Clinical predictive model of new-onset atrial fibrillation in patients with acute myocardial infarction after percutaneous coronary intervention

Abstract

New-onset atrial fibrillation (NOAF) is associated with increased morbidity and mortality. Despite identifying numerous factors contributing to NOAF, the underlying mechanisms remain uncertain. This study introduces the triglyceride-glucose index (TyG index) as a predictive indicator and establishes a clinical predictive model. We included 551 patients with acute myocardial infarction (AMI) without a history of atrial fibrillation (AF). These patients were divided into two groups based on the occurrence of postoperative NOAF during hospitalization: the NOAF group (n = 94) and the sinus rhythm (SR) group (n = 457). We utilized a regression model to analyze the risk factors of NOAF and to establish a predictive model. The predictive performance, calibration, and clinical effectiveness were evaluated using the receiver operational characteristics (ROC), calibration curve, decision curve analysis, and clinical impact curve. 94 patients developed NOAF during hospitalization. TyG was identified as an independent predictor of NOAF and was significantly higher in the NOAF group. Left atrial (LA) diameter, age, the systemic inflammatory response index (SIRI), and creatinine were also identified as risk factors for NOAF. Combining these with the TyG to build a clinical prediction model resulted in an area under the curve (AUC) of 0.780 (95% CI 0.358–0.888). The ROC, calibration curve, decision curve analysis, and clinical impact curve demonstrated that the performance of the new nomogram was satisfactory. By incorporating the TyG index into the predictive model, NOAF after AMI during hospitalization can be effectively predicted. Early detection of NOAF can significantly improve the prognosis of AMI patients.

Introduction

Acute myocardial infarction (AMI) remains one of the leading causes of death globally and is considered a severe cardiovascular disease with a poor prognosis¹. Percutaneous coronary intervention (PCI) and fibrinolysis, through agents like streptokinase, are the primary treatments aimed at restoring blood flow to ischemic tissues². Atrial fibrillation (AF), a commonly encountered arrhythmia in AMI patients, has an incidence of 6–21%³. It has been documented that new-onset atrial fibrillation (NOAF) in AMI patients is associated with a worse prognosis, including extended hospital stays, morbidity and mortality^4,5. Clinical trials also indicate that NOAF significantly worsens both short- and long-term outcomes in patients with ST-segment elevation myocardial infarction (STEMI) during the PCI era^6,7. Reduced cardiac output secondary to impaired atrial contraction, atrioventricular synchronization, and irregular RR intervals may contribute to this poor prognosis⁸. Thus, detecting NOAF is crucial for prognosis. Recent studies have investigated AMI patients with NOAF^6,9–11, identifying several risk factors including age, body mass index (BMI), diabetes, serum albumin and uric acid levels, left atrial (LA) diameter, left ventricular ejection fraction (LVEF), the lymphocyte to C-reactive protein ratio (LCR), left circumflex artery (LCX) stenosis > 50%, and Killip class > II⁹. Xie et al. conducted a study of 600 STEMI patients, finding that the prognostic nutritional index (PNI) is an independent predictor of NOAF in STEMI patients during hospitalization after PCI¹². Despite these findings, the precise mechanisms behind NOAF remain elusive¹³, necessitating an efficient predictive tool for its occurrence post-AMI.

Insulin resistance (IR) is a pathological state where cells fail to respond adequately to insulin. In clinical practice, IR is associated with cardiovascular disease (CVD)¹⁴. The triglyceride-glucose (TyG) index is a recognized reliable and convenient marker of IR¹⁵. Previous research has demonstrated that TyG index’s close association with the recurrence of atrial fibrillation pyanjiuost-radiofrequency catheter ablation, suggesting its potential as an effective predictor of post-infarction atrial fibrillation¹⁶. TyG index plays a key role in predicting CVD, cardiovascular mortality, stroke, hypertension^17–19. To date, the predictive value of the TyG index for atrial fibrillation post-infarction has not been confirmed.

In recent years, the integration of artificial intelligence with medicine has enhanced the use of predictive modeling and other statistical tools in clinical settings, improving disease prediction accuracy. As a graphical form of a clinical prediction model, the Nomogram can quantitatively calculate the probability of an event’s occurrence. Due to its accuracy and intuitiveness, it is used in clinical diagnosis and treatment. Previously, TyG was not included as an important variable in the Nomogram to predict the likelihood of NOAF in patients with AMI after PCI. This study was the first to build a nomogram using clinical data from patients with AMI at our center to explore the impact of the TyG index and other novel indicators on the incidence of NOAF in patients with primary AMI. This study will find high-risk individuals and improve treatment outcomes for patients.

