C-reactive protein-triglyceride glucose index and risk of cognitive decline in Chinese middle-aged and older adults: A 9-year national cohort study from CHARLS

Abstract

Background

Cognitive decline is a pressing global health issue in aging populations. The novel composite biomarker, C-reactive protein-triglyceride glucose index (CTI), may provide a superior tool for early risk stratification.

Objective

This study examined CTI's capacity to independently predict cognitive decline.

Methods

This retrospective cohort analysis utilized data from 5464 middle-aged and older adults without baseline cognitive impairment from the China Health and Retirement Longitudinal Study (CHARLS). The incidence of cognitive decline served as the primary outcome. CTI was calculated as: CTI = 0.412× Ln (CRP [mg/L]) + Ln (TG [mg/dl] ×FPG [mg/dl])/2. Cognitive performance was assessed in 2020 using the Chinese Mini-Mental State Examination (MMSE). Logistic regression and complementary approaches were used to analyze associations.

Results

Over the 9-year follow-up, a significant positive association between CTI and cognitive decline risk was observed. After multifactorial adjustment, individuals in the highest CTI quartile (Q4) demonstrated a 34% elevated risk relative to the lowest quartile (Q1) (HR = 1.34, 95%CI: 1.10–1.63; p = 0.004). A monotonic dose-response gradient was evident (p = 0.005), with restricted cubic splines confirming linearity (p = 0.025). Sensitivity analyses substantiated robustness.

Conclusions

Elevated CTI is independently associated with increased risk of cognitive decline, suggesting its potential as a dual-pathway biomarker for early screening.

Keywords

Alzheimer's disease C-reactive protein-triglyceride glucose index CHARLS cognitive decline inflammation insulin resistance

Introduction

Cognitive impairment represents an urgent global health challenge, profoundly compromising functional independence and well-being while driving escalating healthcare expenditures.^1–3 Its prevalence rises disproportionately with population aging, particularly affecting middle-aged and older adults, with projections indicating sustained growth in future decades. The multifactorial etiology involves dynamic interactions,^4–6 between genetic susceptibility,⁷ environmental exposures,⁸ and modifiable lifestyle components.⁹ Targeted management of adjustable risk factors is therefore essential to mitigate progression trajectories and optimize quality of life in aging populations.

Insulin resistance (IR) constitutes a pathogenic cornerstone of type 2 diabetes and metabolic syndrome, with contemporary research elucidating intricate molecular pathways and emergent therapeutic targets.^10–14 IR mechanistically underpins diverse pathophysiological manifestations, including mitochondrial bioenergetic failure, chronic low-grade inflammation, and gut microbiota dysbiosis.^15–17 The triglyceride-glucose (TyG) index,¹⁸ calculated from fasting triglycerides and glucose concentrations, provides a validated surrogate measure for IR. Accumulating evidence demonstrates TyG's superior predictive capacity over conventional markers for cardiovascular disease (CVD),^19–22 type 2 diabetes mellitus (T2DM), and nonalcoholic fatty liver disease (NAFLD),²³ while its elevation independently forecasts chronic kidney disease (CKD) progression,^24–26 supporting its utility as a pan-systemic metabolic risk indicator. Concurrently, C-reactive protein (CRP) serves as a cardinal systemic inflammation biomarker, implicated in atherosclerosis progression, diabetes pathogenesis, and autoimmune dysregulation.²⁷ Emerging data further link CRP to neuroinflammatory cascades and oncological outcomes.

The C-reactive protein-triglyceride-glucose index (CTI) represents an integrated biomarker synthesizing inflammatory (CRP) and metabolic dysregulation (TyG index) components. Contemporary evidence establishes CTI's superiority over conventional biomarkers in forecasting cardiovascular outcomes, especially within diabetic and obese subgroups. Significant correlations exist between CTI elevation and atherosclerosis severity, NAFLD progression,²⁸ and microvascular compromise development. This metric enables dual-pathway risk stratification through concurrent quantification of inflammatory and metabolic derangements. Notwithstanding CTI's validated cardiometabolic predictive capacity, its relationship with cognitive deterioration, particularly in Asian cohorts, remains uncharted territory. We thus hypothesize that CTI independently predicts cognitive decline by capturing convergent metabolic-inflammatory dysregulation.

To bridge this critical knowledge gap, we leveraged the nationally representative China Health and Retirement Longitudinal Study (CHARLS) cohort to elucidate the complex association between CTI and cognitive deterioration risk.

