, Introduction
Peritoneal dialysis (PD) is an effective treatment to remove uremic toxins, maintain homeostasis, and improve the outcome of patients with end-stage renal disease (ESRD). However, the dialysis efficacy is dependent on the peritoneal structural characteristics that controls the rate of water and solute transport across the semipermeable peritoneal membrane. Peritoneal membrane inflammation plays a critical role in the regulation of peritoneal solute transport and local defense; thus, it can also predict adverse clinical outcomes such as PD-related peritonitis, which causes technique failure and death [1]. Identification of novel biomarkers that can both predict peritoneal inflammation, peritoneal transport characteristics, and outcomes helps in risk stratification and guides individualized management to ameliorate clinical outcomes, as recommended by the European Training and Research in Peritoneal Dialysis Network [2] and International Society for Peritoneal Dialysis (ISPD) [3].
As the peritoneal membrane is surrounded by layers of adipocytes that secrete inflammation-modulating adipokines, the adipokine profiling may be relevant in the peritoneal structure, functionality, and dialysis outcomes [4,5]. Zinc alpha-2-glycoprotein (ZAG) is a lipid-mobilizing adipokine that plays a pivotal role in the inflammatory cascade [6,7]. ZAG level is significantly elevated in patients with ESRD [8–10] and its circulating level predicted survival in ESRD patients [11,12]. Although ZAG could be identified in PD effluent (PDE) [3], its level at adipose tissue and PDE, and their clinical significance towards peritoneal characteristics and outcomes, are not adequately explored. Moreover, the data of circulating ZAG should not be directly extrapolated to model the peritoneal inflammation as systemic and peritoneal inflammation are separate disease entities [13].
The objectives of our study are to quantify ZAG levels at PDE, adipose, and serum samples, to explore their relationship with peritoneal transport state, and to determine their clinical value and relevance to predict the incidence of PD-related peritonitis and peritonitis-free survival in a cohort of patients newly started on PD.
Methods
Study design
This is a single-center prospective cohort study, which is the extension of another study of our unit [12]. The study was approved by the Joint Chinese University of Hong Kong–New Territories East Cluster Clinical Research Ethics Committee (reference No. CREC-2008.554). All study procedures were in compliance with the Declaration of Helsinki. Consecutive adult patients planned for PD were recruited from 1 January 2011 to 31 December 2013 at Prince of Wales Hospital, a university-affiliated tertiary care hospital. Written informed consent was obtained from all subjects.
Peritoneal transport characteristics
A standard peritoneal equilibrium test (PET) by the method of Twardowski et al. [14] was performed at approximately 4 weeks after PD training. Patients must be at euvolemic state (i.e., without features of volume overload such as pedal oedema, shortness of breath, and less than 1.1 L of over-hydration measured by bioimpedance spectroscopy [15]), and peritonitis was excluded by clinical examination and PDE cell count. The peritoneal solute transport rate (PSTR) was evaluated by the three parameters: (1) dialysate-to-plasma ratio of creatinine at 4 hours (D/P4) after correction of glucose interference [16], (2) the mass transfer area coefficients (MTAC) of creatinine normalized for body surface area [17], and (3) plasma protein clearance by the peritoneal membrane (PPCl) [18]. Patients were classified as high, high-average, low-average, and low peritoneal transporter if their D/P4 values were below 0.5, 0.65, 0.8, and above 0.8, respectively.
Specimen collection
After consent, approximately 2 g of pre-peritoneal adipose tissue specimen was obtained during the insertion of a PD catheter by mini laparotomy. The specimen was then processed immediately and stored at –80 °C overnight. Serum and PDE samples were collected during the PET session. These samples were sent to the laboratory for processing immediately or stored at 4 °C overnight. PDE was further centrifuged at 3,000×g for 30 minutes at 4 °C prior to analysis.
Detection of zinc alpha-2-glycoprotein
The methods of RNA extraction were described previously [19]. In short, ZAG messenger RNA expression in the adipose tissue was measured by the real-time quantitative polymerase chain reaction (PCR), using the Applied Biosystems Step One Plus system. Commercially available TaqMan primers and probes, including two unlabeled PCR primers and one fluorescein amidite dye-labeled TaqMan minor groove binder probe, were used (all from Applied Biosystems). The phosphoglycerate kinase-1 (Applied Biosystems) was used as the housekeeping gene. Results were analyzed with Sequence Detection Software version 2.0 (Applied Biosystems), and the relative quantification method by ∆∆Ct was applied for expression of targets in fold compared to the expression detected in samples from healthy subjects.
