Uraemic toxins and cardiovascular disease across the chronic kidney disease spectrum: An observational study

Uraemic toxins and cardiovascular disease across the chronic kidney disease spectrum: An observational study

Accepted Manuscript Uremic toxins and cardiovascular disease across the chronic kidney disease spectrum: an observational study M. Rossi, BS K. Campbe...

842KB Sizes 1 Downloads 15 Views

Accepted Manuscript Uremic toxins and cardiovascular disease across the chronic kidney disease spectrum: an observational study M. Rossi, BS K. Campbell, PhD D. Johnson, MBBS PhD PSM T. Stanton, MBBS PhD E. Pascoe, MBiostat C. Hawley, MBBS M Med Sci G. Dimeski, PhD B. McWhinney, MS J. Ungerer, MBBS N. Isbel, MBBS PhD PII:

S0939-4753(14)00138-0

DOI:

10.1016/j.numecd.2014.04.006

Reference:

NUMECD 1282

To appear in:

Nutrition, Metabolism and Cardiovascular Diseases

Received Date: 28 January 2014 Revised Date:

12 March 2014

Accepted Date: 8 April 2014

Please cite this article as: Rossi M, Campbell K, Johnson D, Stanton T, Pascoe E, Hawley C, Dimeski G, McWhinney B, Ungerer J, Isbel N, Uremic toxins and cardiovascular disease across the chronic kidney disease spectrum: an observational study, Nutrition, Metabolism and Cardiovascular Diseases (2014), doi: 10.1016/j.numecd.2014.04.006. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. 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.

ACCEPTED MANUSCRIPT TITLE PAGE Title: Uremic toxins and cardiovascular disease across the chronic kidney disease spectrum: an observational study

RI PT

Authors: Rossi M, BS1,3,4 Campbell K, PhD1,4

SC

Johnson D, MBBS PhD PSM1,3,4

Pascoe E, MBiostat1 Hawley C, MBBS M Med Sci1,4 Dimeski G, PhD1,6 McWhinney B, MS 7

Isbel N, MBBS PhD1,4 Affiliations:

TE D

Ungerer J, MBBS7

M AN U

Stanton T, MBBS PhD1,5

Australia

EP

School of Medicine1, and Human Movement Studies2, University of Queensland,

AC C

Translational Research Institute3, Brisbane, Queensland, Australia Department of Nephrology4, Cardiology5 and Chemical Pathology6, Princess Alexandra Hospital, Brisbane, Queensland, Australia Department of Chemical Pathology7, Royal Brisbane and Women’s Hospital, Brisbane, Queensland, Australia

1

ACCEPTED MANUSCRIPT

Corresponding author: Megan Rossi, Department if Nephrology, Level 2, ARTS Building, princess Alexandra Hospital,

Fax: 61 731 765 480; Ph: 61 731 765 080 Email: [email protected]

SC

Corresponding author:

RI PT

Ipswich Road, Woolloongabba, Brisbane, Queensland 4102, Australia.

Megan Rossi,

AC C

EP

TE D

cresyl sulphate; uremic toxins

M AN U

Key words: Chronic kidney disease; cardiovascular disease; indoxyl sulphate; p-

2

ACCEPTED MANUSCRIPT ABSTRACT Background and aims: There is a growing body of evidence supporting the nephrovascular toxicity of indoxyl sulphate (IS) and p-cresyl sulphate (PCS).

course of chronic kidney disease (CKD) is lacking.

RI PT

Nonetheless, a comprehensive description of how these toxins accumulate over the

Methods and results: This cross-sectional observational study included a convenience sample of 327 participants with kidney function categorised as normal, non-dialysis

SC

CKD and end-stage kidney disease (ESKD). Participants underwent measurements of serum total and free IS and PCS and assessment of cardiovascular history and structure

M AN U

(carotid intima-media thickness [cIMT, a measure of arterial stiffness]), and endothelial function (brachial artery reactivity [flow-mediated dilation (BAR-FMD); glyceryl trinitrate (BAR-GTN)]).

Across the CKD spectrum there was a significant increase in both total and free IS

TE D

and PCS and their free fractions, with the highest levels observed in the ESKD population. Within each CKD stage, concentrations of PCS, total and free, were significantly greater than IS (all p<0.01). Both IS and PCS, free and total, were

EP

correlated with BAR-GTN (ranging from r=-0.33 to -0.44) and cIMT (r=-0.19 to -

AC C

0.22), even after adjusting for age (all p<0.01). Further, all toxins were independently associated with the presence of cardiovascular disease (all p<0.02). Conclusion: More advanced stages of CKD are associated with progressive increases in total and free serum IS and PCS, as well as increases in their free fractions. Total and free serum IS and PCS were independently associated with structural and functional markers of cardiovascular disease. Studies of therapeutic interventions targeting these uremic toxins are warranted.

3

ACCEPTED MANUSCRIPT INTRODUCTION

Chronic kidney disease (CKD) remains a major public health burden with one in ten adults in developed countries suffering some level of kidney dysfunction[1]. Despite significant advances in CKD treatment, cardiovascular-associated mortality and

RI PT

morbidity continue to rise. The relentless nature of CKD-associated cardiovascular

disease (CVD) and the limited response to traditional cardiovascular risk management in this population have driven research to explore novel risk factors, such as uremic

SC

toxins.

