Acoustic radiation force impulse imaging for assessing graft fibrosis after pediatric living donor liver transplantation: A pilot study

Authors


  • There is no financial support to report, and the authors of this article have no conflicts of interest to disclose.

Abstract

Graft fibrosis is a common finding during protocol biopsy examinations after pediatric liver transplantation. We evaluated the clinical utility of liver stiffness measurements by acoustic radiation force impulse (ARFI) imaging, a novel ultrasound-based elastography method, for assessing graft fibrosis after pediatric living donor liver transplantation (LDLT). We performed 73 liver stiffness measurements by ARFI imaging in 65 pediatric LDLT recipients through the upper midline of the abdomen (midline value) and the right intercostal space (intercostal value) around the time of protocol biopsy examinations. Fifty-nine of these liver stiffness measurements could be compared with histopathological findings. Graft fibrosis was assessed according to the degrees of portal and pericellular fibrosis. Significant fibrosis, which was defined as F2 or worse portal fibrosis and/or moderate or worse pericellular fibrosis, was observed in 14 examinations, which had significantly higher midline (P = 0.005) and intercostal values (P < 0.001) than the others. Liver stiffness measurements by ARFI imaging significantly increased with increases in the portal and pericellular fibrosis grades. For the diagnosis of significant fibrosis, the areas under the receiver operating characteristic curve (AUROCs) were 0.760 (P = 0.005) and 0.849 (P < 0.001) for the midline and intercostal values, respectively. The optimal cutoff values were 1.30 and 1.39 m/second for midline and intercostal values, respectively. Slight but significant elevations were noted in the results of biochemical liver tests: serum levels of γ-glutamyltransferase showed the highest AUROC (0.809, P = 0.001) with an optimal cutoff value of 20 IU/L. In conclusion, liver stiffness measurements by ARFI imaging had good accuracy for diagnosing graft fibrosis after pediatric LDLT. The pericellular pattern of fibrosis was frequently observed after pediatric LDLT, and moderate pericellular fibrosis was detectable by ARFI imaging. Liver Transpl 19:1202–1213, 2013. © 2013 AASLD.

Abbreviations
ACR

acute cellular rejection

ALT

alanine aminotransferase

ARFI

acoustic radiation force impulse

AST

aspartate aminotransferase

AUROC

area under the receiver operating characteristic curve

CI

confidence interval

GGT

γ-glutamyltransferase

ICC

intraclass correlation coefficient

LDLT

living donor liver transplantation

NPV

negative predictive value

PPV

positive predictive value

ROI

region of interest

TE

transient elastography

For more than 20 years, liver transplantation has been the standard therapy for children with life-threatening liver diseases, with 1- and 5-year survival rates greater than 90% and 85%, respectively.[1] Despite normal liver biochemistry results for most long-term survivors, abnormal graft histological findings, such as unexplained chronic hepatitis and graft fibrosis, have been commonly observed during late posttransplant protocol biopsy examinations.[2] In pediatric patients, for whom most transplants are performed for nonrecurring diseases, abnormal graft histological findings are considered indicative of immune-mediated damage.[2, 3] Therefore, protocol biopsy should be considered for liver transplant recipients for determining the immunosuppressant regimen.[4]

Hepatic fibrosis occurs in many chronic liver diseases and is considered the result of the wound-healing response of the liver to repeated injury.[5] Liver biopsy is still the gold-standard method for detecting changes in liver fibrosis; however, it is invasive and is associated with patient discomfort and, rarely, with serious complications.[6] Moreover, the size of the biopsy specimen, which is 1 to 3 cm long and 1.2 to 2 mm in diameter, represents only 1/50,000 of the total mass of the liver,[6] and the histopathological evaluation is essentially subjective, even with the use of a semiquantitative scoring system.[7] Thus, the accuracy of liver biopsy studies is limited by sampling errors[8] and intraobserver and interobserver variability.[7] Therefore, research is underway for the evaluation of noninvasive methods for assessing liver fibrosis.[9]

Acoustic radiation force impulse (ARFI) imaging is a novel ultrasound-based elastography method.[10] To evaluate tissue stiffness, the Virtual Touch tissue quantification system (Siemens Medical Solutions, Mountain View, CA), which uses ARFI technology, has been introduced and is now commercially available in an ultrasound apparatus.[11, 12] Tissue stiffness is measured by ARFI imaging with the simple push of a button while real-time B-mode imaging is performed after the determination of the region of interest (ROI), which is graphically displayed at a size of 10 mm long by 6 mm wide. Tissue in the ROI is mechanically excited with an impulsive acoustic radiation force, which results in the generation of shear waves within the tissue; the velocity of these shear waves, which is proportional to the square root of the tissue elasticity, is quantitatively expressed as the tissue stiffness in meters per second.[9, 10] ARFI imaging has been reported to show good accuracy for the noninvasive diagnosis of liver fibrosis, especially in patients with chronic hepatitis C, chronic hepatitis B, and nonalcoholic steatohepatitis.[13]

