Received: 10 April 2022 | Accepted: 30 June 2022 DOI: 10.1002/hep4.2056 ORIGINAL ARTICLE High frequency of acute decompensation and cancer in patients with compensated cirrhosis due to nonalcoholic fatty liver disease: A retrospective cohort study Octavi Bassegoda1 | Jesús Rivera-­Esteban2 | Isabel Serra3,4 | Rosa Morillas4,5,6 | Teresa Broquetas4,7 | Mercedes Vergara4,8 | Adrià Rodriguez4,9 | Carles Aracil4,10 | Silvia Virolés4,11 | Jose A. Carrión4,7 | Albert Pardo4,9 | Sergio Rodríguez-­Tajes1 | Miquel Serra-­Burriel12 | Juan M. Pericàs2,4 | Salvador Augustin2,4 | Pere Ginès1,4 | Isabel Graupera1,4 1 Servei d'Hepatologia, Hospital Clínic, Facultat de Medicina i Ciències de la Salut, Universitat de Barcelona, Institut d'Investigacions Biomèdiques August Pi i Sunyer, Centro de Investigaciones en Red de Enfermedades Hepáticas y Digestivas (CIBERehd), Barcelona, Spain 2 Servei de Medicina Interna, Hepatologia, Hospital Universitari Vall d'Hebron, Institut de Recerca, Universitat Autònoma de Barcelona, CIBERehd, Instituto de Salud Carlos III, Barcelona, Spain 3 Departament d'hepatologia de l'Hospital Dr Josep Trueta Girona, Secció d'Hepatologia, Girona, Spain 4 Societat Catalana de Digestologia, Acadèmia de Ciències Mèdiques i de la Salut de Catalunya i de Balears, Barcelona, Spain 5 Secció de Hepatologia, Servei Aparell Digestiu, Hospital Germans Trias i Pujol, Barcelona, Spain 6 Universitat Autònoma de Barcelona, CIBERehd, Barcelona, Spain 7 Secció d'Hepatologia, Servei de Digestiu, Hospital del Mar, Institut Hospital del Mar d'Investigacions Mèdiques, Universitat Autònoma de Barcelona, Barcelona, Spain 8 Unidad Hepatología, Servicio Digestivo, Hospital Universitari Parc Taulí, Institut d'Investigació i Innovació Parc Taulí I3PT, Universitat Autònoma de Barcelona, CIBERehd, Sabadell, Spain 9 Servicio de Aparato Digestivo, Hospital Universitari de Tarragona Joan XXIII, Tarragona, Spain 10 Institute of Biomedical Research, Arnau de Vilanova University Hospital, Lleida, Spain 11 Unitat de Digestiu-­Servei Medicina Interna, Hospital de Figueres, Institut d'Investigació Biomèdica de Girona, Girona, Spain 12 Epidemiology, Biostatistics, and Prevention Institute, University of Zurich, Zurich, Switzerland Correspondence Salvador Augustin, Servei de Medicina Abstract Interna –­Hepatologia, Hospital Universitari Vall d'Hebron -­Institut de The natural history of compensated cirrhosis due to nonalcoholic fatty liver Recerca, Universitat Autònoma de disease (NAFLD) has not been completely characterized. The aim of the Barcelona, CIBERehd -­Instituto de Salud Carlos III, Barcelona, Spain. present study was to assess the incidence and risk factors of acute decom- Email: salva.augustin@gmail.com pensation of cirrhosis, hepatocellular carcinoma, and extrahepatic cancers. Isabel Graupera, Servei d'Hepatologia, Hospital Clínic, Facultat de Medicina This was a multicenter, retrospective, cohort study including 449 patients with i Ciències de la Salut, Universitat de Barcelona, IDIBAPS, CIBERehd, compensated cirrhosis due to NAFLD. We calculated cumulative incidences Barcelona, Spain. and used competitive risk analysis to determine the risk factors associated Email: igraupe@clinic.cat with decompensation and cancer development. Over a median of 39 months Funding information the European Regional Development of follow-­up, 124 patients (28%) presented acute decompensation. The most Fund, Grant/Award Number: PI120/00579 Octavi Bassegoda and Jesús Rivera-­Esteban contributed equally to this work as first authors. Pere Ginès and Isabel Graupera contributed equally to this work as senior 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. © 2022 The Authors. Hepatology Communications published by Wiley Periodicals LLC on behalf of American Association for the Study of Liver Diseases. 3212 | wileyonlinelibrary.com/journal/hep4  Hepatology Communications. 2022;6:3212–3222. HEPATOLOGY COMMUNICATIONS     | 3213 and PI18/00862; Fundación de Investigación Sanitaria, ISCIII-­ frequent decompensation was ascites (21%) followed by hepatic encepha- Subdirección General de Evaluación; Contractes Clínic de Recerca Emili lopathy (15%), variceal bleeding (9%), and spontaneous bacterial peritonitis Letang-­Josep Font 2020, Grant/Award (3%). Acute-­on-­chronic liver failure was diagnosed in 6% of patients during Number: CM17/00015 follow-­up. Liver function parameters and specifically an albumin level below 40 g/L were independently associated with an increased risk of decompensa- tion. The presence of ischemic heart disease was independently associated with acute decompensation. Seventy-­eight patients (18%) developed hepa- tocellular carcinoma or extrahepatic cancers during follow-­up (51 and 27, ­respectively). Conclusion: Patients with compensated cirrhosis due to NAFLD are at high risk of severe liver complications, such as the ­development of acute decompensation, in a relative short follow-­up time. This population is at high risk of hepatic and extrahepatic cancers. I NTRO DUCTI O N risk factors for disease progression and decompen- sation is needed. Also, there is a lack of information Cirrhosis is the end stage of chronic liver diseases about the risk of cancer development in this population (CLDs) and is associated with high morbidity and mor- with advanced disease because most of the research tality. Mortality due to cirrhosis has increased by 15% in the field has been performed in low and intermedi- from 2007 to 2017[1] and is expected to increase even ate stages of fibrosis. Moreover, NAFLD is considered further in the future due to an increased prevalence of the hepatic expression of the metabolic syndrome, a nonalcoholic fatty liver disease (NAFLD). NAFLD is condition that can affect many organs. For that rea- the most common cause of CLD worldwide, and re- son, patients with NAFLD and cirrhosis could be at cent data suggest that NAFLD will be responsible for increased risk of acute-­on-­chronic liver failure (ACLF), most liver-­related complications, including hepatocellu- a syndrome characterized by systemic inflammation, lar carcinoma (HCC) and death, in the near future.