- Email: [email protected]
Is Aortic Valve Repair Reproducible? Analysis of the Learning Curve for Aortic Valve Repair
Canadian Journal of Cardiology
-
(2015) 1e8
Clinical Research
Is Aortic Valve Repair Reproducible? Analysis of the Learning Curve for Aortic Valve Repair Tarek Malas, MDCM, MPH,a Richard Saczkowski, CPC, CCP,b Benjamin Sohmer, MD,c Marc Ruel, MD, MPH,a Thierry Mesana, MD, PhD,a Laurent de Kerchove, MD,d Gebrine El Khoury, MD, PhD,d and Munir Boodhwani, MD, MSCa a b c
Division of Cardiac Surgery, University of Ottawa Heart Institute, Ottawa, Ontario, Canada
Division of Cardiac Perfusion, University of Ottawa Heart Institute, Ottawa, Ontario, Canada
Division of Cardiac Anesthesiology, University of Ottawa Heart Institute, Ottawa, Ontario, Canada d
Department of Cardiovascular Surgery, St-Luc Hospital, Brussels, Belgium
ABSTRACT
RESUM E
Background: Aortic valve (AV) preservation and repair, although effective, is performed in a limited number of centres. Lack of wider application might be due to challenges in dissemination of tacit surgical knowledge. We examined the learning curve in 2 centres that initiated dedicated programs in AV repair. Methods: Prospectively collected data on the first 100 (cohort A) and 150 consecutive patients (cohort B) who underwent AV repair surgery were analyzed. Safety end points included mortality, myocardial infarction or stroke, early AV repeat surgery, re-exploration for bleeding, or pacemaker implantation. Efficiency was assessed according to aortic crossclamp and cardiopulmonary bypass times. Efficacy parameters included residual aortic insufficiency or stenosis. Indices of case complexity included presence of severe aortic insufficiency, nontrileaflet AV, and associated cardiac procedures. Each
servation et la re paration de la valve aortique Introduction : La pre alise e dans un nombre limite de centres. (VA), bien qu’efficace, est re fis en Le manque d’application plus large pourrait être dû à des de matière de diffusion des connaissances tacites en chirurgie. Nous la courbe d’apprentissage dans les deux centres qui avons examine des programmes de die s à la re paration de la VA. ont initie thodes : Les donne es recueillies de façon prospective sur les 100 Me paration de VA premiers patients qui ont subi une chirurgie de re te analyse es. Les (cohorte A) et les 150 suivants (cohorte B) ont e curite incluent la mortalite , l’infarctus du myocarde points limites de se re bral, la reprise pre coce d’une chirurgie de ou l’accident vasculaire ce -examen pour saignement, ou l’implantation d’un stimla VA, le re te e value e en fonction du temps de ulateur cardiaque. L’efficience a e e du pontage cardiopulmonaire. Les clampage de l’aorte et la dure
Aortic valve (AV) repair has emerged as a feasible alternative to AV replacement in selected patients with aortic insufficiency (AI). Valve replacement, although effective at treatment of the primary disease, is associated with multiple adverse events including structural valve deterioration, endocarditis, thromboembolism, and the risks and inconveniences associated with lifelong anticoagulation, in the case of mechanical valves. In
contrast, AV repair (AVr) has been demonstrated in multiple cohort studies to have a low risk of valve-related events.1-3 However, AVr is only practiced in a limited number of centres. There might be a number of reasons for this. First, AVr requires understanding and use of a variety of surgical techniques that might include intervention on the valve cusps, annulus, and aortic root, and might be applied to bicuspid and tricuspid AVs. Second, successful repair requires a thorough understanding of functional AV anatomy, mechanisms of AI, and their echocardiographic and intraoperative assessment. Last, AI is a less commonly encountered pathology and the small volume of cases might contribute to slower dissemination of surgical expertise. Because of the limited dissemination of AVr, the reproducibility of the surgical techniques has been questioned. Therefore, the objective of our study was to examine the learning curve in 2 centres that initiated dedicated programs in AVr. We sought to evaluate the safety, efficacy, and
Received for publication April 7, 2015. Accepted May 15, 2015. Winner of the Society of Thoracic Surgery President’s Award. Presented at the Society of Thoracic Surgery 50th Annual Meeting, Orlando Florida (January 25-29, 2014). Corresponding author: Dr Munir Boodhwani, Division of Cardiac Surgery Room H3405, University of Ottawa Heart Institute, 40 Ruskin Street, Ottawa, Ontario K1Y 4W7, Canada. Tel.: þ1-613-761-4313; fax: þ1-613761-5107. E-mail: [email protected] See page 8 for disclosure information.
http://dx.doi.org/10.1016/j.cjca.2015.05.016 0828-282X/Ó 2015 Canadian Cardiovascular Society. Published by Elsevier Inc. All rights reserved.
2
Canadian Journal of Cardiology Volume - 2015
cohort was divided into 3 equal tertiles (T1, T2, and T3). Results: Early mortality was 1% in both cohorts. In cohort A, a total of 12 safety events occurred with a significant reduction in incidence over the tertiles (18%, 15%, and 3%, in T1, T2, and T3, respectively; P ¼ 0.05). In cohort B, 20 safety events occurred in 18 patients with a trend toward reduction of incidence over tertiles (20%, 12%, and 8%, in T1, T2, and T3, respectively; P ¼ 0.14). aortic crossclamp and cardiopulmonary bypass times decreased significantly after T2 in cohort A and T1 in cohort B (P < 0.01). Intraoperative procedural efficacy was similar across tertiles in both cohorts. Conclusions: Procedural safety and efficiency improves with experience whereas efficacy is consistent over time. AV repair is reproducible and appears to have a learning curve of approximately 40-60 cases.
inclurent une insuffisance aortique re siduelle paramètres d’efficacite nose. Les indices de complexite des cas inclurent l’existence ou une ste vère, d’une VA non trifolie e, et des d’une insuffisance aortique se dures cardiaques associe es. Chaque cohorte a e te divise e en proce gaux (T1, T2 et T3). trois terciles e sultats : La mortalite pre coce e tait 1 % dans les deux cohortes. Re ve nements touchant la sûrete ont Dans la cohorte A, un total de 12 e duction significative de l’incidence sur les terciles eu lieu avec une re (18 %, 15 % et 3 %, pour T1, T2, et T3, respectivement, P ¼ 0,05). ve nements lie s à la sûrete sont survenus chez Dans la cohorte B, 20 e duction de l’incidence à travers 18 patients avec une tendance à la re les terciles (20 %, 12 % et 8 %, en T1, T2 et T3, respectivement; P ¼ e du clampage aortique et le temps de circulation 0,14). La dure de façon significative après le T2 pour la extraccorporelle ont diminue de la cohorte A et le T1 pour la cohorte B (P < 0,01). L’efficacite dure per-ope ratoire e tait similaire entre les terciles pour les deux proce cohortes. curite et l’efficience des proce dures s’ame liorent Conclusions : La se rience alors que l’efficacite est constante dans le temps. La avec l’expe restauration de la VA est reproductible et semble correspondre à une courbe d’apprentissage d’environ 40-60 cas.
efficiency of AVr procedures during the learning curve and their effect on perioperative outcomes.
months postoperatively and annually thereafter, were used to assess postoperative AI, LV function, and peak and mean AV gradients. Information collected from operative notes included etiology and pathology of disease process, morphology of AV cusps, surgical techniques used for valve repair or for valve-sparing aortic root replacement (VSARR), number of attempts to repair the AV, duration of cardiopulmonary bypass (CPB) and aortic crossclamp (ACC) times, and any other interventions performed. Data on clinical presentation, reported as New York Heart Association and Canadian Cardiovascular Society classes, and postoperative course in hospital were obtained from the clinical history. In each cohort, patients were divided in chronological order into 3 equal tertiles (T1, T2, and T3, consecutively) and outcomes were compared across each tertile. Safety end points included in-hospital mortality, need for intraoperative reexploration of the AV, peri- or postoperative myocardial infarction, cerebrovascular accident, need for repeat surgery for bleeding, and requirement for permanent pacemaker insertion. Intraoperative re-exploration was performed if the initial repair attempt resulted in recurrent AI or structural defects. Efficiency end points included total duration of CPB and ACC times. Efficacy end points included presence of (1) at least mild AI or stenosis, either intraoperatively or at follow-up, as defined according to the American Heart Association/American College of Cardiology Task Force guidelines for AV disease (2014),4 and (2) mean intraoperative AV gradients > 20 mm Hg after repair. An AV gradient of 20 mm Hg was selected as an end point because it represents a significantly high gradient after AV replacement. Within each cohort, case complexity between tertiles was compared using a number of variables including the presence of severe preoperative AI, having a nontrileaflet AV, and need for other associated cardiac procedures.
