Insulin Resistance as a Predictor of Pulsed Field Ablation Success in Atrial Fibrillation (HOMA-PULSE)

Clinical decision regarding the need for repeat ablation (reablation) based on clinical status and findings during the follow-up examination at 4-5 months after primary PFA ablation. Binary outcome (yes/no). Decision made by the treating physician blinded to HOMA-IR values, based on: recurrence of symptomatic atrial fibrillation, symptomatic atrial tachycardias, or asymptomatic AF recurrence documented on ECG/Holter monitoring. Assessment includes: clinical symptom evaluation (EHRA score), 12-lead ECG, and minimum 24-hour Holter ECG monitoring.

Proportion of enrolled non-diabetic patients with abnormal insulin sensitivity defined as HOMA-IR > 2.5, calculated from fasting glucose and insulin levels measured on the day of ablation.

Odds ratio (OR) for the association between HOMA-IR and the indication for reablation, adjusted for relevant clinical covariates (age, sex, BMI, AF type, left atrial size, LVEF, CHA2DS2-VA score, comorbidities, and pharmacotherapy) using multivariable logistic regression analysis.

4a. Area Under ROC Curve (AUC) of HOMA-IR for Predicting Reablation Indication

Description:

Discriminatory performance of baseline HOMA-IR as a predictor of reablation indication, quantified as the area under the receiver operating characteristic (ROC) curve with 95% confidence interval. An AUC of 0.5 indicates no discrimination; 1.0 indicates perfect discrimination.

Time Frame: Baseline HOMA-IR to reablation decision at 4-5 months

Unit of Measure: AUC (dimensionless, range 0.0-1.0)

4b. Optimal HOMA-IR Cut-Off Value for Predicting Reablation Indication

Description:

Optimal threshold value of baseline HOMA-IR identified from the ROC curve using the Youden index (maximizing sensitivity + specificity - 1). Reported with 95% confidence interval obtained by bootstrapping.

Time Frame: Baseline HOMA-IR to reablation decision at 4-5 months

Unit of Measure: HOMA-IR v

Association between baseline Growth Differentiation Factor 15 (GDF-15) levels and the indication for reablation, including correlation with HOMA-IR (Pearson or Spearman coefficient). Assessment of additive prognostic value of combining GDF-15 with HOMA-IR compared to HOMA-IR alone (comparison of AUC values).

Description:

Discriminatory performance of a multiparametric predictive model integrating baseline HOMA-IR with clinical and biochemical parameters (lipid profile, high-sensitivity C-reactive protein, GDF-15, NT-proBNP, left atrial volume index, age, sex, and body mass index) for predicting reablation indication. The model is developed by multivariable logistic regression with internal validation by bootstrapping (1000 resamples). Discriminatory performance is quantified as the area under the receiver operating characteristic (ROC) curve, reported with 95% confidence interval. An AUC of 0.5 indicates no discrimination; 1.0 indicates perfect discrimination. Calibration (Hosmer-Lemeshow test, calibration slope) and reclassification (NRI, IDI) will be reported in the accompanying publication as secondary analyses.

Discriminatory performance of a deep learning neural network model (convolutional neural network) trained to extract electrophysiological features from intracardiac electrograms recorded during pulsed field ablation. Features include local atrial activation patterns, signal amplitude, fractionation indices, and conduction velocity. The model output is evaluated as a predictor of reablation indication. Performance is quantified as the area under the receiver operating characteristic curve with 95% confidence interval from internal cross-validation. Correlation of the neural network feature score with baseline HOMA-IR (Spearman coefficient) and improvement over HOMA-IR alone (integrated discrimination improvement) will be reported as secondary analyses in the publication.

Spearman rank correlation coefficient between baseline HOMA-IR (calculated from fasting plasma glucose and insulin on the day of ablation) and total pulsed field ablation procedure time (defined as the interval from vascular access to sheath removal, measured by the electrophysiology recording system). The analysis tests whether higher insulin resistance is associated with longer procedure duration, potentially reflecting more complex atrial substrate. Reported with 95% confidence interval. Additional procedural parameters (fluoroscopy time, number of PFA applications, acute pulmonary vein isolation success, need for additional ablation lines) will be reported in the accompanying publication as supplementary correlations.

Spearman rank correlation coefficient between baseline HOMA-IR and left atrial volume index, measured by transthoracic echocardiography using the biplane area-length or Simpson method in apical 4- and 2-chamber views at end-systole and indexed to body surface area. The analysis tests whether higher insulin resistance is associated with greater left atrial remodeling, a key substrate for atrial fibrillation recurrence. Reported with 95% confidence interval. Additional echocardiographic parameters (left atrial area, left ventricular ejection fraction, left atrial strain) will be reported in the accompanying publication as supplementary correlations.

Count of participants with reablation indication at 4-5 months, stratified jointly by baseline HOMA-IR tertile (low, middle, high) and baseline pharmacotherapy class at the time of index ablation (beta-blockers, ACE inhibitors or angiotensin receptor blockers, statins, SGLT2 inhibitors or GLP-1 receptor agonists). The primary analysis evaluates whether pharmacotherapy modifies the association between HOMA-IR and reablation risk. Interaction between HOMA-IR (continuous) and each pharmacotherapy class will be formally tested in logistic regression models with interaction terms, with results reported in the accompanying publication.