Nomogram for major adverse cardiovascular events risk during hospitalization after percutaneous coronary intervention in patients with NSTEMI: Training and validation

Abstract

Objective

This study aimed to develop a nomogram model to predict the risk of in-hospital major adverse cardiovascular events (MACE) following percutaneous coronary intervention (PCI) in Non-ST-segment elevation myocardial infarction (NSTEMI) patients and assess its performance.

Methods

Patient data was collected and individuals were randomly assigned to a training cohort (n = 527) or a validation cohort (n = 227). In the training cohort, LASSO-logistic regression analyses were conducted to identify risk factors associated with MACE in NSTEMI patients. The model's predictive performance, discrimination, and consistency were evaluated using metrics such as the receiver operating characteristic curve, calibration curve, and Decision Curve Analysis. The LASSO-logistic analysis for the training cohort identified BMI (OR:1.49, 95% confidence interval (CI): 1.25–1.78, P = 0.000), adjusted GRACE score (per 10 units GRACE score, adjusted OR [aOR]: 1.20, 95% CI: 1.04–1.37, P = 0.010), and adjusted Gensini score (per 10 units Gensini score, aOR: 1.15, 95% CI: 1.03–1.28, P = 0.013) as predictors of in-hospital MACE for patients with NSTEMI who underwent PCI.

Results

In the development cohort, AUC in the prediction model was 0.871 (95% CI: 0.762–0.980), while in the validation cohort, it was 0.961 (95% CI: 0.927–0.995). The calibration curve and Hosmer-Lemeshow test results indicate that the nomogram was well-calibrated. The DCA curve demonstrates that the DCA map of the nomogram has good clinical application ability. Patients with NSTEMI undergoing PCI are known to have an increased risk of MACE.

Conclusion

The developed nomogram model we established reliably predicts the occurrence of in-hospital MACE in NSTEMI patients undergoing PCI, improving healthcare decision-making accuracy.

Keywords

LASSO-logistic regression model in-hospital MACE NSTEMI PCI nomogram

Highlight

This study is based on the LASSO-logistic regression model to establish a nomogram for predicting in-hospital MACE after PCI in NSTEMI patients.

This model comprises three variables: BMI, adjusted Gensini, and adjusted GRACE.

1. Introduction

Acute coronary syndrome (ACS), encompassing unstable angina (UA), non-ST-segment elevation myocardial infarction (NSTEMI), and ST-segment elevation myocardial infarction (STEMI), remains a major global health burden. The pathological basis of NSTEMI is mainly coronary atherosclerosis, unstable plaque rupture, erosion, vascular spasm, and thrombus formation, leading to myocardial ischemia or necrosis. Clinically, NSTEMI has about twice the incidence of STEMI, with comparable long-term morbidity and mortality. Compared to STEMI, NSTEMI patients are older, have longer symptom-to-consultation times, more atypical symptoms, complications, and poorer prognosis, with higher rates of multivessel lesions and severe stenosis.¹ Despite lower in-hospital mortality, NSTEMI shows higher reinfarction, recurrent angina, and long-term event rates.² Thus, its diagnosis, risk stratification, and treatment remain controversial.

While percutaneous coronary intervention (PCI) effectively restores coronary blood flow and reduces major adverse cardiovascular events (MACE) in NSTEMI.³ The 2021 ACCF/AHA guidelines advocate proactive management for NSTEMI, yet the risk of MACE remains high after PCI.⁴ Studies have shown that diabetic patients with NSTEMI have an increased risk of thrombosis and restenosis after PCI, and complications such as acute and subacute in-stent thrombosis can also jeopardize the success of the procedure.^5,6 Therefore, maintaining vascular patency and effective myocardial perfusion after PCI has become a critical clinical issue.

Recent studies identify multiple risk factors for post-PCI MACE in NSTEMI, including renal function (serum creatinine/albumin),⁷ age (≥80 years),⁸ PCI timing (optimal 3–14 h for high-risk patients),⁹ and demographics (gender/age).¹⁰ Additional predictors include admission mode, C-reactive protein (CRP), total cholesterol (TC), high - density lipoprotein (HDL), and low-density lipoprotein (LDL), and coronary microvascular dysfunction.^11,12

Currently, NSTEMI prediction models mainly focus on mortality or long-term prognosis, with scarce models for in-hospital MACE risk post-PCI. This study aims to fill this gap by developing a nomogram based on risk factors to predict in-hospital MACE after PCI in NSTEMI patients, enabling early and accurate risk management to improve outcomes.

2. Materials and methods

2.1. Patients enrolled

A retrospective data analysis was conducted to gather the clinical characteristics of 754 patients with NSTEMI who underwent PCI at the People's Hospital of Xinjiang Uygur Autonomous Region between January 2017 and December 2021. Patients were randomly divided into a training group (n = 527) and a validation group (n = 227) in a 7:3 ratio.

