Liver-related and Extrahepatic Events in Patients With Non-alcoholic Fatty Liver Disease

A Retrospective Competing Risks Analysis

Grazia Pennisi; Marco Enea; Manuel Romero-Gomez; Mauro Viganò; Elisabetta Bugianesi; Vincent W.-S. Wong; Anna Ludovica Fracanzani; Giada Sebastiani; Jerome Boursier; Annalisa Berzigotti; Mohammed Eslam; Javier Ampuero; Amine Benmassaoud; Claudia La Mantia; Yuly P. Mendoza; Jacob George; Antonio Craxì; Calogero Camma'; Victor de Ledinghen; Salvatore Petta

Disclosures

Aliment Pharmacol Ther. 2022;55(5):604-615. 

In This Article

Results

Baseline Patient Characteristics

Baseline characteristics of the 2135 patients with a histological diagnosis of NAFLD stratified according to the severity of liver fibrosis (F0-F1 = 1136; F2 = 362 and F3-F4 = 637) are shown in Table 1. As expected, mean age was significantly higher in patients with F3-F4 compared to F0-F1 and F2 fibrosis. The prevalence of metabolic comorbidities (obesity, T2D and arterial hypertension) progressively increased according to the severity of liver fibrosis. Previous CVE was more frequent in patients with F3-F4 than in patients with F0-F1 or F2 fibrosis, while the prevalence of previous EHC was not significantly different between the groups.

Similar demographic, metabolic and clinical features were found in the replication cohort (n = 2790, 1365 at low, 286 at intermediate and 1139 at high risk of advanced liver fibrosis by LSM) (Table S1).

Incident LRE and EHE

F0-F1, F2 and F3-F4 patients had median follow-up horizons of 97.0, 79.6 and 65.6 months, respectively. F0-F1 patients experienced 8 LRE and 66 EHE (2.1% and 5.8%, respectively), F2 patients developed 7 LRE and 36 EHE (4.4% and 9.9%, respectively) and F3-F4 patients suffered 81 LRE and 64 EHE (18.8% and 14.6%). In the entire cohort, 30 LRE (2 in F0-F1, 4 in F2 and 24 in F3-F4 patients) and 44 EHE-related (19 in F0-F1, 6 in F2 and 19 in F3-F4 patients) deaths were recorded.

Table 2 shows the crude probability of LRE or EHE as first event, followed by EHE or LRE, respectively, as second event, stratified for fibrosis severity. Notably, patients with F0-F1 and F2 fibrosis had a clinically relevant probability of EHE as first event (5.8% and 9.9%, respectively), despite a very low or low probability of first LRE (0.6% and 1.6%, respectively). Also, in patients with F0-F1 and F2 fibrosis with a first EHE, we confirmed the low probability of LRE as second event (1.5% and 2.7%, respectively). On the other hand, patients with F3-F4 fibrosis had a comparably high probability of both LRE and EHE as first event (12% and 9.4%, respectively), and a similarly high risk of EHE and LRE as second event (5.2% and 6.6%, respectively). Similar data, summarised in Table S2, were found in the replication cohort.

Competing Risk Analysis of LRE and EHE

To estimate the competing risk occurrence of LRE and EHE, 10-year CIFs were calculated. When considered as competitors, the cumulative incidence of not only LRE but also EHE progressively increased over time according to the severity of liver fibrosis.

In F0-F1 patients, the probabilities of LRE and EHE were 0.1% and 0.4% at 12 months, 0.2% and 1.9% at 36 months, 0.2% and 3% at 60 months, 0.4% and 6.6% at 120 months, respectively (Figure 1A) (Table 3). In F2 patients, the probabilities of LRE and EHE were 0.6% and 2.3% at 12 months, 1.6% and 2.6% at 36 months, 2% and 3.8% at 60 months, 4.4% and 11.6% at 120 months, respectively (Figure 1B) (Table 3). Finally, in F3-F4 patients, the probabilities of LRE and EHE were 2.1% and 1.1% at 12 months, 5.7% and 4.3% at 36 months, 9.7% and 6.4% at 60 months, 21% and 10.8% at 120 months, respectively (Figure 1C) (Table 3). Rates of LRE and EHE in the subgroup of patients without histories of previous CVE and/or extrahepatic cancers stratified according to the severity of liver fibrosis are shown in Table 3.

