H-type hypertension and U-shaped body mass index predict atrial fibrillation recurrence after catheter ablation: Insights from a 3-year follow-up

Abstract

Background

H-type hypertension, body mass index (BMI), and metabolic biomarkers are potential biomarkers of long-term (≥3 years) recurrence of post–radiofrequency catheter ablation (RFCA) atrial fibrillation (AF).

Objective

This study aimed to investigate the abovementioned modifiable risk factors in post-RCFA patients.

Methods

Data of 827 patients who underwent first-time RFCA (2016–2021) for AF were retrospectively analyzed. Recurrence was defined as >30-second atrial tachyarrhythmia after a 3-month blanking period. Cox regression and restricted cubic spline models were used to assess the risk factors and nonlinear relationships.

Results

Among the 827 participants, 248 (29.99%) experienced AF recurrence during the follow-up period. Compared with the nonrecurrence group, the recurrence group was older, with higher proportions of smoking, drinking, hypertension (mainly H-type), and diabetes; higher serum glucose and homocysteine levels; and lower serum total cholesterol and triglyceride levels at admission (all P < .05). Adjusted Cox regression multivariate analysis identified smoking, older age, history of diabetes, and H-type hypertension as risk factors for post-RFCA AF recurrence. Restrictive cubic spline analysis revealed a “U-shaped” relationship between BMI and postoperative AF recurrence (analysis of variance test, P < .01). The surv_cutpoint function, which divided BMI and age by the maximum selection rank system, identified optimal outcome-cutoff values (BMI <21.3 kg/m², age ≥70 years), which indicated the highest AF recurrence risk in the Kaplan–Meier survival curve.

Conclusion

H-type hypertension independently predicts post-RFCA AF recurrence. The U-shaped BMI-AF recurrence relationship indicates that both high and low BMI elevate recurrence risk. Clinicians should monitor high-risk patients, optimize weight management, and treat hyperhomocysteinemia, especially in H-type hypertension.

Graphical abstract

Key Findings.

• ▪H-type hypertension was identified as an independent risk factor for long-term atrial fibrillation (AF) recurrence after radiofrequency catheter ablation (RFCA).
• ▪A U-shaped relationship was observed between body mass index (BMI) and AF recurrence, with the highest risk in patients with a BMI of <21.3 kg/m².
• ▪Age of ≥70 years was established as the critical risk threshold for AF recurrence after RFCA.
• ▪Patients with a history of smoking, diabetes, uric acid, and older age had significantly higher recurrence risks, and the role of elevated serum homocysteine, left atrial enlargement, and glucose levels in AF recurrence was highlighted, which suggests potential metabolic intervention targets for AF.
• ▪The need for proactive weight management and close monitoring of high-risk patients, particularly those with hyperhomocysteinemia and H-type hypertension, was emphasized.

Introduction

Atrial fibrillation (AF), the most prevalent sustained cardiac arrhythmia worldwide, confers a higher clinical and economic burden owing to its increasing incidence. AF affects 6% of individuals older than 65 years and approximately 10% of octogenarians.^1,2 In the next 5 decades, AF prevalence is estimated to increase 2-fold, in tandem with population aging.³ This AF epidemic has catastrophic consequences, given that patients face 5- and 3-fold higher risks of stroke- and heart failure-related mortality, respectively, than their counterparts with sinus rhythm,⁴ which underscores the urgent need for effective therapies for AF.

Despite being a cornerstone of interventional rhythm control, radiofrequency catheter ablation (RFCA) has suboptimal long-term efficacy. Even with technological advances, 30%–55% of patients experience recurrence within 3 years,⁵ which frequently warrants repeat procedures that increase health care costs and patient morbidity. Crucially, recurrence is driven by multifactorial pathomechanisms that are poorly predictable based on an analysis of conventional clinical parameters. This highlights a critical knowledge gap in personalized postablation management. A key underexplored area involves metabolic–arrhythmia interactions, particularly H-type hypertension, which is a distinct phenotype defined by coexisting hypertension and hyperhomocysteinemia (HHcy) (serum homocysteine [Hcy] ≥10 μmol/L). Epidemiologically, 75% of Chinese patients with hypertension exhibit this dual risk profile.⁶ Mechanistically, the synergistic effects of pressure overload and Hcy-induced oxidative stress may perpetuate atrial remodeling through fibrosis and endothelial dysfunction.^7,8 Although folate supplementation is potentially effective in reducing the risk of recurrence,⁹ the specific prognostic value of H-type hypertension has not been quantified in large ablation cohorts. Furthermore, the metabolic determinants of recurrence extend beyond hypertension, with considerable controversy regarding the effects of body mass index (BMI). Although obesity is traditionally implicated, emerging research data suggest that underweight patients may paradoxically face elevated risks.^10,11