Inclusion criteria were as follows: (1) Anginal symptoms such as chest pain lasting for more than 30 min and unresponsive to nitroglycerin. (2) ST-segment depression in two or more consecutive ECG leads > 0.1 mV or (and) T-wave limbic symmetry, deep inversion, or dynamic ST-segment changes. (3) Creatine kinase isoenzyme (CK-MB) test more than twice the upper limit of normal with a dynamic evolution and positive troponin. (4) The patient with NSTEMI underwent PCI after admission.

Exclusion criteria were as follows: (1) ST-segment elevation myocardial infarction, heart valve disease, chronic renal failure, pulmonary heart disease, acute pulmonary embolism, and left ventricular hypertrophy due to hypertension. (2) incomplete case information.

2.2. Data collection

Demographic information (age, sex, BMI, race, smoking and drinking history), past medical history (diabetes and hypertension), medications (aspirin, calcium channel blockers, diuretic, β-blocker, angiotensin-converting enzyme inhibitor/angiotensin receptor blocker (ACEI/ARB)), and vital signs (heart rate (HR), pulse pressure, systolic blood pressure (SBP), diastolic blood pressure (DBP)) were obtained from patient electronic medical records. Laboratory tests included calcium channel blocker (CCB), angiotensin-converting enzyme inhibitor (ACEI), white blood cell count (WBC), red blood cell count (RBC), neutrophil count (NEUT), lymphocyte (LY), mononucleosis (MONO), basophil (BASO), hemoglobin (Hb), hematocrit (HCT), red cell volume distribution width (RDW), platelets (PLT), platelet distribution width (PDW), prothrombin activity (PTA), international normalized ratio (INR), fibrinogen (FIB), urea, serum creatinine (Scr), activated partial thromboplastin time (APTT), total cholesterol (TC), glutamic (GLU), alanine transaminase (ALT), aspartate transaminase (AST), AST/ALT, uric acid (UA), glucose (GLU), triglyceride (TG), lactate dehydrogenase (LDH), albumin (ALB), low-density lipoprotein (LDL-C), high-density lipoprotein (HDL-C), albumin-globulin ratio (A/G), MACE, and creatine kinase MB (CK-MB). Additionally, the GRACE score and the Gensini score were collected. The GRACE score was calculated based on eight criteria: age, heart rate, systolic blood pressure, serum creatinine, Killip grade, cardiac arrest prior to admission, ST segment alterations on the ECG, and increased myocardial markers.¹³ The Gensini score is the product of the scores for the degree of each lesion vessel lumen narrowing multiplied by the associated weight coefficient based on the coronary angiography results. First, coronary artery stenosis is classified as having 1, 2, 4, 8, 16, or 32 points depending on the degree of stenosis: ≤ 25%, 26% ∼ 50%, 51% ∼ 75%, 76% ∼ 90%, 91% ∼ 99%, 100%. The weight coefficient is then calculated using the following parameters: the left main trunk multiplied by 5, the distal left anterior descending branch multiplied by 1, the middle section multiplied by 1.5, the proximal segment multiplied by 2.5, the diagonal branch of the first diagonal branch multiplied by 1, the second diagonal branch multiplied by 0.5, the distal left circumflex branch multiplied by 1, the blunt limb multiplied by 1, the proximal multiplied by 2.5, the posterior descending branch multiplied by 1, the posterior lateral branch multiplied by 0.5, and the proximal, middle, distal, and posterior descending branches of the right coronary artery multiplied by 1.¹⁴ The adjusted GRACE score indicates that for every 10-unit increase in the GRACE score, there is a corresponding 10-unit increase in the adjusted Gensini score. All patients’ PCI was done in our hospital's cath lab. All procedures were carried out in accordance with the necessary rules and regulations. In this study, the indicators with missing values greater than 20% were deleted, and the remaining data were filled by the multiple imputation method of the “mice” package.

2.3. The definitions of MACE

In-hospital MACE includes in-hospital cardiac death, malignant arrhythmias, congestive heart failure, cardiogenic shock, cardiac arrest, sudden cardiac death, and stroke.¹⁵ The occurrence of the first adverse event during hospitalization was determined based on comprehensive analysis of the patient's medical records.

2.4. Model training and statistical analysis

First, the data were split into a training cohort (n = 527) and a validation cohort (n = 227) at random in a 7:3 ratio. The statistical analysis was conducted using R (4.1.3). The numerical variables’ distributions were evaluated using the Shapiro-Wilk and Kolmogorov-Smirnov tests. With a normal distribution, continuous variables were written as mean ± standard deviation (X ± S), and the independent samples t-test was performed to compare the two groups. Continuous variables with non-normal distributions were compared between groups using the Mann-Whitney U test, and they were reported as the median (interquartile range) or M (IQR). In order to compare groups, categorical variables were expressed as percentages, and Pearson's chi-square test, continuous-adjusted chi-square test, or Fisher's exact test were used. Statistics were deemed significant at P < 0.05.