[2] organ failure, and high mortality.[11] There are data We know from studies with paired-­liver biopsies that indicating that NAFLD is the fastest rising etiology of fibrosis stage is the most important predictor of mor- cirrhosis associated with ACLF among patients listed tality and progression to cirrhosis.[3–­5] However, there for liver transplantation.[12] No clinical study has so far is a paucity of information on the natural history of ad- addressed the incidence of ACLF or analyzed its pre- vanced NAFLD and the main risk factors for clinical dictive factors in this population. decompensation when cirrhosis is already established, The aims of the present study were (1) to describe mostly because studies in the field have mainly focused the natural history of compensated cirrhosis due to on early stages of the disease. NAFLD in a real-­life setting; and (2) to determine the Compensated cirrhosis is usually a long-­ standing risk factors of acute decompensation (AD), ACLF, sur- and asymptomatic phase followed by a decompen- vival, development of HCC, and extrahepatic cancers sated stage that usually leads to a dramatic increase during follow-­up. in morbidity and mortality.[6] Few studies address the natural history of advanced NAFLD, and these have been hampered by low sample size, short follow-­up, M ATER I A LS A ND M ETHO DS and inclusion of highly selected populations that are not representative of the actual population with NAFLD at Investigators of an association of academic and spe- advanced stages.[3,7–­9] A recent and high-­quality study cialized liver units designed a multicenter observational regarding the natural history of NAFLD has confirmed study involving a cohort of patients diagnosed with that patients with fibrosis stages F3 and F4 are those compensated cirrhosis secondary to NAFLD from nine with the highest risk of decompensation. However, in centers in Spain (see Supporting Materials). this prospective analysis, the population with cirrhosis was limited.[10] Most studies include patients at different stages of advanced fibrosis (F3 and F4), which have Study population different prognosis as recently demonstrated.[3,4,8,9] Therefore, there is a lack of data regarding the nat- The study population included all patients diag- ural history of NAFLD in a real-­life setting and when nosed with compensated cirrhosis due to NAFLD cirrhosis is already established. The identification of from January 2009 to December 2018. Patients were 3214 |    HIGH FREQUENCY OF ACUTE DECOMPENSATION AND CANCER IN PATIENTS identified retrospectively from electronic records at We defined the signs of ultrasonographic portal hy- each center. We searched for a combination of “stea- pertension as the presence of splenomegaly (≥13 cm), tosis,” “steatohepatitis,” “nonalcoholic fatty liver dis- portal-­systemic collaterals, inversion of flow within ease,” and “cirrhosis.” Matching data were individually the portal system, dilatation of the portal vein (diame- analyzed by investigators to account for the fulfillment ter ≥13 mm), or reduced portal vein flow velocity <10 cm/ of the inclusion and exclusion criteria (see below). second.[17,18] Endoscopic signs of portal hypertension Demographic and clinical characteristics; laboratory, were the presence of gastroesophageal varices or ultrasound, or computed tomography scan; and endo- portal hypertensive gastropathy.[18] Hepatic vein cath- scopic and histologic data were recorded. The inclu- eterization and hepatic venous pressure measurement sion date was the date of cirrhosis diagnosis by the (HVPG) were available in 86 patients (19%), and we reference hepatologist; this determined the start of used clinically significant portal hypertension (CSPH), follow-­up time. The maximum time difference accepted defined by HVPG value ≥10 mm Hg,[19] as a diagnostic between all baseline data and the inclusion date was criteria for portal hypertension. 6 months. We included only patients that were followed The primary outcome was acute decompensation up by a hepatologist of the same center after diagno- (AD), defined as new onset ascites; grade 2 hepatic sis. Investigators examined their electronic records in encephalopathy (HE), according to the West-­ Haven search of clinical events after the diagnosis of cirrhosis, classification[13]; portal-­hypertension-­related bleeding ensuring that all patients had a minimum of 1 year of (PHB), confirmed by endoscopy due to the presence follow-­up. Diagnosis, screening, and follow-­up of cir- of bleeding attributed to gastroesophageal varices or rhosis was performed using international guidelines.[13] hypertensive gastropathy and requiring hospital ad- The primary outcome was the development of the first mission; or spontaneous bacterial peritonitis (SBP), acute decompensation of cirrhosis during follow-­up as defined as the presence of ≥250 polymorphonuclear defined by the recent expert recommendations detailed leucocytes/mm3 in ascitic fluid.[14] Secondary out- below.