Methods

Study design

This investigation utilized data from the CHARLS—a prospective national cohort implementing multistage probability proportional-to-size sampling. This study is a retrospective cohort analysis utilizing existing waves of CHARLS data collected from 2011 to 2020. Although CHARLS was designed as a prospective cohort study, the present research employed a retrospective analytical approach, utilizing existing waves of data collected from 2011 to 2020. Wave 1 baseline assessment (2011) enrolled 17,705 participants aged ≥ 45 years across 450 villages within 150 county-level divisions spanning 28 Chinese provinces. Standardized questionnaires administered at enrollment captured core demographic and health parameters. Biennial to triennial follow-ups tracked health outcomes, with five completed survey waves (2011, 2013, 2015, 2018, 2020) comprising the analytical dataset. Longitudinal assessments were conducted at five time points, starting with a baseline in 2011, followed by four follow-up assessments. CTI was calculated using baseline (2011) biomarkers, and cognitive outcomes were evaluated during the 2020 follow-up (9 years later).

Study population

Five waves of data collection were conducted in 2011, 2013, 2015, 2018, and 2020, with the baseline assessments serving as the index time point. The 2020 wave provided the primary cognitive outcome. Figure 1 delineates participant selection. The baseline cohort comprised 17,705 Wave 1 participants. We applied sequential exclusions: 1) Participants with missing information on fasting blood glucose, CRP, or triglycerides (TG) at baseline (N = 8421); 2) 205 aged < 45 years; 3) 138 with baseline cognitive disorders; 4) 1436 lacking cognitive assessment data; 5) 886 missing CTI measurements; 6) 1155 with incomplete covariates. The final analytical cohort included 5464 eligible subjects. The age threshold of 45 years was selected in alignment with the original CHARLS sampling framework, which specifically targets middle-aged and older adults. This threshold corresponds to the typical onset age for investigating age-related cognitive decline in epidemiological studies. Baseline cognitive disorders were defined as: 1) Self-reported physician diagnosis of dementia or mild cognitive impairment (MCI); 2) Medical record verification where available. MCI was included.

Figure 1.

Flow chart of the study population. CTI: C-reactive protein-triglyceride glucose index; FBG: fasting blood glucose; CRP: C-reactive protein; TG, triglycerides.

Calculation of CTI

The CTI index was obtained by using the following formula²⁹: CTI = 0.412×Ln (CRP [mg/L])+Ln (TG [mg/dl]×FPG [mg/dl])/2.

The CTI values were categorized into quartiles (Q1-Q4) based on their distribution within the analytical sample. The quartile cut points were defined as follows: CTI Q1: [3.076, 4.305], CTI Q2: (4.305, 4.665], CTI Q3: (4.665, 5.076], and CTI Q4: (5.076, 7.529]. This stratification facilitated the examination of dose-response relationships while ensuring adequate statistical power within each group.

Assessment of cognitive decline

Cognitive function assessed in 2020 constituted the primary endpoint. Utilizing the validated Chinese Mini-Mental State Examination (MMSE), CHARLS quantified cognition through two domains: 1) Episodic memory: Participants recalled ten Chinese words immediately and after a 5-min delay, scoring one point per correct recall (maximum 20 = immediate + delayed scores). 2) Mental intactness: a) Test of Information and Cognitive Status (TICS) component: Temporal orientation (date/season/weekday) and serial-7 subtraction from 100 (5 trials); b) Visuospatial assessment: Figure replication task (1 point per accurate copy). Mental intactness scores (range 0–11) summed TICS and visuospatial performance. Global cognition (range 0–31) combined both domain scores, with higher values indicating superior function. Cognitive decline was defined as: a) Global score <11; b) Mental intactness score ≤ 2.75 (lowest quartile); c) Episodic memory score ≤5 (lowest quartile).

Assessments of covariates

Trained interviewers systematically captured baseline data through structured instruments, encompassing four domains: 1) Sociodemographic and behavioral characteristics: Sex, age, residential location, educational attainment, marital status, tobacco use, and alcohol consumption patterns. 2) Anthropometric parameters: Height, weight, body mass index (BMI), systolic/diastolic blood pressure (SBP/DBP). 3) Clinical history: Physician-diagnosed CVD, hypertension, diabetes mellitus, dyslipidemia, and corresponding pharmacotherapies. 4) Biochemical assays: Fasting plasma glucose (FPG), total cholesterol (TC), TG, high-/low-density lipoprotein cholesterol (HDL-C/LDL-C), serum creatinine, blood urea nitrogen, CRP, and glycated hemoglobin (HbA1c). All covariates listed above were assessed at baseline (2011) and were treated as time-invariant exposures in the statistical models. Although factors such as lifestyle behaviors are theoretically time-varying, the present analysis utilized their baseline measurements to reflect initial status and ensure consistency with the baseline assessment of the primary exposure (CTI).

