Michel BEAUGRAND

EVALUATION OF LIVER FIBROSIS BLOOD TESTS, LIVER BIOPSY AND FIBROSCAN M. Beaugrand Service d’Hépatologie Hopital J. Verdier BONDY 93143 et Université Paris XIII PHC JANUARY 07

EVALUATION OF FIBROSIS : WHY ? Causal agent Chronic liver Disease Cirrhosis Decompensation Hepatocellular 2 % to 5 % per year carcinoma 2 % to 5 % per year Fibrosis Distorted architecture Portal hypertension Liver failure Carcinogenesis FIBROSIS Death 4 % per year

ASSESSMENT OF FIBROSIS : WHY ? Management of individual patients Significant fibrosis Treatment Cirrhosis Screening for varices and HCC Evaluation of treatments Antiviral and antifibrotic drugs Screening for cirrhosis or extensive fibrosis In high risk patients

EVALUATION OF FIBROSIS : HOW ? - Liver biopsy - Blood tests Genuine serum markers of fibrosis : by products of extracellular matrix metabolism. Probabilistic indexes = surrogate markers. - Fibroscan ( transcient elastography)

LIVER BIOPSY : LIMITATIONS Acceptability - patients are often reluctant - even some doctors are reluctant Cost Morbidity Reliability - liver sample size ideally ≥ 25 mm - pathologist’s experience - suboptimal reproductibility of scoring systems - quantitative assessment unpractical

LENGH OF LIVER BIOPSY Biopsy length Bedossa et al, Hepatology 2003 Liver biopsy is the gold standard for evaluating liver fibrosis but it has some limitations. A shown by bedossa et al using virtual biopsy from liver explants, a 25 mm long biopsy correctly staged in 75% of the cases and 40 mm were necessary to obtain a 100%reliability. Biopsy length Bedossa et al, Hepatology 2003

LIVER BIOPSY : PROS AND CONS . Not influenced by extrahepatic conditions . Allows analysis of elementary lesions and comobidities . May allow assessment of fibrogenesis (molecular biology) CON . Sampling error . Dependant of pathologist’s experience . Unavalaible in large parts of the world

BLOOD TESTS Matrix related PIIINP, collagen type IV, laminin Hyaluronic acid, MMP, TIMP Non maxtrix related AST, ALT, gamma GT Bilirubin, prothrombine, platelets Gammaglobulins, ferritin Alpha 2 macroglobulin, haptoglobin Apo A1, cholesterol, HOMA

BLOOD TESTS Poynard, 1991 Prothrombine, GGT, Apo A1 PGA Bonacini, 1997 AST/ALAT, platelets, prothrombine Imbert-Bismut, 2001 bili, GGT, hapto., a2MG, apoA1 Fibrotest Luo, 2002 glob/alb, platelets, AST/ALT Forns, 2002 age, GGT, cholesterol, platelets Kaul, 2002 sex, ang.spider angiomas, AST, platelets Wai, 2003 AST/platelets APRI Sud, 2004 age, AST, cholesterol, HOMA, alcohol Lainé, 2004 hyaluronate, transferin Rosenberg, 2004 age, hyalur., col IV, col VI, laminin, PIIINP, TIMP-1 ELF Patel, 2004 TIMP-1, a2MG, hyaluronate Fibrospect Leroy, 2004 PIIIMP, MMP1 Hui, 2005 BMI, platelets, albumin, bilirubin Lok, 2005 AST/ALT, platelets, INR Adams, 2005 bili, GGT, hyaluronate, a2MG, age, sex Hepascore Cales, 2005 AST, platelets, prothrombine, hyaluronate, Fibrometre urea, age

Imbert-Bismut et al. Lancet 2001;357:2069-75 BLOOD TESTS Fibrotest ® Metavir index VHC (n= 339) Imbert-Bismut et al. Lancet 2001;357:2069-75

Imbert-Bismut et al. Lancet 2001;357:2069-75 BLOOD TESTS Fibrotest ® HCV (n=339) Métavir ≥ F2 AUC = 0,827 Index < 0,10 NPV 100% Index > 0,60 PPV > 90% Liver Biopsy: - 46% Imbert-Bismut et al. Lancet 2001;357:2069-75