This study addresses key gaps by combining risk assessment and restricted cubic spline (RCS) analyses. First, adjusted Cox models quantified the prognostic value of H-type hypertension. Second, RCS analyses revealed nonlinear links among BMI extremes, Hcy, and arrhythmia recurrence. These insights enhance risk prediction and support personalized metabolic optimization for secondary prevention.

Methods

Study population and definitions

This single-center retrospective cohort study involved an analysis of data from patients with AF who underwent catheter ablation at Zhujiang Hospital, Southern Medical University, between January 2016 and December 2021. Clinical data were retrieved from 3 integrated electronic health record systems: (1) the YiduCloud Medical Data Intelligence Platform (comprising structured demographic and diagnostic data), (2) the institutional Picture Archiving and Communication System (for imaging reports), and (3) the hospital-wide electronic medical record database (containing laboratory results and procedural details). The study protocol was approved (approval number: 2021-KY-143-01) by the Institutional Review Board of Zhujiang Hospital and adhered to the principles of the Declaration of Helsinki.

The research project encompassed 4 methodologically sequential components to ensure comprehensive data integrity and utility: (1) systematic extraction and integration of multisource clinical data from hospital information systems, including electronic health records, laboratory databases, and imaging archives; (2) rigorous standardization processes involving demographic parameter normalization (eg, sex, age, ethnicity), laboratory metric unit conversion (eg, mg/dL to mmol/L for lipid profiles), and diagnostic code alignment with International Classification of Diseases, Tenth Revision, specifications; (3) development of hypertension-specific data schemas incorporating longitudinal blood pressure measurements, antihypertensive medication regimens, and H-type hypertension biomarkers (Hcy ≥10 μmol/L); and (4) implementation of analytical modules within a unified research platform. Leveraging extract-transform-load pipelines through the institutional big data infrastructure enabled seamless synchronization and harmonization of heterogeneous data formats (structured, semistructured, and unstructured) across clinical subsystems. This architecture facilitated automated data aggregation, validation, and cleansing, with Python 3.11.1 scripts (Python Software Foundation, Wilmington, DE) executing quality-control protocols, including outlier detection (eg, exclusion of systolic blood pressure >250 mm Hg), missing-value imputation via k-nearest neighbor algorithms, and temporal consistency verification. The curated dataset, which comprised demographic profiles, longitudinal medical records (admission-to-discharge trajectories), diagnostic classifications, and multidimensional laboratory/imaging reports, was ultimately stored in an SQL Server relational database with role-based access controls, which ensured both scientific rigor and Health Insurance Portability and Accountability Act–compliant data security.¹²

The inclusion criteria were as follows: (1) aged 21–79 years with documented paroxysmal or persistent AF, (2) scheduled for first-time RFCA, (3) demonstrated ≥80% adherence to scheduled follow-ups during preliminary screening, and (4) provided a written informed consent. The exclusion criteria were as follows: (1) congenital heart anomalies, (2) moderate-to-severe valvulopathy (aortic or mitral valve areas <1.5 cm²) or prosthetic valve implantation, (3) cardiomyopathies (ejection fraction <40% or wall thickness >15 mm), and (4) uncontrolled hyperthyroidism (thyroid-stimulating hormone <0.1 mIU/L). According to AF classification, patients were categorized as having persistent AF, defined as AF lasting >7 days, including episodes terminated by pharmacologic or electrical cardioversion after ≥7 days.

Under fluoroscopic guidance, multipolar electrode catheters were percutaneously advanced via femoral venous/arterial or subclavian access to the cardiac chambers. Three-dimensional electroanatomic mapping (CARTO [Biosense Webster, Diamond Bar, CA] or EnSite systems [Abbott, St. Paul, MN]) localized the arrhythmogenic foci, which were ablated using RFA (Stockert generator, 30–35 W, 43°C; Stockert GmbH, Breisgau, Germany) targeting the pulmonary venous ostia, tricuspid isthmus, or left atrial roof lines, as appropriate. Ablation strategies were individualized based on arrhythmogenic substrates and anatomical variations as follows: (1) circumferential pulmonary vein isolation with entrance/exit block verification, (2) cavotricuspid isthmus ablation for typical atrial flutter, and (3) linear lesions for persistent AF substrate modification.

