Noninvasive Evaluation of Hepatic Fibrosis in Hepatitis C Virus-Infected Patients Using Ethoxybenzyl-Magnetic Resonance Imaging

Shunsuke Nojiri; Atsunori Kusakabe; Kei Fujiwara; Noboru Shinkai; Kentaro Matsuura; Etsuko Iio; Tomokatsu Miyaki; Takashi Joh

Disclosures

J Gastroenterol Hepatol. 2013;28(6):1032-1039. 

In This Article

Results

Patient Characteristics

A total of 224 patients were examined who had undergone liver biopsy.

The index was constructed with data from 149 patients (estimation set) and 75 patients (validation set). Patient characteristics at the time of liver biopsy are shown in Table 1. There were no significant differences between the estimation and validation sets in any clinical or biochemical variable or fibrosis stage.

Time-Course of RE in Fibrosis

After EOB injection, RE increased rapidly for the first 35 s and gradually increased until 25 min (Fig. 1b). RE of F0 or F1 was significantly higher than that of F2 at 15, 20, and 25 min.

Because, at any stage (F0-F4), 25-min RE (RE25) was higher than 15-min RE (RE15) and 20-min RE (RE20) was between RE15 and RE25, we considered using 25-min data as the best method to obtain the most significant accuracy. We therefore decided to adopt 25-min data in this study, although the international consensus for the hepatobiliary phase is 20-min post-contrast.

Correlation of MRI Enhancement in Hepatobilialy Phase and Histological Findings

Figure 2 shows box plots of several biochemical fibrosis indexes and several MRI fibrosis indexes. On MRI, the mean contrast enhancement index constantly decreased as the fibrosis stage progressed to a higher stage. Changes in the histological fibrosis stage correlated with changes in RE at 25 min: RE25 (r = −0.478); LM at 25 min: LM25 (r = −0.463); LI at 25 min: LI25 (r = −0.661); LS at 25 min: LS25 (r = −0.376); Forn's Index (r = 0.551); FibroIndex (r = 0.6); Lok index (r = 0.561); and APRI (r = 0.480) (P < 0.001).

Figure 2.

Scores according to METAVIR fibrosis stages. The top and bottom of each box are the 25th and 75th percentiles, respectively. The line through the box is the median, and the error bars are the 5th and 95th percentiles. Spearman correlation coefficients were as follows: 25 m relative enhancement (RE), r = −0.478; 25 m liver-to-muscle ratio (LM), r = −0.463; 25 m liver-to-intervertebral disk ratio (LI), r = −0.661; 25 m liver-to-spleen ratio (LS), r = −0.376; Forn's score, r = 0.551; FibroIndex, r = 0.600; Lok index, r = 0.561; aspartate aminotransferase-to-platelet ratio index (APRI), r = 0.480, P < 0.001 for each correlation test.

Overall Diagnostic Performance of MRI Enhancement and Serum Markers

ROC curves were used for evaluating the overall diagnostic performance of several MRI indexes for different degrees of fibrosis (Estimation: Fig. 3 for F0-F1 versus F2-F4, F0-F2 versus F3-F4, and F0-3 versus F4, Validation: Fig. 4 for F0-F1 versus F2-F4, F0-F2 versus F3-F4, and F0-3 versus F4).

Figure 3.

Receiver operating characteristic curve testing ability of the various methods to discriminate METAVIR fibrosis stage F0/F1 from F2/F3/F4, F0/F1/F2 from F3/F4, and F0-3 from F4 at 25 min in the estimation set. ( ) liver-to-intervertebral disk ratio (LI); ( ) liver-to-spleen ratio (LS); ( ) relative enhancement (RE); ( ) liver-to-muscle ratio (LM).

Figure 4.

Receiver operating characteristic curve testing ability of the various methods to discriminate METAVIR fibrosis stage F0/F1 from F2/F3/F4, F0/F1/F2 from F3/F4, and F0-3 from F4 at 25 min in the validation set. ( ) liver-to-intervertebral disk ratio (LI); ( ) liver-to-spleen ratio (LS); ( ) relative enhancement (RE); ( ) liver-to-muscle ratio (LM).

