Predictive factors for atrial fibrillation recurrence after radiofrequency ablation: a multifactorial approach

Abstract

Background

Radiofrequency ablation is a leading clinical method for restoring sinus rhythm in atrial fibrillation (AF) patients. However, the high recurrence rate and potential complications necessitate careful evaluation of its application.

Aim

This study aims to identify predictive factors for post-ablation AF recurrence by integrating multiple preoperative clinical variables in AF patients.

Methods

A total of 270 patients with non-valvular AF undergoing first-time radiofrequency ablation were categorized into a recurrence group and a non-recurrence group. Univariate and multivariate logistic regression analyses were performed to identify independent risk factors for AF recurrence. Receiver operating characteristic (ROC) curves and the area under the curve (AUC) were used to evaluate the predictive value of related factors and a combined prediction model for AF recurrence.

Results

At one-year follow-up, AF recurred in 100 patients (37.04%). Logistic multifactor regression analysis identified left atrial diameter (LAD), N-terminal pro-brain natriuretic peptide (NT-proBNP), uric acid (UA), left atrial appendage blood flow velocity (LAAV), and early recurrence of AF (ERAF) as independent predictors of AF recurrence. The AUCs of LAD, NT-proBNP, UA and the combined prediction model for predicting AF recurrence after radiofrequency ablation were 0.674, 0.685, 0.652 and 0.785, respectively, with a statistically significant difference between single indicators and the combined model (P < 0.05).

Conclusion

LAD, NT-proBNP, UA, LAAV, and ERAF are independent predictors of atrial fibrillation (AF) recurrence. The combined model of LAD, NT-proBNP, and UA shows better predictive ability than individual factors, offering a more reliable recurrence assessment. Our model, based on these three widely accessible markers, is simple, practical, and easily generalizable. However, the lack of external validation limits its applicability. Future studies should validate the model in independent cohorts to confirm its robustness and generalizability.

Introduction

Atrial fibrillation (AF) stands as the most prevalent tachyarrhythmia encountered in clinical practice, with significant associations to adverse outcomes such as “stroke, cognitive impairment, and heart failure” [1, 2]. This condition is not only disabling but also life-threatening. The global incidence of atrial fibrillation is steadily rising, primarily driven by an aging population and an increasing prevalence of cardiovascular risk factors [3]. As of 2019, approximately 59.7 million individuals worldwide were afflicted by atrial fibrillation and atrial flutter [4]. Current treatment strategies for atrial fibrillation include cardioversion, ventricular rate control, and anticoagulation. The antiarrhythmic drugs such as amiodarone and propafenone remain important guideline-recommended therapies for AF, their use may be limited in some patients by suboptimal efficacy and adverse effects. Although radiofrequency ablation is widely recognized as a leading intervention for restoring sinus rhythm in atrial fibrillation patients [5, 6], its recurrence rate remains notably high, with an estimated efficacy of only 50–80% [7–9]. The anticipation of atrial fibrillation recurrence after radiofrequency ablation and the provision of early intervention have emerged as prominent concerns within the realm of cardiac electrophysiology. Presently, several single-factor indicators have been established to predict the recurrence of atrial fibrillation, including atrial fibrosis [10], atrial fibrillation duration [11], left atrial appendage blood flow velocity [12], and left atrial systolic strain [13], among others. However, the predictive value of single-factor indicators is often suboptimal due to variations in sensitivity, specificity, and numerous clinical confounding factors in AF patients. Thus, the amalgamation of multiple factors may offer heightened sensitivity in forecasting the recurrence rate of atrial fibrillation following radiofrequency ablation. This approach unquestionably holds the potential to empower clinicians in selecting targeted treatment strategies for atrial fibrillation patients.

Aim

The primary objective of this study was to find the factors influencing the outcomes of radiofrequency ablation in patients with atrial fibrillation and to develop a predictive model for postoperative recurrence in atrial fibrillation patients by combining multiple factors.

Materials and methods study population

We assembled a cohort of 356 patients diagnosed with non-valvular AF who underwent their first radiofrequency catheter ablation at the General Hospital of the People’s Liberation Army of China between February 2018 and December 2019. This study is retrospective, based on patients who had already undergone radiofrequency ablation for AF. Follow-up and serial ECGs were performed according to a standardized institutional protocol as part of routine clinical care. Although follow-up data and serial ECGs may appear prospective, the overall design remains retrospective.