Figure 1.

Observed cumulative incidence functions of hepatic and extrahepatic events in patients with F0-F1 fibrosis or at low risk of developing advanced fibrosis (A), F2 fibrosis or at intermediate risk of developing advanced fibrosis (B), and F3-F4 fibrosis or at high risk of developing advanced fibrosis (C)

Finally, in the subgroup of patients with F3-F4 fibrosis, the risk of developing LRE with respect to EHE was lower in F3 while higher in F4 patients. Specifically, in F3 patients, the probabilities of HCC, LD, and EHE were 0.2%, 0.2% and 1.7% at 12 months; 2%, 0.2% and 5.1% at 36 months; 3.3%, 1.9% and 8% at 60 months and 5.4%, 4.8% and 12.8% at 120 months, respectively (Figure S1). Conversely, in F4 patients, the probabilities of HCC, LD and EHE were 2.2%, 1.5% and 0.4% at 12 months; 4.8%, 5.2% and 2.8% at 36 months; 6.4%, 8.3% and 3.8% at 60 months and 10.7%, 20.9% and 8.1% at 120 months, respectively (Figure S1).

Similar cumulative incidences of LRE and EHE stratified for the risk of fibrosis severity were observed in the replication cohort (Figure 1) (Table 3).

Predictors of LRE and EHE

The observed and previously described link between the severity of liver fibrosis and increased competing cumulative incidence of not only LRE but also EHE was underscored by using the cause-specific proportional CSC, that, after adjusting for age, confirmed F2 (HR 5.08, SE 0.42, P < 0.001), F3 (HR 12.91, SE 0.50, P < 0.001) and F4 (HR 39.23, SE 0.39, P < 0.001) fibrosis as independent risk factors for LRE, and F2 (HR 1.6, SE 0.14, P < 0.001) and F3 (HR 1.87, SE 0.19, P < 0.001) fibrosis as independent predictors of EHE. Similar results were obtained by using the FG model (data not shown).

We conducted further analyses by stratifying the population according to liver fibrosis severity. Results of univariate CSC models are reported in Table S3. Using multivariate CSC (Table 4) for LRE and EHE in F0-F1 and F2 patients, age > 50 years (HR 2.7, beta 0.99, P = 0.001) was the only independent predictor of developing LRE; while age > 50 years (HR 2.96, beta 1.08, P < 0.001); previous CVE (HR 2.07, beta 0.73, P = 0.03) and previous EHC (HR 2.36, beta 0.86, P = 0.003) were independent risk factors for incident EHE. When excluding patients with histories of previous CVE and/or extrahepatic cancer, age > 50 years was the only predictor of developing both LRE (HR 2.09, SE 0.32, P = 0.02) and EHE (HR 3.13, SE 0.31, P < 0.001).

The CSC model (Table 4) disclosed that in F3-F4 patients, age > 55 years (HR 1.73, beta 0.55, P = 0.03); obesity (HR 1.52, beta 0.42, P = 0.03); PLT < 150 000/mmc (HR 3.66, beta 1.30, P < 0.001) and log(GGT) (HR 1.77, beta 0.57, P < 0.001) were significantly associated with LRE, while, age > 55 years (HR 1.74, beta 0.55, P = 0.006) and previous CVE (HR 2.51, beta 0.92, P < 0.001) were independent predictors of EHE. In the subgroup of patients without histories of previous CVE and/or EHC, age > 55 years (HR 1.91, SE 0.31, P = 0.03); obesity (HR 1.69, SE 0.20, P = 0.01); PLT < 150 000/mmc (HR 3.86, SE 1.18, P < 0.001) and log(GGT) (HR 1.85, SE 0.75, P < 0.001) were confirmed independent predictors of LRE, while age > 55 years (HR 1.59, SE 0.20, P = 0.02) and serum triglycerides (HR 1.92, SE 0.26, P = 0.01) were independently associated with a higher risk of EHE.

