Abstract
Background:
Magnetic resonance elastography (MRE) can determine presence and stage of liver fibrosis. Data on normative MRE values, while reported in adults, are limited in children.
Purpose:
To determine distribution of MRE-measured liver stiffness in children without liver disease.
Study Type:
Prospective, observational
Population:
81 healthy children, (mean 12.6 ± 2.6 years, range 8–17 years)
Field Strength/Sequence:
3.0 Tesla Signa HDxt, General Electric MR Scanner; 2D GRE MRE sequence
Assessment:
History, examination, laboratory evaluation, and (MR) exams (PDFF and MRE) were performed. MR elastograms were analyzed manually at two reading centers and compared with each other for agreement and with published values in healthy adults and thresholds for fibrosis in adult and pediatric patients.
Statistical Tests:
Descriptive statistics, Bland Altman analysis, t-test to compare hepatic stiffness values with reference standards.
Results:
Stiffness values obtained at both reading centers were similar without significant bias (p=0.362) and with excellent correlation (ICC =0.782). Mean hepatic stiffness value for the study population was 2.45 ± 0.35 kPa (95th percentile 3.19 kPa) which was significantly higher than reported values for healthy adult subjects (2.10 kPa ± 0.23 kPa, p < 0.001). 74–85% of subjects had stiffness measurements suggestive of no fibrosis.
Conclusion:
Mean liver stiffness measured with MRE in this cohort were significantly higher than reported in healthy adults. Despite rigorous screening, some healthy children had stiffness measurements suggestive of liver fibrosis using current published thresholds. Although MRE has the potential to provide noninvasive assessment in patients with suspected hepatic disease, further refinement of this technology will help advance its use as a diagnostic tool for evidence of fibrosis in pediatric populations.
Keywords: pediatric radiology, magnetic resonance imaging, hepatic shear stiffness, quantitative imaging biomarkers
INTRODUCTION
Non-invasive imaging techniques are emerging as methods for both diagnosis and monitoring of liver disease in children. Some of these techniques are designed to assess liver fibrosis, a key determinant of clinical outcome in patients with chronic liver disease.(1) Traditionally, liver biopsy and subsequent histologic evaluation have been used to determine fibrosis stage, yet drawbacks of this approach are well documented, including its invasive nature, sampling variability, and cost. (2–4) Given the importance of determining the presence and stage of liver fibrosis, there is a need for new techniques that can accurately diagnose, stratify, and monitor liver disease progression in pediatric patients.
Magnetic resonance elastography (MRE) is a magnetic resonance (MR) imaging technique that measures the stiffness of liver noninvasively by analyzing the propagation of shear waves transmitted into the abdomen. Studies in adults have demonstrated that MRE has high diagnostic performance for detecting and staging hepatic fibrosis.(5) Furthermore, when compared with other imaging modalities such as ultrasound-based elastography, MRE has demonstrated higher accuracy for identifying liver fibrosis in adults.(6) More recently, in the MAGNET study of children with nonalcoholic fatty liver disease (NAFLD), MRE values in children with known NAFLD showed good interreader agreement, good correlation between reading centers, good correlation with histologic fibrosis stage, and accuracy in detecting liver fibrosis.(7)
Although these early results are promising, published data on MRE in children are scarce. In particular, while normative MRE values have been reported in adults, (8–12) normative data in children are limited.(13) Age-specific normal values are essential for proper interpretation of any diagnostic tool, including liver MRE. As animal studies have demonstrated age-dependent differences in shear stiffness in various soft tissues (14,15), extrapolating hepatic shear stiffness values from studies done in adults to guide the diagnosis of liver fibrosis in children is inappropriate. Without a reliable pediatric standard, physicians risk both under and over diagnosis of liver fibrosis. The study aim therefore was to determine the distribution of MRE-measured liver stiffness in children without liver disease. Secondary aims included evaluating the effect of age within a pediatric range on liver stiffness measurements, and evaluating liver stiffness measurements in relation to reported MRE measured thresholds of fibrosis reported in the pediatric and adult literature.
MATERIALS AND METHODS
Study Cohort
Our study was a single-center, observational, cross-sectional, prospective study of children without liver disease. Children age 8 to 17 years were recruited in the County of San Diego between January 2013 – June 2015 from community health fairs, community health centers, and primary care practices. We excluded anyone with history of liver disease, alcohol use, or inability to undergo MRI (claustrophobia, metal or other implants, body circumference greater than the imaging chamber, weight exceeding scanner table limit). Enrolled subjects were then excluded if liver chemistry was elevated or if any screening test was positive for liver disease. Those remaining subjects who had MRI studies performed were excluded from the analysis if they had incomplete or unanalyzable imaging data, or if they were discovered by MRI to have hepatic steatosis (defined as proton density fat fraction (PDFF) ≥ 5%).(16,17) Because childhood obesity is common and these children have a higher risk for liver disease and thus are more likely to use technology to measure liver stiffness, we chose to include a large number of children with overweight or obesity in our study population.(18) The study protocol was approved by the Institutional Review Board of the study center. The parent or legal guardian of all study participants provided written informed consent. Research participants provided written assent.
