Author + information
- Received November 24, 2010
- Revision received April 19, 2011
- Accepted May 15, 2011
- Published online October 4, 2011.
- Jonathan Buber, MD⁎,
- David Luria, MD⁎,⁎ (, )
- Leonid Sternik, MD⁎,
- Ehud Raanani, MD⁎,
- Micha S. Feinberg, MD⁎,
- Ilan Goldenberg, MD†,
- Eyal Nof, MD⁎,
- Osnat Gurevitz, MD⁎,
- Michael Eldar, MD⁎,
- Michael Glikson, MD⁎ and
- Rafael Kuperstein, MD⁎
- ↵⁎Reprint requests and correspondence:
Dr. David Luria, Leviev Heart Center, Sheba Medical Center, Tel Hashomer 52621, Israel
Objectives The aim of this study was to evaluate whether certain post-Maze left atrial (LA) contractile profiles may pose a risk for ischemic stroke.
Background The mechanical contraction of the left atrium may be modified after the Maze procedure. Whether this imposes a risk for stroke, even in the presence of sinus rhythm and after removal of the LA appendage, is not known.
Methods Clinical, surgery-related, and echocardiographic data from 150 patients who underwent radiofrequency and cryoablation Maze procedures without the use of atrial incisions between 2004 and 2009 and were in sustained sinus rhythm were collected and analyzed. The occurrence of stroke was evaluated by reviewing clinical records. All stroke events were adjudicated by a neurologist.
Results At a mean follow-up time of 24.5 months, 15 patients (10%) had experienced ischemic strokes. Forty-seven patients (31%) had no evidence of LA mechanical contraction at 3 months after surgery (baseline assessment) and on follow-up echocardiography. Multivariate analysis showed that a lack of LA mechanical contraction at baseline was associated with a 5-fold increase in the risk for stroke (p = 0.02) during follow-up. Larger atria imposed a significant risk as well; LA volume index ≥33 ml/m2 was associated with a 3-fold risk increase (p = 0.03). These effects were maintained regardless of the lack of mechanical valve implantation and anticoagulation treatment.
Conclusions Absence of LA contraction and LA volume index ≥33 ml/m2 result in a significant increase in the risk for thromboembolic stroke after the Maze procedure for patients in sinus rhythm.
The surgical Maze procedure is a useful method for sinus rhythm (SR) restoration in patients with atrial fibrillation (AF) undergoing cardiac surgery for other indications. The contemporary “modified” Cox Maze III procedure comprises multiple lines of radiofrequency and cryoablation in the left atrium, aiming to isolate the pulmonary veins and thus to create an electrical separation of the different parts of the atrium (1,2). This method has an excellent success rate for SR maintenance of up to 90% (3–5).
Previous studies have shown that the left atrium might fail to regain its full contractile capability after the Maze procedure and the restoration of SR and thus may remain in a mechanically nonfunctional state (6,7). Although the Maze procedure has been shown to substantially reduce the risk for thromboembolic events (8), it is not known whether this effect also exists for patients with mechanically inactive left atria.
The aim of this study was therefore to determine whether the lack of left atrial (LA) mechanical contraction (LAMC) after the Maze procedure imposes an increased risk for the occurrence of a thromboembolic stroke.
From February 2004 to November 2009, 236 patients underwent the Maze procedure at our institution, using a combination of radiofrequency and cryoablation lesions, as previously described (1). After surgery, patients were followed up at our institutional AF and cardiac surgery clinics, and all data were prospectively obtained during their follow-up visits.
Patients were included in this study if: 1) normal SR was documented beginning at the baseline 3-month post-surgery visit and throughout follow-up (primary analysis) or if normal SR was documented only at the baseline 3-month visit, with follow-up censored upon the development of non-SR (secondary analysis); 2) at least 1 detailed pre-operative and 2 follow-up echocardiographic examinations were performed; and 3) all demographic and clinical data were available. Of the entire Maze-operated cohort, 158 patients (67%) had no evidence of AF or of other atrial tachyarrhythmias during their follow-up. Eight patients (all arriving from different countries to undergo the surgery) were lost to follow-up, leaving 150 patients to be included in this study.
