Abstract
Background
Coronary heart disease (CHD) is a major cause of mortality in critically ill patients, with endothelial dysfunction playing a pivotal role in disease progression. The Endothelial Activation and Stress Index (EASIX), a composite biomarker reflecting endothelial injury and systemic stress, has demonstrated prognostic value across various cardiovascular conditions. Nevertheless, the association of this phenomenon with mortality in CHD patients requiring intensive care remains to be elucidated. This study aims to investigate the association between EASIX and mortality in patients with CHD at intensive care unit (ICU), and to evaluate its potential as a prognostic biomarker for risk stratification.
Methods
A retrospective cohort study was conducted using the Medical Information Mart for Intensive Care IV (MIMIC-IV) database. The study encompassed a total of 2,468 critically ill CHD patients, who were stratified into tertiles based on admission EASIX scores. The application of Cox proportional hazard models and restricted cubic spline regression was utilised for the purpose of evaluating the association between EASIX and all-cause mortality within a 30-, 90- and 365-day timeframe. Subgroup analyses were performed in order to assess potential effect modifications.
Results
The median age of the cohort was found to be 73 years (interquartile range, 64–81 years), with 66.13% of subjects being male (1,632/2,468). Patients in the highest EASIX tertile exhibited significantly elevated 30-day mortality (34.87% vs. 10.45%, P<0.001) and 365-day mortality (38.88% vs. 12.15%, P<0.001) compared to the lowest tertile. Cox regression analysis revealed EASIX to be an independent predictor of 30-day mortality [adjusted hazard ratio (HR): 2.237; 95% confidence interval (CI): 1.534–3.262; P<0.001] and 365-day mortality (adjusted HR: 2.204; 95% CI: 1.553–3.129; P<0.001) following comprehensive adjustment for confounders. The Kaplan-Meier analysis demonstrated a significantly inferior survival probability in the highest EASIX stratum (log-rank P<0.0001). The restricted cubic spline analysis indicated a near-linear dose-response relationship between EASIX and mortality risk (P for non-linearity >0.05).
Conclusions
Elevated EASIX scores independently correlated with increased short- and long-term mortality in critically ill CHD patients, suggesting its utility as a novel prognostic biomarker for risk stratification in this high-risk population. Further prospective validation and investigation of therapeutic implications are warranted.
Keywords: Coronary heart disease (CHD), intensive care unit (ICU), Endothelial Activation and Stress Index (EASIX), mortality, predictor
Highlight box.
Key findings
• In 2,468 critically ill coronary heart disease (CHD) patients, Endothelial Activation and Stress Index (EASIX) independently predicted 30-day [adjusted hazard ratio (HR): 2.237, 95% confidence interval (CI): 1.534–3.262] and 365-day mortality (adjusted HR: 2.204, 95% CI: 1.553–3.129). Patients in the highest EASIX tertile showed 34.87% 30-day mortality vs. 10.45% in the lowest tertile. A linear dose-response relationship was confirmed.
What is known and what is new?
• EASIX has demonstrated prognostic value in hematopoietic stem cell transplantation, hypertensive emergencies and sepsis.
• This study is the first to validate EASIX as an independent mortality predictor specifically in critically ill CHD patients, ranking it third in importance after Acute Physiology Score III and Simplified Acute Physiological Score II by Boruta algorithm.
What is the implication, and what should change now?
• EASIX, calculated from three routine laboratory tests (lactate dehydrogenase, creatinine, platelets), offers a simple, cost-effective tool for early risk stratification in intensive care unit CHD patients. Clinicians should consider incorporating EASIX into routine assessment to identify ultra-high-risk patients within 24 hours of admission and guide intensive monitoring and intervention strategies.
Introduction
Coronary heart disease (CHD) constitutes the predominant cause of cardiovascular morbidity and mortality globally, accounting for approximately 17.8 million deaths annually (1). The pathophysiology of CHD involves complex interactions among atherosclerotic plaque formation, myocardial ischemia, and thrombotic complications, with endothelial dysfunction serving as a critical initiating and perpetuating factor (2). Critically ill CHD patients, particularly those presenting with acute coronary syndromes (ACS) or post-cardiac arrest states, demonstrate exacerbated endothelial injury and systemic inflammatory responses, which significantly influence clinical outcomes (3).
The Endothelial Activation and Stress Index (EASIX) is a clinically accessible composite score calculated from lactate dehydrogenase (LDH), creatinine, and platelet counts, originally developed to prognosticate outcomes in hematopoietic stem cell transplantation (4). Recent evidence has expanded its application to diverse conditions characterized by endothelial perturbation, including coronavirus disease 2019 (COVID-19), sepsis, atrial fibrillation and hypertensive emergencies (5-8).
Endothelial dysfunction in CHD manifests as impaired nitric oxide bioavailability, increased oxidative stress, and upregulated expression of adhesion molecules, collectively promoting plaque vulnerability and adverse cardiac events (9). Inflammatory cascades and oxidative damage create a vicious cycle that accelerates coronary artery disease progression (10). Despite accumulating data supporting EASIX as a mortality predictor in heterogeneous intensive care unit (ICU) populations, no prior investigations have specifically examined its prognostic significance in CHD patients with critical illness. We hypothesized that higher EASIX scores at ICU admission would independently predict increased mortality in this cohort. Therefore, the present study aims to evaluate the prognostic value of EASIX for predicting 30-, 90-, and 365-day mortality in critically ill CHD patients admitted to the ICU, and to explore its potential as a simple and cost-effective tool for early risk stratification. We present this article in accordance with the STROBE reporting checklist (available at https://jtd.amegroups.com/article/view/10.21037/jtd-2026-1-0099/rc).
