The C-reactive protein–albumin–lymphocyte index as a predictor of new-onset atrial fibrillation after ST-segment elevation myocardial infarction: A retrospective cohort study

Abstract

Objective

New-onset atrial fibrillation is a common complication following ST-segment elevation myocardial infarction and is associated with poor prognosis. The C-reactive protein–albumin–lymphocyte index, a composite biomarker reflecting inflammation, nutritional status, and immune function, has not been fully evaluated for its predictive value for new-onset atrial fibrillation.

Methods

This single-center retrospective study included 665 consecutive patients with ST-segment elevation myocardial infarction undergoing primary percutaneous coronary intervention between January 2023 and October 2025. The C-reactive protein–albumin–lymphocyte index was calculated from admission laboratory values. New-onset atrial fibrillation was defined as the first documented episode of atrial fibrillation within 30 days after percutaneous coronary intervention. Multivariable logistic regression, receiver operating characteristic curve analysis, and reclassification metrics were used to assess its predictive performance.

Results

New-onset atrial fibrillation occurred in 56 patients (8.4%). In the multivariable model adjusting for age, heart rate, left atrial diameter, infarct-related artery-right coronary artery, left ventricular ejection fraction, fasting blood glucose, and C-reactive protein–albumin–lymphocyte index, five variables remained independently associated with new-onset atrial fibrillation: left atrial diameter (odds ratio = 1.179), infarct-related artery-right coronary artery (odds ratio = 1.992), left ventricular ejection fraction (odds ratio = 0.933), fasting blood glucose (odds ratio = 1.321), and C-reactive protein–albumin–lymphocyte index (odds ratio = 0.813) (all p < 0.05). Among individual biomarkers, the C-reactive protein–albumin–lymphocyte index demonstrated the highest predictive ability (area under the curve = 0.743; 95% confidence interval: 0.685–0.801). Adding the C-reactive protein–albumin–lymphocyte index to a baseline model comprising age, heart rate, left atrial diameter, infarct-related artery-right coronary artery, left ventricular ejection fraction, and fasting blood glucose significantly improved reclassification (net reclassification improvement = 0.7323; p < 0.001; integrated discrimination improvement = 0.0593; p = 0.005).

Conclusions

In this retrospective single-center cohort, a lower admission C-reactive protein–albumin–lymphocyte index was independently associated with clinically documented 30-day new-onset atrial fibrillation after percutaneous coronary intervention in patients with ST-segment elevation myocardial infarction and may provide incremental predictive information, whereas external validation is required before clinical application.

Introduction

ST-segment elevation myocardial infarction (STEMI) remains a leading cause of sudden cardiac death and cardiovascular mortality worldwide.^1,2 Although timely percutaneous coronary intervention (PCI) effectively restores perfusion to the infarct-related artery and significantly reduces in-hospital mortality, postprocedural complications remain a significant determinant of long-term prognosis.^2,3 New-onset atrial fibrillation (NOAF) is a common complication in patients with STEMI, occurring in approximately 6.3%–8.0% of patients.^4–6 As a prevalent and clinically significant arrhythmia, NOAF can induce rapid ventricular rates, compromise hemodynamic stability, precipitate or exacerbate heart failure, and substantially increase the risk of long-term mortality and major adverse cardiovascular events.^7–9 Therefore, early identification of patients at high risk for NOAF and the implementation of timely preventive strategies are of critical clinical importance.

Growing evidence underscores the central role of systemic inflammation and metabolic dysfunction in the pathophysiology of NOAF.^5,10,11 Elevated inflammatory markers, such as high-sensitivity C-reactive protein (hs-CRP), are closely associated with the development and maintenance of atrial fibrillation (AF). ¹² Inflammatory cell infiltration promotes atrial fibrosis and electrical remodeling, which are key pathophysiological mechanisms of NOAF. ¹³ Serum albumin, which is synthesized by the liver, often decreases during acute myocardial infarction and chronic wasting diseases.^14,15 This reflects both acute inflammatory responses and states of nutritional depletion. Importantly, hypoalbuminemia is independently associated with major adverse cardiovascular events (MACE) and AF recurrence after catheter ablation.^14,15

However, single inflammatory biomarkers have shown limited predictive utility in clinical practice.^16–18 Composite indices, such as the C-reactive protein-to-albumin ratio (CAR) and the lymphocyte-to-CRP ratio, have demonstrated associations with NOAF in patients,^16–18 with STEMI; however, their predictive performance remains moderate, with the area under the receiver operating characteristic curve typically ranging between 0.65 and 0.75.^16–18 This indicates the necessity of developing more comprehensive prognostic biomarkers.