Methods Patient population Patients were included from 2 different tertiary care hospitals with services specialized in cardiac surgery. A total of 100 patients from the Ottawa Heart Institute (cohort A, Ottawa, Canada) between November 2009 and January 2013 and 150 patients from St-Luc Hospital (cohort B, Brussels, Belgium) between December 1999 and November 2003 were included. These cohorts were comprised of consecutive, unselected groups of patients who underwent AV-sparing or valve repair surgery with or without ascending aortic disease, and represented the first AVr surgeries performed at each institution. Patients were not excluded for undergoing concomitant cardiac surgical procedures. Associated procedures included total or hemiarch replacement, coronary artery bypass grafting, atrial fibrillation ablation surgery, or mitral valve repair. Furthermore, patients were not excluded on the basis of age, aortic pathology, or clinical presentation. Institutional review board ethics approval was obtained to access data from patients enrolled in the AVr database. Methods Baseline demographic characteristics, clinical history, surgical details, hospital course, and preoperative, intraoperative, and postoperative echocardiograms were obtained from the electronic medical record retrospectively. Echocardiograms were reviewed and corroborated with surgical findings. Preoperative data included preoperative left ventricular (LV) function, AV dimensions, aortic dimensions, degree of AI, cusp anatomy, and etiology and type of pathology involved. Intraoperative echocardiograms were used to obtain degree of AI after repair, length of coaptation of cusps, and peak and mean AV gradients. Postoperative echocardiograms, usually performed at 3, 6, and 12
Surgical techniques All patients who underwent AVr or root replacement required median sternotomy for surgical access. Cannulation
Malas et al. Analysis of the Learning Curve for AV Repair
technique for CPB was completed either centrally through the ascending aorta or aortic arch, or peripherally through the femoral or axillary arteries. Standard 2-stage venous cannulation was used, unless alternate techniques were required for concomitant procedures. Cardioplegia was administered antegrade through the aortic root or coronary ostia or retrograde into the coronary sinus. After achievement of successful cardiac arrest, the aorta was transsected approximately 1 cm above the sinotubular junction to expose the AV. The anatomy of the functional aortic annulus, cusp tissue, level of coaptation, and aortic root diameters obtained from transesophageal echocardiograms were confirmed with direct surgical inspection. The choice of repair techniques was guided by our repair-oriented classification of AI.5 AVr techniques included but were not limited to cusp resection with primary reapproximation, free margin plication, free margin resuspension, or shaving of calcium deposits. Cusp repair techniques used have been described previously.6,7 Additional techniques for nontricuspid AVs included raphe resection, and bicuspidization or tricuspidization of leaflets.8 If pathology included the aortic root, a valve-sparing aortic root replacement (VSARR) was performed either using the reimplantation or remodelling techniques. In both centres, the reimplantation technique was preferentially used over remodelling for improved stabilization of the ventriculoaortic junction (VAJ).9,10 In cohort B, use of the remodelling technique progressively decreased over the duration of the study.11 Descriptions of both techniques have been previously described.10,12,13 Statistical analysis Baseline data are reported with means and standard deviations for continuous variables and numbers and percentages for categorical variables. Means of continuous variables were compared using a 2-tailed Student t test. Categorical variables were compared using the c2 or Fisher exact test as appropriate. Cumulative sum analysis (CUSUM) was performed for ACC and CPB times and graphs were generated to compare both cohorts.14 Acceptable and unacceptable failure rates were set at 10% and 20%, with reasonable probabilities of type I and II errors set at 10% each. Results Patient characteristics Baseline characteristics for both cohorts are shown in Table 1. A large proportion of patients in both cohorts were male (approximately 80%) with a mean age of 57 14 and 52 16 years in cohorts A and B, respectively. Only a minority of patients had LV dysfunction (LV ejection fraction < 50%) preoperatively, defined according to the 2014 American Heart Association/American College of Cardiology Valve Disease Guidelines.4 There was a larger overall proportion of patients with severe AI in cohort B compared with cohort A (46% vs 61.3%). Left ventricular end-systolic diameter and left ventricular end-diastolic diameter were also similar for cohorts A and B (37.6 8.4 vs 40.3 9.9 mm and 57.3 9.6 vs 60 10.6 mm, respectively).
3
Table 1. Baseline demographic characteristics for both study cohorts Characteristic Male sex Average age, years Preoperative LV dysfunction Normal Mild-moderate Severe Preoperative AI 0 to 2þ 3 to 4þ LVESD, mm LVEDD, mm Etiology Bicuspid Degenerative Aortic dissection Marfan syndrome Unicuspid Endocarditis Aortitis Quadricuspid Lupus Calcific Rheumatic
Cohort A (n ¼ 100)
Cohort B (n ¼ 150)
80 (80) 57.0 13.99
117 (78) 52.5 16
82 (82) 11 (11) 1 (1)
137 (91.3) 12 (8) 1 (0.7)
48 52 37.6 57.3 42 37 14 5 2 2 2 2 1 1 1
(48) (52) 8.4 9.6
58 (38.7) 92 (61.3) 40 10 60 11
(42) (37) (14) (5) (2) (2) (2) (2) (1) (1) (1)
57 76 1 10 0 6 0 1 0 0 2
(38) (50.7) (0.7) (6.7) (0) (4) (0) (0.7) (0) (0) (1.3)
Data are presented as n (%) except where otherwise stated. AI, aortic insufficiency; LV, left ventricular; LVEDD, LV end-diastolic diameter. LVESD, LV end-systolic diameter.
In evaluating complexity between tertiles, we found that the proportions of (1) patients with AI > 1þ, (2) patients with nontrileaflet AVs, and (3) patients who underwent a concomitant cardiac procedure were similar between tertiles in both cohorts (Table 2). Pathology and surgical data A listing of common etiologies for patients are shown in Table 1, with bicuspid AVs found in 42% vs 38% and tricuspid AVs in 56% vs 61.3% of patients in cohorts A and B, respectively. A summary of surgical procedures is shown in Table 3. VSARR was performed in 73% of patients in cohort A compared with 48% in cohort B, whereas subcommissural annuloplasty was performed in 23% vs 54.7%, respectively. Most VSARR procedures were performed using the reimplantation technique. Aortic cusp repair was completed in 73% and 58% of patients in cohorts A and B, respectively. Study end points Safety end points. The median clinical follow-up period was 18 and 91 months for cohorts A and B, respectively. Overall early mortality was 1% in both cohorts. In cohort A, a total of 12 safety events occurred across all tertiles; the number of safety events was significantly lower in T3 compared with in T1 (18% vs 3%; P < 0.05). In cohort B, a total of 20 safety end points occurred; a decreasing trend was observed across tertiles (Fig. 1; P ¼ 0.14). Efficiency end points. Overall mean ACC and CPB times in cohorts A vs B were 136.45 42.9 vs 88.4 36.1 and 185.6 54.2 vs 114.1 47.5 minutes, respectively. ACC and CPB
4
Canadian Journal of Cardiology Volume - 2015
Table 2. Study end points for each tertile in both study cohorts Tertile Cohort A Safety end points, n (%) In-hospital mortality Myocardial infarction Reopening for bleeding Permanent pacemaker insertion Intraoperative AV re-exploration Cerebrovascular accident Efficiency end points, minutes Aortic crossclamp time Cardiopulmonary bypass time Efficacy end points, n (%) Follow-up AI > 2þ Mean AV gradient > 20 Case complexity, n (%) Preoperative AI > 1þ Nontrileaflet AV Associated procedures Cohort B Safety end points, n (%) In-hospital mortality Myocardial infarction Reopening for bleeding Permanent pacemaker insertion Intraoperative AV re-exploration Cerebrovascular accident Efficiency end points, minutes Aortic crossclamp time Cardiopulmonary bypass time Efficacy end points Follow-up AI > 2þ Mean AV gradient > 20 Case complexity Preoperative AI > 1þ Nontrileaflet AV Associated procedures
1
2
3
6 (18) 0 0 3 3 0 0
5 (15) 1 0 2 1 0 1
1 (3) 0 0 1 0 0 0
151.79 40.70 205 55.89
135.97 39.55 186.09 53.05
122.03 44.12 165.56 47.20
2 (6) 1 (6)
0 (0) 0 (0)
0 (0) 0 (0)
24 (72) 16 (48) 10 (30)
19 (57) 14 (42) 18 (54)
16 (47) 14 (41) 13 (38)
10 (20) 1 0 5 0 4 0
6 (12) 0 0 3 3 0 0
4 (8) 0 0 2 1 1 0
100.71 39.17 132.02 56.91
81.72 34.26 104.94 43.42
83.12 32.25 105.76 35.86
5 (10) 0 (0)
10 (20) 2 (4)
4 (8) 4 (8)
41 (82) 16 (32) 15 (30)
48 (96) 23 (46) 13 (26)
46 (92) 19 (38) 20 (40)
Data are presented as n except where otherwise stated. AI, aortic insufficiency; AV, aortic valve.