Methods

Study population

We enrolled 707 consecutive AMI patients > 18 years old who were admitted to the Department of Cardiology at the First Hospital of Jilin University from February 2018 to February 2024. The included patients met the diagnostic criteria for AMI established by the European Society of Cardiology/American College of Cardiology²⁰. Inclusion criteria were: (1) age > 18 years and (2) diagnosed with AMI following PCI. Patients were excluded if they: (1) lacked crucial laboratory results; (2) had severe liver insufficiency or end-stage renal disease; (3) were undergoing thrombolytic therapy or emergent coronary artery bypass grafting (CABG); (4) had a history of AF or atrial flutter; (5) were diagnosed with hyperthyroidism or heart valve disease; or (7) had a history of PCI or AMI. Ultimately, 551 patients were included in the analysis (Fig. 1). Of these, 386 and 165 were allocated to the training and validation cohorts, respectively. The ethics committee of the The First Hospital of Jilin University approved the study, which adhered to the Declaration of Helsinki.

Fig. 1.

Flow chart of the study population. AF atrial fibrillation, SR sinus rhythm.

Definitions

Postoperative NOAF was defined as the occurrence of atrial fibrillation in acute myocardial infarction patients without prior history of the condition, following PCI. It was identified through episodes of AF lasting at least 30 s, detected by continuous telemetry, 12-lead electrocardiogram (ECG), or Holter monitoring during hospitalization. The Killip class was defined as follows: Class I, no signs of heart failure; Class II, rales present in the lungs but covering less than half of the lung field; Class III, rales covering more than half of the lung field; and Class IV, cardiogenic shock with varying degrees of hemodynamic change²¹. There are two equations in our article, such as TyG index and SIRI. The TyG index, a composite value of triglyceride (TG) and fasting blood glucose (FBG) levels, was calculated as ln(TG(mg/dl)×FBG(mg/dl)/2)²². The Systemic Inflammatory Response Index (SIRI) is defined as the product of neutrophils and monocytes divided by lymphocytes²³. SIRI = (neutrophils ×monocytes) / lymphocytes. Coronary artery stenosis was defined as ≥ 50% stenosis in any coronary artery (including the left main artery, left anterior descending branch, left circumflex branch, and right coronary artery) as demonstrated by coronary angiography.

Data collection

We collected clinical information from patient’s medical records, including demographic data (age, sex, heart rate, systolic blood pressure, diastolic blood pressure, smoking and drinking status, and medication use); electrocardiogram (ECG) results; laboratory parameters; angiographic findings; echocardiography results (left atrial (LA) diameter, left ventricular (LV) diameter, and left ventricular ejection fraction (LVEF)); and presence of comorbidities (hypertension, diabetes, history of stroke). Laboratory test results included leukocytes, neutrophils, monocytes, lymphocytes, hemoglobin, serum albumin, fasting blood glucose, total cholesterol, triglycerides, low-density lipoprotein cholesterol, high-density lipoprotein cholesterol, creatinine, uric acid, B-type natriuretic peptide (BNP), and cardiac troponin I (cTnI).

Statistical analysis

AMI patients were randomly divided into a training dataset (70%) and a validation dataset (30%). The training dataset was used to construct the nomogram and the validation dataset served for external validation. The Kolmogorov-Smirnov test assessed the normal distribution of continuous variables, which were described as mean ± standard deviation; non-normally distributed variables were described as median (interquartile range (IQR)) and analyzed using the Mann-Whitney U test. Categorical variables were presented as frequency and assessed with the χ² test. We used software xtitle to select an appropriate TyG index cutoff value. Univariate and multivariate logistic regression models were employed to explore the relationship between the variables and NOAF. Variables identified as potential risk factors in the univariate analysis underwent multicollinearity analysis; those with tolerances < 0.2 or a variance inflation factor (VIF) > 5 were refined for multivariate analysis based on expert judgement to identify independent risk factors for NOAF post-AMI. R version 4.1.0 was employed to construct the nomogram. Using atrial fibrillation as the outcome, the predictive model was developed and evaluated by plotting the receiver operating characteristic (ROC) curve and calculating the area under the curve (AUC) to assess the model’s predictive value. The model’s accuracy was further evaluated using a decision curve analysis (DCA) and clinical impact curves (CICs). Statistical analyses were performed using R statistical software (version 4.0.2), and a two-sided P-value < 0.05 was considered statistically significant.

Results

We enrolled 551 AMI patients without a previous history of AF, including 457 with SR and 94 who developed AF during hospitalization after PCI. The incidence of NOAF was 17.1%. The baseline characteristics are detailed in Table 1. In the SR group, there were 31 patients with elevated serum creatinine, 178 with higher BNP levels, and 41 with increased D-dimer levels; corresponding figures in the NOAF group were 19, 51, and 15, respectively. Among NOAF patients, 64 (68.09%) had left circumflex artery stenosis > 50%, compared to 241 in the SR group. NOAF patients were generally older, had higher heart rates, larger LA diameters, and elevated levels of fasting blood glucose, white blood cell and neutrophil counts, uric acid, aspartate aminotransferase (AST), alanine aminotransferase (ALT), HbA1C, TyG index, and SIRI. Conversely, they had significantly lower admission systolic and diastolic blood pressure, albumin levels, and low-density lipoprotein cholesterol. LVEF was lower in the NOAF group compared to the SR group.

