ABSTRACT INTRODUCTION. *Diana Feier 1, *Monica Lupsor Platon 1,2, Horia Stefanescu 1,3, Radu Badea 1,2

Transient Elastography for the Detection of Hepatocellular Carcinoma in Viral C Liver Cirrhosis. Is there something else than Increased Liver Stiffness? *Diana Feier 1, *Monica Lupsor Platon 1,2, Horia Stefanescu 1,3, Radu Badea 1,2 1) Iuliu Hatieganu University of Medicine and Pharmacy; 2) Medical Imaging Department 3) Hepatology Department Regional Institute of Gastroenterology and Hepatology Cluj-Napoca, Romania Address for correspondence: Horia Stefanescu, MD, PhD Regional Institute of Gastroenterology and Hepatology 19-21 Croitorilor Str, 400162 Cluj-Napoca, Romania hstefanescu@umfcluj.ro ABSTRACT Background & Aims: Liver stiffness (LS) is increased in liver cirrhosis, higher values being associated with complications, among them the development of hepatocellular carcinoma (HCC). However, LS values alone cannot accurately differentiate patients with HCC. Therefore, our aim was to study the performance of LS measurement data and common biomarkers for the detection of HCC in HCV related liver cirrhosis. Methods: We performed a case matching study comparing HCV cirrhotic patients with and without HCC (72 in each group) that were identical in terms of sex, age, BMI and duration of HCV infection. All patients underwent LS measurement, endoscopy, liver imaging and liver function tests. A multiple regression analysis was performed and a HCC detection model was calculated, which was further validated in another group of 40 HCV infected cirrhotics, of whom 52% had HCC. Results: In the HCC group, LS was significantly higher (42 vs 27 kpa, p<0.0001). In the multivariate analysis higher values of LS, alanine-aminotransferase (ALAT), alpha-fetoprotein (AFP) and interquartile range (IQR) of LS measurements were independently associated with the presence of HCC (p<0.0001 for all parameters; Odds Ratios of 8.27, 1.01, 1.04 and 1.16, respectively). The detection model combining the four variables showed a good diagnostic performance in both training and validation groups, with AUROCs of 0.86 and 0.8, respectively. All variables were also positively correlated with tumor size. Conclusion: In HCV related cirrhosis, HCC is associated with increased LS and IQR values and high ALAT and AFP levels. By combining these four parameters into a regression model, liver cancer may be noninvasively predicted with good accuracy. Key words: liver cirrhosis hepatocellular carcinoma liver stiffness interquartile range detection model. Received: 26.06.2013 Accepted: 19.08.2013 *These authors contributed equally to the study INTRODUCTION Early detection of hepatocellular carcinoma (HCC) in patients with chronic hepatitis C virus (HCV) infection represents an emerging health problem. The growing incidence of HCV infection, together with continuous improvements in the clinical management of patients have led to an increasing prevalence of liver cirrhosis, which is the most important pathway towards developing HCC. These patients are the perfect candidates for HCC surveillance, having an annual incidence of HCC ranging between 4.1 and 7% [1]. The prognosis of patients with liver cirrhosis and HCC is strictly related to the liver residual function and tumour size, as stated and extensively validated by the BCLC criteria [2]; thus, non-invasive early detection of HCC is of major importance. Multiple parameters widely available in routine clinical practice were proposed for the detection of early HCC in patients with well-preserved liver function and various algorithms emerged for the diagnosis and follow-up of patients with HCV liver cirrhosis [3, 4]. Some of these models pointed out the individual role of increased liver stiffness (LS) measurements and serological markers as predictive biomarkers of HCC [5], but none of the studies evaluated the combined role of these main predictors. Transient elastography (TE) is an imaging technique that measures LS under the assumption of pure elasticity of the

