Received: 12 January 2022 | Revised: 11 May 2022 | Accepted: 25 May 2022 DOI: 10.1111/liv.15323 ORIGINAL ARTICLE Prevalence estimation of significant fibrosis because of NASH in Spain combining transient elastography and histology José L. Calleja1 | Jesús Rivera-­Esteban2,3 | Rocío Aller4,5 | Marta Hernández-­Conde1 | Javier Abad1 | Juan M. Pericàs2 | Hugo G. Benito5 | Miguel A. Serra6 | Amparo Escudero6 | Javier Ampuero7 | Ana Lucena7 | Yolanda Sánchez7 | María T. Arias-­Loste8 | Paula Iruzubieta8 | Manuel Romero-­Gómez7 | 2,3,9 8 Salvador Augustin | Javier Crespo 1 Department of Gastroenterology and Hepatology, Hospital Universitario Puerta de Hierro, School of Medicine, Universidad Autónoma Madrid, IDIPHIM, Majadahonda, Spain 2 Liver Unit, Vall d'Hebron Hospital Universitari, Vall d'Hebron Institut de Recerca (VHIR), Vall d'Hebron Barcelona Hospital Campus, Barcelona, Spain 3 Universitat Autònoma de Barcelona, Bellaterra, Spain 4 Department of Gastroenterology, Clinic University Hospital, Medical School, Group of Biomedical Research in Critical Care Medicine (BioCritic), University of Valladolid, Valladolid, Spain 5 Research Unit, Clinic University Hospital, Medical School, Institute of Health Sciences of Castille and Leon (IECSCYL), Group of Biomedical Research in Critical Care Medicine (BioCritic), Valladolid, Spain 6 Digestive Medicine Service, Clinic University Hospital, University of Valencia, Valencia, Spain 7 Digestive Diseases Department and CIBERehd, Virgen del Rocío University Hospital, Institute of Biomedicine of Seville, University of Seville, Seville, Spain 8 Gastroenterology and Hepatology Department, Marqués de Valdecilla University Hospital, Clinical and Translational Digestive Research Group, IDIVAL, Santander, Spain 9 Therapeutic Area Cardio-­Metabolism and Respiratory Medicine, Boehringer Ingelheim International GmbH, Ingelheim am Rhein, Germany Correspondence Salvador Augustin, Therapeutic Area Abstract Cardio-­Metabolism and Respiratory Background & Aims: Non-­alcoholic fatty liver disease (NAFLD) has become a major Medicine, Boehringer Ingelheim International GmbH, Ingelheim am Rhein, public health problem, but the prevalence of fibrosis associated with non-­alcoholic Germany. steatohepatitis (NASH) is largely unknown in the general population. This study aimed Email: salva.augustin@gmail.com; salvador.augustin@boehringer-­ingelheim. to provide an updated estimation of the prevalence of NASH fibrosis in Spain. com Methods: This was an observational, retrospective, cross-­sectional, population-­based Jesús Rivera-­Esteban, Liver Unit, study with merged data from two Spanish datasets: a large (N = 12 246) population-­ Department of Internal Medicine, Vall d'Hebron Hospital Universitari, Vall based cohort (ETHON), including transient elastography (TE) data, and a contem- d'Hebron Institut de Recerca (VHIR), Vall porary multi-­centric biopsy-­proven NASH cohort with paired TE data from tertiary d'Hebron Barcelona Hospital Campus. Passeig Vall d'Hebron 119-­129, 08035 centres (N = 501). Prevalence for each NASH fibrosis stage was estimated by crossing Barcelona, Spain. Universitat Autònoma TE data from ETHON dataset with histology data from the biopsy-­proven cohort. de Barcelona, 08193 Bellaterra, Spain. Email: jesusriveraest@gmail.com Abbreviations: ALP, alkaline phosphatase; ALT, alanine aminotransferase; AST, aspartate aminotransferase; AUDIT, Alcohol Use Disorders Identification Test; BMI, body mass index; CAP, controlled attenuation parameter; CI, confidence interval.; GGT, gamma-­glutamyl transferase; HBsAg, hepatitis B surface antigen; HCV, hepatitis C virus; HDL, high-­density lipoprotein; INE, Instituto Nacional Estadística; LDL, low-­density lipoprotein; LSM, liver stiffness measurement; NAFLD, Non-­alcoholic fatty liver disease; NASH, non-­alcoholic steatohepatitis; NCEP-­ATP III, National Cholesterol Education Program-­Adult Treatment Panel III; TE, transient elastography. José L. Calleja and Jesús Rivera-­E steban are co-­f irst authors contributed equally to the manuscript This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited. © 2022 The Authors. Liver International published by John Wiley & Sons Ltd. Liver International. 2022;00:1–10.  wileyonlinelibrary.com/journal/liv | 1 2 | CALLEJA et al. Handling Editor: Dr. Luca Valenti Results: From the patients with valid TE in ETHON dataset (N = 11 440), 5.61% (95% confidence interval [95% CI]: 2.53-­11.97) had a liver stiffness measurement (LSM) ≥ 8 kPa. The proportion attributable to NAFLD (using clinical variables and Controlled Attenuation Parameter) was 57.3% and thus, the estimated prevalence of population with LSM ≥ 8 kPa because of NAFLD was 3.21% (95% CI 1.13–­8.75). In the biopsy-­proven NASH cohort, 389 patients had LSM ≥ 8 kPa. Among these, 37% did not have significant fibrosis (F2-­4). The estimated prevalence of NASH F2-­3 and cirrhosis in Spain's adult population were 1.33% (95% CI 0.29–­5.98) and 0.70% (95% CI 0.10–­4.95) respectively. Conclusions: These estimations provide an accurate picture of the current prevalence of NASH-­related fibrosis in Spain and can serve as reference point for dimensioning the therapeutic efforts that will be required as NASH therapies become available. KEYWORDS hepatic fibrosis, liver biopsy, non-­alcoholic steatohepatitis, transient elastography 1 | I NTRO D U C TI O N Lay summary Non-­alcoholic fatty liver disease (NAFLD) has become a major pub- Non-­alcoholic steatohepatitis (NASH) has become a major lic health problem. Both its prevalence and incidence have risen public health issue worldwide, but the exact prevalence in sharply in the last decades.1–­5 and, by 2030, cirrhosis and hepatocel- the general population of the different stages of liver fibro- lular carcinoma because of NAFLD are expected to increase