Prognostic significance of the C-reactive protein–triglyceride–glucose index for all-cause and cardiovascular mortality in peritoneal dialysis: a multicenter cohort study

Abstract

Chronic low-grade inflammation and insulin resistance (IR) frequently coexist in peritoneal dialysis (PD) and may drive excess mortality. The C-reactive protein–triglyceride–glucose index (CTI) integrates inflammatory (CRP) and metabolic (TyG) signals, but its prognostic value in PD is unknown. In a multicenter cohort of 2,240 Chinese PD patients, CTI was assessed 3 months after PD initiation and dichotomized at the median (5.09). Mortality outcomes were tracked through May 31, 2023. Analyses used Kaplan–Meier, Cox models (with and without propensity score matching [PSM]), prespecified subgroup analyses (sex, age, BMI, diabetes, hypertension), and restricted cubic splines (RCS) for dose–response. Incremental predictive value beyond established risk factors and single markers (hs-CRP, TyG) was evaluated using C-statistics, net reclassification improvement (NRI), and integrated discrimination improvement (IDI). There were 634 deaths (314 CVD). Cox regression analysis showed that a higher CTI was an independent risk factor for all-cause mortality and CVD mortality in PD patients. After full adjustment, compared with patients with CTI < 5.09, those with CTI ≥ 5.09 had a 34% higher all-cause mortality risk (HR 1.34, 95% CI 1.07–1.68) and a 39% higher CVD mortality risk (HR 1.39, 95% CI 1.01–1.93). RCS analysis revealed an approximately linear relationship between increased CTI and the elevated risk of both adverse clinical outcomes (P for nonlinearity > 0.05). The positive association between CTI and mortality was observed in both diabetic and non-diabetic subgroups (e.g., in diabetic patients: all-cause mortality HR 1.72, 95% CI 1.13–2.61; CVD mortality HR 2.11, 95% CI 1.21–3.68), with no statistically significant interaction by diabetes status (P for interaction > 0.05). Incremental predictive analyses showed that CTI had slightly higher C-indices than hs-CRP or TyG and provided only modest improvements in discrimination and reclassification. CTI independently and approximately linearly predicts all-cause and CVD mortality in PD, with consistent associations irrespective of diabetes status. It provides only modest incremental predictive value beyond hs-CRP and TyG, but its simplicity and cost-effectiveness support its potential use as a supplementary marker for risk stratification and individualized PD care.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-025-32217-1.

Introduction

Peritoneal dialysis (PD) is a key renal replacement therapy for patients with end-stage kidney disease (ESKD). Despite advances in technique and management that have improved short-term outcomes, the long-term prognosis of PD remains unsatisfactory: compared with age-matched members of the general population, PD patients have an approximately sevenfold higher risk of all-cause death¹, and cardiovascular disease (CVD) causes account for 47%–52.7% of deaths, making them the leading cause of mortality^2–4. Even after recognition and treatment of traditional risk factors (such as diabetes and hypertension), substantial “residual” cardiovascular (CV) risk persists in PD, suggesting that current risk stratification fails to capture critical nontraditional pathogenic axes.

Accumulating evidence indicates that low-grade chronic inflammation and insulin resistance (IR) commonly coexist and amplify each other in PD, jointly accelerating atherosclerosis and correlating closely with higher risks of all-cause and CVD mortality^5,6. The PD-specific pathological mechanisms include persistent inflammation caused by residual renal function loss and uremic toxin accumulation, as well as glucose absorption and metabolic disorders resulting from long-term exposure to high-glucose dialysate^7,8. These factors collectively form an “inflammation-metabolism” coupling⁹, amplifying CVD risks. Consequently, reliance on traditional risk factors or single biomarkers alone is insufficient; quantitative metrics integrating inflammatory and metabolic information are needed.

C-reactive protein (CRP), a recognized marker of inflammatory burden, has been linked to higher mortality in PD cohorts¹⁰. Chen et al. demonstrated that CRP is an independent predictor of all-cause mortality and major adverse CV events (MACE) in PD patients, with significantly increased risks of mortality and MACE as CRP quintiles rise¹¹. Liu et al.’s study found that CRP is a significant predictor of death in PD patients during a 2-year follow-up¹². The triglyceride-glucose (TyG) index, a convenient surrogate for insulin resistance (IR), is associated with dyslipidemia, oxidative stress, and adverse CVD outcomes^13,14. Previous studies on PD cohorts have confirmed that the TyG index can predict the risks of CVD and all-cause mortality in PD populations¹⁵. However, single metrics reflect only one dimension and cannot quantify the synergy between inflammation and metabolic derangement. The composite C-Reactive Protein–Triglyceride–Glucose Index (CTI), which integrates both inflammation and IR-related information, has shown better predictive performance than single indicators in prospective studies of the general population¹⁶. Its prognostic value and clinical applicability in PD patients, however, remain undefined. This study aims to address this gap by systematically evaluating the association between CTI and all-cause/CVD mortality in PD patients, testing its clinical potential as an accessible, low-cost prognostic tool that integrates multidimensional information to optimize long-term risk stratification and management in PD.