Follow-up and definition of AF recurrence

AF recurrence was rigorously defined as any documented atrial tachyarrhythmia lasting ≥30 seconds on a 12-lead electrocardiogram (ECG) or ambulatory Holter monitoring after the postablation 3-month blanking period. To mitigate early recurrence, all patients received protocol-directed antiarrhythmic drug therapy (class I/III agents) for the initial 3 months, and subsequent antiarrhythmic drug continuation was determined through shared decision making between clinicians and patients. A structured follow-up protocol was implemented as follows: (1) intensive monitoring phase (3–12 months), serial arrhythmia surveillance using 24-hour Holter ECG (quarterly) and 14-day event recorders (semiannually), and (2) long-term phase (>12 months), biannual assessments combining in-person consultations (physical examination, 12-lead ECG, 24-hour Holter monitoring, etc) and telephonic interviews conducted by dedicated cardiac nurses and supplemented by patient-initiated symptom-triggered ECG recordings. In adherence to the modified 2012 Heart Rhythm Society/European Heart Rhythm Association/European Cardiac Arrhythmia Society expert consensus statement guidelines, 24-hour Holter monitoring was conducted semiannually for 2 years, annually for 2–5 years, and biennially after 5 years. Participants received portable ECG devices (eg, KardiaMobile) for real-time arrhythmia documentation and were instructed to promptly report symptomatic episodes to the core laboratory for adjudication of the events.

Statistical analysis

All analyses were performed using R software (v4.2.1; R Foundation for Statistical Computing, Vienna, Austria) with a 2-tailed significance threshold of P < .05. Continuous variables were assessed for normality using the Shapiro–Wilk test. Normally distributed data were expressed as mean ± standard deviation and compared using the Student t test. Non-normally distributed variables were summarized as the median [interquartile range] and analyzed using the Mann–Whitney U test. Categorical variables were presented as frequencies (percentages), with intergroup comparisons conducted using the χ² test or Fisher’s exact test, as appropriate. Missing laboratory values (which occurred in <5% of cases) were addressed via k-nearest neighbor imputation (k = 10), which estimates missing entries based on similarity metrics from complete cases within the same metabolic cluster. This method preserves data distribution patterns while minimizing bias from complete-case deletion, as validated by sensitivity analyses comparing the imputed and nonimputed datasets.

To mitigate confounding bias, we used directed acyclic graphs informed by clinical expertise and literature evidence to select covariates for the multivariable Cox proportional hazards regression. Survival analyses used the survival package (v3.4-0) for time-to-event modeling, and RCSs (3 knots) from the rms package (v6.4.1) were used to examine the nonlinear BMI effects. Spline nonlinearity was formally tested using likelihood ratio tests to compare the linear and spline models. Optimal BMI/age thresholds were determined using the survminer package’s (v0.4.9) maximally selected rank statistics [surv_cutpoint()], and Kaplan–Meier curves were generated using ggsurvplot() to visualize differences in recurrence-free survival.

Results

Baseline characteristics of the study population

The study cohort comprised 827 eligible patients, of whom 248 (29.99%) experienced AF recurrence during the follow-up period. The clinicodemographic profiles demonstrated substantial heterogeneity between the groups (Table 1). Patients with recurrence were significantly older than those without recurrence (64.7 ± 11.8 vs 59.6 ± 11.5 years, P < .001), with a 2.5-fold higher proportion of octogenarians (17.3% vs 6.9%, P < .001). Despite comparable mean BMI in both groups (24.6 ± 3.69 vs 24.9 ± 3.28, P = .205), participants with recurrence exhibited pronounced metabolic and behavioral risk profiles: smoking (33.1% vs 20.0%, P < .001) and alcohol consumption (24.6% vs 17.6%, P = .027) were more prevalent, whereas H-type hypertension (49.2% vs 34.2%, P < .001) and diabetes (37.5% vs 15.2%, P < .001) predominated. In patients with H-type hypertension, the prevalence of AF recurrence was 38.1%, whereas, in patients without H-type hypertension, the prevalence of AF recurrence was 24.9% (P < .001). In the nonrecurrence group, the prevalence of H-type hypertension was 34.2%, and in the recurrence group, it was 49.2%.

Table 1.

