Non-alcoholic fatty liver disease increases the risk of incident chronic kidney disease

Abstract

Background and aim

Non-alcoholic fatty liver disease (NAFLD) is a highly prevalent chronic liver disease. Its role in the development of extrahepatic co-morbidities is under investigation. The impact of NAFLD on the development of chronic kidney disease (CKD) is incompletely understood. The aim of this study was to explore the potential contribution of NAFLD on CKD in Germany.

Methods

The Disease Analyzer Database covering 7.49 million cases in Germany was explored for patients diagnosed with NAFLD between 2000 and 2015 and was matched 1:1 to a cohort without NAFLD. Matching criteria included age, sex, physician, index year and co-diagnoses associated with CKD. The primary outcomes of this study were incidences of CKD and end-stage renal disease.

Results

A total of 48,057 patients with NAFLD were matched to 48,057 patients without NAFLD. Within 10 years of the index date, 17.1% of patients with NAFLD and 11.6% of patients without NAFLD were diagnosed with CKD (p < 0.001). On Cox regression analysis, NAFLD was significantly associated with the incidence of CKD (hazard ratio (HR) = 1.58, p < 0.001). This association remained significant across different age groups and subgroups such as patients with diabetes mellitus or arterial hypertension. There was no association between NAFLD and emerging dialysis therapy (HR = 1.25, p = 0.245).

Conclusions

In this large database analysis in Germany, NAFLD constitutes an independent risk factor for CKD. Patients living with NAFLD should be monitored for a change in kidney function, facilitating therapeutic measures for kidney disease at an early stage.

Introduction

Non-alcoholic fatty liver (NAFL) disease (NAFLD) is the most frequent chronic liver disease in Europe, with an estimated prevalence of 24%.¹,² NAFLD compromises a wide spectrum, ranging from simple steatosis (NAFL) to non-alcoholic steatohepatitis (NASH) and finally liver cirrhosis.³,⁴ The mortality risk of NAFLD is determined by the disease stage and is prominently influenced by the amount of hepatic fibrosis.⁵ The excess mortality in patients living with NAFLD is prominently related non-hepatic co-morbidities.⁶ In a German population, an independent contribution on the emerging risk of cardiovascular disease and cancer has been established.^7–9 Pathophysiological extrahepatic co-morbidities have been suggested to result from chronic low-grade inflammation in metabolically burdened patients (‘metabolic inflammation’) that in part originates in the liver, and as such, NAFLD is a potential co-factor and driver to these underlying co-morbidities.¹⁰

Chronic kidney disease (CKD) represents a significant health burden in Germany, affecting more than 25% of the elderly population (>65 years).¹¹ Overall, CKD causes significant mortality and progression to end-stage renal disease (ESRD), with the need for renal replacement therapy or kidney transplantation constituting a main driver of individual disease burden and socio-economic costs.¹¹ A close link between CKD and metabolic and vascular diseases has been established, and these risk factors are shared by patients living with NAFLD. Thus, the association of CKD with NAFLD could be related to the high prevalence in the general population and joint risk factors or it could exist independently.¹² A recent meta-analysis indicated that there is a link between NAFLD and the development of CKD.¹³ Additionally, a large Asian-based study was able to demonstrate an association between NAFLD and CKD, being consistent across relevant subgroups such as patients with dyslipidaemia or elevated blood pressure.¹⁴ The association of NAFLD and CKD has been established before, but large population-based studies in Germany have not been available. The aim of the current large-scale database study was to investigate the role of NAFLD on emerging CKD, the emergence of ESRD and the identification of additional risk factors which may increase the risk for CKD.

Methods

Database and ethical considerations

This study explores data derived from the Disease Analyzer database (IQVIA), which compiles drug prescriptions, diagnoses and basic medical and demographic data obtained directly and in anonymous format from computer systems used in the practices of general practitioners and specialists. Patient data were analysed as aggregates, with no individual health data available. Related to the analysis of anonymized data, no individual informed consent was obtained. The database covers approximately 3% of all outpatient practices in Germany. Diagnoses (according to International Classification of Diseases, 10th revision (ICD-10)]), prescriptions (according to Anatomical Therapeutic Chemical Classification system) and the quality of reported data are monitored by IQVIA. In Germany, the sampling methods used to select physicians’ practices are updated on a yearly basis in order to obtain a representative database in primary and secondary care.

