In the period from 2008 to 2018, 969 patients underwent a revascularization procedure. Of these, 410 underwent PCI and 559 underwent CABG. The median follow-up of our sample was 5 years (Figure 1). A cohort flow diagram is presented in the supplement eFigure 1.
In the CABG sample, the SSI median was 23 (17-29.5), SSII median 25.4 (19.2-32.8), and rSS median 2 (0-6.5). A LIMA to LAD was present in 97.7%, and 12.3% received an additional arterial graft, 51.3% of patients were diabetic, and complete revascularization was achieved in 40.1% (Table 1). MACCE occurred in 87 (15.6%), death in 36 (6.4%), MI in 31 (5.5%), repeat revascularization in 21 (3.8%), and stroke in 10 (1.8%) patients (Table 2).
In the PCI group, the SSI median was 14 (10-19.1), SSII median 28.7 (23-34.2), and rSS median 4.7 (0-9). DES was used in 37.1% of patients and 60.2% had diabetes; complete revascularization was achieved in 25.9% (Table 1). MACCE occurred in 87 (21.2%), death in 36 (8.8%), MI in 28 (6.8%), repeat revascularization in 47 (11.5%), and stroke in 13 (3.2%) patients (Table 2).
Table 1
Baseline characteristics of CABG and PCI samples.
Variables
|
PCI population – 410
|
CABG – 559
|
P-value
|
Age (IQR)
|
61 (54–67)
|
63 (57–69)
|
< .001
|
Male sex (%)
|
258 (62.9)
|
398 (71.2)
|
< .001
|
Hypertension (%)
|
331 (80.9)
|
444 (79.4)
|
.564
|
Diabetes (%)
|
247 (60.2)
|
287 (51.3)
|
.006
|
Smoke history (%)
|
179 (43.7)
|
308 (55.1)
|
< .001
|
EF (IQR)
|
62,5 (58–69)
|
60 (51–64)
|
< .001
|
SYNTAX score I (IQR)
|
14 (10-19.1)
|
23 (17-29.5)
|
< .001
|
Residual SYNTAX score (IQR)
|
4,7 (0–9)
|
2 (0–6,5)
|
< .001
|
SYNTAX score II (IQR)
|
28,7 (23-34.2)
|
25,4 (19.2–32.8)
|
< .001
|
PAD (%)
|
22 (5.4)
|
67 (12)
|
< .001
|
COPD (%)
|
6 (1.5)
|
27 (4.8)
|
.004
|
GFR (IQR)
|
67 (57–80)
|
69 (58–79)
|
.514
|
LDL (IQR)
|
117 (89–145)
|
106 (83–134)
|
.002
|
Aspirin / Clopidogrel (%)
|
406 (99)
|
553 (98.9%)
|
.882
|
Statin (%)
|
400 (97.6)
|
553 (98.9%)
|
.099
|
Three-vessel disease (%)
|
215 (51)
|
220 (39.4)
|
< .001
|
Left-main (%)
|
15 (3.7)
|
116 (20.8)
|
< .001
|
Complete revascularization (%)
|
106 (25.9)
|
224 (40.1)
|
< .001
|
Drug eluting stent (%)
|
152 (37.1)
|
N.A
|
-
|
On-pump CABG (%)
|
N.A
|
323 (57.8)
|
-
|
Left internal thoracic artery (%)
|
N.A
|
546 (97.7)
|
-
|
Second arterial graft (%)
|
N.A
|
69 (12.3)
|
-
|
Number of grafts (SD)
|
N.A
|
2.9 (± 0.64)
|
-
|
|
Table 2
Event rate in CABG and PCI.
Events
|
CABG (n = 559)
|
PCI (n = 410)
|
P-value
|
MACCE, n (%)
|
87 (15.6)
|
87 (21.2)
|
.008
|
Death, n (%)
|
36 (6.4)
|
36 (8.8)
|
.195
|
MI, n (%)
|
31 (5.5)
|
28 (6.8)
|
.333
|
Revasc, n (%)
|
21 (3.8)
|
47 (11.5)
|
< .001
|
Stroke, n (%)
|
10 (1.8)
|
13 (3.2)
|
.174
|
MACCE: major adverse cardiac and cerebrovascular events; MI: myocardial infarction; Revasc: Repeat revascularization. |
Table 3: Univariate and Multivariate Cox Regression for MACCE and Death.
