Objective: We examined the effects of learning curve on clinical outcomes and operative time in minimally invasive coronary artery bypass grafting (MICS CABG).
Methods: We studied 210 consecutive MICS CABG cases performed by the same surgeon, composed of 3 cardiopulmonary bypass (CPB)–assisted single-vessel small thoracotomy (SVST), 87 off-pump SVST, 51 CPB-assisted multivessel small thoracotomy (MVST), and 69 off-pump MVST. For each MICS CABG technique, the frequency of early clinical events (mortality, reopening, stroke, myocardial infarction, and revascularization) was compared between the first 25 cases and the remainder. Logarithmic curve regression analysis and a cumulative summation technique were performed to assess the correlation between operative time and the performed number of each technique.
Results: There was no mortality, and there were 10 conversions to standard sternotomy, all of which were intended as off-pump MVST (P < 0.001, vs other procedures). Experience was otherwise not associated with perioperative outcome. However, experience numbers correlated with operative time in off-pump SVST and off-pump MVST (122 ± 30 minutes, R2 = 0.18, P < 0.001, and 241 ± 80 minutes, R2 = 0.38, P < 0.001, respectively) but not in CPB-assisted MVST (258 ± 44 minutes, R2 = 0.004, P = 0.7). No complications occurred as a result of CPB assistance.
Conclusions: Minimally invasive coronary artery bypass grafting can be safely initiated, with a very low perioperative risk. Pump assistance is a good strategy to alleviate some of the learning curve and avoid conversions to sternotomy when initiating a multivessel MICS CABG program.
From the *Division of Cardiac Surgery, and †Division of Cardiac Anesthesiology, University of Ottawa Heart Institute, Ottawa, ON Canada; and ‡McLaren Regional Medical Center Family Medicine, Michigan State University College of Human Medicine, Flint, MI USA.
Accepted for publication September 24, 2013.
Presented at the Annual Scientific Meeting of the International Society for Minimally Invasive Cardiothoracic Surgery, June 12–15, 2013, Prague, Czech Republic.
Disclosures: Dai Une, MD, has received honoraria from Medtronic Japan Co, Ltd, Tokyo, Japan. Marc Ruel, MD, MPH, received research support and honoraria from Medtronic, Inc, Minneapolis, MN USA. Harry Lapierre, MD; Benjamin Sohmer, MD; and Vaneet Rai declare no conflicts of interest.
Address correspondence and reprint requests to Marc Ruel, MD, MPH, Division of Cardiac Surgery, University of Ottawa Heart Institute, 40 Ruskin St, Suite 3402, Ottawa, ON Canada. E-mail: firstname.lastname@example.org.
A small thoracotomy technique, which represents a novel technique of minimally invasive coronary artery bypass grafting (MICS CABG) for single-vessel or multivessel revascularization, has been initiated at the University of Ottawa Heart Institute (Ottawa, ON Canada) and Staten Island University Hospital (Staten Island, NY USA) in 2005. This technique may provide the patients with earlier recovery and fewer infections.1 Although the safety and the efficacy of MICS CABG have been reported,2 this technique is still limited to a few experienced surgeons because advanced exposure and grafting techniques are required.
In MICS CABG, surgeons and their surgical team need to master different methods of patient positioning, lung ventilation, intricate mediastinal exposures, and the setup of an epicardial stabilizer/apical positioner. Because the surgical techniques are performed in a narrow tunnel space, surgeons have to get familiar with internal thoracic artery harvest from a lateral approach, distant aortic control and side-clamping, and exposure of target vessels via a small thoracotomy.3,4
Evaluating the learning curve of MICS CABG may help other surgeons, their surgical teams, and institutions initiate this technique safely, thereby allowing more patients to derive the potential benefits of this procedure. We investigated whether learning curve issues increase 1-month clinical events after MICS CABG, and how many cases are required to acquire acceptable skills regarding operative time in MICS CABG, by evaluating the learning curve at one center.