Statistical analysis

Missing data were addressed using a complete case analysis approach. Participants with missing data on any variable required for the analysis (including exposure, outcome, or covariates) were excluded from the final analytical cohort, as detailed in Figure 1. This approach was selected over multiple imputation due to the extensive nature of the missing data and concerns that the data may not be missing at random, which could introduce bias if imputed. While multiple imputation was considered in initial analyses, the final models are based on complete cases to ensure transparency and avoid potentially unreliable imputations.

Baseline characteristics were summarized as mean ± standard deviation (SD) or median with interquartile range (IQR) for continuous measures, and frequency (percentage) for categorical measures. Group comparisons (cognitive decline versus normal cognition) employed Student's t-tests, Wilcoxon rank-sum tests, or χ² tests as appropriate. Primary analyses utilized Cox proportional hazards models to evaluate the associations between baseline CTI quartiles and the incidence of cognitive decline. The time-to-event was defined from the baseline measurement to the assessment conducted in 2020. Time-to-event analyses employed Cox proportional hazards models, with a follow-up duration extending from baseline in 2011 until the identification of cognitive decline in 2020. The model incorporated CTI quartiles as time-invariant exposures and adjusted for relevant covariates. Three progressively adjusted multivariate models examined the associations between CTI and cognitive decline: 1) Model 1: Unadjusted (CTI only); 2) Model 2: Model 1 + age and sex; 3) Model 3: Model 2 + HbA1c, creatinine, alcohol/tobacco use, residence, education, and marital status. We used a generalized linear model with binomial family and logit link to estimate the odds of outcome across CTI quartiles, testing for linear trend. All analyses used R 4.4.2. Two-tailed tests established statistical significance at p = 0.05. The key assumption of the Cox proportional hazards model, the proportionality of hazards, was tested for all variables using Schoenfeld residuals. The global test and tests for individual variables indicated no significant deviations from the proportionality assumption. No influential outliers were identified that substantially altered the model estimates.

Results

Population characteristics

The study comprised 5464 participants from the CHARLS cohort. The overall mean age was 57.06 ± 7.73 years, with 2452 (44.9%) being male. Participants were stratified into two groups: 4098 (75.0%) with normal cognition and 1375 (25.2%) with cognitive decline. Individuals with cognitive decline were more likely to be older, female, rural residents, less educated, and unmarried compared to those with normal cognition. Regarding cardiometabolic profiles, the cognitive decline group exhibited lower levels of CTI, creatinine, urea, and TG, but higher levels of HDL-C, LDL-C, and TC. Notably, systolic blood pressure was elevated in the cognitive decline group, while diastolic pressure showed no significant difference. Lifestyle factors revealed lower rates of current smoking and alcohol consumption in the cognitive decline group. Chronic conditions such as hypertension and diabetes did not differ significantly between groups, whereas dyslipidemia was less prevalent in the cognitive decline group. Body measurements indicated that the cognitive decline group had significantly lower height, weight, and BMI values. Table 1 presents the baseline characteristics of the total analytical cohort (N = 5464) from the 2011 wave, stratified by cognitive status in 2020. All variables were measured at study entry, except for the cognitive outcomes assessed in 2020. Comprehensive demographic and clinical profiles of the study cohort are summarized in Table 1.

Table 1.

Baseline (2011) Characteristics of study participants stratified by 2020 cognitive status.