further sis associated to NASH is largely unknown. In the present worldwide.6–­8 This threat is aggravated by the fact that there are study, we merged data from a large general population-­ currently no approved pharmacologic therapies for non-­alcoholic based dataset and a contemporary multicentric biopsy-­ steatohepatitis (NASH). Accurate and geographically-­specific esti- proven NASH cohort to provide updated prevalence mations of NASH prevalence by fibrosis stage are paramount for the estimates for NASH fibrosis in Spain. These estimates design and implementation of public health measures and therapeu- might be leveraged for designing future interventions for tic strategies. NASH. Nonetheless, the exact prevalence of NAFLD and for the dif- ferent fibrosis stages of NASH remains unknown to date because previous calculations are subject to various problems. First, reports on NAFLD prevalence base their predictions on non-­invasive diag- histological confirmation of fibrosis predictions is scarce.9,10 Finally, 9–­13 14 nostic tests, registry diagnostic codes or on indirect extrap- the attribution of NAFLD causality in those fibrosis-­based popu- olations from basic metabolic demographics.15 All these methods lation studies has been indirect (basically based on comorbidities, have suboptimal sensitivity and probably underestimate the real without concurrent measurement of hepatic steatosis). prevalence of NAFLD. 2,16 In recent years, Controlled Attenuation The aim of the present study was to provide an updated, accu- Parameter (CAP) by transient elastography (TE) has been proven rate, real-­life estimate of the prevalence of NASH-­related fibrosis in more sensitive for the detection of steatosis than ultrasound or Spain. For that purpose, we merged data from a large study on the 10,17–­20 serum-­based scores but reports on NAFLD at the general use of TE for screening of liver disease in the general population of population level using CAP are still scarce and small-­scale. 21 Second, Spain with contemporary data from a multi-­centre cohort of biopsy-­ estimations of NASH at population level have been extrapolated proven NASH from real practice from our country. from either small autopsy or living-­donor series22 or through back-­ calculations crossing population-­scale estimates for NAFLD with non-­contemporary NASH biopsy series23–­25 subject to selection 2 | M E TH O D S and ascertainment biases, leading to overrepresentation of more ad- vanced stages of NASH. An intermediate approach consists of the 2.1 | Aim and study design use non-­invasive tests for estimations of liver fibrosis (with transient elastography and/or serum scores) as proxy for NASH in different This was an observational, retrospective, cross-­sectional, population-­ population-­based studies. 26,27 However, stratification of patients by based, epidemiological study. The main objective of the study was fibrosis stage with these methods has been proven suboptimal and to provide updated estimates for the prevalence of NASH-­related CALLEJA et al. | 3 fibrosis in the general population in Spain, with a special focus on the LSM 9.1 kPa cut-­off, which has been recently suggested as the those stages at higher risk of complications and that would be even- most cost-­efficient LSM threshold for theoretical population-­level tually amenable to receive pharmacologic therapy under the current screening. 28 regulatory framework, that is F2-­4 fibrosis stages. The paper has been performed and written following The general plan for the study (Figure 1) consisted of four steps: the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines. 29 • STEP 1: To estimate the prevalence and distribution of liver fibro- sis through clinically relevant TE ranges in Spain's general popula- tion from a large population-­based cohort (ETHON). 2.2 | Patients • STEP 2: To estimate the prevalence of NAFLD within the sub- set of patients with liver stiffness measurements (LSM) ≥ 8 kPa The study integrates data from two large cohorts: in the same cohort. For this estimation, we used the subcohort from Cantabria, which was the largest subcohort (N = 5090) and 1. Population-­based cohort (ETHON cohort). The PREVHEP-­ contained 99.7% of all valid CAP measurements from the whole ETHON is a Spanish dataset composed of subjects aged 20–­ ETHON dataset. Patients with viral hepatitis (positive anti-­HCV or 79 years selected from the general population of 18 primary HBsAg), high-­risk alcohol consumption (≥ 15 units/week) assessed care centres belonging to three university hospitals from 2015 by AUDIT test and those with the absence of steatosis defined as to 2017. Participants were randomly selected and stratified by CAP < 250 dB/m (except for patients with LSM ≥ 20 kPa in order socioeconomical status, rural/urban setting and age. The geo- to avoid “burn-­out” NASH cirrhosis exclusion) were excluded. The graphical location of Santander, Madrid and Valencia (north, proportion attributable to NAFLD was calculated as the ratio of centre and east, respectively) determinates substantial differ- the remaining patients after these exclusions over the total of pa- ences in terms of weather, lifestyle and food habits, appro- tients with LSM ≥ 8 kPa (Figure 2). priately showing a real-­life picture of Spain at population level. • STEP 3: To describe the distribution of the different liver fibro- Thus, this cohort is considered representative of Spain's general sis stages in a Spanish multi-­centre cohort of biopsy-­proven population and has served as a reference for other observa- NASH with paired TE (LSM and CAP) data, contemporary to the tional, cross-­sectional population-­based studies,30 even though population-­based cohort. the initial study was aimed to investigate the prevalence of • STEP 4: To generate estimations