Abstract
Background
Preservation of residual kidney function (RKF) is critical for patients undergoing peritoneal dialysis (PD). However, the relationship between the rate of RKF decline (RKF slope) and PD technique failure remains incompletely understood, and a clinically actionable threshold is lacking.
Method
This was a single-centre, retrospective cohort study that included 539 patients who underwent peritoneal dialysis catheter placement at the hospital from 1 May 2006 to 31 December 2022. The primary analysis used pre‑specified Cox models to assess the association between RKF decline rate and technique failure (death or transfer to HD). RCS analysis and Kaplan‑Meier curves were also performed.
Results
During a median follow‑up of 30 months, 152 patients (28.2%) in the full cohort experienced technique failure. The median RKF decline rate was −0.14 ml/min/1.73 m2 /month (IQR −0.20 to −0.08). Faster RKF decline was independently associated with a higher risk of technique failure (per 0.1 ml/min/1.73 m2 /month: HR = 1.19, 95% CI 1.05–1.34, p < 0.001). RCS analysis revealed a nonlinear relationship (p = 0.047) with a candidate threshold at −0.20. Patients with decline faster than −0.20 had significantly lower technique failure‑free survival (log‑rank p = 0.027).
Conclusion
There was a nonlinear relationship between the rate of decline in residual kidney function (RKF slope) and the risk of PD patients experiencing technique failure, with a candidate threshold of −0.20 ml/min/1.73 m2 /month. Regularly monitoring the decline in RKF and maintaining it above −0.20 ml/min/1.73 m2 /month may help identify patients at higher risk of technique failure, although external validation is needed.
Supplementary Information
The online version contains supplementary material available at https://doi.org/10.1186/s12882-026-05210-7.
Keywords: Peritoneal dialysis, Residual kidney function, Technique failure, Nonlinear association, Threshold effect
Introduction
The global prevalence of end-stage kidney disease (ESKD) is increasing annually, with the number of patients expected to exceed 14.5 million by 2030 [1]. Peritoneal dialysis (PD) is often suggested as a primary treatment for ESKD because it offers haemodynamic stability, preserves residual kidney function (RKF), and can be conveniently performed at home [2]. However, PD remains underutilized globally, accounting for only approximately 11% of all dialysis and 9% of all kidney replacement therapy [3]. Variations in PD use are substantial across countries and regions. Notably, more than half of the world’s PD patients are concentrated in China, the United States, Mexico, and Thailand, with the highest global PD utilization rate observed in Hong Kong, China [4]. Despite its clinical advantages and expanding utilization in some regions, approximately 30% to 50% of PD patients experience technique failure within 3 to 5 years because of complications such as peritonitis and ultrafiltration failure [5, 6]. Technique failure is defined as the composite endpoint of conversion to haemodialysis (HD) for more than 30 days or death during PD [7]. Technique failure usually manifests as a complication requiring hospitalization or an unexpected conversion to HD. Patients with such conditions have a worse outcome and higher mortality rates than those with scheduled HD [8]. RKF refers to the filtration and endocrine function of residual renal tissue in ESKD patients. Owing to the complexity of assessing endocrine function, the residual glomerular filtration rate is typically used in clinical practice to indicate RKF [9]. Several studies have confirmed that RKF is closely associated with survival, nutritional status and quality of life in PD patients [10, 11]. Loss of RKF not only leads to malnutritional conditions (such as protein-energy wasting) but also is a key predictor of increased mortality risk in PD patients [12].
Therefore, protecting of RKF is a core strategy for improving the clinical prognosis of PD. Notably, a progressive decline in RKF in PD patients has been recognized as a key risk factor for technique failure [13]. More importantly, the rate of decline of RKF—that is, the speed at which it changes over time (typically quantified as a slope, such as ml/min/1.73 m2 /month)—may be a better predictor of poor prognosis than the decline itself and may provide a more sensitive window for clinical intervention [14].
Although multiple cohort studies have confirmed the importance of RKF in prolonging technique survival, there is currently a significant lack of evidence regarding the rate of RKF decline (RKF slope). First, there is controversy regarding the nonlinear relationship between the RKF decline rate and technique failure risk, and quantitative threshold values are lacking [15]. Second, the effect of subgroup heterogeneity is unknown. Key subgroups (such as patients with diabetes) may significantly influence the strength of the association between the rate of RKF decline and technique failure, although the evidence for effect modification by diabetes remains limited [16, 17]. Finally, there is a lack of operational clinical thresholds to guide intervention [12]. Although international guidelines [18] emphasize the importance of monitoring RKF, clear thresholds based on the RKF decline rate have not yet been established in clinical practice, making it difficult to effectively guide targeted interventions to slow RKF decline. Therefore, to bridge these critical gaps and provide actionable clinical insights, this study aimed to elucidate the nonlinear relationship between the RKF decline rate (RKF slope) and the risk of PD patients experiencing technique failure through a single-centre, retrospective cohort analysis and to determine the key threshold values.