Comparison of baseline characteristics between the recurrence and nonrecurrence groups

Variable	Nonrecurrence (n = 579)	Recurrence (n = 248)	P value (overall)
Sex			.434
Male	379 (65.5%)	170 (68.5%)
Female	200 (34.5%)	78 (31.5%)
Age (y)	59.6 (11.5)	64.7 (11.8)	<.001
<60	261 (45.1%)	80 (32.2%)	<.001
60–75	278 (48.0%)	125 (50.6%)
≥75	40 (6.9%)	43 (17.3%)
BMI	24.9 (3.28)	24.6 (3.69)	.205
<24	225 (38.9%)	114 (46.0%)	.064
24–28	263 (45.4%)	91 (36.7%)
≥28	91 (15.7%)	43 (17.3%)
Systolic BP (mm Hg)	128 (18.5)	130 (19.3)	.146
Diastolic BP (mm Hg)	76.4 (11.6)	77.6 (12.0)	.214
Smoking history	116 (20.0%)	82 (33.1%)	<.001
Drinking history	102 (17.6%)	61 (24.6%)	.027
Hypertension history	290 (50.1%)	152 (61.3%)	.004
Hypertension classification			<.001
No hypertension	289 (49.9%)	96 (38.7%)
Simple hypertension	92 (15.9%)	30 (12.1%)
H-type hypertension	198 (34.2%)	122 (49.2%)
Diabetes history	88 (15.2%)	93 (37.5%)	<.001
Admission random glucose (mmol/L)	6.04 (1.83)	6.80 (2.31)	<.001
≥11.1 mmol/L	466 (80.55%)	163 (65.7%)	<.001
Total cholesterol (mmol/L)	4.61 (0.98)	4.16 (0.87)	<.001
<5.2	426 (73.6%)	216 (87.1%)
5.2–6.2	116 (20.9%)	28 (11.3%)
≥6.2	37 (6.4%)	4 (1.6%)
Triglycerides (mmol/L)	1.77 (1.29)	1.60 (1.24)	.075
<1.7	365 (63.0%)	182 (73.4)	.016
1.7–2.3	96 (16.6%)	30 (12.15)
>2.3	118 (20.4%)	36 (14.4%)
HDL cholesterol (mmol/L)	1.24 (0.31)	1.16 (0.27)	<.001
<1	135 (23.3%)	75 (30.2%)	.036
≥1.0	444 (76.7%)	173 (69.8%)
Homocysteine (μmol/L)	11.01 (7.28)	13.92 (6.27)	<.001
<10	211 (36.4%)	60 (24.2%)
≥10	368 (63.6%)	180 (75.8%)
LDL cholesterol (mmol/L)	2.75 (0.78)	2.46 (0.66)	<.001
<1.8	59 (10.2%)	35 (14.1%)	.001
1.8–2.6	205 (35.4%)	112 (45.2%)
2.6–3.4	198 (34.2%)	76 (30.6%)
>3.4	117 (20.2%)	25 (10.1)
Serum potassium (mmol/L)	4.05 (0.53)	4.03 (0.31)	.547
Serum sodium (mmol/L)	141 (2.40)	141 (2.82)	<.001
Serum chloride (mmol/L)	105 (2.84)	104 (3.05)	<.001
Cavotricuspid isthmus block	47 (8.1%)	27 (10.9%)	.201
Left atrial lines block	55 (9.5%)	34 (13.7%)	.073
Estimated glomerular filtration rate∗ (%)	75.3 (1.5)	65.9 (22.2)	<.001
Uric acid	387.06 (108.88)	421.5 (121.62)	<.001
Persistent atrial fibrillation	128 (22.1%)	68 (27.4)	.212
Left atrium size (mm)	37.7 (8.3)	41.1 (8.7)	<.001
Fibrinogen	3.29 (1.7)	3.07 (1.72)	.001
Left ventricular ejection fraction (%)	58.2 (5.9)	53.5 (9.7)	<.001
Follow-up time (mo)	53.9 (31.0)	20.0 (20.6)	<.001

BMI = body mass index; BP = blood pressure; LDL = low-density lipoprotein; HDL = high-density lipoprotein.

^∗Kidney function was evaluated using the baseline creatinine-based estimated glomerular filtration rate according to the Chronic Kidney Disease Epidemiology Collaboration equation.

