Results
Fifty-six patients were included in the original study at our center. Thirty vs 26 patients for the on- respective off-pump groups. All patients were alive 30 days postoperatively. One patient in each group needed early reoperation for bleeding. Two patients developed sternal wound infections, 2/56 (3.6%) and three patients had leg wound infections, 3/56 (5.3%). One patient in the on-pump group needed repeat revascularization with percutaneous coronary intervention (PCI) and one patient in the off-pump group underwent redo surgery due to new coronary lesions and mitral regurgitation. Five patients in the on-pump group and two patients in the off-pump group had died, 7/56 (12.5%). Five out of seven patients died from non-cardiac causes.
The five-year follow-up included 49 patients with 25 vs 24 patients in the on- and off-pump groups respectively. All patients underwent a clinical evaluation according to protocol. Five patients were excluded from CTA due to renal failure and two patients in each group refused to participate in this sub-study. Forty patients underwent CTA with 139 distal anastomoses analyzed, see figure 1. Patient characteristics at five years are reported, see table 1. Similar mean age, gender distribution and most risk factors between the groups. All patients were on anti-platelet therapy. Some numerical differences between the two groups, for example more patients in NYHA class I in the on-pump group. No patient reported pain from the SV harvesting site and only one patient reported numbness. Small coronary targets, ≤ 1mm in diameter, were more frequent in the right coronary territory, 10/41 (24.7%). Six patients received a NT SVG to left anterior descending (LAD) artery due to either borderline stenosis or surgical injury to the LITA. In four patients the NT SVG was used to substitute the LITA in patients with multiple co-morbidities. The mean grafting rate per patient was 3.8 and 3.1 distal anastomoses in the on- and off-pump groups respectively.
The overall five-year patency rate according to the number of distal anastomoses was 123/139, (88.5%) of which 73/80 (91.3%) were in the on-pump group and 50/59 (84.7%) in the off-pump group. The patency rate was higher for the grafts used to the LAD territory than for those grafts used to both the circumflex (Cx) and the right coronary territories, independent of the surgical method used, see figure 2. The total patency rate for LITA was 29/30 (96.7%) and for the NT SVGs 94/109 (86.2%). The patency rate of LITA was similar between the groups, on-pump 16/16 (100%) and off-pump 13/14 (92.8%). The patency rate for NT SVGs was also similar between both surgical procedures, 57/64 (89.1%) in the on-pump group and 37/45 (82.2%) in the off-pump group. However, all NT SVGs that were used to bypass the LAD and the diagonal (D) branches were patent, 20/20 grafts in the on-pump group and12/12 grafts in the off-pump group. The lowest patency for the NT SVG was to the right coronary territory, particularly in off-pump surgery, 16/20 (80.0%) and 10/16 (62.5%) for the on- and off-pump groups respectively, see table 2.
Analyses of a possible superiority in patency for the two surgical procedures stratified to the different target coronary arteries was performed, see table 3. The difference in patency between on- and off-pump was 6.5% (95% CI -7.1 - 20.1; p=0.35), i.e. superiority was not found. For the target coronary arteries, statistically significant differences were found for LAD/D vs Cx (difference 10.6, 95% CI 2.0 – 19.2; p=0.02), LAD/D vs posterior descending artery (PDA) (difference 26.2, 95% CI 10.4 – 41.9; p<0.001) and LAD/D vs Cx or PDA (difference 17.9, 95% CI 9.2 – 26.6; p<0.001). The difference in patency between Cx and PDA was not significant, 15.6 (95% CI -2.2 – 33.4; p=0.09).
As superiority for either of the two surgical procedures was not found, we analyzed a possible equivalence between graft patency in the on- and off-pump groups, see figure 3. We tested an equivalence with a span of at most ± 10 percentage units. As indicated in the figure, the upper 95% CI limit exceeds the limit of 10 % units (15.79) and the lower 95% CI limit is below zero, reaching the value of -2.78. In accordance with Fleming 11 and Blackwelder 12this indicates an inconclusive result, thus neither superiority nor equivalence can be statistically verified, and this is most likely due to a lack of power.