Dissecting Inflammation in ESRD: Do Cytokines and C-Reactive Protein Have a Complementary Prognostic Value for Mortality in Dialysis Patients?

Protocol

The protocol conformed to the ethical guidelines of our institution, and informed consent was obtained from each participant. For hemodialysis (HD) patients, all studies were performed during a midweek nondialysis day; for continuous ambulatory peritoneal dialysis (CAPD) patients at empty abdomen, studies were performed between 8:00 a.m. and 1:00 p.m.

Study Cohort

A total of 217 patients with ESRD (120 men and 97 women) who had been on regular dialysis treatment for at least 6 mo (median duration 43 mo; interquartile range 20 to 101 mo) without history of congestive heart failure and without intercurrent inflammatory illnesses were considered eligible for the study. The main demographic and clinical characteristics of the study population are detailed in Table 1. The prevalence of diabetes in this cohort was 15% (33 of 217 patients).

HD patients were treated thrice weekly with standard bicarbonate dialysis (138 mmol/L Na, 35 mmol/L HCO₃, 1.5 mmol/L K, 1.25 mmol/L Ca, and 0.75 mmol/L Mg) either with cuprophane or semisynthetic membranes (dialysis filters surface area 1.1 to 1.7 m²). The average urea Kt/V in these patients was 1.22 ± 0.27. All patients who were on CAPD were on four exchanges per day with standard dialysis bags. The average weekly Kt/V in these patients was 1.69 ± 0.29. Eighty-six patients were habitual smokers (22 ± 17 cigarettes/d). A total of 110 patients were on treatment with erythropoietin. Ninety-seven patients were being treated with antihypertensive drugs (72 on monotherapy with angiotensin-converting enzyme inhibitors, AT-1 antagonists, calcium channel blockers, or α and β blockers, and 25 were on double or triple therapy with various combinations of these drugs).

Follow-Up

After the initial assessment, patients were followed up for an average of 41 mo (range 0.8 to 70.0 mo). During the follow-up, fatal cardiovascular events (myocardial infarction, electrocardiogram-documented arrhythmia, heart failure, stroke, and other thrombotic events except for arteriovenous fistula thromboses) and death were recorded accurately. Each death was reviewed and assigned an underlying cause by a panel of five physicians. As a part of the review process, all available medical information about each death was collected. This information always included study and hospitalization records. In the case of an out-of-hospital death, family members were interviewed by telephone for better ascertainment of the circumstances surrounding death.

Laboratory Measurements

Blood sampling was performed after an overnight fast between 8:00 a.m. and 10:00 a.m. always during a midweek nondialysis day for HD patients and at empty abdomen for CAPD patients. After 20 to 30 min of quiet resting in semirecumbent position, samples were taken into chilled EDTA Vacutainers, placed immediately on ice, and centrifuged within 30 min at −4°C, and the plasma was stored at −80°C before assay. Serum lipids, albumin, calcium, phosphate, and hemoglobin measurements were made using standard methods in the routine clinical laboratory. The plasma concentrations of asymmetric dimethylarginine and homocysteine were determined as reported elsewhere (4,5). Serum CRP was measured by using a commercially available kit (immunonephelometric method, lower limit of detection 3.5 mg/L; Behring, Scoppito, L’Aquila, Italy). Serum levels of IL-6, IL-1β, IL-18, and TNF-α were measured by ELISA with the use of Quantikine High Sensitivity kits (intra-assay coefficient of variation for these substances 1.6 to 10%; interassay coefficient of variation 3.3 to 10.2%; R&D Systems, Minneapolis, MN).

Statistical Analyses

Data are reported as mean ± SD (normally distributed data), median, and interquartile range (nonnormally distributed data) or as percentage frequency, and comparisons among groups were made by a P for trend. The combined prognostic power of inflammatory proteins (inflammatory burden) for all-cause mortality was analyzed by a score that was calculated as the sum of five biomarkers (CRP, IL-1β, IL-6, IL-18, and TNF-α), each defined in categorical terms (1 = third tertile of the corresponding data distribution; 0 = others two tertiles). We used the third tertile of each biomarker because in a previous article, this functional form of variable provided adequate risk stratification in the same dialysis cohort (1). We constructed six separate Cox models on the basis of either the inflammation score or inflammation biomarkers considered individually. These models included all covariates that were associated to all-cause death with P < 0.10 at univariate Cox regression analysis as well as variables that resulted to be associated with the inflammatory score with P < 0.10 at univariate analysis (Table 1). The treatment modality (HD/CAPD) was always introduced into these Cox models. The Cox models then were compared by using the −2 log likelihood test (6). Furthermore, we performed an alternative analysis by the Bayesian approach (i.e., by comparing the areas under the receiver operating characteristic [ROC] curves of the combined inflammatory load with that of IL-6 in the prediction of death [7]).

Hazard ratios and their 95% confidence intervals were calculated with the use of the estimated regression coefficients and their standard errors in the Cox regression analysis. All calculations were made using a standard statistical package (SPSS for Windows Version 9.0.1; SPSS, Chicago, IL).