Between 2005 and 2012, excluding repeated coronary artery bypass grafting (CABG) cases or previous thoracic surgery, MICS CABG was performed in 210 consecutive patients by the same surgeon (M.R.) (3 cardiopulmonary bypass [CPB]–assisted single-vessel small thoracotomy [SVST], 87 off-pump SVST, 51 CPB-assisted multivessel small thoracotomy [MVST], and 69 off-pump MVST) at the University of Ottawa Heart Institute (Ottawa, ON Canada). There were 10 conversions to sternotomy and 200 patients who had the intended MICS CABG procedure completed. The surgeon is an attending surgeon who had performed more than 400 off-pump CABG cases before initiating MICS CABG. Table 1 shows the preoperative characteristics and the operative details of the 200 patients who had a completed MICS CABG procedure.
The technique of MICS CABG has previously been described.2–4 Briefly, paravertebral thoracic blockage may be given, and patients are intubated with a double-lumen endotracheal tube or a regular endotracheal tube plus a left bronchial blocker to decompress the left lung during the operation. The patients are positioned in a 15- to 30-degree right decubitus position with the right arm extended. If applicable, the right radial artery is harvested. The left arm is elevated over the head. In addition to the chest, both groins, thighs, and legs are draped should femoral cannulation and/or saphenous vein graft harvest be required.
The operation begins with a 4- to 6-cm thoracotomy in the fourth or fifth intercostal space on the left side of the anterior chest and an incision of the pericardium approximately 3 cm anterior to the left phrenic nerve. After the “window” of the small thoracotomy is pulled cephalad and leftward by using the Rultract Skyhook retractor (Rultract, Inc, Independence, OH USA), the left internal thoracic artery is taken down under direct vision, over its usual, entire length. Heparin is given before the left internal thoracic artery is divided. For proximal anastomoses, the minithoracotomy incision is pulled cephalomedially by the retractor. The ascending aorta is mobilized with 2-0 silk stitches on the pericardium around the ascending aorta and by the placement of a gauze on the right of the aorta. After the proximal pulmonary artery is gently displaced leftward and posteriorly by using an epicardial stabilizer, hand-sewn proximal anastomoses are performed on the ascending aorta with a side clamp, as in a regular off-pump CABG.3 We started these techniques for proximal anastomoses in 2006. The epicardial stabilizer is inserted via an 8-mm incision in the seventh intercostal space. Dye is painted on the grafts to avoid twisting. For distal anastomoses, an apical positioner is inserted via a 5-mm subxiphoid incision. If exposure is not sufficiently good, or if the patient is hemodynamically unstable despite low-dose vasopressors, femoral arterial and venous CPB is initiated. When CPB is used, all anastomoses are performed with a beating heart technique without cardiac arrest. All myocardial territories are accessed by using the apical positioner and the epicardial stabilizer, and anastomoses are performed with 7-0 polypropylene sutures by using shunt tubes and a CO2 blower, as in regular off-pump CABG.
Postoperative Therapy and Follow-up
Perioperative death was defined as death within 30 days of operation, from any cause. All patients received postoperative therapy according to current American College of Cardiology/American Heart Association guidelines, which include diabetes management; smoking cessation counseling; and pharmacotherapy including aspirin, β-blockers, angiotensin-converting enzyme inhibitors, and statins as tolerated.5 The patients were followed up at the clinic at 1 month after discharge.
Data were analyzed using the IBM Statistical Package for the Social Sciences version 20.0 (IBM Corp, Armonk, NY USA). Continuous variables are reported as mean ± SD; and categorical variables, as frequency and proportion. Continuous variables were compared using an unpaired t test or the Mann-Whitney U test, whereas categorical variables were compared using the χ2 test with Yates correction, as appropriate.
The frequency of 1-month clinical events (bleeding, graft revision, myocardial infarction, percutaneous coronary intervention, stroke, and mortality), intensive care unit length of stay, hospital length of stay, and operative time were compared between the first 25 cases and the remainder, according to each subgroup: off-pump SVST, CPB-assisted MVST, and off-pump MVST.
Logarithmic curve and linear regression analyses were performed to evaluate the correlation between operative time and experience numbers in each technique, and the model with the best R2 coefficient was chosen as the fitting model for each subgroup.
