We conducted a retrospective cohort analysis of all RA procedures performed at our institution (Centro Hospitalar e Universitário de Coimbra, Portugal) between January 2009 and December 2020. Three hundred eighty-eight patients underwent RA-PCI for at least one lesion, in a total of 410 procedures. All were considered for this analysis, without inclusion/exclusion criteria. The study protocol was written according to the Declaration of Helsinki and approved by the institutional Ethics Committee (ID: CHUC-091-020; 27/08/2020). All patients provided written informed consent regarding the procedure, anonymized data collection, and analysis for research purposes.

Data were collected from our institutional PCI database, where information on all coronary interventions is prospectively recorded. Baseline characteristics were recorded as inserted by the interventional team and included patients’ demographics, relevant past medical history, cardiovascular risk factors, clinical presentation, and angiographic findings on the topography and severity of coronary artery disease (CAD). Technical details on arterial access, rotablation, balloon pre- and post-dilatation, stenting, use of other calcium modifying tools, intravascular imaging, temporary pacing, and hemodynamic support were documented. We performed an exploratory clinical follow-up analysis using our institution's electronic records and telephone interviews when patients were followed elsewhere.

Angiographic success was defined as the presence of less than 30% residual stenosis and grade 3 TIMI (thrombolysis in myocardial infarction) flow in all attempted lesions. Coronary perforation was defined as any dissection or intimal tear that extended through the full thickness of the arterial wall. Slow flow/no-reflow was defined as delayed or absent antegrade flow in the absence of an obstructive lesion. Burr entrapment was defined as any burr that got stuck within or beyond the lesion and could not be retrieved with standard operational methods.19 Two interventional cardiologists retrospectively assessed angiographic outcomes. Emergency coronary artery bypass grafting was defined as any emergency cardiothoracic surgical procedure to solve a PCI complication or after PCI failure. Peri-procedural myocardial infarction (MI) was recorded as registered by the medical team. All recorded events complied with the Fourth Universal Definition of type 4a MI.20 Follow-up data focused on major adverse cardiovascular events (MACE; defined as cardiovascular death, MI, and target lesion revascularization), all-cause mortality, target vessel revascularization and stent thrombosis, defined according to the Academic Research Consortium-2 criteria.21

The procedures were executed at a tertiary-care referral centre by 14 operators. All patients were pre-treated with aspirin and a loading dose of a P2Y12 receptor inhibitor. Intraprocedural anticoagulation was achieved with heparin (target activated clotting time of >250 or 200–250 secs when a glycoprotein IIb/IIIa inhibitor was administered). Access site and sheath size were determined by the anticipated maximum burr size and operator preference.

RA was performed with either the Rotablator™ or the Rotapro™ Rotational Atherectomy Systems (Boston Scientific Corporation, Natick, MA, USA). A continuous infusion of verapamil (5 mg), isosorbide dinitrate (10 mg), and unfractionated heparin (2500 IU) was administered through the device. A conventional 0.014-inch guidewire was used to facilitate navigation through the often-complex anatomy, and posteriorly exchanged for one of the dedicated RotaWire™ guidewires (Boston Scientific) via a microcatheter. As previously described,22 RA was carried out using a pecking motion of the burr, with short runs (15–20 secs) at a relatively low ablation speed (140000–150000 rpm). A reduction of more than 5000 rpm warranted immediate interruption of the ablation run to avoid burr lodging. Despite the preferential use of DES, some patients were exclusively treated with bare-metal stents or drug-coated balloons. Bioresorbable scaffolds were not implanted.

Categorical variables were presented as frequency (percentage), while continuous variables were described as mean±standard deviation or median (interquartile range), according to the normality of the distribution assessed by the Kolmogorov–Smirnov test. Angiographic success and peri-procedural complications were analysed on a per-procedure basis, while follow-up data was analysed on a per-patient basis. Time trends in baseline characteristics and procedural outcomes were examined using the Cochran–Armitage test or the Kendall rank correlation coefficient. One-year event rates were based on Kaplan–Meier estimates. Calendar quartiles were compared using the log-rank test. Multivariate analyses were performed using logistic and Cox regression for procedural and clinical endpoints, respectively. In those models, we included statistically significant variables on univariate analysis and/or known predictors based on previous data. Effect sizes were reported using odds ratio (OR) or hazard ratio (HR) at 95% confidence interval (CI). Statistical analysis was performed with IBM SPSS Statistics, version 26.0 (IBM Corp., Armonk, NY, USA) and R 4.0.3 statistical software (R Core Team, Vienna, Austria), with an alpha significance level set at 0.05.

