Cardiovascular disease relates to intestinal uptake of p-cresol in patients with chronic kidney disease

This is a secondary analysis of the Leuven mild to moderate CKD cohort (clinicaltrials.gov NCT00441623) [3]. Prevalent CKD patients, followed at the nephrology outpatient clinic of the University Hospitals Leuven, 18 year of age or older and able to provide consent, were eligible for inclusion. Patients were screened between November 2005 and September 2006. Data on baseline demographics were collected at time of informed consent. The study was performed according to the Declaration of Helsinki and approved by the ethics committee of the University Hospitals Leuven. Informed consent was obtained from all patients.

At inclusion, blood was taken by venous puncture for measurement of hemoglobin (g/dl), albumin (g/l), C-reactive protein (mg/l), cholesterol (mg/dl), calcium (mg/dl), phosphate (mg/dl), biointact parathyroid hormone (PTH) (ng/l), creatinine (mg/dl), blood urea nitrogen (g/dl) and PCS (μM). Hemoglobin, albumin, C-reactive protein, cholesterol, calcium, phosphate, biointact PTH, creatinine and blood urea nitrogen were all measured using standard laboratory techniques. The eGFR was calculated using the CKD-EPI equation [13]. PCS was quantified using a dedicated high-performance liquid chromatography (HPLC) method as described previously [14]. Assuming steady state conditions, 24 hour intestinal uptake of p-cresol was estimated from 24 hour urinary excretion of PCS. 24 hour urinary collections were sampled when available at time of inclusion. Completeness was assessed using 24 hour urinary creatinine excretion. Collections were considered complete when 24 hour urinary creatinine excretion was within 2 standard deviations of the mean creatinine excretion for the geographical region of this study, derived from the INTERSALT study [15]. Protein intake was calculated according to the formula of Maroni et al. by using 24 hour urinary urea nitrogen excretion and body weight [16, 17].

After inclusion, patients were prospectively followed at the nephrology outpatient clinic at 3- to 6-month intervals until December 31, 2010. Predefined endpoints were prospectively recorded and coded, blinded from clinical and biochemical data. The primary endpoint (cardiovascular event during follow-up) was a composite of death from cardiac causes, non-lethal myocardial infarction, myocardial ischemia, coronary intervention, ischemic stroke, or peripheral vascular disease, whichever occurred first. Only one event per subject was included in the analysis. After review of available information, cause of death was classified as either cardiovascular, infectious, malignancy, or other. Cardiovascular deaths included fatal myocardial infarction, sudden death, and death due to congestive heart failure. Cases of unobserved sudden death were considered cardiovascular death only when other potential causes could be excluded. Otherwise, they were classified as other cause of death. Out-of-hospital deaths were coded after consultation of the general practitioner. Non-lethal cardiovascular events included myocardial infarction, diagnosed based on elevated levels of cardiac enzymes and/or typical electrocardiography changes, myocardial ischemia with typical electrocardiography changes without elevated cardiac enzymes, coronary intervention (thrombolysis, percutaneous coronary intervention, or coronary artery bypass grafting), and ventricular arrhythmia. Ischemic stroke was defined as a neurologic deficit lasting more than 24 hours. Hemorrhagic stroke was excluded from the primary endpoint. Peripheral vascular disease included new-onset ischemic pain in the lower limbs, with abnormal ankle brachial pressure index or radiologic evidence of peripheral vascular disease, new-onset ischemic necrotic lesions, or surgical arterial intervention. Secondary endpoints included overall mortality and progression of renal disease, defined as progression to renal replacement therapy and/or doubling of serum creatinine during follow-up.

Data are expressed as mean (standard deviation) for normally distributed variables or median (interquartile range (IQR)) for non-normally distributed variables. Differences between baseline variables according to tertiles of 24 h urinary excretion of PCS were tested using parametric ANOVA, Kruskal-Wallis or chi-squared test as appropriate. Correlations between 24 h urinary excretion of PCS and other variables were calculated by Spearman’s rank correlation coefficients. To identify independent determinants of 24 h urinary excretion of PCS, multivariate linear regression analysis was performed. Relevant demographic (i.e., age, gender, presence of diabetes mellitus, smoking status, body mass index) and biochemical (i.e., hemoglobin, C-reactive protein (Ln), albumin, eGFR, 24 hour proteinuria (Ln), 24 h protein intake) parameters were first subjected to a backward elimination procedure on P < 0.2, with subsequently, a final backward elimination step on P < 0.05. The Kaplan-Meier method was used to estimate cumulative incidence of the endpoint with the log rank test to compare differences between tertiles of 24 h urinary excretion of PCS. Time to first event analysis was performed using Cox proportional hazards analysis. Besides 24 h urinary excretion of PCS, other relevant demographic (i.e., age, gender, presence of diabetes mellitus, prior cardiovascular disease, smoking status, body mass index, systolic blood pressure, use of statin therapy, use of angiotensin converting enzyme inhitor/receptor blocker therapy) and biochemical (i.e., hemoglobin, albumin, C-reactive protein (Ln), cholesterol, calcium, phosphate, PTH (Ln), creatinine, eGFR, 24 h proteinuria (Ln)) variables were selected from univariate analysis (P < 0.2). With this subset of variables different multivariate models were built and subjected to backward elimination (P < 0.05). Due to the limited number of events (n = 25), no more than three variables (24 h urinary excretion of PCS with two other variables) were included in each model. To test the proportionality assumption, each model was tested against log(time). In case the proportionality assumption was violated, time-dependent covariates were entered into the model. For both Kaplan-Meier and Cox proportional hazard analysis of the primary endpoint, data were censored at start of renal replacement therapy, death other than cardiovascular, loss to follow-up, or at the end of the study observation period. With respect to the secondary endpoints, data were censored at start of renal replacement therapy (for overall mortality), death (for renal disease progression), loss to follow-up, or at study end. For all statistical analysis, P-values less than 0.05 were considered significant. All statistical analyses were performed using SAS (version 9.3, the SAS institute, Cary, NC, USA).

