Long-term Survival in Ischemic Stroke Patients: A Comprehensive Analysis of the TyG-AIP-BMI Composite Index (TabCI) from MIMIC-IV ICU Data

doi:10.21203/rs.3.rs-4872713/v1

Download PDF

Article

Long-term Survival in Ischemic Stroke Patients: A Comprehensive Analysis of the TyG-AIP-BMI Composite Index (TabCI) from MIMIC-IV ICU Data

https://doi.org/10.21203/rs.3.rs-4872713/v1

This work is licensed under a CC BY 4.0 License

You are reading this latest preprint version

Background Insulin resistance-induced metabolic disorders play a crucial role in exacerbating ischemic stroke. This study aims to explore the association between TabCI and long-term mortality risk in severe ischemic stroke patients.

Methods Data from the Medical Information Mart for Intensive Care IV (MIMIC-IV 2.2) database were accessed to retrieve data of ischemic stroke patients. Patients were stratified into four groups based on TabCI quartiles. The study assessed the primary outcome of 180-day all-cause mortality and secondary outcomes including 90-day and 1-year ACM. Kaplan-Meier curves were used to compare outcomes across groups, and lasso regression analysis was employed to select covariates. Multivariable Cox proportional hazards regression models and restricted cubic splines (RCS) were used to explore the association between TabCI and these outcomes. Lastly, interaction and subgroup analyses were conducted to validate the stability of results.

Results A total of 1,141 severe ischemic stroke patients were included, with a mean age of 69 years (interquartile range [IQR]: 59-79), and 565 participants (49.5%) were male. Kaplan-Meier analysis indicated significantly lower long-term survival rates in patients in Q1 and Q3 compared to those in Q2 and Q4. Cox proportional hazards regression analysis adjusted for covariates showed a statistically significant increase in 180-day mortality risk in TabCI quartiles, with Q2 and Q4 groups also exhibiting increased risks at 90 days and 1 year. Additionally, RCS analysis revealed a gradual L-shaped correlation between TabCI and 90-day and 180-day all-cause mortality, with a smooth U-shaped trend observed for 1-year mortality, demonstrating significant non-linearity. Subgroup analysis further indicated an inverse correlation between TabCI and long-term mortality risk in non-Caucasian patients and those using aspirin, as well as negative correlations in TabCI among patients not receiving CRRT for 90-day and 180-day mortality.

Conclusion TabCI could serve as a marker for stratifying long-term risk among severe ischemic stroke patients, although its clinical predictive efficacy for long-term mortality in these patients is limited.

Health sciences/Neurology/Neurological disorders

Health sciences/Endocrinology

Health sciences/Neurology

Health sciences/Risk factors

ischemic stroke

TyG

triglyceride

glucose

MIMIC-IV

Stroke is a leading cause of death and disability worldwide. In China, the lifetime risk and disease burden of stroke are as high as 39.3%, ranking highest globally ^[1].Ischemic stroke, comprising 81.9% of all strokes, is the predominant type ^[2].Large hemisphere ischemic stroke leads to early neurological deficits, progressing to consciousness impairment and rapid onset of brain herniation, known as malignant middle cerebral artery infarction (MMI) ^[3].MMI has a mortality rate of 60.9–78%, often resulting in severe neurological disabilities even for survivors ^[3–5].Early prognosis of these patients aids clinicians in providing targeted treatments.

Insulin resistance, characterized by decreased insulin sensitivity, is a pivotal mechanism in disrupted glucose and lipid metabolism, crucial in the pathogenesis of type 2 diabetes and cardiovascular diseases ^[6].Additionally, insulin resistance is implicated in oxidative stress, inflammatory responses, and neuronal damage ^[7].These pathological changes are all related to the progression of ischemic stroke.TyG, a recently studied biomarker of insulin resistance, not only reflects the level of insulin resistance in patients but also predicts cardiovascular outcomes ^[8].Recent studies have combined TyG with BMI into an index aimed at enhancing the accuracy of predicting cardiovascular events.Recent research indicates that the AIP index, calculated based on the ratio of triglycerides to high-density lipoprotein, is an independent risk factor for cardiovascular events ^[9].However, studies to date have focused on individual or paired combinations of these indices, without confirmation of a combined index incorporating all three.This study aims to integrate these three indices into a formula and explore their correlation with long-term mortality in patients with severe ischemic stroke, aiming to provide a theoretical basis for reliable assessment tools in the long-term follow-up and stratified management of high-risk stroke patients.

2.1 Source of data

This retrospective study utilized data from the publicly accessible Medical Information Mart for Intensive Care IV database (MIMIC-IV, version 2.2).MIMIC-IV is an enhanced version of its predecessor, MIMIC-III, featuring updated data and modifications to some table structures.The dataset comprises clinical information from over 190,000 unique hospitalized patients at Beth Israel Deaconess Medical Center (BIDMC) in Boston, Massachusetts, spanning from 2008 to 2019.The database provides comprehensive details on patient demographics, vital signs, medications, laboratory tests, surgical procedures, diagnoses, treatment plans, and survival outcomes.Access to this data involved completion of the National Institutes of Health (NIH) Human Research Participant Protection training and passing the Collaborative Institutional Training Initiative (CITI) exam (ID: 13371165).Because the database does not contain any protected health information and patients remain anonymous, a waiver of informed consent was obtained.

2.2 Study design and population

This study focused on ischemic stroke patients admitted for the first time to the intensive care unit (ICU).We identified a cohort of 2,770 stroke patients using the search term "Cerebral infarction" in the International Classification of Diseases, 9th and 10th revisions (ICD-9 and ICD-10).Exclusion criteria were established to ensure data accuracy and relevance: (1) Lack of data on triglycerides (TG), fasting blood glucose (FBG), high-density lipoprotein cholesterol (HDL-c), body mass index (BMI), height, or weight on the first day of ICU admission;(2) Patients under 18 years of age;(3) ICU stay less than 24 hours;(4) Patients with multiple ICU admissions for stroke, considering data from the first admission only;(5) Lack of primary outcome indicators (survival status at 90 days, 180 days, and 1 year).Based on these criteria and quartiles of the composite index TabCI including TyG, AIP, and BMI, 1,141 patients were selected and categorized into four groups for further analysis (Fig. 1).