M AN U

Two uremic toxins in particular, indoxyl sulphate (IS) and p-cresyl sulphate (PCS), have attracted attention over the past decade not only for their pro-inflammatory and pro-oxidative properties but their potential therapeutic amenability.[2] Both IS and PCS have demonstrated the ability to induce endothelial dysfunction through

TE D

inhibition of endothelial proliferation and wound repair at uremic concentrations, [3] along with dose-dependent stimulation of oxidative stress and increased expression of inflammatory cytokines [4, 5] in vitro. In addition, the administration of either IS or

EP

PCS in animal models have illustrated significant increases in renal fibrosis and nephrosclerosis, compared to controls.[6] These findings have been supported by

AC C

observational studies which have shown independent associations between at least one or both of these toxins and all-cause mortality[7, 8] and cardiovascular events[9, 10]. Furthermore, controlled clinical intervention studies have demonstrated slowed CKD progression[11] and improvement in endothelial dysfunction[12] with reduction in IS levels following administration of oral adsorbent therapy.

4

ACCEPTED MANUSCRIPT Both IS and PCS are by-products of dietary protein fermentation by bacteria in the large intestine (collectively known as the gut microbiota), which undergo sulphation and hydroxylation in the liver and intestinal wall, respectively, before entering the circulatory system.[13] The toxins’ hydrophobic nature ensures both are

RI PT

predominantly bound to albumin in the systemic circulation and therefore mainly

depend on active renal excretion. Like many uremic toxins, both are detectable in the non-CKD population. However, the decreased renal excretion of the toxins, along

SC

with the CKD-specific gut dysbiosis,[14] resulting in increased bacterial production of IS and PCS, gives rise to the heightened concentrations observed in the CKD

M AN U

population.

IS and PCS are often referred to collectively given their notable similarities including synthesis, transportation (via albumin binding and organic anion transporters) and

TE D

mechanism of action (notably nephrovascular damage). Despite this, important differences between these two toxins have recently been identified, including mechanisms of renal clearance and rates of intestinal absorption [15] as well as

EP

differing concentrations across ethnic groups [16]. These differences may be key to

AC C

explaining the lack of consensus between findings in observational studies reporting associations with one toxin, but not the other. Wu et al.[8] demonstrated that PCS, but not IS, was an independent predictor of all-cause mortality in an Asian pre-dialysis population, whereas Melamed et al.[17] found this association with IS but not PCS in an American hemodialysis cohort. Intervention studies have also drawn conflicting conclusions, indicating potential for population-specific benefit of treatment.[18]

5

ACCEPTED MANUSCRIPT Identifying the differences between these toxins may help to explain some of the conflicting findings between studies as well as providing a better understanding for potential mechanisms of action and opportunities for therapeutic manipulation. This paper aims to compare the total and free serum concentrations of IS and PCS, and the

cardiovascular disease (CVD) markers.

SC

METHODS

RI PT

free fractions across the CKD spectrum, as well as exploring their associations with

Study population

M AN U

Baseline data were analysed from a convenience sample of participants enrolled in one of three trials at a single tertiary centre’s renal outpatient department. All patients from each trial were included in the study, except for when serum samples were not available for analyses (Study 1 n=0, Study 2 n=5, Study 3 n=86). There was no

TE D

systematic exclusion of patients. The three trials included participants from different CKD stages (Study 1: normal kidney function (control), Study 2: moderate kidney dysfunction and Study 3: end stage kidney disease (ESKD) including peritoneal- and

EP

hemo-dialysis). Participants from Study 1 (n=42) were prospective living kidney

AC C

donors. Studies 2 and 3 were participants enrolled in randomised controlled trials of cardiovascular risk modification (n=171 & n=114, respectively). Inclusion into Study 2 required an estimated glomerular filtration rate (eGFR) between 25-75 ml/min/1.73m2 and at least one of the following modifiable risk factors: blood pressure not at target (>130 / 80, or >120/75 for those with diabetes or proteinuria >1g/ 24 hours), overweight (classified as a BMI of 25 or above), poor diabetic control (HbA1c >7%), or hyperlipidemia (defined as low density lipoprotein (LDL) <2.5, or <2.0 in those with diabetes or existing coronary heart disease). Study 3 included

6

ACCEPTED MANUSCRIPT participants with an eGFR below 30 ml/min/1.73m2 or receiving maintenance dialysis.[19] Exclusion criteria for Study 2 were a previous kidney transplant or anticipated dialysis within 6 months. The current study was approved by the

RI PT

institution’s Human Research Ethics Committee (HREC/12/QPAH/216).