Methods
Study design and study population
This was a single-centre, retrospective study that included patients who received PD between 1 May 2006 and 31 December 2022 at Tongji Hospital. The follow-up period was extended to 30 April 2024.The inclusion criteria were as follows: (1) PD placement at Tongji Hospital; (2) patients who had established a file and regular follow-up at the PD centre within the hospital; (3) patients over 18 years old; and (4) patients who had received PD treatment for more than 3 months. The exclusion criteria were as follows: (1) patients who underwent HD for more than 3 months before the start of PD or who had received a kidney transplant; (2) patients whose 24-hour urine volume was < 100 ml or whose estimated glomerular filtration rate (eGFR) was < 2 ml/min/1.73 m2 at the time of PD tube opening; (3) patients whose 24-hour urine volume data were missing; and (4) patients for whom > 40% of data were missing. The study was approved by the Medical Ethics Committee of Tongji Hospital, Tongji Medical College, Huazhong University of Science and Technology (Ethics No. TJ-IRB20220738). As this study was a retrospective observational study without any intervention, informed consent was exempted by the Ethics Committee. This study complied with the STROBE guidelines.
Variables
Residual kidney function
RKF is measured primarily by the glomerular filtration rate, which reflects the kidney’s metabolic filtration capacity. Deterioration of RKF can be categorized into complete loss of RKF and RKF decline rate. Complete loss of RKF is characterized by anuria, as evidenced by a urine volume of less than 100 ml in 24 hours. The RKF decline rate is expressed as the RKF slope, defined as the slope of the curve obtained from a least-squares linear fit of the measured glomerular filtration rate (mGFR) to dialysis time.
The slope was estimated separately for each patient at the individual level. To ensure reliable slope estimation, a minimum of three mGFR measurements (including the baseline value) was required, and all measurements obtained after the occurrence of technique failure (transfer to hemodialysis or death) were excluded. The RKF slope is usually negative. The smaller the value of the RKF slope is, the faster the rate of decline in RKF. We calculated the monthly RKF decline rate as the absolute value of the original RKF slope. This value was then multiplied by 10. Thus, a 1‑unit increase corresponds to a 0.1 ml/min/1.73 m2 /month faster decline. In the tables, this variable is denoted as RKF decline rate (×10).
Although RKF decline may not be strictly linear over an extended period, a linear approximation was chosen for the primary Cox regression model for two reasons. First, the linear slope provides a clinically interpretable measure of the average decline rate, which is commonly used in PD studies. Second, the restricted cubic spline (RCS) analysis (see Results) was performed to complement the linear model by exploring nonlinearity and identifying a candidate threshold.
Technique failure
PD technique failure was defined as a composite endpoint of death or transfer to hemodialysis for more than 30 days, according to the standardized definition proposed by Lan et al. (2016) [7, 19].
Covariates
All baseline data were collected at the time of PD initiation (within 1 month after catheter placement). Demographic, clinical and dialysis-related data at dialysis initiation were collected from electronic patient charts. These data included age, sex, education level, marital status, employment status, place of residence, BMI, laboratory test results, 24-hour urine volume, digestive symptoms, cardiovascular symptoms, peritonitis and oedema.Cardiovascular symptoms were defined as common manifestations of cardiovascular disease, including palpitations, chest tightness and wheezing. Digestive symptoms were defined as common manifestations of digestive disorders, such as nausea, vomiting, diarrhoea and constipation. The diagnosis of peritonitis was based on symptoms and laboratory findings during follow-up. At least two conditions had to be met: (1) clinical manifestations consistent with peritonitis, such as abdominal pain and/or turbidity of the dialysate; (2) a leukocyte count of more than 100/μL in the dialysate (that was maintained for more than 2 h), with more than 50% of the leukocytes being polymorphonuclear; and (3) a positive culture result of the dialysate.
Formula calculation
Residual kidney function (RKF) was assessed by measured glomerular filtration rate (mGFR), calculated as the average of renal urea clearance and renal creatinine clearance according to the method validated by van Olden et al. [20]. The formulae are as follows:
Renal urea clearance (ml/min) = (urine urea / serum urea) × (24-h urine volume / 1440)
Renal creatinine clearance (ml/min) = (urine creatinine / serum creatinine) × (24-h urine volume / 1440)
mGFR (ml/min/1.73 m2) = (renal urea clearance + renal creatinine clearance) / 2, then adjusted by body surface area.
Missing data processing
Variables with more than 40% missing data were excluded from the analysis. For variables with missing data ≤ 40% that were included in the multivariable models (e.g., eGFR, SBP, hemoglobin, albumin), multiple imputation using random forest methods was performed. The proportion of missing data was low for these key covariates (<5% for age, sex, SBP, albumin, hemoglobin, and eGFR; see Supplementary Table S1 for details). RKF slope had no missing data in the primary analysis cohort (n = 433). After imputation, the primary analysis was based on the full cohort of 433 patients.
Statistical analyses
Statistical analyses were conducted using R software (version 4.2.2), GraphPad Prism 8.0 and MSTATA software (http://www.mstata.com). Statistical significance was defined as a p value less than 0.05 (two-sided). Baseline characteristics were categorized into two groups based on the occurrence of technique failure. Normally distributed data are expressed as the means ± standard deviations (SD), nonparametric variables are expressed as quartiles and medians, and categorical parameters are expressed as frequencies and percentages. Comparisons between groups were made using the t test, Wilcoxon rank sum test, or chi-square test. To evaluate the independent association, three hierarchical prespecified Cox models were built without stepwise selection: Model 1 adjusted for age and sex; Model 2 additionally for diabetes, hypertension, and cause of ESKD; Model 3 further for hemoglobin, albumin, eGFR, and SBP. A restricted cubic spline (RCS) with three knots at the 10th, 50th, and 90th percentiles of the RKF decline rate distribution was fitted, adjusting for the same covariates as Model 3. Nonlinearity was tested by the likelihood ratio test. The candidate threshold was identified as the point where the hazard ratio (HR) began to increase steeply on the RCS curve and the lower bound of its 95% confidence interval exceeded 1.0. This approach combines visual assessment of the curve shape with a statistical criterion.