2.3 Data extraction

Data extraction was performed using Navicat Premium (version 16.1.15) with structured query language (SQL).Comprehensive data were collected upon each patient's admission, including demographic variables such as gender, age, height (m), weight (kg), BMI, marital status, and ethnicity; baseline vital signs including average blood pressure (mmHg), average heart rate (beats/min), systolic blood pressure (SBP, mmHg), respiratory rate (breaths/min), diastolic blood pressure (mmHg), and pulse oxygen saturation (SPO2, %); laboratory variables such as triglycerides (TG, mg/dL), fasting blood glucose (FBG, mg/dL), alanine aminotransferase (ALT), C-reactive protein, glycated hemoglobin HbA1c, homocysteine (Hcy), high-density lipoprotein cholesterol, low-density lipoprotein cholesterol, and uric acid; treatments including mechanical ventilation (MT), continuous renal replacement therapy (CRRT), thrombolysis; comorbidities such as hypertension, type 2 diabetes, coronary artery disease, atrial fibrillation, hyperlipidemia, stroke, myocardial infarction; medication usage including alteplase, aspirin, cilostazol, clopidogrel, and tirofiban; scoring systems including Acute Physiology Score III (APS III), Glasgow Coma Scale (GCS), Oxford acute severity of illness score (OASIS), Simplified Acute Physiology Score II (SAPS II), Systemic Inflammatory Response Syndrome (SIRS), Sequential Organ Failure Assessment (SOFA); outcomes such as 90-day, 180-day, and 1-year survival status.All blood indicators were recorded at the first measurement post-ICU admission. The dataset had no variables exceeding 20% missing values, and any missing data were imputed using multiple interpolation techniques.

2.4 Outcomes

This study focuses on assessing all-cause mortality outcomes among stroke patients in the MIMIC-IV database over varying time intervals.The primary endpoint was defined as long-term all-cause mortality at 180 days, with secondary endpoints at 90 days and 1 year.

2.5 TyG-AIP-BMI Composite Index calculation

TyG is calculated using the formula: Ln[FBG (mg/dL) × TG (mg/dL) / 2].AIP is calculated as: log(TG / HDL-c).BMI is calculated using the formula: weight (kg) / (height (m))^2 (kg/m^2).

The TyG-AIP-BMI Composite Index (TabCI) is a composite index based on TyG, AIP, and BMI, calculated as: log((TG * glucose * BMI) / (2*HDL-c)).

2.6 Statistical analysis

In the baseline table, Shapiro-Wilk test was used to assess continuous variables. Variables conforming to normal distribution are presented as mean ± SD, while those not conforming are presented as median [interquartile range (IQR)].Continuous variables were compared using Student's t-test or Mann-Whitney test, depending on their distribution.Categorical variables were presented as frequencies and percentages. Significant differences were assessed using Pearson's chi-square test or Fisher's exact test.

Kaplan-Meier (K-M) curves were used to assess time-to-event outcomes stratified by TabCI.Variables were selected using Least Absolute Shrinkage and Selection Operator (LASSO) with 10-fold cross-validation to identify predictors associated with 1-year, 90-day, and 180-day outcomes.Variables selected through LASSO analysis were included in Cox regression to examine the relationship between TabCI and 1-year, 90-day, and 180-day mortality rates.Final model variables were chosen considering event data availability. Model 1 was unadjusted, while Model 2 adjusted for age + DBP + hr + rr, and Model 3 further adjusted for age + DBP + hr + rr + atrial_fibrillation + coronary_heart_disease + diabetes + myocardial_infarction + alteplase + cilostazol + clopidogrel + tirofiban + mv + apsiii + sapsii. Additionally, TabCI was examined as a continuous variable using restricted cubic splines (RCS) to elucidate its relationship with the risk of outcome variables.In cases of nonlinear correlation, recursive algorithm identified the inflection point of TabCI with long-term all-cause mortality.Stratified analyses and interaction tests were performed based on age (≤ 65 years or > 65 years), gender, race/ethnicity, coronary artery disease, type 2 diabetes mellitus (2-DM), hypertension (HTN), myocardial infarction, aspirin, cilostazol, clopidogrel, tirofiban, CRRT, and MV.All statistical analyses were conducted using R software (version 4.3.1, R Foundation for Statistical Computing, Vienna, Austria). All tests were two-tailed, and p < 0.05 was considered statistically significant.

3.1 Baseline characteristics of study individuals

In this study, out of the 2,770 stroke patients listed in the MIMIC-IV database, 1,141 individuals met the inclusion criteria after screening and were subsequently analyzed (the screening process is depicted in Figure 1).

The mean age of included participants was 69 years (IQR: 59-79). There were 565 male patients, constituting 49.5% of the total, and 749 Caucasian patients, accounting for 65.6% of the total.Participants were categorized into four groups based on quartiles of TabCI at enrollment: Q1 with 286 participants, Q2 with 285 participants, Q3 with 285 participants, and Q4 with 285 participants.Baseline characteristics (Table 1) indicate varying TabCI levels across the four groups: Q1, 7.27 (IQR: 6.99-7.5); Q2, 8.06 (IQR: 7.89-8.23); Q3, 8.7 (IQR: 8.54-8.9); and Q4, 9.71 (IQR: 9.35-10.1).