Indications for catheter ablation in patients with atrial fibrillation

According to contemporary international guidelines, including the 2020 ESC and the 2023 ACC/AHA/HRS recommendations, candidates for AF catheter ablation must meet several criteria. First, patients must have a confirmed diagnosis of paroxysmal or persistent AF associated with symptoms that impair quality of life. Second, catheter ablation is generally indicated for individuals who have failed to respond to, are intolerant of, or prefer not to continue at least one class I or class III antiarrhythmic medication; in selected younger patients without significant structural heart disease, ablation may be considered as first-line therapy. Prior to the procedure, potential reversible triggers of AF should be identified and addressed, and a comprehensive structural cardiac assessment—typically including echocardiographic evaluation of left atrial size and ventricular function—is required. Exclusion of left atrial or left atrial appendage thrombus is mandatory, usually achieved by transesophageal echocardiography or cardiac CT. Adequate peri-procedural anticoagulation must be ensured. Finally, patients should demonstrate an understanding of the risks and benefits of the intervention, adhere to follow-up requirements, and provide informed consent. Collectively, these criteria represent the principal contemporary indications for AF catheter ablation.

After applying exclusion criteria, we removed patients who met any of the following conditions : left atrial diameter ≥ 50 mm confirmed by echocardiography, uncontrolled heart failure, congenital heart disease, valvular heart disease or prior valve replacement, acute coronary syndrome, cardiomyopathy, recent infection, uncontrolled hyperthyroidism, renal insufficiency, previous ablation history, or loss to follow-up for any reason. These patients were categorized into a “recurrence group” and a “non-recurrence group” according to the presence or absence of AF recurrence within one year after the procedure.

The study protocol received approval from the Human Ethics Review Committee of the Chinese People’s Liberation Army General Hospital. All procedures adhered to the ethical standards outlined in the Declaration of Helsinki.

Research data and measurements

The patient’s general information encompasses gender, age, body mass index (BMI), past medical history (including hypertension, diabetes, heart failure, and vascular disease), and a record of drug use (angiotensin-converting enzyme inhibitor (ACEI)/angiotensin II receptor blocker (ARB), CCC calcium antagonist, amiodarone, β-blocker, and statins). Additionally, the HAS-BLED and CHA2DS2-VASc scores were calculated. Imaging data, including left atrial diameter (LAD) and left ventricular ejection fraction (LVEF), were measured via transthoracic echocardiography to assess structural heart changes associated with AF recurrence. Blood biochemical data covers serum creatinine (Scr), serum uric acid (UA), C-reactive protein (CRP), NT-proBNP, triacylglycerol (TG), low-density lipoprotein (LDL-C), and total cholesterol (TC).

Radiofrequency catheter ablation

Radiofrequency catheter ablation was performed under the guidance of the CARTO3 (Biosense Webster, USA) three-dimensional electroanatomic mapping system. All patients underwent preoperative transesophageal echocardiography to exclude left atrial appendage thrombosis, and anticoagulation therapy was administered before the procedure. Activated clotting time (ACT) was continuously monitored during ablation, with intravenous heparin supplemented as needed. The left atrial anatomy and substrate were mapped prior to energy delivery. Radiofrequency ablation was performed using an irrigated ablation catheter at 30–35 W with a temperature limit below 43 °C. Lesion formation was assessed using impedance drop and electrogram attenuation as conventional surrogate indicators of adequate tissue contact. Circumferential pulmonary vein isolation (PVI) was the primary ablation endpoint. Successful PVI required elimination of pulmonary vein potentials and confirmation of bidirectional conduction block between the pulmonary veins and the left atrium. High output pacing along the ablation line was additionally used to verify entrance and exit block and ensure complete PVI. If sinus rhythm was not restored after PVI, synchronized direct-current cardioversion (100–150 J) was performed. Based on substrate characteristics, additional linear ablation—such as a roof line or mitral isthmus line—was performed at the operator’s discretion, and completion of linear lesions was confirmed by disappearance of local electrograms and conduction block testing. Posterior wall isolation was selectively performed when substrate features suggested posterior wall involvement. Intravenous fentanyl was administered for analgesia throughout the procedure. Esophageal temperature was continuously monitored, and ablation was interrupted when temperature exceeded predefined safety thresholds to minimize the risk of thermal injury.

Follow-up and data collection

Telephone follow-up was conducted for all enrolled patients to ascertain that patient underwent 24-hour Holter monitoring (dynamic ECG) every three months during the period from three months to one year after surgery. Patients were also instructed to promptly complete an electrocardiogram if they experienced any symptoms. Dynamic ECG or any ECGs performed during symptomatic episodes were monitored for signs of recurrence. A recurrence of atrial fibrillation was defined as the presence of atrial fibrillation, atrial flutter, or atrial tachycardia lasting at least 30 s [14]. Early recurrence of AF (ERAF)was defined as AF recurrence within a 3-month blanking period, which was not included in the final recurrence. Based on these findings, it was recorded whether patients experienced recurrence during the follow-up period, and research subjects were categorized into two groups: the recurrence group and the non-recurrence group, based on the presence or absence of recurrence.