Table 2 shows the areas under ROC scores of MRI indexes, Forn's score, APRI, FibroIndex, and Lok index for different degrees of fibrosis. For discriminating F0-1 from F2-4, area under receiver operating characteristics (AUROCs) were LI25 (0.88), LS25 (0.77), RE25 (0.83), LM25 (0.83), Forn's Index (0.82), APRI (0.79), FibroIndex (0.87), and Lok index (0.83). For discriminating F0-2 from F3, 4, AUROCs were LI25 (0.87), LS25 (0.72), RE25 (0.81), LM25 (0.75), Forn's Index (0.80), APRI (0.79), FibroIndex (0.85), and Lok index (0.82). For discriminating F0-3 from F4, AUROCs were LI25 (0.87), LS25 (0.69), RE25 (0.74), LM25 (0.78), Forn's Index (0.75), APRI (0.73), FibroIndex (0.84), and Lok index (0.79). In the MRI enhancement index, LI25 had the highest value of AUROC for discriminating significant, extensive fibrosis, and cirrhosis, and this was almost the same or higher than serum markers such as the FibroIndex and Forn's score. The only significant difference was observed between LI25 and LS25 for the diagnosis of cirrhosis (P < 0.05) on estimation. A significant difference was also observed between LI25 and LS15 or LM15 for the diagnosis of extensive fibrosis (P < 0.05) (data not shown).

Table 3 shows AUROC curves scores for different degrees of fibrosis on validation. For discriminating F0, 1 from F2-4, AUROCs were LI25 (0.85), LS25 (0.75), RE25 (0.83), LM25 (0.71), Forn's Index (0.83), APRI (0.80), FibroIndex (0.84), and Lok index (0.80). For discriminating F0-2 from F3, 4, AUROCs were LI25 (0.85), LS25 (0.72), RE25 (0.83), LM25 (0.73), Forn's Index (0.83), APRI (0.80), FibroIndex (0.84), and Lok index (0.78). For discriminating F0-3 from F4, AUROCs were LI25 (0.86), LS25 (0.67), RE25 (0.82), LM25 (0.82), Forn's Index (0.74), APRI (0.78), FibroIndex (0.83), and Lok index (0.82).

Also, in the validation group as well as the estimation group, in the MRI enhancement index, LI25 had the highest value of AUROC for discriminating significant, extensive fibrosis, and cirrhosis, and this was almost the same or higher than serum markers such as the FibroIndex and Forn's score as well as the estimation group.

Sensitivity, Specificity, and Positive and Negative Predictive Values (Tables 4 and 5)

Based on the ROC, two cut-off values were chosen to identify the absence (≧ 1.62, 1.55, and 1.47) and presence (≦ 1.38, 1.38, and 1.315) of significant and extent fibrosis and cirrhosis, respectively. Applying the higher cut-off (≧ 1.62), 9 (23%) of the 39 patients without significant fibrosis in the liver biopsy were correctly identified (number of data not described in the Table). The presence of significant fibrosis, and extent of fibrosis and cirrhosis could be excluded with cut-offs of ≧ 1.62, ≧ 1.55, and ≧ 1.47, respectively, and the negative predictive values were 69.2%, 90%, and 96.5%, respectively. The presence of cirrhosis could be ruled out with high certainty Table 4 and Table 5.

Applying the low cut-off (≦ 1.38), 56 (50.9%) of the 110 patients belonging to the estimation group with significant fibrosis in the liver biopsy were correctly identified. Fifty-six of 57 patients with ≦ 1.38 showed significant fibrosis in the liver biopsy (positive predictive value 98%).

The cut-offs of ≦ 1.38 and ≦ 1.315 also showed high PPV in the extent of fibrosis or cirrhosis but were not so high. The representative cut-off points of significant and extensive fibrosis and cirrhosis were determined as 1.55, 1.51, and 1.38, respectively. Comparing the accuracy of each cut-off point, LI25 had the highest accuracy of significant fibrosis and cirrhosis, but FibroIndex was higher for the extent of fibrosis. These tendencies were also demonstrated in the validation set.

processing....