Characteristic	Overall (N = 5464)	Cognitive normal (N = 4089)	Cognitive decline (N = 1375)	p
Age, y	57.06 ± 7.73	56.21 ± 7.44	59.56 ± 8.04	<0.0001
CRP, mg/dl	2.41 ± 6.72	2.36 ± 6.26	2.55 ± 7.93	0.44
Glucose, mg/dl	108.40 ± 31.19	108.62 ± 31.79	107.74 ± 29.31	0.35
HBA1c, %	5.26 ± 0.76	5.24 ± 0.73	5.31 ± 0.83	0.01
Creatinine, mg/dl	0.77 ± 0.18	0.78 ± 0.19	0.74 ± 0.18	<0.0001
Urea, mg/dl	4.37 ± 1.20	4.42 ± 1.22	4.21 ± 1.13	<0.0001
TC, mg/dl	195.12 ± 38.42	194.11 ± 37.78	198.11 ± 40.13	<0.01
HDL, mg/dl	51.16 ± 14.92	50.51 ± 15.01	53.09 ± 14.49	<0.0001
LDL, mg/dl	118.24 ± 34.72	117.44 ± 34.07	120.62 ± 36.49	<0.01
TG, mg/dl	131.57 ± 108.93	133.83 ± 114.34	124.87 ± 90.69	<0.01
Height, m	1.58 ± 0.08	1.60 ± 0.08	1.55 ± 0.08	<0.0001
Weight, kg	59.76 ± 11.29	61.03 ± 11.12	55.97 ± 10.93	<0.0001
BMI, kg/m²	23.77 ± 3.82	23.94 ± 3.75	23.26 ± 3.98	<0.0001
SBP, mmHg	128.28 ± 20.18	127.45 ± 19.41	130.76 ± 22.12	<0.0001
DBP, mmHg	75.55 ± 11.87	75.69 ± 11.74	75.15 ± 12.22	0.15
Sex				<0.0001
Female	3012 (55.12%)	2071 (50.65%)	941 (68.44%)
Male	2452 (44.88%)	2018 (49.35%)	434 (31.56%)
Marital status				<0.0001
Married	4977 (91.09%)	3800 (92.93%)	1177 (85.60%)
Non-Married	487 (8.91%)	289 (7.07%)	198 (14.40%)
Education level				<0.0001
College and higher	83 (1.52%)	83 (2.03%)
Elementary school and below	3600 (65.89%)	2325 (56.86%)	1275 (92.73%)
High school	1781 (32.60%)	1681 (41.11%)	100 (7.27%)
Residence				<0.0001
rural	3535 (64.70%)	2475 (60.53%)	1060 (77.09%)
urban	1929 (35.30%)	1614 (39.47%)	315 (22.91%)
Smoke				<0.0001
Current smoker	1601 (29.30%)	1274 (31.16%)	327 (23.78%)
Former smoker	419 (7.67%)	350 (8.56%)	69 (5.02%)
Never	3444 (63.03%)	2465 (60.28%)	979 (71.20%)
Drink				<0.0001
Current	2131 (39.00%)	1661 (40.62%)	470 (34.18%)
Never	3333 (61.00%)	2428 (59.38%)	905 (65.82%)
Hypertension				0.23
No	4199 (76.85%)	3159 (77.26%)	1040 (75.64%)
Yes	1265 (23.15%)	930 (22.74%)	335 (24.36%)
Dyslipidemia				<0.01
No	4938 (90.37%)	3668 (89.70%)	1270 (92.36%)
Yes	526 (9.63%)	421 (10.30%)	105 (7.64%)
Diabetes				0.63
No	5178 (94.77%)	3871 (94.67%)	1307 (95.05%)
Yes	286 (5.23%)	218 (5.33%)	68 (4.95%)

All variables were measured at baseline (2011), except cognitive outcomes assessed in 2020.

CRP: C-reactive protein; HbA1c: glycated hemoglobin; TC: total cholesterol; HDL-C: high-density lipoprotein cholesterol; LDL-C: low-density lipoprotein cholesterol; TG: triglycerides; BMI: body mass index; SBP: systolic blood pressure; DBP: diastolic blood pressure.

Associations between CTI and cognitive decline

Cox proportional hazards models with 9-year follow-up demonstrated: Table 2 delineates CTI quartile associations with cognitive decline risk. Progressively higher CTI quartiles conferred incrementally elevated risk. In the unadjusted model (Model 1), over a 9-year follow-up period, Q4 exhibited 21% greater risk versus Q1 (HR = 1.21, 95%CI 1.02–1.44, P = 0.03), demonstrating significant exposure-response gradation (p = 0.02). Age-sex adjustment (Model 2) amplified this association: Q4 risk increased to 43% (HR = 1.43, 95%CI 1.19–1.72, p < 0.001) with persistent trend significance (p = 0.02). Further adjustment for metabolic markers (HbA1c, creatinine), behavioral factors (tobacco/alcohol use), and sociodemographics (residence/education/marital status) in Model 3 attenuated effects modestly, yet Q4 maintained 34% excess risk (HR = 1.34, 95%CI 1.10–1.63, p = 0.004) with robust trend persistence (p = 0.005). Continuous CTI elevation per unit increment showed progressive risk escalation across quartiles (Figure 2). The 34% residual risk in fully adjusted models underscores CTI's independent prognostic value. Significant exposure-response gradients persisted throughout. These findings demonstrate a robust positive association between CTI elevation and cognitive deterioration risk that persisted despite comprehensive covariate adjustment.

Figure 2.

Multi-model analysis of cognitive decline and CTI quartiles. CTI: C-reactive protein-triglyceride glucose index.

Table 2.

Hazard ratios and 95% confidence intervals for cognitive decline risk across CTI quartiles.

Characteristic	Case	Model 1			Model 2			Model 3
Characteristic	Case	HR	95%CI	p	HR	95%CI	p	HR	95%CI	p
CTI (per 1 unit)	5464
CTI quartile
Q1	1365	Ref			Ref			Ref
Q2	1363	0.98	0.82,1.16	0.78	1.09	0.92,1.30	0.32	1.12	0.93,1.35	0.25
Q3	1370	1.05	0.88,1,24	0.59	1.21	1.01,1.45	0.03	1.16	0.96,1.40	0.13
Q4	1366	1.21	1.02,1.44	0.03	1.43	1.19,1.72	<0.001	1.34	1.10,1.63	0.004
P for trend	0.02			0.02			<0.001			0.005

Model 1: unadjusted for CTI.