In addition to the above mentioned regression analyses, the cumulative summation (CUSUM) technique was performed to assess the learning curve of operative time in each technique.6 Failure was defined as a surgery that took longer time than the mean operative time in each procedure. The cumulative incidence of failure was plotted along with the number of experienced operations. We arbitrarily and a priori determined that an acceptable failure rate (p0) was 0.4 and an unacceptable failure rate (p1) was 0.6. The false-positive error rate (type I error rate; α) and the false-negative error rate (type II error rate; β) were set at 0.1. For alert lines, α and β were set at 0.2.
All reported P values are two sided. P values of less than 0.05 are considered significant.
Ten conversions to sternotomy occurred, all before 2010, during operations that were intended as off-pump MVST. The conversion rate of off-pump MVST was 14.5% (10/69), which was significantly higher compared with other groups (10/69 vs 0/141, P < 0.001) and CPB-assisted MVST (10/69 vs 0/51, P = 0.01). The causes of conversion to sternotomy included poor exposure in four cases, hemodynamic instability in four cases, left internal thoracic artery injury in one case, and intolerance to one-lung ventilation in one case. In these 10 sternotomy cases, there were no further major clinical events such as death, myocardial infarction, revascularization, stroke, or reopening for bleeding.
There were no perioperative deaths in this study. Cardiopulmonary bypass assist was used in 54 patients. The comparisons of patient characteristics between CPB-assisted MVST and off-pump MVST are shown in Table 1. The pump-assisted group included more patients with diabetes mellitus, and the pump-assisted patients also had more anastomoses.
Seven patients underwent reopening; five for perioperative bleeding and two for graft revision, and one patient had a perioperative myocardial infarction. The overall median intensive care unit and hospital stay durations were 1 (range, 1–42) day and 5 (range, 3–61) days, respectively. Follow-up to 1 month was 100% complete.
Learning Curve on Clinical Outcomes
Table 2 shows the comparisons of clinical outcomes between the first 25 cases and the remainder, for each technique. In off-pump SVST, there were no significant differences in the frequency of 1-month clinical events, but the median hospital stay of the first 25 cases was significantly shorter than that of the remainder (4 days [range, 3–6] and 5 days [range, 3–61], respectively, P = 0.01), likely reflecting increased preoperative risk in the remainder group. The cases beyond the first 25 include the 30th case, who stayed for 61 days because of pneumonia, and the 58th case, who stayed for 42 days because of complications of another operation. In off-pump MVST, except for conversions to sternotomy, there was no difference in clinical outcomes between the first 25 and subsequent patients. In CPB-assisted MVST, there was no significant clinical outcome difference throughout the study.
Learning Curve on Operative Time
For all groups, the better-fitting model for the correlation between operative time and experience was a logarithmic curve. In off-pump SVST, the operative time was significantly associated with experience (mean ± SD operative time, 122 ± 30 minutes, P < 0.001, R2 = 0.18) (Fig. 1A). In CUSUM analysis, the performance was considered unacceptable between the 6th and 12th cases, whereas the performance reached the acceptable level at the 66th case (Fig. 1B).
In CPB-assisted MVST, the operative time was not significantly associated with experience (mean ± SD operative time, 258 ± 44 minutes, P = 0.7, R2 = 0.004) (Fig. 2A). The CUSUM analysis showed that the cumulative failure never reached the unacceptable level, and the procedure reached the acceptable level at the 16th case (Fig. 2B). After the 16th case, the performance was acceptable except for the 17th, 27th, 29th, 31st, 33rd, 34th, and 35th cases.
In off-pump MVST, the operative time significantly correlated with the experience (mean ± SD operative time, 241 ± 80 minutes, P < 0.001, R2 = 0.38) (Fig. 3A). The CUSUM analysis revealed that the cumulative failure line went above the unacceptable line between the 8th and 16th cases, and the performance came below the acceptable line at the 40th case (Fig. 3B).