Using the candidate threshold identified from the RCS analysis, patients were stratified into rapid decline and slow decline groups; technique failure‑free survival was compared by log‑rank test.
Subgroup analyses (sex, BMI, diabetes, cause of ESKD) used Cox models with interaction terms, adjusted for Model 3 covariates (excluding the stratifying variable). Effect modification was tested by likelihood ratio test.
Two sensitivity analyses addressed the competing risk of death: (1) death‑censored Cox (death censored, transfer to HD as event); (2) Fine‑Gray model (non‑fatal technique failure as primary event, death as competing event). Both were adjusted for Model 3 covariates.
Results
Participant characteristics
After screening according to the inclusion and exclusion criteria, a total of 539 patients were included (Fig. 1). The baseline characteristics of the study cohort are shown in Table 1. The median follow-up time was 30 months (interquartile range [IQR] 20–47). The mean age was 42.4 years, and 51.9% were female. The median individual RKF slope was −0.14 ml/min/1.73 m2 /month (IQR −0.20 to −0.08). Appendix Figure S1 shows the group‑level decline trajectory.
Fig. 1.
Flowchart of the screening process for study participants
Table 1.
Patient demographics and baseline characteristics
| Characteristic | Failure | p-value | |
|---|---|---|---|
| No (n = 387) | Yes (n = 152) | ||
| Sex | 0.342 | ||
| Male | 181 (46.8%) | 78 (51.3%) | |
| Female | 206 (53.2%) | 74 (48.7%) | |
| Age, years | 41 (32, 51) | 45 (35, 52) | 0.030 |
| Education | <0.001 | ||
| Primary school or below | 51 (13.2%) | 41 (27.0%) | |
| Junior high school | 155 (40.1%) | 62 (40.8%) | |
| High school | 86 (22.2%) | 32 (21.1%) | |
| College and above | 95 (24.5%) | 17 (11.2%) | |
| Married | 0.360 | ||
| No | 52 (13.4%) | 16 (10.5%) | |
| Yes | 335 (86.6%) | 136 (89.5%) | |
| Employed | 0.716 | ||
| No | 111 (28.7%) | 46 (30.3%) | |
| Yes | 276 (71.3%) | 106 (69.7%) | |
| Residence | 0.081 | ||
| City | 260 (67.2%) | 90 (59.2%) | |
| Village | 127 (32.8%) | 62 (40.8%) | |
| BMI, kg/m2 | 21.07 (19.19, 23.44) | 22.04 (19.72, 24.54) | 0.007 |
| 24 h urine, ml | 1,200 (900, 1,500) | 1,100 (700, 1,400) | 0.004 |
| eGFR, mL/min/1.73 m2 | 5.40 (4.34, 6.50) | 5.39 (4.60, 6.45) | 0.620 |
| SBP, mmHg | 141 (131, 151) | 143 (134, 156) | 0.101 |
| DBP, mmHg | 91 (84, 99) | 90 (82, 97) | 0.201 |
| Hemoglobin, g/L | 84 (75, 94) | 85 (79, 89) | 0.747 |
| Albumin, g/L | 39.3 (36.8, 41.6) | 38.5 (36.8, 40.5) | 0.030 |
| Urea, mmol/L | 27 (22, 33) | 27 (22, 32) | 0.598 |
| Serum creatinine, mmol/L | 801 (708, 943) | 809 (724, 896) | 0.766 |
| Uric acid, μmol/L | 419 (360, 485) | 426 (395, 466) | 0.184 |
| Bicarbonate, mmol/L | 20.4 (18.0, 22.5) | 20.5 (19.4, 21.3) | 0.486 |
| Potassium, mmol/L | 4.78 (4.29, 5.14) | 4.69 (4.49, 4.99) | 0.605 |
| Sodium, mmol/L | 139.00 (137.60, 140.33) | 139.10 (138.00, 140.18) | 0.382 |
| Calcium, mmol/L | 2.12 (2.03, 2.26) | 2.10 (2.03, 2.19) | 0.118 |
| Phosphorus, mmol/L | 1.85 (1.63, 2.12) | 1.80 (1.65, 2.04) | 0.418 |
| Total cholesterol, mmol/L | 3.87 (3.27, 4.44) | 3.99 (3.57, 4.42) | 0.116 |
| Triglyceride, mmol/L | 1.27 (0.94, 1.82) | 1.24 (1.03, 1.83) | 0.662 |
| Blood glucose, mmol/L | 5.91 (5.26, 6.71) | 6.39 (5.60, 7.23) | <0.001 |
| PTH, pg/ml | 338 (238, 428) | 360 (292, 433) | 0.098 |
| Diabetes | 0.085 | ||
| No | 363 (93.8%) | 136 (89.5%) | |
| Yes | 24 (6.2%) | 16 (10.5%) | |
| Hypertension | 0.202 | ||
| No | 158 (40.8%) | 53 (34.9%) | |
| Yes | 229 (59.2%) | 99 (65.1%) | |
| Cause of ESKD | 0.125 | ||
| Primary glomerular disease | 331 (85.5%) | 121 (79.6%) | |
| Hypertensive nephropathy | 24 (6.2%) | 8 (5.3%) | |
| Diabetic nephropathy | 15 (3.9%) | 10 (6.6%) | |
| Others# | 17 (4.4%) | 13 (8.6%) | |
Note: all values are presented as median (Q1, Q3) or as counts (n (%))
All data are baseline values collected at PD initiation. BMI, Body Mass Index; SBP, Systolic Blood Pressure; DBP, Diastolic Blood Pressure; PTH, Parathyroid hormone; ESKD, End Stage Kidney Disease
#Others include obstructive nephropathy, lupus nephritis, vasculitis, congenital anomalies, and other rare causes
Among the 539 patients, 152 (28.2%) experienced PD technique failure, defined as a composite endpoint of death or transfer to hemodialysis (HD) for more than 30 days. Of these, 98 patients (64.5%) transferred to HD, and 54 patients (35.5%) died (including those who died on PD and those who died within 30 days after transfer to HD). The causes leading to transfer to HD are detailed in the patient flow diagram (Fig. 1).