Because graft fibrosis is often present even when liver biochemistry results are normal, the establishment of a noninvasive method for diagnosing graft fibrosis in pediatric liver transplant recipients would be very useful. The aim of this study was to evaluate the clinical utility of measuring liver stiffness by ARFI imaging in the diagnosis of graft fibrosis after pediatric living donor liver transplantation (LDLT).

PATIENTS AND METHODS

Enrolled Patients and Ethical Considerations

Between July 2011 and July 2012, patients who had undergone pediatric LDLT were prospectively enrolled in this study when they underwent protocol percutaneous liver biopsy (Division of Pediatric Surgery, Department of Surgery, Keio University School of Medicine). Written informed consent was obtained from all patients for participation in the study, liver stiffness measurements by ARFI imaging, and a review of the associated data. We enrolled 65 patients and performed 73 liver biopsy examinations and liver stiffness measurements; 58 patients were examined 1 time, 6 patients were examined 2 times, and 1 patient was examined 3 times. However, we analyzed the data from each patient as independent data. The selection of the study population is shown in Fig. 1. The study protocol conformed to the ethical guidelines of the 1975 Declaration of Helsinki and was approved by the ethics committee at the Keio University School of Medicine (2011-046). The study was registered at the University hospital Medical information Network Clinical Trials Registry (UMIN000005954).

Figure 1.

Selection of the study population. *Fifty-eight patients were examined 1 time, 6 patients were examined 2 times, and 1 patient was examined 3 times. **Fifty-five patients were examined 1 time, and 2 patients were examined 2 times.

Measurements of Liver Stiffness by ARFI Imaging

Measurements of liver stiffness by ARFI imaging were performed by 7 ultrasonologists, who were blinded to the histopathological findings, with the Acuson S2000 (Mochida Siemens Medical Systems, Tokyo, Japan) with a convex probe (4C1) around the time of percutaneous liver biopsy. Liver stiffness was measured through the upper midline of the abdomen (midline value) and the right intercostal space (intercostal value) at depths of 2 and 3 cm, respectively, from the surface of the liver. Before the liver stiffness measurements, the examiner measured the depth of the liver surface from the skin and then carefully marked the ROI at 2 or 3 cm from the liver surface while avoiding large vessels. The measurement through each site was repeated 5 times at rest, without breath holding, and at the end of exhalation. Liver stiffness was defined by the median values measured at each site.

Evaluation of Liver Histopathology

After written informed consent was obtained, percutaneous liver biopsy was performed in clinical practice with an 18-gauge suction needle under ultrasound guidance to determine the immunosuppression regimen. The biopsy specimens, at least 1.0 cm in length, were routinely fixed in formalin, embedded in paraffin, and stained with hematoxylin-eosin, elastica van Gieson, Azan-Mallory, silver impregnation, and cytokeratin 7 for immunohistochemistry (clone OV-TL 12/30, Dako, Glostrup, Denmark). The liver pathology was evaluated by 2 experienced liver histopathologists (W.D. and Y.M.) who were blinded to the liver stiffness measurements by ARFI imaging but were provided the clinical data and previous histopathological findings. Portal fibrosis was evaluated according to the Metavir scoring system as follows: F0 indicated no fibrosis, F1 indicated portal fibrosis without septa, F2 indicated portal fibrosis with rare septa, F3 indicated numerous septa without cirrhosis, and F4 indicated cirrhosis.[14] In addition to portal fibrosis, we also evaluated pericellular fibrosis on the following 4-point semiquantitative scale: no or minimal pericellular fibrosis as no or minimal fibrosis, pericellular fibrosis in some periportal and/or perivenular areas as mild fibrosis, pericellular fibrosis in most periportal and/or perivenular areas as moderate fibrosis (Fig. 2), and diffuse pericellular fibrosis as severe fibrosis. Cases with fibrosis between 2 stages were classified under the higher stage. Significant fibrosis was defined F2 or worse portal fibrosis and/or moderate or worse pericellular fibrosis. The inflammatory activity and steatosis were evaluated with semiquantitative scales as none or minimal, mild, moderate, or severe.

Figure 2.

