Higher liver stiffness scores are associated with early kidney dysfunction in patients with histologically proven non-cirrhotic NAFLD

Journal Pre-proof Higher liver stiffness scores are associated with early kidney dysfunction in patients with histologically proven non-cirrhotic NAFLD Dan-Qin Sun Fang-Zhou Ye Haluk Tarik Kani Jian-Rong Yang Kenneth I. Zheng Hao-Yang Zhang Giovanni Targher Christopher D Byrne Yong-Ping Chen Wei-Jie Yuan Yusuf Yilmaz Ming-Hua Zheng MD PhD

PII:

S1262-3636(19)30183-1

DOI:

https://doi.org/doi:10.1016/j.diabet.2019.11.003

Reference:

DIABET 101132

To appear in:

Diabetes & Metabolism

Received Date:

10 September 2019

Revised Date:

3 November 2019

Accepted Date:

9 November 2019

Please cite this article as: Sun D-Qin, Ye F-Zhou, Kani HT, Yang J-Rong, Zheng KI, Zhang H-Yang, Targher G, Byrne CD, Chen Y-Ping, Yuan W-Jie, Yilmaz Y, Zheng M-Hua, Higher liver stiffness scores are associated with early kidney dysfunction in patients with histologically proven non-cirrhotic NAFLD, Diabetes and Metabolism (2019), doi: https://doi.org/10.1016/j.diabet.2019.11.003

This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of record. This version will undergo additional copyediting, typesetting and review before it is published in its final form, but we are providing this version to give early visibility of the article. Please note that, during the production process, errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain. © 2019 Published by Elsevier.

Higher liver stiffness scores are associated with early kidney dysfunction in patients with histologically proven non-cirrhotic NAFLD Dan-Qin Sun1,2#, Fang-Zhou Ye3,4#, Haluk Tarik Kani5#, Jian-Rong Yang6, Kenneth I. Zheng3, Hao-Yang Zhang7, Giovanni Targher8, Christopher D. Byrne9, Yong-Ping Chen3,10, Wei-Jie Yuan2, Yusuf Yilmaz5,11, Ming-Hua Zheng3,10 1

Department of Nephrology, Affiliated Wuxi No. 2 People’s Hospital of Nanjing Medical University,

Wuxi, China Department of Nephrology, Shanghai General Hospital of Nanjing Medical University, Shanghai, China

3

NAFLD Research Centre, Department of Hepatology, First Affiliated Hospital of Wenzhou Medical

oo f

2

University, Wenzhou, China

School of the First Clinical Medical Sciences, Wenzhou Medical University, Wenzhou, China

5

Department of Gastroenterology, Marmara University School of Medicine, Istanbul, Turkey

6

Department of Clinical Laboratory, First Affiliated Hospital of Wenzhou Medical University, Wenzhou,

e-

pr

4

China

School of Biomedical Engineering, Sun Yat-sen University, Guangzhou, China

8

Division of Endocrinology, Diabetes and Metabolism, Department of Medicine, University and Azienda

Pr

7

9

al

Ospedaliera Universitaria Integrata of Verona, Verona, Italy Southampton National Institute for Health Research Biomedical Research Centre, University Hospital

Jo ur n

Southampton, Southampton General Hospital, Southampton, UK 10

Institute of Hepatology, Wenzhou Medical University, Wenzhou, China

11

Institute of Gastroenterology, Marmara University, Istanbul, Turkey

Short title: Liver stiffness and early kidney disease Co-first Dan-Qin Sun, Fang-Zhou Ye, Haluk Tarik Kani Corresponding author: Ming-Hua Zheng, MD, PhD NAFLD Research Centre, Department of Hepatology, First Affiliated Hospital of Wenzhou Medical University; No. 2 Fuxue Lane, Wenzhou 325000, China

1

Page 1 of 24

E-mail: [email protected] Guarantor of article: Ming-Hua Zheng Received 10 September 2019; Accepted 9 November 2019AUROC: area under the receiver operating characteristic curve BMI: body mass index CI: confidence interval

oo f

CKD: chronic kidney disease eGFR: estimated glomerular filtration rate

pr

EKD: early kidney dysfunction

OR: odds ratio

Jo ur n

LF: liver fibrosis

al

LSM: liver stiffness measurement

Pr

NAFLD: non-alcoholic fatty liver disease

e-

HOMA-IR: homoeostasis model assessment of insulin resistance

TE: transient elastography

2

Page 2 of 24

Abstract Aim. – The association between liver fibrosis (LF), as assessed by either histology or liver stiffness measurement (LSM), and the presence of early kidney dysfunction (EKD) was investigated in this study, as was also the diagnostic performance of LSM for identifying the presence of EKD in patients with nonalcoholic fatty liver disease (NAFLD). Materials and Methods. – A total of 214 adults with non-cirrhotic biopsy-proven NAFLD were recruited from two independent medical centres. Their histological stage of LF was quantified using Brunt’s criteria.

oo f

Vibration-controlled transient elastography (TE), using M-probe (FibroScan®) ultrasound, was performed in 154 patients and defined as significant when LSM was ≥ 8.0 kPa. EKD was defined as the presence of

pr

microalbuminuria with an estimated glomerular filtration rate ≥ 60 mL/min/1.73 m2. Logistic regression modelling was used to estimate the likelihood of having EKD with NAFLD (LSM–EKD model).

e-

Results. – The prevalence of EKD was higher in patients with vs without LF on histology (22.14% vs 4.82%,

Pr

respectively; P < 0.001) and, similarly, EKD prevalence was higher in patients with LSM ≥ 8.0 kPa vs LSM < 8.0 kPa (23.81% vs 6.59%, respectively; P < 0.05). The area under the ROC curve of the LSM–EKD model for identifying EKD was 0.80 (95% CI: 0.72–0.89). LF detected by either method was associated

al

with EKD independently of established renal risk factors and potential confounders.

