Diagnostic performance of magnetic resonance technology in detecting steatosis or fibrosis in patients with nonalcoholic fatty liver disease: A meta-analysis

Abstract

Background: The aim of this study was to evaluate the diagnostic accuracy of magnetic resonance (MR) imaging-based methods for detecting steatosis and fibrosis in nonalcoholic fatty liver disease (NAFLD).

Methods: Data were extracted from research articles obtained after a literature search from multiple electronic databases. Random-effects meta-analyses were performed to obtain overall effect size of the area of operator receiver curve (AUROC), sensitivity and specificity of MR imaging, MR elastography, and MR spectroscopy in detecting or grading steatosis/fibrosis. Meta-analysis of correlation coefficients was performed to have an overall effect size of correlation between MR-based diagnosis and histological diagnosis.

Results: Twenty-one studies (1658 subjects; 45.32 years [95% CI: 35.94, 54.71] of age, 53.67% [45.39, 61.95] males, and 29.98 kg/m [21.93, 38.04] BMI) were included in the meta-analysis. Pooled analyses of the AUROC, specificity, and sensitivity values reported in the individual studies revealed an overall effect sizes of 0.90 (0.88, 0.92), 82.27% (77.74, 86.80), and 86.94% (84.18, 95.28) in the use of any MR-based technique for the diagnosis of NAFLD or its severity. The correlation coefficient between MR-based detection of liver steatosis and histologically measured steatosis was 0.748 (0.706, 0.789) (P < .00001).

Conclusion: MRI-based diagnostic methods are valuable additions in detecting NAFLD or determining the severity of the NAFLD.

Figure 1

A flowchart of study screening and selection process.

Figure 2

A funnel plot showing the outcomes of a publication bias test.

Figure 3

A forest chart showing the overall and subgroup wise effect size of the area of operator receiver curve (AUROC) values obtained by the pooling of effect sizes reported in the individual studies.