Laboratory analyses revealed distinct metabolic patterns in the patients. The recurrence group demonstrated elevated blood glucose (6.80 ± 2.31 vs 6.04 ± 1.83 mmol/L, P < .001) and serum Hcy levels (14.7 ± 5.48 vs 13.6 ± 5.60 μmol/L, P = .031) at admission. Paradoxically, lower atherogenic lipid profiles were observed in this group: total cholesterol (4.16 ± 0.87 vs 4.61 ± 0.98 mmol/L, P < .001), low-density lipoprotein (LDL) cholesterol (2.26 ± 0.66 vs 2.75 ± 0.78 mmol/L, P < .001), and triglyceride levels of ≥2.3 mmol/L (14.4% vs 20.4%, P = .016). Conversely, high-density lipoprotein cholesterol levels showed a marginal increase in participants with AF recurrence (1.26 ± 0.27 vs 1.24 ± 0.31 mmol/L, P < .001).

Multivariable Cox regression analysis

In the multivariate Cox regression analysis (model 1 for H-type hypertension) (Table 2), significant independent predictors of AF recurrence after RFCA included H-type hypertension (hazard ratio [HR] 1.673, 95% confidence interval [CI] 1.146–2.442, ∗P∗ = .008), smoking history (HR 1.986, 95% CI 1.485–2.656, ∗P∗ < .001), age of ≥75 years (vs <60 years: HR 2.212, 95% CI 1.453–3.367, ∗P∗ < .001), elevated uric acid (per unit increase: HR 1.001, 95% CI 1.000–1.002, ∗P∗ = .035), and left atrial enlargement (per mm increase: HR 1.018, 95% CI 1.000–1.037, ∗P∗ = .045). Paradoxically, diabetes mellitus was associated with reduced recurrence risk (HR 0.355, 95% CI 0.271–0.466, ∗P∗ < .001). Moreover, modifiable metabolic factors demonstrated protective effects: BMI of 24–28 kg/m² (vs <24 kg/m²: HR 0.650, 95% CI 0.489–0.865, ∗P∗ = .003) and triglycerides of 1.7–2.3 mmol/L (vs <1.7 mmol/L: HR 0.547, 95% CI 0.366–0.817, ∗P∗ = .003). No significant association was observed for a BMI of ≥28 kg/m² (∗P∗ = .459). Trends toward increased recurrence risk were noted for cavotricuspid isthmus block (HR 1.499, ∗P∗ = .055), left atrial lines block (HR 0.858, P = .064), and simple hypertension (HR 1.214, ∗i∗ = 0.052).

Table 2.

Results of multivariate Cox regression analysis for the identification of risk factors of post-RFCA atrial fibrillation recurrence (model 1 for H-type hypertension)

Variable	P value	HR (95% CI)
Cavotricuspid isthmus block	.075	1.499 (0.992–2.265)
Left atrial lines	.064	0.858 (0.572–1.051)
Estimated glomerular filtration rate∗	.130	0.994 (0.986–1.002)
Uric acid	.035	1.001 (1.000–1.002)
Fibrinogen	.613	1.036 (0.904–1.187)
Left ventricular ejection fraction	.223	0.288 (0.037–2.243)
Diabetes mellitus	<.001	0.355 (0.271–0.466)
BMI, kg/m2
<24	Reference	Reference
24–28	.003	0.650 (0.489–0.865)
≥28	.459	0.872 (0.606–1.254)
Left atrium size (mm)	.045	1.018 (1.000–1.037)
Age, y
<60	Reference	Reference
60–75	.109	1.269 (0.948–1.698)
≥75	<.001	2.212 (1.453–3.367)
Triglycerides, mmol/L
<1.7	Reference	Reference
1.7–2.3	.003	0.547 (0.366–0.817)
≥2.3	.051	0.690 (0.475–1.001)
Smoking history	<.001	1.986 (1.485–2.656)
Persistent atrial fibrillation	.067	1.363 (0.979–1.899)
Simple hypertension	.052	1.214 (0.971–1.693)
H-type hypertension	.008	1.673 (1.146–2.442)

BMI = body mass index; CI = confidence interval; HR = hazard ratio; RFCA = radiofrequency catheter ablation.

^∗Kidney function was evaluated using the baseline creatinine-based estimated glomerular filtration rate according to the Chronic Kidney Disease Epidemiology Collaboration equation.

When serum Hcy levels were included as a covariate in model 2 (Table 3), Hcy was identified as an independent risk factor for AF recurrence (HR 1.043, 95% CI 1.022–1.064, P = .000). The significance of other factors remained largely consistent with that in model 1.

Table 3.