Statistical analysis

Statistical analyses were performed using SPSS 27.0. Data normality was assessed with the Kolmogorov–Smirnov test. Normally distributed variables were expressed as mean ± SD and compared using the independent-samples t test, while non-normally distributed variables were presented as median (Q1, Q3) and analyzed using the Mann–Whitney U test. Categorical variables were expressed as frequencies (%) and compared using the χ² test. Binary logistic regression was used to identify risk factors for atrial fibrillation recurrence after radiofrequency ablation. Candidate variables for multivariable analysis were selected based on a univariate P < 0.20, clinical relevance, and pathophysiological plausibility. Multicollinearity was assessed and highly correlated variables were excluded. To reduce overfitting, the events-per-variable (EPV) ratio was evaluated and exceeded the recommended threshold of 10 in all models. Sensitivity analyses with reduced covariates showed consistent results. ROC curves were used to calculate AUC, sensitivity, and specificity. All tests were two-sided, and P < 0.05 was considered statistically significant.

Results

Baseline characteristics

Following the exclusions, data from the final cohort of 270 patients were included in this study. The cohort was divided into two groups: 158 patients (58.6%) with paroxysmal atrial fibrillation and 112 patients (41.4%) with persistent atrial fibrillation, (Fig. 1). At one-year follow-up, atrial fibrillation (AF) recurred in 100 patients (37.04%).

Fig. 1.

The summary of the process of study participants

Patients experiencing recurrence, whether they had paroxysmal or persistent atrial fibrillation, exhibited several notable characteristics: elevated NT-proBNP levels (P < 0.05), increased left atrial diameter (P < 0.05), higher serum uric acid levels (P < 0.05), slower left atrial appendage flow velocity (P < 0.05) and the history of ERAF. Among patients with persistent atrial fibrillation, significant differences were also observed in body mass index (BMI) and the duration of atrial fibrillation (P < 0.05), (Table 1).

Table 1.

