Inverse Relationship between the Inflammatory Marker Pentraxin-3, Fat Body Mass, and Abdominal Obesity in End-Stage Renal Disease

Abstract

Summary

Background and objectives

Pentraxin-3 (PTX3) belongs to the same pentraxin superfamily of acute-phase reactants as C-reactive protein (CRP). Abdominal fat accumulation in ESRD is considered a chronic inflammatory state, but the relationship of PTX3 to this phenomenon is unknown. This study assesses plausible associations between PTX3 and surrogates of fat mass deposits in dialysis patients.

Design, setting, participants, & measurements

Circulating levels of PTX3, CRP, and IL-6 were cross-sectionally analyzed in relation to anthropometric and nutritional surrogate markers of fat tissue in two cohorts comprising 156 prevalent hemodialysis (HD) and 216 incident dialysis patients.

Results

In both cohorts, PTX3 was negatively associated with body mass index (BMI) and fat body mass index (FBMI) derived from anthropometrics and leptin, whereas there was a positive association with adiponectin. In prevalent HD patients, those with larger waist circumference (above gender-specific median values) had lower PTX3, higher CRP, and higher IL-6 levels. This was also true in multivariate analyses. In both cohorts, multivariate regression analyses showed that PTX3 was negatively and CRP (or IL-6) was positively associated with FBMI.

Conclusions

Although CRP and IL-6 were directly associated with body fat, PTX3 levels showed negative correlations with surrogates of adipose tissue in two independent cohorts of ESRD patients. Understanding the underlying reasons behind these opposite associations may have clinical relevance given the survival advantage described for obese patients on dialysis.

Introduction

Obesity, specifically abdominal obesity, is nowadays regarded as a proinflammatory state that contributes to increased cardiovascular mortality in the general population (1,2). Some but not all studies indicate that this may also be the case in chronic kidney disease (CKD) patients and especially in ESRD (3–7). Pentraxin-3 (PTX3) is a multifunctional soluble pattern recognition receptor modulating the immunoinflammatory response and it belongs to the same pentraxin superfamily of acute-phase reactants as C-reactive protein (CRP) (8,9). However, PTX3 and CRP are produced by different tissues and may be involved in different pathophysiological mechanisms (10). Whereas hepatocytes are the main source of CRP, PTX3 is produced at sites of inflammation by a wide range of different cell types, including endothelial cells and adipocytes (8,9,11). Thus, plasma PTX3, similar to CRP, is positively associated with adhesion molecules and endothelial dysfunction in CKD patients (12–15) and like CRP it is also linked to increased cardiovascular and overall mortality risk (12–15). However, the relationship between PTX3 and the chronic inflammatory state of obesity in CKD remains unknown. Because several different cell types within adipose tissue can produce PTX3 (11), we hypothesized that PTX3 could be a more sensitive marker of inflammation-induced obesity than systemic CRP. Against this background, we studied the association between PTX3 and fat-related parameters in two cohorts of patients: one treated with maintenance hemodialysis (HD) and another with CKD stage 5 patients starting dialysis.

Materials and Methods

Subjects and Study Design

This is a cross-sectional observational study performed in two patient cohorts. The first cohort is composed of 156 prevalent patients undergoing maintenance HD (HD patients, 89 men, median 67 [10th to 90th percentile, 44 to 80] years old, median dialysis vintage 28 [7 to 101] months). The protocol design of this patient material has been described elsewhere in more detail (16,17); in the same patient cohort we observed an association between PTX3 and mortality (15). Patient recruitment was performed at five dialysis units in Stockholm and one at the Uppsala Academic Hospital in Uppsala, Sweden. From the 224 patients included in this cohort, the analysis presented here comprises the 156 patients in whom data on waist circumference (WC), fat body mass index (FBMI), PTX3, IL-6, and CRP were available. Data on WC were the main limiting factor. There were no major differences in clinical and demographic characteristics between the included and the 68 excluded patients (data not shown). The second cohort, with recruitment still ongoing (18,19), is composed of 216 CKD stage 5 patients (128 men, 56 [36 to 68] years old, median GFR 6.3 [3.7 to 9.0] ml/min) referred to start dialysis therapy at the Department of Renal Medicine at Karolinska University Hospital Huddinge in Stockholm, Sweden. In this patient cohort, we also found an association between PTX3 and mortality (12). Among 276 consecutive patients, the analysis presented here comprises the 216 patients in whom data on FBMI, PTX3, IL-6, and CRP were available. Data on FBMI were the main limiting factor. There were no major differences in clinical and demographic characteristics between the included and the excluded 60 patients (data not shown). Inclusion criteria in both cohorts were the absence of clinical signs of acute infection, active vasculitis, hepatitis, or HIV at the time of evaluation and willingness to participate in the study. Additionally, only subjects ≤70 years of age were included in the incident cohort, whereas no age limit was considered in the prevalent cohort. However, only patients on HD >3 months were included in the prevalent cohort. Most patients were on antihypertensive medications as well as phosphate and potassium binders, diuretics, and vitamins B, C, and D supplementation in accordance with the clinical judgment of the treating physician. All subjects gave written informed consent, and the local ethical committees of the involved hospitals approved the protocols.