[14] Secondary outcomes were the occurrence comes were development of HCC, ACLF, liver-­related of ACLF, death, and liver transplantation. Follow-­ up death,[20] death from any cause, cardiovascular events, lasted until the patient died, received a transplant, was and extrahepatic cancer, excluding nonmelanoma skin lost to follow-­up, or at the end of the study follow-­up in cancer. HCC was screened periodically and diagnosed December 2018. by noninvasive dynamic images or by invasive criteria, We excluded patients with alcohol consumption of according to European Association for the Study of the more than 21 standard drinks per week in men and 14 Liver (EASL) guideline recommendations.[21] ACLF was in women or with other concomitant causes of liver dis- defined according to the EASL-­Chronic Liver Failure ease. We excluded all patients who had HCC or any de- (CLIF) Acute-­ on-­Chronic Liver Failure in Cirrhosis compensation of liver disease at the time of diagnosis (CANONIC) study criteria described elsewhere.[11,13] of cirrhosis or previously. Patients with prior liver trans- Cardiovascular event was defined as acute coronary plantation were also excluded as were patients with ad- syndrome or acute stroke. See Supporting Materials for vanced extrahepatic diseases (metastatic cancer and further definitions. advanced heart or respiratory failure) that would impact When multiple decompensating events occurred in short-­term survival. the same patient, the earliest event was adopted for the analysis. The study was approved by the institutional review Definitions board at all participating centers and was conducted in compliance with the Declaration of Helsinki (Internal Diagnosis of NAFLD cirrhosis was made according to code HCP:2019/635). any of the following: (1) liver biopsy with ≥5% steatosis and/or steatohepatitis and a nonalcoholic steatohepa- titis Clinical Research Network score of 4 (F4)[15] or Statistical analysis cryptogenic cirrhosis in a patient with obesity, type 2 diabetes, or metabolic syndrome and no other detect- Normally distributed variables were reported as able etiology of liver disease; (2) radiologic or endo- means ± SD and non-­normally distributed variables as scopic signs of cirrhosis and portal hypertension in a median and interquartile range (IQR). We compared patient with obesity, type 2 diabetes, or metabolic syn- continuous variables using the two-­tailed Student t-­test. drome and no other detectable etiology of liver disease; Categorical variables were compared using the chi-­ (3) presence of steatosis by imaging techniques and square test. We calculated incidence rate as the number liver stiffness measurement (LSM) ≥18 kPa in a patient of new events during the study follow-­up divided by the with obesity, type 2 diabetes, or metabolic syndrome in sum of the years at risk for each patient. When analyzing the absence of other etiologies of cirrhosis and heart the cumulative probability of presenting AD, we found failure.[16] liver transplantation and non-­liver-­related deaths were HEPATOLOGY COMMUNICATIONS     | 3215 TA B L E 1 Baseline characteristics of the study population considered as competing events. We used Fine and Gray's proportional subhazards for multivariate mod- Variable n = 449 (%) eling. All variables clinically and statistically different Age, years 65 (14) (p < 0.1) were considered for multivariate modeling. The Sex, male 234 (52) number of variables included in the initial models followed Race (Caucasian/Black/Asian) 430/10/9 the rule of one variable per 10 outcomes. To avoid collin- Ethnicity (White/Latino/African/Asian) 408/22/10/9 earity, we did not include scores and variables included Current alcohol consumptiona 95 (21) in those scores in the same model. We then chose the Weekly standard drinks among consumers 7 (8) best predictive models, using a backward stepwise re- gression method that included competitive risk analysis. Current smoker 45 (10) Significance levels to enter and drop model variables BMI (kg/m2) 31 (7) were adopted as 5% and 10%, respectively. Area under Obesity 286 (64) the receiver operating characteristic curves (AUROCs) Overweight 134 (30) were calculated using Youden index to choose the best Arterial hypertension 329 (73) cut-­off point for discrimination. All confidence intervals Diabetes mellitus 338 (76) (CIs) and significance tests were two sided with a level of Dyslipidemia 250 (56) significance of 0.05. We used STATA/IC v.15.1 software Previous stroke 11 (3) to perform all statistical analyses. Ischemic heart disease 39 (9) Bilirubin (mg/dl) 0.8 (0.6) Albumin (g/L) 41 (6) R ESULTS INR 1.1 (0.2) Platelets (109/L) Characteristics of the study population 126 (85) Glycated hemoglobin (%) 6.5 (2) We identified 548 patients with NAFLD cirrhosis be- HDL-­cholesterol (mg/dl) 46 (19) tween 2009 and 2018. Of those, 99 were excluded: 84 Triglycerides 125 (85) due to the presence of previous hepatic decompensa- ALT (U/L) 38 (31) tion or because the diagnosis was made at the time of Alkaline phosphatase (U/L) 107 (71) the first decompensation and 15 because of the pres- Creatinine (mg/dl) 0.8 (0.3) ence of HCC. Of the 449 patients included, 165 (37%) Child- ­Pugh, A5/A6/B7/B8 333/58/17/4 patients were diagnosed based on histologic criteria, MELD 8 (2) 263 (58%) patients were diagnosed because they had NAFLD fibrosis score 1.3 (1.8) signs of cirrhosis and endoscopic or ultrasonographic FIB-­4 3.5 (2.7) signs of portal hypertension, and 22 (5%) patients were Liver stiffnessb 22 (17) diagnosed based on an LSM above 18 kPa. Baseline CAPc 296 (98) characteristics of patients at the time of diagno- d