patients met criteria for hepatic decompensation. Among 297 patients with available longitudinal data with a median (IQR) of 4.2 (5.0) years of follow-up, 25 incident cases met criteria for hepatic decompensation. A positive MEFIB-Index [HR = 49.22 (95% CI: 6.23-388.64, p < 0.001)] and a positive MAST-Score [HR = 3.86 (95% CI: 1.46-10.17, p < 0.001)] were statistically significant predictors of the incident hepatic decompensation. MEFIB-Index (c-statistic: 0.89, standard error (SE) = 0.02) was statistically superior to the MAST-Score (c-statistic: 0.81, SE = 0.03) (p < 0.0001) in predicting hepatic decompensation. CONCLUSION: A combination of MRI-based biomarker and blood tests, MEFIB-Index and MAST-Score can predict the risk of hepatic decompensation in patients with MASLD.

Increased accuracy in identifying NAFLD with advanced fibrosis and cirrhosis: independent validation of the Agile 3+ and 4 scores.

BACKGROUND AND AIMS: We explored 2 novel scores, Agile 3+ and 4, to identify advanced fibrosis (≥F3) and cirrhosis (F4), respectively, in NAFLD and compared their diagnostic performances to liver stiffness measurement (LSM) by vibration-controlled transient elastography and fibrosis-4 index (FIB-4) (for Agile 3+). APPROACH AND RESULTS: This multicenter study included 548 NAFLD patients with laboratory testing, liver biopsy, and vibration-controlled transient elastography within 6 months. Agile 3+ and 4 were applied and compared with FIB-4 or LSM alone. Goodness of fit was evaluated using a calibration plot and discrimination using area under the receiver operating curve. Area under the receiver operating curves was compared using the Delong test. Dual cutoff approaches were applied to rule out and rule in ≥F3 and F4. Median (interquartile range) age was 58 (15) years. Median body mass index was 33.3 (8.5) kg/m2. Fifty-three percent had type 2 diabetes, 20% had F3, and 26% had F4. Agile 3+ demonstrated an area under the receiver operating curve of 0.85 (0.81; 0.88) similar to that of LSM [0.83 (0.79; 0.86), p=0.142] but significantly higher than that of FIB-4 [0.77 (0.73; 0.81), p<0.0001). Agile 4's area under the receiver operating curve [0.85 (0.81; 0.88)] was similar to that of LSM [0.85 (0.81; 0.88), p=0.065). However, the percentage of patients with indeterminate results was significantly lower with Agile scores compared with FIB-4 and LSM (Agile 3+: 14% vs. FIB-4: 31% vs. LSM: 13%, p<0.001; Agile 4: 23% vs. LSM: 38%, p<0.001). CONCLUSIONS: Agile 3+ and 4 are novel vibration-controlled transient elastography-based noninvasive scores that increase accuracy in the identification of advanced fibrosis and cirrhosis respectively and are ideal for clinical use due to a lower percentage of indeterminant outputs compared with FIB-4 or LSM alone.

Type 2 diabetes, hepatic decompensation, and hepatocellular carcinoma in patients with non-alcoholic fatty liver disease: an individual participant-level data meta-analysis.

BACKGROUND: Data are scarce regarding the development of hepatic decompensation in patients with non-alcoholic fatty liver disease (NAFLD) with and without type 2 diabetes. We aimed to assess the risk of hepatic decompensation in people with NAFLD with and without type 2 diabetes. METHODS: We did a meta-analysis of individual participant-level data from six cohorts in the USA, Japan, and Turkey. Included participants had magnetic resonance elastography between Feb 27, 2007, and June 4, 2021. Eligible studies included those with liver fibrosis characterisation by magnetic resonance elastography, longitudinal assessment for hepatic decompensation and death, and included adult patients (aged ≥18 years) with NAFLD, for whom data were available regarding the presence of type 2 diabetes at baseline. The primary outcome was hepatic decompensation, defined as ascites, hepatic encephalopathy, or variceal bleeding. The secondary outcome was the development of hepatocellular carcinoma. We used competing risk regression using the Fine and Gray subdistribution hazard ratio (sHR) to compare the likelihood of hepatic decompensation in participants with and without type 2 diabetes. Death without hepatic decompensation was a competing event. FINDINGS: Data for 2016 participants (736 with type 2 diabetes; 1280 without type 2 diabetes) from six cohorts were included in this analysis. 1074 (53%) of 2016 participants were female with a mean age of 57·8 years (SD 14·2) years and BMI of 31·3 kg/m2 (SD 7·4). Among 1737 participants (602 with type 2 diabetes and 1135 without type 2 diabetes) with available longitudinal data, 105 participants developed hepatic decompensation over a median follow-up time of 2·8 years (IQR 1·4-5·5). Participants with type 2 diabetes had a significantly higher risk of hepatic decompensation at 1 year (3·37% [95% CI 2·10-5·11] vs 1·07% [0·57-1·86]), 3 years (7·49% [5·36-10·08] vs 2·92% [1·92-4·25]), and 5 years (13·85% [10·43-17·75] vs 3·95% [2·67-5·60]) than participants without type 2 diabetes (p<0·0001). After adjustment for multiple confounders (age, BMI, and race), type 2 diabetes (sHR 2·15 [95% CI 1·39-3·34]; p=0·0006) and glycated haemoglobin (1·31 [95% CI 1·10-1·55]; p=0·0019) were independent predictors of hepatic decompensation. The association between type 2 diabetes and hepatic decompensation remained consistent after adjustment for baseline liver stiffness determined by magnetic resonance elastography. Over a median follow-up of 2·9 years (IQR 1·4-5·7), 22 of 1802 participants analysed (18 of 639 with type 2 diabetes and four of 1163 without type 2 diabetes) developed incident hepatocellular carcinoma. The risk of incident hepatocellular carcinoma was higher in those with type 2 diabetes at 1 year (1·34% [95% CI 0·64-2·54] vs 0·09% [0·01-0·50], 3 years (2·44% [1·36-4·05] vs 0·21% [0·04-0·73]), and 5 years (3·68% [2·18-5·77] vs 0·44% [0·11-1·33]) than in those without type 2 diabetes (p<0·0001). Type 2 diabetes was an independent predictor of hepatocellular carcinoma development (sHR 5·34 [1·67-17·09]; p=0·0048). INTERPRETATION: Among people with NAFLD, the presence of type 2 diabetes is associated with a significantly higher risk of hepatic decompensation and hepatocellular carcinoma. FUNDING: National Institute of Diabetes and Digestive and Kidney Diseases.