In the LASSO-logistic regression analysis, variables that might affect the study's findings were incorporated based on the training cohort. The “glmnet” package in R was used for LASSO regression. The optimal parameter (lambda) was selected through ten-fold cross-validation. The minimum average error was used to determine the optimal lambda value. Variables with non-zero coefficients after LASSO regression were further analyzed by univariate and multivariate logistic regression. The performance of the training and validation cohorts’ models was evaluated using the subject operating characteristic curve (ROC), calibration curve, and Decision Curve Analysis (DCA). The nomogram was constructed based on the results of the LASSO-logistic regression, and the probability of an event was calculated by summing the points of all variables.

3. Results

3.1. Baseline characteristics

Figure 1 demonstrates that a total of 754 patients were included in this study based on predefined inclusion and exclusion criteria. These patients were subsequently divided into a training cohort (n = 527) and a validation cohort (n = 227). Within each cohort, patients were further categorized into MACE and non-MACE groups based on the presence or absence of in-hospital MACE. Among the patients in the training cohort, 417 (79.1%) were males, 160 (30.4%) individuals had diabetes, 298 (56.5%) had hypertension, and 273 (51.8%) had a history of smoking. Baseline characteristics and outcomes were well-balanced between the two cohorts (Table 1).

Figure 1.

Research process diagram.

Table 1.

Baseline characteristics of patients in the development and validation cohorts.