BLOOD TESTS Ref. Test AUC Construction Validation Imbert-Bismut, 2001 HCV : F0-1 vs F2-3-4 Ref. Test AUC Construction Validation Imbert-Bismut, 2001 Fibrotest 0.83 0.85 Forns, 2002 0.86 0.81 Wai, 2003 APRI 0.80 0.88 Patel, 2004 Fibrospect Rosenberg, 2004 ELF 0.78 Leroy, 2004 0.82 Sud, 2004 0.84 Adams, 2005 Hepascore

HCV : F0-1-2-3 vs F4 (cirrhosis) BLOOD TESTS HCV : F0-1-2-3 vs F4 (cirrhosis) Ref. Test AUC Construction Validation Kaul, 2002 0.94 Wai, 2003 APRI 0.89 Rosenberg, 2004 ELF Le Calvez, 2004 Fibrotest 0.92 Adams, 2005 Hepascore Lok, 2005 0.78 0.81

BLOOD TESTS : PROS AND CONS . Easy, non invasive,not too costy . Allow to split patients in 3 groups - those without significant fibrosis - those with extensive fibrosis or cirrhosis - intermediate CON . Studied mainly in naive patients with HCV chronic hepatitis . External validation lacking (except fibrotest) . Require standardisation of biochemical methods . Poorer performances in HBV patients or coinfected HCV-HIV . Possible influence of extrahepatic conditions . No international consensus

ELASTOMETRY (FIBROSCAN) 2.5 cm 4 cm 1 cm  Volume LB x 100 Probe Sandrin et al. Ultrasound Med Biol 2003;29:1-8

HOW TO MEASURE ELASTICITY ? To generate an elastic Shear vawe To measure its spead Vs Elasticity E  VS2

Volume of exploration > 3 cm3 Probe position 2.5 cm 4 cm 1 cm  Volume of exploration Probe

PATIENTS WITH HCV CHRONIC HEPATITIS + patients with no ascites 251 patients included 53 patients excluded biopsy not suitable for fibrosis stage assessment: less than 10 portal tracts in the absence of cirrhosis 23 patients excluded: unreliable stiffness measurement: success rate less than 60% upon 10 measurements Small biopsy 126 patients Large biopsy 125 patients 327 patients consecutive patients were included from 4 french liver units. There was one exclusion criteria which was the presence of ascitis because elastic waves do not propagate through liquids. 23 stiffness meazsurement were unsuccessfull (too narrow intercostal space or too wide adipose tissue). 53 patients were further exlcluded because the biopsy was too small to accurately grade fibrosis

Fibrosis stage (METAVIR) BOX PLOTS. N=251 100 Legend maximum Stiffness (kPa) (logarithmic scale) 10 median IQR On this figure, median value of liver stiffness are compared with consensus METAVIR stage. Liver stiffness was significantly correlated to fibrosis stage You can see that patients with F4 metavir score which means cirrhosis are well separated from other stages but overlaps are observed between F1 and 2 and F2 and F3. minimum 1 0-1 2 3 4 Fibrosis stage (METAVIR)

ROC CURVES - F≥2 : 0.79 (0.73-0.84) AUROC ( CONFIDENCE INTERVALS 95%)

UNI AND MULTIVARIATE ANALYSIS Univariate analysis (Kendall’s coefficient) Fibrosis Activity Steatosis Stiffness r 0.55 0.21 0.19 p <0.0001 0.0003 0.0008 Fibrosis r - 0.36 0.17 p - <0.0001 0.008 On the other hand we checked that activity (which represents inflammatory infiltrate) and steatosis did not significantly affect liver stiffness measurements. I Multivariate analysis (multiple regression) Only fibrosis was significantly correlated to liver stiffness measurement.

VALIDATION OF DIAGNOSIS ACCURACY IN AN INDEPENDENT HCV POPULATION Total number of included patients: 639 Number of unreliable liver samples: 86 (13%) Number of unreliable LSM: 59 (9%) Patients kept for statistical analysis : 494 Area under ROC curves (95% confidence interval) F01 versus F234 = 0.84 (0.80-0.87) F012 versus F34 = 0.93 (0.90-0.95) F0123 versus F4 = 0.96 (0.94-0.98) METAVIR F 1 2 3 4 % 6 39 31 10 14 A 56 35 Steatosis S 1-10 11-30 31-100 47 27 15 Univariate Spearman correlation METAVIR F: 0.70 (p << 0.001) METAVIR A: 0.45 (p << 0.001) Steatosis: 0.35 (p << 0.001)