Table 1.

The baseline characteristics of the patients.

Variables	SR (N = 457)	NOAF (N = 94)	P value
Demographic data
Age (years)	58.83 ± 10.61	65.71 ± 10.91	< 0.001
Males (n%)	323 (70.68)	61 (64.89)	0.266
SBP (mmHg)	131.00 (120.00–146.00)	123.00 (107.00–135.00)	< 0.001
DBP (mmHg)	80.00 (70.00–87.00)	68.50 (63.00–82.00)	< 0.001
HR (bpm)	78.00 (70.00–84.00)	80.00 (74.00–92.00)	0.002
Smoking (n%)	194 (42.45)	42 (44.68)	0.691
Alcohol intake (n%)	88 (19.26)	12 (12.77)	0.137
Drug use in hospital
ACEI/ARB (n%)	240 (52.52)	47 (50.00)	0.656
Beta blocker (n%)	213 (46.61)	45 (47.87)	0.823
Antidiabetic drugs (n%)	125 (27.35)	30 (31.91%)	0.37
Laboratory data
FBG (mmol/L)	5.98 (5.21–7.81)	7.66 (6.17–10.59)	< 0.001
Albumin (g/L)	37.70 (35.20–40.00)	36.95 (34.82–39.45)	0.047
WBC (/L)	9.01 (7.25–11.26)	10.79 (8.62–13.77)	< 0.001
Neutrophil (/L)	6.50 (4.71–8.64)	8.91 (6.38–11.01)	< 0.001
Lymphocyte (/L)	1.64 (1.18–2.22)	1.48 (1.04–2.36)	0.3
Monocyte (/L)	0.53 (0.39–0.68)	0.58 (0.41–0.80)	0.064
Hemoglobin (g/L)	146.00 (135.00-158.00)	142.50 (129.75-156.75)	0.312
Platelet (/L)	231.00 (194.00-270.00)	226.50 (194.25–259.00)	0.561
Uric acid (µmol/L)	334.00 (275.00-402.00)	366.50 (303.00-452.00)	0.008
AST (U/L)	61.30 (28.20-139.60)	138.40 (51.25-310.35)	< 0.001
ALT (U/L)	28.30 (18.70–45.10)	40.60 (25.35–71.05)	< 0.001
cTnI (ng/mL)	2.26 (0.31–9.85)	4.75 (0.30–19.30)	0.153
Triglyceride (mmol/L)	1.67 (1.13–2.46)	1.77 (1.23–2.47)	0.344
TC (mmol/L)	5.01 (4.35–5.66)	4.72 (4.04–5.64)	0.075
LDL-C (mmol/L)	3.18 (2.69–3.67)	2.83 (2.37–3.57)	0.015
HDL-C (mmol/L)	1.08 (0.93–1.26)	1.07 (0.96–1.21)	0.997
HbA1C (%)	6.00 (5.60–7.30)	6.40 (5.82–7.70)	0.021
Creatinine (n%)	31 (6.78)	19 (20.21)	< 0.001
BNP (n%)	178 (38.95)	51 (54.26)	0.006
D-D (n%)	41 (8.97)	15 (15.96)	0.041
Echocardiography data
Lvef (%)	58.00 (55.00–61.00)	55.50 (47.25-57.00)	< 0.001
Left atrium diameter (mm)	34.00 (32.00–37.00)	37.00 (33.00–40.00)	< 0.001
Left ventricular diameter (mm)	49.00 (46.00–51.00)	49.50 (47.00-52.75)	0.041
TyG index	7.43 (6.99–7.95)	7.64 (7.32–8.23)	0.003
SIRI	2.05 (1.22–3.19)	3.00 (1.81–5.55)	< 0.001
AMI type, (n%)
STEMI (n%)	218 (47.70)	44 (46.81)	0.874
Comorbidities
Hypertension (n%)	221 (48.36)	52 (55.32)	0.219
Diabetes mellitus (n%)	132 (28.88)	33 (35.11)	0.23
Stroke (n%)	20 (4.38)	8 (8.51)	0.096
Killip ≥ III (n%)	95 (20.79%)	37 (39.36%)	< 0.001
Angiographic data: Coronary artery stenosis > 50%, (n%)
LM (n%)	7 (1.53)	2 (2.13)	0.678
LAD (n%)	318 (69.58)	71 (75.53)	0.249
LCX (n%)	241 (52.74)	64 (68.09)	0.006
RCA (n%)	263 (57.55)	60 (63.83)	0.26

SR sinus rhythm, NOAF new-onset atrial fibrillation, SBP systolic blood pressure, DBP diastolic blood pressure, ACEI angiotensin-converting enzyme inhibitor, ARB angiotensin receptor blocker, FBG fast blood glucose, WBC white blood cell, AST aspartate aminotransferase, ALT alanine aminotransferase, CTnI troponin I, TC total cholesterol, HDL-c high-density lipoprotein cholesterol, LDL-c low-density lipoprotein cholesterol, BNP B-type natriuretic peptide, D-D d-dimer, TyG index triglyceride-glucose index, Lvef left ventricular ejection fraction, SIRI systemic immune response index, STEMI ST-segment elevation myocardial infarction, LM left main artery, LAD left anterior descending coronary artery, LCX left circumflex coronary artery, RCA right coronary artery.