Model 2: Model 1 + sex, age.

Model 3: Model 2 + HBA1c, Creatinine, drink, smoke, residence, education level, marital status.

CTI: C-reactive protein-triglyceride glucose index; HR: hazard ratio; CI: confidence interval; HbA1c: glycated hemoglobin.

We present the results of the restricted cubic spline analysis, which examines the linear association between the CTI and the risk of cognitive decline (Figure 3). The analysis includes three models with varying levels of adjustment. In the unadjusted model (Model 1), the overall association was statistically weak (p = 0.0622), and the non-linear association was also not significant (p = 0.7488), indicating a slight but non-significant increase in the risk of cognitive decline with increasing CTI. In the model adjusted for sex and age (Model 2), the overall association became statistically significant (p = 0.0004), while the non-linear association remained non-significant (p = 0.2813), suggesting a more pronounced linear increase in the risk of cognitive decline with higher CTI levels. In the fully adjusted model (Model 3), which included additional covariates such as HbA1c, creatinine, drinking, smoking, residence, education level, and marital status, the overall association remained statistically significant (p = 0.0246), but the non-linear association was still not significant (p = 0.3562), indicating a modest increase in the risk of cognitive decline with CTI. These findings suggest that there is a linear relationship between CTI and the risk of cognitive decline, with the association being most robust in the model adjusted for sex and age. Collectively, the RCS analysis did not detect a significant non-linear relationship, supporting the use of a linear model to describe the association between CTI and cognitive decline risk.

Figure 3.

Restricted Cubic Spline Analysis of CTI's Dose-Response Relationship with Cognitive Decline Risk. CTI, C-reactive protein-triglyceride glucose index.

Sensitivity analysis

Sensitivity analyses consistently corroborated primary findings, affirming result robustness: In Model 1 (unadjusted), Q4 conferred 21% elevated cognitive decline risk versus Q1 (HR = 1.21, 95%CI 1.02–1.44, p = 0.03) with significant exposure-response gradient (p = 0.02); Model 2 (age-sex adjusted) showed Q4 risk escalation to 43% (HR = 1.43, 1.19–1.72, p < 0.001) while maintaining significant trend (p = 0.02); despite attenuation from metabolic (HbA1c, creatinine), behavioral (smoking, alcohol), and sociodemographic (residence/education/marital status) adjustments in Model 3 (fully adjusted), Q4 retained 34% excess risk (HR = 1.34, 1.10–1.63, p = 0.004) with persistent graded association (p = 0.005).

Discussion

Our findings should be interpreted within the methodological framework of this retrospective cohort analysis utilizing CHARLS data. Although the original CHARLS study was designed prospectively, the present investigation employed existing waves of data from 2011 to 2020 to examine associations between baseline CTI and incident cognitive decline.

This pioneering large-scale investigation establishes CTI as a quantifiable predictor of cognitive deterioration risk, revealing a robust monotonic dose-response relationship. Elevated CTI levels demonstrate significant associations with accelerated cognitive decline, supporting its utility as a promising novel biomarker. The clinical relevance of CTI for early risk stratification warrants substantive consideration.

CTI, developed by Ruan for cancer prognosis prediction,¹⁷ integrates the CRP and triglyceride glucose index (TyG index) as biomarkers of inflammation and IR, respectively. Accumulating evidence supports its clinical utility across diverse pathologies. Significant stroke risk associations are established in normoglycemic/prediabetic individuals and hypertensive populations,^29,30 alongside enhanced coronary heart disease detection.³¹ CTI demonstrates linear diabetes risk prediction with gender-specific manifestations: male testosterone decline/erectile dysfunction and female endometriosis risk.^32–35 Positive correlations with depressive symptom severity further confirm neuropsychiatric applications.³⁶ This expanding evidence base underscores CTI's versatility as a multi-system biomarker.

The CTI confers distinct advantages over isolated TyG index or CRP measurements by enabling integrated metabolic-inflammatory profiling. Unlike TyG's exclusive focus on IR risk or CRP's singular reflection of inflammation, CTI captures synergistic pathophysiological processes. Sole reliance on individual biomarkers provides unidimensional risk assessment, whereas CTI's dual-pathway integration yields multidimensional risk stratification. Crucially, CTI accounts for inter-individual biomarker variance by concurrently evaluating metabolic (TyG) and inflammatory (CRP) axes from complementary biological perspectives. This synthesis enhances predictive utility for high-risk subpopulation identification.