Materials and methods

Study design and population

This observational cohort enrolled 2,366 PD patients from 10 PD centers in China. Each center prospectively recorded baseline data and follow-up information according to a uniform protocol; a centralized team managed the database. Inclusion criteria were: (1) age ≥ 18 years; and (2) PD duration ≥ 3 months. Exclusion criteria were: (1) malignancy, tuberculosis, or other chronic wasting diseases; and (2) acute infections such as pneumonia or peritonitis. After rigorous screening, 2,240 patients were included and followed until May 31, 2023, or until a study endpoint occurred (death, kidney transplantation, conversion to hemodialysis, transfer to another center, or loss to follow-up) (Fig. 1). All participants provided written informed consent. The study adhered to the Declaration of Helsinki and received ethics approval from the relevant committees(https://jamanetwork.com/journals/jama/fullarticle/1760318).

Fig. 1.

The flow chart of the study illustrates the selection and exclusion process of patients. PD, peritoneal dialysis; CTI index, c-reactive protein–triglyceride–glucose index, CVD, cardiovascular disease.

Data collection

At 3 months after initiation of regular PD, the following were collected: sex, age, height, weight, body mass index (BMI), histories of diabetes, hypertension, and CVD, baseline medication use (including statins, antidiabetic therapy such as insulin, and ACE inhibitors), as well as laboratory variables including serum creatinine (Scr), hemoglobin (Hb), uric acid (UA), high-sensitivity CRP (hs-CRP), residual kidney function (RKF), estimated glomerular filtration rate (eGFR), and total Kt/V, among other basic data. Laboratory tests were performed using standard methods at each PD center.

CVD was defined according to the International Classification of Diseases, Tenth Revision (ICD-10), and included heart failure, myocardial infarction, stable or unstable angina, arrhythmias, cerebrovascular disease, and peripheral vascular disease.

Total Kt/V was calculated and recorded using PD Adequest software (Baxter Healthcare).

Definition of the CTI

The TyG index was computed as ln[triglycerides (mg/dL) × fasting plasma glucose (FPG) (mg/dL) / 2]¹⁷.

The CTI was defined as 0.412 × ln[CRP (mg/L)] + ln[triglycerides (mg/dL) × FPG (mg/dL) / 2]¹⁸.

All PD patients were dichotomized by the median CTI (low CTI: <5.09; high CTI: ≥5.09).

Study outcomes

Primary and secondary outcomes were all-cause mortality and CVD mortality, respectively. All-cause mortality included death from any cause. CVD mortality was defined as death due to heart failure, myocardial infarction, malignant arrhythmias, sudden cardiac death, cerebrovascular disease, or peripheral vascular disease. Survival status and causes of death were ascertained through regular outpatient visits to the PD centers (every 1–3 months) and nurse-led telephone follow-up. Follow-up ended on May 31, 2023.

Statistical analysis

Analyses were performed using SPSS (version 31.0) and R (version 4.4.1). Continuous variables are presented as mean ± standard deviation (SD) or median (interquartile range, IQR) for skewed distributions. Between-group comparisons used the t-test or Mann–Whitney U test, as appropriate. Categorical variables are reported as counts (percentages) and compared with χ² tests. Patients were divided into two groups by the median CTI.

Propensity score matching (PSM) was applied to balance baseline covariates between groups. We performed 1:1 nearest-neighbor matching without replacement on the logit of the propensity score, using a caliper width equal to 0.2 of the standard deviation of the logit-transformed propensity score. Covariate balance before and after matching was evaluated using standardized mean differences (SMDs) and corresponding SMD plots; after matching, the absolute SMD for all covariates was < 0.10, indicating adequate balance between groups. Kaplan–Meier curves with log-rank tests compared cumulative survival between CTI groups.

Univariable and multivariable Cox proportional hazards models were used to estimate hazard ratios (HRs) and 95% confidence intervals (CIs) for all-cause and CVD mortality. Three models were fitted: Model 1 was an unadjusted model; Model 2 was adjusted for age and sex; Model 3 was a fully adjusted model including the following covariates: age, sex, BMI, DM, hypertension, CVD, white blood cells, platelets, Hb, albumin, serum creatinine, serum uric acid, serum sodium, serum calcium, serum phosphorus, iPTH, Total Kt/V, eGFR, RKF.

To assess the incremental predictive value of CTI beyond established risk factors and commonly used biomarkers, we developed multivariable Cox models that included established risk factors together with CTI, hs-CRP, or TyG, individually and in combination. Discrimination and reclassification were evaluated using Harrell’s C-statistic (C-index), net reclassification improvement (NRI), and integrated discrimination improvement (IDI).

Restricted cubic spline (RCS) models evaluated potential nonlinear dose–response relationships between continuous CTI and outcomes. A two-sided P < 0.05 was considered statistically significant.