No significant differences were observed between the groups in terms of gender, history of hypertension, diabetes, stroke, smoking, alcohol consumption, drug use, levels of total cholesterol, triglycerides, platelet count, lymphocytes, monocytes, hemoglobin, high-density lipoprotein cholesterol, cTnI, AMI type, or the proportion of patients with stenosis > 50% in the left main, left anterior descending, or right coronary arteries.

There were 165 diabetic patients, with 132 in the NOAF group and 33 in the non-atrial fibrillation group. Regarding the use of anti-diabetic drugs, 155 patients were on medication, including 30 in the NOAF group and 125 in the sinus rhythm group.

The study comprised 386 patients in the training cohort and 165 in the validation cohort (Table 2).

Table 2.

Validation cohort and training cohort.

Characteristic	Validation cohort	Training cohort	P value
N	165	386
Age (years)	60.61 ± 10.95	59.74 ± 10.98	0.395
Males (n%)	119 (72.12%)	265 (68.65%)	0.417
SBP (mmHg)	130.00 (117.00-142.00)	130.00 (118.00-145.00)	0.821
DBP (mmHg)	78.00 (70.00–87.00)	79.00 (68.25-86.00)	0.728
HR (bpm)	78.00 (72.00–85.00)	78.00 (70.00–84.00)	0.035
Smoking (n%)	71 (43.03%)	165 (42.75%)	0.951
Alcohol intake (n%)	31 (18.79%)	69 (17.88%)	0.799
Drug use in hospital
ACEI/ARB (n%)	77 (46.67%)	210 (54.40%)	0.096
Beta blocker (n%)	77 (46.67%)	181 (46.89%)	0.961
Antidiabetic drugs (n%)	57 (34.55%)	98 (25.39%)	0.029
Laboratory data
FBG (mmol/L)	6.33 (5.28–8.86)	6.25 (5.25–8.13)	0.458
Albumin (g/L)	36.90 (34.50–39.50)	37.95 (35.30–40.00)	0.023
WBC (/L)	8.85 (7.42–11.38)	9.50 (7.43–11.82)	0.316
Neutrophil (/L)	6.52 (4.61–9.27)	7.02 (4.99-9.00)	0.465
Lymphocyte (/L)	1.48 (1.07–2.12)	1.71 (1.18–2.28)	0.026
Monocyte (/L)	0.52 (0.37–0.70)	0.54 (0.40–0.71)	0.57
Platelet (/L)	227.00 (189.00-260.00)	232.00 (195.00-271.00)	0.286
Mono	0.52 (0.37–0.70)	0.54 (0.40–0.71)	0.57
Hemoglobin (g/L)	146.00 (135.00-159.00)	146.00 (135.00-157.75)	0.893
Uric acid (µmol/L)	327.00 (271.00-402.00)	341.50 (282.25-407.75)	0.294
AST (U/L)	61.60 (27.50-155.80)	71.05 (30.25-172.35)	0.317
ALT (U/L)	27.10 (16.30–49.10)	31.05 (20.70–47.40)	0.067
cTnI (ng/mL)	2.03 (0.41–11.40)	2.59 (0.28–11.17)	0.978
Triglyceride (mmol/L)	1.66 (1.11–2.28)	1.71 (1.18–2.56)	0.357
TC (mmol/L)	4.64 (4.04–5.41)	5.08 (4.42–5.78)	< 0.001
LDL-C (mmol/L)	2.90 (2.52–3.47)	3.22 (2.70–3.71)	0.001
HDL-C (mmol/L)	1.06 (0.92–1.23)	1.08 (0.95–1.26)	0.402
HbA1C (%)	6.20 (5.70–7.50)	6.10 (5.62–7.30)	0.273
Creatinine (n%)	26 (15.76%)	36 (9.33%)	0.753
BNP (n%)	67 (40.61%)	162 (41.97%)	0.766
D-D (n%)	16 (9.70%)	40 (10.36%)	0.813
Echocardiography data
Lvef (%)	58.00 (56.00–61.00)	57.00 (54.00–60.00)	0.106
Left atrium diameter (mm)	35.00 (32.00–37.00)	34.00 (31.00–38.00)	0.707
Left ventricular diameter (mm)	48.00 (46.00–51.00)	49.00 (46.00–52.00)	0.432
TyG	7.40 (7.02–7.96)	7.48 (7.02–7.97)	0.618
SIRI	2.22 (1.27–4.05)	2.15 (1.31–3.42)	0.532
STEMI (n%)	87 (52.73%)	175 (45.34%)	0.112
Atrial_fibrillation	29 (17.58%)	65 (16.84%)	0.833
Comorbidities
Hypertension (n%)	78 (47.27%)	195 (50.52%)	0.485
Diabetes mellitus (n%)	59 (35.76%)	106 (27.46%)	0.051
Stroke (n%)	6 (3.64%)	22 (5.70%)	0.313
Killip ≥ III (n%)	32 (19.39%)	100 (25.91%)	0.101
Angiographic data: coronary artery stenosis > 50% (n%)
LM (n%)	3 (1.82%)	6 (1.55%)	0.823
LAD (n%)	113 (68.48%)	276 (71.50%)	0.476
LCX (n%)	95 (57.58%)	210 (54.40%)	0.493
RCA (n%)	90 (54.55%)	233 (60.36%)	0.204

