This was a prospective observational study. Study participants were recruited from January 2014 to April 2015 in the PD center of Renji Hospital, School of Medicine, Shanghai Jiao Tong University and prospectively followed up to the primary end point or the end of the study (1 October 2018). All procedures in this study were following the ethical standards of the responsible committee on human experimentation and the Helsinki Declaration of 1975. The study protocol was approved by the Human Research Ethics Committee of Renji Hospital, School of Medicine, Shanghai Jiao Tong University. All participants gave their written informed consent.

Stable patients with PD were eligible for this study. Exclusion criteria included patients with acute coronary syndrome, acute heart failure, peritonitis, exit-site infection, or other infectious complications within 4 weeks of enrollment or with underlying active malignancy, chronic liver disease, systemic lupus erythematosus requiring immunosuppression, systemic vasculitis, chronic rheumatic heart disease, and congenital heart disease; the presence of systemic inflammatory disease; patients who refused to give consent; or patients with incomplete data. Based on the inclusion and exclusion criteria, 318 patients with PD were enrolled in our study. All patients were dialyzed using conventional lactate-buffered glucose-based PD solutions (Dianeal®, Baxter, China).

The following data were collected at the time of enrollment: age, gender, height, weight, PD duration, underlying disease of ESRD, and the presence of comorbidity that includes diabetes and CVD. CVD was defined as the presence of ischemic heart disease, history of angina, previous myocardial infarction, coronary artery bypass surgery or stenting, class III–IV congestive heart failure, cerebrovascular event, transient ischemic attack, or peripheral arterial disease with or without amputation. Diabetes was defined either as a comorbid disease or as the etiology of ESRD. Whether taking angiotensin-converting enzyme inhibitors/angiotensin II receptor blockers (ACEI/ARB) or not was also collected.

Peritoneal protein clearance was calculated by the following equation (17): 24-h dialysate protein loss/(serum albumin/0.4783) and expressed as milliliters of plasma per day. Peritoneal total protein losses were measured in 24-h effluent using the Biuret reaction analytical method. Dialysate protein loss was measured by pyrogallol red colorimetry [Total Protein UC FS, DiaSys Diagnostic Systems (Shanghai) CO., LTD].

Solute clearance was determined by measuring total weekly urea clearance (Kt/V) and creatinine clearance (Ccr) using standard methods (18). Weekly Ccr was normalized to 1.73 m² of body surface area. In total, 24 h urine and ultrafiltration were recorded. The contribution of urea clearance by PD was estimated separately. For the evaluation of residual renal function (RRF), we used the estimated glomerular filtration rate, calculated as an average of 24-h urine urea and Ccr (19).

At enrollment, a standard peritoneal equilibration test (PET) was performed on each patient. Mass transfer area coefficient of creatinine (MTACcr) and 4 h D/P creatinine ratio were used to evaluate the peritoneal solute transport rate. MTACcr was calculated using the simplified Garred equation based on the data obtained from the PET test (20): MTACcr (ml/min) = Vd/t ln (Vi·P/(Vd (P–Dt))), where Vd is the drained volume, t is the dwell time (240 min), Vi is the instilled dialysate volume, P is the plasma concentration, and Dt is the dialysate concentration at the end of dwell time, normally determined in the dialysate after drainage. A 4h D/P creatinine ratio was the ratio of dialysate creatinine to plasma creatinine at 4 h in PET.

At the time of enrollment, the fasting venous blood sample of each patient was collected for the measurement of Angpt-1, Angpt-2, and soluble Tie-2 (sTie-2). Meanwhile, these cytokines in the overnight drained PD effluent were also determined. All of these cytokines were detected using an enzyme-linked immunosorbent assay (ELISA) kit (R&D Systems Inc, Minneapolis, MN, USA). All samples were run simultaneously and in duplicate to avoid intra- and inter-assay variations. The dialysate cytokines levels were influenced by ultrafiltration volume and by peritoneal solute transport rate, so the intraperitoneal cytokines levels were assessed as appearance rate, which was calculated as dialysate concentration times the drained volume divided by the dwell time and expressed as picogram per minute (pg/min). The following biochemical parameters were measured: high sensitivity C-reactive protein (hs-CRP), serum albumin, and brain natriuretic peptide (BNP). hs-CRP was measured by using the Tina-quant CRP (Latex) ultra-sensitive assay (D & P Modular analyzer, Roche Diagnostics GmbH, Mannheim, Germany). Serum albumin was measured using the bromocresol purple method, while BNP was measured using a fluorescence immunoassay with the Triage BNP Test (Biosite, San Diego, CA, USA).

All patients were followed up prospectively from the enrollment of the study until the primary end point, PD cessation, or to the end of the study (1 October 2018). The primary end point included cardiovascular events and mortality from all causes. For patients who developed multiple cardiovascular events, analysis was limited to the first one. Cardiovascular events included acute myocardial ischemic event, sustained atrial or ventricular arrhythmia, class III–IV congestive heart failure, stroke, peripheral vascular disease, and sudden death, which was defined and diagnosed clinically as unexpected natural mortality within 1 h from symptom onset and without a prior condition that would appear fatal.