The C-reactive protein–albumin–lymphocyte index (CALLY) index—calculated from CRP, albumin, and lymphocyte counts—represents a novel composite marker that simultaneously reflects systemic inflammation, nutritional status, and immune response.^19–22 Previous studies have linked the CALLY index to clinical outcomes in cardiovascular disease.^19–22 For instance, Deng et al.²² reported its association with AF recurrence after radiofrequency ablation. Additionally, the index has been correlated with hospital length of stay in patients with heart failure and the incidence of major adverse cardiovascular and renal events in patients with STEMI.^19–22 Nevertheless, its predictive value for NOAF following PCI in patients with STEMI has not been established. This study aims to evaluate the value of the CALLY index in predicting the occurrence of NOAF in patients with STEMI.

Methods

Study population

This single-center retrospective study enrolled consecutive patients undergoing PCI for STEMI at the Second Affiliated Hospital of Soochow University between January 2023 and October 2025. All patients were treated according to current STEMI guidelines. The inclusion criteria were as follows: age ≥18 years; confirmed STEMI (typical chest pain ≥20 min, ST-segment elevation ≥0.1 mV in ≥2 contiguous leads, and elevated troponin); and successful PCI. The exclusion criteria included severe hepatic or renal dysfunction (eGFR <30 mL/min/1.73m² or Child-Pugh C), active infection, autoimmune disease, malignancy, thyroid dysfunction, severe valvular heart disease or congenital heart disease, prior AF, and missing key data. The Institutional Review Board of the Second Affiliated Hospital of Soochow University approved the protocol (JD-HG-2025-105). Because this was a retrospective analysis of deidentified clinical data with no additional patient interventions, the requirement for written informed consent was waived. The study was conducted in accordance with the principles set forth in the Declaration of Helsinki of 1975, as revised in 2024. After screening, 665 patients were included in the analysis (Figure 1).

Figure 1.

Study flowchart showing the screening and enrollment process of STEMI patients undergoing primary PCI.

PCI: percutaneous coronary intervention; STEMI: ST-segment elevation myocardial infarction.

PCI procedure

After standard antiplatelet loading (aspirin 300 mg plus clopidogrel 300 mg or ticagrelor 180 mg), arterial access was obtained via the radial or femoral approach. Coronary angiography identified the infarct-related artery, with lesion severity assessed by quantitative coronary angiography. Balloon angioplasty was performed first to restore flow, followed by stenting (preferably with drug-eluting stents) to achieve optimal revascularization (residual stenosis <20% and thrombolysis in myocardial infarction (TIMI) grade 3 flow). Postprocedural anticoagulation with unfractionated heparin or bivalirudin was administered according to protocol, and long-term dual antiplatelet therapy was initiated.

Data collection

Data were obtained from standardized case report forms and electronic medical records. Collected information included demographic data (age, sex, height, weight, and smoking history) and past medical history (hypertension, diabetes, stroke, heart failure, etc.), with all diagnoses adhering to relevant international clinical guidelines.

Fasting venous blood samples were obtained after an overnight fast (≥8 h) for routine laboratory assessments, including lipid profile (total cholesterol, triglycerides, LDL-C, and HDL-C) and fasting blood glucose (FBG). All other laboratory parameters—including high-sensitivity C-reactive protein (hs-CRP), serum albumin, lymphocyte count, estimated glomerular filtration rate (eGFR), and cardiac injury markers (high-sensitivity troponin T (hs-TnT) and N-terminal pro-B-type natriuretic peptide (NT-proBNP))—were derived from a single venous blood sample drawn immediately upon emergency department admission, after the electrocardiographic diagnosis of STEMI was confirmed, and, critically, before any therapeutic intervention (including PCI, anticoagulation, or antiplatelet administration). Therefore, the three components of the CALLY index (hs-CRP, albumin, and lymphocyte count) were all measured from this same preintervention blood draw, ensuring that they accurately reflect the patient’s baseline inflammatory, nutritional, and immune status at the time of initial presentation, unaffected by postintervention physiological alterations. Of note, because fasting status cannot be reliably ascertained in the acute STEMI setting, the term “fasting” applies only to the lipid profile and FBG and not to the emergency blood tests (hs-CRP, albumin, lymphocyte count, eGFR, hs-TnT, and NT-proBNP).

Within 24 h postoperatively, cardiac structure and function were assessed via transthoracic echocardiography. Left ventricular ejection fraction (LVEF) was measured using the two-dimensional Simpson method. Left atrial diameter (LA-D) was measured in the parasternal long-axis view at end-systole.

Concurrently, prehospitalization medication history (antiplatelet agents, β-blockers, statins, ACEI/ARB, etc.), coronary angiography findings (distribution of infarct-related arteries), and Killip cardiac function classification were recorded. The CALLY index was calculated as follows: ALB (g/L) × lymphocyte count (10⁹/L) ÷ (CRP (mg/L) × 10).

All patient details, including direct identifiers and potentially identifying information in images, were deidentified to ensure that patients could not be identified.