Antiarrhythmic treatment was discontinued after surgery and not renewed. In addition, 28 patients with normal SR at the baseline visit who developed atrial arrhythmias during follow-up (all within 6 months of follow-up) were included in a secondary analysis (comprising a total of 178 patients with normal SR at baseline). Anticoagulation treatment with warfarin was discontinued 3 months after surgery and was not renewed for patients who: 1) were not implanted mechanical prosthetic valves; and 2) lacked other indications for anticoagulation treatment.
For patients implanted with mechanical valves, treatment with warfarin was continued for the entire follow-up period, and for this study population, this was the only indication for anticoagulation. All other study patients were treated with either clopidogrel or aspirin.
Baseline demographic and clinical data were prospectively collected for all the patients. The CHA2DS2-VASc score (consisting of congestive heart failure; hypertension; age [2 points if ≥75 years]; diabetes mellitus; stroke, transient ischemic attack, or thromboembolism [2 points]; vascular disease; age [65 to 74 years]; and female sex category) was retrospectively calculated for each patient (9). All patients underwent baseline electrocardiographic and echocardiographic examinations before surgery. Follow-up was conducted at our institution's outpatient clinics. A complete physical examination, review of hospitalization records after index surgery (if extant), 12-lead electrocardiography, and 24-h Holter monitoring were performed as described previously. Detailed echocardiographic studies were performed at 3 and 12 months after surgery and annually thereafter. All available clinical data from visits not performed at our institution, including electrocardiography and Holter monitoring, were obtained.
Echocardiography was performed using commercially available systems (Philips iE33, Philips Medical Systems, Bothell, Washington; Vivid 7, GE Vingmed Ultrasound AS, Horten, Norway; Vivid I, GE, Tirat Carmel, Israel; or Sonos 7500, Hewlett-Packard, Andover, Massachusetts), and echocardiograms were reviewed by an experienced echocardiographer (R.K.). LA measurements were performed by two physicians (J.B. and R.K.) unaware of each other's readings, according to the recommendations of the American Society of Echocardiography for chamber quantification (10). LA anteroposterior diameter was measured using 2-dimensional echocardiography at the end of left ventricular systole in the long-axis parasternal view. LA volume was measured using Simpson's method from the 4-chamber apical view, indexed to body surface area (left atrial volume index [LAVI]), and divided into quartiles at post-surgical measurement of ≤21, 21.1 to 27, 27.1 to 32.9, and ≥33 ml/m2. Care was taken to exclude the pulmonary veins from the LA tracing.
Transmitral peak velocity and the velocity-time integral of the early (E) and late (A) filling waves were measured in the apical 4-chamber view. The absence of an A wave on the tracings was considered as the absence of mechanical atrial contraction. All patients were in SR during the echocardiographic examinations.
The study was approved by the Sheba Medical Center Institutional Review Board.
End point evaluation
The end point of this study was defined as the occurrence of an ischemic stroke suspected to be of thromboembolic origin at any time during the follow-up period. A stroke was defined as the sudden onset of a focal neurologic deficit in a location consistent with the territory of a major cerebral artery and lasting longer than 24 h. All patients who had strokes during the follow-up period were hospitalized at our institution at the time of the events. The adjudication of stroke events was based on the diagnosis of an expert neurologist during the index hospital stay. Brain computed tomography was performed, and hemorrhagic stroke was excluded in all cases at the time of arrival to the hospital. SR was confirmed for all patients at the time of the occurrence of stroke by performing repeated electrocardiography for the entire duration of the stroke-related hospital stay.