times significantly decreased after T2 in cohort A (P ¼ 0.005; P ¼ 0.002) and T1 in cohort B (P ¼ 0.01; P ¼ 0.009). To compare patients with similar operative procedures, patients in each cohort were divided into 2 groupsdthose who underwent aortic root interventions vs those who did not and ACC and CPB times were compared across the same tertiles. Although this is a crude method to render the procedures comparable because of the wide range of concomitant procedures involved (such as coronary artery bypass graft, mitral valve interventions, etc), we found that in both Table 3. Surgical data for both cohorts Cohort A (n ¼ 100) Valve-sparing aortic root replacement Reimplantation Remodelling Subcommissural annuloplasty Aortic cusp repair Free margin plication Free margin resuspension Triangular resection Patch repair Data are presented as n (%).
73 71 2 23 73 61 17 16 6
(73) (71) (2) (23) (73) (61) (17) (16) (6)
Cohort B (n ¼ 150) 72 32 40 82 87 15 43 4 13
(48) (21.3) (26.6) (54.7) (58) (10) (28.7) (2.6) (8.7)
cohorts there was a decrease in ACC and CPB times across tertiles (Table 4). In cohort A, the decrease in ACC and CPB times from T1 to T3 for aortic root interventions had P values of 0.07 and 0.003, whereas for nonaortic root interventions the decrease in ACC and CPB had P values of 0.02 and 0.09, respectively. In cohort B, the decrease in ACC and CPB times from T1 to T2 had P values of 0.02 and 0.09, whereas for nonaortic root interventions the decrease in ACC and CPB from T1 to T3 had P values of 0.55 and 0.59, respectively. CUSUM analysis curves were generated for ACC and CPB times for both cohorts (Figs. 2 and 3). Patients were subgrouped into those who underwent only AVr and those who underwent VSARR with or without AVr. CUSUM analysis showed that, in both cohorts, neither CPB nor ACC times crossed the upper boundary limit into the unacceptable failure rate across all tertiles. Efficacy end points. Follow-up time was substantially longer in cohort B vs cohort A (91 months vs 18 months). In cohort A, only 3 patients had either AI > 2þ or a mean AV gradient > 20 mm Hg on follow-up. In cohort B, this occurred evenly across tertiles without any noticeable trends.
Malas et al. Analysis of the Learning Curve for AV Repair
5
Figure 1. Top to bottom: safety, efficiency, and efficacy end points for both study cohorts. ACC, aortic crossclamp; AI, aortic insufficiency; CPB, cardiopulmonary bypass; Echo, echocardiographic; grad, gradient; Intraop, intraoperative; T, tertile.
We sought to assess the learning curve by evaluating 3 different parameters, namely, safety, efficiency, and efficacy. We observed that the learning curve was characterized by a significant decrease in safety end points after T2 in cohort A (66 patients) and T1 in cohort B (50 patients). Of note, most safety events centred around the need for pacemaker insertion and reopening for bleeding rather than myocardial infarction, cerebrovascular accident, or mortality. Although the number of permanent pacemaker insertions was low, this complication is expected because of the proximity of the conduction system to local manipulation at the repair boundaries. In cohort B, we identified a decrease in AV re-exploration rates after the first tertile. Thus, results of this analysis suggest that it takes
Discussion Because AVr is performed in a limited number of centres, its reproducibility has been questioned. To address this issue, we examined the learning curve in 2 different centres that initiated dedicated programs in AV preservation and repair surgery. We evaluated safety, efficacy, and efficiency end points over time. Our analysis demonstrated that AVr is, in fact, reproducible, showing similar rates of efficacy across both cohorts. Safety and efficiency parameters appear to improve with an increasing number of cases with an estimated learning curve of approximately 40-60 cases. This information might have implications for the management of case volumes and need for mentorship for centres that initiate programs in AVr.
Table 4. Comparison of ACC and CPB times (minutes) grouped according to nature of intervention Aortic root intervention Cohort A ACC CPB Cohort B ACC CPB
Nonaortic root intervention
T1
T2
T3
T1
T2
T3
160.3 39.0 221.9 49.9
143.3 37.9 192.5 55.0
140.9 36.4 180.6 42.3
120.3 32.0 145.3 29.3
116.4 39.2 169 45.7
82.5 31.1 134.1 42.4
114 36.4 148.1 56.8
95.3 23.2 121.7 35.3
100.6 27.7 119 31.1
67.5 23.3 91.8 32.6
72.7 37.6 93.7 45.2
72.4 30.4 97.6 36.6
ACC, aortic crossclamp; CPB, cardiopulmonary bypass; T, tertile.
6
Canadian Journal of Cardiology Volume - 2015
Figure 2. Cumulative sum analysis charts for aortic crossclamp (ACC) and (CPB) times for cohort A. Patients are subgrouped into those who had valve-sparing aortic root replacement (VSARR) with or without (þ/) aortic valve repair (AVr) and AVr only.
Figure 3. Cumulative sum analysis charts for aortic crossclamp (ACC) and (CPB) times for cohort B. Patients are subgrouped into those who had valve-sparing aortic root replacement (VSARR) with or without (þ/) aortic valve repair (AVr) and AVr only.
