Inflammatory markers as mortality predictors in continuous ambulatory peritoneal dialysis patients

Abstract

Background/Aim: Besides peritonitis, the most common complication, indicators of chronic inflammation are also present in patients treated by peritoneal dialysis. The aim of this study was to analyze the predictive value of inflammatory parameters on mortality of continuous ambulatory peritoneal dialysis (CAPD) patients. Methods: Eighty-seven patients (57 males), aged from 30 to 85 [62.92 (10.61)] years who had been treated by a chronic program of CAPD for 3–113 months were analyzed. The basal period lasted 3 months with a follow-up of 30 months. Clinical parameters, dialysis adequacy and laboratory parameters including some inflammatory markers: serum amyloid-A (SAA), high sensitive C-reactive protein (hs-CRP), fibrinogen, erythrocyte sedimentation rate (ESR) and leukocytes were determined for each patient. Cox regression analysis selected the parameters of univariate and multivariate survival analysis.

Results: During the follow-up period, 37 patients (42.5%) died. Univariate analysis selected the following potential mortality predictors (p < 0.10): age, months on CAPD, residual urine output, presence of cerebrovascular insult (CVI), KT/V, serum urea and albumin concentrations, SAA, hs-CRP, fibrinogen and ESR. In the multivariate survival analysis four models were created, each with a single inflammatory parameter. In all of these models, besides the age and CVI, inflammatory parameters were the most significant mortality predictors. When the inflammatory markers were analyzed altogether, multivariate analysis established that independent mortality predictors in this group of patients were: SAA, age and CVI. Conclusion: It may be concluded that in this studied group treated by CAPD, SAA was the most significant independent mortality predictor among the analyzed inflammatory markers.

• CAPD
• chronic inflammation
• C-reactive protein
• mortality
• serum amyloid-A

Introduction

Although improvements in patient care and dialysis technology have lead to an increase in life expectancy for chronic dialysis patients over the last 40 years, the majority of them die within 5 years of therapy.1 The most common cause of death (about 50%) is cardiovascular disease, which are 10–30 folds more frequent than in the general population in the same age group.2–6

Recently, many studies have indicated some novel risk factors that are far more common in dialysis patients than in the general population7^,8: oxidative stress, hyperhomocysteinemia and chronic inflammation.1 Moreover, inflammatory markers as predictors of mortality in dialysis patients have become an important topic in the nephrology literature.9

In 1995, Bergstrom and colleagues10 first reported an association between increased mortality and C-reactive protein (CRP), a marker of inflammation. Many later studies reached the same conclusion in patients with end stage renal disease (ESRD)11–13 and chronic kidney disease.14 Investigations on other positive acute phase proteins as mortality predictors, such as serum amyloid-A (SAA) and fibrinogen, are scarce, especially in subjects on peritoneal dialysis.

In these patients it is not clear if the cause of inflammation is dialysis per se or uremia per se. Some authors have found that the state of chronic inflammation indicates bio-incompatibility of peritoneal dialysis fluids.15 Also, the decrease of residual renal function was connected with a stronger inflammatory response in peritoneal dialysis patients with higher concentrations of CRP, soluble vascular cell adhesion molecule-1 or fibrinogen.16–18

The aim of this study was to analyze the predictive value of inflammatory parameters: SAA, CRP, fibrinogen, erythrocyte sedimentation rate (ESR) and leukocytes, on mortality of continuous ambulatory peritoneal dialysis patients.