LIVER BIOPSIES > 30 mm - 103 Patients - Results Causes : 71 VHC 14 VHB 15 VHC+HIV 2 VHB+HIV 1 VHC+VHB Fibrosis Score : F0 F1 F2 F3 F4 N 9 58 14 7 15 - Results  F2  F3 = F4 AUROC 0.94 0.95 0.93

CUT-OFF VALUES* The optimum thresholds were chosen to maximize the sum of sensitivity and specificity. Threshold Sensitivity Specificity LR (kPa) F  2 8.7 0.55 0.84 3.5 F  3 9.6 0.84 0.85 5.7 F = 4 14.5 0.84 0.94 13.0 Choosing tresholds to maximize sensitivity and specificity 8,7 kpascal defined F superior or equal to 2 with a 55 percent sensitivity and 84% specificity etc…. * Obtained by the jack-knife method.

FIBROSIS AREA (morphometry) - Patients 69 patients with chronic hepatitis C without ascites, and previous anti-viral treatment METAVIR F1 F2 F3 F4 20 21 5 23 - Results Spearman correlation test p < 0.001 Parameters r Fibrosis area METAVIR score 0.66 Elasticity METAVIR score 0.65 Elasticity fibrosis area 0.74  Liver elasticity is closely correlated to fibrosis area.

FIBROSCAN VERSUS BLOOD TESTS FibroTest APRI Elasticité (kPa) Score METAVIR de fibrose Score METAVIR de fibrose Score METAVIR de fibrose Castera Al. Gastroenterology 2005

CONCORDANCE WITH LIVER BIOPSY AUROC F01/F234 F012/F34 F0123/F4 APRI 0.78 0.84 0.83 FibroTest 0.85 0.90 0.87 FibroScan 0.83 0.90 0.95 Combinaison 0.88 0.95 0.95 FibroTest+FibroScan Percentage of concording results F01/F234 F0123/F4 FibroTest 80 81 FibroScan 73 83 Combinaison 84 95 FibroTest+FibroScan

PROPOSED COMBINATION OF NON INVASIVE METHODS FibroScan + FibroTest Concordance Discordance Fibrose minime (FS < 7.1 kPa + FT < F2) Fibrose modérée (FS  7.1 kPa + FT  F2) Fibrose sévère (FS  9.5 kPa + FT  F3) Biopsy Follow-up Treatment Treatment HCC screening treatment Castera et Al. Gastroenterlogy 2005

FIBROSCAN IN HBV PATIENTS 15 non interpretable biopsies 14 LSM considered as non reliable Statistical analysis on 173 patients AUROC F01 versus F234: 0.81 (0.73-0.86) F012 versus F34: 0.93 (0.88-0.96) F0.123 versus F4: 0.93 (0.82-0.98)

FIBROSCAN PROS AND CONS -easy, quick, not too costy - very specific for extensive fibrosis or cirrhosis - Allows to split patients between 3 groups . Without significant fibrosis . With extensive fibrosis or cirrhosis . Intermediate - closely related to the area of fibrosis in patients with chronic hepatitis CON - high rate of failure in obese patients - doesn’t take in account liver architecture - disturbed in acute conditions

SCREENING IN HIGH RISK PATIENTS LSM Blood tests Confirmation of cirrhosis LB LSM>13 kPa oui Absence of cirrhosis Suspected cirrhosis non 227 patients in alcoholic abstinence program 41 34 33

FOLLOW-UP OF LSM IN TREATED PATIENTS WITH CHRONIC HEPATITIS C BEFORE TREATMENT END OF TREATMENT 6 MONTHS AFTER All (n=85) 14.1 ± 7.2 10.9 ± 6.5 11.2 ± 8.6 SVR 12.0 ± 6.7 9.1 ± 3.7 7.5 ± 2.4 RR 14.6 ± 5.6 11.5 ± 5.0 12.8 ± 7.2 NR 16.9 ± 8.1 13.3 ± 9.1 16.1 ± 12.2

CONCLUSION 1) Liver biopsy has been partly challenged by non invasive methods for assessment of liver fibrosis 2) Non invasive methods have their own limitations : 2 might be better than one 3) Fibroscan could become a useful tool for assessing fibrosis variations