Accuracy and reproducibility of transient elastography for the diagnosis of fibrosis in pediatric nonalcoholic steatohepatitis

Authors

  • Valerio Nobili,

    Corresponding author
    1. Liver Unit, “Bambino Gesù” Children's Hospital and Research Institute, Rome, Italy
    • Liver Unit, Research Institute, Bambino Gesù Children's Hospital, Piazza S. Onofrio, 4, 00165 Rome, Italy
    Search for more papers by this author
    • fax: 39-0668592192.

  • Francesco Vizzutti,

    1. Department of Internal Medicine, Higher Education and Transfer DENOThe, University of Florence/Azienda Ospedaliero Universitaria Careggi Firenze (AOUC), Florence, Italy
    Search for more papers by this author
  • Umberto Arena,

    1. Department of Internal Medicine, Higher Education and Transfer DENOThe, University of Florence/Azienda Ospedaliero Universitaria Careggi Firenze (AOUC), Florence, Italy
    Search for more papers by this author
  • Juan G. Abraldes,

    1. Hepatic Hemodynamic Laboratory, Liver Unit, IMDiM, Hospital Clínic, Centro de Investigación Biomédica en red de Enfermedades Hepáticas y Digestivas (Ciberehd), University of Barcelona, Barcelona, Spain
    Search for more papers by this author
  • Fabio Marra,

    1. Department of Internal Medicine, Higher Education and Transfer DENOThe, University of Florence/Azienda Ospedaliero Universitaria Careggi Firenze (AOUC), Florence, Italy
    2. Center for Research, Higher Education and Transfer DENOThe, University of Florence/Azienda Ospedaliero Universitaria Careggi Firenze (AOUC), Florence, Italy
    Search for more papers by this author
  • Andrea Pietrobattista,

    1. Liver Unit, “Bambino Gesù” Children's Hospital and Research Institute, Rome, Italy
    Search for more papers by this author
  • Rodolfo Fruhwirth,

    1. Radiology Department “Bambino Gesù” Children's Hospital and Research Institute, Rome, Italy
    Search for more papers by this author
  • Matilde Marcellini,

    1. Liver Unit, “Bambino Gesù” Children's Hospital and Research Institute, Rome, Italy
    Search for more papers by this author
  • Massimo Pinzani

    1. Department of Internal Medicine, Higher Education and Transfer DENOThe, University of Florence/Azienda Ospedaliero Universitaria Careggi Firenze (AOUC), Florence, Italy
    2. Center for Research, Higher Education and Transfer DENOThe, University of Florence/Azienda Ospedaliero Universitaria Careggi Firenze (AOUC), Florence, Italy
    Search for more papers by this author

  • Potential conflict of interest: Nothing to report.

Abstract

Transient elastography (TE) has received increasing attention as a means to evaluate disease progression in chronic liver disease patients. In this study, we assessed the value of TE for the prediction of fibrosis stage in a cohort of pediatric patients with nonalcoholic steatohepatitis. Furthermore, TE interobserver agreement was evaluated. TE was performed in 52 consecutive biopsy-proven nonalcoholic steatohepatitis patients (32 males, 20 females, age 13.6 ± 2.44 years). The area under the receiver operating characteristic curves for the prediction of “any” (≥1), significant (≥2), or advanced fibrosis (≥3) were 0.977, 0.992, and 1, respectively. Calculation of multilevel likelihood ratios showed that TE values <5, <7, and <9 kPa, suggest the presence of “any” fibrosis, significant fibrosis, and advanced fibrosis, respectively. TE values between 5 and 7 kPa predict a fibrosis stage of 1, but with some degree of uncertainty. TE values between 7 and 9 kPa predict fibrosis stages 1 or 2, but cannot discriminate between these two stages. TE values of at least 9 kPa are associated with the presence of advanced fibrosis. The intraclass correlation coefficient for absolute agreement was 0.961. Conclusion: TE is an accurate and reproducible methodology to identify pediatric subjects without fibrosis or significant fibrosis, or with advanced fibrosis. In patients in which likelihood ratios are not optimal to provide a reliable indication of the disease stage, liver biopsy should be considered when clinically indicated. (HEPATOLOGY 2008.)