Patients and methods

Patients

The study group was selected from outpatients on continuous ambulatory peritoneal dialysis (CAPD) treatment (4–5 × 2 l exchanges per day) who attended the Nephrology Clinic of the Clinical Center of Serbia in Belgrade for their monthly control. CAPD outpatients were prospectively involved in this investigation. Individuals who received CAPD treatment for more than 3 months (3–113 months), who had no acute inflammatory disease currently or 3 months before (i.e., the patients were clinically asymptomatic, with negative cultures of peritoneal dialysis effluent, urine, throat and nose swabs) or malignant disease and patients who were not “high transporters” were included in the study. Thus, 87 participants (57 males), aged from 30 to 85 [62.9 (10.6)] years were examined. The underlying kidney disorders which had led to ESRD were nephrosclerosis (35 patients), diabetic nephropathy (28 patients), tubulointerstitial nephritis (11 patients), glomerulonephritis (7 patients), polycystic kidney disease (3 patients) and unknown kidney disease (3 patients).

All patients provided written informed consent prior to their enrolment in the study, which was planned according to the ethical guidelines following the Declaration of Helsinki (sixth version). The institutional review committee approved our study protocol thereby following local biomedical research regulations.

Methods

The basal period lasted 3 months followed by monitoring for 30 months. Clinical parameters (cardiovascular morbidity, blood pressure, and residual urine output), dialysis adequacy (KT/V and ∑Ccr) and laboratory parameters, including the inflammatory markers: SAA, high sensitive (hs) CRP, fibrinogen and ESR were obtained for each patient, when they were included in the study. After that the patients had monthly clinical and laboratory checks. Due to limited resources it was not possible to do repeated measurements of SAA.

Clinical parameters

On the day of control all patients underwent a physical examination including determination of body weight (kg) and height (cm). Body mass index (BMI) was calculated according to the formula: weight (kg)/height² (m²).19

Blood pressure was measured twice within 10 min and the mean value was considered as representative for the systolic and diastolic blood pressure of each patient. Pulse pressure was calculated as systolic blood pressure minus diastolic blood pressure. The mean arterial pressure (MAP) was calculated as 2/3 diastolic blood pressure plus 1/3 systolic blood pressure.

Patients with a history or signs of cerebrovascular insult (CVI)—stroke, ischemic heart disease (IHD): myocardial infarction and/or angina pectoris or congestive heart failure (CHF) was considered to have these cardiovascular morbidities. Patients with these cardiovascular morbidities were marked as 1 and the others with 0.

Laboratory methods

Blood samples were obtained from fasting patients on the control day for the following measurements: complete blood count (CBC), serum concentrations of urea, creatinine, albumin, calcium, phosphate, hs-CRP, SAA and fibrinogen and ESR.

Hematological profiles were determined using an LH 750 hematology analyzer (Beckman Coulter Inc., Brea, CA). Creatinine, urea, albumin, calcium and phosphate were analyzed employing routine methods (Olympus System Reagents using an Olympus analyzer AU 2700, Hamburg, Germany). Intact parathyroid hormone (iPTH) in plasma was measured using an electrochemiluminescence method (ESLIA) (Roche-Diagnostics, Mannheim, Germany).

Fibrinogen was measured by the Clauss method (BCS Siemens, Germany). hs-CRP and SAA were assayed immuno nephelometrically (Dade-Behring, BN II, Marburg, Germany). The reference ranges for acute phase proteins were: fibrinogen 1.8–4.1 g/L, hs-CRP 0.00–3.00 mg/L and SAA 0.0–6.8 mg/L.

Dialysis adequacy

Adequacy of dialysis was calculated from 24-h dialysate and urine collections for urea and creatinine. Peritoneal creatinine clearance (pCcr) was corrected to a body surface area of 1.73 m² and urea clearance was expressed as the dialysis dose (KT/V) using the Watson formula for body water. Renal creatinine clearance (rCcr) was calculated from the mean of creatinine and urea clearances corrected to a body surface area of 1.73 m², while total Ccr was calculated as the sum of pCcr and rCcr. Total weekly Ccr and weekly pCcr were expressed in L/1.73 m².20

Statistical methods

Continuous variables are expressed as the mean (standard deviation) if normally distributed or median (interquartile value) for other distributions. Frequencies are given as a percentage (number). Differences in continuous variables between the analyzed groups were tested using Student’s t-test for normally distributed variables and the Mann–Whitney U test for other distributions. Association was evaluated using Pearson’s coefficients for normally distributed variables and Spearman correlation for other distributions. The parameters were tested in the baseline period. The cut-off point for statistical significance was 0.05.

