Intraindividual Interleukin-6 Variations on the Cardiovascular Prognosis of Patients with Chronic Renal Disease

Abstract

In chronic kidney disease (CKD) patients on dialysis, plasma interleukin (IL)-6 levels predict mortality better than other markers. Impact of intraindividual changes of inflammatory markers on cardiovascular (CV) events in CKD patients is unknown. The aim of this study is to demonstrate the relation between CV outcomes and variations of C-reactive protein (CRP), IL-6, IL-1β, and tumor necrosis factor (TNF)-α in CKD. Ninety patients (mean age: 68.5 ± 12.8 years) at different stages (1–4) of CKD were evaluated. Serum CRP, IL-6, IL-1β, and TNF-α were measured basally and after taking statins or angiotensin II receptor blockers. Three patterns were defined for each marker (baseline, mean of two measurements, and variation of the marker: increase or decrease after 6 months). During follow-up (mean time: 72.7 ± 19.8 months), 14 patients died, 11 were included on dialysis program, and 29 suffered a CV event. Patients with persistently elevated IL-6 values had higher risk to develop CV events [OR = 1.21 (1.11–1.32), p = 0.001]. Mean of two measurements of IL-6 was a better predictor for events than a single measurement of IL-6, CRP, TNF-α, and IL-1β. A mean of two determinations of plasma IL-6 greater than 6 pg/mL and previous peripheral vascular disease was related to an increased risk for CV events [2.34 (1.05–5.22), p = 0.037 and 2.95 (1.27–6.93), p = 0.011, respectively] in an adjusted Cox regression model. IL-6 is a better inflammatory marker than CRP, TNF-α, and IL1β at predicting CV events in CKD nondialysis patients. Mean of two measurements is better than simple determinations at predicting CV outcome.

• cardiovascular events
• chronic kidney disease
• inflammation
• interleukin-6
• prognosis

INTRODUCTION

Inflammation is a major risk factor for mortality and cardiovascular (CV) complications in patients with chronic kidney disease (CKD).^1,2 In patients with CKD, a state of persistent, low-grade inflammation is commonly observed.³ The reasons for the association of CKD with this state are complex. This is due to multiple underlying factors, including uremic environment, increased circulating pro-inflammatory cytokines, increased oxidative stress, chronic infections, and so on.⁴ This state of chronic inflammation is evidenced by elevated levels of a number of biomarkers.⁵ Oxidative stress results from an imbalance between the generation of oxidant compounds and the lack of antioxidant compounds.⁶ The prevalence of CV risk factors in patients with CKD could partially explain the higher rates of oxidative stress, although CKD itself can increase these rates.⁷ The consequences of increased oxidative stress are endothelial dysfunction and left ventricular hypertrophy, both of which affect total survival and CV event-free survival in CKD patients.^8,9

In recent years, there have been a large number of studies to find out the best inflammatory marker for predicting the prognosis in patients with CKD.^10,11

In 2008, a large genetic study of patients with ischemic heart disease¹² showed that C-reactive protein (CRP) polymorphisms are not in themselves associated with an increased risk of ischemic vascular disease; therefore, available evidence suggested that although CRP is a strong risk marker it is not a risk factor of CV disease.¹³ As studies suggested that CRP is a risk marker rather than risk factor, other inflammatory proteins may be more relevant as causative inducers of CV disease. One such inflammatory protein may be interleukin (IL)-6. IL-6 is a 26-kDa protein produced by the liver and lymphocytes and activated macrophages, and it is considered crucial in the acute-phase inflammatory response. In contrast to variations in the CRP gene, IL-6 polymorphisms seem to be important genetic factors in premature coronary artery disease.¹⁴ A common G/C polymorphism at position −174 in the promoter region of the IL-6 gene has been associated with higher risk for CV disease incidence.¹⁵ Thus, IL-6 may directly act as a promoter of atherosclerosis, as the IL-6 gene has functional variants that could affect both inflammation and risk for CV disease among dialysis patients. However, the role of the −174 G/C variant in CV risk is uncertain.¹⁶