284 Feier et al tissue [6]. Transient elastography accurately estimates and stages liver fibrosis, so that it is considered a surrogate for liver biopsy from this point of view. Transient elastography has been extensively validated over the last years, mainly in patients with chronic hepatitis C and it has been proved that its best accuracy is for detecting liver cirrhosis [7]. In the setting of liver cirrhosis, increased LS is associated with episodes of decompensation (high grade oesophageal varices - EV and/or bleeding, development of ascites) as well as with the presence of HCC [8], meaning that increased LS value alone cannot be an accurate predictor for HCC. The main drawbacks of TE are the technique difficulties in evaluating obese patients and the overestimation of liver fibrosis stage due to inflammation [9-13], central venous pressure [14], or cholestasis [15]. Transient elastography appears to be also influenced by the confounding effects resulting from attenuation and propagation of the acoustic waves in visco-elastic materials [11]. The aim of the present study was to study the performance of LS measurement data and of common biochemical parameters for the diagnosis of HCC in HCV related liver cirrhosis. MATERIAL AND METHODS Patients A cohort of consecutive patients previously diagnosed with HCV liver cirrhosis (either biopsy proven, or having unequivocal clinical, biological and imaging features) was retrospectively included in the study according to following criteria: age between 18 and 80 years, positive anti-hcv antibodies for at least 6 months, positive HCV RNA and compensated disease (Child-Pugh A or B with no ascites). All patients were naïve for the antiviral treatment. We excluded from the study patients with other etiology of chronic liver diseases such as: hepatitis B or HIV coinfection, ethanol induced or autoimmune liver disease, Wilson s disease, hemochromatosis or α1-antitripsin deficiency. They were matched by sex, age, BMI and years of HCV infection with another cohort of patients having early and very early HCC (BCLC class 0 or A). The same inclusion criteria were used as far liver cirrhosis was concerned. All HCC patients were inside the Milan criteria, meaning a single tumor of 5 cm or less in diameter, or no more than three nodules, each 3 cm or less in diameter [16, 17]. The two cohorts were compared in a case-matching manner and formed together the study group. The validation group consisted of another set of consecutive HCV patients having either liver cirrhosis alone or associated with early HCC. They were prospectively included using the same inclusion criteria as the study group, in order to validate the findings that emerged from the previous analysis. In the day of their participation in the study, a blood sample was obtained from all patients after 8 hours of overnight fasting for routine biomarkers: aspartate aminotransferase (ASAT), alanine aminotransferase (ALAT), gama-glutamyltranspeptidase (GGT), alkaline phosphatase (AP), total bilirubin, platelets count, and alpha-fetoprotein (AFP). On the same day, TE and imaging investigations were performed for assessing the presence of focal lesions. Patients in the HCC group were eventually excluded, if six months after their participation in the study, they exceeded the Milan criteria (either because of the size, or the number of the focal lesions) at a follow-up ultrasound evaluation. All patients also underwent oesophagoscopy for the assessment and grading of EV either during current admission, or in the previous 6 months interval. When present, EV were graded according to their size as follows: (i) grade 1: small, straight EV; (ii) grade 2: enlarged, tortuous EV occupying less than one third of the lumen; and (iii) grade 3: large, coil-shaped EV occupying more than one third of the lumen. The study was designed in full accordance with the Declaration of Human Rights (Helsinki, 1975) and its further revisions and was previously approved by the Hospital Ethics Committee. All patients gave their written informed consent before enrolling into the study. Liver stiffness measurement Liver stiffness measurements were performed in the right liver lobe using one-dimension transient (impulsional) elastography (FibroScan, Echosens, Paris, France) following the technical background and examination procedure as previously described [18]. Liver stiffness measurement was performed only after ultrasound guidance, in order to avoid the presence of focal liver lesions into the acquisition window. The medium probe was used for all patients. The results were expressed in kilopascals (kpa). The median value of 10 successful measurements was kept as a representative of the LS, according to the manufacturer s recommendations and previous evidence [19]. The success rate was calculated as the number of validated LS measurements divided by the number of total measurements. Only examinations with a success rate of at least 60% were further analysed [12]. Since we wanted to fully investigate the influence of stiffness related parameters, the interquartile range (IQR) lower than 30% of the median value was not considered an exclusion criteria, as previous studies suggested [20, 21]. All LS measurements were performed by experienced operators (DF, ML), with more than 2,000 examinations performed at the time