Methods
Study population
This retrospective cohort study utilized the Medical Information Mart for Intensive Care IV (MIMIC-IV) database (version 3.0), which is publicly available and has been approved by the Institutional Review Boards (IRBs) of the Massachusetts Institute of Technology (MIT) and Beth Israel Deaconess Medical Center (BIDMC) (Protocol No. 2001-P-001699) (11). Given that all data in the MIMIC-IV database are fully de-identified and contain no personal identifiable information (PII) of patients or healthcare providers, the IRBs of both institutions waived the requirement for written informed consent from individual participants. One author (S.D.) completed the required human subject research training (record ID: 14012091). All research procedures were conducted in accordance with the Declaration of Helsinki and its subsequent amendments, and the relevant ethical guidelines for observational studies using de-identified clinical data.
We identified adult patients (≥18 years) admitted to the ICU between 2008–2022 with a primary diagnosis of CHD according to International Classification of Diseases (ICD)-9/10 codes. Exclusion criteria comprised: (I) patients under 18 years old at the time of first admission; (II) multiple ICU admissions (only first admission retained); (III) missing data on LDH, creatinine or platelet within 24 hours of admission; (IV) outlier data on LDH, creatinine or platelet. Outlier data were defined based on data quantiles and the interquartile range (IQR), data points beyond Q1−1.5IQR or Q3+1.5IQR were considered outliers. Missing data were present, however, for every variable the proportion of missing values did not exceed 5%. We retained the missing values without imputation. The final cohort included 2,468 patients stratified into EASIX tertiles (Figure 1).
Figure 1.
Flowchart of the selection of patients. ICU, intensive care unit; LDH, lactate dehydrogenase; MIMIC-IV, Medical Information Mart for Intensive Care IV.
Data collection
Data extraction employed PostgreSQL (version 13.7.2) and Navicat Premium (version 16). Variables encompassed: (I) demographics: age, sex, race, body mass index (BMI); (II) comorbidities: respiratory failure, atrial fibrillation, hypertension, acute kidney injury (AKI), stroke, hypertension, type 2 diabetes mellitus (T2DM), heart failure, myocardial infarction; (III) laboratory parameters: hemoglobin, red blood cells (RBCs), white blood cells (WBCs), international normalized ratio (INR), prothrombin time (PT), activated partial thromboplastin time (APTT), fasting blood glucose (FBG), potassium, sodium, serum creatinine, LDH, and platelet; (IV) severity scores: Sequential Organ Failure Assessment (SOFA), Simplified Acute Physiological Score II (SAPS II), Acute Physiology Score III (APS III), Oxford Acute Severity of Illness Score (OASIS); and (V) therapeutic: invasive ventilation, antihypertensive drug, glucocorticoid. EASIX was calculated as: [LDH (U/L) × creatinine (mg/dL)] / platelet (109/L) using admission values within 24 hours (4). This composite index encapsulates key pathophysiological processes: LDH reflects cellular death and tissue hypoperfusion, creatinine mirrors renal microcirculatory integrity, and platelet count indicates consumptive coagulopathy and endothelial interaction (12). The primary outcomes were 30- and 365-day all-cause mortality, and the secondary outcome was 90-day all-cause mortality.
Statistical analysis
Continuous variables were expressed as mean ± standard deviation (SD) or median (IQR) based on normality (Kolmogorov-Smirnov test). Categorical variables were presented as frequencies (percentages). Given EASIX’s skewed distribution, log2 transformation was applied before analysis (4). Inter-group comparisons utilized analysis of variance (ANOVA), Kruskal-Wallis, or χ2 tests as appropriate. Survival analysis employed Kaplan-Meier curves with log-rank tests. Cox proportional hazards models calculated hazard ratios (HRs) and 95% confidence intervals (CIs) across three hierarchical models: Model 1 (unadjusted); Model 2 (demographics); Model 3 (adjust all parameters). The proportional hazards assumption was verified using Schoenfeld residuals. Restricted cubic spline regression with four knots assessed non-linear relationships. Subgroup analyses examined effect modification by age (<65 vs. ≥65 years), sex, BMI (<25 vs. ≥25 kg/m2), atrial fibrillation, stroke, T2DM and heart failure status. The Boruta algorithm identified variable importance for mortality prediction. All statistical analyses were performed using R version 4.3.2 and DecisionLinnc 1.1 software, with a P value of less than 0.05 being considered statistically significant. DecisionLinnc 1.1 is a data analysis platform integrating multiple programming languages and providing a visual interface for processing data and performing analyses (13).
Results
Baseline characteristics
The median age was 73.0 (IQR, 64.0–81.0) years, with 1,632 (66.1%) male patients. The overall 30-, 90-, and 365-day mortality rates were 20.99%, 22.65% and 23.99%, respectively. Patients in the highest EASIX tertile (T3) were older, predominantly male, and exhibited higher prevalence of respiratory failure, atrial fibrillation, AKI and T2DM (Table 1). Laboratory indices revealed significant trends across tertiles: platelet counts decreased (T1: 224.5×109/L vs. T3: 133.0×109/L, P<0.001), while LDH (T1: 215.0 U/L vs. T3: 435.0 U/L, P<0.001) and creatinine (T1: 0.9 mg/dL vs. T3: 1.9 mg/dL, P<0.001) progressively increased. Severity scores demonstrated analogous deterioration (SOFA: T1 =4.0 vs. T3 =8.0, P<0.001).
