You are here

  • Home
  • Archive
  • Volume 72, Issue 2
  • Non-alcoholic fatty liver disease and risk of new-onset heart failure: an updated meta-analysis of about 11 million individuals

Article Text

Article menu
Hepatology
Original research
Non-alcoholic fatty liver disease and risk of new-onset heart failure: an updated meta-analysis of about 11 million individuals
  1. Alessandro Mantovani1,
  2. Graziana Petracca1,
  3. Alessandro Csermely1,
  4. Giorgia Beatrice1,
  5. Stefano Bonapace2,
  6. Andrea Rossi3,
  7. Herbert Tilg4,
  8. Christopher D Byrne5,
  9. Giovanni Targher1
  1. 1 Endocrinology and Metabolism, Department of Medicine, University of Verona, Verona, Italy
  2. 2 Division of Cardiology, IRCSS Sacro Cuore - Don Calabria Hospital, Negrar (VR), Italy
  3. 3 Section of Cardiology, Department of Medicine, University of Verona, Verona, Italy
  4. 4 Department of Internal Medicine I, Gastroenterology, Hepatology, Endocrinology and Metabolism, Medizinische Universitat Innsbruck, Innsbruck, Austria
  5. 5 Faculty of Medicine, University of Southampton, Southampton, UK
  1. Correspondence to Professor Giovanni Targher, Endocrinology and Metabolism, University of Verona Department of Medicine, Verona 37129, Italy; giovanni.targher{at}univr.it

Abstract

Objective Recent studies reported an association between non-alcoholic fatty liver disease (NAFLD) and increased risk of new-onset heart failure (HF). However, the magnitude of the risk and whether this risk changes with severity of liver disease remains uncertain. We performed a meta-analysis of observational studies to quantify the magnitude of the association between NAFLD and risk of new-onset HF.

Design We systematically searched Scopus, Web of Science and PubMed from database inception to March 2022 to identify eligible observational studies, in which NAFLD was diagnosed by serum biomarkers/scores, International Classification of Diseases (ICD) codes, imaging techniques or liver histology. The primary outcome was new-onset HF, as assessed mainly by ICD codes. Data from selected studies were extracted, and meta-analysis was performed using random-effects models to obtain summary hazard ratios (HRs) with 95% CIs.

Results We identified 11 longitudinal cohort studies with aggregate data on 11 242 231 middle-aged individuals from different countries and 97 716 cases of incident HF over a median of 10 years. NAFLD was associated with a moderately higher risk of new-onset HF (pooled random-effects hazard ratio 1.50, 95% CI 1.34 to 1.67, p<0.0001; I 2=94.8%). This risk was independent of age, sex, ethnicity, adiposity measures, diabetes, hypertension and other common cardiovascular risk factors. Sensitivity analyses did not change these results. The funnel plot did not show any significant publication bias.

Conclusion NAFLD is associated with a 1.5-fold higher long-term risk of new-onset HF, regardless of the presence of diabetes, hypertension and other common cardiovascular risk factors. However, the observational design of the studies does not allow for proving causality.

  • NONALCOHOLIC STEATOHEPATITIS
  • CARDIOVASCULAR COMPLICATIONS
  • CARDIOVASCULAR DISEASE

Data availability statement

All data relevant to the study are included in the article or uploaded as supplementary information.

https://doi.org/10.1136/gutjnl-2022-327672

Statistics from Altmetric.com

    Request Permissions

    If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

    WHAT IS ALREADY KNOWN ON THIS TOPIC

    • Recent observational studies have reported an association between non-alcoholic fatty liver disease (NAFLD) and higher risk of new-onset heart failure (HF). However, the magnitude of the risk and whether this risk changes with severity of liver disease remain uncertain.

    WHAT THIS STUDY ADDS

    • This updated meta-analysis involving over 11 million middle-aged individuals from different countries shows that the risk of developing incident HF is 1.5-fold higher in individuals with NAFLD during a median follow-up of 10 years.

    • The risk of incident HF is independent of age, sex, adiposity measures, diabetes, hypertension and other common cardiovascular risk factors.

    • This risk also appears to increase further with greater severity of NAFLD, especially with higher fibrosis stage.

    HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE AND/OR POLICY

    • Healthcare professionals should be aware that risk of new-onset HF is moderately higher in patients with NAFLD. Because of the link between the two conditions, more careful surveillance of these patients will be needed.

    Introduction

    Non-alcoholic fatty liver disease (NAFLD) has become one of the most common causes of chronic liver diseases worldwide (affecting up to ~30% of the world’s adults), and its global prevalence is expected to rise sharply over the next decade.1–3 The clinical burden of NAFLD is not only restricted to its liver-related complications (non-alcoholic steatohepatitis (NASH), cirrhosis or hepatocellular carcinoma). Strong evidence now indicates that NAFLD is part of a multisystem disease that adversely affects several extrahepatic organs, including the heart and vasculature.4 5 In fact, NAFLD not only promotes accelerated coronary atherosclerosis but also confers an increased risk of myocardial abnormalities (cardiac remodelling and hypertrophy) and certain cardiac arrhythmias (mostly permanent atrial fibrillation), which may precede and/or promote the development of new-onset heart failure (HF).6 7 HF is increasingly becoming a major public and economic health burden worldwide. Its global prevalence is increasing due to the ageing of the population, and HF is strongly associated with a higher risk of hospitalisations and all-cause mortality.8

    To our knowledge, there are only two meta-analyses that have recently assessed the association between NAFLD and the risk of developing incident HF.9 10 These two meta-analyses showed that NAFLD is significantly associated with a greater risk of incident HF. However, these two meta-analyses included a relatively small number of studies (with a maximum of five longitudinal studies) with a relatively modest sample size.9 10 In addition, these two meta-analyses did not address the question of whether the strength of any association between NAFLD and risk of new-onset HF was affected by severity of NAFLD. Notably, as will be discussed further, in recent months, a number of new cohort studies have examined the association between NAFLD and the risk of new-onset HF.