Variables	Description	Training group(N = 527)	Validation group (N = 227)	P
Sex n (%)	Female	110 (20.9%)	56 (24.7%)	0.290
	male	417 (79.1%)	171 (75.3%)
Age n (%)	≤60 (years)	259 (49.1%)	99 (43.6%)	0.188
	>60 (years)	268 (50.9%)	128 (56.4%)
BMI (kg/m²)	Mean ± SD	26.4 ± 3.4	26.5 ± 3.2	0.741
Hypertension n (%)	NO	229 (43.5%)	111 (48.9%)	0.194
	YES	298 (56.5%)	116 (51.1%)
Diabetes n (%)	NO	367 (69.6%)	156 (68.7%)	0.869
	YES	160 (30.4%)	71 (31.3%)
Smoking n (%)	NO	254 (48.2%)	117 (51.5%)	0.445
	YES	273 (51.8%)	110 (48.5%)
Drinking n (%)	NO	416 (78.9%)	181 (79.7%)	0.881
	YES	111 (21.1%)	46 (20.3%)
Number of vessel lesions	1	51 (9.7%)	19 (8.4%)	0.315
n (%)	2	150 (28.5%)	77 (33.9%)
	3	326 (61.9%)	131 (57.7%)
Adjusted GRACE	Mean ± SD	12.5 ± 3.5	12.8 ± 3.4	0.355
Adjusted Gensini	Mean ± SD	5.5 ± 4.0	5.7 ± 4.2	0.652
SBP (mmHg)	Mean ± SD	129.5 ± 20.4	128.5 ± 19.4	0.494
DBP (mmHg)	Mean ± SD	77.2 ± 12.9	75.4 ± 11.2	0.054
Pulse pressure	Mean ± SD	52.4 ± 16.2	53.1 ± 16.3	0.584
HR (bpm)	Mean ± SD	78.8 ± 15.3	77.8 ± 12.6	0.345
Beta blockers n (%)	NO	476 (90.3%)	202 (89%)	0.669
	YES	51 (9.7%)	25 (11%)
CCB n (%)	NO	423 (80.3%)	176 (77.5%)	0.451
	YES	104 (19.7%)	51 (22.5%)
ACEI n (%)	NO	416 (78.9%)	174 (76.7%)	0.547
	YES	111 (21.1%)	53 (23.3%)
Diuretic n (%)	NO	490 (93%)	216 (95.2%)	0.337
	YES	37 (7%)	11 (4.8%)
Asplin n (%)	NO	7 (1.3%)	2 (0.9%)	0.878
	YES	520 (98.7%)	225 (99.1%)
WBC n(%)	≤12.22 10⁹/L	381 (72.3%)	159 (70%)	0.588
	>12.22 10⁹/L	146 (27.7%)	68 (30%)
NEUT n (%)	≤7.26 10⁹/L	381 (72.3%)	160 (70.5%)	0.675
	>7.26 10⁹/L	146 (27.7%)	67 (29.5%)
LY n (%)	≤4 10⁹/L	523 (99.2%)	225 (99.1%)	1.000
	>4 10⁹/L	4 (0.8%)	2 (0.9%)
MONO n(%)	≤0.8 10⁹/L	435 (82.5%)	188 (82.8%)	1.000
	>0.8 10⁹/L	92 (17.5%)	39 (17.2%)
BASO n (%)	≤0.1 10⁹/L	522 (99.1%)	221 (97.4%)	0.147
	>0.1 10⁹/L	5 (0.9%)	6 (2.6%)
RBC n (%)	≤4 10¹²/L	494 (93.7%)	216 (95.2%)	0.554
	>4 10¹²/L	33 (6.3%)	11 (4.8%)
Hb n (%)	≤110 g/L	466 (88.4%)	208 (91.6%)	0.237
	>110 g/L	61 (11.6%)	19 (8.4%)
HCT n (%)	≤35%	512 (97.2%)	223 (98.2%)	0.537
	>35%	15 (2.8%)	4 (1.8%)
RDW n (%)	≤15.5 fL	517 (98.1%)	223 (98.2%)	1.000
	>15.5 fL	10 (1.9%)	4 (1.8%)
PLT n (%)	≤100 10⁹/L	455 (86.3%)	184 (81.1%)	0.082
	>100 10⁹/L	72 (13.7%)	43 (18.9%)
PDW n (%)	≤9 fL	500 (94.9%)	219 (96.5%)	0.442
	>9 fL	27 (5.1%)	8 (3.5%)
PTA n (%)	≤130%	496 (94.1%)	207 (91.2%)	0.190
	>130%	31 (5.9%)	20 (8.8%)
INR n (%)	≤1.5	494 (93.7%)	209 (92.1%)	0.498
	>1.5	33 (6.3%)	18 (7.9%)
FIB n (%)	≤4 g/L	400 (75.9%)	166 (73.1%)	0.474
	>4 g/L	127 (24.1%)	61 (26.9%)
APTT n (%)	≤40 s	489 (92.8%)	213 (93.8%)	0.717
	>40 s	38 (7.2%)	14 (6.2%)
UREA n (%)	≤7.1 mmol/L	475 (90.1%)	201 (88.5%)	0.599
	>7.1 mmol/L	52 (9.9%)	26 (11.5%)
SCr n (%)	≤72.4 umol/L	498 (94.5%)	210 (92.5%)	0.379
	>72.4 umol/L	29 (5.5%)	17 (7.5%)
UA n (%)	≤416 umol/L	267 (50.7%)	113 (49.8%)	0.886
	>416 umol/L	260 (49.3%)	114 (50.2%)
GLU n (%)	≤6.1 mmol/L	178 (33.8%)	79 (34.8%)	0.850
	>6.1 mmol/L	349 (66.2%)	148 (65.2%)
TG n (%)	≤1.67 mmol/L	369 (70%)	171 (75.3%)	0.163
	>1.67 mmol/L	158 (30%)	56 (24.7%)
TC n (%)	≤5.6 mmol/L	481 (91.3%)	207 (91.2%)	1.000
	>5.6 mmol/L	46 (8.7%)	20 (8.8%)
HDL_c n (%)	≤1.8 mmol/L	423 (80.3%)	186 (81.9%)	0.664
	>1.8 mmol/L	104 (19.7%)	41 (18.1%)
LDL_c n (%)	≤3.12 mmol/L	388 (73.6%)	171 (75.3%)	0.689
	>3.12 mmol/L	139 (26.4%)	56 (24.7%)
ALB n (%)	≤51 g/L	61 (11.6%)	31 (13.7%)	0.497
	>51 g/L	466 (88.4%)	196 (86.3%)
A/G n (%)	≤2.5	394 (74.8%)	174 (76.7%)	0.646
	>2.5	133 (25.2%)	53 (23.3%)
AST n (%)	≤52.94 U/L	225 (42.7%)	89 (39.2%)	0.418
	>52.94 U/L	302 (57.3%)	138 (60.8%)
ALT n (%)	≤29.83 U/L	287 (54.5%)	116 (51.1%)	0.442
	>29.83 U/L	240 (45.5%)	111 (48.9%)
AST/ALT n (%)	≤2.5	301 (57.1%)	146 (64.3%)	0.077
	>2.5	226 (42.9%)	81 (35.7%)
LDH n (%)	≤3 U/L	277 (52.6%)	118 (52%)	0.947
	> 3 U/L	250 (47.4%)	109 (48%)
CKMB n (%)	≤111.3 ng/mL	248 (47.1%)	129 (56.8%)	0.017
	>111.3 ng/mL	279 (52.9%)	98 (43.2%)
MACE	NO	510 (96.8%)	222 (97.8%)	0.596
	YES	17 (3.2%)	5 (2.2%)

