Received: 7 October 2020 | Revised: 23 August 2021 | Accepted: 1 September 2021 DOI: 10.1111/liv.15058 ORIGINAL ARTICLE Elastography is unable to exclude cirrhosis after sustained virological response in HCV-­infected patients with advanced chronic liver disease Teresa Broquetas1,2 | Paula Herruzo-­Pino2 | Zoe Mariño3 | Dolores Naranjo4 | Mercedes Vergara5 | Rosa Mª Morillas6 | Xavier Forns3 | José A. Carrión1,2 1 Liver Section, Gastroenterology Department, Hospital del Mar, Barcelona, Abstract Spain Background: Liver fibrosis and transient elastography (TE) correlation in hepatitis 2 Universitat Autònoma de Barcelona (UAB), C virus (HCV)-­infected patients with compensated advanced chronic liver disease Departament de Medicina de la UAB, Barcelona, Spain (cACLD) after the sustained virological response (SVR) is unknown. Aims: To evaluate TE accuracy at identifying cirrhosis 3 years after HCV-­eradication. 3 Liver Unit, Hospital Clínic de Barcelona, Centro de Investigación Biomédica en Red de Enfermedades Hepáticas y Digestivas Methods: Prospective, multi-­centric study including HCV-­c ACLD patients before (CIBERehd), IDIBAPS, University of direct-­acting antivirals (DAA). Diagnostic accuracy of TE (area under ROC, AUROC) Barcelona, Barcelona, Spain 4 to identify cirrhosis 3 years after SVR was evaluated. Gastrointestinal and Hepatobiliary Pathology Section, Department of Pathology, Hospital Results: Among 746 HCV-­infected patients (95.4% with TE ≥10 kPa), 76 (10.2%) del Mar, Barcelona, Spain underwent a liver biopsy 3 years after SVR. Before treatment, 46 (63%) showed 5 Liver Unit, Digestive Disease Department, a TE>15 kPa. The TE before DAA was the best variable for predicting cirrhosis Parc Taulí Sabadell Hospital Universitari, Institut d’Investigació i Innovació Parc Taulí (METAVIR, F4) after SVR (AUROC = 0.79). Liver function parameters, serological non-­ I3PT, Centro de Investigación Biomédica invasive tests (APRI and FIB-­4), and TE values improved after SVR. However, liver en Red de Enfermedades Hepáticas y Digestivas (CIBERehd), Universitat biopsy 3 years after HCV elimination (median time = 38.4 months) showed cirrhosis in Autònoma de Barcelona, Barcelona, Spain 6 41 (53.9%). Multivariate analysis (OR (95% CI), P) showed that HCV-­genotype 3 (20.81 Hepatology Department, Hospital Germans Trias i Pujol, Centro de Investigación (2.12-­201.47), .009), and TE before treatment (1.21 (1.09-­1.34), <.001) were the only Biomédica en Red de Enfermedades variables associated with cirrhosis after SVR. However, the accuracy of TE after SVR Hepáticas y Digestivas (CIBERehd), Badalona, Spain was poor (AUROC = 0.75) and 6 (27.3%) out of 22 patients with a TE <8 kPa had cir- rhosis. Similar results were found with APRI and FIB-­4 scores. Correspondence José A. Carrión, Liver Section, Conclusions: Cirrhosis is present, 3 years after SVR, in more than half of HCV-­c ACLD Gastroenterology Department, Hospital patients even with the normalisation of liver function parameters, serological non-­ del Mar, 25-­29 Passeig Marítim; 08003 Barcelona, Spain. invasive tests and TE values. The low diagnostic accuracy of non-­invasive methods Email: 95565@parcdesalutmar.cat; jcarrion@ after SVR reinforces the need for long-­term surveillance. psmar.cat Abbreviations: ALT, alanine aminotransferase; APRI, AST to platelet ratio; AST, aspartate aminotransferase; AUROC, area under the receiver operating characteristics; BMI, body mass index; cACLD, compensated advanced chronic liver disease; CSPH, clinically significant portal hypertension, N, number; DAA, direct-­acting antiviral; EoT, end of treatment; FIB-­4, fibrosis-­4 index; HCC, hepatocellular carcinoma; HCV, hepatitis C virus; HIV, human immunodeficiency virus; INR, international normalised ratio; IQR, interquartile range; MELD, model for end-­s tage liver disease; NASH CRN, NASH clinical research network; NITs, serological non-­invasive tests; NPV, negative predictive value; PLT, platelet; PPV, positive predictive value; SAEs, serious adverse events; SVR, sustained virological response; T2DM, type 2 diabetes mellitus; TE, transient elastography; γGT, gamma-­glutamyl transpeptidase. Teresa Broquetas and Paula Herruzo-­Pino collaborated equally The corresponding author of this manuscript certifies that the contributors' and conflicts of interest statements included in this paper are correct and have been approved by all co-­authors. This is an open access article under the terms of the Creative Commons Attribution-­NonCommercial-­NoDerivs License, which permits use and distribution in any medium, provided the original work is properly cited, the use is non-­commercial and no modifications or adaptations are made. © 2021 The Authors. Liver International published by John Wiley & Sons Ltd. Liver International. 2021;41:2733–2746.  wileyonlinelibrary.com/journal/liv | 2733 2734 | BROQUETAS et al. Funding information KEYWORDS None. cirrhosis, fibrosis, hepatitis C, liver biopsy, sustained virological response, transient Handling Editor: Alessio Aghemo elastography 1 | I NTRO D U C TI O N Key points Hepatitis C virus (HCV) infection is a major public health prob- • Our study shows that neither transient elastography nor lem affecting over 71 million people worldwide, and it is responsi- serological non-­invasive tests as APRI or FIB-­4 are re- ble for more than 399 000 deaths from liver cancer and cirrhosis liable enough to accurately identify liver fibrosis stage yearly.1 after sustained virological response in patients with ad- The evaluation of liver fibrosis in patients with chronic HCV in- vanced chronic liver disease (cACLD). fection is crucial, since the late stages of the disease are related to • In spite of finding significant improvements in liver func- poorer prognosis. 