In order to analyze the risk of death, survival analysis was obtained with the Cox proportion hazard model. The outcome was death and all the other patients were censored. The primary dependent variable was the time to death measured in months. Variables that were significant in univariate analysis (p < 0.10) were entered into the forward stepwise Cox proportion hazard model to determine the adjusted hazard ratio (HR) and select the independent predictors of mortality. The Cox proportional hazard model used time-independent covariates and the forward stepwise method. Proportional hazards were tested by fitting time-varying models. The variables were considered possibly collinear if r > 0.5 or ρ > 0.5. Because the analyzed inflammatory parameters were collinear, in the multivariate survival analysis four models were created, each with a single marker. Due to the small patient group it was not possible to analyze different causes of mortality.

All calculations were performed using SPSS 15 software.

Results

In a prospective cohort study, 87 patients (57 males), aged from 30 to 85 years, who had been treated by a chronic program of CAPD for 3–113 months, were analyzed. Demographic data and the primary kidney disease for all patients at the time of inclusion in the study are presented in Table 1. Nephrosclerosis and diabetic nephropathy were responsible for ESRD in 72.4% of the patients.

Table 1. Demographic variables and diagnosis.

Parameter	Value
Sex
Men	57
Women	30
Age (years)^a	62.9 (10.6)
iTime on dialysis (months)^b	30 (15–49)
Kidney disease^c
Nephrosclerosis	40.2 (35)
Diabetic nephropathy	32.2 (28)
Chronic interstitial nephropathy^d	12.6 (11)
Chronic glomerulonephritis	8.0 (7)
Polycystic kidney disease	3.5 (3)
Unknown	3.5 (3)

Notes: ^aMean (SD).

^bMedian (interquartile value).

^cPercentage (count).

^dChronic interstitial nephropathy includes Balkan endemic nephropathy, chronic interstitial nephritis and chronic pyelonephritis.

Clinical and laboratory data and inflammatory markers are shown in Table 2. Although the median level of SAA was in the normal range, 40 patients had SAA levels above upper limit of normal. Fifty-six patients had higher hs-CRP values than normal. SAA showed a wider range of concentrations than hs-CRP (0.7–117.0 mg/L vs. 0.2–53.7 mg/L) (data not shown).

Table 2. Clinical and laboratory data and inflammatory parameters.

Parameters	Value
BMI (kg/m^2)^a	24.7 (3.9)
Residual urine output (mL)^b	500 (300–1000)
Ultrafiltration (mL)^b	1100 (800–1300)
MAP (mmHg)^a	96 (13)
Congestive heart failure (CHF)^c	12.6 (11)
Ischemic heart disease (IHD)^c	22.9 (20)
Cerebrovascular insult (CVI)^c	8.0 (7)
Number of peritonitis^b	1 (0–3)
KT/V^a	2.09 (0.44)
∑Ccr (L/1.73 m^2)^a	66.8 (16.3)
Hemoglobin (g/L)^a	110.9 (12.1)
Urea (mmol/L)^a	18.5 (4.9)
Creatinine (µmol/L)^a	777 (226)
Calcium (mmol/L)^a	2.28 (0.18)
Phosphate (mmol/L)^a	1.56 (0.35)
Albumin (g/L)^a	31.5 (3.6)
iPTH (pg/ml)^b	235 (144–326)
Leukocytes (×10^9/L)^a	8.17 (2.31)
ESR (mm/h)^b	64 (44–92)
Fibrinogen (g/L)^a	6.2 (1.5)
hs-CRP (mg/L)^b	5.6 (1.8–10.7)
SAA (mg/L)^b	6.69 (3.30–18.90)

Notes: KT/V, dialysis dose; Ccr, total weekly creatinine clearance; MAP, mean arterial pressure; BMI, body mass index; iPTH, intact parathyroid hormone; ESR, erythrocyte sedimentation rate; hs-CRP, high sensitive C-reactive protein; SAA, serum amyloid-A.