Table 1. Baseline characteristics stratified by EASIX tertiles.
| Characteristics | Overall (n=2,468) | T1 (n=823) | T2 (n=822) | T3 (n=823) | P value |
|---|---|---|---|---|---|
| Age (years) | 73.0 (64.0–81.0) | 72.0 (62.0–81.0) | 74.0 (64.0–82.0) | 74.0 (66.0–82.0) | 0.001 |
| Sex | |||||
| Female | 836 (33.9) | 344 (41.8) | 249 (30.3) | 243 (29.5) | <0.001 |
| Male | 1,632 (66.1) | 479 (58.2) | 573 (69.7) | 580 (70.5) | |
| Race | |||||
| Black | 154 (6.2) | 42 (5.1) | 56 (6.8) | 56 (6.8) | 0.58 |
| White | 2,039 (82.6) | 691 (84.0) | 674 (82.0) | 674 (81.9) | |
| Others | 275 (11.1) | 90 (10.9) | 92 (11.2) | 93 (11.3) | |
| BMI (kg/m2) | 27.7 (24.3–31.8) | 27.0(24.0–31.2) | 28.0 (24.3–32.3) | 28.1 (24.6–32.2) | 0.02 |
| Comorbidities | |||||
| Respiratory failure | 1,137 (46.1) | 264 (32.1) | 388 (47.2) | 485 (58.9) | <0.001 |
| Atrial fibrillation | 1,090 (44.2) | 311 (37.8) | 372 (45.3) | 407 (49.5) | <0.001 |
| Hypertension | 900 (36.5) | 414 (50.3) | 276 (33.6) | 210 (25.5) | <0.001 |
| AKI | 1,244 (50.4) | 182 (22.1) | 436 (53.0) | 626 (76.1) | <0.001 |
| Stroke | 249 (10.1) | 75 (9.1) | 91 (11.1) | 83 (10.1) | 0.42 |
| T2DM | 963 (39.0) | 251 (30.5) | 335 (40.8) | 377 (45.8) | <0.001 |
| Heart failure | 1,310 (53.1) | 323 (39.3) | 473 (57.5) | 514 (62.5) | <0.001 |
| Myocardial infarction | 856 (34.7) | 211 (25.6) | 300 (36.5) | 345 (41.9) | <0.001 |
| Illness severity scores | |||||
| SOFA | 6.0 (3.0–9.0) | 4.0 (2.0–6.0) | 5.0 (3.0–8.0) | 8.0 (6.0–11.0) | <0.001 |
| APS III | 47.0 (35.0–62.0) | 38.0 (29.0–49.0) | 46.0 (36.0–58.0) | 58.0 (47.0–74.0) | <0.001 |
| SAPS II | 40.0 (32.0–50.0) | 35.0 (28.0–43.0) | 39.0 (32.0–48.0) | 46.0 (38.0–57.0) | <0.001 |
| OASIS | 34.0 (28.0–40.0) | 32.0 (26.0–38.0) | 33.0 (28.0–39.0) | 36.0 (30.0–43.0) | <0.001 |
| Laboratory tests | |||||
| Hemoglobin (g/dL) | 10.4 (9.0–12.0) | 10.7 (9.2–12.2) | 10.5 (9.0–12.0) | 10.1 (8.7–11.7) | <0.001 |
| RBC (×106/µL) | 3.5 (3.0–4.0) | 3.6 (3.2–4.1) | 3.5 (3.0–4.0) | 3.4 (2.9–3.9) | <0.001 |
| WBC (×103/µL) | 11.4 (8.5–15.6) | 11.1 (8.4–14.5) | 11.4 (8.6–15.6) | 11.8 (8.4–16.8) | <0.001 |
| INR | 1.3 (1.2–1.6) | 1.2 (1.1–1.4) | 1.3 (1.2–1.5) | 1.4 (1.2–1.9) | <0.001 |
| PT (sec) | 14.2 (12.8–17.1) | 13.5 (12.4–15.3) | 14.2 (12.8–16.5) | 15.6 (13.5–20.1) | <0.001 |
| APTT (sec) | 36.0 (29.3–54.8) | 33.5 (28.3–49.1) | 36.0 (29.2–53.6) | 39.2 (30.9–62.5) | <0.001 |
| FBG (mg/dL) | 135.3 (112.0–176.0) | 125.0 (107.0–151.0) | 138.0 (113.5–177.9) | 147.4 (118.1–196.4) | <0.001 |
| Potassium (mEq/L) | 4.2 (3.9–4.5) | 4.1 (3.8–4.4) | 4.2 (3.9–4.5) | 4.3 (4.0–4.7) | <0.001 |
| Sodium (mEq/L) | 138.5 (135.7–141.0) | 138.5 (136.0–140.5) | 138.5 (136.0–141.0) | 138.3 (135.0–141.2) | 0.98 |
| Serum creatinine (mg/dL) | 1.2 (0.9–1.8) | 0.9 (0.7–1.0) | 1.2 (1.0–1.6) | 1.9 (1.3–2.7) | <0.001 |
| EASIX | 1.13 (0.21–2.09) | –0.12 (–0.57 to 0.21) | 1.13 (0.84–1.39) | 2.49 (2.09–3.14) | <0.001 |
| LDH (U/L) | 296.0 (219.0–435.0) | 215.0 (177.0–272.0) | 300.8 (239.1–407.8) | 435.0 (313.5–618.3) | <0.001 |
| Platelet (×109/L) | 180.0 (133.0–237.0) | 224.5 (174.8–287.8) | 180.7 (147.0–229.5) | 133.0 (93.9–182.3) | <0.001 |
| Therapeutic | |||||
| Invasive ventilation | 2,212 (89.6) | 722 (87.7) | 739 (89.9) | 751 (91.3) | 0.06 |
| Antihypertensive drug | 2,107 (85.4) | 692 (84.1) | 712 (86.6) | 703 (85.4) | 0.35 |
| Glucocorticoid | 647 (26.2) | 161 (19.6) | 217 (26.4) | 269 (32.7) | <0.001 |
| Events | |||||
| 30-day mortality | 518 (20.99) | 86 (10.45) | 145 (17.64) | 287 (34.87) | <0.001 |
| 90-day mortality | 559 (22.65) | 96 (11.66) | 161 (19.59) | 302 (36.70) | <0.001 |
| 365-day mortality | 592 (23.99) | 100 (12.15) | 172 (20.92) | 320 (38.88) | <0.001 |
Data are presented as median (IQR) or n (%). AKI, acute kidney injury; APS III, Acute Physiology Score III; APTT, activated partial thromboplastin time; BMI, body mass index; EASIX, Endothelial Activation and Stress Index; FBG, fasting blood glucose; INR, international normalized ratio; IQR, interquartile range; LDH, lactate dehydrogenase; OASIS, Oxford Acute Severity of Illness Score; PT, prothrombin time; RBC, red blood cell; SAPS II, Simplified Acute Physiological Score II; SOFA, Sequential Organ Failure Assessment; T, tertile; T2DM, type 2 diabetes mellitus; WBC, white blood cell.