Minimally invasive coronary artery bypass grafting is still limited to experienced surgeons and their teams because of the need for advanced techniques. To promote safe initiation of this operation, this study examined learning curves of SVST and MVST, with or without CPB assistance. There were more conversions to sternotomy in patients intended to undergo off-pump MVST. In addition, the multivessel off-pump technique was associated with a learning curve regarding operative time.
Conversion to Sternotomy
All conversion cases were intended as off-pump MVST. Of 10 conversion cases, 8 cases were due to poor exposure or hemodynamic instability. Had these patients been converted to CPB-assisted MVST immediately, it is likely that sternotomy would have been avoided in several cases. This has been our policy since 2010, and no further conversion occurred since.
Learning Curve and Clinical Outcomes
In this study, there were no differences in the frequency of early clinical events (with the exception of conversion to sternotomy for off-pump MVST) between the first 25 cases and subsequent cases in off-pump SVST, CPB-assisted MVST, and off-pump MVST. These results indicate that an attending surgeon and his/her team members who have sufficient experience with off-pump coronary artery bypass grafting can initiate MICS CABG with very low short-term risk, although patient selection also has to be considered. To avoid clinical events due to the learning curve, we started these small thoracotomy approaches with lower-risk patients who generally had preserved left ventricular function.2 In addition, this technique was performed only for elective or semiurgent cases and not used for emergent cases. Along with the development of experience of the surgeon and the surgical team, the proportion of MICS CABG gradually increased from less than 20% to greater than 40% of elective surgical cases between 2005 and 2008, without additional risks.2
Learning Curve and Operative Time
Off-Pump SVST Versus Off-Pump MVST
Operative time significantly correlated with experience in off-pump SVST and off-pump MVST. Compared with off-pump SVST, off-pump MVST had a steeper learning curve with better model fitting. These findings imply that multivessel targets are more steeply associated with a learning curve affecting operative time. In the thoracotomy approach, the left anterior descending artery is generally right near the incision, which does not require a new technique for exposure. On the other hand, the circumflex and right coronary branches need experience and advanced techniques for exposure and anastomoses without CPB. These difficulties in exposure and anastomoses may have caused the steeper learning curve of off-pump MVST.
Cardiopulmonary Bypass–Assisted MVST Versus Off-Pump MVST
This study showed a significant difference of learning curves on operative time between CPB-assisted and off-pump MVST. The use of CPB assistance was decided before or during the operation, more frequently for patients with diabetes mellitus and/or for those requiring more anastomoses. This may be because patients with diabetes mellitus are more likely to have diffuse coronary disease and smaller targets,7 thereby requiring better exposure. Cardiopulmonary bypass assistance enables the surgeon to perform MVST for difficult target vessels without risks or additional operative time attributed to learning curve issues. Cardiopulmonary bypass assistance can provide surgeons not only with good exposure and stable hemodynamics but also with more space for surgical performance in the closed chest. Interestingly, the mean operative time of off-pump MVST was shorter than that of CPB-assisted MVST in all cases. This indicates that it takes extra time (approximately 15–30 minutes) for pump setup and removal and that operative times may be similar between CPB-assisted MVST and off-pump MVST after the learning curve period, with the exception of mandatory pump setup and CPB wean-off time.
How Many Cases Are Required for Optimal Efficiency?
In CUSUM analyses, operative time reached the acceptable level at the 66th, 16th, and 40th cases in off-pump SVST, CPB-assisted MVST, and off-pump MVST, respectively. All techniques were initiated near concomitantly, and we did not have external guidance about how to develop MICS CABG. From the learning curve analyses on operative time, we recommend that approximately 60 SVST cases and 15 CPB-assisted MVST cases be completed by the same surgeon and team before initiating off-pump MVST. In addition, one should not hesitate to use CPB for at least 120 cases of MICS CABG including all techniques. After this period, team training, team function, patient decision making, postoperative care, and surgical skills all seem collectively acceptable.
This is a single-center and single-surgeon study; consequently, the findings are not necessarily generalizable. Patients amenable to a MICS CABG procedure may, in some instances, be selected. The outcome of the CUSUM analyses for the operative time may sometimes be dependent on the patient rather than on the surgeon. Slow procedures for certain patients are technically not a “failure” and may not represent a clinically relevant outcome; however, it is an economic indicator and possibly a surrogate of the difficulty of the operation. We did not measure intraoperative graft flow or graft patency, which may be another good index of clinical outcomes. Finally, the decision to perform conversion to sternotomy in some early-phase patients may have been influenced by worldwide inexperience with the operation, which is now better understood and accepted.