For the primary analysis, we included patients with at least three mGFR measurements (including baseline), yielding a cohort of 433 patients. In this cohort, the median number of mGFR measurements was 9 (IQR 5–13), and the median observation period for slope calculation was 16 months (IQR 7–28). Among these, 119 patients experienced technique failure. Their baseline characteristics were similar to those of the full cohort (Table 1).
PD patients were divided into two groups, the technique failure group and the technique survival group, on the basis of whether technique failure occurred. Patients with technique failure were older and had lower education levels, higher BMIs, lower 24-hour urine volumes, and lower albumin levels. Metabolic parameters also differed, with higher blood glucose in the technique failure group. The technique failure group had a higher incidence of peritonitis during the follow-up period. The failure group also exhibited a faster RKF decline rate (i.e., more negative RKF slope). These follow-up characteristics (peritonitis, anuria, and RKF slope) are summarized in Table 2. No significant differences were found in sex, marital status, employment status, eGFR, or most biochemical parameters.
Table 2.
Follow-up characteristics of the study cohort
| Characteristic | Failure | p-value | |
|---|---|---|---|
| No (n = 387) | Yes (n = 152) | ||
| Anuria during follow-up | 0.076 | ||
| No | 289 (74.7%) | 102 (67.1%) | |
| Yes | 98 (25.3%) | 50 (32.9%) | |
| Follow-up time, months | 30 (21, 46) | 30 (15, 53) | 0.521 |
| Peritonitis | <0.001 | ||
| No | 329 (85.0%) | 100 (65.8%) | |
| Yes | 58 (15.0%) | 52 (34.2%) | |
| RKF slope, ml/min/1.73 m2/M | −0.13 (−0.20, −0.07) | −0.16 (−0.21, −0.11) | 0.008 |
Note: all values are presented as median (Q1, Q3) or as counts (n (%)). Baseline anuria excluded. Anuria during follow-up: 24 h urine volume < 100 ml. RKF slope: individual linear regression of measured glomerular filtration rate (mGFR) over time. Follow-up time: from PD initiation to event or censoring. RKF, residual kidney function
Association between RKF decline rate and PD technique failure
The association between RKF decline rate and PD technique failure was first examined using pre-specified multivariable Cox regression models (Table 3). After adjustment for age and sex (Model 1), each 0.1 ml/min/1.73 m2 /month faster decline in RKF was associated with a 17% increase in the risk of technique failure (HR 1.17, 95% CI 1.03–1.32, p = 0.012). Additional adjustment for diabetes, hypertension, and cause of ESKD (Model 2) yielded an HR of 1.18 (95% CI 1.05–1.33, p = 0.006). In the fully adjusted Model 3 (additionally including albumin, hemoglobin, eGFR, and SBP), the HR was 1.19 (95% CI 1.05–1.34, p = 0.006). These results indicate that a faster RKF decline is independently associated with a higher risk of PD technique failure under a linear assumption.
Table 3.
Pre‑specified multivariable Cox regression models for the association between RKF decline rate and PD technique failure
| Characteristic | Model 1 | Model 2 | Model 3 | |||
|---|---|---|---|---|---|---|
| HR (95% CI) | p-value | HR (95% CI) | p-value | HR (95% CI) | p-value | |
| RKF decline rate (×10) | 1.17 (1.03, 1.32) | 0.012 | 1.18 (1.05, 1.33) | 0.006 | 1.19 (1.05, 1.34) | 0.006 |
| RKF decline rate (×10) (quartiles) | ||||||
| Q1 | Reference | Reference | Reference | |||
| Q2 | 2.26 (1.37, 3.74) | 0.001 | 2.30 (1.38, 3.83) | 0.001 | 2.35 (1.40, 3.93) | 0.001 |
| Q3 | 1.70 (1.02, 2.82) | 0.040 | 1.84 (1.10, 3.07) | 0.019 | 1.82 (1.09, 3.05) | 0.022 |
| Q4 | 2.37 (1.41, 3.97) | 0.001 | 2.61 (1.55, 4.41) | <0.001 | 2.69 (1.57, 4.60) | <0.001 |
| P for trend | 0.007 | 0.002 | 0.002 | |||
Note: Model 1 adjusted for age and sex. Model 2 additionally adjusted for diabetes, hypertension, and cause of ESKD. Model 3 further adjusted for hemoglobin, albumin, eGFR, and SBP. Abbreviations: CI, confidence interval; HR, hazard ratio; RKF, residual kidney function; ESKD, end-stage kidney disease; eGFR, estimated glomerular filtration rate; SBP, systolic blood pressure. The RKF decline rate (×10) was derived from the absolute RKF slope multiplied by 10; each 1‑unit increase represents a 0.1 ml/min/1.73 m2 /month faster decline
Nonlinear relationship and candidate threshold
To explore whether the linear assumption underlying the Cox model was adequate, we performed a restricted cubic spline (RCS) analysis adjusting for the same covariates as in Model 3. The RCS curve revealed a nonlinear association between RKF decline rate and the hazard of technique failure (P for nonlinearity = 0.047; Fig. 2). The hazard ratio remained relatively stable when the decline rate was below approximately −0.20 ml/min/1.73 m2 /month, and then increased steeply beyond this point. This inflection point at −0.20 was therefore identified as a candidate clinical threshold.