Results of multivariate Cox regression analysis for the identification of risk factors of post-RFCA atrial fibrillation recurrence (model 1 for H-type homocysteine)

Variable	P value	HR (95% CI)
Cavotricuspid isthmus block	.059	1.554 (0.984–2.457)
Left atrial lines	.067	1.243 (0.912–1.996)
Estimated glomerular filtration rate∗	.117	0.993 (0.985–1.002)
Uric acid	.203	1.001 (1.000–1.002)
Fibrinogen	.653	1.029 (0.894–1.247)
Left ventricular ejection fraction	.171	0.257 (0.036–1.858)
Diabetes mellitus	<.001	0.415 (0.312–0.551)
BMI, kg/m2
<24	Reference	Reference
24–28	.040	0.720 (0.527–0.985)
≥28	.353	0.828 (0.556–1.233)
Left atrium size (mm)	.129	1.015 (0.995–1.035)
Age, y
<60	Reference	Reference
60–75	.888	1.023 (0.742–1.412)
≥75 y	.024	1.696 (1.071–2.686)
Triglycerides, mmol/L
<1.7	Reference	Reference
1.7–2.3	.023	0.615 (0.403–0.936)
≥2.3	.185	0.763 (0.512–1.139)
Smoking history	<.001	1.786 (1.314–2.428)
Persistent atrial fibrillation	.067	1.363 (0.979–1.899)
Homocysteine	<.001	1.043 (1.022–1.064)

BMI = body mass index; CI = confidence interval; HR = hazard ratio; RFCA = radiofrequency catheter ablation.

^∗Kidney function was evaluated using the baseline creatinine-based estimated glomerular filtration rate according to the Chronic Kidney Disease Epidemiology Collaboration equation.

Nonlinear association of BMI with AF recurrence

BMI demonstrated a U-shaped relationship with AF recurrence risk (P_nonlinear < .01) (Figure 1). Compared with the reference group (24 kg/m² ≤ BMI <28 kg/m²), underweight patients (BMI <21.3 kg/m²) had a 32% higher risk of AF recurrence (HR 1.32, 95% CI 1.12–1.56), whereas obesity (BMI ≥28 kg/m²) showed neutral effects (HR 0.944, P = .756). RCS analysis confirmed this nonlinearity, with the nadir risk of AF recurrence at a BMI of 24–28 kg/m².

Figure 1.

Restricted cubic spline analysis showing a “U-shaped” relationship between BMI and the postoperative risk of AF recurrence, with significant differences in the nonlinear relationship (P for nonlinearity < .01). AF = atrial fibrillation; BMI = body mass index; CI = confidence interval.

Furthermore, Hcy levels demonstrated a nonlinear relationship with the AF recurrence risk after RFCA (P_nonlinear < .001) (Figure 2). RCS analysis indicated that the risk of AF recurrence started increasing significantly with Hcy levels of >12.2 μmol/L. The RCS curve further confirmed that Hcy levels of <12.2 μmol/L were associated with relatively stable and lower AF recurrence risks, and levels beyond this cutoff value conferred a steep, dose-dependent increase in the risk of AF recurrence.

Figure 2.

Threshold effect of serum homocysteine levels on atrial fibrillation recurrence risk after RFCA. Restricted cubic spline analysis (P for nonlinearity < .001) identified a critical threshold at 12.2 μmol/L, below which risk was stable and lower and above which risk increased dose dependently. CI = confidence interval; RFCA = radiofrequency catheter ablation.

Risk stratification by BMI and age

Maximally selected rank statistics identified critical thresholds for the risk of AF recurrence: BMI of 21.3 kg/m² and age of 70 years. Kaplan–Meier analysis revealed a striking divergence in recurrence-free survival (log-rank P < .0001) (Figure 3). The highest risk cohort—patients with a BMI of <21.3 kg/m² and age of ≥70 years—exhibited a 3.1-fold increased hazard compared with younger individuals with a BMI of ≥21.3 kg/m².

Figure 3.

Kaplan–Meier survival curves based on the optimal cutoff points for BMI and age identified by the surv_cutpoint function. The group with a BMI of <21.3 kg/m² and age of ≥70 years had the highest risk of AF recurrence (log-rank analysis showed significant intergroup differences among the survival curves, P < .0001). BMI = body mass index.

