The potential value of serum neuron-specific enolase in screening for silent cerebral infarction following radiofrequency catheter ablation for atrial fibrillation

Abstract

Background

Silent cerebral infarction (SCI) is a frequent perioperative complication of catheter ablation for atrial fibrillation (AF), yet its detection currently relies on magnetic resonance imaging (MRI) — a modality not routinely accessible for all patients. This preliminary study investigated whether neuron-specific enolase (NSE), a serum biomarker of neuronal injury, could serve as a practical screening tool for perioperative SCI following radiofrequency ablation for AF.

Methods

This exploratory study enrolled 97 patients undergoing first-time radiofrequency catheter ablation for AF. Based on MRI performed 24–48 h post-procedure, patients were classified into the SCI and non-SCI groups. Serum NSE levels were measured before ablation (NSE_pre), immediately after ablation (NSE_ipost), and 24 h post-ablation (NSE_{24h post}). Multivariate logistic regression, receiver operating characteristic (ROC) curve analysis, net reclassification improvement (NRI), and integrated discrimination improvement (IDI) were employed to assess the association between NSE and SCI.

Results

Among the 97 patients (63 paroxysmal AF, 34 persistent AF), SCI was detected in 15 patients (15.5%). In the overall cohort, serum NSE levels increased significantly immediately after ablation compared with baseline and remained elevated at 24 h post-procedure (both P < 0.001). NSE_ipost levels were significantly higher in the SCI group than in the non-SCI group (P = 0.034), whereas NSE_pre levels were comparable between groups. Multivariate logistic regression identified a greater increase in NSE from pre- to immediate post-ablation (ΔNSE_ipost–pre) as significantly associated with SCI (OR = 1.31, 95%CI 1.09–1.62, P = 0.016). ROC analysis revealed that ΔNSE_ipost–pre, with an optimal cutoff value of 4.6 ng/mL, yielded a sensitivity of 93.3%, a specificity of 39.0%, and an area under the curve of 0.69 for detecting SCI. Incorporating ΔNSE_ipost–pre significantly improved the model’s discriminative performance (NRI = 0.748, 95% CI 0.232–1.064, P < 0.001; IDI = 0.084, 95% CI 0.024–0.114, P = 0.003).

Conclusion

Serum NSE exhibits high sensitivity for detecting SCI following AF ablation, suggesting its potential as a post-procedural screening tool to help identify patients who may benefit from confirmatory MRI, particularly in resource-constrained settings. Given the exploratory nature of this study and its methodological limitations, these findings should be considered hypothesis-generating and warrant validation in larger, well-designed prospective cohorts.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12872-026-05789-7.

Introduction

Catheter ablation is the first-line treatment for symptomatic, drug-refractory atrial fibrillation (AF) [1]. Although the use of irrigated-tip catheters has substantially reduced the incidence of symptomatic cerebral infarction [2], silent cerebral infarction (SCI) remains a frequent perioperative complication [3]. Recent studies report that SCI occurs in up to 51.7% of patients following AF ablation [4–9], raising concerns about its potential long-term neurocognitive consequences [10–13].

Currently, diffusion-weighted magnetic resonance imaging (DW-MRI) is the gold standard for detecting ischemic stroke—including SCI [14]. However, the routine clinical application of MRI is constrained by several practical limitations: it is time-consuming, expensive, and contraindicated in patients with certain implantable devices or claustrophobia. These barriers preclude its use as a universal screening tool, highlighting the need for a simple, accessible, and cost-effective alternative for perioperative SCI assessment.

Neuron-specific enolase (NSE), a cytoplasmic glycolytic enzyme predominantly localized in neurons and neuroectodermal cells, has emerged as a well-established biomarker of neuronal injury. Upon neuronal damage and disruption of the blood–brain barrier, NSE is released into the cerebrospinal fluid and subsequently enters the systemic circulation. With a biological half-life of approximately 24 h, serum NSE levels have been shown to correlate with the extent of brain injury in various clinical settings, including cardiovascular surgery, traumatic brain injury, and cardiac arrest [15–17]. Based on this evidence, current international guidelines recognize its prognostic value in survivors of cardiac arrest [18]. In the specific context of AF ablation, Acibuca et al. observed that periprocedural NSE levels exceeded the normal reference limit in approximately one-third of patients undergoing catheter ablation for paroxysmal AF, a frequency comparable to the reported incidence of subclinical embolism [15]. This finding suggested that NSE elevation might reflect ablation-related cerebral injury. However, that study had two important limitations: it did not employ MRI to confirm the presence of SCI, and it included only patients with paroxysmal AF, leaving the question open for persistent AF populations.

To address this gap, the present study used DW-MRI—the reference standard for SCI diagnosis—to monitor perioperative changes in serum NSE levels. Specifically, we aimed to investigate whether NSE could serve as a practical screening biomarker for SCI in patients undergoing catheter ablation for AF. Given the preliminary and exploratory nature of this investigation, our findings are intended to generate hypotheses and inform the design of future large-scale prospective studies.