^aMean (SD).

^bMedian (interquartile value).

^cPercentage (count).

Testing differences among the variables when patients were grouped by sex showed significantly higher creatinine and urea concentrations in men. In addition, men had a significantly lower KT/V index (Table 3). However, there were no significant differences in inflammatory markers, age, dialysis vintage or cardiovascular morbidity between men and women (data not shown).

Table 3. Significant differences among variables in CAPD patients grouped by sex (t-test).

Variables	Men	Women	p
Creatinine (µmol/L)	818 (240)	697 (176)	<0.01
Urea (mmol/L)	19.4 (4.6)	16.6 (5.1)	<0.05
KT/V	1.98 (0.45)	2.31 (0.35)	<0.01

Notes: p, significance of coefficient; KT/V, dialysis dose.

The group of patients with CVI had significantly lower levels of albumin and urea compared with patients without CVI (Table 4). Among inflammatory parameters, only hs-CRP was markedly elevated in patients with CVI when compared to those without CVI. Patients with CHF had significantly higher MAP and fibrinogen values compared with those without CHF (Table 4). There was no significant difference between the groups of patients with and without IHD. Moreover, sex frequencies in the IHD, CHF and CVI groups did not differ significantly from the others (Fisher’s exact test) (data not shown).

Table 4. Significant differences among variables in CAPD patients grouped by cardiovascular morbidity.

Variables	With CVI	Without CVI	Test	p
Cerebrovascular insut (CVI)
Albumin (g/L)	27.4 (3.6)	31.9 (3.4)	t-Test	<0.05
Urea (mmol/L)	13.5 (4.4)	18.9 (4.8)	t-Test	<0.05
hs-CRP (mg/L)	17.5 (11.6)	10.1 (4.8)	Mann–Whitney U test	<0.05
Congestive heart disease (CHF)
MAP (mmHg)	104 (12)	95 (13)	t-Test	<0.05
Fibrinogen (g/L)	7.1 (1.3)	6.0 (1.5)	Mann–Whitney U test	<0.05

Notes: p, significance of coefficient; hs-CRP, high sensitive C-reactive protein; MAP, mean arterial pressure.

Correlation testing revealed marked associations among the inflammatory markers (Table 5). Leukocyte counts correlated with other inflammatory parameters but the associations were not strong. Fibrinogen was collinear with hs-CRP and ESR. SAA was collinear with hs-CRP and correlated with ESR. ESR and hs-CRP were also correlated.

Table 5. Significant correlations among inflammatory markers in the analyzed CAPD patients.

Variables	SAA	hs-CRP	Fibrinogen	ESR
hs-CRP	ρ = 0.62 p < 0.01
Fibrinogen	ρ = 0.69 p < 0.01	ρ = 0.57 p < 0.01
ESR	ρ = 0.49 p < 0.01	ρ = 0.42 p < 0.01	ρ = 0.65 p < 0.01
Leukocytes	ρ = 0.21 p < 0.05	ρ = 0.30 p < 0.01	r = 0.31 p < 0.01	r = 0.27 p < 0.05

Notes: SAA, serum amyloid-A; hs-CRP, high sensitive C-reactive protein; ESR, erythrocyte sedimentation rate; ρ, Spearman correlation coefficient; r, Pearson correlation coefficient; p, significance of coefficient.

Correlations among inflammatory markers and clinical and laboratory variables revealed only a few significant associations: ESR and hemoglobin, fibrinogen and phosphate. However, albumin, as a negative acute phase reactant, correlated significantly with leukocytes, fibrinogen, SAA and hs-CRP (Table 6).