Endoscopic signs of portal hypertension 192 (63) sis of cirrhosis (inclusion date) are shown in Table 1. Gastroesophageal varices 95 (31) Cardiometabolic risk factors were highly prevalent. Portal hypertensive gastropathy 97 (31) Almost 10% of the cohort had previous ischemic heart disease. As expected for a cohort of patients with com- Ultrasonographic signs of portal hypertension 293 (68) pensated cirrhosis, liver function tests were preserved HVPG (mm Hg)e 13 (7) and most patients (87%) were classified as Child-­Pugh Liver biopsy available 175 (39) A. Patients had a median Model for End-­stage Liver Steatosis grade 2–­3f 60 (48) Disease (MELD) score of 8 (IQR, 2). Two thirds of the Ballooning grade 2 25 (20) cohort had either endoscopic or ultrasonographic signs Lobular inflammation grade 2 24 (20) of portal hypertension (63% and 68%, respectively). Abbreviations: ALT, alanine aminotransferase; AST, aspartate aminotransferase; BMI, body mass index; CAP, controlled attenuation parameter; FIB-­4, fibrosis-­4; HDL, high-­density lipoprotein; HVPG, hepatic Frequency and risk factors of AD venous pressure gradient; INR, international normalized ratio; MELD, Model for End-­Stage Liver Disease; NAFLD, nonalcoholic fatty liver disease. a Consumption inferior to 21 standard drinks per week in men and 14 in Median follow-­ up time of the cohort was 39 months women. Higher consumption was an exclusion criterion as described in (IQR, 22–­66). During follow-­up, 124 patients (28%) de- Materials and Methods. b Available in 208 patients. veloped a first AD. Of the 124 patients who developed c Available in 127 patients. AD, 68 (55%) developed more than one AD either si- d Screening for varices with gastroscopy within 1 year of admission in a total multaneously with the first event or during follow-­up. of 306 patients. Ascites was the most frequent AD (21% of the overall e Available in 86 patients. patients) followed by HE (15%), PHB (9%), and SBP f Individual data about steatosis, ballooning, and lobular inflammation grades were available in 126, 123, and 122 patients, respectively. (3%) (Table 2). 3216 |    HIGH FREQUENCY OF ACUTE DECOMPENSATION AND CANCER IN PATIENTS T A B L E 2 Acute decompensations of cirrhosis, deaths, and liver transplantation frequencies of the patients included specificity of 0.78 and 0.53, respectively. Comparison of cumulative incidences of developing AD categorized Outcome n = 449 (%) by the aforementioned predictors is shown in Figure 2. Patients with any hepatic decompensation 124 (28) We performed subgroup analysis of the prognostic Total acute decompensationsa 211 impact of the presence of signs of portal hypertension by endoscopy (i.e., gastroesophageal varices and/or Ascites 94 portal hypertensive gastropathy). We found a strong Hepatic encephalopathy 61 association between the presence of endoscopic signs Portal hypertension-­related bleeding 41 of portal hypertension and the probability of decom- Spontaneous bacterial peritonitis 15 pensation after adjusting for MELD, serum albumin, Acute- ­on- ­chronic liver failure 26 (6) and presence of ischemic heart disease (Figure S2; Grade 1/2/3 13/7/6 Table S2). Of the 449 included patients, 86 patients had Deaths 68 (15) an assessment of HVPG and 62 (72%) patients had CSPH. As expected, 27 (43%) patients among those Cause of death with CSPH developed AD during follow-­up compared Liver failure related 26 (38) to none of those without CSPH (p < 0.001). Hepatocellular carcinoma 13 (19) Two-­hundred and eight patients had LSM avail- Extrahepatic cancer 9 (13) able, and we found that higher values of liver stiffness Cardiovascular related 10 (15) were independently associated with the development Others b 10 (15) of AD after adjusting for other covariates (Table S3). Interestingly, among the patients with controlled at- Liver transplantation 20 (5) tenuation parameter (CAP) assessment (n = 127), we a Multiple decompensations occurred in some patients. Here, we report the found a negative association between CAP values total number of decompensating events (not patients) during follow-­up. b Other causes of death were acute infections not related to acute and the probability of AD at follow-­up after adjusting decompensation of cirrhosis, decompensation of chronic lung disease, and for other covariates. Consistently, the presence of unknown. histologic steatosis stage 2 or 3 at univariate analysis (Table 1) was associated with a lower probability of de- The incidence rate of presenting a first AD event veloping AD. was 0.08 events per person-­ year of follow-­ up (95% CI, 0.07–­0.10). Patients with Child A5 had an AD in- cidence rate of 0.06 events per person-­year (95% CI, Frequency and risk factors of HCC and 0.05–­0.08) compared to the rest of the cohort (Child A6 other types of cancer and Child B) with 0.16 events per person-­year (95% CI, 0.12–­0.21). The incidence rate ratio was 2.65 (95% CI, Fifty-­one patients (12%) developed HCC during follow- 1.81–­3.84). Considering extrahepatic deaths and liver ­up (Table S1; Figure S1). Most patients presented with transplantation as competing events, the cumulative early stages of HCC, mainly Barcelona Clinic Liver incidence of presenting an AD at 2 years of follow-­up Cancer stages 0 and A. The only variable indepen- was 15% (95% CI, 12–­18). The cumulative incidence of dently associated with an increased probability of HCC AD is shown in Figure 1, and the univariate description development was serum albumin after adjusting for of the population categorized according to the develop- MELD and age at diagnosis