Laboratory assessment

Venous blood was collected from all patients following an overnight fast. Serum

SC

creatinine, cholesterol, low-density lipoproteins (LDL), triglycerides, high-density lipoproteins (HDL), albumin, phosphate Beckman DxC800 general chemistry

M AN U

analysers (Beckman coulter Brea, CA, USA).and haemoglobin using a Sysmex XE5000 haematology analyser (Sysmex Cooperation, Kobe, Japan) . The eGFR was calculated using the Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) formula. CKD stage was defined by the KDIGO criteria for CKD diagnosis.[20]

TE D

Uremic toxins

Serum total and free concentrations of both uremic toxins, IS and PCS, were analysed

EP

by ultra-performance liquid chromatography (UPLC) using a fluorescence detection method (Waters Corporation, Milford, MA, USA). This recently validated method

AC C

allowed for low limits of detection (as low as 0.1 µmol/L) as detailed by Pretorius et al.[21] The free fraction of each toxin was defined as a percentage of total concentration (free serum concentration divided by the total concentration multiplied by 100). Samples were run in duplicates and the coefficient of variation (CV) for the assays ranged from 1.8 to 2.9%.

Vascular assessment

7

ACCEPTED MANUSCRIPT Two markers of cardiovascular risk, brachial artery reactivity (BAR) and carotidintima-media thickness (cIMT) were measured. BAR, a non-invasive marker of endothelial dysfunction, was examined using a 12 MHZ ultrasound probe (Philips IE33). Both endothelium-dependent vasodilatation

RI PT

(reactive hyperaemia, flow-mediated dilatation [BAR-FMD]) and endothelium-

independent vasodilatation (after administration of sublingual glycerol trinitrate

[BAR-GTN]) were measured. The BAR-FMD and BAR-GTN were expressed as a

SC

percentage change from resting.

cIMT was used to assess arterial structure. This involved longitudinal images of the

M AN U

extracranial carotid arteries using the 12MHz probe in three imaging planes with automated edge tracking software of the R-wave from the ECG (IMT plug-in, Qlab). The reported cIMT was averaged from six measurements with an intra-observer variation of 0.03 ± 0.02 mm (CoV 5%).

TE D

All measurements were performed according to the American Society of Echocardiography recommendations.

EP

Risk factor assessment

AC C

Smoking history was defined by self report and categorised into two groups: never or previous/current. Blood pressure was recorded as the average of 3 seated blood pressures and was measured pre dialysis (after at least a 2-day break from hemodialysis (HD)) in all HD patients. Presence of CVD was defined as a history of myocardial infarction, heart failure, angina, coronary artery bypass, coronary or carotid angioplasty stenting, stroke, transient ischemic attack, peripheral vascular disease, renovascular disease, abdominal aortic aneurysm repair, or amputation due to ischemia.

8

ACCEPTED MANUSCRIPT

Anthropometry assessment Body weight was measured using Wedderburn SK-VET (Sydney, Australia) calibrated electronic scales correct to 0.1kg. Height was measured at baseline using a

RI PT

wall-mounted stadiometer. Body Mass Index (BMI) was calculated as weight (kg)/height2 (m) and categorised according to the World Health Organisation’s

SC

(WHO) definitions.

Statistical analysis

M AN U

Summary statistics for patients’ characteristics were expressed as mean (standard deviation) for normally distributed continuous data, median (inter-quartile range [IQR]) for skewed continuous data and frequencies (percentages) for categorical data. All continuous variables were assessed for normality and transformed as appropriate.

TE D

Toxin levels below the detection limit were given the value of half the lower limit of detection. Within each CKD stage, serum levels of IS, PCS and protein binding were compared using paired t-test. Analysis of each variable across CKD stages was

EP

performed using analysis of variance (ANOVA), or Kruskal-Wallis non-parametric

AC C

ANOVA where variances were unequal. Post-hoc analyses of pair-wise comparisons were adjusted using Bonferroni’s correction. Sensitivity analyses to determine whether there was a significant difference in the serum concentrations of the toxins between the Caucasian and non-Caucasian sub-populations were undertaken using independent samples t-tests. Associations between the uremic toxins, kidney function (eGFR) and cardiovascular markers (cIMT and BAR) were assessed using Pearson’s correlation coefficients. Multivariable linear and logistic regression analyses were performed to determine whether uremic toxins were independently associated with

9

ACCEPTED MANUSCRIPT cardiovascular markers and the presence of CVD. Covariates were chosen based on a priori set of traditional risk factors reported in the CKD literature, which included age, gender, diabetes, statin therapy, BMI, smoking history, serum albumin and systolic blood pressure .[22] Variance inflation factor was used to assess collinearity and final

RI PT

models were assessed for goodness-of-fit using Hosmer-Lemeshow’s test. The null hypothesis was rejected at the 0.05 level. All of the statistical analyses were

SC

performed using Stata (version 12, 2012, Statacorp, College Station, TX, USA).

Patient characteristics

M AN U

RESULTS

Three hundred and twenty seven participants, with kidney function ranging from normal to end stage (including both peritoneal dialysis [PD] and HD), were included in the analysis. Participant demographics are described in Table 1. Fifty percent (165

Uremic toxins

TE D

of the 327 participants) had evidence of CV disease.

EP

Across the CKD spectrum there was a significant increase in both total and free IS,

AC C

with the highest levels observed in the dialysis population (Figure 1). Similarly, total and free PCS increased across the spectrum, although serum levels peaked earlier in the CKD stage 5 pre-dialysis population (Figure 2). Within each CKD stage, total and free concentrations of PCS were significantly greater than those for total and free IS across controls and CKD stages 2 to 5 (all p<0.002 and all p<0.01, respectively). There were no significant differences between PCS and IS concentrations in either PD or HD patients.

