Remnant cholesterol inflammation index may be a superior biomarker to C-reactive protein-triglyceride glucose index in predicting carotid atherosclerosis risk Authors: Nimei Zeng, Ting Tian, Quan Zhou, Yuanli Li, Ruixin Zhou, Fangfei Xie, Renfang Han, Yi Wang, Yun Wang, Wei Ni, Jingyi Fan Published: 2026 ABSTRACT AbstractObjectiveWe aimed to investigate and compare the performance of C-reactive protein-triglyceride glucose index (CTI) and remnant cholesterol inflammation index (RCII) in carotid atherosclerotic plaque (CAP) risk prediction.MethodsWe conducted a case-control study (2,888 age- and sex-matched pairs) and a prospective cohort study of 4,418 participants without CAP (median follow-up: 767 days). Multivariable logistic and Cox regression were used to assess the associations of CTI and RCII with CAP. Receiver operating characteristic (ROC), net reclassification index (NRI) and integrated discrimination improvement (IDI) were calculated.ResultsIn the case-control study, increased CTI (OR: 1.45, 95% CI: 1.16–1.80) and RCII (OR: 1.95, 95% CI: 1.66–2.29) were significantly associated with increased CAP risk. Similar associations with CAP were also observed in the cohort for CTI (HR: 1.79, 95% CI: 1.32–2.44) and RCII (HR: 2.02, 95% CI: 1.60–2.54). Moreover, adding CTI and RCII showed improvements in CAP risk prediction beyond the traditional model, with the area under the ROC curve of 0.681 for CTI, 0.704 for RCII and 0.702 for RCII+CTI (p < 0.05). RCII showed a superior ability to identify high-risk individuals with CAP compared with CTI, with a significant NRI of 0.119 (95% CI: 0.045–0.191) and IDI of 0.006 (95% CI: 0.002–0.011).ConclusionsBoth CTI and RCII were significantly associated with increased risk of CAP, and RCII may be a superior biomarker to CTI in CAP risk prediction. Future large-scale cohort studies are needed to validate the predictive value of CTI and RCII for subclinical atherosclerosis.Clinical trial numberNot applicable.Supplementary informationThe online version contains supplementary material available at 10.1186/s12902-026-02353-4.Keywords: C-reactive protein-triglyceride glucose index, Remnant cholesterol inflammation index, Carotid atherosclerotic plaque, Risk prediction Introduction Atherosclerosis is the leading cause of cardiovascular disease (CVD) worldwide, including coronary heart disease, ischemic stroke and peripheral arterial disease [1]. Carotid atherosclerotic plaque (CAP), as the manifestation of atherosclerosis in carotid arteries which can be early detected by using non-invasive ultrasonography, can serve as a basis for CVD prevention [2]. The global prevalence of CAP was estimated to be 21.1% in 2020, translating to a substantial burden of potential cardiovascular events [3]. Inflammation plays a central role in the pathophysiological process of atherosclerosis development [4]. High sensitivity C-reactive protein (hsCRP) is a well-established inflammatory biomarker that has been used as an indicator for CVD risk in clinical practice [5, 6]. Several studies showed that hsCRP was associated with CAP incidence and instability [7, 8], but the findings are inconsistent [9, 10]. Reliable biomarkers are still needed to identify high-risk individuals of CAP and to guide risk management in clinical practice. Triglyceride-glucose (TyG) index has been recognized as an alternative biomarker of insulin resistance (IR), which contributes to the development of atherosclerosis [11]. Accumulating epidemiological evidence demonstrated that TyG index was associated with carotid atherosclerosis and its progression, including increased carotid intima-media thickness, CAP formation and stenosis [12–16]. A recent study showed a co-exposure effect of TyG index and hsCRP on cardiovascular events, recommending the combined assessment of TyG index and hsCRP to further stratify cardiovascular risk [17]. C-reactive protein-triglyceride glucose index (CTI) is a composite biomarker derived from integrating TyG index and hsCRP, which has demonstrated significant predictive value for worse outcome in patients with cancer [18, 19]. Since then, multiple studies have evaluated the potential value of CTI for CVD risk prediction. These studies indicated that CTI was associated with incident stroke and CVD in the Chinese population from the CHARLS study [20, 21]. In addition, it was also associated