Atrial fibrillation recurrence and baseline features

Characteristics	Paroxysmal AF				Persistent AF
	Total (n = 158)	Recurrence (n = 55)	No Recurrence (n = 103)	p-value	Total (n = 112)	Recurrence (n = 45)	No Recurrence (n = 67)	p-value
Male, n (%)	99 (62.66)	36 (65.45)	66 (64.08)	0.743	80 (71.43)	38 (84.44)	42 (62.69)	0.172
Age [years]	60.42 ± 10.87	61.11 ± 9.54	59.93 ± 11.74	0.487	59.52 ± 10.00	58.03 ± 59.53	60.43 ± 10.26	0.304
BMI [kg/m2]	25.63 ± 3.14	25.46 ± 3.19	25.72 ± 3.72	0.571	26.73 ± 3.554	27.36 ± 3.33	25.75 ± 3.11	0.045*
AF duration (months)	——	——	——	——	22 (6–60)	30 (15–108)	12 (6–48)	0.005**
Hypertension, n (%)	81 (72.32)	25 (45.45)	56 (54.37)	0.318	73 (65.18)	28 (62.22)	45 (67.16)	0.686
Diabetes, n (%)	36 (22.78)	13 (23.63)	26 (25.24)	0.689	23 (51.11)	7 (15.56)	16 (23.88)	0.345
CHD, n (%)	39 (24.68)	11 (20.00)	28 (17.48)	0.341	29 (25.89)	7 (15.56)	22 (32.84)	0.291
HFrEF, n (%)	8 (5.06)	3 (5.45)	5 (4.85)	0.999	9 (8.04)	3 (6.67)	6 (8.96)	0.738
Stroke or TIA, n (%)	22 (13.92)	7 (12.73)	14 (13.59)	0.380	16 (14.28)	7 (15.56)	9 (13.43)	0.294
Vascular disease, n (%)	12 (7.59)	4 (7.27)	8 (7.77)	0.474	9 (8.04)	4 (8.89)	5 (7.46)	0.461
CHA2DS2-VASC score	2.0 [1.0–3.0]	2.0 [1.0–3.0]	2.0 [1.0–3.0]	0.769	2.0 [1.0–3.0]	3.0 [1.0–3.3]	2.0 [1.0–3.0]	0.545
HAS-BLED score	1.0 [0.0–2.0]	1.0 [0.0–2.0]	1.0 [0.0–2.0]	0.715	1.0 [0.0–2.0]	1.0 [0.0–2.0]	1.0 [0.0–2.0]	0.675
Clinical laboratory date
NT-proBNP [pg/ml]	86.0 [47.6–325.0]	271.7 [87.8–533.6]	68.0 [43.7–149.1]	<0.001**	293.4 [115.6–529.7]	342.3 [251.2–608.5]	185.4 [99–501.4.]	0.001**
CRP	0.10 [0.06–0.29]	0.10 [0.06–0.32]	0.10 [0.07–0.24]	0.817	0.13 [0.09–0.35]	0.10 [0.07–0.36]	0.23 [0.10–0.34]	0.720
UA (umol/L)	325.1 ± 80.9	349.5 ± 88.2	310.9 ± 73.1	0.004**	368.9 ± 108.1	413.4 ± 112.6	330.9 ± 88.8	0.008**
Scr (umol/L)	81.2 ± 15.6	81.9 ± 16	80.9 ± 15.3	0.700	82.1 ± 15.7	79.6 ± 14.9	83.6 ± 16.2	0.281
TG (mmol/L)	1.34 [0.98–1.83]	1.49 [1.00–1.83]	1.31 [0.91–1.71]	0.118	1.30 [0.88–1.69]	1.25 [0.85–1.48]	1.31 [0.89–1.89]	0.308
TC (mmol/L)	3.98 ± 0.88	4.03 ± 0.96	3.95 ± 0.84	0.624	4.04 ± 0.94	4.08 ± 0.96	4.02 ± 0.94	0.796
LDL-C (mmol/L)	2.46 ± 0.81	2.51 ± 0.89	2.43 ± 0.76	0.592	2.57 ± 0.84	2.63 ± 0.87	2.54 ± 0.83	0.645
N (109/L)	0.56 ± 0.10	0.56 ± 0.10	0.56 ± 0.09	0.999	0.58 ± 0.08	0.58 ± 0.10	0.58 ± 0.07	0.904
L (109/L)	0.33 ± 0.09	0.33 ± 0.10	0.32 ± 0.09	0.624	0.32 ± 0.08	0.32 ± 0.09	0.32 ± 0.07	0.851
RDW (%)	12.6 [12.0–13.2]	12.6 [12.1–13.3]	12.5 [12.0–13.1]	0.571	12.5 [12.1–13.1]	12.5 [12.2–12.9]	12.6 [12.1–13.2]	0.300
Echocardiographic date
LAD (mm)	36.0 [34.0–40.0]	37.0 [34.5–42.0]	35.0 [33.0–38.0]	0.006**	42.0 [39.0–47.0]	46.5 [41.8–51.0]	40.0 [37.0–43.5]	<0.001**
LVD (mm)	45.0 [42.0–48.0]	45.0 [42.0–47.0]	46.0 [43.0–48.0]	0.247	47.0 [44.0–51.0]	46.5 [41.8–51.0]	47.0 [44.0–51.0]	0.725
LVEF (%)	61.0 [57.0–64.0]	60.0 [56.0–64.0]	61.0 [58.0–65.0]	0.142	58.0 [54.0–62.0]	59.0 [54.0–62.0]	58.0 [53.0–62.0]	0.962
LAAV (m/s)	0.53 ± 0.18	0.47 ± 0.19	0.58 ± 0.16	<0.001**	0.42 ± 0.17	0.35 ± 0.15	0.46 ± 0.17	0.005*
Medications, n (%)
NOAC	107 (67.72)	40 (72.72)	67 (65.05)	0.374	84 (75.00)	30 (66.67)	54 (80.60)	0.120
Amiodarone	88 (55.70)	30 (54.54)	58 (56.31)	0.868	66 (58.93)	23 (51.11)	43 (64.20)	0.240
Propafenone	27 (17.09)	8 (14.55)	19 (18.45)	0.659	25 (22.32)	8 (17.78)	17 (25.37)	0.367
β-blockers	37 (23.42)	9 (16.36)	28 (27.18)	0.167	41 (36.61)	12 (26.67)	29 (43.28)	0.109
ACEI/ARB	36 (22.78)	10 (18.18)	26 (25.24)	0.426	37 (33.04)	11 (24.44)	26 (38.81)	0.152
Statins	64 (40.51)	19 (34.55)	45 (43.68)	0.309	55 (49.11)	18 (40.00)	37 (55.22)	0.127
Antiplatelet	8 (5.06)	2 (3.64)	6 (5.83)	0.714	8 (7.14)	2 (4.44)	6 (8.96)	0.472
Ablational procedures, n (%)
Additional roof ablation	69 (43.67)	23 (41.82)	46 (44.66)	0.740	52 (46.43)	17 (37.78)	35 (52.23)	0.176
ERAF, n (%)	38 (24.05)	19 (34.55)	19 (18.45)	0.034*	46 (10.41)	28 (62.22)	18 (26.87)	0.004**

BMI  Body mass index, CHD Coronary heart disease, HFrEF Heart failure with reduced ejection fraction, CHA2DS2-VASc Congestive heart failure (CHF), hypertension (HT), age ≥ 75 years [doubled], diabetes mellitus (DM), stroke [doubled], vascular disease, age 65 − 74 years, sex category [female], HAS-BLED Elderly, alcohol use, history of impaired kidney or hepatic performance, history of hypertension, stroke history, history of major bleeding, the usage of medications predisposes to bleeding, an international normalized ratio that is labile, NT-proBNP N-terminal pro-brain natriuretic peptide, CRP c reactive protein. UA, uric acid, Scr Blood creatinine, TG Triglyceride, TC Total cholesterol, LDL Low density lipoprotein, LAD Left atrial diameter, LVD Left ventricular diameter, LVEF Left ventricular ejection fraction, ARB Angiotensin receptor blocker, ACEI Angiotensin-converting enzyme inhibitors, NOAC New oral anticoagulants, ERAF Early recurrence of atrial fibrillation