Nonalcoholic fatty liver disease (NAFLD) is the hepatic manifestation of the metabolic syndrome, a cluster of abnormalities related to insulin resistance, which is frequently associated with obesity. Given the strong association of NAFLD with increased body mass index (BMI) and the considerable increase in the prevalence of overweight among children and adolescents,1 NAFLD represents an emerging clinical problem affecting a substantial proportion of these subjects (2.6%–9.8%),2–3 especially in the presence of obesity.4 The high prevalence of NAFLD, and the likelihood of evolution to cirrhosis and its complications warrant increased attention toward this disorder.5–9 Disease progression depends on the presence of hepatocellular damage, inflammation and fibrogenesis which define a pathological entity called nonalcoholic steatohepatitis (NASH).

Currently, histopathological analysis of liver tissue represents the only means to assess fibrosis in NAFLD.10 In the past decade, major efforts have been directed at identifying noninvasive methods for the assessment of liver fibrosis in different chronic liver diseases (CLD) including NAFLD.11 These efforts assume a particular relevance in the pediatric setting, where the use of liver biopsy is perceived as bearing higher risks and is less acceptable than in adults.

A noninvasive medical device based on transient elastography (TE) (Fibroscan; Echosens, Paris, France) has received increasing attention. This system has been proposed for the measurement of liver stiffness, considered to be a direct consequence of the fibrotic evolution of CLD. Accordingly, several studies suggesting a clinical usefulness of TE for the prediction of liver fibrosis in CLD have been published (for a review see Pinzani et al.11). Notably, as recently reported by Yoneda et al.,12 the extent of steatosis does not seem to affect TE even in a cohort of adult patients with NASH. Recently, encouraging results concerning the feasibility and accuracy of TE in the prediction of fibrosis in a cohort of miscellaneous pediatric liver diseases were provided by de Lédinghen et al.13

The aim of the present study was to assess the accuracy of TE in identifying different degrees of fibrosis in a cohort of consecutive children and adolescents with NASH, by using a multilevel likelihood ratios (LRs) analysis approach. Furthermore, interobserver agreement of TE measurements was evaluated.

Abbreviations

BMI, body mass index; CLD, chronic liver diseases; IQR, interquartile range; LR, likelihood ratio; NAFLD, nonalcoholic fatty liver disease; NASH, nonalcoholic steatohepatitis; TE, transient elastography.

Patients and Methods

Patients.

A total of 67 consecutive children and adolescents, 35 males and 32 females, with suspected NASH were referred to Bambino Gesù Children's Hospital and Research Institute between July 15, 2007 and January 15, 2008 for staging and grading of the disease. All the enrolled subjects had serum aminotransferases either persistently or intermittently elevated (at least two abnormal determinations within 6 months prior to enrollment), associated with diffusely hyperechogenic liver tissue at ultrasound examination (so-called “bright liver”), and hyperinsulinism. Insulin resistance was assessed by the homeostatic model assessment14 and insulin sensitivity index composite.15 After histopathological evaluation, the presence of NASH was confirmed in 52 patients (32 males and 20 females; mean age of 13.6 ± 2.44 years; age range, 4–17 years), which were enrolled in the present study. Subjects with at least one of the following conditions were excluded from the study: cardiopulmonary disease, chronic renal failure, recent-active infections, chronic inflammatory diseases, autoimmune diseases, use of anti-inflammatory drugs, abnormal international normalized ratio, and/or platelet count below 60 × 109/L.16 Secondary causes of steatosis17 including alcohol abuse (≥140 g/week), total parenteral nutrition or rapid weigh loss, endocrinological diseases, inborn disorders, inflammatory bowel disease, and the use of drugs known to cause steatosis were excluded in all cases. No patient had been previously subjected to bariatric/abdominal surgery. The concomitant presence of other acute or CLD, particularly viral hepatitis, cytomegalovirus, and Epstein-Barr virus infections; vascular diseases of the liver; biliary tract disorders; and autoimmune, genetic, and other metabolic liver diseases were ruled out using standard clinical and laboratory evaluation as well as through histological examination of the liver biopsy.