with heart failure, coronary heart disease (CHD), all-cause and CVD mortality among US adults from NHANES [22–24]. Remnant cholesterol (RC) is an emerging biomarker that contributes to the residual risk of CVD and subclinical atherosclerosis, despite the optimal level of low-density lipoprotein cholesterol (LDL-C) [25–27]. A review from observational and genetic studies revealed a causal relationship between elevated RC and low-grade inflammation, suggesting that RC combined with low-grade inflammation could further increase the risk of CVD [28]. Recently, Chen et al. [29] integrated RC and hsCRP as a novel biomarker named remnant cholesterol inflammation index (RCII) to reflect the co-effect of RC and low-grade inflammation [29]. It was reported that RCII could provide comprehensive assessment of incident stroke compared to evaluating RC and hsCRP individually [29]. In addition, RCII could offer incremental predictive value for stroke prognosis beyond RC and hsCRP alone [30]. Although aforementioned population-based studies have confirmed that CTI and RCII were effective predictors for both incidence and prognosis of CVD, the association of CTI and RCII with CAP among the general population remains unclear. Moreover, no studies have directly compared the performance of these two composite biomarkers in CAP risk prediction within the same population. Atherosclerosis develops early and persists silently over decades before the onset of clinical cardiovascular events. Therefore, exploring the predictive value of these biomarkers for subclinical atherosclerosis is more meaningful in early prevention of CVD. To bridge this knowledge gap, we aimed to investigate the association of CTI and RCII with risks of CAP and compared the performance of the two composite biomarkers in CAP risk prediction, thereby providing evidence for the risk management of subclinical atherosclerosis in practice. Materials and methods Study design and population We conducted a two-stage study based on the Health Examination project from 2020 to 2024, recruiting participants from the general population at the Health Management Center of Affiliated Suzhou Hospital of Nanjing Medical University. The study process is illustrated in Fig. 1. A total of 13,198 individuals aged 18 to 80 years were enrolled. All participants underwent face-to-face interviews and comprehensive physical examinations. We excluded participants with (i) missing clinical biomarker data (n = 265), (ii) a history of cancers (n = 26), or (iii) a history of CVDs (n = 66), resulting in a final cohort of 12,841 participants for further analysis. Flow chart of the study In the first stage of the study, we adopted a case-control design matching controls to CAP cases in a 1:1 ratio according to age and gender. The matching procedure was conducted using the “MatchIt” package in R, with a caliper width set to 0.2 to exclude any pairs whose propensity scores differed by more than 20% of the standard deviation, thereby ensuring matching precision. This process resulted in a total of 2,888 well-matched case-control pairs for subsequent analysis. In the second stage, we conducted a prospective cohort study involving the 9,521 participants who were free of CAP from the first stage. Follow-up data were collected through periodic health examinations. Notably, individuals enrolled in 2024 were excluded due to insufficient time for routine follow-up. Exclusions were applied as follows: (i) participants enrolled in 2024 (n = 1,604); (ii) those who did not complete follow-up health examinations (n = 2,819); and (iii) those without carotid ultrasound examination during follow-up (n = 680). After these exclusions, 4,418 participants were included in the final analysis. The baseline characteristics of the analyzed cohort (n = 4,418) were systematically compared with those of participants lost to follow-up (n = 3,499) in Supplementary Table 1. No statistically significant differences were observed between the two groups in demographic characteristics, metabolic indicators, comorbidities, or lifestyle factors (all p > 0.05). These findings suggest that, with respect to the measured key baseline variables, loss to follow-up was likely random. Therefore, its potential impact on the internal validity of this study is probably limited, although residual selection bias due to unmeasured confounders