Clinical Data Collection
Clinical data were obtained for each research subject at a fasting intake visit that included a history and physical exam performed by a physician investigator and basic laboratory studies. Birth date and sex were recorded. Parents and participants self-identified race and ethnicity. Weight and height were measured on a clinical scale and clinical stadiometer. Body mass index (BMI) was calculated as weight (kilograms) divided by height squared (meters). Fasting laboratory assays included: complete blood count, alanine aminotransferase (ALT), aspartate aminotransferase, gamma glutamyl transferase (GGT), total bilirubin, direct bilirubin, albumin, and alkaline phosphatase. Viral hepatitis was excluded by testing (Hepatitis B Virus surface antigen, Hepatitis C Virus antibody). Alcohol use was excluded by interview. MRI was then performed on research subjects not excluded for abnormal liver chemistries or history of liver disease.
Magnetic Resonance Examination
MR examinations were performed on a 3.0T scanner (Signa HDxt, General Electric Healthcare, Wakeshaw WI). To reduce confounding effects, research subjects fasted for a minimum of four hours prior to imaging.(19) Positioning on the scanning table was standardized; all patients were supine, feet first with a dielectric pad placed on the anterior body wall between the patient and an 8-element torso phase-array surface coil overlying the liver. MR exams included PDFF, R2*, and stiffness determinations as described below.
Determination of Hepatic PDFF and Iron Quantification
Image Acquisition
Hepatic PDFF was determined using a magnitude-based confounder-corrected chemical-shift-encoded acquisition and reconstruction technique. As previously described, the technique used a low (10 degree) flip angle relative to repetition time (150 ms) to minimize T1 bias while permitting quantification of both R2* relaxometry and fat water signal oscillation. Six gradient recalled echoes were collected at successive nominally out-of-phase and in-phase echo times to permit computation of T2*-corrected fat fraction.(20) R2* was performed to confirm the absence of iron overload.(21) One to two breathholds were required for whole-liver coverage. The scanner computer, using a 6-fat peak model to correct for fat spectral complexity, generated parametric PDFF and R2* maps automatically.(22, 23)
Image Analysis
Data were transferred offline for manual analysis by image analysts (each with ≥ 6 months of training under the supervision of radiology investigator (C.S.) at Reading Center One, where PDFF and R2* values were measured by placing one circular region of interest (ROI) with 1 cm radius manually in each of the hepatic Couinaud segments on the maps. Bile ducts, major vessels, liver edges, artifacts were carefully avoided. The average PDFF and R2* from the 9 ROIs were calculated by averaging mean values from individual ROIs. (23)
Determination of Hepatic Stiffness
Image Acquisition
Hepatic stiffness was measured by MRE using a previously described technique.(24,25) Briefly, a passive acoustic driver (Resoundant, Rochester MN) was placed against the lower right chest at the level of the xiphoid in the midclavicular line, and continuous vibrations of 60 Hz were applied. Subjects were provided verbal instructions along with exaggerated physical demonstrations of breath holds prior to acquisition of images, which were obtained on expiration. A 2D gradient-recalled-echo MRE sequence (parameters provided in Table 1) was performed with acquisition of four separate slices in the axial plane through the widest portion of the liver. Field of view ranged from 36×36 cm to 48×48 cm and was adjusted by the technologist to accommodate the body habitus of each subject.
Table 1.
MR Elastography Imaging Parameters
| Imaging Parameter | Value |
|---|---|
| Repetition time (ms) | 50 |
| Flip angle (degrees) | 30 |
| Echo time (ms) | 20.2 |
| Receive bandwidth (± kHz) | 31.25 |
| Motion encoding direction | Z (superior to inferior) |
| Number of phase offsets | 4 |
| Number of slices | 4 |
| Slice thickness (mm) | 10 |
| Interslice gap (mm) | 10 |
| Field of view (cm) | 36×36 cm to 48×48 cm |
| Reconstruction Matrix | 256 × 64 |
| Acceleration factor (anterior → posterior direction) | 2 |
| Number of breath holds | 4 |
| Duration of breath holds (sec) | 16 |
ms = milliseconds; kHz = kilohertz; mm = millimeter; cm = centimeter; sec = seconds
From the acquired magnitude and phase images, the MR scanner computer automatically generated wave images. Elastograms were generated depicting the spatial distribution of the shear stiffness (i.e., magnitude of the complex modulus, |G*|) in kilopascals.(23) Additionally, standard confidence masks for stiffness measurements were automatically generated by software using the multimodel direct inversion algorithm provided in commercial versions of MRE, with values ranging from zero to one based on correlation coefficients from regression analysis, with a value of one indicating a highly reliable hepatic stiffness estimate. Only pixels with confidence values >0.95 were utilized in the creation of confidence masks.