Table 6. Significant correlations among inflammatory markers and clinical and laboratory variables in the analyzed CAPD patients.

Variables	SAA	Hs-CRP	Fibrinogen	ESR	Leukocytes
Hemoglobin	n.s.	n.s.	n.s.	ρ = −0.27 p < 0.05	n.s.
Phosphate	n.s.	n.s.	r = 0.22 p < 0.05	n.s.	n.s.
Albumin	ρ = −0.29 p < 0.01	ρ = −0.29 p < 0.01	r = −0.28 p < 0.01	n.s.	r = −0.23 p < 0.05

Notes: SAA, serum amyloid-A; hs-CRP, high sensitive C-reactive protein; ESR, erythrocyte sedimentation rate; n.s., non-significant; ρ, Spearman correlation coefficient; r, Pearson correlation coefficient; p, significance of coefficient.

During the follow-up period of 30 months, 37 (42.5 %) of our patients died (10 diabetics—11.5%). Fourteen patients (16.1%) stopped peritoneal dialysis and started hemodialysis (most of them due to peritonitis, only one because of a large umbilical hernia), one patient (1.15%) received a transplant and one patient (1.15%) went to another dialysis center. The causes of death were cardiovascular events (heart failure, myocardial infarction, arrhythmia or CVI) in 14 (37.9%) patients, sepsis in 6 (16.2%) patients, malignancy in 4 (10.6%) patients, ulcus ventriculi perforatio in 2 (5.4%) patients, coma hyperglycemia in 1 (2.7%) patient and unknown (at home) in 10 (27%) patients.

Univariate analysis selected potential mortality predictors: age (years), months on dialysis, residual urine output, CVI, KT/V, serum albumin and urea concentrations, SAA, hs-CRP, fibrinogen and ESR (Table 7).

Table 7. Univariate analysis selected the following potential mortality predictors (p < 0.10).

Variable	Hazard ratio (95% CI)	p
Age (years)	1.06 (1.02–1.11)	<0.01
Months on dialysis	1.01 (0.99–1.02)	0.07
Residual urine output	0.99 (0.99–1.00*)	0.02
CVI	6.31 (2.62–15.18)	<0.01
KT/V	0.46 (0.20–1.06)	0.07
Urea	0.92 (0.85–0.99)	0.03
Albumin	0.90 (0.83–0.98)	0.02
SAA	1.01 (1.00–1.02)	<0.01
hs-CRP	1.02 (1.00*–1.03)	<0.01
Fibrinogen	1.34 (1.09–1.65)	<0.01
ESR	1.01 (1.00*–1.02)	0.01

Notes: Hazard ratio, Exp(B); 95% CI, 95% confidence interval of Exp(B); p, significance of coefficient, *, >1.00; CVI, cerebrovascular insult; KT/V. dialysis dose; SAA, serum amyloid-A; hs-CRP, high sensitive C-reactive protein; ESR, erythrocyte sedimentation rate.

As the inflammatory parameters were collinear in the multivariate survival analysis, four models were created (SAA, hs-CRP, fibrinogen, and ESR). In all four models, besides age and CVI, the inflammatory parameters were the most significant mortality predictors (Table 8). Dialysis vintage appeared as an additional parameter in the fibrinogen model, as well as residual urine output in the ESR model.

Table 8. Multivariate analysis separately for four models: SAA, hs-CRP, fibrinogen, and ESR.