Malas et al. Analysis of the Learning Curve for AV Repair
somewhere around 40-60 cases to reduce complications associated with AVr to a low, stable rate. In a similar fashion, we observed a statistically significant improvement in efficiency end points with CPB and ACC times after T2 in cohort A (66 patients) and T1 in cohort B (50 patients). To assess progression of the efficiency parameters with time, we used a CUSUM analysis. This method has been previously used in studies that evaluated the need for proctoring in the start of a transcatheter aortic valve implantation program,15 and to assess the learning curve for off-pump coronary artery bypass grafting and outcomes.16,17 For example, CUSUM charts have been used as performance monitoring tools to establish that off-pump coronary artery bypass grafting incorporation into residency training is safe.18 In our study, CUSUM analysis for ACC and CPB (Figs. 2 and 3) indicate no excessive failures in patients with VSARR and/or AVr. Based on acceptable type I and type II statistical errors, we defined 2 boundary lines, H0 and H1, which define acceptable and unacceptable failure rates in ACC or CPB, or that for which performance is considered probably acceptable (H0) or unacceptable (H1). Thus, when we aligned consecutive procedure numbers in a chronological fashion along the x-axis against the cumulative sum on the y-axis, we could assess performance of each procedure compared with the acceptable boundary lines. Any transgression beyond H1 indicates an unacceptably high ACC or CPB time compared with other values. Because no lines crossed the upper decision limits in all graphs, no statistical inference can be made about failures in the learning curve. In other words, none of the data points indicate unacceptably high CPB or ACC values in the learning curve. As for efficacy end points, although we could not make any inferences on efficacy of repair leading to recurrent AI or AV gradients, follow-up in both cohorts was limited in duration; future data might potentially elucidate differences in tertiles. However, repair rates in both cohorts have been good with only a small proportion of patients having significant recurrent AI or increased valve gradient values. It is important to note the differences in the 2 cohorts. The first important difference is the time period over which the procedures were performed. Cohort B preceded cohort A by approximately a decade and this experience was accrued when AVr was in its infancy as a discipline. Many important lessons have since been learned. An important implication of this is the longer follow-up time available for cohort B. Second, related to the differences outlined herein, there are some differences in surgical strategy between cohorts. A larger proportion of patients underwent VSARR in cohort A compared with cohort B (73% vs 48%), with more patients who underwent subcommissural annuloplasty in cohort B. This represents a shift in the treatment paradigm in stabilization of the AV annulus that has occurred over time. Although dilatation of the sinotubular junction is typically repaired with a supracoronary aortic replacement, the optimal approach to remodelling and stabilization of the VAJ continues to be debated.19 In patients who present with a dilated VAJ, but a normal or mildly enlarged aortic root, a more aggressive approach using VSARR with the reimplantation technique has increasingly been favoured over subcommissural annuloplasty.11,20,21 This is particularly true in patients with bicuspid AVs, dilated functional annulus, connective tissue disease, or in young patients.
7
Another important observation noted between cohorts is the larger proportion of patients who underwent any type of cusp repair in cohort A (73% vs 58%). With increasing experience with AVr, there has been greater appreciation of cusp pathology that might be either the primary cause of AI, might be unmasked after annular and root interventions, or might be induced because of technical errors. Correction of this cusp pathology is critical to a durable AVr and we advocate an aggressive approach to the diagnosis and treatment of associated cusp disease. This analysis provides certain insights that might be useful to surgeons and centres that initiate programs in AV preservation and repair surgery. First, in both centres, the approach used has been a programmatic one, empowering teams led by surgeons that include cardiologists, anaesthesiologists, and echocardiographers, and allied health professionals. Second, for surgeons who initiate new programs, having adequate numbers of eligible patients can be an important challenge. If the case volume is 5-10 cases per year, then it can take 5 or more years to progress through the 40-60 cases required for the learning curve. In these contexts, there is a balance between waiting for the ideal patient to perform an AVr to using a more inclusive strategy for patient selection. We would favour the latter. It is also useful within a program to triage all potentially eligible cases to a single surgeon, when case volume is expected to be a limiting factor. Last, the role of mentorship in safely and successfully traversing the learning curve cannot be overemphasized. Like the adoption of any new cardiac surgical procedure, this mentorship can take the form of informal case discussions, review of echocardiograms and other imaging information, formal proctoring of cases, and importantly, discussion of failures. Seeking and developing a strong mentorship relationship can be critical to the success of a program. Limitations Limitations of this study include the modest sample size used in both cohorts and a smaller duration of follow-up for cohort A. This reflects the difference in time of inception of the AV programs at each institution. Furthermore, the patients in both cohorts represent a heterogeneous population with a wide mix of pathologies, posing challenges in comparison of techniques across tertiles and between cohorts. The notion that AVr comprises a number of surgical techniques, rather than being a single entity, further adds to this challenge. Last, it is important to note that improvement in outcomes might also be influenced by an improvement in patient selection and optimization of postoperative care.22 Conclusions In 2 independent centres that initiated dedicated programs in AVr, we found that procedural safety and efficiency improved with experience and efficacy was consistent over time. AVr is safe and reproducible and appears to have a learning curve of approximately 40-60 cases. Acknowledgements The authors thank Kathryn McLean, Sophia Chaudry, and Karen Holmes, University of Ottawa Heart Institute, for their assistance in maintaining the patient database.
8
Disclosures The authors have no conflicts of interest to disclose. References 1. Price J, De Kerchove L, Glineur D, et al. Risk of valve-related events after aortic valve repair. Ann Thorac Surg 2013;95:606-13. 2. Aicher D, Fries R, Rodionycheva S, et al. Aortic valve repair leads to a low incidence of valve-related complications. Eur J Cardiothorac Surg 2010;37:127-32. 3. Saczkowski R, Malas T, de Kerchove L, El Khoury G, Boodhwani M. Systematic review of aortic valve preservation and repair. Ann Cardiothorac Surg 2013;2:3-9. 4. Nishimura RA, Otto CM, Bonow RO, et al. 2014 AHA/ACC guideline for the management of patients with valvular heart disease: a report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines. J Am Coll Cardiol 2014;63:e57-185. 5. Boodhwani M, de Kerchove L, Glineur D, et al. Repair-oriented classification of aortic insufficiency: impact on surgical techniques and clinical outcomes. J Thorac Cardiovasc Surg 2009;137:286-94. 6. Langer F, Aicher D, Kissinger A, et al. Aortic valve repair using a differentiated surgical strategy. Circulation 2004;110(11 suppl 1): II-67-73.
Canadian Journal of Cardiology Volume - 2015 11. de Kerchove L, Boodhwani M, Glineur D, et al. Valve sparing-root replacement with the reimplantation technique to increase the durability of bicuspid aortic valve repair. J Thorac Cardiovasc Surg 2011;142: 1430-8. 12. Boodhwani M, de Kerchove L, El Khoury G. Aortic root replacement using the reimplantation technique: tips and tricks. Interact Cardiovasc Thorac Surg 2009;8:584-6. 13. Schäfers HJ, Aicher D. Root remodeling for aortic root dilatation. Ann Cardiothorac Surg 2013;2:113-6. 14. Kestin IG. A statistical approach to measuring the competence of anaesthetic trainees at practical procedures. Br J Anaesth 1995;75:805-9. 15. Glauber M, Murzi M, Cerillo AG. Is proctoring mandatory when starting a TAVI program? Ann Cardiothorac Surg 2012;1:190-3. 16. Novick RJ, Fox SA, Stitt LW, et al. Effect of off-pump coronary artery bypass grafting on risk-adjusted and cumulative sum failure outcomes after coronary artery surgery. J Cardiovasc Surg 2002;17:520-8. 17. Caputo M, Reeves BC, Rogers CA, Ascione R, Angelini GD. Monitoring the performance of residents during training in off-pump coronary surgery. J Thorac Cardiovasc Surg 2004;128:907-15. 18. Murphy GJ, Rogers CA, Caputo M, Angelini GD. Acquiring proficiency in off-pump surgery: traversing the learning curve, reproducibility, and quality control. Ann Thorac Surg 2005;80:1965-70.
7. Aicher D, Schäfers HJ. Aortic valve repairdcurrent status, indications, and outcomes. Semin Thorac Cardiovasc Surg 2012;24:195-201.
19. de Kerchove L, El Khoury G. Anatomy and pathophysiology of the ventriculo-aortic junction: implication in aortic valve repair surgery. Ann Cardiothorac Surg 2013;2:57-64.
8. Aicher D, Schäfers HJ. Bicuspidization of the regurgitant unicuspid aortic valve. Multimed Man Cardiothorac Surg 2010;2010. mmcts.2009. 004069.
20. David TE. Aortic valve sparing operations: outcomes at 20 years. Ann Cardiothorac Surg 2013;2:24-9.
9. Erasmi AW, Sievers HH, Bechtel JF, et al. Remodeling or reimplantation for valve-sparing aortic root surgery? Ann Thorac Surg 2007;83:S752-6.
21. Vallabhajosyula P, Komlo C, Szeto WY, et al. Root stabilization of the repaired bicuspid aortic valve: subcommissural annuloplasty versus root reimplantation. Ann Thorac Surg 2014;97:1227-34.
10. Boodhwani M, El Khoury G. Aortic valve repair. Oper Tech Thorac Cardiovasc Surg 2009;14:266-80.
22. Colquhoun PH. CUSUM analysis of J-pouch surgery reflects no learning curve after board certification. Can J Surg 2008;51:296-9.