Ascertainment of NOAF. During hospitalization, all patients underwent continuous bedside electrocardiographic monitoring from admission until discharge. Standard 12-lead electrocardiograms were obtained at admission and after PCI, with additional electrocardiographic examinations performed in patients with arrhythmia-related symptoms or an abnormal rhythm detected by bedside monitoring. Suspected episodes were confirmed using a 12-lead electrocardiogram or 24-h Holter monitoring.

After discharge, patients were followed for 30 days through outpatient visits and/or structured telephone interviews. Information regarding palpitations, dyspnea, dizziness, syncope, emergency department visits, rehospitalization, and newly diagnosed AF was collected. Available outpatient and hospitalization records were also reviewed. Patients with suspected arrhythmic symptoms underwent further electrocardiographic or Holter evaluation.

NOAF was defined as the first documented episode of AF lasting at least 30 s within 30 days after PCI in a patient without a known history of AF.

The reporting of this study conforms to the STROBE guidelines. ²³

Statistical analyses

Data analyses were conducted using SPSS 27.0 (IBM Corp., Armonk, NY, USA) and R 4.3.1 (R Foundation for Statistical Computing, Vienna, Austria). Graphical visualizations were generated using GraphPad Prism 10.0 (GraphPad Software, San Diego, CA, USA). A two-sided P < 0.05 was considered statistically significant.

Continuous variables were assessed for normality using the Kolmogorov–Smirnov test. Data are presented as mean ± SD for normally distributed variables or median (interquartile range (IQR)) for non-normally distributed variables. Categorical variables are expressed as frequencies and percentages. Group differences between patients with and without NOAF were evaluated using the independent-samples t-test or Mann–Whitney U test for continuous variables, and the χ² test or Fisher's exact test for categorical variables, as appropriate.

Age and heart rate were forced into the multivariable model regardless of their univariable significance, based on their established clinical relevance. The remaining candidate variables with P < 0.05 in the univariable analyses were entered into the multivariable model. The CALLY index, as the primary exposure of interest, was included in the multivariable model irrespective of its univariable p value. All selected variables were entered simultaneously using the Enter method.

For clarity, two related models were distinguished. Model 2 included age, heart rate, LA-D, IRA-RCA, LVEF, FBG, and the CALLY index and was used to assess the independent association between CALLY and NOAF. Model 1 included the same six clinical covariates without CALLY (age, heart rate, LA-D, IRA-RCA, LVEF, and FBG). The CALLY-augmented prediction model was obtained by adding the CALLY index to this baseline model. Incremental discrimination and reclassification were evaluated by comparing these two prediction models using the DeLong test, continuous net reclassification improvement (NRI), and integrated discrimination improvement (IDI).

Internal validation using 1000 bootstrap resamples was performed specifically for the CALLY-augmented prediction model. Model performance was assessed using the optimism-corrected area under the curve (AUC), calibration intercept and slope, Brier score, and bootstrap-corrected calibration curve.

An exploratory sex-disaggregated analysis was performed to evaluate potential effect modification. The distributions of the CALLY index and its components were compared between sexes. A multiplicative interaction term (sex $\times$ CALLY) was introduced into the fully adjusted logistic regression model. Sex-specific odds ratios (ORs) were derived from this interaction model. Given the limited number of NOAF events in the female subgroup, these analyses were interpreted as exploratory.

Results

Patients characteristics

Among 665 patients with STEMI who underwent direct PCI, 56 (8.4%) developed NOAF. Compared with patients without NOAF, those with NOAF were older (66.59 ± 10.38 years vs. 63.37 ± 13.07 years; p = 0.034), had a faster heart rate (81.61 ± 15.38 bpm vs. 77.23 ± 14.00 bpm; p = 0.027), and exhibited lower serum albumin levels (41.16 ± 4.78 g/L vs. 43.69 ± 3.81 g/L; p < 0.001), lower lymphocyte counts (1.55 ± 0.66 × 10⁹/L vs. 1.89 ± 0.93 × 10⁹/L; p = 0.008), larger left atrial diameters (36.71 ± 4.19 mm vs. 33.66 ± 3.77 mm; p < 0.001), higher FBG levels (5.97 ± 1.32 mmol/L vs. 5.46 ± 1.08 mmol/L; p = 0.001), higher median high-sensitivity C-reactive protein levels (4.25 mg/L vs. 1.30 mg/L; p < 0.001), lower median CALLY index values (1.45 vs. 6.53; p < 0.001), higher median N-terminal pro-brain natriuretic peptide levels (3413.28 pg/mL vs. 1999.00 pg/mL; p = 0.008), lower median LVEF (50.00% vs. 52.00%; p < 0.001), and a higher proportion of patients with right coronary artery involvement (48.21% vs. 32.68%; p = 0.019). There were no statistically significant differences between the two groups in terms of sex, body mass index, blood pressure, blood lipids, past medical history (hypertension, diabetes, and stroke), smoking history, baseline medication use, or Killip cardiac function classification (all p > 0.05) (Table 1).