3.2. LASSO-logistic regression analysis of risk factors

Potentially relevant risk factors were subjected to dimensionality reduction using LASSO regression in order to extract significant predictive factors and prevent overfitting. The optimal parameter (lambda) in the LASSO model was selected through ten-fold cross-validation. The minimum average error was used to determine the optimal value of the model (Figure 2). The results of the LASSO regression showed that when lambda was set to 0.0085, LY, ALB, AST, AST/ALT, CKMB, HR, adjusted GRACE, adjusted Gensini, and BMI were identified as risk factors influencing the occurrence of in-hospital MACE in patients. Furthermore, based on the univariate logistic regression analysis of the aforementioned risk factors, the following factors were found to have a significant association with in-hospital MACE (P < 0.05): LY, ALB, AST, AST/ALT, HR, adjusted GRACE, adjusted Gensini, and BMI. Subsequently, multivariate logistic regression analysis was performed using the aforementioned risk variables. It was depicted that adjusted GRACE (per 10 units Gensini score, aOR: 1.20, 95% confidence interval (CI): 1.04–1.37, P = 0.01), adjusted Gensini (OR: 1.15, 95% CI: 1.03–1.28, P = 0.013), and BMI (OR: 1.49, 95% CI: 1.25–1.78, P = 0.000) (Table 2).

Figure 2.

Variable screening on Lasso regression. (A) The characteristics of the variation in the coefficient of variables; (B) Identifying the best value of the parameter λ in the Lasso regression model via the cross-validation approach.

Table 2.

Univariable and multivariable analysis of independent risk factors associated with in-hospital MACE.

	Univariate analysis			Multivariate analysis
Variables	OR	CI	P	OR	CI	P
LY	33.87	4.46–257	0.001	NA	NA	NA
ALB	0.22	0.08–0.62	0.004	0.39	0.11–1.33	0.133
AST	5.83	1.32–25.75	0.02	4.37	0.92–20.74	0.064
AST/ALT	3.32	1.15–9.57	0.026	NA	NA	NA
CKMB	55446955.33	0–Inf	0.987	NA	NA	NA
HR	1.02	1–1.05	0.048	1.02	0.99–1.05	0.135
AdjustedGRACE	1.27	1.13–1.44	0	1.20	1.04–1.37	0.01
AdjustedGensini	1.17	1.07–1.28	0	1.15	1.03–1.28	0.013
BMI	1.51	1.29–1.77	0	1.49	1.25–1.78	<0.001

The adjusted GRACE score reflects the overall risk status of patients, integrating multiple clinical parameters such as age, heart rate, and serum creatinine. A higher adjusted GRACE score indicates a greater risk of MACE. And the adjusted Gensini score quantifies the severity of coronary artery atherosclerosis. A higher score means more severe coronary artery stenosis and a greater burden of atherosclerotic lesions. After PCI, although the blood flow was initially improved, due to the large atherosclerotic burden, the risk of reocclusion and MACE remained high. BMI is an important indicator related to body fat content. Obesity (high BMI) is associated with increased inflammation, abnormal lipid metabolism, and endothelial dysfunction, all of which can promote the development of atherosclerosis and increase the risk of MACE.

3.3. Prediction model training and validation

The above independent predictors were incorporated into the nomogram (Figure 3). The prognostic model for in-hospital MACE demonstrated an area under the curve (AUC) of 0.871 (95% confidence interval: 0.762–0.980), a sensitivity of 0.882, and a specificity of 0.780 in the training cohort, as illustrated in Figure 4A. In the validation cohort, the model exhibited an AUC of 0.961 (95% CI: 0.927–0.995), a sensitivity of 1.000, and a specificity of 0.914 (Figure 4B). Figure 5 illustrates the calibration curve of the model, revealing a high degree of concordance between the predictions generated by the nomogram model and the actual observations in both the training cohort (Figure 5A) and the validation cohort (Figure 5B). Notably, the calibration curves closely approximate diagonal lines for both the training and validation groups. The Hosmer-Lem show test indicated no statistically significant differences (training cohort: χ2 = 14.695, p = 0.100; validation cohort: χ2 = 1.711, p = 0.995), affirming the excellent fit of the nomogram model to the data. Finally, Decision Curve Analysis (DCA) curves were constructed to illustrate the model's favorable performance in a clinical setting, demonstrating a superior net benefit. In the training and validation cohort DCAs, the model offered a net benefit over the “happen-all” or “happen-none” strategies. (Figure 5C, D), demonstrating that our nomogram was clinically useful.

Figure 3.

Nomogram for the prognosis of in-hospital MACE after PCI in patients with NSTEMI, which developed in the training cohort and included BMI, adjusted Gensini, adjusted GRACE.

Figure 4.

ROC curves for the training group and the validation group. Diagonal line represents the reference (AUC = 0.5, no discriminative power), AUC >0.7 indicates excellent discriminative ability of the model.

Figure 5.

(A) Calibration curves for the training group. (B) Calibration curves for the validation group. (C) Decision-curve analysis of the development group; (D) Decision-curve analysis of the validation group. Calibration Curves (5A, 5B): The diagonal line represents ideal calibration. Closer alignment between the predicted probability and the ideal line indicates higher calibration accuracy. Decision Curve Analysis (5C, 5D): The nomogram model demonstrates clinical utility when its curve lies above the ‘treat all’ (gray line) and ‘treat none’ (horizontal dashed line) strategies.