of cirrhosis. Besides HCC, ment of AD is shown in Table 3. Sex differences in AD 27 (6%) patients presented extrahepatic cancers during and HCC development are shown in Figure S1. follow-­up, colorectal cancer being the most frequent. We found that fibrosis-­ 4 (FIB-­4) score and liver function as assessed by MELD score and albumin were independently associated with the development Frequency and risk factors of ACLF of AD. Interestingly, the presence of previous ischemic development heart disease was also independently associated with AD development. On multivariable modeling (Table 4), During the study follow-­up, 26 patients (6%) developed the best predictive model of AD development included ACLF: 13 patients with grade 1, seven grade 2, and six serum albumin, MELD score, FIB-­4, and the presence grade 3. The incidence rate of ACLF development was of ischemic heart disease. Ischemic heart disease re- low, namely 0.015 events per person-­ year (95% CI, mained an independent predictor even after adjustment 0.010–­0.022). All patients but one presented AD be- by the presence of diabetes mellitus (which was not as- fore ACLF. The majority (70%) of patients who devel- sociated with the development of AD). Albumin alone oped ACLF died. Variables independently associated had the best discriminative capacity, with an AUROC with the development of ACLF were serum albumin and of 0.70. The best cutoff to discriminate patients with MELD score. Results of the multivariate modeling are AD was 40 g/L (95% CI, 38.2–­41.4), with sensitivity and shown in Table S4. HEPATOLOGY COMMUNICATIONS     | 3217 F I G U R E 1 Cumulative incidence of presenting HCC, AD alone, or a combination of HCC or AD. AD, acute decompensation; HCC, hepatocellular carcinoma. Survival analysis patients (28%) developed AD during the follow-­up. Our data contrast with previous studies where the incidence During a median follow-­up time of 39 months (IQR, 22–­ of AD reported was lower. Sanyal et al.[9] reported a 66), we observed 68 deaths (15%), 20 (5%) patients proportion of 19% of liver-­related events (LREs) in pa- received liver transplants, and five were lost to follow- tients with compensated cirrhosis during 30.9 months ­up. The detailed causes of mortality in our cohort are of follow-­up (range, 0.1–­45.5); Angulo et al.,[8] in the shown in Table 2. The majority of deaths were liver re- Prognostic Relevance of Liver Histology in NAFLD lated (57%): 26 (38%) due to complications of cirrhosis (PREHLIN) study, detected 24% of LREs in 12.6 years and 13 (19%) due to HCC. The second most frequent with a sample enriched with patients at low and inter- cause of mortality was cardiovascular events (n = 10; mediate stages of fibrosis; and Bhala et al.[7] detected 15% of deaths) followed by extrahepatic cancer in nine 19% of LREs in patients with advanced fibrosis during patients (13%). Survival analysis is shown in Figure S3. a median follow-­ up of 86 months. All these studies Death incidence rate was 0.04 (95% CI, 0.03–­0.05) were performed in highly selected populations exclud- events per person-­year. Variables independently as- ing patients with advanced liver dysfunction. Indeed, sociated with mortality were age at diagnosis, serum Calzadilla-­Bertot and colleagues,[22] in a recent predic- albumin, and MELD score. Detailed results of the sub- tive study including patients with compensated cirrho- hazards modeling are shown in Table S5. sis due to NAFLD, found 27% of events in 5.6 years (range, 2.4–­14.1). Moreover, analyzing a derivation co- hort to validate their model using less stringent exclu- D I SCUSS I O N sion criteria in liver function, they found 54% of events during a follow-­up time similar to that of the current This study shows the natural history of NAFLD cirrhosis study.[22] In a recent prospective study of the full spec- in a real-­life setting with the largest cohort of patients trum of NAFLD with a median follow-­up time of 4 years, with NAFLD with compensated cirrhosis analyzed to the authors found 37 hepatic decompensating events, date. Our study has three main findings: (1) we found 17 events among patients with F4 stage, and 17 among a significant frequency of AD and cancer development; F3.[10] This latter study reported only three events (2) predictors of AD were mainly related to liver func- among patients with lower fibrosis stages, underlining tion; and (3) we described the frequencies of develop- the importance of fibrosis stage as a predictive factor. ment and risk factors of ACLF in this population. However, the relatively low number of patients with Data from nine centers provided insight into the in- cirrhosis (n = 167) and the low number of total events cidence of AD of cirrhosis during a median follow-­up hampered the development of a predictive analysis of time of 39 months (IQR, 44). Nearly one third of our hepatic decompensation in this important study. 