2 Patients with cirrhosis can develop portal hyper- tion parameters, NITs and TE values after HCV clearance tension, oesophageal varices, hepatocellular carcinoma (HCC), and in HCV-­c ACLD patients, cirrhosis and advanced fibrosis liver failure. Therefore, they require stricter follow-­up, with abdomi- are present in 50% and 70%, respectively, after 3 years nal ultrasound every six months and upper gastrointestinal endosco- of viral elimination, being recommendable a continuous pies every 2 years. 2,3 Liver biopsy still remains the gold standard for surveillance after sustained virological response. fibrosis stage assessment, 2 with multiple staging systems, such as the METAVIR4 and Laennec5 scales. Since liver biopsy is an invasive method implying risks, the current guidelines recommend that HCV-­ infected patients with compensated advanced chronic liver disease still remain at risk of developing HCC.17 On the other hand, The (cACLD) should initially be assessed and classified using non-­invasive American Gastroenterological Association Institute Guidelines on 6 methods. Both, transient elastography (TE) that assesses liver stiff- the Role of Elastography in the Evaluation of Liver Fibrosis hy- ness, and serological non-­invasive tests (NITs) as AST to platelet pothesise that a post-­t reatment TE cut-­off of 9.5 kPa can rule out ratio (APRI) and fibrosis-­4 index (FIB-­4), have been shown to be use- advanced liver fibrosis after SVR and allow these patients to be ful at identifying patients with cirrhosis before antiviral treatment.3,7 discharged, although the level of evidence of such recommenda- The TE is the simplest, best evaluated and most widely used elas- tion is low.7 tographic technique and it is considered the standard non-­invasive The primary endpoint of our study was to evaluate TE and NITs method in clinical practice. It offers an excellent correlation with accuracy at identifying cirrhosis 3 years after HCV elimination com- liver fibrosis, well-­established quality criteria and a high diagnostic pared to liver biopsy. The secondary objectives were to assess the 4,8 reliability for identifying patients with cirrhosis. Therefore, a TE impact of HCV eradication, and the variables associated with cirrho- cut-­off value of 14 kPa identifies and excludes early cirrhosis with a sis 3 years after SVR. positive predictive value (PPV) of 80%-­90% and negative predictive value (NPV) of 94%-­96% respectively.9,10 However, the diagnostic accuracy of TE and NITs to predict residual fibrosis decreased in 2 | M ATE R I A L S A N D M E TH O DS cirrhotic patients after the sustained virological response (SVR) to interferon.11,12 2.1 | Study design and participants Direct-­a cting antiviral (DAA) therapy achieves SVR rates of >95% with only 8-­12 weeks of treatment. 3 The SVR is associated This is an open, multi-­centric, prospective study concerning four with a decrease in mortality and morbidity,13 by reducing the risk public hospitals in Catalonia: Hospital del Mar (Barcelona), Hospital of liver complications.14 The HCV elimination normalises trans- Clínic (Barcelona), Hospital Germans Trias-­Pujol (Badalona) and aminases, resolves necroinflammation, and improves TE and NITs Consorci Sanitari Parc Taulí (Sabadell). 11,12 independently of baseline fibrosis. However, in patients with Inclusion criteria were: age >18 years old, history of viremic cACLD or cirrhosis, liver remodelling has not been widely studied HCV infection with a TE >9 kPa compatible with advanced fibrosis after virological cure15 and many patients remain with portal hy- (METAVIR, F3-­4)9,10,19,20 prior to DAA treatment and having achieved pertension,16 risk of HCC17 and oesophageal varices.18 European SVR between July 2015 and July 2016. Patients were enrolled after Association for the Study of the Liver guidelines recommend that 3 years from end of treatment (EoT) (from April 2018 to April 2019) patients with suspected advanced fibrosis (METAVIR score F3-­ and agreed to have a liver biopsy and a new TE evaluation. 4),19 cACLD or cirrhosis before antiviral treatment should continue Exclusion criteria were: absence of informed consent, no to undergo abdominal ultrasounds every 6 months because they SVR, HBV-­coinfection, TE value without quality criteria, clinical BROQUETAS et al. | 2735 decompensation, HCC or comorbidities limiting survival, large oe- Laennec grading system.5 Bridging fibrosis was defined by METAVIR sophageal varices, anticoagulant treatment or severe thrombopenia score as F3 and cirrhosis as F4. Advanced fibrosis includes F3 and (platelet -­PLT-­count <60 × 109/L), or less than 3 years of follow-­up. F4. The Laennec grading system scored patients with cirrhosis as mild (F4A), moderate (F4B) and severe (F4C). Necro-­inflammatory activity was assessed according to the METAVIR score and the stea- 2.2 | Clinical variables and follow-­up tosis percentage followed the NASH CRN system.4 Serious adverse events (SAEs) associated with liver biopsy defined as any condition Data were recorded at 3 different time-­points: before DAA treat- causing an increase in hospitalisation (pain, haemoperitoneum, bil- ment (baseline), at SVR (12 or 24 weeks after the EoT) and 3 years iary leakage, surgery or death) were registered. after EoT. Variables evaluated at baseline and 3 years after EoT were anthropometric and demographic characteristics, concomitant medication, blood test, NITs (APRI and FIB-­4); Child-­Pugh and MELD 2.5 | Ethical aspects of the study (model for