EASIX and mortality associations
Non-survivors exhibited significantly higher EASIX scores compared to survivors [1.96 (0.96–2.84) vs. 0.93 (0.09–1.83), P<0.001] based on 30-day outcomes (Table S1). Mortality analyses (Tables S2,S3) revealed that non-survivors had consistently higher EASIX values than survivors at 90- and 365-day outcomes. The Boruta algorithm identified EASIX as a “Confirmed” predictor of 30-day mortality, with importance Z-score ranking third after APSIII and SAPS II (Figure 2).
Figure 2.
Variable importance ranking for 30-day mortality by Boruta algorithm. AKI, acute kidney injury; APS III, Acute Physiology Score; EASIX, Endothelial Activation and Stress Index; FBG, fasting blood glucose; INR, international normalized ratio; LDH, lactate dehydrogenase; OASIS, Oxford Acute Severity of Illness Score; PT, prothrombin time; RBC, red blood cell; SAPS II, Simplified Acute Physiological Score II; SOFA, Sequential Organ Failure Assessment; WBC, white blood cell.
Kaplan-Meier analysis revealed significant mortality gradients across EASIX tertiles for both 30- and 365-day endpoints (log-rank P<0.001) (Figure 3). Patients in the highest EASIX tertile exhibited significantly elevated 30-day mortality (34.87% vs. 10.45%, P<0.001) and 365-day mortality (38.88% vs. 12.15%, P<0.001) compared to the lowest tertile. Similar findings were observed for 90-day mortality (Figure S1).
Figure 3.
Kaplan-Meier survival curves comparing EASIX tertiles. (A) Death within 30 days after admission. (B) Death within 365 days after admission. EASIX, Endothelial Activation and Stress Index.
Cox regression demonstrated consistent associations across all models (Table 2). In the fully-adjusted Model 3, each log2-unit increase in EASIX conferred a 52.2% higher mortality risk at 30days (HR: 1.522; 95% CI: 1.346–1.72; P<0.001) and 47.4% higher mortality risk at 365 days (HR: 1.474; 95% CI: 1.315–1.653; P<0.001). Compared to T1, T3 patients exhibited a 2.237-fold elevated mortality risk at 30 days (HR: 2.237; 95% CI: 1.534–3.262; P<0.001) and 2.204-fold elevated mortality risk at 365 days (HR: 2.204; 95% CI: 1.553–3.129; P<0.001) after multivariable adjustment. These patterns remained consistent at the intermediate 90-day time point, confirming the robust prognostic value of EASIX.
Table 2. Cox proportional hazards analysis for short- and long-term mortality.