    We have therefore undertaken an updated systematic review and meta-analysis of observational cohort studies to quantify the long-term risk of new-onset HF in individuals with NAFLD. We also aimed to explore whether the severity of NAFLD was associated with a higher risk of new-onset HF and to test the effect of recognised cardiometabolic risk factors as moderator variables on the association between NAFLD and risk of new-onset HF.

    Methods

    Search strategy and selection criteria

    We systematically searched three large electronic databases (Scopus, Web of Science and PubMed) from database inception to 21 March 2022, using predefined key words, to identify observational cohort studies examining the risk of new-onset HF events among adult (age ≥18 years) individuals with and without NAFLD. Search free-text terms were “non-alcoholic fatty liver disease” (“NAFLD” OR “fatty liver” OR “non-alcoholic steatohepatitis” OR “NASH”) AND risk of “incident heart failure” OR “heart failure”. Searches were restricted to human studies. Studies in languages other than English were also excluded. We also reviewed references from original papers and review articles to identify further eligible studies not covered by our original database searches. This systematic review was performed according to the updated Preferred Reporting Items for Systematic Reviews and Meta-Analysis statement.11 Because the included studies were observational in design, we followed the reporting proposed by the Meta-analysis of Observational Studies in Epidemiology for the meta-analysis of these studies.12

    Studies were included in the meta-analysis if they met the following criteria: (1) observational cohort studies assessing the association between NAFLD and risk of new-onset HF; (2) studies reporting HRs with 95% CIs for the outcome of interest; and (3) studies diagnosing NAFLD with serum liver enzyme levels, serum biomarkers/scores (eg, fatty liver index (FLI) or other algorithmically derived scores to diagnose the presence of hepatic steatosis), International Classification of Diseases (ICD) codes, imaging techniques or liver histology, in keeping with the diagnostic criteria for NAFLD. Participants included in the meta-analysis were both adult men and women without any restriction in terms of race or ethnicity.

    Criteria for exclusion of the studies from the meta-analysis were as follows: (1) congress abstracts, case reports, reviews, practice guidelines, and case–control or cross-sectional studies; (2) cohort studies which did not specifically report any HR and 95% CIs for the outcome of interest; (3) cohort studies that included only patients with NAFLD or NASH without a non-steatotic control group; and (4) cohort studies conducted in patients with type 2 diabetes, patients with established chronic or acute HF, or patients with established or suspected coronary artery disease.

    Data extraction and quality assessment

    Data from studies eligible for the aggregate data meta-analysis were extracted by three authors independently (AM, AC and GT). Disagreements at this level were resolved by consensus and a fourth author if needed (GP). For all studies, we extracted data on first author, publication year, study design, study country, population characteristics, methods used for NAFLD diagnosis, duration of follow-up, outcomes of interest, matching and confounding factors included in multivariable regression analyses. In case of multiple publications, we included the most up-to-date or comprehensive information.

    Each eligible study was assessed for quality by using the Newcastle-Ottawa Scale (NOS) by two independent reviewers (AM and GT), with disagreements resolved through consensus. The NOS uses a star system to evaluate a study in three domains: selection of participants (assigning a maximum of four stars), comparability of study groups (assigning a maximum of two stars) and ascertainment of outcomes of interest (assigning a maximum of three stars). Therefore, nine stars reflect the highest quality. We judged studies that received a score of 8 or 9 stars to be at low risk of bias, studies that scored 7 stars to be at medium risk, and those that scored ≤6 stars to be at high risk of bias. We recorded the reviewing authors’ judgements regarding the three NOS domains, in the risk of bias tool in the Review Manager software of the Cochrane collaboration.13

    Data synthesis and analysis

    The primary outcome of interest of the meta-analysis was the risk of developing new-onset HF among individuals with NAFLD compared with NAFLD-free controls. In eligible studies, the ascertainment of incident HF outcomes was based mainly on the International Classification of Diseases, 9th Revision (ICD-9) or the International Classification of Diseases, 10th Revision (ICD-10) codes or ascertained by revision of medical records. The HRs and 95% CIs were considered as the effect size for each eligible study. In the case of studies reporting several HRs with varying degrees of covariate adjustment, HRs that reflected the maximum extent of adjustment for potential confounding factors were extracted. The adjusted HRs of all eligible studies were then pooled, and an overall estimate of the effect size was calculated using a random-effects model, as this methodology considers any differences between studies, even if there is no statistically significant heterogeneity.13

    Visual inspection of the forest plot was used to assess statistical heterogeneity. This was also assessed with the I 2 statistics, which provides an estimate of the percentage of variability across eligible studies that is due to heterogeneity rather than chance alone. Heterogeneity was considered to be low if I 2 is <25%, moderate if I 2 is between 25% and 75%, and high if I 2 is >75%.14 The risk of publication bias was examined using the funnel plot and the Egger’s regression test.13

