Remnant cholesterol inflammatory index and cardiovascular disease among adults with cardiovascular-kidney-metabolic syndrome stages 0–3: a nationwide prospective cohort study

Abstract

The remnant cholesterol inflammatory index (RCII), integrating lipid and inflammatory pathways, may serve as a novel biomarker for cardiovascular risk. Yet, its role in adults across cardiovascular-kidney-metabolic (CKM) syndrome stages 0–3 remains unclear. We included 6,062 participants aged ≥ 45 years and free of cardiovascular disease at baseline from the China Health and Retirement Longitudinal Study (CHARLS). RCII was calculated as remnant cholesterol × high-sensitivity C-reactive protein/10 and log-transformed (lnRCII). Incident cardiovascular disease (CVD) events were ascertained through follow-up interviews (2011–2020). Cox proportional hazards models estimated hazard ratios (HRs) for CVD across lnRCII tertiles and per 1-standard-deviation (SD) increase, adjusting sequentially for demographic and lifestyle factors. Restricted cubic splines, Kaplan–Meier curves, and subgroup analyses were performed. During a median 9-year follow-up, 1751 CVD events occurred (cumulative incidence 28.9%). Higher lnRCII was consistently associated with elevated CVD risk. Each 1-SD increment corresponded to a 9% higher risk in crude and minimally adjusted models (HR  1.09, 95% CI 1.06–1.12), and a 5% higher risk after full adjustment (HR 1.05, 95% CI 1.01–1.08). Cox proportional hazards regression analysis demonstrated that, after adjustment for confounders, the risk of CVD was significantly higher in Q3 than in Q1 (HR 1.22, 95% CI 1.09–1.37, P < 0.001). Spline analysis indicated a linear dose–response, and subgroup findings were broadly consistent across demographic and clinical strata. In additional analyses, lnRCII’s ability to identify cardiovascular events is very poor (AUC = 0.557). After adding lnRCII to the traditional risk model, the C index, net reclassification improvement and comprehensive identification improvement have little change. Elevated lnRCII was associated with an increased risk of incident CVD among adults with CKM stages 0–3. However, its predictive performance was limited, so RCII should currently be regarded as an epidemiologic marker rather than a stand-alone tool for individual risk stratification.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-025-32281-7.

Introduction

The cardiovascular-kidney-metabolic (CKM) syndrome, a novel disease concept proposed by the American Heart Association (AHA) in October 2023, is defined as a health disorder—a systemic condition arising from the interactive interplay between metabolic dysfunction, chronic kidney disease (CKD), and cardiovascular disease(CVD)¹.CKM syndrome has emerged as a prevailing clinical paradigm globally. The pathophysiological mechanism of CKM syndrome is more complicated, showing a vicious cycle of “metabolic abnormality-renal injury-cardiovascular injury”. Metabolic dysfunction (such as insulin resistance and dyslipidemia) can damage renal filtration function by inducing oxidative stress and endothelial dysfunction², and the progression of chronic kidney disease will further aggravate the accumulation of metabolic toxins and inflammatory state, which will worsen the structure and function of the cardiovascular system³. The pathological interaction of multiple organs greatly increases the risk of cardiovascular events and pushes cardiovascular mortality to the forefront of disease burden^4,5.

Remnant cholesterol (RC) mainly exists in lipoprotein residues rich in triglycerides, which is an important indicator of abnormal lipid metabolism⁶. In the context of metabolic disorders, the scavenging ability of RC is significantly reduced, which can accelerate the progression of atherosclerosis by damaging the vascular endothelium, promoting foam cell formation, and activating an inflammatory response^7,8. Previous studies have clearly confirmed that RC is an independent risk factor of atherosclerotic CVD, especially in people with multi-system disorders such as CKM, which is more closely related to CVD^9–11. Remnant cholesterol inflammation index (RCII) is a composite index that integrates RC level and inflammatory biomarkers¹². In view of the synergistic pathogenic characteristics of lipid disorder and chronic inflammation in CKM syndrome, the inflammatory reaction caused by abnormal metabolism and the inflammatory amplification effect caused by renal function damage overlap each other, making it difficult to fully reflect the complex risk of the disease by relying solely on RC detection¹³. RCII may represent a composite measure that captures both lipid load and inflammatory state, potentially reflecting the risk of CVD in CKM patients.