Baseline clinical characteristics are depicted in Table 1. Mean age was 72.3±9.3 years; patients were more often male (74.0%), and had high proportions of hypertension (85.6%), diabetes (53.6%), chronic kidney disease (26.3%), prior MI (33.8%), and impaired ejection fraction (37.6%). Initial presentation was an acute coronary syndrome in 33.8% of patients. The prevalence of dyslipidaemia (ptrend <0.001), hypertension (ptrend=0.008), and prior PCI (ptrend=0.035) increased significantly over the study period.

Multivessel disease was predominant, with 149 patients (38.4%) having two-vessel and 127 three-vessel CAD (32.7%). Significant left main disease (i.e., ≥50% stenosis) was present in 50 patients (12.2%). CAD complexity increased over time, including median SYNTAX (Synergy between Percutaneous Coronary Intervention with TAXUS and Cardiac Surgery) score (ptrend=0.003) and proportion of American College of Cardiology/American Heart Association types B2 and C lesions (up to 91.8%, ptrend=0.003). Table 2 summarizes the target lesion characteristics.

Procedural characteristics and their temporal changes are presented in Table 3. RA was employed upfront in 324 cases (79.0%), staged in 22 cases (5.4%) and performed as an ad hoc bailout strategy in the remainder. Reasons for crossover to RA were the inability to cross the lesion with a balloon or stent (67.7%), inadequate balloon expansion (28.1%) and balloon rupture (3.1%).

Fluoroscopy time increased (ptrend=0.047) and the amount of contrast dye declined (ptrend=0.015) throughout the study period. Transradial approach and the use of <7Fr sheaths increased significantly over time (ptrend=0.003 and ptrend <0.001, respectively). Regarding rotablation, the use of burr diameters over 1.25 mm decreased over time (ptrend <0.001) in parallel with a slight reduction in burr/artery ratio, from 0.52±0.08 to 0.45±0.06 (ptrend <0.001). A single burr was sufficient in most cases (86.1%), with the employment of more than one burr being residual in the last triennium (3.2%).

Intravascular imaging was roughly used in one out of ten procedures; an increase in optical coherence tomography utilization was observed up to 4.8% in the last three years. In 63 cases (15.4%), other calcium modifying tools were employed along with RA, specifically cutting balloons (n=60), scoring balloons (n=2), and intravascular lithotripsy (n=1).

A total of 700 stents were implanted, 92.6% of which were DES (100% after 2014, ptrend <0.001). High-pressure post-dilatation was undertaken in 68.3%, with growing frequency over time, reaching 81% in the last three years (ptrend <0.001). A similar growing trend was observed in terms of total stent length, rising to 44.0±21.8 mm (ptrend=0.006). Glycoprotein IIb/IIIa inhibitors were seldom used overall and never administered from 2013 onwards (ptrend <0.001). Prophylactic temporary pacing and hemodynamic support were exceptional (three and four procedures, respectively).

Early and long-term outcomes are detailed in Table 4. The angiographic success rate was consistently high throughout the study period (96.6% overall; ptrend=0.938; Figure 2). Despite several runs with the smallest burr, six lesions remained uncrossable, deeming those procedures unsuccessful. Peri-procedural complications were recorded in 37 procedures (9.0%), with a downward trend over time (ptrend=0.029; Figure 2). The most common complication was coronary dissection, which generally required additional stenting (82% of dissections). Six deaths (1.5%) occurred prior to hospital discharge, one of which during the procedure (cardiac arrest following acute coronary syndrome presentation requiring PCI to the proximal left anterior descending artery). All cases of burr entrapment and perforation were successfully managed percutaneously. Univariate analysis identified acute coronary syndrome at presentation and bailout RA as predictors of peri-procedural complications, yet only bailout RA remained significant in the multivariate model (OR: 3.60; 95% CI: 1.55–8.35; p=0.003; Table S1).