Table 1. Baseline Characteristics and Outcomes by Quartiles of TabCI

	ALL	Q1	Q2	Q3	Q4	P value
	N=1141	N=286	N=285	N=285	N=285
Status_90d	0.24 (0.43)	0.29 (0.45)	0.22 (0.42)	0.27 (0.44)	0.20 (0.40)	0.047
Time_90d	90.0 [90.0;90.0]	90.0 [67.6;90.0]	90.0 [90.0;90.0]	90.0 [61.3;90.0]	90.0 [90.0;90.0]	0.051
Status_180d	0.30 (0.46)	0.35 (0.48)	0.26 (0.44)	0.34 (0.47)	0.26 (0.44)	0.015
Time_180d	180 [97.4;180]	180 [67.6;180]	180 [147;180]	180 [61.3;180]	180 [170;180]	0.015
Status_1year	0.37 (0.48)	0.40 (0.49)	0.29 (0.46)	0.41 (0.49)	0.37 (0.48)	0.013
Time_1year	365 [97.4;365]	365 [67.6;365]	365 [147;365]	365 [61.3;365]	365 [170;365]	0.015
Age	69.0 [59.0;79.0]	71.0 [58.0;80.0]	72.0 [61.0;80.0]	69.0 [62.0;81.0]	66.0 [56.0;73.0]	<0.001
Bmi	27.9 [24.4;32.6]	24.2 [20.7;27.7]	27.6 [24.8;30.8]	29.0 [25.5;33.4]	31.4 [27.3;35.7]	<0.001
Gender:						<0.001
Male	565 (49.5%)	106 (37.1%)	141 (49.5%)	141 (49.5%)	177 (62.1%)
Female	576 (50.5%)	180 (62.9%)	144 (50.5%)	144 (50.5%)	108 (37.9%)
Weight	74.0 [62.0;88.0]	62.0 [52.0;74.0]	75.0 [66.0;86.0]	75.0 [64.0;90.0]	84.0 [68.0;100]	<0.001
Marital_status:						<0.001
Single	320 (28.0%)	82 (28.7%)	58 (20.4%)	70 (24.6%)	110 (38.6%)
Divorced/Widowed	255 (22.3%)	77 (26.9%)	82 (28.8%)	59 (20.7%)	37 (13.0%)
Married	566 (49.6%)	127 (44.4%)	145 (50.9%)	156 (54.7%)	138 (48.4%)
Race:						<0.001
White	749 (65.6%)	163 (57.0%)	179 (62.8%)	207 (72.6%)	200 (70.2%)
No White	392 (34.4%)	123 (43.0%)	106 (37.2%)	78 (27.4%)	85 (29.8%)
DBP	130 [120;145]	130 [118;144]	132 [120;148]	132 [120;143]	132 [120;145]	0.39
SBP	76.0 [67.0;82.0]	75.0 [64.0;82.0]	78.0 [70.0;82.0]	75.0 [68.0;82.0]	76.0 [64.5;85.0]	0.456
HR	83.0 [73.0;97.0]	82.0 [67.0;96.0]	83.0 [73.0;97.0]	83.0 [73.0;94.0]	86.0 [75.0;101]	0.037
RR	19.0 [15.0;22.0]	18.0 [15.0;22.0]	18.0 [15.0;21.0]	19.0 [16.0;22.0]	20.0 [17.0;22.0]	0.002
ALT	20.0 [15.0;31.0]	18.0 [13.0;26.0]	20.0 [14.0;28.0]	20.0 [15.0;32.8]	25.0 [17.0;36.8]	<0.001
Glucose	113 [96.0;156]	99.0 [89.0;110]	104 [91.0;125]	119 [101;163]	177 [132;273]	<0.001
Hba1c	5.90 [5.40;6.80]	5.60 [5.23;6.00]	5.80 [5.40;6.20]	6.00 [5.50;6.90]	6.90 [6.00;8.80]	<0.001
HDL-c	48.0 [37.0;62.0]	65.0 [53.0;76.0]	52.0 [43.0;62.0]	43.0 [36.0;53.0]	33.0 [28.0;42.0]	<0.001
TG	116 [85.0;168]	72.0 [58.0;87.8]	107 [87.0;125]	137 [107;170]	198 [150;257]	<0.001
Atrial fibrillation	1141 (100%)	286 (100%)	285 (100%)	285 (100%)	285 (100%)	.
Coronary heart disease:						<0.001
No	584 (51.2%)	182 (63.6%)	155 (54.4%)	128 (44.9%)	119 (41.8%)
Yes	557 (48.8%)	104 (36.4%)	130 (45.6%)	157 (55.1%)	166 (58.2%)
Diabetes:						<0.001
No	569 (49.9%)	212 (74.1%)	177 (62.1%)	127 (44.6%)	53 (18.6%)
Yes	572 (50.1%)	74 (25.9%)	108 (37.9%)	158 (55.4%)	232 (81.4%)
Hyperlipidemia	1141 (100%)	286 (100%)	285 (100%)	285 (100%)	285 (100%)	.
Hypertension						0.032
No	323 (28.3%)	94 (32.9%)	63 (22.1%)	86 (30.2%)	80 (28.1%)
Yes	818 (71.7%)	192 (67.1%)	222 (77.9%)	199 (69.8%)	205 (71.9%)
Myocardial infarction						<0.001
No	765 (67.0%)	210 (73.4%)	206 (72.3%)	186 (65.3%)	163 (57.2%)
Yes	376 (33.0%)	76 (26.6%)	79 (27.7%)	99 (34.7%)	122 (42.8%)
Stroke	1141 (100%)	286 (100%)	285 (100%)	285 (100%)	285 (100%)	.
Alteplase						0.812
No	865 (75.8%)	217 (75.9%)	221 (77.5%)	216 (75.8%)	211 (74.0%)
Yes	276 (24.2%)	69 (24.1%)	64 (22.5%)	69 (24.2%)	74 (26.0%)
Aspirin						0.132
No	104 (9.11%)	34 (11.9%)	29 (10.2%)	21 (7.37%)	20 (7.02%)
Yes	1037 (90.9%)	252 (88.1%)	256 (89.8%)	264 (92.6%)	265 (93.0%)
Cilostazol						<0.001
No	1125 (98.6%)	286 (100%)	285 (100%)	283 (99.3%)	271 (95.1%)
Yes	16 (1.40%)	0 (0.00%)	0 (0.00%)	2 (0.70%)	14 (4.91%)
Clopidogrel						0.001
No	694 (60.8%)	186 (65.0%)	181 (63.5%)	183 (64.2%)	144 (50.5%)
Yes	447 (39.2%)	100 (35.0%)	104 (36.5%)	102 (35.8%)	141 (49.5%)
Tirofiban						0.011
No	1137 (99.6%)	286 (100%)	285 (100%)	285 (100%)	281 (98.6%)
Yes	4 (0.35%)	0 (0.00%)	0 (0.00%)	0 (0.00%)	4 (1.40%)
CRRT						0.047
No	1071 (93.9%)	267 (93.4%)	276 (96.8%)	268 (94.0%)	260 (91.2%)
Yes	70 (6.13%)	19 (6.64%)	9 (3.16%)	17 (5.96%)	25 (8.77%)
Intravenous thrombolysis	1138 (100%)	285 (100%)	285 (100%)	285 (100%)	283 (100%)	.
MV						0.001
No	589 (51.6%)	162 (56.6%)	162 (56.8%)	144 (50.5%)	121 (42.5%)
Yes	552 (48.4%)	124 (43.4%)	123 (43.2%)	141 (49.5%)	164 (57.5%)
APS III	38.0 [28.0;51.0]	38.0 [29.0;53.0]	36.0 [26.0;45.0]	38.0 [28.0;51.0]	41.0 [31.0;51.0]	0.015
GCS	15.0 [14.0;15.0]	15.0 [14.0;15.0]	15.0 [14.0;15.0]	15.0 [14.0;15.0]	15.0 [15.0;15.0]	0.008
OASIS	29.0 [24.0;36.0]	30.0 [26.0;37.0]	30.0 [24.0;35.0]	30.0 [24.0;36.0]	28.0 [24.0;35.0]	0.032
SAPS II	33.0 [26.0;41.0]	32.0 [26.0;39.8]	34.0 [27.0;41.0]	33.0 [26.0;41.0]	34.0 [27.0;42.0]	0.557
SIRS	2.00 [1.00;3.00]	2.00 [1.00;3.00]	2.00 [2.00;3.00]	2.00 [1.00;3.00]	2.00 [2.00;3.00]	0.149
SOFA	3.00 [2.00;6.00]	3.00 [1.00;5.00]	3.00 [2.00;5.00]	3.00 [2.00;6.00]	4.00 [2.00;6.00]	0.073
TyG	8.96 (0.76)	8.19 (0.34)	8.65 (0.31)	9.10 (0.34)	9.88 (0.65)	<0.001
TabCI	8.37 [7.75;9.08]	7.27 [6.99;7.50]	8.06 [7.89;8.23]	8.70 [8.54;8.90]	9.71 [9.35;10.1]	<0.001