of the study in patients with chronic liver diseases. Statistical analysis Mean LS values for the two groups were compared and introduced in a regression model of diagnosis. Data was analysed using MedCalc 9.3.9.0. package for Windows. Data in text and tables are reported as mean (95%CI) or percentage (%). Demographic variables were assessed using descriptive statistics. Differences in mean values were tested using the t-test. The predictive value of biochemical and imaging parameters on HCC diagnosis was investigated through uniand multivariate regression analysis. Variables that showed a significant relationship (p<0.05) with the presence of HCC in univariate analysis were included in multivariate regression analysis, in order to evaluate their simultaneous influence. We computed the model s R-squared, the standardized and unstandardized weights, partial correlation coefficients and the p-values for all the variables included in the analysis. The relationship between different parameters and tumor size

Sero-elastographic detection model for HCC in HCV related cirrhosis 285 was tested through non-parametric Spearman correlation coefficients (r). The diagnostic performance of the LS and the regression model were assessed using the receiver operating characteristic (ROC) curves analysis. Optimal cut-off values were chosen by using a common optimization step that maximized the Youden index for predicting HCC. The sensitivity (Se), specificity (Sp), positive predictive value (PPV), negative predictive value (NPV) and likelihood ratios (LR) were computed from the same data, without further adjustments. A p value of less than 0.05 was considered to indicate statistical significance. RESULTS Each cohort of the case-matching study consisted of 72 HCV cirrhotic patients, with or without HCC nodules. Another 40 HCV liver cirrhosis patients were included in the validation group. Clinical and biochemical characteristics of the study patients are summarized in Table I. Liver stiffness values in the study population In the case-matching group LS values ranged from 23 to 75 kpa, with a mean value of 27 kpa in patients without HCC, significantly lower than the mean LS value in those with HCC - 42 kpa (p<0.001), as shown in Fig. 1. Liver stiffness alone showed a poor performance for the diagnosis of HCC, as it results from the ROC curve analysis (Fig. 2A). Regression analysis for HCC detection In the univariate analysis ALAT, AP, GGT, platelets count, AFP, large oesophageal varices, LS and IQR were statistically significantly different in patients with HCC vs those without (Table I). In the multiple regression analysis, only four variables statistically correlated with the presence of HCC: LS, IQR, AFP and ALAT, of which LS had the highest odds ratio (OR) (Table II). The four parameters together explained 64.5% of the variance of the HCC (R2=0.645, p<0.0001) and uniquely explained 45.45% of the amount of R2, with LS making the largest unique contribution; the difference of 19.05% (64.5% - Fig. 1. Box plots of liver stiffness values according to HCC presence. The top and the bottom of the boxes are the first and third quartiles, respectively. The length of the box thus represents the interquartile range within which 50% of the values were located. The line through the middle of each box represents the median. The error shows the minimum and maximum values (range). 45.45%) was accounted for by the joint contribution of the four parameters. Using the coefficients obtained in the regression analysis, the prediction model computed from this study can be expressed as follows: Regression equation (LogitHCC)=1.92*LS+1.49*IQR+0.04 *AFP+0.02*ALAT -3.54 Analysing the performance of the predictive model for assessing the presence of HCC in the study population using the ROC curve method, we obtained a cut-off value of 0.5, a probability of 86% that a randomly selected individual from the HCC group has a test result indicating higher suspicion than that for a randomly chosen individual from the negative group (Fig. 2B, Table III). We also analysed the relationship between the HCC size and each parameter included in the regression equation and we found a high positive correlation for LS (r=0.77, p<0.0001) and IQR (r=0.71, p<0.0001) and a moderate but significant correlation with the values of AFP (r=0.42, p=0.03). Fig. 2. Comparative analysis of LS alone and prediction model computed based on the regression equation for assessing the presence of HCC in the study population A: Receiver operating characteristic (ROC) curve for LS as unique predictor of HCC. B: Receiver operating characteristic (ROC) curve for the regression model in HCC assessment.

