Correspondence to:
M. José Fernández–Reyes, md, Servicio de Nefrología, Hospital General de Segovia, Carretera de Avila s/n, Segovia, Spain.
email: capd@hulp.insalud.es

C-Reactive protein, mortality, inflammation, peritoneal dialysis

Serum albumin is a remarkable predictor of morbidity and mortality in dialysis patients [1]. Some data suggest, however, that hypoalbuminemia may be secondary to non nutritional factors, specifically inflammation [2,3]. Bergström et al. [2] showed that serum albumin levels are predicted by C-reactive protein (CRP), and that CRP levels are better predictors of mortality than are albumin levels. Moreover, some authors found an independent association between CRP levels and morbidity and mortality in both hemodialysis (HD) [4–7] and peritoneal dialysis (PD) patients [8].

Although the relationships among CRP level, serum albumin, and morbidity and mortality in dialysis patients have been accepted, the cause for CRP elevation and the inflammatory mechanism causing the increase in mortality remain to be determined. End-stage renal disease (ESRD) patients have higher CRP plasma levels than does a healthy population [9], and HD patients have even higher levels than do PD or pre-dialysis patients [9,10]. This difference has been attributed to the HD process as a stimulus for inflammation. In addition, the use of "less biocompatible" membranes [9,11] and techniques associated with retrofiltration [12] cause a chronic inflammatory response.

The goal of this study was to determine, based on CRP plasma levels, the degree of inflammation in HD patients using low-permeability polysulfone membranes, and that in PD patients. A secondary objective was to study factors associated with hypoalbuminemia and inflammation in both populations.

We studied 69 stable dialysis patients, 32 on HD and 37 on PD; age range, 50 – 80 years; mean age, 69.9 ± 8.2 years. Of the 69 patients, 56% were men, and 17% had diabetes. The study patients had been treated with dialysis for a mean of 27 months (range: 1 – 122 months). At the time of inclusion in the study, the patients showed no signs or symptoms of acute inflammatory disease.

All HD patients were treated three times per week with bicarbonate dialysate. The dialysate flow was 500 mL/min and the blood flow ranged between 300 mL/min and 350 mL/min. The duration of each HD session was adjusted to obtain Kt/V values higher than 1.2. All patients were treated with low-permeability polysulfone membranes: F8HPS and F10HPS (Fresenius Medical Care, Bad Homburg, Germany), with ultrafiltration coefficients of 11.1 mL/h/mmHg and 18 mL/h/mmHg respectively. The water analyses during the study period and the previous year found dialysate bacteria < 10 CFU/mL and endotoxin levels < 0.25 IU/mL. Of the patients on PD, 27 were on continuous ambulatory peritoneal dialysis (CAPD), and 10 were on automated peritoneal dialysis (APD).

Biochemical and hematologic parameters were measured by routine methodologies and included urea, albumin, total protein, cholesterol, serum iron, transferrin, ferritin, hemoglobin, and leukocytes. Plasma levels of CRP were measured by nephelometry, considering non quantifiable normal levels lower than 0.6 mg/dL. A higher level was interpreted as indicative of inflammation, and CRP was quantified.

Upon entrance into the study, an Overall Comorbidity Index (OCI) was scored for each patient. This index includes 24 pathologic conditions defined by specific criteria and scored from 0 – 4: 0 for absence, 1 for presence without repercussion, and 2 – 4 for a presence that causes mild (score 2), moderate (score 3), or severe (score 4) physical limitation. From the OCI, and including angina, arrhythmia, congestive cardiac failure, hypertension, and peripheral vascular or cerebral diseases, a cardiovascular comorbidity index (CCI) was also calculated [13].