4. Discussion

In this study, we took into account five risk factors, including adjusted GRACE, adjusted Gensini, and BMI, for in-hospital MACE after PCI in patients with NSTEMI. In both the training cohort (AUC = 0.871, 95% CI: 0.762–0.980) and validation cohort (AUC = 0.961, 95% CI: 0.927–0.995), the constructed clinical nomogram model showed strong predictive ability.

NSTEMI involves partial arterial occlusion, subendocardial ischemia, and myocardial necrosis. Unstable plaques with thrombus can cause reocclusion and reinfarction. Even with adequate blood flow, overlying thrombus on unstable plaques may trigger reinfarction.¹⁶ In recent decades, While STEMI mortality has declined due to advancements in early treatment, NSTEMI mortality remains unchanged.¹⁷ International guidelines recommend invasive therapy for intermediate/high-risk patients, but low-risk and elderly patients with comorbidities lack clear guidance.¹⁸ Furthermore, existing prognostic stratification tools demonstrate limited prediction for in-hospital mortality.

When compared with existing prediction models for NSTEMI patients, such as the TIMI score and PURSUIT score, our model exhibits several notable advantages. The TIMI score primarily focuses on culprit lesion characteristics—including the degree of coronary artery stenosis and myocardial injury markers and short-term treatment responses, while overlooking comprehensive patient parameters such as renal function and non-culprit vessel lesion burden.^19,20 While PURSUIT score integrates clinical parameters including age, HR, and SBP for 30-day adverse event prediction, yet critically omits two prognostic determinants: coronary atherosclerotic burden (multivessel disease severity) and metabolic biomarkers (BMI). This limitation restricts its utility in guiding PCI optimization and postprocedural metabolic interventions.^20,21 Our model incorporates the adjusted GRACE score, the adjusted Gensini score, and BMI.

Among them, the Gensini score comprehensively quantifies coronary atherosclerotic burden through angiography, incorporating even mild stenosis (>0% diameter narrowing). A novel adaptation, the non-culprit Gensini score, quantifies atherosclerotic burden in non-culprit lesions by subtracting the culprit lesion score, addressing angiographic limitations in visualizing distal segments beyond occluded culprit lesions.^22,23 This may explain the poor prognosis in nonobstructive coronary NSTEMI patients. Post-PCI, dislodged microthrombi can embolize non-culprit arteries, causing blood flow disruption and in-hospital MACE. Numerous studies show patients with greater coronary artery disease burden face higher recurrent ischemic events (e.g., myocardial infarction, stroke, death) than those with less severe burden. In NSTEMI patients, multiple coronary lesions (stenosis/blockage areas) strongly correlate with increased adverse cardiovascular events, and more lesions elevate myocardial blood flow reduction and ischemic event risks.^24,25

The GRACE score is general clinical guidelines for risk stratification of patients with NSTEMI, which is superior to other methods in the prediction of the risk of cardiovascular events during hospitalization and at 6 months and 1 year after discharge, which is a classic scoring system for the assessment of short- and long-term prognosis.²⁶ Prior research has associated GRACE scores with interventional timing outcomes in NSTEMI²⁷ and confirmed their independent predictive value for MACE incidence, even after confounder adjustment.²⁸ Aligned with this evidence, our model incorporates the GRACE score as a core predictor.

Obesity, a modifiable coronary artery disease risk factor, promotes cardiovascular pathophysiology through adipose tissue dysfunction.²⁹ Dysregulated adipocyte-derived mediators (pro-inflammatory cytokines, atherogenic lipids, and altered adipokines including leptin and resistin) contribute to endothelial dysfunction and plaque instability.³⁰ Excess adiposity critically influences inflammatory cascades and plaque destabilization, synergistically increasing myocardial injury risks even with optimal therapies.³¹ While our data confirm elevated BMI correlates with MACE, an “obesity paradox” still persists—a study demonstrates improved short-term coronary artery disease prognosis in obese versus lean patients.³² This discrepancy likely reflects BMI's inability to discriminate fat/lean mass or assess central adiposity—a key driver of metabolic risk. While epidemiological studies suggest BMI may protect in ACS, clinical evidence consistently links its elevation to long-term adverse events.

This study has several limitations. Firstly, the data was collected from only one center, the People's Hospital of Xinjiang Uygur Autonomous Region, which may introduce selection bias and limit the generalizability of the results. The characteristics of patient populations, treatment methods, and disease management models may vary among different hospitals, thus restricting the universal applicability of the findings. Secondly, although 754 patients were included, the sample size is relatively limited for a complex cardiovascular disease study. It may not be sufficient to comprehensively capture all potential variations in risk factors and individual differences among patients. A relatively small sample size could affect the stability and precision of the model's estimates, increasing the randomness and uncertainty of the results. Finally, although BMI, the adjusted Gensini score, and the adjusted GRACE score were identified as the main predictors, there may be other important factors not included in the analysis.

