Quantification of Hepatic Steatosis with MRI Quantification of
Transcripción
Quantification of Hepatic Steatosis with MRI Quantification of
Quantification of Hepatic Steatosis with MRI Scott B. Reeder, MD, PhD International Society for Magnetic Resonance in Medicine Sociedad Mexicana de Radiologia e Imagen (SMRI) No se puede mostrar la imagen. Puede que su equipo no tenga suficiente memoria para abrir la imagen o que ésta esté dañada. Reinicie el equipo y , a continuación, abra el archiv o de nuev o. Si sigue apareciendo la x roja, puede que tenga que borrar la imagen e insertarla de nuev o. Mexico City June 5, 2014 100% 0% Department of Radiology University of Wisconsin Madison, WI Disclosure UW receives support from GE and Bracco Investigational Pulse Sequences Learning Objectives • Understand the importance and prevalence of Fatty liver disease • Understand the basic requirements of quantitative imaging biomarkers • Be familiar with emerging quantitative biomarkers of liver fat using MRI Case: 61yo obese female • Obese, type II diabetes • No known liver disease, No EtOH • Presents with cryptogenic cirrhosis • Develops HCC 1 year after presentation • Necessitated liver transplant Presumed Etiology: Non-Alcoholic Fatty Liver Disease Non-Alcoholic Fatty Liver Disease Non-Alcoholic Fatty Liver Disease (NAFLD) • First described by Ludwig et al (Mayo Clin Proc 1980) • Most common cause of chronic liver disease – 30% of people in the USA (100 million) have fatty liver disease (Harrison et al, ClinLivDis 2004) – 10% of all children have fatty liver disease (Schwimmer et al, Semin Liver Dis 2007) • Fatty liver can progress to injury and scarring, leading to – Cirrhosis – Liver failure – Hepatocellular carcinoma (HCC) • Fatty Liver Disease: a feature of the “Metabolic Syndrome” – Obesity, Diabetes (type II) – Increasing cause of cancer, cardiovascular disease, ? Diabetes type II – Underlying etiology: Insulin Resistance Does Fat Matter? • Elevated risk of cirrhosis, liver failure, HCC • NAFLD is an independent risk factor for cardiovascular disease – Schindhelm et al 2007 Curr Diab Rep • Steatosis may be causative of DM type II - Ekstedt et al 2006 Hepatology • 6-fold increase in malignancy - Rubenstein et al Hepatology 2008 Yes, Liver Fat Matters … NAFLD: Pathophysiology “2nd Hit” Metabolic Syndrome Insulin Resistance Fatty Liver (steatosis) ? Iron ? Genetic ? Infection Fibrosis (NASH) Cirrhosis & Liver Failure Improving insulin sensitivity reduces steatosis Treatment: • Exercise • Caloric restriction • Drugs • Surgery Treatment Early Diagnosis with MRI Normal Liver Intracellular accumulation of triglycerides in hepatocytes Quantitative biomarkers are needed for … detection of disease staging of disease treatment monitoring Fatty Liver Disease: Terminology NAFLD - Isolated Steatosis NASH NASH is aggressive subset of NAFLD NASH has worse prognosis, but isolated steatosis is not “benign” or “simple” -Steatosis -Ballooning degeneration -Inflammation -Fibrosis Key Clinical Issues: • Identification of NASH • Risk factors for progression to NASH Non-Alcoholic Fatty Liver Disease At the University of Wisconsin, fatty liver disease is … - 2nd leading cause of liver failure - 3rd leading indication for liver transplantation - Expected to be the leading cause of liver failure in a decade Prevalence of Childhood Obesity in the USA, 1971-2006 Current Diagnosis of Fatty Liver: Biopsy • Expensive • Risky (death 1:10000, hospitalization 1:1000) • Sampling errors - Liver disease is patchy - Sampling 1/50,000th of the organ inherently flawed - Agreement between two adjacent biopsies = 62% (Ratziu et al 2005) Same Patient Biopsy 1 Biopsy 2 • Utility of biopsy very limited in children - Missed or delayed diagnosis Histology images courtesy Claude Sirlin, MD Detection of Fat with MRI: In and Opposed Phase (2-point Dixon) Opposed-Phase (W-F) In-Phase (W+F) Two Point Dixon Imaging IP = 4.6ms (at 1.5T) Signal IP IP OP IP