Magnetic resonance elastography in evaluation of liver fibrosis in children with chronic liver disease

All continuous variables, except the BMI Z-scores, followed an abnormal distribution. A total of 52 patients, 31 male (59.6%) were included in the study. The majority of the biopsies were done for diagnosis of liver disease etiology. The median age was 11.8 years (IQR 8.0–15.6). The median time between MRE and liver biopsy was one day (IQR 0.0–2.0). Table 1 shows the patients’ demographic and clinical characteristics.

Table 1 Patient demographics and characteristics

MRE-SS could not be obtained from three patients as insufficient spleen parenchyma was included in the MR elastography images. Liver fat content and R2* could not be obtained from one patient as a result of obvious motion artifacts. MRE-LS values (kilopascal-kPa) were significantly different between Ishak fibrosis stages (p = 0.036), while MRE-SS values did not differ between fibrosis stages (p = 0.712). Liver stiffness on MRE was significantly higher in Group 2 than in Group 1 (3.99 ± 1.05 kPa vs. 2.50 ± 0.64, p < 0.001). Although statistically insignificant, spleen stiffness was also higher in Group 2 (6.26 ± 4.25 kPa vs. 3.70 ± 2.01 kPa, p = 0.065).

MRE-LS and spleen stiffness were weakly correlated (rs = 0.297, p = 0.038). MRE-LS was moderately correlated with the Ishak fibrosis score (r = 0.553, p < 0.001) and HAI (r = 0.406, p = 0.003), and weakly correlated with AST (r = 0.364, p = 0.008), steatosis (r = − 0.330, p = 0.018), and R2* (r = − 0.326, p = 0.02). The abnormal AST and higher HAI grade proportions were significantly higher in patients with higher Ishak fibrosis scores (p = 0.004 and p = 0.01, respectively). Only Ishak fibrosis score was a significant predictor of MRE-LS values in the multiple linear regression analysis. MRE-SS values were weakly correlated with Ishak fibrosis scores (r = 0.357, p = 0.012) (Additional file 1). After controlling for the effects of HAI, hepatic steatosis, AST level, and R2*, the strength and direction of the relationship between MRE-LS and Ishak fibrosis stage did not change (r = 0.496, p < 0.001). In contrast, the correlation strength between MRE-SS and Ishak fibrosis stages increased (r = 0.452, p = 0.002).

The diagnostic cut-off values of liver stiffness and spleen stiffness on MRE to differentiate each liver fibrosis stage (F0 vs. F1–6, F0–1 vs. F2–6, F0–3 vs. F4–6, F0–4 vs. F5–6, and F0–5 vs. F6) are given in Table 2. With a cut-off value of 2.65 kPa, the AUC was 0.633 (p = 0.302), and its sensitivity, specificity, PPV, NPV, and accuracy for differentiating no fibrosis from any fibrosis were 40.8%, 100%, 100%, 9.4%, and 44.2%, respectively. The same parameters for spleen stiffness were 0.638 (p = 0.526), 100%, 33.3%, 95.8%, 100%, and 95.9%, respectively. When no fibrosis-minimal fibrosis (F0–1) was compared with a higher degree of fibrosis (F ≥ 2), with a cut-off value of 2.14 kPa, the AUC was 0.760 (p = 0.003), and its sensitivity, specificity, PPV, NPV, and accuracy were 91.4%, 58.8%, 82.1%, 76.9%, and 80.8%, respectively (Fig. 1. Receiver operating characteristic curve for differentiation of Ishak stage 0–1 from stage 2 or higher). The same parameters for spleen stiffness were 64.2% (p = 0.112), 73.5%, 53.3%, 78.1%, 47.1%, and 67.3%, respectively.

Table 2 Diagnostic performance of liver stiffness and spleen stiffness to detect hepatic fibrosisFig. 1figure 1

Receiver operating characteristic curve for differentiation of Ishak stage 0–1 from stage 2 or higher. *Area Under Curve 0.760 (95% CI 0.618–0.901) (p = 0.003)

When comparing Groups 1 and 2, with a cut-off value of 2.97 kPa for the liver, the AUC was 0.905 (p < 0.001), and the sensitivity, specificity, PPV, NPV, and accuracy were 90.9%, 82.9%, 58.8%, 97.1%, and 84.6%, respectively [Fig. 2. Receiver operating characteristic curve for differentiation of Group 1 (Ishak stage 0–1–2–3) from Group 2 (Ishak stage 4–5–6)]. With a cut-off value of 6.26 kPa for MRE-SS, the AUC, sensitivity, specificity, PPV, NPV, and accuracy for differentiating the same groups were 0.684 (p = 0.083), 45.5%, 94.7%, 71.4%, 85.7%, and 83.7%, respectively. The AUC values for T1, T2, T2*, and cT1 were not statistically significant.

Fig. 2figure 2

Receiver operating characteristic curve for differentiation of Group 1 (Ishak stage 0–1-2–3) from Group 2 (Ishak stage 4–5–6). *Area Under Curve 0.905 (95% CI 0.814–0.996) (p < 0.001)

When patients were divided into two groups with fibrosis stages 0–1–2–3–4 and 5–6 (cirrhosis) according to the Ishak scoring system, with a cut-off value of 3.15 kPa for the liver, the AUC was 0.898 (p < 0.001) and its sensitivity, specificity, PPV, NPV, and accuracy were 100.0%, 80.9%, 35.7%, 100.0%, and 82.7%, respectively. With a cut-off value of 8.26 kPa for spleen, AUC was 0.825 (p = 0.002), and its sensitivity, specificity, PPV, NPV, and accuracy were 60.0%, 97.7%, 75.0%, 95.6%, and 93.9%, respectively. The AUC values for T1, T2, T2*, and cT1 were not statistically significant. However, there were only five patients in the second group.

Only six patients had a BMI Z-score of >  + 2. Those with a BMI Z-score >  + 2 had a higher liver fat ratio of ≥ 5% (hepatic steatosis) than those with a BMI Z-score ≤ 2 (p = 0.047), and the median liver fat ratio (2.7% vs. 17.6%, p = 0.012) and median R2* value (25.3 vs 30.6 kPa, p = 0.009) were higher in obese patients. The other variables, including MRE-LS and MRE-SS, were not significantly different between the two groups. Nevertheless, there were no obese patients with Ishak fibrosis stage ≥ 4.

留言 (0)

沒有登入
gif