286 Feier et al Table I. Baseline characteristics of the patients included in the study* Patients Case-matching group characteristics Liver cirrhosis without HCC Liver cirrhosis with HCC p Validation group Nr. 72 72 1 40 Age 62.23 (59.21-66.25) 62.23 (59.21-66.25) 1 63.21 Male sex 48.61% 48.61% 1 54% BMI 27.43 (24.11 29.44) 27.21 (24.21 29.11) 0.72 26.4 ASAT (U/l) 89.26 (74.11-100.40) 92.26 (76.81-107.67) 0.15 88.21 ALAT (U/I) 77.38 (63.72-91.04) 98.36 (82.33-114.39) <0.0001 81.22 AP (U/I) 270.07 (241.32-298.82) 358.72 (291.53-425.91) <0.0001 220.21 GGT (U/I) 116.82 (81.05-152.59) 151.85 (110.34-193.36) <0.0001 88 Total bilirubin (mg/dl) 0.98 (0.86-1.09) 1.73 (1.45-2.02) <0.0001 0.76 Platelets (10 3 /mm 3 ) 125.38 (110.84-39.91) 124.35 (96.88-151.82) <0.0001 132 AFP (U/I) 16.38 (12.18-20.57) 119(28.77-209.28) 0.0001 78.23 Presence of HCC Nr. of patients (%) 0 (0) 72 (100) 21(52.5) Nr. of nodules - 105 65 Mean diameter (mm) - 24.12 (14.21-38.33) 26.11 Oesophageal varices grade 1 (%) 14 (19.44) 9 (12.5) 0.004 9 (22.5) grade 2,3 (%) 4 (5.55) 10 (13.88) 14 (35) Liver stiffness measurement LS (kpa) 26.90 (23.97-29.84) 41.75 (36.04-47.46) <0.0001 37 IQR% 22.14 (17.88-26.39) 39.15 (24.98-53.32) 0.001 32.4 Success rate 86 (91-100) 83.81 (81-100) 0.6 91.2 * Data are expressed as absolute numbers, percentage or mean values. CI - confidence interval, ASAT - aspartate aminotransferase, ALAT - alanine aminotransferase, AP alkaline phosphatase, GGT - gamma-glutamyltranspeptidase, AFP - alpha-fetoprotein, LS - liver stiffness, IQR - Interquartile range Table II. Parameters independently associated with presence of HCC Main OR* 95%CI Coefficient Std. Error P predictors LSM 8.27 2.58-26.46 1.92 0.018 0.0001 IQR 1.16 1.04-1.29 1.49 0.055 0.0001 ALAT 1.01 0.96 0.99 0.02 0.009 0.004 AFP 1.04 1.01-1.08 0.04 0.01 0.009 OR Odds Ratio, CI - confidence interval, Std. Error- Standard Error, LS - liver stiffness, IQR - Interquartile Range, ALAT - alanine aminotransferase, AFP alpha-fetoprotein * OR for the independent variables gives the relative amount by which the odds of HCC increase or decrease when the value of the independent variable is increased by 1 unit Validation of the HCC model In the validation group, the mean LS value was 37 kpa, ranging between 23 and 75 kpa. Assessing the performance of the predictive model in the validation population and plotting the ROC curve for the prediction of HCC, using the same cut-off value, we still obtained a good AUROC and comparable Se, Sp, PPV and NPV figures (Fig. 3, Table III). Fig. 3. Validation (ROC curve analysis) of the prediction model in a new set of patients. DISCUSSION The present paper confirms the current knowledge that increased LS, AFP and serum transaminases are predictors of the presence of HCC in HCV liver cirrhosis patients. Besides

Sero-elastographic detection model for HCC in HCV related cirrhosis 287 Table III. The diagnostic performance of the regression model for the prediction of HCC in the study group and the validation group. HCC vs no HCC study group HCC vs no HCC validation group Se (%) (95%CI) 73.81 (58-86.1) 78.05 (62-89.1) Sp (%) (95%CI) 96.15 (86.8-99.5) 69.70 (59.8-84.5) +LR 34.63 28.63 -LR 0.27 0.31 PPV (%) 97.18 72 NPV (%) 78.07 76 AUROC (95%CI) 0.86 (0.76-0.92) 0.80 (0.65-0.89) Standard Error 0.04 0.05 p <0.0001 <0.0001 Se -sensitivity, Sp - specificity, PPV- positive predictive value, NPV - negative predictive value, CI - confidence interval, +LR - Positive likelihood ratio, -LR - Negative likelihood ratio, AUROC - area under ROC curve, SE - standard error that, we demonstrated that increased IQR might be also a predictor for HCC, raising the question of a stiffness shadow generated by HCC into the cirrhotic liver. We also managed to compute a prognostic model based on regression analysis, that predicted the presence of HCC in both training and validation cohort with good accuracy. Liver stiffness measurement using