OP IP OP Echo Time (TE) OP = 2.3ms (at 1.5T) IP OP Patterns of Steatosis Patient 2 Diffuse Diffuse with mass like sparing Patient 3 In-Phase Geographic Patient 4 Patient 1 Opposed-Phase Lobar … Others … Hamer et al Radiographics 2006 Fat Quantification: Why Quantitative? • Qualitative methods limited • Quantification needed for thresholds - Thresholds for early diagnosis - Different thresholds for different applications - eg. screening vs diagnosis • Treatment monitoring • Surrogate biomarkers in drugs studies • Standardization - across protocols, platforms, sites - widespread clinically applicability requires standardization - Standardization requires a quantitative biomarker Most biomarkers in medicine are quantitative – why not MRI ? Classes of Fat Quantification Methods 1. With/without fat suppression - eg. compare T2 without and out fat saturation 2. “Magnitude MRI” (MRI-M) - Two or more magnitude images acquired in/opposed phase 3. “Complex MRI” (MRI-C) - Chemical shift based water-fat separation from complex source images Imaging Methods for Quantifying NAFLD add… Water Fat Measured signal Chemical Shift Based Fat-Water Separation Water Fat 100% Fat Fraction Fat-Fraction independent of coil sensitivity 0% Opposed-Phase OP = W-F In-Phase IP = W+F IP - OP) F ( h= = W+F 2 IP 50% 35% 0% IOP Fat Signal Fraction Map Quantitative Biomarkers of Steatosis Confounding Sources of Bias • Quantitative MRI biomarker for fat requires consideration of … – – – – – T1 bias T2* decay Multiple fat peaks Noise bias Eddy Currents MRI-M MRI-C MRI-C has more Potential Sources of Bias, but has Larger Dynamic Range: 0-100% Fat Fraction Sources of Bias: Noise • Using magnitude images to calculate fat-fraction – Leads to bias from non-zero noise at low fat-fractions Magnitude discrimination (Liu et al, 2007) η = 1-W/(W+F) Genetic ? Severe Hemochromatosis Steatosis ? Opposed In Phase Phase? TE=4.8ms Opposed In-Phase? Phase TE=2.4ms For IOP imaging, fat and iron have opposite effects! Simultaneous Estimation: T2*, Water, Fat • • • • • Combined T2* into signal model Yu et al JMRI 2007 (MRI-C) Bydder et al MRI 2008 (MRI-M) O’Regan 2009 Radiology (MRI-C) Permits simultaneous calculation of water, fat and T2* Water Fat R2* Yu et al, MRM 2007 Sources of Bias: Multiple Peaks of Fat • Many metabolites have more than one spectral peak – Fat has multiple spectral peaks, several near water – Leads to incomplete separation of water and fat – Source of “gray” fat on many fat suppression methods 210Hz Water -47Hz 23Hz 117Hz 159Hz 236Hz Sources of Bias: Multiple Peaks of Fat • “Spectral modeling” – Must know the relative amplitudes a priori – Frequencies and relative amplitudes constant (Hamilton et al) – No additional information required – Bydder et al 2008 – Yu et al 2008 T1 Weighted IDEAL-FSE (Yu et al) Same Window/Level Water With spectral modeling Confounder-Corrected MRI: MRI-C vs MRS r2=0.91 slope=0.66 ± 0.003, p<10-17 intercept=2.2% ± 0.3%, p<10-14 r2=0.86 slope=0.91 ± 0.05, p=0.08 Intercept=4.6% ± 0.5%, p<10-14 No MP No T2* With MP No T2* r2=0.76 slope=0.71 ± 0.05, p=10-6 intercept=-0.3% ± 0.5%, p=0.50 No MP With T2* r2=0.99 slope=1.00 ± 0.01, p=0.77 Intercept=0.2% ± 0.1%, p=0.19 With MP With T2* Confounder-Corrected MRI: MRI-C vs MRS MRI PDFF vs. Spectroscopy PDFF 35.0 30.0 MRI PDFF 25.0 20.0 15.0 y = 0.9853x + 0.5933 R² = 0.97639 10.0 5.0 0.0 0.0 5.0 10.0 15.0 20.0 25.0 30.0 35.0 Spectroscopy PDFF Siemens 3.0T widebore Duke University Fananapazir et al ISMRM 2013 Confounder-Corrected MRI: MRI-C vs MRS • • • • Three sites 7 magnets 1.5T, 3T Two vendors Data courtesy Claude Sirlin, MD Different Methods to Measure Unconfounded Fat Fraction No se puede mostrar la imagen. Puede que su equipo no tenga suficiente memoria para abrir la imagen o que ésta esté dañada. Reinicie el equipo y , a continuación, abra el archiv o de nuev o. Si sigue apareciendo la x roja, puede que tenga que borrar la imagen e insertarla de nuev o. No