Diagnostic Accuracy of Non-Invasive Diagnostic Tests for Nonalcoholic Fatty Liver Disease: A Systematic Review and Network Meta-Analysis.

Abstract

Purpose: In recent decades, numerous non-invasive tests (NITs) for diagnosing nonalcoholic fatty liver disease (NAFLD) have been developed, however, a comprehensive comparison of their relative diagnostic accuracies is lacking. We aimed to assess and compare the diagnostic accuracy of various NITs for NAFLD using network meta-analysis (NMA).

Materials and methods: We conducted a systematic search in seven databases up to April 2024 to identify studies evaluating the diagnostic values of NITs, with liver biopsy as the gold standard. The participants included patients with suspected or confirmed NAFLD, irrespective of age, sex, ethnicity. Statistical analysis was conducted using R 4.0.3 for Bayesian NMA and STATA 17.0 for pairwise meta-analysis. Sensitivity, specificity, diagnostic odds ratio (DOR), area under the receiver operating characteristic curve (AUC), and superiority index were calculated. Bayesian calculations were performed using the Rstan package, specifying parameters like MCMC chain count, iteration count, and operational cycles. The methodological quality of included studies was assessed using the QUADAS-2 tool.

Results: Out of 15,877 studies, 180 were included in the quantitative synthesis, and 102 were used in head-to-head meta-analyses. For diagnosing steatosis stage 1, Hydrogen Magnetic Resonance Spectroscopy (H-MRS, DOR 15,745,657.6, 95% CI 17.2-1,014,063.59) proved to be the most accurate. For significant fibrosis, HRI leading (DOR 80.94, 95% CI 6.46-391.41), For advanced fibrosis, CK-18 showed the highest performance (DOR 102654.16, 95% CI 1.6-134,059.8). For high-risk NASH, Real-Time Elastography showing the highest performance (DOR 18.1, 95% CI 0.7-96.33). Meta-regression analyses suggested that variability in the diagnostic accuracy of NITs for NAFLD may result from differences in study design, thresholds, populations, and performance indicators.

Conclusion: We conducted a network meta-analysis to rank the accuracy of these tests. While some results are promising, not all NITs demonstrate substantial accuracy, highlighting the need for validation with larger datasets. Future research should concentrate on studying the thresholds of NITs and enhancing the clarity of methodological reporting.