10

ACCEPTED MANUSCRIPT Both toxins (total and free) were moderately correlated with eGFR (ranging from r=0.57 to -0.70, all p<0.001). Strong correlations were observed between the free and total levels of each toxin (IS: r=0.96; PCS: r=0.87) and moderate to high correlations were observed between free and total IS and PCS (ranging from r=0.62 to 0.84) (all

RI PT

p<0.001).

There was no significant difference in the serum concentrations of either toxin

SC

between the Caucasian and non-Caucasian sub-populations by CKD stage (p>0.07).

Protein binding

M AN U

The unbound free fractions of both IS and PCS were increased in the ESKD groups (CKD stage 5, PD and HD) compared to those with normal renal function and CKD stages 2-4 (all p<0.01) (Figure 3). Relative to controls, there was a higher proportion of unbound IS observed in CKD stage 4 (p<0.01) and stage 3 (p<0.001). Within the

TE D

ESKD groups there was only a significant difference between the protein binding for PCS in CKD stage 5 and PD (p<0.03). Within each CKD stage there was a greater

EP

free fraction of IS than PCS (all p<0.001) (refer to Figure 3).

AC C

Uremic toxins, cardiovascular markers and cardiovascular disease Total and free serum IS concentrations had small to moderate correlations with BARGTN (r=-0.42, r=-0.44 respectively) and cIMT (r=0.20, r=0.21) (all p<0.001). Likewise, PCS, total and free, were similarly correlated with both BAR-GTN (r= 0.33, r=-0.39 respectively) and cIMT (r= 0.18, r=0.19) (all p<0.01). These associations remained even after adjusting for a well known risk factors of age, gender, diabetes, statin therapy, BMI, smoking history, serum albumin and systolic

11

ACCEPTED MANUSCRIPT blood pressure (all p<0.01, refer to Supplement Table 1). There was no association between BAR-FMD and either toxin. In this cohort the toxins were both significantly associated with clinically evident CVD after adjusting for traditional risk factors(refer to Table 2). In fact for every

AC C

EP

TE D

M AN U

SC

respectively, in the odds of having CVD.

RI PT

10umol/L increase in free serum IS and PCS there was 47% and 34% increase,

12

ACCEPTED MANUSCRIPT DISCUSSION This is the first study to compare the serum concentration and free fraction of IS and PCS across the CKD spectrum. Overall, the highest concentrations of both toxins were observed in ESKD along with a significant increase in the free fraction. Within

RI PT

each CKD stage in the pre-dialysis population (CKD stages 2-5) serum concentrations of PCS were significantly higher than IS, whereas the free fraction was significantly

lower suggesting different binding capacities. In addition, both toxins were associated

SC

with cardiovascular markers of endothelial dysfunction (BAR-GTN) and atheroma

M AN U

burden (cIMT), and were independently associated with CVD.

IS and PCS both originate from colonic bacterial proteolytic fermentation and are both excreted via renal tubule secretion. Nonetheless, there are key differences in their generation pathways including different amino acids and gut bacterial species

TE D

involved [2] and different tubular transport affinities for their urinary excretion[15]. These factors may contribute to their distinguishable patterns of serum accumulation

EP

observed in this study.

AC C

The ratio of IS to PCS (IS:PCS) was notably lower in this study compared to studies in Asian populations. In fact, across the literature there is a notable difference between the IS:PCS ratio in Asian and Caucasian populations, where the IS:PCS ratio was reported to be three times greater in an Asian population at similar CKD stages.[16] The contrasting IS:PCS ratio may be attributable to the different diets typical of these populations, which is known to influence both the type of substrate available for fermentation (amino acids) as well as the diversity of the gut microbiota[23] and therefore the generation rates of these toxins. In the current study, 14% of the

13

ACCEPTED MANUSCRIPT population were non-Caucasians, yet sensitivity analysis indicated the IS:PCS ratio was no different to the rest of the cohort (p=0.162). Whether this is attributable to the influence of a Western diet or to a type 2 statistical error related to small sample size is uncertain. It should also be noted that the controls did not contain any Asian

RI PT

participants.

The low IS:PCS ratio exemplified in this study offers a potential explanation as to

why studies based in Asia, with high IS:PCS ratio, have demonstrated a benefit for

SC

delaying progression of CKD (measured by reciprocal serum creatinine),[11, 24] by

lowering IS achieved with oral adsorbent therapy, whereas studies based in America,

M AN U

where serum concentrations were less than half, had no clear benefit on renal function [18, 25]. This may suggest that absolute serum concentrations are a more important consideration than relative reductions in serum concentrations when evaluating and comparing the efficacy of therapeutic interventions between different studies.

TE D

The present study was the first to report serum concentrations of free and total levels of IS and PCS across the CKD spectrum using the same validated assay. Therefore the results maybe useful as a starting reference to compare other studies’ concentrations,

EP

although greater participant numbers within each CKD stage are needed to provide a

AC C

valid reference point.