At present, there is a lack of research on the relationship between RCII and CVD risk in the CKM population, especially prospective studies among individuals with CKM stages 0–3. The value of RCII in this context therefore remains uncertain. Leveraging data from a large, nationally representative prospective cohort, we aimed to characterise the association between RCII and incident CVD among adults with CKM stages 0–3, who are at increased cardiometabolic risk but free of established CVD at baseline. By situating RCII within the cardiovascular–kidney–metabolic (CKM) framework, this study evaluates whether an integrative lipid–inflammation index is related to cardiovascular risk along the earlier stages of the CKM continuum. Such evidence may help to clarify the potential role of lipid–inflammation interactions in the development of CVD in CKM and provide a useful basis for future research.

Methods

Study design and population

We leveraged data from the China Health and Retirement Longitudinal Study (CHARLS), a nationally representative longitudinal cohort of middle-aged and older Chinese adults. Designed by Peking University’s National School of Development, CHARLS employed a multi-stage stratified probability sampling framework across 150 counties and 450 village committees in 28 provinces. The baseline survey (Wave 1) was conducted from June 2011 to March 2012, establishing a core database through standardized interviews, physical examinations, and biospecimen collection from 17,708 community-dwelling participants aged ≥ 45 years. Follow-up surveys were conducted in Wave 2 (2013), Wave 3 (2015), Wave 4 (2018), and Wave 5 (2020), documenting health transitions. Peking University’s Institutional Review Board approved the protocol (IRB00001052-11015), and all participants provided written informed consent¹⁴. For the current prospective analysis, participants were excluded based on the following criteria: (1) age < 45 years at baseline; (2) Miss TC, HDL-C, LDL-C, hs-CRP; (3)Participants with CKM stages 0–3 at baseline were included, while those with CKM stage 4 were excluded; (4)Participants who missing visits from 2011 to 2020. The final analytical cohort consisted of 6062 eligible participants (Fig. 1).

Fig. 1.

Flow chart of study subjects.

Calculation of RCII

The RCII was constructed as an integrated metric combining metabolic and inflammatory markers, defined as the product of RC and high-sensitivity C-reactive protein (hs-CRP) divided by 10 in the present study¹³. Fasting venous blood specimens were collected and processed by professional medical personnel from the research team at a centralized laboratory in accordance with standardized operational procedures. Blood lipid parameters, including total cholesterol (TC), low-density lipoprotein cholesterol (LDL-C) and high-density lipoprotein cholesterol (HDL-C), were quantified using enzymatic colorimetric assays. Levels of hs-CRP (mg/L) were measured via an immunoturbidimetric method on a clinical chemistry analyzer (Hitachi 7180).

RC (mg/dL) was calculated using the formula: RC = TC − (HDL-C + LDL-C).RCII was computed as RCII = RC(mg/dL)×hs-CRP(mg/L)/10. Due to the skewed distribution of RCII, values were natural log-transformed (lnRCII) for all statistical analyses. Participants with missing data for any of the components (TC, HDL-C, LDL-C, or hs-CRP) were excluded to ensure the validity of index calculations.

Definition of CKM syndrome stages 0–3

In accordance with the definition of CKM syndrome outlined by the AHA, the syndrome is categorized into stages 0 to 4 based on the severity of cardiovascular, renal, and metabolic system involvement¹⁵. Stage 0 is characterized by the absence of any risk factors, with individuals presenting normal weight glucose levels, blood pressure (BP), lipid profiles, and renal function, along with no signs of CVD. Stage 1 includes obese individuals who have impaired glucose metabolism, which serves as an indicator of excessive or dysfunctional adiposity. Stage 2 involves individuals at moderate to high risk of CKD (estimated glomerular filtration rate (eGFR) 30–60 ml/min/1.73 m² and/or with self-reported CKD diagnosis) together with metabolic risk factors such as hypertriglyceridemia, hypertension (HTN), metabolic syndrome, and type 2 diabetes (T2D). Stage 3 comprises individuals with subclinical CVD, including those at high 10-year CVD risk (Framingham risk score ≥ 21.5/21.6 in females/males) or those with extremely high-risk CKD (eGFR < 30 ml/min/1.73 m²). The Framingham risk score evaluates CVD risk using multiple risk factors. Finally, stage 4 consists of individuals with clinical CVD. Since this study focused on incident CVD, participants with baseline clinical CVD (stage 4) were excluded, and analyses were conducted among individuals with CKM stages 0–3 as a single combined population. Table S1 provides a detailed description of the Methods for evaluating CKM syndrome.