Jo ur n

Conclusion. – LF was independently associated with EKD in patients with biopsy-proven NAFLD. Thus, TE-measured LSM, a widely used technique for quantifying LF, can accurately identify those patients with NAFLD who are at risk of having EKD.

Abbreviations ACR: albumin-to-creatinine ratio

Keywords: Early kidney dysfunction; Liver fibrosis; Liver stiffness measurement; Non-alcoholic fatty 3

Page 3 of 24

Jo ur n

al

Pr

e-

pr

oo f

liver disease

4

Page 4 of 24

Introduction Non-alcoholic fatty liver disease (NAFLD), a major cause of liver-related mortality, also increases the risk of developing chronic kidney disease (CKD) [1–3]. CKD, a general term used for heterogeneous diseases affecting both kidney structure and function, is also often associated with an increased risk of cardiovascular events and all-cause mortality [4]. Progression to end-stage kidney disease can only be treated by chronic dialysis or renal transplantation, thereby placing a considerable economic burden on health services [5]. Early stages of CKD are often asymptomatic and may be detected only incidentally

oo f

during evaluation for other diseases [6]. Yet, as early kidney dysfunction (EKD) may be reversible, its early detection is clinically important.

pr

In recent years, several studies have examined the association between NAFLD and risk of CKD, and supported the conclusion that the presence and severity of NAFLD, diagnosed by either imaging or

e-

histology, is associated with an increased prevalence and incidence of CKD [1, 7, 8]. While these studies

Pr

have provided evidence of a strong association between the presence and severity of NAFLD and risk of stage 3–5 CKD [7–9], there are so far very few studies that have examined the association between NAFLD

60 mL/min/1.73 m2.

al

and EKD, defined as the presence of microalbuminuria with an estimated glomerular filtration (eGFR) ≥

Jo ur n

Transient elastography (TE) has been extensively used as a non-invasive technique for measuring liver fibrosis (LF), which in itself may identify a subgroup of patients with NAFLD who are at higher risk of liver-related clinical outcomes, including the need for liver transplantation [10–12]. However, the diagnostic value of liver stiffness measurement (LSM), obtained from TE, for identifying EKD among patients with NAFLD is currently uncertain. For this reason, the major aim of our present study was to examine the association between LF severity (whether assessed invasively by liver histology or noninvasively by TE) and the presence of EKD in patients with biopsy-proven NAFLD. In addition, our study also investigated the diagnostic performance of both LF assessed by histology and LSM assessed by TE in predicting the presence of EKD in this patient population.

Materials and Methods 5

Page 5 of 24

Study design A total of 742 adult patients who had undergone liver biopsy were identified at two medical centres: First Affiliated Hospital of Wenzhou Medical University in China; and Marmara University School of Medicine in Istanbul, Turkey. Of these patients, 214 with biopsy-confirmed NAFLD were included in the study (Fig. 1). At both centres, patients who had any of the following criteria were excluded: excessive alcohol consumption (> 140 g/week for men, > 70 g/week for women); urinary tract infections; fever; and menstruation. Participants with other known causes of chronic liver disease (such as viral hepatitis,

oo f

autoimmune hepatitis, primary biliary cholangitis, Wilson’s disease) were also excluded from our analyses on the basis of liver biopsy results as well as tests for serological markers of hepatitis A, B and C, antimitochondrial antibodies, smooth muscle cell antibodies, antinuclear antibodies, anti-liver kidney

pr

microsomal antibodies and levels of caeruloplasmin, as were those with a previous history of cirrhosis of

e-

any aetiology, CKD stages 3–5, macroalbuminuria (urinary albumin-to-creatinine ratio > 300 mg/g) or other known causes of kidney disease.

Pr

Written informed consent was obtained from each patient before participating in the study. As all personal information was removed, study participants were identified by only their health examination number. In

al

Wenzhou, the research protocol was approved by the ethics committee of the First Affiliated Hospital of

Jo ur n

Wenzhou Medical University, while the local Institutional Review Board approved the study in Istanbul. Liver biopsy and transient elastography Ultrasound-guided liver biopsy was performed under sedation using a 16-gauge Hepafix needle, and all biopsy specimens were placed in formalin solution for fixation, embedded in paraffin blocks and stained with haematoxylin and eosin or Masson’s trichrome. All liver pathology slides (from both Wenzhou and Istanbul) were assessed by a single experienced liver pathologist, who was blinded to the laboratory’s and participants’ clinical data. The stage of LF on histology was quantified using Brunt’s criteria as follows [13]: stage 1, zone 3 perisinusoidal fibrosis; stage 2, as for stage 1 with portal fibrosis; stage 3, as for stage 2 with bridging fibrosis; and stage 4, cirrhosis. By study design, patients with cirrhosis were excluded because such patients have low skeletal muscle mass and lower serum creatinine levels, making accurate

6

Page 6 of 24

estimation of eGFR more difficult in that patient group. The remaining patients were stratified into two groups according to their LF histological stage: no LF (stage F0) and LF (stages F1–F3). In addition, 154 of the 214 (72%) patients were also assessed by vibration-controlled TE (FibroScan®, Echosens, Paris, France). This technique uses a probe that generates a pressure wave and an ultrasound wave: propagation of the pressure wave depends on the stiffness of liver tissue, whereas propagation of the ultrasound wave is largely affected by the amount of adipose tissue. Patients were predicted to have significant LF if they had an LSM ≥ 8.0 kPa as measured by M-probe ultrasound, shown to be equivalent

oo f

to Kleiner stage ≥ F2 on liver histology [10, 14]. Clinical and laboratory data

pr

Anthropometric and clinical parameters and comorbidities were recorded for all participants. Body mass index (BMI) was calculated as weight (in kg) divided by height (in m2). Blood pressure (BP) was measured

e-

using an automated sphygmomanometer with the subject sitting in a quiet environment for at least 10 min.