of the real prevalence for the HCV in the general population.31 different NASH-­related fibrosis stages for Spain's general pop- ulation by crossing the estimated prevalence of NAFLD drawn This cohort was the basis for the estimation of TE values for the from ETHON cohort with the probability of each specific stage of Spanish general population as well as for the estimation of the at- fibrosis observed in the biopsy-­proven NASH cohort in patients tributable causal weight of NAFLD within the range of LSM ≥8 kPa. with LSM ≥ 8 kPa. 2. Biopsy-­p roven NASH cohort. This dataset was a cross-­ As secondary outcome, using the same step by step approach sectional, retrospective cohort of real clinical practice pa- detailed above, we aimed at describing prevalence estimates using tients with biopsy-­p roven NASH from two hospitals from the ETHON cohort (Marqués de Valdecilla, Cantabria and Puerta de Hierro, Madrid) and three additional tertiary centres (Virgen del Rocío, Seville; Clínico Universitario, Valladolid; and Vall d'Hebron, Barcelona). These centres had been collecting data prospectively from NAFLD patients for several studies and for Spain's NAFLD National Registry (Hepamet). 32 Patient selection was based on what is esti- mated as current clinical practice in tertiary centres in Spain, where patients are referred to liver clinics based on a combination of altered liver function tests and/or finding of steatosis in liver ultrasound. Once in the clinics, patients with LSM < 8 kPa are considered low risk and are less likely to undergo liver biopsy, and in patients with LSM ≥ 8 kPa liver biopsy is individually decided for disease staging and for inclusion consideration in clinical trials. For the present study, we only included consecutive patients with paired data from liver biopsy and TE within a < 12-­m onth F I G U R E 1 Study design and cohorts. period obtained with Fibroscan devices equipped with M 4 | CALLEJA et al. F I G U R E 2 Flow chart for the COHORT 1 (ETHON) - CANTABRIAN calculation of the proportion attributable subcohort with valid LSM to non-­alcoholic fatty liver disease n=5090 (NAFLD) in the population-­based cohort. CAP, controlled attenuation parameter; CAP not available, n=376 HBsAg, hepatitis B surface antigen; HCV, hepatitis C virus; LSM, liver stiffness measurements; NAFLD, non-­alcoholic LSM < 8 kPa, n=4524 fatty liver disease. CANTABRIAN subcohort with LSM ≥ 8 kPa and CAP n=190 Viral hepatitis (anti-HCV or HBsAg), n=16 Risk alcohol consumption (≥ 15 units/week), n=20 CAP < 250 dB/m (except patients with LSM ≥ 20kPa), n=45 PROPORTION ATTRIBUTABLE TO NAFLD 57.3% (109/190*100) and XL probes, spanning from 2015 to 2020. Only liver excluded from final analyses under the assumption that these pa- samples clearly interpretable for the pathologists were in- tients would not be referred to tertiary centres to undergo liver bi- cluded in the study. opsy in real clinical practice. Categorical data are presented as number (percentage). The biopsy-­proven cohort was used to translate TE estimations Continuous data are presented as mean ± standard deviation for the general Spanish population into histology-­based estimations and median (interquartile range). A p < .05 was considered sta- for NASH diagnosis and its different fibrosis stages. tistically significant. Missing values were kept as missing, and no specific statistical procedures were used for imputations. Data were collected and edited using Microsoft Excel (version 2.3 | Procedures Microsoft Office Pro 2019). Statistical analyses were performed using PAWS Statistics (version 19.0; SPSS Inc., Hong Kong) TE measurements, liver biopsy and histological evaluation were per- software. formed according to current standards.33,34 A detailed description of procedures is provided in Supplementary Material. 3 | R E S U LT S 2.4 | Statistical analysis 3.1 | Patients characteristics from the population-­ based transient elastography cohort (ETHON cohort) For demographic extrapolations and comparisons with the Spanish general population, publicly available data from Instituto Nacional From the complete ETHON dataset of 12 246 individuals, 806 (6.6%) de Estadística (INE) were used. INE is a legally independent admin- were excluded because of non-­valid or indeterminate LSM. istrative institution which serves as main repository of demographic For the 11 440 patients included in the analysis (Table 1) me- data for Spain, including key health indicators. Patients with non-­ dian age was 51 years, 58% were women and 88% were Caucasian. valid LSM and those with LSM < 8 kPa from both cohorts will be Median BMI was 26.1 kg/m2 and 60% were either overweight or CALLEJA et al. | 5 TA B L E 1 Baseline characteristics of the patients from the TA B L E 2 Prevalence of different ranges of LSM in the general general population (ETHON) cohort included in the analysis population (ETHON) cohort General population cohort N = 11 440 LSM ranges Patients, n (%) Age, years 51 (42–­60) < 8 kPa 10 797 (94.39) LSM ≥ 8 kPa, n (%) Male, n (%) 4792 (41.9) 8–­10 kPa 344 (3.00) 643 (5.61) LSM ≥ 10 kPa, Geographic subcohort, n (%) n (%) Cantabria 5090 (44) 10–­15 kPa 185 (1.62) 299 (2.60) Madrid 4088 (36) 15–­20 kPa 36 (0.31) Valencia 2262 (20) ≥ 20 kPa 78 (0.68) Caucasian, n (%) 10 058 (87.9) Total 11 440 (100) Alcohol risk consumption, n (%) 442 (3.9) Abbreviation: LSM, liver stiffness measurements. Body mass index (BMI), kg/m2 26.1 (23.3–­29.3) Weight, n (%) Normal weight (<25 kg/m2) 3846 (40.4) Subjects from Cantabria represented 44% of ETHON dataset. Overweight (≥25–­<30 kg/m ) 2 3597 (37.8) Within this subcohort, 23.4% were obese, 61.4% presented a BMI ≥ 25 kg/m2 and 20% met metabolic syndrome criteria. Median LSM Obesity (≥30 kg/m2) 2075 (21.8) and CAP values from Cantabria were 4.4 kPa and 247 dB/m respec- Waist