Author + information
- Received May 11, 2015
- Revision received July 20, 2015
- Accepted July 21, 2015
- Published online September 29, 2015.
- Tirone E. David, MD∗ (, )
- Carolyn M. David, BN,
- Cedric Manlhiot, BSc,
- Jack Colman, MD,
- Andrew M. Crean, MD and
- Timothy Bradley, MB, ChB
- Divisions of Cardiovascular Surgery and Cardiology, Peter Munk Cardiac Centre, Toronto General Hospital, University of Toronto, Toronto, Ontario, Canada
- ↵∗Reprint requests and correspondence:
Dr. Tirone E. David, Division of Cardiovascular Surgery, University of Toronto, 200 Elizabeth Street, 4N453, Toronto, Ontario M5G 2C4, Canada.
Background In many cardiac units, aortic valve-sparing operations have become the preferred surgical procedure to treat aortic root aneurysm in patients with Marfan syndrome, based on relatively short-term outcomes.
Objectives This study examined the long-term outcomes of aortic valve-sparing operations in patients with Marfan syndrome.
Methods All patients with Marfan syndrome operated on for aortic root aneurysm from 1988 through 2012 were followed prospectively for a median of 10 years. Follow-up was 100% complete. Time-to-event analyses were calculated using the Kaplan-Meier method with log-rank test for comparisons.
Results A total of 146 patients with Marfan syndrome had aortic valve-sparing operations. Reimplantation of the aortic valve was performed in 121 and remodeling of the aortic root was performed in 25 patients. Mean age was 35.7 ± 11.4 years and two-thirds were men. Nine patients had acute, 2 had chronic type A, and 3 had chronic type B aortic dissections before surgery. There were 1 operative and 6 late deaths, 5 caused by complications of dissections. Mortality rate at 15 years was 6.8 ± 2.9%, higher than the general population matched for age and sex. Five patients required reoperation on the aortic valve: 2 for endocarditis and 3 for aortic insufficiency. Three patients developed severe, 4 moderate, and 3 mild-to-moderate aortic insufficiency. Rate of aortic insufficiency at 15 years was 7.9 ± 3.3%, lower after reimplantation than remodeling. Nine patients developed new distal aortic dissections during follow-up. Rate of dissection at 15 years was 16.5 ± 3.4%.
Conclusions Aortic valve-sparing operations in patients with Marfan syndrome were associated with low rates of valve-related complications in long-term follow-up. Residual and new aortic dissections were the leading cause of death.
Aortic valve-sparing operations were developed to preserve the native aortic valve in patients with aortic root aneurysm (1,2). Their application in patients with Marfan syndrome was initially questioned because of the presence of abnormal fibrillin in the aortic valve (3). Early on in our experience, we treated only patients with aortic root aneurysms with normal aortic cusps and closely monitored post-operative valve function with periodic echocardiography. As our confidence in these operations increased, we expanded their use to patients with mild aortic cusp abnormalities, such as prolapse caused by elongation of free margins and stress fenestrations in the commissural areas caused by dilation of the aortic root. Elongation of the free margin of an aortic cusp was corrected by shortening its central portion along the nodule of Arantius, and large stress fenestrations by weaving a double layer of a fine expanded polytetrafluoroethylene suture along the free margin from commissure to commissure. These techniques of cusp repair proved to be durable; thus, the indications for aortic valve-sparing operations expanded accordingly (4,5).
Aortic valve-sparing operations have now become the preferred surgical procedure in our center to treat young patients with aortic root aneurysm (5) and approximately one-third of them had the diagnosis of Marfan syndrome according to Ghent criteria (6,7). This study examines the long-term outcomes of aortic valve-sparing operations in patients with Marfan syndrome.
From September 1988 through December 2012, a total of 149 patients with the diagnosis of Marfan syndrome by Ghent criteria (6,7) were operated on for aortic root aneurysm or Stanford type A aortic dissection at Toronto General Hospital by 1 surgeon (T.E.D.). The original Ghent criteria (6) were used retrospectively to identify patients operated on before 1997, then prospectively from 1997 through 2010; the revised criteria have been used since then (7). Three patients had aortic root replacement with a conduit containing a mechanical valve because of grossly abnormal aortic cusps and were excluded. The remaining 146 patients had aortic valve-sparing operations. This study was approved by the University Health Network Research Ethics Board and patients' consent was waived.