Several risk factors, including age, systolic and diastolic blood pressures, heart rate, fasting blood glucose, white blood cell and neutrophil counts, albumin, uric acid, AST, ALT, monocytes, D-dimer, serum creatinine, LA diameter, Lvef, left circumflex artery stenosis > 50%, Killip ≥ III, TyG index, and SIRI, were identified as potentional predictors of NOAF (all P values < 0.05) (Table 3). These variables were then subjected to multivariable logistic backward stepwise regression analysis to identify independent risk factors for NOAF. Multivariate analyses revealed that age (OR: 1.065; 95% CI: 1.032–1.099), serum creatinine (OR: 2.584; 95% CI: 1.095–6.097), TyG index (OR: 1.981; 95% CI: 1.344–2.92), LA diameter (OR: 1.118; 95% CI: 1.054–1.186), and SIRI (OR: 1.102; 95% CI: 1.021–1.19) were independent predictors of NOAF (Table 4). These variables were used to construct a novel nomogram (Fig. 2). The ROC curve demonstrated good predictive performance, with an AUC of 0.780 (95% CI: 0.538–0.888) in the training cohort and 0.804 (95% CI: 0.690–0.838) in the validation cohort (Fig. 3). The calibration curve showed a high concordance between predicted and observed risks of NOAF, with an average absolute error of 0.019 in the training cohort and 0.029 in the validation cohort (Figs. 4, 5). Decision curve analysis (DCA) revealed that the nomogram provided significant net clinical benefits (Figs. 6, 7), and the clinical impact curve (CIC) confirmed the model’s high clinical efficacy.

Table 3.

Potential clinical predictors for NOAF after AMI univariate logistic regression analysis.

Characteristics	OR	95% CI	P-value
Age (years)	1.059	1.03–1.088	0
Sex (n%)	0.948	0.536–1.677	0.855
SBP (mmHg)	0.978	0.964–0.991	0.001
DBP (mmHg)	0.964	0.943–0.985	0.001
HR (bpm)	1.047	1.025–1.07	0
Smoking (n%)	1.181	0.692–2.017	0.543
Alcohol intake (n%)	0.598	0.272–1.318	0.203
ACEI/ARB (n%)	1.027	0.602–1.754	0.921
Beta blocker (n%)	1.117	0.654–1.907	0.687
Antidiabetic drugs (n%)	1.162	0.62–2.181	0.639
FBG (mmol/L)	1.146	1.059–1.239	0.001
Albumin (g/L)	0.907	0.846–0.974	0.006
WBC (/L)	1.211	1.121–1.307	0
Neutrophil (/L)	1.224	1.129–1.326	0
Lymphocyte (/L)	0.985	0.715–1.355	0.924
Monocyte (/L)	2.823	1.234–6.455	0.014
Hemoglobin (g/L)	0.994	0.979–1.01	0.459
Platelet (/L)	1	0.997–1.004	0.849
Uric acid (µmol/L)	1.004	1.002–1.006	0.001
AST(U/L)	1.004	1.002–1.006	0
ALT(U/L)	1.024	1.014–1.034	0
cTnI (ng/mL)	1.018	0.998–1.038	0.076
Triglyceride (mmol/L)	1.538	1.102–2.146	0.011
TC (mmol/L)	0.869	0.675–1.119	0.276
LDL-C (mmol/L)	0.754	0.537–1.058	0.103
HDL-C (mmol/L)	0.724	0.26–2.019	0.538
HbA1C (%)	1.132	0.956–1.34	0.148
Creatinine (n%)	3.731	1.792–7.765	0
BNP (n%)	1.654	0.968–2.824	0.066
D-D (n%)	2.722	1.318–5.622	0.007
Lvef (%)	0.928	0.897–0.959	0
Left atrium diameter (mm)	1.137	1.077–1.202	0
Left ventricular diameter (mm)	1.054	0.994–1.118	0.075
TyG index	1.538	1.102–2.146	0.011
SIRI	1.128	1.051–1.211	0.001
STEMI (n%)	0.77	0.447–1.324	0.344
Hypertension (n%)	1.174	0.687–2.005	0.556
Diabetes mellitus (n%)	1.014	0.559–1.84	0.963
Stroke (n%)	1.939	0.729–5.155	0.185
Killip ≥ III (n%)	2.617	1.5–4.566	0.001
LM stenosis > 50% (n%)	0	0–Inf	0.988
LAD stenosis > 50% (n%)	0.958	0.533–1.721	0.886
LCX stenosis > 50% (n%)	1.956	1.113–3.44	0.02
RCA stenosis > 50% (n%)	1.463	0.832–2.573	0.187