Typical moderate pericellular fibrosis as per our evaluation. Perisinusoidal strands of collagen deposition were observed around the portal tract. An elastica van Gieson stain was used (original magnification ×100).

Analysis of Other Findings and Post Hoc Exclusion

Clinical findings, radiological findings, and the results of blood tests at the time of examinations were retrospectively collected. Post hoc exclusion was performed on the basis of these findings and not on the basis of histopathological findings. Previous studies indicated that liver stiffness measurements by ARFI imaging were affected by inflammation, congestion, and steatosis[15-17]; therefore, the following cases were excluded from the analysis. Although all liver biopsies were planned on a protocol basis, 2 examinations incidentally showed elevations in the results of biochemical liver tests and acute cellular rejection (ACR) during histopathological analysis. Radiological findings were indicative of fatty liver in 2 examinations: they showed moderate steatosis during the histopathological analysis, and one of them additionally had mild ACR. Hepatic vein stenosis was observed in 2 patients (2 examinations); one patient had abdominal distension, whereas the other was asymptomatic. In addition, because pediatric LDLT was mostly performed with the left lobe or lateral segment graft, 2 recipients of right liver grafts were excluded. The findings in these excluded cases were separately analyzed. For the histopathological examination, 6 biopsy samples containing fewer than 3 portal tracts were considered inadequate and were excluded from the analysis. In all, we excluded 14 examinations from the analysis of the relationships between liver stiffness and histopathological findings, and 59 examinations from 57 patients were analyzed (Fig. 1); 55 patients were examined 1 time, and 2 patients were examined 2 times.

Statistical Analysis

To evaluate the intraobserver reliability of the measurements of liver stiffness by ARFI imaging, the values measured at each site were assessed with intraclass correlation coefficients (ICCs). The categorical data were presented as frequencies for the subjects. Fisher's exact test was used for comparing the categorical data. The continuous data were presented as medians and ranges. Differences between the continuous data were analyzed with the Mann-Whitney U test or the Kruskal-Wallis test. The correlations between continuous data were assessed with the Pearson correlation coefficient. The accuracy of liver stiffness measurements by ARFI imaging and biochemical liver tests for the diagnosis of significant fibrosis was assessed with an area under the receiver operating characteristic curve (AUROC) analysis. The cutoff values were determined by the maximization of the sum of the sensitivity and specificity on Youden's index.[18] P values less than 0.05 were considered statistically significant. The statistical analysis was performed with SPSS 20.0 (IBM SPSS, Chicago, IL).

RESULTS

Reliability of Measurements of Liver Stiffness by ARFI Imaging

Liver stiffness measurements through the upper midline of the abdomen and the right intercostal space (repeated 5 times through each site) were recorded for 69 and 70 examinations, respectively. The ICCs for the midline and intercostal values were 0.931 [95% confidence interval (CI) = 0.902-0.954, P < 0.001] and 0.925 (95% CI = 0.893-0.950, P < 0.001), respectively, which indicated excellent intraobserver reliability. We employed the median values measured through each site for further analysis because they were convenient for practical use and had sufficient diagnostic power.

Association Between Liver Stiffness Measurements by ARFI Imaging and Histopathological Findings

This analysis was based on the results of 59 examinations. The median number of portal tracts in the liver biopsy specimens was 6 (range = 3-20). Portal fibrosis was graded F0 in 27 cases, F1 in 28 cases, and F2 in 4 cases. Pericellular fibrosis was absent or minimal in 8 cases, mild in 38 cases, and moderate in 13 cases. In all, significant fibrosis (F2 or worse portal fibrosis and/or moderate or worse pericellular fibrosis) was observed in 14 specimens.

Figure 3 shows box plots for the midline values and intercostal values versus the grades of portal fibrosis and pericellular fibrosis. The midline values differed significantly with the portal fibrosis grade (P = 0.003, Kruskal-Wallis test) but did not differ significantly with the pericellular fibrosis grade (P = 0.05). The intercostal values showed significant differences with the portal fibrosis grade (P = 0.003) and the pericellular fibrosis grade (P = 0.004). The midline and intercostal values were significantly higher for the patients with F1 portal fibrosis (P = 0.02) and the patients with F2 portal fibrosis (P = 0.02 and P = 0.03, respectively) versus the patients with F0 portal fibrosis. The intercostal values were also significantly higher for the group with moderate pericellular fibrosis versus the group with no or minimal pericellular fibrosis (P = 0.02) and the group with mild pericellular fibrosis (P = 0.007).

Figure 3.