Image Analysis
Images were transferred offline for manual analysis by analysts at two different reading centers (with ≥ 6 months experience under the supervision of radiology investigator (C.S.) at Reading Center One, and by radiology investigator with 20 years of experience (K.G.) at Reading Center Two). Reading centers were blinded to the other center’s results and to the clinical history of the patient population. At both centers, the analysts used a custom software package (MRE Quant, Mayo Clinic; Rochester, MN) to manually draw ROIs on each magnitude image with reliable pixels, avoiding large blood vessels, bile ducts, and edges of the liver parenchyma. Regions of the liver with low magnitude signal to noise ratio defined as < 3 were excluded.(8) Wave images were then co-localized to the corresponding ROIs, which were further modified manually so that only areas with identified parallel propagating waves were included. ROIs were automatically propagated onto the elastograms. Hepatic stiffness measurements were calculated based on averaged values from the elastograms within the intersection of manually drawn ROIs and confidence masks.
Descriptive Statistics and Correlation Analysis
Demographics and descriptive statistics for the entire study population were reported. Categorical data were summarized as percentages and counts. One-way analysis of variance was used to calculate mean, median, interquartile ranges, and standard deviations of all continuous data. Success rate of MRI completion, defined as completion of MRE acquisition sequences without interruption between start and end of MRE acquisition, was determined.
To assess intercenter agreement for hepatic shear stiffness estimation, Bland-Altman analysis was performed with computation of bias (mean of the paired differences) and its significance, standard deviation (SD) of the differences, and 95% limits of agreement. Linear regression was performed to assess proportional bias. Intraclass correlation coefficient (ICC) was calculated for liver stiffness values between centers. Interpretation of ICC reliability was as follows: 0–0.39, poor; 0.40–0.59, fair; 0.60– 0.74, good; and 0.75–1.0, excellent.(27)
Comparison of Liver Stiffness Measurements with Reported Values in Healthy Subjects and Reported Thresholds of Hepatic Fibrosis
After assessing adequate intercenter agreement, liver stiffness scores from both reading centers were averaged. Distribution, mean, and 95th percentile of liver stiffness scores were calculated. Spearman’s rank order correlation analysis was then performed to determine any association between variables of interest and average liver stiffness measurements of subjects. A literature review was performed and studies were selected based upon study population (healthy participants) and employment of the current standard MRE technology. Five publications were found that met the criteria, and pooled mean hepatic stiffness scores from these publications was calculated.(8–12) Independent samples T-test was performed to evaluate for differences in mean liver stiffness between the pooled healthy population and our study population.
To evaluate cut-offs for fibrosis within the confines of limited pediatric data, we made comparisons to both adult and pediatric studies. In the adult literature, a recent pooled meta-analysis of adults proposed a cut off of ≥ 2.61 kPa for detecting any fibrosis (stage 0 vs stage 1–4) (5). In the pediatric literature, the MAGNET study used two reading centers with proposed cut-off values of 2.69 and 2.77 kPa respectively as optimal values for detecting any fibrosis (stage 0 vs stage 1–4). We averaged these two values, using ≥ 2.74 kPa as a proposed cut-off for detection of fibrosis.(7) Percent of children in our population correctly identified by these published cut-off values was calculated. Last, binary linear regression models were constructed to evaluate for predictor variables of abnormal liver stiffness using reported adult and pediatric thresholds and variables of interest.
Comparison of Shear Stiffness Values By Age of Study Participants
Children in the study population were divided into age groups using 13 years of age as the cut off between children and adolescents to allow for a similar number of subjects in each group. Liver stiffness values between age groups (children versus adolescents) were compared with calculation of means of the average liver stiffness measurement for each group. Bootstrapping methods were used to compare 95th percentile measurements of younger versus older children.
Statistical analysis was performed using IBM SPSS Statistics for Windows, Version 24.0 (Armonk, NY: IBM Corp) and R 3.5.0 (Vienna, Austria, R foundation for statistical computing). A p value < 0.05 was considered significant for all inference testing. Study sample size was based upon a power calculation (effect size 0.4, α 0.1, power 0.80) showing that a minimum of 77 participants were required to estimate the population distribution of MRE shear stiffness in children without liver disease.
RESULTS
Study Population Descriptive Statistics and Correlation between Reading Centers
There were 122 children screened for study participation; 41 were excluded due to possible liver disease (elevated AST, ALT or GGT (n=9), or elevated MRI-PDFF (n=32)). (Figure 1) All children had R2* values < 60 Hertz, excluding iron overload. Descriptive statistics for the study population are presented in Table 2. The mean (± SD) patient age in the analysis cohort at the time of MRI was 12.6 ± 2.6 years, mean ALT (± SD) was 15.9 ± 6.6 U/L, and mean MRI PDFF (± SD) was 2.8 ± 1.1 %. MRI success rate was 100%, with all 81 participants tolerating MRI acquisition. All MR examinations were determined to be complete by both reading centers.
Figure 1.
Study flow diagram demonstrating process or identification, screening, exclusion and inclusion of study participants.
Table 2.