-
(2015) 1e8
Clinical Research
Is Aortic Valve Repair Reproducible? Analysis of the Learning Curve for Aortic Valve Repair Tarek Malas, MDCM, MPH,a Richard Saczkowski, CPC, CCP,b Benjamin Sohmer, MD,c Marc Ruel, MD, MPH,a Thierry Mesana, MD, PhD,a Laurent de Kerchove, MD,d Gebrine El Khoury, MD, PhD,d and Munir Boodhwani, MD, MSCa a b c
Division of Cardiac Surgery, University of Ottawa Heart Institute, Ottawa, Ontario, Canada
Division of Cardiac Perfusion, University of Ottawa Heart Institute, Ottawa, Ontario, Canada
Division of Cardiac Anesthesiology, University of Ottawa Heart Institute, Ottawa, Ontario, Canada d
Department of Cardiovascular Surgery, St-Luc Hospital, Brussels, Belgium
ABSTRACT
RESUM E
Background: Aortic valve (AV) preservation and repair, although effective, is performed in a limited number of centres. Lack of wider application might be due to challenges in dissemination of tacit surgical knowledge. We examined the learning curve in 2 centres that initiated dedicated programs in AV repair. Methods: Prospectively collected data on the first 100 (cohort A) and 150 consecutive patients (cohort B) who underwent AV repair surgery were analyzed. Safety end points included mortality, myocardial infarction or stroke, early AV repeat surgery, re-exploration for bleeding, or pacemaker implantation. Efficiency was assessed according to aortic crossclamp and cardiopulmonary bypass times. Efficacy parameters included residual aortic insufficiency or stenosis. Indices of case complexity included presence of severe aortic insufficiency, nontrileaflet AV, and associated cardiac procedures. Each
servation et la re paration de la valve aortique Introduction : La pre alise e dans un nombre limite de centres. (VA), bien qu’efficace, est re fis en Le manque d’application plus large pourrait être dû à des de matière de diffusion des connaissances tacites en chirurgie. Nous la courbe d’apprentissage dans les deux centres qui avons examine des programmes de die s à la re paration de la VA. ont initie thodes : Les donne es recueillies de façon prospective sur les 100 Me paration de VA premiers patients qui ont subi une chirurgie de re te analyse es. Les (cohorte A) et les 150 suivants (cohorte B) ont e curite incluent la mortalite , l’infarctus du myocarde points limites de se re bral, la reprise pre coce d’une chirurgie de ou l’accident vasculaire ce -examen pour saignement, ou l’implantation d’un stimla VA, le re te e value e en fonction du temps de ulateur cardiaque. L’efficience a e e du pontage cardiopulmonaire. Les clampage de l’aorte et la dure
Aortic valve (AV) repair has emerged as a feasible alternative to AV replacement in selected patients with aortic insufficiency (AI). Valve replacement, although effective at treatment of the primary disease, is associated with multiple adverse events including structural valve deterioration, endocarditis, thromboembolism, and the risks and inconveniences associated with lifelong anticoagulation, in the case of mechanical valves. In
contrast, AV repair (AVr) has been demonstrated in multiple cohort studies to have a low risk of valve-related events.1-3 However, AVr is only practiced in a limited number of centres. There might be a number of reasons for this. First, AVr requires understanding and use of a variety of surgical techniques that might include intervention on the valve cusps, annulus, and aortic root, and might be applied to bicuspid and tricuspid AVs. Second, successful repair requires a thorough understanding of functional AV anatomy, mechanisms of AI, and their echocardiographic and intraoperative assessment. Last, AI is a less commonly encountered pathology and the small volume of cases might contribute to slower dissemination of surgical expertise. Because of the limited dissemination of AVr, the reproducibility of the surgical techniques has been questioned. Therefore, the objective of our study was to examine the learning curve in 2 centres that initiated dedicated programs in AVr. We sought to evaluate the safety, efficacy, and
Received for publication April 7, 2015. Accepted May 15, 2015. Winner of the Society of Thoracic Surgery President’s Award. Presented at the Society of Thoracic Surgery 50th Annual Meeting, Orlando Florida (January 25-29, 2014). Corresponding author: Dr Munir Boodhwani, Division of Cardiac Surgery Room H3405, University of Ottawa Heart Institute, 40 Ruskin Street, Ottawa, Ontario K1Y 4W7, Canada. Tel.: þ1-613-761-4313; fax: þ1-613761-5107. E-mail: [email protected] See page 8 for disclosure information.
http://dx.doi.org/10.1016/j.cjca.2015.05.016 0828-282X/Ó 2015 Canadian Cardiovascular Society. Published by Elsevier Inc. All rights reserved.
2
Canadian Journal of Cardiology Volume - 2015
cohort was divided into 3 equal tertiles (T1, T2, and T3). Results: Early mortality was 1% in both cohorts. In cohort A, a total of 12 safety events occurred with a significant reduction in incidence over the tertiles (18%, 15%, and 3%, in T1, T2, and T3, respectively; P ¼ 0.05). In cohort B, 20 safety events occurred in 18 patients with a trend toward reduction of incidence over tertiles (20%, 12%, and 8%, in T1, T2, and T3, respectively; P ¼ 0.14). aortic crossclamp and cardiopulmonary bypass times decreased significantly after T2 in cohort A and T1 in cohort B (P < 0.01). Intraoperative procedural efficacy was similar across tertiles in both cohorts. Conclusions: Procedural safety and efficiency improves with experience whereas efficacy is consistent over time. AV repair is reproducible and appears to have a learning curve of approximately 40-60 cases.
inclurent une insuffisance aortique re siduelle paramètres d’efficacite nose. Les indices de complexite des cas inclurent l’existence ou une ste vère, d’une VA non trifolie e, et des d’une insuffisance aortique se dures cardiaques associe es. Chaque cohorte a e te divise e en proce gaux (T1, T2 et T3). trois terciles e sultats : La mortalite pre coce e tait 1 % dans les deux cohortes. Re ve nements touchant la sûrete ont Dans la cohorte A, un total de 12 e duction significative de l’incidence sur les terciles eu lieu avec une re (18 %, 15 % et 3 %, pour T1, T2, et T3, respectivement, P ¼ 0,05). ve nements lie s à la sûrete sont survenus chez Dans la cohorte B, 20 e duction de l’incidence à travers 18 patients avec une tendance à la re les terciles (20 %, 12 % et 8 %, en T1, T2 et T3, respectivement; P ¼ e du clampage aortique et le temps de circulation 0,14). La dure de façon significative après le T2 pour la extraccorporelle ont diminue de la cohorte A et le T1 pour la cohorte B (P < 0,01). L’efficacite dure per-ope ratoire e tait similaire entre les terciles pour les deux proce cohortes. curite et l’efficience des proce dures s’ame liorent Conclusions : La se rience alors que l’efficacite est constante dans le temps. La avec l’expe restauration de la VA est reproductible et semble correspondre à une courbe d’apprentissage d’environ 40-60 cas.
efficiency of AVr procedures during the learning curve and their effect on perioperative outcomes.
months postoperatively and annually thereafter, were used to assess postoperative AI, LV function, and peak and mean AV gradients. Information collected from operative notes included etiology and pathology of disease process, morphology of AV cusps, surgical techniques used for valve repair or for valve-sparing aortic root replacement (VSARR), number of attempts to repair the AV, duration of cardiopulmonary bypass (CPB) and aortic crossclamp (ACC) times, and any other interventions performed. Data on clinical presentation, reported as New York Heart Association and Canadian Cardiovascular Society classes, and postoperative course in hospital were obtained from the clinical history. In each cohort, patients were divided in chronological order into 3 equal tertiles (T1, T2, and T3, consecutively) and outcomes were compared across each tertile. Safety end points included in-hospital mortality, need for intraoperative reexploration of the AV, peri- or postoperative myocardial infarction, cerebrovascular accident, need for repeat surgery for bleeding, and requirement for permanent pacemaker insertion. Intraoperative re-exploration was performed if the initial repair attempt resulted in recurrent AI or structural defects. Efficiency end points included total duration of CPB and ACC times. Efficacy end points included presence of (1) at least mild AI or stenosis, either intraoperatively or at follow-up, as defined according to the American Heart Association/American College of Cardiology Task Force guidelines for AV disease (2014),4 and (2) mean intraoperative AV gradients > 20 mm Hg after repair. An AV gradient of 20 mm Hg was selected as an end point because it represents a significantly high gradient after AV replacement. Within each cohort, case complexity between tertiles was compared using a number of variables including the presence of severe preoperative AI, having a nontrileaflet AV, and need for other associated cardiac procedures.
