Association between the aggregate index of systemic inflammation and 28- or 365-day mortality in acute myocardial infarction patients: insights from the MIMIC-IV database

Abstract

Background

Systemic inflammation plays a key role in acute myocardial infarction (AMI). The Aggregate Index of Systemic Inflammation (AISI), a composite measure of neutrophils, monocytes, platelets, and lymphocytes, may offer a more comprehensive assessment of inflammation than single biomarkers. However, its association with short- and long-term mortality in AMI patients remains underexplored.

Methods

This retrospective study analyzed data from 969 AMI patients in the Medical Information Mart for Intensive Care(MIMIC) -IV database, stratified by AISI levels into tertiles. Cox regression and Kaplan-Meier survival analyses assessed the relationship between AISI and mortality. Subgroup analyses examined age, gender, and comorbidity effects, and the predictive value of AISI was evaluated using Receiver Operating Characteristic(ROC) curves.

Result

This study enrolled 969 patients with AMI, stratified into group1(low), group2(medium), and group3(high) tertiles. The group3 exhibited significantly higher 28-day mortality (18.6% vs. 5.6% in low/medium groups) and 365-day mortality (27.2% vs. 13.3% in the low group) (both P < 0.001). Survival analysis demonstrated an inverse correlation between AISI levels and survival probability (P < 0.001). Restricted cubic spline(RCS) models revealed a nonlinear association between AISI and mortality (28-day: likelihood ratio test P = 0.001; 365-day: P = 0.002). Multivariable Cox regression confirmed a dose-response relationship between AISI tertiles and mortality risk (trend P < 0.001 for both endpoints). Subgroup analyses highlighted greater sensitivity to AISI in female patients (28-day: interaction P = 0.041; 365-day: P = 0.011), with ROC curve analysis showing enhanced predictive performance in females (28-day AUC: 0.759 vs. 0.669, P = 0.019; 365-day AUC: 0.760 vs. 0.730, P = 0.038).

Conclusion

In this study, AISI exhibited a nonlinear association with 28-day and 365-day mortality in AMI patients. Higher AISI levels were significantly linked to increased mortality risk. Notably, female patients showed greater sensitivity to elevated AISI, suggesting its potential value in risk stratification and clinical management.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12872-025-05401-4.

Introduction

Coronary artery disease (CAD) remains one of the most significant threats to human health, imposing a growing burden on healthcare expenditures and global social productivity [1]. Acute myocardial infarction (AMI), the most severe manifestation of CAD, carries the highest mortality rate within this disease spectrum. Despite substantial advances in revascularization techniques and antithrombotic therapies, AMI patients continue to experience high rates of hospital readmission and mortality [2]. Early identification of high-risk patients is therefore crucial for optimizing treatment strategies and improving survival outcomes. Inflammation plays a pivotal role in AMI pathogenesis, contributing not only to acute myocardial injury but also significantly influencing long-term cardiovascular remodeling and complications [3]. Current prognostic models incorporate inflammatory markers such as the neutrophil-lymphocyte ratio (NLR) and platelet-lymphocyte ratio (PLR) [4, 5]. However, these individual markers may not fully capture the complexity of systemic inflammation. The Aggregate Index of Systemic Inflammation (AISI), a composite metric integrating neutrophil, monocyte, platelet, and lymphocyte counts, provides a more comprehensive assessment of inflammatory burden [6]. Although AISI has potential clinical value, studies exploring the association between AISI and short-term (28-day) and long-term (365-day) mortality in AMI patients remain limited. Furthermore, the differential impact of AISI across various subgroups (such as age, gender, and comorbidities) remains unclear. Addressing this research gap could provide valuable insights into risk stratification and personalized management for AMI patients. Despite its potential clinical utility, research investigating the association between AISI and short-term (28-day) and long-term (365-day) mortality in AMI patients remains limited. Furthermore, the differential impact of AISI across demographic and clinical subgroups (e.g., age, gender, comorbidities) remains unelucidated. Addressing these knowledge gaps could enhance risk stratification and facilitate personalized management for AMI patients. This study aimed to evaluate the association between AISI and 28-day and 365-day mortality in AMI patients and to explore subgroup-specific differences. Additionally, the study assessed the incremental value of AISI in predictive models, providing important references for clinical risk assessment and patient management.

Methods

Data source

This study utilized a large, publicly available database—Medical Information Mart for Intensive Care (MIMIC)—which contains extensive clinical data on hospital stays for patients admitted to a tertiary academic medical center in Boston. Data extraction from the MIMIC-IV database (version 2.2, which includes ICU patients admitted to BIDMC between 2008 and 2019) was performed using Structured Query Language (SQL) to ensure efficient and accurate retrieval of relevant information.

Population selection criteria and extracted variables

Inclusion criteria: Patients aged 18–100 years, diagnosed with acute myocardial infarction based on International Classification of Diseases, 9th and 10th Revisions(ICD-9 and ICD-10), with at least one measurement of neutrophil count, monocyte count, platelet count, and lymphocyte count after admission. Exclusion criteria: Patients with concomitant sepsis or history of malignant tumors. The extracted variables included patient demographics (age and gender), counts of neutrophils, monocytes, platelets, and lymphocytes, comorbidities (hypertension, type 2 diabetes mellitus, heart failure, and stroke), other laboratory parameters (uric acid, total bilirubin, direct bilirubin, indirect bilirubin, Alanine Aminotransferase(ALT), Aspartate Aminotransferase(AST), blood urea nitrogen, serum creatinine, triglycerides, total cholesterol, High-Density Lipoprotein Cholesterol(HDL-c), Low-Density Lipoprotein Cholesterol(LDL-c), lactate, partial pressure of oxygen, partial pressure of carbon dioxide, red blood cell count, and hemoglobin), and medication use (aspirin, metoprolol, sacubitril/valsartan, statins, and enoxaparin). All laboratory results were obtained from the first measurement after admission. The AISI value was calculated using the formula: (neutrophil count × monocyte count × platelet count)/lymphocyte count(Fig. 1).

Fig. 1.

