HBVandHCV-Direct Comparison of Diagnostic Performance of Transient Elastography

From Liver International

Direct Comparison of Diagnostic Performance of Transient Elastography in Patients With Chronic Hepatitis B and Chronic Hepatitis C

Ana-Carolina Cardoso; Roberto J. Carvalho-Filho; Christiane Stern; Alexandrine Dipumpo; Nathalie Giuily; Marie-Pierre Ripault; Tarik Asselah; Nathalie Boyer; Olivier Lada; Corinne Castelnau; Michelle Martinot-Peignoux; Dominique-Charles Valla; Pierre Bedossa; Patrick Marcellin

Discussion Only
View Full Text Provided By Medscape
This study was specifically designed to evaluate and compare the diagnostic performance of transient elastography (TE) for the assessment of liver fibrosis in HBV and HCV subjects and was carried out in a single centre and included a large cohort of consecutive subjects who underwent a liver biopsy (n = 565). Its results confirm that TE measurement is an accurate tool for the non-invasive diagnosis of liver fibrosis in patients with chronic viral hepatitis, either related to HBV or HCV.

As expected, TE values significantly increased with higher fibrosis stages, both in HCV and HBV patients (Fig. 3), with TE values being not significantly different between groups for a given fibrosis stage. Specifically regarding HCV patients, TE exhibited a good diagnostic performance in predicting significant fibrosis, and an excellent performance for the diagnosis of advanced fibrosis and cirrhosis, as previously shown in previous studies,^[10,15] with area under the receiver-operating characteristics (AUROCs) comparable to these studies (0.868 ± 0.019, 0.894 ± 0.020 and 0.947 ± 0.027 respectively).
In contrast to chronic hepatitis C, there are few studies with appropriate methodology on the accuracy of TE in patients with chronic hepatitis B.^{[11, 12, 16, 17]} In a multicentre French study, Marcellin et al.^[11] evaluated 173 HBV patients and found AUROCs of 0.810 (95% CI 0.730–0.860) for predicting significant fibrosis, and 0.930 (95% CI 0.820–0.980) for estimating the presence of cirrhosis, by using 7.2 and 11.0 kPa as cut-offs respectively. In the present study, performed in a different cohort of patients, TE showed a diagnostic performance comparable to that previously observed, being highly accurate in discriminating patients with significant fibrosis (0.867 ± 0.026), with even higher AUROCs for predicting the presence of advanced fibrosis (0.902 ± 0.029) and cirrhosis (0.935 ± 0.024).
It has been shown that TE exhibits a similar performance for predicting significant fibrosis and higher accuracy to identify cirrhosis, as compared to other non-invasive tests.^[18-20] This has been further confirmed by Castéra et al.^[21] in a recent study with 298 HCV patients, in which the performances of TE and of several serum non-invasive biomarkers for predicting the presence of cirrhosis were compared. In this study, TE was the most accurate method for identifying cirrhosis (METAVIR F4). In addition, the highly validated TE cut-off of 12.5 kPa exhibited a positive predictive value (PPV) of 85% to predict the presence of cirrhosis and a negative predictive value (NPV) of 95% to exclude cirrhosis, with LR(+) and LR(−) of 16.6 and 0.18 respectively.

Likewise, Chan et al.^[12] observed in HBV patients a PPV of 82% for predicting cirrhosis and a NPV of 87% to exclude cirrhosis when using a cut-off of 12.0 kPa for patients with ALT <1 ×ULN [LR(+) = 12.9 and LR(−) = 0.42], and a PPV of 78% and a NPV of 92% when using a cut-off of 13.4 kPa for subjects with ALT 1–5 ×ULN (LR(+) = 11.1 and LR(−) = 0.27). As compared to the results of these two studies, the TE cut-off values used in our study (12.5 kPa for HCV patients and 11.0 kPa for HBV subjects) had similar and high NPVs for excluding cirrhosis (98%), although these cut-offs showed lower PPVs to predict F4 stage (58% for HCV and 39% for HBV,Table 2). This might be attributable to different prevalences of cirrhosis in the study by Castéra et al. (23%),^[19] in the study by Chan et al. (25%),^[12] and in this study (8%) (Fig. 2). In fact, this hypothesis is further emphasized by the observation, in our study, of very similar likelihood ratios [LR(+) = 14.65 and LR(−) = 0.17 for HCV; LR(+) = 7.34 and LR(−) = 0.28 for HBV], diagnostic parameters known to be less susceptible to the prevalence of the target disorder (e.g. a given fibrosis stage) in the study population.