Ascertainment of outcomes

The primary endpoint was incident CVD. Aligning with established literature^16,17, CVD is based on a heart attack or stroke diagnosed by a self-reported doctor. In each wave, participants were asked, “Have you been told by your doctor that you have been diagnosed with a heart condition/stroke ?” “Are you now undergoing any of the following treatments (Taking Chinese Traditional Medicine/Taking Western Modern Medicine/Other treatments/None of the Above) to treat heart condition/stroke or its complications?” Or “Compared to before, how’s your heart condition/stroke doing now?” Participants who reported being diagnosed with a heart condition/stroke, who indicated specific treatment for a heart condition/stroke or those who reported a current condition, were considered to have CVD.

Data collection

Comprehensive data were collected through standardized interviews, physical examinations, and centralized laboratory analyses. Demographic characteristics included age, gender (male/female), residence (city/village), education level (below primary/primary/middle/high school+), marital status (married/other), and lifestyle factors (smoking, drinking). Anthropometric measurements comprised body mass index (BMI) and waist circumference. Comorbidity profiles documented physician-diagnosed hypertension, diabetes, dyslipidemia, cancer, liver disease, kidney disease, and digestive disease via self-reported physician diagnosis. Laboratory assessments covered complete blood counts (WBC, platelets), glucose metabolism (fasting glucose, HbA1c), lipid profiles (TG, TC, HDL-C, LDL-C, remnant cholesterol), inflammatory markers (hs-CRP), and renal function (BUN, eGFR, uric acid, serum creatinine), with all assays conducted under ISO 15,189 accreditation.

Statistical analysis

Given the skewed distribution of the RCII, all statistical analyses were performed using the lnRCII. Participants were categorized into three tertiles (Q1–Q3) based on lnRCII levels, with Q1 serving as the reference group. To test for linear trends across tertiles, the median lnRCII value of each tertile was assigned to all participants within that group, and this median value was incorporated into the model as a continuous variable.

Cox proportional hazards models were used to assess the association between lnRCII and incident CVD, with hazard ratios (HRs) and 95% confidence intervals (CIs) calculated. Follow-up duration was defined as the period from baseline to the first occurrence of CVD or the final follow-up (2020). Three sequential adjusted models were constructed to control for confounding factors: Crude Model: No covariates adjusted. Model 1 (demographic and lifestyle-adjusted model): Adjusted for age, sex, residence (urban/rural), smoking status (yes/no), drinking status (yes/no), marital status (married/other), and educational level (below primary school/primary school/junior high school/senior high school and above). Model 2 (fully adjusted model): Further adjusted for BMI, hypertension (yes/no), diabetes (yes/no) and dyslipidemia (yes/no) in addition to the covariates in Model 2. In Model 2, covariates were selected based on prior literature and their established relevance to cardiometabolic risk. We intentionally did not include renal function parameters (e.g., eGFR, uric acid, BUN) as additional covariates, because kidney function is already embedded within the CKM staging framework. Including these parameters could result in overadjustment or collinearity, potentially attenuating the true association between RCII and cardiovascular outcomes within the CKM construct. The proportional hazards assumption of the Cox models was verified using Schoenfeld residuals, with no significant violations observed (all p > 0.05).

Restricted cubic spline (RCS) models with 3 knots (placed at the 10th, 50th, and 90th percentiles of lnRCII) were employed to explore potential non-linear associations between lnRCII and incident CVD. These models were adjusted for all covariates in Model 3, and formal non-linearity tests were conducted to determine whether the association deviated from linearity.

Kaplan–Meier (K–M) survival curves were plotted to compare CVD-free survival probabilities across the three lnRCII tertiles, and log-rank tests were used to assess the significance of survival differences between groups.

