Background
Surgical closure of the left atrial appendage (LAA) was first reported by John Madden in 1949 ( Fig. 40.1 ) In that publication, Madden stated, “ Since a thrombus is the precursor of every arterial embolus, the ideal prophylaxis for recurrent arterial emboli should be the removal of the thrombus together with its site of origin .” Despite Madden’s recommendation, little interest was shown in the LAA or its relation to strokes for the next five decades ( Fig. 40.2 ). In 1996, Blackshear and Odell reported that the left atrium (LA) was involved in up to 57% of strokes in patients with AF with rheumatic heart disease and in up to 91% of patients with AF without rheumatic heart disease.
John L. Madden, who in 1949 published the first report on surgical closure of the left atrial appendage to prevent systemic thromboembolism.
The number of scientific publications on the left atrial appendage (LAA) from Madden’s 1949 article through 2011. Note that the bars do not represent the same timeframes.
(Modified from Chatterjee S, Alexander JC, Pearson PJ, Feldman T. Left atrial appendage occlusion: lessons learned from surgical and transcatheter experiences. Ann Thorac Surg. 2011;92(6):2283–2292.)
In 1999, we reported the first large clinical series of LAA occlusions (LAAOs) by either surgical amputation or endocardial surgical closure as a part of the Maze-I,-II, and-III procedures. The LAA occlusions were accomplished in the Maze-I, Maze-II, and the cut-and-sew version of the Maze-III procedure by surgical amputation ( Fig. 40.3 ) and by cryosurgical isolation and endocardial suture closure in the minimally invasive cryosurgical version of the Maze-III (CryoMaze-III) procedure ( Fig. 40.4 ). Surprisingly, we found that perioperative strokes associated with both routine and complex cardiac surgical procedures were virtually eliminated when the Maze procedure was performed concomitantly with those procedures ( Fig. 40.5 ). It was well documented that postoperative atrial fibrillation (POAF) was associated with an increase in perioperative strokes, , so one might attribute this dramatic reduction in perioperative stroke rate to the Maze procedure having prevented POAF. However, the Maze procedure does not prevent POAF as evidenced by 37% of the 265 patients who had either type of Maze-III procedure experienced POAF within the first 2 weeks after surgery, peaking at 4 days. Thus the Maze procedure alone could not have been responsible for the lack of perioperative strokes. The only plausible explanation for the dramatic decrease in perioperative stroke was amputation or closure of the LAA during the procedure.
In the original cut-and-sew Maze-III procedure, forceps were inserted into the orifice of the left atrial appendage (LAA) to grasp its endocardial tip and invert it into the left atrium for better visualization and determination of the exact location of its base. The LAA was then excised at its base, and the defect was closed in two layers with sutures.
In the minimally invasive CryoMaze-III procedure that replaced the cut-and-sew Maze procedure in 1997, the left atrial appendage (LAA) was first inverted to check for thrombi and then returned to its normal position. A cryolesion was placed around the circumference of the LAA orifice with a 2-cm-diameter nondisposable cryoprobe to electrically isolate the entire LAA from the remainder of the heart (left panel). The LAA was left intact, and its orifice was closed in two layers of sutures (right panel). Thus, after both procedures, not only was the LAA excluded from the systemic circulation, but the myocardium of the LAA was also electrically isolated from the myocardium of the left atrium (LA). In some of the early minimally invasive CryoMaze-III procedures, the cryolesion around the LAA orifice was replaced by a single linear lesion out into the LAA. In retrospect, this was a mistake because even though the LAA was excluded from the systemic circulation by suture closure, the suture line alone did not electrically isolate the LAA from the LA. Fortunately, this alteration did not change the efficacy of the CryoMaze-III procedure.
Upper panel, Perioperative stroke rates after the most common cardiac surgical procedures according to the 1998 Society of Thoracic Surgeons National Cardiac Surgery Database plus the stroke rate when the Maze procedure was performed with or without these procedures. Lower panel, Same information in bar graph form. See text for further discussion. AVR, aortic valve replacement; CABG, coronary artery bypass grafting; MV, mitral valve; MVR, mitral valve replacement.
The CHADS 2 and the CHA 2 DS 2 -VASc scoring systems for stroke risk in patients with atrial fibrillation (AF) had not been devised at the time of our 1999 report. However, the risk of stroke associated with AF was well established at that time based on a number of clinical trials, a history of previous stroke or transient ischemic attack (TIA) and anticoagulation status. Therefore, the number of long-term strokes expected over the 12-year follow-up period if these patients had not undergone a Maze procedure could be calculated. By categorizing the patients into groups with different stroke risks, it was predicted that 68 strokes would have occurred over the ensuing 12 years in these 265 patients if they not undergone a Maze procedure ( Fig. 40.6 ). In reality, only one stroke occurred during that time even though 65% of the patients were not anticoagulated. In 2003, Prasad et al. reported a longer follow-up of the same cohort of patients showing that after 15 years, 99.3% of the patients were free of stroke after a Maze procedure ( Fig. 40.7 ). Again, many attributed this low incidence of long-term stroke to the success of the Maze procedure in preventing AF. Although LAA closure was clearly responsible for the decrease in perioperative strokes, the decrease in long-term strokes could have been due to the low recurrence rate of AF, LAA occlusion, or a synergistic combination of both. See Chapter 43 for a detailed update on this subject.