Methods

Study population and protocol

Between January 2022 and December 2022, a total of 269 consecutive patients with symptomatic AF referred for catheter ablation at the Affiliated Hospital of Xuzhou Medical University were initially evaluated for enrollment. Of these, 97 patients (63 with paroxysmal AF, 34 with persistent AF) ultimately met the study criteria and were included in the final analysis. The reasons for exclusion were as follows: valvular heart disease (n = 13), recent cerebral infarction before the procedure (n = 3), contraindications to MRI (n = 5), previous transcatheter ablation (n = 9), left atrial appendage thrombus (n = 3), refusal to undergo MRI (n = 88), and refusal to undergo serum NSE monitoring (n = 51) (Fig. 1). A comprehensive comparison of baseline characteristics between the excluded and included populations is presented in Supplementary Table 1, which revealed no statistically significant differences between the two groups (all P > 0.05).

Fig. 1.

Flow chart

For each patient, the following clinical data were collected: age; sex; body mass index (BMI, calculated as weight in kilograms divided by the square of height in meters); medical history, including type and duration of AF; thromboembolic risk assessed by the CHA₂DS₂-VASc score [1]; physical examination findings; and neurological status. Transesophageal echocardiography (TEE) was performed prior to the procedure to exclude left atrial thrombus. Cerebral MRI was conducted 24–48 h after ablation. Serum NSE levels were measured at three predefined time points: before ablation (NSE_pre), immediately after ablation (NSE_ipost), and 24 h post-ablation (NSE_{24h post}). Based on MRI findings, patients were divided into two groups: the SCI group and non-SCI group. Patients in the SCI group underwent repeated physical and neurological examinations after the procedure.

All participants provided written informed consent prior to enrollment, and the study protocol was approved by the Institutional Review Board of the Affiliated Hospital of Xuzhou Medical University (Approval No. XYFY-2019-KL-156).

Ablation procedure

All ablation procedures were performed by experienced electrophysiologists. A decapolar catheter was positioned in the coronary sinus via either the left subclavian or femoral venous approach. Using a double transseptal technique, an 8.5 F Swartz sheath was advanced into the left atrium to accommodate a circular multipolar diagnostic catheter (Lasso, Biosense Webster, Diamond Bar, CA) and a 3.5 mm open-irrigation-tip ablation catheter (ST, Biosense Webster, Diamond Bar, CA, USA). Following the initial transseptal puncture, a weight-adjusted bolus of unfractionated heparin was administered. Activated clotting time (ACT) was measured every 15 min thereafter, with additional heparin given as needed to maintain ACT above 250 s, in accordance with the Chinese guidelines for catheter ablation [19]. A three-dimensional electroanatomical map of the left atrium and pulmonary veins was constructed using the Carto^® mapping system (Biosense Webster Inc., Diamond Bar, CA, USA). Circumferential pulmonary vein isolation (CPVI) was performed in all patients using a point-by-point ablation approach guided by the ablation index (AI). Targeting AI values were 450–500 on the anterior wall and 350–400 on the posterior wall. Ablation settings included a temperature limit of 43 °C, a power limit of 35 W, and irrigation flow rate of 17 mL/min during radiofrequency delivery. Electrical isolation of the pulmonary veins was confirmed by the achievement of bidirectional conduction block.

If AF persisted after CPVI, electrical cardioversion was performed. Subsequently, substrate mapping of the left atrium was carried out during sinus rhythm using the ablation catheter, and additional ablation was applied to regions of low voltage. Based on individual electrophysiological findings, further ablation strategies were employed in selected patients, including linear lesions (anterior line, cavotricuspid isthmus line, and mitral isthmus line), superior vena cava isolation, and posterior wall box isolation. Upon completion of ablation, intravenous isoproterenol was administered, followed by a bolus of adenosine triphosphate (ATP) to verify durable pulmonary vein isolation. Finally, programmed atrial stimulation and burst pacing were performed to assess the inducibility of arrhythmias.

Periprocedural anticoagulation

All patients received pre-procedural anticoagulation with a novel oral anticoagulant (NOAC), specifically either rivaroxaban (15 mg once daily; n = 81) or dabigatran (110 mg twice daily; n = 16). NOAC therapy was discontinued on the morning of the procedure. Following the initial transseptal puncture, an intravenous bolus of unfractionated heparin was administered based on the baseline ACT and body weight. ACT was subsequently monitored every 15 min throughout the procedure, with supplemental heparin boluses given as needed to maintain the target ACT. According to the Chinese guidelines for catheter ablation, an intra-procedural ACT target of 250–350 s is recommended for this population, taking into account the higher bleeding risk generally observed in Asian patients during anticoagulant therapy [19]. Accordingly, this study strictly adhered to the 250–350 s ACT target range specified in these national guidelines [19].