Statistical analysis was performed using the SPSS 6.12 software program (SPSS Inc., Chicago, IL, U.S.A.). The Student t-test was used for comparison of unpaired data, and the Mann–Whitney test was used for nonparametric data comparison. The Fisher exact test and the chi-square test were used to estimate associations among qualitative variables. A linear regression analysis was also performed using Pearson or Spearman coefficients, depending on normal or non normal distribution. A logistic regression analysis was performed to determine possible influences of multiple variables on CRP (at values higher and lower than 1 mg/dL). Values of p less than 0.05 were considered significant.

Of the patients studied, 53% showed CRP plasma levels higher than 0.6 mg/dL; in 36% of the patients, the levels were higher than 1 mg/dL (Table I). No significant differences were seen between the dialysis groups in regard to these percentages. In fact, mean values of CRP higher than 0.6 mg/dL were not different between the HD and PD patients (2 ± 1.3 mg/dL for HD vs 1.8 ± 1.3 mg/dL for PD).

Table II shows the principal differences between patients with CRP levels higher and lower than 1 mg/dL. Comorbidity indices, protein intake, and erythropoietin (EPO) dose were similar. Nonetheless, patients with CRP levels higher than 1 mg/dL showed lower serum albumin, iron, hemoglobin, and transferrin levels, and higher ferritin values and leukocyte counts.

Logistic regression analysis found that CRP levels higher and lower than 1 mg/dL were significantly associated with serum albumin [p = 0.01; odds ratio (OR): 0.15], serum iron (p = 0.006; OR: 0.96), transferrin (p = 0.004; OR: 0.97), and hemoglobin (p = 0.02; OR: 0.67). The other quantitative and qualitative variables were not significantly related to CRP values.

For CRP values higher than 0.6 mg/dL, the correlation between CRP and serum albumin was negative and significant (r = –0.65; p < 0.001).

In dialysis patients, CRP has become one of the best predictors of morbidity and mortality [2,4–6]. Cardiovascular disease continues to be the principal cause of death in these patients. Evidence is growing for an inflammatory component in the atherosclerosis process. Elevated CRP has been confirmed as an independent cardiovascular risk factor in healthy people [14] and in ESRD patients [6,15]. Owing to the important repercussions of these factors on dialysis survival, any measure designed to estimate and control the process should be considered important.

Levels of CRP in renal patients are possibly higher than in the general population. The reasons have not been identified, but the uremic state or the renal disease itself may be the cause [9]. Haubitz et al. [9] found that these high levels are even higher among HD patients relative to PD or pre-dialysis patients. The elevation was confirmed by the increased levels after HD was initiated in 8 patients. The group concluded that PD represents a lower degree of inflammation than does HD. However, that study was performed on HD patients using cuprophane membranes. The same authors had previously demonstrated CRP elevation with the use of cuprophane membranes [16].

McIntyre et al. [10] also found higher CRP levels in HD patients than in PD patients , although the HD process with modified cellulose did not increase such levels. The researchers suggest that the use of dialysate with non pure water may be related to the phenomenon of chronic inflammation. Other authors suggest that the acquisition of endotoxins via retrofiltration from dialysate prepared using non pure water may be the inflammatory stimulus [12].

The patients included in the present study were treated with biocompatible membranes (polysulfone) with a low degree of permeability to avoid retrofiltration, and the water used was relatively pure. Possibly as a consequence of these precautions, we found no differences in CRP plasma levels between our HD and PD patients.

Several studies have demonstrated a relationship between serum albumin and CRP levels in both PD and HD [2,3,7,8]. Patients on PD show serum albumin levels that are lower than those in patients on HD. This difference could be explained by over-hydration [18] and high peritoneal permeability [19,20]. The association observed between ultrafiltration capacity and serum albumin corroborates this hypothesis.

The frequent elevation of CRP plasma levels observed in HD and PD patients suggests the presence of silent inflammatory states in the two populations. Hemodialysis performed with biocompatible, low-permeability membranes is not associated with CRP plasma levels higher than those seen in PD. Hypoalbuminemia is related with CRP levels in both groups. Levels of serum albumin, slightly lower in PD patients, are also associated with diminished peritoneal ultrafiltration capacity.