The present study is also the first to illustrate that the free fractions of both toxins, particularly IS, progressively increased as kidney function declined from normal through stages 2-5 CKD to dialysis-dependent end-stage kidney failure. It has been suggested that the free serum concentrations of IS and PCS are better predictors of poor outcomes,[9, 26] possibly reflecting the fact that the free fractions represent the metabolically active components of the toxins. With this in mind, the different

14

ACCEPTED MANUSCRIPT proportions of non-protein-bound IS and PCS across the CKD spectrum may be important for understanding the efficacy of different therapeutic opportunities and for comparing the potential toxicity between IS and PCS. Watanabe et al have previously demonstrated that both IS and PCS share the same serum albumin binding site and

RI PT

have relatively similar binding constants for the high affinity binding site, Sudlow II.[27] However, the binding constant of the low affinity site for PCS was 2.5

fold greater than that for IS. [27] This phenomenon may have accounted for the

SC

findings of higher free fractions of IS relative to PCS with more advanced CKD in the present study, such that at the higher serum concentrations observed in the more

M AN U

advanced stages of CKD, PCS displaced a greater proportion of IS resulting in a higher free fraction of IS. It is possible therefore that IS may play a more predominant

TE D

role in nephrovascular toxicity in the later stages of CKD, as opposed to PCS.

The associations reported in this study between toxin concentration and both surrogate markers of CVD (cIMT and BAR-GTN) and presence of CVD, supports the growing

EP

body of observational evidence for the vascular toxicity of IS and PCS in the CKD population.[7, 28] Recently, investigations into the vascular toxicity of IS and PCS

AC C

have extended beyond the CKD population, with a number of studies demonstrating their predictive power for cardiac events in CVD populations.[29, 30] Furthermore, a cohort of 145 patients with coronary artery disease and preserved kidney function (defined as eGFR>60) demonstrated a similar association between serum IS and cIMT (p=0.004) to that found in this study.[31] Although not definitive, this may suggest the relationship between IS and PCS and CVD is more than just a reflection of kidney function.

15

ACCEPTED MANUSCRIPT This observational study has several limitations. The population was a convenience sample from three parent studies recruited from a single nephrology outpatient department. This led to an uneven distribution of participants within each CKD stage and therefore limited the capacity to explore associations within each CKD stage. The

RI PT

cross-sectional nature of the study limited the ability to infer a causal relationship

between toxins and CVD and therefore the possibility that these toxins in vivo are inert biomarkers of kidney function can not be rejected by these results alone.

SC

Furthermore, both IS and PCS are influenced by dietary intake and, given intake data was not available for this cohort, the generalisability of the results to other

M AN U

populations maybe limited. Studies investigating the association between serum concentrations of IS and PCS and different dietary intakes across ethic groups, as well as intervention studies that manipulate the serum concentrations and measure the

TE D

subsequent effect on CVD, are needed.

In conclusion, more advanced stages of CKD are associated with progressive increases in total and free serum concentrations of IS and PCS, as well as increases in

EP

the free factions of IS and PCS. Total and free serum concentrations of IS and PCS

AC C

were associated with structural and functional markers of cardiovascular disease independent of other traditional risk factors. The key differences between the serum concentrations of IS and PCS, as well as protein binding within and across CKD stages, may help to further our understanding of these toxins and explain some of the conflicting findings in the literature. In addition, these important differences may provide valuable insight when exploring opportunities for therapeutic manipulation. Intervention studies exploring therapies that can effectively lower IS and PCS levels

16

ACCEPTED MANUSCRIPT will be required to determine the role of IS and PCS in the causal pathway of CVD in CKD.

ACKNOWLEDGEMENTS:

RI PT

Erin Howden and Dr Brian A Haluska for their assistance with the arterial stiffness measurements and Dr Omar Kaisar and research nurses for their involvement in data

AC C

EP

TE D

M AN U

SC

collection.