Variables
|
MACCE
|
Death
|
Univariate analysis
|
Multivariate analysis
|
|
|
HR (CI 95%)
|
P value
|
HR (95%CI)
|
p value
|
HR (CI 95%)
|
P value
|
HR (95%CI)
|
P value
|
HTN
|
1.416 (0.939-2.135)
|
.097
|
|
|
1.591 (0.816-3.102)
|
.173
|
|
|
DM
|
1.272 (0.938-1.725)
|
.121
|
|
|
1.574 (0.964-2.569)
|
.070
|
|
|
Smoke history
|
0.861 (0.640-1.160)
|
.325
|
|
|
1.061 (0.771-1.460)
|
.716
|
|
|
SSI (each point)*
|
1.014 (0.998-1.030)
|
.089
|
|
|
1.010 (0.985-1.036)
|
.449
|
|
|
rSS (each point)*
|
1.043 (1.018-1.067)
|
.001
|
1.042 (1.017-1.067)
|
0.001
|
1.013 (0.973-1.055)
|
.534
|
|
|
Age (each point)*
|
1.010 (0.993-1.027)
|
.243
|
|
|
1.022 (0.995-1.050)
|
.110
|
|
|
Three-vessel
|
1.119 (0.831-1.507)
|
.460
|
|
|
1.117 (0.703-1.774)
|
.640
|
|
|
PAD
|
1.079 (0.662-1.757)
|
.761
|
|
|
1.527 (0.783-2.978)
|
.214
|
|
|
COPD
|
0.299 (0.074-1.206)
|
.090
|
|
|
0.047 (<0.001-10.033)
|
.264
|
|
|
GFR (each point)*
|
0.992 (0.983-1.000)
|
.060
|
|
|
0.984 (0.971-0.998)
|
.025
|
0.987 (0.973-1.001)
|
.029
|
LDL (each point)*
|
0.998 (0.994-1.001)
|
.193
|
|
|
0.998 (0.992-1.003)
|
.403
|
|
|
EF (each point)*
|
0.973 (0.959-0.987)
|
<.001
|
0.973 (0.960-0.987)
|
<0.001
|
0.972 (0.951-0.993)
|
.010
|
0.972 (0.951-0.993)
|
.009
|
Female sex
|
1.153 (0.855-1.554)
|
.350
|
|
|
0.865 (0.538-1.389)
|
.548
|
|
|
Left-main
|
1.107 (0.734-1.669)
|
.627
|
|
|
1.126 (0.592-2.140)
|
.717
|
|
|
*Variables were analyzed continuously. COPD: chronic obstructive pulmonary disease; DM: diabetes mellitus; GFR: Glomerular filtration rate; HTN: hypertension; LDL: low-density lipoprotein; MACCE: major cardiovascular and cerebrovascular events; PAD: peripheral artery disease, NA: not available.
Before stratification according to SYNTAX scores, a comparison between CABG and PCI whole cohorts found a significant lower difference for MACCE (15.6% versus 21.2%, P log-rank =.008; adjusted hazard ratio, 1.984; 95% CI, 1.134-3.470; P=.016) and repeat revascularization (3.8% versus 11.5%, P log-rank <.001; adjusted hazard ratio, 4.356; 95% CI, 1.749-10.860; P=.002; respectively) in the surgical group. The death rate was similar between PCI versus CABG (8.8% versus 6.4%, P log-rank =.195) (Figure 1 and Table 2).
1. SYNTAX score I
Analyses of intraobserver (kappa:0.604, 95% CI, 0.269-0.787, P<.001) and interobserver variability (kappa:0.660, 95% CI, 0.390-0.825, P<.001) for the SSI were at least moderate. The SSI tertiles obtained were a low SSI ≤15 (n=328), intermediate SSI 15-24 (n=340), and high SSI >24 (301). CABG sample had 89, 214, and 256 patients in low, intermediate, and high SSI tertiles, respectively. In PCI, 239, 126, and 45 patients were found in low, intermediate, and high SSI tertiles.
The composite outcome in the lower tertile of SSI was not significantly different between CABG and PCI: 12 (13.5%) versus 43 (18%), P=.136. The only significant difference found was for the individual analysis of repeat revascularization: n=3 (3.4%) in CABG versus 27 (11.3%) in PCI (P log-rank=.021; adjusted hazard ratio, 7.071; CI 95%, 1.367-36.561; P=.020). No difference was found in the individual analysis of death (P=.227), myocardial infarction (P=.637), or stroke (P=.222) (eTable 1 in the supplement).
In the intermediate SSI tertile, the primary outcome was significantly different between CABG and PCI, with an event rate of 25 (11.7%) in CABG versus 29 in PCI (23%) (P log-rank=.002; adjusted hazard ratio, 2.704; 95% CI, 0.967-7.563; P=.58). There was a difference for myocardial infarction: n=8 (3.7%) in CABG versus n=15 (11.9%) (log-rank, P=.002; adjusted hazard ratio, 3.131; 95% CI, 0.556-17.637; P=.196). Repeat revascularization rate was also higher in PCI: n=7 (3.3%) in CABG versus n=12 (9.5%) in PCI (log-rank, P=.008; adjusted hazard ratio, 3.061; 95% CI, 0.587-15.963; P=.184). No difference was found in the individual analysis of death (P=.201) or stroke (P=.419) (eTable 1 in the supplement).
MACCE rate was significantly different between CABG and PCI in the higher tertile of SSI: n=50 (19.5%) in CABG versus n=15 (33.3%) in PCI (P log-rank=.016; adjusted hazard ratio, 5.157; 95% CI, 1.698-15.659; P=.004). Repeat revascularization was higher in PCI, n=9 (20%) versus CABG, n=10 (3.9%) (log rank P<.001; adjusted hazard ratio, 89.909; 95% CI, 11.672-692.570; P<.001). Stroke rate was also significantly different: n=6 (2.3%) in CABG and 5 (11.1%) in PCI (log rank P=.004; adjusted HR, 10.132; 95% CI, 0.592-179.726; P=.110). No difference was found in the individual analysis of death (P=.325) or myocardial infarction (P=.986) (eTable 1 in the supplement).
