Prevalence and clinical correlates of metabolic dysfunction-associated steatotic liver disease in heart failure

Setting and design

Between 2012 and 2017, symptomatic (Stage C) patients with HF from a multinational ASIAN-HF registry were prospectively recruited across 46 medical centres in 10 Asian countries¹⁷ (India, Singapore, Taiwan, Malaysia, Indonesia, Hong Kong, Korea, Thailand, Philippines and Japan). Site selection in ASIAN-HF was made according to the size of the country, geographical location of the site in the country, patient population served, HF patient volume and the presence of echocardiography expertise.¹⁸

Asian patients aged >18 years and diagnosed with HF with reduced ejection fraction (HFrEF, ejection fraction (EF) <40%) and preserved EF (HFpEF, EF ≥50%) were included. All patients had a validated clinical diagnosis of HF based on symptoms according to the Framingham Heart Study criteria, with HF decompensation within 6 months.¹⁹ Those with severe valvular heart disease as the primary cause of HF, life-threatening comorbidity with life-expectancy of <1 year, unable to give consent or actively participating in a clinical therapeutic trial, were excluded. Patients were then allocated into two groups according to the presence of MASLD. Individuals with missing data necessary for determining MASLD status were not included.

Demographics, clinical characteristics, functional status, past medical history, quality of life (QoL) and blood chemistry were recorded at recruitment. Transthoracic echocardiography data were also obtained at baseline. Coronary artery disease was defined as angiographic evidence of significant coronary obstruction, history of myocardial infarction or previous revascularisation. Diabetes mellitus, atrial fibrillation, previous stroke were identified based on previous diagnosis. CKD was defined as estimated glomerular filtration rate (eGFR) <60 mL/min per 1.73 m² based on the Modification of Diet in Renal Disease study equation.¹⁸

Study definition

Serum biomarkers have been recommended as a feasible tool in large scale screening for hepatic steatosis where cost and availability of imaging service are prohibitive.²⁰ MASLD was defined as evidence of hepatic steatosis accompanied by at least one of five cardiometabolic risk factors: (1) high body mass index (BMI) ≥23 kg/m² in Asians or BMI ≥25 kg/m² in whites; (2) high serum fasting glucose ≥5.6 mmol/L, or glycated haemoglobin ≥5.7%, or previously diagnosed type 2 diabetes mellitus; (3) elevated blood pressure ≥130/85 mm Hg, or on specific antihypertensive drug treatment; (4) high plasma triglycerides ≥1.70 mmol/L, or on lipid-lowering treatment; and (5) low plasma high-density lipoprotein cholesterol ≤1.0 mmol/L in males, or ≤1.3 mmol/L in females, or on lipid-lowering treatment.²¹ Considering the availability of obesity measures, other than BMI, we used waist-to-height ratio (WHtR) >0.5²² as a marker of central adiposity replacing BMI in an alternative MASLD definition (MASLD_WHtR).

The Hepatic Steatosis Index (HSI) is a non-invasive serum-based test that is widely used to screen for hepatic steatosis, and has been externally validated in cohorts with ultrasound-determined hepatic steatosis.²³ The HSI uses serum transaminases (alanine aminotransferase (ALT) and aspartate aminotransferase (AST)) and is calculated by: HSI=8×(ALT[IU/L]/AST[IU/L])+BMI[kg/m²] + 2 × female gender+2 × presence of diabetes mellitus. Hepatic steatosis was defined as HSI >36.²³ QoL was measured using the self-administered Kansas City Cardiomyopathy Questionnaire²⁴ (KCCQ). KCCQ domain scores range from 0 to 100, with higher scores representing better QoL.

Outcomes

The primary outcome was 1-year all-cause mortality. The secondary outcome was a composite of time to all-cause mortality or hospitalisation for HF. The cause of death was adjudicated by an independent committee in accordance with the US Food and Drug Administration’s standardised event definitions.²⁵

External validation analysis

The results of this study were validated in two HF populations representing predominantly Chinese and White European ethnicity respectively: CDARS, an independent territory-wide population cohort of 28 233 patients with HF by Hong Kong Hospital Authority; and BIOSTAT-CHF, a cohort of new-onset or worsening HF across 11 European countries.

In CDARS, adult patients aged 18 years or above with HF, determined by the International Classification of Diseases (ICD-9: 398.91, 402.0, 402.01, 402.1, 402.11, 402.9, 402.91, 404, 404.01, 404.03, 404.1, 404.11, 404.13, 404.9, 404.91, 404.93, 425, 428; and ICD-10: I11.0, I13.0, I13.2, I50.0, I50.1, I50.9, I42, I43), as the primary cause for hospitalisation between 1996 and 2020, were included. In BIOSTAT-CHF, patients with new-onset or worsening signs and/or symptoms of HF from 11 European countries were included.²⁶ Patients with missing data for MASLD categorisation were excluded. The patient selection process for each cohort, including the number screened, excluded and analysed, is detailed in online supplemental figure 1.