circumference, cm 90 (80–­99) tively. As shown, baseline features between the Cantabrian subco- Type 2 diabetes, n (%) 1540 (13.5) hort and the whole ETHON dataset were comparable. Arterial hypertension, n (%) 5206 (53.6) Metabolic estimates from ETHON cohort were consistent with Dyslipidemia, n (%) 7418 (64.8) available data from Spain's general population, where 22.9%, 13.8% Metabolic syndrome, n (%) 1764 (15.4) and 42.6% have been estimated to present obesity, diabetes and ar- Fasting glucose, (mg/dl) 86 (79–­96) terial hypertension respectively.36–­39 Total cholesterol, (mg/dl) 197 (174–­222) HDL, (mg/dl) 57 (47–­68) LDL, (mg/dl) 113 (91–­135) 3.2 | STEP 1: TE ranges Triglycerides, (mg/dl) 114 (78–­172) Creatinine, mg/dl 0.78 (0.67–­0.91) The first step consisted of the estimation of LSM ranges in the gen- AST, (U/L) 22 (19–­27) eral population. Out of the 11 440 individuals with reliable TE from ETHON cohort, 5.61% (95%CI 2.53–­11.97) had LSM ≥ 8 kPa and ALT, (U/L) 20 (16–­28) 2.60% (95%CI 0.83–­7.88) presented LSM ≥ 10 kPa (Table 2). Within ALP, (U/L) 68 (56–­83) the LSM ≥ 8 kPa subgroup, 53% (344/643*100) of patients fell in the GGT, (U/L) 20 (14–­33) 8–­10 kPa range. Bilirubin, mg/dl 0.50 (0.40–­0.65) Albumin, g/dl 4.5 (4.3–­4.6) Platelets, x10E9/L 241 (205–­282) 3.3 | STEP 2: Estimation of the prevalence FIB–­4 index 0.99 (0.72–­1.36) attributable to NAFLD within the LSM ≥8 HBsAg positive, n (%) 90 (0.8) kPa subcohort Anti-­HCV positive, n (%) 143 (1.3) Liver stiffness, (kPa) 4.5 (3.6–­5.6) The second step was to estimate the aetiologic relative weight of CAP, (dB/m) a 247 (209–­293) NAFLD within the LSM ≥ 8 kPa population. For this specific purpose, as mentioned above, we used the subcohort from Cantabria, which Note: Risk alcohol consumption: ≥15 units of alcohol/week. Hypertension: ≥ 140/90 mmHg or requiring treatment; type 2 diabetes: was the largest subcohort and contained 4714/4728 (99.7%) CAP as a fasting plasma glucose ≥126 mg/dl or a non-­fasting plasma glucose measurements from the whole ETHON dataset. ≥180 mg/dl or requiring treatment.; dyslipidemia: serum triglycerides As seen in Figure 2, the proportion attributable to NAFLD was ≥150 mg/dl and/or total cholesterol >200 mg/dl, LDL >130 mg/dl, 57.3% (109/190*100), calculated as the ratio of the remaining sub- HDL < 40 mg/dl in men and < 50 mg/dl in women or requiring treatment. a jects after exclusion of patients with viral hepatitis, high-­risk alco- Data from Cantabrian subcohort, N = 4714. hol consumption and those with the absence of steatosis assessed by CAP, over the total of patients with LSM ≥ 8 kPa and available obese. Diabetes prevalence was 13.5% and 5206 patients (53.6%) CAP. Thus, the prevalence of individuals with LSM ≥ 8 kPa because had arterial hypertension. Almost 16% of the population met NCEP-­ of NAFLD in the general population was estimated to be 3.21% ATP III criteria for metabolic syndrome.35 (5.61*0.573). 6 | CALLEJA et al. 3.4 | Patient characteristics from the biopsy-­ The distribution of fibrosis by different LSM ranges for proven NASH cohort the remaining 389 patients are shown in Figure 3 and Table 3. Among patients with LSM ≥ 8 kPa, 37% (143/389) had F0-­1 and, The real clinical practice dataset was composed of 501 consecu- for those patients with LSM ≥ 10 kPa, as much as 44% (123/277) tive patients with histological confirmation of NASH and paired did not show advanced fibrosis (F3-­4) at histology. As seen, the TE measurements from five tertiary hospitals in Spain. The main proportion of patients with advanced fibrosis increased at each clinical features were usual in European NASH cohorts. Median LSM interval, but even at the highest interval (LSM ≥ 20 kPa), age was 59 years and 57% were males. Median BMI was 32 kg/m2 , there was a substantial proportion of patients without cirrhosis 66% of patients had obesity and 44% type 2 diabetes. Median time (39%). between TE and liver biopsy was 2.29 months (IQR 1.12–­4.41) Regarding liver biopsy quality parameters, median tissue length and only 12.5% (63/501) of patients presented a time interval > among patients with LSM ≥ 8 kPa was 23.0 mm (IQR 19.0–­28.0), and 6 months between procedures. A detailed description of the base- only 5.1% of samples were fragmented into more than two pieces. line characteristics from the biopsy-­p roven NASH cohort and the Of note, the subset of patients within the highest LSM interval comparison between both study cohorts are provided in Table S1. (≥ 20 kPa) showed similar quality data (21.5 mm median size -­IQR 17.0–­27.0-­and 4.5% fragmentation respectively). 3.5 | STEP 3: Description of the distribution of histological NASH-­related fibrosis according to 3.6 | STEP 4: Prevalence estimation of NASH-­ different LSM ranges related fibrosis stages for Spain's general population (2015–­2020) Among the 501 patients of the whole biopsy-­proven NASH cohort, 112 patients had LSM < 8 kPa. Among these patients, 92 (82%) were The main study outcome was calculated by merging the main results F0-­1, 15 presented F2, only 5 were F3 and no patients had cirrhosis from the two cohorts. The prevalence for each NASH-­related fibro- on histology. sis stage in Spain's general population was estimated by crossing the 100% F I G U R E 3 Distribution of liver fibrosis within increasing ranges of LSM in the 90% biopsy-­proven cohort of patients with 80% non-­alcoholic steatohepatitis (NASH) 70% F4 % Patients 60% 50% F3 40% F2 30% F1 20% F0 10% 0% 8-10 kPa 10-15 kPa 15-20 kPa ≥ 20 kPa Liver stiffness TA B L E 3 Distribution of histological LSM ranges, n (%) liver fibrosis according to LSM in the LSM ≥ LSM ≥ biopsy-­proven NASH cohort Fibrosis 8–­10 kPa 10–­15 kPa 15–­20 kPa ≥ 20 kPa 8 kPa 10 kPa F0 25 (22.32) 23 (14.84) 3 (5.66) 1 (1.44) 52 (13.37) 27 (9.75) F1 37 (33.04) 41 (26.45) 