Sex differences in procedural and clinical outcomes following rotational atherectomy

Abstract Aim Evaluate sex differences in procedural net adverse clinical events and long‐term outcomes following rotational atherectomy (RA). Methods and Results From August 2010 to 2016, 765 consecutive patients undergoing RA PCI were followed up for a median of 4.7 years. 285 (37%) of subjects were female. Women were older (mean 76 years vs. 72 years; p < .001) and had more urgent procedures (64.6 vs. 47.3%; p < .001). Females received fewer radial procedures (75.1 vs. 85.1%; p < .001) and less intravascular imaging guidance (16.8 vs. 25.0%; p = .008). After propensity score adjustment, the primary endpoint of net adverse cardiac events (net adverse clinical events: all‐cause death, myocardial infarction, stroke, target vessel revascularization plus any procedural complication) occurred more often in female patients (15.1 vs. 9.0%; adjusted OR 1.81 95% CI 1.04–3.13; p = .037). This was driven by an increased risk of procedural complications rather than procedural major adverse cardiac events (MACE). Specifically, women were more likely to experience coronary dissection (4.6 vs. 1.3%; p = .008), cardiac tamponade (2.1 vs. 0.4%; p = .046) and significant bleeding (BARC ≥2: 5.3 vs. 2.3). Despite this, overall MACE‐free survival was similar between males and females (adjusted HR 1.03; 95% CI 0.80–1.34; p = .81). Procedural complications during RA were associated with almost double the incidence of MACE at long‐term follow‐up (HR 1.92; 95% CI 1.34–2.77; p < .001). Conclusion Women may be at greater risk of procedural complications following rotational atherectomy. These include periprocedural bleeding episodes and coronary perforation leading to cardiac tamponade. Despite this, the adjusted overall long‐term survival free of major adverse cardiac events was similar between males and females.

Conclusion: Women may be at greater risk of procedural complications following rotational atherectomy. These include periprocedural bleeding episodes and coronary perforation leading to cardiac tamponade. Despite this, the adjusted overall long-term survival free of major adverse cardiac events was similar between males and females.

K E Y W O R D S
calcific coronary disease, complex percutaneous coronary intervention, gender differences, rotational atherectomy

| INTRODUCTION
Cardiovascular disease is the leading cause of death in women. 1 Following percutaneous coronary intervention (PCI), women suffer disproportionately high rates of death, 2 bleeding 3 and complications including myocardial infarction (MI) and stroke. 4 Furthermore, female gender is an independent predictor of death, MI, stent thrombosis and target lesion revascularization after PCI of calcific coronary lesions. 5 Rotational atherectomy (RA) may be necessary in calcific coronary lesions to prevent stent underexpansion or malapposition that are associated with higher rates of restenosis or target vessel failure. 6 Women requiring treatment with RA present a unique challenge, often presenting later, with more advanced coronary artery disease, and smaller caliber peripheral and coronary arteries. Women may be particularly susceptible to complications following rotational atherectomy (RA) due to gender-specific differences in arterial access, bleeding and coronary pathophysiology. This has not been studied in a contemporary cohort, we therefore designed a study to analyze gender differences in procedural net adverse clinical events (NACE) and assess effects on long-term outcomes following RA.

| Study population
Over 6 years from August 2010, 16,198 PCIs were analyzed at our high volume regional center (serving a population of~2 million). Seven hundred sixty-five consecutive RA procedures were included representing 4.7% of all PCIs performed. Data was obtained from a PCI database with paired analysis of electronic health records, a nationwide electronic portal and sourced individual patient case notes. The institutional review board approved the study and use of patient data.

| Variables
Baseline characteristics (demographic, clinical, and procedural) were extracted from our PCI database as entered by the operator. Long-term follow-up data (minimum 12 months) was obtained through an electronic health database including nationwide electronic portal with individual source patient files where needed. Diabetes Mellitus included patients with diet controlled diabetes and those on treatment according to WHO criteria. Left ventricular ejection fraction (LVEF) was stratified into three categories-good (LVEF > 50%), moderately impaired (30-50%), severely impaired (LVEF < 30%). Arterial access was recorded as radial where the rotablation procedure was performed exclusively via the radial artery. This included patients who had a concomitant intra-aortic balloon pump (IABP) inserted via the femoral artery. Significant renal impairment was defined as active renal replacement therapy or a serum creatinine >200 μmoL/L (2.26 mg/dL). Patients with missing data were prospectively recorded in a Appendix S1. Data were assumed to be missing at random allowing them to be imputed with the use of multivariate imputation by chained equations.