cannot be entirely excluded. All research was conducted in accordance with Declaration of Helsinki. This study was approved by the Ethics Committee of the Affiliated Suzhou Hospital of Nanjing Medical University (Ethical Approval Number: K-2026-161-K01), all participants received and signed written informed consent. Data collection Information on smoking, alcohol consumption, tea drinking, and physical exercise was collected through face-to-face interviews. The questionnaire also covered medical history. Age and gender data were retrieved from the hospital registration system. Height and weight were measured using an Omron ultrasonic weight scale (model HNH-318). During measurement, participants were instructed to stand fully upright without bending or overextending, with eyes level and head in a neutral position. Body mass index (BMI) was calculated as weight in kilograms divided by the square of height in meters (kg/m2). Waist circumference (WC) was measured twice at the mid-point between the lowest rib and the iliac crest during gentle expiration. The average of two measurements was used. Systolic and diastolic blood pressure (SBP and DBP) were measured twice with an electronic sphygmomanometer, and the mean value was recorded. Blood samples were collected after at least 10 hours of fasting and analyzed by the Department of Laboratory Medicine. Serum levels of hsCRP, fasting plasma glucose (FPG), glycated hemoglobin (HbA1c), triglycerides (TG), total cholesterol (TC), high-density lipoprotein cholesterol (HDL-C), and LDL-C were determined using a Beckman Coulter AU5800 automated biochemical analyzer. Hypertension was defined as either a self-reported history of hypertension, SBP ≥ 140 mmHg, or DBP ≥ 90 mmHg [31]. Diabetes was defined as a self-reported history of diabetes, FPG ≥ 7.0 mmol/L, or HbA1c ≥ 6.5% [32]. CTI and RCII calculation The CTI and RCII indexes were calculated according to the established equation: TyG = Ln (TG [mg/dL] × FPG [mg/dL]/2) CTI = 0.412 × Ln (hsCRP [mg/L]) + TyG RC = TC (mg/dL) - (HDL-C [mg/dL] + LDL-C [mg/dL]) RCII = RC (mg/dL) × hsCRP (mg/L)/10 The normality of continuous variables was assessed by using the Shapiro-Wilk test. CTI followed a normal distribution, whereas RCII showed significant skewness; thus, natural log-transformation (LnRCII) was applied for all subsequent analyses [33]. Atherosclerosis assessment Carotid ultrasound was performed by trained and qualified clinical sonographers using a GE HealthCare Voluson Premier-R2 color Doppler ultrasound system with a probe frequency of 7.5–10.0 MHz. The measured parameters included: (i) vascular internal diameter and carotid intima-media thickness (CIMT; 3 locations: common carotids, bifurcations and internal carotid arteries); (ii) the location, size, morphology, number, and echogenicity of plaques, as well as the presence of filling defects in color flow signals, with measurement of the vascular stenosis rate. CAP was defined as a focal CIMT ≥ 1.5 mm encroaching into the lumen, or at least 0.5 mm of local protrusion or 50% local thickening compared with the surrounding CIMT [34]. For the cohort study, follow-up information was obtained from the participant’s periodic health examination, and the diagnostic procedures were as described above. The date of complete follow-up was December 31, 2024, with a median follow-up of 767 (interquartile range: 399–1420) days. Statistical analysis The basic characteristics of the study participants were summarized using median (lower quartile, upper quartile) for continuous variables and number (percentage) for categorical variables. Differences were evaluated using Wilcoxon test for continuous variables and Chi-square test for categorical variables. Multivariable logistic regression models were performed to compute odds ratios (ORs) and 95% confidence intervals (95% CIs). While for incident CAPs, we used Cox proportional hazards regression models to estimate hazard ratios (HRs) and 95% CIs. Two models were constructed for analyses. Model 1 was adjusted for gender and age. Model 2 was further adjusted for BMI, WC, SBP, DBP, LDL-C, TG, HDL-C, smoking, alcohol consumption, tea drinking, physical exercise, history of diseases (hypertension and diabetes). Additionally, based on the quartiles of CTI or RCII, participants were divided into four groups. Using the lowest quartile as