TE is recognised as accurately assessing the stage of liver fibrosis in patients with chronic HCV infection [12, 18], recorded values increasing as the liver disease progresses, and the highest LS being specific for cirrhotic patients with associated HCC [10]. Therefore, it was thought that LS measurement might be a valuable noninvasive tool for assessing the presence of, or the risk of developing liver cancer. The first studies reported a cut-off value of 53.7 kpa as suggestive for the presence of HCC in HCV cirrhotic patients [10], especially if total serum bilirubin was higher than 1.0 mg/dl [20]. However, the proposed threshold was not independently validated, and ranged between 12.5 and 53.7 kpa in other reports [5, 20, 21]. In Foucher s study [10] the 53.7 kpa value was obtained from a subset of only 19 patients with HCC, and the Se% and Sp% were 37% and 87%, respectively, with PPV and NPV of 30% and 90%. In addition, the study population in all mentioned papers was heterogeneous (HBV, HCV and ethanol), whereas our study included only HCV patients. For all these reasons we chose to determine our own cut-off value (38.5 kpa). Optimal cut-off values were chosen by using a common optimization step that maximized the Youden index and Se, Sp, PPV and NPV were computed from the same data, without further adjustments. Nevertheless, LS assessed by magnetic resonance elastography is not associated with the development of HCC in subjects with compensated cirrhosis [22]. None of these studies have been designed to evaluate the accuracy of more predictive parameters, in addition to LS. Therefore, we analysed a series of simple and widely available biochemical parameters together with LS and the variance of the measurement (IQR) and we performed a logistic regression analysis. Between the investigated parameters, LS measurement is the most important predictor, as suggested by the squared partial correlation. Taking into account the cut-off values obtained for LS measurements alone, we can suggest that a LS greater than 38 kpa in an HCV infected cirrhotic patient is an important indicator to further investigate the presence of a liver tumour (OR=8.27). Another important and surprising parameter was the IQR percentage of LS measurements. An IQR greater than 39% of median LS value resulted in being a good indicator and an important predictor of HCC presence. Interquartile range represents the interval including 50% of patients above and below the median. According to the manufacturer s suggestion, the IQR/LS should be less than 30% of the median value [19]. Only a few studies investigated this issue, therefore the interpretation of results is derived more from personal experience and from the manufacturer s advice. However, even lower values (<20%) [23] have been proposed in order to obtain a better concordance with liver biopsy. If gross technical operation errors are eliminated, than an IQR value greater than 30% indicates that structure disintegration is present in the liver. By finding a significantly higher IQR value (39.15% vs. 22.14%; p=0.001) in the HCC group, it means that this tissue unrest is due to the presence of HCC itself. It cannot be stated that HCC increases the LS because of its structure, as it is well known that liver cancer is a soft tumour [24]. More probably, it is the shear stress produced by the chaotic growth of the tumour inside of an already hard medium (the cirrhotic tissue) that induces the LS and the IQR increment. This finding supports the hypothesis of Mueller and Sandrin, according to which not only matrix but also pressure-associated conditions influence LS [25]. This inhomogeneous distribution of LS appears to generate a stiffness shadow, a finding that may have important clinical implications: if a certain patient with HCV cirrhosis, followed up by TE among other methods, suddenly develops an increase in LS median value, as well as in IQR, it may