17

ACCEPTED MANUSCRIPT REFERENCES

AC C

EP

TE D

M AN U

SC

RI PT

1. El Nahas AM, Bello AK. Chronic kidney disease: the global challenge. The Lancet 2005;365(9456):331-340. 2. Rossi M, Campbell K, L., Johnson D, W. Indoxyl sulphate and p-cresyl sulphate: therapeutically modifiable nephrovascular toxins. OA Nephrology 2013;1(2):13. 3. Dou L, Bertrand E, Cerini C, et al. The uremic solutes p-cresol and indoxyl sulfate inhibit endothelial proliferation and wound repair. Kidney Int. 2004;65(2):442-51. 4. Motojima M, Hosokawa A, Yamato H, et al. Uremic toxins of organic anions up-regulate PAI-1 expression by induction of NF-kappaB and free radical in proximal tubular cells. Kidney Int. 2003;63(5):1671-80. 5. Watanabe H, Miyamoto Y, Honda D, et al. p-Cresyl sulfate causes renal tubular cell damage by inducing oxidative stress by activation of NADPH oxidase. Kidney Int. 2013;83(4):582-92. 6. Sun CY, Chang SC, Wu MS. Uremic Toxins Induce Kidney Fibrosis by Activating Intrarenal Renin-Angiotensin-Aldosterone System Associated Epithelial-to-Mesenchymal Transition. PLoS One 2012;7(3):e34026. 7. Barreto FC, Barreto DV, Liabeuf S, et al. Serum Indoxyl Sulfate Is Associated with Vascular Disease and Mortality in Chronic Kidney Disease Patients. Clin. J. Am. Soc. Nephrol. 2009;4(10):1551-1558. 8. Wu IW, Hsu KH, Lee CC, et al. p-Cresyl sulphate and indoxyl sulphate predict progression of chronic kidney disease. Nephrol. Dial. Transplant. 2010;26(3):938-947. 9. Wu IW, Hsu KH, Hsu HJ, et al. Serum free p-cresyl sulfate levels predict cardiovascular and all-cause mortality in elderly hemodialysis patients-a prospective cohort study. Nephrol. Dial. Transplant. 2012;27(3):1169-75. 10. Lin CJ, Liu HL, Pan CF, et al. Indoxyl sulfate predicts cardiovascular disease and renal function deterioration in advanced chronic kidney disease. Arch. Med. Res. 2012;43(6):451-6. 11. Owada A, Nakao M, Koike J, et al. Effects of oral adsorbent AST-120 on the progression of chronic renal failure: a randomized controlled study. Kidney Int. Suppl. 1997;63:S188-90. 12. Yu M, Kim YJ, Kang DH. Indoxyl sulfate-induced endothelial dysfunction in patients with chronic kidney disease via an induction of oxidative stress. Clin. J. Am. Soc. Nephrol. 2011;6(1):30-9. 13. Smith EA, Macfarlane GT. Formation of Phenolic and Indolic Compounds by Anaerobic Bacteria in the Human Large Intestine. Microb. Ecol. 1997;33(3):180-8. 14. Vaziri ND, Wong J, Pahl M, et al. Chronic kidney disease alters intestinal microbial flora. Kidney Int. 2013;83(2):308-15. 15. Poesen R, Viaene L, Verbeke K, et al. Renal Clearance and Intestinal Generation of p-Cresyl Sulfate and Indoxyl Sulfate in CKD. Clin. J. Am. Soc. Nephrol. 2013. 16. Lin CJ, Chen HH, Pan CF, et al. p-Cresylsulfate and indoxyl sulfate level at different stages of chronic kidney disease. J. Clin. Lab. Anal. 2011;25(3):191-7. 17. Melamed ML, Plantinga L, Shafi T, et al. Retained organic solutes, patient characteristics and all-cause and cardiovascular mortality in hemodialysis: results from the retained organic solutes and clinical outcomes (ROSCO) investigators. BMC Nephrol. 2013;14(1):134. 18. Schulman G, Berl T, Beck GJ, et al. EPPIC (Evaluating Prevention of Progression In Chronic Kidney Disease): Results from 2 Phase III, Randomized, Placebo-Controlled, Double-Blind Trials of AST-120 in Adults with CKD ASN 2012;A:PO1106. 19. Isbel NM, Haluska B, Johnson DW, et al. Increased targeting of cardiovascular risk factors in patients with chronic kidney disease does not improve atheroma burden or cardiovascular function. Am. Heart J. 2006;151(3):745-53. 20. KDIGO. KDIGO 2012 Clinial Practice Guidelines for the Evaluation and Management of Chronic Kidney Disease. Kidney Int. 2013;3(1). 21. Pretorius C, J., McWhinney B, C., Sipinkoski B, et al. Reference ranges and biological variation of free and total serum indoxyl- and p-cresyl sulphate measured with a rapid UPLC fluorescence detection method. Clin. Chim. Acta 2013;419:122-126. 22. Petchey WG, Hawley CM, Johnson DW, et al. Multimodality vascular imaging in CKD: divergence of risk between measured parameters. Nephrology, dialysis, transplantation : official publication of the European Dialysis and Transplant Association - European Renal Association 2012;27(3):1004-12. 23. Wu GD, Chen J, Hoffmann C, et al. Linking long-term dietary patterns with gut microbial enterotypes. Science 2011;334(6052):105-8.

18

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

24. Niwa T, Nomura T, Sugiyama S, et al. The protein metabolite hypothesis, a model for the progression of renal failure: an oral adsorbent lowers indoxyl sulfate levels in undialyzed uremic patients. Kidney Int. Suppl. 1997;62:S23-8. 25. Schulman G, Agarwal R, Acharya M, et al. A multicenter, randomized, double-blind, placebocontrolled, dose-ranging study of AST-120 (Kremezin) in patients with moderate to severe CKD. Am. J. Kidney Dis. 2006;47(4):565-77. 26. Liabeuf S, Barreto DV, Barreto FC, et al. Free p-cresylsulphate is a predictor of mortality in patients at different stages of chronic kidney disease. Nephrol. Dial. Transplant. 2010;25(4):1183-91. 27. Watanabe H, Noguchi T, Miyamoto Y, et al. Interaction Between Two Sulfate Conjugated Uremic Toxins, p-cresyl Sulfate and Indoxyl Sulfate, During Binding with Human Serum Albumin. Drug Metab. Dispos. 2012. 28. Moradi H, Sica DA, Kalantar-Zadeh K. Cardiovascular burden associated with uremic toxins in patients with chronic kidney disease. Am. J. Nephrol. 2013;38(2):136-48. 29. Shimazu S, Hirashiki A, Okumura T, et al. Association between indoxyl sulfate and cardiac dysfunction and prognosis in patients with dilated cardiomyopathy. Circ. J. 2013;77(2):390-6. 30. Hsu CC, Lu YC, Chiu CA, et al. Levels of indoxyl sulfate are associated with severity of coronary atherosclerosis. Clin. Invest. Med. 2013;36(1):E42-9. 31. Sato B, Yoshikawa D, Ishii H, et al. Indoxyl sulfate, a uremic toxin, and carotid intima-media thickness in patients with coronary artery disease. Int. J. Cardiol. 2013;163(2):214-6.