Pr

Hypertension was defined as BP ≥ 130/85 mmHg or use of any antihypertensive drugs. Diabetes was defined as a fasting plasma glucose level ≥ 7.0 mmol/L or HbA1c ≥ 6.5%, or use of antihyperglycaemic agents. Hyperuricaemia was defined as serum uric acid levels > 420 μmol/L in men and > 360 μmol/L in

al

women, or use of allopurinol. Dyslipidaemia was defined by any of the following criteria: total cholesterol

Jo ur n

> 5.17 mmol/L; triglycerides (TG) > 1.70 mmol/L; high-density lipoprotein (HDL) cholesterol < 1.0 mmol/L in women and < 1.3 mmol/L in men; and low-density lipoprotein (LDL) cholesterol ≥ 3.4 mmol/L, or use of any lipid-lowering drugs. Homoeostasis model assessment of insulin resistance (HOMA-IR) score was calculated using fasting glucose and insulin measurements as follows: [fasting insulin (mU/mL) × fasting glucose (mmol/L)/22.5]. Biochemical laboratory parameters included the measurement of albumin, alanine aminotransaminase (ALT), aspartate aminotransferase (AST), glucose, blood urea nitrogen, creatinine, uric acid, lipids, and white blood cell, red blood cell and platelet counts. All of these parameters were measured using an automated immunochemical analyzer (AxSYM, Abbott Laboratories, Chicago, IL, USA) and standard laboratory methods. EKD was defined as the presence of microalbuminuria (urinary albumin-to-creatinine ratio 30–300 mg/g) with an eGFR value ≥ 60 mL/min/1.73 m2, as estimated by the Chronic Kidney Disease Epidemiology Collaboration (CKD–EPI) equation [15]. 7

Page 7 of 24

Statistical analysis Data are presented as means ± standard deviation (SD) or as frequencies (n, %). Baseline clinical and biochemical characteristics of the study population were compared using one-way analysis of variance (ANOVA) for continuous variables and χ2 testing for categorical variables. The association between LF assessed by either histology or TE and the presence of EKD in patients with NAFLD was tested using an unadjusted logistic regression model, and also two multivariable logistic regression models with prespecified adjustments for (i) age, gender, ethnicity, waist circumference and hip circumference, and

oo f

further adjustments for (ii) hyperuricaemia, dyslipidaemia, hypertension, diabetes and HOMA-IR score. In addition, to identify any possible predictors of EKD, two logistic regression models were constructed to

pr

determine the probability of each patient with NAFLD of having EKD as follows: (i) the LF identified on histology (LFH–EKD) model; and (ii) the LF identified by TE (LSM–EKD) model. Both models used the

e-

following equation: probability = exp(c)/[1 + exp(c)], where c is 1.63 × LF histology (stages F1–F3 = 1, stage F0 = 0) – 0.84 × gender (male = 1, female = 0) + 0.02 × age – 3.45 in the LFH–EKD model, or c is

Pr

1.43 × LF–TE (LSM ≥ 8 kPa = 1, LSM < 8 kPa = 0) – 1.70 × gender (male = 1, female = 0) + 0.01 × age – 2.207 in the LSM–EKD model. Moreover, discrimination was examined in the context of logistic regression

al

analyses using the area under the receiver operating characteristic curve (AUROC) and MedCalc version

Jo ur n

12.7 software (MedCalc Software, Ostend, Belgium). The optimal cut-off point was identified based on the maximum Youden index (sensitivity + specificity – 1). All statistical tests were two-sided and a P value < 0.05 (two-tailed) was considered statistically significant. These statistical analyses were conducted using SPSS version 22.0 software (IBM Corp., Armonk, NY, USA).

Results

Among the 214 patients with biopsy-proven NAFLD included in the present study, the prevalence of LF and EKD were 15.4% and 61.2%, respectively. Patients in the LF group (as assessed by liver biopsy) had higher adiposity measures (BMI, waist and hip circumferences) and lower eGFR levels than those in the non-LF group (all P < 0.05). In contrast, no significant differences were found in serum uric acid levels, plasma lipid profiles or HOMA-IR scores between these two groups of patients (Table S1; see supplementary materials associated with this article online). 8

Page 8 of 24

Table I presents the main clinical and biochemical characteristics of NAFLD patients stratified by EKD status: those with EKD were more likely to be older, centrally obese and hypertensive, with lower eGFR levels and higher histological stages of LF (P < 0.05). Mean values of systolic and diastolic BP as well as HOMA-IR score were higher in patients with EKD compared with those without EKD (all P < 0.05). Those with EKD also had greater prevalences of diabetes and hypertension than those without EKD. To verify the relationship between LF and EKD, the prevalence of EKD was evaluated in our NAFLD participants, and proved to be higher in the LF vs non-LF group (22.14% vs 4.82%, respectively; P < 0.001).

oo f

In particular, histology-measured LF was significantly associated with the presence of EKD [adjusted odds ratio (OR): 4.33, 95% confidence interval (CI): 1.25–15.03; P = 0.021] even after adjusting for age, gender, ethnicity, waist circumference, hip circumference, hypertension, dyslipidaemia, hyperuricaemia, diabetes

pr

and HOMA-IR score (Table II).

e-

In addition, 72% of patients (n = 154) also underwent TE examination. The prevalence of TE-measured significant LF (defined as LSM ≥ 8.0 kPa and stage ≥ F2 on histology) was lower than that measured by

Pr

liver biopsy (40.90% vs 61.21%, respectively). Moreover, TE-measured LF demonstrated good diagnostic performance for predicting LF histological stage (stages F1–F3; AUROC: 0.75, 95% CI: 0.68–0.82; Fig.