In CKD patients on dialysis, plasma IL-6 levels have been shown to better predict death than interleukin (IL)-1β, tumor necrosis factor (TNF)-α, and CRP levels.¹⁷ However, only one study has demonstrated the predictive role of IL-6 in nondialysis CKD patients.¹⁸

Robust evidence concerning single measurements of various inflammatory biomarkers as independent predictors of CV risk and mortality in CKD patients has justified their use to identify patients at increased risk. However, until now only few studies have addressed the relationship between longitudinal inflammatory variation and CV risk, and these studies have been performed among dialysis patients.^19,20

The aim of this study is to analyze the best inflammatory marker of CV outcome in nondialysis CKD patients and to examine whether the simple determination of an inflammatory marker is a better predictor than the changes of the marker over a 6-month period.

MATERIAL AND METHODS

Tis study is a post hoc analysis that includes patients who participated in two clinical trials carried out in outpatient Nephrology consultation in 2004. The objective of the first study was to evaluate the anti-inflammatory role of 20 mg of atorvastatin²¹ in patients with renal disease, and the second study analyzed the anti-inflammatory role and the effect of olmesartan (40 mg; angiotensin II receptor blocker) on metabolic syndrome.²² All included patients had renal disease in any stage and did not have hospital admissions or CV events 3 months prior to inclusion in the study. Patients with low-density lipoprotein (LDL)-cholesterol between 100 and 200 mg/dL were included in the first study. Hypertensive nondiabetic patients (blood pressure ≥ 130/85 mm Hg) were included in the second study. Patients with active infection or neoplasm were excluded from the study. Data from 90 patients were analyzed, excluding six patients who did not have measurements of all of the inflammatory parameters over the course of follow-up (Figure 1).

Figure 1.  Flow chart.

Demographic variables were collected: age, sex, etiology of renal disease, systolic and diastolic blood pressures, and history of CV risk: diabetes, ischemic heart disease, heart failure, cerebrovascular accidents, and peripheral artery disease. In addition, we recorded concomitant medication: use of renin–angiotensin–aldosterone system blockers, antiplatelet medications, and statins.

The following variables were measured using routine standardized methods at baseline and at 6 months after receiving treatment with 20 mg daily of atorvastatin or 40 mg daily of olmesartan: hemoglobin, serum creatinine, uric acid, total cholesterol, LDL-cholesterol, high-density lipoprotein (HDL)-cholesterol, and triglycerides. Albuminuria (g/day) in urine was measured from 24 h urine using the immunonephelometry method. In addition, the following inflammation markers were measured at baseline and at 6 months: high-sensitivity CRP (hs-CRP), IL-6, IL-1β, and TNF-α. The obtained samples were immediately centrifuged at 4°C and 1500 rpm for 10 min, and the supernatant was separated into different aliquots, which were frozen at −80°C until they were tested. hs-CRP was measured using immunoassay based on the turbidimetric method in a Hitachi automatic analyser (Sigma Chemical Co., St. Louis, MO, USA). IL-6, IL-1β, and TNF-α were measured using an enzyme immunoassay. Specific monoclonal antibodies were used for the different cytokines (R&D Systems, Minneapolis, MN, USA). Glomerular filtration rate (GFR) was estimated using the abbreviated Modification Diet Renal Disease-4 (MDRD-4) formula.

Final Variables and Follow-Up

Patients were followed-up for a mean period of 72.7 ± 19.8 months. During follow-up, fatal and nonfatal CV events and all-cause mortality were recorded. CV events were considered in the following cases: sudden death, acute myocardial infarction, coronary bypass, or clinical angina; congestive heart failure was diagnosed by the presence of acute pulmonary edema and an echocardiogram with ventricular systolic dysfunction (left ventricular ejection fraction (LVEF) 45%); ischemic or hemorrhagic cerebrovascular accident and acute peripheral vascular disease diagnosed by stenosis of the primary arteries or lower extremities confirmed by arteriogram and/or need for amputation.

Each CV event was recorded and reviewed by the same nephrologist. This information always included a hospitalization report and, in the case of out-of-hospital death, telephone information with direct family members was recorded.

