Introduction: Several papers have focused on the learning curve in endovascular repair of abdominal (EVAR) and thoracoabdominal aortic aneurysms (F/B-EVAR), providing valuable information. However, studies on the thoracic endovascular aortic repair (TEVAR) learning curve are still lacking. We aimed to analyze the learning curve for TEVAR procedures performed in a single center over 25 years.

Methods: This was a retrospective, observational, single-center study involving TEVAR procedures undertaken between May 1996 and September 2021. Data about procedural variables, 30-day clinical success, major adverse events (MAEs), and reintervention rates were subdivided into four quartiles of experience (Q1-Q4). Cumulative sum (CUSUM) charts were elaborated for the entire center experience as well as for the first and second implanting physicians. The primary outcome of this study was to evaluate the learning curve of TEVAR procedures performed in our center using the CUSUM method. This technique was used to generate a graphical depiction of the relationship between case number (on the x-axis) and 30-day MAE rates (on the y-axis), highlighting the number of procedures needed for achieving an acceptable MAE rate. The learning curve target achievement was defined as a CUSUM graph dropping below the alert MAE line. Scatterplots coupled with lowness smoothing were employed to provide a graphical representation of the trend between procedural variables and cumulative experience.

Results: 390 consecutive TEVAR procedures were analyzed. Within the study cohort 204 patients were treated for descending thoracic aortic aneurysms, 94 for type B aortic dissections, 29 for blunt traumatic aortic injuries and 63 for endoleaks correction. The mean follow-up was 4.3 ± 4.0 years. During the study period, a rise of urgent/emergent cases involving patients with more comorbidities was observed (6.2% in Q1 vs 42.3% in Q4; 20.6% in Q1 vs 23.7% in Q4, respectively; P< 0.000). However, the 30-day clinical success increased from Q1-2 to Q3-4 (85.9 vs 90.3%, P = 0.190) and thirty-day mortality decreased from 11.4% in Q1-2 to 3.6% in Q3-4 (P = 0.003). A significant reduction in recourse to additional maneuvers (21.5% in Q1-2 vs 13.3% in Q3-4, P = 0.033), operative time (Q1 = 139.8 ± 65.5 min vs Q4 = 76.7 ± 43.7 min, P = 0.001), fluoroscopy time (Q1 = 15.1 ± 8.8 min vs Q4 = 7.1 ± 5.1 min, P< 0.000) and contrast volume (Q1 = 244.0 ± 112.1 ml vs Q4 = 104.3 ± 46.1 ml, P< 0.000) was also noted. Additional access vessels-related maneuvers (P = 0.015), operative (P = 0.001) and fluoroscopy time (P = 0.054) and contrast volume (P = 0.042) were identified as independent risk factors for MAEs. About 75 TEVAR procedures had to be performed for the center's curve to assume a significant trend below the desired MAEs threshold. In the current study MAE incidence was 13.6%. Late endoleak incidence decreased from 26.8% in Q1 to 11.5% in Q4 (P = 0.061). Operative time ( P= 0.021), fluoroscopy time (P = 0.004) and contrast volume (P = 0.016) were individuated as independent risk factors for late endoleak occurrence through uni- and multivariate Cox regression. In particular, a 1.3-fold increase in endoleak risk was observed for both each 60 minutes of additional operative time (HR 1.3; 95%CI 1.0-1.7; P = 0.021) and for each 100 mL of additional contrast medium (HR 1.3; 95%CI 1.0-1.6; P = 0.016), along with 1.4-fold risk increment with every 10 minutes increase in fluoroscopy time (HR 1.4; 95%CI 1.1-1.7; P = 0.004).

Conclusion: Our TEVAR experience demonstrates improvements in terms of MAEs, clinical success, procedural outcomes, endoleak occurrence, and aortic-related survival. The learning curve of the first implanting physician was achieved roughly at the 75th treated patient. However, the learning curve of the surgeons coming thereafter proved to be significantly shorter. Such results suggest how specific endovascular training in this procedure is mandatory to become an effective TEVAR performer.