| Rotational atherectomy procedure
RA was performed using the Boston Scientific Rotablator ® system. Burr speeds were routinely between 150,000 and 170,000 rpm with smooth back and forth pecking motion performed in runs of up to 20 s duration. Occasionally, further acceleration was used to cross resistant lesions with a maximum speed of 200,000 rpm. Access site and sheath size were determined by individual operator preference.
The maximum burr size used and the maximum external sheath size (which corresponds to arteriotomy size, a more important predictor of vascular access complications than guiding catheter diameter) were recorded. In the case of sheathless guiding catheters, the equivalent maximum external diameter was recorded (e.g., Asahi Sheathless™ 7.5F guiding catheter has outer diameter of 2.5 mm and is thus considered <7F arteriotomy). Anticoagulation with intravenous unfractionated heparin (UFH) was used in all cases maintaining an activated clotting time >250 s. Additional glycoprotein IIb/IIIa antagonists were used in selected cases at operator discretion.

| Outcomes
The primary endpoint of this study was the composite incidence of net adverse clinical events (NACE occurring within 24 hr of PCI (BARC 2 or greater), arrhythmia (ventricular fibrillation, ventricular tachycardia, pulseless electrical activity (PEA) cardiac arrest or bradycardia requiring temporary pacing line insertion). The occurrence of procedural complications were identified and verified by an independent cardiologist who was not involved in the procedure against the objective criteria above. The outcomes were evaluated at 30-day, 1-year and longer-term follow-up until August 2018.

| Statistical analysis
Results are reported as mean (±SD) for parametric data and median (25th, 75th percentile) for data that was not normally distributed. The χ 2 test (or Fisher's exact test for infrequent events) was used to assess for differences between categorical variables. Univariate odds ratios with 95% confidence intervals were calculated to measure strength of association between categorical variables. One-way ANOVA was used to assess for differences between means of continuous normally distributed variables. All multiple comparisons were adjusted using the Benjamin-Hochberg correction. All analyses were two-tailed analysis with significance considered as p value <.05 to be significant.
Propensity scores were created for female and male groups and incorporated to adjust for baseline differences in a multivariable logistic regression model for procedural NACE. The following variables were used to calculate the propensity score: Age, procedure urgency/ACS, previous stroke or myocardial infarction, renal impairment, diabetes mellitus, hypertension, left ventricular function, previous coronary bypass grafting, use of intravascular ultrasound for PCI, maximum burr size ≥1.75 mm, maximum arteriotomy (sheath ≥7F), left main lesion location, number of vessels undergoing HSRA. Covariate balance between groups was evaluated by the Wald chi-square statistic before and after propensity score adjustment. After adjusting for propensity score, none of the variables used to create the propensity score were found to be significantly different between the male and female groups ( Figure 1).
Logistic regression was used to measure the adjusted odds ratio (OR) and 95% confidence intervals (CI). Multivariate regression models were used to determine predictors of procedural NACE adjusting for

| Baseline characteristics
Seven hundred sixty-five consecutive patients undergoing RA were studied and followed up over a median duration of 4.7 years. There were five high volume operators (>75 RA procedures: MME, KGO, SW, MML, PR) who performed 624 (82%) of all RA procedures, and a total of ten main consultant operators. The mean age of patients was 73 ± 9 years (37% female). There were no patients unaccounted for at long-term follow-up.
The temporal distribution of HSRA cases over the course of the study is illustrated in Figure S1. Baseline demographics and procedural details are shown in Table 1. Women undergoing RA were older (mean 76 ± 8 years vs. 72 ± 9 years in men; p < .001) with more urgent procedures (64.6 vs. 47.3%; p < .001) and a higher prevalence of hypertension (80 vs. 67%; p = .002). Women had lower rates of previous CABG, but otherwise similar baseline demographics factors including anatomical lesion location and renal function (Table 1). After propensity score adjustment, there were no differences in the baseline characteristics of patients studied ( Figure 1).

| Procedural characteristics
Radial access was used less frequently in women (75.1 vs. 85.1%, p < .001, Figure 2A), with the femoral approach therefore significantly more common (OR 1.91, 95% CI 1.33 to 2.76, p < .001; Figure 2A). Intravascular imaging (IVUS) guidance (16.8 vs. 25%, p = .008) and large burr sizes (1.75 mm or greater) (30 vs. 40%; p = .018) were used less frequently in women compared with men. Anticoagulation with UFH was F I G U R E 1 Propensity adjustment for baseline differences between female and male groups [Color figure can be viewed at wileyonlinelibrary.com] used uniformly in all patients. Glycoprotein IIB/IIIA inhibitor use was similar between the groups (11 vs. 12%; p = .830). The use of temporary venous pacing upfront was exceptionally low (n = 4; 0.5%) and bailout use of temporary pacing line was needed in only 6 (0.8%) subjects.