Analysis of 5464 CHARLS participants revealed a robust independent association between elevated CTI levels and cognitive deterioration risk, with restricted cubic spline (RCS) analysis confirming a significant positive linear exposure-response relationship. These findings position CTI reduction as a viable preventive strategy for cognitive preservation in aging populations. Notably, the lower smoking/alcohol consumption rates observed in the cognitive decline group suggest lifestyle modifications (particularly smoking cessation and moderated alcohol intake) may mediate neuroprotection. Furthermore, the higher cognitive decline prevalence among rural residents likely reflects tripartite disadvantages in healthcare accessibility, health education, and lifestyle patterns, collectively delineating actionable targets for public health interventions.

Systolic hypertension management emerges as a critical intervention focus, given significantly elevated SBP in cognitive decline groups versus cognitively normal counterparts (p < 0.001), contrasting with non-significant diastolic pressure differences. This implicates SBP as a pivotal modifiable risk factor, necessitating stringent systolic control in hypertensive populations for cognitive preservation. Paradoxically, the cognitive decline cohort exhibited elevated high-/low-density lipoprotein (HDL-C/LDL-C) and TC alongside reduced TG, revealing complex lipid-cognition interactions warranting mechanistic investigation. Anthropometrically, significantly lower height, weight, and BMI in cognitive decline groups suggest malnutrition may accelerate cognitive deterioration, positioning nutritional optimization as a potential prevention strategy.

The strong correlation between elevated CTI indices and cognitive decline may stem from the synergistic effects of chronic low-grade inflammation and metabolic dysregulation, which are hallmarks of the two major neurodegenerative pathologies. The following hypothesis elucidates this correlation more effectively: 1) Inflammatory pathways^6,27: As an acute-phase reactive protein, systemic inflammation reflected by CRP may act through the following pathways: a) Disrupting the integrity of the blood-brain barrier through the activation of matrix metalloproteinases. b) Promoting microglial activation and amyloid-β deposition. 2) Metabolic dysregulation^11,28: The TyG component of CTI captures IR, the effects of which include reduced cerebral glucose uptake, leading to insufficient neuronal energy supply, exacerbated tau protein hyperphosphorylation, and the development of Alzheimer-like pathological changes. Notably, we found that SBP was significantly elevated in the cognitive decline group, suggesting a possible vascular mechanism. Hypertension may amplify the effects of CTI by promoting cerebral small vessel disease,¹ a known causative agent of vascular cognitive decline.

While the TyG index and CRP alone predict cognitive risk, the strength of the CTI lies in its bi-directional capture of metabolic-inflammatory interactions. This includes direct metabolic parameters such as FPG and TG, which are critical for diabetes-related dementia subtypes.¹⁰ This dual assessment makes the CTI particularly suitable for primary care scenarios. A single CTI calculation, which requires only routine testing, can effectively identify high-risk patients in need of neuropsychological evaluation.

The negative association of TG with cognitive decline may appear paradoxical; however, it aligns with recent studies indicating that low TG levels in cognitively impaired individuals may reflect malnutrition, a known characteristic of advanced dementia.⁹ This indicates that the predictive utility of CTI may vary depending on the stage of the disease. Malnutrition may alter the pattern of dyslipidemia in the later stages of the disease. For instance, reduced dietary fat intake may lead to a decline in TG in patients with advanced cognitive impairment. This suggests that CTI may have greater predictive value in the pre-malnutritional stage. Elevated HDL-C may represent a compensatory anti-inflammatory response,²³ as HDL particles transport CRP-scavenging proteins. Notably, the NHANES cohort observed a similar lipid pattern when examining the association between CTI and depression, suggesting the presence of systemic metabolic-inflammatory dysregulation.³⁶

This study establishes CTI as a validated metric for predicting cognitive deterioration in community-dwelling populations, offering a clinically actionable tool for early risk stratification. Against the backdrop of escalating global cognitive decline burden, timely identification of high-risk individuals becomes imperative. Our findings advocate integrating CTI monitoring into routine practice, maintaining levels within optimal thresholds to mitigate cognitive impairment progression in a cost-effective preventive approach for aging societies.

This investigation offers three key methodological strengths: (1) Novelty as the first large-scale evidence establishing CTI-cognitive decline associations, yielding novel mechanistic and clinical implications; (2) Robust design via a prospective nationwide cohort with representative sampling (n = 5464), ensuring high validity; and (3) Clinical translatability wherein CTI's operational simplicity facilitates seamless integration into routine practice. Collectively, these advantages position CTI as a scalable biomarker for population-level cognitive risk stratification.