the reference, the ORs and HRs of upper quartiles were calculated. Restricted cubic spline (RCS) analysis with four knots placed at the 5th, 35th, 65th, and 95th percentiles was also used to assess the potential nonlinear relationship between the two indexes and the risk of the CAP. Subgroup analyses were performed based on gender, age (younger: 0.05). In the cohort study, significant baseline differences were observed in all variables between incident CAP cases and non-cases (all p < 0.05). Similarly, CTI [8.91 (8.44, 9.41) vs. 8.65 (8.13, 9.21), p < 0.001] and RCII [1.23 (0.55, 1.94) vs. 1.00 (0.31, 1.77), p < 0.001] were significantly elevated in the incident CAP group (Table 1; Supplementary Figure 2). Table 1Baseline characteristics of study populationVariablesCase-control studyCohort studyAllCasesControlsPAllIncident CAPWithout CAPPN57762888288844188573561CTI8.99 (8.51,9.52)9.06 (8.57,9.58)8.93 (8.46,9.45)<0.0018.70 (8.19,9.25)8.91 (8.44,9.41)8.65 (8.13,9.21)<0.001RCII1.49 (0.73,2.22)1.67 (0.84,2.40)1.35 (0.63,2.04)<0.0011.04 (0.35,1.83)1.23 (0.55,1.94)1.00 (0.31,1.77)<0.001Age, years56.63 (50.31,62.74)56.68 (50.15,63.60)56.60 (50.42,62.07)0.07943.85 (34.67,53.20)53.72 (44.98,62.04)41.51 (33.28,50.55)<0.001BMI, kg/m224.80 (23.00,26.60)24.80 (23.00,26.60)24.80 (23.10,26.60)0.84124.20 (22.20,26.30)24.70 (22.40,26.60)24.10 (22.00,26.20)<0.001WC, cm87.00 (81.00,93.00)87.00 (81.00,93.00)87.00 (81.00,93.00)0.36285.00 (76.00,91.00)87.00 (80.00,92.00)83.00 (75.00,90.00)<0.001TG, mg/dL132.85 (94.77,189.54)134.63 (96.54,191.31)130.20 (93.00,187.80)0.036116.91 (81.48,171.83)133.74 (93.00,193.97)112.48 (79.71,166.51)<0.001TC, mg/dL201.47 (176.72,228.15)204.95 (178.66,232.12)198.76 (175.56,223.90)<0.001194.90 (172.85,220.03)203.40 (177.50,228.15)193.30 (171.30,217.70)<0.001HDL-C, mg/dL49.50 (42.54,58.01)49.11 (42.54,57.71)49.50 (42.54,58.39)0.46550.27 (42.92,59.17)48.72 (42.92,57.62)50.66 (43.31,59.17)0.001LDL-C, mg/dL117.94 (99.00,136.89)119.10 (99.00,138.83)116.78 (98.61,134.96)0.001115.62 (98.32,133.80)122.58 (105.96,142.31)113.30 (97.06,131.87)<0.001SBP, mmHg131.00 (120.00,143.00)132.00 (122.00,145.00)130.00 (119.00,141.00)<0.001123.00 (113.00,134.00)128.00 (118.00,142.00)121.00 (111.00,132.00)<0.001DBP, mmHg80.00 (73.00,87.00)80.00 (73.00,87.00)79.00 (72.00,86.00)0.02675.00 (68.00,83.00)78.00 (71.00,86.00)74.00 (67.00,82.00)<0.001FPG, mg/dL98.46 (91.22,109.26)99.18 (91.44,110.52)97.56 (90.90,107.82)<0.00193.78 (87.66,101.34)95.94 (89.46,105.66)93.42 (87.30,100.44)<0.001HbA1c, %5.51 (5.24,5.89)5.54 (5.28,5.96)5.47 (5.21,5.83)<0.0015.26 (5.05,5.52)5.38 (5.13,5.72)5.24 (5.03,5.48)<0.001hsCRP, mg/L1.41 (0.83,2.50)1.53 (0.88,2.75)1.28 (0.79,2.24)<0.0011.10 (0.70,1.92)1.26 (0.77,2.16)1.07 (0.68,1.83)<0.001RC, mg/dL31.32 (21.27,43.70)33.26 (22.82,46.40)29.39 (20.11,40.99)<0.00126.30 (17.79,36.35)27.84 (18.95,37.12)25.91 (17.79,36.35)0.005Sex0.909<0.001Male, n (%)4050 (70.12)2027 (70.19)2023 (70.05)2720 (61.57)596 (69.54)2124 (59.65)Female, n (%)1726 (29.88)861 (29.81)865 (29.95)1698 (38.43)261 (30.46)1437 (40.35)Smoking<0.001<0.001No, n (%)3845 (66.57)1794 (62.12)2051 (71.02)3686 (83.43)672 (78.41)3014 (84.64)Yes, n (%)1931 (33.43)1094 (37.88)837 (28.98)732 (16.57)185 (21.59)547 (15.36)Alcohol drinking<0.001<0.001No, n (%)4443 (76.92)2153 (74.55)2290 (79.29)3770 (85.33)682 (79.58)3088 (86.72)Yes, n (%)1333 (23.08)735 (25.45)598 (20.71)648 (14.67)175 (20.42)473 (13.28)Tea drinking0.604<0.001No, n (%)2132 (36.91)1076 (37.26)1056 (36.57)2310 (52.29)390 (45.51)1920 (53.92)Yes, n (%)3644 (63.09)1812 (62.74)1832 (63.43)2108 (47.71)467 (54.49)1641 (46.08)Physical exercise0.2020.017No, n (%)1550 (26.84)797 (27.60)753 (26.07)1324 (29.97)286 (33.37)1038 (29.15)Yes, n (%)4226 (73.16)2091 (72.40)2135 (73.93)3094 (70.03)571 (66.63)2523 (70.85)History ofhypertension<0.001<0.001No, n (%)3453 (59.78)1605 (55.57)1848 (63.99)3482 (78.81)559 (65.23)2923 (82.08)Yes, n (%)2323 (40.22)1283 (44.43)1040 (36.01)936 (21.19)298 (34.77)638 (17.92)History ofdiabetes<0.001<0.001No, n (%)4990 (86.39)2416 (83.66)2574 (89.13)4194 (94.93)783 (91.37)3411 (95.79)Yes, n (%)786 (13.61)472 (16.34)314 (10.87)224 (5.07)74 (8.63)150 (4.21)CTI: C-reactive protein-triglyceride glucose index; RCII: natural log-transformed remnant cholesterol inflammation index; BMI: body mass index; WC: waist circumference; TG: triglycerides; TC: total cholesterol; HDL-C: high-density lipoprotein cholesterol; LDL-C: low-density lipoprotein cholesterol; SBP: systolic blood pressure; DBP: diastolic blood pressure; FPG: fasting plasma glucose; HbA1c: glycated hemoglobin; hsCRP: high sensitivity C-reactive protein; RC: remnant cholesterolNumerical variables are presented as median (lower quartile, upper quartile), and categorical variables are presented as count (percentage)RCII was natural log-transformed for all analyses Baseline characteristics of study population History of hypertension History of diabetes CTI: C-reactive protein-triglyceride glucose index; RCII: natural log-transformed remnant cholesterol inflammation index; BMI: body mass index; WC: waist circumference; TG: triglycerides; TC: total cholesterol; HDL-C: high-density lipoprotein cholesterol; LDL-C: low-density lipoprotein cholesterol; SBP: systolic blood pressure; DBP: diastolic blood pressure; FPG: fasting plasma glucose; HbA1c: glycated hemoglobin; hsCRP: high sensitivity C-reactive protein; RC: remnant cholesterol Numerical variables are presented as median (lower quartile, upper quartile), and categorical variables are presented as count (percentage) RCII was natural log-transformed for all analyses Associations between CTI or RCII and CAP As shown in Fig. 2A and B, per standard deviation (SD) increase in CTI (OR: 1.26, 95% CI: 1.15–1.38, p < 0.001) and RCII (OR: 1.30, 95% CI: 1.22–1.38, p < 0.001) was significantly associated with increased risks of CAP in the case-control analysis. When both indexes were categorized into quartiles, a significant graded increase in CAP risk was observed from the first to the fourth quartile for CTI and RCII (P for trend ≤ 0.001). Compared with participants in the lowest quartile, those in the highest quartile had significantly higher risks of CAP, with ORs of 1.45 (95% CI: 1.16–1.80) and 1.95 (95% CI: 1.66–2.29), respectively. RCS analysis revealed a linear dose-response relationship between CTI and CAP risk (P for nonlinear = 0.391; Fig. 3A), while a nonlinear association was observed for RCII (P for nonlinear = 0.023; Fig. 3B), indicating a steeper increase in risk at higher levels of RCII. Associations of CTI and RCII with CAP risk: (A-B) case-control study; (C-D) cohort study. CTI: C-reactive protein-triglyceride glucose index; RCII: natural log-transformed remnant cholesterol inflammation index; OR: odds ratio; HR: hazard ratio; 95% CI: 95% confidence interval. Data are presented as count, OR (95%CI) or HR (95%CI). RCII was natural log-transformed for all analyses. Model 1 was adjusted for gender and age. Model 2 was further adjusted for BMI, WC, SBP, DBP, LDL-C, TG, HDL-C, smoking, alcohol consumption, tea drinking, physical exercise, hypertension and diabetes Restricted cubic spline regression results of associations between CTI and RCII with CAP risk: (A-B) case-control study; (C-D) cohort study. CTI: C-reactive protein-triglyceride glucose index; RCII: natural log-transformed remnant cholesterol inflammation index; OR: odds ratio; HR: hazard ratio; 95% CI: 95% confidence interval. ORs (95%CI) and HRs (95%CI) were derived from restricted cubic spline regression, with knots placed at the 5th, 35th, 65th and 95th percentiles of CTI or RCII. Panels were adjusted for the same variables as model 2 in Fig. 2. RCII was natural log-transformed for all analyses As of December 31, 2024, a total of 857 incident CAP cases were identified among 4,418 eligible participants, contributing 10,484 person-years of follow-up. Consistent with the first stage, per SD increase in CTI (HR: 1.39, 95% CI: 1.22–1.59, p < 0.001) and RCII (HR: 1.31, 95% CI: 1.20–1.43, p < 0.001) was significantly associated with an elevated risk of CAP (Fig. 2C–D). Compared with the lowest quartile, participants in the highest quartile of both CTI (HR: 1.79, 95% CI: 1.32–2.44) and RCII (HR: 2.02, 95% CI: 1.60–2.54) had a significantly increased risk of CAP. Furthermore, significant dose-response trends were observed across quartiles (P for trend ≤ 0.001). RCS analysis also showed a linear relationship between CTI and CAP risk (P for nonlinear = 0.942; Fig. 3C), as well as a nonlinear association between RCII and CAP risk (P for nonlinear = 0.013; Fig. 3D). Subgroup, sensitivity analyses and predictive ability comparison Subgroup analyses were conducted to assess whether the associations between CTI or RCII