Between November 2005 and September 2006, 548 prevalent patients with CKD K/DOQI stage 1 – 5, followed at the nephrology outpatient clinic of the University Hospitals Leuven, Belgium were found eligible to be enrolled in the Leuven mild to moderate CKD study. 499 patients providing informed consent were included [3]. Of these, urinary collections were available for 264 patients and considered complete for 203 patients. No further follow-up was scheduled for 3 patients, making a total of 200 patients to be included in the prospective cohort study (Figure 1). Baseline characteristics of the study population are presented in Table 1. Glomerular disease was the most prevalent underlying renal disease (37.5%), followed by undetermined cause (35.5%), vascular disease (9.0%), and diabetic nephropathy (6.5%). Apart from a small, but significant age (median 4 years younger, P 0.01) and systolic blood pressure (median 3 mmHg lower, P 0.05) difference, we observed no significant differences between the current study population and the original patient cohort.

24 h urinary excretion of PCS amounted to a median of 457.47 μmol (IQR 252.68 – 697.17). There was a moderate correlation between 24 h urinary excretion of PCS and serum PCS (Spearman’s rank correlation ρ 0.64, P < 0.0001) (Table 2). In addition, there were significant, though minor correlations between 24 h urinary excretion of PCS and 24 h intake of protein (ρ 0.30, P < 0.0001) as well as presence of diabetes mellitus (ρ 0.18, P < 0.0001). Patients with diabetes mellitus had a significantly higher 24 h urinary excretion of PCS than patients without diabetes mellitus (median 668.20 μmol (IQR 403.05 – 1161.10) vs. 427.02 μmol (IQR 681.36 – 238.37), P 0.009). Although there was a trend of increasing 24u urinary excretion of PCS with higher 24 h proteinuria, higher age, prior cardiovascular disease, and lower eGFR, these correlations did not reach significance.

In multivariate analysis, independent determinants of 24 h urinary excretion of PCS were age (β 4.30, P 0.03), presence of diabetes mellitus (β 202.67, P 0.01), body mass index (β -19.59, P 0.003), hemoglobin (β - 41.01, P 0.01) and 24 h protein intake (β 9.21, P < 0.0001) (Model R ² 0.22) (Table 3).

After a median follow-up of 52 months (IQR 20.8 – 57.7), 25 patients reached the combined primary endpoint, including 3 fatal and 22 non-fatal cardiovascular events (Table 4). Patients were censored at start of renal replacement therapy (n = 45), death other than cardiovascular (n = 15), loss to follow-up (n = 21), and at the end of the study observation period. Cumulative incidence of the primary endpoint in tertiles of urinary excretion of PCS was estimated with the Kaplan-Meier method (Figure 2). Log rank test was significant for differences between these 3 groups (P 0.037). In univariate Cox proportional hazard analysis, 24 h urinary excretion of PCS was directly associated with cardiovascular disease during follow-up (Hazard ratio (HR) per 100 μmol increase 1.112, P 0.002, HR highest vs. lowest tertile 3.011, P 0.03). Other significant variables include age (HR 1.064, P 0.002), systolic blood pressure (HR 1.021, P 0.04), prior cardiovascular disease (HR 5.880, P < 0.0001), presence of diabetes mellitus (HR 4.420, P 0.0003), albumin (HR 0.872, P 0.0003), eGFR (HR 0.973, P 0.02), 24 h proteinuria (Ln) (HR 1.298, P 0.009). We then built different multivariate models, each consisting of 3 variables (24 h urinary excretion of PCS and 2 other variables) (Table 5). In each model 24 h urinary excretion of PCS remained a significant predictor of cardiovascular events during follow-up. We also built sequential models with addition of variables that were considered confounders a priori, i.e., age, presence of diabetes mellitus, protein intake and eGFR. Again, 24 h urinary excretion of PCS remained associated with cardiovascular events during follow-up (HR 1.103 (1.006 – 1.209), P 0.04).

We also explored the relationship between 24 h urinary excretion of PCS and overall mortality, as well as renal disease progression. In this cohort, we observed a total of 21 deaths (5 cardiovascular, 7 oncologic, 1 infectious and 8 other deaths), again censored at start of renal replacement therapy and loss to follow-up. In univariate Cox proportional hazard analysis, 24 h urinary excretion of PCS was directly associated with overall mortality (HR per 100 μmol increase 1.090, P 0.02). Other significant variables include age (HR 1.072, P 0.0003), diabetes mellitus (HR 4.011, P 0.002), eGFR (HR 0.966, P 0.01), systolic blood pressure (HR 1.025, P 0.02), hemoglobin (HR 0.743, P 0.02), C-reactive protein (Ln) (HR 1.532, P 0.02) and PTH (Ln) (HR 1.680, P 0.01). In addition, we built different multivariate models with 24 h urinary excretion of PCS remaining a significant determinant of overall mortality in all trivariate-adjusted models (Table 6). As for cardiovascular disease, we also built sequential models with addition of variables that were considered confounders a priori, i.e., age, presence of diabetes mellitus, protein intake and eGFR. Again, there was a clear, albeit non-significant, relationship between 24 h urinary excretion of PCS and overall mortality (HR 1.098 (0.987 – 1.222), P 0.09).