Model	Variables	Hazard ratio (95% CI)	p
SAA	Age	1.06 (1.02–1.11)	<0.01
	CVI	5.84 (2.32–14.71)	<0.01
	SAA	1.01 (1.00*–1.02)	<0.01
hs-CRP	Age	1.06 (1.01–1.10)	<0.01
	CVI	5.35 (2.08–13.73)	<0.01
	hs-CRP	1.02 (1.00*–1.03)	0.03
Fibrinogen	Age	1.06 (1.02–1.11)	<0.01
	CVI	5.26 (1.89–14.63)	<0.01
	Months on dialysis	1.01 (1.00–1.03)	0.03
	Fibrinogen	1.41 (1.12–1.78)	<0.01
ESR	Age	1.05 (1.01–1.10)	<0.01
	CVI	6.06 (2.28–16.04)	<0.01
	Residual diuresis	0.99 (0.99–1.00^**)	0.02
	ESR	1.01 (1.00*–1.03)	<0.01

Notes: Hazard ratio, Exp(B); 95% CI, 95% confidence interval of Exp(B); p, significance of coefficient; *, >1.00; **, <1.00; SAA, serum amyloid-A; hs-CRP, high sensitive C-reactive protein; ESR, erythrocyte sedimentation rate; CVI, cerebrovascular insult.

When the inflammatory parameters were analyzed altogether, multivariate analysis established that independent mortality predictors in this group of patients were: age, CVIs and SAA (Table 9). Increasing SAA by 1 mg/L increased peritoneal dialysis patient mortality by 1%.

Table 9. Multivariate analysis – all inflammatory factors together.

Variables	Hazard ratio (95% CI)	p
Age (years)	1.06 (1.02–1.11)	<0.01
CVI	5.83 (2.31–14.68)	<0.01
SAA	1.01 (1.00*–1.03)	<0.01

Notes: Hazard ratio, Exp(B); 95% CI, 95% confidence interval of Exp(B); p, significance of coefficient; *, >1.00; CVI, cerebrovascular insult; SAA, serum amyloid-A.

Discussion

In our present study 87 patients receiving CAPD treatment were monitored for 30 months in order to analyze the predictive value of inflammatory parameters on their mortality. Correlation testing revealed marked associations among the inflammatory markers. Among all analyzed parameters univariate analysis selected the following potential mortality predictors (p < 0.10): age, months on CAPD, residual urine output, presence of CVI, KT/V, serum urea and albumin concentrations, SAA, hs-CRP, fibrinogen and ESR (Table 7). In the multivariate survival analysis four models, each with a single inflammatory parameter, were created. In all of these models, besides age and CVI, inflammatory parameters were the most significant mortality predictors (Table 8). When the inflammatory markers were analyzed altogether, multivariate analysis established that among inflammatory parameters the independent mortality predictor was SAA (Table 9).

SAA and CRP belong to group 3 acute phase proteins and their serum concentrations are significantly elevated (few hundred times) in acute inflammation. SAA and hs-CRP usually increase in parallel, although SAA may be a more sensitive inflammatory marker.21^,22 As in the experimental study of Christensen and colleagues,23 SAA also showed a wider range of concentration than hs-CRP in our study.

Investigations in the general population,24^,25 as well as in nephrology patients,11^,26 indicated that CRP was a strong predictor of cardiovascular mortality and overall mortality.27 In nephrology patients SAA is a less commonly used marker of inflammation than CRP, maybe because SAA assays were not widely accessible in the past. A few studies have compared CRP and SAA concentrations in peritoneal dialysis patients28 and hemodialysis patients.1^,29 Simic-Ogrizovic and colleagues1 showed that SAA was a better mortality predictor than hs-CRP in patients on hemodialysis, together with hypoalbuminemia and hyperphosphatemia. Moreover, Zimmerman and colleagues13 and Hung and colleagues,30 found that CRP was a better predictor of cardiovascular mortality than SAA, but they measured SAA by ELISA. Other authors, including Simic-Ogrizovic and colleagues1 used nephelometric assays which have higher quality, great sensitivity, excellent reproducibility and rapid automation for both hs-CRP and SAA.31 We also measured hs-CRP and SAA by nephelometry.