19

ACCEPTED MANUSCRIPT TABLES Table 1: Demographic and clinical characteristics of study population

CKD stage Controls n=42

Stage 2 n=10

Stage3 n=134

Stage 4 n=37

Age (years)

60 ± 12

53 ± 10

54±10

61±10

64±10

Male (%)

178 (54)

10 (24)

5(50)

81 (60)

22 (59)

White Caucasian (%)

281 (86)

42 (100)

8 (80)

114 (85)

33(92)

Glomerulonephritis

50 (18)

NA

2 (20)

15 (11)

8(22)

Hypertension/vascular

44 (15)

NA

0 (0)

25(19)

5(14)

Diabetic nephropathy

79 (28)

NA

2(20)

39(29)

13 (35)

Cause of Kidney disease

Stage 5 n=17 68±10

PD n=33

HD n=54

61±15

56±14

RI PT

All n=327

12 (71)

19 (58)

29 (54)

15(88)

26(79)

42 (78)

1(6)

7(21)

17(31)

1(6)

8(24)

5(9)

7(41)

4(12)

14(26)

28±5

26±5

28±7

30 ± 7

26±3

34±8

33±6

31±6

Diabetes (%)

110 (34)

0(0)

2(20)

59(44)

20 (54)

7(41)

5(15)

17(31)

Smoking history (%)

185 (57)

16 (38)

4(40)

89 (67)

23(66)

8 (47)

19 (58)

26 (48)

Systolic blood pressure

139 ± 21

125 ±11

126±17

140±21

142±20

146±18

140±20

147±25

eGFR (mL/min/1.73m2)

46±25

93±12

64±4

NA

Cholesterol (mmol/L)

4 .5 ± 1.1

5.1±1.0

4.8±0.6

Low density lipoproteins (mmol/L)

2.5±0.9

3.2±0.9

3.0±0.7

Triglycerides (mmol/L)

1.6 (1.0-2.3)

0.8(0.7-1.3)

Haemoglobin (g/L)

126±17

138±10

Albumin (g/L)

38 ± 4

40 ± 2

37±3

M AN U

Biochemical

SC

BMI (kg/m2)

41±8

23±5

10±3

NA

4.4±1.0

4.5±1.4

4.2±1.1

4.6±1.4

4.2±0.8

2.4±0.9

2.4±1.1

2.1±0.7

2.4±1.0

2.2±0.7

1.5(0.9-1.7)

1.5(1-2.2)

1.9(1.3-2.5)

1.3(1-2.5)

1.8(1.3-2.5)

1.9(1.3-2.6)

140±21

133±15

121±13

114±13

113±12

114±14

38±4

38±4

40±3

36±3

39±3

9(7-13)

16(11-26)

54(39-65)

83(56-136)

116(79-172)

0.3(0.2-0.5)

0.9(0.5-2.2)

8.6(5.4-10.9)

17.0(10.7-26.1)

27.5(13.9-36.9)

Uremic toxins (µmol/L) 13 (7-57)

4(2-7)

6(4-9)

Free IS

0.5 (0.2-9.7)

0.1(0.1-0.1)

0.2(0.2-0.3)

Total PCS

70 (26-128)

17(12-27)

TE D

Total IS

27(16-41)

47(26-79)

99(74-124)

241(179-291)

158(72-201)

155(111-205) 28.1(16.2-40.9)

Free PCS

2.1 (0.9-16.5

0.8(0.5-1.1)

0.8(0.3-1.4)

1.4(0.8-2.1)

4.2(2.4-8.3)

28.2(19.9-33.8)

34.4(13.6-49.5)

IS:PCS ratio

0.3 (0.2-0.5)

0.2(0.1-0.3)

0.2(0.1-0.5)

0.2(0.1-0.3)

0.2(0.1-0.3)

0.2(0.2-0.3)

0.5 (0.3-1.2)

0.8 (0.5-1.1)

CVD (%)

165 (50)

2(5)

2(20)

73 (55)

24 (65)

8 (47)

17 (52)

39 (72)

cIMT (mm) n=281

0.65 ±0.13

0.58±0.09

0.64±0.13

0.65±0.12

0.70±0.15

0.70±0.12

0.64±0.10

0.70±0.15

BAR-FMD∆ (%) n=232

4 (2- 7)

6(4-10)

4(0-5)

3(1.5-6)

4(2-7)

3(1-5)

4(2-7)

3(2-7)

BAR-GTN∆ (%) n=221

12 (5-18)

18(14-24)

10(8-20)

12(7-17)

6(2-12)

4(1-8)

5(3-12)

3(2-10)

EP

Vascular parameters

AC C

Values are expressed as numbers with proportions (%) , mean ±s.d., or median (inter quartile range) CKD, chronic kidney disease; NA, not applicable; BMI, body mass index; eGFR, estimated glomerular function; IS, indoxyl sulphate; PCS, p-cresyl sulphate; CVD, cardiovascular disease; cIMT, carotid-intima-thickness; BARFMD∆, brachial arterial reactivity, flow-mediated dilation change; GTN, glycerol trinitrate.