Subgroup analyses were performed to investigate the potential effect modification of key variables on the association between lnRCII and CVD. Subgroups were defined by sex (male/female), smoking status (yes/no), marital status (married/other), educational level (low: below primary school/primary school; high: junior high school/senior high school and above), BMI (< 25/25–30/≥30 kg/m²), hypertension (yes/no), diabetes (yes/no), dyslipidemia (yes/no), and CKM stage (stages 0/1/2/3). Interaction terms (lnRCII × subgroup variable) were included in the Cox proportional hazards models, and likelihood ratio tests were used to calculate p-values for interactions.

All statistical analyses were conducted using R software (version 4.4.2) and Free Statistics software (version 2.1.1; Beijing Free Clinical Medical Technology Co., Ltd.). Prior to formal analysis, data distribution was evaluated via normality tests: normally distributed continuous variables were presented as mean ± standard deviation (SD), while skewed continuous variables were expressed as median (interquartile range, IQR). Between-group comparisons for normally distributed variables were performed using one-way analysis of variance (ANOVA), and for skewed variables using the Kruskal–Wallis rank sum test. Categorical variables were described as frequency (percentage), with between-group comparisons conducted using the chi-squared test; Fisher’s exact test was used if expected frequencies were < 5.

Results

Baseline characteristics

A total of 6062 participants were included at baseline. Table 1 summarises baseline characteristics by tertiles of lnRCII. Individuals in the highest tertile (Q3) tended to live in rural areas and exhibited a clearly more adverse metabolic and inflammatory profile compared with those in the lowest tertile (Q1). Specifically, BMI, waist circumference, triglycerides, total cholesterol, fasting glucose, HbA1c, hsCRP, uric acid, platelets, and remnant cholesterol all rose progressively across tertiles (P < 0.001). Conversely, HDL-C declined steadily with increasing lnRCII.The distribution of comorbidities mirrored these biochemical trends. The prevalence of hypertension, diabetes and dyslipidaemia increased monotonically from Q1 to Q3, while proportions of participants in advanced CKM stages were also higher in the upper tertiles. By contrast, age and sex distributions were similar across groups, suggesting that differences were not driven simply by demographic structure. Drinking behaviour was less frequent in Q3, whereas smoking patterns were broadly similar. Collectively, these patterns indicate that elevated lnRCII identifies a subgroup with clustered metabolic abnormalities and systemic inflammation, consistent with its hypothesised pathophysiological role.

Table 1.

Baseline characteristics of the study individuals in CVD incidence.