indicate the presence of HCC and may speed up the referral to other diagnostic techniques. In our study, the presence of HCC was associated with high ALAT values, as previously reported [26], an association stronger if the ALAT increase is persistent [27]. Surprisingly, we found no correlation between large (grade 2 and 3) EV and the presence of HCC, although it is known that the two conditions are usually associated and each of them predict a poor outcome of the other [28]. As a common practice, AFP is widely used for HCC diagnosis, despite its low sensitivity and specificity [29]. However, high AFP levels at baseline were associated with an increased risk for HCC in HCV infected patients [30]. In our study, increased AFP was an independent predictor of HCC presence, but not an important variable in the regression model. Our findings support the idea that each independent predictor, LS, IQR, ALAT and AFP, have a large role in the detection of HCC. The prognostic model is based on logistic regression, because this analysis finds the best fitting (yet biologically reasonable) model to describe the relationship between the outcome (HCC) and all independent variables. The four identified parameters predicted 64.5% of the HCC, with LS having the highest predictive power. The size and the direction of the relationship suggest that higher LS values are obtained for patients with HCC. The prognosis model using this formula had a good diagnostic performance, with an AUROC of 0.86 in the training set, and 0.8 in the validation one.

288 Feier et al The efficacy of a predictor should be assessed under ideal conditions, which we tried to reproduce in a case-matching study by assuring a homogeneous background noise in the two groups [31]. By trying to generate the same baseline conditions, we also overcame some of the critical aspects of case-control studies: the selection of cases, recall bias, and surveillance bias. Therefore, we enrolled only patients under surveillance according to the latest recommendations [2, 32] and with the same frequency. Additionally, we matched the patients according to age, sex, BMI and history oh HCV infection - because recent reports suggest a direct correlation between these parameters and the presence of HCC [33-36]. In this way, we could overcome the issue of BMI, which strongly determines the outcome of LS measurements [37]. To our knowledge, this is the first study that reports the combined role of LS measurement (median value and variance - IQR) and serum markers (AFP and ALAT) in the prediction of HCC in chronic HCV patients, with good performance. However, the most important finding of the study seems to be the increased IQR as a predictor of HCC presence. Since TE is a common examination in outpatient settings, the finding of an elevated LS value (e.g. >38 kpa, as our data suggest) as well as of an increased IQR during the follow-up of a patient with compensated liver cirrhosis may raise awareness for early HCC and may hasten the referral to confirmatory imaging techniques. This will not become a screening method, but will definitely improve the screening performance, by selecting the high risk patients for further confirmatory investigations. CONCLUSION In patients with HCV related cirrhosis, the presence of HCC is associated with increased LS and IQR values, as well as with high ALAT and AFP serum levels. Combining these four parameters into a model based on logistic regression analysis, liver cancer may be predicted with good accuracy. Stiffness changes, however, seem to be more complex, generating a shadow that might be characteristic for HCC, but this finding needs to be further investigated. Nevertheless, we believe that such an approach could be used to identify patients at high risk or, on the contrary, patients with no risk of HCC in order to limit the screening and decrease the number of useless procedures. Acknowledgements: This work was partially funded by the European Social Fund through the POSDRU 107/1.5/S/78702 Project. The authors would like to thank Dr. Bogdan Procopet for the stimulating discussions and valuable ideas that he offered during preparation of the manuscript. Conflicts of interest: None to declare. REFERENCES 1. Lok AS, Seeff LB, Morgan TR, et al. Incidence of hepatocellular carcinoma and associated risk factors in hepatitis C-related advanced liver disease. Gastroenterology 2009;136:138-148. 