al

2A), and the association between TE-measured LF and EKD was consistent with that observed with liver

Jo ur n

biopsy results. Notably, patients with LSM ≥ 8.0 kPa had a greater prevalence of EKD (23.81%) compared with 6.59% in those with LSM < 8.0 kPa (P < 0.05; Table III). As shown in Table II and similar to the link between histology-measured LF and EKD, the association between TE-measured LF and risk of EKD remained statistically significant (adjusted OR: 4.73, 95% CI: 1.19–28.48; P = 0.03) even after adjustment for age, gender, ethnicity, waist circumference, hip circumference, hypertension, dyslipidaemia, hyperuricaemia, diabetes and HOMA-IR score. To further examine the diagnostic accuracy of LF measured on histology and LF measured by TE for identifying EKD, two further models were constructed—the LFH–EKD model and the LSM–EKD model—as already described in the statistical analysis section above and using three variables (gender, age and either LF stage or LSM score) that are widely available in clinical practice. As depicted in Fig. 2B, 2C, the LSM–EKD model accurately identified the presence of EKD (AUROC: 0.80, 95% CI: 0.72–0.89; P < 9

Page 9 of 24

0.05) and was slightly better in identifying EKD compared with the similar LFH–EKD model. When an optimal cut-off point of 0.11 was used (as per the best Youden index), the associated sensitivity and specificity for the LSM–EKD model was 90.5% and 62.4%, respectively.

Discussion The novel findings of the present cross-sectional study are: (i) the severity of LF, whether assessed by

oo f

histology or by TE, was significantly associated with an increased risk of prevalent EKD (defined as the presence of microalbuminuria with an eGFR value ≥ 60 mL/min/1.73 m2) even after adjusting for age, gender, ethnicity, adiposity measures, hypertension and other established renal risk factors; (ii) the

pr

significant association between TE-measured LF and risk of EKD was consistent with histological data

e-

obtained by liver biopsy; and (iii) both LFH–EKD and LSM–EKD models demonstrated good diagnostic performance for accurately identifying EKD. These results are considered clinically important because TE,

who are at high risk of having EKD.

Pr

a reliable, non-invasive method for staging LF, also appears to accurately identify those NAFLD patients

al

These present findings expand the results of several cross-sectional Asian studies recently published by

Jo ur n

other investigators. In a cohort of 1763 Chinese patients with type 2 diabetes in Hong Kong, Yeung et al. [14] reported that advanced fibrosis assessed by TE was independently associated with an increased risk of albuminuria ≥ 30 mg/g (adjusted OR: 1.52, 95% CI: 1.02–2.28; P = 0.039). Likewise, another study of 1439 Chinese patients reported that increased LSM assessed by TE was a potential indicator of CKD (defined as an eGFR ≤ 60 mL/min/1.73 m2) in patients with ultrasound-detected NAFLD [12]. However, it should be noted that neither of these studies used liver biopsy for diagnosing and staging NAFLD, nor did these two studies examine the association between LF severity and risk of EKD. Moreover, our present study found that the significant association between the severity of LF (whether assessed by histology or TE) and risk of EKD persisted even after adjusting for age, gender, adiposity measures, hypertension, HOMA-IR score and other established risk factors for CKD. In fact, many of the risk factors for NAFLD have the potential to influence the development of EKD. First, 10

Page 10 of 24

our present results found that patients with NAFLD and EKD were more insulin-resistant than those without EKD. It is well known that insulin resistance plays a pivotal role in the development and progression of kidney disease by worsening renal haemodynamics [3, 16, 17], which can lead to endothelial dysfunction, increased vascular permeability and sustained glomerular hyperfiltration that can ultimately result in abnormal urinary albumin excretion. Insulin resistance can also alter renal cellular metabolism and stimulate mesangial hyperplasia [18]. Furthermore, against the background of increased risk for cardiovascular disease commonly seen in patients with NAFLD, the potential for endothelial dysfunction

oo f

and renal vascular damage may also become more prominent [19, 20]. Moreover, activation of the nuclear factor (NF)-κB pathway in non-alcoholic steatohepatitis (NASH) has been shown to increase transcription of a variety of proinflammatory genes that might further amplify systemic chronic inflammation, and this

pr

mechanism could play a pathogenetic role in the development of CKD [21]. Finally, NASH with varying levels of LF exacerbates systemic and hepatic insulin resistance, triggers atherogenic dyslipidaemia, and

e-

releases a variety of procoagulant, pro-oxidant and profibrogenic mediators that may, in turn, promote the

Pr

development and progression of CKD [7, 20, 22, 23].

Thus, our study has shown that assessment of LSM is useful for accurately identifying those patients with

al

NAFLD at particular risk of EKD. Also, TE has the advantage of providing quantitative measurements of liver stiffness, and this reproducible methodology has been validated as a good proxy for LF in various

Jo ur n

different patient populations [24]. Although liver biopsy is considered the ‘gold standard’ for staging LF, such an invasive procedure has the potential risk of acute complications and is not favoured by patients [25]. As a consequence, non-invasive TE is now becoming a widely accepted technology for use in both clinical practice and research [24, 26, 27]. Regarding the cut-off used in our study, it was found that the LSM score had appropriate discriminatory power for predicting LF histological stage. In addition, the association between TE-measured LF and risk of EKD was consistent with data obtained by liver biopsy. By combining LSM with age and gender, our novel LSM–EKD model also demonstrated good performance in identifying EKD. This finding is clinically relevant as the LSM–EKD model could also be used by clinicians to screen for risk of CKD in patients with NAFLD. 11

Page 11 of 24

Our study has several important limitations. First, its cross-sectional design does not allow temporal and causal relationships to be established between LF severity and presence of EKD. Second, only a single measurement of both albuminuria and eGFR levels was available in this study. Third, there was no available information about the specific renal pathology associated with NAFLD in our participants and, in addition, any patients treated with potentially nephrotoxic drugs were excluded. However, those treated with antihyperglycaemic or antihypertensive drugs that may have been potentially nephroprotective were included (although use of those drugs was adjusted for in our multivariable logistic regression models). Finally, there may have been a selection bias, as a considerable number of subjects were excluded due to

oo f

our stringent exclusion criteria.