| Procedural outcomes
After propensity score adjustment, the primary study endpoint (procedural NACE) was over 80% more likely in females than males (15.1 vs. 9.0%; adjusted OR 1.81, 1.04 to 3.13, p = .037; Table 2, Figure 2B). To assess the goodness of fit and diagnostic accuracy of the model, we assessed area under the ROC (AUC: Harrell's c-statistic).
The multivariate fitted regression model showed moderate discrimination potential with an AUC of 0.68 ± 0.03; p < .001 ( Figure S2).

| Predictors of procedural net adverse cardiac events
The other independent predictors of adverse procedural events during RA included intervention involving the left main coronary artery

| Clinical events and long-term survival
At thirty days, MACE rates were similar in both groups with 18 (6.3%) females and 36 (7.5%) males. Long-term major adverse cardiac events at median 4.7 years follow-up were observed more frequently in females (47.0 vs. 38.5%; HR 1.25; 1.00-1.56; Figure 4). However after adjusting for propensity score and baseline factors, the long-term clinical outcomes were similar between groups (HR 1.03; 0.80-1.34; p = .813; Table 4). There were no differences in the adjusted overall mortality of patients according to gender at thirty day or longer-term follow-up. The mortality at 30-days (3.9% females vs. 5.2% males) and

| DISCUSSION
In this contemporary cohort of patients undergoing predominantly transradial RA, we showed that female sex is independently associated with the occurrence of procedural net adverse cardiac events (NACE). This was predominantly driven by an excess of periprocedural bleeding and coronary events, including dissection and perforation leading to tamponade.
Importantly, this effect was seen after propsensity score adjustment and was not explained by differences in periprocedural anticoagulation strategies or baseline comorbidities. Despite the increase in complications, female gender was not independently associated with short term or longterm mortality after RA. However this study did show that procedural NACE were prognostically important and independently associated with reduced long-term survival after RA PCI.

| Rotational atherectomy procedural complications: Females at high risk
Our study provides real-world evidence of the risk/treatment paradox in women undergoing complex PCI with RA. Women are at greatest risk of complications but are under-served in the best practices known to reduce risk (e.g., less radial approach, less intravascular imaging guidance). A recent MATRIX sub-study confirmed that women have a higher risk of severe bleeding and access site complications in the setting of ACS. Radial access was an effective method to reduce these complications, as well as impacting on the composite ischemic, and ischemic or bleeding endpoints. 11 A large contemporary UK cohort study confirmed that radial access was used less often for RA in women versus men (31.6 vs. 37.1%; p < .001) with both of these rates being approximately half the rate of radial access in our centre. 12 Radial access for complex procedures including RA is increasing which may help reduce complications in this high risk cohort. 13 While larger studies provide power to assess hard binary endpoints such as mortality, it is plausible that The overall coronary artery perforation rate in our complex patient cohort was 2.9%. In addition to device perforations, this included wire perforations or minor contained perforations thought to relate more to the coronary calcification and post dilation than the atherectomy itself. 14 Perforations in other cohorts have been reported as "device perforations" rather than "any perforation" and are reported in up to 2.5% of RA 15-17 and 1.9% of orbital atherectomy procedures. 18 Whether female gender is independently predictive of PCI complications is controversial and varies according to the population studied. 19 Large studies show no overall difference in mortality after adjustment for age and other baseline differences between men and women undergoing PCI. 20

| Limitations
This study represents the largest contemporary cohort of majority

| CONCLUSIONS
Women may be at greater risk of net adverse cardiac events (NACE) after rotational atherectomy. These risks include periprocedural bleeding episodes and coronary perforation leading to cardiac tamponade.
The adjusted overall long-term survival is not different between males and females. Efforts to close the gender gap in NACE should be focused on strategies to reduce bleeding, enhancing best practice transradial PCI incorporating accurate vessel sizing to avoid coronary perforations.

ACKNOWLEDGMENTS
We thank our patients, administrative staff, nurses and doctors notably who supported this study and the British Heart Foundation for financial support. T.J.F. is supported by the British Heart Foundation RE/13/5/30177, PG-17-25-32884. No companies were involved in this study. The authors are solely responsible for the design and conduct of the study, all analyses, the drafting and editing of the paper and its final contents.

ETHICS AND DECLARATION OF HELSINKI
This study complies with the Declaration of Helsinki and has full local approval for using subject data.

IMPACT ON DAILY PRACTICE
Female gender is an independent predictor of net adverse clinical events following the rotational atherectomy. Efforts to close the gender gap should focus on strategies to reduce bleeding, performing transradial PCI with consideration of adjunctive intracoronary imaging to accurately size vessels.