Figure 2.

Trends in peri-procedural outcomes over the study period (2009–2020). Angiographic success rate (grey bars) remained unchanged over time (ptrend=0.938), while the incidence of peri-procedural complications (black line) declined (ptrend=0.029). RA-PCI: percutaneous coronary intervention with rotational atherectomy.

(0.11MB).

Follow-up data were available for 357 patients (92.0%). The incidence of MACE was 27.7% at a median follow-up period of 40 (16–76) months and was mainly driven by cardiovascular death (20.4%). The rate of all-cause death was 31.9%. At one year, MACE, all-cause mortality, and target vessel revascularization occurred in 12.1%, 6.4%, and 5.6% of patients, respectively, with no significant changes during the study period. Multivariable risk adjustment confirmed this finding (Table S2). Chronic kidney disease (HR: 1.98; 95% CI: 1.05–3.742; p=0.035) and multivessel disease (HR: 2.58; 95% CI: 1.09–6.126; p=0.032; Table S3) were significantly associated with the occurrence of MACE during follow-up.

Angiographic success rate was high (96.6% in 410 cases) in support of previous studies, despite variation in success definition.16–18,31 Even though RA has been recognized as an important risk factor for coronary perforation,35 only two cases were recorded in our cohort. Overall, we found a relatively low and declining rate of peri-procedural complications, which may reflect the learning curve,32 as well as the strict compliance with current best practices: use of small burrs (maximum burr size of <1.75 mm in 88.0% of our cases; burr-to-artery ratio of 0.49±0.07), rotational speed <180000 rpm, pecking burr motion, short ablation runs, and avoidance of abrupt decelerations.15

While crossing over to RA can enable procedural success, this approach may increase fluoroscopy time, contrast medium volume, and complication rates.28,36 Indeed, a bailout strategy was an independent predictor of peri-procedural complications in our cohort. This could reflect the risk of extending pre-existing dissections caused by repeated high-pressure pre-dilatation or specialty balloon use. Further studies are needed to determine if a staged RA approach would be superior in this setting to allow “healing” of such dissections,28 which is not an uncommon strategy in our practice.

Theoretically, RA can act both in favour and against DES. On the one hand, RA smoothens the lumen and helps to fracture calcium rings, facilitating lesion crossing with minimal polymer disruption, as well as optimal stent expansion and apposition. On the other hand, thermal injury and mechanical vessel trauma may trigger an exaggerated neointimal response and increase restenosis.37 In our cohort, follow-up MACE was 27.2% at 40 months and was mainly driven by cardiovascular death (20.4%), as reported in the Euro4C registry for one-year MACE.31 Chronic kidney disease and multivessel disease were significant predictors of MACE. Due to the exploratory and retrospective nature of long-term outcomes analysis, only 92.0% of the cohort (80.8% at one year) had complete follow-up data. As a referral centre, our institutional records are more likely to be updated in patients with recurrent events/interventions, therefore adverse clinical outcomes may be overestimated. In light of the high clinical complexity and this information bias, the MACE rate is acceptable.

This analysis has some limitations. First, data quality may be suboptimal due to missing data and heterogeneity among interventional team members introducing the data. Second, we were unable to account for all potential confounding factors, including those prompting planned RA employment, as we lack information on non-RA procedures. Third, due to the low number of outcomes, high uncertainty is brought into the multivariable logistic regression on peri-procedural complications, demonstrated by the broad CI of predictor variables. Fourth, the incidence of peri-procedural MI was lower than previously reported,30 due to the lack of data to retrospectively adjudicate this outcome. Lastly, the true effect of the bias created by the missing follow-up data is hard to estimate, but possibly would drive the MACE rate down.

The retrospective observational design of this study makes it prone to bias, thus demanding careful interpretation and limiting data extrapolation. However, the lack of exclusion criteria offered the opportunity to characterize an all-comers population undergoing RA, providing a comprehensive evaluation of the utilization and safety of this device in routine clinical practice.