Methods Patient population Patients were included from 2 different tertiary care hospitals with services specialized in cardiac surgery. A total of 100 patients from the Ottawa Heart Institute (cohort A, Ottawa, Canada) between November 2009 and January 2013 and 150 patients from St-Luc Hospital (cohort B, Brussels, Belgium) between December 1999 and November 2003 were included. These cohorts were comprised of consecutive, unselected groups of patients who underwent AV-sparing or valve repair surgery with or without ascending aortic disease, and represented the first AVr surgeries performed at each institution. Patients were not excluded for undergoing concomitant cardiac surgical procedures. Associated procedures included total or hemiarch replacement, coronary artery bypass grafting, atrial fibrillation ablation surgery, or mitral valve repair. Furthermore, patients were not excluded on the basis of age, aortic pathology, or clinical presentation. Institutional review board ethics approval was obtained to access data from patients enrolled in the AVr database. Methods Baseline demographic characteristics, clinical history, surgical details, hospital course, and preoperative, intraoperative, and postoperative echocardiograms were obtained from the electronic medical record retrospectively. Echocardiograms were reviewed and corroborated with surgical findings. Preoperative data included preoperative left ventricular (LV) function, AV dimensions, aortic dimensions, degree of AI, cusp anatomy, and etiology and type of pathology involved. Intraoperative echocardiograms were used to obtain degree of AI after repair, length of coaptation of cusps, and peak and mean AV gradients. Postoperative echocardiograms, usually performed at 3, 6, and 12
Surgical techniques All patients who underwent AVr or root replacement required median sternotomy for surgical access. Cannulation
Malas et al. Analysis of the Learning Curve for AV Repair
technique for CPB was completed either centrally through the ascending aorta or aortic arch, or peripherally through the femoral or axillary arteries. Standard 2-stage venous cannulation was used, unless alternate techniques were required for concomitant procedures. Cardioplegia was administered antegrade through the aortic root or coronary ostia or retrograde into the coronary sinus. After achievement of successful cardiac arrest, the aorta was transsected approximately 1 cm above the sinotubular junction to expose the AV. The anatomy of the functional aortic annulus, cusp tissue, level of coaptation, and aortic root diameters obtained from transesophageal echocardiograms were confirmed with direct surgical inspection. The choice of repair techniques was guided by our repair-oriented classification of AI.5 AVr techniques included but were not limited to cusp resection with primary reapproximation, free margin plication, free margin resuspension, or shaving of calcium deposits. Cusp repair techniques used have been described previously.6,7 Additional techniques for nontricuspid AVs included raphe resection, and bicuspidization or tricuspidization of leaflets.8 If pathology included the aortic root, a valve-sparing aortic root replacement (VSARR) was performed either using the reimplantation or remodelling techniques. In both centres, the reimplantation technique was preferentially used over remodelling for improved stabilization of the ventriculoaortic junction (VAJ).9,10 In cohort B, use of the remodelling technique progressively decreased over the duration of the study.11 Descriptions of both techniques have been previously described.10,12,13 Statistical analysis Baseline data are reported with means and standard deviations for continuous variables and numbers and percentages for categorical variables. Means of continuous variables were compared using a 2-tailed Student t test. Categorical variables were compared using the c2 or Fisher exact test as appropriate. Cumulative sum analysis (CUSUM) was performed for ACC and CPB times and graphs were generated to compare both cohorts.14 Acceptable and unacceptable failure rates were set at 10% and 20%, with reasonable probabilities of type I and II errors set at 10% each. Results Patient characteristics Baseline characteristics for both cohorts are shown in Table 1. A large proportion of patients in both cohorts were male (approximately 80%) with a mean age of 57 14 and 52 16 years in cohorts A and B, respectively. Only a minority of patients had LV dysfunction (LV ejection fraction < 50%) preoperatively, defined according to the 2014 American Heart Association/American College of Cardiology Valve Disease Guidelines.4 There was a larger overall proportion of patients with severe AI in cohort B compared with cohort A (46% vs 61.3%). Left ventricular end-systolic diameter and left ventricular end-diastolic diameter were also similar for cohorts A and B (37.6 8.4 vs 40.3 9.9 mm and 57.3 9.6 vs 60 10.6 mm, respectively).
3
Table 1. Baseline demographic characteristics for both study cohorts Characteristic Male sex Average age, years Preoperative LV dysfunction Normal Mild-moderate Severe Preoperative AI 0 to 2þ 3 to 4þ LVESD, mm LVEDD, mm Etiology Bicuspid Degenerative Aortic dissection Marfan syndrome Unicuspid Endocarditis Aortitis Quadricuspid Lupus Calcific Rheumatic
Cohort A (n ¼ 100)
Cohort B (n ¼ 150)
80 (80) 57.0 13.99
117 (78) 52.5 16
82 (82) 11 (11) 1 (1)
137 (91.3) 12 (8) 1 (0.7)
48 52 37.6 57.3 42 37 14 5 2 2 2 2 1 1 1
(48) (52) 8.4 9.6
58 (38.7) 92 (61.3) 40 10 60 11
(42) (37) (14) (5) (2) (2) (2) (2) (1) (1) (1)
57 76 1 10 0 6 0 1 0 0 2
(38) (50.7) (0.7) (6.7) (0) (4) (0) (0.7) (0) (0) (1.3)
Data are presented as n (%) except where otherwise stated. AI, aortic insufficiency; LV, left ventricular; LVEDD, LV end-diastolic diameter. LVESD, LV end-systolic diameter.
In evaluating complexity between tertiles, we found that the proportions of (1) patients with AI > 1þ, (2) patients with nontrileaflet AVs, and (3) patients who underwent a concomitant cardiac procedure were similar between tertiles in both cohorts (Table 2). Pathology and surgical data A listing of common etiologies for patients are shown in Table 1, with bicuspid AVs found in 42% vs 38% and tricuspid AVs in 56% vs 61.3% of patients in cohorts A and B, respectively. A summary of surgical procedures is shown in Table 3. VSARR was performed in 73% of patients in cohort A compared with 48% in cohort B, whereas subcommissural annuloplasty was performed in 23% vs 54.7%, respectively. Most VSARR procedures were performed using the reimplantation technique. Aortic cusp repair was completed in 73% and 58% of patients in cohorts A and B, respectively. Study end points Safety end points. The median clinical follow-up period was 18 and 91 months for cohorts A and B, respectively. Overall early mortality was 1% in both cohorts. In cohort A, a total of 12 safety events occurred across all tertiles; the number of safety events was significantly lower in T3 compared with in T1 (18% vs 3%; P < 0.05). In cohort B, a total of 20 safety end points occurred; a decreasing trend was observed across tertiles (Fig. 1; P ¼ 0.14). Efficiency end points. Overall mean ACC and CPB times in cohorts A vs B were 136.45 42.9 vs 88.4 36.1 and 185.6 54.2 vs 114.1 47.5 minutes, respectively. ACC and CPB
4
Canadian Journal of Cardiology Volume - 2015
Table 2. Study end points for each tertile in both study cohorts Tertile Cohort A Safety end points, n (%) In-hospital mortality Myocardial infarction Reopening for bleeding Permanent pacemaker insertion Intraoperative AV re-exploration Cerebrovascular accident Efficiency end points, minutes Aortic crossclamp time Cardiopulmonary bypass time Efficacy end points, n (%) Follow-up AI > 2þ Mean AV gradient > 20 Case complexity, n (%) Preoperative AI > 1þ Nontrileaflet AV Associated procedures Cohort B Safety end points, n (%) In-hospital mortality Myocardial infarction Reopening for bleeding Permanent pacemaker insertion Intraoperative AV re-exploration Cerebrovascular accident Efficiency end points, minutes Aortic crossclamp time Cardiopulmonary bypass time Efficacy end points Follow-up AI > 2þ Mean AV gradient > 20 Case complexity Preoperative AI > 1þ Nontrileaflet AV Associated procedures
1
2
3
6 (18) 0 0 3 3 0 0
5 (15) 1 0 2 1 0 1
1 (3) 0 0 1 0 0 0
151.79 40.70 205 55.89
135.97 39.55 186.09 53.05
122.03 44.12 165.56 47.20
2 (6) 1 (6)
0 (0) 0 (0)
0 (0) 0 (0)
24 (72) 16 (48) 10 (30)
19 (57) 14 (42) 18 (54)
16 (47) 14 (41) 13 (38)
10 (20) 1 0 5 0 4 0
6 (12) 0 0 3 3 0 0
4 (8) 0 0 2 1 1 0
100.71 39.17 132.02 56.91
81.72 34.26 104.94 43.42
83.12 32.25 105.76 35.86
5 (10) 0 (0)
10 (20) 2 (4)
4 (8) 4 (8)
41 (82) 16 (32) 15 (30)
48 (96) 23 (46) 13 (26)
46 (92) 19 (38) 20 (40)
Data are presented as n except where otherwise stated. AI, aortic insufficiency; AV, aortic valve.