    To explore the possible sources of heterogeneity across the studies and to test the robustness of the observed associations, we performed subgroup analyses by study country, length of follow-up, modality of HF diagnosis, modality of NAFLD diagnosis, and even after excluding studies using serum gamma-glutamyl transferase (GGT) levels for diagnosing NAFLD. Univariable metaregression analyses were also performed to test the effect of specific moderator variables (ie, age, sex, ethnicity, body mass index, plasma total cholesterol levels, and percentage of hypertension, diabetes or prior acute myocardial infarction at baseline) on the effect size for the incidence of NAFLD-related HF events. Finally, we tested for possible excessive influence of individual studies using a meta-analysis influence test that eliminated each of the included studies one at a time. All statistical tests were two-sided, and a p value of <0.05 (two-tailed) was considered statistically significant. For analyses, we used Stata V.16.1 and its meta-analysis package and R software V.4.1.0 (R Foundation for Statistical Computing, Vienna, Austria) with meta and metafor packages. The protocol of this systematic review was registered in advance on Open Science Framework (number osf.io/yu8zq.

    Results

    Characteristics of eligible studies

    Figure 1 shows the results of the literature research and study selection. Based on the titles and abstracts of 490 selected citations (after excluding duplicates), we initially identified 17 potentially eligible studies from PubMed, Scopus or Web of Science databases prior to 21 March 2022.15–31 In online supplemental table 1, the syntax used and the records identified through database searches are presented. After examining the full text of these 17 potentially eligible studies, we excluded six studies, mainly due to unsatisfactory inclusion criteria or unsatisfactory outcome measures (as specified in online supplemental table 2).26–31 As a result of these exclusions, we identified 11 unique, observational cohort studies for inclusion in the meta-analysis.

    Supplemental material

    Figure 1
    Figure 1

    The PRISMA flow diagram for search and selection processes of the meta-analysis. From: Page MJ, McKenzie JE, Bossuyt PM, et al. The PRISMA 2020 statement: an updated guideline for reporting systematic reviews. BMJ 2021;372:n71. doi: 10.1136/bmj.n71. For more information, visit: http://www.prisma-statement.org/

    The main characteristics of the 11 eligible cohort studies are summarised in table 1. Overall, these studies had aggregate data on 11 242 231 middle-aged individuals (50.1% women, mean age 55 years, mean BMI 26.4 kg/m2). There were a total of 2 944 058 (26.2%) individuals with NAFLD at baseline and 97 716 incident cases of ICD-diagnosed HF over a median follow-up of 10.0 years (IQR 8–14 years). Most of these studies recruited participants either from general populations or large health examination check-up programmes, in which NAFLD was diagnosed by serum GGT levels, serum biomarkers/scores (FLI score), ICD codes, imaging techniques or liver histology. Four studies were carried out in Europe (Sweden, Finland and UK); four studies were carried out in the USA; and three studies were carried out in South Korea. Three studies used serum GGT levels for the diagnosis of NAFLD; four studies used the FLI score; two studies used ICD codes; one study used CT; and one study used liver biopsy. The assessment of incident HF events on follow-up was based on ICD-9 or ICD-10 codes (n=7 studies), Hospital International Classification of Diseases Adapted codes (n=1 study), or ascertained by revision of primary care records and/or hospitalisation records (n=3 studies).

    View this table:
    Table 1

    Eligible cohort studies examining the association between NAFLD and risk of developing incident HF (ordered by publication year, n=11 studies)

    As reported in online supplemental figure 1, two studies received at least eight stars on the NOS (ie, studies at relatively low risk of bias); six studies received seven stars (studies at medium risk of bias); and three studies received six stars (studies at high risk of bias), thus indicating an overall medium–low risk of bias.

    NAFLD and risk of new-onset HF

    The distribution of eligible studies by estimate of the association between NAFLD and the risk of developing new-onset HF is plotted in figure 2.

    Figure 2
    Figure 2

    Forest plot and pooled estimates of the effect of NAFLD on the risk of developing new-onset heart failure in 11 eligible cohort studies, stratified by methods used for the diagnosis of NAFLD. NAFLD, non-alcoholic fatty liver disease.

    The presence of NAFLD was associated with a moderately increased risk of incident HF (n=11 studies; pooled random-effects HR 1.50, 95% CI 1.34 to 1.67, p<0.0001; I 2=94.8%). Notably, since we have always used the fully adjusted HR estimates for each eligible study (as specified in table 1), this pooled random-effects HR was independent of age, sex, adiposity measures, diabetes and other common cardiovascular risk factors. As also shown in figure 2, the association between NAFLD and risk of new-onset HF was consistent even when the comparison was stratified by modality of NAFLD diagnosis (also reported in table 2), except for the single cohort study that used CT for diagnosing NAFLD in 719 older subjects. Additionally, in this study, the diagnosis of NAFLD was based on cardiac CT that examined only a single hepatic lobe (therefore being less accurate for complete structural assessment of the liver).

    View this table:
    Table 2

    Subgroup analyses: associations between NAFLD and risk of new-onset HF, stratified by study country, median length of follow-up, modality of HF diagnosis, modality of NAFLD diagnosis or after excluding cohort studies using serum GGT levels for diagnosing NAFLD

    Subgroup analyses and metaregressions

    To explore the possible sources of heterogeneity across the eligible studies, we undertook subgroup analyses (table 2). Notably, the association between NAFLD and risk of new-onset HF was consistent in all subgroups examined. In particular, the pooled random-effects HRs were essentially comparable after stratification by study country, follow-up length, modality of HF diagnosis or after excluding the three cohort studies using serum GGT levels for diagnosing NAFLD.