Anthropometric Evaluation and Nutritional Status

Body weight, body mass index (BMI; in kg/m²), and anthropometric measurements were taken on a dialysis day immediately after the dialysis session for the prevalent HD patients, and, for the incident dialysis patients, at the time of or within 1 week of blood sample collection. Fat mass was assessed according to the formula of Durnin et al. (20), which is based on the four skinfold thicknesses (biceps, triceps, subscapular, and suprailiac), and was measured with a skinfold caliper (Cambridge Scientific Instruments, Cambridge, MD). Fat BMI (FBMI, in kg/m²) was calculated from these values. WC was measured at a level midway between the lowest lateral border of the ribs and the uppermost lateral iliac crest. Subjective global nutritional assessment (21) was done on the same occasion as blood sampling and used as a surrogate of protein-energy wasting (defined as a subjective global nutritional assessment score ≥2).

Blood Sampling and Biochemical Measurements

Blood samples were collected before the HD session after the longest interdialytic period for the prevalent HD patients or under fasting conditions early in the morning for the incident dialysis patients and stored at −70°C if not analyzed immediately. In both cohorts, plasma PTX3 was determined from frozen samples by and ELISA kit (Perseus Proteomics, Tokyo, Japan). The intra-assay coefficient of variation was 4.1%, and the interassay coefficient of variation was within 4.3%. Cross-reactivity with CRP and serum amyloid protein was <0.1 ng/ml. High-sensitivity CRP was measured by the nephelometry method. IL-6 was quantified in serum by an immunometric assay on an Immulite Analyzer according to the instructions of the manufacturer (Siemens Medical Solutions Diagnostics, Los Angeles, CA). Determinations of serum albumin (bromcresol purple), cholesterol, triglycerides, and HDL cholesterol were performed by routine procedures in the Department of Clinical Chemistry at Karolinska University Hospital or Uppsala Academic Hospital. Additionally, serum leptin levels in both cohorts and plasma adiponectin levels in the cohort of HD patients were measured by a RIA kit (Linco Research, Inc., Saint Charles, MO). Plasma adiponectin levels in the cohort of incident dialysis patients were determined by ELISA (Linco Research, Inc.). GFR in incident dialysis patients was calculated by the mean of renal urea and creatinine clearances from a 24-hour urine correction.

Statistical Analyses

Values are expressed as median (10th to 90th percentile) or percentage, as appropriate. Statistical significance was set at the level of P < 0.05. Spearman rank correlation analysis was used to determine associations between two parameters. Comparisons between two groups were assessed with a nonparametric Mann–Whitney test for continuous variables and a chi-squared test for nominal variables. Differences among four groups were analyzed by a nonparametric Kruskal–Wallis test. Linear regression models were fitted to assess determinants of FBMI and WC. These models included PTX3, CRP, or IL-6 and potential confounders of their association with fat deposits. Other possible confounders considered are age and gender (both influencing FBMI and WC) as well as cardiovascular disease and diabetes (influencing both inflammation and body composition) and finally GFR or dialysis vintage, depending on the patient material. CRP, IL-6, and PTX3 values were log₂ transformed. All analyses were performed using STATA version 11.1 (Stata Corporation, College Station, TX).

Results

General characteristics and laboratory data for the two cohorts studied are depicted in Table 1 with univariate correlates of PTX3. The median levels of plasma PTX3 were 10.5 (5.7 to 22.9) ng/ml in prevalent HD patients and 5.5 (2.3 to 15.1) ng/ml in the incident dialysis cohort. In univariate analysis, PTX3 showed significant negative associations with BMI, FBMI, leptin, and serum albumin, whereas it was positively associated with adiponectin in both cohorts. Additionally, WC also negatively associated with PTX3 in the prevalent HD cohort. In both cohorts, PTX3 associated positively with CRP and IL-6 levels.

Table 1.