end-­stage liver disease) scores; TE values; indirect data of portal hypertension (ultrasound or endoscopic) and Baveno VI crite- This study was conducted in accordance with national and inter- ria to avoid endoscopy.6,18 Variables at SVR were TE and blood tests. national guidelines (deontological code, Declaration of Helsinki Included patients did undergo 6-­month abdominal ultrasounds, 1975) and legal regulations concerning confidential data (LOPD -­ regular upper endoscopy (every 12-­24 months) according to interna- Spanish Organic Law 15/1999, of 13 December, on the Protection of 21 6,18 tional recommendations and Baveno criteria. The liver biopsy Personal Data). It was approved by the ‘Comité Ético de Investigación was performed 3 years after EoT. Clínica del Parc de Salut Mar’ (Parc de Salut Mar Clinical Research Patients with a TE > 15 kPa before antiviral treatment were Ethics Committee) at the Hospital del Mar with reference number considered patients highly suggestive of cACLD,6 and those with a 2017/7492/I, and in the Ethic Committees of the rest of participant 9,10 TE ≥14 kPa with a high suspicion of cirrhosis. Patients were fol- hospitals. lowed-­up until 1 April 2020 or early discontinuation if voluntary withdrawal, de novo clinical decompensation, HCC, transplantation or death. 2.6 | Sample size and statistical analysis The sample size was calculated based on D'Ambrosio et al11 to de- 2.3 | Transient Elastography and serological non-­ termine the diagnostic accuracy of TE to identify cirrhosis after SVR invasive tests (area under the receiver operating characteristics, AUROC = 0.77) with a null hypothesis of AUROC = 0.5, and the publication of Maylin The TE measurements were made using FibroScan® (Echosens). et al27 to obtain the ratio of negative patients (without cirrhosis) and Valid measurements were those that fulfilled the quality criteria (me- positive (with cirrhosis) 3 years after SVR (ratio = 9:5). A type I (alpha) dian value of 10 valid measurements with a ratio >60% and an inter- error level of 0.05 and a type II (beta) error level of 0.10 were con- quartile range (IQR)/TE < 0.3). 22 Probe M or XL were used according sidered, so the number of patients with liver biopsy required was at to the patient's body mass index (BMI). 23 Previously published TE least 48. It was estimated that up to 90% of patients might refuse to cut-­offs suggestive of cACLD (TE = 10-­15 kPa) or highly suggestive perform the liver biopsy. Therefore, it was planned to include more (TE > 15 kPa),6 and those diagnostic of cirrhosis (TE ≥ 14 kPa)9,10 than 480 patients from April 2018 to April 2019 to demonstrate the 24 25 were evaluated. Serological NIT as APRI and FIB-­4 were tested. diagnostic reliability of TE before and after SVR. Reference cut-­offs of APRI for exclude/include cirrhosis (<1/>2)24 Quantitative variables were expressed as medians (range) and and FIB-­4 for exclude/include advanced fibrosis (<1.45/>3.25)25 qualitative variables as proportions. The differences between pa- were assessed. tients with TE suggestive of cACLD (TE 10-­15 kPa) vs. those with TE highly suggestive of cACLD (TE > 15 kPa) and patients with sus- pected cirrhosis (TE ≥ 14 kPa) vs those with TE < 14 kPa before 2.4 | Liver biopsy and histological evaluation antiviral treatment were analysed with different statistical tests depending on the characteristics of the variables: Fisher's test was Liver biopsy was performed 3 years after EoT. Biopsy was carried out used for qualitative variables, and non-­parametric tests were used with abdominal ultrasound guidance using a 16G needle, in accord- for quantitative variables (the Mann-­Whitney U test for indepen- ance with the protocol of each hospital. The samples were formalin-­ dent data and the Wilcoxon test for dependent data). Both base- fixed and paraffin-­embedded and both haematoxylin-­eosin and line and 3-­year differences among patients who had F4 after SVR 26 Masson's Trichrome stains were evaluated. Histological evaluation and those who had F0-­3 were analysed by univariate analysis. All was centralised and performed by a double-­blind reference pathol- variables with statistical significance (P < .05) were included in a ogist (DN) who did not know the patient's clinical/analytical data. multivariate progressive logistic regression analysis to determine Stage of fibrosis was assessed according to the METAVIR score 4 and independent predictors of post-­treatment cirrhosis. Diagnostic 2736 | BROQUETAS et al. accuracy to identify cirrhosis by TE and NITs (APRI, FIB-­4) before more frequent alcohol consumption, with higher median γGT val- and 3 years after antiviral treatment was evaluated with AUROC ues. However, there were no significant differences between groups curves. The AUROC curves were compared using the Hanley-­ in other variables. The proportion of missing values was 1.6% and McNeil method. 28 the mechanism was completely at random. Given the small propor- Sample size and comparisons between AUROC curves were cal- tion of missing values, no post hoc method was necessary. 