times significantly decreased after T2 in cohort A (P ¼ 0.005; P ¼ 0.002) and T1 in cohort B (P ¼ 0.01; P ¼ 0.009). To compare patients with similar operative procedures, patients in each cohort were divided into 2 groupsdthose who underwent aortic root interventions vs those who did not and ACC and CPB times were compared across the same tertiles. Although this is a crude method to render the procedures comparable because of the wide range of concomitant procedures involved (such as coronary artery bypass graft, mitral valve interventions, etc), we found that in both Table 3. Surgical data for both cohorts Cohort A (n ¼ 100) Valve-sparing aortic root replacement Reimplantation Remodelling Subcommissural annuloplasty Aortic cusp repair Free margin plication Free margin resuspension Triangular resection Patch repair Data are presented as n (%).
73 71 2 23 73 61 17 16 6
(73) (71) (2) (23) (73) (61) (17) (16) (6)
Cohort B (n ¼ 150) 72 32 40 82 87 15 43 4 13
(48) (21.3) (26.6) (54.7) (58) (10) (28.7) (2.6) (8.7)
cohorts there was a decrease in ACC and CPB times across tertiles (Table 4). In cohort A, the decrease in ACC and CPB times from T1 to T3 for aortic root interventions had P values of 0.07 and 0.003, whereas for nonaortic root interventions the decrease in ACC and CPB had P values of 0.02 and 0.09, respectively. In cohort B, the decrease in ACC and CPB times from T1 to T2 had P values of 0.02 and 0.09, whereas for nonaortic root interventions the decrease in ACC and CPB from T1 to T3 had P values of 0.55 and 0.59, respectively. CUSUM analysis curves were generated for ACC and CPB times for both cohorts (Figs. 2 and 3). Patients were subgrouped into those who underwent only AVr and those who underwent VSARR with or without AVr. CUSUM analysis showed that, in both cohorts, neither CPB nor ACC times crossed the upper boundary limit into the unacceptable failure rate across all tertiles. Efficacy end points. Follow-up time was substantially longer in cohort B vs cohort A (91 months vs 18 months). In cohort A, only 3 patients had either AI > 2þ or a mean AV gradient > 20 mm Hg on follow-up. In cohort B, this occurred evenly across tertiles without any noticeable trends.
Malas et al. Analysis of the Learning Curve for AV Repair
5
Figure 1. Top to bottom: safety, efficiency, and efficacy end points for both study cohorts. ACC, aortic crossclamp; AI, aortic insufficiency; CPB, cardiopulmonary bypass; Echo, echocardiographic; grad, gradient; Intraop, intraoperative; T, tertile.
We sought to assess the learning curve by evaluating 3 different parameters, namely, safety, efficiency, and efficacy. We observed that the learning curve was characterized by a significant decrease in safety end points after T2 in cohort A (66 patients) and T1 in cohort B (50 patients). Of note, most safety events centred around the need for pacemaker insertion and reopening for bleeding rather than myocardial infarction, cerebrovascular accident, or mortality. Although the number of permanent pacemaker insertions was low, this complication is expected because of the proximity of the conduction system to local manipulation at the repair boundaries. In cohort B, we identified a decrease in AV re-exploration rates after the first tertile. Thus, results of this analysis suggest that it takes
Discussion Because AVr is performed in a limited number of centres, its reproducibility has been questioned. To address this issue, we examined the learning curve in 2 different centres that initiated dedicated programs in AV preservation and repair surgery. We evaluated safety, efficacy, and efficiency end points over time. Our analysis demonstrated that AVr is, in fact, reproducible, showing similar rates of efficacy across both cohorts. Safety and efficiency parameters appear to improve with an increasing number of cases with an estimated learning curve of approximately 40-60 cases. This information might have implications for the management of case volumes and need for mentorship for centres that initiate programs in AVr.
Table 4. Comparison of ACC and CPB times (minutes) grouped according to nature of intervention Aortic root intervention Cohort A ACC CPB Cohort B ACC CPB
Nonaortic root intervention
T1
T2
T3
T1
T2
T3
160.3 39.0 221.9 49.9
143.3 37.9 192.5 55.0
140.9 36.4 180.6 42.3
120.3 32.0 145.3 29.3
116.4 39.2 169 45.7
82.5 31.1 134.1 42.4
114 36.4 148.1 56.8
95.3 23.2 121.7 35.3
100.6 27.7 119 31.1
67.5 23.3 91.8 32.6
72.7 37.6 93.7 45.2
72.4 30.4 97.6 36.6
ACC, aortic crossclamp; CPB, cardiopulmonary bypass; T, tertile.
6
Canadian Journal of Cardiology Volume - 2015
Figure 2. Cumulative sum analysis charts for aortic crossclamp (ACC) and (CPB) times for cohort A. Patients are subgrouped into those who had valve-sparing aortic root replacement (VSARR) with or without (þ/) aortic valve repair (AVr) and AVr only.
Figure 3. Cumulative sum analysis charts for aortic crossclamp (ACC) and (CPB) times for cohort B. Patients are subgrouped into those who had valve-sparing aortic root replacement (VSARR) with or without (þ/) aortic valve repair (AVr) and AVr only.