Characteristics	Total(n = 6062)	Q1(n = 2021)	Q2(n = 2021)	Q3(n = 2020)	P
Age, median, years	57.0 (51.0, 64.0)	57.0 (51.0, 63.0)	57.0 (51.0, 64.0)	58.0 (52.0, 64.0)	0.012
Gender, n(%)					0.179
Female	3281 (54.1)	1097 (54.3)	1063 (52.6)	1121 (55.5)
Male	2781 (45.9)	924 (45.7)	958 (47.4)	899 (44.5)
Residence, n (%)					< 0.001
City	918 (15.1)	247 (12.2)	299 (14.8)	372 (18.4)
Village	5144 (84.9)	1774 (87.8)	1722 (85.2)	1648 (81.6)
Smoking, n(%)					0.217
No	4250 (70.1)	1436 (71.1)	1388 (68.7)	1426 (70.6)
Yes	1812 (29.9)	585 (28.9)	633 (31.3)	594 (29.4)
Drinking, n(%)					0.038
No	4023 (66.4)	1304 (64.5)	1339 (66.3)	1380 (68.3)
Yes	2039 (33.6)	717 (35.5)	682 (33.7)	640 (31.7)
Marital status, n (%)					0.504
Married	5192 (85.6)	1716 (84.9)	1740 (86.1)	1736 (85.9)
Others	870 (14.4)	305 (15.1)	281 (13.9)	284 (14.1)
Education level, n(%)					0.865
Below Primary School	2889 (47.7)	960 (47.5)	975 (48.2)	954 (47.2)
Primary School	1329 (21.9)	446 (22.1)	434 (21.5)	449 (22.2)
Middle School	1225 (20.2)	419 (20.7)	394 (19.5)	412 (20.4)
High school and above	619 (10.2)	196 ( 9.7)	218 (10.8)	205 (10.1)
BMI, median(IQR), kg/m²	23.2 (21.0, 25.9)	22.0 (20.1, 24.1)	23.5 (21.2, 26.0)	24.4 (22.0, 27.3)	< 0.001
Comorbidity
Hypertension, n (%)					< 0.001
No	4485 (74.0)	1637 (81.0)	1497 (74.1)	1351 (66.9)
Yes	1577 (26.0)	384 (19.0)	524 (25.9)	669 (33.1)
Diabetes, n (%)					< 0.001
No	5724 (94.4)	1946 (96.3)	1913 (94.7)	1865 (92.3)
Yes	338 (5.6)	75 (3.7)	108 (5.3)	155 (7.7)
Dyslipidemia, n (%)					< 0.001
No	5532 (91.3)	1909 (94.5)	1866 (92.3)	1757 (87.0)
Yes	530 (8.7)	112 (5.5)	155 (7.7)	263 (13.0)
Cancer, n (%)					0.912
No	6018 (99.3)	2007 (99.3)	2007 (99.3)	2004 (99.2)
Yes	44 (0.7)	14 (0.7)	14 (0.7)	16 (0.8)
Liver disease, n (%)					0.545
No	5877 (96.9)	1953 (96.6)	1965 (97.2)	1959 (97.0)
Yes	185 (3.1)	68 (3.4)	56 (2.8)	61 (3.0)
Kidney disease, n (%)					0.667
No	5733 (94.6)	1905 (94.3)	1911 (94.6)	1917 (94.9)
Yes	329 (5.4)	116 (5.7)	110 (5.4)	103 (5.1)
Digestive system disease, n (%)					0.016
No	4787 (79.0)	1553 (76.8)	1617 (80.0)	1617 (80.0)
Yes	1275 (21.0)	468 (23.2)	404 (20.0)	403 (20.0)
Laboratory tests
WBC, median(IQR),10⁹/L	6.0 (5.0, 7.2)	5.5 (4.6, 6.5)	6.0 (5.0, 7.1)	6.5(5.5, 7.8)	< 0.001
Platelets, median(IQR),10⁹/L	207.0(162.0, 254.0)	200.0(160.0, 249.0)	207.0(162.0, 252.0)	213.0(164.0, 262.0)	< 0.001
FPG, median(IQR), mg/dL	103.0(95.0, 113.9)	100.8(93.6, 109.1)	102.2(95.0, 112.5)	107.1 (97.9, 121.9)	< 0.001
HbA1c, median(IQR),%	5.1 (4.9, 5.4)	5.1 (4.8, 5.3)	5.1 (4.9, 5.4)	5.2 (4.9, 5.6)	< 0.001
TG, median(IQR), mg/dL	108.9(76.1, 162.0)	77.9 (61.1, 101.8)	116.8(85.0, 155.8)	162.0(108.0,236.5)	< 0.001
TC, mean (SD), mg/dL	195.4 (37.7)	189.1 (34.8)	195.5 (36.2)	201.5 (40.7)	< 0.001
HDL-C, mean(SD), mg/dL	51.3 (15.1)	59.0 (14.0)	51.1 (13.8)	43.8 (13.5)	< 0.001
LDL-C, mean(SD), mg/dL	117.0 (35.1)	117.4 (31.8)	119.4 (34.1)	114.2 (38.8)	< 0.001
hs-CRP, median(IQR), mg/L	1.0(0.5, 2.0)	0.5(0.3, 0.7)	0.9(0.6, 1.4)	2.7(1.6, 4.8)	< 0.001
BUN, median(IQR), mg/dL	15.2(12.6,18.3)	15.4(12.7,18.8)	15.2(12.7,18.4)	14.9(12.5,17.8)	0.001
eGFR, mean(SD), mL/min/1.73 m²	108.7 (28.4)	111.9 (27.1)	107.5 (27.4)	106.8 (30.3)	< 0.001
UA, median(IQR), mg/dL	4.3 (3.6, 5.2)	4.0 (3.3, 4.7)	4.3 (3.6, 5.2)	4.6 (3.8, 5.5)	< 0.001
Scr, mean(SD), mg/dL	0.8 (0.2)	0.8 (0.2)	0.8 (0.2)	0.8 (0.2)	< 0.001
RC, median(IQR), mg/dL	20.9(12.0,33.6)	11.2(6.6,16.6)	22.4(15.1,31.7)	34.0(22.4,52.3)	< 0.001
lnRCII, median(IQR), mg/dL	0.8 (1.5)	−0.7 (1.0)	0.7 (0.3)	2.4 (0.8)	< 0.001
CKM stage, n (%)					< 0.01
0	432 ( 7.1)	256 (12.7)	125 ( 6.2)	51 ( 2.5)
1	929 (15.3)	514 (25.4)	282 (14.0)	133 ( 6.6)
2	3194 (52.7)	899 (44.5)	1099 (54.4)	1196 (59.2)
3	1507 (24.9)	352 (17.4)	515 (25.5)	640 (31.7)