3.2 Main Results

Kaplan-Meier curves indicated changes in ACM across quartiles of TabCI at 90 days, 180 days, and 1 year (Figure 2).Importantly, there was no discernible trend of positive or negative correlation between long-term survival rates and quartiles of TabCI, whether for the primary outcome of 180-day mortality risk or secondary outcomes at 90 days or 1 year.However, patients in the Q1 and Q3 groups exhibited significantly lower long-term survival rates compared to those in the Q2 and Q4 groups, with corresponding p-values of 0.049 for the primary outcome at 180 days and 0.015 and 0.014 for the secondary outcomes at 90 days and 1 year, respectively (Figure 2).

Subsequently, we assessed the clinical predictive performance of TabCI for mortality using receiver operating characteristic (ROC) curves. However, the area under the curve (AUC) values indicated limited predictive ability: AUC for primary outcome at 180 days was 0.543, and for secondary outcomes, AUC was 0.541 at 90 days and 0.561 at 1 year (Figure 3). Then, We grouped the participants based on the quartiles of TabCI and observed the predictive effects of these four groups on long-term mortality events (Fig. S1). However, compared to the traditional TyG index, the TabCI index shows a slight advantage in predicting long-term all-cause mortality in stroke patients. The AUC values for the TyG index in predicting long-term all-cause mortality are as follows: 90 days: 0.506; 180 days: 0.505; 1 year: 0.482 (Figure S2).

3.3 Correlation between TabCI index and clinical outcomes in stroke patients

All variables from the baseline table were included in a LASSO regression analysis, followed by ten-fold cross-validation, resulting in λ value of 0.02100445 (Figure 4). Ultimately, 15 prognostically relevant covariates were selected, including age, DBP, heart rate, respiratory rate, atrial fibrillation, coronary heart disease, diabetes, myocardial infarction, alteplase use, cilostazol use, clopidogrel use, tirofiban use, mechanical ventilation, APACHE III score, and SAPS II score.

To explore the independent effect of TabCI on long-term all-cause mortality in stroke patients, we employed three Cox proportional hazards regression models (Table 2). Model 1 was unadjusted for covariates, Model 2 adjusted for age, DBP, heart rate, and respiratory rate, and Model 3 further adjusted for age, DBP, heart rate, respiratory rate, atrial fibrillation, coronary heart disease, diabetes, myocardial infarction, alteplase use, cilostazol use, clopidogrel use, tirofiban use, mechanical ventilation, APACHE III score, and SAPS II score.

In Model 3, we observed that the hazard ratios (HRs) and 95% confidence intervals (CIs) for TabCI predicting all-cause mortality at 90 days, 180 days, and 1 year were 0.74 (0.64-0.86), 0.77 (0.68-0.88), and 0.85 (0.76-0.95) respectively, with corresponding P-values of <0.001, <0.001, and 0.004 (Table 2).

Using the lowest quartile of TabCI (Q1) as reference, the hazard ratios (HRs) and 95% CIs for Q2/Q3/Q4 predicting 180-day all-cause mortality were 0.55 (0.39-0.76), 0.72 (0.52-0.99), and 0.47 (0.33-0.68) respectively, with corresponding P-values of <0.001, 0.041, and <0.001, indicating statistically significant differences.