29 Only ® culated with MedCalc v19.1.3 (MedCalc Software). All other analy- 2 variables had missing data above 10% (HCV viral load, and alpha-­ ses were performed with SPSS® v20.0 (SPSS Inc). fetoprotein) and were excluded from the analysis. All patients from the final sample (n = 76) were Child-­Pugh A (5) with a median MELD score of 6 (Table 1). No patients presented 3 | R E S U LT S SAEs resulting from the liver biopsy. Most of the included patients (n = 73, 96.1%) showed a TE > 10 kPa before antiviral treatment 3.1 | Baseline characteristics of the patients and were classified as patients with cACLD: 27 (37%) with TE = 10-­ 15 kPa and 46 (63%) with a TE > 15 kPa.6 A total of 746 HCV-­infected patients with TE > 9 kPa (95.4% with Table 2 shows the differences between patients with TE sug- TE ≥ 10 kPa) before DAA treatment were evaluated. Among these gestive of cACLD (TE = 10-­15 kPa) and those highly suggestive of patients, 689 (94.8%) were Child-­Pugh A and 38 (5.2%) Child-­Pugh cACLD (TE ≥ 15 kPa). Patients with a TE highly suggestive of cACLD B. The median (range) MELD was 6 (6-­36). From the total cohort, (n = 46) compared to those with only suggestive cACLD (n = 27) 670 (89.8%) patients were excluded from performing a liver bi- had higher bilirubin levels, increased hepatic echogenicity, TE and opsy 3 years after SVR. Figure 1 shows the detailed flowchart of FIB-­4 score values. Moreover, these patients showed lower platelet HCV-­infected candidates and the final sample of 76 evaluable counts and the proportion of patients who did not show the Baveno subjects. The differences between the patients with (n = 76) and VI criteria to avoid endoscopy6 was higher. There were no significant without (n = 670) a liver biopsy 3 years after EoT are depicted in differences between both groups in other variables. Similar results Table 1. Patients who underwent a liver biopsy had a higher per- were found after categorising the patients as suspected cirrhosis centage of human immunodeficiency virus (HIV) co-­infection and with a TE ≥ 14 kPa9,10 (Table S1). FIGURE 1 Study flowchart BROQUETAS et al. | 2737 TA B L E 1 Baseline characteristics before antiviral treatment (N = 746) No liver Biopsy, n = 670 Liver Biopsy, n = 76 Characteristics N (%) All patients, N = 746 (89.8) (10.2) P value Males. n (%) 415 (55.7) 367 (54.9) 48 (63.2) NS Age. y 63 (18-­87) 64 (18-­87) 59 (25-­8 0) NS BMI, kg/m2 (n = 656) 26.5 (13.1-­46.4) 26.4 (13.1-­46.4) 26.6 (16.8-­41.4) NS 2 Obesity (BMI > 30 kg/m ), n (%) 161 (24.5) 139 (23.9) 22 (29.3) NS T2DM, n (%) (n = 743) 162 (21.8) 145 (21.7) 17 (22.4) NS Arterial hypertension, n (%) (n = 715) 206 (28.8) 186 (29.0) 20 (27.0) NS Alcohol abuse, n (%) (>20/30 g/dL) 54 (7.3) 48 (7.2) 6 (7.9) .048 (n = 744) HIV-­coinfection, n (%) (n = 743) 30 (4.0) 21 (3.1) 9 (12) .002 HCV viral load, Log10 IU/mL (n = 737) 6.1 (1.9-­7.8) 6.1 (1.9-­7.8) 6.1 (3.95-­7.38) NS HCV genotype, n (%) (n = 737) NS 1a 114 (15.5) 95 (14.4) 19 (25) 1b 483 (65.5) 446 (67.5) 37 (48.7) 2 18 (2.4) 16 (2.4) 2 (2.6) 3 72 (9.8) 62 (9.4) 10 (13.2) 4 50 (6.8) 42 (6.4) 8 (10.5) TE before treatment, kPa (n = 715) 16.0 (9.1-­75.0) 15.7 (9.1-­75.0) 19.9 (9.6-­43.6) NS <10 kPa, n (%) 33 (4.6) 30 (4.7) 3 (3.9) NS 10-­15 kPa, n (%) 302 (42.2) 275 (43) 27 (35.5) >15 kPa, n (%) 380 (53.1) 334 (52.3) 46 (60.5) Baveno VI criteria, n (%) (n = 742) NS No (TE > 20 or PLT < 150) 500 (67.6) 444 (66.7) 56 (73.7) Yes (TE < 20 and PLT > 150) 242 (32.6) 222 (33.3) 20 (26.3) Abdominal ultrasound, n (%) (n = 733) Nodular liver surface 84 (11.5) 71 (10.8) 13 (17.1) NS Spleen size >12 cm 305 (41.8) 275 (42.1) 30 (39.5) NS Increased hepatic echogenicity 144 (19.6) 130 (19.8) 14 (18.4) NS Oesophageal varices, n (%) (n = 384) 194 (50.5) 171 (51.5) 23 (44.2) NS AST, IU/L (n = 736) 66.0 (14-­670) 66 (14-­670) 64.5 (18-­333) NS ALT, IU/L (n = 740) 71.0 (6-­632) 70 (6-­632) 76.5 (16-­369) NS γGT, IU/L (n = 742) 75.5 (13-­1867) 73 (13-­1867) 98.5 (18-­830) .002 Bilirubin, mg/dL (n = 741) 0.7 (0.2-­9.9) 0.7 (0.2-­9.9) 0.72 (0.20-­6.4) NS INR (n = 730) 1.1 (0.9-­4.9) 1.1 (0.9-­4.9) 1.1 (0.9-­1.4) NS Albumin, g/dL (n = 740) 4.2 (2.5-­5.2) 4.2 (2.5-­5.1) 4.2 (3.5-­5.2) NS Cholesterol, mg/dL (n = 629) 157 (52-­292) 157 (52-­292) 158 (80-­286) NS Child-­Pugh score (n = 727) NS A (5, 6) 689 (94.8) 613 (94.2) 76 (100) B (7-­9) 38 (5.2) 38 (5.8) 0 (0) MELD score (n = 685) 6 (6-­36) 6 (6-­36) 6 (6-­11) NS PLT, 109/L (n = 743) 138 (10-­439) 133 (10-­439) 132 (31-­386) NS <130, n (%) 354 (47.6) 317 (47.5) 37 (48.7) NS APRI score (n = 735) 1.5 (0.1-­23.3) 1.5 (0.1-­23.3) 1.5 (0.28-­10.8) NS <1, n (%) 236 (32.1) 217 (32.9) 19 (25.0) NS 1-­2, n (%) 236 (32.1) 203 (30.8) 33 (43.4) (Continues) 2738 | BROQUETAS et al. TA B L E 1 (Continued) No liver Biopsy, n = 670 Liver Biopsy, n = 76 Characteristics N (%) All patients, N = 746 (89.8) (10.2) P value >2, n (%) 263 (35.8) 239 (36.3) 24 (31.6) FIB-­4 score (n = 561) 4.0 (0.3-­50.1) 4.0 (0.3-­50.1) 4.0 (0.9-­18.1) NS <1.45, n (%) 44 (7.8) 41 (8.5) 3 (3.9) NS 1.45-­3.25, n (%) 177 (31.6) 149 (30.7) 28 (36.8) >3.25, n (%) 340 (60.6) 295 (60.8) 45 (59.2) Direct-­acting antiviral (DAA) NS SOF/LDV 201 (27.5) 180 (28.0) 21 (28.0) SOF+DCV 80 (10.9) 69 (10.5) 11 (14.7) SOF+SMV 81 (11.1) 70 (10.7) 11 (14.7) PTV/r/O/D 333 (45.5) 308 (46.9) 25 (33.3) Others 37 (5.1) 30 (4.6) 7 (9.3) Ribavirin, n (%) 367 (61.9.0) 320 (61.9) 47 (61.8) NS SVR12, n (%) 737 (99.1) 661 (99.0) 76 (100) NS Note: Liver biopsies have been collected 3 y after SVR. Abbreviation: ALT, alanine aminotransferase; APRI, AST to platelet ratio; AST, aspartate aminotransferase; BMI, body mass index; DCV, daclatasvir; FIB-­4, fibrosis-­4 index; HCV, hepatitis C virus; HIV, human immunodeficiency virus; INR, international normalised ratio; LDV, ledipasvir; MELD, Model for End-­Stage Liver Disease; N, number; PLT, platelets; PTV/r/O/D, paritaprevir/ritonavir/ombitasvir/dasabuvir; SMV, simeprevir; SOF, sofosbuvir; SVR, sustained viral response; T2DM, type 2 diabetes mellitus; TE, transient elastography; y, years; γGT, gamma-­glutamyl transpeptidase. 