Serum and PDE ZAG levels were measured by the commercially available enzyme-linked immunosorbent assay (ELISA) kit (Alpha-2 Glycoprotein/Glycoprotein/ZAG/A2GP1 Human ELISA kit, BMS 2201; Invitrogen) following the manufacturer’s instructions. All assays were performed in duplicate. The detection limit of ZAG was 0.174 ng/mL.
Clinical parameters
Clinical information including demographics and laboratory results were obtained by chart review. To evaluate the systemic inflammation, we measured the serum albumin by the bromocresol purple method, and high-sensitive C-reactive protein (hsCRP), ferritin, hemoglobin, and lipid profile at baseline. Comorbidity load was assessed by the Charlson comorbidity index (CCI) [20]. We also assessed the carotid-femoral pulse wave velocity (PWV), which reflects arterial stiffness [21]. It was measured by a validated Vicorder device (SMT Medical GmbH & Co.) and analyzed by the Complior Analyse program (Artech Medical). Pressure sensors were placed over the neck (carotid artery) and groin (femoral artery), and the PWV was calculated by dividing the distance between the sensors by the time corresponding to the period separating the start of the rising phase of the carotid and femoral pulse wave.
Outcome measures
All subjects were followed for 2 years. The overall clinical management was decided by the attending physician and was not affected by the study. The primary outcome was 2-year peritonitis rate as defined by the ISPD [22] and was expressed as the number of episodes per patient-year of follow-up. The secondary outcome was peritonitis-related hospitalization, peritonitis-free, and overall survival. Receiving kidney transplantation, permanent conversion to hemodialysis, recovery of renal function, loss to follow-up, and transfer to another dialysis center were considered censoring events.
Statistical analysis
Statistical analysis was performed by IBM SPSS for Mac version 27 (IBM Corp.). Descriptive data were presented as mean ± standard deviation. Baseline clinical parameters and peritonitis rate were compared by the Student t test, one-way analysis of variance, Mann-Whitney U test, Kruskal-Wallis test, and chi-square test, while the correlations were analyzed by Pearson and Spearman rank correlation as appropriate. Kaplan-Meier plots were constructed for peritonitis-free survival with the comparison by the log-rank test. Multivariate linear and Cox regression models were then constructed to identify significant predictors of peritoneal transporter characteristics, peritonitis, peritonitis-associated hospitalization, and peritonitis-free survival. Known contributing factors were added to these models. The final p-value of <0.05 was considered as statistically significant. All probabilities were two-tailed.
Results
Patient characteristics
We recruited 148 consecutive patients in the study. Twenty-two patients were excluded as their PDE were unavailable for ZAG quantification. Therefore, 126 patients were enrolled and followed up (Fig. 1). Their mean age was 58.4 ± 11.7 years, and 96 (76.2%) were male. The average ZAG levels in serum, PDE, and adipose tissue samples were 74.4 ± 21.5 μg/mL, 1.0 ± 1.2 μg/mL, and 15.3 ± 140.5-fold, respectively. PDE ZAG correlated with adipose ZAG (r = 0.207, p = 0.02), but not with serum ZAG (p = 0.30). Their baseline clinical characteristics and details of dialysis therapy are summarized and compared in Tables 1 and 2. In short, patients with high PDE ZAG had a higher D/P4, MTAC, and a lower serum albumin level. PDE ZAG also correlated weakly with serum hsCRP and albumin (Supplementary Table 1, available online).
Peritoneal characteristics
The average D/P4, MTAC, and PPCl were 0.69 ± 0.12, 10.8 ± 5.0 mL/min/1.73 m2, and 89.1 ± 40.8 mL/day, respectively (Table 2). Nine (5.4%) and 25 patients (15.0%) were classified as low and high peritoneal transporter respectively. We identified strong and significant correlations between PDE ZAG with D/P4 (p < 0.001) and MTAC (p < 0.001), but not with PPCl (p = 0.2) (Table 3). There was a graded increase in PDE ZAG level across peritoneal transporter state (p = 0.002) (Fig. 2). On the other hand, adipose ZAG only correlated with MTAC (p = 0.004) but not with D/P4 and PPCl. Serum ZAG did not correlate with any of the PSTR parameters (Table 3). In the multivariate linear regression model, PDE ZAG independently predicted D/P4 and MTAC (Table 4). In the same model, the significance of serum albumin and PPCl disappeared in the multivariate model. Serum and adipose ZAG did not predict PSTR: D/P4 (serum ZAG, p = 0.20; adipose ZAG, p = 0.70), MTAC (serum ZAG, p = 0.40; adipose ZAG, p = 0.50), and PPCl (serum ZAG, p = 0.30; adipose ZAG, p = 0.60) in the univariate model.