Liver stiffness measurements by ARFI imaging versus the grades of 2 patterns of fibrosis. Liver stiffness was expressed as the velocity of the shear wave (meters per second). Measurements were taken at 2 sites: those at the upper midline of the abdomen are presented as midline values, and those at the right intercostal space are presented as intercostal values. Box plots show median liver stiffness values with interquartile ranges. Error bars indicate the smallest and largest values within 1.5 box lengths of the upper and lower quartiles. Outliers represent the individual points. Midline values were significantly higher for the groups with F1 and F2 portal fibrosis versus the F0 group. Intercostal values were significantly higher for the F1 and F2 groups versus the F0 group. Intercostal values were also significantly higher for the group with moderate pericellular fibrosis versus the groups with no or minimal pericellular fibrosis or mild pericellular fibrosis. The statistical analysis was performed with the Kruskal-Wallis test.*P < 0.05; **P < 0.01.

In addition to fibrosis grading, histopathological findings revealed mild inflammatory activity in 17 specimens and moderate inflammatory activity in 6 specimens. Inflammatory activities were nonspecific and were mostly observed in the portal tracts. Mild steatosis (<25%) was observed in 6 specimens. Liver stiffness measurements of the midline and intercostal values showed no significant differences according to the grade of inflammatory activity or steatosis (Fig. 4).

Figure 4.

Liver stiffness measurements by ARFI imaging versus the grade of inflammatory activity or steatosis. Liver stiffness was expressed as the velocity of the shear wave (meters per second). Measurements were performed at 2 sites: those at the upper midline of the abdomen are presented as midline values, and those at the right intercostal space are presented as intercostal values. Box plots show median liver stiffness values with interquartile ranges. Error bars indicate the smallest and largest values within 1.5 box lengths of the upper and lower quartiles. Outliers represent the individual points. Midline and intercostal liver stiffness values showed no significant differences according to the grade of inflammatory activity or steatosis. The statistical analysis was performed with the Kruskal-Wallis test or the Mann-Whitney U test.

Patient Characteristics Grouped by Significant Fibrosis

Patient characteristics and histopathological findings grouped by the presence or absence of significant fibrosis are shown in Table 1. Patients who underwent LDLT for primary sclerosing cholangitis and hepatoblastoma showed no evidence of recurrence. The age at LDLT was significantly lower for the patients with significant fibrosis (P = 0.02, Mann-Whitney U test). ABO incompatibility was significantly associated with significant fibrosis (P = 0.001, Fisher's exact test). The grades of inflammatory activity and steatosis showed no significant intergroup differences (P = 0.58 and P > 0.99, respectively; Fisher's exact test).

Table 1. Patient Characteristics and Histopathological Findings Grouped by Significant Fibrosis
 No Significant Fibrosis (n = 45)Significant Fibrosis (n = 14)P Value
  1. NOTE: F2 portal fibrosis and moderate pericellular fibrosis were categorized as significant fibrosis. The statistical analysis was performed with Fisher's exact test or the Mann-Whitney U test.

  2. a

    The data are presented as medians and ranges.

Patient characteristics   
Male sex [n (%)]20 (44.4)8 (57.1)0.54
Age at examination (years)a10.1 (1.7–23.5)7.0 (3.7–22.7)0.19
Age at LDLT (years)a1.9 (0.2–16.8)1.0 (0.3–10.3)0.02
Time from LDLT (years)a6.0 (0.2–15.1)6.0 (2.0–12.5)0.73
ABO incompatibility [n (%)]2 (4.4)6 (42.9)0.001
Original disease [n (%)]  0.88
Biliary atresia31 (68.9)10 (71.4) 
Fulminant hepatic failure6 (13.3)2 (14.3) 
Wilson's disease3 (6.7)0 (0) 
Alagille syndrome2 (4.4)1 (7.1) 
Primary sclerosing cholangitis1 (2.2)1 (7.1) 
Hepatoblastoma2 (4.4)0 (0) 
Histopathological findings   
Portal fibrosis [n (%)]  0.002
F024 (53.3)3 (21.4) 
F121 (46.7)7 (50.0) 
F20 (0)4 (28.6) 
Pericellular fibrosis [n (%)]  <0.001
None or minimal8 (17.8)0 (0) 
Mild37 (82.2)1 (7.1) 
Moderate0 (0)13 (92.9) 
Inflammatory activity [n (%)]  0.58
Mild12 (26.7)5 (35.7) 
Moderate4 (8.9)2 (14.3) 
Mild steatosis [n (%)]5 (11.1)1 (7.1)>0.99

