Coronary bypass versus percutaneous intervention: sex matters. The impact of gender on long-term outcomes of coronary revascularization†

Abstract

INTRODUCTION

Coronary artery bypass surgery (CABG) has historically been reported to have higher operative mortality for women than for men [1]. Whether this discrepancy is due to differential risk factors [2, 3] or unidentified sex-specific considerations has not been determined [4, 5]. Data regarding the impact of sex on the results of percutaneous coronary intervention (PCI) has been more heterogeneous, with contemporary risk adjusted studies demonstrating comparable outcomes in men and women [6–8]. What has been much less well defined is the potential differential impact of sex on the comparative outcomes of CABG versus PCI as an optimal therapy for patients with multi-vessel coronary artery disease [9].

Although prospective randomized controlled trials (RCT) represent the ‘gold standard’ of clinical evidence, the necessarily highly selective nature of study entry criteria may limit their external validity. It is essential to corroborate these findings with registry data from the ‘real-world’ clinical experience in order to fully elucidate the applicability of their findings. We therefore investigated the impact of sex on the relative outcomes of CABG versus PCI in a prospective community-based multi-centre registry.

METHODS

As previously published [10, 11], this was a multi-centre registry recording data on consecutive patients undergoing coronary revascularization (CR) procedures from eight centres. There were no pre-specified selection criteria and patients were treated according to physician preference. Data were collected locally and returned to a central repository where they were entered into a proprietary database, as previously described. Data definitions followed the Society of Thoracic Surgeons adult cardiac surgery database version 2.52 active at the time of study entry. The protocol was approved by the Institutional Review Board for North Texas at Medical City Dallas Hospital with a waiver of consent.

Data were collected over a 6-month period from 1 February 2004, until 31 July 2004. All collected data were subjected to quality checks and data validation tests.

Follow-up was completed in 2011, when the database was locked. Follow-up included Social Security Death Index (SSDI) validation of all mortalities, hospital and physician inquiries and letters and telephone calls to the patients. Patients with a diagnosis of ST elevation myocardial infarction (STEMI) or status listed as salvage were deleted from the data set for analysis [11]. Major adverse cardiac events (MACE) collected were all-cause mortality, non-fatal myocardial infarction (MI) and any repeat revascularization procedure.

Data were summarized and compared using χ2 statistics for discrete variables (Fisher’s exact test when cell counts were < 5) and t tests or for continuous data. The propensity matching used the preoperative risk parameters outlined in Table 1 and accounted for nesting of patients in hospitals. The scores were put through a 1-to-1 greedy matching algorithm that produced two equivalent data sets for PCI and CABG patients with 5+ years of follow-up. Survival curves were derived using Kaplan–Meier methods and compared using the log rank test. Hazard ratios were calculated using the Cox proportional hazard model. All statistical calculations used SAS 9.4 (SAS Institute, Cary, NC).

RESULTS

Patient population

There were a total of 3153 patients entered into the registry of whom 2162 (68.6%) were men and 991(31.4%) were women. In the CABG group (967), there were 673 (69.6%) men and 294 (30.4%) women, while in the PCI group (2186) there were 1489 men (68.1%) and 697 (31.9%) women. Patient characteristics are well outlined in Table 1. Of note, in both the CABG and PCI populations, women were older than men, had a higher incidence of hypertension, diabetes and cerebrovascular disease, while men were more likely to be cigarette smokers and tended to have more prior cardiac interventions and previous myocardial infarctions. Among the CABG patients, 475/673 (70.6%) of the men and 214/294 (72.8%) of the women received at least one internal mammary artery (IMA) graft. Bilateral IMA (BIMA) grafting was used in 62/673 (9.2%) of the men and only 9/294 (3.1%) of the women. Overall follow-up was 63.5 ± 27.9 months (median, 79.7 months).

Percutaneous coronary interventions versus coronary artery bypass grafting

Comparing PCI with CABG for the male patients, we discovered that the 5-year actuarial MACE-free survival (±standard error [SE]) 77.1 ± 1.1 vs 83.1 ± 1.5. with a hazard ratio (HR) of 0.588 (95% confidence interval [CI]: 0.491–0.704; P < 0.001) . For female patients, results were quite different. Five year MACE-free survival, PCI versus CABG, was 75.0 ± 1.6 vs 74.5 ± 2.5, with a HRof 0.869 (95% CI: 0.687–1.100; P = 0.24). MACE free survival for all four groups are depicted in Fig. 1.