Brain magnetic resonance imaging

All patients underwent cerebral MRI using a 3.0-T scanner (Siemens, Germany) at 24–48 h post-ablation to detect new ischemic lesions. The imaging protocol included the following sequences: T1-weighted imaging (repetition time [TR] 2003 ms, echo time [TE] 15 ms, inversion time [TI] 750 ms, slice thickness 6 mm); T2-weighted imaging (TR 3864 ms, TE 105 ms, TI 750 ms, slice thickness 6 mm); axial T2-weighted fluid-attenuated inversion recovery (FLAIR; TR 8000 ms, TE 100 ms, TI 2200 ms, slice thickness 6 mm); and diffusion-weighted echo-planar imaging (DWI; TR 5000 ms, TE 100 ms, TI 750 ms, slice thickness 6 mm). All images were independently reviewed by two radiologists who were blinded to the clinical and laboratory data. Discrepancies between the readers were resolved by consensus discussion or, if necessary, by consultation with a third senior neuroradiologist. Acute SCI was defined as a new focal hyperintensity area on DWI with corresponding hypointensity on the apparent diffusion coefficient (ADC) map, thereby excluding T2 shine-through artifacts [14]. The size and location of each diffusion abnormality were documented.

Serum NSE monitoring

Blood samples were collected from each patient for NSE measurement. After allowing the samples to clot at room temperature for 30–60 min, they were centrifuged at 3000×g for 15 min. The resulting serum was then aliquoted and stored at − 80 °C until analysis. Serum NSE concentrations were quantified using an immunoluminometric assay (Roche NSE; Roche Diagnostics, Mannheim, Germany). According to the manufacturer’s specifications, the upper limit of the normal reference range for serum NSE is 17 ng/mL [15].

Statistical analysis

Statistical analyses were performed using R software (version 3.6.3). For continuous variables, normality was assessed using the Shapiro–Wilk test. Variables with P > 0.05 were considered normally distributed and were summarized as means ± standard deviation (SD); between-group comparisons were performed using independent-samples t-test. Variables with P ≤ 0.05 were considered non-normally distributed and were reported as medians with interquartile ranges (IQR); between-group comparisons were performed using the Mann–Whitney U test. For paired measurements, the paired t-test was applied to normally distributed data, whereas the Wilcoxon signed-rank test was used for non-normally distributed data. Categorical variables were expressed as numbers and percentages and were compared using Pearson’s chi-square test or Fisher’s exact test, as appropriate. To identify factors associated with SCI, all candidate variables were first entered into univariate logistic regression analysis. Variables with P < 0.05 in the univariate analysis were then included in the multivariate logistic regression analysis using the forward stepwise selection. The performance of individual factors and combined models for discriminating SCI was assessed using receiver operating characteristic (ROC) curve analysis. Net reclassification improvement (NRI) and integrated discrimination improvement (IDI) were calculated to evaluate improvements in risk classification and overall discriminative ability, respectively. As this was an exploratory analysis, multiple hypothesis tests were conducted for various NSE-derived indicators without adjustment for multiple comparisons; consequently, the identified associations carry a risk of type I error and should be interpreted with caution. A two-sided P value < 0.05 was considered statistically significant.

Results

Baseline and procedural characteristics

The final study cohort included 97 patients with a median age of 62.0 years (IQR 55.0–67.0), including 66 males (68.0%) and 63 patients with paroxysmal AF (64.9%). Detailed comparisons of baseline and procedural characteristics stratified by SCI status are presented in Table 1. Baseline clinical and demographic characteristics were generally similar between the SCI and non-SCI group, with the exception of left ventricular ejection fraction (LVEF), which was significantly lower in the SCI group [52% (IQR 50–62) vs. 59% (IQR 55–62), P = 0.038]. Although baseline ACT values and total heparin administration were comparable, the median intraprocedural ACT was significantly lower in the SCI group [245 s (IQR 235–259) vs. 260 s (IQR 240–278); P = 0.013]. Of note, none of the 21 patients whose ACT remained above 275 s throughout the entire procedure developed SCI. The type of ablation protocol and the rate of electrical cardioversion used to restore sinus rhythm were also similar between the groups (Table 1).Throughout the procedure, systolic blood pressure was maintained at ≥ 90 mmHg and mean arterial pressure remained > 65 mmHg in all patients, with no episodes of persistent hypotension observed. No significant differences were observed between the two groups in systolic blood pressure, diastolic blood pressure(Table 1).

Table 1.