*P < 0.05, **p < 0.01

Recurrence of AF in multivariate and univariate analysis

Multivariate logistic regression analysis identified several independent risk factors for atrial fibrillation recurrence after radiofrequency ablation, including LAD (OR 1.117, 95% CI 1.023–1.162, P < 0.01), LAAV (OR 0.064, 95% CI 0.005–0.803, P < 0.05), NT-proBNP (OR 1.020, 95% CI 1.001–1.050, P < 0.01), UA (OR 1.050, 95% CI 1.030–1.130, P < 0.01) and ERAF (OR: 1.523, 95% CI (1.124–6.091), P < 0.05). (Table 2).

Table 2.

Univariate and multivariate analyses of atrial fibrillation recurrence

Variables	Univariate analysis			Multivariate analysis
	OR	95%CI	P value	OR	95%CI	P value
Age (years)	1.005	0.980–1.030	0.703
BMI (kg/m2)	0.930	0.859–1.007	0.730
Hypertension	0.735	0.439–1.231	0.242
Diabetes mellitus	0.822	0.440–1.534	0.538
CHD	0.923	0.482–1.768	0.809
LAD	1.111	1.049–1.176	<0.001*	1.117	1.023–1.162	0.001*
LVD	0.964	0.908–1.025	0.241
LVEF	0.982	0.944–1.021	0.354
LAAV	0.750	0.600–0.930	0.004*	0.064	0.005–0.803	0.033*
NT-proBNP	1.001	1.000–1.002	0.006*	1.020	0.001–1.050	0.004*
CRP	1.023	0.685–1.529	0.911
UA	1.007	1.004–1.010	<0.001*	1.050	1.030–1.130	0.002*
TG	1.166	0.827–1.645	0.380
ACEI/ARB	0.913	0.522–1.597	0.749
Statin	1.016	0.603–1.712	0.952
ERAF	1.287	0.046–5.673	0.001*	1.523	1.124–6.091	0.037*

AF Atrial fibrillation, HR Hazard ratio, CI Confidence interval, BMI Body mass index, CHD Coronary heart disease, NT-proBNP N-terminal pro-brain natriuretic peptide, LAD Left atrial diameter, LVD Left ventricular diameter, LVEF Left ventricular ejection fraction, LAAV Left atrial appendage peak flow velocity, CRP c reactive protein, UA Uric acid, TG Triglyceride, ARB Angiotensin receptor blocker, ACEI Angiotensin-converting enzyme inhibitor. ORs for NT-proBNP and UA are expressed per 10-unit increase

*P < 0.05

LAD, NT-proBNP, UA, and combined predictive model predictive value

The logistic equation that combines LAD, NT-proBNP, and UA using logistic regression is as follows: Logistic (P) = 0.111LAD + 0.002NT-proBNP + 0.005UA − 6.872. The predictive model’s accuracy was assessed using the Hosmer-Lemeshow goodness of fit test (χ2: 7.556, P: 0.478). ROC curves were generated for LAD, NT-proBNP, UA, and the combined predictive model to predict atrial fibrillation recurrence following radiofrequency ablation. The combined predictive model demonstrated superior predictive value compared to any individual factor, with statistically significant differences (P < 0.01) (Tables 3 and 4; Fig. 2).

Table 3.

The ability of different indexes to predict the recurrence of atrial fibrillation after radiofrequency ablation

	Sensitivity (%)	Specificity	Cutoff value	AUC	95%CI		P value
					Lower bound	Upper bound
LAD	0.636	0.671	39.50	0.674	0.606	0.742	<0.001
NT-proBNP	0.545	0.788	253.30	0.685	0.617	0.752	<0.001
UA	0.596	0.740	363.35	0.652	0.580	0.723	<0.001
Combined model	0.657	0.842	——	0.785	0.752	0.845	<0.001

LAD Left atrial diameter, NT-proBNP N-terminal pro-brain natriuretic peptide, UA Uric acid

Table 4.

Comparison of areas under the ROC curve of different factors

	AUC Difference	Std. Error Difference	95%CI		Z	P value
			Lower bound	Upper bound
LAD-NT-proBNP	−0.010	0.262	−0.096	0.076	−0.237	0.813
UA-NT-proBNP	−0.033	0.267	−0.1.28	0.062	−0.682	0.495
UA-LAD	−0.023	0.267	−0.117	0.072	−0.469	0.639
LAD- Combined model	−0.111	0.254	−0.174	−0.048	−3.450	0.001
NT-proBNP- Combined model	−0.101	0.253	−0.160	−0.042	−3.343	0.001
UA- Combined model	−0.134	0.258	−0.205	−0.062	−3.677	<0.001

LAD Left atrial diameter, NT-proBNP N-terminal pro-brain natriuretic peptide, UA Uric acid

Fig. 2.