3.2 | Impact of SVR on liver function, metabolic-­ follow-­up. Supporting Information about patients with or without related factors and portal hypertension oesophageal varices is depicted more in detail in Table S3. No pa- tient from our sample (n = 76) developed clinical decompensation or Hepatitis C virus elimination was associated with a global improve- HCC during follow-­up. ment in liver function parameters, NITs and TE values 3 years after SVR (P < .001 in all cases) (Table 3). Body mass index evaluation before antiviral treatment was 3.3 | Histological evaluation after SVR available in 75 (98.7%) patients and 22 of them (29.3%) had obe- sity (BMI ≥ 30 kg/m2). Patients with obesity showed a higher Liver biopsy was performed after 3 years of EoT in the whole proportion of T2DM (40.9% vs. 15.1%; P = .031), and higher sample (median time = 38.4 months). Liver biopsies (median levels of gamma-­g lutamyl transpeptidase (γGT) before and after sample size = 24.5 mm) showed absent (A0) or minimal (A1) SVR. Among these obese patients, 19 (86.4%) persisted with necro-­inflammatory activity in most patients (n = 73, 96%) and BMI > 30 kg/m2 , 10 (45.5%) with T2DM and 13 (59.1%) with non-­significant steatosis (<5% of the hepatic parenchyma) in 55 arterial hypertension after HCV-­e radication. After SVR, obese (72.4%) patients. Nevertheless, F4 was observed in 41 (53.9%) patients showed higher TE values, higher rate of steatosis in ul- and F3 in 12 (15.8%) patients. The fibrosis stage distribution ac- trasound and higher proportion of patients with >5% of steatosis cording to Laennec scale is described more in detail in Table 2 in liver biopsy. However, obese and non-­o bese patients showed and Table S1. Among patients with cirrhosis (n = 41), the Laennec similar rate of cirrhosis 3 years after SVR. Additional information grading system identified F4A in 25 (61%), F4B in 10 (24.4%) and is depicted in Table S2. F4C in 6 (14.6%). No changes in obesity, type 2 diabetes mellitus (T2DM), or alcohol abuse rates were observed in the whole sample after SVR (Table 4). However, the use of antihypertensive therapy increased (from 20 3.4 | Characteristics of patients with F4 after SVR to 37 patients) after SVR. The Baveno VI criteria to avoid endos- copy were present in 20 patients before antiviral treatment and in 42 Patients with F4 after SVR (n = 41, 53.9%) showed significant differ- after SVR (Table 4). Therefore, oesophageal varices were evaluated ences compared to those with F0-­3 (n = 35, 46.1%; Table 4). Before in 52 before DAA and in 25 after HCV eradication. Oesophageal var- antiviral treatment, patients with F4 after SVR were younger and had ices were present at baseline in 23 (30.3%) but only in 11 (14.5%) higher rates of HCV-­infection by genotype 3, higher bilirubin levels after SVR. In patients without oesophageal varices before antiviral and international normalised ratio (INR) values, lower platelet count, treatment, advanced liver fibrosis was present after SVR in 66% of higher APRI and TE values and higher rates of splenomegaly. Three them and 1 (14.3%) out of 7 developed oesophageal varices during years after HCV elimination, patients with F4 had higher values of BROQUETAS et al. | 2739 TA B L E 2 Characteristics of included patients with cACLD according to TE value before treatment (N = 73) Characteristics before antiviral treatment, N (%) All patients, N = 73 TE 10-­15 kPa, n = 27 (37) TE > 15 kPa, n = 46 (63) P value Males, n (%) 46 (63) 21 (77.8) 25 (54.3) NS Age, y 59.2 (25.8-­8 0.4) 57.7 (34.1-­78.3) 60.4 (25.8-­8 0.4) NS BMI, kg/m2 (n = 72) 26.5 (16.8-­41.4) 26.8 (16.8-­39.8) 26.5 (17.9-­41.4) NS 2 Obesity (BMI > 30 kg/m ), n (%) 20 (27.8) 8 (29.6) 12 (26.7) NS T2DM, n (%) 16 (21.9) 5 (18.5) 11 (23.9) NS Arterial hypertension, n (%) 19 (22.5) 8 (29.6) 11 (23.9) NS Alcohol abuse (>20/30 g/dL), n (%) 6 (8.2) 1 (3.7) 5 (10.9) NS HIV-­coinfection, n (%) (n = 72) 9 (12.5) 6 (22.2) 3 (6.7) NS HCV genotype 3, n (%) 10 (13.7) 5 (18.5) 5 (10.9) NS TE before treatment, kPa 20 (10-­43.6) 12 (10-­14.3) 21.5 (15.4-­43.6) <.001 Baveno VI criteria, n (%) <.001 No (TE > 20 or PLT < 150) 55 (75.3) 12 (44.4) 43 (93.5) Abdominal ultrasound, n (%) Nodular liver surface 13 (18.5) 5 (18.5) 8 (17.4) NS Spleen size >12 cm 29 (39.7) 9 (33.3) 20 (43.5) NS Increased hepatic echogenicity 14 (19.2) 2 (7.4) 12 (26.1) .05 Oesophageal varices, n (%) (n = 52) 22 (43.1) 5 (35.7) 17 (45.9) NS AST, IU/L 64 (18-­333) 56 (18-­208) 68.5 (34-­333) NS ALT, IU/L 78 (16-­369) 83 (16-­327) 67 (29-­369) NS γGT, IU/L 98 (20-­830) 87 (25-­660) 98 (20-­830) NS Bilirubin, mg/dL 0.7 (0.2-­1.7) 0.6 (0.3-­1.6) 0.8 (0.2-­1.7) .043 INR 1.1 (0.9-­1.4) 1.0 (0.9-­1.2) 1.1 (1.0-­1.4) .001 Cholesterol, mg/d (n = 72) 156 (80-­236) 156 (115-­211) 157 (80-­236) NS Albumin, g/dL 4.2 (3.5-­5.2) 4.3 (3.7-­4.9) 4.2 (3.5-­5.2) NS Child-­Pugh score A (5) 73 (100) 27 (100) 46 (100) NS MELD score (n = 72) 6 (6-­11.4) 6 (6-­8.7) 6 (6-­11.4) NS 9 PLT, 10 /L 132 (31-­386) 156 (31-­386) 117 (58-­3 09) .012 APRI score 1.54 (0.3-­10.8) 1.19 (0.3-­5.6) 1.67 (0.4-­10.8) NS <1, n (%) 18 (24.7) 10 (37) 8 (17.4) >2, n (%) 23 (31.5) 7 (25.9) 16 (34.8) FIB-­4 score 3.8 (0.96-­18.1) 2.6 (1.0-­18.8) 4.3 (1.6-­10.2) .007 <1.45, n (%) 3 (4.1) 3 (11.1) 0 (0.0) >3.25, n (%) 43 (58.9) 10 (37) 33 (71.7) Liver biopsy 3 y after antiviral treatment Obesity (BMI > 30 kg/m2), n (%) 20 (27.8) 8 (29.6) 12 (26.7) NS T2DM, n (%) 17 (23.3) 5 (18.5) 12 (26.1) NS Arterial hypertension, n (%) 35 (47.9) 14 (51.9) 21 (45.7) NS Alcohol abuse (>20/30 g/dL), n (%) 2 (2.7) 1 (3.7) 1 (2.2) NS Time to liver biopsy (months) 38.7 (32-­49.1) 38.5 (32.9-­46.2) 39.2 (32-­49) NS Liver biopsy length (mm) 24 (6-­4 0) 32 (12-­4 0) 17 (6-­39) .015 Liver biopsy length ≥15 mm, n (%) 64 (87.7) 25 (92.6) 39 (84.8) NS Hepatitis activity (METAVIR), n (%) NS No activity (A0) 27 (37) 9 (33.3) 18 (39.1) Mild activity (A1) 43 (58.9) 18 (66.7) 25 (54.3) (Continues) 2740 | BROQUETAS et al. TA B L E 2 (Continued) Characteristics before antiviral treatment, N (%) All patients, N = 73 TE 10-­15 kPa, n = 27 (37) TE > 15 kPa, n = 46 (63) P value Moderate activity (A2) 2 (2.7) 0 (0.0) 2 (4.3) Severe activity (A3) 1 (1.4) 0 (0.0) 1 (2.2) Steatosis (NASH CRN), n (%) NS <5% 53 (72.6) 21 (77.8) 32 (69.6) 5%-­3 0% 17 (23.3) 5 (18.5) 12 (26.1) 30%-­60% 3 (4.1) 1 (3.7) 2 (4.3) Fibrosis stage (Laeenec), n (%) .003 F0 2 (2.7) 2 (7.4) 0 (0) F1 9 (12.3) 5 (18.5) 4 (8.7) F2 11 (15.1) 6 (22.2) 5 (10.9) F3 11 (15.1) 6 (22.2) 5 (10.9) F4A 24 (32.9) 5 (18.5) 19 (41.3) F4B 10 (13.7) 1 (3.7) 9 (19.6) F4C 6 (8.2) 2 (7.4) 4 (8.7) Cirrhosis (METAVIR, F4), n (%) 40 (54.8) 8 (29.6) 32 (69.6) .001 Clinical decompensation, n (%) 0 (0) 0 (0) 0 (0) NS Abbreviations: ALT, alanine aminotransferase; APRI, AST to platelet ratio; AST, aspartate aminotransferase; BMI, body mass index; FIB-­4, fibrosis-­4 index; HCV, hepatitis C virus; HIV, human immunodeficiency virus; INR, international normalised ratio; MELD, Model for End-­Stage Liver Disease; N, number; NASH CRN, NASH Clinical Research Network; PLT, platelets; T2DM, type 2 diabetes mellitus; TE, transient elastography; y, years; γGT, gamma-­glutamyl transpeptidase. TA B L E 3 Impact of HCV eradication on liver function, liver Multivariate analysis (OR (95%CI), P) showed that HCV-­infection stiffness and NITs in patients with liver biopsy (n = 76) by genotype 3 (20.81 (2.12-­201.47), .009), and TE (kPa) before antiviral treatment (1.21 (1.09-­1.34), <.001) were the only vari- Before antiviral 3 y after antiviral P treatment treatment value ables independently associated with cirrhosis 3 years after SVR. Characteristics of HCV-­genotype 3 infected patients are reported AST, IU/L 64.5 (18-­333) 21 (9-­68) <.001 in Table S4. ALT, IU/L 76.5 (16-­369) 17 (6-­70) <.001 γGT, IU/L 98.5 (18-­830) 25.5 (8-­157) <.001 Albumin, g/dL 4.2 (3.5-­5.2) 4.6 (3.6-­5.2) <.001 3.5 | Diagnostic accuracy of non-­invasive methods Cholesterol, 158 (80-­236) 182 (110-­239) <.001 to identify F4 after SVR mg/dL PLT, 109/L 132 (31-­386) 163 (80-­3 03) <.001 Patients with a TE ≥ 15 kPa before antiviral treatment showed higher Bilirubin, mg/dL 0.7 (0.2-­1.7) 0.5 (0.1-­2.3) <.001 proportion of F4 after SVR compared to those with a TE = 10-­15 kPa INR 1.1 (0.9-­1.4) 1.1 (1.0-­1.3) NS (69.6% vs 29.6%; P = .001) (Table 2). None of the patients (0%) with TE, kPa 19.9 (9.6-­43.6) 10.2 (4.5-­37.4) <.001 TE ≥ 15 kPa before DAA completely eliminated liver fibrosis (F0) APRI score 1.53 (0.3-­10.8) 0.4 (0.1-­2.1) <.001 3 years after HCV eradication. The absence of fibrosis was observed FIB-­4 score 3.8 (1.0-­18.1) 1.9 (0.6-­5.1) <.001 in two patients (7.4%) with TE = 10-­15 kPa (P = .003). Similar re- MELD score 6.0 (6.0-­11.4) 6.0 (6.0-­11.2) NS sults were found in patients with suspected cirrhosis before antivi- Abbreviations: ALT, alanine aminotransferase; APRI, AST to platelet ral treatment (TE ≥ 14 kPa) compared to those with a TE < 14 kPa ratio; AST, aspartate aminotransferase; FIB-­4, fibrosis-­4 index; INR, (Table S1). However, one out of three patients with a TE 9.6-­9.9 kPa international normalised ratio; MELD, Model for End-­Stage Liver before antiviral treatment showed F4 after HCV eradication. Disease; PLT, platelets; TE, transient elastography; γGT, gamma-­ Diagnostic accuracy of TE, APRI and FIB-­4 before and 3 years glutamyl transpeptidase. after SVR to identify patients with F4 was evaluated with the AUROC curve and the comparative analysis with the Hanley-­McNeil aspartate aminotransferase (AST), alanine aminotransferase (ALT), method28 (Figure 2). The best variable for predicting F4 after SVR γGT, lower platelet count, higher APRI and TE values and higher was TE before the antiviral treatment (AUROC = 0.79), being the rates of splenomegaly than those with F0-­3. accuracy lower for APRI (AUROC = 0.71) and FIB-­4 (AUROC = 0.64) BROQUETAS et al. | 2741 TA B L E 4 Characteristics of included patients (N = 76) and variables associated with METAVIR F4, 3 y after antiviral treatment Characteristics before antiviral METAVIR F0-­3 (n = 35), METAVIR F4 (n = 41), treatment 46.1% 53.9% P value OR (95% CI), P Males, n (%) 23 (65.7) 25 (61.0) NS Age, y 68.0 (34.3-­8 0.7) 54.9 (26.3-­77.6) <.001 <65, n (%) 15 (42.9) 33 (80.5) .001 2 Obesity (BMI > 30 kg/m ), n (%) 8 (22.9) 14 (35) NS (n = 75) T2DM, n (%) 6 (17.1) 11 (26.8) NS Arterial hypertension, n (%) 10 (28.6) 10 (24.4) NS Alcohol abuse (>20/30 g/dL), n (%) 1 (2.9) 5 (12.2) NS Child-­Pugh score A (5) 35 (100) 41 (100) NS MELD score (n = 75) 6 (6-­11.4) 6 (6-­8.7) NS HIV-­coinfection, n (%) 4 (11.4) 5 (12.5) NS HCV Genotype 3, n (%) 1 (2.9) 9 (22.0) .017 20.8 (2.12-­201), .009 AST, IU/L 60 (18-­208) 72 (34-­333) NS ALT, IU/L 70 (16-­327) 82 (29-­369) NS γGT, IU/L 87 (18-­660) 105 (43-­830) NS Bilirubin, mg/dL 0.6 (0.3-­1.62) 0.9 (0.2-­1.7) .023 INR 1.06 (0.9-­1.2) 1.1 (1-­1.4) .022 Albumin, g/dL 4.2 (3.7-­4.9) 4.2 (3.5-­5.2) NS Abdominal ultrasound, n (%) Nodular liver surface 7 (20) 6 (14.6) NS Spleen size >12 cm 8 (22.9) 22 (53.7) .006 Increased hepatic echogenicity 6 (17.1) 8 (19.5) NS Oesophageal varices, n (%) (n = 52) 6 (28.6) 17 (54.8) .061 PLT, 109/L 160 (31-­386) 115 (58-­253) .001 <160, n (%) 16 (45.7) 34 (82.9) .001 APRI score 1.1 (0.28-­6.2) 1.8 (0.49-­10.8) .002 <1, n (%) 16 (45.7) 3 (7.3) .001 >2, n (%) 7 (20.0) 17 (41.5) FIB-­4 score 3.4 (1.1-­18.8) 4.6 (1.0-­11.0) NS <1.45, n (%) 2 (5.7) 1 (2.4) NS >3.25, n (%) 16 (45.7) 29 (70.7) TE before treatment, kPa 14 (9.8-­3 0.8) 