Fig. 2.
Kaplan-Meier curves for PD technique failure according to RKF decline rate. Patients were stratified using the candidate threshold of −0.20 ml/min/1.73 m²/month: rapid decline (RKF slope ≤ −0.20) and slow decline (RKF slope > −0.20). The log-rank test yielded a p-value of 0.027. Tick marks indicate censored patients. Numbers at risk are shown below the x-axis
Kaplan-Meier analysis using the candidate threshold
Using the candidate threshold of −0.20 ml/min/1.73 m2 /month, patients were stratified into rapid decline (≤−0.20) and slow decline (>−0.20) groups. Kaplan‑Meier curves demonstrated that the rapid decline group had a significantly lower technique failure‑free survival compared with the slow decline group (log‑rank test: χ2 = 4.888, p = 0.027; Fig. 3).
Fig. 3.
Restricted cubic spline (RCS) curve for the association between RKF decline rate and PD technique failure, adjusted for the same covariates as in Model 3. The solid line represents the hazard ratio (HR); shaded area indicates 95% confidence interval. A nonlinear association was observed (PP for nonlinearity = 0.047). The inflection point at approximately −0.20 ml/min/1.73 m²/month was identified as a candidate clinical threshold
Subgroup and sensitivity analyses
Subgroup analyses for sex, BMI, diabetes, and cause of ESKD (Table 4) showed no significant interactions (all P for interaction > 0.05), indicating that the association between faster RKF decline and PD technique failure was consistent across these subgroups. Point estimates were similar across sexes and BMI categories. In diabetic patients (n = 40) and in the merged ‘others’ category for ESKD cause (n = 87), the estimates were imprecise due to small sample sizes, and no significant interactions were detected. However, given the small sample sizes in these subgroups, the absence of significant interactions should not be interpreted as definitive evidence of no effect modification.
Table 4.
Subgroup analysis of the association between RKF decline rate (×10) and technique failure
| Subgroup | N | Adjusted HR (95% CI) | Pvalue | Pfor interaction |
|---|---|---|---|---|
| Sex | 0.602 | |||
| Female | 280 | 1.13 (0.90–1.43) | 0.295 | |
| Male | 259 | 1.21 (1.05–1.39) | 0.008 | |
| BMI_ group | 0.195 | |||
| ˂ 20 | 169 | 1.09 (0.77–1.56) | 0.618 | |
| 20–25 | 283 | 1.15 (0.96–1.37) | 0.121 | |
| ≥25 | 87 | 1.47 (1.05–2.06) | 0.026 | |
| Diabetes | 0.902 | |||
| No | 499 | 1.19 (1.05–1.35) | 0.006 | |
| Yes | 40 | 1.07 (0.58–1.96) | 0.823 | |
| Cause of ESKD | 0.275 | |||
| Primary glomerular disease | 452 | 1.16 (1.02–1.33) | 0.028 | |
| Others# | 87 | 1.32 (0.93–1.89) | 0.122 |
Note: #Others include hypertensive nephropathy, diabetic nephropathy, obstructive nephropathy, lupus nephritis, vasculitis, congenital anomalies, and other rare causes.All subgroup models were adjusted for the same covariates as Model 3 (age, sex, diabetes, cause of ESKD, baseline eGFR, SBP, hemoglobin, albumin), except that the stratifying variable was not included as a covariate in its own subgroup analysis. The HRs correspond to a 0.1 ml/min/1.73 m²/month increase in RKF decline rate (calculated as the absolute RKF slope multiplied by 10)
To address the competing risk of death, we performed two sensitivity analyses (Supplementary Table S2 and S3). In the death-censored Cox model, the HR for RKF decline rate was 1.23 (95% CI 1.07–1.42, p = 0.004). In the Fine-Gray competing-risk model, the subdistribution HR was 1.23 (95% CI 1.08–1.39, p = 0.001). Both results were consistent with the primary analysis.
Discussion
This study aimed to explore the relationship between the rate of decline in RKF and PD patients experiencing technique failure. Multifactorial Cox regression analysis indicated that the rate of decline in RKF is an independent predictor of PD patients experiencing technique failure. In addition, RCS analysis confirmed a nonlinear relationship between the two variables, identifying a candidate threshold of −0.20 ml/min/1.73 m2 /month. When the decline rate exceeded this threshold, the risk of technique failure increased steeply as the decline rate became more negative (faster decline). These findings provide a potential rate threshold that may help identify high-risk patients, although external validation is required.