Results

Baseline characteristics

Among 2,240 included patients, 1,274 (56.88%) were male, with a mean age of 50.78 ± 14.59 years. Histories of diabetes, CVD, and hypertension were present in 20.98%, 15.40%, and 74.42% of patients, respectively. Based on the median CTI (5.09), patients were classified into low (< 5.09) and high (≥ 5.09) CTI groups. Significant between-group differences were observed for age, sex, BMI, diabetes, CVD history, and other clinical indices (white blood cells, platelets, hemoglobin, albumin, serum creatinine, uric acid, serum calcium, and eGFR) (Table 1). RKF was comparable between the two CTI groups. Regarding baseline medication use, the proportions of patients receiving ACE inhibitors and statins were similar, whereas insulin use was more frequent in the high-CTI group, in line with the higher prevalence of diabetes in this group (Table 1).

Table 1.

Baseline characteristics of all patients in this study.

Characteristic	Overall (n = 2240)	Lower CTI index (n = 1120)	Higher CTI index (n = 1120)	P-value
Demographics
Age (years)	51.00 (40.00–62.00)	48.00 (38.00–59.00)	54.00 (43.00–64.00)	< 0.001
Male (%)	1274 (56.88)	613 (54.73)	661 (59.02)	0.041
BMI (kg/m2 )	21.78 (19.84–24.17)	21.30 (19.56–23.42)	22.27 (20.20–24.80)	< 0.001
Comorbid
Diabetes (%)	470 (20.98)	194 (17.32)	276 (24.64)	< 0.001
Hypertension (%)	1667 (74.42)	822 (73.39)	845 (75.45)	0.286
History of CVD (%)	345 (15.40)	157 (14.00)	188 (16.80)	0.079
Laboratory variables
White blood cells (*109 /L)	6.30 (5.06–7.78)	6.06 (4.82–7.34)	6.63 (5.34–8.29)	< 0.001
Platelets (*109 /L)	197.00 (152.00-247.00)	194.00 (150.00-238.25)	202.00 (153.00-256.00)	0.008
HGB (g/L)	94.00 (78.00-112.00)	97.00 (79.00-115.00)	92.00 (77.00-108.00)	< 0.001
Albumin (g/L)	35.50 (31.50–39.00)	36.27 (32.60-39.98)	34.70 (30.60–38.20)	< 0.001
Serum creatinine (µmol/L)	763.00 (587.00-992.00)	792.00 (611.50-1025.28)	726.50 (568.00-955.25)	< 0.001
Uric acid (µmol/L)	403.00 (336.00-476.00)	394.50 (331.00-467.25)	413.00 (345.75–485.00)	< 0.001
Serum sodium (mmol/L)	140.30 (138.00-142.50)	140.45 (138.80-142.60)	140.00 (138.00-142.30)	0.035
Serum calcium (mmol/L)	2.14 (1.97–2.30)	2.16 (2.00-2.31)	2.12 (1.95–2.27)	< 0.001
Serum potassium (mmol/L)	4.08 (3.59–4.61)	4.10 (3.58–4.60)	4.03 (3.60–4.62)	0.383
Serum phosphorus (mmol/L)	1.59 (1.30–1.95)	1.60 (1.32–1.97)	1.58 (1.27–1.93)	0.182
iPTH (pg/mL)	190.56 (72.55-363.45)	203.59 (68.20-387.55)	180.00 (78.61-340.28)	0.130
Total Kt/V	2.16 (1.79–2.66)	2.22 (1.82–2.69)	2.12 (1.74–2.64)	0.033
eGFR (mL/min/1.73m2)	6.04 (4.54–8.26)	5.79(4.40–7.98)	6.23(4.69–8.50)	< 0.001
RKF (mL/min/1.73m2)	4.03 (2.11–8.11)	4.01 (2.10–7.53)	4.04 (2.11–8.90)	0.254
Medication use
CCB (%)	1615 (72.10)	824 (73.60)	791 (70.60)	0.132
ACEI (%)	125 (5.60)	69 (6.20)	56 (5.00)	0.269
ARB (%)	733 (32.70)	372 (33.20)	361 (32.20)	0.653
EPO (%)	1466 (65.40)	767 (68.50)	699 (62.40)	0.003
Iron preparations (%)	1285 (57.40)	660 (58.90)	625 (55.80)	0.146
Calcitriol (%)	745 (33.30)	422 (37.70)	323 (28.80)	< 0.001
Insulin (%)	282 (12.60)	107 (9.60)	175 (15.60)	< 0.001
Statins (%)	340 (15.20)	162 (14.50)	178 (15.90)	0.377

Data are shown as medians (25th-75th percentile) or n (%).

CTI: C-reactive protein-triglyceride-glucose index; BMI: body mass index; CVD: cardiovascular disease; HGB: hemoglobin, iPTH: serum intact parathyroid hormone; Kt/V: K, dialyzer clearance of urea; t: dialysis time; V: volume of distribution of urea; eGFR: estimated glomerular filtration rate; RKF: residual kidney function; CCB: Calcium Channel Blockers; ACEI: Angiotensin-Converting Enzyme Inhibitors; ARB: Angiotensin Receptor Blockers; EPO: Erythropoietin.