Statistical Analysis

Values are expressed as the mean ± standard deviation or median (interquartile range) if the variables did not follow a normal distribution. The Kolmogorov–Smirnov test was used to evaluate normality of the distribution of the parameters. To analyze the difference between patients who had suffered a CV event versus the group of patients who did not have events, the chi-squared test and Student’s t-test were used for univariate analysis in the case of normally distributed variables. Mann–Whitney U-test was used for parameters with a non-Gaussian distribution. We assumed a significant statistical difference when p < 0.05.

To analyze the role of inflammatory biomarkers in the CV outcome, we used three marker measurement models: baseline measurement, mean of the measurements (baseline and after treatment with statins or angiotensin II receptor blockers during 6 months), and the model’s variation marker over the course of 6 months (as increase, decrease, or remain the same).

The prognostic power of CV risk factors and inflammation markers on CV events was evaluated using a multivariate Cox regression model. All the covariables that predicted with a p-value <0.1 for CV events and/or mortality were introduced into the model. The regression coefficients and their standard deviations were calculated using the SPSS 16.0^® (Chicago, IL, USA) statistical package.

RESULTS

Ninety patients (63 men and 27 women) with a median age of 72 years (63–77 years) and an estimated GFR of 40.1 ± 19.7 mL/min/1.73 m² were analyzed. The patients were distributed based on CKD stages, that is, 9 patients with stage 1 or 2 CKD, 48 patients with stage 3 CKD, and 33 patients with stage 4 CKD. The etiology of renal disease was glomerulonephritis in 26 patients, polycystic renal diseases in 11 patients, vascular renal diseases in 23 patients, interstitial chronic renal diseases in 12 patients, diabetic nephropathy in 2 patients, and miscellaneous nephropathies in 16 patients. The baseline demographic variables, laboratory parameters, and inflammation markers are shown in Tables 1 and 2. The distribution of inflammation markers based on the stages of renal disease is shown in Table 3. In relation to concomitant medication: 93% of patients use renin–angiotensin–aldosterone system blockers, 32% of patients use anti-platelet medications, and 72% of patients use statins.

Table 1.  Demographic and analytical characteristics of patients included in the study: baseline characteristics.

Baseline parameters
Age (years)^a	72 (63–77)
Cr (mg/dL)	1.8 ± 0.7
Diabetes (yes/no)	10/80
Ischaemic heart disease (yes/no)	11/79
PVD (yes/no)	12/78
Hb (g/dL)^a	13 (12–14)
Uric acid (mg/dL)	7.6 ± 2.2
Cholesterol (mg/dL)	216 ± 37
HDL-cholesterol (mg/dL)	56 ± 17
LDL-cholesterol (mg/dL)	135 ± 35
Triglycerides (mg/dL)	143 ± 56

Notes: ^aValues expressed as median and range. Cr, serum creatinine; PVD, peripheric vascular disease; HB, serum hemoglobin; HDL, high-density lipoproteins; LDL, low-density lipoproteins.

Table 2.  Inflammatory parameters: baseline inflammatory parameters.

	Events
	Yes (N = 61)	No (N = 29)	p
Baseline IL-6 (pg/mL)	7.55 ± 8.9	5.36 ± 3.27	0.05
Mean IL-6 (pg/mL)	7.46 ± 5.42	5.09 ± 2.47	0.005
Baseline IL-1β (pg/mL)	1.77 ± 0.77	1.61 ± 0.66	Ns
Mean IL-1β (pg/mL)	1.59 ± 0.63	1.46 ± 0.65	Ns
Baseline TNF-α (pg/mL)	7.48 ± 3.99	7.17 ± 3.22	Ns
Mean TNF-α (pg/mL)	6.90 ± 2.94	6.88 ± 2.72	Ns
Baseline CRP (mg/L)	8.11 ± 8.48	5.49 ± 4.69	0.05
Mean CRP (mg/L)	6.41 ± 5.97	4.82 ± 3.79	Ns
Albuminuria (mg/day)^a	22.0 (0.0–607.8)	38.0 (0.0–263.1)	Ns

Notes: ^aValues are expressed as median (interquartile range). Ns, nonsignificant; IL-6, interleukin-6; IL-1β, interleukin-1β; TNF-α, tumor necrosis factor-α; CRP, C-reactive protein.