Two types of aortic valve-sparing operations were performed in these patients (Figure 1): reimplantation of the aortic valve (the David operation ) and remodeling of the aortic root (the Yacoub operation ). Reimplantation of the aortic valve was performed by detaching the coronary arteries from the aneurysmal aortic sinuses, leaving a small rim of tissue around their orifices; a narrow rim of aortic sinus tissue was also left attached to the aortic annulus. Next, a tubular Dacron graft of appropriate diameter was sutured on the outside of the left ventricular outflow tract immediately below the level of the nadir of the aortic annulus. The remnants of aortic sinuses were sutured inside the graft, recreating the crescent shape of the aortic annulus. The cusps were examined to ensure that their coaptation level was well above the level of the nadir of the aortic annulus. If 1 or more aortic cusps were prolapsing, their free margin was shortened by plication along the nodule of Arantius. Additionally, if large fenestrations were present on the free margin along the commissural areas, a double layer of a fine expanded polytetrafluoroethylene suture was weaved along the free margin of the cusp from commissure to commissure. Finally, the coronary arteries were reimplanted on the Dacron graft. During the second decade of experience with this operation, neoaortic sinuses were created by using a slightly larger graft and reducing its diameter at the subannular and sinotubular junction levels (Figure 1).
Remodeling of the aortic root was performed by preparing the aortic root as described previously and tailoring a tubular Dacron in such a way as to recreate the aortic sinuses and suturing it to the remnants of aortic sinuses and aortic annulus. If the aortic annulus was deemed dilated at the time of surgery, an external annuloplasty with a Dacron band along the fibrous components of the left ventricular outflow tract was performed. The aortic cusps and coronary arteries were handled similarly as with the reimplantation procedure.
All patients have been followed prospectively with echocardiographic studies and periodic computed tomography or cardiac magnetic resonance imaging of the entire aorta every 3 to 5 years, unless the patient had a history of aortic dissection or another pre-existing aneurysm, in which case imaging of the aorta was performed more often. Given that aortic insufficiency (AI) was assessed by echocardiography, we used the date of the first post-operative study to reveal AI greater than mild as the time of event. Although the apparition of AI might predate the echocardiogram, using the first report of the abnormality is an adequate and only available surrogate.
Data are described as mean ± SD (continuous variables) or frequencies (categorical variables) as appropriate. Rates of mortality, reoperations, AI, and aortic dissections were calculated using the Kaplan-Meier method with log-rank test for comparisons between groups. To account for the concomitant probability of death or other adverse event, competing risk analysis was used. In competing risk analysis, at time zero, all patients were free from adverse events; patients transiting to either death or other adverse event, whichever occurred first and at the same time, were removed from the proportion of patients who were alive and free from adverse events. Sex-specific life tables from the Province of Ontario from the 2000 to 2002 period (available from Statistics Canada) were used to estimate predicted survival of the patient cohort with yearly estimates fitted with an exponential regression model. All statistical analyses were performed using SAS statistical software version 9.3 (SAS Institute, Inc., Cary, North Carolina).
A total of 146 patients were included in this study. Clinical follow-up was 100% complete to a median of 10.0 years (interquartile range: 3.6 to 14.0 years) and echocardiographic follow-up was complete to a median of 9.3 years (interquartile range: 3.2 to 13.4 years). Follow-up for this study was closed on December 31, 2014.
Table 1 describes the pre-operative clinical, echocardiographic, and angiographic data. Table 2 provides the surgical data. In addition to the aortic valve-sparing operation, one-fourth of the patients had other procedures, mitral valve (MV) repair being the most common.
There was 1 operative death caused by an acute type B aortic dissection and malperfusion on the day after surgery. Ten patients required reopening of the chest because of post-operative bleeding, pericardial tamponade, or cardiac arrest. In 1 of these patients, the right lobe of the liver was torn during resuscitation and was successfully repaired. Transfusion of homologous blood products (packed cells, fresh frozen plasma, or platelets) was needed in 63 of 146 (43%) patients. Seventeen patients developed transient atrial fibrillation and 1 patient required a permanent transvenous pacemaker for sick sinus syndrome (this patient had concomitant MV repair and the maze procedure for chronic atrial fibrillation). There was no other perioperative complication. Post-operatively, patients were advised to take a beta-blocker or, if not tolerated, an alternative antihypertensive agent.
Late mortality and morbidity
There were 6 late deaths: 4 caused by complications of aortic dissections and 2 by complications of chronic obstructive pulmonary disease in patients with severe kyphoscoliosis. The mortality rate (± SE) at 15 years was 6.8 ± 2.9% (Table 3). In the Kaplan-Meier analysis of mortality rate over time, the study population was compared with the general population of Ontario matched for age and sex (Figure 2).