Baseline and procedural characteristics

Parameter	SCI group (n = 15)	Non-SCI group (n = 82)	P value
Age, years	63.0(IQR 57.0–69.0)	61.5(IQR 55.0–67.0)	0.660
Gender, n(%)
Male	12(80.0%)	54(65.9%)	0.436
Female	3(20.0%)	28(34.1%)
AF type, n(%)
Paroxysmal	10(66.7%)	53(64.6%)	0.879
Persistent	5(33.3%)	29(35.4%)
Duration of AF, months	2.0(IQR 1.0–24.0)	12.0(IQR 1.0–36.0)	0.102
BMI, kg/m2	25.7(IQR 24.6–29.3)	26.0(IQR 23.9–28.5)	0.611
CHF, n(%)	4(26.7%)	8(9.8%)	0.161
Hypertension, n(%)	4(26.7%)	34(41.5%)	0.280
DM, n(%)	1(6.7%)	12(14.6%)	0.674
History of stroke/TIA, n(%)	3(20.0%)	5(6.1%)	0.197
CHD, n(%)	1(6.7%)	16(19.5%)	0.404
Pre-operation anticoagulation, n(%)			0.984
Dabigatran	3(20.0%)	13(15.9%)
Rivaroxaban	12(80.0%)	69(84.1%)
CHA2DS2-VaSc score	1.0(IQR 1.00–3.0)	2.0 (IQR 1.0–3.0)	0.806
GFR, mL/min	95.8 ± 23.4	105.8 ± 20.5	0.091
LVEF, %	52.0(IQR 50.0–62.0)	59.0(IQR 55.0–62.0)	0.038*
LAD, mm	41.5 ± 4.5	41.2 ± 6.1	0.855
LVEDD, mm	49.0(IQR 45.0–50.0)	48.0(IQR 45.8–51.0)	0.912
Total procedural time, min	192.8 ± 29.0	186.6 ± 41.9	0.584
LA operation time, min	151.4 ± 31.2	139.6 ± 36.3	0.226
Total amount of Heparin, U	10433.0 ± 3005.2	10969.5 ± 3272.9	0.556
Load amount of Heparin, U	7000(IQR 6000–8000)	7000(IQR 6000–8000)	0.367
Baseline ACT, s	155.1 ± 28.0	156.7 ± 28.5	0.941
Median ACT, s	245.0(IQR 235.0–259.0)	260.0(IQR 240.0–278.0)	0.013*
Periprocedural SBP, mmHg	137.9 ± 14.2	140.3 ± 17.2	0.602
Periprocedural DBP, mmHg	89.4 ± 6.8	87.3 ± 11.6	0.492
Periprocedural Heart rate, bpm	79.0(IQR 75.0–89.0)	80.0(IQR 71.0–94.0)	0.996
Type of Ablation protocol, n(%)
PVI	7(46.7%)	43(52.4%)	0.681
PVI+ additional ablation	8(53.3%)	39(47.6%)
Need for cardioversion, n(%)	11(73.3%)	42(51.2%)	0.114

CHF Congestive heart failure, DM Diabetes mellitus, TIA Transient ischemic attack, CHD Coronary heart disease, GFR Glomerular filtration rate, LVEF Left ventricular ejection fraction, LAD Left atrial diameter, LVEDD Left ventricular end-diastolic diameter, LA operation time: the duration of the ablation catheter remaining in the left atrial since the ablation catheter entered the left atrial, including left atrial linear ablation and left atrial matrix improvement, ACT Activated clotting time, SBP Systolic blood pressure, DBP Diastolic blood pressure, PVI Pulmonary vein isolation

Values are given as mean ± SD, median (IQR), or n(%)

*P < 0.05

MRI-related characteristics

All ablation procedures were successfully completed, and no patient exhibited clinically evident neurological deficits. Post-ablation diffusion-weighted MRI (DWI) revealed 18 new focal hyperintense lesions with corresponding hypotintensity on ADC maps in 15 patients (15.5%), consistent with a diagnosis of SCI (Fig. 2). Among these 15 patients, 12 presented with a single lesion and 3 with two lesions. The median lesion diameter was 3.8 mm (range 2.1–11.8 mm). Of the 18 lesions detected, 12 (66.7%) were located in the right cerebral hemisphere.

Fig. 2.

Magnetic resonance example. A Diffusion-weighted image and (B) Apparent diffusion Coefficient (ADC) maps of one patient; arrows indicate the acute cerebral infarctions post-ablation. A, B an acute lesion located in the right occipital lobe

Serum NSE kinetics

In the overall study population, serum NSE levels increased from 6.6 ± 2.7 ng/mL before ablation (NSE_pre) to 12.5 ± 4.6 ng/mL immediately after ablation (NSE_ipost), and decreased to 8.9 ± 3.5 ng/mL at 24 h post-ablation (NSE_{24h post}). Both NSE_ipost and NSE_{24h post} were significantly higher than NSE_pre (both P < 0.001).

When stratified by SCI status, pre-ablation NSE levels were comparable between the SCI and non-SCI groups (7.3 ± 1.8 ng/mL vs. 6.5 ± 2.8 ng/mL, P = 0.299). However, immediately after ablation, NSE levels were significantly higher in the SCI group than in the non-SCI group (14.8 ± 3.5 ng/mL vs. 12.1 ± 4.6 ng/mL, P = 0.034) (Table 2). Based on NSE measurements at different time points, the following derived variables were calculated: the absolute change from pre-ablation to immediately post-ablation(ΔNSE_ipost–pre = NSE_ipost – NSE_pre); the absolute change from pre-ablation to 24 h post-ablation (ΔNSE_{24h post–pre} = NSE_{24h post} – NSE_pre); and the corresponding percentage changes (%ΔNSE_ipost–pre and %ΔNSE_{24h post–pre} ). A significant difference between the two groups was observed for ΔNSE_ipost–pre, with higher values in the SCI group [7.2 ng/mL (IQR 5.6–8.3) vs. 5.3 ng/mL (IQR 3.3–7.4), P = 0.023].

Table 2.