Characteristics of Study Population
| Total Study Population | Age 8–12 years | Age 13–17 years | |
|---|---|---|---|
| Age in years (mean ± SD) | 12.6 ± 2.6 | 10.5 ± 1.6 | 14.7 ± 1.4 |
| Male (mean age in years ± SD) | 12.6 ± 2.5 | 10.6 ± 1.4 | 14.6 ± 1.5 |
| Female (mean age in years ± SD) | 12.6 ± 2.8 | 10.1 ± 1.8 | 14.8 ± 1.4 |
| Male (number, (%)) | 49 (60 %) | 26 (63 %) | 23 (57 %) |
| Female (number, (%)) | 32 (40 %) | 15 (37 %) | 17 (43 %) |
| Height (cm, mean ± SD (range)) | 157.3 ± 13.9 (119.0 – 183.2) | 148.5 ± 12.0 (119.0 – 170.3) | 166.4 ± 9.1 (150.1 – 183.2) |
| Weight (kg, mean ± SD (range)) | 63.5 ± 21.5 (21.5 – 139.5) | 56.3 ± 19.3 (21.5 – 94.0) | 71.0 ± 21.4 (38.0 – 139.5) |
| BMI (mean ± SD (range)) | 25.2 ± 6.6 (14.3 – 50.0) | 24.8 ± 5.9 (14.3 – 37.1) | 25.6 ± 7.3 (15.0 – 50.0) |
| BMI percentile (mean ± SD (range)) | 78.7 ± 29.4 (0.9 – 99.7) | 84.7 ± 23.7 (8.6 – 99.6) | 72.5 ± 33.4 (0.9 – 99.7) |
| BMI z score (mean ± SD (range)) | 1.2 ± 1.2 (−2.4 – 2.8) | 1.5 ± 1.0 (−1.4 – 2.7) | 0.9 ± 1.3 (−2.38 – 2.79) |
| AST (U/L, mean ± SD (range)) | 22.4 ± 5.7 (8 – 38) | 23.1 ± 5.6 (11 – 38) | 21.8 ± 5.8 (8 – 38) |
| ALT (U/L, mean ± SD (range)) | 15.9 ± 6.6 (6 – 35) | 16.2 ± 5.7 (9 – 34) | 15.5 ± 7.5 (6 – 35) |
| GGT (U/L, mean ± SD (range)) | 17.3 ± 6.3 (7 – 49) | 16.9 ± 4.7 (7 – 28) | 17.6 ± 7.7 (7 – 49) |
| PDFF (%, mean ± SD (range)) | 2.8 ± 1.1 (1.0 – 4.6) | 2.5 ± 0.4 (1.4 – 4.6) | 2.4 ± 0.3 (1.1 – 4.5) |
SD = standard deviation; % = percent; cm = centimeters; kg = kilograms, BMI = body mass index; AST = aspartate aminotransferase; ALT = alanine aminotransferase; GGT = gamma-glutamyl transferase; U/L = units/liter; PDFF = proton density fat fraction
Figure 2 displays a Bland-Altman plot comparing liver stiffness values from both reading centers. The difference in mean ROI size was statistically significant between Reading Centers One and Two (2254 vs 1232, p < 0.001), but this difference was not associated with differences in hepatic shear stiffness measurements (r = −0.05, p = 0.68). Liver stiffness values obtained at both reading centers were similar, with Reading Center One underestimating stiffness by 0.03kPa on average compared to Reading Center Two, but this bias was not significant (p = 0.362). Linear regression indicated no proportional bias (p = 0.065) between center readings. The ICC between reading centers was excellent (ICC =0.782) with 95% CI of 0.661 – 0.860.
Figure 2.
Bland Altman plot of hepatic shear stiffness values from Reading Center One and Reading Center Two in kilopascals (kPa). Bland Altman Bias (mean of the paired differences) and its p-value (p-value for paired t-test), standard deviation (SD) of the differences and limits of agreement (LOA) (LOA = Bias ±1.96*SD) are shown.
The distribution of averaged MR-determined liver stiffness scores is shown in Figure 3. Mean liver stiffness value for the study population was 2.45 ± 0.35 kPa (95th percentile 3.19 kPa). Spearman’s rank order correlation analysis between liver stiffness values and variables of interest is presented in Table 3. There was no significant correlation between liver stiffness and AST, ALT, GGT, BMI, sex, or age. Additionally, there was no significant difference in hepatic shear stiffness measurements when comparing children with and without obesity (p = 0.113).
Figure 3.
The distribution of average MR determined shear stiffness scores in kilopascals (kPa) for healthy children without liver disease.
Table 3.