Figure 1:

Freedom from MACE in male and female non-STEMI patients undergoing coronary revascularization with CABG versus PCI. MACE: major adverse cardiac events; STEMI: ST elevation myocardial infarction; CABG: coronary artery bypass surgery; PCI: percutaneous coronary interventions.

Open in new tabDownload slide

Because of treatment assignment was by physician preference rather than random selection, CABG and PCI groups were markedly different in many respects as has been previously reported for this registry [11]. We therefore used propensity matching arrive at two well-matched patient groups consisting of 434 male and 185 female patients in each CABG and PCI group (Tables 2 and 3). Overall results for PCI versus CABG, were similar to those for the unmatched patients. For men, the 5-year actuarial MACE-free survival was 69.5 ± 2.2 vs 79.5 ± 2.0, HR 0.548 (95% CI: 0.424–0.682; P < 0.001). For the matched PCI/CABG female groups we found no difference in the 5-year MACE-free survival:68.1 ± 3.4 vs 69.4 ± 3.4, HR 0.752 (95% CI: 0.540–1.049; P= 0.093) (See Fig. 2 for PCI versus CABG MACE-free survival in matched male [panel A] and female [panel B] patients).

Figure 2:

(A) Freedom from MACE in male patients well-matched for preprocedural risk factors between CABG versus PCI. (B) Freedom from MACE in female patients well-matched for preprocedural risk factors between CABG versus PCI. MACE: major adverse cardiac events; CABG: coronary artery bypass surgery; PCI: percutaneous coronary interventions.

Open in new tabDownload slide

Table 2:

Open in new tab

Baseline characteristics of matched male CABG and PCI patient groups

Variable .	CABG .	PCI .	P-value .	Std-Diff .
Number of patients .	434 .	434 .	.	.
Males	100%  (434/434)	100%  (434/434)	—	—
Pre-op Stroke	5.1%  (22/434)	4.8%  (21/434)	0.88	0.022
Previous CV intervention	29.0%  (126/434)	28.8%  (125/434)	0.94	0.051
Previous CABG surgery	7.8%  (34/434)	7.6%  (33/434)	0.90	−0.036
Previous valve surgery	0.2%  (1/434)	0	0.32	0.063
Previous PCI procedure	23.3%  (101/434)	22.4%  (97/434)	0.75	0.059
Hypertensive	76.5%  (332/434)	74.4%  (323/434)	0.48	−0.054
Angina	84.8%  (368/434)	82.8%  (356/430)	0.43	−0.087
Heart failure	6.0%  (26/434)	5.3%  (23/434)	0.66	−0.010
Renal failure	5.1%  (22/434)	4.8%  (21/434)	0.88	0.010
On dialysis	1.4%  (6/434)	1.4%  (6/431)	0.99	0.057
Pre-op MI	29.1%  (126/433)	25.6%  (111/434)	0.24	−0.037
Arrhythmia	5.1%  (22/434)	3.7%  (16/434)	0.32	−0.045
Diabetics on insulin	9.4%  (41/434)	8.8%  (38/434)	0.72	−0.025
Triple vessel disease	34.3%  (149/434)	28.6%  (124/434)	0.068	−0.119
Preop inotrope use	0.7%  (3/434)	0.5%  (2/434)	0.65	0.031
Elective status	59.0%  (256/434)	61.5%  (267/434)	0.73	0.103
Family history	45.9%  (199/434)	47.0%  (204/434)	0.73	0.023
Cerebrovascular disease	10.1%  (44/434)	9.0%  (39/434)	0.56	−0.032
Peripheral arterial disease	12.2%  (53/434)	10.8%  (47/434)	0.52	−0.080
Current/recent smoker	53.7%  (233/434)	51.2%  (222/434)	0.46	−0.120
Dyslipidaemia	56.2%  (244/434)	59.0%  (256/434)	0.41	0.009
Preop beta blockade	60.1%  (261/434)	58.8%  (255/434)	0.68	0.023
Age  ± SD	63.85  ± 10.27	64.75  ± 11.89	0.23	0.073
Ejection fraction  ± SD	51.03  ± 10.90	51.94  ± 13.79	0.28	0.014