Thus, further research is now needed to test the associations observed here between LF severity and risk of

pr

EKD in other ethnic cohorts, and to better investigate the incidence rates of progression of EKD to

e-

macroalbuminuria, impaired eGFR and end-stage renal disease in patients with NAFLD. In conclusion, the results of the present study suggest a significant positive association between LF severity,

Pr

as assessed by either invasive liver biopsy or non-invasively by FibroScan, and the presence of EKD in patients with non-cirrhotic biopsy-proven NAFLD that is also independent of the usual renal risk factors,

al

such as age, gender, ethnicity, measures of adiposity, hypertension, dyslipidaemia, hyperuricaemia,

Jo ur n

diabetes and HOMA-IR score. In addition, our results suggest that the evaluation of LF by either histology or TE may be of value for accurately identifying those NAFLD patients at increased risk of EKD, and that identifying such NAFLD patients might also assist in the referral of these patients to tertiary-care centres for more careful surveillance (as proposed in Fig. 3). Appendix supplementary material Supplementary

materials

(Table

S1)

associated

with

this

article

can

be

found

at

http://www.scincedirect.com at doi . . . Funding sources This work was supported by grants from the National Natural Science Foundation of China (81500665), High Level Creative Talents from Department of Public Health in Zhejiang Province, Project of New 12

Page 12 of 24

Century 551 Talent Nurturing in Wenzhou, and General Program of Science and Technology Development Foundation of Nanjing Medical University (2017NJMU168). G.T. is supported in part by grants from the University School of Medicine of Verona, Italy. C.D.B. is supported in part by the Southampton NIHR Biomedical Research Centre, UK (IS-BRC-20004). Authors’ contributions Dan-Qin Sun and Ming-Hua Zheng designed the study, prepared the figures, reviewed the results, interpreted data and wrote the manuscript. Yusuf Yilmaz, Haluk Tarik Kani, Fang-Zhou Ye, Kenneth I.

oo f

Zheng and Yong-Ping Chen collected data. Jian-Rong Yang contributed reagents and performed the experiments. Hao-Yang Zhang and Wei-Jie Yuan performed the statistical analyses. Christopher D. Byrne,

pr

Giovanni Targher and Kenneth I. Zheng proofread the manuscript. All authors contributed important

e-

intellectual content to the manuscript and approved its submission for publication.

Pr

Disclosure statement

None of the authors has any potential conflicts of interest to disclose.

Jo ur n

20–23 September 2018.

al

Some of these data were presented as a poster at the EASL NAFLD Summit, held in Geneva, Switzerland,

Acknowledgments

We thank Xiao-Yan Pan and Chen-Fei Zheng, who contributed by their patient referrals; Pei-Wu Zhu who performed partial experiments; Xiao-Dong Wang who analyzed the pathological biopsies; and also Professor Ji-Min Liu, liver pathologist at McMaster University in Ontario, Canada, who was responsible for quality control of the pathology data for the Wenzhou cohort.

13

Page 13 of 24

References

[7] [8]

[9] [10]

[11] [12] [13] [14]

[15] [16] [17] [18]

[19] [20] [21] [22] [23]

[24]

oo f

[6]

pr

[5]

e-

[4]

Pr

[3]

al

[2]