The clinical characteristics of study subjects stratified by the presence or absence of LAMC and by the occurrence of stroke for patients without LAMC were compared using chi-square tests for categorical variables and t tests or Mann-Whitney-Wilcoxon tests for continuous variables. For time-to-event analyses, the Kaplan-Meier method was used, and groups were compared using log-rank statistics. Time 0 was defined as the echocardiographic examination performed 3 months after surgery, because at this time point, anticoagulation therapy was discontinued for the patient group not implanted with prosthetic valves. None of the strokes occurred within this time period. The Cox proportional hazards regression model was used to calculate the risk for 3 pre-specified covariates of the study end point. The 3 covariates were assessed first separately using univariate analysis, and only those with p values <0.10 were included in the final multivariate model. The pre-specified covariates included were lack of LA mechanical function, LAVI ≥33 ml/m2, and a CHA2DS2-VASc score ≥2. LAVI was dichotomized at the upper quartile (≥33 ml/m2) and compared with the 3 lower quartiles (<33 ml/m2) for our primary analysis. In a secondary analysis, event rates were also assessed by individual LAVI quartiles. Echocardiographic parameters used for all survival analyses were obtained at the baseline 3-month echocardiographic assessment. Interaction-term analysis was used to assess the risk for stroke by atrial size in patients with absent LAMC by including an interaction term for LAMC (yes or no) by LAVI (≥33 ml/m2) in the multivariate model. A 2-sided p value < 0.05 was used to denote statistical significance.
Of the 150 study patients, 47 (31%) had no evidence of LAMC at 3-month echocardiographic assessment after index surgery (Fig. 1). Follow-up echocardiograms obtained at 12 months and yearly thereafter showed consistent findings. Thus, none of the patients without LAMC at baseline developed atrial contraction during follow-up, and all patients with LAMC at baseline continued to show evidence of atrial contraction on follow-up echocardiography. Baseline characteristics of the study patients stratified by the presence of LAMC are presented in Table 1. Median follow-up periods were similar for the 2 groups (17 months for patients without LAMC and 19 months for those with LAMC). Forty-one patients continued their anticoagulation treatment during the entire follow-up period, all because of prosthetic mechanical valve implantation. Of these, 11 patients had no evident LAMC (constituting 23% of this group), and 30 had intact LAMC (constituting 29% of this group) (p = NS). None of the patients included in the analysis was implanted with a permanent pacemaker.
Mechanical atrial activity and the occurrence of thromboembolic stroke
Fifteen patients (10% of the entire study sample) experienced ischemic strokes during follow-up, while no hemorrhagic strokes occurred. Proof of left atrial appendage (LAA) obliteration existed in 5 patients who had strokes and was provided via transesophageal echocardiography performed at a mean time of 32 ± 8 h after admission. All 5 patients belonged to the group of patients with absent LAMC, and 3 of them belonged to the group with LAVI ≥33 ml/m2 as well. In none of these patients was flow beyond the line of LAA occlusion observed, and in 4 of them, the residual stump measured <1 cm, both findings suggesting adequate obliteration. There was no evidence of thrombi in the left atria in these examinations, yet in 3 patients, severe swirling was observed.
Ten patients without LAMC (21%) had strokes, compared to 5 (5%) with intact LAMC. Kaplan-Meier estimates of event-free outcomes in the 2 groups are shown in Figure 2A. At 2-year follow-up, the cumulative probability of survival without stroke was 43% in patients without LAMC, compared with 95% in those with intact LAMC (log-rank p = 0.003 for the overall difference in event rates during follow-up).
Three patients who had strokes were treated with anticoagulation at the time of their strokes (constituting 7% of the entire anticoagulated group), of whom 2 had absent LAMC, compared with 12 patients who had strokes and were not receiving anticoagulation treatment (constituting 11% of the entire group not receiving anticoagulation; p = NS). Of the 2 patients receiving anticoagulation without LAMC, 1 had an international normalized ratio within the therapeutic range (3.8), while the other had a considerably lower international normalized ratio (1.2).