Variable .	CABG .	PCI .	P-value .	Std-Diff .
Number of patients .	434 .	434 .	.	.
Males	100%  (434/434)	100%  (434/434)	—	—
Pre-op Stroke	5.1%  (22/434)	4.8%  (21/434)	0.88	0.022
Previous CV intervention	29.0%  (126/434)	28.8%  (125/434)	0.94	0.051
Previous CABG surgery	7.8%  (34/434)	7.6%  (33/434)	0.90	−0.036
Previous valve surgery	0.2%  (1/434)	0	0.32	0.063
Previous PCI procedure	23.3%  (101/434)	22.4%  (97/434)	0.75	0.059
Hypertensive	76.5%  (332/434)	74.4%  (323/434)	0.48	−0.054
Angina	84.8%  (368/434)	82.8%  (356/430)	0.43	−0.087
Heart failure	6.0%  (26/434)	5.3%  (23/434)	0.66	−0.010
Renal failure	5.1%  (22/434)	4.8%  (21/434)	0.88	0.010
On dialysis	1.4%  (6/434)	1.4%  (6/431)	0.99	0.057
Pre-op MI	29.1%  (126/433)	25.6%  (111/434)	0.24	−0.037
Arrhythmia	5.1%  (22/434)	3.7%  (16/434)	0.32	−0.045
Diabetics on insulin	9.4%  (41/434)	8.8%  (38/434)	0.72	−0.025
Triple vessel disease	34.3%  (149/434)	28.6%  (124/434)	0.068	−0.119
Preop inotrope use	0.7%  (3/434)	0.5%  (2/434)	0.65	0.031
Elective status	59.0%  (256/434)	61.5%  (267/434)	0.73	0.103
Family history	45.9%  (199/434)	47.0%  (204/434)	0.73	0.023
Cerebrovascular disease	10.1%  (44/434)	9.0%  (39/434)	0.56	−0.032
Peripheral arterial disease	12.2%  (53/434)	10.8%  (47/434)	0.52	−0.080
Current/recent smoker	53.7%  (233/434)	51.2%  (222/434)	0.46	−0.120
Dyslipidaemia	56.2%  (244/434)	59.0%  (256/434)	0.41	0.009
Preop beta blockade	60.1%  (261/434)	58.8%  (255/434)	0.68	0.023
Age  ± SD	63.85  ± 10.27	64.75  ± 11.89	0.23	0.073
Ejection fraction  ± SD	51.03  ± 10.90	51.94  ± 13.79	0.28	0.014

CABG: coronary artery bypass grafting; PCI: percutaneous coronary intervention; MI: myocardial infarction; SD: standard deviation. Std-Diff: standardized mean difference.

Table 2:

Open in new tab

Baseline characteristics of matched male CABG and PCI patient groups

Variable .	CABG .	PCI .	P-value .	Std-Diff .
Number of patients .	434 .	434 .	.	.
Males	100%  (434/434)	100%  (434/434)	—	—
Pre-op Stroke	5.1%  (22/434)	4.8%  (21/434)	0.88	0.022
Previous CV intervention	29.0%  (126/434)	28.8%  (125/434)	0.94	0.051
Previous CABG surgery	7.8%  (34/434)	7.6%  (33/434)	0.90	−0.036
Previous valve surgery	0.2%  (1/434)	0	0.32	0.063
Previous PCI procedure	23.3%  (101/434)	22.4%  (97/434)	0.75	0.059
Hypertensive	76.5%  (332/434)	74.4%  (323/434)	0.48	−0.054
Angina	84.8%  (368/434)	82.8%  (356/430)	0.43	−0.087
Heart failure	6.0%  (26/434)	5.3%  (23/434)	0.66	−0.010
Renal failure	5.1%  (22/434)	4.8%  (21/434)	0.88	0.010
On dialysis	1.4%  (6/434)	1.4%  (6/431)	0.99	0.057
Pre-op MI	29.1%  (126/433)	25.6%  (111/434)	0.24	−0.037
Arrhythmia	5.1%  (22/434)	3.7%  (16/434)	0.32	−0.045
Diabetics on insulin	9.4%  (41/434)	8.8%  (38/434)	0.72	−0.025
Triple vessel disease	34.3%  (149/434)	28.6%  (124/434)	0.068	−0.119
Preop inotrope use	0.7%  (3/434)	0.5%  (2/434)	0.65	0.031
Elective status	59.0%  (256/434)	61.5%  (267/434)	0.73	0.103
Family history	45.9%  (199/434)	47.0%  (204/434)	0.73	0.023
Cerebrovascular disease	10.1%  (44/434)	9.0%  (39/434)	0.56	−0.032
Peripheral arterial disease	12.2%  (53/434)	10.8%  (47/434)	0.52	−0.080
Current/recent smoker	53.7%  (233/434)	51.2%  (222/434)	0.46	−0.120
Dyslipidaemia	56.2%  (244/434)	59.0%  (256/434)	0.41	0.009
Preop beta blockade	60.1%  (261/434)	58.8%  (255/434)	0.68	0.023
Age  ± SD	63.85  ± 10.27	64.75  ± 11.89	0.23	0.073
Ejection fraction  ± SD	51.03  ± 10.90	51.94  ± 13.79	0.28	0.014