General characteristics and univariate Spearman correlations with circulating PTX3

	Prevalent Hemodialysis Patients		Incident Dialysis Patients
	n = 156	ρ (P)a	n = 216	ρ (P)a
Age (years)	67 (44 to 80)	0.14 (0.09)	56 (36 to 68)	0.16 (0.02)
Gender (% men)	57	–	59	–
Diabetes mellitus (%)	22	–	32	–
Cardiovascular disease (%)	61	–	39	–
Dialysis vintage (months)	28 (7 to 101)	0.04 (0.63)	NA	NA
GFR (ml/min)	NA	NA	6.3 (3.7 to 9.0)	−0.10 (0.16)
Body mass index (kg/m2)	23.9 (18.8 to 31.1)	−0.35 (<0.001)	24.0 (19.8 to 29.9)	−0.14 (0.04)
Fat body mass index (kg/m2)	7.9 (4.4 to 12.4)	−0.28 (<0.001)	5.7 (4.5 to 12.0)	−0.17 (0.01)
WC (cm)	100 (84 to 117)	−0.23 (0.004)	NE	NE
Wasting (SGA ≥ 2, %)	47	–	29	–
Serum albumin (g/L)	35 (29 to 40)	−0.38 (<0.001)	33 (26 to 40)	−0.39 (<0.001)
Cholesterol (mmol/L)	4.3 (3.0 to 5.7)	0.03 (0.70)	4.9 (3.4 to 7.2)	−0.05 (0.50)
Triglyceride (mmol/L)	1.5 (0.8 to 3.3)	−0.20 (0.01)	1.8 (0.9 to 3.5)	−0.10 (0.15)
HDL cholesterol (mmol/L)	1.3 (0.9 to 2.1)	0.28 (<0.001)	1.3 (0.8 to 2.0)	0.05 (0.46)
Leptin (ng/ml)	17.1 (3.7 to 113.9)	−0.25 (0.002)	12.0 (2.7 to 54.8)b	−0.18 (0.02)
Adiponectin (ng/ml)	20.7 (9.6 to 37.5)	0.25 (0.002)	25.0 (11.4 to 74.3)c	0.24 (0.01)
PTX3 (ng/ml)	10.5 (5.7 to 22.9)	–	5.5 (2.3 to 15.1)	–
CRP (mg/L)	6.0 (0.9 to 34.6)	0.19 (0.02)	4.8 (0.6 to 43.2)	0.36 (<0.001)
IL-6 (pg/ml)	8.5 (3.3 to 24.7)	0.24 (0.002)	6.7 (2.2 to 18.6)	0.37 (<0.001)

Data presented as median (10th to 90th percentile); NA, not applicable; NE, nonexistent; PTX3, pentraxin-3; CRP, C-reactive protein; WC, waist circumference.

^aSpearman univariate correlation with PTX3 levels.

^bn = 159.

^cn = 125.

Measurements of WC were only available in the prevalent HD material. In univariate analysis, PTX3 associated negatively with WC (ρ = −0.23; P = 0.004), whereas CRP showed a positive nonsignificant trend (ρ = 0.15; P = 0.08) and IL-6 presented a positive significant association (ρ = 0.19; P = 0.01). When patients were dichotomized according to gender-specific median values of WC, individuals with a large WC (≥101 cm [for men], ≥97 cm [for women]) presented at the same time significantly lower PTX3 (P = 0.01; Figure 1), higher CRP (P = 0.03; Figure 1), and higher IL-6 (P = 0.01; Figure 1) levels. The opposing associations between PTX3 and CRP (or IL-6) with WC persisted after adjustment for age, gender, diabetes mellitus, cardiovascular disease, and dialysis vintage in multiple regression models (Table 2).

Figure 1.

Plasma pentraxin-3 (PTX3), C-reactive protein (CRP), and IL-6 levels in relation to gender-specific waist circumference (WC) categories (≤101 cm [men], and ≤97 cm [women]) in 156 prevalent hemodialysis patients. Seventy-one patients were included in the low WC group and 85 in the high WC group.

Table 2.