| Mortality | Model 1 | Model 2 | Model 3 | |||||
|---|---|---|---|---|---|---|---|---|
| HR (95% CI) | P value | HR (95% CI) | P value | HR (95% CI) | P value | |||
| 30-day mortality | ||||||||
| Continuity | 1.523 (1.436–1.614) | <0.001 | 1.54 (1.451–1.635) | <0.001 | 1.522 (1.346–1.72) | <0.001 | ||
| T1 (n=823) | 1.00 | 1.00 | 1.00 | |||||
| T2 (n=822) | 1.778 (1.362–2.322) | <0.001 | 1.786 (1.365–2.337) | <0.001 | 1.426 (1.043–1.952) | 0.03 | ||
| T3 (n=823) | 3.911 (3.073–4.977) | <0.001 | 3.95 (3.095–5.04) | <0.001 | 2.237 (1.534–3.262) | <0.001 | ||
| P for trend | <0.001 | <0.001 | <0.001 | |||||
| 90-day mortality | ||||||||
| Continuity | 1.503 (1.420–1.591) | <0.001 | 1.519 (1.434–1.61) | <0.001 | 1.499 (1.333–1.686) | <0.001 | ||
| T1 (n=823) | 1.00 | 1.00 | 1.00 | |||||
| T2 (n=822) | 1.774 (1.378–2.284) | <0.001 | 1.79 (1.387–2.308) | <0.001 | 1.426 (1.059–1.919) | 0.02 | ||
| T3 (n=823) | 3.732 (2.966–4.696) | <0.001 | 3.782 (2.998–4.772) | <0.001 | 2.162 (1.509–3.098) | <0.001 | ||
| P for trend | <0.001 | <0.001 | <0.001 | |||||
| 365-day mortality | ||||||||
| Continuity | 1.512 (1.431–1.597) | <0.001 | 1.526 (1.443–1.614) | <0.001 | 1.474 (1.315–1.653) | <0.001 | ||
| T1 (n=823) | 1.00 | 1.00 | 1.00 | |||||
| T2 (n=822) | 1.824 (1.426–2.334) | <0.001 | 1.844 (1.439–2.364) | <0.001 | 1.465 (1.097–1.958) | 0.01 | ||
| T3 (n=823) | 3.836 (3.064–4.802) | <0.001 | 3.891 (3.101–4.883) | <0.001 | 2.204 (1.553–3.129) | <0.001 | ||
| P for trend | <0.001 | <0.001 | <0.001 | |||||
Model 1: unadjusted crude model. Model 2: adjusted for sex, age, race, BMI. Model 3: adjusted for factors in Model 2 and respiratory failure, atrial fibrillation, hypertension, AKI, stroke, hypertension, T2DM, heart failure, myocardial infarction, hemoglobin, RBC, WBC, PT, APTT, FBG, potassium, sodium, serum creatinine, platelet, SOFA, SAPS II, APS III, OASIS, invasive ventilation, antihypertensive drug, glucocorticoid. AKI, acute kidney injury; APS III, Acute Physiology Score III; APTT, activated partial thromboplastin time; BMI, body mass index; CI, confidence interval; FBG, fasting blood glucose; HR, hazard ratio; OASIS, Oxford Acute Severity of Illness Score; PT, prothrombin time; RBC, red blood cell; SAPS II, Simplified Acute Physiological Score II; SOFA, Sequential Organ Failure Assessment; T, tertile; T2DM, type 2 diabetes mellitus; WBC, white blood cell.
Dose-response relationship
Using Cox Model 3 (adjusted for sex, age, race, BMI, respiratory failure, atrial fibrillation, hypertension, AKI, stroke, hypertension, T2DM, heart failure, myocardial infarction, hemoglobin, RBC, WBC, PT, APTT, FBG, potassium, sodium, serum creatinine, platelet, SOFA, SAPS II, APS III, OASIS, invasive ventilation, antihypertensive drug, glucocorticoid), RCS regression confirmed a linear dose-response relationship between EASIX and 30-day as well as 365-day all-cause mortality (P for non-linear >0.05) (Figure 4). A consistent pattern was observed at the intermediate 90-day time point (Figure S2).
Figure 4.
Restricted cubic spline analysis of EASIX and 30- and 365-day all-cause mortality in critically ill CHD based on Cox proportional hazards Model 3. (A) 30-day all-cause mortality. (B) 365-day all-cause mortality. Cox proportional hazards model 3: adjusted for sex, age, race, BMI, respiratory failure, atrial fibrillation, hypertension, AKI, stroke, hypertension, T2DM, heart failure, myocardial infarction, hemoglobin, RBC, WBC, PT, APTT, FBG, potassium, sodium, serum creatinine, platelet, SOFA, SAPS II, APS III, OASIS, invasive ventilation, antihypertensive drug, glucocorticoid. AKI, acute kidney injury; APS III, Acute Physiology Score III; APTT, activated partial thromboplastin time; BMI, body mass index; CHD, coronary heart disease; CI, confidence interval; EASIX, Endothelial Activation and Stress Index; FBG, fasting blood glucose; OASIS, Oxford Acute Severity of Illness Score; PT, prothrombin time; RBC, red blood cell; SAPS II, Simplified Acute Physiological Score II; SOFA, Sequential Organ Failure Assessment; T2DM, type 2 diabetes mellitus; T2DM, type 2 diabetes mellitus; WBC, white blood cell.
Subgroup analyses
We performed stratified analyses to examine the consistency between EASIX and 30- and 365-day mortality association across key clinical subgroups defined by age (<65 vs. ≥65 years), sex (female vs. male), BMI (<25 vs. ≥25 kg/m2), atrial fibrillation, stroke, T2DM and heart failure. Using the fully adjusted model 3, we assessed these relationships at 30 and 365 days post-CHD, with T1 serving as the reference category (Figure 5). P for interaction values was computed to assess effect modification across strata. Overall, a consistent dose-dependent relationship was observed, with progressively elevated mortality risk from T2 to T3 across most subgroups. Notably, BMI demonstrated a statistically significant interaction with EASIX for both 30-day (P=0.03) and 365-day mortality (P=0.04), whereas no significant interactions were detected for other variables (all P>0.05), suggesting the prognostic utility of EASIX is largely independent of these clinical characteristics, except for potential effect modification by BMI status. Similar patterns were observed for 90-day mortality (Figure S3), confirming the robust prognostic value of EASIX across all time points.
Figure 5.
Subgroup analysis forest plots for 30- and 365-day mortality. (A) 30-day all-cause mortality; (B) 365-day all-cause mortality. BMI, body mass index; CI, confidence interval; HR, hazard ratio; Ref, reference; T, tertile; T2DM, type 2 diabetes mellitus.