and CAP were modified by specific characteristics, including age, gender, BMI, WC, smoking, alcohol consumption, tea drinking, physical exercise, hypertension and diabetes. The distributions of CTI and RCII across subgroups in both case-control and cohort analyses are presented in Supplementary Tables 2 and 3. As shown in Supplementary Figure 3, the associations between CTI or RCII and CAP risk remained significant in most subgroups, indicating the robustness of the findings. Significant interactions between both CTI and RCII with age were observed in both stages, with stronger effects evident in the younger group (all Pinteraction < 0.05). In the case-control study, CTI exhibited heterogeneity across sex subgroups (Pheterogeneity = 0.024); however, this effect was not replicated in the interaction analysis (Pinteraction = 0.127). In addition, an interaction between RCII and BMI stratification was detected in the cohort study (Pheterogeneity = 0.039; Pinteraction = 0.018). No other subgroups showed significant effect modification (all Pheterogeneity > 0.05; all Pinteraction > 0.05). We conducted several sensitivity analyses to assess the robustness of the associations. In the case-control study, the associations remained significant after excluding participants with missing covariates (Supplementary Table 4) or controls with 1.0 ≤ CIMT <1.5 mm (Supplementary Table 5). Similarly, in the cohort study, consistent results were observed after excluding individuals with missing covariates (Supplementary Table 6) or incident CAP cases that occurred within the first year of follow-up (Supplementary Table 7). Time-dependent ROC analysis showed that adding CTI and RCII to a standard model significantly improved predictive accuracy for CAP risk, with an area under the curve (AUC) of 0.681 for the CTI model, 0.704 for the RCII model and 0.702 for the CTI+RCII model (Fig. 4A). Compared with the CTI model, the RCII model showed a statistically significant improvement in both NRI (NRI = 0.119, 95% CI: 0.045–0.191; p = 0.001) and IDI (IDI = 0.006, 95% CI: 0.002–0.011; p = 0.006) (Fig. 4B). Adding RCII to the CTI model significantly improved risk prediction, with an NRI of 0.111 (95% CI: 0.035–0.187) and an IDI of 0.005 (95% CI: 0.001–0.012) (Fig. 4C). These results support that RCII may be a superior predictor of CAP. Comparison of predictive performance for CAP. CAP: carotid atherosclerotic plaque; ROC: receiver operating characteristic; CTI: C-reactive protein-triglyceride glucose index; RCII: natural log-transformed remnant cholesterol inflammation index; AUC: area under the curve; NRI: net reclassification index; IDI: integrated discrimination improvement. A: Time-dependent ROC curves for standard, CTI, RCII, and CTI+RCII models; B: NRI and IDI comparing RCII vs CTI; C: NRI and IDI for adding RCII to the CTI model. RCII was natural log-transformed for all analyses Discussion In the present study, we found that CTI and RCII were significantly associated with increased risk of CAP among the general population. In addition, both CTI and RCII showed improvements in the prediction of CAP risk beyond traditional multi-factorial models, with RCII demonstrating a more pronounced effect. Furthermore, we observed a nonlinear dose-response relationship between RCII and CAP, indicating that higher RCII was associated with a steeper increase in risk of CAP. Numerous studies have consistently shown that CTI was significantly associated with incident stroke in people with different diseases and metabolic statuses [20, 35, 36]. Other studies also showed linear associations of CTI with all-cause and CVD mortality in patients with cardiovascular-kidney-metabolic syndrome [21], as well as in those with CHD and diabetes [37]. A cohort study showed that the integration of the CTI and C-reactive protein-albumin-lymphocyte index could further enhance the discriminatory ability for incident CHD among patients with metabolically heterogeneous obesity [38]. In addition, CTI could significantly improve CHD identification when introducing CTI to the Framingham risk score among American general population [24]. In our study, we found a linear dose-response association between CTI and CAP, indicating that CTI was an indicator for increased risk of CAP. These findings are consistent with previous studies, suggesting the sustained role of CTI in