Association of RCLL index with CVD in CKM syndrome patients

In this study, a total of 1751 participants developed CVD. We constructed three Cox proportional hazards models to evaluate the association between lnRCII and CVD risk (Table 2). Across all models, the results consistently demonstrated a robust and statistically significant positive association (P < 0.001).In the crude model, compared with the lowest tertile (Q1), participants in Q2 and Q3 had a 25% (HR 1.25, 95% CI 1.11–1.39, P < 0.001) and 43% (HR 1.43, 95% CI 1.28–1.59, P < 0.001) higher risk of CVD, respectively. In Model 1 (adjusted for age, sex, residence, smoking, drinking, marital status, and education), the estimates attenuated slightly but remained significant: Q2 vs. Q1: HR 1.21 (95% CI 1.09–1.36, P < 0.001); Q3 vs. Q1: HR  1.36 (95% CI 1.22–1.52, P < 0.001). In Model 2 (further adjusted for BMI, hypertension, diabetes, dyslipidaemia, cancer, liver disease, kidney disease, and digestive disorders), the associations weakened modestly but remained robust: Q2 vs. Q1: HR 1.16 (95% CI 1.04–1.30, P = 0.005); Q3 vs. Q1: HR 1.22 (95% CI 1.09–1.37, P < 0.001). Overall, lnRCII showed a stable positive association with CVD risk, and the trend remained significant after sequential adjustment for multiple potential confounders.

Table 2.

Multivariable Cox regression analysis of LnRCII association with CVD incidence with CKM syndrome (0–3 stages).

InRCII	Crude model		Model 1		Model 2
	95%CI	P	95%CI	P	95%CI	P
CVD incidence
Continuous	1.09 (1.06, 1.13)	< 0.001	1.09 (1.06, 1.12)	< 0.001	1.05 (1.01–1.08)	0.005
Q1	Ref		Ref		Ref
Q2	1.25 (1.11, 1.39)	< 0.001	1.21 (1.09, 1.36)	< 0.001	1.16(1.04, 1.30)	0.005
Q3	1.51 (1.35–1.70)	< 0.001	1.36 (1.22, 1.52)	< 0.001	1.22(1.09, 1.37)	< 0.001
p for trend		< 0.001		< 0.001		< 0.001

Crude model: no covariates were adjusted; Model 1: Adjusted for Age, Gender, Residence, Smoking, drinking, Marital, Education; Model 2: Adjusted for Age, Gender, Residence, Smoking, drinking, Marital, Education, BMI, Hypertension, Diabetes, Dyslipidemia; lnRCII: Natural-log-transformed remnant cholesterol inflammatory index.

RCS and threshold effect analysis

Restricted cubic spline analyses provided a more nuanced view of the risk gradient. Figure 2 illustrates a steadily increasing hazard of CVD across the observed lnRCII range, with no evidence of a threshold effect. However, the association was slightly attenuated after further adjustment for clinical covariates (Model 2 P-overall = 0.024).

Fig. 2.

RSC with 3 knots showing the connection between lnRCII and CVD incidence. Crude model: no covariates were adjusted; Model 1: Adjusted for Age, Gender, Residence, Smoking, drinking, Marital, Education; Model 2: Adjusted for Age, Gender, Residence, Smoking, drinking, Marital, Education, BMI, Hypertension, Diabetes, Dyslipidemia.