Continuous data were expressed as mean ± SD or median (interquartile range) depending on the distribution of data. Patients were stratified by dichotomization of their Pcl level. Demographic features were compared with independent sample t-tests, Mann-Whitney U tests, or chi-squared tests, depending on whether the variable was normally distributed, skewed or categorical. Meanwhile, Spearman correlations of Pcl and other variables were performed. Survival and development of cardiovascular events of patients above and below the median value of Pcl were compared by the Kaplan-Meier method and log-rank test. The univariate Cox proportional hazards model was used to identify predictors of cardiovascular events and survival. The factors of p < 0.05 in the univariate Cox analysis of cardiovascular events, all-cause mortality, and clinical variables that may affect end point further entered the Cox regression analysis. We plotted scaled Schoenfeld residuals vs. time for all variables and computed their correlation against time to confirm that all variables considered in the Cox regression analysis met the assumptions of proportional hazards. The multivariate linear stepwise regression model analyzed variables that determined Pcl. A value of p < 0.05 was considered to be statistically significant. Statistical analysis was performed using SPSS software, version 24.0 (SPSS, Inc., Chicago, IL, USA).

A total of 318 patients were enrolled, which consisted of 163 (51.2%) men with a mean age of 56.7 ± 14.3 y and a median PD duration of 31.5 (12.1–57.2) months. The causes of ESRD were chronic glomerulonephritis in 96 patients (30.2%), diabetic nephropathy in 43 patients (13.5%), hypertensive nephrosclerosis in 10 patients (3.1%), polycystic kidney disease in 9 patients (2.8%), obstructive nephropathy in 4 patients (1.3%), tubulointerstitial nephritis in 4 patients (1.3%), others were 25 patients (7.9%), and the renal diagnosis was unknown in 127 patients (39.9%). Among these patients, 82 patients (25.7%) were comorbid with diabetes and 91 patients (28.6%) had a history of CVD.

The median Pcl level was 68.5 (52.1–90.3) ml/day in the whole cohort. Patients were dichotomized into two groups by the median of Pcl, namely, those with Pcl < 68.5 ml/day (low Pcl group) and those with Pcl ≥ 68.5 ml/day (high Pcl group). As shown in Tables 1, 2, participants in the high Pcl group are more likely to be older, have a greater proportion of background CVD, with higher BMI, BNP, pulse pressure (PP) lower serum albumin, higher MTACcr, and D/Pcr. Although the high Pcl group had increased total Ccr (p = 0.029) and peritoneal Ccr (p = 0.023), the Kt/V was not significantly different between the two groups. Besides, the high Pcl group had higher dialysate Angpt-2AR (p = 0.002), dialysate Angpt-2/1 ratio (p = 0.006), and dialysate sTie2 AR (p < 0.001).

However, there was no significant difference in serum Angpt-Tie2 system cytokine levels between patients in the two groups. No significant difference was seen in RRF, fluid removal (such as daily urine volume and daily ultrafiltration), PD duration, and proportion of patients who did continuous ambulatory peritoneal dialysis (CAPD) and taking ACEI/ARB (shown in Table 1).

The multivariate stepwise linear regression analysis showed that serum albumin, MTACcr, and BMI were independently associated with Pcl, after adjustment for age, background CVD, BNP, dialysate Angpt-2AR, and PP (shown in Table 3).

After follow-up for 43.9 (24.2–50.3) months, 63 patients experienced one or more cardiovascular events. The first cardiovascular events included 29 strokes, 16 acute myocardial ischemic events, 9 sudden death, 5 peripheral arterial disease events, and 4 class III–IV congestive heart failure. During the study period, 81 patients (25.5%) had died. Causes of mortality included 21 strokes, 21 infections (7 peritonitis episodes and 14 other infections), 10 sudden death, 9 acute myocardial ischemic events, 4 gastrointestinal bleeding, 3 malignancies (1 lung carcinoma and 2 stomach carcinoma), 3 heart failure episodes, 2 peripheral arterial disease, and 8 patients with unknown causes.

As shown in Figure 1, with comparison to the patients in the low Pcl group, a significant increase in cardiovascular events (log-rank = 4.667, p = 0.031; shown in Figure 1A) and all-cause mortality (log-rank = 8.115, p = 0.004; shown in Figure 1B) can be observed in patients in the high Pcl group.

After adjusting for age, background CVD, diabetes, BMI, serum albumin, serum Angpt-2, and Pcl remained an independent predictive factor for cardiovascular events [HR = 1.006 (1.000–1.012), p = 0.038; shown in Table 4], mainly in patients with PD without diabetes [1.008 (1.002–1.014), p = 0.005; shown in Table 5]. In addition, after adjusting for age, background CVD, diabetes, BMI, serum albumin, serum Angpt-2, RRF, and PD duration, the predictive effect of Pcl on all-cause mortality was lost (shown in Table 6).