the pathogenesis of subclinical atherosclerosis and cardiovascular events. Besides, several studies showed that RCII was associated with stroke incidence and prognosis [29, 30]. Meanwhile, RCII was shown to be associated with all-cause mortality in US and China, as well as increased risks of CVD and cancer mortality in US population [33]. To date, limited evidence from population-based cohort studies is available on RCII. Our findings expand the current knowledge on the positive association between RCII and CAP risk. Furthermore, we compared the predictive value between RCII and CTI for CAP risk, demonstrating that RCII had superior ability to identify high-risk individuals with CAP compared with CTI. We hypothesize that RC may play a more crucial role in the pathological mechanism of CAP formation than insulin resistance, and a comprehensive assessment of RC and hsCRP may better reflect the atherogenic process, but the specific mechanisms have not been elucidated. Further studies are needed to test this hypothesis. Atherosclerosis is a chronic, inflammatory process initiated by endothelial dysfunction, and followed by lipid accumulation, activation of inflammatory pathways and plaque formation [39]. Insulin resistance status accelerates atherosclerosis through the induction of endothelial damage and cellular dysfunction, which is the initial step of atherogenic process [40]. Recent observational study showed that RC could contribute to CVD mortality through two pathways: an indirect effect mediated by insulin resistance, and a direct atherogenic effect by stimulating diverse inflammatory molecules and cells [41, 42]. A large prospective study showed that hsCRP contributed minorly to the association between RC and major adverse cardiovascular events, suggesting the importance of addressing both RC and low-grade inflammation in the clinical management of CVD [43]. More experimental studies are needed to clarify the mechanisms underlying the complex correlation among the RC, insulin resistance, inflammation and atherosclerotic damage. Notably, we observed significant interactions between age and CTI, as well as RCII in the subgroup analysis of both study stages. The risk of CAP increased with higher levels of CTI and RCII to a greater extent in younger individuals (below median age) than in the older population. Meanwhile, the significant association of RCII with CAP remained in the older population. These results were consistent with previous findings, such as those from the NHANES in US, showing stronger associations of CTI and RCII with all-cause mortality in younger individuals [21, 33]. An 18-year cohort study revealed that the presence of CAP in adolescents and young adults was strongly associated with subsequent CVD and all-cause mortality, suggesting that atherosclerosis begins early in life, and that interventions targeting adolescents and young adults to slow the progression of CAP could help prevent or delay cardiovascular events [44]. Taken together, our findings raise the possibility that CTI and RCII might be more suitable for indicating high risk of CAP at an early stage in younger populations. However, the interpretation remains speculative, and the reasons behind the observed stronger associations in younger individuals require further investigation. To our knowledge, this is the first study to report associations of CTI and RCII with CAP in a large-scale population, and we provided new insights that RCII might be a superior biomarker to CTI in predicting CAP risk. In addition, we adopted a two-stage design with both case-control study and prospective cohort study, which strengthens causal inference and the reliability of our conclusions. From a clinical implementation perspective, although our study does not yet provide definitive thresholds for RCII, the observed dose-response relationship suggests its potential utility in CAP risk stratification. Importantly, calculated RCII could be readily obtained from standard lipid profiles and inflammatory assessments without additional cost, making it suitable for routine medical examinations. Our findings also raise the hypothesis that individuals with elevated RCII might benefit from strategies targeting both remnant cholesterol and inflammation. Clinical studies have demonstrated that anti-inflammatory therapy targeting inflammatory pathways