Table 4.

Independent clinical predictors for NOAF after AMI multivariate analysis.

Characteristics	OR	CI	P
Age (years)	1.065	1.032–1.099	0
Creatinine (n%)	2.584	1.095–6.097	0.03
Left atrium diameter (mm)	1.118	1.054–1.186	0
TyG index	1.981	1.344–2.92	0.001
SIRI	1.102	1.021–1.19	0.013

Fig. 2.

Nomogram for calculating risk score and predicting the incidence of new-onset atrial fibrillation (NOAF) in acute myocardial infarction (AMI) patients.

Fig. 3.

Receiver operating characteristic (ROC) curve for the nomogram to predict the incidence of new-onset atrial fibrillation (NOAF) in patients with acute myocardial in the training cohort and validation cohort.

Fig. 4.

Calibration curve for the nomogram to predict the incidence of new-onset atrial fibrillation (NOAF) in patients with acute myocardial infarction (AMI) in training cohort.

Fig. 5.

Calibration curve for the nomogram to predict the incidence of new-onset atrial fibrillation (NOAF) in patients with acute myocardial infarction (AMI) in validation cohort.

Fig. 6.

Decision curve analysis (DCA) for the incidence of new-onset atrial fibrillation (NOAF) in patients with acute myocardial infarction (AMI) in the training cohort, demonstrating the net benefit of using the nomogram.

Fig. 7.

Decision curve analysis (DCA) for the incidence of new-onset atrial fibrillation (NOAF) in patients with acute myocardial infarction (AMI) in the validation cohort, demonstrating the net benefit of using the nomogram.

Discussion

Our study established and validated a predictive model for NOAF in 551 AMI patients post-PCI, identifying age, serum creatinine, the TyG index, LA diameter, and the SIRI as independent risk factors. We developed a nomogram model using these predictors, which demonstrated good calibration, predictive performance, and clinical utility.