2. SYNTAX score II
The SSII tertiles obtained were a low SSII ≤23 (n=330), intermediate SSII 23-31.3 (n=320), and high SSII >31.3 (319). CABG group had 89, 214, and 256 patients in low, intermediate, and high SSII tertiles, respectively. In PCI, 239, 126, and 45 patients were in low, intermediate, and high SSII tertiles.
The composite outcome in the lower tertile of SSII was not significantly different between CABG and PCI: 26 (11.4%) versus 19 (18.6%) (P=.055). Repeat revascularization was higher in PCI: n=3 (1.3%) events in CABG versus n=27 (11.3%) in PCI (P log-rank<.001; adjusted hazard ratio, 5.402; 95% CI, 0.472-61.822; P=.175). No difference was found in the individual analysis of death (P=.113), MI (P=.799), and stroke (P=.644) (eTable 2 in the supplement).
In the intermediate SSII tertile, the primary outcome was not significantly different between CABG and PCI with an event rate of 31 (17.8%) in CABG versus 28 (19.2%) in PCI (P=.568). No difference was found for death (P=.295), MI (P=.329), repeat revascularization (P=.156), or stroke, n=4 (P=0.773) (eTable 2 in the supplement).
MACCE rate was statistically similar between CABG and PCI in the higher tertile of SSII: n=30 (19.1%) in CABG versus n=40 (24.7%) in PCI (P=.132). A significant difference was found only for repeat revascularization: n=7 (4.5%) in CABG versus n=23 (14.2%) in PCI (P log-rank=.002; adjusted hazard ratio, 12.607; 95% CI, 2.848-55.800; P=.001). There was no difference in the individual analysis of death (P=.172), myocardial infarction (P=.972), or stroke (P=.234) (eTable 2 in the supplement).
3. Residual SYNTAX score
A correlation evaluation between SSI and rSS was performed for both surgical procedures. A moderate correlation was found after PCI (R=0.686, P<.001), but it was weak after CABG (R=0.200, P<.001).
Residual SYNTAX score tertiles were low rSS=0 (n=330), intermediate rSS 0-6 (n=337), and high rSS >6 (302). CABG group had 224, 195, and 140 patients in low, intermediate, and high SSII tertiles, respectively. In PCI, 106, 142, and 162 patients were in low, intermediate, and high SSII tertiles.
MACCE rate of low rSS tertile was n=26 (11.6%) in CABG and n=14 (13.2%) in PCI (P=0.533). No difference was found in the individual analysis of death (P=.795), MI (P=.673), repeat revascularization (P=.275), or stroke (P=0.407) (eTable 3 in the supplement).
Intermediate rSS tertile had a primary outcome rate not significantly different between CABG and PCI: n=33 (16.9%) in CABG versus n=35 (24.6%) in PCI (P=.066). The only difference found was for repeat revascularization: n=8 (4.1%) in CABG versus n=22 (15.5%) in PCI (log-rank, P<.001; adjusted hazard ratio, 6.008; 95% CI, 1.435-25,152; P=.014). No difference occurred for death (P=.269), MI (P=.663), or stroke (P=.751) (eTable 3 in the supplement).
In the higher tertile of rSS, the MACCE rate was statistically similar between groups: n=28 (20%) in CABG versus n=38 (23.5%) in PCI (P=.291). A significant difference was found for repeat revascularization: n=7 (5%) in CABG versus n=21 (13%) in PCI (P log-rank=.012; adjusted hazard ratio, 12.604; 95% CI, 2.524-62.949; P=.002). No difference occurred in the analysis of death (P=.497), MI (P=.621), or stroke (P=.571) (eTable 3 in the supplement).
4. SYNTAX scores accuracy, subgroup, and multivariate analysis
Independent predictors for MACCE in our sample were evaluated in a model testing each component of SSII with rSS, SSI, and other baseline variables. In this sample, the rSS as a continuous variable was an independent predictor for MACCE (hazard ratio, 1.042; 95% CI, 1.017-1.067; P=.001) as was EF. For the analysis of death, EF and GFR were the only independent predictors (Table 3). In the subgroup analysis, there was no interaction among the tertiles of the 3 SYNTAX scores and the revascularization procedure, but an interaction for the reduction of MACCE in CABG compared with PCI was found for diabetes (P=.034) (Figure 2).
ROC curves were built to examine the accuracy for MACCE and death in our sample. SSI AUC was 0.540 (95% CI, 0.494-0.586; P=.095) for MACCE and 0.592 (95% CI, 0.448-0.590; P=.592) for death. SSII AUC was 0.567 (95% CI, 0.522-0.613; P=.005) for MACCE and 0.609 (95% CI, 0.541-0.677; P=.002) for death. RSS AUC was 0.581 (95% CI, 0.535-0.627, P=.001) for MACCE and 0.528 (95% CI, 0.462-0.595; P=.034) for death.