Laboratory tests, including levels of bilirubin, albumin, alanine aminotransferase, aspartate aminotransferase, gamma-glutamyl transpeptidase, alkaline-phosphate, creatinine, fasting glucose, ferritin, cholesterol and triglycerides, insulin, platelet count, international normalized ratio, and oral glucose tolerance test, were performed in all patients within 1 week prior the enrollment. Afterward, all patients were prescribed a balanced, low-caloric diet, as detailed elsewhere,18 to attain a negative caloric balance. All patients underwent TE measurement within 6 months from liver biopsy.

The study protocol conformed to the ethical guidelines of the 1975 declaration of Helsinki (revision of Edinburgh, 2000) and was performed according to the recommendations of the Ethics Committee of the Bambino Gesù Children's Hospital and Research Institute. An informed consent was a priori obtained from a responsible guardian.

Ultrasound-Assisted Liver Biopsy.

Liver biopsy was performed in all subjects, after an overnight fast, using an automatic core biopsy 18-Gauge needle (Biopince, Amedic, Sweden) under general anaesthesia19 and ultrasound guidance. A Sonoline Omnia Ultrasound machine (Siemens, Germany) provided with a 5-MHz probe (5.0 C 50; Siemens) and with a biopsy adaptor was employed. Two biopsy passes within different liver segments were performed for each subject. The length of liver specimen (in millimeters) was recorded. Only samples with a length ≥15 mm and including at least 10–11 complete portal tracts were considered adequate for the purpose of the study. Biopsies were evaluated by a single experienced liver pathologist blinded to TE results. Biopsies were routinely processed (that is, formalin-fixed and paraffin-embedded): sections of liver tissue, 5-μm thick, were stained with hematoxylin-eosin, Masson trichrome, Van Gieson, periodic acid Schiff stain after diastase digestion, and Prussian blue stain. Immunohistochemical staining with antibodies to alpha-1-antitrypsin was used to exclude alpha-1-antitrypsin deficiency–associated liver disease. Fibrosis was scored using the Brunt classification:20 0, no fibrosis; 1, zone 3 perisinusoidal fibrosis; 2, as above with portal fibrosis; 3, as above with bridging fibrosis; and 4, cirrhosis. In this study, fibrosis stages were arbitrary stratified on the basis of Brunt's scoring system, unless indicated otherwise, as: “any” fibrosis (≥1), significant fibrosis (≥2), and advanced fibrosis (≥3). NASH was defined as the presence of steatosis associated with at least two of the following features: lobular inflammation, hepatocyte ballooning with or without Mallory bodies, and fibrosis (Table 1).

Table 1. Histological Findings
Grade/StageSteatosis, n (%)Lobular Inflammation, n (%)Ballooning, n (%)Fibrosis, n (%)
  1. Steatosis in liver specimens was arbitrarily scored by percentage of hepatocytes with fat deposits as: 0, <5%; 1, 5%–33%; 2, 34%–66%; and 3, >66%. Lobular inflammation was graded on a 4-point scale on a 200× field as: 0, no foci; 1, <2 foci; 2, 2–4 foci; and 3, >4 foci. Hepatocyte ballooning was graded from 0 to 2 as: 0, none; 1, few ballooned cells; and 2, many/prominent ballooned cells. Fibrosis was scored according to Brunt et al.20 n, number of patients.

02 (4)24 (48)11 (22)
116 (32)32 (64)18 (36)27 (54)
230 (60)16 (32)8 (16)7 (14)
34 (8)2 (4)
43 (6)

TE.

TE was performed by using the FibroScan apparatus (kindly provided by Axsan, Milan, Italy), which consists of a 3,5-MHz ultrasound transducer probe mounted on the axis of a vibrator. Mild amplitude and low frequency vibrations (50 Hz) are transmitted to the liver tissue, inducing an elastic shear wave that propagates through the underlying liver tissue. The velocity of the wave is directly related to tissue stiffness. With the patient lying in dorsal decubitus with the right arm in the maximal abduction, ultrasound examination previously identified an adequate section of liver tissue free of large vascular structures and gallbladder in the intercostal space on the right lobe, and a skin mark was made to guide the position of the TE transducer. The tip of FibroScan transducer was covered with a drop of gel and placed perpendicularly in the intercostal space. Moreover, under control time motion, the operator checked the correct position of the probe during TE examination.21, 22 Stiffness was measured on a cylinder of hepatic tissue of 1 cm in diameter and 2–4 cm in length. The procedures were performed by two independent investigators (F.V. and U.A.) who were blinded to the clinical and histological data. The interobserver agreement of TE was evaluated in 31 subjects (62%) studied on the same day and using the same Fibroscan apparatus. The operators had previously performed at least 100 TE determinations in patients with CLD. As previously described22 and as suggested by the provider of the instrumentation, we considered representative measurements the median value of 10 successful acquisitions with a success rate of at least 60%, and with an interquartile range (IQR) lower than 30%.