21.3 (9.6-­43.6) <.001 1.21 (1.09-­1.34), <.001 <10 kPa, n (%) 2 (5.7) 1 (2.4) .002 10-­15 kPa, n (%) 19 (54.3) 8 (19.5) >15 kPa, n (%) 14 (40) 32 (78) Baveno VI criteria, n (%) .002 No (TE > 20 or PLT < 150) 20 (57.1) 36 (87.8) Characteristics 3 y after antiviral treatment Obesity (BMI > 30 kg/m2), n (%) 9 (25.7) 13 (32.5) NS T2DM, n (%) 6 (17.1) 12 (29.3) NS Arterial hypertension, n (%) 19 (54.3) 18 (43.9) NS Alcohol abuse (>20/30 g/dL), n (%) 1 (2.9) 1 (2.4) NS AST, IU/L 20 (12-­37) 24 (9-­68) .01 ALT, IU/L 15 (6-­56) 21 (11-­70) .001 (Continues) 2742 | BROQUETAS et al. TA B L E 4 (Continued) Characteristics before antiviral METAVIR F0-­3 (n = 35), METAVIR F4 (n = 41), treatment 46.1% 53.9% P value OR (95% CI), P γGT, IU/L 22 (8-­90) 37 (12-­157) .006 Bilirubin, mg/dL (n = 75) 0.5 (0.3-­1.5) 0.6 (0.1-­2.3) NS INR (n = 75) 1.08 (1-­1.2) 1.05 (0.9-­1.3) NS Albumin, g/dL (n = 74) 4.5 (3.6-­5.2) 4.6 (3.9-­5.2) NS PLT, 109/L 190 (97-­3 03) 143 (80-­296) .004 Abdominal ultrasound, n (%) (n = 74) Nodular liver surface 1 (2.9) 3 (7.5) NS Spleen size >12 cm 4 (11.8) 13 (32.5) .035 Increased hepatic echogenicity 4 (11.8) 11 (27.5) .093 Oesophageal varices, n (%) (n = 25) 1 (9.1) 10 (50.0) NS APRI score 0.2 (0.1-­0.7) 0.4 (0.1-­2.1) NS <1, n (%) 35 (100) 39 (95.1) NS >2, n (%) 0 (0.0) 1 (2.4) FIB-­4 score 2.0 (0.8-­3.4) 2.5 (0.5-­5.5) NS <1.45, n (%) 12 (34.3) 14 (34.1) NS >3.25, n (%) 2 (5.7) 6 (14.6) TE before liver biopsy, kPa 8.6 (4.5-­15.2) 13.2 (5.4-­37.4) <.001 <8 kPa, n (%) 15 (45.5) 6 (14.6) .003 >8 kPa, n (%) 18 (54.5) 35 (85.4) Baveno VI criteria, n (%) (n = 74) .004 No (TE > 20 or PLT < 150) 9 (25.7) 25 (61.0) Abbreviations: ALT, alanine aminotransferase; APRI, AST to platelet ratio; AST, aspartate aminotransferase; FIB-­4, fibrosis-­4 index; HCV, hepatitis C virus; HIV, human immunodeficiency virus; INR, international normalised ratio; N, number; PLT, platelets; T2DM, type-­2 diabetes mellitus; TE, transient elastography; y, years; γGT, gamma-­glutamyl transpeptidase. F I G U R E 2 AUROC curves of TE accuracy for the diagnosis of cirrhosis in patients with liver biopsy. The corresponding areas under the AUROC curves were 0.79 (95% CI: 0.68-­0.87) for TE before DAA, 0.75 (95% CI: 0.63-­0.84) for TE after SVR, 0.71 (95% CI: 0.59-­ 0.81) for APRI before DAA, 0.71 (95% CI: 0.60-­0.81) for APRI after SVR, 0.64 (95% CI: 0.52-­0.75) for FIB-­4 before DAA, and 0.55 (95% CI: 0.43-­0.67) for FIB-­4 after SVR. The AUROC curves were compared using the Hanley-­McNeil method28 and the differences statistically significant are depicted before treatment, and TE (AUROC = 0.75), APRI (AUROC = 0.71) and 5 (26.3%) F3, meaning that up to 8 (42%) patients with an APRI and FIB-­4 (AUROC = 0.55) 3 years after SVR. The best cut-­off of <1 and TE > 10 kPa before treatment, showed F3-­4, 3 years after TE before antiviral treatment to identify patients with F4 after SVR SVR. Among 24 patients with APRI > 2 before antiviral therapy, 17 was 20.9 kPa (sensitivity 61% and specificity 88.6%), but the cut-­off (70.8%) had F4 and 4 (16.7%) F3 thereafter. Out of 33 patients with with the highest NPV was 12 kPa (NPV = 88%; Figure 3). Among intermediate APRI, 21 (63.6%) had F4 and 3 (9.1%) had F3. Similar 19 patients with APRI <1 before treatment, 3 (15.8%) revealed F4 results were found with FIB-­4 test (Figure S1). BROQUETAS et al. | 2743 F I G U R E 3 Best cut-­off points for TE (kPa) before DAA treatment. The best cut-­off of TE before antiviral treatment to identify patients with cirrhosis after SVR was 20.9 kPa (sensitivity 61% and specificity 88.6%), but the cut-­off with the highest negative predictive value (NPV) was 12 kPa (NPV = 88%) F I G U R E 4 Best cut-­off points for TE (kPa) 3 y after SVR. The best cut-­off of TE 3 y after SVR to identify cirrhosis was 12.5 kPa (sensitivity 56.1% and specificity 88.6%), but the value with the highest NPV was 8 kPa (NPV = 77%) The elimination of HCV after antiviral treatment produced a sig- treatment and 3 years after SVR with DAA and liver biopsy. We have nificant decrease in TE, APRI and FIB-­4 values. The best cut-­off for shown that neither TE nor NITs as APRI or FIB-­4 are reliable enough the TE 3 years after SVR to identify F4 was 12.5 kPa (sensitivity to accurately identify liver fibrosis stage after SVR. In spite of finding 56.1% and specificity 88.6%), but the value with the highest NPV significant improvements in liver function parameters and TE values was 8 kPa (NPV = 77%; Figure 4). In consequence, 12 (37.5%) out after HCV clearance, cirrhosis was present in more than half of the of 32 patients with a TE 3 years after EoT <9.5 kPa showed F4 and patients and advanced fibrosis in 70%. 19 (59.4%) F3-­4. Among 22 patients with a TE value <8 kPa 3 years Our study assessed a large multicentre cohort of HCV-­infected after SVR, 12 (54.5%) revealed F3-­4: 6 (27.3%) F3 and 6 (27.3%) F4 cACLD patients before antiviral treatment. Total cohort showed in- (Figure S2). Similarly, NITs experienced a significant decrease after direct signs of portal hypertension in more than half of them and HCV elimination. We did not find differences in FIB-­4 or APRI values one-­third were obese patients with T2DM. Patients with suspected after SVR between patients with F4 and those F0-­3. cirrhosis or with a TE value highly suggestive of cACLD before anti- viral treatment6 showed the highest proportion of F4, 3 years after SVR. Obese patients showed higher values of TE, increased hepatic 4 | D I S CU S S I O N echogenicity and steatosis despite HCV eradication, but no differ- ences in liver fibrosis stage were found. Unfortunately, the accep- Our work is probably the largest