Secondary outcomes showed that for 90-day all-cause mortality, the hazard ratios were 0.61 (0.43-0.88), 0.73 (0.51-1.04), and 0.45 (0.29-0.68) with corresponding P-values of 0.009, 0.085, and <0.001. For 1-year all-cause mortality, the hazard ratios were 0.56 (0.41-0.77), 0.77 (0.57-1.03), and 0.57 (0.41-0.78) with corresponding P-values of <0.001, 0.076, and <0.001. Results indicated statistical significance for all groups except Q3 in 90 days and 1 year.These findings were consistent with Kaplan-Meier curve analysis, showing that stroke patients with TabCI indices in the ranges of (7.75-8.37) and (9.08-13.1) faced higher long-term all-cause mortality risks compared to those in the ranges of (5.88-7.75) and (8.37-9.08) (Table 2).

Table 2.Multivariable Cox proportional hazard models for long-term all-cause mortality

Variable	Model Ⅰ HR	95% CI	P-value	Model Ⅱ HR	95% CI	P-value	Model Ⅲ HR	95% CI	P-value
90-d mortality
TabCI	0.9	0.80-1.01	0.071	0.84	0.73-0.95	0.007	0.74	0.64-0.86	<0.001
Q1
Q2	0.74	0.53-1.02	0.068	0.68	0.47-0.96	0.03	0.61	0.43-0.88	0.009
Q3	0.94	0.69-1.29	0.7	0.91	0.65-1.27	0.6	0.73	0.51-1.04	0.085
Q4	0.66	0.47-0.93	0.017	0.61	0.42-0.9	0.011	0.45	0.29-0.68	<0.001
180-d mortality
TabCI	0.92	0.83-1.02	0.11	0.86	0.76-0.96	0.01	0.77	0.68-0.88	<0.001
Q1
Q2	0.68	0.51-0.92	0.013	0.60	0.44-0.84	0.002	0.55	0.39-0.76	<0.001
Q3	0.96	0.72-1.27	0.8	0.89	0.66-1.20	0.5	0.72	0.52-0.99	0.041
Q4	0.69	0.51-0.94	0.017	0.63	0.45-0.88	0.007	0.47	0.33-0.68	<0.001
1-year mortality
TabCI	1.01	0.92-1.11	0.8	0.97	0.87-1.07	0.5	0.85	0.76-0.95	0.004
Q1
Q2	0.68	0.51-0.91	0.008	0.62	0.46-0.84	0.002	0.56	0.41-0.77	<0.001
Q3	1.05	0.81-1.36	0.7	0.98	0.74-1.29	0.9	0.77	0.57-1.03	0.076
Q4	0.88	0.67-1.14	0.3	0.83	0.62-1.11	0.2	0.57	0.41-0.78	<0.001

3.4 The detection of nonlinear relationships

RCS analysis reveals a gradual L-shaped relationship between TabCI and all-cause mortality at 90 and 180 days among stroke patients. The presence of a threshold effect is evident from the graph, indicating that beyond a certain TabCI threshold, the risk of death does not significantly increase with further TabCI elevation.Conversely, a gradual U-shaped relationship is observed in 1-year all-cause mortality, suggesting that both low and high TabCI levels increase the risk of death at 1 year.Specifically, significant nonlinearity is observed only in the RCS curve for 1-year all-cause mortality, with P nonlinear = 0.0035 and P overall = 0.105 (Figure 5C).However, similar significant trends were not observed in the RCS curves for 90-day and 180-day all-cause mortality among stroke patients (90 days: P nonlinear = 0.227, P overall = 0.083; 180 days: P nonlinear = 0.127, P overall = 0.08), as shown in Figures 5A and 5B.

3.5 Subgroup Analysis

Subgroup analyses and interaction tests were conducted based on age (≤65 years or >65 years), gender, ethnicity, coronary heart disease, type 2 diabetes, hypertension, myocardial infarction, aspirin use, cilostazol use, clopidogrel use, tirofiban use, CRRT (continuous renal replacement therapy), and MV (mechanical ventilation) (Figure 6).

The study findings reveal a significant association between lower mortality risk in subgroups of non-white stroke patients and TabCI, with HR (95% CI) of 0.68 (0.55-0.84) at 90 days, 0.75 (0.63-0.9) at 180 days, and 0.79 (0.67-0.94) at 1 year.Conversely, higher mortality rates in subgroups of white stroke patients are associated with TabCI, with HR (95% CI) of 1.04 (0.9-1.19) at 90 days, 1.03 (0.91-1.17) at 180 days, and 1.14 (1.02-1.27) at 1 year, all statistically significant (90 days [P<0.001], 180 days [P=0.001], 1 year [P=0.006]).

Regarding aspirin use, significant associations are observed between TabCI and lower mortality rates in subgroups of stroke patients receiving aspirin antiplatelet therapy, with HR (95% CI) of 0.86 (0.76–0.98) at 90 days, 0.88 (0.79–0.98) at 180 days, and 0.97 (0.88–1.07) at 1 year.Conversely, higher mortality rates in subgroups of stroke patients not receiving aspirin antiplatelet therapy are associated with TabCI, with HR (95% CI) of 1.25 (0.9-1.73) at 90 days, 1.34 (0.99-1.83) at 180 days, and 1.42 (1.06-1.91) at 1 year, all statistically significant (90 days [P=0.029], 180 days [P=0.007], 1 year [P=0.008]).

Additionally, among stroke patients not receiving CRRT treatment, TabCI significantly reduces the risk of death at 90 days and 180 days, with HR (95% CI) of 0.83 (0.73–0.95, P=0.007) and 0.88 (0.78–0.98, P=0.032), respectively.Notably, among stroke patients not receiving CRRT treatment, lower TabCI does not significantly benefit 1-year mortality risk (HR 0.99, 95% CI 0.9–1.1, P=0.412), indicating no statistically significant effect.