Results of NSE monitoring

Parameter	SCI group (n = 15)	Non-SCI group (n = 82)	P value
NSEpre, ng/mL	7.3 ± 1.8	6.5 ± 2.8	0.299
NSEipost, ng/mL	14.8 ± 3.5	12.1 ± 4.6	0.034*
NSE24h post, ng/mL	9.8(IQR 7.8–12.3)	8.8(IQR 6.3–10.9)	0.084
ΔNSEipost–pre, ng/mL	7.2(IQR 5.6–8.3)	5.3(IQR 3.3–7.4)	0.023*
ΔNSE24h post–pre, ng/mL	2.4(IQR 1.4–4.4)	1.9(IQR 0.9–2.7)	0.057
%ΔNSEipost–pre, %	120.3(IQR 54.9–141.9)	81.9(IQR 49.8–115.3)	0.151
%ΔNSE24h post–pre, %	40.6(IQR 21.8–69.9)	26.8(IQR 16.8–48.7)	0.281

NSE_pre: the NSE level determined before the procedure; NSE_ipost: the NSE level determined immediately at the end of the procedure; NSE_{24h post}: the NSE level determined at 24 h after the procedure; ΔNSE_ipost–pre༚NSE_ipost – NSE_pre ; ΔNSE_ipost–pre༚NSE_{24h post} – NSE_pre ; %ΔNSE_ipost–pre༚[NSE_ipost – NSE_pre]/NSE_pre× 100% ; %ΔNSE_{24h post–pre} ༚[NSE_{24h post}– NSE_pre]/NSE_pre× 100%

Values are given as mean (SD), median (IQR), or n(%)

*P < 0.05.

Exploratory analyses were performed to examine the relationship between ΔNSE_ipost–pre and SCI lesion characteristics. No significant difference in ΔNSE_ipost–pre was observed between patients with a single lesion (n = 12) and those with multiple lesions (n = 3) (P = 0.773). Furthermore, linear correlation analysis revealed no significant correlation between ΔNSE_ipost–pre and the maximum diameter of SCI lesions (r = 0.12, P = 0.764).

Potential role of serum NSE in risk stratification for SCI

Variables with P < 0.05 in the univariate analysis were entered into multivariate logistic regression analysis, with the presence of SCI on MRI as the dependent variable (Table 3). Multivariate analysis identified that a lower intra-procedural median ACT (Odds ratio [OR] = 0.96, 95% CI 0.93–0.99, P = 0.028), a lower LVEF (OR = 0.90, 95% CI 0.83–0.97, P = 0.011), and a greater absolute change from pre- to post-ablation NSE (ΔNSE_ipost–pre; OR = 1.31, 95% CI 1.05–1.62, P = 0.016) were significantly associated with an increased risk of SCI.

Table 3.

Univariate and multivariate logistic regression analysis

Variable	Univariate analysis			Multivariate analysis
	Odds ratio	95%CI	P value	Odds ratio	95%CI	P value
Age, years	1.02	0.96–1.07	0.577
Male/Female	2.07	0.54–7.96	0.288
AF type
Paroxysmal	1.37	0.39–4.75	0.622
Persistent
Duration of AF, months	0.99	0.97–1.04	0.280
BMI, kg/m2	1.04	0.90–1.19	0.634
CHF, n(%)	3.36	0.87–13.07	0.080
Hypertension, n(%)	0.51	0.05–1.75	0.286
DM, n(%)	0.58	0.11–3.22	0.537
History of stroke/TIA, n(%)
CHD, n(%)	0.30	0.04–2.41	0.254
Pre-operation anticoagulation, n(%)
Dabigatran	0.75	0.19–3.05	0.691
Rivaroxaban
CHA2DS2-VaSc score	1.01	0.72–1.43	0.946
GFR, ml/min	0.98	0.95–1.00	0.094
LVEF, %	0.91	0.85–0.95	0.016	0.90	0.83–0.97	0.011*
LAD, mm	1.12	0.92–1.11	0.853
LVEDD, mm	1.03	0.93–1.14	0.623
Total procedural time, min	1.00	0.99–1.02	0.580
LA operation time, min	1.01	0.99–1.03	0.227
Total amount of Heparin, U	1.00	1.00–1.00	0.552
Load amount of Heparin, U	1.00	1.00–1.00	0.537
Baseline ACT, s	0.99	0.98–1.02	0.940
Median ACT, s	0.97	0.95–0.99	0.038	0.96	0.93–0.99	0.028*
Periprocedural SBP, mmHg	0.99	0.96–1.02	0.598
Periprocedural DBP, mmHg	1.02	0.97–1.07	0.488
Periprocedural Heart rate, bpm	0.99	0.96–1.03	0.691
Type of Ablation protocol, n(%)
CPVI	0.79	0.26–2.39	0.681
CPVI+ extra ablation
Need for cardioversion, n(%)	3.11	0.91–10.59	0.069
NSEpre, ng/ml	1.12	0.90–1.40	0.298
NSEipost, ng/ml	1.15	1.01–1.32	0.040
NSE24h post, ng/ml	1.06	0.77–1.43	0.067
ΔNSEipost–pre, ng/mL	1.28	1.03–1.49	0.024	1.31	1.09–1.62	0.016*
ΔNSE24h post–pre, ng/mL	1.02	0.68–1.93	0.052
%ΔNSEipost–pre, %	0.99	0.62–1.57	0.956
%ΔNSE24h post–pre, %	1.45	0.46–4.49	0.527