The shape of the curve suggested an approximately linear association, and formal testing confirmed that the non-linearity component was not statistically significant. Importantly, the risk increase was evident even within the lower-to-middle spectrum of lnRCII, reinforcing the notion that there is no safe range once systemic inflammation and remnant cholesterol burden are elevated.

K–M survival curves

K-M curves of CVD-free survival stratified by lnRCII tertiles showed early and sustained divergence(Fig. 3). By year 5, survival probabilities were already noticeably lower in Q3 compared with Q1, and the gap widened progressively with longer follow-up. The intermediate group (Q2) showed a trajectory between Q1 and Q3 but consistently below the lowest tertile. Log-rank testing confirmed that the differences across groups were highly significant (P < 0.001). These findings complement the Cox regression results and highlight the long-term prognostic separation associated with elevated lnRCII.In additional analyses, the discriminative performance of lnRCII for incident CVD was poor. The area under the ROC curve was approximately 0.55 for lnRCII, with similarly modest values for hsCRP and RC, indicating only limited ability to distinguish between individuals who did and did not develop CVD. When lnRCII was added to the fully adjusted base model, the increases in C-index (ΔC-index = 0.004), 8-year AUC (0.55 vs. 0.54), net reclassification improvement (NRI = 0.036), and integrated discrimination improvement (IDI = 0.002) were numerically small. These findings suggest that, despite statistically significant associations, the incremental predictive value of lnRCII for individual risk stratification is limited (Supplementary Marital: Table S5 and Fig. S2).

Fig. 3.

Kaplan–Meier survival curves showing the CVD-free survival rates for the different InRCII classification groups.

Subgroup analyses

We further evaluated whether the association between lnRCII and CVD varied across clinically relevant subgroups (Fig. 4). Consistent positive associations were observed in men and women, in smokers and non-smokers, across categories of BMI, and among those with and without hypertension, diabetes or dyslipidaemia. Effect sizes were generally similar, with HRs ranging from 1.06 to 1.14 per 1-SD increment of lnRCII. Tests for interaction showed no significant heterogeneity for sex (P = 0.386), smoking (P = 0.113), BMI (P = 0.660), hypertension (P = 0.889), diabetes (P = 0.801), dyslipidaemia (P = 0.904) or CKM stage (P = 0.457). The observed interactions by marital status and education were exploratory and may reflect chance findings given the limited statistical power and the issue of multiplicity inherent in subgroup analyses.

Fig. 4.

Subgroup analysis of the association between lnRCII and incident CVD. Hazard ratios (HRs) and 95% confidence intervals (CIs) per 1-standard-deviation increase in lnRCII are shown across subgroups defined by sex, smoking status, marital status, education, BMI, hypertension, diabetes, dyslipidaemia, and CKM stage. Models were adjusted for covariates as specified. P values for interaction are presented.

Discussion

Our findings demonstrate that higher RCII levels were significantly associated with increased risk of CVD across adults with CKM stages 0–3. The RCS analysis further revealed a robust dose–response relationship between lnRCII and incident CVD, suggesting that the risk elevation was consistent across the exposure distribution without clear evidence of a threshold effect. These results suggest that RCII is associated with cardiovascular risk in individuals without advanced CKM stage. By integrating both lipid metabolism and systemic inflammation, RCII reflects two key biological processes central to atherosclerosis, and its levels are associated with cardiovascular risk beyond traditional factors.

Our observations are consistent with and extend previous research that has independently linked remnant cholesterol and systemic inflammation with adverse cardiovascular outcomes. Remnant cholesterol has been increasingly recognized as a causal risk factor for atherosclerotic cardiovascular disease, independent of low-density lipoprotein cholesterol (LDL-C)^18–20. Similarly, elevated hsCRP has been shown to predict incident CVD across diverse populations. Recent studies combining lipid and inflammation indices, such as the TyG index, the C-reactive protein–triglyceride glucose index, and the atherogenic index of plasma (AIP), have reported significant associations with coronary heart disease, stroke, and all-cause mortality^21–24. By integrating remnant cholesterol and hsCRP into a single metric, RCII provides an innovative biomarker that unites two key biological axes of atherosclerosis.