    As shown in online supplemental figure 2, the results of univariable metaregression analyses to examine the effect of potential moderator variables showed a significant inverse association between age (p=0.020) or the proportion of those with pre-existing type 2 diabetes (p=0.022) and the risk of NAFLD-related HF events. Conversely, metaregression analyses did not reveal any significant effect of sex, ethnicity, body mass index, plasma total cholesterol level or percentage of hypertension or prior acute myocardial infarction at baseline on the association between NAFLD and risk of new-onset HF.

    We also tested for the possibility of excessive influence of individual cohort studies using an influence test that eliminated each of the included studies one at a time. Notably, eliminating each of the studies in turn from our pooled primary analysis did not reveal any significant influence on the overall risk of new-onset HF (online supplemental figure 3).

    Online supplemental figure 4 shows that there was no significant asymmetry of the funnel plot, thus suggesting that publication bias was unlikely (p=0.527 by Egger’s regression test).

    NAFLD severity and risk of new-onset HF

    The risk for incident HF appeared to further increase with more advanced liver disease, especially with higher levels of liver fibrosis, as assessed by non-invasive fibrosis biomarkers or histology (n=2 studies; pooled random-effects HR 1.76, 95% CI 0.75 to 4.36; I 2=96.9%). However, we believe that the results for the association between the severity of NAFLD and incident HF can only be discussed as a systematic review, due to insufficient data suitable for combining the studies in a meta-analysis. As shown in table 1, there are only two cohort studies (from South Korea and Sweden) that examined the association between NAFLD severity and risk of new-onset HF. The study by Park et al 23 in Korea reported that among participants with NAFLD, advanced liver fibrosis (as non-invasively assessed by BARD score ≥2) was associated with a higher risk for new-onset HF (adjusted HR 1.12, 95% CI 1.04 to 1.20). The study by Simon et al 25 in Sweden reported that rates of incident HF increased progressively with the severity of NAFLD histology, with the highest incidence rates observed with non-cirrhotic fibrosis (adjusted HR 2.04, 95% CI 1.66 to 2.51) and cirrhosis (adjusted HR 2.83, 95% CI 2.08 to 3.85), after adjusting for age, sex and common cardiometabolic risk factors.

    Discussion

    In this updated meta-analysis of 11 observational cohort studies (involving a total of about 11 million middle-aged individuals from different countries with ~98 000 cases of new-onset HF events over a median follow-up of 10 years), we found that NAFLD was significantly associated with a 1.5-fold higher risk of new-onset HF events (pooled random-effects HR 1.50, 95% CI 1.34 to 1.67, p<0.0001). Importantly, this risk was independent of age, sex, ethnicity, adiposity measures, pre-existing diabetes, hypertension and other common cardiovascular risk factors. Sensitivity analyses did not modify these results. In particular, the magnitude of this risk remained essentially unchanged even when the comparison was stratified by study country, follow-up duration, modality of HF diagnosis and modality of NAFLD diagnosis, or when we excluded from the analysis the three cohort studies using serum GGT levels for the diagnosis of NAFLD. Our metaregression analyses also reported a significant inverse association between the proportion of patients with pre-existing diabetes at baseline and the risk of developing NAFLD-related HF events. This finding could be due, at least in part, to the fact that the coexistence of type 2 diabetes at baseline, which is a strong risk factor for HF,32 could attenuate the effect of NAFLD itself as a risk factor for new-onset HF, or alternatively, subjects with diabetes are often treated with a range of cardiovascular risk-reduction medications that could attenuate the risk of CVD and subsequent development of HF in this patient group. A similar hypothesis can also be proposed to explain the observed inverse association for age; older age, which is a strong risk factor for HF, could attenuate the effect of NAFLD itself as a risk factor for new-onset HF, and older people are more likely to be treated with medications that may attenuate risk of CVD and development of HF. It should also be noted that since NAFLD is also associated with a greater risk of developing incident type 2 diabetes,33 it is possible that during follow-up, a larger number of patients will develop type 2 diabetes. Accordingly, the risk of new-onset HF should not only be simply tested in relation to the presence of diabetes at entry, but also as an effect of incident diabetes. Unfortunately, almost all of the eligible studies (except for one study17) did not have any data on occurrence of type 2 diabetes during the follow-up period (defined as a time-varying covariate). Finally, in this meta-analysis, we found that the risk of incident HF appeared also to increase further with greater severity of NAFLD, especially with higher fibrosis stage (as reported in the two eligible cohort studies using liver biopsy or non-invasive scores of advanced fibrosis).23 25 These latter observations are consistent with those of previous meta-analyses reporting a significant association between the presence and severity of NAFLD and the risk of developing adverse cardiovascular outcomes,34 atrial fibrillation,35 chronic kidney disease36 or other extrahepatic complications.33 37 These latter observations are also indirectly supported by longitudinal studies showing that increased fibrosis-4 index or other non-invasive liver fibrosis risk scores are associated with a higher risk of HF hospitalisation and adverse cardiovascular outcomes in cohorts of patients with established NAFLD/NASH.31 Similarly, these results are supported by data from cohorts of patients with established HF, especially in those with preserved ejection fraction (with heart failure preserved ejection fraction (HFpEF)).38 39 However, this question remains to be fully resolved, and further studies are required to prove whether the severity of liver disease in NAFLD further amplifies the increased risk of developing incident HF.