This study introduces a novel index, TabCI, combining TyG, AIP, and BMI for the first time, and explores its correlation with long-term mortality risk in severe ischemic stroke patients.TyG was initially identified as a pathophysiological marker for obesity, metabolic syndrome, and insulin resistance in diabetes ^[10].Insulin resistance is closely associated with stroke and serves as an independent predictor of stroke onset and progression in hospitalized patients ^[11^-12].Recently, TyG has been reported as a clinical prognostic indicator for cardiovascular disease incidence and mortality risk ^[13].This index primarily incorporates fasting glucose and serum triglycerides, two key laboratory indicators closely associated with stroke incidence and progression.Blood glucose levels are associated with multiple pathological changes leading to stroke, such as arteriosclerosis and microcirculatory disturbances ^[14].Analysis of multicenter RCT data from Solitaire Flow Restoration With Intention for Thrombectomy (SWIFT) indicates that participants with baseline serum glucose levels >7.8 mmol/L (140 mg/dL) have a higher risk of functional deterioration at 3 months ^[15].The relationship between TGs and ischemic stroke has been extensively studied, and a large body of evidence indicates that hypertriglyceridemia is a risk factor for ischemic stroke, and the multifactory-adjusted hazard ratio (HR) and 95% confidence interval (CI) for ischemic stroke are 1.07 (95%CI, 95%CI) for every 1 mmol/l increase in triglycerides. 1.05 to 1.09) ^[16].Combining TyG with obesity metrics such as Body Mass Index (BMI), Waist Circumference (WC), and Waist-to-Height Ratio (WtHR) significantly enhances the accuracy of insulin assessment ^[17].

This study categorized subjects into four groups based on TabCI quartiles and examined their long-term mortality rates.Our Kaplan-Meier survival analysis revealed no significant correlation between TabCI quartiles and long-term survival rates at 90 days, 180 days, or 1 year.However, notably, patients in Q1 and Q3 exhibited significantly lower long-term survival rates compared to those in Q2 and Q4 at these time points.This suggests higher survival rates among severe ischemic stroke patients when TabCI ranges between 7.89-8.23 and 9.35-10.1.In contrast to traditional TyG index, we conducted ROC curve analyses using overall TabCI and quartiles to further explore its diagnostic efficacy for predicting long-term survival rates in severe ischemic stroke patients. Although TabCI did not demonstrate significant specificity in this regard, it outperformed TyG when compared.Remarkably, TabCI showed superior diagnostic efficacy over TyG for predicting long-term survival rates in severe ischemic stroke patients. This superiority may be attributed to TabCI's integration with the Atherogenic Index of Plasma (AIP).The plasma atherogenic index (AIP) is a novel biomarker in today's era of cardiovascular disease, proven to be positively associated with the risk of cardiovascular disease mortality ^[18]. Our TabCI index combines the TyG index with two other metabolic markers, thereby enhancing the accuracy of prognosis diagnosis for cardiovascular disease patients to some extent.

Furthermore, this study explored TabCI levels at 90 days, 180 days, and 1 year, revealing a gradual L-shaped correlation trend between TabCI levels and all-cause mortality among stroke patients at 90 and 180 days. This trend indicates a threshold effect, suggesting that lower TabCI levels moderately increase the risk of all-cause mortality at these time points. Conversely, a U-shaped correlation trend was observed for 1-year all-cause mortality, indicating that both very low and very high TabCI levels elevate the risk of death within a year. Therefore, controlling TabCI within a specific range may reduce long-term mortality rates in stroke patients. These findings provide theoretical support for developing clinical guidelines aimed at reducing mortality risk in severe stroke patients. Hence, TabCI holds promise as a reliable assessment tool for long-term follow-up care and stratified management of severe stroke patients across different risk profiles.

Finally, to gain deeper insights into the impact of TabCI on long-term mortality among critically ill stroke patients, we conducted subgroup analyses adjusting for multiple covariates.We observed a significant association between lower TabCI and increased mortality rates at 180 days, 90 days, and 1 year among non-Caucasian stroke patients who received aspirin intervention.Timely antiplatelet therapy is universally recognized by guidelines to improve neurological outcomes and reduce mortality risks for ischemic stroke patients. Guidelines recommend administering aspirin within 48 hours of ischemic stroke onset, aligning with overall patient benefits ^[19].Previous studies have indicated that compared to Caucasians, African-American, Hispanic, and other ethnic/racial groups have a higher compliance rate with fenofibrate oral intake, leading to a greater reduction in triglyceride levels ^[20]. This suggests higher adherence to treatment regimens among these groups compared to Caucasians.Furthermore, we observed a significant inverse relationship between higher TabCI and reduced 90-day and 180-day mortality risks among stroke patients who did not receive CRRT treatment. Conversely, this trend was reversed in stroke patients who received CRRT, possibly due to renal impairment and the severity of their condition. Renal dysfunction often accompanies metabolic abnormalities, which can exacerbate dyslipidemia and hyperglycemia, thereby increasing adverse cardiovascular events ^[21].

It is important to note the limitations of this study. Firstly, it is a single-center retrospective study relying on observational data retrieved from the MIMIC-IV database, which presents challenges in establishing definitive causal relationships. Secondly, the study included a relatively small sample size of only 1141 cases, all of whom were critically ill ischemic stroke patients treated in the United States. To comprehensively evaluate the predictive performance of this index in the real world, exploration among other mild stroke or TIA patients is warranted. Additionally, the inclusion was limited to the United States, with no reported studies on Asians, particularly Chinese individuals. Lastly, our primary outcome measure was long-term all-cause mortality rather than stroke-related mortality, which may be influenced by numerous uncontrollable factors. Despite adjusting for many variables and conducting subgroup analyses, the potential impact of unmeasured confounders cannot be completely ruled out. Future research will involve large-scale longitudinal studies among Chinese and other Asian populations to strengthen the conclusions of this study.

In conclusion, TabCI shows a gradual L-shaped correlation between severe ischemic stroke and 90-day and 180-day all-cause mortality, while displaying a flattened U-shaped trend in one-year all-cause mortality, suggesting TabCI could serve as a marker for long-term risk stratification in severe ischemic stroke patients.However, TabCI exhibits limited clinical predictive efficacy for long-term mortality in severe ischemic stroke patients.It is important to note that this study represents preliminary findings. Further rigorous research in diverse populations is needed to validate our discoveries and assess the general applicability and precision of TabCI as a prognostic tool.