RCII may reflect the synergistic atherogenic effects of cholesterol-rich remnant lipoproteins and chronic low-grade inflammation. Remnant lipoproteins penetrate the arterial intima more efficiently than LDL particles, are avidly taken up by macrophages without prior oxidative modification, and accelerate foam cell formation. In parallel, CRP is not merely a marker but may act as a mediator of vascular inflammation, promoting endothelial dysfunction, leukocyte recruitment, and complement activation. Elevated RCII therefore identifies individuals with both lipid overload and heightened inflammatory milieu, a combination that may destabilize atherosclerotic plaques and increase the propensity for thrombotic events. Novel lipid-lowering therapies, such as PCSK9 inhibitors, fibrates, and angiopoietin-like protein 3 (ANGPTL3) inhibitors, substantially lower remnant cholesterol, while anti-inflammatory agents, including colchicine and IL-1β antagonists, have demonstrated cardiovascular benefit by reducing residual inflammatory risk. The dual targeting of lipid and inflammatory pathways may therefore offer synergistic risk reduction, and RCII levels may help identify individuals who could benefit from combined interventions targeting lipid and inflammatory pathways. However, cardiovascular events in this study were identified based on self-reported, physician-diagnosed conditions during follow-up, which may have introduced a certain degree of misclassification bias. Although validation studies in the CHARLS cohort have demonstrated reasonable agreement between self-reported and hospital-recorded cardiovascular outcomes, underreporting and recall errors are inevitable. Such non-differential misclassification would likely bias the associations toward the null, leading to underestimation of the true strength of the relationship between RCII and CVD risk and partially explaining the modest hazard ratios (HRs ≈ 1.05–1.22) observed in this study.

In conclusion, our study provides strong evidence that RCII is an integrative biomarker associated with cardiovascular risk, combining two pathophysiological domains central to atherosclerosis. These results emphasize the importance of targeting both lipid metabolism and inflammation in the prevention of CVD and highlight RCII as a promising marker for risk stratification and therapeutic monitoring in individuals with early CKM syndrome.

Strength and limitation

This study benefits from a large, nationally representative cohort with long-term follow-up and comprehensive covariate assessment, which supports the robustness of the estimates. However, several limitations should be acknowledged. First, as an observational study, although multiple covariates were adjusted for, residual confounding cannot be completely excluded, and further prospective and mechanistic research is needed to clarify whether the observed associations are causal. Second, the CHARLS database focuses on middle-aged and older adults in China, which may limit the generalizability of our findings to younger populations and to other countries or ethnic groups. Third, the accuracy of the results is influenced by disease diagnoses that rely on self-reported physician diagnoses during follow-up interviews, which may introduce misclassification bias. Fourth, the predictive performance of lnRCII for incident CVD was limited, and the incremental changes in discrimination and reclassification metrics after adding lnRCII to conventional models were small, indicating that its stand-alone clinical usefulness for individual risk prediction is restricted. Further studies in diverse cohorts are needed to validate RCII as an epidemiologic marker of cardiovascular risk and to determine whether it provides meaningful incremental information beyond established risk scores.

Conclusions

Our findings indicate that higher RCII levels are independently and approximately linearly associated with the incidence of CVD among adults with CKM stages 0–3. RCII appears to capture lipid- and inflammation-related cardiovascular risk beyond traditional cardiometabolic factors; however, the magnitude of association and its predictive performance are modest, so RCII should currently be regarded as an epidemiologic marker rather than a stand-alone tool for individual risk stratification.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

Z.Y., L.N., Y.Z.: data collection, data analysis, manuscript writing; X.C.: perform interval-censored analysis; L.W.: data analysis, manuscript editing; Q.Z. and J.F.: data collection; J.W.: project development, manuscript editing. All authors have read and approved this manuscript.

Funding

This work was supported by the Scientific Research Project of Shanxi Provincial Health and Wellness Committee (2024040).

Data availability

Data supporting the results of this study are available from the CHARLS repository. The data can be accessed by registering and submitting a request through the official CHARLS website at http://charls.pku.edu.cn.

Declarations

Ethics approval and consent to participate

The data used in this study were approved by the Biomedical Ethics Review Board of Peking University (approval number: IRB00001052-11015). Written informed consent was gained from all participants. All human research methods described in the manuscript were in accordance with national laws and the 1964 Declaration of Helsinki and its later amendments.

Competing interests

The authors declare no competing interests.