Targher G, Chonchol MB, Byrne CD. CKD and nonalcoholic fatty liver disease. Am J Kidney Dis 2014;64:63852. Paik J, Golabi P, Younoszai Z, Mishra A, Trimble G, Younossi ZM. Chronic kidney disease is independently associated with increased mortality in patients with nonalcoholic fatty liver disease. Liver Int 2019;39:34252. Yilmaz Y, Alahdab YO, Yonal O, Kurt R, Kedrah AE, Celikel CA, et al. Microalbuminuria in nondiabetic patients with nonalcoholic fatty liver disease: association with liver fibrosis. Metabolism 2010;59:1327-30. Ramspek CL, de Jong Y, Dekker FW, van Diepen M. Towards the best kidney failure prediction tool: a systematic review and selection aid. Nephrol Dial Transplant 2019 pii: gfz018. Crews DC, Bello AK, Saadi G, for the World Kidney Day Steering C. Burden, Access, and Disparities in Kidney Disease. Am J Nephrol 2019;49:254-61. Coresh J, Heerspink HJL, Sang Y, Matsushita K, Arnlov J, Astor BC, et al. Change in albuminuria and subsequent risk of end-stage kidney disease: an individual participant-level consortium meta-analysis of observational studies. Lancet Diabetes Endocrinol 2019;7:115-27. Targher G, Byrne CD. Non-alcoholic fatty liver disease: an emerging driving force in chronic kidney disease. Nat Rev Nephrol 2017;13:297-310. Mantovani A, Zaza G, Byrne CD, Lonardo A, Zoppini G, Bonora E, et al. Nonalcoholic fatty liver disease increases risk of incident chronic kidney disease: A systematic review and meta-analysis. Metabolism 2018;79:64-76. Chinnadurai R, Ritchie J, Green D, Kalra PA. Non-alcoholic fatty liver disease and clinical outcomes in chronic kidney disease. Nephrol Dial Transplant 2019;34:449-57. Roulot D, Costes JL, Buyck JF, Warzocha U, Gambier N, Czernichow S, et al. Transient elastography as a screening tool for liver fibrosis and cirrhosis in a community-based population aged over 45 years. Gut 2011;60:977-84. Castera L, Friedrich-Rust M, Loomba R. Noninvasive assessment of liver disease in patients with nonalcoholic fatty liver disease. Gastroenterol 2019;156:1264-81 e4. Qin S, Wang S, Wang X, Wang J. Liver stiffness assessed by transient elastography as a potential indicator of chronic kidney disease in patients with nonalcoholic fatty liver disease. J Clin Lab Anal 2019;33:e22657. Brunt EM, Janney CG, Di Bisceglie AM, Neuschwander-Tetri BA, Bacon BR. Nonalcoholic steatohepatitis: a proposal for grading and staging the histological lesions. Am J Gastroenterol 1999;94:2467-74. Yeung MW, Wong GL, Choi KC, Luk AO, Kwok R, Shu SS, et al. Advanced liver fibrosis but not steatosis is independently associated with albuminuria in Chinese patients with type 2 diabetes. J Hepatol 2017;68:14756. Levey AS, de Jong PE, Coresh J, El Nahas M, Astor BC, Matsushita K, et al. The definition, classification, and prognosis of chronic kidney disease: a KDIGO Controversies Conference report. Kidney Int 2011;80:17-28. Spoto B, Pisano A, Zoccali C. Insulin resistance in chronic kidney disease: a systematic review. Am J Physiol Renal Physiol 2016;311:F1087-F108. Kobayashi H, Tokudome G, Hara Y, Sugano N, Endo S, Suetsugu Y, et al. Insulin resistance is a risk factor for the progression of chronic kidney disease. Clin Nephrol 2009;71:643-51. Parvanova AI, Trevisan R, Iliev IP, Dimitrov BD, Vedovato M, Tiengo A, et al. Insulin resistance and microalbuminuria: a cross-sectional, case-control study of 158 patients with type 2 diabetes and different degrees of urinary albumin excretion. Diabetes 2006;55:1456-62. Kronenberg F. Emerging risk factors and markers of chronic kidney disease progression. Nat Rev Nephrol 2009;5:677-89. Targher G, Lonardo A, Byrne CD. Nonalcoholic fatty liver disease and chronic vascular complications of diabetes mellitus. Nat Rev Endocrinol 2018;14:99-114. Willy JA, Young SK, Stevens JL, Masuoka HC, Wek RC. CHOP links endoplasmic reticulum stress to NF-kappaB activation in the pathogenesis of nonalcoholic steatohepatitis. Mol Biol Cell 2015;26:2190-204. Ix JH, Sharma K. Mechanisms linking obesity, chronic kidney disease, and fatty liver disease: the roles of fetuin-A, adiponectin, and AMPK. J Am Soc Nephrol 2010;21:406-12. Oberg BP, McMenamin E, Lucas FL, McMonagle E, Morrow J, Ikizler TA, et al. Increased prevalence of oxidant stress and inflammation in patients with moderate to severe chronic kidney disease. Kidney Int 2004;65:1009-16. Chow JC, Wong GL, Chan AW, Shu SS, Chan CK, Leung JK, et al. Repeating measurements by transient

Jo ur n

[1]

14

Page 14 of 24

oo f pr ePr

[27]

al

[26]

Jo ur n

[25]

elastography in non-alcoholic fatty liver disease patients with high liver stiffness. J Gastroenterol Hepatol 2019;34:241-8. Araujo AR, Rosso N, Bedogni G, Tiribelli C, Bellentani S. Global epidemiology of non-alcoholic fatty liver disease/non-alcoholic steatohepatitis: What we need in the future. Liv Int 2018;38 Suppl 1:47-51. Dong H, Xu C, Zhou W, Liao Y, Cao J, Li Z, et al. The combination of 5 serum markers compared to FibroScan to predict significant liver fibrosis in patients with chronic hepatitis B virus. Clin Chim Acta 2018;483:14550. Siddiqui MS, Vuppalanchi R, Van Natta ML, Hallinan E, Kowdley KV, Abdelmalek M, et al. VibrationControlled Transient Elastography to Assess Fibrosis and Steatosis in Patients With Nonalcoholic Fatty Liver Disease. Clin Gastroenterol Hepatol 2019;17:156-63 e2.

15

Page 15 of 24

FIGURE LEGENDS Fig. 1. Study flow diagram: 742 biopsied participants were initially enrolled, but 528 did not meet the inclusion criteria and were excluded, leaving 214 non-cirrhotic patients with biopsy-proven non-alcoholic fatty liver disease (NAFLD) who were ultimately included in the study (154 patients also underwent transient elastography). eGFR: estimated glomerular filtration rate. Fig. 2. Area under the receiver operating characteristic curve (AUROC) analyses of the diagnostic performance of liver stiffness measurement (LSM) scores in identifying (A) liver fibrosis on histology

oo f

(LFH), and models of (B) early kidney dysfunction (LSM–EKD) and (C) LFH–EKD in patients with biopsy-proven NAFLD.

pr

Fig. 3. Proposed algorithm for management of early kidney dysfunction (EKD) in patients with nonalcoholic fatty liver disease (NAFLD). LSM: liver stiffness measurement; ACR: albumin-to-creatinine

Jo ur n

al

Pr

e-

ratio; eGFR: estimated glomerular filtration rate.