Table 3.  Inflammatory parameters as a function of chronic kidney disease stages.

	hs-CRP (mg/L)	IL-1β (pg/mL)	IL-6 (pg/mL)	TNF-α (pg/mL)
Stage 1 or 2 (n = 9)	7.65 ± 4.9	1.55 ± 0.65	3.93 ± 2.64	6.25 ± 2.84
Stage 3 (n = 48)	6.20 ± 6.20	1.69 ± 0.73	6.75 ± 7.23	7.07 ± 3.5
Stage 4 (n = 33)	5.23 ± 5.32	1.72 ± 0.67	5.10 ± 3.05	7.77 ± 3.83

Note: IL-6, interleukin-6; IL-1β, interleukin-1β; TNF-α, tumor necrosis factor-α; hs-CRP, high-sensitive C-reactive protein.

After a follow-up period of 72.7 ± 19.7 months, 29 patients suffered a CV event (3 – sudden death, 9 – ischemic heart disease, 8 – heart failure, 5 – cerebrovascular accident, and 4 – acute peripheral artery disease), 14 patients died, and 11 patients started dialysis program. Of the 14 deaths, 10 deaths were due to CV causes (3 – sudden death, 2 – heart failure, 1 – ischemic heart disease, 2 – peripheral vascular disease, and 2 – cerebrovascular accidents) and 4 deaths were due to non-CV causes (sepsis, neoplasia, respiratory insufficiency, and gastrointestinal hemorrhage).

Inflammation Markers and Fatal and Nonfatal CV Events

In the univariate analysis, older patients (73.6 ± 9.8 vs. 66.2 ± 13.5 years, p = 0.007) and patients with a history of peripheral vascular disease (75% vs. 25%, p = 0.001) had more CV events (see Table 4). Baseline IL-6, median of two measurements of IL-6 (previous and after treatment with atorvastatin and olmesartan), and baseline hs-PCR levels were higher in the patients who had a CV event (see Table 5).

Table 4.  Demographic and analytical characteristics of patients included in the study: Differences between patients who suffered and not suffered cardiovascular events. Univariable analysis (Student’s t).

	Events
	Yes (N = 61)	No (N = 29)	p
SBP (mmHg)	155 ± 25	151 ± 20	NS
DBP (mmHg)	78 ± 12	80 ± 11	NS
Creatinine (mg/dL)	1.72 ± 0.62	1.83 ± 0.73	NS
MDRD (mL/min/1.73 m2)	39.4 ± 19.6	45.0 ± 23.3	NS
Cholesterol (mg/dL)	212 ± 35	217 ± 38	NS
HDL cholesterol (mg/dL)	54 ± 15	56 ± 17	NS
LDL cholesterol (mg/dL)	134 ± 34	135 ± 35	NS
Triglycerides (mg/dL)	130 ± 46	148 ± 60	NS
Uric acid (mg/dL)	7.4 ± 2.4	7.6 ± 2.1	NS
Age (years)	73.6 ± 9.8	66.2 ± 13.5	0.007
PVD (%)	75%	25%	0.001

Note: SBP, systolic blood pressure; DBP, diastolic blood pressure; MDRD, Modification Diet Renal Disease; PVD, peripheric vascular disease.

Table 5.  Inflammatory parameters: Inflammatory parameters in patients who had cardiovascular events versus patients without cardiovascular events. Univariate Analysis.

Inflammatory parameters
Baseline IL-6 (pg/mL)	4.8 (3.3–7.1)
Mean IL-6 (pg/mL)	5.8 ± 3.8
Baseline IL-1β (pg/mL)	1.6 ± 0.6
Median IL-1β (pg/mL)	1.5 ± 0.6
Baseline TNF-α (pg/mL)	7.2 ± 3.4
Median TNF-α (pg/mL)	6.8 ± 2.7
Baseline CRP^a (mg/L)	4.0 (2.3–6.9)
Mean CRP^a (mg/L)	3.8 (2.0–6.0)
Albuminuria^a (mg/day)	22 (0–300)

Notes: ^aValues expressed as median and range. IL-6, interleukin-6; IL-1β, interleukin-1β; TNF-α, tumor necrosis factor-α; CRP, C-reactive protein.