3218 |    HIGH FREQUENCY OF ACUTE DECOMPENSATION AND CANCER IN PATIENTS TA B L E 3 Baseline characteristics of patients who developed acute decompensation of cirrhosis during follow-­up compared to patients who did not Acute decompensation No decompensation Variables n = 124 (%) n = 325 (%) p-­valuea Age, years 67 (14) 64 (13) 0.101 Sex, female 55 (44) 160 (49) 0.291 b Current alcohol consumption 23 (19) 72 (22) 0.558 Weekly standard drinks among consumers 7 (5) 7 (9) 0.627 Current smoker 10 (8) 35 (11) 0.643 BMI (kg/m2) 31 (6) 32 (8) 0.494 Obesity 78 (63) 208 (64) 0.963 Arterial hypertension 91 (73) 238 (74) 0.434 Diabetes mellitus 97 (78) 241 (75) 0.321 Dyslipidemia 63 (50) 187 (58) 0.287 Ischemic heart disease 17 (14) 22 (7) 0.008* Bilirubin (mg/dl) 0.9 (0.7) 0.7 (0.5) <0.001* Albumin (g/L) 38 (7) 42 (6) <0.001* INR 1.2 (0.2) 1.1 (0.2) 0.005* Platelets (109/L) 107 (56) 135 (84) 0.022* Glycated hemoglobin (%) 6.4 (1.6) 6.5 (2) 0.517 HDL-­cholesterol (mg/dl) 44 (19) 46 (18) 0.539 Triglycerides (mg/dl) 115 (64) 129 (93) 0.135 AST (U/L) 47 (27) 40 (25) <0.001* ALT (U/L) 36 (30) 38 (32) 0.043* Creatinine (mg/dl) 0.8 (0.4) 0.8 (0.3) 0.915 Child-­Pugh Class, A5/A6/B7/B8 74/25/10/4 259/33/7/0 <0.001* MELD 9 (4) 8 (2) <0.001* NAFLD fibrosis score 1.8 (1.5) 1.1 (1.8) <0.001* FIB-­4 4.7 (3.3) 3.0 (2.5) <0.001* Liver stiffnessc 24 (31) 21 (12) 0.003* d CAP 255 (128) 309 (88) <0.001* e Endoscopic signs of portal hypertension 73 (81) 119 (55) <0.001* Gastroesophageal varices 38 (42) 59 (27) Portal hypertensive gastropathy 35 (39) 60 (28) f Ultrasonographic signs of portal hypertension 78 (77) 170 (67) 0.163 HVPG (mm Hg)g 15 (8) 11 (8) <0.001* Histologic diagnosis 53 (35) 134 (41) 0.153 Steatosis grade ≥2 12 (34) 51 (53) 0.014*,h Ballooning grade 2 5 (14) 21 (22) 0.207 Lobular inflammation grade 2 (17) 21 (22) 0.317 Abbreviations: ALT, alanine aminotransferase; AST, aspartate aminotransferase; BMI, body mass index; CAP, controlled attenuation parameter; FIB-­4, fibrosis-­4; HDL, high-­density lipoprotein; HVPG, hepatic venous pressure gradient; INR, international normalized ratio; MELD, Model for End-­Stage Liver Disease; NAFLD, nonalcoholic fatty liver disease. a p-­values calculated in univariate Fine and Gray's regression analysis. b Consumption inferior to 21 standard drinks in men and 14 in women. c Available in 208 patients. d Available in 127 patients. e Available in a total of 306 patients. f Available in 353 patients. g Available in 86 patients. h Grades 2–­3 compared to combined steatosis grade 1 and no steatosis. *Significant at p < 0.05 in univariate analysis. HEPATOLOGY COMMUNICATIONS     | 3219 TA B L E 4 Univariate and multivariate competing risk analysis for acute decompensation development Univariate analysis Multivariate analysis Variable sHR (95% CI) p-­value Variables sHR (95% CI) p-­value Age, years 1.02 (1.00–­1.04) 0.101 Model 1 Current alcohol 0.88 (0.56–­1.36) 0.558 Ischemic heart disease 2.45 (1.35– ­4.45) 0.001* consumptiona Weekly standard drinks 0.99 (0.93–­1.05) 0.627 Bilirubin (mg/dL) 1.15 (0.94–­1.41) 0.183 among consumers BMI (kg/m2) 0.99 (0.96–­1.02) 0.494 Albumin 0.90 (0.87– ­0.93) <0.001* Obesity 1.01 (0.75–­1.35) 0.963 INR 1.51 (0.95–­2.41) 0.078 9 Arterial hypertension 1.18 (0.78–­1.78) 0.434 Platelets (10 /L) 0.99 (0.99–­1.00) 0.278 Diabetes mellitus 1.25 (0.81–­1.92) 0.321 AST 1.02 (1.01–­1.03) 0.003* Dyslipidemia 0.83 (0.58–­1.17) 0.287 ALT 0.99 (0.98–­1.01) 0.092 Ischemic heart disease 2.12 (1.23–­3.65) 0.007* Bilirubin (mg/dL) 1.48 (1.22–­1.81) <0.001* Model 2 Albumin (g/L) 0.89 (0.87– ­0.92) <0.001* Ischemic heart disease 2.08 (1.15–­3.77) 0.015* INR 1.65 (1.16–­2.35) 0.005* Albumin 0.90 (0.87– ­0.93) <0.001* 9 * Platelets (10 /L) 0.996 (0.99–­1.00) 0.022 MELD 1.07 (1.01–­1.13) 0.017* Glycated hemoglobin (%) 0.940 (0.78–­1.13) 0.517 FIB- ­4 score 1.05 (1.01–­1.08) 0.010* HDL-­cholesterol (mg/dL) 0.995 (0.98–­1.01) 0.539 Triglycerides (mg/dL) 0.997 (0.99–­1.01) 0.135 Model 3 * ALT (U/L) 1.01 (1.00–­1.01) <0.001 Ischemic heart disease 2.17 (1.23–­3.82) 0.007* * AST (U/L) 1.003 (1.00–­1.01) 0.043 NAFLD fibrosis score 1.31 (1.13–­1.51) <0.001* Creatinine (mg/dL) 1.03 (0.59–­1.82) 0.915 Child- ­Pugh score 2.02 (1.56–­2.63) <0.001* * Child- ­Pugh score 2.16 (1.79–­2.67) <0.001 MELD 1.10 (1.05–­1.15) <0.001* NAFLD fibrosis score 1.44 (1.26–­1.65) <0.001* FIB-­4 1.08 (1.05–­1.11) <0.001* Abbreviations: ALT, alanine aminotransferase; AST, aspartate aminotransferase; BMI, body mass index; CAP, controlled attenuation parameter; CI, confidence interval; FIB-­4, fibrosis-­4; HDL, high-­density lipoprotein; HVPG, hepatic venous pressure gradient; INR, international normalized ratio; MELD, Model for End-­ Stage Liver Disease; NAFLD, nonalcoholic fatty liver disease; sHR, subhazard ratio. a Consumption inferior to 21 standard drinks in men and 14 in women. *Significant at p < 0.05 in univariate analysis. In our cohort, liver function as assessed mainly by are needed to identify factors associated with worse albumin levels and MELD score was the best predic- hepatic outcomes. tor of AD, as has been reported for other etiologies of Information about the development of ACLF in this cirrhosis.[6] In our real-­life cohort, each drop of 1 g/L population, which occurred in 6% of the cohort, is also in serum albumin increased the risk of decompen- described for the first time. Studies addressing the in- sation by 10%. Interestingly, we found that the pres- cidence of ACLF are mainly focused on patients with ence of ischemic heart disease almost doubled the alcohol-­related cirrhosis. We hypothesized that the bur- risk of presenting AD. However, individual parameters den of cardiometabolic disease of patients with NAFLD of cardiometabolic risk, such as the presence of dia- with cirrhosis could increase the risk of organ failure betes mellitus, dyslipidemia, or arterial hypertension, and thus ACLF development. However, we could not were not found to be predictors of decompensation. find any metabolic factors associated with the develop- This could be related to the high prevalence of these ment of ACLF that were predicted only by liver function features in our cohort. A history of previous cardio- tests. More studies are needed to address the inci- vascular events may reflect