20

ACCEPTED MANUSCRIPT Table 2: Univariate and multivariable logistic regression analyses evaluating the relationship between presence of cardiovascular disease and uremic toxins as predictor variables across the CKD spectrum Presence of cardiovascular disease Multivariable

OR

95% CI

P-value

Adjusted OR*

Total indoxyl sulphate (µmol/L)

1.05

1.00-1.09

0.035

1.10

Free indoxyl sulphate (µmol/L)

1.25

1.05-1.48

0.010

1.47

Total p-cresyl sulpahte (µmol/L)

1.05

1.02-1.08

0.001

1.05

Free p-cresyl sulpahte (µmol/L)

1.25

1.09-1.44

0.002

1.34

95% CI

P-value

RI PT

Univariate

Predictor variables in final model

1.04-1.16

≤0.001

1.18-1.82

≤0.001

1.01-1.09

0.007

1.12-1.60

0.001

AC C

EP

TE D

M AN U

SC

Multivariable model included age, gender, smoking history (no/yes), statin therapy (no/yes), albumin, systolic blood pressure, body mass index (BMI), diabetes (no/yes) and toxin (entered separately) Note: When total IS and PCS were added in the same model only total IS remained significant. Odds ratio (OR) and Confidence Intervals (CI) presented are for the odds of a 10µmol/L increase in the toxins.

21

ACCEPTED MANUSCRIPT FIGURE LEGENDS Figure 1: Serum indoxyl sulphate across the CKD spectrum ^p<0.02 versus controls; # p<0.001 versus controls & CKD stage 2 & 3; +p<0.001 versus controls & CKD stages 2-4; *p£0.002 versus controls & CKD stages 2-5. Whiskers represent 5-95percentile. Levels were tranformed prior

Figure 2: Serum p-cresyl sulphate across the CKD spectrum

RI PT

to ANOVA and pairwise testing.

2-4. Whiskers represent 5-95percentile. Comparison were made using Kruskal-Wallis and

M AN U

Mann-Whitney for pair-wise comparisons.

SC

#p<0.001 versus controls; +p<0.02 versus controls & CKD stage 2 & 3; *p£0.002 versus controls & CKD stages

Figure 3: Free (non-protein-bound) fractions of indoxyl sulphate and p-cresyl sulphate across the CKD spectrum. *p<0.01 versus control & CKD stages 2-4.

AC C

EP

TE D

Bars represent medians.

22

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

ACCEPTED MANUSCRIPT

Supplement table 1: Correlation matrix of uremic toxinsa and kidney function (r) Free IS (µmol/L)

Total PCS (µmol/L)

Free PCS (µmol/L)

Total IS (µmol/L)

1.00

Free IS (µmol/L)

0.96

1.00

Total PCS (µmol/L)

0.64

0.62

1.00

0.79

0.84

0.87

1.00

-0.67

-0.70

-0.57

-0.68

Free PCS (µmol/L) 2

eGFR (mL/min/1.73m )

AC C

EP

TE D

M AN U

SC

p<0.001 for all correlations a Transformed prior to Pearson correlation r, Correlation coefficient

RI PT

Total IS (µmol/L)

ACCEPTED MANUSCRIPT

Carotid-intima-thicknessa (mm) Predictor variables in final model

Univariate Std β

P-value

Multivariable Std β

P-value

RI PT

Supplement table 1: Univariate and multivariable logistic regression analyses evaluating the relationship between surrogate markers of cardiovascular disease and uremic toxins as predictor variables across the CKD spectrum Brachial arterial reactivity- glycerol trinitratea ∆ (%) Univariate Multivariable Std β

P-value

Std β

P-value

AC C

EP

TE D

M AN U

SC

Total indoxyl sulphate 0.13 0.032 0.15 0.005 -0.32 ≤0.001 -0.26 ≤0.001 (µmol/L) Free indoxyl sulphate 0.16 0.008 0.14 0.005 -0.26 ≤0.001 -0.17 0.009 (µmol/L) Total p-cresyl sulpahte 0.17 0.005 0.11 0.039 -0.33 ≤0.001 -0.22 0.001 (µmol/L) Free p-cresyl sulpahte 0.16 0.010 0.11 0.28 -0.29 ≤0.001 -0.18 0.006 (µmol/L) Multivariable model included age, gender, smoking history (no/yes), statin therapy (no/yes), albumin, systolic blood pressure, body mass index (BMI), diabetes (no/yes) and toxin (entered separately) Note: When total IS and PCS were added in the same model only total IS remained significant for both models. Standardised betas are presented are for a 10µmol/L increase in the toxins. a Data transformed (log for cIMT, square root for BAR) prior to regression

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

Supplement figure 1: Correlation between uremic toxins and surrogate markers of cardiovascular disease