ROC curve for predicting postoperative recurrence of AF

Discussion

AF represents the most prevalent tachyarrhythmia encountered in clinical practice. The advent of radiofrequency ablation has significantly enhanced the capacity to restore sinus rhythm in patients with AF, thereby reducing the morbidity and mortality associated with complications. However, the inherent recurrence rate of AF following ablation places additional physical and financial stress on patients. Hence, it is imperative to identify the factors influencing AF recurrence after radiofrequency ablation, which can enable clinicians to identify high-risk patients and customize personalized treatment strategies.

This study comprehensively analyzed clinical data from patients with AF and uncovered significant insights. Irrespective of whether the patients had persistent or paroxysmal AF, larger LAD, higher levels of NT-proBNP and UA, slower LAAV and the history of ERAF were linked to a heightened likelihood of AF recurrence following ablation. Notably, in patients with persistent AF, additional statistically significant associations were observed with body mass index (BMI) and the duration of AF (P < 0.05).

The enlargement of the LAD is closely linked to the incidence of AF. An increase in LAD is often accompanied by structural remodeling of the atria, including hypertrophy, necrosis, excessive stretching of atrial myocytes, and fibrosis of the interstitial myocardium. These changes affect the electrophysiological properties of the atria, promoting the onset of atrial fibrillation. Furthermore, the increased incidence of AF in individuals with alcohol consumption and obesity can be attributed to the resulting left atrial enlargement [15, 16]. Previous studies have suggested that LAD enlargement holds predictive value for the recurrence of AF after radiofrequency ablation. In this study, we verified this conclusion again. By plotting receiver operating characteristic curves and calculating the predictive value of LAD for post-ablation AF recurrence, our findings indicate that when LAD exceeds 39.5 mm, the area under the curve, sensitivity, and specificity for predicting post-ablation AF recurrence are 67.4%, 63.6%, and 67.1%, respectively. Moreover, for every 1 mm increase in LAD, there is an associated 11.7% increase in the risk of post-ablation AF recurrence. Left atrial enlargement is considered a hallmark of left atrial remodeling, leading to atrial conduction abnormalities and the accumulation of the substrate for arrhythmia, consequently elevating the risk of AF recurrence. LAD is a crucial parameter in diagnosing and managing atrial fibrillation, with its enlargement being strongly correlated with AF recurrence post-ablation. Further research in this area is essential for a better understanding of the relationship between LAD and atrial fibrillation. This, in turn, can enhance prevention and treatment strategies for AF and reduce the occurrence of related complications. Clinical practitioners are encouraged to recognize the significance of left atrial enlargement and incorporate it into the management plans for atrial fibrillation patients, ultimately offering more effective diagnostic and treatment approaches.

NT-proBNP, a precursor hormone primarily secreted by the atria, increases in response to elevated atrial pressure, atrial matrix remodeling (fibrosis and hypertrophy), and inflammation. Studies have shown that the elevated level of NT-proBNP in patients without atrial fibrillation is significantly related to the burden of atrial fibrillation [17] and is a sign of increased risk of atrial fibrillation [18]. The higher preoperative NT-proBNP level in patients with atrial fibrillation is an independent risk factor for recurrence after radiofrequency ablation of fibrillation [19]. In addition, some scholars have found that patients whose NT-proBNP levels drop by more than 30% within 1 year after radiofrequency ablation have a very low probability of recurrence of atrial fibrillation 1 year after surgery, which can be considered a sign of successful radiofrequency ablation of atrial fibrillation [20]. Our study shows that the level of NT-proBNP in patients with atrial fibrillation is directly related to the recurrence of atrial fibrillation after radiofrequency ablation. When NT-proBNP surpasses 253.3 pg/mL, the area under the curve, sensitivity, and specificity for predicting postoperative AF recurrence are 68.5%, 54.5%, and 78.8%, respectively. Moreover, for every 10 pg/mL increase in NT-proBNP, the likelihood of AF recurrence following ablation rises by 2%. Based on a comprehensive review of previous research and the results of our own study, we believe that increased levels of NT-proBNP may be associated with deterioration of atrial structure and function and is involved in the mechanisms underlying the occurrence and recurrence of AF. In clinical practice, the measurement of NT-proBNP can assist physicians in better diagnosing AF, assessing the severity of the condition, and guiding treatment strategies, thereby enhancing the management and prevention of AF in patients and reducing the occurrence of related complications. Further research will contribute to a deeper understanding of the role of NT-proBNP in AF, providing more evidence for the diagnosis and treatment of AF.

