Biomarkers for staging fibrosis and non-alcoholic steatohepatitis in non-alcoholic fatty liver disease (the LITMUS project): a comparative diagnostic accuracy study

Vali, Yasaman; Lee, Jenny; Boursier, Jérôme; Petta, Salvatore; Wonders, Kristy; Tiniakos, Dina; Bédossa, Pierre; Geier, Andreas; Francque, Sven; Allison, Mike; Papatheodoridis, Georgios; Cortez‐Pinto, Helena; Pais, Raluca; Dufour, Jean‐François; Leeming, Diana Julie; Harrison, Stephen A.; Chen, Yu; Cobbold, Jeremy; Pavlides, Michael; Holleboom, Adriaan G.; Yki‐Järvinen, Hannele; Crespo, Javier; Karsdal, Morten; Ostroff, Rachel; Zafarmand, Mohammad Hadi; Torstenson, Richard; Duffin, Kevin L.; Yunis, Carla; Brass, Clifford A.; Ekstedt, Mattias; Aithal, Guruprasad P.; Schattenberg, Jörn M.; Bugianesi, Elisabetta; Romero–Gómez, Manuel; Ratziu, Vlad; Anstee, Quentin M.; Bossuyt, Patrick M.

doi:10.1016/s2468-1253(23)00017-1

Cited by 71 publications

(30 citation statements)

References 31 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…An external validation study found that MACK-3 (fasting glucose and insulin, AST, and CK-18) also had an AUC of 0.80 for detecting at-risk NASH. More recently, the SomaSignal test developed based on elastic net produced an AUC of 0.76 44,45 . All of these multi-marker scores include more novel biomarkers, which come with the cost of additional testing.…”

Section: Discussionmentioning

confidence: 99%

“…More recently, the SomaSignal test developed based on elastic net produced an AUC of 0.76. [44,45] All of these multi-marker scores include more novel biomarkers, which come with the cost of additional testing. Our at-risk NASH model performed well, relying only on clinical data, highlighting a potential advantage of utilizing machine learning.…”

Section: Discussionmentioning

confidence: 99%

“…We analyzed data from 966 participants in the Liver Investigation: Testing Marker Utility in Steatohepatitis (LITMUS) metacohort. [23] These participants were recruited from 12 centers in 9 countries across Europe between 2010 and 2019 and included adults with biopsy-confirmed NAFLD with available clinical, laboratory, and biomarker data within 6 months of biopsy. Serum samples drawn within 6 months of biopsy and stored at −80°C were also available.…”

Section: Methodsmentioning

confidence: 99%

See 2 more Smart Citations

Machine learning algorithm improves the detection of NASH (NAS-based) and at-risk NASH: A development and validation study

et al. 2023

Self Cite

View full text Add to dashboard Cite

Background and Aims: Detecting NASH remains challenging, while at-risk NASH (steatohepatitis and F ≥ 2) tends to progress and is of interest for drug development and clinical application. We developed prediction models by supervised machine learning techniques, with clinical data and biomarkers to stage and grade patients with NAFLD.Approach and Results: Learning data were collected in the Liver Investigation: Testing Marker Utility in Steatohepatitis metacohort (966 biopsyproven NAFLD adults), staged and graded according to NASH CRN. Conditions of interest were the clinical trial definition of NASH (NAS ≥ 4;53%), atrisk NASH (NASH with F ≥ 2;35%), significant (F ≥ 2;47%), and advanced fibrosis (F ≥ 3;28%). Thirty-five predictors were included. Missing data were handled by multiple imputations. Data were randomly split into training/validation (75/25) sets. A gradient boosting machine was applied to develop 2 models for each condition: clinical versus extended (clinical and biomarkers). Two variants of the NASH and at-risk NASH models were constructed: direct and composite models. Clinical gradient boosting machine models for steatosis/inflammation/ballooning had AUCs of 0.94/0.79/0.72. There were no improvements when biomarkers were included. The direct Abbreviations: ALT, alanine aminotransferase; AST, aspartate aminotransferase; FIB-4, Fibrosis-4; GGT, gamma-glutamyl transferase; PRO-C3; Amino-terminal propeptide of procollagen type III; P3NP, Amino-terminal propeptide of type III procollagen; AUC, Area under the receiver operating characteristic curve; Carboxyterminal propeptides of procollagen type IV (PRO-C4) and VI (PRO-C6);

show abstract

Section: Discussionmentioning

confidence: 99%

Section: Discussionmentioning

confidence: 99%

Section: Methodsmentioning

confidence: 99%

See 1 more Smart Citation

Machine learning algorithm improves the detection of NASH (NAS-based) and at-risk NASH: A development and validation study

et al. 2023

Self Cite

View full text Add to dashboard Cite

show abstract

“…The most recent reviews of NITs in T2D concluded unanimously that there is an unmet need to perform specific prospective studies in outpatients with T2D, using head-to-head comparison, to compare the numerous blood and imaging tests already available in subjects at risk of MASLD. 1,2,[5][6][7][8][9][10][11]27 Here, despite limitations, we succeed to address several unmet needs: first, a prospective head-to-head comparison of widespread NITs focusing in patients with T2D; second, we optimised the performance of an already validated NashFibroTest TA B L E 2 Diagnostic performances of optimised NashFibroTest-T2D versus original NashFibroTest.…”