Study population

This retrospective cohort study included adult patients (≥18 years) with an initial diagnosis of NAFLD (ICD-10: K75.8, K76.0) in 1262 general practices in Germany between January 2000 and December 2015 (index date; Figure 1). A further inclusion criterion was an observation time of at least 12 months prior to the index date. Patients with cancer diagnoses (ICD-10: C00–C99), other liver disorders (ICD-10: K70–K79) and CKD (ICD-10: N18), other renal failure (ICD-10: N17, N19) or diabetic nephropathy (ICD-10: E10.2, E11.2, E14.2) prior to the index date were excluded.

Figure 1.

Selection of study patients within the Disease Analyzer Database.

NAFLD patients were matched to non-NAFLD patients by age, sex, physician, index year, diabetes, hypertension, obesity, ischaemic heart diseases, glomerular diseases and renal tubulo-interstitial diseases. For the controls, the index date was that of a randomly selected visit between January 2000 and December 2015 (Figure 1).

Study outcomes and covariates

The main outcome of the study was the cumulative incidence of CKD (ICD 10: N18, Z49) as a function of NAFLD. In diabetes patients, the definition of CKD additionally included diabetic nephropathy after the index date (ICD-10: E10.2, E11.2, E14.2). Patients were followed for up to 10 years. Additionally, the cumulative incidence of dialysis therapy (ICD 10: Z49) was estimated.

Statistical analyses

Differences in the sample characteristics between those with and those without NAFLD were tested using chi-square tests for categorical variables and Wilcoxon tests for continuous variables. Hazard regression models were conducted to study the association between the NAFLD and CKD/dialysis therapy. These models were performed separately for patients in five age groups, female and male patients as well as patients with diabetes, obesity, hypertension and ischaemic heart diseases. p-Values  < 0.05 were considered statistically significant. Analyses were carried out using SAS v9.4 (SAS Institute, Cary, NC).

Results

Basic characteristics of the study cohort

The present study included 48,057 patients with NAFLD and 48,057 patients without NAFLD. The mean age of the patients was 58.8 years (standard deviation (SD) = 13.7 years), and 47.1% were women. After matching, arterial hypertension was the most frequent metabolic co-morbidity in both groups (58.3%). Other baseline characteristics of the study patients are summarized in Table 1. Next, we explored the plausibility of NAFLD coding in the database. In NAFLD patients, the hazard ratio (HR) of cirrhosis was 2.36 (95% confidence interval (CI) 1.98–2.82, p < 0.001), and the HR for cirrhosis with associated complications was 2.81 (95% CI 1.75–4.52, p < 0.001).

Table 1.

Basic characteristics of the study sample (after 1:1 matching by age, sex, physician, index year, diabetes, hypertension, obesity, ischaemic heart diseases, glomerular diseases and renal tubulo-interstitial diseases).

Variable	Proportion affected among patients with NAFLD (%), N = 48,057	Proportion affected among patients without NAFLD (%), N = 48,057	p
Age (years), M (SD)	58.8 (13.7)	58.8 (13.7)	1.000
Age 18–40 years	9.9	9.9	1.000
Age 41–50 years	17.3	17.3
Age 51–60 years	25.8	25.8
Age 61–70 years	25.2	25.2
Age >70 years	21.8	21.8
Women	47.1	47.1	1.000
Men	52.9	52.9
Diagnosed prior to or on the index date
Diabetes	25.2	25.2	1.000
Obesity	22.1	22.1	1.000
Hypertension	58.3	58.3	1.000
Ischaemic heart disease	16.7	16.7	1.000
Glomerular or renal tubulo-interstitial diseases	2.1	2.1	1.000

Proportion of patients in % shown, unless otherwise indicated.

NAFLD: non-alcoholic fatty liver disease; SD: standard deviation.

Table 2.

Association between NAFLD and the incidence CKD diagnoses and dialysis therapy in patients followed in general practices in Germany (Cox regression models).

	CKD		Dialysis
Cohort	HR (95% CI)	p	HR (95% CI)	p
All patients	1.58 (1.51–1.66)	<0.001	1.25 (0.86–1.83)	0.245
Age 18–40 years	1.96 (1.37–2.81)	<0.001	2.82 (0.76–10.42)	0.120
Age 41–50 years	2.16 (1.77–2.62)	<0.001	2.59 (0.52–12.93)	0.248
Age 51–60 years	1.61 (1.45–1.79)	<0.001	1.31 (0.59–2.92)	0.515
Age 61–70 years	1.51 (1.39–1.64)	<0.001	1.18 (0.62–2.23)	0.616
Age >70 years	1.52 (1.42–1.63)	<0.001	0.76 (0.36–1.62)	0.479
Women	1.61 (1.51–1.71)	<0.001	1.08 (0.61–1.92)	0.793
Men	1.56 (1.46–1.66)	<0.001	1.40 (0.84–2.33)	0.193
Diabetes patients	1.48 (1.39–1.58)	<0.001	0.82 (0.45–1.50)	0.515
Obesity patients	1.35 (1.24–1.47)	<0.001	0.83 (0.40–1.75)	0.630
Hypertension patients	1.55 (1.47–1.63)	<0.001	1.23 (0.76–1.96)	0.391
Ischaemic heart disease patients	1.32 (1.22–1.44)	<0.001	1.19 (0.51–2.77)	0.685