Discussion

The progression and recurrence of AF are driven by complex interactions between cardiovascular risk factors and atrial substrate remodeling. Our study demonstrates that the optimized management of modifiable metabolic factors—particularly H-type hypertension and body-weight extremes—may substantially attenuate the postablation risk of AF recurrence. Among 827 patients with AF who underwent RFCA, multivariate Cox regression revealed that H-type hypertension was an independent predictor of recurrence, whereas conventional hypertension showed no significant association. Notably, RCS analysis uncovered a U-shaped BMI-recurrence relationship, with a nadir risk at a BMI of 24–28 kg/m² and peak hazard in underweight older individuals.

Revisiting obesity and AF recurrence: Beyond linear assumptions

Although obesity has long been recognized as a key driver of AF pathogenesis through epicardial adipose tissue–mediated structural remodeling,¹³ our findings challenge the conventional linear BMI-risk paradigm. The U-shaped relationship between BMI and postablation recurrence risk (P_nonlinear < .01) revealed 2 critical insights.

Obesity-driven arrhythmogenesis

Excessive adiposity (BMI ≥28 kg/m²) perpetuates AF recurrence via paracrine signaling (eg, interleukin-6, leptin) and direct epicardial fat infiltration, which disrupts connexin-43 gap junctions and prolongs atrial conduction times.¹⁴

Underweight paradox

Counterintuitively, patients with a BMI of <21.3 kg/m² exhibited a 32% higher recurrence risk (HR 1.32, 95% CI 1.12–1.56) than the reference group (24 ≤ BMI < 28 kg/m²). This may reflect malnutrition-induced sarcopenia, where skeletal muscle loss reduces the production of antifibrotic myokines (eg, irisin), whereas hypokalemia and hypomagnesemia destabilize action potential dynamics.^15,16

These bidirectional risks necessitate a paradigm shift from universal weight loss to BMI-stratified management. For obesity (BMI ≥28 kg/m²), bariatric surgery reduced AF burden by 48% in the LEGACY trial,¹⁴ likely through adipokine normalization. For underweight patients (BMI <21.3 kg/m²), protein supplementation (≥1.2 g/kg/d) and resistance training preserved lean mass and attenuated recurrence in pilot studies.¹⁷ This precision approach mirrors the principles of chronic disease management that have been applied in hypertension and diabetes, where individualized targets (eg, hemoglobin A1C [HbA1c] <7%) optimize outcomes.¹⁸ Future guidelines should integrate BMI thresholds with comorbidity profiles to formulate patient-specific lifestyle recommendations.

Hcy metabolism and H-type hypertension: Implications for AF recurrence

Hcy, a sulfur-containing amino acid derived from methionine demethylation,¹⁸ accumulates in the systemic circulation owing to disruptions in the remethylation and transsulfuration pathways. HHcy worsens cardiovascular pathology through 4 key mechanisms: It causes oxidative endothelial injury by generating reactive oxygen species and depleting nitric oxide, triggers vascular remodeling via nuclear factor kappa B activation and matrix metalloproteinase-9 upregulation, creates a prothrombotic state by modifying fibrinogen, and induces lipid peroxidation by altering Hcy-LDL complexes.¹⁹ In China, HHcy is comorbid with hypertension in 75% of patients and defines the distinct phenotype of H-type hypertension.²⁰ Our findings corroborate the arrhythmogenic role of H-type hypertension: patients with H-type hypertension exhibited a higher AF recurrence risk than those with isolated hypertension. In addition, when serum Hcy levels were included as a covariate, Hcy emerged as an independent risk factor for AF recurrence (HR 1.043). RCS analysis revealed a nonlinear relationship between Hcy levels and AF recurrence risk (P_nonlinear < .001). The risk of AF recurrence increased significantly when Hcy levels surpassed 12.2 μmol/L, with levels below this threshold associated with lower and more stable recurrence risks. This synergy likely stems from Hcy-based amplification of angiotensin II–induced calcium mishandling (eg, RyR2 hyperphosphorylation and SERCA2a downregulation), which destabilizes atrial electrophysiology.²¹ Routine Hcy measurement should be mandated in patients with hypertension and AF, particularly MTHFR C677T carriers, to enable early identification of high-risk subgroups; targeted intervention with folate supplementation (0.8 mg/d) reduced AF recurrence by 38% in H-type hypertension cohorts,²⁰ whereas methionine-restricted diets (<1.5 g/d) attenuated postablation Hcy surges in pilot trials.²¹