Previous studies have consistently shown that RKF is significantly associated with patient survival rates [11, 21]. Differences in how endpoints are defined may impact study outcomes. In this study, the technique failure endpoint was a composite of death or permanent transition to HD > 30 days. A significant proportion of PD patients switched to haemodialysis because of PD-specific complications (such as peritonitis), and baseline RKF may have limited the ability to predict such events. Second, patients with extremely low RKF may die earlier, thereby reducing their chances of experiencing nonfatal technique failure. These findings suggest that the dynamic changes in RKF are more relevant than its static initial level in predicting technique survival. One study [12]reported that a rapid decline in RKF increases the risk of technique failure (excluding death) and that its predictive value is superior to that of the baseline RKF. Another study [13] also confirmed that a decline in RKF during the first year is a predictor of technique failure in nondiabetic populations. The present study extends these observations by confirming the core role of the RKF slope and its nonlinear characteristics in an exploratory subgroup analysis of diabetic patients, using a composite endpoint (death or transfer to HD). However, given the small sample size of the diabetic subgroup (n = 40), these findings should be interpreted cautiously.
The observed nonlinear relationship between the rate of RKF decline and PD technique failure suggests that once RKF loss exceeds a candidate threshold (approximately −0.20 ml/min/1.73 m2 /month), the underlying pathophysiological processes may accelerate and become self‑perpetuating (Fig. 4). Several interconnected mechanisms may explain this threshold effect. There are many reasons why the rapid decline in RKF leads to technique failure. First, a decrease in RKF leads to a reduction in fluid clearance capacity. Although adjustments in the prescription of PD (e.g., high-glucose dialysate) can increase ultrafiltration [22], studies have demonstrated that this approach is often insufficient to fully compensate for fluid imbalance [22, 23], leaving patients at significant risk of fluid overload. Volume overload is directly associated with poor blood pressure control and cardiovascular events [23]. In addition, volume overload may increase the risk of bacterial translocation and peritonitis through secondary intestinal oedema [24], which is among the main causes of technique failure [25]. Studies by Ng et al. have also confirmed that increased extracellular water is associated with an increased risk of death [26]. Second, a decrease in RKF weakens the ability to clear uraemic toxins, inflammatory factors, and oxidative stress products [27]. The accumulation of these substances can induce local and systemic inflammatory states. Local peritoneal inflammation (even if it does not reach the level of peritonitis) can drive fibrosis, angiogenesis, and epithelial‒mesenchymal transition, impairing peritoneal ultrafiltration function and creating a vicious cycle [28]. Systemic inflammation further exacerbates poor outcomes. Third, rapid RKF decline may contribute to endothelial dysfunction through oxidative stress and reduced nitric oxide bioavailability, which compromises peritoneal vascular integrity and angiogenic capacity. This vascular pathology, combined with chronic inflammation and volume overload, accelerates peritoneal fibrosis and ultrafiltration failure. It has been shown that residual renal function is independently associated with endothelial dysfunction in patients receiving PD [29]. These pathways may explain why patients with a steep RKF decline (more negative than −0.20) face a substantially higher risk of technique failure. The threshold may represent a tipping point where compensatory mechanisms (e.g., increased peritoneal ultrafiltration) become exhausted and organ damage becomes irreversible.
Fig. 4.
Proposed mechanisms linking rapid RKF decline to PD technique failure. Rapid RKF decline triggers three interconnected pathways: volume overload, systemic inflammation, and endothelial dysfunction. These processes synergistically promote peritoneal fibrosis and ultrafiltration failure, ultimately leading to PD technique failure (transfer to HD or death). The threshold of −0.20 ml/min/1.73 m²/month may represent a tipping point where these pathways become self-perpetuating
Limitations
This study has several limitations. First, its single‑centre, retrospective design may introduce selection bias and limit generalizability to other populations or centres with different patient characteristics. Second, despite adjustment for major covariates, we could not capture all potential confounders, including diuretic use, intradialytic hypotension, nephrotoxic medications, and PD modality (CAPD vs. APD). Third, RKF was assessed using urinary creatinine clearance, which may underestimate true GFR compared to isotopic methods, although this approach is consistent with routine clinical practice. Fourth, the candidate threshold of −0.20 ml/min/1.73 m2 /month was derived from a single cohort and requires external validation in independent populations. Fifth, peritonitis and anuria are time‑dependent events that occur during follow‑up. Because the exact timing of these events was not available in our retrospective database, we could not perform a formal time‑dependent Cox analysis. Their incidence is reported descriptively in Table 2, and they were not included as fixed covariates in the primary models to avoid time‑dependent bias. Future prospective, multicentre studies with detailed event timing are needed to validate our findings and to properly assess the impact of peritonitis and anuria on the association between RKF decline and technique failure. Although missing data were imputed using random forest methods, the possibility of residual bias due to missing information cannot be completely excluded.
Conclusion
This study demonstrated that the rate of RKF decline, rather than its baseline level, is a key dynamic indicator for predicting technique failure. We revealed that there is a nonlinear relationship between the two and identified −0.20 ml/min/1.73 m2 /month as the candidate threshold. When the rate of RKF decline faster than this threshold (i.e., decline is more rapid), the risk of technique failure increases steeply. These findings highlight the clinical value of regularly monitoring RKF trends. Maintaining the RKF decline rate above this threshold (i.e., slowing the decline) may help reduce the risk of technique failure, although external validation is required. Future studies should validate this threshold in multicentre cohorts and assess whether interventions based on it can improve long-term patient outcomes.
Electronic supplementary material
Below is the link to the electronic supplementary material.