Predicted CIFs for LRE and EHE in F0-F1, F2 and F3-F4 patients were obtained from the models, and classes of patients at high or low risk of events according to presence or absence of all independent risk factors were considered. Specifically, in F0-F1 patients the probability of EHE ranged from 0.18%, 0.96%, 1.55% and 3.56% at 12, 36, 60 and 120 months, respectively, in the low- risk class; to 2.61%, 13.06%, 20.24% and 40.89% at 12, 36, 60 and 120 months, respectively, in the high-risk class (Figure 2A). In F2 patients, the probability of EHE ranged from 1.03%, 1.18%, 1.75% and 5.69% at 12, 36, 60 and 120 months, respectively, in the low-risk class, to 13.96%, 15.80%, 22.61% and 56.83% at 12, 36, 60 and 120 months, respectively, in the high-risk class (Figure 2B). LRE risk also increased from the low-risk to the high-risk class in both F0-F1 and F2 patients but remained negligible in F0-F1 (Figure 2A) and low in F2 (Figure 2B). In patients with F3-F4 fibrosis, the probability of LRE ranged from 0.39%, 1.10%, 1.94% and 4.69% at 12, 36, 60 and 120 months, respectively, in the low-risk class; to 7.10%, 18.31%, 29.01% and 51.97% at 12, 36, 60 and 120 months, respectively, in the high-risk class (Figure 2C). On the other hand, the probability of EHE ranged from 0.69%, 2.57%, 4.09% and 7.24% at 12, 36, 60 and 120 months, respectively, in the low-risk class, to 2.90%, 9.78%, 14.33% and 21.04% at 12, 36, 60 and 120 months, respectively, in the high-risk class (Figure 2C). Similar results were obtained when replacing the CSC model with the FG model (Table 4 and Figure S2). Notably, probabilities of events predicted by classical Cox models were slightly (for F0-F2) to moderately (for F3-F4 patients) biased, compared to results obtained by CSC and FG models (Figures S3-S5).

Figure 2.

Predicted cumulative incidence functions of hepatic and extrahepatic events by cause-specific Cox hazard model in patients with F0-F1 fibrosis or at low risk of developing advanced fibrosis (A), F2 fibrosis or at intermediate risk of developing advanced fibrosis (B), and F3-F4 fibrosis or at high risk of developing advanced fibrosis (C)

Finally, both the CSC and FG models yielded congruous results in the replication cohort, as summarised in Table S4. Specifically, by using the CSC (Table S4) for LRE and EHE in patients at low/intermediate risk of advanced fibrosis, age > 50 years (HR 1.3, beta 0.26, P = 0.01) was the only independent predictor of LRE, while age > 50 years (HR 2.13, beta 0.76, P < 0.001), and previous CVE (HR 2.48, beta 0.91, P < 0.001) were independent risk factors for EHE. The CSC model (Table S4) disclosed that for F3-F4 patients, age > 55 years (HR 1.95, beta 0.67, P < 0.001), PLT < 150 000/mmc (HR 5.02, beta 1.61, P < 0.001) and log(GGT) (HR 1.31, beta 0.27, P = 0.003) were significantly associated with LRE, while, age > 55 years (HR 2.06, beta 0.72, P = 0.005) and previous CVE (HR 2.37, beta 0.86, P = 0.004) were independent predictors of EHE. The obtained predicted CIFs for LRE and EHE in patients at low, intermediate and high risk of advanced fibrosis are reported in Figure 2. Similar results were obtained when replacing the CSC model with the FG model (Table S4 and Figure S2).

Subgroup analyses were performed in patients without histories of previous CVE and/or EHC. In this sub-cohort, in patients at low-intermedium risk of advanced fibrosis by LSM, age > 50 years (HR 1.49, SE 0.52, P < 0.001) was independent predictor of LRE, and age > 50 years (HR 1.47, SE 0.32, P < 0.001) and higher log(GGT) serum levels (HR 1.21, SE 0.15, P < 0.001) were independent predictors of EHE. In patients at high risk of advanced fibrosis by LSM, age > 55 years (HR 1.80, SE 0.27, P = 0.001) and PLT < 150 000 mmc (HR 5.27, SE 0.21, P < 0.001) were independent predictors of LRE, and age > 55 years (HR 1.89, SE 0.33, P = 0.01) was an independent predictor of EHE.

processing....