Variable .	CABG .	PCI .	P-value .	Std-Diff .
Number of patients .	434 .	434 .	.	.
Males	100%  (434/434)	100%  (434/434)	—	—
Pre-op Stroke	5.1%  (22/434)	4.8%  (21/434)	0.88	0.022
Previous CV intervention	29.0%  (126/434)	28.8%  (125/434)	0.94	0.051
Previous CABG surgery	7.8%  (34/434)	7.6%  (33/434)	0.90	−0.036
Previous valve surgery	0.2%  (1/434)	0	0.32	0.063
Previous PCI procedure	23.3%  (101/434)	22.4%  (97/434)	0.75	0.059
Hypertensive	76.5%  (332/434)	74.4%  (323/434)	0.48	−0.054
Angina	84.8%  (368/434)	82.8%  (356/430)	0.43	−0.087
Heart failure	6.0%  (26/434)	5.3%  (23/434)	0.66	−0.010
Renal failure	5.1%  (22/434)	4.8%  (21/434)	0.88	0.010
On dialysis	1.4%  (6/434)	1.4%  (6/431)	0.99	0.057
Pre-op MI	29.1%  (126/433)	25.6%  (111/434)	0.24	−0.037
Arrhythmia	5.1%  (22/434)	3.7%  (16/434)	0.32	−0.045
Diabetics on insulin	9.4%  (41/434)	8.8%  (38/434)	0.72	−0.025
Triple vessel disease	34.3%  (149/434)	28.6%  (124/434)	0.068	−0.119
Preop inotrope use	0.7%  (3/434)	0.5%  (2/434)	0.65	0.031
Elective status	59.0%  (256/434)	61.5%  (267/434)	0.73	0.103
Family history	45.9%  (199/434)	47.0%  (204/434)	0.73	0.023
Cerebrovascular disease	10.1%  (44/434)	9.0%  (39/434)	0.56	−0.032
Peripheral arterial disease	12.2%  (53/434)	10.8%  (47/434)	0.52	−0.080
Current/recent smoker	53.7%  (233/434)	51.2%  (222/434)	0.46	−0.120
Dyslipidaemia	56.2%  (244/434)	59.0%  (256/434)	0.41	0.009
Preop beta blockade	60.1%  (261/434)	58.8%  (255/434)	0.68	0.023
Age  ± SD	63.85  ± 10.27	64.75  ± 11.89	0.23	0.073
Ejection fraction  ± SD	51.03  ± 10.90	51.94  ± 13.79	0.28	0.014

CABG: coronary artery bypass grafting; PCI: percutaneous coronary intervention; MI: myocardial infarction; SD: standard deviation. Std-Diff: standardized mean difference.

Table 3:

Open in new tab

Baseline characteristics of matched female CABG and PCI patient groups

Variable .	CABG .	PCI .	.	.
Number of patients .	185 .	185 .	P-Value .	Std-Diff .
Females	100%  (185/185)	100%  (185/185)		—
Pre-op stroke	9.2%  (17/185)	7.6%  (14/185)	0.57	−0.080
Previous CV intervention	18.9%  (35/185)	21.6%  (40/185)	0.52	0.083
Previous CABG surgery	3.8%  (7/185)	4.3%  (8/185)	0.79	0.025
Previous valve surgery	0	0	—	—
Previous PCI procedure	16.2%  (30/185)	17.3%  (32/185)	0.78	0.060
Hypertensive	85.4%  (158/185)	83.2%  (154/185)	0.57	−0.106
Angina	84.8%  (156/184)	88.1%  (163/185)	0.35	0.127
Heart failure	11.4%  (21/184)	10.8%  (20/185)	0.85	−0.018
Renal failure	4.3%  (8/185)	5.4%  (10/185)	0.63	0.050
On dialysis	2.2%  (4/185)	1.6%  (3/184)	0.71	−0.086
Pre-op MI	23.8%  (43/181)	28.6%  (53/185)	0.29	0.166
Arrhythmia	3.8%  (7/185)	4.9%  (9/185)	0.61	0.028
Diabetics on insulin	15.7%  (29/185)	13.0%  (24/185)	0.46	−0.048
Triple vessel disease	30.3%  (56/185)	27.6%  (51/185)	0.57	−0.120
Pre-op inotrope use	0.5%  (1/185)	1.6%  (3/185)	0.32	0.061
Elective status	50.8%  (94/185)	58.4%  (108/185)	0.29	0.132
Family history	50.3%  (93/185)	51.4%  (95/185)	0.84	−0.054
Cerebrovascular disease	15.1%  (28/185)	13.0%  (24/185)	0.55	−0.032
Peripheral arterial disease	13.0%  (24/185)	14.1%  (26/185)	0.76	−0.085
Current/recent smoker	42.2%  (78/185)	43.2%  (80/185)	0.83	−0.076
Dyslipidaemia	60.0%  (111/185)	62.7%  (116/185)	0.59	0.225
Pre-op beta blockade	60.0%  (111/185)	63.8%  (118/185)	0.45	0.056
Age  ± SD	66.72  ± 10.93	68.23  ± 12.35	0.21	0.079
Ejection fraction  ± SD	52.56  ± 11.28	52.99  ± 13.48	0.74	0.039