Statistical analysis

Normally and non-normally distributed continuous variables were reported as means (SD) and median (IQR), respectively. Categorical variables were presented as numbers (percentage). Differences between MASLD and non-MASLD groups were evaluated with χ² test and independent t-test for categorical and continuous variables respectively. Kaplan Meier curves were constructed for 1-year all-cause mortality, comparing patients with HF with and without MASLD. For cause-specific mortality endpoints (cardiovascular death and HF-related death), competing risk regression was performed using the Fine and Gray method, with other causes of death categorised as competing events. The results were reported as subdistribution HRs (SHR). For HF hospitalisation, death from any cause was categorised as a competing event, and SHR were similarly estimated using the Fine and Gray method. Cox proportional hazards models were used to calculate HRs of 1-year all-cause mortality, and the composite outcome of all-cause mortality or hospitalisation for HF, adjusting for age, sex, enrolment type, geographical region, HF subtype, systolic blood pressure, heart rate, BMI, previous coronary artery disease, diabetes mellitus, atrial fibrillation, CKD and ACE inhibitors (ACEI)/angiotensin receptor blockers (ARB), beta-blocker or mineralocorticoid receptor antagonist use. Covariates for multivariable adjustment were selected a priori based on established clinical and prognostic predictors in HF described in prior studies.^{18 22 27} Age-adjusted Cox regression was also performed for the individual components of the MASLD definition to examine the association with 1-year mortality, as well as the composite outcome of 1-year HF hospitalisation or death. For outcomes with violated proportionality assumptions, Cox regression was performed, stratified by time periods of ≤30 days and >30 days to 1 year.

To determine if the body anthropometric criterion affects the relationship between MASLD and outcomes, further analyses were performed using WHtR instead of BMI, as a marker of central obesity in the MASLD definition.²² The association of MASLD_WHtR and HF outcomes was examined. A two-tailed p-value of ≤0.05 was considered statistically significant. Statistical analysis was performed using Stata (V.16.1 StataCorp).

Patient and public involvement

Patients and members of the public were not involved in the design, conduct, reporting or dissemination plans of this research. This study was a secondary analysis of de-identified data from established HF registries and cohorts.

Baseline characteristics

A total of 2698 patients with HF and available MASLD status were included in the study analysis (mean age 64±13 years, 26.7% women). The mean follow-up duration was 2.0±1.1 years. In the HF cohort, 34.4% (n=927) had evidence of MASLD. Patients with HF and MASLD were younger (59.3±13.1 vs 65.6±12.8 years, p<0.001), slightly more likely to be women (31.1% vs 24.4%, p<0.001) and have HFpEF (25.2% vs 17.1%, p<0.001), compared with those with HF without MASLD (table 1). Among Asian ethnicities, Malays had the highest prevalence of MASLD (51.6%), followed by Chinese (44.2%) and Indian (42.3%).

Patients with HF and MASLD had higher left ventricular EF (LVEF) (35.1±16.4% vs 32.2±14.9%, respectively, p<0.001) and more concentric remodelling/concentric hypertrophy compared with individuals with HF without MASLD. In addition, patients with MASLD were more likely to have hypertension, diabetes and be prescribed with ACEI/ARB and statins compared with those without MASLD (table 1).

Clinical outcomes

In the ASIAN-HF cohort, all-cause mortality rates were lower in MASLD compared with non-MASLD at 1-year (7.1% vs 11.7%, p<0.001). In patients with HF, MASLD was independently associated with lower mortality risk (adjusted HR (aHR) 0.58, 95% CI 0.39 to 0.86, p=0.007) (table 2). The relationship between MASLD and mortality was not significantly modified by sex (p_interaction=0.085), HF subtype (p_interaction=0.119), diabetes mellitus (p_interaction=0.058), eGFR (p_interaction=0.082) or obesity (p_interaction=0.646). Kaplan-Meier curve (figure 1) demonstrated favourable survival in MASLD versus non-MASLD over 1 year (p=0.002). Notably, the survival advantage of MASLD persisted over the full study duration, with MASLD remaining independently associated with lower all-cause mortality (aHR 0.68, 95% CI 0.52 to 0.88, p=0.003) and the composite of death or HF hospitalisation (aHR 0.78, 95% CI 0.65 to 0.94, p=0.008) (online supplemental table 1; online supplemental figure 2). Stratified analysis based on time period revealed that the favourable prognosis of MASLD was only observed >30 days to 1 year from the onset of HF (aHR 0.50, 95% CI 0.32 to 0.77, p=0.002), but not within the initial 30 days (aHR 1.32, 95% CI 0.50 to 3.49, p=0.581). In post hoc sensitivity analyses with additional adjustments for smoking status and alcohol use, MASLD remained associated with lower risks of 1-year mortality (HR 0.57; 95% CI 0.38 to 0.85; p=0.005) and 1-year composite outcomes (HR 0.74; 95% CI 0.58 to 0.94; p=0.015). The direction and magnitude of these associations were consistent with the primary analysis, supporting the robustness of our findings.