This study presents several limitations. First, while the MMSE is widely utilized in epidemiological research, its exclusive application without additional cognitive assessments may have constrained our capacity to identify specific domain impairments. Second, the potential for unmeasured confounding factors, such as APOE ε4 status and dietary patterns. Third, the observational design of the study limits the ability to draw causal inferences, as the observed associations may be influenced by residual confounding. Fourth, since the CHARLS participants were exclusively Chinese adults, the generalizability of the findings to other ethnic groups necessitates further investigation. Fifth, our use of a complete case analysis to handle missing data may have introduced selection bias and potentially limited the generalizability of our findings, if the excluded participants systematically differed from those included. Although multiple imputation was initially considered, the pattern and extent of missingness led us to favor a more conservative complete case approach. Nevertheless, these constraints do not compromise the core findings regarding CTI's prognostic utility for cognitive deterioration.

Conclusion

In summary, this study confirms a significant monotonic dose-response relationship between elevated CTI levels and cognitive deterioration risk. This association persisted after comprehensive adjustment for key confounders including demographic characteristics, lifestyle factors, and metabolic markers, solidifying CTI's robustness as an innovative composite biomarker. Our findings support CTI's clinical implementation for early detection of high-risk individuals, enabling targeted preventive strategies. Future research should validate these associations across multi-ethnic cohorts and develop public health interventions focused on CTI modulation.

Footnotes

Acknowledgements

This investigation leveraged data from the China Health and Retirement Longitudinal Study (CHARLS). The authors gratefully acknowledge the CHARLS research team and all study participants for their invaluable contributions.

ORCID iDs

Linyi Yang

Linlin Li

Yunwei Zhang

Ethical considerations

The CHARLS study protocol received ethical clearance from Peking University's Institutional Review Board under approval codes IRB00001052-11015 (household survey) and IRB00001052-11014 (biological specimens).

Consent to participate

Written informed consent was obtained from all participants prior to study enrollment.

Consent for publication

Informed consent for publication was obtained from all participants during the original CHARLS data collection process.

Author contribution(s)

Linyi Yang: Conceptualization; Data curation; Formal analysis; Methodology; Software; Supervision; Visualization; Writing – original draft; Writing – review & editing.

Jingyan Zeng: Conceptualization; Data curation; Formal analysis; Methodology; Software; Visualization; Writing – original draft; Writing – review & editing.

Linlin Li: Conceptualization; Data curation; Writing – original draft.

Yunwei Zhang: Conceptualization; Data curation; Writing – original draft.

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.

Declaration of conflicting interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Data availability statement

The datasets analyzed during this investigation are publicly accessible through the CHARLS repository: .

References

Mok

VCT

Cai

Markus

. Vascular cognitive impairment and dementia: mechanisms, treatment, and future directions. Int J Stroke 2024; 19: 838–856.

Ghondaghsaz

Khalaji

Mahalleh

, et al. Exploring the association between cognitive decline and triglyceride-glucose index: a systematic review and meta-analysis. Brain Behav 2024; 14: e70131.

Shi

Tian

Wei

, et al. Meditation for subjective cognitive decline, mild cognitive impairment and Alzheimer’s disease: a systematic review and meta-analysis of randomized controlled trials. Front Public Health 2025; 13: 1524898.

Parker

Davidson

Osmulski

, et al. Proteasome augmentation mitigates age-related cognitive decline in mice. Aging Cell 2025; 24: e14492.

Bowie

Low

Rubenstein

, et al. Neurovascular mechanisms of cognitive aging: sex-related differences in the average progression of arteriosclerosis, white matter atrophy, and cognitive decline. Neurobiol Dis 2024; 201: 106653.

Mekhora

Lamport

Spencer

. An overview of the relationship between inflammation and cognitive function in humans, molecular pathways and the impact of nutraceuticals. Neurochem Int 2024; 181: 105900.

Olejnik

Golenia

Małyszko

. The potential role of microbiota in age-related cognitive decline: a narrative review of the underlying molecular mechanisms. Int J Mol Sci 2025; 26: 1590.

Wang

, et al. The impact of multidomain interventions on cognitive and physical function in older adults with subjective cognitive decline: a meta-analysis and systematic review. PeerJ 2025; 13: e19588.

Chen

. Association between social engagement decline and cognitive function changes: mediating effect of depressive symptoms. Aging Clin Exp Res 2024; 37: 7.

10.

Elkanawati

Sumiwi

Levita

. Impact of lipids on insulin resistance: insights from human and animal studies. Drug Des Devel Ther 2024; 18: 3337–3360.

11.

Xie

Pan

, et al. Type 2 diabetes mellitus in adults: pathogenesis, prevention and therapy. Signal Transduct Target Ther 2024; 9: 262.

12.

Sun

Wang

, et al. Association of insulin resistance with cardiovascular disease and all-cause mortality in type 1 diabetes mellitus: systematic review and meta-analysis. Eur Heart J 2024; 45: ehae666.

13.

Sancar

Birkenfeld

. The role of adipose tissue dysfunction in hepatic insulin resistance and T2D. J Endocrinol 2024; 262: e240115.

14.

