Abstract and Introduction
Abstract
Background & Aim: Non-alcoholic fatty liver disease (NAFLD), and especially fibrotic non-alcoholic steatohepatitis, is associated with high risks of liver-related events (LRE) and extrahepatic events (EHE). We evaluated the competitive risk occurrence of LRE and EHE in a large cohort of biopsy-proven NAFLD stratified according to baseline severity of fibrosis.
Methods: Two thousand one hundred thirty-five patients with biopsy-proven NAFLD were enrolled. Observed cumulative incidence functions (CIFs) were used to evaluate the risk of LRE and EHE; cause-specific Cox model and predicted CIFs were fitted to identify predictors of LRE and EHE. A replication cohort of NAFLD patients with liver fibrosis severity estimated by liver stiffness measurement by transient elastography was also enrolled.
Results: Observed CIFs indicated that the 60-month probabilities of LRE and EHE were 0.2% and 3% in F0-F1, 2% and 3.8% in F2 and 9.7% and 6.4% in F3-F4 patients, respectively. The cause-specific Cox model indicated that in F0-F1 and F2 patients, age > 50 years (HR 2.7) was the only predictor of LRE, while age > 50 years (HR 2.96), previous cardiovascular events (CVE, HR 2.07), and previous extra-hepatic cancer (HR 2.36) were independent risk factors for EHE. In F3-F4 patients, age > 55 years (HR 1.73), obesity (HR 1.52), PLT < 150 000/mmc (HR 3.66) and log(GGT) (HR 1.77) were associated with LRE, while age > 55 years (HR 1.74) and previous CVE (HR 2.51) were independent predictors of EHE. Predicted CIFs for HE and EHE in F0-F1, F2 and F3-F4 patients stratified the risk of events. The results were externally replicated.
Conclusion: The likelihood of EHE in NAFLD patients is relevant and increases according to the severity of liver fibrosis, while the risk of LRE is negligible in F0-F1, low but clinically relevant in F2 and high in F3-F4 patients.
Introduction
Non-alcoholic fatty liver disease (NAFLD) is the most emergent cause of liver disease, affecting one quarter of the world population, with prevalence predicted to increase to 33.5% within the next few years, in line with the global epidemics of obesity and type 2 diabetes (T2D).[1,2] Natural history studies of NAFLD patients show that the relative risk of liver-related complications has increased when compared to control populations.[3–9] The two most frequent complications are cardiovascular events (CVE) and extra-hepatic cancers (EHC). NAFLD is becoming the most prevalent cause of liver-related events (LRE) that include liver decompensation (LD) and hepatocellular carcinoma (HCC), leading to end-stage liver disease and liver transplantation.[10–13] In parallel to the upsurge of LRE, a recent American analysis also reported increased mortality due to EHC and CVE in patients with NAFLD.[14]
In this complicated scenario, multiple studies have shown that—as expected—the risk of developing LRE is higher in NAFLD patients with advanced liver fibrosis or cirrhosis.[15] Furthermore, in the same clinical setting, other studies have reported an increased risk of both CVE and EHC in patients with more advanced fibrosis.[16] Consistent with these results, two recent meta-analyses have shown that the risk of liver-related and overall mortality progressively increases according to the baseline severity of liver fibrosis.[17,18]
Despite the clinically relevant and heterogeneous burden of hepatic and extrahepatic complications of NAFLD and the effect of fibrosis on their occurrence, no definitive evidence exists regarding the natural history of NAFLD when LRE and extra-hepatic events (EHE) are considered as competing risks. This is a crucial point because, in complex epidemiological and clinical contexts, competing risks modelling is a powerful statistical method to adjust for the impact of clinical decision making in the presence of multiple major clinical endpoints.[19] The rationale for using the competing risk approach for the study of the natural history of NAFLD patients lies in the fact that a competing risk is an event that either precludes or modifies the risk of occurrence of an event of interest (eg, LRE vs EHE, as in the present study).[19] Furthermore, from a methodological point of view, time-to-event analysis using either the naive Kaplan-Meier estimator or the PH Cox model to estimate the probability of occurrence of the event of interest in the presence of competing risks could yield biased estimates. Cumulative incidence function (CIF) and competing risks models such as the cause-specific Cox (CSC) and/or the Fine-Gray (FG) models should be applied because they account for the effects of competing risks.[20]
In the present study, we evaluated the probability of developing LRE and EHE, considered as competing risks, in a large multicentre cohort of patients with a histological diagnosis of NAFLD stratified by fibrosis stage. We replicated these results in a large cohort of NAFLD patients in whom liver fibrosis was assessed non-invasively.