Correlation of Clinical Parameters of Interest with Liver Stiffness in Children Free of Liver Disease
| Variable of Interest | Rank Correlation |
|---|---|
| BMI | r= −0.15; p= 0.169 |
| BMI percentile | r = −0.11; p = 0.314 |
| BMI z-score | r = −0.11; p = 0.311 |
| ALT (U/L) | r = 0.05; p = 0.638 |
| AST (U/L) | r = 0.15; p = 0.172 |
| GGT (U/L) | r = 0.22; p = 0.052 |
| Age (years) | r = −0.16; p = 0.160 |
| Sex | r = −0.086; p = 0.466 |
BMI = body mass index; ALT = alanine aminotransferase; AST = aspartate aminotransferase; GGT = gamma-glutamyl transferase; U/L = units/liter; r = Spearman rank order coefficient
Comparison of MR Liver Stiffness Measurements with Reported Values in Healthy Subjects and Reported Thresholds of Hepatic Fibrosis
Characteristics of studies in healthy adults using current standard MRE technology are presented in Table 4. These studies were pooled together to yield 154 adults free of liver disease with a mean liver stiffness of 2.10 ± 0.23 kPa. Compared to these healthy adults, mean liver stiffness in our population of children (2.45 ± 0.35 kPa) was significantly higher (p < 0.001).(8–12) Assuming no liver disease in our carefully screened study population, the correct classification rate as true negative was determined using adult and pediatric thresholds for liver fibrosis reported in the literature. Hsu et al. reported 2.61 kPa as an optimal threshold to distinguish between no fibrosis and stage 1 or greater fibrosis in adult populations, yielding a false-positive rate of 8.7%.(5) This fibrosis threshold when used in our study population correctly identified 74% of healthy subjects. Similarly, Schwimmer et al reported 2.74 kPa as an optimal threshold to distinguish between no fibrosis and stage 1 or greater fibrosis in children, yielding a false-positive rate of 6.1%.(7) This threshold when used in our study population correctly identified 85% of healthy subjects.
Table 4.
Characteristics of MRE Determined Shear Stiffness Studies in Adult Populations
| Study | Subjects | Population characteristics | Mean Age (SD) in years | MRI Field Strength and Sequence | Mean Hepatic Shear Stiffness in kilopascals ± SD (range) |
|---|---|---|---|---|---|
| Yin et al 2011 | n= 20 | Healthy adult volunteers with no history of liver disease or abnormal liver chemistries | 43.6 (17.3) | 1.5T, 2D GRE | 2.16 ± 0.24 (not reported) |
| Lee at al 2013 | n= 49 | Adult liver donors without history of liver disease, normal liver chemistries, no medications, no history of drug or alcohol abuse | 30.6 (9.6) | 1.5T, 2D GRE | 2.12 ± 0.04 (1.54 – 2.87) |
| Venkatesh et al 2014 | n=53 | Healthy Asian adult volunteers with no previous history of hospitalizations, personal or family history of liver disease, alcohol abuse, no medications | 41.8 (not reported) | 1.5T, 2D GRE | 2.09 ± 0.22 (1.1 – 3.08) |
| Rusak et al 2014 | n=20 | Healthy adults with normal appearing liver on ultrasound, with normal ALT, no risk factors for liver disease, no hospitalizations, no alcohol abuse, normal weight, no surgical history | 39.1 (not reported) | 1.5T, 2D GRE | 2.30 ± 0.32 (1.1 – 3.08) |
| Trout et al 2016 | n= 24 | Self-reported healthy adult volunteers | 38.8 (9) | 1.5T, 2D GRE | 1.95 ± 0.27 (1.52 – 2.68) |
SD = standard deviation; ALT = alanine aminotransferase; T = Tesla; 2D = two-dimensional; GRE = gradient recalled echo
Binary linear regression models were constructed to evaluate for variables that predicted liver stiffness above reported thresholds for detection of fibrosis. When using the pediatric threshold for detection of fibrosis, only GGT was a significant predictor with good fit (odds ratio 1.13, p =.024, CI 1.015 – 1.246). When using the adult threshold for detection of fibrosis, no variables were significant predictors.
Age Comparison of MR liver stiffness
Mean liver stiffness value was higher in children than in adolescents, however this difference was not statistically significant (children 2.51 ± 0.35 kPa vs. adolescents 2.38 ± 0.34 kPa, p = 0.08). Similarly, children had a higher 95th percentile estimate calculated by bootstrapping methods when compared to adolescents, however, the estimated 95th percentile measurements were overlapping (children 3.11–3.28 kPa vs. adolescents 2.70–3.35 kPa.) Box plots of the distribution of MRE-determined liver stiffness values in children and adolescents are shown in Figure 4. Box plots showing the distribution of age by MRE-determined stiffness values above or below thresholds for the detection of liver fibrosis are shown in Figure 5.
Figure 4.
Box plot of distribution of MR determined shear stiffness scores (in kilopascals (kPa)) in healthy children vs. adolescents.
Figure 5.
Box plot of distribution of MR determined liver stiffness above and below adult (a) and pediatric (b) fibrosis thresholds by age in years.
DISCUSSION
We performed MRE on 81 carefully phenotyped children who were recruited from the community and were free from liver disease by history, exam and laboratory analyses. All participants had normal fat fraction measured by MRI-PDFF. We demonstrated that MRE was well tolerated and yielded quality images with excellent correlation between reading centers. The mean liver stiffness in this population was 2.45 ± 0.35 kPa. Depending upon the threshold used, 74 – 85% of healthy children free from liver disease were correctly identified as true negatives using MRE determined liver stiffness.