The results of liver stiffness measurements and biochemical liver tests are presented in Table 2. In all, 57 midline values and 56 intercostal values were obtained for liver stiffness measurements by ARFI imaging. The midline values were 1.26 m/second (range = 0.94-1.94 m/second) and 1.50 m/second (range = 1.15-1.87 m/second) for the groups without and with significant fibrosis, respectively, and the intercostal values were 1.205 m/second (range = 0.82-1.63 m/second) and 1.495 m/second (range = 1.20-1.76 m/second), respectively. Thus, both values were significantly higher in the group with significant fibrosis versus the group without significant fibrosis (P = 0.005 and P < 0.001, respectively). The results of biochemical liver tests were obtained for all examinations (n = 59) and revealed that the serum levels of aspartate aminotransferase (AST), alanine aminotransferase (ALT), and γ-glutamyltransferase (GGT) were significantly higher in the patients with significant fibrosis versus the patients without significant fibrosis (P = 0.01, P = 0.003, and P = 0.001, respectively).

Table 2. Results of Liver Stiffness Measurements and Biochemical Liver Tests
 Mean ± Standard Deviation (n)No Significant Fibrosis (n = 45)aSignificant Fibrosis (n = 14)aP Value
  1. NOTE: The liver stiffness was measured through the upper midline of the abdomen (midline values) and the right intercostal space (intercostal values). The results of biochemical liver tests were obtained for all examinations (n = 59). The statistical analysis was performed with the Mann-Whitney U test.

  2. a

    The data are presented as medians and ranges.

  3. b

    n = 44.

  4. c

    n = 13.

  5. d

    n = 12.

  6. e

    The normal ranges are 0.4–1.3 mg/dL for total bilirubin, 0.0–0.2 mg/dL for direct bilirubin, 10–35 IU/L for AST, 5–40 IU/L for ALT, 10–90 IU/L for GGT in males, and 5–40 IU/L for GGT in females.

ARFI imaging    
Midline value (m/second)1.346 ± 0.253 (57)1.26 (0.94–1.94)b1.50 (1.15–1.87)c0.005
Intercostal value (m/second)1.292 ± 0.195 (56)1.205 (0.82–1.63)b1.495 (1.20–1.76)d<0.001
Biochemistrye    
Total bilirubin (mg/dL)0.82 ± 0.28 (59)0.8 (0.3–1.6)0.9 (0.5–1.4)0.10
Direct bilirubin (mg/dL)0.02 ± 0.07 (59)0.0 (0.0–0.4)0.0 (0.0–0.2)0.35
AST (IU/L)29.1 ± 9.1 (59)26 (13–59)34 (24–51)0.01
ALT (IU/L)20.5 ± 11.8 (59)15 (9–80)25.5 (12–42)0.003
GGT (IU/L)32.5 ± 43.0 (59)16 (9–275)27 (13–110)0.001

Accuracy of Liver Stiffness Measurements by ARFI Imaging and Biochemical Liver Tests for the Diagnosis of Significant Fibrosis

Figure 5 shows receiver operating characteristic curves for liver stiffness measurements by ARFI imaging and biochemical liver tests for the diagnosis of significant fibrosis. The AUROCs were 0.760 for midline values (P = 0.005); 0.849 for intercostal values (P < 0.001); and 0.729 (P = 0.01), 0.767 (P = 0.003), and 0.809 (P = 0.001) for serum AST, ALT, and GGT levels, respectively. Table 3 shows AUROCs and cutoff values of liver stiffness measurements by ARFI imaging and biochemical liver tests. The indices of diagnostic accuracy, including the sensitivities, specificities, positive predictive values (PPVs), and negative predictive values (NPVs), are also documented. The optimal cutoff values were 1.30 m/second for midline values; 1.39 m/second for intercostal values; and 24, 25, and 20 IU/L for serum AST, ALT, and GGT levels, respectively. Intercostal values showed the highest AUROC and relatively high values for the specificity (81.8%) and PPV (52.9%). However, intercostal values alone still had a high false-positive rate (8/17 examinations or 47.1%).

Figure 5.

Receiver operating characteristic curves of liver stiffness measurements by ARFI imaging and biochemical liver tests for the diagnosis of significant fibrosis together with reference lines (gray lines). Biochemical liver tests included measurements of serum levels of AST, ALT, and GGT. The AUROCs were 0.760 for midline values (P = 0.005), 0.849 for intercostal values (P < 0.001), 0.729 for serum AST levels (P = 0.01), 0.767 for serum ALT levels (P = 0.003), and 0.809 for serum GGT levels (P = 0.001).