Overall Inflammatory Burden and Mortality: −2 Log-Likelihood Statistics

During the follow-up period, 112 patients died, 65 of them (58% of total deaths) of cardiovascular causes. The independent prognostic value of overall inflammation burden for all-cause mortality was analyzed in a multiple Cox regression analysis that adjusted for a series of traditional and nontraditional risk factors. In this analysis, the hazard ratios of all-cause mortality increased in a dose-response manner according to the number of altered (third tertile) biomarkers so that the relative risks of patients who were exposed to one, two, three, and four or more increased levels of these inflammatory proteins were 1.48, 1.64, 2.76, and 3.05 times higher, respectively, than in those in the reference category (Table 2). However, the explained variation in mortality that was attributable to overall inflammatory burden was marginally (P = 0.06) higher than that provided by IL-6 (9.1 versus 6.1%; P = 0.06), which by far resulted to be the biomarker that showed the highest (P < 0.01) prognostic power for death (CRP +3.0%; IL-18 +2.9%; IL-1β +0.5%; and TNF-α +0.5%).

ROC Curves Analysis

The prediction power for mortality of overall inflammation burden and of IL-6 was compared further by the analysis of ROC curves. By this approach, the areas under the corresponding ROC curves both were significantly superior to the threshold of diagnostic indifference (P = 0.03 and P = 0.04, respectively) but identical between them (0.59 ± 0.04 versus 0.59 ± 0.04).

Overlapping and Complementary Role of Inflammatory Proteins and Atherosclerosis

Mainly because of the intellectual drive by Ross (3), atherosclerosis now is seen as an inflammatory disease, and substantial evidence has been accrued that the inflammation is a major contributor to arterial disease in patients with ESRD (1,8). Proinflammatory and anti-inflammatory cytokines are crucial elements in such process because the balance and the interplay of these proteins control the direction, amplitude, and duration of inflammation as well as tissue remodeling. Cytokines and acute-phase response proteins intervene at various levels in the atherosclerosis process. Thus, CRP seems to be implicated in endothelial dysfunction, foam cell formation, and inhibition of progenitor cell survival and differentiation (2). IL-6, a ubiquitous cytokine that is fundamental for leukocyte and endothelial cell activation, is highly expressed at the shoulder of the atherosclerotic plaque, where it may induce plaque instability by driving the local synthesis of matrix metalloproteinases, monocyte chemotactic protein-1, and TNF-α (9). Furthermore, IL-6 stimulates fibrinogen synthesis in the liver via IL-6–responsive sequences in the promoter region of the fibrinogen gene, thereby enhancing the risk for thrombosis (10). IL-1β (11) and IL-18 (12) play a relevant role in the inflammatory component of atherosclerosis in experimental models. TNF-α, besides being involved in plaque instability, is implicated in myocardial dysfunction and remodeling after an acute ischemic insult (13). The disparate interferences of cytokines and CRP with atherothrombosis suggest that the combined measurement of these biomarkers may provide prognostic information that is superior to what can be derived by individual biomarkers.

Individual and Combined Prediction Power of Inflammation Biomarkers in ESRD

In patients who have ESRD without clinically apparent inflammatory complications, independent of exposure to dialysis membranes and/or contaminated dialysis fluid, plasma levels of IL-6, IL-1β, and TNF-α and of CRP are raised substantially (8). The detrimental effect of inflammation in ESRD seems particularly prominent in the cardiovascular system because high CRP and high cytokine levels signal a high risk for mortality and incident cardiovascular complications (1,14–19). We previously showed that the plasma concentrations of IL-6, CRP, and other cytokines are mutually interrelated in dialysis patients without overt inflammatory processes (1). The hypothesis that the overall inflammation burden that is estimated by the combined measurement of cytokines and CRP has not been examined so far in patients with ESRD. To address this problem, we constructed a composite score based on the categorization of each inflammation marker into tertiles. We did so because this categorization allowed adequate risk stratification in our study cohort (1). Somewhat unexpected, this analysis showed that the combined use of these biomarkers provides very small, if any, gain in prediction power for death in comparison with that provided by the sole IL-6. The “inflammation score” was expected to capture the prognostic power of nonoverlapping or partially overlapping pathogenetic pathways that lead to arterial injury better than IL-6 or other individual cytokines. In other words, we hypothesized that this score reflects several critical aspects of the offending potential of inflammation, from endothelial dysfunction and plaque formation to plaque instability and atherothrombosis. Contrary to our expectations, the sole measurement of IL-6 was almost as informative as the composite inflammation score. IL-6 perhaps is the most investigated cytokine in human diseases (20). This protein has a very marked pleiotropy, being involved not only in inflammation but also in the regulation of several fundamental organ functions. Apart from the action on multipotent hematopoietic cell progenitors and on B cells or from its stimulatory role for the synthesis of other cytokines and acute-phase reactants in the liver, IL-6 affects myocardial cell growth, stimulates thermogenesis, suppresses the thyroid axis, and induces growth hormone secretion (20). This large variety of effects makes IL-6 unique among proinflammatory cytokines and indicates that it represents not only a potential agent of organ damage in pathophysiologic conditions but also a potent marker of the overall severity of the inflammation process. Our observation that IL-6 captures almost entirely the predictive power of inflammation in ESRD is in keeping with findings in the general population, in which IL-6 emerged as the sole inflammation marker to predict progression of atherosclerosis in an analysis that considers also CRP, adhesion molecules, and E-selectin (21). The marked pleiotropy of IL-6 makes it the best inflammation marker in ESRD, its prediction power being almost equal to what can be derived from simultaneous measurements of several other cytokines and CRP.