Our multivariate analysis in four models, one for each inflammatory parameter showed that, besides age and CVI, these variables were the most significant mortality predictors. However, dialysis vintage appeared in the fibrinogen model as an additional parameter, as well as residual urine output in the ESR model (Tables 8 and 9). When all inflammatory factors were analyzed together, multivariate analysis established that the independent mortality predictors for this group of patients were: SAA, age and CVI (Table 9).

Although proinflammatory cytokines like IL-1, IL-6, TNF-ά have much better prognostic utility when compared with acute phase response proteins,32^,33 Zoccali and colleagues34 found that the cost of measuring IL-6 is about four times that of determining hs-CRP. However, as in the study of Simic-Ogrizovic and colleagues1 on hemodialysis patients, SAA was a better predictor of mortality than hs-CRP in our subjects on peritoneal dialysis.

The nephrology literature35–37 shows that hypoalbuminemia is a well known independent predictor of mortality in patients with ESRD, which was also found here with univariate analysis (Table 7). Hypoalbuminemia is also seen in cardiovascular disease (as found in our CVI patients) (Table 4) and it is considered to be related to malnutrition and inflammation.38 In multivariate analysis hypoalbuminemia was one of the best mortality predictors in ESRD patients on hemodialysis in two studies by Simic-Ogrizovic and colleagues1^,39 as well as in many other investigations.26^,35^,40^,41 In our study, hypoalbuminemia was not selected as a mortality predictor in multivariate analysis.

Cardiovascular diseases often appear as a risk factor for mortality.3–6^,35^,41 However, in our study both univariate analysis and multivariate analysis selected only CVI as a mortality predictor (Tables 7–9).

Correlations among inflammatory parameters were strong, as can be seen from our results (Table 5), so they were all important markers of inflammation in these patients. Negative correlations between inflammatory parameters and albumin (Table 6) are also expected, because albumin is not only a nutritional parameter but also a negative acute phase reactant.38 However, in spite of their very important role in the mortality of our participants, inflammatory parameters correlated only with serum phosphate among the clinical parameters (Table 6). No association was found with adequacy of dialysis, residual urine output or hypertension and other parameters except for the expected negative correlation between ESR and hemoglobin (Table 6). Association of inflammatory parameters (CRP) with phosphate has already been reported42 but in our study only fibrinogen was significantly correlated with phosphate (Table 6).

During the analyzed period 37 (42.5%) patients died (11.5% diabetics). The most common causes of mortality were cardiovascular events (37.9%). Unsal and colleagues40 found that 85.2% patients on peritoneal dialysis survived after 1 year of dialysis and 66.5% after 3 years, which is better than in our patients after 2.5 years (57.5%). In another study concerning patients on peritoneal dialysis Yang and colleagues41 reported that 72% diabetics and 92% non-diabetics survived after 2 years, and 63% diabetics and 87% non-diabetics survived after 3 years. Survival after 2.5 years in our diabetic and non-diabetic patients was: 54.3% versus 64.3%. Simic-Ogrizovic and colleagues1 found greater 3 year-survival in hemodialysis patients (75.1%) probably because our peritoneal dialysis patients were older with many comorbidities even at the beginning of peritoneal dialysis, and most of them had diabetic nephropathy and nephrosclerosis (72.2%) (Table 1).

A drawback of this study is the relatively small number of patients for survival analysis. However, this analysis was reliable according to the narrow confidence intervals of the inflammation parameters and age.

All the analyzed inflammatory parameters were independent mortality predictors except for leukocytes. However, in the studied CAPD patients SAA was the most significant independent mortality predictor among the analyzed inflammatory factors. Besides SAA, age and the presence of CVI were also independent mortality predictors in our group of peritoneal dialysis patients. As assays for SAA are now more available, SAA could be more widely used as a predictable inflammatory marker in peritoneal dialysis patients.

Declaration of interest

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

This work was supported by the Ministry of Science and Technology, Republic of Serbia, Contract No. 175089.