Table 6.  Cox regression models for evaluating the predictive value of the different model of inflammatory parameters.

		HR (95% CI)	p
IL-6 (pg/mL)	Baseline	0.991 (0.905–1.084)	0.838
	Mean	1.203 (1.032–1.402)	0.018
	Variation	0.662 (0.339–1.289)	0.225
	Age	1.051 (1.004–1.101)	0.034
IL-1β (pg/mL)	Baseline	1.728 (0.671–4.450)	0.257
	Mean	0.708 (0.277–1.813)	0.472
	Variation	0.489 (0.174–1.373)	0.175
	Age	1.064 (1.019–1.110)	0.005
TNF-α (pg/mL)	Baseline	1.029 (0.782–1.354)	0.838
	Mean	1.031 (0.773–1.375)	0.837
	Variation	1.079 (0.362–3.218)	0.891
	Age	1.053 (1.010–1.099)	0.016
CRP (mg/L)	Baseline	1.118 (0.955–1.309)	0.165
	Mean	0.626 (0.771–1.170)	0.626
	Variation	0.706 (0.246–2.022)	0.516
	Age	1.044 (1.001–1.088)	0.047

Note: HR, hazard ratio; 95% CI, 95% confidence interval; IL-6, interleukin-6; IL-1β, interleudin 1β; TNF-α, tumor necrosis factor-α; CRP, C-reactive proteinuiria.

In the unadjusted analysis, intraindividual variation in inflammation markers after treatment with statins or olmesartan was not associated with CV events. A nonsignificant tendency was found for having a greater number of events in patients in whom IL-6 levels increased after treatment for 6 months (Figure 2).

Figure 2.  Kaplan-Meyer for intra-individual variation in inflammation markers after treatment with statins or olmesartan and association with cardiovascular events. (A) = IL-1; (B) = IL-6; (C) = TNF; (D) = PCR. P = ns.

Only the mean of IL-6 levels independently predicted CV events in the age-adjusted Cox regression model (Table 6). In a Cox regression model adjusted for renal function (estimated GRF), age, albuminuria, and ischemic heart disease, mean of IL-6 [OR: 2.34 (1.05–5.22), p = 0.037] and history of peripheral vascular disease [OR: 2.95 (1.27–6.93), p = 0.011] increased the relative risk of suffering a CV event. Figure 3 shows the CV event-free survival in the group of patients with a mean of two IL-6 measurements above the mean (>6 pg/mL) compared with those whose levels resulted lower (log rank: 8.26, p < 0.04).

Figure 3.  Cox regression for free-cardiovascular event survival in patients with mean of IL-6 over and under 6 pg/mL, adjusted for age, renal function, albuminuria and history of cardiovascular disease.

DISCUSSION

The impact of intraindividual inflammation marker variations on the CV outcome of nondialysis CKD patients is unknown. Studies performed to date have been focused on patients in dialysis.^23–25 This is the first study that analyzes the influence of intraindividual variations of different inflammatory markers on the CV prognosis after treatment with statins and olmesartan in patients with different stages of chronic renal disease.

Our results confirm that the best inflammatory marker for predicting CV outcome in CKD patients is IL-6, as has been described in the population on dialysis.^26–28 Moreover, the patients with persistently elevated IL-6 levels after treatment with anti-inflammatory drugs (statins or angiotensin II receptor blockers) will have a higher probability of a CV event (Figure 3).