During follow-up, 1 patient with chronic aortic dissection and intimal flaps in the aortic arch suffered a stroke with complete recovery; another patient with infective endocarditis caused by Staphylococcus aureus suffered an embolic stroke with partial neurologic recovery. In addition, 5 patients suffered a transient ischemic attack. Rate of thromboembolism at 15 years was 4.1 ± 2.0%.
Two patients developed infective endocarditis with aortic root abscess both caused by S. aureus at 12 and 23 years after surgery. Both patients were successfully treated with antibiotics and aortic root replacement using an aortic valve homograft.
Five patients required reoperation on the reconstructed aortic root: the 2 patients described previously for infective endocarditis and 3 for AI. The 2 patients with endocarditis had aortic root replacement with aortic homograft and the other 3 had aortic root replacement with a Dacron conduit containing a mechanical valve. All patients survived reoperation. Rate of reoperation on the aortic root at 15 years was 4.9 ± 2.5% (Table 3, Central Illustration). Reoperations were more common in patients who had the remodeling procedure (reimplantation vs. remodeling; hazard ratio: 0.16; 95% confidence interval: 0.01 to 0.54; log-rank p = 0.01). In addition to reoperations on the aortic root, 7 patients had replacement of the entire thoracic and abdominal aorta because of distal aortic aneurysms: 1 died, 1 became paraplegic, and 5 did well. The patient who suffered paraplegia developed aortic root abscess years later. Two patients required MV repair for new severe mitral regurgitation.
During the entire follow-up, 3 patients developed severe AI and underwent reoperations. Another 4 developed moderate and 3 developed mild-to-moderate AI. The remaining 135 operative survivors had no, trace, or mild AI. Rate of AI greater than mild at 15 years was 7.9 ± 3.3% (Table 3). Figure 3 shows the rate of AI greater than mild in all patients and also according to the type of aortic valve-sparing operation (reimplantation vs. remodeling, hazard ratio: 0.43; 95% confidence interval: 0.10 to 1.03; log-rank p = 0.08). Additionally, 2 patients developed severe aortic root abscess years later. Two patients required MV repair for new severe mitral (both underwent MV repair) and 9 developed moderate aortic root abscess years later. Two patients required MV repair for new severe mitral and continue under surveillance.
Before aortic root surgery, 11 patients had Stanford type A and 3 had type B aortic dissections. Postoperatively, 9 patients developed new aortic dissections: 1 in the aortic arch at 8 years, and 8 type B at Day 1 to 23 years after the aortic valve-sparing operation. Two patients with type B aortic dissections died during the acute phase because of malperfusion. Only 1 of the 9 patients with new dissections was not on a beta-blocker or alternative antihypertensive agent when the new aortic dissection occurred. Rate of all aortic dissections in the competing risk analysis is shown in the Central Illustration; the rate of aortic dissection at 15 years was 16.5 ± 3.4% (Table 3). In addition to aortic dissections, 3 other patients developed chronic thoracic or abdominal aneurysms.
Before the advent of open-heart surgery, the survival of patients with Marfan syndrome was greatly reduced because of aortic dissections or rupture of the aorta (8). Replacement of the aortic root with a conduit containing a valve (Bentall procedure ) improved that dismal prognosis. The Bentall procedure is usually performed with a tubular Dacron graft containing a mechanical valve but it can also be done with an aortic valve bioprosthesis or aortic homograft (3). Replacement of the entire aortic root, including the aortic valve cusps, was the standard treatment of aortic root aneurysms until aortic valve-sparing operations were developed (1,2).
This study examined the long-term outcomes of aortic valve-sparing operations in adult patients with Marfan syndrome. The number of adverse events was relatively small, limiting the value of statistical analyses, but confirms the safety and efficacy of aortic valve-sparing operations. Our operative mortality was <1%; other surgeons have reported equally low operative mortality after these operations (10–13). Mortality at 15 years was 6.8%, slightly higher than the general population matched for age and sex. However, this mortality rate seems much lower than that reported for patients treated with the Bentall procedure (3,13). In a series of 271 patients with Marfan syndrome from Johns Hopkins Hospital treated mostly with Bentall procedures with mechanical valves, the mortality rate at 15 years was approximately 24% (13). The survival advantage in our series of aortic valve-sparing operations is also evident compared with that from a multicenter series of 675 patients with Marfan syndrome treated largely with Bentall operation (3). Although the patient profiles may have been different in those studies, the risk factors that are known to affect late survival after cardiac surgery, such as age, aortic dissections, mitral insufficiency, and coronary artery disease, seem to be similar to ours. Only a randomized clinical trial could determine unequivocally whether there is indeed a survival benefit associated with aortic valve-sparing operations compared with the Bentall procedure. However, such a trial is unlikely to ever be conducted given that aortic valve-sparing operations have already become the preferred procedure in most centers.