5. Conclusion

In summary, the developed nomogram model based on the predictors of adjusted GRACE, adjusted Gensini, and BMI offers significant assistance to healthcare providers in the risk assessment and management of NSTEMI patients who have undergone PCI. Clinicians can input the relevant data of patients into the nomogram model to obtain a comprehensive risk score. Based on this score, patients can be stratified into different risk levels. For patients with a relatively high predicted risk of MACE, the doctor may consider adding antiplatelet or lipid-lowering drugs with stronger effects. At the same time, to enhance the clinical applicability and robustness of the nomogram, future research should prioritize multi-center collaboration (e.g., cooperation with the First Affiliated Hospital of Xinjiang Medical University, Xinjiang Uygur Autonomous Region People's Hospital, and Kashgar Prefecture Hospital) to validate the model in a larger cohort, addressing regional and ethnic variability. Meanwhile, integrating machine learning algorithms (such as XGBoost) with advanced imaging biomarkers (such as plaque characteristics derived from coronary CTA) and circulating biomarkers (such as lipoprotein(a)) is recommended to improve predictive accuracy. Finally, subgroup-specific analyses (such as elderly patients with diabetes) could be conducted, and real-time risk assessment tools (such as mobile applications) should be developed to optimize personalized management and clinical decision-making for NSTEMI patients after PCI.

Footnotes

Acknowledgments

We appreciate the support of the People's Hospital of Xin jiang Uygur Autonomous Region for our work.

Author contributions

Feng-xia Wang is the corresponding author who supervised the project and designed the study. Xiang-yu Dong and Ting Wang were responsible for analyzing and interpreting the data and wrote the manuscript who contribute equally as first authors.Yu-juan Yuan,Yong Liu, Zhong-xing Xu, Zi-long Zhang and Yan Feng were responsible for literature retrieval, data extraction, and literature quality evaluation. All authors have read and approved the manuscript.

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The Program of High-level Medical and Health Talents in the “Tianshan Yingcai” Project of the Third Batch of the “2+5” Key Talent Plan in Xinjiang Uygur Autonomous Region – Young and Middle-aged Backbone Medical Talents (No.TSYC202401B073), sponsored by Feng-xia Wang; Add The in-hospital project of the People's Hospital of Xinjiang Uygur Autonomous Region (No.20240118), sponsored by Xiang-yu Dong.

Declaration of conflicting interests

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

References

Bhatt

Lopes

Harrington

. Diagnosis and treatment of acute coronary syndromes: a review. JAMA 2022; 327: 62.

Cordero

Rodríguez-Mañero

Bertomeu-González

, et al. Managing NSTEMI in older patients. Lancet. 2021; 397: 370–371.

Omerovic

Råmunddal

Petursson

, et al. Percutaneous vs. surgical revascularization of non-ST-segment elevation myocardial infarction with multivessel disease: the SWEDEHEART registry. Eur Heart J. 2025; 46: 518–531.

Jernberg

Hasvold

Henriksson

, et al. Cardiovascular risk in post-myocardial infarction patients: nationwide real world data demonstrate the importance of a long-term perspective. Eur Heart J 2015; 36: 1163–1170. 20150113.

Wilson

Mone

Kansakar

, et al. Diabetes and restenosis. Cardiovasc Diabetol. 2022; 21: 23.

Cha

J-J

Bae

Park

D-W

, et al. Clinical outcomes in patients with delayed hospitalization for non–ST-segment elevation myocardial infarction. J Am Coll Cardiol. 2022; 79: 311–323.

Long

Wen

Zhang

. Serum albumin-to-creatinine ratio predicts one-year mortality in elderly patients with non-ST-elevation acute coronary syndrome after percutaneous coronary intervention: a prospective cohort analysis. Med Sci Monit 2024; 30: e945516. 20241018.

Wang

, et al. Differential impacts of invasive percutaneous coronary intervention and conservative strategy on elderly patients with non-ST-segment elevation myocardial infarction: an analysis of short-term and long-term survival. Br J Hosp Med (Lond) 2024; 85: 1–13. 20241125.

Zheng

Xia

, et al. Optimal timing of invasive intervention for high-risk non-ST-segment-elevation myocardial infarction patients. J Geriatr Cardiol 2024; 21: 807–815.

10.

Shi

, et al. Timing of percutaneous coronary intervention and risk of new-onset acute ischemic stroke in non-ST elevation myocardial infarction: a retrospective cohort study insight into the national inpatient sample database (2016-2019). Health Sci Rep 2024; 7: e70029. 20240918.

11.

Liu

, et al. Development and validation of a machine learning-based readmission risk prediction model for non-ST elevation myocardial infarction patients after percutaneous coronary intervention. Sci Rep 2024; 14: 13393. 20240611.

12.