We found that mean liver stiffness in healthy children was higher than reported values in healthy adults. However, there was no clear association of liver stiffness measurements with age in our study. Previous studies evaluating liver stiffness in healthy children have suggested a possible association between liver stiffness and age. Most pediatric data evaluating liver stiffness in healthy children to date have employed either transient elastography (TE) or acoustic radiation force impulse (ARFI) methodology. Three out of four studies that used TE in children reported that TE-measured liver stiffness values increased with age.(28–31) However, this association was not observed when using ARFI-determined liver stiffness, arguably a better comparison with MRE determined liver stiffness secondary to its higher diagnostic accuracy and ability for direct visualization of the liver region of interest being measured.(32) Two out of three studies of ARFI-determined liver stiffness in healthy children demonstrated no significant correlation in ARFI-determined liver stiffness score and age; the third study did not analyze correlation of liver stiffness with respect to age.(33–35) There has been only one previous study utilizing MRE-determined liver stiffness in healthy children (13), and there was no significant difference in stiffness based on age in that pediatric population, which is similar to our study result. Mean hepatic shear stiffness values in that population were 2.2 ± 0.3 in 13 children (5–13 years) and 2.2 ± 0.2 in 11 adolescents (14–18 years). These values were lower than those obtained in our study population, however the study did not employ the same MRE technology currently used in clinical practice and was limited by a small sample size.(13) Based on our data, age did not seem to significantly influence liver stiffness as assessed by MRE, and thus age does not adequately explain the higher stiffness in children compared to adults.
Up to 85% of children free from liver disease in our study population were correctly identified using published MRE determined thresholds for fibrosis (2.74 kPa in children, 2.61 kPa in adults). The subset of healthy children with liver stiffness values above these thresholds was an unexpected finding. Given the rigorous screening performed prior to obtaining MRE in this study population, it is unlikely that these children had any fibrosis. Whether stiffness values in this subset of our population were a result of true physiologic difference or due to technical variations in liver stiffness measurements is unclear. Multiple extrahepatic factors have been proposed as sources of variability in MRE shear stiffness measurements, one concern especially applicable to younger children is compliance with breath-holding techniques during image acquisition.(13,36,37) Recent studies comparing free breathing, respiratory navigated breathing, and breathholding (inspiratory and expiratory) have suggested that inability to comply with expiratory breathhold could result in nonlinear tissue elasticity during the respiratory cycle leading to increased variation in stiffness measurements. Although children were taught breathing techniques prior to image acquisition in our study, efficacy of expiratory breathholds was not measured. Therefore, evaluating stiffness measurements throughout the respiratory cycle may provide a more robust evaluation of the biomechanics of hepatic tissues in children.(36) Using current protocols the majority of children were correctly classified, however, further refinements will be needed to understand and close the remaining gaps in MRE hepatic stiffness measurements for liver fibrosis in children.
Strengths of this study are that it is the largest sample size to date of healthy subjects undergoing MRE acquisition. All subjects were carefully phenotyped with imaging, history, physical and laboratory values to exclude liver disease. Additionally, the MRE images were analyzed separately at two different experienced centers and demonstrated excellent correlation, suggesting that imaging performed at outside centers may produce similar results. One potential limitation is that the study population oversampled children with obesity, however comparison of liver stiffness values in children with and without obesity demonstrated no significant difference. Another potential limitation is that the quality of breathholding was not measured, which may have influenced shear stiffness values. Furthermore, because biopsies were not clinically indicated, stiffness values in this study population could not be compared with histology for validation of findings. Our study provides normative values using 2D GRE MRE. There are other forms of MRE including 2D SE-EPI MRE and 3D SE-EPI MRE. Establishing normative values using these other MRE methods will require further research.
In conclusion, our study confirms that MRE is well tolerated in children 8 years and older. The mean values of liver stiffness measured with MRE in this cohort were significantly higher than reported in adult studies on healthy volunteers. However, within the pediatric population, increase in age did not adequately explain stiffness variability. In addition, despite rigorous screening, some healthy children had stiffness measurements suggestive of liver fibrosis using current published thresholds. It is unclear whether the source of these elevated measurements should be attributed to a true physiologic difference between children and adults that remains to be investigated, or to a measurement error, suggesting the need for further scrutiny of MRE acquisition techniques in a pediatric population. Although MRE has the potential to provide noninvasive assessment in patients with suspected hepatic disease, further refinement of this technology will help advance its use as a diagnostic tool for evidence of fibrosis in pediatric populations.
Acknowledgments
Grant Support: National Institute of Health Grant ULTR000100, ULTR001442, EB001981, and EB017197. The content is solely the responsibility of the author and does not necessarily represent the official views of the NIH.