such as IL-1β, IL-18 and IL-6 can complement traditional lipid-lowering treatment and delay the clinical complications of atherosclerosis [45]. Therefore, RCII may be a useful biomarker for risk management of CAP. Future studies should establish validated cut-off values and evaluate whether adding RCII to existing risk scores improves clinical decision-making. Some limitations deserve to mention in our study. First, this is a single-center study, thus this limits the generalizability of the findings to broader populations. Second, we did not collect the information on lipid-lowering therapy, which constitutes an important unmeasured confounder for lipid-derived indices such as RCII and CTI. This is particularly relevant for RCII because remnant cholesterol is more sensitive to statins and fibrates. Even though the overall treatment rate for lipid-lowering drugs is low in China (only 3.6% of adults and 5.5% of those at high cardiovascular risk) [46], treated individuals likely have a different risk profile that may bias our estimates, especially the comparison between RCII and CTI. Therefore, future independent validation in cohorts with medication records is needed. Third, the loss of follow-up was substantial, which may raise concerns about selection bias. Our cohort consisted of individuals who voluntarily participated in annual health check-ups and follow-up attendance depended on participants’ own willingness, thus the loss to follow-up was unlikely to be strongly associated with CAP status or CTI/RCII levels. Fourth, a median follow-up of 767 days is relatively short for assessing atherosclerosis progression. Nevertheless, we still observed 857 CAP events during follow-up, representing an incidence of 19.4% (857/4418). Fifth, our study only focused on the presence of CAP, more detailed characteristics such as intra-plaque hemorrhage and lipid-rich core were not assessed; therefore, we could not explore the association of CTI and RCII with these specific plaque features in greater depth. In conclusion, CTI and RCII were significantly associated with increased risk of CAP, and RCII may be a superior biomarker to CTI in predicting risk of CAP among the general population. Future cohort studies in diverse populations are needed to validate the predictive value of RCII on subclinical atherosclerosis. Electronic supplementary material Below is the link to the electronic supplementary material. Supplementary Material 1 Acknowledgements The authors would like to thank all the participants in the Health Examination project, as well as all the research staff and students who participated in this study. Authors’ contributions Nimei Zeng drafted the manuscript, and contributed to analysis and interpretation of the data; Ting Tian optimized the graph visualization; Quan Zhou and Yuanli Li investigated and supervised the original data; Ruixin Zhou verified all the data; Fangfei Xie, Renfang Han and Yi Wang critically revised the manuscript; Yun Wang and Jingyi Fan contributed to funding acquisition; Wei Ni contributed to project management; Jingyi Fan contributed to the conception and design of the work. Funding This work was funded by Suzhou Gusu Medical Youth Talent (GSWS2023109), Suzhou Applied Basic Research (Medical and Health) Technology Innovation Project (SYWD2024144) and Suzhou Municipal Health Commission Cadre Healthcare Program (SZBJ202504). Data availability All data generated during this study are included in the manuscript and supporting files. Data will be made available on request. Declarations Ethical approval and consent to participate All research was conducted in accordance with Declaration of Helsinki. This study was approved by the Ethics Committee of the Affiliated Suzhou Hospital of Nanjing Medical University (Ethical Approval Number: K-2026-161-K01). All participants received and signed written informed consent. Consent for publication Not applicable. 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. Nimei Zeng and Ting Tian contributed equally to this work. Contributor Information Wei Ni, Email: 1010702982@qq.com. Jingyi Fan, Email: fanjingyi@njmu.edu.cn. References Associated Data Supplementary Materials Supplementary Material 1 Data Availability Statement All data generated during this study are included in the manuscript and supporting files. Data will be made available on request.