Accuracy of Magnetic Resonance Imaging in Diagnosis of Liver Iron Overload: A Systematic Review and Meta-analysis  Maria Sarigianni, Aris Liakos, Efthymia Vlachaki, Paschalis Paschos, Eleni Athanasiadou, Victor M. Montori, Mohammad Hassan Murad, Apostolos Tsapas  Clinical Gastroenterology and Hepatology  Volume 13, Issue 1, Pages 55-63.e5 (January 2015) DOI: 10.1016/j.cgh.2014.05.027 Copyright © 2015 AGA Institute Terms and Conditions

Figure 1 Flow diagram of study selection process. Clinical Gastroenterology and Hepatology 2015 13, 55-63.e5DOI: (10.1016/j.cgh.2014.05.027) Copyright © 2015 AGA Institute Terms and Conditions

Figure 2 Graphic display of Quality Assessment of Diagnostic Accuracy Studies 2 (QUADAS-2) results. Methodologic design and reporting quality of studies included in the meta-analysis according to risk of bias and applicability concerns using the QUADAS-2 tool. Judgments about each methodologic quality item are presented as percentages across all included studies. Clinical Gastroenterology and Hepatology 2015 13, 55-63.e5DOI: (10.1016/j.cgh.2014.05.027) Copyright © 2015 AGA Institute Terms and Conditions

Figure 3 Forest plot of individual study estimates of sensitivity and specificity. Accuracy of MRI for estimating liver iron overload in patients with hereditary hemochromatosis or transfusional overload (thalassemia, sickle cell disease, myelodysplastic syndrome). Only studies reporting results for greater than 7 mg Fe/g dry liver weight biopsy threshold were included. CI, confidence interval; FN, false negative; FP, false positive; TN, true negative; TP, true positive. Clinical Gastroenterology and Hepatology 2015 13, 55-63.e5DOI: (10.1016/j.cgh.2014.05.027) Copyright © 2015 AGA Institute Terms and Conditions

Figure 4 Hierarchical SROC (HSROC) curve of sensitivity vs specificity. Performance of MRI for estimating liver iron concentration in patients with hereditary hemochromatosis or transfusional overload (thalassemia, sickle cell disease, myelodysplastic syndrome). Only studies reporting results for greater than 7 mg Fe/g dry liver weight biopsy threshold were included. Each circle represents a study, with the size being proportional to the study size. The solid curve represents the HSROC curve. Clinical Gastroenterology and Hepatology 2015 13, 55-63.e5DOI: (10.1016/j.cgh.2014.05.027) Copyright © 2015 AGA Institute Terms and Conditions

Figure 5 Conditional probability plots to explore the clinical utility of MRI. Clinical utility of (A) T2 SE and (B) T2* GRE sequences for detecting patients with liver iron overload (>7 mg Fe/g dry liver weight) in all possible pretest (prior) probabilities. CI, confidence interval. Clinical Gastroenterology and Hepatology 2015 13, 55-63.e5DOI: (10.1016/j.cgh.2014.05.027) Copyright © 2015 AGA Institute Terms and Conditions

Supplementary Figure 1 Hierarchical SROC (HSROC) curve of sensitivity vs specificity for different MRI modalities. Performance of (A) T2 SE and (B) T2* GRE sequences for estimating the liver iron concentration in patients with hereditary hemochromatosis or transfusional overload (thalassemia, sickle cell disease, myelodysplastic syndrome). Each circle represents a study. The solid curve and respective solid points represent the HSROC curve and summary estimates of test performance, respectively. The dashed outlines surrounding them represent the 95% confidence region of these summary estimates. Clinical Gastroenterology and Hepatology 2015 13, 55-63.e5DOI: (10.1016/j.cgh.2014.05.027) Copyright © 2015 AGA Institute Terms and Conditions

Supplementary Figure 2 Hierarchical SROC (HSROC) curve of sensitivity vs specificity for different liver iron concentration thresholds. Performance of MRI for estimating the liver iron concentration in patients with hereditary hemochromatosis or transfusional overload (thalassemia, sickle cell disease, myelodysplastic syndrome) in studies reporting results with a biopsy positivity threshold of (A) 2 and (B) 15 mg Fe/g dry liver weight. Each circle represents a study, with the size being proportional to the study size. The solid curve represents the HSROC curve. Clinical Gastroenterology and Hepatology 2015 13, 55-63.e5DOI: (10.1016/j.cgh.2014.05.027) Copyright © 2015 AGA Institute Terms and Conditions

Supplementary Figure 3 Deeks’ funnel plot23 and asymmetry test for the detection of publication bias. Linear regression of the logistic diagnostic odds ratios against the inverse root of effective sample sizes (ESS). Clinical Gastroenterology and Hepatology 2015 13, 55-63.e5DOI: (10.1016/j.cgh.2014.05.027) Copyright © 2015 AGA Institute Terms and Conditions