prospective multi-­centric study tance to undergo the liver biopsy after HCV elimination was low of HCV-­c ACLD patients with TE paired measurements before in real clinical practice. Nevertheless, the multicentre origin of the 2744 | BROQUETAS et al. samples and the comparative baseline analysis demonstrated that their FIB-­4 score decreases, and authors recommend the indefinite patients who underwent the liver biopsy were representative of the HCC surveillance.36 total cohort. Recent studies have shown that the risk of persistent portal Our work has demonstrated that HCV genotype 3 and high hypertension and developing liver-­related complications still exist TE values before treatment were independently associated with after SVR in patients with HCV-­ cACLD. Lens et al16 showed the cirrhosis 3 years after SVR. Patients infected by genotype 3 were persistence of clinically significant portal hypertension (CSPH) at younger and had a higher percentage of HIV-­coinfection. The SVR24 in up to 80% of patients with CSPH before DAA, but TE val- HCV-­genotype 3 and HIV coinfection have been associated with ues were not sufficiently discriminatory to rule out CSPH. Similarly, a faster liver fibrosis progression, higher degree of portal hyper- our group18 has demonstrated that HCV-­cirrhotic patients can de- 30-­3 3 tension and an increased incidence of HCC. However, pro- velop high-­risk gastroesophageal varices after SVR, especially in spective studies evaluating fibrosis regression after SVR in HIV those with high TE values or gastroesophageal varices before treat- and HCV coinfected patients are scarce, especially after DAA ment. In consonance, the current study has demonstrated that 1 out introduction. Labarga et al demonstrated progression in 7.2% of of 7 HCV-­c ACLD patients without oesophageal varices can develop HIV/HCV patients despite SVR with pegylated interferon/ribavi- ‘the novo’ varices despite HCV elimination. However, none of our rin (median time: 4.4 years). 34 However, this study was retrospec- patients developed clinical decompensation or HCC, although this tive and only 32% had advanced fibrosis. Our prospective study could be associated with the restrictive exclusion criteria for the has demonstrated the presence of advanced fibrosis 3 years after liver biopsy. Recently, Pons et al37 have shown that HCC is the most SVR in most HIV co-­infected patients and in all previously in- frequent liver-­related complication after SVR, and authors recom- fected by GT3. mend indefinite surveillance for HCC in patients with HCV-­ cACLD The best TE cut-­off point before antiviral treatment to differ- before treatment. In this line, our study has demonstrated that 70% entiate our patients with cirrhosis after SVR was 20.9 kPa (sensi- of HCV-­c ACLD patients have F3-­F4 after 3 years of viral eradication, tivity 61% and specificity 88.6%). However, the cut-­off point with and therefore, ultrasound follow-­up is recommended. the highest negative predictive value was 12 kPa (NPV = 88%) and The limitations of our study were (a) the lack of biopsies be- 1 out of 3 patients with a TE 9.6-­9.9 kPa showed F4 after SVR. The fore antiviral therapy that prevented us to analyse the exact rate elimination of the HCV infection produced a significant decrease in of fibrosis regression/progression after SVR; however, this fact was ALT, AST, FIB-­4, APRI and TE values 3 years after antiviral treat- expected since current international guidelines9,10 accept TE as the ment. However, liver biopsies showed cirrhosis in 54% of them. A best non-­invasive method to identify patients with cirrhosis before systematic review including 24 studies with paired TE before and antiviral therapy; (b) the low patients’ acceptance rate for liver bi- after SVR has shown that liver stiffness decreases around 3 kPa in opsies in clinical practice after HCV eradication, which implied a re- 35 the first 6-­12 months following HCV clearance. Our study showed duced sample size and (c) the lack of external validation, which do that 47% of patients with a baseline TE compatible with advanced not allow to give solid recommendations regarding the optimal fol- fibrosis (TE > 9.5 kPa) presented values <9.5 kPa after SVR.35 low-­up of these patients. Indeed, we can only suggest upper endos- Likewise, the best cut-­off point to differentiate patients with liver copy for variceal surveillance in the group of patients at higher risk cirrhosis decreased to 12.5 kPa 3 years after SVR (sensitivity 56.1% of cirrhosis after SVR, but this would clearly need external validation and specificity 88.6%) and the lowest cut-­off of 8 kPa only showed a in a larger sample. moderate NPV (77%). Although this early decrease in TE is related to In spite of this, we believe that our study has important strengths. improvements in inflammation and necrosis, there are few evidences This is probably the largest prospective multi-­centric study of HCV-­ that suggest an improvement in the stage of liver fibrosis.7 A recent cACLD patients evaluated 3 years after DAA with simultaneous liver pilot study evaluating 15 paired biopsies by morphometry confirms biopsy, TE and non-­invasive tests. Moreover, the multicentre origin fibrosis regression after SVR, but showed important discordances of the histological samples, the good quality and the centralised eval- with the decrease in TE values: 13 out of 15 patients with high TE uation, made biopsied patients as representative of the total cohort. values before treatment, showed a TE <9.5 kPa after SVR, but only In conclusion, our study has shown that despite significant im- 4 had low liver fibrosis stages (