Table 1.

Baseline characteristics of ST-segment elevation myocardial infarction patients with and without NOAF.

Variables	Total (n = 665)	No NOAF (n = 609)	NOAF (n = 56)	p
Age, years	63.64 ± 12.89	63.37 ± 13.07	66.59 ± 10.38	0.034
Female, n (%)	185 (27.82)	164 (26.93)	21 (37.50)	0.091
BMI, kg/m2	24.58 ± 3.38	24.54 ± 3.29	24.94 ± 4.29	0.509
SBP, mmHg	126.37 ± 21.79	126.55 ± 21.33	124.50 ± 26.42	0.502
DBP, mmHg	78.63 ± 13.70	78.71 ± 13.63	77.68 ± 14.54	0.589
Heart rate, bpm	77.60 ± 14.16	77.23 ± 14.00	81.61 ± 15.38	0.027
WBC, 109/L	9.06 ± 3.28	8.98 ± 3.31	9.87 ± 2.82	0.053
N, 109/L	7.85 ± 3.92	7.77 ± 4.01	8.65 ± 2.69	0.110
ALB, g/L	43.47 ± 3.96	43.69 ± 3.81	41.16 ± 4.78	<.001
L, 109/L	1.86 ± 0.91	1.89 ± 0.93	1.55 ± 0.66	0.008
LA-D, mm	33.91 ± 3.90	33.66 ± 3.77	36.71 ± 4.19	<.001
HGB, g/L	140.37 ± 16.59	140.24 ± 16.62	141.84 ± 16.26	0.490
Plt, 109/L	214.89 ± 59.67	215.84 ± 59.59	204.50 ± 60.03	0.174
Total cholesterol, mmol/L	4.38 ± 1.06	4.40 ± 1.06	4.27 ± 1.16	0.407
Triglycerides, mmol/L	1.35 ± 0.77	1.34 ± 0.79	1.46 ± 0.47	0.261
LDL-C, mmol/L	2.75 ± 0.91	2.75 ± 0.90	2.72 ± 0.94	0.815
HDL-C, mmol/L	1.06 ± 0.29	1.06 ± 0.28	1.01 ± 0.31	0.221
FBG, mmol/L	5.51 ± 1.11	5.46 ± 1.08	5.97 ± 1.32	0.001
hs-CRP, mg/L	1.50 (0.90, 5.50)	1.30 (0.75, 5.20)	4.25 (1.00, 9.53)	<.001
CALLY index	6.23 (1.31, 9.41)	6.53 (1.49, 9.80)	1.45 (0.63, 5.19)	<.001
hs-TnT, ng/L	3086.0 (1243.0, 5627.0)	2974.0 (1225.0, 5600.0)	3474.7 (1335.8, 5687.0)	0.685
NT-proBNP, pg/mL	2056.0 (1083.29, 4267.7)	1999.0 (1064.4, 4084.7)	3413.3 (1747.6, 5196.2)	0.008
LVEF, %	51.00 (49.00, 55.00)	52.00 (49.00, 56.00)	50.00 (41.75, 52.00)	<.001
Hypertension, n (%)	269 (40.45)	242 (39.74)	27 (48.21)	0.216
Diabetes mellitus, n (%)	145 (21.80)	132 (21.67)	13 (23.21)	0.789
Stroke, n (%)	73 (10.98)	70 (11.49)	3 (5.36)	0.160
Smoking, n (%)	271 (40.75)	251 (41.22)	20 (35.71)	0.423
Aspirin, n (%)	615 (92.48)	561 (92.12)	54 (96.43)	0.365
P2Y12, n (%)	638 (95.94)	583 (95.73)	55 (98.21)	0.584
β-blockers, n (%)	538 (80.90)	488 (80.13)	50 (89.29)	0.095
Statins, n (%)	627 (94.29)	572 (93.92)	55 (98.21)	0.306
ACEI/ARB, n (%)	347 (52.18)	321 (52.71)	26 (46.43)	0.368
Killip class, n (%)				0.097
I	558 (83.91)	516 (84.73)	42 (75.00)
II	44 (6.62)	40 (6.57)	4 (7.14)
III	20 (3.01)	16 (2.63)	4 (7.14)
IV	43 (6.47)	37 (6.08)	6 (10.71)
IRA
LAD, n (%)	347 (52.18)	324 (53.20)	23 (41.07)	0.082
LCX, n (%)	75 (11.28)	71 (11.66)	4 (7.14)	0.307
RCA, n (%)	226 (33.98)	199 (32.68)	27 (48.21)	0.019
Others, n (%)	17 (2.56)	15 (2.46)	2 (3.57)	0.952