    To our knowledge, our meta-analysis assessing the association between NAFLD and the long-term risk of new-onset HF is the largest and most comprehensive assessment of this association to date. The results of our meta-analysis extend the observations of two recent smaller meta-analyses.9 10 In the first meta-analysis including four longitudinal studies (published until March 2021), Alon et al 9 reported that NAFLD (as detected by either FLI score, ICD codes or CT) was associated with a~60% increased risk of new-onset HF (pooled random-effects HR 1.62, 95% CI 1.43 to 1.84; I 2=97%). However, no studies using liver biopsy for diagnosing NAFLD were included in this meta-analysis. In addition, among the four eligible studies included in the meta-analysis, the authors also included a cohort study by VanWagner et al,28 who examined the longitudinal association of NAFLD with subclinical changes in myocardial structure and function but not with new-onset HF. In the second meta-analysis that incorporated five longitudinal studies involving about 1.4 million subjects (published until 12 December 2021), Salah et al 10 reported that NAFLD was associated with a higher risk of new-onset HF (pooled random-effects HR 1.60, 95% CI 1.24 to 2.05; I 2=97%). Four of these five cohort studies have been also included in our meta-analysis; we excluded a small cohort study by Vita et al 27 that included only patients who were suspected of having coronary artery disease. In contrast to our meta-analysis, in this meta-analysis the authors pooled the unadjusted HRs for each eligible study, so we cannot exclude that the association they observed between NAFLD and risk of new-onset HF could be partly dependent on coexisting cardiometabolic risk factors. That said, in the two aforementioned meta-analyses, the number of eligible studies was too small for performing subgroup analyses and metaregressions, or for formally testing the existence of any significant publication bias. Compared with data from these two smaller meta-analyses,9 10 we have more than doubled the total number of eligible studies (by including a total of 11 large cohort studies published until March 21, 2022); and we have also increased the sample size about 8–9 times (by including about 11.2 million people from different countries with aggregate data on ~98 000 cases of new-onset HF). In addition, we have also been able to test the effect of various established risk factors for HF as potential moderator variables, in the association between NAFLD and risk of new-onset HF.

    It is beyond the scope of this meta-analysis to discuss the putative underlying mechanisms involved in the development of NAFLD-related HF, and these have been discussed in a recent review of the subject.40 There are probably several pathophysiological mechanisms by which NAFLD increases the risk of cardiac (functional, structural and arrhythmic) complications that may promote the development of HF; NAFLD (especially advanced NAFLD with increasing levels of liver fibrosis) exacerbates systemic insulin resistance, promotes atherogenic dyslipidaemia and releases multiple thrombogenic molecules, proinflammatory factors and oxidative stress biomarkers.6 40 In addition, there is growing scientific interest in the link between these two conditions because newer antihyperglycaemic agents (such as sodium–glucose cotransporter-2 inhibitors and glucagon-like peptide-1 receptor agonists) exerting some benefit on hepatic fat content and histological resolution of NASH41 have also documented clinically meaningful benefits on cardiovascular outcomes, including risk of HF hospitalisation, regardless of the diabetes status.40 42 That said, further research is needed to better decipher the complex pathophysiological mechanisms by which NAFLD may contribute to increase risk of new-onset HF.

    The present meta-analysis has some important limitations, which are strictly inherent to the design of the eligible studies. First, the observational design of the eligible studies precludes us from assessing causality. Second, although most of these studies adjusted the results for age, sex and common cardiometabolic risk factors, the possibility of unmeasured and residual confounding cannot be ruled out. In particular, the large majority of these studies did not have any echocardiographic data for examining the association between NAFLD and different categories of left ventricular ejection fraction. As specified in table 1, the only eligible study22 that examined this association showed that NAFLD was associated with a higher risk of developing HFpEF (adjusted HR 1.24, 95% CI 1.14 to 1.34) versus HF with a reduced ejection fraction (adjusted HR 1.09, 95% CI 0.98 to 1.20). Third, most of the eligible studies have adjusted the results for pre-existing diabetes or plasma glucose concentrations at entry, but they did not adjust also for the effect of change in glucose tolerance status on the risk of new-onset HF during the follow-up period. Fourth, although we used a random-effects model, the interpretation of some results in this meta-analysis (like previously published meta-analyses9 10) should be made cautiously, given the significant heterogeneity observed in the pooled primary analysis. We explored possible sources of heterogeneity using subgroup analyses and metaregressions. It is reasonable to hypothesise that the significant heterogeneity likely reflects differences in the characteristics of study populations, study country, modality of NAFLD diagnosis and severity of NAFLD. However, future individual participant data meta-analyses will provide a more detailed analysis of the possible sources of heterogeneity. Finally, we also included three cohort studies that used serum GGT levels as a proxy for diagnosing NAFLD. That said, our subgroup analyses showed that the exclusion of these three cohort studies from the analysis did not substantially change the significant association between NAFLD and risk of new-onset HF (table 2).

    Our study also has important strengths. Our meta-analysis includes data from large cohort studies from Asia, Europe and the USA, and the included subjects are likely to be an accurate reflection of individuals worldwide with NAFLD, who are seen in routine clinical practice. The large number of incident cases of new-onset HF at follow-up also provides sufficiently high statistical power to quantify the magnitude of the association between NAFLD and the risk of new-onset HF. In addition, the overall quality of studies included in the meta-analysis was acceptable, according to the NOS scale. Finally, we did not extensively search for ‘grey literature’, so a possible selective reporting bias of studies cannot be definitely excluded. However, we believe that our comprehensive search has made it unlikely that any published studies were missed. Moreover, visual inspection of the funnel plot and the Egger’s regression test did not show any significant publication bias.