After PSM, 1,082 patients remained. Baseline characteristics were well balanced between CTI groups post-matching, whereas notable imbalances were present before matching (Table 2).

Table 2.

Baseline characteristics of the matched cohorts.

Characteristic	After matching
	Overall (n = 1082)	Lower CTI index (n = 541)	Higher CTI index (n = 541)	P-value
Demographics
Age (years)	52.00 (42.00–62.00)	51.00 (42.00–61.00)	53.00 (42.00–63.00)	0.313
Male (%)	585 (54.07)	291 (53.79)	294 (54.34)	0.903
BMI (kg/m2 )	21.67 (19.85–24.18)	21.15 (19.72–23.42)	22.06 (20.13–24.69)	0.850
Comorbid
Diabetes (%)	245 (22.64)	116 (21.44)	129 (23.84)	0.383
Hypertension (%)	851 (78.65)	425 (78.56)	426 (78.74)	1.000
History of CVD (%)	132 (12.20)	62 (11.46)	70 (12.94)	0.516
Laboratory variables
White blood cells (*109 /L)	6.33 (5.10–7.72)	6.37 (5.10–7.64)	6.28 (5.11–7.78)	0.621
Platelets (*109 /L)	199.00 (153.25–254.00)	199.00 (159.00-245.00)	199.00 (150.00-259.00)	0.861
HGB (g/L)	93.00 (76.25-109.75)	93.00 (77.00-110.00)	92.00 (76.00-109.00)	0.801
Albumin (g/L)	36.35 (32.60–40.00)	36.50 (33.00-39.90)	36.10 (32.30–40.00)	0.609
Serum creatinine (µmol/L)	722.50 (566.73-932.75)	730.00 (569.00-937.20)	710.00 (564.00-917.00)	0.635
Uric acid (µmol/L)	402.50 (340.00-476.00)	397.00 (340.00-472.00)	408.00 (340.00-479.00)	0.471
Serum sodium (mmol/L)	140.00 (138.00-142.00)	140.00 (138.20–142.00)	140.00 (138.00-142.00)	0.189
Serum calcium (mmol/L)	2.16 (2.00-2.32)	2.15 (1.99–2.32)	2.16 (2.00-2.33)	0.659
Serum potassium (mmol/L)	4.00 (3.50, 4.60)	4.05 (3.50, 4.60)	4.00 (3.50, 4.62)	0.816
Serum phosphorus (mmol/L)	1.60 (1.32–1.95)	1.61 (1.34–1.95)	1.60 (1.30–1.93)	0.256
iPTH (pg/mL)	198.74 (90.62-379.88)	199.20 (83.70–381.00)	197.78 (95.90-371.40)	0.875
Total Kt/V	2.25 (1.88–2.73)	2.32 (1.89–2.77)	2.17 (1.87–2.70)	0.079
eGFR (mL/min/1.73m2 )	5.93 (4.41–7.94)	5.86 (4.41–7.83)	6.01 (4.46–8.02)	0.617
RKF(mL/min/1.73m2 )	3.38 (2.03–6.06)	3.64 (2.12–6.02)	3.20 (1.91–6.07)	0.054
Medication use
CCB (%)	833 (77.00)	421 (77.80)	412 (76.20)	0.563
ACEI (%)	65 (6.00)	36 (6.70)	29 (5.40)	0.443
ARB (%)	397 (36.70)	200 (37.00)	197 (36.40)	0.900
EPO (%)	843 (77.90)	424 (78.40)	419 (77.40)	0.769
Iron Preparations (%)	691 (63.90)	352 (65.10)	339 (62.70)	0.448
Calcitriol (%)	302 (27.90)	184 (34.00)	118 (21.80)	< 0.001
Insulin (%)	146 (13.50)	66 (12.20)	80 (14.80)	0.247
Statins (%)	143 (13.20)	76 (14.00)	67 (12.40)	0.473

Data are shown as medians (25th-75th percentile) or n (%).

CTI: C-reactive protein-triglyceride-glucose index; BMI: body mass index; CVD: cardiovascular disease; HGB: hemoglobin, iPTH: serum intact parathyroid hormone; Kt/V: K, dialyzer clearance of urea; t: dialysis time; V: volume of distribution of urea; eGFR: estimated glomerular filtration rate; RKF: residual kidney function; CCB: Calcium Channel Blockers; ACEI: Angiotensin-Converting Enzyme Inhibitors; ARB: Angiotensin Receptor Blockers; EPO: Erythropoietin.

Kaplan–Meier survival analysis

Over a median follow-up of 64.00 (25.75, 96.00) months, 634 deaths (28.30%) occurred overall, including 314 CVD deaths (14.02%). Treating CTI as a categorical variable, and using data after PSM for Kaplan–Meier survival curve analysis, the high-CTI group showed significantly higher risks of all-cause mortality (log-rank, P < 0.001) and CVD mortality (log-rank, P = 0.014) than the low-CTI group (Fig. 2).

Fig. 2.