CKD: chronic kidney disease; HR: hazard ratio; CI: confidence interval.

Role of NAFLD on the incidence of CKD

Within 10 years of the index date, 17.1% of patients with NAFLD and 11.6% of patients without NAFLD were diagnosed with CKD (log-rank p < 0.001; Figure 2). On regression analyses, NAFLD was significantly associated with the incidence of CKD (HR = 1.58, p < 0.001). Sensitivity analyses showed the strongest association among patients aged 41–50 years (HR = 2.16, p < 0.001), followed by patients in the 18–40 age group (HR = 1.96, p < 0.001). The association between NAFLD and the development of CKD remained significant across subgroups with relevant co-morbidities, including diabetes mellitus (HR = 1.48, p < 0.001), obesity (HR = 1.35, p < 0.001) and arterial hypertension (HR = 1.55, p < 0.001). An association was also observed in patients with ischaemic heart disease (HR = 1.31; p < 0.001).

Figure 2.

Kaplan–Meier curves for time to chronic kidney disease diagnosis in patients with and without non-alcoholic fatty liver disease (NAFLD).

Association of NAFLD and incidence of haemodialysis

Within 10 years of the index date, 0.2% of patients with NAFLD and 0.2% of patients without NAFLD were diagnosed with CKD (log-rank p = 0.447; Figure 3). No association was observed between NAFLD and haemodialysis therapy (HR = 1.25, p = 0.245; Table 2).

Figure 3.

Kaplan–Meier curves for time to begin dialysis therapy in patients with and without NAFLD.

Discussion

In this large database study in Germany, the risk of emerging CKD was specifically increased in patients with NAFLD. The association between NAFLD and the incidence of CKD remained independently associated across different age groups, both sexes and the presence of diabetes, obesity, hypertension and ischaemic heart disease. In contrast, in the current analysis, NAFLD was not independently associated with an emerging risk for ESRD requiring haemodialysis.

The pathogenesis underlying NAFLD is multifactorial and involves various metabolic conditions.⁵,¹⁵ The most frequently encountered co-morbidities include obesity, diabetes mellitus type 2 and hyperlipidaemia – all of which constitute metabolic syndrome. There is a growing body of evidence that NAFLD, especially the inflammatory form of NASH, may be associated with CKD.¹⁶ A recent meta-analysis included nine observational studies encompassing a total of 96,595 individuals with 32,898 cases of NAFLD. Herein, an association of NAFLD with the incidence of CKD (HR = 1.37) in Asia and Europe was observed.¹³ In the current study, NAFLD patients had a 58% increased risk for the development of CKD when compared to patients without liver disease. Additionally, we were able to demonstrate that this association remained significant across high-risk subgroups, including diabetes mellitus type 2 (HR = 1.48) and arterial hypertension (HR = 1.55). In a recent multinational cohort study, the HR for CKD was 1.79 in patients with diabetes mellitus type 2.¹⁷ In a cohort study of patients with NAFLD conducted in Korea, an increased risk for CKD was observed in patients with NASH versus non-inflammatory fatty liver disease.¹⁴ In the current study, we observed the highest incidence of CKD in the 41–50 age group (HR = 2.16), followed by the 18–40 age group (HR-1.96). Importantly, older patients exhibited lower HRs, suggesting that NAFLD adds to the risk of CKD at a time when traditional risk factors are not as prominent.

Multiple pathophysiological mechanisms are under debate which may link NAFLD to CKD. The liver is a central regulator of the metabolism, controls gluconeogenesis and is a major side for synthesis of fatty acids. Several studies have highlighted metabolic imbalances in NAFLD, and the emerging chronic, subclinical inflammation (‘metabolic inflammation’) may be an underlying co-factor in determining the risk of developing CKD.¹¹,¹⁸,¹⁹ In this context, the renin–angiotensin system (RAS) is a potential link between NAFLD and CKD, as RAS activation has been implied in the production of pro-inflammatory cytokines, in particular interleukin-6 that promotes oxidative stress.²⁰ Thus, RAS activation in NAFLD can support subclinical organ dysfunction, eventually leading to CKD.