ALB: albumin; BMI: body mass index; CALLY: C-reactive protein–albumin–lymphocyte index; DBP: diastolic blood pressure; FBG: fasting blood glucose; HDL-C: high-density lipoprotein cholesterol; HGB: hemoglobin; hs-CRP: high-sensitivity C-reactive protein; hs-TnT: high-sensitivity troponin T; IRA: infarct-related artery; L: lymphocyte; LAD: left anterior descending artery; LA-D: left atrial diameter; LCX: left circumflex artery; LDL-C: low-density lipoprotein cholesterol; LVEF: left ventricular ejection fraction; N: neutrophil; NOAF: new-onset atrial fibrillation; NT-proBNP: N-terminal pro-B-type natriuretic peptide; Plt: platelet; RCA: right coronary artery; SBP: systolic blood pressure; TC: total cholesterol; TG: triglycerides; WBC: white blood cell.

Association between clinical variables and NOAF: Univariate and multivariate logistic regression

Univariate logistic regression analysis revealed that heart rate, left atrial diameter, albumin, lymphocyte count, FBG, high-sensitivity C-reactive protein, CALLY index, N-terminal pro-brain natriuretic peptide, LVEF, and right coronary artery as the infarct-related artery were all significantly associated with NOAF (all p < 0.05). In the multivariable logistic regression analysis, we used the Enter method, entering all variables with p < 0.05 in the univariate analysis (including heart rate, LA-D, FBG, CALLY index, NT-proBNP, LVEF, and IRA-RCA), together with age (forced into the model based on clinical relevance), simultaneously into the model. Seven covariates (age, heart rate, LA-D, IRA-RCA, LVEF, FBG, and CALLY index) were entered into the final multivariable logistic regression model. Among these, five variables remained statistically significant independent predictors of NOAF: LA-D (OR, 1.179; 95% CI, 1.092–1.274), IRA-RCA (OR, 1.992; 95% CI, 1.063–3.733), LVEF (OR, 0.933; 95% CI, 0.888–0.979), FBG (OR, 1.321; 95% CI, 1.048–1.665), and decreased CALLY index (OR, 0.813; 95% CI, 0.741–0.892); all p values <0.05. Age (OR, 1.013; 95% CI, 0.982–1.044; p = 0.418) and heart rate (OR, 1.019; 95% CI, 0.997–1.042; p = 0.089) were retained in the model but did not reach statistical significance (Table 2).

Table 2.

Univariate and multivariate logistic regression analysis of factors associated with NOAF.

Variables	Univariate		Multivariate
	OR (95%CI)	p	OR (95%CI)	p
Age, years	1.021 (0.998–1.044)	0.075	1.013 (0.982–1.044)	0.418
Sex, female	1.628 (0.921–2.878)	0.094
BMI, kg/m2	1.034 (0.955–1.120)	0.408
SBP, mmHg	0.996 (0.983–1.008)	0.501
DBP, mmHg	0.994 (0.975–1.015)	0.588
Heart rate, bpm	1.022 (1.002–1.042)	0.028	1.019 (0.997–1.042)	0.089
WBC, 109/L	1.078 (0.999–1.164)	0.054
N, 109/L	1.041 (0.988–1.096)	0.135
ALB,	0.856 (0.800–0.916)	<.001
L, 109/L	0.552 (0.359–0.851)	0.007
LA-D, mm	1.220 (1.135–1.311)	<.001	1.179 (1.092–1.274)	<0.001
HGB, g/L	1.006 (0.989–1.023)	0.490
Plt,109/L	0.997 (0.992–1.001)	0.173
Total cholesterol, mmol/L	0.893 (0.683–1.167)	0.406
Triglycerides, mmol/L	1.207 (0.869–1.676)	0.262
LDL-C, mmol/L	0.964 (0.710–1.309)	0.814
HDL-C, mmol/L	0.473 (0.145–1.539)	0.214
FBG, mmol/L	1.366 (1.122–1.662)	0.002	1.321 (1.048–1.665)	0.019
hs-CRP, mg/L	1.192 (1.122–1.267)	<.001
CALLY index	0.805 (0.738–0.878)	<.001	0.813 (0.741–0.892)	<0.001
Peak hs-TnT, ng/L	1.000 (1.000–1.000)	0.972
Peak NT-proBNP, pg/mL	2.026 (1.122–3.655)	0.019
LVEF	0.921 (0.888–0.956)	<.001	0.933 (0.888–0.979)	0.005
Hypertension, n (%)	1.412 (0.816–2.444)	0.218
Diabetes mellitus, n (%)	1.092 (0.571–2.092)	0.790
Stroke, n (%)	0.436 (0.133–1.432)	0.171
Smoking, n (%)	0.792 (0.448–1.401)	0.424
Aspirin, n (%)	2.310 (0.547–9.764)	0.255
P2Y12, n (%)	2.453 (0.327–18.407)	0.383
β-blockers, n (%)	2.066 (0.866–4.931)	0.102
Statins, n (%)	3.558 (0.479–26.431)	0.215
ACEI/ARB, n (%)	0.778 (0.449–1.346)	0.369
Killip class>1	1.849 (0.971–3.521)	0.061
IRA-LAD	0.613 (0.352–1.069)	0.084
IRA-LCX	0.583 (0.205–1.660)	0.312
IRA-RCA	1.918 (1.106–3.328)	0.020	1.992 (1.063–3.733)	0.032