Flowchart of population selection.AISI(Aggregate Index of Systemic Inflammation)

Outcomes

The outcomes were short-term (28-day) and long-term (365-day) mortality.

Statistic analysis

All participants were separated into tertiles according to the AISI levels: group1 (AISI < 405.1, n = 323), group2 (405.1 ≤ AISI < 1144.8, n = 323), and group3 (1144.8 ≤ AISI < 31861.6, n = 323), and the characteristics were depicted. Categorical variables were shown as numbers (n) and percentages (%) and were assessed using chi-square tests. Continuous variables were presented as the mean ± standard deviation for a normal distribution or the median (interquartile range[IQR]) for a non-normal distribution. They were analyzed using one-way Analysis of Variance (ANOVA) and the ruskal–Wallis test, respectively. Kaplan-Meier survival curves were constructed for three patient groups stratified by AISI levels. Differences in survival probabilities were compared using the log-rank test, with significance confirmed at p < 0.05. An enhanced Cox proportional hazards regression model with restricted cubic splines was employed to evaluate the potential nonlinear relationship between AISI and mortality risk. Non-linearity of the association was confirmed at p < 0.05. In the Cox regression models, AISI tertiles were incorporated as an ordinal categorical variable, and the P-value for linear trend was calculated via the Wald test. We assessed the proportional hazards assumption by using the Schoenfeld residuals test. Harrel’s C-statistics and time-dependent Area Under the Curve (AUC) with cut-offs at 28 and 365 days were performed for further analysis. Adjustments were made for age, gender, and comorbidities (hypertension, type 2 diabetes mellitus, stroke, and heart failure) to isolate the specific effect of AISI. To ensure the model’s stability and reliability, multicollinearity diagnostics were performed. Variance Inflation Factors (VIFs) were calculated for all covariates included in the regression models. All VIF values were below 5, confirming the absence of significant multicollinearity, thereby validating the independence of the variables and enhancing the credibility of the model coefficients. After z-score standardization of AISI values, the differential effects of AISI on 28-day and 365-day mortality were evaluated across predefined subgroups. P for interaction were used to determine whether the effect of AISI differed significantly across these subgroups. The discriminative ability of predictive models for 28-day and 365-day mortality was evaluated using ROC curve analysis, comparing the performance of models with and without the inclusion of AISI. The AUC comparisons between the models were evaluated with DeLong’s test. Improvements in predictive performance were confirmed as statistically significant at p < 0.05, emphasizing the incremental predictive value of AISI. All analyses were conducted using the R Statistical Software (http://www.R-project.org, Te R Foundation), Free Statistics software versions 1.8.

Result

Characteristics of the study patients by AISI

A total of 969 patients were included, with a median age of 69.0 years (IQR: 60.0, 77.0) and a male-to-female ratio of 2.09. Table 1 presents the baseline characteristics of the study patients stratified by AISI tertiles. Patients were divided into three equal groups based on their AISI level: Group 1 (Low AISI), Group 2 (Medium AISI), and Group 3 (High AISI). The median age was similar across the three AISI groups, with no statistically significant differences (p = 0.585). Similarly, gender distribution was balanced across groups, with approximately two-thirds of patients being male in each group (p = 0.914). Notable variations existed in comorbidity prevalence. Hypertension and type 2 diabetes mellitus (T2DM) were significantly more prevalent in patients with lower AISI levels. Group 1 had the highest prevalence of hypertension (46.4%) and T2DM (43.7%), while Group 3 had the lowest rates for both conditions (hypertension: 26.6%, T2DM: 29.7%; p < 0.001 for both). Conversely, heart failure was significantly more common in Group 3 (59.1%; p < 0.001). Stroke prevalence was similar across all groups (p = 0.858). Several laboratory parameters differed significantly between groups. Group 3 had significantly higher ALT, AST, urea nitrogen, and creatinine levels compared to Groups 1 and 2 (p < 0.001), indicating worse hepatic and renal function with higher inflammation. Group 3 patients also exhibited the highest blood lactate levels (1.9 mmol/L [IQR: 1.3, 2.9]; p < 0.001) and lowest partial pressure of oxygen (PaO₂: 98.0 mm Hg [IQR: 56.0, 230.0]; p < 0.001), indicating more pronounced hypoxemia. While aspirin and statin use did not differ significantly across groups, metoprolol and enoxaparin usage varied significantly. The highest metoprolol usage was observed in Group 2 and the highest enoxaparin usage in Group 3 (p < 0.001 and p = 0.020, respectively). Sacubitril/valsartan use was limited (0.3% overall), with no significant differences among groups. Non-ST-segment elevation MI (NSTEMI) predominated overall (79.8%, 773/969), but its prevalence progressively decreased with higher AISI tertiles: 85.8% (277/323) in Group 1 vs. 71.2% (230/323) in Group 3 (p < 0.001). Conversely, ST-segment elevation MI (STEMI) frequency significantly increased across tertiles, from 14.2% (46/323) in Group 1 to 28.8% (93/323) in Group 3 (p < 0.001), suggesting a graded association between systemic inflammation and STEMI incidence. PCI rates exhibited a stepwise increase with rising AISI tertiles: 11.1% (36/323) in Group 1, 16.1% (52/323) in Group 2, and 20.1% (65/323) in Group 3 (p = 0.007). This trend aligns with the higher proportion of STEMI cases in the high-AISI group, which typically require urgent revascularization(Table 1).

Table 1.