Surgery for aortic root aneurysm in patients with Marfan syndrome is recommended when the transverse diameter of the aortic root reaches 50 mm or even smaller (45 mm) when there is a family history of aortic dissection or pregnancy is intended (14,15). Patients with Marfan syndrome are usually in their third or fourth decades of life when this surgery is required. Bioprosthetic aortic valves and aortic homografts are not durable in young adults (16,17). Mechanical valves are durable, but require lifelong anticoagulation with drugs like warfarin, which are associated with a constant risk of bleeding and thromboembolism (18). Thus, there is a preference for aortic valve-sparing operations in young patients because of their proven durability and lack of requirement for anticoagulation. Pre-operative AI is not a contraindication to an aortic valve-sparing operation as long as the cusps are reasonably elastic and the cusp prolapse is addressed during the procedure. Actually, more than one-half of our patients needed 1 or more cusps repaired at the time of the aortic valve-sparing procedure.
A review and meta-analysis by Benedetto et al. (19) compared 972 patients with Marfan syndrome who had undergone the Bentall procedure with 413 who had undergone aortic valve-sparing operations. During a relatively short duration of follow-up, they found that the probability of reoperation was 4 times higher after the aortic valve-sparing operation than after the Bentall procedure (1.3%/year vs. 0.3%/year). However, risk of thromboembolism after the aortic valve-sparing operation was lower than after the Bentall procedure (0.3%/year vs. 0.7%/year), whereas the composite risk of reoperation, thromboembolism, and bleeding was similar. The authors concluded that aortic valve-sparing operations represent a valuable option, but should be used with caution in patients for whom the durability of the repair was questionable.
A recent report from a prospective, international cardiac surgery registry comparing 239 aortic valve-sparing operations with 77 Bentall procedures in patients with Marfan syndrome showed similar survival at 1 year, but after aortic valve-sparing operations, 7% had AI of at least moderate severity and 1 patient required aortic valve reoperation for valve failure (20). Such a high 1-year post-operative failure rate is worrisome and contrasts with our findings, perhaps emphasizing the need for centralizing care for patients with Marfan syndrome. In centers having a large experience with this type of surgery, such as Cleveland Clinic, Johns Hopkins Hospital, and Stanford University, the freedom from reoperation is reportedly above 90% at 10 years (10–12).
In our experience, the rate of AI greater than mild was approximately 8% at 15 years and the rate of reoperation on the aortic root was even lower at just under 5% (Central illustration). We were unable to identify predictors of failure, likely because of the sample size and small number of adverse events. However, reimplantation of the aortic valve was associated with better outcomes than remodeling of the aortic root in our series and in a previous meta-analysis (19). We believe that remodeling of the aortic root is inappropriate for patients with Marfan syndrome because the aortic annulus can dilate years after the operation and cause AI (21). An external annuloplasty band along the fibrous components of the left ventricular outflow tract was added in some patients as illustrated in Figure 1, but our experience is inadequate to draw conclusions on its effectiveness in preventing annular dilation.
Common in patients with Marfan syndrome, chest wall deformities can make the aortic root operation difficult. We have never added correction of the chest wall deformity during aortic valve-sparing operations but we have had to do the operation through a left anterior thoracotomy in 2 patients because of severe pectus excavatum.
In patients with Marfan syndrome, aortic root involvement is the principal cause of death because of proximal aortic dissections and rupture (15). Replacement of the aortic root and ascending aorta prevents proximal aortic dissections and increases the lifespan of patients with Marfan syndrome (13). Aortic valve-sparing operations may further increase longevity, but late survival is shorter than in the general population, largely because of complications of residual dissections and development of new ones in the remaining aorta. The recognition of this problem is not new (22,23). A recent observational study by den Hartog et al. (23) on 600 patients with Marfan syndrome from a Dutch registry examined the issue of distal aortic dissections. They excluded patients who had previous aortic dissections but included 194 patients who had prophylactic aortic root surgery, either before or during the median observation time of 6 years. Distal aortic dissections occurred in 54 patients for an annualized rate of 1.5%. Multivariate analysis identified previous aortic root surgery and a descending thoracic aorta diameter cut-off of ≥27 mm as independent predictors of distal dissections. However, only 53% of the distal dissections occurred in aortas with diameters ≥27 mm. The authors speculated that replacement of the ascending aorta with a noncompliant Dacron graft may result in higher pulsatile forces on the aortic arch and proximal descending thoracic aorta, increasing the risk of dissection. Most patients (73%) were taking a beta-blocker and the remainder were on another antihypertensive agent at the time of the dissection. The study suggested no solution to the problem of distal aortic dissections, but the findings emphasize the importance of monitoring the distal aorta in all patients with Marfan syndrome, particularly those with prior aortic root surgery or dissections, with periodic cardiac magnetic resonance imaging or computed tomography imaging of the entire aorta in specialized centers.