ALB: albumin; BMI: body mass index; CALLY: C-reactive protein–albumin–lymphocyte index; CI: confidence interval; DBP: diastolic blood pressure; FBG: fasting blood glucose; HDL-C: high-density lipoprotein cholesterol; HGB: hemoglobin; hs-CRP: high-sensitivity C-reactive protein; hs-TnT: high-sensitivity troponin T; IRA: Infarct-related artery; L: lymphocyte; LAD: Left anterior descending artery; LA-D: left atrial diameter; LCX: Left circumflex artery; LDL-C: low-density lipoprotein cholesterol; LVEF: left ventricular ejection fraction; N: neutrophil; NOAF: new-onset atrial fibrillation; NT-proBNP: N-terminal pro-B-type natriuretic peptide; OR: odds ratio; Plt: platelet; RCA: Right coronary artery; SBP: systolic blood pressure; TC: total cholesterol; TG: triglycerides; WBC: white blood cell.

Sensitivity analysis using individual CALLY components

In the component-based sensitivity analysis, CRP was positively associated with NOAF (adjusted OR, 1.175; 95% CI, 1.091–1.265; p < 0.001), whereas albumin was inversely associated with NOAF (adjusted OR, 0.853; 95% CI, 0.787–0.923; p < 0.001). Lymphocyte count showed an inverse association that did not reach statistical significance (adjusted OR, 0.634; 95% CI, 0.389–1.031; p = 0.066). No meaningful multicollinearity was observed (VIF range, 1.012–1.096) (Supplementary Table S2a).

The component-based model yielded an AUC of 0.850 (95% CI, 0.796–0.903), compared with 0.838 (95% CI, 0.786–0.889) for the CALLY-based model. This small numerical difference was not statistically significant (DeLong p = 0.515), indicating comparable discriminative performance between the two modeling approaches (Supplementary Table S2b).

Sex-disaggregated analysis

NOAF occurred in 21 of 185 women (11.4%) and 35 of 480 men (7.3%; p = 0.091). The admission CALLY index did not differ significantly between women and men (6.143 (1.292–9.328) vs. 6.253 (1.318–9.604); p = 0.757), and no sex-related differences were observed in CRP, serum albumin, or lymphocyte count (Supplementary Table 1). In the adjusted interaction model, the sex × CALLY interaction was not significant (OR, 0.931; 95% CI, 0.761–1.140; p = 0.490). The association between a lower CALLY index and NOAF was directionally consistent in men (OR, 0.834; 95% CI, 0.747–0.931) and women (OR, 0.777; 95% CI, 0.655–0.921).

Relationship between left atrial diameter and the CALLY index

Spearman correlation analysis demonstrated a weak inverse correlation between left atrial diameter and the CALLY index (r = −0.117, p = 0.020) (Figure 2).

Figure 2.

Correlation between left atrial diameter and the CALLY index.

CALLY: C-reactive protein–albumin–lymphocyte index.

Receiver operating characteristic (ROC) analysis for identifying NOAF

Among single parameters for identifying NOAF, the C-reactive protein–albumin–lymphocyte index demonstrated the highest predictive efficacy, with an AUC of 0.743 (95% CI: 0.685–0.801). At a cutoff value of 5.694, its sensitivity was 87.5% and specificity was 61.7% (P < 0.001). The AUC for high-sensitivity C-reactive protein was 0.692 (95% CI: 0.619–0.765), whereas the AUCs for albumin and lymphocyte count were 0.652 (95% CI: 0.564–0.739) and 0.611 (95% CI: 0.535–0.687), respectively. All indicators showed statistically significant differences from random prediction (all p < 0.01) (Table 3, Figure 3).

Table 3.

ROC analysis of individual biomarkers for predicting NOAF.

Variables	AUC	95%CI	Cutoff	Sensitivity	Specificity	p
CALLY index	0.743	0.685–0.801	5.694	0.875	0.617	<0.001
ALB	0.652	0.564–0.739	39.05	0.429	0.913	<0.001
hs-CRP	0.692	0.619–0.765	2.09	0.661	0.657	<0.001
Lym	0.611	0.535–0.687	1.65	0.643	0.565	0.006

ALB: albumin; AUC: area under the curve; CALLY: C-reactive protein–albumin–lymphocyte index; CI: confidence interval; hs-CRP: high-sensitivity C-reactive protein; Lym: lymphocyte; NOAF: new-onset atrial fibrillation; ROC: receiver operating characteristic.