Statistical Analysis.

The comparison between clinical and laboratory variables and TE values between fibrosis stages was conducted with analysis of variance. Post hoc comparisons were conducted with the least significant t test. Linear trends were explored with polynomial contrasts.23 The discriminative ability of TE to predict “any” fibrosis (≥1), significant fibrosis (≥2), and advanced fibrosis (≥3) was assessed by receiver operating characteristic curve analysis. Multilevel LRs were used to explore the relationship between stiffness and fibrosis stage across the whole spectrum of TE values. The advantage of this approach is that, unlike sensitivity, specificity, and positive and negative predictive values, computation of LRs does not require dichotomization of test results, which may discard useful diagnostic information.24–25 The LR for each category was calculated by dividing the percentage of patients with the target condition (fibrosis ≥1, ≥2, or ≥3) in that category by the percentage without the condition in that category. Confidence intervals of 90%26 were calculated using the iterative method suggested by Gart and Nam27 with the StatsDirect statistical software (StatsDirect Ltd., Cheshire, UK). LRs above 10 and below 0.1 are considered to provide strong evidence to rule in or rule out diagnoses, respectively.24, 25 Multivariable logistic regression analysis was used to evaluate whether any clinical or laboratory variables could add to the diagnostic value of TE. Interobserver reproducibility was estimated with the intraclass correlation coefficient for absolute agreement.28 Values greater than 0.75 indicate an excellent agreement. All reported P values are two-sided. Only P values ≤ 0.05 were considered statistically significant. Statistical analysis was performed with the SPSS 15.0 package (SPSS, Chicago, IL).

Results

Characteristics of Patients.

TE was feasible in 50 NASH patients and these were included in the statistical analysis (96%). In two patients, TE determination failed, likely due to high BMI (>35 kg/m2). The major clinical and biochemical parameters of the subjects included in the analysis are listed in Table 2. All the enrolled patients showed insulin resistance, but none had diabetes. All subjects fulfilled the histopathological requirements, that is, the length of liver specimens was on average 19.2 ± 2.4 mm, and included 15.4 ± 2.9 complete portal tracts. All histological sections were considered adequate for evaluation with the scoring systems employed by the pathologist. In the group of subjects with advanced fibrosis/cirrhosis, none had a clinical history of clinical decompensation (that is, ascites, portal hypertension related bleeding, or encephalopathy). Considering all investigated subjects, the mean decrease of BMI from biopsy to TE measurement was of 2 ± 2 kg/m2 (ranging from −4.4 to to +0.2 kg/m2), and the mean decrease in alanine aminotransferase levels did not reach a level of significance (66 ± 75 U/L with extremes of 16–330 U/L vs. 80 ± 61 U/L with extremes of 14–388 U/L at the time of liver biopsy). Mean interval between liver biopsy and TE measurement was 4 ± 1.7 months. Mean success rate and the mean IQR of TE measurements were 88.8% and 16%, respectively. Figure 1 shows mean TE values for each stage of fibrosis.

Table 2. Clinical and Laboratory Findings of the Patients Enrolled in the Study
VariableFibrosis (n = pts,%)
0 (n = 11, 22)1 (n = 27, 54)2 (n = 7, 14)3–4 (n = 5, 10)P (ANOVA)
  • Results are expressed as mean ± standard deviation (SD). Fibrosis was scored according to Brunt classification. Patients were considered overweight or obese if the BMI was greater than the 85th and equal to the 97th or greater than the 97th percentile, respectively.

  • *

    Chi-square.