Inflammation and oxidative stress play pivotal roles in the onset, maintenance, and postoperative recurrence of AF [21]. Xanthine oxidase has been identified as significantly linked to inflammation and oxidative stress. Serum uric acid, the end product of xanthine oxidase-catalyzed purine catabolism, is implicated in endothelial cell dysfunction and reflects the body’s inflammatory and oxidative stress status. Numerous studies have underscored the significance of elevated serum uric acid in cardiovascular risk factors and diseases, such as hypertension, heart failure, and coronary artery disease [22, 23]. Study have confirmed that serum uric acid is associated with the incidence of atrial fibrillation [24]. A Japanese study that included 285,882 subjects found that serum uric acid is an independent risk factor for the occurrence of atrial fibrillation [25], and when serum uric acid is > 315umol/L in men and > 297.5umol/L in women, the incidence of atrial fibrillation increases significantly. A cohort study based on medicare data, in US, in which a total of 1,647,812 subjects were included and followed up for 6 years found that after multivariable adjustment, among people over 64 years old, subjects with high uric acid levels had twice the risk of an atrial fibrillation event (HR: 1.92 (1.88–1.96), P < 0.001) [26]. In addition, large-sample cohort studies conducted in Norway, South Korea, and the United Kingdom have also shown that high serum uric acid is a risk factor for the incidence of atrial fibrillation [27–29]. Our study shows that high serum uric acid is a predictive factor for postoperative recurrence of atrial fibrillation. When serum uric acid exceeds 363.35 µmol/L, the area under the curve, sensitivity, and specificity for predicting postoperative AF recurrence are 65.2%, 59.6%, and 74.0%, respectively. With each 10 µmol/L rise in serum uric acid, the risk of postoperative AF recurrence increases by 5%. Taking into consideration the results of the aforementioned studies, we believe that elevated serum uric acid may influence the electrophysiological characteristics of the heart by inducing inflammation and oxidative stress, leading to the disruption of myocardial cell electrical activity and an increased susceptibility to AF. Additionally, it may also exert an impact on AF’s onset and maintenance by affecting endothelial function and metabolic processes within the cardiovascular system. Therefore, in clinical practice, the monitoring of serum uric acid should be emphasized, particularly for high-risk patients, in order to timely adjust treatment strategies, enhance AF prevention and management, and reduce the occurrence of related complications.

A multifactor logistic regression analysis revealed that LAD, NT-proBNP, and UA are robust predictors of AF recurrence after radiofrequency ablation. Building upon these findings, we constructed a composite prediction model for postoperative AF recurrence. The area under the ROC curve, sensitivity, and specificity of this composite prediction model were 78.5%, 65.7%, and 84.2%, respectively. Comparing the AUC of individual factors with the composite prediction model, a statistically significant difference was observed (P < 0.01), underscoring the enhanced predictive capability of the composite model in comparison to any single factor. The composite prediction model emerged as a potent tool for forecasting AF recurrence following radiofrequency ablation. In recent years, multiple models have been introduced to improve prediction of AF recurrence after catheter ablation, incorporating echocardiographic, biochemical, or machine-learning–derived variables. Zhao et al. developed a nomogram for persistent AF incorporating left atrial volume index, left atrial appendage function, BNP and inflammatory markers, achieving an AUC of 0.893 [30]. Askarinejad et al. applied machine-learning techniques to clinical and ECG data and reported an AUC of 0.96 for early recurrence prediction [31]. Similarly, Duan et al. utilized multimodal clinical and imaging features to predict recurrence in paroxysmal AF, reaching an AUC of approximately 0.91 [32]. These studies demonstrate the trend toward increasingly complex, multi-parameter models aimed at capturing atrial structural and functional remodeling more comprehensively. However, despite their strong statistical performance, these newer models generally rely on advanced imaging, extensive laboratory testing, or machine-learning frameworks that may not be readily available in many clinical settings. Their complexity can limit applicability, and external validation remains limited for most existing tools. In comparison, the present study proposes a simple and clinically practical model based solely on three routinely obtained pre-procedural variables—LAD, NT-proBNP, and UA. Despite its simplicity, the model achieved an AUC of 0.785, which is comparable to or slightly higher than commonly used risk scores such as APPLE, CAAP-AF, BASE-AF2 and MB-LATER, while requiring fewer parameters and no specialized testing. Notably, the identification of serum uric acid as an independent predictor provides additional insight into the role of oxidative stress and metabolic factors in AF recurrence, a dimension not commonly integrated into prior prediction tools. The ease of obtaining LAD, NT-proBNP and UA in routine practice gives the present model a distinct advantage in real-world applicability, offering clinicians an efficient tool for pre-procedural risk stratification. This is particularly valuable in resource-limited settings or centres without advanced imaging or computational infrastructure. Nonetheless, as with most existing models, the present analysis is limited by its single-center retrospective design. External validation across broader populations and diverse ethnic groups will be crucial to confirming its generalizability.