2. Bruix J, Sherman M; American Association for the Study of Liver Diseases. Management of hepatocellular carcinoma: an update. Hepatology 2011;53:1020-1022. 3. Kanwal F, Hoang T, Kramer JR, et al. Increasing prevalence of HCC and cirrhosis in patients with chronic hepatitis C virus infection. Gastroenterology 2011;140:1182-1188.e1. 4. Wang CH, Mo LR, Chang KK, Lin RC, Kuo JJ. A cohort study to investigate hepatocellular carcinoma risk in hepatitis C patients. Hepatogastroenterology 2011;58:904-908. 5. Nahon P, Kettaneh A, Lemoine M, et al. Liver stiffness measurement in patients with cirrhosis and hepatocellular carcinoma: a case-control study. Eur J Gastroenterol Hepatol 2009;21:214-219. 6. Sandrin L, Fourquet B, Hasqueoph JM, et al. Transient elastography: a new noninvasive method for assessment of hepatic fibrosis. Ultrasound Med Biol 2003; 29:1705 1713. 7. Friedrich Rust M, Ong MF, Martens S, et al. Performance of transient elastography for the staging of liver fibrosis: a meta-analysis. Gastroenterology 2008;134:960-974. 8. Masuzaki R, Tateishi R, Yoshida H, et al. Prospective risk assessment for hepatocellular carcinoma development in patients with chronic hepatitis C by transient elastography. Hepatology 2009;49:1954-1961. 9. Tapper EB, Cohen EB, Patel K, et al. Levels of alanine aminotransferase confound use of transient elastography to diagnose fibrosis in patients with chronic hepatitis C virus infection. Clin Gastroenterol Hepatol 2012;10:932-937.e1. 10. Foucher J, Chanteloup E, Vergniol J, et al. Diagnosis of cirrhosis by transient elastography (FibroScan): a prospective study. Gut 2006;55:403-408. 11. Ganne-Carrie N, Ziol M, de Ledinghen V, et al. Accuracy of liver stiffness measurement for the diagnosis of cirrhosis in patients with chronic liver diseases. Hepatology 2006;44: 1511-1517. 12. Kettaneh A, Marcellin P, Douvin C, et al. Features associated with success rate and performance of FibroScan measurements for the diagnosis of cirrhosis in HCV patients: a prospective study of 935 patients. J Hepatol 2007;46:628-634. 13. Lupşor M, Badea R, Stefănescu H, et al. Analysis of histopathological changes that influence liver stiffness in chronic hepatitis C. Results from a cohort of 324 patients. Gastrointestin Liver Dis 2008;17:155-163. 14. Millonig G, Friedrich S, Adolf S, et al. Liver stiffness is directly influenced by central venous pressure. J Hepatol 2010;52:206-210. 15. Millonig G, Reimann FM, Friedrich S, et al. Extrahepatic cholestasis increases liver stiffness (FibroScan) irrespective of fibrosis. Hepatology 2008;48:1718-1723. 16. Mazzaferro V, Regalia E, Doci R, et al. Liver transplantation for the treatment of small hepatocellular carcinomas in patients with cirrhosis. N Engl J Med 1996;334:693-699. 17. Singal A, Volk ML, Waljee A, et al. Meta-analysis: surveillance with ultrasound for early-stage hepatocellular carcinoma in patients with cirrhosis. Aliment Pharmacol Ther 2009;30:37-47. 18. Ziol M, Handra-Luca A, Kettaneh A, et al. Noninvasive assessment of liver fibrosis by measurement of stiffness in patients with chronic hepatitis C. Hepatology 2005;41:48 54. 19. Castera L, Forns X, Alberti A. Non-invasive evaluation of liver fibrosis using transient elastography. J Hepatol 2008;48:835 847. 20. Akima T, Tamano M, Hiraishi H. Liver stiffness measured by transient elastography is a predictor of hepatocellular carcinoma development in viral hepatitis. Hepatol Res 2011;41:965-970.