While there is no specific treatment for CKD in NAFLD, treatments for NAFLD are emerging and could potentially impact the progression to CKD. Most prominently, lifestyle changes aiming at weight loss are recommended for patients with NAFLD treatment.³ More recently, the FXR-agonist obeticholic acid was shown to induce fibrosis regression in the absence of worsening of steatohepatitis in patients with NASH.²¹ It will be interesting to determine if this liver-specific treatment will also be beneficial to prevent extrahepatic co-morbidities in NAFLD. On the other hand, safety concerns related to temporal worsening of blood lipid profiles with increasing low-density lipoprotein as such extrahepatic side effects may be a limitation for using this drug in high-risk populations.²²

The terminal complication of CKD is haemodialysis. In a recent analysis including 161 patients with NAFLD and pre-existing CKD, an increased incidence of haemodialysis was observed.²² This was not seen in the current study and is likely related to the differences between the two study populations at baseline and the duration of follow-up. In the current analysis, we chose to exclude patients with pre-existing CKD and followed them over a maximum of 10 years. In the entire population and observation period, 1.2% of patients required haemodialysis (66 in the NAFLD group and 45 in the non-NAFLD group; p>0.05). Considering this low incidence rate of haemodialysis (1.2%), it is conceivable that we included predominantly early CKD in NAFLD that did not reach ERSD in the observation time. Thus, despite the large study population, we did not observe an impact of NALFD on emerging ESRD in this population after excluding patients with prior kidney disease over the observation time of 10 years.

This study has inherent limitations related to mis- or under-coding and thus a potential disease spectrum bias in which only advanced or clinical obvious cases were recorded. In the absence of lab results or imaging studies, we could not confirm the validity of the NAFLD codes in the Disease Analyzer Database. However, the plausibility of the coding is suggested by the significantly increased HR for cirrhosis compared to the non-NAFLD cohort. In addition, a recent analysis confirmed an acceptable accuracy for the codes used in the Disease Analyzer Database.²³ The overall prevalence of NAFLD in the Disease Analyzer Database was 3.3%, clearly below the expected prevalence.¹ Furthermore, misclassification bias is among the concerns, and no radiological or ultrasound-based data to assess hepatic steatosis are available. Due to the lack of laboratory data, we were unable to stratify NAFLD based on non-invasive surrogates to define advanced disease stages or to define the stage of CKD impairment. The total number of deaths is also not registered in the database, and thus survival data could not be calculated. Likewise, additional risk factors for CKD, including family history, smoking or medication, were not available in this analysis. Finally, the analysis constitutes an exploratory observation that highlights an independent association between CKD and NAFLD but does not show a causal link. Among the strengths of this analysis is the exploration of 96,114 patients treated in primary and secondary care in Germany. The German Disease Analyzer database combines data from primary and secondary care in Germany and is adjusted to represent outpatient practice and thus provides a representative database. With appropriate matching, potential confounders were reduced.

In conclusion, this large-scale database research study adds to the growing body of evidence suggesting an independent association of NAFLD with emerging CKD. The strongest association was found in middle-aged patients. This association remained robust across relevant high-risk groups, including patients living with diabetes mellitus type 2 or arterial hypertension. Thus, the current work underlines the need to assess patients with NAFLD for CKD.

Declaration of conflicting interests

The authors declared the following potential conflicts of interest with respect to the research, authorship, and/or publication of this article: K.K. is an employee of IQVIA. J.M.S. has acted as consultant to Genfit, Gilead Sciences, Intercept Pharmaceuticals, IQVIA, Madrigal, Pfizer, Novartis, Roche and Siemens Healthineers and has received research funding from Gilead Sciences. All other authors have nothing to declare.

Ethics approval

Anonymized electronic medical records were analyzed as aggregates with no identifiable, individual health data available. Accordingly, no study-specific ethical board approval was required (Ethikkommission der Landesärztekammer Rheinland-Pfalz).

Funding

The authors disclosed receipt of the following financial support for the research, authorship and/or publication of this article: This work was funded in part by intramural funds of the University Medical Centre, Mainz, Germany.

Informed consent

Individual informed consent or IRB approval was not obtained.

ORCID iD

Jörn M Schattenberg https://orcid.org/0000-0002-4224-4703