Baseline characteristics of the study patients by tertiles of AISI

Baseline characteristics	Total (n = 969)	1 (n = 323)	2 (n = 323)	3 (n = 323)	p
Age, Median (IQR)	69.0 (60.0, 77.0)	68.0 (60.0, 76.0)	68.0 (60.0, 77.0)	70.0 (60.0, 76.5)	0.585
Gender, n (%)					0.914
Male	655 (67.6)	218 (67.5)	216 (66.9)	221 (68.4)
Female	314 (32.4)	105 (32.5)	107 (33.1)	102 (31.6)
Hypertension, n (%)	368 (38.0)	150 (46.4)	132 (40.9)	86 (26.6)	< 0.001
T2DM, n (%)	370 (38.2)	141 (43.7)	133 (41.2)	96 (29.7)	< 0.001
Stroke, n (%)	63 (6.5)	23 (7.1)	20 (6.2)	20 (6.2)	0.858
Heart failure, n (%)	436 (45.0)	101 (31.3)	144 (44.6)	191 (59.1)	< 0.001
Uric acid, Mean ± SD	7.9 ± 3.9	9.0 ± 4.1	7.6 ± 5.3	7.2 ± 3.2	0.363
Bilirubin total, Median (IQR), (mg/dL)	0.5 (0.4, 0.8)	0.5 (0.4, 0.7)	0.5 (0.4, 0.8)	0.6 (0.4, 0.8)	0.049
Bilirubin direct, Median (IQR), (mg/dL)	0.7 (0.4, 1.3)	0.8 (0.5, 1.4)	0.6 (0.3, 0.9)	0.9 (0.5, 1.7)	0.181
Bilirubinindirect, Median (IQR), (mg/dL)	0.9 (0.5, 1.6)	0.8 (0.6, 1.5)	1.2 (0.6, 2.0)	0.7 (0.4, 1.1)	0.117
ALT, Median (IQR), (IU/L)	28.0 (17.0, 52.0)	25.0 (17.0, 38.5)	26.0 (16.0, 47.0)	36.0 (21.0, 98.0)	< 0.001
AST, Median (IQR), (IU/L)	36.0 (23.0, 91.0)	30.0 (22.0, 55.5)	33.0 (22.0, 67.0)	63.0 (29.5, 189.0)	< 0.001
Blood urea nitrogen, Median (IQR), (mg/dL)	19.0 (15.0, 30.0)	18.0 (14.0, 28.0)	19.0 (14.0, 26.5)	22.0 (17.0, 35.0)	< 0.001
Creatinine, Median(IQR), (mg/dL)	1.0 (0.8, 1.4)	1.0 (0.8, 1.3)	1.0 (0.8, 1.3)	1.2 (0.9, 1.6)	< 0.001
TG, Median (IQR), (mg/dL)	122.0 (86.0, 164.0)	113.5 (79.2, 148.8)	136.0 (96.0, 184.0)	117.5 (90.0, 162.0)	0.054
Cholesterol total, Median (IQR), (mg/dL)	154.0 (127.8, 188.2)	162.0 (132.0, 193.0)	145.0 (125.0, 185.0)	157.0 (127.5, 187.5)	0.450
HDL-c, Median(IQR), (mg/dL)	43.0 (34.0, 53.0)	47.0 (35.0, 57.0)	43.0 (35.8, 51.2)	41.5 (32.0, 51.0)	0.153
LDL-c, Median(IQR), (mg/dL)	80.0 (60.0, 111.0)	91.5 (66.8, 113.5)	74.5 (57.8, 105.5)	81.5 (60.8, 107.5)	0.195
PaCO2,Median (IQR), (mmHg)	40.0 (36.0, 45.0)	40.0 (36.0, 44.0)	40.0 (36.0, 45.0)	41.0 (35.0, 46.0)	0.258
PaO2, Median (IQR), (mmHg)	198.0 (72.8, 353.0)	306.0 (139.0, 395.5)	218.0 (80.0, 357.0)	98.0 (56.0, 230.0)	< 0.001
Lactate, Median(IQR), (mmol/L)	1.5 (1.2, 2.1)	1.4 (1.1, 1.8)	1.4 (1.1, 1.9)	1.9 (1.3, 2.9)	< 0.001
RBC, Median (IQR), (m/uL)	4.1 (3.5, 4.6)	4.1 (3.5, 4.5)	4.2 (3.6, 4.6)	4.1 (3.4, 4.7)	0.466
Hemoglobin, Median(IQR), (g/dL)	12.2 (10.4, 13.8)	12.1 (10.4, 13.7)	12.4 (11.1, 14.0)	12.0 (9.9, 13.8)	0.087
Aspirin, n (%)	883 (91.1)	290 (89.8)	302 (93.5)	291 (90.1)	0.183
Metoprolol, n (%)	787 (81.2)	260 (80.5)	283 (87.6)	244 (75.5)	< 0.001
Sacubitrilvalsartan, n (%)	3 (0.3)	0 (0)	2 (0.6)	1 (0.3)	0.784
Statin, n (%)	137 (14.1)	48 (14.9)	49 (15.2)	40 (12.4)	0.538
Enoxaparin, n (%)	39 (4.0)	5 (1.5)	16 (5)	18 (5.6)	0.020
STEMI or NSTEMI, n (%)					< 0.001
NSTEMI	773 (79.8)	277 (85.8)	266 (82.4)	230 (71.2)
STEMI	196 (20.2)	46 (14.2)	57 (17.6)	93 (28.8)
PCI, n (%)					0.007
No	816 (84.2)	287 (88.9)	271 (83.9)	258 (79.9)
Yes	153 (15.8)	36 (11.1)	52 (16.1)	65 (20.1)

Data are shown as mean ± standard deviation (SD) or median (IQR) for continuous variables and proportions (%) for categorical variables. Abbreviations: IQR: Interquartile Range; T2DM: Type 2 Diabetes Mellitus; SD: Standard Deviation; ALT: Alanine Aminotransferase; AST: Aspartate Aminotransferase; TG: Triglycerides; HDL-c: High-Density Lipoprotein Cholesterol; LDL-c: Low-Density Lipoprotein Cholesterol; PaCO2: Partial Pressure of Carbon Dioxide; PaO2: Partial Pressure of Oxygen; RBC: red blood cell; STEMI: ST-segment elevation myocardial infarction; NSTEMI: Non-ST-segment elevation myocardial infarction; PCI: Percutaneous coronary intervention

ASIS and 28- or 365-Day mortality in the overall patients

The 28- or 365-day mortality rates following acute myocardial infarction were compared among the groups in Table 2. A statistically significant difference was observed in 28-day mortality across the groups (p < 0.001). Both Group 1 and Group 2 had the same 28-day mortality rate of 5.6% (18 patients), whereas Group 3 had a notably higher rate of 18.6% (60 patients). Similarly, for 365-day mortality, a statistically significant difference was seen among the groups (p < 0.001). At 365 days, the cumulative mortality rate was lowest in Group 1 at 13.3% (43 patients), followed by 17.6% (57 patients) in Group 2, and highest in Group 3 at 27.2% (88 patients)(Table 2).