Section: Discussionmentioning

confidence: 99%

Prospective direct comparison of non‐invasive liver tests in outpatients with type 2 diabetes using intention‐to‐diagnose analysis

Poynard,

Deckmyn,

Peta

et al. 2023

Aliment Pharmacol Ther

Self Cite

View full text Add to dashboard Cite

SummaryBackgroundNo prospective diagnostic studies have directly compared widespread non‐invasive liver tests in patients with type 2 diabetes (T2D) using the intention‐to‐diagnose method for each of the three main histological features of metabolic dysfunction associated steatotic liver disease ‐ namely fibrosis, metabolic dysfunction‐associated steatohepatitis (MASH), and steatosis.AimsTo compare the performance of nine tests using the intention‐to‐diagnose rather than the standard method, which would exclude non‐evaluable participantsMethodsBiopsy was used as the reference with predetermined cut‐offs, advanced fibrosis being the main endpoint. The Nash‐FibroTest panel including FibroTest‐T2D, SteatoTest‐T2D and MashTest‐T2D was optimised for type 2 diabetes. FibroTest‐T2D was compared to vibration‐controlled transient elastography stiffness (VCTE), two‐dimensional shear‐wave elastography stiffness (TD‐SWE), and Fibrosis‐4 blood test. NashTest‐T2D was compared to aspartate aminotransferase. SteatoTest‐T2D was compared to the controlled attenuation parameter and the hepatorenal gradient.ResultsAmong 402 cases, non‐evaluable tests were 6.7% for VCTE, 4.0% for hepatorenal gradient, 3.2% for controlled attenuation parameter, 1.5% for TD‐SWE, 1.2% for NashTest‐T2D, and 0.02% for Fibrosis‐4, aspartate aminotransferase and SteatoTest‐T2D. The VCTE AUROC for advanced fibrosis was over‐estimated by 6% (0.83 [95% CI: 0.78–0.87]) by standard analysis compared to intention‐to‐diagnose (0.77 [0.72–0.81] p = 0.008). The AUROCs for advanced fibrosis did not differ significantly in intention‐to‐diagnose between FibroTest‐T2D (0.77; 95% CI: 0.73–0.82), VCTE (0.77; 95% CI: 0.72–0.81) and TD‐SWE(0.78; 0.74–0.83) but were all higher than the Fibrosis‐4 score (0.70; 95% CI all differences ≥7%; p ≤ 0.03). For MASH, MashTest‐T2D had a higher AUROC (0.76; 95% CI: 0.70–0.80) than aspartate aminotransferase (0.72; 95% CI: 0.66–0.77; p = 0.035). For steatosis, AUROCs did not differ significantly between SteatoTest‐T2D, controlled attenuation parameter and hepatorenal gradient.ConclusionsIn intention‐to‐diagnose analysis, FibroTest‐T2D, TD‐SWE and VCTE performed similarly for staging fibrosis, and out‐performed Fibrosis‐4 in outpatients with type 2 diabetes. The standard analysis over‐estimated VCTE performance.ClinicalTrial.gov: NCT03634098.

show abstract

“…However, recent data demonstrated that the FIB-4 index is not a good predictor of the histological grade of fibrosis, although it is a very good indicator for the detection of NAFLD as a metabolic disorder [10,13].…”

mentioning

confidence: 99%

Liver fibrosis scores to predict cardiovascular events in hypertensive patients: old wine in new skins?

Kasper,

Demir,

Steffen

2023

Journal of Hypertension

View full text Add to dashboard Cite

Biomarkers for staging fibrosis and non-alcoholic steatohepatitis in non-alcoholic fatty liver disease (the LITMUS project): a comparative diagnostic accuracy study

Cited by 71 publications

References 31 publications

Machine learning algorithm improves the detection of NASH (NAS-based) and at-risk NASH: A development and validation study

Machine learning algorithm improves the detection of NASH (NAS-based) and at-risk NASH: A development and validation study

Prospective direct comparison of non‐invasive liver tests in outpatients with type 2 diabetes using intention‐to‐diagnose analysis

Liver fibrosis scores to predict cardiovascular events in hypertensive patients: old wine in new skins?

Contact Info

Product

Resources

About