In addition, we show that other inflammation markers such as hs-CRP, TNF-α, and IL-1β do not predict the CV outcome in CKD patients not in dialysis. Despite the fact that many comparative studies have suggested that IL-6 might be the best outcome predictor in early and advanced CKD,^21,22 CRP measurement is still the prototypic marker of uremic inflammation²⁹ owing to the widespread availability of this method. Nevertheless, the majority of studies on CKD have been based on patients in stage 5. To the best of our knowledge, there are only four studies published on patients with early CKD. In three of them, CRP was a predictive marker of CV events.^12,30–32 Only Barreto et al.¹² have performed a study in CKD nondialysis (stages 2–5) patients and have measured CRP, IL-6, and TNF-α. In this report, IL-6 was the best outcome predictor in CKD patients not included in the dialysis program. Recent reports indicate that IL-6, not TNF-α, and IL-1β might be involved in CV pathologies, perhaps even directly by affecting the atherosclerosis burden, as high plasma IL-6 levels have been related to the risk of coronary death and major coronary events in patients with unstable angina.³³ Furthermore, IL-6 mRNA has been detected in coronary plaque samples obtained from patients having undergone heart transplantation or atherectomy.³⁴ This hypothesis is corroborated by the fact that higher plasma IL-6 levels predict the risk of myocardial infarction among apparently healthy men³⁵ and have been associated with severe congestive heart failure.³⁶

Follow-up periods for studies in which single measurements of inflammatory biomarkers were used range from 2 to 10 years. From a clinical point of view, it would be incorrect to assume that a single CRP measurement would predict the probability of death within 10 years. Indeed, conditional or time-stratified analyses showed that CRP is an excellent predictor of short-term risk (1-year follow-up).^37,38 However, the association between CRP and mortality diminishes over an extended period (2–3 years of follow-up) when other comorbidities and conditions influence the patients’ prognosis.³⁹ Therefore, a single measurement should only be taken into account for the short-term outcome. The robust evidence concerning single measurements of various inflammatory biomarkers as independent predictors of comorbidities and mortality in CKD patients has justified their use for identification of patients at increased risk.⁴⁰ Nevertheless, uremic patients are exposed to a low level of persistent inflammation with significant intraindividual and interindividual variations.⁴¹ Intraindividual variability is associated with the presence and influence of intercurrent events and different responses of the immune system.⁴² In 1800 healthy Japanese patients who were tested twice during a 1-year period, increasing age was associated with larger intraindividual CRP variation in men than in women.⁴³ Proinflammatory cytokines are also subject to a considerable day-to-day variation, and intraindividual IL-6 variability was reported in healthy adults.⁴⁴ To the best of our knowledge, only seven studies,^{13,14,17,18,20,32,45} all examining patients on dialysis, have addressed the consequences of fluctuating levels of CRP, IL-6, or TNF-α on mortality or CV disease. The follow-up time of these studies varied from 29 to 38 months. All but one were based on the CRP measurement except the most recently published study by Meuwese et al.,⁴⁶ which was based on CRP and inflammatory cytokines. Additionally, all inflammatory markers with a persistently high elevation predict mortality better than a single baseline measurement. Our study, for the first time, analyzes the role of variation of different inflammatory markers on CV prognosis, including fatal and nonfatal CV events. This is the first study that includes patients in whom a therapeutic intervention has been performed with an anti-inflammatory effect and it is the first analysis with a long follow-up time, greater than 3 years.

Our study is not free of limitations. The primary limitation is the small number of patients included, which may put the results in doubt. However, the only study that has measured the influence of cytokines on the prognosis of patients in predialysis to date is the study by Barreto et al.,¹² which included patients in dialysis and patients not in dialysis, of which 72 patients had moderate CKD. Results were similar to ours with respect to the predictive value of IL-6 versus other markers. In addition, the analyzed groups were heterogeneous. However, our study also offers some advantages such as serial measurements of inflammation markers after two therapeutic interventions: treatment with statins and renin–angiotensin–aldosterone system blockers. Both groups of drugs have been shown to have an anti-inflammatory effect, reducing marker levels.^15,16

In conclusion, IL-6 is the best inflammation marker in the long-term prediction of CV prognosis in patients with CKD who are not on dialysis. Persistently elevated levels of IL-6 predict CV prognosis better than a single baseline measurement of this cytokine. Moreover, persistently elevated IL-6 values after taking statins or angiotensin II receptor blockers predict CV events in CKD patients.

Declaration of interest: The authors report no conflicts of interest. The authors alone are responsible for the content and writing of the paper.