Kaplan-Meier survival analysis curves stratified by the median of CTI index for all-cause mortality (A), and CVD mortality (B). CTI index, c-reactive protein–triglyceride–glucose index, CVD, cardiovascular disease.

Association of CTI with all-cause and CVD mortality

Table 3 summarizes associations between CTI and mortality outcomes. In Cox models before and after PSM, higher CTI was positively associated with increased risks of both all-cause and CVD mortality. These associations persisted after additional adjustment for age, sex, and all covariates. Results were consistent whether CTI was modeled categorically or continuously. Per 1-unit increase in CTI, the risk of all-cause death increased by 22% (95% CI 1.03 - 1.43) and CVD death by 33% (95% CI 1.05 - 1.69). After full adjustment, compared with CTI <5.09, CTI ≥5.09 was associated with a 34% higher risk of all-cause death (95%CI: 1.07 - 1.68) and a 39% higher risk of CVD death (95% CI 1.01 - 1.93).

Table 3.

Association between CTI and mortality in unmatched and matched cohorts.

	Before matching			After matching
	CTI < 5.09	CTI ≥ 5.09		CTI < 5.09	CTI ≥ 5.09		HR for continuous CTI
All-cause mortality
Events/total	254/1120	380/1120	p-value	132/541	192/541	p-value		p-value
Model 1	Ref	1.49 (1.27 ~ 1.75)	< 0.001	Ref	1.48 (1.19 ~ 1.85)	< 0.001	1.37 (1.16 ~ 1.62)	< 0.001
Model 2	Ref	1.18 (1.01 ~ 1.39)	0.041	Ref	1.40 (1.13 ~ 1.75)	0.003	1.21 (1.03 ~ 1.43)	0.022
Model 3	Ref	1.12 (0.94 ~ 1.33)	0.177	Ref	1.34 (1.07 ~ 1.68)	0.010	1.22 (1.03 ~ 1.43)	0.019
CVD mortality
Events/total	123/1120	191/1120	p-value	63/541	92/541	p-value		p-value
Model 1	Ref	1.55 (1.24 ~ 1.95)	< 0.001	Ref	1.49 (1.08 ~ 2.06)	0.014	1.45 (1.14 ~ 1.85)	0.003
Model 2	Ref	1.27 (1.01 ~ 1.60)	0.042	Ref	1.42 (1.03 ~ 1.95)	0.033	1.29 (1.01 ~ 1.64)	0.038
Model 3	Ref	1.20 (0.95 ~ 1.53)	0.129	Ref	1.39 (1.01 ~ 1.93)	0.046	1.33 (1.05 ~ 1.69)	0.018

Model 1: Unadjusted for any covariates.

Model 2: Adjusted for sex, age.

Model 3: Model 2 plus BMI, DM, hypertension, CVD, white blood cells, platelets, HGB, albumin, serum creatinine, serum uric acid, serum sodium, serum calcium, serum phosphorus, iPTH, Total Kt/V, eGFR, RKF.

DM: diabetes mellitus; HR: hazard ratio.

In analyses stratified by diabetes status using matched data, CTI ≥ 5.12 in diabetic patients was significantly associated with both all-cause mortality (HR = 1.72, 95% CI: 1.13–2.61) and CVD mortality (HR = 2.11, 95% CI: 1.21–3.68). Among non-diabetic patients, CTI ≥ 5.08 was associated with all-cause mortality (HR = 1.37, 95% CI: 1.05–1.78), with a similar but non-significant trend for CVD mortality. These estimates remained broadly stable after additional adjustment for age, sex, and other covariates (Fig. 3 and Supplementary Fig. 1).

Fig. 3.

Cox proportional hazards regression models of different CTI index levels and all-cause mortality (A), and of CVD mortality (B). CTI index, c-reactive protein–triglyceride–glucose index, CVD, cardiovascular disease.

Discrimination and reclassification analyses

In multivariable models adjusted for established risk factors, we performed head-to-head comparisons of models including CTI, hs-CRP, and TyG for all-cause and CVD mortality, using the C-index, NRI, and IDI. For all-cause mortality, the C-index of the CTI model (0.7118) was slightly higher than that of the hs-CRP (0.7086) and TyG (0.7103) models, and adding CTI to models based on hs-CRP or TyG increased the C-index only modestly (by ≈ 0.002–0.004), with overlapping confidence intervals. NRI values were small and not statistically significant for both all-cause and CVD mortality. In contrast, IDI showed small but statistically significant improvements when CTI was added to models based on hs-CRP or TyG for all-cause mortality (IDI ≈ 0.0028–0.0052, p < 0.01), whereas IDI gains for CVD mortality were modest and not statistically significant. Overall, these results suggest that CTI performs at least as well as hs-CRP or TyG alone and provides modest incremental discriminative information, but does not dramatically improve discrimination or reclassification (Supplementary Table 1).