more established cardio- dence of ACLF among patients with decompensated vascular damage in those patients that could therefore cirrhosis due to NAFLD. impact hepatic outcomes. More studies taking into ac- Interestingly, we also found a high incidence of neo- count the dynamic changes in diabetes control and plastic events in this cohort of patients. Taken together, other metabolic parameters in patients with NAFLD HCC or extrahepatic cancers were detected in 18% of 3220 |    HIGH FREQUENCY OF ACUTE DECOMPENSATION AND CANCER IN PATIENTS F I G U R E 2 Cumulative incidence of presenting AD according to the presence of low serum albumin, low MELD score, low FIB-­4, and the presence of ischemic heart disease. AD, acute decompensation; Alb, albumin; FIB-­4, fibrosis-­4; MELD, Model for End-­Stage Liver Disease. the cohort, revealing a dramatic susceptibility to severe limitations of this technique, we performed a sensitivity health issues in a relatively short follow-­up time in this analysis excluding those patients from the analysis, but population. we observed no differences in the results. Moreover, The impact of alcohol use on the natural history data on LSM and HVPG measurement were not avail- of cirrhosis is a very interesting research topic in able for the majority of patients, and potential con- the NAFLD field. In our analysis, moderate alcohol founding factors associated with the outcome, such as consumption below the limits of the alcohol risk con- treatment with nonselective beta blockers, glucagon-­ sumption definition to define patients with ALD [15,23] like peptide-­1 agonists, vitamin E, or statins, were not was not associated with a worse outcome. More stud- recorded. The probability of missing an AD event in our ies are needed to address this important issue and retrospective cohort is low because of the low rate of establish the importance of any amount of alcohol patients lost to follow-­up. consumption for decompensation risk in patients with In conclusion, patients with compensated cirrhosis NAFLD. due to NAFLD present a high risk of hepatic AD and We are aware that our study has some limitations cancer development, indicating that this population is mainly related to its retrospective design and the relative at high risk of severe health issues in a relatively short short follow-­up period. Our cohort is overrepresented follow-­up period. A simple serum parameter, such as by patients with compensated advanced cirrhosis as an albumin level below 40 g/L, confers a higher risk of two thirds of our patients presented signs of portal hy- presenting AD compared to patients with values above pertension. Whether our results can be extrapolated to this cut-­off point. The presence of previous ischemic patients with earlier stages of compensated cirrhosis is heart disease as well as FIB-­4 and MELD scores also not known. On the other hand, 22 patients in our study predict the development of hepatic decompensation. were diagnosed based only on LSM criteria. Given the The incidence of ACLF in this population is no different HEPATOLOGY COMMUNICATIONS     | 3221 compared to published data from other etiologies and analysis for the Global Burden of Disease Study 2017. Lancet. 2018;392:1736–­88. Erratum in: Lancet. 2018;392:2170. is predicted by MELD score and albumin levels. 2. Younossi ZM, Stepanova M, Ong J, Trimble G, AlQahtani S, Younossi I, et al. Nonalcoholic steatohepatitis is the most rap- AUTHOR CONTRIBUTIONS idly increasing indication for liver transplantation in the United Conceptualization: Octavi Bassegoda, Isabel Graupera, States. Clin Gastroenterol Hepatol. 2021;19:580–­9.e5. Salvador Augustin, Pere Ginès. 3. Vilar-­Gomez E, Calzadilla-­ Bertot L, Wai-­ Sun Wong V, Castellanos M, Aller-­de la Fuente R, Metwally M, et al. Fibrosis Data acquisition: Octavi Bassegoda, Jesús Rivera-­ severity as a determinant of cause-­specific mortality in patients Esteban, Isabel Serra, Rosa Morillas, Teresa Broquetas, with advanced nonalcoholic fatty liver disease: a multi-­national Mercedes Vergara, Adrià Rodriguez, Carles Aracil, Silvia cohort study. Gastroenterology. 2018;155:443–­57.e17. Virolés, Jose A Carrión, Albert Pardo, Sergio Rodríguez-­ 4. Ekstedt M, Hagström H, Nasr P, Fredrikson M, Stål P, Kechagias Tajes, Miquel Serra-­ Burriel, Juan M Pericàs, Pere S, et al. Fibrosis stage is the strongest predictor for disease-­ specific mortality in NAFLD after up to 33 years of follow-­up. Ginès, Salvador Augustin, Isabel Graupera. Statistical Hepatology. 2015;61:1547–­5 4. analysis: Octavi Bassegoda, Isabel Graupera. Funding 5. Taylor RS, Taylor RJ, Bayliss S, Hagström H, Nasr P, acquisition: Isabel Graupera, Salvador Augustin, Pere Schattenberg JM, et al. Association between fibrosis stage Ginès. Methodology: Octavi Bassegoda, Jesús Rivera-­ and outcomes of patients with nonalcoholic fatty liver disease: Esteban, Albert Pardo, Salvador Augustin, Isabel a systematic review and meta-­ analysis. Gastroenterology. 2020;158:1611–­25.e12. Graupera. Analysis and interpretation of data: Octavi 6. D'Amico G, Garcia-­ Tsao G, Pagliaro L. Natural history and Bassegoda, Jesús Rivera-­ Esteban, Albert Pardo, prognostic indicators of survival in cirrhosis: a systematic re- Salvador Augustin, Isabel Graupera, Pere Ginès. view of 118 studies. J Hepatol. 2006;44:217–­31. Writing original draft: Octavi Bassegoda, Albert Pardo, 7. Bhala N, Angulo P, van der Poorten D, Lee E, Hui JM, Saracco Salvador Augustin, Isabel Graupera. Review and edit- G, et al. The natural history of nonalcoholic fatty liver disease with advanced fibrosis or cirrhosis: an international collabora- ing: Octavi Bassegoda, Jesús Rivera-­Esteban, Isabel tive study. Hepatology. 