Acknowledgements
The authors thank the nursing staff of the peritoneal dialysis center for their assistance in data collection. AI assistance: ChatGPT-4 was used for language polishing only. The authors take full responsibility for the content of this manuscript.
Author contributions
Yulin Xu and Xiaoqin Liu performed the data analysis and drafted the manuscript. Shuwang Ge and Ying Yao conceived the study, supervised the analysis, and substantively revised the manuscript. All authors read and approved the final manuscript.
Funding
This work was supported by grants from the Tongji Hospital Research Fund Project (2022C04, 2022D22). The funders had no role in the study design, data collection, analysis, interpretation, or decision to publish.
Data availability
The data that support the findings of this study are available from the corresponding author upon reasonable request.
Declarations
Ethics approval and consent to participate
The study protocol was approved by the Medical Ethics Committee of Tongji Hospital, Tongji Medical College, Huazhong University of Science and Technology (Approval No.: TJ-IRB20220738). Due to the retrospective nature of the study, the requirement for informed consent was waived by the ethics committee. All procedures were performed in accordance with the ethical standards of the institutional research committee and with the 1964 Helsinki declaration and its later amendments.
Consent for publication
Not applicable, as no individual person’s data are presented in this manuscript.
Competing interests
The authors declare no competing interests.
Footnotes
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Yulin Xu and Xiaoqin Liu have contributed equally to this work.
Contributor Information
Shuwang Ge, Email: geshuwang@tjh.tjmu.edu.cn.
Ying Yao, Email: yaoyingkk@126.com.
References
- 1.Nephrology ISo. Second edition of ISN global kidney health Atlas now available for download 2019 [Available from: https://www.doi.org/www.theisn.org/blog/2019/08/23/second-edition-of-isn-global-kidney-health-atlas-now-available-for-download/.
- 2.Li PK, Chow KM, Van de Luijtgaarden MW, Johnson DW, Jager KJ, Mehrotra R, et al. Changes in the worldwide epidemiology of peritoneal dialysis. Nat Rev Nephrol. 2017;13(2):90–103. [DOI] [PubMed] [Google Scholar]
- 3.Bello AK, Okpechi IG, Osman MA, Cho Y, Cullis B, Htay H, et al. Epidemiology of peritoneal dialysis outcomes. Nat Rev Nephrol. 2022;18(12):779–93. 10.1038/s41581-022-00623-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Bello AK, Levin A, Lunney M, Osman MA, Ye F, Ashuntantang GE, et al. Status of care for end stage kidney disease in countries and regions worldwide: international cross sectional survey. BMJ. 2019;367:l 5873. 10.1136/bmj.l5873. [DOI] [PubMed]
- 5.Cnossen TT, Usvyat L, Kotanko P, van der Sande FM, Kooman JP, Carter M, et al. Comparison of outcomes on continuous ambulatory peritoneal dialysis versus automated peritoneal dialysis: results from a USA database. Perit Dial Int. 2011;31(6):679–84. 10.3747/pdi.2010.00004. [DOI] [PubMed] [Google Scholar]
- 6.Cho Y, See EJ, Htay H, Hawley CM, Johnson DW. Early peritoneal dialysis technique failure: review. Perit Dial Int. 2018;38(5):319–27. 10.3747/pdi.2018.00017. [DOI] [PubMed] [Google Scholar]
- 7.Lan PG, Clayton PA, Johnson DW, McDonald SP, Borlace M, Badve SV, et al. Duration of hemodialysis following peritoneal dialysis cessation in Australia and New Zealand: proposal for a standardized definition of technique failure. Perit Dial Int. 2016;36(6):623–30. 10.3747/pdi.2015.00218. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Li WY, Wang YC, Hwang SJ, Lin SH, Wu KD, Chen YM. Comparison of outcomes between emergent-start and planned-start peritoneal dialysis in incident ESRD patients: a prospective observational study. BMC Nephrol. 2017;18(1):359. 10.1186/s12882-017-0764-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Pérez Fontán M, Remón Rodríguez C, da Cunha Naveira M, Borràs Sans M, Rodríguez Suárez C, Quirós Ganga P, et al. Baseline residual kidney function and its ensuing rate of decline interact to predict mortality of peritoneal dialysis patients. PLoS One. 2016;11(7):e0158696. 10.1371/journal.pone.0158696. [DOI] [PMC free article] [PubMed]
- 10.Tanriover C, Ucku D, Basile C, Tuttle KR, Kanbay M. On the importance of the interplay of residual renal function with clinical outcomes in end-stage kidney disease. J Nephrol. 2022;35(9):2191–204. 10.1007/s40620-022-01388-9. [DOI] [PubMed] [Google Scholar]
- 11.Marrón B, Remón C, Pérez-Fontán M, Quirós P, Ortíz A. Benefits of preserving residual renal function in peritoneal dialysis. Kidney Int. 2008;73:S42–51. 10.1038/sj.ki.5002600. [DOI] [PubMed]
- 12.Liao CT, Chen YM, Shiao CC, Hu FC, Huang JW, Kao TW, et al. Rate of decline of residual renal function is associated with all-cause mortality and technique failure in patients on long-term peritoneal dialysis. Nephrol Dial Transpl. 2009;24(9):2909–14. 10.1093/ndt/gfp056. [DOI] [PubMed] [Google Scholar]