    In conclusion, the results of this large and updated meta-analysis provide evidence that NAFLD is moderately associated with a higher risk of new-onset HF over a median follow-up of 10 years. Importantly, the increase in risk of new-onset HF with NAFLD is independent of age, sex, ethnicity, adiposity measures, diabetes, hypertension and other coexisting cardiovascular risk factors. Our data also suggest that the magnitude of this increase in risk seems to parallel the severity of NAFLD, especially the severity of liver fibrosis. We believe that the results of our meta-analysis further highlight the need for a patient-centred, multidisciplinary and holistic approach to manage both liver disease and cardiovascular risk in people with NAFLD.5 Future high-quality prospective and mechanistic studies are certainly needed to better elucidate the existing but complex link between NAFLD and risk of new-onset HF; to examine whether genetic polymorphisms that increase risk of NAFLD (eg, the patatin-like phosholipase domain-containing 3 (PNPLA3) gene variant I148M and the transmembrane 6 superfamily member 2 (TM6SF2) E167K variant) can modify risk of HF and to test whether resolution or improvement of NAFLD (achieved by the use of newer antihyperglycaemic drugs, such as glucagon-like peptide-1 receptor agonists or sodium-glucose cotransporter-2 inhibitors) may reduce risk of new-onset HF.

    Data availability statement

    All data relevant to the study are included in the article or uploaded as supplementary information.

    Ethics statements

    Patient consent for publication

    Ethics approval

    Ethical approval is not requested for a meta-analysis.