Discussion
This study systematically evaluated the predictive value of the EASIX for short-term and long-term outcomes in critically ill patients with CHD admitted to ICU for the first time, based on an analysis of 2,468 patients from the MIMIC-IV database. The main findings showed that an EASIX calculated within 24 hours of admission was an independent predictor of all-cause mortality at 30, 90, and 365 days. Patients in the highest tertile had an approximately 2.2-fold higher mortality risk than those in the lowest tertile, and this association remained robust after full adjustment for multidimensional confounding factors. Further analysis using restricted cubic splines revealed an approximately linear dose-response relationship between EASIX and mortality. The Boruta algorithm identified EASIX as the third most important variable for predicting 30-day mortality, ranking behind only the APS III and the SAPS II. These results support the use of EASIX as a composite biomarker reflecting endothelial injury and systemic stress status, with potential clinical utility in the risk stratification of critically ill CHD patients.
EASIX was initially developed by for predicting transplant-associated thrombotic microangiopathy and mortality in patients undergoing hematopoietic stem cell transplantation (HSCT) (4). Recent studies have extended its application to disease states centered on endothelial dysfunction, such as COVID-19 (7), sepsis (14), atrial fibrillation (6), ischemic stroke (15) and hypertensive emergency (8). EASIX assesses the severity of endothelial dysfunction by integrating indicators reflecting renal function (creatinine), cellular damage (LDH), and coagulation/inflammatory status (platelets). Endothelial activation is a core mechanism in many pathological processes, such as transplant complications, infection, and immunotherapy toxicity. Elevated EASIX levels indicate endothelial barrier disruption and an increased risk of microthrombus formation (16-18). The present study is the first to apply EASIX to the population of CHD patients in ICUs, and the results are consistent with those of the aforementioned studies (19-23). Notably, although the pathophysiological mechanisms of CHD involve multiple processes including atherosclerotic plaque formation, myocardial ischemia, and thrombotic complications, endothelial dysfunction remains a key link in initiating and sustaining disease progression (24). In CHD, progressive endothelial cell stress may promote plaque instability, microvascular obstruction, and amplification of systemic inflammatory responses, ultimately leading to multiple organ failure (25). The vicious cycle formed by the inflammatory cascade and oxidative damage can accelerate the progression of coronary artery lesions (10). By integrating three indicators—LDH, creatinine, and platelet count—EASIX is precisely able to capture the core elements of this pathological process: elevated LDH indicates cell death and tissue hypoperfusion, serum creatinine levels reflect the integrity of renal microcirculation, and decreased platelet count suggests consumptive coagulopathy and activation of endothelial-platelet interactions (16-18,26).
In the present study, patients in the highest EASIX tertile exhibited a typical “endothelial injury phenotype”: platelet count was 40.7% lower than that in the lowest tertile (133×109/L vs. 224.5×109/L), LDH level was 102.3% higher (435 vs. 215 U/L), creatinine level doubled (1.9 vs. 0.85 mg/dL), accompanied by significant deterioration in SOFA score, and the prevalence of AKI and respiratory failure. This pattern of synchronous multisystem dysregulation is highly consistent with the pathophysiological characteristics of systemic endothelial dysfunction in critically ill CHD patients (27). The Boruta algorithm ranked EASIX as more important than traditional laboratory indicators (e.g., creatinine, WBC) and age, suggesting that as a composite index, it may capture systemic pathological information that cannot be reflected by individual variables. Compared with traditional scoring systems, EASIX only requires three routine laboratory tests for calculation, offering convenient clinical access and significant cost-effectiveness, which is particularly important in resource-limited medical settings.
The findings of this study offer multiple implications for the clinical management of critically ill CHD patients. Firstly, EASIX can serve as an early risk identification tool. The 30-day mortality rate of patients in the highest tertile (EASIX >2.089) reached 34.87%, compared with only 10.45% in the lowest tertile. This significant difference suggests that clinicians can identify ultra-high-risk subgroups within 24 hours of ICU admission, thereby triggering more aggressive monitoring and intervention strategies, such as enhanced hemodynamic support, optimization of oxygen supply-demand balance, and early initiation of renal replacement therapy. Secondly, the existence of a linear dose-response relationship indicates that EASIX can not only be used for binary risk stratification but also as a continuous variable for dynamic risk assessment, providing a quantitative basis for adjusting the intensity of individualized treatment. Thirdly, subgroup analysis revealed that the prognostic value of EASIX was more pronounced in patients with BMI <25 kg/m2. This finding may be related to insufficient nutritional reserves and more severe inflammatory responses in underweight patients, suggesting that EASIX has stronger discriminative power in vulnerable populations and facilitates the realization of precision medicine.
From the perspective of research tools, EASIX can serve as a valid covariate for patient stratification in clinical trials. In randomized controlled trials (RCTs) evaluating novel anti-inflammatory, antithrombotic, or endothelial protective therapies, incorporating EASIX as a stratification factor can balance baseline risks between groups and enhance statistical power. Furthermore, given its core attribute of reflecting endothelial function, EASIX may potentially be used to monitor the efficacy of targeted therapies for endothelial injury in the future, although this application requires validation by prospective studies.
This study has several methodological strengths. Firstly, it features a large and well-defined sample size. As a high-quality ICU cohort, the MIMIC-IV database provides abundant clinical variables and long-term follow-up data, enhancing statistical power. Secondly, the study covers comprehensive time points by evaluating outcomes at 30, 90, and 365 days, confirming that the predictive value of EASIX is temporally stable—suitable for both short-term prognosis assessment and long-term risk prediction. Thirdly, the statistical analysis methods are rigorous: a multilevel Cox proportional hazards model was used to sequentially adjust for confounding factors, ranging from demographic characteristics to comprehensive laboratory indicators and disease severity scores, ensuring the robustness of the results; restricted cubic spline analysis objectively verified the linear relationship, avoiding selection bias from artificial cutoffs; the Boruta algorithm, based on the principle of random forests, ranked variable importance, reducing the false-positive risk associated with traditional univariate analysis. Fourthly, subgroup analyses covered key clinical characteristics, and interaction tests revealed no significant heterogeneity (except for BMI), indicating that EASIX has broad applicability and is not significantly influenced by factors such as age, gender, or comorbidities.