  • P = 0.023 for the linear trend.

  • P = 0.021 for the linear trend.

  • §

    P < 0.05 versus fibrosis stage 0.

  • P < 0.05 versus fibrosis stage 1.

  • P < 0.05 versus fibrosis stage 2.

  • Abbreviations: BMI, body mass index; AST, alanine aminotransferase; ALT, aspartate aminotransferase; INR, international normalized ratio; γ-GT, gamma-glutamyl transpeptidase; HOMA, homeostasis model assessment; ISI, insulin sensitivity index; TE, transient elastography; kPa, kilopascal; n = pts, number of patients per class.

Age (years)13 ± 413 ± 214 ± 314 ± 20.378
Male gender, n (%)4 (36)18 (67)6 (86)3 (60)0.153*
BMI (kg/m2)24 ± 626 ± 427 ± 626 ± 60.479
Obese, n (%)5 (46)14 (52)2 (29)3 (60)0.664*
Overweight, n (%)9 (82)21 (78)4 (57)3 (60)0.577*
ALT (U/L)66 ± 3472 ± 38102 ± 84136 ± 1410.116
AST (U/L)46 ± 1949 ± 1970 ± 4768 ± 470.143
INR1.12 ± 0.191.13 ± 0.21.05 ± 0.160.99 ± 0.210.482
Bilirubin (mg/dL)0.98 ± 0.790.6 ± 0.180.63 ± 0.150.62 ± 0.290.073
Albumin (g/dL)4.26 ± 0.314.33 ± 0.294.22 ± 0.234.48 ± 0.280.429
γ-GT (U/L)20 ± 1523 ± 1836 ± 3332 ± 280.323
Fasting glucose (mg/dL)79 ± 976 ± 879 ± 1082 ± 80.430
Cholesterol (mg/dL)148 ± 23169 ± 36181 ± 48176 ± 520.263
ISI3.59 ± 3.981.94 ± 1.042.24 ± 0.541.75 ± 0.620.187
HOMA17.19 ± 20.928.76 ± 4.8710.99 ± 4.128.24 ± 4.280.154
Triglycerides (mg/dL)75 ± 28117 ± 78124 ± 83171 ± 1050.116
Platelet count (×109/L)333 ± 83343 ± 80321 ± 61340 ± 620.911
TE (kPa)4.4 ± 0.66.1 ± 0.88§8.6 ± 0.83§20.4 ± 7.69§<0.000001
Figure 1.

Box plots of TE values in relation to the degree of fibrosis. Liver stiffness values are reported on the y axis in kPa, and, on the x axis, the degree of fibrosis according to Brunt scoring system. Top and bottom of boxes are first and second quartiles, respectively. Length of box represents the IQR within which 50% of the values are located. The line through the middle of each box represents the median. Error bars show minimum and maximum values.Abbreviations: kPa, kilopascal; TE, transient elastography.

Accuracy of TE for Prediction of “Any” Fibrosis (≥1).

The accuracy of TE for the diagnosis of “any” fibrosis is given in Fig. 2A. The data-driven best cutoff for the diagnosis of “any” fibrosis was 5.1 kPa (Table 3). Multilevel LRs (Table 4) show that only values lower than 5 kPa suggested the absence of “any” fibrosis. Values between 5 and 7 kPa suggested the presence of fibrosis, but only values higher or equal to 7 kPa were highly indicative of fibrosis.

Figure 2.

Receiver operating characteristic (ROC) curve showing the prediction of (A) “any” fibrosis (≥1), (B) significant fibrosis (≥2), and (C) advanced fibrosis (≥3) with TE in the whole patient population. The ideal area under the curve is 1.00. The straight line represents that based on chance alone (area under the curve, 0.50). The areas under the ROC curves are as follows: (A) 0.97 (90% CI, 0.90–0.99); (B) 0.99 (90% CI, 0.92–0.99); and (C) 1 (90% CI, 0.94–1). The best cutoff values for each curve are indicated with black dots (see Table 3). Abbreviation: TE, transient elastography.