    References

    1. ↵
      2. Cotter TG ,
      3. Rinella M
      . Nonalcoholic fatty liver disease 2020: the state of the disease. Gastroenterology 2020;158:1851–64.doi:10.1053/j.gastro.2020.01.052
      OpenUrlCrossRefPubMed
    2. ↵
      2. MH L ,
      3. Yeo YH ,
      4. Li X , et al
      . Global NAFLD prevalence: a systematic review and meta-analysis. Clin Gastroenterol Hepatol 2019;2021.
    3. ↵
      2. Estes C ,
      3. Razavi H ,
      4. Loomba R , et al
      . Modeling the epidemic of nonalcoholic fatty liver disease demonstrates an exponential increase in burden of disease. Hepatology 2018;67:123–33.doi:10.1002/hep.29466
      OpenUrlCrossRefPubMed
    4. ↵
      2. Byrne CD ,
      3. Targher G
      . NAFLD: a multisystem disease. J Hepatol 2015;62:S47–64.doi:10.1016/j.jhep.2014.12.012
      OpenUrlCrossRefPubMed
    5. ↵
      2. Targher G ,
      3. Tilg H ,
      4. Byrne CD
      . Non-alcoholic fatty liver disease: a multisystem disease requiring a multidisciplinary and holistic approach. Lancet Gastroenterol Hepatol 2021;6:578–88.doi:10.1016/S2468-1253(21)00020-0
      OpenUrlPubMed
    6. ↵
      2. Targher G ,
      3. Byrne CD ,
      4. Tilg H
      . NAFLD and increased risk of cardiovascular disease: clinical associations, pathophysiological mechanisms and pharmacological implications. Gut 2020;69:1691–705.doi:10.1136/gutjnl-2020-320622
      OpenUrlAbstract/FREE Full Text
    7. ↵
      2. Anstee QM ,
      3. Mantovani A ,
      4. Tilg H , et al
      . Risk of cardiomyopathy and cardiac arrhythmias in patients with nonalcoholic fatty liver disease. Nat Rev Gastroenterol Hepatol 2018;15:425–39.doi:10.1038/s41575-018-0010-0
      OpenUrl
    8. ↵
      2. McDonagh TA ,
      3. Metra M ,
      4. Adamo M
      . ESC guidelines for the diagnosis and treatment of acute and chronic heart failure. Eur Heart J 2021;2021:3599–726.
      OpenUrl
    9. ↵
      2. Alon L ,
      3. Corica B ,
      4. Raparelli V
      . Risk of cardiovascular events in patients with non-alcoholic fatty liver disease: a systematic review and meta-analysis. Eur J Prev Cardiol 2021.
    10. ↵
      2. Salah HM ,
      3. Pandey A ,
      4. Van Spall HGC , et al
      . Meta-Analysis of nonalcoholic fatty liver disease and incident heart failure. Am J Cardiol 2022;171:180–1.doi:10.1016/j.amjcard.2022.02.012 pmid:http://www.ncbi.nlm.nih.gov/pubmed/35305785
      OpenUrlPubMed
    11. ↵
      2. Page MJ ,
      3. McKenzie JE ,
      4. Bossuyt PM , et al
      . The PRISMA 2020 statement: an updated guideline for reporting systematic reviews. BMJ 2021;372:n71.doi:10.1136/bmj.n71 pmid:http://www.ncbi.nlm.nih.gov/pubmed/33782057
      OpenUrlFREE Full Text
    12. ↵
      2. Stroup DF ,
      3. Berlin JA ,
      4. Morton SC , et al
      . Meta-Analysis of observational studies in epidemiology: a proposal for reporting. meta-analysis of observational studies in epidemiology (moose) group. JAMA 2000;283:2008–12.doi:10.1001/jama.283.15.2008 pmid:http://www.ncbi.nlm.nih.gov/pubmed/10789670
      OpenUrlCrossRefPubMedWeb of Science
    13. ↵
      2. Cumpston M ,
      3. Li T ,
      4. Page MJ , et al
      . Updated guidance for trusted systematic reviews: a new edition of the Cochrane Handbook for systematic reviews of interventions. Cochrane Database Syst Rev 2019;10:ED000142.doi:10.1002/14651858.ED000142 pmid:http://www.ncbi.nlm.nih.gov/pubmed/31643080
      OpenUrlCrossRefPubMed
    14. ↵
      2. Higgins JPT ,
      3. Thompson SG
      . Quantifying heterogeneity in a meta-analysis. Stat Med 2002;21:1539–58.doi:10.1002/sim.1186
      OpenUrlCrossRefPubMedWeb of Science
    15. ↵
      2. Dhingra R ,
      3. Gona P ,
      4. Wang TJ , et al
      . Serum gamma-glutamyl transferase and risk of heart failure in the community. Arterioscler Thromb Vasc Biol 2010;30:1855–60.doi:10.1161/ATVBAHA.110.207340 pmid:http://www.ncbi.nlm.nih.gov/pubmed/20539015
      OpenUrlAbstract/FREE Full Text
    16. ↵
      2. Wannamethee SG ,
      3. Whincup PH ,
      4. Shaper AG , et al
      . γ-Glutamyltransferase, hepatic enzymes, and risk of incident heart failure in older men. Arterioscler Thromb Vasc Biol 2012;32:830–5.doi:10.1161/ATVBAHA.111.240457
      OpenUrlAbstract/FREE Full Text
    17. ↵
      2. Wang Y ,
      3. Tuomilehto J ,
      4. Jousilahti P , et al
      . Serum γ-glutamyltransferase and the risk of heart failure in men and women in Finland. Heart 2013;99:163–7.doi:10.1136/heartjnl-2012-302972
      OpenUrlAbstract/FREE Full Text
    18. ↵
      2. Shah RV ,
      3. Anderson A ,
      4. Ding J , et al
      . Pericardial, But Not Hepatic, Fat by CT Is Associated With CV Outcomes and Structure: The Multi-Ethnic Study of Atherosclerosis. JACC Cardiovasc Imaging 2017;10:1016–27.doi:10.1016/j.jcmg.2016.10.024 pmid:http://www.ncbi.nlm.nih.gov/pubmed/28330662
      OpenUrlAbstract/FREE Full Text
    19. ↵
      2. Allen AM ,
      3. Therneau TM ,
      4. Mara KC , et al
      . Women with nonalcoholic fatty liver disease lose protection against cardiovascular disease: a longitudinal cohort study. Am J Gastroenterol 2019;114:1764–71.doi:10.14309/ajg.0000000000000401
      OpenUrl
    20. ↵
      2. Roh J-H ,
      3. Park J-H ,
      4. Lee H , et al
      . Higher fatty liver index is associated with increased risk of new onset heart failure in healthy adults: a nationwide population-based study in Korea. BMC Cardiovasc Disord 2020;20:204.doi:10.1186/s12872-020-01444-x
    21. ↵
      2. Lee H ,
      3. Lee Y-ho ,
      4. Kim SU , et al
      . Metabolic Dysfunction-Associated fatty liver disease and incident cardiovascular disease risk: a nationwide cohort study. Clinical Gastroenterology and Hepatology 2021;19:2138–47.doi:10.1016/j.cgh.2020.12.022
      OpenUrl
    22. ↵
      2. Fudim M ,
      3. Zhong L ,
      4. Patel KV , et al
      . Nonalcoholic fatty liver disease and risk of heart failure among Medicare beneficiaries. J Am Heart Assoc 2021;10:e021654.doi:10.1161/JAHA.121.021654
    23. ↵
      2. Park J ,
      3. Kim G ,
      4. Kim H , et al
      . The association of hepatic steatosis and fibrosis with heart failure and mortality. Cardiovasc Diabetol 2021;20:197.doi:10.1186/s12933-021-01374-8
    24. ↵
      2. Zou B ,
      3. Yeo YH ,
      4. Cheung R , et al
      . Fatty liver index and development of cardiovascular disease: findings from the UK Biobank. Dig Dis Sci 2021;66:2092–100.doi:10.1007/s10620-021-06954-y