Table 2.

Incidence of 28-day and 365-day mortality among all patients during follow-up

mortality	Total (n = 969)	1 (n = 323)	2 (n = 323)	3 (n = 323)	p
28-day mortality, n (%)	96 (9.9)	18 (5.6)	18 (5.6)	60 (18.6)	< 0.001
365-day mortality, n (%)	188 (19.4)	43 (13.3)	57 (17.6)	88 (27.2)	< 0.001

Data are shown as proportions (%)

Survival analysis for patients stratified into three groups based on AISI levels

The groups, labeled AISI Group 1, Group 2, and Group 3, represented ascending levels of AISI. The analysis revealed a clear trend: Group 1 consistently showed the highest survival probability across the observation period, followed by Group 2, while Group 3 demonstrated the lowest survival probability. This trend suggested a strong correlation between higher AISI levels and increased mortality risk both within 28 days and at 365 days post-acute myocardial infarction. This association was statistically significant (p < 0.0001), supporting the hypothesis that elevated AISI levels were linked to poorer survival outcomes(Fig. 2).

Fig. 2.

Kaplan-Meier survival curves for patients stratified into three groups based on AISI levels, showing survival probability over time (in days). Figure 2a represents 28-day survival, while Fig. 2b depicts 365-day survival

The nonlinear relationship between AISI and 28-day or 365-day mortality

To explore the potential nonlinear relationship between the AISI and mortality, we employed RCS within Cox proportional hazards models (Fig. 3).

Fig. 3.

The nonlinear relationship between AISI and the 28-day(3a) or 365-day(3b) mortality hazard ratio. These models are adjusted for potential confounding factors, including age, gender, hypertension, diabetes, stroke, and heart failure, to isolate the effect of AISI

28-Day mortality analysis

Optimal Knot Configuration: Based on Akaike Information Criterion (AIC) and Bayesian Information Criterion (BIC), the 28-day mortality model utilized 5 knots (3 internal knots). Internal knots were positioned at AISI values of 308.94, 686.26, and 1619.10, with boundary knots at the minimum (0.00) and maximum (31,861.56) observed AISI values. Model Comparison: Compared to a linear Cox model assuming a proportional relationship between AISI and 28-day mortality, the RCS model demonstrated significantly improved fit (likelihood ratio test [LRT], P = 0.001), indicating a statistically significant nonlinear association.

365-Day mortality analysis

Knot Configuration: Consistent with the 28-day model, the 365-day model also selected 5 knots (3 internal knots at identical positions: 308.94, 686.26, 1619.10) and boundary knots at 0.00 and 31,861.56. Model Comparison: The RCS model significantly outperformed the linear assumption model (LRT, P = 0.002), confirming a nonlinear relationship between AISI and 365-day mortality.

These results suggest that the association of AISI with both 28-day and 365-day mortality is nonlinear. Incorporating nonlinear spline functions provides a more accurate representation of AISI’s relationship with mortality risk compared to linear assumptions.

Multivariable Cox regression analysis for association between AISI tertiles and 28-day or 365-day mortality

The AISI tertiles were included as an ordinal categorical variable in the analyses. The P value for linear trend was calculated using the Wald test. As shown in Tables 3 and 4 (Multivariable Cox regression results), the column labeled “P for trend” indicates a statistically significant increasing trend in mortality risk with higher AISI tertiles for both 28-day (P for trend < 0.001) and 365-day outcomes (P for trend < 0.001), strongly supporting the existence of a dose-response relationship.

Table 3.

Multivariable Cox regression analysis for association between AISI tertiles and 28-day mortality

group	n.total	n.event_%	crude.HR_95CI	crude.P_value	adj.HR_95CI	adj.P_value	P for trend
group1	323	18 (5.6)	1(Ref)		1(Ref)		< 0.001
group2	323	18 (5.6)	1 (0.52 ~ 1.91)	0.989	0.91 (0.47 ~ 1.75)	0.777
group3	323	60 (18.6)	3.57 (2.11 ~ 6.04)	< 0.001	2.89 (1.68 ~ 4.98)	< 0.001

Crude model: no covariates were adjusted; Adjusted model: adjusted for age, gender, hypertension, T2DM, stroke, and heart failure. HR: Hazard Ratio; CI: Confidence Interval

Table 4.

Multivariable Cox regression analysis for association between AISI tertiles and 365-day mortality

group	n.total	n.event_%	crude.HR_95CI	crude.P_value	adj.HR_95CI	adj.P_value	P for trend
group1	323	43 (13.3)	1(Ref)		1(Ref)		< 0.001
group2	323	57 (17.6)	1.34 (0.9 ~ 1.99)	0.150	1.24 (0.84 ~ 1.86)	0.282
group3	323	88 (27.2)	2.29 (1.59 ~ 3.3)	< 0.001	1.93 (1.33 ~ 2.82)	0.001

Crude model: no covariates were adjusted; Adjusted model: adjusted for age, gender, hypertension, T2DM, stroke, and heart failure

Proportional hazards assumption and Time-dependent ROC analysis

In the assessment of the proportional hazards assumption via Schoenfeld residual tests, we observed distinct patterns for 28-day and 365-day mortality outcomes(Table 5).For the association between AISI and 28-day mortality, the proportional hazards assumption was not violated (P = 0.923), supporting the validity of the standard Cox proportional hazards model. In contrast, the assumption was violated for 365-day mortality (P = 0.011).

Table 5.