Cox regression analysis demonstrated that MASLD was associated with reduced HF hospitalisation or death after adjustment (aHR 0.75, 95% CI 0.59 to 0.96, p=0.020) in patients with HF. The association between MASLD and outcomes was significantly modified by the presence of diabetes mellitus (p_interaction=0.036), where MASLD was independently associated with reduced death or HF hospitalisation risk in patients without diabetes mellitus (aHR 0.60, 95% CI 0.37 to 0.96, p=0.034) but not in those with diabetes mellitus (aHR 0.80, 95% CI 0.61 to 1.06, p=0.239). There was no effect modification by sex (p_interaction=0.057), HF subtype (p_interaction=0.490), obesity (p_interaction=0.989) or eGFR (p_interaction=0.082). The overall clinical score of KCCQ was lower in MASLD compared with non-MASLD (67.4±24.7 vs 70.7±23.3, respectively, p=0.004) (table 3). Cox regression analysis of the individual components of the MASLD definition demonstrated that the paradoxical short-term survival was largely driven by hepatic steatosis (aHR 0.63, 95% CI 0.47 to 0.84, p=0.002) and, to a non-significant extent, obesity (aHR 0.88, 95% CI 0.68 to 1.13, p=0.305). However, these protective trends were not observed with 1-year HF hospitalisation or death (online supplemental table 2).

Stratified analyses demonstrated that the favourable survival association with MASLD was directionally consistent across BMI categories and HF severity strata, including EF subtype (HFrEF and HFpEF) and New York Heart Association functional class, although statistical significance was not uniformly reached across all subgroups, likely reflecting limited sample sizes and event rates (online supplemental table 3).

Validation cohorts

Of the 28 132 Asian patients with HF from CDARS, 39.1% (n=10 998) satisfied criteria for MASLD. Individuals with MASLD were younger (p<0.001) than those without MASLD (online supplemental table 2). The risk of all-cause mortality at 1 year was lower in MASLD than non-MASLD (HR 0.74, 95% CI 0.73 to 0.76, p<0.001). Kaplan-Meier analysis revealed favourable survival in the MASLD group compared with the non-MASLD group over time (p<0.001) (figure 2). After adjusting for confounders, MASLD was an independent predictor of lower all-cause mortality (aHR 0.82, 95% CI 0.80 to 0.84, p<0.001). The association between MASLD and outcomes was significantly modified by the presence of diabetes mellitus (p_interaction<0.001), but not by sex (p_interaction=0.837), obesity (p_interaction=0.412) or eGFR (p_interaction=0.255). In CDARS, MASLD was independently associated with reduced all-cause death risk in patients with (aHR 0.75, 95% CI 0.71 to 0.79, p<0.001) and without (aHR 0.80, 95% CI 0.76 to 0.84, p<0.001) diabetes mellitus. Moreover, in the CDARS cohort we observed that this paradoxical trend in MASLD extended to cardiovascular death (aHR 0.77, 95% CI 0.74 to 0.81, p<0.001), HF-related death (aHR 0.78, 95% CI 0.72 to 0.86, p<0.001), as well as HF-related hospitalisation (aHR 0.94, 95% CI 0.91 to 0.96, p<0.001) (online supplemental table 5).

In the European BIOSTAT-CHF cohort (n=1593), 60% satisfied criteria for MASLD. Individuals with HF and MASLD were younger (73.2±10.2 vs 75.5±10.8 years, p<0.001) and more likely to be women (36.3% vs 31.4%, p=0.043) compared with those without MASLD (online supplemental table 6). Patients with HF and MASLD had similar LVEF (41.5±18.2% vs 41.9±21.0%, p=0.816). Kaplan-Meier analysis demonstrated favourable survival in the MASLD group compared with the non-MASLD group (p=0.0063) (figure 3). The lower risk of 1-year mortality with MASLD (HR 0.78, 95% CI 0.65 to 0.93, p=0.005) became non-significant after adjusting for confounders (aHR 1.08, 95% CI 0.67 to 1.73, p=0.750). There was no interaction of MASLD with HF subtype, sex, diabetes mellitus, obesity or eGFR, for both all-cause death as well as the composite outcome of death and/or HF hospitalisation (online supplemental table 7).