Figure 3.

Receiver operating characteristic (ROC) curves of individual biomarkers for predicting new-onset atrial fibrillation (NOAF).

ROC analysis of combined models for predicting NOAF

Model 1 (including age, heart rate, LA-D, IRA-RCA, LVEF, and FBG) achieved an AUC of 0.783 (95% CI: 0.719–0.847), with sensitivity of 83.9% and specificity of 62.7%. After adding the CALLY index (Model 2), the AUC increased to 0.838 (95% CI: 0.787–0.889), with sensitivity of 73.2% and specificity of 80.8%. Both models were statistically significant (P < 0.001) (Table 4, Figure 4).

Table 4.

ROC analysis of combined models for predicting NOAF.

Variables	AUC	95%CI	Sensitivity	Specificity	p
Model 1	0.783	0.719–0.847	0.839	0.627	<0.001
Model 2	0.838	0.787–0.889	0.732	0.808	<0.001

Model 1: Age, heart rate, LA-D, IRA-RCA, LVEF, and FBG.

Model 2: Age, heart rate, LA-D, IRA- RCA, LVEF, FBG, and CALLY index.

AUC: area under the curve; CI: confidence interval; NOAF: new-onset atrial fibrillation; ROC: receiver operating characteristic.

Figure 4.

Receiver operating characteristic (ROC) curves of combined models for predicting new-onset atrial fibrillation (NOAF).

Internal validation using 1000 bootstrap resamples demonstrated excellent model performance, with an optimism-corrected AUC of 0.846, a corrected calibration slope of 1.110, a calibration intercept of −0.000, and a corrected Brier score of 0.0608, indicating limited overfitting and good calibration.

Discrimination accuracy and reclassification improvement of the CALLY Index for NOAF

Adding the CALLY index to the baseline model (which included age, heart rate, LA-D, LVEF, FBG, and IRA-RCA) significantly improved the prediction of NOAF. The category-free NRI was 0.7323 (95% CI: 0.4762–0.9993, p < 0.001), and the IDI was 0.0593 (95% CI: 0.0277–0.1101, p = 0.005) (Table 5).

Table 5.

Discrimination accuracy and reclassification of CALLY index for the NOAF.

	NRI		IDI
	Estimate (95% CI)	p	Estimate (95% CI)	p
Model of characteristics routinelya	Reference	−	Reference	−
Model of characteristics routinely + CALLY index	0.7323 (0.4762–0.9993)	<0.001	0.0593 (0.0277–0.1101)	0.005

CALLY: C-reactive protein–albumin–lymphocyte index; CI: confidence interval; IDI: integrated discrimination improvement; NOFA: new-onset atrial fibrillation; NRI: net reclassification improvement.

^aModel of characteristics routinely included age, heart rate, LA-D, LVEF, FBG, and IRA-RCA.

Discussion

The key findings of this study are as follows: (a) the CALLY index was independently associated with NOAF after PCI in patients with ST-segment elevation myocardial infarction; (b) it demonstrated better predictive performance for AF than individual inflammatory or metabolic biomarkers; and (c) adding the CALLY index to a clinical model containing established risk factors significantly improved risk reclassification and discrimination.

Although PCI has significantly reduced in-hospital mortality in patients with ST-STEMI, acute-phase complications including tachyarrhythmias, remain significant contributors to adverse outcomes.^1,2 In our study, NOAF occurred in approximately 8.4% of patients with STEMI. The associated rapid ventricular rate can precipitate heart failure, exacerbate myocardial ischemia, and worsen clinical prognosis. ²⁴ Therefore, early identification of high-risk factors for NOAF is of considerable clinical importance.

Recent research has increasingly highlighted the significance of the CALLY index in cardiovascular disease.^19,20 A study by Liu et al.¹⁹ demonstrated its predictive value for MACEs in patients with heart failure and diabetes. In hypertensive populations, a lower CALLY index has been closely associated with an increased risk of stroke. ²¹ Deng et al.²² identified the CALLY index as an independent risk factor for AF recurrence following radiofrequency ablation in patients with AF. In our study, we found that the CALLY index had significant predictive value for NOAF in patients with STEMI. This association may be explained by the following pathophysiological mechanisms.