Sero-elastographic detection model for HCC in HCV related cirrhosis 289 21. Kuo YH, Lu SN, Hung CH. Liver stiffness measurement in the risk assessment of hepatocellular carcinoma for patients with chronic hepatitis. Hepatol Int 2010;4:700-706. 22. Anaparthy R, Talwalkar JA, Yin M, Roberts LR, Fidler JL, Ehman RL. Liver stiffness measurement by magnetic resonance elastography is not associated with developing hepatocellular carcinoma in subjects with compensated cirrhosis. Aliment Pharmacol Ther 2011;34:83-91. 23. Lucidarme D, Foucher J, Le Bail B, et al. Ratio interquartile range / median value of liver stiffness measurement is a key factor of accuracy of transient elastography (FIBROSCAN ) for the diagnosis of liver fibrosis. Hepatology 2007;46 Suppl:A318. 24. Bruix J, Castells A, Bosch J, et al. Surgical resection of hepatocellular carcinoma in cirrhotic patients: prognostic value of preoperative portal pressure. Gastroenterology 1996;111:1018-1022. 25. Mueller S, Sandrin L. Liver stiffness: a novel parameter for the diagnosis of liver disease. Hepatic Medicine: Evidence and Research 2010;2:49 67. 26. Kobayashi M, Suzuki F, Akuta N, et al. Development of hepatocellular carcinoma in elderly patients with chronic hepatitis C with or without elevated aspartate and alanine aminotransferase levels. Scand J Gastroenterol 2009;44:975-983. 27. Miyakawa K, Tarao K, Ohshige K, et al. High serum alanine aminotransferase levels for the first three successive years can predict very high incidence of hepatocellular carcinoma in patients with Child Stage A HCV-associated liver cirrhosis. Scand J Gastroenterol 2009;44:1340-1348. 28. Chang CJ, Hou MC, Liao WC, Lee FY, Lin HC, Lee SD. Risk factors of early re-bleeding and mortality in patients with ruptured gastric varices and concomitant hepatocellular carcinoma. J Gastroenterol 2012;47:531-539. 29. Zhang G, Ha SA, Kim HK, et al. Combined analysis of AFP and HCCR-1 as an useful serological marker for small hepatocellular carcinoma: a prospective cohort study. Dis Markers 2012;32:265-271. 30. Tateyama M, Yatsuhashi H, Taura N, et al. Alpha-fetoprotein above normal levels as a risk factor for the development of hepatocellular carcinoma in patients infected with hepatitis C virus. J Gastroenterol 2011;46:92-100. 31. Altman DG. Practical statistics for medical research. London: Chapman &Hall Gardner MJ, Altman DG. Gli intervalli di confidenza. Rome: Il pensiero scientifico editore 1991. 32. European Association For The Study Of The Liver; European Organisation For Research And Treatment Of Cancer. EASL EORTC clinical practice guidelines: management of hepatocellular carcinoma. J Hepatol 2012;56:908 943. 33. Walter SR, Thein HH, Gidding HF, et al. Risk factors for hepatocellular carcinoma in a cohort infected with hepatitis B or C. J Gastroenterol Hepatol 2011;26:1757-1764. 34. Wang CH, Mo LR, Chang KK, Lin RC, Kuo JJ. A cohort study to investigate hepatocellular carcinoma risk in hepatitis C patients. Hepatogastroenterology 2011;58:904-908. 35. Kanwal F, Hoang T, Kramer JR, et al. Increasing prevalence of HCC and cirrhosis in patients with chronic hepatitis C virus infection. Gastroenterology 2011;140:1182-1188. 36. Akiyama T, Mizuta T, Kawazoe S, et al. Body mass index is associated with age-at-onset of HCV-infected hepatocellular carcinoma patients. World J Gastroenterol 2011;17:914-921. 37. Castéra L, Foucher J, Bernard PH, et al. Pitfalls of liver stiffness measurement: a 5-year prospective study of 13,369 examinations. Hepatology 2010;51:828-835.