LVEF : Left ventricular ejection fraction; ACT: activated clotting time; NSE_ipost :the NSE level determined at immediately at the end of the procedure; NSE_{24h post} :the NSE level determined at 24 h after the procedure; ΔNSE_ipost–pre: NSE_ipost – NSE_pre; ΔNSE_ipost–pre༚NSE_24hpost – NSE_pre; %ΔNSE_ipost–pre༚[NSE_ipost – NSE_pre]/NSE_pre× 100%; %ΔNSE_{24hpost–pre} ༚[NSE_24hpost– NSE_pre]/NSE_pre× 100%

Values are given as mean (SD), median (IQR), or n(%)

*P < 0.05

ROC curve analysis showed that ΔNSE_ipost–pre, with a cutoff value of 4.60 ng/mL, yielded a sensitivity of 93.3% and a specificity of 39.0% for detecting SCI (AUC = 0.69). Subsequently, a reference model was constructed based on the multivariate analysis results, incorporating median ACT and LVEF. This model demonstrated adequate fit (Hosmer–Lemeshow χ² = 6.836, P > 0.05) and exhibited moderate discriminative ability for SCI after AF ablation (AUC = 0.759, 95% CI 0.647–0.875, P < 0.001). When ΔNSE_ipost–pre was added to the reference model, the enhanced model also showed good fit (Hosmer–Lemeshow χ² = 7.415, P > 0.05) and significantly improved predictive performance (AUC = 0.832, 95% CI 0.728–0.925, P < 0.001).

The NRI was 0.758 (95% CI 0.232–1.264, P < 0.001), and the IDI was 0.084 (95% CI 0.037–0.112, P = 0.003), indicating that the addition of ΔNSE_ipost–pre substantially enhanced the model’s ability to discriminate patients who developed SCI after AF ablation (Fig. 3; Tables 4 and 5).

Fig. 3.

ROC curves of ΔNSE_ipost–pre, Median ACT + LVEF, ΔNSE_ipost–pre + Median ACT + LVEF. ΔNSE_ipost–pre: NSE_ipost – NSE_pre

Table 4.

ROC curve analysis for SCI

	AUC	95%CI	P value	Sensitivity,%	Specificity,%
ΔNSEipost−pre	0.69	0.57–0.82	0.010	93.3	39.0
Reference modela	0.76	0.65–0.88	< 0.001	98.9	45.0
Reference model+ΔNSEipost−pre	0.83	0.73–0.93	< 0.001	86.7	68.3

^aReference model included Median ACT and LVEF; ΔNSE_ipost–pre: NSE_ipost – NSE_pre

Table 5.

Discrimination accuracy and reclassification of factors for SCI

	NRI		IDI
	Estimate(95%CI )	P value	Estimate(95%CI )	P value
Reference modela	Reference		Reference
Reference model+ΔNSEipost−pre	0.748 (0.232–1.264)	< 0.001	0.084(0.037–0.112)	0.003

^aReference model included MedianACT and LVEF; ΔNSE_ipost−pre: NSE_ipost – NSE_pre;

Discussion

This pilot study explored the association between serum NSE levels and the occurrence of SCI following catheter ablation for AF. The main observations are as follows: First, postoperative SCI was identified in a considerable proportion of patients, confirming its status as a frequent procedural complication. Second, serum NSE levels increased significantly during the procedure, and this elevation was more pronounced in patients who developed SCI, suggesting a potential link between NSE release and ablation-related cerebral injury. Third, although NSE exhibited high sensitivity (93.3%) for detecting SCI, its modest overall discriminatory ability (AUC 0.69) and low specificity (39.0%) indicate that it is not suitable as a standalone diagnostic tool. Instead, NSE may have value as an adjunctive screening marker to identify patients who could benefit from further MRI evaluation. Given the exploratory nature of this study and its methodological limitations, these findings should be considered hypothesis-generating and warrant validation in larger prospective cohorts.

Catheter ablation has become a standard therapeutic intervention for symptomatic AF. As an invasive procedure targeting the left atrium, it intrinsically increases the risk of thromboembolism, particularly cerebral infarction [20]. Although symptomatic cerebral infarction is rare, the incidence of SCI following AF ablation is substantially higher, with reported rates ranging from 2% to 51.7% [4, 21]. Considerable research has focused on identifying risk factors associated with ablation-related SCI. The use of irrigated-tip ablation catheters has been shown to reduce thrombus formation [3], while appropriate anticoagulation remains the cornerstone of SCI prevention. Previous studies have suggested that a lower ACT does not increase the risk of symptomatic thromboembolic events [22]; however, this criterion may underestimate the true thromboembolic burden, as it excludes silent infarcts. The current European Society of Cardiology (ESC) guidelines recommend maintaining an ACT > 300 s during ablation [1]. However, given the higher bleeding risk observed in Asian populations during anticoagulant therapy, the Chinese guidelines recommend an ACT range of 250–350 s at that time to balance thromboprophylaxis against bleeding risk [19].