To evaluate the possible association between LAMC and the occurrence of stroke in a manner that removed any possible influence of both anticoagulation treatment and the presence of a prosthetic mechanical valve (which alone may be a source of embolization), we also examined the relationship between LAMC and stroke in patients who did not meet these criteria. A similar relationship existed within this subgroup. The cumulative probability of survival without stroke was 45% in patients without LAMC, compared with 95% in those with intact LAMC (log-rank p = 0.01 for the overall difference in event rates during follow-up) (Fig. 2B). Multivariate Cox proportional hazards regression analysis showed that the absence of LAMC was independently associated with a nearly 5-fold (95% confidence interval [CI]: 1.19 to 19; p = 0.02) increase in the risk for stroke for the entire study population and with a 4.1-fold (95% CI: 1.07 to 16.3; p = 0.04) increase in risk for patients not on anticoagulation treatment (Table 2).
LAVI and the occurrence of thromboembolic stroke
Of the entire study population, 35 patients had LAVI ≥33 ml/m2, of whom 8 (23%) had ischemic strokes. During follow-up, the cumulative probability of no stroke event was 55% in patients with LAVI ≥33 ml/m2, compared with 94% in those with LAVI <33 ml/m2 (log-rank, p = 0.01) (Fig. 2C). This effect was more moderate for the subpopulation of patients not receiving anticoagulation therapy (log-rank p = 0.07; data not shown). Multivariate Cox proportional hazards regression analysis showed that LAVI ≥33 ml/m2 was associated with >3-fold risk increase for stroke in the entire study population (95% CI: 1.10 to 9.48; p = 0.03). To confirm the relationship between higher atrial volume and the risk for thromboembolic stroke, we performed a separate multivariate model, incorporating LAVI as a continuous variable. In this separate model, higher LAVI was similarly independently associated with an increased risk for stroke (hazard ratio [HR]: 1.03; 95% CI: 1.01 to 1.07; p = 0.02; data not shown). A similar trend, but without statistical significance, also existed for patients not receiving anticoagulation therapy with LAVI ≥33 ml/m2 (HR: 3.33; 95% CI: 0.93 to 11.9; p = 0.06) (Table 2).
Risk for stroke by LAVI in patients without LAMC
In the total study population, the only parameters that were significantly different between those who did and did not develop strokes were the presence or absence of LAMC and LAVI ≥33 ml/m2. The CHA2DS2-VASc score was not found to be associated with an increased risk for stroke in the entire cohort. In Table 3, a comparison of selected parameters among the group of patients without LAMC stratified by the occurrence of stroke is shown. The only parameter statistically different between the groups was LAVI ≥33 ml/m2. Interaction-term analysis showed that the risk for stroke in patients without LAMC was similar between those with LAVI ≥33 ml/m2 and those with LAVI <33 ml/m2 (p for the difference = 0.75). However, the lack of a statistically significant interaction in the effect of LAVI on the risk for stroke in patients without LAMC may have been due to the limited number of end point events in the present analysis (n = 15).
Risk for stroke in patients who developed atrial arrhythmias before and during follow-up
Because of the very high rate of post-Maze stroke events in the study subjects, we also evaluated stroke incidence in the 78 patients who developed episodes of AF or other atrial arrhythmias and were excluded from the analysis to determine whether these patients sustained a similar risk for stroke. Follow-up data were available for 100% of these patients, of whom 5 (6%) had strokes. However, information regarding the international normalized ratio and the mechanism and adjudication of the stroke event was available for only 3 patients. For the majority of these patients, atrial arrhythmias were designated as permanent, so evaluation of the LAMC profile on echocardiography was not possible, because of a lack of effective LA contraction secondary to the presence of these arrhythmias. Thus, stroke incidence in the post-Maze atrial arrhythmia group was similar to that in the group with intact LAMC and significantly lower than that in the group with absent LAMC (p < 0.001 for between-group difference; data not shown). Of the aforementioned 78 patients, 28 had normal SR at the baseline 3-month visit but developed atrial arrhythmias during follow-up. Within this subgroup, 27 patients (96%) had echocardiographic evidence of intact LAMC at baseline and throughout follow-up, and 1 patient without evidence of LAMC at baseline experienced a stroke during follow-up (after the development of AF). When this patient subgroup was added to the primary cohort, a secondary multivariate analysis (which included all 178 patients who had normal SR at the baseline, with follow-up censored upon the development of AF) showed consistent results: a lack of LAMC was associated with a nearly significant 4-fold increase in the risk for stroke (HR: 3.83; 95% CI: 1.21 to 12.10), and LAVI >33 ml/m2 was associated with a significant 2.5-fold risk increase (HR: 2.48; 95% CI: 1.01 to 4.90).