6 (11.32) 7 (10.15) 91 (23.39) 54 (19.49) F2 28 (25.00) 28 (18.07) 7 (13.20) 7 (10.15) 70 (18.00) 42 (15.16) F3 19 (16.96) 43 (27.74) 17 (32.08) 12 (17.39) 91 (23.39) 72 (26.00) F4 3 (2.68) 20 (12.90) 20 (37.74) 42 (60.87) 85 (21.85) 82 (29.60) Total 112 (100) 155 (100) 53 (100) 69 (100) 389 (100) 277 (100) Abbreviation: LSM, liver stiffness measurements. CALLEJA et al. | 7 estimated prevalence of NAFLD with LSM ≥ 8 kPa (3.21%) drawn 4 | DISCUSSION from ETHON dataset with the different fibrosis stages probabilities observed in the biopsy-­proven NASH cohort (detailed in Table 3—­ In the present study, we provide an updated estimation of the prev- LSM ≥ 8 kPa column). The final calculations from our study are alence of NASH-­related fibrosis in Spain's general population, by shown in Table 4 (i.e., prevalence of F2 fibrosis stage resulted in combining TE data from a large population-­based screening cohort, 0.58% = 3.21*0.18). Finally, the estimated prevalence of NASH with and histological information from a contemporary multi-­centre co- significant fibrosis F2-­4 in Spain was 2.03 (95% CI 0.56–­7.05), 1.33 hort of biopsy-­proven NASH from Spain. We observed that 5.61% (95% CI 0.29–­5.98) of the Spanish population aged 20–­79 years pre- of the general population had LSM ≥ 8 kPa, 57.3% of which (3.21%) sented NASH F2-­3 and 0.70 (95% CI 0.10–­4.95) was estimated to was attributable to NAFLD. By crossing these estimates with the have NASH cirrhosis. corresponding probabilities for the different fibrosis stages by LSM Additional estimations using the LSM ≥ 10 kPa threshold and al- intervals drawn from the biopsy-­proven cohort, we estimated that ternative approximations to the proportion attributable to NAFLD the current prevalence of NASH with significant fibrosis in Spain was using more or less conservative CAP thresholds (275 or 220 dB/m, 2.03% (1.33% NASH F2-­3 and 0.70% NASH cirrhosis). respectively) are provided in Tables S2–­S5. The estimates on NASH with significant fibrosis from our study are lower than those provided in previous epidemiological reports from other European cohorts,28,40–­42 even though the prevalence of LSM 3.7 | Prevalence estimates with LSM ≥ 8 kPa in the ETHON population is almost the same. The most likely 9.1 kPa threshold reason for these differences stems from the fact that fibrosis estima- tions in those studies were fundamentally TE-­based (backed up by the The LSM ≥ 9.1 kPa threshold has been suggested as the most cost-­ absence of liver biopsies or a relatively small number of them) and, in effective cut-­off for a theoretical TE-­based screening programme the present study, TE estimations were back-­tested against a large of liver fibrosis in populations with lower prevalence of risk alco- contemporary histological dataset, which clearly suggests that current hol consumption, such as Spain. 28 Applying the same methodol- TE-­based definitions are overestimating the prevalence of fibrosis in ogy used for the 8 kPa threshold we found that in ETHON cohort, NASH. In fact, 37% of the patients in our study with LSM ≥ 8 kPa did prevalence of LSM ≥ 9.1 kPa was 3.36%. In our biopsy-­proven co- not have “significant fibrosis,” and 44% of patients with LSM ≥ 10 kPa hort, 66.5% (223/335) of patients had significant fibrosis (F2-­4) and did not have “advanced fibrosis” at histology. A recent paper,28 inte- 49.8% (167/335) presented advanced fibrosis (F3-­4). The propor- grating population-­based TE data with histological data from 6 inde- tion attributable to NAFLD resulted in 55.2% of cases of all Spain's pendent cohorts, with 6300 patients and 350 biopsies, showed an population with LSM ≥ 9.1 kPa, and thus 1.85% of Spain's popula- estimated prevalence of F2-­4 in the general population (all aetiologies tion would be assumed to have NAFLD, with 1.10% estimated to combined) of 3.9%. In our study, there were more than 12 000 patients have F2-­4. in the population-­based cohort and more than 500 biopsies, which Of note, raising the threshold from 8.0 to 9.1 kPa in a hypotheti- served at narrowing confidence intervals for the different degrees cal screening plan would represent a 39.9% relative reduction in the of fibrosis that could be expected at each TE interval, providing more proportion of patients that would be targeted. However, according precise references for the population-­based estimates. As a practical to our biopsy-­proven cohort, in the subcohort of 54 patients with consequence, we firmly believe that current TE-­based definitions of LSM 8–­9.1 kPa, 9 (16.6%) had advanced fibrosis. The advantages and “significant” and “advanced” fibrosis (LSM ≥ 8 kPa and 10 kPa, respec- risks of using the 9.1 kPa threshold should be weighted in dedicated tively) should be avoided, since they provide unrealistic overestima- cost-­effective studies. tions of the actual prevalence of the corresponding fibrosis stages. TA B L E 4 Prevalence estimation of the Nash fibrosis prevalence (%) different NASH fibrosis stages in Spain's general population Fibrosis LSM ≥ 8 kPa (95%CI) Significant Intermediate Cirrhosis (F4) fibrosis stages (F2-­F3) (F2-­F4) F0 0.43 (0.04–­4.49) F1 0.75 (0.11–­5.04) F2 0.58 (0.07–­4.75) 2.03 1.33 (0.29–­5.98) F3 0.75 (0.11–­5.04) (0.56–­7.05) F4 0.70 (0.10–­4.95) 0.70 (0.10–­4.95) Total 3.21 (1.13–­8.75) Abbreviations: CI, confidence interval; LSM, liver stiffness measurements; NASH, non-­alcoholic steatohepatitis. 