Peritonitis-related clinical outcomes
All patients were followed for 222.7 patient-years. Forty patients (31.7%) developed a total of 66 episodes of peritonitis—33, 12, 10, and one episodes were caused by Gram-positive organisms, Gram-negative organisms, mixed growth of organisms, and mycobacterium species, respectively. Ten episodes were culture negative. The overall peritonitis rates, expressed as episodes per patient-year of follow-up, were 0.7 ± 1.8 overall, 1.0 ± 2.4 in the high PDE ZAG group, and 0.3 ± 0.6 in the low PDE ZAG group (p = 0.048). PDE ZAG strongly and positively correlated with overall peritonitis episodes, and peritonitis episodes caused by Gram-positive, Gram-negative and mixed bacterial organisms (Supplementary Table 2, available online). However, PDE ZAG levels did not differ significantly in patients who developed peritonitis with concomitant exit site infection (Supplementary Table 3, available online). In multivariate linear regression analysis, PDE ZAG remained as the only independent predictor of peritonitis episodes (unstandardized B = 0.455, p < 0.001) and peritonitis-associated hospitalization (unstandardized B = 0.337, p < 0.001) after adjusting for age, sex, serum inflammatory markers (albumin, hsCRP), CCI, peritoneal characteristics (D/P4, MTAC, and PPCl), adipose and serum ZAG.
PDE ZAG also correlated with peritonitis-associated treatment outcome. In short, PDE ZAG negatively correlated with dialysate leukocyte count at the 3rd, 5th, and 10th day of peritonitis, which indicated patients with high PDE ZAG had a more rapid and favorable response to treatment. Patients with high PDE ZAG levels also had shorter peritonitis-related hospitalization stays (Supplementary Table 4, available online). However, patients who developed at least one episode of relapsing peritonitis, as defined by the ISPD guideline [22], had significantly higher PDE and adipose ZAG levels (Supplementary Table 5, available online).
Peritonitis-free survival
At follow-up, five patients received kidney transplantation, three patients were transferred to hemodialysis (two due to peritoneal failure, one due to a PD access problem), and 19 patients died. The causes of death were cardiovascular disease (n = 7), cerebrovascular accident (n = 3), infection (n = 7), malignancy (n = 1), and uncertain (n = 1) (Fig. 1). Seventy-six surviving patients (60.3%) were free of peritonitis. Their clinical characteristics are summarized in Supplementary Table 6 (available online). In short, surviving patients who remained peritonitis-free had a lower baseline D/P4 (p = 0.04). Patients with low PDE ZAG had superior peritonitis-free survival (68.3% vs. 52.4%; log-rank test, p = 0.03) (Fig. 3). However, the peritonitis-free rate did not differ with respect to the adipose ZAG (p = 0.80) and serum ZAG levels (p = 0.80). While PDE ZAG (p = 0.002), serum albumin (p = 0.03), D/P4 (p = 0.03), and CCI (p = 0.04) predicted peritonitis-free survival in the univariate model, PDE ZAG remained as the only independent predictor of such (p = 0.02) (Table 5) in the multivariate analysis. Serum (p = 0.90) and adipose ZAG (p = 0.90) did not predict peritonitis-free survival in the univariate model.
The overall survival rate was numerically lower in patients with high PDE ZAG, although the difference did not reach statistical significance (81.0% vs. 88.9%; log-rank test, p = 0.20). In the multivariate model, PDE ZAG did not predict overall survival (p = 0.90).
Discussion
In our study, the association between PDE ZAG and peritonitis-related clinical outcomes, such as peritonitis and peritonitis-free survival in a cohort of incident PD patients, was identified. In addition, we also quantified ZAG in PDE, determined its cellular origin from adipose tissue, and explored the relation between ZAG and peritoneal transport characteristics. To our knowledge, this is the first report that simultaneously analyzed and dissected the internal relationship of ZAG level at different body origins. Our results supplement the existing data on adipokine profiling in outcome prediction and suggest the potential utility of ZAG quantification as a risk stratification tool in patients with advanced renal failure.