IR is known to be associated with several diseases, including hypertension and CAD^19,24. The TyG index serves as a reliable surrogate marker of IR¹⁵. Research has shown that the TyG index is positively associated with metabolic risk factors and cardiovascular outcomes. For instance, Xin et al. pointed out that both the baseline TyG index and the high trajectories of TyG index growth were associated with the risk of hypertension²⁵. Previous study indicated that people who are over 45 years old, with higher baseline TyG index and an increasing trend in the TyG index may have increased risk of stroke²⁶. Additionally, a systematic review and meta-analysis concluded that the TyG index played a significant role in assessing the risk of heart failure (HF) incidence and adverse outcomes in different populations²⁷. Zhao et al. found that the TyG index was positively associated with higher incidence of chest pain and all-cause mortality not only in participants with chest pain but also in those without chest pain²⁸. Moreover, Xue et al. carried out a study of 1,727 adults to validate that TyG index-related parameters (TyG-WHtR, TyG-BMI, and TyG-WC) had high clinical diagnostic values in NAFLD, MAFLD, liver fibrosis, and moderate-to-advanced fibrosis²⁹. Clinical research revealed that the TyG index was positively associated with the impaired cardiovascular fitness in non-diabetic young populations, especially in males³⁰. Previous literature demonstrated that the TyG index played a significant role in predicting the normalization of blood glucose status in Chinese prediabetes populations. When TyG is above 8.88, there was a negative correlation between TyG and the conversion of glucose status from prediabetes to normoglycemia³¹.Santulli et al. demonstrated TyG index was a risk factor in the development of cognitive and physical impairments, among elders with prefrail hypertension³². A study validated that higher TyG index had a higher prevalence of symptomatic CAD patients. The application of the TyG index in cardiovascular disease (CVD) patients is susceptible to bias from diabetes and hyperlipidemia. Thus, these factors must be well managed to validate its role as a biomarker. It is imperative not to deduce reverse causality when employing the TyG index for CVD patients³³. Li et al. analyzed 1516 patients with symptomatic CAD at Tianjin Union Medical Center from January 2016 to December 2022, finding that higher TyG index values were predictive of a high rate of coronary lesions and plaques³⁴. The TyG index has also been recognized as an effective indicator of postoperative NOAF following septal myectomy³⁵. Furthermore, Wang et al. conducted a retrospective survey of 2242 AF patients over a one-year follow-up period, from June 2018 to January 2022, at two hospitals in China, concluding that the TyG index could be independently associated with AF recurrence following ablation³⁶. Our study showed that the TyG index was positively associated with NOAF in patients with AMI. Several mechanisms may explain the correlation between the TyG index and AF. Primarily, insulin resistance (IR) can lead to excessive lipid accumulation in cardiomyocytes, resulting in a phenomenon known as “cardiotoxicity”. This accumulation causes cellular dysfunction, cardiomyocyte apoptosis, and impaired myocardial metabolism, potentially altering the function and structure of cardiomyocytes and increasing the risk of AF³⁷. IR also promotes CaMKIIδ oxidation, leading to abnormal intracellular calcium homeostasis and atrial structural remodeling³⁸. Animal studies have demonstrated that IR impairs the transport and expression of the major myocardial isoform of the glucose transporter protein (GLUT) 4 and the novel subtype GLUT8, heightening susceptibility to AF. IR is associated with various aspects of left atrial (LA) remodeling, including increased oxidative stress, elevated expression of hyperphosphorylated calcium-related proteins, and cardiac fibrosis³⁹. Increased oxidative stress and inflammation related to impaired IR and insulin secretion contribute to atrial electrical remodeling, LA fibrosis, and the formation of LA low-pressure areas¹⁶. Chen et al. found that an elevated TyG index was an independent risk factor for AF among non-diabetic patients but was not associated with AF in diabetic patients. The main reason is that the development of AF increases when the insulin resistance homeostasis model (HOMA-IR) level is between 1 and 2.5, and then it reaches a plateau. A nonlinear relationship exists between IR and AF in non-diabetic patients. In diabetic patients, a higher level of IR does not correlate with AF⁴⁰. A previous study demonstrated that the TyG index serves as a surrogate marker of IR in patients with AF⁴¹.

At multivariate analysis, age was an independent risk factor for NOAF in our study. Sarduhave et al. identified advancing age as an independent risk factor for the recurrence of AF within 1 Year of Catheter Ablation⁴². With aging, the myocardium undergoes anatomical and electrophysiological changes, including the loss of lateral electrical connections between myofibers and reduced electrical conduction in the sinoatrial node, atrioventricular node, and atria⁴³. Left atrial diameter has been recognized as an independent predictor for NOAF, with increased left atrial diameter indicating progressive dilatation and remodeling of the left atrial myocardium, which serves as a substrate for AF initiation and maintenance⁴⁴.

As we know, inflammation plays a role in the development and prognosis of AMI, with certain inflammatory markers, such as C-reactive protein and the SII, correlated with both short- and long-term prognosis of myocardial infarction^45,46. Previous studies have shown that the SIRI can predict prognosis in cardiovascular diseases^47,48. Furthermore, a recent retrospective study of 526 ischemic stroke patients confirmed that SIRI is an independent predictor of AF⁴⁸. SIRI encompasses neutrophils, lymphocytes, and monocytes. Monocyte aggregation leads to the production of inflammatory factors, such as interleukin (IL)-6 and IL-1β, which contribute to atrial structural and electrical remodeling, ultimately leading to atrial fibrillation⁴⁹. Additionally, neutrophils can produce proinflammatory cytokines, proteases, peroxidases, and reactive oxygen species, which may result in atrial fibrosis⁵⁰. Finally, T cells and B cells, which are lymphocytes, play roles in AF—T cells primarily regulate the innate immune response, and B cells may affect the condition by secreting autoantibodies. Concerning creatinine, renal dysfunction is associated with both electrical and structural remodeling of the LA, potentially leading to new-onset POAF⁵¹.

We constructed a new nomogram that includes the TyG index, SIRI, creatinine, age, and LA diameter. Our model demonstrates high clinical predictive performance and is a practical fit. The nomogram can identify patients who are at high risk of NOAF, facilitating early intervention in these patients to improve clinical outcomes.

Serum album, D-dimer, blood pressure, monocyte count, Killip ≥ III, and left circumflex artery stenosis > 50% were included in the univariate analysis; however, none were incorporated into our model.

Previous studies validated that serum albumin levels are a critical independent marker for AF recurrence following pulmonary vein isolation (PVI) ablation, with lower levels associated with significantly more recurrences⁵². In our study, serum albumin was negatively associated with NOAF. D-dimer was positively associated with NOAF in our study. A clinical study identified elevated D-dimer levels as an independent factor for NOAF in patients with AMI⁵³.