Variable .	CABG .	PCI .	.	.
Number of patients .	185 .	185 .	P-Value .	Std-Diff .
Females	100%  (185/185)	100%  (185/185)		—
Pre-op stroke	9.2%  (17/185)	7.6%  (14/185)	0.57	−0.080
Previous CV intervention	18.9%  (35/185)	21.6%  (40/185)	0.52	0.083
Previous CABG surgery	3.8%  (7/185)	4.3%  (8/185)	0.79	0.025
Previous valve surgery	0	0	—	—
Previous PCI procedure	16.2%  (30/185)	17.3%  (32/185)	0.78	0.060
Hypertensive	85.4%  (158/185)	83.2%  (154/185)	0.57	−0.106
Angina	84.8%  (156/184)	88.1%  (163/185)	0.35	0.127
Heart failure	11.4%  (21/184)	10.8%  (20/185)	0.85	−0.018
Renal failure	4.3%  (8/185)	5.4%  (10/185)	0.63	0.050
On dialysis	2.2%  (4/185)	1.6%  (3/184)	0.71	−0.086
Pre-op MI	23.8%  (43/181)	28.6%  (53/185)	0.29	0.166
Arrhythmia	3.8%  (7/185)	4.9%  (9/185)	0.61	0.028
Diabetics on insulin	15.7%  (29/185)	13.0%  (24/185)	0.46	−0.048
Triple vessel disease	30.3%  (56/185)	27.6%  (51/185)	0.57	−0.120
Pre-op inotrope use	0.5%  (1/185)	1.6%  (3/185)	0.32	0.061
Elective status	50.8%  (94/185)	58.4%  (108/185)	0.29	0.132
Family history	50.3%  (93/185)	51.4%  (95/185)	0.84	−0.054
Cerebrovascular disease	15.1%  (28/185)	13.0%  (24/185)	0.55	−0.032
Peripheral arterial disease	13.0%  (24/185)	14.1%  (26/185)	0.76	−0.085
Current/recent smoker	42.2%  (78/185)	43.2%  (80/185)	0.83	−0.076
Dyslipidaemia	60.0%  (111/185)	62.7%  (116/185)	0.59	0.225
Pre-op beta blockade	60.0%  (111/185)	63.8%  (118/185)	0.45	0.056
Age  ± SD	66.72  ± 10.93	68.23  ± 12.35	0.21	0.079
Ejection fraction  ± SD	52.56  ± 11.28	52.99  ± 13.48	0.74	0.039

CABG: coronary artery bypass grafting; PCI: percutaneous coronary intervention; MI: myocardial infarction. SD: standard deviation; Std-Diff: standardized mean difference.

Table 3:

Open in new tab

Baseline characteristics of matched female CABG and PCI patient groups

Variable .	CABG .	PCI .	.	.
Number of patients .	185 .	185 .	P-Value .	Std-Diff .
Females	100%  (185/185)	100%  (185/185)		—
Pre-op stroke	9.2%  (17/185)	7.6%  (14/185)	0.57	−0.080
Previous CV intervention	18.9%  (35/185)	21.6%  (40/185)	0.52	0.083
Previous CABG surgery	3.8%  (7/185)	4.3%  (8/185)	0.79	0.025
Previous valve surgery	0	0	—	—
Previous PCI procedure	16.2%  (30/185)	17.3%  (32/185)	0.78	0.060
Hypertensive	85.4%  (158/185)	83.2%  (154/185)	0.57	−0.106
Angina	84.8%  (156/184)	88.1%  (163/185)	0.35	0.127
Heart failure	11.4%  (21/184)	10.8%  (20/185)	0.85	−0.018
Renal failure	4.3%  (8/185)	5.4%  (10/185)	0.63	0.050
On dialysis	2.2%  (4/185)	1.6%  (3/184)	0.71	−0.086
Pre-op MI	23.8%  (43/181)	28.6%  (53/185)	0.29	0.166
Arrhythmia	3.8%  (7/185)	4.9%  (9/185)	0.61	0.028
Diabetics on insulin	15.7%  (29/185)	13.0%  (24/185)	0.46	−0.048
Triple vessel disease	30.3%  (56/185)	27.6%  (51/185)	0.57	−0.120
Pre-op inotrope use	0.5%  (1/185)	1.6%  (3/185)	0.32	0.061
Elective status	50.8%  (94/185)	58.4%  (108/185)	0.29	0.132
Family history	50.3%  (93/185)	51.4%  (95/185)	0.84	−0.054
Cerebrovascular disease	15.1%  (28/185)	13.0%  (24/185)	0.55	−0.032
Peripheral arterial disease	13.0%  (24/185)	14.1%  (26/185)	0.76	−0.085
Current/recent smoker	42.2%  (78/185)	43.2%  (80/185)	0.83	−0.076
Dyslipidaemia	60.0%  (111/185)	62.7%  (116/185)	0.59	0.225
Pre-op beta blockade	60.0%  (111/185)	63.8%  (118/185)	0.45	0.056
Age  ± SD	66.72  ± 10.93	68.23  ± 12.35	0.21	0.079
Ejection fraction  ± SD	52.56  ± 11.28	52.99  ± 13.48	0.74	0.039

Variable .	CABG .	PCI .	.	.
Number of patients .	185 .	185 .	P-Value .	Std-Diff .
Females	100%  (185/185)	100%  (185/185)		—
Pre-op stroke	9.2%  (17/185)	7.6%  (14/185)	0.57	−0.080
Previous CV intervention	18.9%  (35/185)	21.6%  (40/185)	0.52	0.083
Previous CABG surgery	3.8%  (7/185)	4.3%  (8/185)	0.79	0.025
Previous valve surgery	0	0	—	—
Previous PCI procedure	16.2%  (30/185)	17.3%  (32/185)	0.78	0.060
Hypertensive	85.4%  (158/185)	83.2%  (154/185)	0.57	−0.106
Angina	84.8%  (156/184)	88.1%  (163/185)	0.35	0.127
Heart failure	11.4%  (21/184)	10.8%  (20/185)	0.85	−0.018
Renal failure	4.3%  (8/185)	5.4%  (10/185)	0.63	0.050
On dialysis	2.2%  (4/185)	1.6%  (3/184)	0.71	−0.086
Pre-op MI	23.8%  (43/181)	28.6%  (53/185)	0.29	0.166
Arrhythmia	3.8%  (7/185)	4.9%  (9/185)	0.61	0.028
Diabetics on insulin	15.7%  (29/185)	13.0%  (24/185)	0.46	−0.048
Triple vessel disease	30.3%  (56/185)	27.6%  (51/185)	0.57	−0.120
Pre-op inotrope use	0.5%  (1/185)	1.6%  (3/185)	0.32	0.061
Elective status	50.8%  (94/185)	58.4%  (108/185)	0.29	0.132
Family history	50.3%  (93/185)	51.4%  (95/185)	0.84	−0.054
Cerebrovascular disease	15.1%  (28/185)	13.0%  (24/185)	0.55	−0.032
Peripheral arterial disease	13.0%  (24/185)	14.1%  (26/185)	0.76	−0.085
Current/recent smoker	42.2%  (78/185)	43.2%  (80/185)	0.83	−0.076
Dyslipidaemia	60.0%  (111/185)	62.7%  (116/185)	0.59	0.225
Pre-op beta blockade	60.0%  (111/185)	63.8%  (118/185)	0.45	0.056
Age  ± SD	66.72  ± 10.93	68.23  ± 12.35	0.21	0.079
Ejection fraction  ± SD	52.56  ± 11.28	52.99  ± 13.48	0.74	0.039

CABG: coronary artery bypass grafting; PCI: percutaneous coronary intervention; MI: myocardial infarction. SD: standard deviation; Std-Diff: standardized mean difference.