Proportional hazards assumption for AISI between 28-day and 365-day mortality

Variable	event	chisq	df	p.value
AISI	28-day mortality	0.009	1	0.923
	365-day mortality	6.424	1	0.011

Adjusting for covariates including age, gender, hypertension, T2DM, stroke, and heart failure

We have conducted a comprehensive analysis based on Harrell’s C-index and plot time-dependent AUC curves, adjusting for covariates including age, gender, hypertension, T2DM, stroke, and heart failure(Table 6). For 28-day prediction: AUC = 0.681, indicating moderate discriminatory ability of the model in the short term, consistent with the MaxYouden index (0.327). For 365-day prediction: AUC = 0.613, suggesting stronger predictive performance for the short term. We selected time-dependent AUC and Harrell’s C-statistic as the primary validation metrics for this analysis because they directly quantify the discriminatory ability of AISI in predicting time-to-event outcomes, aligning with our study objectives. While alternative metrics such as the Brier score (assessing overall prediction error) and net reclassification improvement (NRI, evaluating incremental predictive value) offer utility in specific contexts, they are less central to our primary focus on evaluating the fundamental discriminatory capacity of AISI.

Table 6.

Time-dependent ROC analysis

time.n	Cases	survivor	Censored	AUC	MaxYouden
28days	96	872	1	0.681	0.327
365days	188	781	0	0.613	0.205

Adjusting for covariates including age, gender, hypertension, T2DM, stroke, and heart failure

Figure 4 illustrates the time-dependent ROC curves for predicting 28-day and 365-day mortality using the AISI-based model. The model demonstrated moderate discriminatory ability for 28-day mortality (AUC = 0.681), with peak predictive accuracy occurring at early time points. In contrast, predictive performance for 365-day mortality was lower (AUC = 0.613), and the curve exhibited a gradual decline over time.

Fig. 4.

Time-dependent AUC with cut-offs at 28 and 365 days

Subgroup analysis for association between standardized AISI and 28- or 365-day mortality

After z-score standardization of AISI values, Fig. 5 examined the impact of AISI on 28-day mortality across various subgroups. Age: Standardized AISI significantly affected 28-day mortality in older patients (≥ 60 years; p < 0.001). The P for interaction was 0.920, indicating no significant difference in AISI’s effect between age groups. Gender: AISI significantly influenced 28-day mortality in both males and females, with a stronger effect observed in females (P for interaction = 0.041). Hypertension: AISI had a significant impact in both non-hypertensive and hypertensive patients, with no significant interaction effect (p = 0.305). Diabetes: Among diabetic and non-diabetic patients, AISI was significant in both groups, with no significant interaction effect (p = 0.631). Stroke history: For participants with or without a history of stroke, the interaction term (p = 0.588) indicated no significant difference in the effect of AISI. Heart failure: For patients with or without heart failure, AISI did not significantly impact mortality, as indicated by non-significant P for interaction. AMI subtype: The interaction between AISI and AMI subtype (STEMI vs. NSTEMI) was non-significant (P for interaction = 0.467), indicating that the prognostic impact of AISI on 28-day mortality did not differ statistically between STEMI and NSTEMI patients. Overall, these findings suggest that while AISI has a significant impact in several subgroups, only gender shows a statistically significant differential effect, with female patients exhibiting a stronger association with increased mortality risk. Figure 6 presents the impact of AISI on 365-day mortality across subgroups. Age: AISI significantly increased mortality in older patients (≥ 60 years; p < 0.001) but had no significant effect on younger patients (≤ 60 years; P for interaction = 0.566). Gender: AISI had no significant impact on males but significantly elevated risk in females (p < 0.001), with a significant interaction (p = 0.011), suggesting a higher risk for females. Hypertension: AISI significantly affected non-hypertensive but not hypertensive individuals (P for interaction = 0.319). Diabetes: AISI showed a significant effect in non-diabetic but not diabetic patients, with no significant interaction (p = 0.471). Stroke history: AISI was significant in patients without stroke but not in those with stroke (P for interaction = 0.298). Heart failure: AISI had a significant impact on patients both with and without heart failure, with no significant differential effect (P for interaction = 0.097). AMI subtype: No significant interaction was observed between AISI and AMI subtype (P for interaction = 0.535), indicating that the relationship between AISI and 365-day mortality risk did not differ statistically between STEMI and NSTEMI patients. These findings suggest that AISI’s effect on long-term mortality varies by gender, with no significant modification by age, hypertension, diabetes, stroke history, or heart failure status. In summary, female patients with acute myocardial infarction demonstrated greater sensitivity to AISI, resulting in a significantly increased risk of both short-term and long-term mortality(Detailed data are provided in Supplementary File S7).

Fig. 5.

Subgroup analysis for association between AISI level (Standardized by Z-Score)and 28-day mortality

Fig. 6.

Subgroup analysis for association between AISI and 365-day mortality

The P for interaction reported in Figs. 5 and 6 indicate whether the strength of the association between AISI and clinical outcomes differs statistically significantly across the specified subgroups, as assessed using likelihood ratio tests within Cox proportional hazards regression models.

Multicollinearity diagnostic conclusion

We conducted multicollinearity diagnostics on the covariates and found that all VIF values were below 5, confirming the absence of significant multicollinearity in this model(Table 7).

Table 7.

Multicollinearity diagnostic conclusion

Term1	coeff1	Change.percentage1	Term2	coeff2	Change.percentage2	VIF	colinearity	select	select.VIF
Crude	0	Ref.	Full	0	Ref.	1.032	0	Ref.	Ref.
age	0	−4.7	age	0	3.1	1.088	0	No	No
gender	0	2.3	gender	0	−2.9	1.043	0	No	No
hypertension	0	−6	hypertension	0	0.8	1.178	0	No	No
T2DM	0	−1.1	T2DM	0	1.7	1.006	0	No	No
stroke	0	−0.1	stroke	0	0	1.019	0	No	No
heartfailure	0	−10.8	heartfailure	0	3.6	1.255	0	Yes	Yes