Table 3. Accuracy of Liver Stiffness Measurements by ARFI Imaging and Biochemical Liver Tests for the Diagnosis of Significant Fibrosis
 AUROC95% CIP ValueCutoffSensitivity (%)Specificity (%)PPV (%)NPV (%)
  1. a

    The normal ranges are 10–35 IU/L for AST, 5–40 IU/L for ALT, 10–90 IU/L for GGT in males, and 5–40 IU/L for GGT in females.

ARFI imaging        
Midline value0.7600.629–0.8920.0051.30 m/second92.363.642.996.6
Intercostal value0.8490.736–0.962<0.0011.39 m/second75.081.852.992.3
Biochemistrya        
AST (IU/L)0.7290.597–0.8610.0124 IU/L1004034.1100
ALT (IU/L)0.7670.629–0.9060.00325 IU/L57.186.757.186.7
GGT (IU/L)0.8090.686–0.9320.00120 IU/L92.966.746.496.8
Combination of tests        
Intercostal value ≥ 1.39 m/second and midline value ≥ 1.30 m/second    63.690.763.690.7
Intercostal value ≥ 1.39 m/second and GGT ≥ 20 IU/L    66.790.988.991.5

Correlations Between Liver Stiffness Values and Serum GGT Levels

The midline values and intercostal values are plotted in the left panel of Fig. 6. A significant correlation between the liver stiffness values measured through the 2 sites was observed (r = 0.398, P = 0.003, Pearson correlation coefficient). However, some examinations showed a large difference between the midline values and the intercostal values. Four of the 5 examinations in which the intercostal values were 1.39 m/second or higher and the midline values were less than 1.30 m/second did not show significant fibrosis. The serum GGT levels are plotted against the intercostal values in the right panel of Fig. 6. The correlation was not significant (P = 0.64). Because the sensitivity of serum GGT levels of 20 IU/L or higher for significant fibrosis was high (92.9%), 7 of 8 examinations that afforded false-positive intercostal values showed serum GGT levels < 20 IU/L. The combination of intercostal and midline values had a specificity of 90.7% and a PPV of 63.6%, and the combination of the intercostal value and the serum GGT level had a specificity of 90.9% and a PPV of 88.9% (Table 3).

Figure 6.

Intercostal values versus midline values (left) and serum GGT levels (right). A significant correlation between liver stiffness values measured through 2 sites was observed (r = 0.398, P = 0.003, Pearson correlation coefficient). The correlation between the intercostal values and the serum GGT levels was not significant (P = 0.64). The examinations that showed no significant fibrosis are represented by circles, and those with significant fibrosis are represented by crosses. The gray lines represent optimal cutoff values for the diagnosis of significant fibrosis determined from the receiver operating characteristic curves. Among the examinations that showed false-positive results with the intercostal values, 4 of 5 and 7 of 8 revealed negative values for the midline values and the serum GGT levels, respectively.

Liver Stiffness Values in the Excluded Examinations

Fourteen liver stiffness values were obtained from 8 excluded examinations (examinations excluded for inadequate biopsy specimens were not included). These values were compared to those obtained for the examinations included in the study and were classified by portal and pericellular fibrosis grading. According to portal fibrosis grading, 9 of the 14 values were outside the interquartile range. Similarly, 9 of the 14 values were outside the interquartile range for pericellular fibrosis grading. In detail, 2 examinations with hepatic vein stenosis showed higher liver stiffness values than the upper quartile in all grades. From the 2 examinations with right liver grafts, only intercostal values were obtained, and these values were within the interquartile range. No clear tendency was noted for the examinations with ACR and/or fatty liver. Detailed data are shown in Fig. 7.

Figure 7.

Liver stiffness values for the excluded examinations. The liver stiffness values of the excluded examinations were plotted, compared with the liver stiffness values obtained for the included examinations, and classified according to the portal and pericellular fibrosis grades. The liver stiffness values of all examinations with hepatic vein stenosis were higher than the upper quartile. Right liver grafts showed liver stiffness values within the interquartile range of intercostal values. No obvious tendency was noted for the examinations with ACR and/or fatty liver.