The predicted stroke curves for the patients in our original Maze series divided into five groups based on their known risk factors and thus the predicted incidence of stroke over the ensuing 12 years had they not undergone a Maze procedure. The high-risk patients were subdivided on the basis of those who were not anticoagulated (group I) and those who were anticoagulated (group III). The lower-risk patients were also subdivided into those who were not anticoagulated (group II) and those who were anticoagulated (group IV). Group V represented patients with “lone atrial fibrillation” (AF) representing the lowest-risk group for having a stroke. The actual observed stroke rate for all of these patients after the Maze procedure is the lowest line on the graph. Overall, the contemporary knowledge of stroke risks at that time predicted that 68 strokes would have occurred over the ensuing 12 years had their AF not been ablated. The total number of strokes that actually occurred in all five combined groups after the Maze procedure was one stroke in 12 years. See text for further discussion.
(Modified from Cox JL, Ad N, Palazzo T. Impact of the Maze procedure on the stroke rate in patients with atrial fibrillation. J Thor Cardiovasc Surg . 1999;118:833–840.)
The 15-year freedom from stroke after the Maze procedure was 99.3%. See text for further discussion.
Evolution of the Surgical Management of the Left Atrial Appendage
Interest in the LAA increased in the second decade of the 21st century as cardiologists and particularly surgeons came to appreciate the relationship between LAA thrombosis and strokes associated with AF as evidenced by the rapid growth in the number of annual LAA publications ( Fig. 40.8 ). It had already been established that approximately 20% of all strokes are associated with AF and that most of the thrombi responsible for strokes in patients with AF originate in the LAA. As a result, surgeons were encouraged to become more aggressive in amputating or closing the LAA in patients with AF undergoing cardiac surgery.
Annual number of scientific publications on the left atrial appendage (LAA) from 1955 through 2022. Note the continued increased interest in the LAA after 2011 illustrated in Fig. 40.2 .
In 2005, Healey et al. published the first randomized controlled trial (now referred to as the LAAOS-I trial) to evaluate whether surgeons could reliably close the LAA surgically. There were two arms in the trial, one in which the LAA was closed with sutures placed epicardially using the surgeon’s preferred technique and the other in which the LAA was closed using an epicardial stapling device. The LAAOS trial showed that surgeons could not close the LAA safely and reliably by either of these techniques. In the epicardial suture closure group, 57% of the patients had a residual connection between the LAA and the LA, and in the surgical stapling group, there was a residual pouch greater than 1 cm in depth at the base of the LAA in 28% of patients that proved to be thrombogenic. Interestingly, in patients who had successful LAA closures by either technique (i.e., complete closure without residual pouches >1 cm), the stroke rate was reduced, though those patients could not be subjected to statistical analysis because of the structure of the trial. The LAAOS-I trial demonstrated definitively that surgeons could not close the LAA reliably even when using their best direct surgical techniques.
The findings of the LAAOS-I trial were confirmed by Kanderian and coworkers’ 2008 Cleveland Clinic trial and Cullen and coworkers’ 2016 Mayo Clinic trial. In the latter, LAA surgical amputation was included in the study, and interestingly, there was roughly a 20% postoperative thrombus rate along the LAA suture line closure with approximately 12% of the thrombi being in actual contact with the systemic circulation. This study in particular showed that although amputation of the LAA is obviously simple and cheap, it may not be as safe as most surgeons believe it to be. Fortunately, these problems were overcome with the introduction of the surgical AtriClip (AtriCure, Inc.), which is discussed in detail in Chapters 44 and 45 .
Left Atrial Appendage Closure in Cardiac Surgery Patients With Atrial Fibrillation
The question of whether or not patients undergoing cardiac surgery with a history of AF should have their LAAs closed was addressed in a 2017 report from the Society of Thoracic Surgeons Adult Cardiac Surgery Database by Friedman et al. The study included 10,524 patients, 37% of whom had LAA occlusion at the time of surgery and 63% of whom did not. The study showed superior results for thromboembolism (TE), an end point that included TE, hemorrhagic stroke, or death, and for all-cause mortality in the patients with AF who had LAA closure at the time of surgery ( Fig. 40.9 ). Likewise, a 2019 study by Yao et al. of 75,782 patients also demonstrated an advantage in cumulative risk by closing the LAA in cardiac surgery patients with AF ( Fig. 40.10 ). A meta-analysis of 280,585 patients by Gutierrez et al. also showed results that favored closure of the LAA in cardiac surgery patients who had preoperative AF. Finally, the issue of whether to close the LAA in cardiac surgery patients with AF was settled by the multicenter, prospective, randomized LAAOS-III trial reported by Whitlock et al. in 2021. The LAAOS-III trial documented that LAA occlusion in cardiac surgery patients with AF decreases subsequent strokes over a mean of 3.8 years ( Fig. 40.11 ). This seminal trial is discussed in detail by Whitlock in Chapter 46 .