VIF Variance Inflation Factor

ROC curve analysis for predicting 28- or 365-day mortality in female patients

The ROC curve analysis for predicting 28-day mortality in female patients revealed that Model 1 demonstrated superior discriminatory performance, with an AUC of 0.759 (95% CI: 0.677–0.842). In comparison, Model 2 achieved a moderately good predictive ability, reflected by an AUC of 0.669 (95% CI: 0.587–0.751), though its performance was inferior to Model (1) Notably, the inclusion of AISI in Model 1 significantly enhanced Model 1’s predictive capability, as evidenced by the higher AUC value. The difference in AUC values between the two models was statistically significant (P = 0.019), underscoring the incremental predictive value of AISI for 28-day mortality. These findings highlighted that the improvement observed with the incorporation of AISI was robust. The ROC curve analysis for predicting 365-day mortality in female patients demonstrated similar findings. Model 1 achieved an AUC of 0.760 (95% CI: 0.700–0.819), indicating superior discriminatory power. In contrast, Model 2 showed a slightly lower predictive performance, with an AUC of 0.730 (95% CI: 0.668–0.792). The inclusion of AISI in Model 1 led to a significant enhancement in predictive ability compared to Model (2) The observed improvement was statistically significant, with a P-value of 0.038, further supporting the incremental value of AISI in improving long-term mortality prediction. These findings aligned with the results from the 28-day mortality analysis, highlighting the consistent utility of AISI in enhancing predictive models(Fig. 7).

Fig. 7.

The ROC curves of AISI as a marker to predict 28-day(7a)and 365-day (7b)mortality. Model 1: baseline risk factors + AISI. Model 2: baseline risk factors only. Baseline risk factors include age, hypertension, T2DM, stroke and heart failure

Discussion

The prognostic utility of complete blood cell count (CBC) in patients with AMI is often overlooked by clinicians. In contrast, composite inflammatory indices derived by integrating multiple hematological parameters are considered to more accurately reflect the inflammatory status than single biomarkers [7]. Based on existing studies, Several composite indices calculated from CBC, such as the NLR, PLR, lymphocyte-to-monocyte ratio (LMR), HALP (hemoglobin, albumin, lymphocyte, and platelet) score [8], systemic immune-inflammation index (SII), systemic inflammation response index (SIRI), and pan-immune-inflammatory value (PIV), are often used to assess the immune and inflammatory status of the body [4, 9, 10]. AISI is a relatively new composite marker that incorporates neutrophils, lymphocytes, monocytes, and platelets to provide a more integrated evaluation of systemic inflammation. AISI was initially used to identify patients at risk of prolonged hospital stay after open thoracic surgery [6], but subsequent studies have confirmed its reliable predictive value in assessing the prognosis of various systemic diseases, including diabetes [9], malignant tumors (such as esophageal cancer [11], prostate cancer [12], etc.), intracerebral hemorrhage [10], COVID-19 [13], and hemophagocytic lymphohistiocytosis [14]. A recent single-center study demonstrated a positive association between the AISI and major adverse cardiovascular and cerebrovascular events (including all-cause in-hospital mortality, revascularization, new-onset atrial fibrillation of any etiology, venous thromboembolism, and stroke) in AMI patients [15]. AISI incorporates four key hematologic components—neutrophils, lymphocytes, monocytes, and platelets—thereby providing a more comprehensive characterization of the complex inflammatory-immune network following AMI. Although SIRI (neutrophils × monocytes/lymphocytes) and AISI overlap in cellular components, AISI’s inclusion of platelets contributes critically to thrombotic processes and inflammatory amplification, potentially enabling more comprehensive risk stratification for atherosclerotic cardiovascular events. The study by Murat and Bektas et al. has confirmed that PIV (which shares its formula with AISI) outperforms NLR, PLR, and SII in predicting adverse events in STEMI patients [7]. Of particular note, our subgroup analysis revealed that AISI exhibited enhanced predictive performance specifically in female AMI patients—a sex-specific relationship not consistently observed in prior investigations of established inflammatory markers such as SIRI and NLR [16]. Our study is among the first to employ AISI as a composite inflammatory marker to explore its nonlinear relationship with mortality in AMI patients. Although the association between AISI and mortality exhibited a nonlinear pattern, the overall dose-response relationship remained statistically significant. This suggests that varying pathophysiological mechanisms such as excessive inflammation at high AISI levels or immune suppression at low levels may contribute to mortality risk, warranting further investigation. For risk stratification, we adopted a threshold value from Group 3 (AISI ≥ 1144.8) to categorize the study population into high- and low-risk subgroups. ROC curve analysis indicated that this stratification provided a significant improvement in predicting 28-day mortality (AUC increase: Δ = 0.048, p = 0.012), although it did not offer a statistically significant incremental value for predicting 365-day mortality (Δ = 0.009, p = 0.12) (Supplementary File S5). These findings support the hypothesis that AISI measured at admission may serve as a clinically relevant marker for assessing short-term mortality risk, potentially reflecting the acute inflammatory response during the early phase of AMI. This interpretation is further supported by the results from the time-dependent ROC analysis.