DISCUSSION

Graft fibrosis is a common finding during protocol biopsy examinations long after pediatric liver transplantation.[2] Evans et al.[3] and Scheenstra et al.[19] reported that 42 of 62 patients (68%) and 38 of 55 patients (69%), respectively, had developed graft fibrosis (as assessed by the extent of portal fibrosis) 10 years after pediatric liver transplantation. Fouquet et al.[20] reported that 15 of 67 patients (22%) had centrilobular fibrosis 10 years after pediatric liver transplantation for biliary atresia. Portal fibrosis–based liver fibrosis staging systems, such as those reported by Ishak[21] and the Metavir Study Group,[14] are widely used, even in studies on pediatric liver transplant recipients.[22-24] However, these systems were originally developed for hepatitis C, in which fibrosis is initially portal-based, and they do not account for other patterns of fibrosis.[25] The pericellular pattern of fibrosis, which was assessed in the current study, has been classically described in alcoholic liver disease[26] and nonalcoholic steatohepatitis.[27] For liver allografts, the pericellular pattern of fibrosis has been described in fibrosing cholestatic hepatitis developing in transplant recipients with recurrent hepatitis B or C.[28, 29] Recent reports indicate that centrilobular perisinusoidal fibrosis occurs in pediatric liver transplant recipients in association with tacrolimus withdrawal or donor-specific anti-human leukocyte antigen antibodies.[30, 31] In our study, pericellular fibrosis was observed in most examinations (51 of 59 examinations or 86%). Significant fibrosis, mainly moderate pericellular fibrosis, was often associated with ABO incompatibility, and this implicates immunological mechanisms. Because ABO-incompatible LDLT was performed in early childhood in the significant fibrosis group, the age at LDLT was lower in that group versus the group without significant fibrosis (data not shown). Although the terms are different (pericellular versus centrilobular perisinusoidal fibrosis), these fibrosis patterns seem to be similar and probably express the same phenomenon. It is suggested that graft fibrosis after pediatric liver transplantation should not be assessed only by the extent of portal fibrosis. Recently, Venturi et al.[23] developed a novel histological scoring system based on fibrosis in 3 areas: portal tracts, sinusoids, and centrilobular veins. The current study shows that moderate pericellular fibrosis is detectable by ARFI imaging and indicates that pericellular fibrosis is a significant change after pediatric LDLT.

For the noninvasive diagnosis of liver allograft fibrosis, transient elastography (TE; FibroScan, Echosens, Paris, France), an ultrasound-based method similar to ARFI imaging, has been reported to have good accuracy in detecting cirrhosis or significant fibrosis due to recurrent hepatitis C after liver transplantation.[32] However, there is some concern about the use of TE for the diagnosis of graft fibrosis after pediatric LDLT. A left liver graft, which is usually used for pediatric LDLT, often deviates from the right subphrenic space to the left. Because TE is performed through the right intercostal space and is not performed under real-time B-mode imaging, the measurement of liver stiffness by TE may be more difficult for pediatric liver transplant recipients with a partial liver versus those with whole or native livers. Additionally, TE measures liver stiffness 2.5 to 6.5 cm below the skin,[13] which is rather deep for pediatric patients. Venturi et al.[23] described no correlation between TE and the fibrosis score in 38 pediatric liver transplant recipients, including 24 recipients with partial grafts. Thus, we think that ARFI imaging is more feasible than TE after pediatric LDLT, although further studies comparing ARFI imaging and TE are required. During liver stiffness measurements by ARFI imaging, a few movement artifacts were occasionally seen in small children; however, liver stiffness measurements showed excellent reliability, as indicated by the ICCs, even in small children (data not shown). ICCs range between 0 (no reliability) and 1 (perfect reliability), and ICCs ≥ 0.75 are generally considered to signify excellent reliability.[33] Thus, the ICCs of this study (0.931 for the midline value and 0.925 for the intercostal value) indicate that liver stiffness measurements by ARFI imaging have excellent reliability in the pediatric liver transplant population. The results of excellent reliability for liver stiffness measurements by ARFI imaging are consistent with previous reports on liver stiffness measurements by ARFI imaging in healthy children of various ages (range = 0-17 years).[34, 35]

In our analysis, all examinations were performed for patients who had undergone liver transplantation with left lobe or lateral segment grafts. In fact, we proposed that liver stiffness measurements in recipients with a left liver graft were easier and more accurate through the upper midline of the abdomen versus the right intercostal space. However, the intercostal values of liver stiffness actually showed better diagnostic accuracy than the midline values, and this is consistent with the findings of a previous study of adult native liver stiffness measurements by ARFI imaging.[12] We have previously reported that midline values are higher than intercostal values, probably because of the direct pressure of the ultrasound probe on the liver.[11] However, in this study, some examinations showed intercostal values obviously higher than midline values, and intercostal values were often found to afford false-positive results. Sometimes, it was difficult to observe the left liver graft through the right intercostal space. This difficulty or the unique shape of the partial liver allograft might have affected intercostal values. Another possibility is that the segmental injury due to the complicated reconstruction of bile ducts or blood vessels affected the liver stiffness and resulted in sampling error. The diagnostic accuracy was improved with the consideration of both measurements.