8 | CALLEJA et al. The second reason for the different prevalence of NASH F2-­4 is general population level. This issue can be partially mitigated in two the more comprehensive attribution of aetiology made in our study, ways. First, by adjusting CAP values according to influencing covari- where we estimated that approximately 60% of patients with LSM ables, such as type-­2 diabetes, BMI and liver aetiology.18 Second, as ≥ 8 kPa from the general population had NAFLD. In the paper by provided in the present study, by generating alternative estimations Caballeria et al.,40 attribution of NAFLD aetiology was made by ex- using different CAP cutoffs with greater or lesser sensitivity (includ- clusion of with viral hepatitis and patients with high-­risk alcohol con- ing the 275 dB/m threshold suggested in the latest clinical practice sumption, resulting in a remarkably high aetiologic weight of NAFLD guideline).17 Moreover, the estimation on the relative weight of (near 90%). On the other hand, in the other two large European stud- NAFLD would require adjustment by the alcoholic liver disease bur- ies,41,42 NAFLD was suspected basically on clinical grounds, yielding den in each country. There will be also patients with other liver ae- a much lower estimated prevalence (32%–­42% for the whole LSM tiologies, although their reported frequencies in primary care series spectrum). By contrast, to estimate the proportion attributable to are very low (<2%) and should not distort significantly large-­scale NAFLD we added the use of CAP in nearly 5000 patients from a rep- epidemiological estimations.43 Finally, the ethnic background in our resentative subcohort, enabling the extrapolation of the results to study is predominantly Caucasian, so extrapolation to populations the whole dataset and, thus, to the general population. We believe with more diverse ethnicities should be done with caution. that the combination of CAP along clinical and laboratory values in In conclusion, our large-­scale epidemiological study, merging the present study likely provides the most accurate estimation for population-­based TE data with histological data from real-­world the attribution of aetiology in population-­based studies to date. practices, provides accurate and updated estimates of the current The study has nonetheless some limitations that should be taken prevalence of NASH-­related fibrosis in Spain, with a special focus on into account when interpreting the transferability of our results to those stages that should be the target of early detection and referral real practice decisions. The main limitation is the risk of overestima- policies to identify high-­risk patients who will benefit more from the tion of the prevalence of F2-­4 stages, as consequence of unavoidable pharmacological therapies under development. referral and selection bias. It has been consistently shown that NASH Until large-­scale prospective epidemiological data with biopsy prevalence is overestimated in tertiary settings. 2 Local initiatives for confirmation becomes available, our results could serve as refer- structured early detection and referral have been developed in re- ence points for both assessing the current burden of disease and for cent years in the biopsy-­proven cohort centres and may have helped modelling changes in time, so health policies can be designed and at buffering that risk. However, the efficiency of systematic refer- adapted accordingly, and besides for dimensioning the therapeutic ral is still suboptimal, and thus mitigation of referral bias is probably efforts that will be required as the time when we have effective modest. The risk of selection bias is also unavoidable in this kind of treatments to treat NASH approaches. retrospective studies since decision for biopsy was not standardized and could have been influenced by other criteria different from LSM. AC K N OW L E D G E M E N T S However, it should be noted that current practice in all participating The authors thank Intercept Pharma Spain and EY Spain, which centres is quite homogenous and representative of what is done in helped at processing databases and providing descriptive analyses. daily clinical practice in Spain. In fact, the 2 main centres recruiting Neither Intercept Pharma Spain nor EY Spain funded the study or for the ETHON cohort were also part of the contemporary biopsy-­ had a role in the study design, interpretation of the data, writing or proven cohort (Cantabria, Madrid). In any case, the ideal approach to revision of the manuscript or the decision to submit the manuscript minimize these biases would be the implementation of prospective for publication. The authors also thank to LITMUS Consortium (Liver large-­scale screening programmes, with per-­protocol liver biopsies Investigation: Testing Marker Utility in Steatohepatitis LITMUS above a pre-­specified LSM threshold. Such a study is already ongo- 777377 IMI-­2 European Commission). ing (LiverScreen, H2020-­EU ID 847989), but results of such large efforts will take long. Thus, until this sort of large-­scale prospec- F U N D I N G I N FO R M AT I O N tive data becomes available, the estimations provided in the present The work was independent of all funding. JRE is a PhD student study remain as the most accurate approximations to the prevalence at Universitat Autònoma de Barcelona, Spain. JC is a receptor of of NASH with significant fibrosis in a European population. a grant of Fondo Investigaciones Sanitarias (FIS): PI18/01304. Other limitations should be considered. Biopsy reading was Immunomediated Nonalcoholic SteaTohepatItis; prevalence and not centralized and could be subject of interobserver variability, al- CharacTerization. INSTInCT. though we believe this is a reflection of what should be expected in real-­life situations. We acknowledge that the attribution of the C O N FL I C T O F I N T E R E S T relative weight of NAFLD as aetiology was imperfect. The choice JLC received consulting and lecture fees for Intercept, Echosens of the CAP threshold was arbitrary, since there is no clear consen- and Gilead Sciences. JMP received consulting, educational or sus in the literature and there are no clear published reports on the research grants for Gilead, NovoNordisk, Novartis, Boehringer-­ specificity and positive predictive values of CAP estimates for pa- Ingelheim, BMS, Pfizer, Accelerate, Astellas, ViiV, Janssen, MSD, tients with higher LSM.10 It can be also argued about the positive Abbie. MRG received consulting fees from Alpha-­sigma, Allergan, predictive value of 250 dB/m CAP threshold to identify steatosis at BMS, Boehringer-­Ingelheim, Gilead, Intercept, Kaleido, MSD, CALLEJA et al. | 9 5. Younossi Z, Tacke F, Arrese M, et al. Global perspectives on non- Pfizer, Prosciento, Shionogi, Sobi, Zydus and research Grants: alcoholic fatty liver disease and nonalcoholic steatohepatitis. Gilead, Intercept. SA changed affiliations after finalizing the manu- Hepatology. 