Because matching can limit sample size as well as limiting analysis to the subgroups which matched, multivariable Cox regression was performed on both the entire unmatched male and female patient groups to determine if selection of PCI versus CABG influenced outcome when adjusting for differences between the CABG and PCI patient populations. In the men, PCI versus CABG was associated with increased MACE, HR = 1.97 (95% CI: 1.58, 2.44; P < 0.001), whereas in women CR selection did not impact outcome, HR = 1.21 (95% CI: 0.90, 1.63; P = 0.21). (Supplementary Table S1)

Men versus women

In order to better clarify the nature of the difference in outcome between men and women, we then performed propensity matching with sex as the dependent variable to arrive at two matched groups of male/female patients, 255 of each sex for CABG and 618 for PCI (Table 3). When matched for risk factors we did not identify a sex-specific treatment difference for either CABG or PCI. For CABG, when comparing propensity-score matched men with women, the 5-year actuarial MACE-free survival was 79.1 ± 2.6% vs 73.6 ± 2.8%, HR 1.301 (95% CI: 0.933–1.815; P = 0.12). Similarly, for the PCI group, male versus female, 5-year MACE-free survival was 72.6 ± 1.8% versus 69.9 ± 1.9%, HR 1.126 (95% CI: 0.939–1.350; P = 0.20) (See Fig. 3 for actuarial MACE-free survival of male/female matched CABG [panel A] and PCI [panel B] patients.)

Figure 3:

(A) Freedom from MACE in CABG patients well matched for risk factors between men and women. (B) Freedom from MACE in PCI patients well matched for risk factors between men and women. MACE: major adverse cardiac events; CABG: coronary artery bypass surgery; PCI: percutaneous coronary interventions.

Open in new tabDownload slide

DISCUSSION

Although there has been a dramatic decrease in cardiac mortality in the past two decades, cardiovascular disease remains the leading killer of both men and women in most of the western world. In Europe, this illness accounts for 56% of deaths in women and 43% in men [12]. However, women tend to be underrepresented in both prospective and retrospective studies of therapeutic approaches [13]. Since it is now well known that there are marked sex-specific differences in both the presentation and pathophysiology of cardiovascular disease, it is extremely important to evaluate the potential role of gender in determining therapeutic outcome [14–16]. This need is particularly pressing in situations in which there is considerable professional equipoise, such as the optimal interventional approach for patients with advanced multi-vessel coronary artery disease. The sheer number of prospective, retrospective and ongoing trials on this topic attests to the clinical ambivalence regarding this topic [17]. The question is not which is better, CABG or PCI, but rather, which is better for a given patient—and what are the factors that contribute to finding the answer. It for these reasons that we investigated the data from a ‘real world’ all-comer multi-centre community-based registry to address the hypothesis that gender will impact the relative outcome of CABG versus PCI for patients with an acute non-STEMI manifestation of multi-vessel coronary artery disease. Since the patient population was somewhat heterogeneous, with both elective and acute coronary syndrome patients and since sample size would not have permitted meaningful sex-specific analyses of multiple subgroups, multiple risk factors (urgency, previous MI, haemodynamic instability, inotrope usage, number of diseased vessels) were used to adjust for patient characteristics. As the SYNTAX score had not yet been introduced at the time of data collection, we were not able to analyse data according to lesion severity—a feature which clearly warrants further investigation in future gender-specific analyses [18].

As is common in other registries, slightly less than one third (31.4%) of our study population were women, with a relatively similar distribution among the CABG (30.4%) and PCI (31.9%) populations [13]. Prevalence of risk factors was also similar to that reported elsewhere, with the female patients being older and having a higher incidence of cerebrovascular disease, hypertension and heart failure, while the men were more commonly smokers who had had more prior procedures and tended to have a lower ejection fraction [15].

Although the CABG population had an improved MACE-free survival, when examined by gender, it appears as though all of the benefit for CABG emerged from the male population. Outcomes between PCI and CABG were virtually indistinguishable in the female population and appeared similar to that for PCI in the male population. Even among well-matched male and female populations, men appeared to benefit from CABG relative to PCI but women did not. Although it has been well-described that women have a higher operative mortality than men for CABG [1–5]. There is mounting evidence that long-term survival may actually be more favourable for women, perhaps related to female longevity in general [19–22]. Therefore, it might be anticipated that the initial benefit for male gender that we observed in the CABG population would be ameliorated over time. The fact that this was not apparent in our data may relate to the older age and associated risk factors of our female population at study entry, as well possibly as the length of follow-up. It is therefore notable that when male and female patients were well matched, there was no difference in outcome between the two interventional strategies.

It might be argued that the size of the female cohort in this study, as is so common in other studies, is simply too small to discern significant differences in therapeutic strategy. Perhaps this might explain why recently reported shorter follow-up of a larger clinical registry drew somewhat different conclusions [23]. Although this is certainly a possibility, the homogeneity of the findings in the matched female CABG/PCI comparisons really gives no signal to suggest that a larger sample size would have altered the findings.