Table 3. Diagnostic Accuracy of TE.
FibrosisCutoff (kPa)S (%)Sp (%)PPV (%)NPV (%)+LR−LR
  1. Diagnostic accuracy of TE in predicting “any” fibrosis (≥1), significant fibrosis (≥2), and advanced fibrosis (≥3). Performance of the selected best TE cutoff values are indicated. Abbreviations: S, sensitivity; Sp, specificity; PPV, positive predictive value; NPV, negative predictive value; +LR, positive likelihood ratio; −LR, negative likelihood ratio; TE, transient elastography; kPa, kilopascal; Inf., infinity.

≥15.197 (93–100)91 (77–100)97 (93–100)91 (77–100)10.72 (3.02–47.44)0.02 (0.006–0.12)
≥27.4100 (82–100)92 (82–97)80 (59–92)100 (93–100)12.67 (5.23–30.12)0.00 (0–0.20)
≥310.2100 (65–100)100 (94–100)100 (65–100)100 (94–100)Inf. (15.95–Inf.)0.00 (0–0.35)
Table 4. Multilevel LRs for the Prediction of Fibrosis Stages
TE (kPa)LRs for Different Fibrosis Stages
“Any”SignificantAdvanced
  1. LRs above 10 and below 0.1 provide strong evidence to rule in or rule out diagnoses, respectively. Fibrosis stages were classified as: “any” fibrosis, ≥1; significant fibrosis, ≥2; advanced fibrosis, and ≥3. Abbreviations: LR, likelihood ratio; TE, transient elastography; kPa, kilopascal; Inf., infinity.

<50.028 (0.006–0.119)0 (0–0.212)0 (0–0.369)
≥5 and <75.923 (1.614–26.566)0 (0–0.212)0 (0–0.369)
≥7 and <9Inf. (2.170- Inf.)3.167 (1.300–7.398)0 (0–0.369)
≥9Inf. (2.170- Inf.)Inf. (8.680- Inf.)22.50 (6.825–62.255)

Accuracy of TE for Prediction of Significant Fibrosis (≥2).

The accuracy of TE for the diagnosis of significant fibrosis is illustrated in Fig. 2B. The data-driven best cutoff for the diagnosis of significant fibrosis was 7.4 kPa (Table 3). Multilevel LRs (Table 4) show that TE values lower than 7 kPa suggested the absence of significant fibrosis, whereas values between 7 and 9 kPa were indeterminate for the diagnosis of significant fibrosis.

Accuracy of TE for Prediction of Advanced Fibrosis (≥3).

The accuracy of TE for the diagnosis of advanced fibrosis is shown in Fig. 2C. The data-driven best cutoff for the diagnosis of advanced fibrosis was 10.2 kPa (Table 3). Multilevel LRs (Table 4) show that values higher or equal to 9 kPa were indicative of advanced fibrosis, whereas values lower than 9 kPa suggested the absence of advanced fibrosis.

Multivariable analysis showed that no clinical-analytical variable could add to the discriminative value of TE for any stage of fibrosis investigated.

Interobserver Agreement in TE Determination.

Reproducibility of TE measurements could be estimated in 31 patients (10, 14, 2, and 5 subjects for Brunt's stages 0, 1, 2, and 3–4, respectively) that were explored by both operators (F.V. and U.A.). Reproducibility was excellent, as indicated by an intraclass correlation coefficient for absolute agreement of 0.96 (90% confidence interval [CI] = 0.92–0.97). No significant differences in clinical and biochemical variables were observed when compared to subjects investigated by a single operator. In particular, the mean BMI was not statistically different in the two groups (25 ± 6 and 26 ± 4).

Complications Related to the Procedures.

No major complications were associated with percutaneous liver biopsy. A total of 15 subjects (30%) experienced self-limiting abdominal and/or right shoulder pain, and six subjects (12%) required a single dose of intravenous analgesic drug (Tramadol). There were no complications associated with TE measurements.

Discussion

Pediatric obesity is an increasingly common problem in Western countries. While the majority of obese children with liver disease will have simple NAFLD, identification of those at risk of progressive liver fibrosis is important to determine prognosis and to offer available interventions. Although histopathological analysis of liver tissue represents the gold standard for NASH and fibrosis stage assessment, its use in a pediatric setting is less tolerable than in adults. Along these lines, it would be ideal to develop a noninvasive methodology to discriminate, among obese/overweight pediatric subjects with NAFLD, those who already have fibrosis.