The Killip class assesses cardiac function, with grades III and IV indicating worse cardiac function than grades I and II. In our study, Killip ≥ III was positively related to NOAF⁹. Earlier research identified Killip ≥ III as a strong risk factor for NOAF. A correlation between left circumflex artery stenosis > 50% and NOAF was observed in our study. Several reasons may explain this association. Significant stenosis can lead to temporary or permanent mitral valve insufficiency, burdening the⁵⁴. Left circumflex artery occlusion may affect the LA and promote atrial ischemia, but the mechanism requires further exploration⁹.

Regarding monocyte count, it was identified as a risk factor in our study. Ren et al. conducted a study on 91 AF patients, revealing that monocyte count is an important factor mediating inflammation and oxidative stress reactions and an independent predictor of AF recurrence⁵⁵.

Prior study demonstrated that elevated blood pressure and increased the risk of atrial fibrillation and there was some increase in risk even within the normal range of systolic and diastolic blood pressure⁵⁶. In our ariticle, both systolic and diastolic blood pressure were negatively correlated with NOAF, with higher pressures observed in the sinus rhythm (SR) group. Patients in the SR group had higher blood pressure for several reasons. Firstly, prior studies demonstrated that patients with NOAF had lower blood pressure than those with sinus rhythm. Secondly, Low blood pressure in patients with AMI may indicate that the heart’s pumping function is weakened.Once the pumping function of the heart is weakened, it will lead to insufficient blood perfusion throughout the body, especially to the heart itself, resulting in atrial ischemia. Atrial ischemia is an important factor in the occurrence of atrial fibrillation, as ischemia can lead to prolonged atrial conduction time and increased conduction heterogeneity, thereby increasing the susceptibility and perpetuation of atrial fibrillation^57,58.

Previous literature indicated that in patients with diabetes complicated by myocardial infarction, Sodium-glucose co-transporter 2 inhibitors (SGLT2-i) can significantly reduce the incidence of atrial fibrillation, ventricular tachycardia and ventricular fibrillation during hospitalization⁵⁹. In this paper, we did not reach this conclusion. Our sample size is relatively small, so high-quality and large sample studies are needed to further study whether there is an intrinsic relationship between the SGLT2-i and arrhythmic events .

Limitations

This study has several limitations. Firstly, it is a single-center study with a small sample size, which may not represent the entire patient population. Therefore, our model should be validated in a larger cohort. Secondly, as a retrospective observational study without randomization or intervention, it is not possible to eliminate all unmeasurable confounding factors, introducing potential bias. Thirdly, some patients were not included because of missing data, which resulted in a selection bias. But the bias has little impact on the results. Moreover, although we have adjusted for confounding factors, some risk factors that were not included in our study may also play an important role. Due to the lack of long-term follow-up, we have not been able to study the relationship between risk factors and long-term prognosis. In subsequent work, we will supplement the data to make the article more comprehensive. However, the study still offers significant advantages, such as investigating new-onset atrial fibrillation following PCI in patients with acute myocardial infarction and introducing the relatively novel TyG indicators. Our article is the first to incorporate the TyG index into a predictive model for NOAF after AMI, and this model demonstrates good discrimination and clinical utility. This approach enables early detection of atrial fibrillation high-risk individuals and the implementation of measures to reduce the occurrence of adverse events.

Conclusion

In our study, 94 AMI patients developed NOAF during hospitalization after PCI. Multivariate analysis showed that age (OR: 1.065; 95% CI: 1.032–1.099), serum creatinine (OR: 2.584; 95% CI: 1.095–6.097), TyG index (OR: 1.981; 95% CI: 1.344–2.92), LA diameter (OR: 1.118; 95% CI: 1.054–1.186), and SIRI (OR: 1.102; 95% CI: 1.021–1.19) were independent predictors of NOAF (Table 4). The prediction model yielded an area under the curve (AUC) of 0.780 (95% CI: 0.358–0.888). Our data indicate that the nomogram effectively identifies high-risk patients for atrial fibrillation and demonstrates relatively good discrimination, calibration, and clinical utility. We hope that our model will bring convenience to the clinic work. However, due to the inherent limitations in our article, further research with larger samples and high-quality literature is still required to confirm the validity of our conclusions.

Author contributions

W.X.D wrote the main manuscript text and Z.W. prepared Figs. 2, 3, 4, 5, 6 and 7. W.Q.W prepared Tables 1, 2, 3 and 4. Y.X.Y is charge of data curation. W.J.Y. and Y.S. are charge of methodology. T.Q. reviewd the manuscript.

Data availability

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

Declarations

Competing interests

The authors declare no competing interests.

Ethical approval

The study was approved by the ethics committee of the The First Hospital of Jilin University (ethical approval number: 2024-KS-207).

Informed consent

The studies involving humans were approved by the Medical Ethics Committee of the The First Hospital of Jilin University. The studies were conducted in accordance with the local legislation and institutional requirements. The participants provided their written informed consent to participate in this study.