References
- 1.Altamirano-Barrera A, Barranco-Fragoso B, Méndez-Sánchez N. Management strategies for liver fibrosis. Ann Hepatol. 2017; 16(1): 48–56. doi: 10.5604/16652681.1226814. [DOI] [PubMed] [Google Scholar]
- 2.Regev A, Berho M, Jeffers LJ, et al. Sampling error and intraobserver variation in liver biopsy in patients with chronic HCV infection. Am J Gastroenterol. 2002; 97(10):2614–2618. doi: 10.1111/j.1572-0241.2002.06038. [DOI] [PubMed] [Google Scholar]
- 3.Ratziu V, Charlotte F, Heurtier A, et al. Sampling variability of liver biopsy in nonalcoholic fatty liver disease. Gastroenterology. 2005; 128:1898–1906. doi: 10.1053/j.gastro.2005.03.084. [DOI] [PubMed] [Google Scholar]
- 4.Bravo AA, Sheth SG, Chopra S. Liver Biopsy. N Engl J Med [Internet]. 2001; 344(7):495–500. [DOI] [PubMed] [Google Scholar]
- 5.Hsu C, Caussy C, Imajo K, et al. Magnetic Resonance vs Transient Elastography Analysis of Patients With Nonalcoholic Fatty Liver Disease: A Systematic Review and Pooled Analysis of Individual Participants. Clin Gastroenterol Hepatol [Internet]. 2018. June 14 [cited 2018 Oct 2]. doi: 10.1016/j.cgh.2018.05.059. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Yoon JH, Lee JM, Joo I, et al. Hepatic Fibrosis: Prospective Comparison of MR Elastography and US Shear Wave Elastography for Evaluation. Radiology 2014; 273(3): 772–782. doi: 10.1148/radiol.14132000. [DOI] [PubMed] [Google Scholar]
- 7.Schwimmer JB, Behling C, Angeles JE, et al. Magnetic resonance elastography measured shear stiffness as a biomarker of fibrosis in pediatric nonalcoholic fatty liver disease. Hepatology 2017; 66(5):1474–1485. doi: 10.1002/hep.29241. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Yin M, Talwalkar JA, Glaser KJ, et al. Dynamic Postprandial Hepatic Stiffness Augmentation Assessed With MR Elastography in Patients With Chronic Liver Disease. AJR. 2011; 197 (1):64–70. doi: 10.2214/AJR.10.5989. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Lee DH, Lee JM, Han JK, Choi BI. MR elastography of healthy liver parenchyma: Normal value and reliability of the liver stiffness value measurement. J Magn Reson Imaging. 2013;38(5):1215–1223. doi: 10.1002/jmri.23958. [DOI] [PubMed] [Google Scholar]
- 10.Venkatesh SK, Wang G, Teo LLS, Ang BWL. Magnetic resonance elastography of liver in healthy asians: Normal liver stiffness quantification and reproducibility assessment. J Magn Reson Imaging. 2014;39(1):1–8. doi: 10.1002/jmri.24084. [DOI] [PubMed] [Google Scholar]
- 11.Rusak G, Zawada E, Lemanowicz A, Serafin Z. Whole-organ and segmental stiffness measured with liver magnetic resonance elastography in healthy adults: significance of the region of interest. Abdom Imaging. 2014; 40(4):776–782. doi: 10.1007/s00261-014-0278-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Trout AT, Serai S, Mahley AD, et al. Liver Stiffness Measurements with MR Elastography: Agreement and Repeatability across Imaging Systems, Field Strengths, and Pulse Sequences. Radiology. 2016; 281(3):793–803. doi: 10.1148/radiol.2016160209. [DOI] [PubMed] [Google Scholar]
- 13.Etchell E, Jugé L, Hatt A, Sinkus R, Bilston LE. Liver Stiffness Values Are Lower in Pediatric Subjects than in Adults and Increase with Age: A Multifrequency MR Elastography Study. Radiology. 2017; 283 (1):222–230. doi: 10.1148/radiol.2016160252. [DOI] [PubMed] [Google Scholar]
- 14.Clarke EC, Cheng S, Bilston LE. The mechanical properties of neonatal rat spinal cord in vitro, and comparisons with adult. J Biomech. 2009; 42 (10):1397–1402. doi: 10.1016/j.jbiomech.2009.04.008. [DOI] [PubMed] [Google Scholar]
- 15.Alnaqeeb MA, Al Zaid NS, Goldspink G. Connective tissue changes and physical properties of developing and ageing skeletal muscle. J Anat. 1984; 139 ( Pt 4):677–689. [PMC free article] [PubMed] [Google Scholar]
- 16.Awai HI, Newton KP, Sirlin CB, Behling C, Schwimmer JB. Evidence and recommendations for imaging liver fat in children, based on systematic review. Clin Gastroenterol Hepatol. 2014; 12(5):765–773. doi: 10.1016/j.cgh.2013.09.050. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Cui J, Philo L, Nguyen P, et al. Sitagliptin vs. placebo for non-alcoholic fatty liver disease: A randomized controlled trial. J Hepatol. 2016; 65(2):369–376. doi: 10.1016/j.jhep.2016.04.021. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Ogden CL, Carroll MD, Lawman HG et al. Trends in Obesity Prevalence Among Children and Adolescents in the United States, 1988–1994 Through 2013–2014. JAMA. 2016; 15(21):2292–9. doi: 10.1001/jama2016.6361. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Zhang J, Arena C, Pednekar A, et al. Short-term repeatability of magnetic resonance elastography at 3.0T: effects of motion-encoding gradient direction, slice position, and meal ingestion. J Magn Reson Imaging. 2016;43(3):704–712. doi: 10.1002/jmri.25035. [DOI] [PubMed] [Google Scholar]
