Associations between MRI T1 mapping, liver stiffness, quantitative MRCP, and laboratory biomarkers in children and young adults with autoimmune liver disease

This single-center, prospective, cross-sectional, Health Insurance Portability and Accountability Act (HIPAA)-compliant pilot study was approved by the institutional review board at Cincinnati Children's Hospital Medical Center (CCHMC).

Patients up to 25 years of age with AILD (including AIH, PSC, and ASC) that were enrolled into an institutional AILD registry were included in this study. A group of pediatric hepatologists at CCHMC assigned the registry participants with a diagnosis of AIH, PSC, or ASC according to clinical guidelines [3, 19,20,21]. Written informed consent, and assent as appropriate, were obtained. Patients with any other form of liver disease were excluded from the registry.

MRI acquisition protocol

All registry participants underwent research MRI examinations of the liver at the time of registry enrollment. Subsequent research MRI examinations using the same protocol and scanner were performed 12 and 24 months later. The current study includes baseline MRI examinations acquired on a 1.5 T scanner (Ingenia; Philips Healthcare, Best, The Netherlands) using a 16-channel phased-array anterior (torso) surface coil as well as scanner table built-in spine coils (12-channels). The imaging protocol and parameters for the broader registry study have been previously published [2, 19]. Pertinent quantitative sequences for the current study include the following:

(1)

Three-dimensional (3D) fast spin-echo (FSE) MRCP using respiratory triggering; the belt used to detect breathing was placed over the upper abdomen.

(2)

Iron (T2*)-corrected T1 mapping (cT1) of the liver using a modified Look Locker (MOLLI) pulse sequence approach and complex chemical shift MRI [6].

(3)

Two-dimensional (2D) gradient recalled echo (GRE) MRE of the liver (active driver frequency = 60 Hz), with the passive driver placed over the right upper quadrant of the abdomen and secured in place with a Velcro strap.

Clinical and laboratory data collection

Pertinent demographic and clinical data were recorded from the AILD registry. The following clinical data were recorded: patient age, sex, clinical diagnosis (AIH, PSC, or ASC), time between diagnosis and research MRI (in months), and presence of inflammatory bowel disease (IBD). The following laboratory measurements also were recorded from the time of the research MRI: total bilirubin, alanine aminotransferase (ALT), aspartate aminotransferase (AST), gamma-glutamyl transferase (GGT), platelets, fibrosis-4 (FIB4) score, AST to platelet ratio index (APRI), alkaline phosphatase (ALP), and matrix metalloproteinase 7 (MMP7), a novel marker of bile duct injury and liver fibrosis in AILD [22]. Methods for calculating FIB4 score and APRI have been previously published [23].

MR image post-processing

Perspectum Ltd. (Oxford, United Kingdom) provided image post-processing tools and analyses for quantitative 3D MRCP metrics and iron-corrected T1 (cT1) relaxation measurements at no cost through a formal research agreement. Their analysts were blinded to all patient information, including clinical diagnosis as well as laboratory, clinical, and other imaging data.

3D MRCP

3D MRCP images were processed by a single operator at Perspectum using the MRCP + tool. This operator has more than 5 years of experience using MRCP + . To achieve this, bile ducts were identified through a series of signal enhancement and thresholding steps to detect tubular structures and generate a 3D model of the biliary tree [19]. The following metrics were then quantified: biliary tree volume; maximum and median diameters of the common bile duct, left hepatic bile duct, and right hepatic bile duct; total number of ducts in the 3D biliary tree model, defined as the total number of branches in the modeled tree; total length of ducts; total number of strictures, defined as local minima that were more than 30% narrower than neighboring maxima, total length of duct strictures; total number of dilations, defined as local maxima that were more than 30% wider than neighboring minima; and total length of duct dilations [19]. Diameters and lengths were reported in millimeters (mm).

cT1

Region-of-interest (ROI) and whole liver cT1 values were provided by Perspectum using LiverMultiScan [6, 24, 25], with values derived by a single observer with more than 5 years of experience using LiverMultiScan. For ROI measurements, three equally sized circular ROIs, with a diameter of 15 mm, were drawn in the right hepatic lobe of the liver on one representative axial slice, while avoiding major blood vessels and bile ducts. Voxel-by-voxel cT1 values were also calculated for the whole liver on four representative axial slices. The mean cT1 for the three circular ROIs and the mean, median, and interquartile range (IQR) of whole liver cT1 measurements across the four axial slices were calculated.

2D GRE MRE

MREplus + prototype software (Resoundant Inc., Rochester, Minnesota) was utilized to measure liver stiffness values from MRE data. Under the supervision of an image analyst in the Department of Radiology who was blinded to clinical data and other imaging data, ROIs were automatically placed on elastograms with 90% confidence threshold masks across four representative axial slices. The weighted mean of the mean liver shear stiffness values (in kPa) across the four axial slices was calculated for each patient.

Representative images of the quantitative MRI sequences assessed in this study are provided in Fig. 1.

Fig. 1figure 1

13-year-old male with autoimmune sclerosing cholangitis. a Maximum intensity projection 3D MRCP image. b Corresponding 3D biliary tree model extracted from 3D MRCP image using MRCP + . c MR elastogram of the liver (units of kPa). d Iron-corrected T1 (cT1) map of the liver (units of ms)

Statistical analysis

The patient cohort was divided into two groups, (1) AIH and (2) PSC/ASC. Patients with PSC/ASC were grouped into one cohort due to similarities in MRCP findings and clinical outcomes [3, 19]. Continuous variables were summarized as median, first quartile, and third quartile values. Variables that were not normally distributed were log transformed using base10 for analyses (Supplementary Table 1). Group differences in age, MRE liver stiffness, cT1, quantitative MRCP metrics, and laboratory values between the AIH and PSC/ASC groups were examined using the Mann–Whitney U test. Categorical variables were presented as frequency counts and percentages; group differences were determined using Chi-square/Fisher’s exact test, as applicable.

Five variables were considered primary outcome variables: liver stiffness, ROI-based mean cT1, whole liver mean cT1, whole liver median cT1, and whole liver cT1 IQR. The following variables were considered potential predictor variables: demographic data, quantitative MRCP metrics, and serum laboratory values. Univariate associations between outcomes and predictor variables for the entire study cohort and for the AIH and PSC/ASC groups individually were assessed using Pearson correlation coefficients (r). Relationships between three individual outcome variables (liver stiffness, whole liver mean cT1, and whole liver cT1 IQR) and potential predictor variables were further examined using multiple linear regression models, adjusting for patient age, sex, the presence of IBD, specific diagnosis (AIH vs. PSC/ASC), and time from diagnosis to research MRI examination. Stratified multiple linear regression models by diagnosis (AIH vs PSC/ASC) were also run, while also adjusting for patient age, sex, presence of IBD, and time from diagnosis to research MRI examination.

p values were not adjusted for multiple comparisons as this was an exploratory assessment. p less than 0.05 was considered statistically significant. All statistical testing was carried out on SAS version 9.4 (SAS Institute, Cary, North Carolina).

留言 (0)

沒有登入
gif