2019;69(6):2672-­2682. script and its submission to the journal. Boehringer Ingelheim did 6. Estes C, Anstee QM, Arias-­Loste MT, et al. Modeling NAFLD dis- ease burden in China, France, Germany, Italy, Japan, Spain, United not fund the study and did not have any role in the study design, Kingdom, and United States for the period 2016–­2030. J Hepatol. analysis or interpretation of the data, writing or revision of the 2018;69(4):896-­904. manuscript or the decision to submit the manuscript for publica- 7. Younossi Z, Stepanova M, Ong JP, et al. Nonalcoholic steatohepati- tion. Other potential conflicts of interest SA: consulting fees from tis is the fastest growing cause of hepatocellular carcinoma in liver Boehringer Ingelheim, Ferrer, Gilead, Intercept, IQVIA, Novartis, transplant candidates. Clin Gastroenterol Hepatol. 2019;17(4):748-­ 755.e3. Pfizer; speaking fees from Allergan, Gilead, MSD and Novartis; 8. Younossi ZM, Otgonsuren M, Henry L, et al. Association of non- travel expenses from Gilead, MSD, Janssen, Genfit, Bayer and alcoholic fatty liver disease (NAFLD) with hepatocellular carci- Ferring; grant support from Gilead. JC received consulting fees noma (HCC) in the United States from 2004 to 2009. Hepatology. from Gilead, Intercept, MSD, Abbie; grant support from Gilead, 2015;62(6):1723-­1730. 9. Vilar-­Gomez E, Chalasani N. Non-­invasive assessment of non-­ MSD. alcoholic fatty liver disease: clinical prediction rules and blood-­ based biomarkers. J Hepatol. 2018;68(2):305-­315. E T H I C S S TAT E M E N T 10. Castera L, Friedrich-­Rust M, Loomba R. Noninvasive assessment The study protocol was approved by Vall d'Hebron Ethics of liver disease in patients with nonalcoholic fatty liver disease. Gastroenterology. 2019;156(5):1264-­1281.e4. Committee for Clinical Research (PR[AG]655/2020) and con- 11. Younossi ZM, Noureddin M, Bernstein D, et al. Role of noninva- formed to the ethical guidelines of the 1975 Declaration of sive tests in clinical gastroenterology practices to identify pa- Helsinki. tients with nonalcoholic steatohepatitis at high risk of adverse outcomes: expert panel recommendations. Am J Gastroenterol. 2021;116(2):254-­262. PAT I E N T C O N S E N T FO R P U B L I C AT I O N 12. Castera L, Yuen Chan HL, Arrese M, et al. EASL-­ALEH Clinical Not required. Practice Guidelines: non-­invasive tests for evaluation of liver dis- ease severity and prognosis. J Hepatol. 2015;63(1):237-­264. P E R M I S S I O N TO R E P R O D U C E M AT E R I A L F R O M 13. Huber Y, Schulz A, Schmidtmann I, et al. Prevalence and risk factors of advanced liver fibrosis in a population-­based study in Germany. OT H E R S O U R C E S Hepatol Commun. 2022;6:1457-­1466. The material in this paper is original and does not come from other 14. Alexander M, Loomis AK, Fairburn-­Beech J, et al. Real-­world data sources. reveal a diagnostic gap in non-­alcoholic fatty liver disease. BMC Med. 2018;16(1):130. ORCID 15. Yamazaki H, Tsuboya T, Tsuji K, Dohke M, Maguchi H. Independent association between improvement of nonalcoholic fatty liver dis- José L. Calleja https://orcid.org/0000-0002-2265-6591 ease and reduced incidence of type 2 diabetes. Diabetes Care. Jesús Rivera-­Esteban https://orcid.org/0000-0003-4357-8817 2015;38(9):1673-­1679. Marta Hernández-­Conde https://orcid. 16. Vernon G, Baranova A, Younossi ZM. Systematic review: The epi- org/0000-0002-7153-0493 demiology and natural history of non-­alcoholic fatty liver disease and non-­alcoholic steatohepatitis in adults. Aliment Pharmacol Ther. Juan M. Pericàs https://orcid.org/0000-0002-3645-3293 2011;34(3):274-­285. Javier Ampuero https://orcid.org/0000-0002-8332-2122 17. Berzigotti A, Tsochatzis E, Bousier J, et al. EASL Clinical Practice María T. Arias-­Loste https://orcid.org/0000-0001-8864-3833 Guidelines on non-­invasive tests for evaluation of liver disease se- Paula Iruzubieta https://orcid.org/0000-0001-9476-1801 verity and prognosis -­2021 update. J Hepatol. 2021;75(3):659-­689. 18. Eddowes PJ, Sasso M, Allison M, et al. Accuracy of fibroscan con- Manuel Romero-­Gómez https://orcid.org/0000-0001-8494-8947 trolled attenuation parameter and liver stiffness measurement in Salvador Augustin https://orcid.org/0000-0002-3515-9033 assessing steatosis and fibrosis in patients with nonalcoholic fatty Javier Crespo https://orcid.org/0000-0001-8248-0172 liver disease. Gastroenterology. 2019;156(6):1717-­1730. 19. Karlas T, Petroff D, Sasso M, et al. Individual patient data meta-­ analysis of controlled attenuation parameter (CAP) technology for REFERENCES assessing steatosis. J Hepatol. 2017;66(5):1022-­1030. 1. Younossi Z, Anstee QM, Marietti M, et al. Global burden of NAFLD 20. Bril F, Ortiz-­Lopez C, Lomonaco R, et al. Clinical value of liver ultra- and NASH: trends, predictions, risk factors and prevention. Nat Rev sound for the diagnosis of nonalcoholic fatty liver disease in over- Gastroenterol Hepatol. 2018;15(1):11-­20. weight and obese patients. Liver Int. 2015;35(9):2139-­2146. 