2011;54:1208–­16. Serra, Rosa Morillas, Teresa Broquetas, Mercedes 8. Angulo P, Kleiner DE, Dam-­Larsen S, Adams LA, Bjornsson Vergara, Adrià Rodriguez, Carles Aracil, Silvia Virolés, ES, Charatcharoenwitthaya P, et al. Liver fibrosis, but no other Jose A Carrión, Albert Pardo, Sergio Rodríguez-­Tajes, histologic features, is associated with long-­term outcomes of Miquel Serra-­ Burriel, Juan M Pericàs, Pere Ginès, patients with nonalcoholic fatty liver disease. Gastroenterology. 2015;149:389–­97.e10. Salvador Augustin, Isabel Graupera. Study supervision: 9. Sanyal AJ, Harrison SA, Ratziu V, Abdelmalek MF, Diehl AM, Pere Ginès, Isabel Graupera. All authors reviewed and Caldwell S, et al. The natural history of advanced fibrosis due approved the final manuscript. to nonalcoholic steatohepatitis: data from the simtuzumab tri- als. Hepatology. 2019;70:1913–­27. F U N D I N G I N F O R M AT I O N 10. Sanyal AJ, Van Natta ML, Clark J, Neuschwander-­Tetri BA, Diehl A, Dasarathy S, et al. Prospective study of outcomes Hospital Clínic de Barcelona, “Contractes Clínic de in adults with nonalcoholic fatty liver disease. N Engl J Med. Recerca Emili Letang-­Josep Font 2020”; Grant Number: 2021;385:1559– ­69. Catalonian Society of Digestology, Grant Number: 11. Moreau R, Jalan R, Gines P, Pavesi M, Angeli P, Cordoba J, Instituto Carlos III Rio Hortega (ISCIII); Fellowship et al. Acute-­on-­chronic liver failure is a distinct syndrome that Number: CM17/00015; Fundación de Investigación develops in patients with acute decompensation of cirrhosis. Gastroenterology. 2013;144:1426–­37.e1–­9. Sanitaria, ISCIII-­Subdirección General de Evaluación, 12. Sundaram V, Jalan R, Shah P, Singal AK, Patel AA, Wu T, and the European Regional Development Fund; Project et al. Acute on chronic liver failure from nonalcoholic fatty Numbers: PI18/00862, PI120/00579. liver disease: a growing and aging cohort with rising mortality. Hepatology. 2021;73:1932– ­4 4. CONFLICT OF INTEREST 13. European Association for the Study of the Liver. EASL Clinical Practice Guidelines for the management of patients with de- The authors declare no conflict of interest with respect compensated cirrhosis. J Hepatol. 2018;69:406–­60. Erratum to this investigation. in: J Hepatol. 2018;69:1207. 14. Solà E, Pose E, Campion D, Piano S, Roux O, Simon-­Talero ORCID M, et al. Endpoints and design of clinical trials in patients with Octavi Bassegoda https://orcid. decompensated cirrhosis: position paper of the LiverHope Consortium. J Hepatol. 2021;74:200–­19. org/0000-0002-6036-6026 15. Chalasani N, Younossi Z, Lavine JE, Charlton M, Cusi K, Rinella Jesús Rivera- ­Esteban https://orcid. M, et al. The diagnosis and management of nonalcoholic fatty org/0000-0003-4357-8817 liver disease: practice guidance from the American Association Mercedes Vergara https://orcid. for the Study of Liver Diseases. Hepatology. 2018;67:328–­57. org/0000-0002-6971-8657 16. Eddowes PJ, Sasso M, Allison M, Tsochatzis E, Anstee QM, Sheridan D, et al. Accuracy of FibroScan controlled attenua- Pere Ginès https://orcid.org/0000-0003-4657-4504 tion parameter and liver stiffness measurement in assessing steatosis and fibrosis in patients with nonalcoholic fatty liver REFERENCES disease. Gastroenterology. 2019;156:1717–­30. 1. GBD 2017 Causes of Death Collaborators. Global, regional, 17. Berzigotti A, Seijo S, Reverter E, Bosch J. Assessing portal hy- and national age-­ sex-­specific mortality for 282 causes of pertension in liver diseases. Expert Rev Gastroenterol Hepatol. death in 195 countries and territories, 1980–­2017: a systematic 2013;7:141–­55. 3222 |    HIGH FREQUENCY OF ACUTE DECOMPENSATION AND CANCER IN PATIENTS 18. De Franchis R, Baveno VI Faculty. Expanding consensus management of non-­alcoholic fatty liver disease. J Hepatol. in portal hypertension: report of the Baveno VI Consensus 2016;64:1388– ­402. Workshop: stratifying risk and individualizing care for portal hy- pertension. J Hepatol. 2015;63:743–­52. 19. Bosch J, Abraldes JG, Berzigotti A, García-­Pagan JC. The clin- S U P P O R T I N G I N F O R M AT I O N ical use of HVPG measurements in chronic liver disease. Nat Additional supporting information can be found online in Rev Gastroenterol Hepatol. 2009;6:573–­82. 20. Asrani SK, Larson JJ, Yawn B, Therneau TM, Kim WR. the Supporting Information section at the end of this article. Underestimation of liver-­related mortality in the United States. Gastroenterology. 2013;145:375– ­82.e1–­2. 21. European Association for the Study of the Liver. EASL clini- cal practice guidelines: management of hepatocellular carci- How to cite this article: Bassegoda O, Rivera- noma. J Hepatol. 2018;69:182–­ 236. Erratum in: J Hepatol. Esteban J, Serra I, Morillas R, Broquetas T, 2019;70:817. 22. Calzadilla-­Bertot L, Vilar-­Gomez E, Wong VW, Romero-­Gomez Vergara M, High frequency of acute M, Aller-­de la Fuente R, et al. ABIDE: an accurate predictive decompensation and cancer in patients with model of liver decompensation in patients with non-­alcoholic compensated cirrhosis due to nonalcoholic fatty fatty liver-­related cirrhosis. Hepatology. 2021;73:2238–­50. liver disease: A retrospective cohort study. Hepatol 23. European Association for the Study of the Liver (EASL); Commun. 2022;6:3212–­3222. https://doi. European Association for the Study of Diabetes (EASD); European Association for the Study of Obesity (EASO). org/10.1002/hep4.2056 EASL–­ E ASD–­ E ASO Clinical Practice Guidelines for the