Minimally invasive coronary artery bypass grafting can be safely initiated without mortality or additional morbidity due to learning curve. Pump assistance may be used without additional risk and represents a good strategy to avoid a steep learning curve and the possibility of conversion to sternotomy. Operative time reached acceptable level at the 66th case in off-pump SVST, the 16th case in CPB-assisted MVST, and the 40th case in off-pump MVST. We believe that these findings may help guide surgeons, their teams, and institutions initiate a multivessel MICS CABG program.
1. Lapierre H, Chan V, Sohmer B, Mesana TG, Ruel M. Minimally invasive coronary artery bypass grafting via a small thoracotomy versus off-pump: a case-matched study. Eur J Cardiothorac Surg
. 2011; 40: 804–810.
2. McGinn JT Jr, Usman S, Lapierre H, Pothula VR, Mesana TG, Ruel M. Minimally invasive coronary artery bypass grafting: dual-center experience in 450 consecutive patients. Circulation
. 2009; 120 (suppl): S78–S84.
3. Chan V, Lapierre H, Sohmer B, Mesana TG, Ruel M. Handsewn proximal anastomoses onto the ascending aorta through a small left thoracotomy during minimally invasive multivessel coronary artery bypass grafting: a stepwise approach to safety and reproducibility. Semin Thorac Cardiovasc Surg
. 2012; 24: 79–83.
4. Ruel M, Chan V, Lapierre H, et al. Multi-vessel small thoracotomy (MVST) coronary artery bypass grafting. In: Sellke FW, Ruel M, eds. Atlas of Cardiac Surgical Techniques
. Philadelphia, PA: Elsevier Science; 2009: 83–94.
5. Hillis LD, Smith PK, Anderson JL, et al. 2011 ACCF/AHA Guideline for Coronary Artery Bypass Graft Surgery: a report of the American College of Cardiology Foundation/American Heart Association Task Force on Practice Guidelines. Circulation
. 2011; 124: e652–e735.
6. Yap CH, Colson ME, Watters DA. Cumulative sum techniques for surgeons: a brief review. ANZ J Surg
. 2007; 77: 583–586.
7. Kapur A, Malik IS, Bagger JP, et al. The Coronary Artery Revascularisation in Diabetes (CARDia) trial: background, aims, and design. Am Heart J
. 2005; 149: 13–19.
This is an excellent report from Dr. Dai Une and his colleagues of the University of Ottawa Heart Institute examining the effects of learning curve on clinical outcomes and operative time in minimally invasive coronary artery bypass grafting. Two hundred ten consecutive patients were studied undergoing a combination of cardiopulmonary bypass assisted and off-pump single-vessel and multi-vessel small thoracotomy approaches. In this series, there was no mortality and 10 conversions to standard sternotomy. Experience correlated with operative time in off-pump single- and multi-vessel small thoracotomy cases but not in the cardiopulmonary bypass-assisted cases. The authors concluded that minimally invasive coronary bypass grafting can be safely initiated with low perioperative risk and an acceptable learning curve. Pump assistance was felt to be a good strategy to shorten the learning curve and help avoid conversion to sternotomy when initiating a multi-vessel minimally invasive coronary bypass graft program.
While this is a single-center, single-surgeon study which may not be generalizable to all institutions, this study does show that a team experienced with sternotomy off-pump coronary bypass grafting can initiate a small thoracotomy approach with low short-term risk. There were no differences in the frequency of early clinical events between the first 25 cases and subsequent cases for the small thoracotomy approaches. Off-pump multi-vessel small sternotomy cases had a steeper learning curve, as expected. The study would have been strengthened by measurement of intraoperative graft flow for graft patency as an outcome variable. However, the authors are to be congratulated for their careful analysis of the learning curve and for this excellent contribution to the literature.