2. Younossi ZM, Koenig AB, Abdelatif D, Fazel Y, Henry L, Wymer 21. Fabrellas N, Hernández R, Graupera I, et al. Prevalence of hepatic M. Global epidemiology of nonalcoholic fatty liver disease—­Meta-­ steatosis as assessed by controlled attenuation parameter (CAP) in analytic assessment of prevalence, incidence, and outcomes. subjects with metabolic risk factors in primary care. A population-­ Hepatology. 2016;64(1):73-­8 4. based study. PLoS One. 2018;13(9):e0200656. 3. Allen AM, Therneau TM, Larson JJ, Coward A, Somers VK, Kamath 22. Lee JY, Kim KM, Lee SG, et al. Prevalence and risk factors of non-­ PS. Nonalcoholic fatty liver disease incidence and impact on met- alcoholic fatty liver disease in potential living liver donors in Korea: abolic burden and death: a 20 year-­community study. Hepatology. A review of 589 consecutive liver biopsies in a single center. J 2018;67(5):1726-­1736. Hepatol. 2007;47(2):239-­244. 4. Cotter TG, Rinella M. Nonalcoholic fatty liver disease 2020: the 23. Singh S, Allen AM, Wang Z, Prokop LJ, Murad MH, Loomba R. state of the disease. Gastroenterology. 2020;158(7):1851-­1864. Fibrosis Progression in Nonalcoholic Fatty Liver vs Nonalcoholic 10 | CALLEJA et al. Steatohepatitis: A Systematic Review and Meta-­analysis of Paired-­ 36. Soriguer F, Goday A, Bosch-­Comas A, et al. Prevalence of diabetes Biopsy Studies. Clin Gastroenterol Hepatol. 2015;13(4):643-­654.e9. mellitus and impaired glucose regulation in Spain: the di@bet.es 24. Bedossa P, Patel K. Biopsy and noninvasive methods to assess study. Diabetologia. 2012;55:88-­93. progression of nonalcoholic fatty liver disease. Gastroenterology. 37. Gutiérrez-­Fisac JL, Guallar-­C astillón P, León-­Muñoz LM, Graciani 2016;150(8):1811-­1822.e4. A, Banegas JR, Rodríguez-­Artalejo F. Prevalence of general and ab- 25. Wong VWS, Wong GLH, Choi PCL, et al. Disease progression of dominal obesity in the adult population of Spain, 2008-­2010: the non-­alcoholic fatty liver disease: a prospective study with paired ENRICA study. Obes Rev. 2012 Apr;13(4):388-­392. liver biopsies at 3 years. Gut. 2010;59(7):969-­974. 38. Aranceta-­Bartrina J, Pérez-­Rodrigo C, Alberdi-­Aresti G, Ramos-­ 26. Harris R, Harman DJ, Card TR, Aithal GP, Guha IN. Prevalence of Carrera N, Lázaro-­Masedo S. Prevalence of general obesity and clinically significant liver disease within the general population, as abdominal obesity in the spanish adult population (aged 25-­ defined by non-­invasive markers of liver fibrosis: a systematic re- 64 years) 2014-­2015: the ENPE study. Rev Esp Cardiol (Engl Ed). view. Lancet Gastroenterol Hepatol. 2017;2(4):288-­297. 2016;69(6):579-­587. 27. Ginès P, Graupera I, Lammert F, et al. Screening for liver fibrosis 39. Menendez E, Delgado E, Fernandez-­Vega F, et al. Prevalence, diag- in the general population: a call for action. Lancet Gastroenterol nosis, treatment, and control of hypertension in Spain. Results of Hepatol. 2016;1(3):256-­260. the di@bet.es study. Rev Esp Cardiol (Engl Ed). 2016;69(6):572-­578. 28. Serra-­Burriel M, Graupera I, Torán P, et al. Transient elastography for 40. Caballería L, Pera G, Arteaga I, et al. High prevalence of liver fibrosis screening of liver fibrosis: Cost-­effectiveness analysis from six pro- among european adults with unknown liver disease: a population-­ spective cohorts in Europe and Asia. J Hepatol. 2019;71(6):1141-­1151. based study. Clin Gastroenterol Hepatol. 2018;16(7):1138-­1145.e5. 29. von Elm E, Altman DG, Egger M, Pocock SJ, Gøtzsche PC, 41. Roulot D, Costes J-­L , Buyck J-­F, et al. Transient elastography as a Vandenbroucke JP. The Strengthening the Reporting of Observational screening tool for liver fibrosis and cirrhosis in a community-­based Studies in Epidemiology (STROBE) statement: guidelines for report- population aged over 45 years. Gut. 2011;60(7):977-­984. ing observational studies. Lancet. 2007;370(9596):1453-­1457. 42. Koehler EM, Schouten JNL, Hansen BE, et al. Prevalence and risk 30. Llop E, Iruzubieta P, Perelló C, et al. High liver stiffness values factors of non-­alcoholic fatty liver disease in the elderly: results by transient elastography related to metabolic syndrome and from the Rotterdam study. J Hepatol. 2012;57(6):1305-­1311. harmful alcohol use in a large Spanish cohort. United European. 43. Armstrong MJ, Houlihan DD, Bentham L, et al. Presence and se- Gastroenterol J. 2021;9:1-­10. verity of non-­alcoholic fatty liver disease in a large prospective pri- 31. Crespo J, Cuadrado A, Perelló C, et al. Epidemiology of hepatitis C mary care cohort. J Hepatol. 2012;56(1):234-­240. virus infection in a country with universal access to direct-­acting antiviral agents: Data for designing a cost-­effective elimination pol- icy in Spain. J Viral Hepat. 2020;27(4):360-­370. S U P P O R T I N G I N FO R M AT I O N 32. Ampuero J, Aller R, Gallego-­Durán R, et al. Significant fibrosis pre- Additional supporting information may be found in the online dicts new-­onset diabetes mellitus and arterial hypertension in pa- version of the article at the publisher’s website. tients with NASH. J Hepatol. 2020;73(1):17-­25. 33. Boursier J, Zarski JP, de Ledinghen V, et al. Determination of reli- ability criteria for liver stiffness evaluation by transient elastogra- phy. Hepatology. 2013;57(3):1182-­1191. 34. Kleiner DE, Brunt EM, Van Natta M, et al. Design and validation of How to cite this article: Calleja JL, Rivera-­Esteban J, Aller R, a histological scoring system for nonalcoholic fatty liver disease. et al. Prevalence estimation of significant fibrosis because of Hepatology. 2005;41(6):1313-­1321. 35. Tobin J, Zazula T. National cholesterol education program (ncep). In: NASH in Spain combining transient elastography and Anderson NB, ed. Encyclopedia of health and behavior. Vol 1. SAGE histology. Liver Int. 2022;00:1-10. doi: 10.1111/liv.15323 Publications, Inc; 2004:541-­533.