In our study, patients in the high AISI group exhibited increased short-term and long-term mortality rates, aligning with findings from another independent center [15]. The association between elevated AISI levels and higher mortality may reflect the detrimental impact of systemic inflammation on myocardial tissue and vascular integrity. Evidence from animal models supports this notion, showing that systemic inflammation can promote abnormal platelet aggregation and adhesion to endothelial cells, resulting in localized ischemia, hypoxia, microthrombus formation, and ultimately tissue necrosis [17]. Lymphocyte apoptosis, coupled with elevated levels of monocytes and neutrophils, contributes to atherosclerotic plaque rupture and thrombosis, thereby increasing the risk of adverse cardiovascular events by activating the inflammatory response [15, 17]. Michael A. Matter et al. summarized the inflammatory response in AMI into three phases [3]: “Good”: Some cytokines, such as interleukin(IL) −10 and IL-2, mitigate inflammation by suppressing pro-inflammatory mediators like Tumor Necrosis Factor (TNF)-α, Monocyte Chemoattractant Protein (MCP)−1, and IL-8. They also inhibit extracellular matrix remodeling while promoting the activation of regulatory T cells (Tregs), T-helper 2 (Th2) cells, and natural killer (NK) cells. These actions foster M2 macrophage polarization, which supports cardiomyocyte repair and healing; “Bad”: This phase is characterized by persistent low-grade inflammation during the late stages of AMI, involving M1 macrophages, foam cells, and polymorphonuclear neutrophils (PMNs), which adversely affect patient outcomes; “Ugly”: This phase refers to the intense and acute inflammatory response that occurs during the early stage of AMI. After plaque rupture and thrombus formation, PMNs are rapidly activated, and monocytes are recruited to the site of injury. This inflammatory cascade is further amplified by the formation of neutrophil extracellular traps (NETs), which contribute to tissue damage through an oxidative burst. Concurrently, pro-inflammatory cytokines such as IL-1 and IL-6 surge, intensifying the destructive inflammatory response. In our study, we observed a nonlinear association between AISI values and both 28-day and 365-day mortality. This may reflect the complex and dynamic nature of inflammation in AMI. Low AISI values likely indicate a mild inflammatory state that has not reached a clinically damaging threshold. However, once the AISI exceeds the reference value of 686.26, the hazard ratio increases steeply, suggesting that elevated AISI levels may reflect the explosive inflammatory response characteristic of early AMI. Although targeting post-AMI inflammation presents a promising therapeutic strategy, its clinical efficacy remains a subject of ongoing debate. Preclinical studies [18–22] have shown that NLRP3 inflammasome inhibitors, such as MCC950, OLT1177, and Oridonin, can reduce the release of pro-inflammatory factors (e.g., IL-1β and IL-18) and improve AMI outcomes. Additionally, Metformin, Methotrexate, and TAK-242 (a Toll-like Receptor 4(TLR4) inhibitor) can modulate TLR4-dependent monocyte/macrophage-mediated inflammation, suppressing the production of IL-6 and TNF-α, thereby alleviating AMI injury [23–25]. Clinical trials have explored the potential of targeting inflammation in AMI, with mixed outcomes. The CANTOS trial demonstrated that Canakinumab, an interleukin-1β (IL-1β) inhibitor, significantly reduced major cardiovascular events in AMI patients, however, this benefit came at the cost of an increased risk of infections [26]. In the COLCOT trial, early use of colchicine (a drug with broad anti-inflammatory effects, including inhibition of the NLRP3 inflammasome) in AMI patients led to a reduction in cardiovascular events but was associated with a slight increase in pneumonia risk [27]. Notably, COLCOT also revealed that while colchicine showed some benefit, the overall impact of Canakinumab was limited, and Methotrexate, a non-specific anti-inflammatory drug, had no significant effect on cardiovascular outcomes [27]. The LoDoCo2 trial confirmed the anti-inflammatory effects of colchicine in chronic coronary artery disease [28]. Although the therapeutic benefit of inflammation-targeted interventions in AMI remains uncertain, these findings do not negate the pivotal role of inflammation in atherosclerosis. As such, further clinical research is essential to determine whether inflammation represents a viable and effective therapeutic target in this context [29].

Moreover, the observed gender-specific differences offer important insights, suggesting that females may experience heightened inflammatory responses. Sex-based biological factors are known to significantly influence the nature and magnitude of inflammatory reactions [30]. A study by Suarez et al. [31] reported that female patients with peripheral artery disease exhibit greater platelet reactivity, increased thrombus strength, and reduced platelet inhibition in response to antiplatelet therapy, indicating a potential risk of under-treatment in thrombo-prophylaxis among women. Acute systemic inflammation may induce microvascular dysfunction through increased endothelial permeability [32]. Notably, ischemic conditions caused by coronary microvascular dysfunction are more commonly observed in women [33, 34], which may explain why AISI appears to be a more sensitive predictor of adverse cardiovascular outcomes in the female population. Therefore, the sex-specific differences observed in this study can be summarized as follows: “Biological Determinants”: Estrogen plays a protective role in cardiovascular health through several mechanisms. It inhibits the Nuclear Factor-kappa B (NFκB) signaling pathway, thereby suppressing the release of pro-inflammatory cytokines [35, 36]. It also regulates oxidative stress levels, alleviating myocardial inflammation. Additionally, estrogen promotes the synthesis of nitric oxide (NO), improving vascular endothelial function [37]. Estrogen exerts cardioprotective effects through its anti-inflammatory, antioxidant, and endothelial-stabilizing properties, helping to reduce cardiomyocyte apoptosis and necrosis after AMI. However, these protective effects are closely linked to estrogen levels, which decline with age. Premenopausal women, who have higher circulating estrogen levels, are theoretically more resilient to the inflammatory response triggered by AMI, thereby exhibiting lower levels of pro-inflammatory cytokines and reduced myocardial injury. In contrast, postmenopausal women experience a significant drop in estrogen, diminishing these protective effects and leading to poorer outcomes following AMI [37]. In this study, most female patients (295 out of 314, or 93.9%) were over the age of 50, placing them in the perimenopausal or postmenopausal stage, a period marked by declining estrogen levels. This hormonal shift may increase their susceptibility to heightened inflammatory responses during AMI. “Clinical Presentation and Management Differences”: Female patients with AMI more frequently present with spontaneous coronary artery dissection (SCAD) or coronary microvascular dysfunction (CMD)—pathophysiological processes primarily characterized by endothelial dysfunction and diffuse inflammation, which demonstrate close alignment with the AISI measurements. This study included 655 male and 314 female patients. While coronary angiography and PCI rates were comparable between the two groups, analysis of pharmacological treatments (limited to aspirin and metoprolol due to data availability) revealed significantly higher utilization rates of both medications among male patients (Supplementary file S10). Thus, the potential influence of sex-based differences in clinical practice—particularly in the adjustment of antiplatelet agents and cardioprotective drug dosing—on the relationship between inflammatory responses and clinical outcomes cannot be excluded.

