Mechanistic insights into the learning curve of cryoballoon ablation: ablation temperature mediates the effect of operator experience on radiation exposure

Vollständiger Abstract

Worum geht es in dieser Arbeit?

Background Cryoballoon ablation (CBA) for atrial fibrillation (AF) is associated with reduced procedural radiation exposure as operator experience accumulates, yet the mechanistic underpinning of this improvement remains incompletely characterized. We hypothesized that ablation temperature parameters, reflecting balloon-tissue contact quality, serve as a quantifiable mediator linking operator experience to radiation dose. Methods We retrospectively analyzed 804 consecutive patients undergoing first-time CBA for symptomatic AF at a single tertiary center. Cumulative sum (CUSUM) analysis identified learning curve inflection points. Patients were stratified into early ( n = 385) and plateau ( n = 419) phases. Multivariable linear regression, restricted cubic spline (RCS) analysis, and formal causal mediation analysis (bootstrap 1,000 iterations) were performed to evaluate worst pulmonary vein (PV) minimum temperature as a mediator between learning curve phase and log-transformed radiation dose. Results CUSUM analysis identified learning curve inflection points at case 385 (procedural time), 301 (radiation dose), and 329 (fluoroscopy time). Compared with the early phase, median radiation dose in the plateau phase decreased by 15.4% (506 vs. 428 mGy, p < 0.001), procedural time by 27.0% (89 vs. 65 min, p < 0.001), and fluoroscopy time by 26.3% (19 vs. 14 min, p < 0.001). In multivariable analysis (adjusted R 2 = 0.509), worst PV minimum temperature was an independent predictor of radiation dose (exp[β] = 1.012 per °C, 95% CI 1.008–1.017, p < 0.001), as was plateau phase (exp[β] = 0.895, 95% CI 0.847–0.946, p < 0.001). BMI exhibited a significant nonlinear association with radiation dose ( p for nonlinearity < 0.001), with an inflection threshold near 25 kg/m 2 . Mediation analysis demonstrated that worst PV minimum temperature mediated 14.5% (95% CI 5.5%–35.4%) of the total learning curve effect on radiation dose (average causal mediation effect [ACME] p < 0.001; performed on 734 patients with complete mediator data). Conclusions Greater operator experience was associated with lower radiation exposure and more favorable cryoablation temperature profiles. Worst PV nadir temperature statistically mediated a modest proportion of the association between learning curve phase and radiation dose, although most of the association remained unexplained by this temperature metric. These findings support a potential role of balloon–tissue contact quality in experience-related procedural improvement but should be interpreted as hypothesis-generating rather than evidence of a causal mechanism or improved clinical outcomes.