An issue that has been a matter of dispute is whether or not TE would reveal a similar diagnostic performance in patients with chronic hepatitis B as compared with patients with chronic hepatitis C. Theoretically, interferences in TE accuracy could be exerted by a macronodular pattern of cirrhosis and by the fluctuation of hepatic necroinflammatory activity (often expressed by exacerbations in serum aminotransferases levels), features more frequently observed in chronic hepatitis B. In addition, in HBV infection, a more precise fibrosis staging is desirable for therapeutic decisions, particularly to identify proper candidates for antiviral therapy [septal fibrosis (F2) is usually required], as well as to select the most adequate drug to be used (interferon is generally not indicated for cirrhotic subjects). In a study published by Ogawa et al.^[22] they evaluated TE in 68 patients with chronic HBV hepatitis and 161 with chronic HCV hepatitis. These authors found a statistically significant correlation of TE values with liver fibrosis in both HBV and HCV hepatitis (r = 0.559 and r = 0.686 respectively). However, a direct comparison of accuracies between groups has not been done in that study. In 2010, Sporea et al.^[23] assessed TE in 140 patients with HBV infection and 317 patients with HCV infection. These authors found a significant correlation between liver stiffness and fibrosis in both groups, with a higher correlation coefficient in HCV patients (r = 0.578 vs. r = 0.408, P = 0.02). Nevertheless, the small number of cirrhotic HBV patients might have biased the comparison (7 vs. 39 HCV). In addition, likelihood ratios have not been presented, making it difficult to draw any firm conclusions. In the present study, by direct comparison, the AUROCs of TE were not significantly different for predicting significant fibrosis, advanced fibrosis and cirrhosis in HBV and HCV patients (Fig. 4). In addition, as shown inTable 2, TE displayed comparable accuracy parameters in both groups. Taken together, these results show that TE exhibits similar overall diagnostic performance in HBV infection as compared to HCV patients.

It is well known that the accumulation of extracellular matrix is the major determinant of liver stiffness. However, it is conceivable that cellular oedema and necroinflammatory changes could also influence TE measurement. This concept is supported by prior studies that observed increasing values of liver stiffness in the presence of high grades of necroinflammatory activity.^{[12, 24–26]} In the present study, as expected, HBV subjects with elevated ALT showed higher TE values. This could be explained, at least in part, by an association between higher ALT levels and significant necroinflammatory activity or steatosis. In fact, we observed an increase of TE values in parallel with the degree of necroinflammatory activity in both HCV and HBV patients (Fig. 3), as well as higher TE values in the presence of more significant hepatic steatosis. Nevertheless, the comparison of diagnostic accuracies of TE in subjects with different ALT levels, as assessed by AUROC analysis, showed no influence of ALT activity on the overall performance of TE in predicting significant fibrosis, advanced fibrosis and cirrhosis. Likewise, comparable AUROCs were found for the diagnosis of significant fibrosis, advanced fibrosis and cirrhosis irrespective of the presence of significant activity or significant hepatic steatosis, in both HCV and HBV patients. However, we cannot exclude the possibility that the association between TE values and steatosis in HBV subjects might have been attenuated by the lower prevalence of moderate/significant steatosis as compared to HCV patients. In fact, in a recently published study by Fraquelli et al.,^[27] severe/moderate necroinflammatory activity was associated with fibrosis overestimation in HBV patients, while severe/moderate steatosis and severe/moderate necroinflammatory activity were independently associated with fibrosis overestimation in HCV subjects.

To evaluate whether the use of fixed cut-off points proposed by Marcellin et al.^[11] would exhibit the same accuracy of TE measurement for the prediction of fibrosis stage in HBV carriers with normal or elevated ALT levels, we performed a comparison with the ALT-specific values defined by Chan et al.^[12] For predicting significant fibrosis, both cut-off sets performed similarly in estimating the severity of liver fibrosis in HBV patients with either normal or elevated ALT (Table 3). The cut-off values proposed by Chan et al. did not exhibit higher performances for predicting fibrosis stage. The performances were slightly higher for predicting advanced fibrosis and cirrhosis in patients with ALT between 1 and 5 ×ULN, however, the differences were not significant. This discrepancy between our results and those of Chan et al. might be because of different populations studied. The use of cut-offs more adapted to the interference of necroinflammatory activity (expressed by higher ALT levels) on TE measurement, instead of fixed values, needs to be validate in further studies, in different populations.

In conclusion, in HBV patients, TE measurement reliably predicts the absence or the presence of significant fibrosis, advanced fibrosis or cirrhosis, and exhibits similar diagnostic performance as compared to HCV subjects. In patients with chronic hepatitis B, the use of TE cut-off values adjusted to ALT level for estimating liver fibrosis needs to be validated.

Continue Reading....