In the present study, although no symptomatic thromboembolic events occurred, the observed incidence of SCI was 15.5%. This rate is consistent with findings from a larger investigation by our research team, which reported a mean ACT significantly lower in the SCI group compared with the non-SCI group (246.4 s vs. 254.9 s, P < 0.05) and an overall SCI incidence of 15.8% [23]. Similarly, Gaita et al. [24] studied 232 patients with paroxysmal or persistent AF using the same ACT target (> 250 s) and reported an SCI incidence of 14%. That study further demonstrated that SCI occurred more frequently in patients with ACT < 250 s and identified intra-procedural ACT as an independent factor associated with cerebral thromboembolism P = 0.0087, consistent with our findings. In this exploratory analysis, ACT was significantly associated with the occurrence of perioperative SCI (OR = 0.96; 95% CI 0.93–0.99, P = 0.028), suggesting that higher ACT values may confer a protective effect. This association may be partly attributable to the generally low and highly variable ACT levels observed in our cohort, which could have amplified the observed relationship. Of note, none of the 21 patients whose ACT remained above 275 s throughout the procedure developed SCI. Based on this exploratory observation, a preliminary hypothesis is proposed: in certain low-risk populations, a lower anticoagulation intensity might be sufficient to prevent thromboembolic events while potentially reducing bleeding risk. However, this hypothesis requires rigorous validation through well-designed, large-scale randomized controlled trials or prospective cohort studies. Until confirmatory evidence is available, current clinical anticoagulation practices should not be altered.

In our study, left ventricular ejection fraction (LVEF) remained significantly associated with SCI after adjustment for confounding factors, which aligns with the findings of Martinek et al. [25]. LVEF is a standard measure of left ventricular systolic function and impaired left ventricular function may therefore contribute to an increased risk of cerebral thromboembolic events.

Although extended ablation could theoretically increase embolic risk by prolonging procedure time and expanding the area of endocardial injury, clinical evidence regarding this association remains inconsistent [8, 25–27]. In this pilot study, the ablation strategy (pulmonary vein isolation alone vs. pulmonary vein isolation with additional ablation) was not significantly associated with the occurrence of SCI. One possible explanation is that, with standard and effective anticoagulation, extending ablation time and enlarging the ablation area may not substantially elevate embolic risk, a hypothesis consistent with earlier findings [28, 29]. However, given the limited sample size—particularly the small number of SCI events—the statistical power to detect significant differences for some risk factors was likely reduced. Therefore, larger prospective studies are warranted to confirm the relationship between ablation strategy and SCI risk, and to enable more robust multivariate analyses of relevant risk factors.

NSE, with a biological half-life of approximately 24 h, is considered a specific and sensitive biomarker of cerebral injury [15]. It has demonstrated prognostic value in various clinical settings, including the prediction of neurological outcomes after cardiovascular surgery, prognostication after cardiac arrest [16, 17], and assessment of neurocognitive impairment or neurological sequelae in stroke patients [15, 30, 31].

In this exploratory study, we monitored perioperative changes in serum NSE levels. The observed association between NSE_ipost, ΔNSE_ipost–pre, and SCI should be interpreted cautiously, as multiple comparisons were performed without adjustment, increasing the risk of type I error. Although both indicators differed significantly between the groups, these findings require validation in larger prospective studies. Following ablation, serum NSE levels exceeded the upper reference limit (17 ng/mL) in 17 patients (17.5%), including 6 patients in the non-SCI group. This observation raises the possibility that NSE may detect subtle neuronal or glial injury not visible on diffusion-weighted MRI, such as transient ischemic insults affecting small vessels below the resolution of conventional MRI. Indeed, prior studies have shown that DW-MRI identifies acute ischemic lesions in only 13–67% of patients with transient ischemic attacks [32]. It is also worth noting that the 6-mm slice thickness used in our MRI protocol may have limited the detection of small infarcts, which could partly explain elevated NSE levels in some patients without radiologically evident SCI. To further explore the biological plausibility of NSE as a biomarker of neuronal injury, we analyzed its relationship with SCI lesion burden, defined by the number and size of lesions. No significant correlation was observed, which may be attributable to the limited sample size, imprecise lesion volume measurements, or the possibility that NSE levels correlate more strongly with injuries not visible on DWI. Future prospective studies with larger cohorts should employ high-resolution MRI with three-dimensional volumetric measurements to enable a more precise evaluation of the dose–response relationship between NSE levels and the true burden of neuronal injury. ROC curve analysis was performed to assess the potential role of NSE in risk stratification for SCI. ΔNSE_ipost–pre alone exhibited high sensitivity (93.3%) but modest specificity (39.0%) and overall discriminatory ability (AUC 0.69, 95% CI 0.57–0.82), supporting its potential utility as an adjunctive screening tool rather than a standalone predictor. Given the current lack of effective early biochemical indicators, the high sensitivity of NSE suggests its utility for ruling out SCI, as patients with low NSE levels have a very low probability of developing the condition. When ΔNSE_ipost–pre was integrated into the reference model (incorporating ACT and LVEF), discriminative performance improved significantly. The AUC increased from 0.76 to 0.83 (P < 0.001), indicating enhanced discriminative capacity. Furthermore, the NRI was 0.758 (95% CI 0.232–1.264, P < 0.001), and the IDI was 0.084 (95% CI 0.037–0.112, P = 0.003), suggesting that the addition of ΔNSE_ipost–pre improved risk classification and model accuracy.