se puede mostrar la imagen. Puede que su equipo no tenga suficiente memoria para abrir la imagen o que ésta esté dañada. Reinicie el equipo y , a continuación, abra el archiv o de nuev o. Si sigue apareciendo la x roja, puede que tenga que borrar la imagen e insertarla de nuev o. Images courtesy Claude Sirlin, MD Different Methods to Measure Unconfounded Fat Fraction Images courtesy Claude Sirlin, MD Results: MRI vs. Biopsy 20 MRI-C = 0.32 * Biopsy + 1.5 R² = 0.91 MRI (%) 15 10 5 0 0% -5 15% 30% 45% Biopsy 60% 75% Results: MRI vs Biopsy Biopsy sample 4 regions - Aperio Quantification Image pro plus Segmentation Image pro plus Results: Biopsy quantification 20 R2 = 0.99 y = 0.984x + 0.629 R² = 0.986 slope = 0.98 ± 0.03 intercept = 0.63 ± 0.25 MRI (%) 15 10 5 0 0 -5 5 10 15 Biopsy Quantification (%) 20 Differentiate grade 2 vs 3 steatosis with 93% sensitivity, 85% specificity (cutoff of PDFF of 15%). AUC = 0.97 (no fibrosis), 0.861 (with fibrosis) n=86 Idilman et al Radiology 2013 AUC 0.989 for 0 vs >=1 0.825 for 0,1 vs 2, 3 0.892 for 0, 1, 2 vs 3 n=77 Tang et al Radiology 2013 PDFF Study of Health in Pomerania Jens Kühn, MD University of Greifswald 40 46 37 42 n=2,814 40 42 40 39 Med. PDFF = 4% (0-46%) mild moderate high = 27.7% (n=779) = 10.6% (n=298) = 1.7% (n= 49) 44 40 40% Hepatic Steatosis 41 yo M with h/o diabetes and Dyslipidemia: Triglyceridemia (TG=1022) 100% 31% 51% 35% 55% 10/09/09 (baseline) 0% 12/15/09 After Treatment Note decrease in fat-fraction and decrease in the size of the liver Case: 46yo M elevated AST/ALT, 36 lbs weight gain (4 years), presumed NAFLD Clinical Question - NASH vs Steatosis? Opposed-Phase In-Phase Case: 51yo F Acute Alcoholic Steatohepatitis T2W SSFSE HU = 16 (non-con) CT Hepatomegaly Steatosis In-phase Opposed Phase Case: 31 yo man with family Hx of hemochromatosis, elevated ferritin. MRI ordered to r/o iron overload Opposed Phase In Phase Conventional IOP Imaging Signal dropout on opposed phase imaging consistent with steatosis only 250 s-1 100% Complex MRI Severe steatosis: PDFF = 28% (normal < 5-6%) 28% 90s-1 0 s-1 0% Fat-Fraction Map Mild iron overload: R2*=90s-1 (normal < 50-60s-1) R2* Map Diagnosis: NAFLD and hemochromatosis (Iron overload missed on IOP imaging) Case: 31 yo man with family Hx of hemochromatosis, elevated ferritin. MRI ordered to r/o iron overload H&E (4x) Perl’s Blue (10x) Case: 10 yo boy with abdominal pain T2 with Fat-Sat Out of Phase Coronal T2 SSFSE In Phase Case: 10 yo boy with abdominal pain 10kPa 100% 200 s-1 22% 0% 40ms 0 s-1 Biopsy: severe steatohepatitis with severe fibrosis, just short of cirrhosis Case: 10 yo male with abdominal pain Grade 2 steatosis, Grade 3 steatohepatitis, stage 3-4 fibrosis Findings and clinical history consistent with NASH Case: 16 yo girl with PCOS 3.6kPa 100% 200s-1 31% 0% 20ms 0s-1 Biopsy: severe steatohepatitis with mild fibrosis Study: Adolescent Girls 3T 35% Mean Fat Fraction 30% r2 = 0.98 Slope: 1.03 (p<0.001) Intercept: 0.73 (p=0.01) 25% 20% 15% 10% 5% 0% 0% 5% 10% 15% 20% 25% 30% 35% MRS Spectroscopy Fat-Fraction (%) Rehm et al ISMRM 2011 Study: Adolescent Girls 100% 3% 32% 17% 0% kg/m2 BMI =35 ALT = 19 U/L Insulin = 25 μU/mL kg/m2 BMI =31 ALT = 48 U/L Insulin = 46 μU/mL kg/m2 BMI =30 ALT = 29 U/L Insulin = 71 μU/mL Good correlation with ALT Moderate correlation with insulin Poor correlation of liver fat with BMI Rehm et al ISMRM 2011 Learning Objectives • Understand the importance and prevalence of Fatty liver disease • Understand the basic requirements of quantitative imaging biomarkers • Be familiar with emerging quantitative biomarkers of liver fat using MRI Thank you! • • • • • • • • • • • • Rashmi Agni, MD Jean Brittain, PhD Diego Hernando, PhD Cathy Hines, PhD Jen Kuhn, MD Ben Landgraf Alejandro Roldan, PhD Claude Sirlin, MD Oliver Wieben, MD Emily Winslow, MD Huanzhou Yu, PhD Jennifer Rehm, MD Grant Support • WARF Accelerator • NIH: RO1 DK083380 RO1 DK088925 RC1 EB010384 RO1 DK096169