6.1 Contributors

Hu Y.Q. designed the study, wrote the first draft of the manuscript and verified the underlying data. Hu Y.Q., Li J.X.conducted the statistical analysis, Hu Y.Q. and Wang C.H. played a role in the acquisition of the data and analyses, and participated in data interpretation. Hu Y.Q. edited the figures and tables. Cheng J.W. directed the study design and funded the study. All authors revised and approved the final manuscript. The guarantor (Cheng J.W.) confirms that all listed authors meet the authorship criteria and that no others meeting the criteria have been omitted.

6.2 Data Sharing Statement

All data are publicly available in MIMIC-IV (https://charls.charlsdata.com).

6.3 Declaration of Competing Interests

All authors declare the following: no support from any organization for the submitted work; no financial relationships with any organizations that might have an interest in the submitted work in the previous three years; and no other relationships or activities that could appear to have influenced the submitted work.

6.4 Acknowledgments

I would like to extend my sincere gratitude to the corresponding author of this study, Jiwei Cheng,for her instructive advice and useful suggestions on this study. I am deeply grateful of her funding and help in the completion of this study.

6.5 Funding

This study was supported by the

1.Project of “XingLin Scholars Training” of Chengdu University of Traditional Chinese Medicine (No. YYZX2022170)

2.Training Plan of “100 professionals” of Shanghai Putuo District Central Hospital (No.2022-RCJC-05)

3.Shanghai Putuo District Health System Clinical Characteristic Special Disease Construction Project (No. 2023tszb04)

4. Science and Technology Innovation Project of Putuo District Health System (No. ptkwws202301)

6.6 Ethics approval and consent to participate

The data used in this study are all from the publicly available MIMIC-IV database. Because the database does not contain any protected health information and patients remain anonymous, a waiver of informed consent was obtained.

6.7 Consent for publication

Not applicable, this study did not contain any details, images, or videos relating to an individual person.