Based on the high sensitivity and significant predictive value of AISI in female AMI patients observed in this study, the following potential clinical suggestions are proposed: “AISI Monitoring”: Measure Evaluate AISI levels at admission for early risk stratification and perform regular monitoring throughout the treatment course to track inflammatory trends. This can aid in guiding timely therapeutic adjustments. “Focus on Sex-Specific Management”:Accumulating evidence suggests that women with arterial disease demonstrate enhanced platelet reactivity, increased thrombus strength, and a blunted response to antiplatelet agents. Furthermore, coronary microvascular dysfunction is more prevalent in female patients. Notably, perimenopausal and postmenopausal women exhibit declining estrogen levels, which may amplify their susceptibility to inflammatory activation during acute myocardial infarction. Taken together, these observations support the consideration of tailored therapeutic strategies for women with AMI, which may include intensified antithrombotic regimens, interventions aimed at improving microvascular function, and exploratory use of estrogen replacement to attenuate inflammation. However, the efficacy and safety of such sex-specific approaches await confirmation in future large-scale multicenter trials. “Multidisciplinary Collaboration”: Promote collaboration among cardiology, rehabilitation medicine, clinical psychology, nutrition, and gynecology to deliver holistic care. Comprehensive rehabilitation strategies should include personalized exercise programs [38],psychological support [39], anti-inflammatory dietary recommendations (e.g., the Mediterranean diet [40]), and evaluating the effects of sex hormone levels and menstrual cycles on treatment [41]. “Long-term Management”: Overall, incorporating AISI into post-discharge follow-up protocols may help monitor chronic inflammation and assess the long-term risk of cardiovascular events. However, we acknowledge that the single-center, retrospective design of this study limits the generalizability of our findings. To enhance broader applicability, validation through larger multicenter studies involving diverse populations is necessary. Additionally, the potential for AISI-guided therapeutic approaches—especially in female AMI patients—warrants further investigation.

Limitations

This study is entirely based on the MIMIC-IV database, and diagnoses were identified through coding rather than clinical adjudication (data distribution plots are provided in Supplementary File S8), which limits the generalizability of the findings to broader and more diverse populations. Although we adjusted for comorbidities and demographic variables and addressed potential bias related to aspirin use (Supplementary File S9), other potential confounders—such as socioeconomic status, lifestyle factors, and genetic predispositions—were not accounted for. In addition, data on specific medications (e.g., corticosteroids, other nonsteroidal anti-inflammatory drugs, and immunosuppressants) were not comprehensively recorded, which may affect the robustness of the results. Furthermore, this study lacked a direct comparison with existing inflammatory markers and lacked GRACE and TIMI scores. Given that these markers and scores represent the current routine clinical standards for risk stratification in AMI, this limitation may affect the immediate translatability of our findings into clinical practice. Future studies should validate whether AISI provides incremental predictive value beyond these conventional tools. Finally, although this study highlights the potential clinical utility of AISI in inflammation monitoring and risk stratification, due to insufficient supporting evidence, we are currently unable to provide specific recommendations for clinical intervention. Further research is needed to explore this aspect in greater detail.

Conclusion

In this study, AISI shows a nonlinear association with 28-day and 365-day mortality in AMI patients. Elevated AISI levels are significantly associated with increased mortality risk, underscoring the potential of AISI as a novel prognostic biomarker for both short- and long-term mortality. Female AMI patients are more sensitive to elevated AISI levels, showing a stronger link between high AISI and mortality. This highlights its clinical importance in enhancing risk stratification and patient management.

Abbreviations

AIC

Akaike Information Criterion

AISI

Aggregate Index of Systemic Inflammation

ALT

Alanine Aminotransferase

AMI

Acute myocardial infarction

ANOVA

Analysis of Variance

AST

Aspartate Aminotransferase

AUC

Area Under the Curve

BIC

Bayesian Information Criterion

CAD

Coronary artery disease

CBC

Complete blood cell count

CMD

Coronary microvascular dysfunction

HDL-c

High-Density Lipoprotein Cholesterol

ICD

International Classification of Diseases

IL

Interleukin

IQR

Interquartile Range

LDL-c

Low-Density Lipoprotein Cholesterol

LMR

Lymphocyte-to-monocyte ratio

LRT

Likelihood ratio test

MIMIC

Medical Information Mart for Intensive Care

NETs

Neutrophil extracellular traps

NFκB

Nuclear Factor-kappa B

NLR

Neutrophil-lymphocyte ratio

NK

Natural killer

NO

Nitric oxide

NRI

net reclassification improvement

NSTEMI

Non-ST-segment elevation MI

PaCO2

Partial Pressure of Carbon Dioxide

PaO2

Partial Pressure of Oxygen

PCI

Percutaneous coronary intervention

PIV

Pan-immune-inflammatory value

PLR

Platelet-lymphocyte ratio

PMNs

Polymorphonuclear neutrophils

RBC

Red blood cell

RCS

Restricted cubic spline

ROC

Receiver Operating Characteristic

SCAD

Spontaneous coronary artery dissection

SD

Standard deviation

SII

Systemic immune-inflammation index

SIRI

Systemic inflammation response index

SQL

Structured Query Language

STEMI

ST-segment elevation MI

T2DM

Type 2 diabetes mellitus

TG

Triglycerides

Th2

T-helper 2

TLR4

Toll-like Receptor 4

TNF

Tumor Necrosis Factor

VIFs

Variance Inflation Factors

Authors’ contributions

Kangzheng Yuan, Fangmei Liu, Jian Wang and Qingchi Liao contributed to the conception and design of the study. Yu Huang, Qing Ye and Yang Xiao performed the data extraction from the MIMIC-IV database. Xinjie Sun, Min Deng and Lei Wen carried out the statistical analyses and interpreted the results. Kangzheng Yuan, Fangmei Liu and Jian Wang were responsible for drafting the initial manuscript. Qingchi Liao critically revised the manuscript for important intellectual content. All authors contributed to the review and approval of the final manuscript. Kangzheng Yuan supervised the entire project and ensured the study’s methodological rigor.

Data availability

The datasets generated and/or analyzed during the current study are available in the MIMIC-IV database [42].

Declarations

Ethics approval and consent to participate

This study was conducted using data from the publicly available MIMIC-IV database. The data were de-identified, and the Beth Israel Deaconess Medical Center’s ethical committee waived the requirement for informed consent. The author Kangzheng Yuan successfully completed the “Protecting Human Research Participants” course offered by the National Institutes of Health (Record ID: 60220095) to access the database.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.