In addition, the current study shows that the results of biochemical liver tests, such as those for serum AST, ALT, and GGT levels, have sufficient diagnostic power for significant graft fibrosis. Because liver fibrosis is a wound-healing process, graft fibrosis indicates that liver grafts are exposed to chronic injury, which probably results in slight immune-mediated damage, as indicated by slight elevations (within normal limits) in the results of biochemical liver tests. As indicated by AUROC analyses, liver stiffness measurements by ARFI imaging added minimal supplemental predictive value to the results of biochemical liver tests. However, all the cutoff values of biochemical liver tests were within normal limits, and this was consistent with previous reports showing slight elevations in the results of biochemical liver tests with abnormal graft histology.[3, 19] On the other hand, the cutoff value for liver stiffness measurements by ARFI imaging (1.39 m/second for intercostal values) was clearly abnormal. A previous meta-analysis revealed an optimal cutoff value of 1.34 m/second for fibrosis ≥ F2 in adult patients.[13] In pediatric liver disease, a cutoff value of 1.34 m/second for fibrosis > F0 was reported, but the results were limited by a small sample size and heterogeneous patient backgrounds.[36] The current study indicates that a combination of serum GGT levels and intercostal values provides a more accurate diagnosis for significant graft fibrosis after pediatric LDLT.

Our study does have some limitations. First, the sample population was relatively small; in particular, only a few patients showed progressive portal fibrosis (F2 in only 4 cases). Thus, we cannot clearly assess the ability of ARFI imaging to predict portal fibrosis. However, the midline and intercostal values were significantly different between the patients with F0 and F1 portal fibrosis. Because our previous study indicated that ARFI imaging could not differentiate between F0 and F1 fibrosis in adult patients,[11] the current study suggests that the expansion of fibrosis to areas other than the portal tract influences liver stiffness measurements by ARFI imaging in patients undergoing pediatric LDLT. Similarly, we could not derive further information on the factors affecting liver stiffness measurements, such as inflammation, steatosis, and congestion, because of the small sample size. Similarly to inflammation or steatosis indicated only by histopathological analysis, clinically suggestive ACR (inflammation) and steatosis also showed no clear effect on liver stiffness measurements. However, hepatic vein stenosis, especially in the symptomatic patient, indicated clearly elevated liver stiffness values. Second, the measurements of liver stiffness were made by 7 ultrasonologists. Because ARFI imaging might need more experience than TE,[13] measurements by several examiners might lead to heterogeneous results. Third, in the present study, liver biopsy was a reference parameter. Because biopsy itself is limited by sampling error[8] and observer variability,[7] the histological results are probably subject to misses and false-positive results. Moreover, the biopsy specimens investigated in this study (at least 1.0 cm long and containing 3 portal tracts) were small for an accurate diagnosis.[8] However, we have performed protocol biopsy annually in most pediatric transplant recipients because serial investigations are important even if a single specimen is small. Although the diagnostic accuracy of ARFI imaging was probably impaired by these limitations, the current study suggests that the measurement of liver stiffness by ARFI imaging is a useful alternative to protocol biopsy, if not the only test required, and that this diagnostic method can serve as a bridge between biopsies or as a guide for liver biopsy. However, the histopathological study of liver-biopsy specimens appears to be of continued value for interventions in the presence of inflammation, which still cannot be appropriately detected by elastography and clinical biochemistry testing, even in combination. The current study also suggests that ARFI imaging might play an adjuvant role in the diagnosis of hepatic vein stenosis.

In conclusion, liver stiffness measurements by ARFI imaging showed good accuracy in diagnosing graft fibrosis after pediatric LDLT. Measurement through the right intercostal space was more feasible than measurement through the midline. A pericellular pattern of fibrosis was frequently observed after pediatric LDLT, and moderate pericellular fibrosis was detectable by ARFI imaging. Slightly elevated levels of biochemical liver test results, even if they are within the normal limits, may be associated with significant graft fibrosis. Further investigations are required to estimate the diagnostic accuracy and reproducibility of liver stiffness measurements by ARFI imaging after pediatric LDLT.

ACKNOWLEDGMENT

The authors thank Dr. Atsuhiro Arisue and Dr. Mototoshi Kato for cooperating with this study and Ms. Mieko Tsuchida, Ms. Yuko Shirahata, Ms. Yoshiko Shiomi, Ms. Chigusa Mochizuki, Dr. Akihisa Ueno, and Mr. Yuji Masuda for performing the liver stiffness measurements via ARFI imaging.

Ancillary