GBD 2016 Lifetime Risk of Stroke Collaborators, Feigin VL, Nguyen G, Cercy K, Johnson CO, Alam T, et al. Global, regional, and country-specific lifetime risks of stroke, 1990 and 2016. N Engl J Med. (2018) 379:2429– 37. doi: 10.1056/NEJMoa1804492
Chen Y, Wright N, Guo Y, Turnbull I, Kartsonaki C, Yang L, et al. Mortality and recurrent vascular events after first incident stroke: a 9-year communitybased study of 0·5 million Chinese adults. Lancet Glob Health. (2020) 8:e580–90. doi: 10.1016/S2214-109X(20)30069-3
Zhao J, Su YY, Zhang Y, Zhang YZ, Zhao R, Wang L, Gao R, Chen W, Gao D. Decompressive hemicraniectomy in malignant middle cerebral artery infarct: a randomized controlled trial enrolling patients up to 80 years old. Neurocrit Care. 2012 Oct;17(2):161-71. doi: 10.1007/s12028-012-9703-3. PMID: 22528280.
Jüttler E, Schwab S, Schmiedek P, Unterberg A, Hennerici M, Woitzik J, Witte S, Jenetzky E, Hacke W; DESTINY Study Group. Decompressive Surgery for the Treatment of Malignant Infarction of the Middle Cerebral Artery (DESTINY): a randomized, controlled trial. Stroke. 2007 Sep;38(9):2518-25. doi: 10.1161/STROKEAHA.107.485649. Epub 2007 Aug 9. PMID: 17690310.
Vahedi K, Vicaut E, Mateo J, Kurtz A, Orabi M, Guichard JP, Boutron C, Couvreur G, Rouanet F, Touzé E, Guillon B, Carpentier A, Yelnik A, George B, Payen D, Bousser MG; DECIMAL Investigators. Sequential-design, multicenter, randomized, controlled trial of early decompressive craniectomy in malignant middle cerebral artery infarction (DECIMAL Trial). Stroke. 2007 Sep;38(9):2506-17. doi: 10.1161/STROKEAHA.107.485235. Epub 2007 Aug 9. PMID: 17690311.
Ormazabal V, Nair S, Elfeky O, Aguayo C, Salomon C, Zuñiga FA. Association between insulin resistance and the development of cardiovascular disease. Cardiovasc Diabetol. 2018 Aug 31;17(1):122. doi: 10.1186/s12933-018-0762-4. PMID: 30170598; PMCID: PMC6119242.
Ding PF, Zhang HS, Wang J, Gao YY, Mao JN, Hang CH, Li W. Insulin resistance in ischemic stroke: Mechanisms and therapeutic approaches. Front Endocrinol (Lausanne). 2022 Dec 15;13:1092431. doi: 10.3389/fendo.2022.1092431. PMID: 36589857; PMCID: PMC9798125.
Hill MA, Yang Y, Zhang L, Sun Z, Jia G, Parrish AR, Sowers JR. Insulin resistance, cardiovascular stiffening and cardiovascular disease. Metabolism. 2021 Jun;119:154766. DOI: 10.1016/j.metabol.2021.154766.
Qu L, Fang S, Lan Z, Xu S, Jiang J, Pan Y, Xu Y, Zhu X, Jin J. Association between atherogenic index of plasma and new-onset stroke in individuals with different glucose metabolism status: insights from CHARLS. Cardiovasc Diabetol. 2024 Jun 21;23(1):215. doi: 10.1186/s12933-024-02314-y. PMID: 38907337.
Hill MA, Yang Y, Zhang L, Sun Z, Jia G, Parrish AR, Sowers JR. Insulin resistance, cardiovascular stiffening and cardiovascular disease. Metabolism. 2021 Jun;119:154766. DOI: 10.1016/j.metabol.2021.154766.
Huang Q, Yin L, Liu Z, Wei M, Feng J, Huang Q, Liu Y, Liu Z, Xia J. Association of novel lipid indicators with the risk of stroke among participants in Central China: a population-based prospective study. Front Endocrinol (Lausanne). 2023 Oct 2;14:1266552. doi: 10.3389/fendo.2023.1266552. PMID: 37850101; PMCID: PMC10577285.
Yang Y, Huang X, Wang Y, Leng L, Xu J, Feng L, Jiang S, Wang J, Yang Y, Pan G, Jiang B, Wang Y, Chen L. The impact of triglyceride-glucose index on ischemic stroke: a systematic review and meta-analysis. Cardiovasc Diabetol. 2023 Jan 6;22(1):2. doi: 10.1186/s12933-022-01732-0. PMID: 36609319; PMCID: PMC9825038.
Yu H , Tao L , Li Y G ,et al.Association between triglyceride-glucose index trajectories and carotid atherosclerosis progression[J].Cardiovascular Diabetology, 2023, 22(1).DOI:10.1186/s12933-023-01847-y.
Sacco S, Foschi M, Ornello R, De Santis F, Pofi R, Romoli M. Prevention and treatment of ischaemic and haemorrhagic stroke in people with diabetes mellitus: a focus on glucose control and comorbidities. Diabetologia. 2024 Jul;67(7):1192-1205. doi: 10.1007/s00125-024-06146-z. Epub 2024 Apr 16. PMID: 38625582; PMCID: PMC11153285.
Kim JT, Jahan R, Saver JL; SWIFT Investigators. Impact of Glucose on Outcomes in Patients Treated With Mechanical Thrombectomy: A Post Hoc Analysis of the Solitaire Flow Restoration With the Intention for Thrombectomy Study. Stroke. 2016 Jan;47(1):120-7. doi: 10.1161/STROKEAHA.115.010753. Epub 2015 Dec 10. PMID: 26658447.
Gu X, Li Y, Chen S, Yang X, Liu F, Li Y, Li J, Cao J, Liu X, Chen J, Shen C, Yu L, Huang J, Lam TH, Fang X, He Y, Zhang X, Lu X, Wu S, Gu D. Association of Lipids With Ischemic and Hemorrhagic Stroke: A Prospective Cohort Study Among 267 500 Chinese. Stroke. 2019 Dec;50(12):3376-3384. doi: 10.1161/STROKEAHA.119.026402. Epub 2019 Oct 29. PMID: 31658904.
Bala C, Gheorghe-Fronea O, Pop D, Pop C, Caloian B, Comsa H, Bozan C, Matei C, Dorobantu M. The Association Between Six Surrogate Insulin Resistance Indexes and Hypertension: A Population-Based Study. Metab Syndr Relat Disord. 2019 Jul/Aug;17(6):328-333. doi: 10.1089/met.2018.0122. Epub 2019 Apr 29. PMID: 31034338.
Rabiee Rad M, Ghasempour Dabaghi G, Darouei B, Amani-Beni R. The association of atherogenic index of plasma with cardiovascular outcomes in patients with coronary artery disease: A systematic review and meta-analysis. Cardiovasc Diabetol. 2024 Apr 2;23(1):119. doi: 10.1186/s12933-024-02198-y. PMID: 38566139; PMCID: PMC10986012.
Jauch EC, Saver JL, Adams HP Jr, Bruno A, Connors JJ, Demaerschalk BM, Khatri P, McMullan PW Jr, Qureshi AI, Rosenfield K, Scott PA, Summers DR, Wang DZ, Wintermark M, Yonas H; American Heart Association Stroke Council; Council on Cardiovascular Nursing; Council on Peripheral Vascular Disease; Council on Clinical Cardiology. Guidelines for the early management of patients with acute ischemic stroke: a guideline for healthcare professionals from the American Heart Association/American Stroke Association. Stroke. 2013 Mar;44(3):870-947. doi: 10.1161/STR.0b013e318284056a. Epub 2013 Jan 31. PMID: 23370205.
Cromer SJ, Thaweethai T, Wexler DJ. Racial/ethnic and socioeconomic disparities in achievement of treatment goals within a clinical trial: a secondary analysis of the ACCORD trial. Diabetologia. 2023 Dec;66(12):2261-2274. doi: 10.1007/s00125-023-05997-2. Epub 2023 Sep 16. PMID: 37715820; PMCID: PMC10942722.
Ndumele CE, Neeland IJ, Tuttle KR, Chow SL, Mathew RO, Khan SS, Coresh J, Baker-Smith CM, Carnethon MR, Després JP, Ho JE, Joseph JJ, Kernan WN, Khera A, Kosiborod MN, Lekavich CL, Lewis EF, Lo KB, Ozkan B, Palaniappan LP, Patel SS, Pencina MJ, Powell-Wiley TM, Sperling LS, Virani SS, Wright JT, Rajgopal Singh R, Elkind MSV, Rangaswami J; American Heart Association. A Synopsis of the Evidence for the Science and Clinical Management of Cardiovascular-Kidney-Metabolic (CKM) Syndrome: A Scientific Statement From the American Heart Association. Circulation. 2023 Nov 14;148(20):1636-1664. doi: 10.1161/CIR.0000000000001186. Epub 2023 Oct 9. PMID: 37807920.

No competing interests reported.

Supplementarymaterial.docx

Download PDF

Reviews received at journal
23 Aug, 2024
Reviewers agreed at journal
20 Aug, 2024
Reviewers agreed at journal
19 Aug, 2024
Reviewers invited by journal
18 Aug, 2024
Editor assigned by journal
18 Aug, 2024
Editor invited by journal
14 Aug, 2024
Submission checks completed at journal
13 Aug, 2024
First submitted to journal
07 Aug, 2024

You are reading this latest preprint version

Long-term Survival in Ischemic Stroke Patients: A Comprehensive Analysis of the TyG-AIP-BMI Composite Index (TabCI) from MIMIC-IV ICU Data

Status:

Version 1

Abstract

Figures

1 Introduction

2 Methods

2.1 Source of data

2.2 Study design and population

2.3 Data extraction

2.4 Outcomes

2.5 TyG-AIP-BMI Composite Index calculation

2.6 Statistical analysis

3 Result

4 Discussion

5 Conclusion

Declarations

References

Additional Declarations

Supplementary Files

Status:

Version 1