- 13.Wang J, Xie X, Yan X, Yang X, Zhang X, Chen J, et al. A fast decline of residual renal function in the first year is a predictor for early withdrawal from peritoneal dialysis in non-diabetic patients. Kidney Blood Press Res. 2019;44(1):12–21. 10.1159/000497807. [DOI] [PubMed] [Google Scholar]
- 14.Herget-Rosenthal S, von Ostrowski M, Kribben A. Definition and risk factors of rapidly declining residual renal function in peritoneal dialysis: an observational study. Kidney Blood Press Res. 2012;35(4):233–41. 10.1159/000332887. [DOI] [PubMed] [Google Scholar]
- 15.Hu SL, Joshi P, Kaplan M, Lefkovitz J, Poenariu A, Dworkin LD, et al. Rapid change in residual renal function decline is associated with lower survival and worse residual renal function preservation in peritoneal dialysis patients. Perit Dial Int. 2017;37(4):477–81. 10.3747/pdi.2016.00211. [DOI] [PubMed] [Google Scholar]
- 16.Danni S, Fei X. Analysis of predictive factors for the decline of residual renal function in new peritoneal dialysis patients. Chin J Nephrol. 2015;31(4):246–250. 10.3760/cma.j.issn.1001-7097.2015.04.002 [DOI]
- 17.Zhou D, Lei H, Wu S, Yang W, Cui W, Li L, et al. Influencing factors for residual kidney function in incident peritoneal dialysis patients: a systematic review and meta-analysis. Ren Fail. 2023;45(2):2286328. 10.1080/0886022X.2023.2286328. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Brown EA, Blake PG, Boudville N, Davies S, de Arteaga J, Dong J, et al. International society for peritoneal dialysis practice recommendations: prescribing high-quality goal-directed peritoneal dialysis. Perit Dial Int. 2020;40(3):244–53. 10.1177/0896860819895364. [DOI] [PubMed] [Google Scholar]
- 19.Dong X, Wu H, Ye H, Yi C, Diao X, Liu R, et al. Incidence and risk factors associated with technique failure in the first year of peritoneal dialysis: a single center retrospective cohort study in Southern China. BMC Nephrol. 2022;23(1):207. 10.1186/s12882-022-02833-4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.van Olden RW, Krediet RT, Struijk DG, Arisz L. Measurement of residual renal function in patients treated with continuous ambulatory peritoneal dialysis. J Am Soc Nephrol. 1996;7(5):745–50. 10.1681/ASN.V75745. [DOI] [PubMed] [Google Scholar]
- 21.Chung SH. Association between residual renal function, inflammation and patient survival in new peritoneal dialysis patients. Nephrol Dial Transpl. 2003;18(3):590–97. 10.1093/ndt/18.3.590. [DOI] [PubMed] [Google Scholar]
- 22.Wang AY, Woo J, Wang M, Sea MM, Sanderson JE, Lui SF, et al. Important differentiation of factors that predict outcome in peritoneal dialysis patients with different degrees of residual renal function. Nephrol Dial Transpl. 2005;20(2):396–403. 10.1093/ndt/gfh331. [DOI] [PubMed] [Google Scholar]
- 23.Wang AY, Wang M, Woo J, Lam CW, Lui SF, Li PK, et al. Inflammation, residual kidney function, and cardiac hypertrophy are interrelated and combine adversely to enhance mortality and cardiovascular death risk of peritoneal dialysis patients. J Am Soc Nephrol. 2004;15(8):2186–94. [DOI] [PubMed] [Google Scholar]
- 24.Santhakumaran T, Samad N, Fan SL. Hydration status measured by BCM: a potential modifiable risk factor for peritonitis in patients on peritoneal dialysis. Nephrol (Carlton). 2016;21(5):404–09. 10.1111/nep.12622. [DOI] [PubMed] [Google Scholar]
- 25.de Moraes TP, Figueiredo AE, de Campos LG, Olandoski M, Barretti P, Pecoits-Filho R. Characterization of the BRAZPD II cohort and description of trends in peritoneal dialysis outcome across time periods. Perit Dial Int. 2014;34(7):714–23. 10.3747/pdi.2013.00282. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Ng JK, Kwan BC, Chow KM, Pang WF, Cheng PM, Leung CB, et al. Asymptomatic fluid overload predicts survival and cardiovascular event in incident Chinese peritoneal dialysis patients. PLoS One. 2018;13(8):e0202203. [DOI] [PMC free article] [PubMed]
- 27.Pecoits-Filho R, Heimbürger O, Bárány P, Suliman M, Fehrman-Ekholm I, Lindholm B, et al. Associations between circulating inflammatory markers and residual renal function in CRF patients. Am J Kidney Dis. 2003;41(6):1212–18. 10.1016/S0272-6386(03)00353-6. [DOI] [PubMed] [Google Scholar]
- 28.Song Q, Yang X, Shi Y, Yan H, Yu Z, Li Z, et al. High intraperitoneal interleukin-6 levels predict ultrafiltration (UF) insufficiency in peritoneal dialysis patients: a prospective cohort study. Front Med. 2022;9:836861. 10.3389/fmed.2022.836861. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Han SH, Lee SC, Kang EW, Park JK, Yoon HS, Yoo TH, et al. Reduced residual renal function is associated with endothelial dysfunction in patients receiving peritoneal dialysis. Perit Dial Int. 2012;32(2):149–58. 10.3747/pdi.2010.00111. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
The data that support the findings of this study are available from the corresponding author upon reasonable request.