Our study also identified LAAV and ERAF as significant predictors of AF recurrence after radiofrequency ablation. Slower LAAV was associated with a higher risk of recurrence (OR: 0.064, 95% CI 0.005–0.083, P < 0.05), with an optimal cut-off value of 0.44 m/s, aligning with previous research [33]. ERAF is commonly observed following atrial fibrillation ablation, and both prior studies [34] and our findings suggest that patients with ERAF are less likely to maintain sinus rhythm compared to those without ERAF. Thus, ERAF serves as a reliable predictor of recurrence. This supports the notion that ERAF is not solely the result of acute thermal injury or inflammation from ablation energy but may also involve pulmonary vein-left atrium reconnection [35], which could contribute to late recurrence in ERAF patients. Understanding the underlying mechanisms of LAAV and ERAF in AF recurrence could inform the development of targeted strategies to reduce recurrence rates. However, the invasive nature of procedures required to obtain LAAV and ERAF, along with their associated risks, precludes their inclusion in the composite prediction model. Additionally, variations in radiofrequency (RF) power settings may influence lesion characteristics and recurrence risk [36]. In our cohort, however, all procedures were conducted using standardized power settings, which lacked sufficient variability to assess the impact of power-dependent effects. Future studies should explore the role of different RF power strategies in AF recurrence, including those described in recent literature.

Study limitations

This study has several limitations. First, its single-center retrospective design and relatively small sample size may limit the generalizability of the findings. Second, the cohort consisted exclusively of Chinese patients, and differences in genetic background, lifestyle, and comorbidities across ethnicities may affect the broader applicability of the model. Third, AF recurrence may have been underestimated because follow-up relied primarily on scheduled Holter monitoring and symptom-driven ECGs, which may miss asymptomatic episodes. Finally, external validation was not performed, and the model’s performance in other populations or clinical settings remains to be confirmed. Future studies should include external validation in independent cohorts to better establish the robustness and clinical utility of the model.

Conclusions

In conclusion, this study confirms that LAD, NT-proBNP, UA, LAAV, and ERAF are independent predictors of atrial fibrillation recurrence after radiofrequency catheter ablation. More importantly, our work provides a novel and simplified prediction model based solely on three universally accessible pre-procedural markers—LAD, NT-proBNP, and UA, offering a fully pre-procedural, highly practical, and easily generalizable approach. This represents a meaningful innovation, as it allows clinicians to identify high-risk patients before the ablation procedure, improving decision-making, patient selection, and individualized treatment planning. Furthermore, by integrating structural remodeling (LAD), neurohormonal activation (NT-proBNP), and metabolic/oxidative stress (UA), our model captures multiple mechanistic pathways contributing to AF recurrence—an aspect not simultaneously addressed by existing models. These strengths highlight the unique contribution of our study and support the potential of our model as an efficient, accessible, and clinically relevant tool for predicting AF recurrence across diverse clinical settings. However, the absence of external validation is a key limitation of this study. Future research should focus on validating the model in independent cohorts to ensure its broader applicability.

Abbreviations

AF

Atrial fibrillation

ACEI

Angiotensin-converting enzyme inhibitor

ACT

Activated clotting time

AHA

American Heart Association

ARB

Angiotensin II receptor blocker

AUC

Area under the curve

BMI

Body mass index

BNP

B-type natriuretic peptide

CARTO3

Three-dimensional electroanatomic mapping system

CHD

Coronary heart disease

CI

Confidence interval

CRP

C-reactive protein

ECG

Electrocardiogram

ERAF

Early recurrence of atrial fibrillation

HFrEF

Heart failure with reduced ejection fraction

LAD

Left atrial diameter

LAAV

Left atrial appendage flow velocity

LDL-C

Low-density lipoprotein cholesterol

LVD

Left ventricular diameter

LVEF

Left ventricular ejection fraction

NOAC

Non–vitamin K antagonist oral anticoagulant

NT-proBNP

N-terminal pro–B-type natriuretic peptide

OR

Odds ratio

PVI

Pulmonary vein isolation

RF

Radiofrequency

ROC

Receiver operating characteristic

Scr

Serum creatinine

SD

Standard deviation

TC

Total cholesterol

TG

Triglyceride

TIA

Transient ischemic attack

UA

Uric acid

Authors’ contributions

ZNN， LXY and BYL conceived and designed the study. BYL, ZZF, WCC, LXY and ZC performed the data analysis, applying statistical methods and interpreting the results. ZNN and BYL wrote the manuscript. All authors read and approved the final manuscript.

Funding

The authors declare that no funds, grants, or other support were received during the preparation of this manuscript.

Data availability

No datasets were generated or analysed during the current study.

Declarations

Ethics approval and consent to participate

Not applicable. This study was a retrospective analysis utilizing data collected from routine clinical practice. And the informed consent was not obtained in this case, as the study uses anonymized data and does not involve direct patient intervention. All patient data were thoroughly anonymized prior to analysis. It was granted an exemption from ethical approval by the Institutional Review Chinese PLA General Hospital. Every step undertaken in this study adhered to the principles outlined in the Declaration of Helsinki.

Consent for publication

All authors approved the final manuscript and the submission to this journal.

Competing interests

The authors declare no competing interests.