Subgroup analyses

After adjusting for confounding factors, we performed prespecified subgroup analyses stratified by age, sex, BMI, diabetes, and hypertension to further examine the relationship between the median-based categorical CTI and mortality outcomes. The forest plot (Fig. 3) showed generally consistent positive associations between higher CTI and both all-cause and CVD mortality across most subgroups, and no statistically significant interaction was detected for any subgroup (all P for interaction > 0.05).

RCS analyses

RCS models demonstrated that higher CTI was associated with increased risks of all-cause and CVD mortality (both P < 0.001), with an approximately linear increase as CTI rose (nonlinearity P = 0.478 and 0.837, respectively). The association with CVD mortality was at least as strong as that for all-cause mortality (Fig. 4), underscoring the need for vigilant CVD monitoring in patients with elevated CTI.

Fig. 4.

Restricted cubic spline curve of all-cause mortality based on CTI index (A), and CVD mortality (B). CI, confidence interval, CTI index, c-reactive protein–triglyceride–glucose index, CVD, cardiovascular disease.

Discussion

In this national multicenter PD cohort, we are the first to show that the composite CTI—integrating inflammation (hs-CRP) and IR (TyG)—is independently and approximately linearly associated with higher risks of all-cause and CVD death in PD. Across 2,240 patients with long-term follow-up, higher baseline CTI tracked with steadily increasing risks of both endpoints; this association showed the same trend in the DM subgroup; and the above results remained robust and reliable after PSM and multivariable adjustment. These findings align with recent reports in the general population and disease-specific cohorts showing that CTI or its components (hs-CRP and TyG) predict adverse CVD outcomes^16,17. Consistently, Tang et al. reported stable risk stratification by CTI for all-cause and CVD mortality in patients with coronary artery disease and type 2 diabetes¹⁹. In PD—a population characterized by the coexistence of inflammation and metabolic abnormalities—the composite phenotype that integrates inflammation-IR signaling (that is, CTI) may help to capture part of the “residual risk” that is not fully explained by traditional factors, by integrating both inflammation and insulin resistance into a single metric.

By combining hs-CRP (inflammation) and the TyG index (metabolic derangement), CTI offers a new perspective on the dual action of inflammation and metabolism. In PD, both inflammation and IR are pervasive. Prior studies have linked hs-CRP to CVD mortality in PD¹² and TyG to CVD events, death, and adverse outcomes in kidney disease^13,20. Zhou et al. showed that higher TyG predicted greater all-cause mortality in 6,697 patients with chronic heart failure²¹. In PD specifically, chronic exposure to high-glucose dialysate, glucose absorption, and uremia-related mechanisms jointly drive selective IR, lipotoxicity, oxidative stress, and endothelial dysfunction, thereby promoting CVD events^9,22. Yet most studies have focused on single indicators and cannot capture their synergy. Using a weighted logarithmic model to combine hs-CRP and TyG into CTI, we observed that each 1-unit increase in CTI corresponded to a significant, near-linear rise in mortality, with improved precision for predicting CVD death and a more comprehensive assessment of all-cause mortality risk in PD.

The inflammation–metabolism coupling in PD also has population-specific features: prolonged high-glucose dialysate exposure and glucose absorption exacerbate IR⁹, while uremic toxins, peritonitis, and catheter-related events sustain low-grade inflammatory activation^23–25. These processes amplify one another, forming the canonical “Inflammation-Metabolism-Atherosclerosis” pathway. The observed CTI–mortality association quantitatively reflects this pathological axis. As a surrogate of IR, TyG has been repeatedly validated^14,26 and associated with adverse outcomes in the general population, CKD, and PD. In PD, higher TyG at PD initiation has been linked to increased short-term CVD mortality²⁷; in real-world Continuous Ambulatory Peritoneal Dialysis (CAPD) cohorts, elevated TyG correlates positively not only with all-cause and CVD mortality but also with technique failure and peritonitis²⁸. Furthermore, TyG combined with BMI (TyG-BMI) has shown predictive value for death and CVD events in PD and CKD populations^29,30. Collectively, these data suggest that IR-centered metabolic dysfunction pervades the course of renal failure and relates to mortality, whereas CTI extends TyG by adding an inflammatory dimension, which may improve its ability to capture the combined inflammatory–metabolic burden³¹.

Given the high prevalence of diabetes in PD and its strong links to both all-cause and CVD mortality, diabetes is a key modifier when examining the CTI–mortality relationship. Chronic inflammation and IR in diabetes exacerbate CVD risk^32,33. In our subgroup analyses, higher CTI was significantly associated with both all-cause and CVD mortality among diabetic PD patients, and a similar positive association was observed in non-diabetic patients. Although the hazard ratios were numerically larger in the diabetic subgroup, formal tests for interaction between CTI and diabetes status were not statistically significant; therefore, these subgroup differences should be interpreted cautiously and regarded as hypothesis-generating rather than definitive evidence of effect modification. This pattern is broadly consistent with previous evidence that a higher CRP–TyG index is associated with adverse CVD outcomes in patients with coronary heart disease and type 2 diabetes¹⁹, supporting the concept that combined inflammatory–metabolic burden is particularly relevant in diabetes. In this context, CTI may be especially useful for risk assessment in diabetic PD patients because of their high baseline CVD risk and the coexistence of inflammation and insulin resistance, but overall our findings support CTI as a broadly applicable prognostic marker in PD irrespective of diabetes status.