Chen

Zhang

Yidilisi

, et al. Combined risk estimates of diabetes and coronary angiography-derived index of microcirculatory resistance in patients with non-ST elevation myocardial infarction. Cardiovasc Diabetol 2024; 23: 300. 20240816.

13.

Wang

, et al. Plasma biomarkers and plaque strain predict long-term cardiovascular events in patients with acute coronary syndrome. Sci China Life Sci. 2020; 63: 269–278.

14.

Rampidis

Benetos

Benz

, et al. A guide for gensini score calculation. Atherosclerosis 2019; 287: 181–183.

15.

Vernooij

Van Klei

Moons

, et al. The comparative and added prognostic value of biomarkers to the revised cardiac risk index for preoperative prediction of major adverse cardiac events and all-cause mortality in patients who undergo noncardiac surgery. Cochrane Database Syst Rev. 2021; 12: CD013139.

16.

Feistritzer

H-J

Meyer-Saraei

Lober

, et al. Long-term outcome after thrombus aspiration in non-ST-elevation myocardial infarction: results from the TATORT-NSTEMI trial: thrombus aspiration in acute myocardial infarction. Clin Res Cardiol. 2020; 109: 1223–1231.

17.

Ram

Fisman

Tenenbaum

, et al. Revascularization outcomes in diabetic patients presenting with acute coronary syndrome with non-ST elevation. Cardiovasc Diabetol. 2022; 21: 75.

18.

Collet

J-P

Thiele

Barbato

, et al. 2020 ESC guidelines for the management of acute coronary syndromes in patients presenting without persistent ST-segment elevation. Eur Heart J. 2021; 42: 1289–1367.

19.

Kola

Shuka

Meyers

, et al. OMI/NOMI: time for a new classification of acute myocardial infarction. J Clin Med 2024; 13: 5201.

20.

Paredes

Rocha

de Carvalho

, et al. Improvement of CVD risk assessment tools’ performance through innovative patients’ grouping strategies. Annu Int Conf IEEE Eng Med Biol Soc 2012; 2012: 5907–5910.

21.

Isilak

Kardesoglu

Aparci

, et al. Comparison of clinical risk assessment systems in predicting three-vessel coronary artery disease and angiographic culprit lesion in patients with non-ST segment elevated myocardial infarction/unstable angina pectoris. Kardiol Pol 2012; 70: 242–250.

22.

Charach

Blatt

Jonas

, et al. Using the gensini score to estimate severity of STEMI, NSTEMI, unstable angina, and anginal syndrome. Medicine (Baltimore). 2021; 100: e27331.

23.

Tanaka

Miki

Akahori

, et al. Comparison of coronary atherosclerotic disease burden between ST-elevation myocardial infarction and non-ST-elevation myocardial infarction: non-culprit Gensini score and non-culprit SYNTAX score. Clin Cardiol. 2021; 44: 238–243.

24.

Baumann

AAW

Mishra

Worthley

, et al. Management of multivessel coronary artery disease in patients with non-ST-elevation myocardial infarction: a complex path to precision medicine. Ther Adv Chronic Dis. 2020; 11: 204062232093852.

25.

Takeji

Shiomi

Morimoto

, et al. Differences in mortality and causes of death between STEMI and NSTEMI in the early and late phases after acute myocardial infarction. PLOS ONE 2021; 16: e0259268.

26.

Kumar

Ashok

Saghir

, et al. Prognostic value of GRACE score for in-hospital and 6 months outcomes after non-ST elevation acute coronary syndrome. Egypt Heart J 2021; 73: 22.

27.

Wang

J-X

Gao

Xiao

J-Y

, et al. The relevance of interventional time and clinical outcomes in patients with NSTEMI based on the GRACE score. Appl Bionics Biomech. 2022; 2022: 1–8.

28.

Chotechuang

Phrommintikul

Kuanprasert

, et al. GRACE Score and cardiovascular outcomes prediction among the delayed coronary intervention after post-fibrinolytic STEMI patients in a limited PCI-capable hospital. Open Heart 2020; 7: e001133.

29.

Long

Yang

, et al. Smartphone-based interventions in the secondary prevention of cardiovascular disease: a meta-analysis. Technol Health Care 2023; 31: 1935–1948.

30.

Petito

Cioffi

Magnacca

, et al. Adipose tissue remodeling in obesity: an overview of the actions of thyroid hormones and their derivatives. Pharmaceuticals (Basel, Switzerland) 2023; 16: 572. 2023/04/28.

31.

Madala

Franklin

Chen

, et al. Obesity and age of first non-ST-segment elevation myocardial infarction. J Am Coll Cardiol 2008; 52: 979–985. 2008/09/13.

32.

Sungur

Simsek

, et al. Body fat percentage and infarct size in patients with non-ST segment elevation myocardial infarction. North Clin Istanbul 2023; 10: 567–574. 2023/10/13.