Epidemiologic studies have revealed that, compared with individuals without diabetes, patients with diabetes have a 40% higher AF incidence, with an annual risk escalation of 3%.^22,23 This association is mediated through multifactorial pathways: hyperglycemia-induced oxidative stress accelerates atrial fibrosis via advanced glycation end product and its receptor axis activation, insulin resistance promotes lipotoxicity-driven mitochondrial dysfunction, and autonomic imbalance enhances ectopic pulmonary venous activity.²³ Notably, abnormal glucose metabolism exacerbates intra-atrial conduction heterogeneity (low-voltage zones: 28% vs 12% in nondiabetics, P < .01) and doubles postablation recurrence rates.²⁴ Despite controversy regarding ablation efficacy in patients with diabetes, a 1121-patient cohort found no increased AF risk after atrial flutter ablation; furthermore, our multivariate Cox analysis confirmed diabetes as an independent predictor of recurrence (HR 2.482, P < .001), which aligns with the results of 83% of previous studies.²⁵ These findings underscore stringent glycemic control (HbA1c <7%) as a pivotal requirement for secondary prevention, which can be augmented by sodium-glucose cotransporter-2 inhibitors to attenuate AF burden through ketone-mediated electrophysiological stabilization.²³

Age-dependent AF recurrence and geriatric management strategies

Although the incidence of AF escalates exponentially with age and doubles every decade,²⁶ the relationship between aging and postablation recurrence remains contentious. Despite previously reported comparable long-term outcomes in older patients (age ≥75 years) with paroxysmal AF undergoing ablation,²⁶ our data revealed a 2.6-fold higher recurrence risk in this population (HR 2.591, P < .001), which was attributable to their elevated comorbidity burden and prevalent H-type hypertension (49.2% vs 34.2%). Despite increased procedural complexities,²⁷ ablation remains beneficial (78% symptom improvement at 3 years).²⁶ Tailored interventions include intensified monitoring with quarterly 7-day Holter tests, optimizing comorbidities by targeting HbA1c of <7% and Hcy of <10 μmol/L, and reducing QT-prolonging antiarrhythmics while favoring dronedarone or ablation-first strategies. These approaches address age-related vulnerabilities such as fibrosis, metabolic issues, and drug response variability to lower recurrence risks.

Emerging evidence implicates dysregulated lipid metabolism as a novel determinant of postablation AF recurrence. Elevated triglyceride levels (>2.3 mmol/L) and lipoprotein(a) potentiate arrhythmogenesis through oxidized phospholipid-mediated NLRP3 inflammasome activation in atrial macrophages, whereas low high-density lipoprotein cholesterol (<1.0 mmol/L) reflects impaired reverse cholesterol transport, which exacerbates atrial oxidative stress.^28,29 Although corroborated by findings from European cohorts that demonstrate U-shaped LDL cholesterol–AF relationships,30, 31, 32 our findings contradict conventional cardioprotective lipid paradigms; therefore, mechanistic studies are required to delineate the causal pathways (eg, lipidomics profiling of ceramide-to-oxysterol ratios) and refine risk stratification.

This study provides critical insights into the metabolic predictors of AF recurrence. Nonetheless, this study had a few limitations. First, the retrospective design inherently confers a risk of residual confounding from unmeasured variables (eg, dietary folate or B-vitamin status, MTHFR genotype prevalence, duration of AF history, DNA methylation at the PITX2 loci,³³ and CHA₂DS₂-VASc), which is a concern that is partially mitigated by multivariable adjustment but precludes causal inference.³⁴ Second, temporal heterogeneity in ablation techniques (2016–2021) introduced variability, given that early cases used conventional radiofrequency ablation without contact-force sensing, whereas later procedures incorporated cryoballoon pulmonary vein isolation, which demonstrated superior durability in recent meta-analyses.³⁵ Third, asymptomatic recurrences were likely underdetected owing to the reliance on intermittent Holter monitoring (sensitivity: 68% vs 96% for implantable loop recorders),³⁶ which potentially attenuated the observed risk estimates.

Conclusion

This study comprehensively investigated the risk factors for AF recurrence after RFCA in a large cohort of patients with detailed clinical and laboratory data. Our primary findings indicate that H-type hypertension is a significant independent risk factor for AF recurrence, highlighting the importance of HHcy treatment in postablation management of patients with AF. The nonlinear “U-shaped” relationship between BMI and AF recurrence suggests that both underweight and overweight states may contribute to an increased AF recurrence risk. This study revealed that smoking, older age, and a history of diabetes are significant predictors of AF recurrence, reinforcing the need for comprehensive lifestyle and chronic disease management in these patients.