Multiple regression model predicting waist circumference (per centimeter of increase) in 156 prevalent hemodialysis patients including either CRP or IL-6

	β	Coefficient (95% CI)	p-values
Including CRP (adjusted r2 = 0.16)
age, per year	0.065	0.07 (−0.09;0.23)	0.40
gender, male	0.069	2.06 (−2.30;6.42)	0.35
diabetes, presence	0.338	11.90 (6.64;17.29)	<0.0001
cardiovascular disease, presence	0.010	0.26 (−4.43;4.96)	0.91
dialysis vintage, months	−0.037	−0.01 (−0.05;0.03)	0.62
PTX3, per doubling ng/ml	−0.208	−3.52 (−6.72;−0.28)	0.008
CRP, per doubling mg/L	0.138	0.99 (−0.10;2.01)	0.060
Including IL-6 (adjusted r2 = 0.19)
age, per year	0.037	0.04 (−0.12;0.20)	0.63
gender, male	0.064	1.91 (−2.38;6.21)	0.38
diabetes, presence	0.318	11.26 (5.98;16.53)	<0.0001
cardiovascular disease, presence	−0.022	−0.67 (−5.40;4.04)	0.77
dialysis vintage, months	−0.051	−0.01 (−0.06;0.03)	0.49
PTX3, per doubling ng/ml	−0.227	−3.84 (−7.07;−0.61)	0.004
IL-6, per doubling pg/ml	0.222	2.68 (0.98;4.38)	0.007

PTX3, IL-6, and CRP were log₂ transformed in both models. CI, confidence interval; CRP, C-reactive protein; PTX3, pentraxin-3.

In each patient cohort, FBMI was concomitantly associated with lower PTX3 and higher CRP levels in univariate analysis (Table 1), which also held true in multivariate models including both markers as well as age, gender, presence of diabetes and cardiovascular disease, and GFR or dialysis vintage (Table 3). As a sensitivity analysis, when multivariate models were repeated including IL-6 instead of CRP, the statistically significant inverse association between FBMI and PTX3 was maintained (data not shown).

Table 3.

Multiple regression model predicting fat body mass index (per kg/m² of increase) in prevalent and incident dialysis patients

	β	Coefficient (95% CI)	P
Prevalent hemodialysis, n = 156 (adjusted r2 = 0.30)
age, per year	−0.060	−0.01 (−0.04; 0.01)	0.40
gender, male	−0.403	−2.63 (−3.51; −1.75)	<0.0001
diabetes, presence	0.269	2.08 (1.02; 3.15)	<0.0001
cardiovascular disease, presence	0.084	0.55 (−0.39; 1.50)	0.24
dialysis vintage, months	−0.094	−0.01 (−0.01; 0.01)	0.17
PTX3, per doubling ng/ml	−0.289	−1.08 (−1.68; −0.48)	<0.0001
CRP, per doubling mg/L	0.162	0.26 (0.02; 0.50)	0.02
Incident dialysis, n = 216 (adjusted r2 = 0.31)
age, per year	0.344	0.08 (0.05; 0.11)	<0.0001
gender, male	0.325	1.94 (1.20; 2.67)	<0.0001
diabetes, presence	0.168	1.05 (0.26; 1.85)	0.09
cardiovascular disease, presence	0.006	0.04 (−0.81; 0.88)	0.93
GFR, ml/min	−0.005	−0.01 (−0.16; 0.15)	0.92
PTX3, per doubling ng/ml	−0.212	−0.61 (−0.98; −0.25)	0.002
CRP, per doubling mg/L	0.147	0.20 (0.01; 0.40)	0.031

PTX3 and CRP were log₂ transformed. CI, confidence interval; PTX3, pentraxin-3; CRP, C-reactive protein.

Discussion

PTX3 is produced by various types of cells and increases rapidly in response to primary local activation of inflammation and innate immunity (8–11). This specificity opposes that of its homologous CRP, which is exclusively synthesized by the liver mainly in response to IL-6 and reflecting the overall inflammatory status. Because abdominal adipose tissue serves as a crucial site in the generation of inflammatory responses and mediators, we expected to find a positive association between PTX3 and fat-related markers in CKD patients. Instead, and contrary to our hypothesis, this study reports inverse relationships between PTX3 and BMI, FBMI, and WC. Given the presumed proinflammatory nature of PTX3, such results were unexpected, although mathematically it is in the nature of mutual correlations that, in the case of three factors, the correlation between two pairs could be positive and one pair negative. Such novel observation in ESRD patients is nevertheless in line with previous small studies in nonrenal populations: Yamasaki et al. (22) reported an inverse association between PTX3 and BMI in healthy Japanese individuals, and a brief report (23) in healthy populations describes opposed associations of PTX3 and CRP in relation to components of the metabolic syndrome. Our study adds the value of confirming our findings in two independent CKD patient cohorts and demonstrating these inverse associations with different surrogates of total and abdominal adiposity, even in multivariate models that include, among others, the inflammatory marker CRP. Interestingly, during 2 weeks of bed rest in healthy volunteers, longitudinal changes in fat mass directly correlated with CRP and IL-6 but inversely correlated with changes in PTX3 (24). However, this evidence is opposed by the findings of Shim et al. (25), who assessed markers of total, visceral, and abdominal fat by anthropometrics and computed tomography in 40 subjects with ST-segment elevation myocardial infarction. Although BMI or WC did not associate with PTX3 in that study, visceral and abdominal fat by computed tomography were positively associated with circulating PTX3 levels in univariate analysis. No association was seen with subcutaneous fat. The lack of multivariate adjustment in that study raises the question of whether other factors may have confounded the reported associations. It is also possible that the association of this acute-phase protein with fat deposits may have been influenced by the blood sampling during the acute phase of myocardial infarction. Altogether, this evidence illustrates that we still do not understand the mechanisms that regulate PTX3 or the biologic role of PTX3.