Inflammation is a well-established risk factor for NOAF. ²⁵ CRP, an acute-phase protein synthesized by the liver in response to infection, injury, or autoimmune activity, typically peaks 1.5–2 days after myocardial infarction, with levels correlating positively with the extent of myocardial injury.^26,27 Previous studies have identified CRP as an independent risk factor for NOAF in patients with acute coronary syndrome. ²⁷ This finding is consistent with our observation of significantly elevated CRP levels in the NOAF group. Additionally, Lanca Bastiancic et al. ²⁸ found that elevated hs-CRP levels constitute an independent risk factor for AF recurrence following radiofrequency ablation. CRP promotes the release of interleukin-1 (IL-1) and tumor necrosis factor by macrophages,^29,30 enhances activation of the nuclear factor-κB (NF-κB) and transforming growth factor-β/signal modulator-activated protein kinase (TGF-β/SMAD) signaling pathways, thereby accelerating myocardial fibrosis and subsequently promoting the onset and maintenance of AF.^29–31 Beyond promoting myocardial fibrosis, recent evidence suggests that CRP is closely associated with microcirculatory dysfunction, which may exacerbate atrial ischemia and promote electrical remodeling, further sustaining AF. ³²

In contrast, during the acute phase of myocardial infarction, systemic inflammation and elevated cortisol induce lymphocyte apoptosis, resulting in lymphocytopenia. ³³ This impairs immune regulatory functions, exacerbates local inflammation and oxidative stress, and promotes myocardial fibrosis and microcirculatory dysfunction.^34,35 Collectively, these factors increase susceptibility to AF. Several recent studies have also linked lymphocytopenia to adverse outcomes in patients with STEMI.^34,36

Hypoalbuminemia reflects both malnutrition and inflammatory activity. ³⁷ Importantly, low albumin increases susceptibility to oxidative stress, exacerbates endothelial dysfunction, and promotes foam cell formation and inflammation.^37,38 Unlike some prior studies,^22,39 our study found significantly lower albumin levels in patients with NOAF, which may reflect differences in study populations. In acute STEMI, pronounced inflammation shifts hepatic synthesis toward positive acute-phase proteins, such as CRP, thereby reducing albumin production. ⁴⁰

The component-based sensitivity analysis further supported the biological coherence of the CALLY index, as CRP and albumin showed significant associations with NOAF in the expected directions, whereas lymphocyte count showed a directionally consistent but nonsignificant inverse association. Although the model incorporating the three components separately yielded a numerically higher AUC than the CALLY-based model (0.850 vs. 0.838), the difference was small and not statistically significant. Because the component-based approach estimates separate coefficients for three variables, whereas the CALLY index summarizes the same domains within a single composite measure, these findings suggest that CALLY may provide a more parsimonious representation of inflammatory, nutritional, and immune status without a detectable loss of discriminative performance.

Women showed a numerically higher incidence of NOAF than men, although neither the admission CALLY index nor its association with NOAF differed significantly by sex. These exploratory findings should be interpreted cautiously because only 21 NOAF events occurred among women. Rapid AF may aggravate myocardial ischemia by increasing oxygen demand, shortening diastolic coronary perfusion, and reducing cardiac output, mechanisms that overlap with the supply–demand imbalance underlying type 2 myocardial infarction. Sex-related differences in this setting have been increasingly recognized. ⁴¹ However, type 2 myocardial infarction was not adjudicated in the present cohort; therefore, our findings cannot establish NOAF as a trigger of type 2 myocardial infarction, and further adequately powered studies are required.

Finally, consistent with previous studies, we confirmed that left atrial diameter (LA-D), IRA- RCA, LVEF, and FBG are significant risk factors for NOAF.^42–44 As a marker of atrial structural remodeling,^16,45 LA-D demonstrated a weak but statistically significant inverse correlation with the CALLY index in our study. Although this cross-sectional association was weak (r = −0.117), it does not establish a causal relationship, and its clinical relevance remains uncertain. Furthermore, adding the CALLY index to the baseline model comprising age, heart rate, LA-D, IRA-RCA, LVEF, and FBG significantly improved risk reclassification for NOAF, as reflected by an NRI of 0.7323 and an IDI of 0.0593 (both p < 0.05).

This study has several limitations. First, the single-center retrospective design may introduce selection bias and unmeasured confounding. Second, NOAF was ascertained mainly by routine electrocardiography; although Holter monitoring was performed in some patients, systematic ambulatory surveillance was not implemented, potentially resulting in missed asymptomatic or paroxysmal episodes. Third, the CALLY index was based on a single admission measurement, reflecting status at presentation rather than the dynamic inflammatory response, and its temporal stability is unknown. Fourth, the limited number of events (n = 56) may increase the risk of overfitting, and external validation is lacking. Finally, the sex-disaggregated analysis was underpowered because of the few events among women (n = 21) and should be interpreted cautiously.

Conclusions

In this retrospective single-center cohort, a lower admission CALLY index was independently associated with clinically documented 30-day NOAF after PCI in patients with STEMI and provided incremental predictive information; however, external validation is required before clinical application.