For more than 20 years, beta-blocker therapy has been the mainstay of medical care for Marfan syndrome, based on the trial reported by Shores et al. (24). Since 2006, when Habashi et al. (25) first reported that aortic aneurysm in a mouse model of Marfan syndrome can be prevented by the antagonism of increased transforming growth factor-beta signaling afforded by losartan, a number of human clinical drug trials have been performed. In adult patients with Marfan syndrome, Groenink et al. (26) demonstrated that losartan reduced the rates of preoperative aortic root dilation and post-operative aortic arch dilation compared with placebo. Lacro et al. (27) found no significant difference in the rate of aortic root dilation in children and young adults with Marfan syndrome treated with losartan compared with atenolol. Therefore, it seems that the ongoing improvement in life expectancy for patients with Marfan syndrome remains primarily caused by the increased frequency of early diagnosis, more effective monitoring, and elective aortic root aneurysm repair.
This prospective observational study has limitations largely because of its sample size and the relatively small number of adverse events. Early on the issue of annular dilation was not recognized and it is possible that some failures were related to the wrong technique of aortic valve-sparing operation. In addition, one surgeon performed all procedures, and the results may not be generalized.
Aortic valve-sparing operations, particularly the reimplantation procedure, are attractive and durable choices to treat young adults with aortic root aneurysm associated with Marfan syndrome. This type of surgery is complex and should be performed in centers with a high volume of aortic root surgery and experience with aortic valve repair. In addition, patients with Marfan syndrome must be followed throughout their lifetime by a multidisciplinary group of professionals including cardiologists, cardiac surgeons, and geneticists with protocols targeted at detecting distal aortic complications.
COMPETENCY IN MEDICAL KNOWLEDGE: Aortic valve-sparing operations to treat aortic root aneurysm in patients with Marfan syndrome at a center of excellence were associated with low rates of valve-related complications during the first 15 years of observation, during which aortic dissection was the leading cause of death. Repair of aortic root aneurysms in these patients reduced the risk of rupture of the replaced segment of the aorta and valve-related adverse events, but did not prevent distal aortic dissection.
TRANSLATIONAL OUTLOOK: More research is needed to delineate the clinical factors and molecular mechanisms responsible for aortic dissection in patients with aortic disease associated with Marfan syndrome.
This study was funded by the Academic Enrichment Fund of the Division of Cardiovascular Surgery of Peter Munk Cardiac Centre, University Health Network, Toronto, Ontario, Canada. All authors have reported that they have no relationships relevant to the contents of this paper to disclose.
- Abbreviations and Acronyms
- aortic insufficiency
- mitral valve
- Received May 11, 2015.
- Revision received July 20, 2015.
- Accepted July 21, 2015.
- American College of Cardiology Foundation
- Loeys B.L.,
- Dietz H.C.,
- Braverman A.C.,
- et al.
- Bentall H.H.,
- De Bono A.
- Jondeau G.,
- Detaint D.,
- Tubach F.,
- et al.
- Une D.,
- Ruel M.,
- David T.E.
- Fukushima S.,
- Tesar P.J.,
- Pearse B.,
- et al.
- Hammermeister K.,
- Sethi G.K.,
- Henderson W.G.,
- et al.
- Benedetto U.,
- Melina G.,
- Takkenberg J.J.,
- et al.
- Coselli J.S.,
- Volguina I.V.,
- LeMaire S.A.,
- et al.,
- for the Aortic Valve Operative Outcomes in Marfan Patients Study Group
- Yetman A.T.,
- Roosevelt G.E.,
- Veit N.,
- Everitt M.D.
- den Hartog A.W.,
- Franken R.,
- Zwinderman A.H.,
- et al.
- Habashi J.P.,
- Judge D.P.,
- Holm T.M.,
- et al.
- Groenink M.,
- den Hartog A.W.,
- Franken R.,
- et al.