Cross-sectional studies in patients with CKD have shown that PTX3 levels increase along with renal function loss and share strong independent links with endothelial dysfunction and albuminuria (12–15,26). In dialysis patients, increased PTX3 levels strongly predicted outcome and did so independently of traditional risk factors and CRP (12–15). Pharmacologically, short-term angiotensin converting enzyme inhibitor treatment significantly improved flow-mediated dilation while normalizing PTX3 and urinary protein excretion in patients with type 2 diabetes with proteinuria (27), whereas the improvement in flow-mediated dilation after treatment with antihypertensive drugs in diabetic CKD stage 1 patients with hypertension was independently associated with PTX3 normalization (28). Altogether, these results suggest that increased PTX3 may play a role in the pathogenesis of endothelial dysfunction/cardiovascular disease.

Interestingly, recent observations have led some authors to postulate that the increase in PTX3 should be regarded as an attempt to counterbalance the overactivation of the proinflammatory cascade rather than as a harmful response. Indeed, mice with induced PTX3 genes are more resistant to endotoxic shock induced by lipopolysaccharide and to polymicrobial sepsis caused by cecal ligation and puncture (29). PTX3-deficient mouse models of acute myocardial infarction caused by coronary artery ligation developed more myocardial damage, and this phenotype was reversed by exogenous PTX3 (30). Double knockout mice lacking ApoE and PTX3 displayed an increment in aortic lesion size and a higher inflammatory response as compared with ApoE knockout mice expressing PTX3 (31). Taken together, PTX3 may play tissue-protective and anti-inflammatory roles. In human studies, endurance exercise, which has been shown to improve cardiovascular profile, was also associated with an acute increase in systemic PTX3 in healthy volunteers (32). HDL cholesterol stimulated PTX3 mRNA expression and protein release in endothelial cells, leading the authors to speculate on an involvement of PTX3 in the atheroprotective effects of HDL cholesterol (33). The positive correlation between HDL cholesterol and PTX3 observed in our study may agree with this. Also, and consistent with a previous report in healthy men (23), our study shows a direct relationship between PTX3 and adiponectin, an adipokine with purported antiatherosclerotic properties (34,35). In the general population, systemic clinical hypoadiponectinemia is closely associated with obesity, type 2 diabetes, and coronary artery disease, and the increased oxidative stress in adipose tissue that accumulates in obesity associates with decreased adiponectin production (36). In CKD patients, reduction in oxidative stress through blockade of the angiotensin II type 1 receptor ameliorates adiponectin concentration (37), and oxidative stress is an important determinant of low fat adiponectin expression (38). A similar oxidative stress-related mechanism might also, at least in part, affect PTX3 production in adipocytes. Whether these reported effects may contribute to explain the observed negative associations with surrogates of fat mass in our study requires further investigation. We hope that the description of this unexpected but robust finding may trigger further mechanistic studies on the topic.

Although confirmation of our findings in two carefully phenotyped materials strengthens the analysis, the cross-sectional study design cannot denote causality in the observed associations. The lack of WC data in the incident dialysis cohort is a limitation, and in future studies it would be interesting to assess these relationships in cohorts with alternative methods to estimate abdominal fat, such as a dual-energy x-ray absorptiometry scan or computed tomography. Finally, and to conclude, this study shows that although CRP and IL-6 were directly associated with body fat, PTX3 levels exhibited negative correlations with surrogates of adipose tissue in two independent cohorts of ESRD patients. Understanding the mechanisms behind this observation may have clinical and/or therapeutic relevance given the survival advantage described for obese patients on dialysis.

Disclosures

Baxter Healthcare Corporation employs Bengt Lindholm. Peter Stenvinkel is a member of the scientific advisory board of Gambro AB. The other authors have no conflicts of interest to declare.