Among other noninvasive methodologies, TE was recently proposed to investigate the degree of fibrotic evolution of CLD in a pediatric setting. De Lédinghen et al.13 provided evidence on the usefulness of TE for the prediction of fibrosis in a cohort of 116 patients with miscellaneous pediatric CLD. However, in that study, liver biopsy was available only in 33 subjects and it was not stated whether patients with NAFLD were included. Accordingly, the present study was specifically designed to provide information on the clinical utility of this methodology in a pediatric population affected by NASH. It should be noted that the population analyzed in our study is representative of a highly selected cohort typical of a specialized tertiary care referral center, and that the conclusions of the study cannot be applied to pediatric populations seen in primary care settings. In addition, the relatively small number of subjects recruited, reflecting the difficulty of obtaining liver biopsy samples and an accurate clinical framing even in a specialized pediatric referral center, does not allow us to reach definitive conclusions.

The results reported herein clearly indicate that while TE mean values are significantly different among closest Brunt's stages of fibrosis, an overlap is observed among patients with lower degrees of liver fibrosis (stages 0 and 1 and stages 1 and 2). However, TE appears to reliably predict the presence of “any” fibrosis, significant fibrosis, and advanced fibrosis.

To explore the whole spectrum of TE measurements, the analysis of multilevel LRs was introduced to estimate the likelihood of having a target degree of fibrosis according to TE measurements. Indeed, the use of multilevel LRs analysis is a more informative and correct approach since it explores the whole spectrum of TE measurements.24, 25 As a result, TE values of <5, <7, and <9 kPa suggest the presence of “any” fibrosis, significant fibrosis, and advanced fibrosis, respectively. TE values between 5 and 7 kPa predict fibrosis stage 1, but with some degree of uncertainty. TE values between 7 and 9 kPa predict fibrosis stages 1 or 2, but cannot discriminate between these two stages. TE values of at least 9 kPa are associated with the presence of advanced fibrosis. Of note, no clinical or analytical variable considered in the present study contributed to the predictive value of TE.

Data obtained in a similar cohort of NAFLD pediatric subjects29 indicate than in spite of advice on lifestyle modification no effect on fibrosis stage was observed in follow-up biopsies after a period of 24 months. Therefore, the time between biopsy and TE measurements (≤6 months) should not have had any impact on the fibrosis stage. On the contrary, the temporal gap between liver biopsy and TE determination did not allow us to reliably assess the possible interfering role of steatosis, necroinflammatory activity, and hepatocyte ballooning in our cohort of patients. Nevertheless, the degree and/or fluctuations of necroinflammatory activity during the course of NASH are generally moderate.30 Indeed, aminotransferases were only moderately increased and lower degrees of inflammation at histology were observed in our study population. Therefore, a relevant interfering role of inflammation in this scenario was not expected. Concerning liver steatosis the absence of an interfering role on TE measurements has been already defined in adults affected by NASH12 or chronic viral hepatitis.21, 31

Our findings on TE reproducibility are consistent with those of Fraquelli et al.32 in an adult population of CLD patients. In our cohort, the high level of interobserver agreement was maintained in spite of high BMI values, probably because of the use of ultrasound for guidance. From a technical point of view it is of note that de Lédinghen et al.13 used a specific pediatric probe, while the standard probe, indicated for adults, was successfully employed in our patients cohort as demonstrated by the IQR of TE measurements.

In summary, TE is an accurate and reproducible methodology to identify, in children and adolescents affected by NASH, those without any degree of fibrosis or significant fibrosis, or with advanced fibrosis. In subjects in whom LRs are not optimal to provide a reliable indication of the disease stage, liver biopsy should be considered when clinically indicated. However, TE is unlikely to show a similarly good performance for screening purposes in an unselected overweight/obese pediatric population. Therefore, the results of the study offer a promising background for further validating TE for the stratification of NAFLD subjects in childhood, including larger primary care and nonselected obese children and adolescents, before TE can be recommended as a screening test alone or in combination with other noninvasive methodologies.

Acknowledgements

We are indebted to all the patients and their legal guardians who participated in this study.

Ancillary