Importantly, the modestly better performance of CTI compared with single markers in our analyses does not negate the independent value of hs-CRP or TyG. Prior PD and CKD studies have shown that time-weighted or repeated hs-CRP measurements outperform single-time assessments for long-term mortality prediction³⁴; TyG has been linked to all-cause and CVD death and even infectious complications such as PD-related peritonitis²⁸. In our study, a single CTI measurement at 3 months after PD initiation already yielded stable, linear associations, suggesting that even within a one-time measurement framework, the composite “inflammation-IR signal” has good discriminative capacity. Nevertheless, short-term fluctuations in inflammation and insulin resistance due to infections, dialysate adjustments, or disease progression may not be fully captured by this single time point, and time-averaged or serial CTI measurements could better reflect the true long-term exposure. Future work incorporating time-weighted CTI or serial measurements may further enhance predictive performance.

From a clinical perspective, CTI has two potential applications: (1) Risk stratification. Beyond traditional variables (age, CVD history, blood pressure, albumin, Kt/V, etc.), CTI can serve as a “second-tier” marker to identify otherwise occult high-risk individuals and to guide the intensity and timing of secondary CVD prevention. Concretely, patients with high baseline CTI could be triaged for earlier and more intensive metabolic optimization (tightened glycemic control, review and adjustment of lipid-lowering therapy), enhanced infection surveillance and prevention, or quicker consideration of glucose-sparing PD prescriptions (e.g., increased use of icodextrin or amino-acid solutions) to reduce peritoneal glucose exposure — interventions that have shown metabolic benefits in PD cohorts^35,36. (2) Monitoring and intervention assessment. Because both CTI dimensions (inflammation and IR) are obtainable from routine labs, CTI is amenable to integration into PD follow-up alongside dialysis prescriptions, dialysate selection, nutrition and exercise interventions, lipid and glycemic management, and infection prevention—functioning as a “command dashboard” for comprehensive care.

Existing peritoneal dialysis (PD) risk models (e.g., CANUSA-derived indices and the NECOSAD cohort model) are primarily based on traditional clinical variables such as age, comorbidities, nutritional indicators, and dialysis adequacy. While these models exhibit good explanatory power for patient survival and technique failure, there remains room for improvement—particularly in capturing the residual risk arising from the “inflammation-metabolism” interaction^37–40. Given that the present study demonstrates the C-reactive protein-triglyceride-glucose index (CTI) can reflect composite signals of inflammation and insulin resistance, CTI holds promise as a candidate variable to be incorporated into these traditional models, aiming to evaluate its actual added value in identifying high-risk PD patients.

Our findings support CTI as a potential biomarker to enable early identification of high-risk PD patients, particularly those with diabetes, for whom early intervention may delay all-cause and CVD mortality and improve quality of life. Current PD management focuses largely on blood pressure and glycemic control while under-addressing the combined abnormalities of inflammation and metabolism. Incorporating CTI provides a more holistic assessment to optimize therapeutic strategies and outcomes. Future research should validate CTI in other renal replacement modalities and explore its combined use with additional biomarkers. Studies of CTI dynamics and long-term trajectories may clarify its value for ongoing risk monitoring in PD.

Despite the promising results, CTI provides modest incremental value over traditional markers like hs-CRP or TyG. Although CTI offered slightly better discrimination, particularly in all-cause mortality, the improvements in C-statistics and NRI were small. This indicates that CTI adds some predictive benefit, but the increase in overall predictive power is relatively modest. Although this study demonstrated that CTI is an effective predictor of mortality, the use of a single baseline measurement may limit its ability to capture long-term fluctuations in inflammation and insulin resistance. Future studies incorporating time-averaged or serial CTI measurements may enhance the predictive accuracy and better reflect the true long-term risk. These results highlight that while CTI may be a useful supplementary tool, its practical utility in clinical settings requires further validation, particularly with regard to its long-term predictive capacity.

Although this multicenter cohort was based on retrospective records and established a temporal sequence in which baseline CTI (measured 3 months after PD initiation) preceded the outcome, a causal relationship cannot be firmly inferred. The inflammatory component of CTI (CRP) is an acute-phase reactant and may be elevated by subclinical or undocumented infections; therefore, reverse causality and residual confounding cannot be completely ruled out. Moreover, we assessed CTI only once and were unable to account for dynamic changes in inflammation and insulin resistance over time. Prior PD studies have suggested that time-averaged or serial CRP is superior to a single baseline value for mortality prediction⁴¹, and similar longitudinal assessments of CTI are needed to better capture long-term exposure.