- 20.Yokoo T, Shiehmorteza M, Hamilton G, et al. Estimation of hepatic proton-density fat fraction by using MR imaging at 3.0 T. Radiology. 2011;258(3):749–759. doi: 10.1148/radiol.10100659. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Hernando D, Levin YS, Sirlin CB, et al. Quantification of Liver Iron with MRI: State of the Art and Remaining Challenges. J Magn Reson Imaging. 2014; 40(5):1003–1021. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Hamilton G, Yokoo T, Bydder M, et al. In vivo characterization of the liver fat 1H MR spectrum. NMR Biomed [Internet]. 2011; 24(7):784–790. doi: 10.1002/nbm.1622. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Mamidipalli A, Hamilton G, Manning P, et al. Cross-sectional correlation between hepatic R2* and proton density fat fraction in children with hepatic steatosis. J Magn Reson Imaging. 2018; 47(2):418–424. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Yin M, Talwalkar JA, Glaser KJ, et al. Assessment of Hepatic Fibrosis With Magnetic Resonance Elastography. Clin Gastroenterol Hepatol. 2007; 5(10):1207–1213. doi: 10.1016/j.cgh.2007.06.012 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Venkatesh SK, Yin M, Ehman RL. Magnetic resonance elastography of liver: Technique, analysis, and clinical applications. J Magn Reson Imaging. 2013; 37(3):544–555. doi: 10.1002/jmri.23731. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Venkatesh SK, Ehman RL. Magnetic resonance elastography of abdomen. Abdom Imaging. 2015;40(4):745–759. doi: 10.1007/s00261-014-0315-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Hallgren KA. Computing inter-rater reliability for observational data: an overview and tutorial. Tutor Quant Methods Psychol 2012;8(1):23–34. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Lewindon PJ, Balouch F, Pereira TN, et al. Transient liver elastography in unsedated control children: Impact of age and intercurrent illness. Journal of Paediatrics and Child Health 52 (2016) 637–642. [DOI] [PubMed] [Google Scholar]
- 29.Tokuhara CD, Cho Y, Shintaku H. Transient Elastography-Based Liver Stiffness Age-Dependently Increases in Children PLOS ONE. doi: 10.1371/journal.pone.0166683. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Engelmann G, Gebhardt C, Wenning D, et al. Feasibility study and control values of transient elastography in healthy children. Eur J Pediatr 2012; 171: 353–360. doi: 10.1007/s00431-011-1558-7. [DOI] [PubMed] [Google Scholar]
- 31.Goldschmidt I, Streckenbach C, Dingemann C, et al. Application and limitations of transient liver elastography in children. J Pediatr Gastroenterol Nutr 2013; 57: 109–113. doi: 10.1097/MPG.0b013e31829206a0. [DOI] [PubMed] [Google Scholar]
- 32.Cui J, Heba E, Hernandez C et al. MRE is superior to ARFI for the diagnosis of fibrosis in patients with biopsy- proven NAFLD: A prospective study. Hanquinet S, Hepatology (2016) 63(2):453–461. doi: 10.1002/hep.28337. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Hanquinet S, Courvoisier D, Kanavaki A, Dhouib A, Anooshiravani M. Acoustic radiation force impulse imaging— normal values of liver stiffness in healthy children. Pediatr Radiol (2013) 43:539–544. doi: 10.1007/s00247-012-2553-5. [DOI] [PubMed] [Google Scholar]
- 34.Lee MJ, Kim MJ, Han KH, Yoon CS. Age-related changes in liver, kidney, and spleen stiffness in healthy children measured with acoustic radiation force impulse imaging. European J of Radiology 82(2013)290–294. [DOI] [PubMed] [Google Scholar]
- 35.Noruegas MJ, Matos H, Goncalves I, et al. Acoustic radiation force impulse-imaging in the assessment of liver fibrosis in children. Pediatr Radiol (2011) 42:201–204. [DOI] [PubMed] [Google Scholar]
- 36.Wang K, Manning P, Szeverenyi N, et al. Repeatability and reproducibility of 2D and 3D hepatic MR elastography with rigid and flexible drivers at end-expiration and end-inspiration in healthy volunteers. Abdom Radiol. 2017; 42(12):2843–2854. doi: 10.1007/s00261-017-1206-4 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Murphy IG, Graves MJ, Reid S, et al. Comparison of breath-hold, respiratory navigated and free-breathing MR elastography of the liver. Magn Reson Imaging. 2017; 37:46–50. : 10.1016/j.mri.2016.10.011. [DOI] [PubMed] [Google Scholar]