Jiang

W-Z

Fan

Z-L

M-L

, et al. Association between insulin resistance and multiple chronic diseases: a cross-sectional study from CHARLS. J Health Popul Nutr 2025; 44: 148.

15.

Wen

Huang

, et al. Long-term insulin resistance is associated with frailty, frailty progression, and cardiovascular disease. J Cachexia Sarcopenia Muscle 2024; 15: 1578–1586.

16.

Guo

, et al. Insulin resistance and sarcopenia: a prognostic longitudinal link to stroke risk in middle-aged and elderly Chinese population. BMC Public Health 2024; 24: 2757.

17.

Ruan

G-T

Xie

H-L

Zhang

H-Y

, et al. A novel inflammation and insulin resistance related indicator to predict the survival of patients with cancer. Front Endocrinol 2022; 13: 905266.

18.

Sun

, et al. Triglyceride glucose (TyG) index: a promising biomarker for diagnosis and treatment of different diseases. Eur J Intern Med 2025; 131: 3–14.

19.

Qiu

Aimaiti

Chen

, et al. Associations of TyG index with coronary heart disease risk and coronary artery sclerosis severity in OSA. Diabetol Metab Syndr 2024; 16: 301.

20.

Yang

Ren

, et al. The interaction between triglyceride-glucose index and visceral adiposity in cardiovascular disease risk: findings from a nationwide Chinese cohort. Cardiovasc Diabetol 2024; 23: 427.

21.

Zhang

Wang

, et al. Association of the triglyceride–glucose index with all-cause and cause-specific mortality: a population-based cohort study of 3.5 million adults in China. Lancet Reg Health West Pac 2024; 49: 101135.

22.

Tian

Chen

Xia

, et al. Pathways from insulin resistance to incident cardiovascular disease: a Bayesian network analysis. Cardiovasc Diabetol 2024; 23: 421.

23.

Zhou

Lin

Weng

, et al. Correlation between hs-CRP-triglyceride glucose index and NAFLD and liver fibrosis. BMC Gastroenterol 2025; 25: 252.

24.

Chen

Liu

, et al. Comparison of TyG and newly TyG related indicators for chronic kidney diseases estimation in a Chinese population. Diabetes Metab Syndr Obes 2024; 17: 3063–3075.

25.

Wang

Zhou

, et al. Association between triglyceride-glucose index and chronic kidney disease: results from NHANES 1999–2020. Int Urol Nephrol 2024; 56: 3605–3616.

26.

Chen

Yang

, et al. Triglyceride–glucose index and prognosis in individuals afflicted with heart failure and chronic kidney disease. ESC Heart Failure 2024; 11: 3120–3132.

27.

Roy

Zeller

Nero

, et al. C-reactive protein: the nexus between inflammation and protein misfolding diseases. Front Immunol 2025; 16: 1612703.

28.

Shi

Chen

Zhang

, et al. Multi-dimensional comparison of abdominal obesity indices and insulin resistance indicators for assessing NAFLD. BMC Public Health 2024; 24: 2161.

29.

Huo

Tang

Liu

, et al. Association between C-reactive protein-triglyceride glucose index and stroke risk in different glycemic status: insights from the China health and retirement longitudinal study (CHARLS). Cardiovasc Diabetol 2025; 24: 142.

30.

Tang

Wang

, et al. C-reactive protein-triglyceride glucose index predicts stroke incidence in a hypertensive population: a national cohort study. Diabetol Metab Syndr 2024; 16: 277.

31.

Zhang

, et al. Usefulness of C-reactive protein-triglyceride glucose index in detecting prevalent coronary heart disease: findings from the national health and nutrition examination survey 1999–2018. Front Cardiovasc Med 2024; 11: 1485538.

32.

Shan

Liu

Gao

. Application of the C-reactive protein-triglyceride glucose index in predicting the risk of new-onset diabetes in the general population aged 45 years and older: a national prospective cohort study. BMC Endocr Disord 2025; 25: 126.

33.

Mei

Zhang

, et al. Association between the C-reactive protein-triglyceride glucose index and erectile dysfunction in US males: results from NHANES 2001–2004. Int J Impot Res 2025; 37: 612–622.

34.

Zhang

Chen

, et al. Association between C-reactive protein-triglyceride glucose index and testosterone levels among adult men: analyses of NHANES 2015–2016 data. Sex Med 2025; 13: qfaf012.

35.

Ren

Zhang

, et al. Association between the C-reactive protein-triglyceride–glucose index and endometriosis: a cross-sectional study using data from the national health and nutrition examination survey, 1996–2006. BMC Women Health 2025; 25: 13.

36.

Huang

You

Zhang

, et al. Association between C-reactive protein-triglyceride glucose index and depressive symptoms in American adults: results from the NHANES 2005 to 2010. BMC Psychiatry 2024; 24: 890.