In this single-center study, we examined for the first time the risk for ischemic stroke in 150 patients who underwent surgery that included the modified Cox Maze III procedure and subsequently remained in SR. Our main findings are that: 1) for one-third of the patients with sustained SR, no mechanical activity of the left atrium was observed during 2 years of follow-up; 2) a lack of LAMC was found to be independently associated with a 5-fold increase in the risk for stroke; 3) LAVI ≥33 ml/m2 was found to be independently associated with a 3-fold increase in the risk for stroke; and 4) these effects existed regardless of the absence of a mechanical prosthetic valve and anticoagulation treatment.
Previous data on LA performance after the Maze procedure
Several previous studies have addressed the issue of post-Maze LA function and filling capacities. Feinberg et al. (7) found that only 60% of patients who underwent the Maze procedure had evidence of LA contraction after a mean follow-up period of 8 ± 7 months and subsequently confirmed this observation by additional echocardiographic techniques (11). Yashima et al. (6) reported on the early and late echocardiographic follow-up of 35 patients with chronic AF who underwent the Maze procedure at the time of other surgery. On early post-operative echocardiography, 5 patients exhibited no evidence of LA contraction, which later improved to some extent but was still depressed at the end of the follow-up period. Our study is consistent with these findings, showing that for a significant percent of patients undergoing the Maze procedure, LA “electromechanical dissociation” may occur. Apparently, the less extensive lesions associated with the modified Cox Maze III procedure, as well as the use of radiofrequency and cryoablation techniques, do not protect against this LA standstill phenomenon. It is likely that this condition undergoes spontaneous resolution in some patients, while in others, it may be long lasting or even become sustained.
Two important findings of our study should be further emphasized. First, although the LAA is considered to be the main source of emboli during AF, its removal during surgery (as performed in 95% of our cohort) was not protective against thromboembolic complications. The underlying mechanism might be the formation of a thrombus in the noncontractile LA space or in the appendage stump due to its incomplete removal. This phenomenon is somewhat similar to that occurring at the time of LA stunning after cardioversion of AF lasting longer than 48 h (12–14). Second, no relation was found in this study between the recently established CHA2DS2-VASc score for stroke prediction and the risk for events in patients without LAMC (Table 3). This finding suggests that LA mechanical standstill is an independent and powerful risk factor for a thromboembolic stroke, which should be accounted for when considering anticoagulation treatment for post-Maze patients.
As observed in previous studies (6,7), patients who underwent the Maze procedure may also obtain a decreased, rather than absent, LA contractile profile. This group of patients was identified among our study subjects as well but was not evaluated separately for the occurrence of stroke. Further studies concerning outcomes in this subpopulation are certainly required.
The modified Cox Maze III procedure is currently considered a safe and effective procedure for the maintenance of SR, and its performance was found to obtain beneficiary effects in prior studies (15) as well as in our study population. These effects, although not addressed in the present analysis, are of considerable importance and include an increase in the left ventricular ejection fraction, freedom from the need for antiarrhythmic drugs, freedom from palpitations, and a decrease in New York Heart Association functional class. Nevertheless, on the basis of the results of the present study, absence of LA contraction and an enlarged left atrium (defined as LAVI ≥33 ml/m2) on consecutive post-operative echocardiograms is associated with an increased risk for subsequent cerebral thromboembolic events. The absence of LAMC after the Cox Maze procedure should alert physicians to the possible risk for future strokes, and further studies are indicated to examine the need for chronic anticoagulation treatment for patients with LA mechanical standstill.