In addition, although we adjusted for serum albumin and residual kidney function as surrogates of nutritional status and residual clearance, these markers may not fully capture their complex effects on both CTI and prognosis. We also did not collect several variables that may influence both CTI and prognosis, such as detailed PD modality (CAPD vs. APD), and dialysate glucose exposure. These factors have been associated with inflammatory burden, metabolic status, and outcomes in PD^42–46. Although we attempted to mitigate confounding by excluding patients with overt infections and known malignancies, restricting the cohort to patients who had survived at least 3 months on PD and were clinically stable at the time of CTI assessment, and using propensity score matching plus multivariable adjustment, unmeasured or imprecisely measured confounders (e.g., occult infections, frailty, recent hospitalizations) may still partly explain the observed associations. Future prospective studies should systematically collect these variables and apply time-updated models to more precisely delineate the independent effect of CTI.

Finally, kidney transplantation, conversion to hemodialysis, transfer to another center, and loss to follow-up were treated as censoring events in the cause-specific Cox models, and formal competing-risk regression analyses were not performed. This may have affected the estimation of absolute risk, particularly in the context of frequent competing events such as technique failure, and future studies using comprehensive competing-risk models are warranted. In conclusion, CTI is a promising composite index that provides modest incremental value in predicting mortality in PD patients. While the improvements in predictive performance are relatively small, CTI offers a practical tool for integrating inflammation and insulin resistance into routine clinical practice. Future research should explore its application in longitudinal studies with time-averaged measurements to further assess its predictive power and clinical utility.

Conclusions

This study is the first to validate the prognostic value of CTI in PD. Higher CTI was independently and approximately linearly associated with increased risks of all-cause and CVD mortality in PD patients. Similar positive associations were observed in both diabetic and non-diabetic subgroups, although formal interaction by diabetes status was not statistically significant. As a novel composite index reflecting both inflammatory and metabolic abnormalities, CTI offers a practical tool for CVD health monitoring and mortality risk assessment in PD. Our results provide a new perspective for CVD risk evaluation in PD and lay the groundwork for individualized management strategies. Given its accessibility and low cost, CTI may be considered for routine follow-up to enable early identification of high-risk patients and dynamic management. Future studies should further test CTI’s applicability across diverse patient groups and define specific strategies for its use in clinical practice.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

BMI

Body mass index

CAPD

Continuous ambulatory peritoneal dialysis

CKD

Chronic kidney disease

CRP

C-reactive protein

CTI

C-reactive protein–triglyceride–glucose index

CV

Cardiovascular

CVD

Cardiovascular diseases

DM

Diabetes mellitus

eGFR

estimated glomerular filtration rate

ESKD

End-stage kidney disease

FPG

Fasting plasma glucose

Hb

Hemoglobin

hs-CRP

High-sensitivity CRP

ICD

International classification of diseases

IDI

Integrated discrimination improvement

IQR

Interquartile range

IR

Insulin resistance

MACE

Major adverse CV events

NRI

Net reclassification improvement

PD

Peritoneal dialysis

PSM

Propensity score matching

RCS

Restricted cubic spline

RKF

Residual kidney function

Scr

Serum creatinine

SD

Standard deviation

TyG

Triglyceride–glucose

UA

Uric acid

Author contributions

Mengting Wang conceived the article, composed the main text, and generated (Figs. 1 and 2). Wenlong Qiu led the critical revision of the study design and statistical framework, participated in drafting, and generated the remaining figures and supplements. Rui Chu, and Lu Li participated in the investigation, review, and editing, and also compiled the supplementary figures and tables. Qingdong Xu contributed to the review and editing. Yueqiang Wen contributed to the review and editing. Xianfeng Wu contributed to the review and editing. Xiaojiang Zhan contributed to the review and editing. Fenfen Peng contributed to the review and editing. Xiaoyang Wang contributed to the review and editing. Juan Wu contributed to the review and editing. Ning Su contributed to the review and editing. Xiaoran Feng contributed to the review and editing. Xingming Tang contributed to the review and editing. Qian Zhou contributed to the review and editing. Bin Wu contributed to the review and editing. Na Tian was responsible for funding acquisition, project administration, supervision, review, and editing, and also carried out a critical review and amendment of the essential intellectual content. The authors certify that this manuscript, together with its related data, graphical materials, and tabular information, has not been previously published and is not under consideration for publication elsewhere.

Funding

This study was supported by the Project of the National Natural Science Foundation of China (Project No.: 82360153), the Ningxia Autonomous Region Key R&D Program of Ningxia Science and Technology Department (Project No.: 2022BEG03120), the Key Project of Ningxia Natural Science Foundation (Project No.: 2022AAC02062), and the Ningxia Natural Science Project of Ningxia Science and Technology Department (Project No.: 2021AAC03378).

Data availability

All data that support the findings of this study are available from the corresponding author, upon reasonable request.

Declarations

Competing interests

The authors declare no competing interests.

Ethics statement

All procedures involving human participants in this study were performed in accordance with the ethical standards of the institution and conducted in line with the 1975 Declaration of Helsinki. The study has been approved by the Ethics Committee of the General Hospital of Ningxia Medical University (application ID: 2022-410-01). All participants provided written informed consent to participate in this study.

Consent to participate

Informed consent was obtained from all individual participants included in the study.