Patients with high PSTR are susceptible to dialysis inadequacy and impaired dialysis ultrafiltration, as dialysate glucose is rapidly absorbed into the systemic circulation, which further leads to reabsorption of dialysate back into circulation when the osmotic gradient is dissipated. Intraperitoneal inflammation is an important determinant of PSTR, but it is poorly reflected by serum biomarkers [1]. Cellular compartment of PDE offers the most comprehensive view and insight into the determinants, cascades, and effects of intraperitoneal inflammation [23]. Cytokines driving the peritoneal fibrotic process are good effluent biomarkers to predict the PSTR [13] and peritonitis [24]. In our study, we identified a significant correlation between PDE ZAG and PSTR. ZAG is a 40 kDa-sized protein secreted from adipocytes, which functions as a lipid-mobilizing agent and maintains the balance of adiposity and wasting [12]. ZAG also modulates the inflammatory axis [25] through TNF-α [6] and amine oxidase 3 [26], which are also actively involved in the process of peritoneal inflammation and fibrosis.
Epithelial-mesenchymal transition (EMT) is the hallmark of peritoneal membrane disruption and fibrosis [27]. A few studies have underlined ZAG as a key player in the process of malignant cell progression, migration, and invasion in cancers [28] through induction of the EMT. In addition, ZAG activates transforming growth factor-beta (TGF-β) [29] and causes fibroblast activation and peritoneal fibrosis, which eventually reduces the peritoneal efficacy. In our study, the association between ZAG and PSTR was largely confined to D/P4 and MTAC, which are surrogate markers of small-sized solute transfer. Since the molecular size of ZAG corresponds to the size of small pores as described in the classical three-pore model of the semipermeable peritoneal membrane [30], the transfer rate of such a molecule across the peritoneum is therefore greatly affected by even a tiny change in the pore size.
Peritonitis is the commonest cause of peritoneal sclerosis, technique failure, and death even after resolution of peritonitis [31]. This highlights the importance to identify high risk individuals so that close monitoring can be provided to improve their survival [2]. In our univariate analysis (Table 4), serum albumin and PSTR predicted peritonitis, which is congruent with the data reported in the existing literature [24]. Peritoneal inflammation not only reduces peritoneal efficacy, as discussed, but also disrupts peritoneal equilibrium and induces a substantial amount of protein loss to the dialysate [32]. Nonetheless, the significances of albumin and PSTR were lost in the multivariate model after taking ZAG into consideration. Indeed, zinc is a crucial micronutrient for the functioning of the immune system against infection. Zinc controls the recruitment of polymorphs and the chemotactic response of macrophages and neutrophils through neutrophil extracellular traps, which result in phagocytosis and intracellular killing of bacterial pathogens [33]. ZAG itself also impairs the immune response by activation of TGF-β and Smad signaling [34,35], and mimics the adiponectin-related pathways that only blunts the effect of nuclear factor-κB/c-Jun N-terminal kinase signaling [25]. The inflammomodulatory effect of ZAG through its interplay with the amine oxidase copper-containing 3 also modifies the T-lymphocyte priming and activating response during active infection [26]. As a result, high circulating ZAG was reported in patients with refractory and life-threatening pneumonia [36]. Nonetheless, ZAG impairs the intestinal barrier function by modifying the intercellular tight junctions [37] and fosters the bacterial translocation, causing peritonitis [38] through the succinate-induced intestinal dysbiosis [39]. ZAG itself is also a lipid-mobilizing factor that promotes lipolysis and wasting [12], which are risk factors for infection and technique failure risks [40]. These explain the association between PDE ZAG and peritonitis demonstrated in our study.
Our study has a few limitations. Since this is a cohort study, we were unable to establish causality. Our sample size is small, and we recruited incident but not prevalent dialysis patients, which may reduce the generalizability. However, this can be our advantage as the peritoneal structure is yet to be altered by glucose degradation products from glucose-containing dialysis fluid. Evaluation of incident PD patients can permit us a more precise and accurate assessment of peritoneal structure with fewer confounders. We also did not determine the levels of downstream mediators and other associated adipokines; thus, we are unable to establish the complete pathogenic pathway. Furthermore, ZAG levels could be transiently affected by any acute illness such as infection considering it as an acute phase protein. However, the chance of such bias is low as patients who were medically unfit, such as those with ongoing sepsis, were excluded at recruitment. Peritonitis was also carefully excluded during the collection of PDE. Having said that, our study is the first and the largest cohort to quantify ZAG levels at different body origins and explore the interrelation.
In conclusion, our study highlights the role of peritoneal ZAG in the control and maintenance of peritoneal transport. PDE ZAG assay may represent a novel method to evaluate the peritoneal transport state, compared to the traditional PET test, which is tedious and time-consuming. The questions on the feasibility of PDE ZAG assaying to serially monitor peritoneal transport state, and ZAG-based therapy to improve outcomes deserves further clarification with large prospective clinical studies.
PDF Links
PubReader
ePub Link
XML Download
Full text via DOI
Download Citation
Supplement table 1
Print