The limitations of this study should be fully considered when interpreting the results. Firstly, the retrospective design inherently limits causal inference. Despite comprehensive adjustments, residual confounding from unmeasured variables (e.g., severity of coronary artery lesions, specific treatment strategies, timing of infection source control) cannot be completely ruled out. Prospective cohort studies are necessary to validate causal relationships. Secondly, although the MIMIC-IV database is large-scale, the 2,468 patients ultimately included in this analysis accounted for only 10.7% of the initially screened population, primarily due to missing key variables in 16,206 patients. This selective inclusion may introduce survival bias and overestimate the predictive performance of EASIX. Since LDH is not a routine test in clinical practice, future studies need to evaluate its generalizability in cohorts with complete data. Thirdly, the single-center data source (Beth Israel Deaconess Medical Center, Boston, USA) restricts the external validity of the results. Differences in clinical practices across regions, ethnicities, and healthcare systems may affect the optimal cutoff value and predictive efficacy of EASIX. Fourthly, this study only used baseline EASIX values calculated within 24 hours of ICU admission and did not assess its dynamic changes. Endothelial injury is a dynamic process, and serial monitoring of EASIX may provide more abundant time-dependent prognostic information. Fifthly, although outliers were excluded, EASIX is calculated as the product of three indicators, meaning measurement errors in any single parameter may be amplified. The potential impact of extreme values on the results requires attention.
Future research should focus on the following directions: (I) conduct multicenter, prospective cohort studies to validate the predictive performance of EASIX across diverse healthcare settings and ethnic populations, and explore ethnicity-specific cutoff values; (II) design nested studies to assess the incremental predictive value of dynamic changes in EASIX (e.g., daily monitoring) for prognosis, as well as whether a decrease in EASIX following therapeutic interventions is associated with improved survival; (III) further investigate the biological mechanisms underlying the interaction between BMI and EASIX, particularly the regulatory roles of muscle mass, adipokines, and inflammatory status; (IV) evaluate in RCTs whether EASIX-guided treatment strategies (e.g., intensified antithrombotic, anti-inflammatory, or endothelial protective therapy) can improve clinical outcomes, and validate its potential as a therapeutic target; (V) explore the combined application of EASIX with other emerging biomarkers (e.g., soluble thrombomodulin, von Willebrand factor), to construct more accurate prognostic models related to endothelial injury.
Conclusions
This study confirms that EASIX is an independent predictor of short-term and long-term mortality in CHD patients admitted to ICUs. Its predictive performance is robust, with an approximately linear dose-response relationship with mortality. As a simple and cost-effective composite biomarker, EASIX can effectively capture the systemic pathological status characterized by endothelial injury, microcirculatory disorders, and coagulation dysfunction, providing a novel tool for early clinical risk stratification. Despite limitations such as the retrospective design and selection bias, the current evidence sufficiently supports the routine monitoring of EASIX in critically ill CHD patients. Future prospective studies are needed to further validate its clinical utility and explore its potential value as an indicator for monitoring treatment responses and a therapeutic target.
Supplementary
The article’s supplementary files as
Acknowledgments
None.
Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments.
Footnotes
Reporting Checklist: The authors have completed the STROBE reporting checklist. Available at https://jtd.amegroups.com/article/view/10.21037/jtd-2026-1-0099/rc
Funding: The study was supported by Henan Provincial Medical Science and Technology Key Research and Development Program Joint Construction Project (No. LHGJ20240156).
Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://jtd.amegroups.com/article/view/10.21037/jtd-2026-1-0099/coif). The authors have no conflicts of interest to declare.