Analysis using WHtR

WHtR was measured in 782 patients in the ASIAN-HF cohort, among whom 37.6% (n=294) had MASLD_WHtR. When using the MASLD_WHtR criterion, 0.4% (n=3) of the ASIAN-HF cohort was classified differently compared with the standard MASLD definition. Individuals with MASLD_WHtR were younger (p<0.001), more likely to be women (p=0.006) and more likely to have the HFpEF phenotype (p=0.002) compared with the non-MASLD_WHtR group (table 4). Cox regression revealed a borderline significant paradoxical relationship of MASLD_WHtR with all-cause mortality in univariable analysis (HR 0.64, 95% CI 0.40 to 0.98, p=0.042) which became attenuated and did not reach statistical significance after multivariable adjustment (aHR 0.89, 95% CI 0.48 to 1.63, p=0.700) (table 5). The relationship between MASLD_WHtR and mortality was not significantly modified by sex (p_interaction=0.580), HF subtype (p_interaction=0.142), diabetes mellitus (p_interaction=0.430), eGFR (p_interaction=0.533) or obesity (p_interaction=0.675). Similar trends were seen in the relationship between MASLD_WHtR and the composite outcome of mortality or HF hospitalisation (aHR 0.92, 95% CI 0.62 to 1.50, p=0.855). Of note, Cox regression using BMI and the standard MASLD definition in the same subgroup of 782 individuals similarly demonstrated the paradoxical relationship in MASLD and 1-year mortality in univariable analysis (aHR 0.63, 95% CI 0.40 to 0.98, p=0.039), but not in multivariable analysis (aHR 0.89, 95% CI 0.48 to 1.63, p=0.695).

Similar results were observed in CDARS and BIOSTAT-CHF. Among 28 132 individuals with available WHtR in CDARS, 31.9% (n=8961) had MASLD_WHtR (online supplemental table 8), with the MASLD_WHtR criterion reclassifying 2.9% (n=839) of patients compared with the standard MASLD definition. Among 1593 individuals with available WHtR in BIOSTAT-CHF, 57.4% (n=915) had MASLD_WHtR (online supplemental table 9), with the MASLD_WHtR criterion reclassifying 2.6% (n=42) of patients compared with the standard MASLD definition. Similar survival associations were observed in each cohort after replacing the standard MASLD definition with the alternative MASLD_WHtR criterion (online supplemental tables 10 and 11).

Strength and limitations

The strength of this study lies in the representativeness of this multinational cohort, across a vast geographical area in Asia, with the additional independent external validation cohorts, that would enhance the representation of various demographic subpopulation groups, spanning across diverse geographical, ethnic and socio-economic demographics. However, there are several limitations. First, the validated HSI was used as a surrogate marker of hepatic steatosis as routine liver imaging was not feasible across all participating centres within the multinational real-world registry. While biopsy-proven fatty liver remains the gold standard for diagnosis, it is seldom practicable for all patients presenting with HF in routine clinical settings. Accordingly, we cannot completely rule out the possibility that the elevated levels of transaminases in some individuals might be contributed by the sequelae of venous congestion. Notably, the HSI has not been specifically validated against biopsy or imaging-determined hepatic steatosis in a HF cohort, which should be considered when interpreting the study findings. Moreover, individuals with more severe HF may be less likely to meet the HSI-based MASLD definition owing to cardiac cachexia, reduced hepatic fat accumulation and attenuated levels of transaminases, potentially enriching the non-MASLD group with higher risk individuals, and may contribute to the observed survival paradox. Therefore, these findings should be validated in future large prospective HF cohorts with systematic concomitant evaluation of MASLD, preferably by liver elastography, to better evaluate the diagnostic and prognostic impact of the MASLD criterion in the HF cohort.⁴⁹ Second, the observational nature of the multinational studies precludes conclusions on causality, and residual confounding may still be present. Third, HF was identified using ICD-9 and ICD-10 discharge codes in the CDARS validation cohort, which are subject to the inherent limitations of administrative coding such as potential misclassification resulting from variability in coding practices, undercoding of clinically diagnosed HF and the inability to ascertain LVEF classification or diagnostic certainty. Nevertheless, the consistency of the MASLD-outcome association across the prospectively recruited ASIAN-HF cohort, in which HF was determined by validated clinical criteria coupled with echocardiographic confirmation and the CDARS cohort provides reassurance regarding the robustness of the findings to differences in case ascertainment methodology.