Test-retest repeatability of MR elastography for noninvasive liver fibrosis assessment in hepatitis C

Abstract

Purpose: To conduct a rigorous evaluation of the repeatability of liver stiffness assessed by MR elastography (MRE) in healthy and hepatitis-C-infected subjects.

Materials and methods: A biopsy-correlated repeatability study using four-slice MRE was conducted in five healthy and four HCV-infected subjects. Subjects were scanned twice on day 1 and after 7-14 days. Each slice was acquired during a 14-s breath-hold with a commercially available acquisition technique (MR-Touch, GE Healthcare). Results were analyzed by two independent analysts.

Results: The intraclass correlation coefficient (ICC) was 0.85 (90% confidence interval [CI]: 0.71 to 0.98) for the between-scan average of maximum stiffness within each slice and 0.88 (90% CI: 0.78 to 0.99) for the average of mean stiffness within each slice for the primary analyst. For both analysts, the average of the mean liver stiffness within each slice was highly reproducible with ICC of 0.93 and 0.94. Within-subject coefficients of variation ranged from 6.07% to 10.78% for HCV+ and healthy subjects.

Conclusion: MRE is a highly reproducible modality for assessing liver stiffness in HCV patients and healthy subjects and can discriminate between moderate fibrosis and healthy liver. MRE is a promising modality for noninvasive assessment of liver fibrosis (CLINICALTRIALS.GOV IDENTIFIER: NCT00896233).

Figure 1

Figure 1A. Axial magnitude (M), wave (W), and elastogram (E) images of A. a healthy 60-year-old female (M001) and B. an HCV-positive cirrhotic 60-year-old female (M010) from the fourth slice acquired across four timepoints, T1 – T4. The outlined areas represent “individual” regions of interest selected to exclude areas of interference and portal areas/blood vessels. Stiffness values, standard deviations, and number of pixels (n) in each ROI for each timepoint in A. are: T1: 1.84 ± 0.31 kPa, n = 2692 T2: 1.87 ± 0.40 kPa n = 3093 T3: 1.84 ± 0.28 kPa, n = 2455 T4: 2.00 ± 0.30 kPa, n = 3522 Figure 1B. Axial magnitude (M), wave (W), and elastogram (E) images of A. a healthy 60-year-old female (M001) and B. an HCV-positive cirrhotic 60-year-old female (M010) from the fourth slice acquired across four timepoints, T1 – T4. The outlined areas represent “individual” regions of interest selected to exclude areas of interference and portal areas/blood vessels. Stiffness values, standard deviations, and number of pixels (n) in each ROI for each timepoint in B. are: T1: 5.85 ± 1.16 kPa, n = 1802 T2: 5.49 ± 0.89 kPa n = 3232 T3: 5.21 ± 0.93 kPa, n = 2475 T4: 5.12 ± 0.77 kPa, n = 1472

Figure 2

MRE-derived mean liver stiffness (kPa) per subject as assessed by 2 independent analysts, designated as A1 (circles) and A2 (squares). “Common” ROI were analyzed across 4 timepoints, T1 – T4 as labelled. Subjects 1 – 5 were healthy. Subjects 6 – 10 were HCV+ with METAVIR F4 (6, 8, and 10) and F2 (8). Subject 7 discontinued from the study and data are not provided.

Figure 3

Display of stiffness values normalized to pixel count by timepoint (T1 – T4) for a healthy 60-year-old female (M001) and an HCV-positive 60-year-old female (M010) with fibrosis stage F4.

Figure 4

Hepatic fat fraction from 2-point Dixon A. healthy 60-year-old female (1001) with mean fat fraction of -1.8 ± 0.46% (essentially 0), and and B. HCV-positive 60-year-old female (1010) with mean fat fraction of 19 ± 0.35%.