It should be noted that, due to the limited number of SCI events (n = 15), the multivariate regression models may be susceptible to overfitting despite meeting the goodness-of-fit criteria. Therefore, these model results should be considered exploratory and require validation in larger prospective cohorts. Prior to integrating NSE into routine clinical workflows, large-scale prospective studies are needed to determine its optimal cut-off value, clarify the ideal timing of measurement, and evaluate its relationship with long-term outcomes.

Limitations

Several limitations of this study should be acknowledged. First, the absence of preoperative cranial MRI as a baseline control represents a major limitation. Although the reported incidence of pre-existing SCI among candidates for AF ablation without a history of stroke is relatively low (approximately 1.6–8%) [33, 34], this does not exclude the possibility that some of the small lesions observed post-procedurally may have been present before ablation. Future well-designed prospective studies should routinely incorporate preoperative and postoperative multimodal MRI to more accurately identify new SCI lesions and evaluate their clinical significance. Second, this study lacked follow-up imaging to assess the persistence or evolution of SCI lesions over time. Whether a lesion persists, resolves, or progresses is crucial for understanding its clinical relevance. Future studies should implement long-term imaging follow-up protocols to evaluate the natural history of these small SCI lesions and their potential long-term impact on cognitive function. Third, the relatively small sample size (97 patients; enrollment rate 36%) limits the robustness of the findings and increases the risk of type II error. Over half of screened patients declined MRI or NSE testing, raising concerns about potential systematic selection bias—even though baseline characteristics showed no significant differences between enrolled and excluded patients. Unmeasured confounders may further limit generalizability of the results. Therefore, these findings should be interpreted cautiously and validated in larger, multicenter studies with higher enrollment rates. Fourth, as a non-specific biomarker, NSE levels may be influenced by confounding factors such as hemolysis, electrical cardioversion, and intraprocedural hypoperfusion. Although the hemolysis index was not routinely assessed in this study, samples with visible hemolysis were excluded. Intraoperative blood pressure remained stable throughout the procedure, and the use of electrical cardioversion was balanced between the two groups, thereby minimizing their potential impact. Nevertheless, future studies should rigorously control for these confounders and consider incorporating additional biomarkers, such as S100B, to enhance the specificity and accuracy of cerebral injury detection.

Conclusion

SCI remains a frequent complication following AF ablation. The findings of this pilot study indicate that, although the overall discriminatory ability of serum NSE for detecting SCI is moderate, its high sensitivity supports its potential utility as an adjunctive screening tool. In clinical practice, NSE measurement may serve as a post-procedural screening test, with cranial MRI reserved for patients exhibiting significant NSE elevation. This strategy may facilitate the identification and management of asymptomatic cerebral injury, particularly in resource-limited settings. Furthermore, incorporating NSE into risk prediction models may enhance their ability to stratify patients according to SCI risk. Given the exploratory nature of this study and its inherent methodological limitations, these findings should be considered preliminary and warrant validation in larger, well-designed prospective studies.

Abbreviations

SCI

Silent cerebral infarction

AFCA

Atrial fibrillation catheter ablation

NSE

Neuron-specific enolase

MRI

Magnetic resonance imaging

ROC

Receiver operating characteristic

NRI

Net Reclassification Index

IDI

Integrated Discrimination Improvement

ACT

Activated clotting time

AUC

Area under the curve

AF

Atrial fibrillation

LAA

Left atrial appendage

CPVI

Circumferential pulmonary vein isolation

SD

Standard deviation

IQR

Interquartile range

OR

Odds ratio

LVEF

Left ventricular ejection fraction

Authors’ contributions

Beibei Gao and LiqiGe drafted the manuscript and analysed the data. Chunqing Wang and Xudong Zhao collected the data. Zhongxiao Liu and Dandan Zuo analysed the magnetic resonance images.Fei Li and Chaoqun Zhang designed the study. Fei Li and Chaoqun Zhang revised the manuscript. All authors read and approved the final manuscript.

Funding

This work was partly supported by the Xuzhou Science and Technology Bureau Key R&D Program (Social Development Project, Grant No.KC23276).

Data availability

The datasets used and/or analyzed during the present study are available from the corresponding author on reasonable request.

Declarations

Ethics approval and consent to participate

The Institutional Review Board (IRB) of the Affiliated Hospital of Xuzhou Medical University approved this study protocol (XYFY-2019-KL-156), and all participants provided written informed consent prior to enrollment, and our study was performed in compliance with the ethical principles of the Declaration of Helsinki.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.