References
- 1.Roth GA, Mensah GA, Johnson CO, et al. Global Burden of Cardiovascular Diseases and Risk Factors, 1990-2019: Update From the GBD 2019 Study. J Am Coll Cardiol 2020;76:2982-3021. 10.1016/j.jacc.2020.11.010 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Hadi HA, Carr CS, Al Suwaidi J. Endothelial dysfunction: cardiovascular risk factors, therapy, and outcome. Vasc Health Risk Manag 2005;1:183-98. [PMC free article] [PubMed] [Google Scholar]
- 3.Libby P, Buring JE, Badimon L, et al. Atherosclerosis. Nat Rev Dis Primers 2019;5:56. 10.1038/s41572-019-0106-z [DOI] [PubMed] [Google Scholar]
- 4.Luft T, Benner A, Jodele S, et al. EASIX in patients with acute graft-versus-host disease: a retrospective cohort analysis. Lancet Haematol 2017;4:e414-23. 10.1016/S2352-3026(17)30108-4 [DOI] [PubMed] [Google Scholar]
- 5.Xu HB, Ye Y, Xue F, et al. Association Between Endothelial Activation and Stress Index and 28-Day Mortality in Septic ICU patients: a Retrospective Cohort Study. Int J Med Sci 2023;20:1165-73. 10.7150/ijms.85870 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Zuo P, Zhu H, Sun C, et al. Association between endothelial activation and stress index and mortality in critically ill patients with atrial fibrillation: In MIMIC database: A Retrospective Cohort Study. PLoS One 2026;21:e0342664. 10.1371/journal.pone.0342664 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Kalicińska E, Biernat M, Rybka J, et al. Endothelial Activation and Stress Index (EASIX) as an Early Predictor for Mortality and Overall Survival in Hematological and Non-Hematological Patients with COVID-19: Multicenter Cohort Study. J Clin Med 2021;10:4373. 10.3390/jcm10194373 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Dong G, Wu T, Gu X, et al. Endothelial Activation and Stress Index Predicts All-Cause and Cardiovascular Mortality in Hypertensive Individuals: A Nationwide Study. J Clin Hypertens (Greenwich) 2025;27:e70057. 10.1111/jch.70057 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Deanfield JE, Halcox JP, Rabelink TJ. Endothelial function and dysfunction: testing and clinical relevance. Circulation 2007;115:1285-95. 10.1161/CIRCULATIONAHA.106.652859 [DOI] [PubMed] [Google Scholar]
- 10.Jin Q, Liu T, Qiao Y, et al. Oxidative stress and inflammation in diabetic nephropathy: role of polyphenols. Front Immunol 2023;14:1185317. 10.3389/fimmu.2023.1185317 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Johnson AEW, Bulgarelli L, Shen L, et al. MIMIC-IV, a freely accessible electronic health record dataset. Sci Data 2023;10:1. 10.1038/s41597-022-01899-x [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Gavriilaki E, Sakellari I, Gavriilaki M, et al. A New Era in Endothelial Injury Syndromes: Toxicity of CAR-T Cells and the Role of Immunity. Int J Mol Sci 2020;21:3886. 10.3390/ijms21113886 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Team DC. Decisionlinnc is a Platform That Integrates Multiple Programming Language Environments and Enables Data Processing, Data Analysis, and Machine Learning Through a Visual Interface. Hangzhou: Statsape Co. Ltd.; 2023. [Google Scholar]
- 14.Yang C, Zhou S, Wen C, et al. The endothelial activation and stress index as a predictor of 28-day mortality in pulmonary sepsis: a retrospective two-cohort analysis. Front Med (Lausanne) 2026;13:1714682. 10.3389/fmed.2026.1714682 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Yu RY, Liu ZN, Wang Y, et al. The endothelial activation and stress index as a key prognostic marker in severe ischemic stroke. Eur J Med Res 2026;31:241. 10.1186/s40001-025-03823-6 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Pennisi M, Sanchez-Escamilla M, Flynn JR, et al. Modified EASIX predicts severe cytokine release syndrome and neurotoxicity after chimeric antigen receptor T cells. Blood Adv 2021;5:3397-406. 10.1182/bloodadvances.2020003885 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Peña M, Salas MQ, Mussetti A, et al. Pretransplantation EASIX predicts intensive care unit admission in allogeneic hematopoietic cell transplantation. Blood Adv 2021;5:3418-26. 10.1182/bloodadvances.2021004812 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Greenbaum U, Strati P, Saliba RM, et al. CRP and ferritin in addition to the EASIX score predict CAR-T-related toxicity. Blood Adv 2021;5:2799-806. 10.1182/bloodadvances.2025018619 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Yu JW, Yang CJ, Yang JF, et al. Relationship of endothelial activation and stress index on concurrent acute kidney injury in patients with heart failure during intensive care unit: a competing risk model analysis. BMC Cardiovasc Disord 2025;26:22. 10.1186/s12872-025-05403-2 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Cai S, Lin S, Yang B, et al. Association between endothelial activation and stress index and all-cause mortality in intensive care unit patients with intracerebral hemorrhage: A retrospective cohort study. Neurosurg Rev 2025;49:29. 10.1007/s10143-025-03968-2 [DOI] [PubMed] [Google Scholar]
- 21.Kordelas L, Terzer T, Gooley T, et al. EASIX-1year and late mortality after allogeneic stem cell transplantation. Blood Adv 2023;7:5374-81. 10.1182/bloodadvances.2022008617 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Liu L, Xiang S, Cao J, et al. EASIX score predicts non-relapse mortality and severe infection in elderly patients undergoing allo-HSCT. Br J Haematol 2025;207:498-508. 10.1111/bjh.20210 [DOI] [PubMed] [Google Scholar]
- 23.Rehn P, Tsoukakis I, Full P, et al. Preoperative Endothelial Activation and Stress Index as a predictor of in-hospital mortality and postoperative complications in cardiac surgery with cardiopulmonary bypass: A retrospective cohort study. Eur J Anaesthesiol 2025. [Epub ahead of print]. doi: . 10.1097/EJA.0000000000002313 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Smyk JM, Danielecka Z, Kotowska M, et al. Cardiovascular risks and endothelial dysfunction in reproductive-age women with endometriosis. Sci Rep 2024;14:24127. 10.1038/s41598-024-73841-7 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Frangogiannis NG. Cardiac fibrosis. Cardiovasc Res 2021;117:1450-88. 10.1093/cvr/cvaa324 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Crea F, Libby P. Acute Coronary Syndromes: The Way Forward From Mechanisms to Precision Treatment. Circulation 2017;136:1155-66. 10.1161/CIRCULATIONAHA.117.029870 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Reventun P, Sánchez-Esteban S, Cook-Calvete A, et al. Endothelial ILK induces cardioprotection by preventing coronary microvascular dysfunction and endothelial-to-mesenchymal transition. Basic Res Cardiol 2023;118:28. 10.1007/s00395-023-00997-0 [DOI] [PMC free article] [PubMed] [Google Scholar]