Perhaps the most interesting findings of this study are that the entire long-term outcome benefit for CABG versus PCI was due to that which was observed in the male population. Women presenting with non-STEMI multi-vessel disease appear not to do as well as men, but do equally well with CABG or PCI. When the male population is well matched to the female for clinical risk factors, the benefit for CABG over PCI appears to disappear. Although the original male sample size was diminished by this match, at very least the difference appears to be diminished if not removed altogether. Therefore, some if not most of the sex difference in clinical outcome appears to be related to risk profile rather than a substantial underlying difference in pathophysiology. This finding may well be due to the fact that this population was already preselected for those patients who underwent a revascularization procedure. IMA grafting was less common in this study population than generally reported for that time, but was comparable between the sexes [24]. Although the use of BIMA grafting was more common in men, the small difference is unlikely to have accounted for the differential outcome of CABG surgery. The risk factors associated with MACE were not identical between sexes. Age, congestive heart failure previous PCI and previous stroke were associated with the primary outcome in both groups; however, additional risk factors were notable in men—urgent status, previous CABG, inotrope usage and hypertension, while hyperlipidaemia appeared to impact the outcome in women but not in men. The different risk profile and stage of presentation in the female population likely accounted for the major portion of these differences. These findings strongly support the conclusion that the influence of sex, whether as an independent factor or as a marker of associated risk factors, is essential to consider in the evaluation of comparative outcomes for different strategies of coronary revascularization.

LIMITATIONS

As with any retrospective registry trial, this study was subject to the limitation of selection bias. Perhaps this selection bias was even more marked by the absence of a ‘Heart Team’ approach to revascularization in the era of this study. Although as an inclusive analysis of a real-world population, the analysis was not limited by the pre-specified selection bias which leaves most RCT’s in this field with only 3–8% of screened patients being enrolled, there are nonetheless treatment decisions which necessarily bias treatment decisions [25, 26]. Even though statistical modelling was successful in arriving at groups that were well-matched for identified variables, the potential impact of variables not included in these models (frailty, psycho-social factors, patient preference, etc.) cannot be discounted.

Since this was a cross-sectional study, the incidence of MACE may have been underestimated. Once patients were identified as having died by SSDI, the families were not contacted and therefore the incidence of events prior to death may have been underreported.

Although this study included both patients with drug-eluting and bare metal stents, as well as on- and off-pump CABG, patient entry was prior to the era of second- generation drug eluting stents. Recent evidence suggests that the incremental long-term benefit seen in previous studies for CABG over PCI may no longer be applicable with the use of newer second-generation drug-eluting stents [27]. Whether or not gender remains an important differentiating factor with this emerging form of therapy remains to be determined. Similarly, it is unclear whether or not more uniform use of BIMA grafting would have ameliorated or even eliminated the gender differences in surgical outcomes, as has been suggested by others [28] The reluctance of the surgical community to embrace multiple arterial grafting makes unlikely that this BIMA grafting will impact gender differences in the near future [29]. Although it cannot be assumed that the results of this study would necessarily be reflected in contemporary practice, the purpose of this report is to highlight the gender differences in results of coronary revascularization. Of note, the most recent 2014 ESSC/EACTS Guidelines on myocardial revascularization, although dealing extensively with the topic of CABG versus PCI, do not explore the potential impact of gender on decision-making—highlighting the potential significance of these findings and further emphasizing the need for ongoing research on this topic as therapeutic advances are made [30].

Lastly, unfortunately, this registry was not designed to capture the incidence of stroke, which, in many studies has been a distinguishing feature between CABG and PCI. Even though our initial study reported a relatively low incidence of periprocedural stroke (0.7% CABG vs 0.1% PCI, P = 0.03), long-term risk could not be evaluated [10].

CONCLUSION

In conclusion, in the era of which saw the introduction of first generation drug-eluting stents and off-pump CABG, patient sex—whether primarily or through associated risk factors–had a major impact on the long-term outcomes of CABG versus PCI for patients with multi-vessel coronary artery disease. Future studies should therefore carefully account for the impact of gender in the evaluation of outcomes comparing different therapeutic options.

SUPPLEMENTARY MATERIAL

Supplementary material is available at EJCTS online.

FUNDING

This work was supported by unrestricted educational grants from the Florida Heart Research Institute and the Cardiopulmonary Research Science and Technology Institute.

Conflict of interest: none declared.

REFERENCES

Author notes