16

Page 16 of 24

Jo ur n

al

Pr

e-

pr

oo f

Figr-1

17

Page 17 of 24

Jo ur n

al

Pr

e-

pr

oo f

Figr-2

18

Page 18 of 24

oo f Jo ur n

al

Pr

e-

pr

Figr-3

19

Page 19 of 24

Table I Clinical and biochemical characteristics of patients (n = 214) with biopsy-proven non-alcoholic fatty liver disease (NAFLD) stratified by early kidney dysfunction (EKD)

Demographic parameters Age (years)

EKD (n = 33)

P

41.54 ± 13.32

48.21 ± 10.84

0.007

128 (70.71) 27.82 ± 5.29 101.29 ± 8.67 93.44 ± 11.47

14 (42.42) 30.51 ± 6.57 105.08 ± 14.05 99.06 ± 16.34

0.002 0.010 0.043 0.019 0.071

44 (24.31) 137 (75.69)

13 (39.39) 20 (60.61)

124.23 ± 18.14 78.46 ± 10.81 36 (19.89) 51 (28.18)

134.13 ± 15.44 83.33 ± 10.33 18 (54.55) 21 (63.64)

0.005 0.022 < 0.001 < 0.001

6.42 ± 1.23 6.32 ± 3.18 73.66 ± 17.84 390.27 ± 101.70 2.50 ± 1.51 5.58 ±1.36 2.94 ± 1.19 1.02 ± 0.22 46.53 ± 3.46 91.41 ± 19.20

0.373 0.127 0.017 0.496 0.447 0.076 0.616 0.540 0.989 < 0.001 < 0.001

Pr

e-

pr

6.15 ± 1.51 5.67 ± 1.67 66.96 ± 14.03 372.46 ± 145.17 2.30 ± 1.35 5.06 ± 1.34 3.04 ± 1.01 1.06 ± 0.26 46.52 ± 3.76 109.53 ± 16.90 160 (88.40) 21 (11.60) 4.87 ± 6.32

16 (48.48) 17 (51.52) 7.81 ± 10.50

79 (43.65) 64 (35.36) 22 (12.15) 16 (8.84)

4 (12.12) 18 (54.55) 5 (15.15) 6 (18.18)

al

Jo ur n

Male gender, n (%) Body mass index (kg/m2) Hip circumference (cm) Waist circumference (cm) Ethnicity White, n (%) Yellow, n (%) Clinical parameters Systolic BP (mmHg) Diastolic BP (mmHg) Hypertension, n (%) Type 2 diabetes, n (%) Laboratory parameters Haemoglobin A1c (%) Fasting glucose (mmol/L) Creatinine (μmol/L) Uric acid (μmol/L) Triglycerides (mmol/L） Total cholesterol (mmol/L) LDL cholesterol (mmol/L) HDL cholesterol (mmol/L) Albumin (g/L) eGFR (mL/min/1.73 m2) eGFR category, n (%) ≥ 90 mL/min/1.73 m2 < 90 mL/min/1.73 m2 HOMA-IR score LF stage on biopsy, n (%) F0 F1 F2 F3

No EKD (n = 181)

oo f

Variable

0.034 0.006

Data are presented as means ± standard deviation (SD) or n (%); BP: blood pressure; LDL/HDL: low-density/high-density lipoprotein; eGFR: estimated glomerular filtration rate; HOMA-IR: homoeostasis model assessment of insulin resistance; LF: liver fibrosis

20

Page 20 of 24

Table II Association between early kidney dysfunction and liver fibrosis assessed by either biopsy or transient elastography (FibroScan®) Liver biopsy (n = 214) Unadjusted Model 1 Adjusted Model 2 Adjusted Model 3 FibroScan (n = 154) Unadjusted Model 1 Adjusted Model 2 Adjusted Model 3

B

Wald

OR

95% CI

P

1.73 1.56 1.47

9.70 6.91 5.32

5.62 4.55 3.99

1.89–16.63 1.46–14.25 1.17–13.60

0.002 0.009 0.027

1.49 1.25 1.76

8.32 3.87 4.73

4.43 3.50 6.00

1.61–12.17 1.02–12.05 1.23–29.26

0.010 0.047 0.027

oo f

Model 1: univariable logistic regression model; Model 2: including age, gender, ethnicity*, waist circumference, hip circumference; Model 3: including Model 2 covariates plus hyperuricaemia*, hyperlipidaemia*, hypertension*, diabetes*, HOMA-IR;

Jo ur n

al

Pr

e-

pr

* Included as dichotomous values

21

Page 21 of 24

Jo ur n

al

Pr

e-

pr

oo f

Table III Clinical and biochemical characteristics of patients (n = 154) with non-alcoholic fatty liver disease (NAFLD) stratified by liver fibrosis (LF) assessed by transient elastography (FibroScan) Variable No LF Significant LF P (n = 91) (n = 63)* Demographic parameters Age (years) 41.70 ± 12.57 42.81 ± 13.27 0.600 Male gender, n (%) 64 (70.33) 38 (60.32) 0.196 Body mass index (kg/m2) 25.88 ± 3.64 31.07 ± 5.67 < 0.001 Hip circumference (cm) 98.47 ± 7.89 106.48 ± 11.62 < 0.001 Waist circumference (cm) 89.09 ± 8.81 101.94 ± 14.54 < 0.001 Ethnicity 0.001 White, n (%) 8 (8.79) 39 (61.90) Yellow, n (%) 83 (91.21) 24 (38.10) Clinical parameters Systolic BP (mmHg) 123.17 ± 15.40 125.90 ± 20.61 0.355 Diastolic BP (mmHg) 77.87 ± 9.59 82.63 ± 11.31 0.006 Hypertension, n (%) 16 (17.58) 19 (30.16) 0.067 Type 2 diabetes, n (%) 23 (25.27) 28 (44.44) 0.013 Early kidney dysfunction, n (%) 6 (6.59) 15 (23.81) 0.002 Laboratory parameters Haemoglobin A1c (%) 5.99 ± 1.40 6.17 ± 1.14 0.455 Fasting glucose (mmol/L) 5.67 ± 1.64 5.87 ± 2.28 0.552 Creatinine (μmol/L) 68.69 ± 12.52 68.38 ± 13.66 0.887 Uric acid (μmol/L) 394.14 ± 94.90 412.31 ± 113.28 0.306 Triglycerides (mmol/L) 2.30 ± 1.21 2.59 ± 1.76 0.228 Total cholesterol (mmol/L) 4.84 ± 1.19 5.39 ± 1.38 0.009 HDL cholesterol (mmol/L) 1.03 ± 0.25 1.07 ± 0.31 0.093 LDL cholesterol (mmol/L) 2.94 ± 1.00 3.27 ± 1.02 0.052 Albumin (g/L) 46.85 ± 3.88 46.81 ± 3.60 0.902 eGFR category, n (%) 0.033 ≥ 90 mL/min/1.73 m2 80 (87.91) 47 (74.60) < 90 mL/min/1.73 m2 11 (12.09) 16 (25.40) HOMA-IR score 4.61 ± 7.25 6.43 ± 5.86 0.117 FibroScan parameters CAP (dB/m) 303.05 ± 40.51 325.51 ± 38.43 0.001 LSM (kPa) 5.50 ± 1.10 11.82 ± 3.50 < 0.001 Liver histology Steatosis, n (%) < 0.001 1 34 (37.36) 9 (14.29) 2 36 (39.56) 23 (36.51) 3 12 (13.19) 29 (46.03) Hepatocyte ballooning, n (%) < 0.001 0 35 (38.46) 2 (3.17) 1 51 (56.04) 34 (53.97) 2 5 (5.49) 27 (42.86) Lobular inflammation, n (%) < 0.001 0 13 (14.29) 3 (4.76) 1 63 (69.23) 28 (44.44) 2 14 (15.38) 22 (34.92) 3 1 (1.09) 10 (15.87) Fibrosis, n (%) < 0.001 0 49 (53.85) 8 (12.70) 1 35 (38.46) 25 (39.68) 2 5 (5.49) 15 (23.81) 3 2 (2.20) 15 (23.81)