      OpenUrl
    25. ↵
      2. Simon TG ,
      3. Roelstraete B ,
      4. Hagstrom H , et al
      . Non-alcoholic fatty liver disease and incident major adverse cardiovascular events: results from a nationwide histology cohort. Gut 2021.
    26. ↵
      2. Dunn MA ,
      3. Behari J ,
      4. Rogal SS , et al
      . Hepatic steatosis in diabetic patients does not predict adverse liver-related or cardiovascular outcomes. Liver Int 2013;33:1575–82.doi:10.1111/liv.12285
      OpenUrl
    27. ↵
      2. Vita T ,
      3. Murphy DJ ,
      4. Osborne MT , et al
      . Association between nonalcoholic fatty liver disease at CT and coronary microvascular dysfunction at myocardial perfusion PET/CT. Radiology 2019;291:330–7.doi:10.1148/radiol.2019181793
      OpenUrl
    28. ↵
      2. VanWagner LB ,
      3. Wilcox JE ,
      4. Ning H , et al
      . Longitudinal association of Non‐Alcoholic fatty liver disease with changes in myocardial structure and function: the cardia study. J Am Heart Assoc 2020;9:e014279.doi:10.1161/JAHA.119.014279
    29. ↵
      2. Ichikawa K ,
      3. Miyoshi T ,
      4. Osawa K , et al
      . Prognostic value of non-alcoholic fatty liver disease for predicting cardiovascular events in patients with diabetes mellitus with suspected coronary artery disease: a prospective cohort study. Cardiovasc Diabetol 2021;20:8.doi:10.1186/s12933-020-01192-4
    30. ↵
      2. Lee C-O ,
      3. Li H-L ,
      4. Tsoi M-F , et al
      . Association between the liver fat score (LFS) and cardiovascular diseases in the National health and nutrition examination survey 1999–2016. Ann Med 2021;53:1067–75.doi:10.1080/07853890.2021.1943514
      OpenUrl
    31. ↵
      2. Vieira Barbosa J ,
      3. Milligan S ,
      4. Frick A , et al
      . Fibrosis-4 index can independently predict major adverse cardiovascular events in nonalcoholic fatty liver disease. Am J Gastroenterol 2022;117:453–61.doi:10.14309/ajg.0000000000001606
      OpenUrl
    32. ↵
      2. Dauriz M ,
      3. Mantovani A ,
      4. Bonapace S , et al
      . Prognostic impact of diabetes on long-term survival outcomes in patients with heart failure: a meta-analysis. Diabetes Care 2017;40:1597–605.doi:10.2337/dc17-0697
      OpenUrlAbstract/FREE Full Text
    33. ↵
      2. Mantovani A ,
      3. Petracca G ,
      4. Beatrice G , et al
      . Non-alcoholic fatty liver disease and risk of incident diabetes mellitus: an updated meta-analysis of 501 022 adult individuals. Gut 2021;70:962–9.doi:10.1136/gutjnl-2020-322572
      OpenUrlAbstract/FREE Full Text
    34. ↵
      2. Mantovani A ,
      3. Csermely A ,
      4. Petracca G , et al
      . Non-alcoholic fatty liver disease and risk of fatal and non-fatal cardiovascular events: an updated systematic review and meta-analysis. Lancet Gastroenterol Hepatol 2021;6:903–13.doi:10.1016/S2468-1253(21)00308-3
      OpenUrlPubMed
    35. ↵
      2. Mantovani A ,
      3. Dauriz M ,
      4. Sandri D , et al
      . Association between non‐alcoholic fatty liver disease and risk of atrial fibrillation in adult individuals: an updated meta‐analysis. Liver Int 2019;39:758–69.doi:10.1111/liv.14044
      OpenUrl
    36. ↵
      2. Mantovani A ,
      3. Petracca G ,
      4. Beatrice G , et al
      . Non-alcoholic fatty liver disease and risk of incident chronic kidney disease: an updated meta-analysis. Gut 2022;71:156-162.doi:10.1136/gutjnl-2020-323082 pmid:http://www.ncbi.nlm.nih.gov/pubmed/33303564
      OpenUrlPubMed
    37. ↵
      2. Mantovani A ,
      3. Petracca G ,
      4. Beatrice G , et al
      . Non-Alcoholic fatty liver disease and increased risk of incident extrahepatic cancers: a meta-analysis of observational cohort studies. Gut 2022;71:778–88.doi:10.1136/gutjnl-2021-324191 pmid:http://www.ncbi.nlm.nih.gov/pubmed/33685968
      OpenUrlAbstract/FREE Full Text
    38. ↵
      2. Peters AE ,
      3. Pandey A ,
      4. Ayers C , et al
      . Association of liver fibrosis risk scores with clinical outcomes in patients with heart failure with preserved ejection fraction: findings from TOPCAT. ESC Heart Failure 2021;8:842–8.doi:10.1002/ehf2.13250
      OpenUrl
    39. ↵
      2. Yoshihisa A ,
      3. Sato Y ,
      4. Yokokawa T , et al
      . Liver fibrosis score predicts mortality in heart failure patients with preserved ejection fraction. ESC Heart Failure 2018;5:262–70.doi:10.1002/ehf2.12222
      OpenUrl
    40. ↵
      2. Mantovani A ,
      3. Byrne CD ,
      4. Benfari G , et al
      . Risk of heart failure in patients with nonalcoholic fatty liver disease: JACC review topic of the week. J Am Coll Cardiol 2022;79:180–91.
      OpenUrl
    41. ↵
      2. Mantovani A ,
      3. Byrne CD ,
      4. Targher G
      . Efficacy of peroxisome proliferator-activated receptor agonists, glucagon-like peptide-1 receptor agonists, or sodium-glucose cotransporter-2 inhibitors for treatment of non-alcoholic fatty liver disease: a systematic review. Lancet Gastroenterol Hepatol 2022;7:367–78.doi:10.1016/S2468-1253(21)00261-2 pmid:http://www.ncbi.nlm.nih.gov/pubmed/35030323
      OpenUrlPubMed
    42. ↵
      2. Brown E ,
      3. Heerspink HJL ,
      4. Cuthbertson DJ , et al
      . SGLT2 inhibitors and GLP-1 receptor agonists: established and emerging indications. The Lancet 2021;398:262–76.doi:10.1016/S0140-6736(21)00536-5
      OpenUrl

    Supplementary materials

    • Supplementary Data

      This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

    Footnotes

    • Contributors AM and GT designed the study, analysed the data and did the figures. AM, GP, AC and GT did the literature search, with arbitration by AC. GB, AR, SB, HT and CDB interpreted the data. GT and CDB wrote the manuscript. All authors reviewed and approved the final version of the manuscript. GT is the guarantor and accepts full responsibility for the work and/or the conduct of the study, had access to the data and controlled the decision to publish.

    • Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.

    • Competing interests None declared.

    • Patient and public involvement Patients and/or the public were not involved in the design, conduct, reporting or dissemination plans of this research.

    • Provenance and peer review Not commissioned; externally peer reviewed.

    • Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.