The data for this study were retrospectively analyzed, and groups were not randomized into anticoagulation treatment arms to determine the preventive effect of anticoagulation on patients without LAMC. International normalized ratios were not available for some of the study patients, so we were unable to evaluate whether a possible relationship existed between event occurrence and the adequacy of anticoagulation. To overcome this limitation, data were provided on the subgroup of patients who were not on anticoagulation treatment.
Asymptomatic AF could have occurred, because AF recognition was based on symptoms and periodic Holter examinations. It is possible therefore that our results were overestimated, because asymptomatic paroxysmal AF may still have been present at the time of the stroke but not appeared on the later electrocardiographic tracings obtained during the hospital stay.
The contribution of a decreased, rather than absent, LA contractile profile to the occurrence of stroke was not assessed in the present study, and this situation might by itself possess a risk for the occurrence of such events.
The majority of our study patients did not undergo transesophageal echocardiography to ascertain complete LAA obliteration or to directly visualize LA thrombus at the time of the stroke event. However, adequate LAA obliteration was demonstrated for all 5 patients who did undergo transesophageal echocardiography during stroke-related hospital stay. In addition, currently available data indicate that successful LAA obliteration rates tend to significantly increase along with increased surgeon experience and may be as high as 90% (16). In our study patients, all but 5 LAA obliterations were carried out by a single surgeon, suggesting a very high level of skill. In addition, contradictory results have been reported from a single randomized controlled trial (consisting of 77 patients) (17) and several observational studies (18,19) that aimed to evaluate the actual protective effect of complete LAA obliteration from subsequent thromboembolic stroke events for patients in SR. We therefore believe that the lack of complete data regarding the obliteration of the LAA in this study should not be considered an obstacle in the interpretation and future implications of its results.
Although the present study represents a large population of patients who were prospectively followed up after Maze procedures, the number of end point events in the study was limited (n = 15). Accordingly, the multivariate models were adjusted for only 2 pre-specified predictors, resulting in only modest overfitting. As mentioned previously, the lack of a statistically significant interaction in the effect of LAVI on the risk for stroke in patients without LAMC may be because the study was not adequately powered to detect interaction effects.
For patients who undergo the modified Cox Maze III procedure, the absence of LAMC is common and is independently associated with an increased risk for thromboembolic stroke. This risk is sustained regardless of the presence of stable SR. In addition, LAVI ≥33 ml/m2 is independently associated with increased risk for thromboembolic stroke. We therefore assume that for patients who undergo Maze procedures, routine echocardiographic follow-up examinations should be performed, with an emphasis on recording the contractility profile and volume measurement of the left atrium. Anticoagulant treatment should be strongly considered for patients with absent LAMC and with LAVI ≥33 ml/m2.
Dr. Raanani is a medical consultant for Edwards Lifesciences and Medtronic. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose. Drs. Buber and Luria contributed equally to this work.
- Abbreviations and Acronyms
- atrial fibrillation
- confidence interval
- hazard ratio
- left atrial
- left atrial appendage
- left atrial mechanical contraction
- left atrial volume index
- sinus rhythm
- Received November 24, 2010.
- Revision received April 19, 2011.
- Accepted May 15, 2011.
- American College of Cardiology Foundation
- Yashima N.,
- Nasu M.,
- Kawazoe K.,
- Hiramori K.
- Lang R.M.,
- Bierig M.,
- Devereux R.B.,
- et al.,
- Chamber Quantification Writing Group,
- American Society of Echocardiography's Guidelines and Standards Committee,
- European Association of Echocardiography
- Ad N.,
- Henry L.,
- Hunt S.
- Onalan O.,
- Crystal E.
- Juratli N.,
- Wilkoff B.L.,
- Tchou P.,
- et al.
- Garcia-Fernandez M.A.,
- Perez-David E.,
- Quiles J.,
- et al.