Data are presented as means ± standard deviation (SD) or n (%); * predicted to have LF if liver stiffness measurement (LSM) score ≥ 8.0 kPa;

22

Page 22 of 24

Jo ur n

al

Pr

e-

pr

oo f

BP: blood pressure; HDL/LDL: high-density/low-density lipoprotein; eGFR: estimated glomerular filtration rate; HOMA-IR: homoeostasis model assessment of insulin resistance; CAP: controlled attenuation parameter

23

Page 23 of 24

Jo ur n

al

Pr

e-

pr

oo f

Table S1 Clinical and biochemical characteristics of patients (n = 214) with non-alcoholic fatty liver disease (NAFLD) stratified by liver fibrosis (LF) assessed by liver biopsy Variable No LF LF P (n = 83) (n = 131) Demographic parameters Age (years) 41.12 ± 12.87 43.49 ± 13.33 0.201 Male gender, n (%) 60 (75) 82 (62.59) 0.144 Height (cm) 166.82 ± 9.83 165.90 ± 10.30 0.521 Weight (kg) 74.16 ± 12.63 80.41 ± 17.04 0.005 Body mass index (kg/m2) 26.78 ± 5.64 29.16 ± 5.36 0.002 Hip circumference (cm) 98.70 ± 7.73 103.94 ± 10.39 < 0.001 Waist circumference (cm) 89.51 ± 9.37 97.42 ± 13.25 < 0.001 Clinical parameters Systolic BP (mmHg) 126.67 ± 15.91 124.98 ± 19.40 0.512 Diastolic BP (mmHg) 77.84 ± 9.40 80.02 ± 11.66 0.157 Hypertension, n (%) 13 (15.66) 41 (31.30) 0.010 Type 2 diabetes, n (%) 18 (21.69) 54 (41.22) 0.005 Laboratory parameters Haemoglobin A1c (%) 6.14 ± 1.71 6.23 ± 1.28 0.675 Fasting glucose (mmol/L) 5.78 ± 1.85 5.74 ± 2.05 0.926 Creatinine (μmol/L) 66.87 ± 13.89 68.71 ± 15.41 0.380 Uric acid (μmol/L) 398.24 ± 95.60 403.83 ± 103.96 0.703 Triglycerides (mmol/L) 2.32 ± 1.54 2.35 ± 1.24 0.756 Total cholesterol (mmol/L) 5.10 ± 1.35 5.17 ± 1.38 0.868 HDL cholesterol (mmol/L) 1.06 ± 0.27 1.04 ± 0.26 0.705 LDL cholesterol (mmol/L) 3.01 ± 0.96 3.03 ± 1.10 0.894 Albumin (g/L) 45.95 ± 3.58 46.89 ± 3.75 0.072 eGFR (mL/min/1.73 m2) 110.55 ± 15.06 104.33 ± 19.97 0.016 eGFR category, n (%) 0.001 ≥ 90 mL/min/1.73 m2 77 (92.77) 99 (75.57) < 90 mL/min/1.73 m2 6 (7.23) 32 (24.43) Urinary ACR 10.74 ± 26.10 24.45 ± 48.55 0.019 Urinary ACR category, n (%) 0.001 ACR ≥ 30 but < 300 mg/g 4 (4.82) 29 (22.14) ACR < 30 mg/g 79 (95.18) 102 (77.86) HOMA-IR score 4.59 ± 7.44 5.80 ± 6.95 0.240 FibroScan parameters* CAP score 300.70 ± 44.56 318.87 ± 38.00 0.008 LSM score 6.25 ± 2.68 9.17 ± 4.13 < 0.001

Data are presented as means ± standard deviation (SD) or n (%); * only patients who underwent transient elastography (n = 154); BP: blood pressure; HDL/LDL: high-density/low-density lipoprotein; eGFR: estimated glomerular filtration rate; ACR: albumin-to-creatinine ratio; HOMA-IR: homoeostasis model assessment of insulin resistance; CAP: controlled attenuation parameter; LSM: liver stiffness measurement

24

Page 24 of 24