Malas et al. Analysis of the Learning Curve for AV Repair
somewhere around 40-60 cases to reduce complications associated with AVr to a low, stable rate. In a similar fashion, we observed a statistically significant improvement in efficiency end points with CPB and ACC times after T2 in cohort A (66 patients) and T1 in cohort B (50 patients). To assess progression of the efficiency parameters with time, we used a CUSUM analysis. This method has been previously used in studies that evaluated the need for proctoring in the start of a transcatheter aortic valve implantation program,15 and to assess the learning curve for off-pump coronary artery bypass grafting and outcomes.16,17 For example, CUSUM charts have been used as performance monitoring tools to establish that off-pump coronary artery bypass grafting incorporation into residency training is safe.18 In our study, CUSUM analysis for ACC and CPB (Figs. 2 and 3) indicate no excessive failures in patients with VSARR and/or AVr. Based on acceptable type I and type II statistical errors, we defined 2 boundary lines, H0 and H1, which define acceptable and unacceptable failure rates in ACC or CPB, or that for which performance is considered probably acceptable (H0) or unacceptable (H1). Thus, when we aligned consecutive procedure numbers in a chronological fashion along the x-axis against the cumulative sum on the y-axis, we could assess performance of each procedure compared with the acceptable boundary lines. Any transgression beyond H1 indicates an unacceptably high ACC or CPB time compared with other values. Because no lines crossed the upper decision limits in all graphs, no statistical inference can be made about failures in the learning curve. In other words, none of the data points indicate unacceptably high CPB or ACC values in the learning curve. As for efficacy end points, although we could not make any inferences on efficacy of repair leading to recurrent AI or AV gradients, follow-up in both cohorts was limited in duration; future data might potentially elucidate differences in tertiles. However, repair rates in both cohorts have been good with only a small proportion of patients having significant recurrent AI or increased valve gradient values. It is important to note the differences in the 2 cohorts. The first important difference is the time period over which the procedures were performed. Cohort B preceded cohort A by approximately a decade and this experience was accrued when AVr was in its infancy as a discipline. Many important lessons have since been learned. An important implication of this is the longer follow-up time available for cohort B. Second, related to the differences outlined herein, there are some differences in surgical strategy between cohorts. A larger proportion of patients underwent VSARR in cohort A compared with cohort B (73% vs 48%), with more patients who underwent subcommissural annuloplasty in cohort B. This represents a shift in the treatment paradigm in stabilization of the AV annulus that has occurred over time. Although dilatation of the sinotubular junction is typically repaired with a supracoronary aortic replacement, the optimal approach to remodelling and stabilization of the VAJ continues to be debated.19 In patients who present with a dilated VAJ, but a normal or mildly enlarged aortic root, a more aggressive approach using VSARR with the reimplantation technique has increasingly been favoured over subcommissural annuloplasty.11,20,21 This is particularly true in patients with bicuspid AVs, dilated functional annulus, connective tissue disease, or in young patients.
7
Another important observation noted between cohorts is the larger proportion of patients who underwent any type of cusp repair in cohort A (73% vs 58%). With increasing experience with AVr, there has been greater appreciation of cusp pathology that might be either the primary cause of AI, might be unmasked after annular and root interventions, or might be induced because of technical errors. Correction of this cusp pathology is critical to a durable AVr and we advocate an aggressive approach to the diagnosis and treatment of associated cusp disease. This analysis provides certain insights that might be useful to surgeons and centres that initiate programs in AV preservation and repair surgery. First, in both centres, the approach used has been a programmatic one, empowering teams led by surgeons that include cardiologists, anaesthesiologists, and echocardiographers, and allied health professionals. Second, for surgeons who initiate new programs, having adequate numbers of eligible patients can be an important challenge. If the case volume is 5-10 cases per year, then it can take 5 or more years to progress through the 40-60 cases required for the learning curve. In these contexts, there is a balance between waiting for the ideal patient to perform an AVr to using a more inclusive strategy for patient selection. We would favour the latter. It is also useful within a program to triage all potentially eligible cases to a single surgeon, when case volume is expected to be a limiting factor. Last, the role of mentorship in safely and successfully traversing the learning curve cannot be overemphasized. Like the adoption of any new cardiac surgical procedure, this mentorship can take the form of informal case discussions, review of echocardiograms and other imaging information, formal proctoring of cases, and importantly, discussion of failures. Seeking and developing a strong mentorship relationship can be critical to the success of a program. Limitations Limitations of this study include the modest sample size used in both cohorts and a smaller duration of follow-up for cohort A. This reflects the difference in time of inception of the AV programs at each institution. Furthermore, the patients in both cohorts represent a heterogeneous population with a wide mix of pathologies, posing challenges in comparison of techniques across tertiles and between cohorts. The notion that AVr comprises a number of surgical techniques, rather than being a single entity, further adds to this challenge. Last, it is important to note that improvement in outcomes might also be influenced by an improvement in patient selection and optimization of postoperative care.22 Conclusions In 2 independent centres that initiated dedicated programs in AVr, we found that procedural safety and efficiency improved with experience and efficacy was consistent over time. AVr is safe and reproducible and appears to have a learning curve of approximately 40-60 cases. Acknowledgements The authors thank Kathryn McLean, Sophia Chaudry, and Karen Holmes, University of Ottawa Heart Institute, for their assistance in maintaining the patient database.
8
Disclosures The authors have no conflicts of interest to disclose. References 1. Price J, De Kerchove L, Glineur D, et al. Risk of valve-related events after aortic valve repair. Ann Thorac Surg 2013;95:606-13. 2. Aicher D, Fries R, Rodionycheva S, et al. Aortic valve repair leads to a low incidence of valve-related complications. Eur J Cardiothorac Surg 2010;37:127-32. 3. Saczkowski R, Malas T, de Kerchove L, El Khoury G, Boodhwani M. Systematic review of aortic valve preservation and repair. Ann Cardiothorac Surg 2013;2:3-9. 4. Nishimura RA, Otto CM, Bonow RO, et al. 2014 AHA/ACC guideline for the management of patients with valvular heart disease: a report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines. J Am Coll Cardiol 2014;63:e57-185. 5. Boodhwani M, de Kerchove L, Glineur D, et al. Repair-oriented classification of aortic insufficiency: impact on surgical techniques and clinical outcomes. J Thorac Cardiovasc Surg 2009;137:286-94. 6. Langer F, Aicher D, Kissinger A, et al. Aortic valve repair using a differentiated surgical strategy. Circulation 2004;110(11 suppl 1): II-67-73.
Canadian Journal of Cardiology Volume - 2015 11. de Kerchove L, Boodhwani M, Glineur D, et al. Valve sparing-root replacement with the reimplantation technique to increase the durability of bicuspid aortic valve repair. J Thorac Cardiovasc Surg 2011;142: 1430-8. 12. Boodhwani M, de Kerchove L, El Khoury G. Aortic root replacement using the reimplantation technique: tips and tricks. Interact Cardiovasc Thorac Surg 2009;8:584-6. 13. Schäfers HJ, Aicher D. Root remodeling for aortic root dilatation. Ann Cardiothorac Surg 2013;2:113-6. 14. Kestin IG. A statistical approach to measuring the competence of anaesthetic trainees at practical procedures. Br J Anaesth 1995;75:805-9. 15. Glauber M, Murzi M, Cerillo AG. Is proctoring mandatory when starting a TAVI program? Ann Cardiothorac Surg 2012;1:190-3. 16. Novick RJ, Fox SA, Stitt LW, et al. Effect of off-pump coronary artery bypass grafting on risk-adjusted and cumulative sum failure outcomes after coronary artery surgery. J Cardiovasc Surg 2002;17:520-8. 17. Caputo M, Reeves BC, Rogers CA, Ascione R, Angelini GD. Monitoring the performance of residents during training in off-pump coronary surgery. J Thorac Cardiovasc Surg 2004;128:907-15. 18. Murphy GJ, Rogers CA, Caputo M, Angelini GD. Acquiring proficiency in off-pump surgery: traversing the learning curve, reproducibility, and quality control. Ann Thorac Surg 2005;80:1965-70.
7. Aicher D, Schäfers HJ. Aortic valve repairdcurrent status, indications, and outcomes. Semin Thorac Cardiovasc Surg 2012;24:195-201.
19. de Kerchove L, El Khoury G. Anatomy and pathophysiology of the ventriculo-aortic junction: implication in aortic valve repair surgery. Ann Cardiothorac Surg 2013;2:57-64.
8. Aicher D, Schäfers HJ. Bicuspidization of the regurgitant unicuspid aortic valve. Multimed Man Cardiothorac Surg 2010;2010. mmcts.2009. 004069.
20. David TE. Aortic valve sparing operations: outcomes at 20 years. Ann Cardiothorac Surg 2013;2:24-9.
9. Erasmi AW, Sievers HH, Bechtel JF, et al. Remodeling or reimplantation for valve-sparing aortic root surgery? Ann Thorac Surg 2007;83:S752-6.
21. Vallabhajosyula P, Komlo C, Szeto WY, et al. Root stabilization of the repaired bicuspid aortic valve: subcommissural annuloplasty versus root reimplantation. Ann Thorac Surg 2014;97:1227-34.
10. Boodhwani M, El Khoury G. Aortic valve repair. Oper Tech Thorac Cardiovasc Surg 2009;14:266-80.
22. Colquhoun PH. CUSUM analysis of J-pouch surgery reflects no learning curve after board certification. Can J Surg 2008;51:296-9.