The Rheumatic Mitral Valve and Repair Techniques in Children Afksendiyos Kalangos The mitral valve is the most commonly affected valve in acute and chronic rheumatic heart disease in the first and second decades of life. Pure or predominant mitral regurgitation with non-significant stenosis (mitral valve area > 1.5 cm 2 on echocardiography) is the most frequently encountered valvular dysfunction in children. In our experience, based on 428 children operated between 1993 and 2011 at our institution, functional classification based on leaflet motion assessed by echocardiography and reconfirmed peroperatively revealed pure annulus dilatation (type I) in 7% of patients, anterior leaflet prolapse (type IIa) in 33%, combination of anterior leaflet pseudoprolapse with restricted motion of the posterior leaflet (type pseudoiia/ IIIp) in 34%, and restricted anterior and posterior leaflet motion (type IIIa/p) in 26%. Patients with type III were older than those with type IIa and type pseudoiia/iiip. Different techniques can be used to repair rheumatic mitral valve lesions: prolapse of the anterior leaflet caused by chordal elongation or rupture can be treated by chordal shortening, chordal transfer, or artificial chordal replacement; restricted motion of the anterior and/or posterior leaflet can be treated by commissurotomy, splitting of the papillary muscles, resection of the secondary, or sometimes primary posterior chordae, posterior leaflet free edge suspension, leaflet thinning, and leaflet enlargement using autologous pericardium. Because mitral annulus dilatation is present in almost all patients with mitral regurgitation, concomitant ring annuloplasty offers more stability in valve repair, improving long-term outcome. The major causes for failure of rheumatic mitral valve repair are the presence of ongoing rheumatic inflammation at the time of surgery, use of inappropriate techniques, technical failures requiring early reoperation, lack of concomitant ring annuloplasty, and progression of leaflet and chordal disease further resulting in more leaflet retraction, thickening, and deformity. Freedom from reoperation depends on mitral regurgitation functional type, the type IIa and type pseudoiia/iiip having a better long-term outcome than type I and type III, in our series. In conclusion, mitral valve repair should be a preferred strategy in children with rheumatic heart disease whenever feasible, providing stable actuarial survival with fewer thromboembolic complications in a pediatric population noncompliant to anticoagulation. Semin Thorac Cardiovasc Surg Pediatr Card Surg Ann 15:80-87 2012 Elsevier Inc. All rights reserved. Rheumatic fever and rheumatic heart disease still remain a public health problem in most developing countries and are a significant cause of cardiovascular morbidity and mortality. Many affected children require surgery within 5 to 10 years of diagnosis. 1 Rheumatic heart disease results from an inflammatory process generated by an autoimmune reaction, From the University Hospital of Geneva, Division of Cardiovascular Surgery, Faculty of Medicine, University of Geneva, Switzerland. Dr Kalangos reports receiving royalties provided by the intellectual property of biodegradable ring manufactured by the Parvulus Switzerland SA. Address correspondence to Afksendiyos Kalangos, MD, PhD University Hospital of Geneva, Division of Cardiovascular Surgery, 4 Rue Gabrielle- Perret-Gentil, 1211 Geneva 14, Switzerland; E-mail: afksendyios. kalangos@hcuge.ch triggered by the group A streptococcus, against the different cardiac layers and anatomic structures. Transformation of the inflammatory process into fibrosis over time is the main histopathologic factor responsible for impairing valve function, the mitral valve being the most frequently affected of the heart valves. Valvular fibrosis can be exacerbated by recurrent inflammatory reaction during repeat episodes of rheumatic fever if left untreated. Fibrosis affects both the mitral valve tissue itself and the components of the subvalvular apparatus by increased tissue thickness, gradual retraction and deformation, and subsequent predisposition to calcification over time. The different types of resulting chronic mitral valve dysfunction are combined stenosis and regurgitation, pure regurgitation, and, less frequently, pure stenosis. In some cases, the inflammatory process itself can 80 1092-9126/12/$-see front matter 2012 Elsevier Inc. All rights reserved. doi:10.1053/j.pcsu.2012.01.013
Rheumatic mitral valve and repair techniques 81 exceptionally be the origin of acute mitral valve regurgitation by acute dilatation of the left ventricle because of carditis or by rupture of weakened primary chordae tendineae. Twenty-two out of the 428 children (5%) who underwent mitral valve repair over the past 18 years at our institution were operated on with active rheumatic carditis, as based on clinical examination, serological criteria, and macroscopic appearance of the heart at the time of surgery. 2 For all of these children, emergency surgery was required because of hemodynamic instability. Based on the detailed retrospective analysis of our rheumatic pediatric population (age 16 years) who underwent mitral valve repair over the past 18 years, this article addresses the different mechanisms responsible for acute or chronic mitral dysfunction and describes various mitral repair techniques and the corresponding long-term results of surgical repair. Mechanisms of Mitral Valve Dysfunction Based on Carpentier s Classification Type I: Annular dilatation According to the functional type of mitral valve dysfunction established by Carpentier, 3 30 out of 428 patients (7%) had type I mitral regurgitation. In pure type I dysfunction, the main mechanism is rapid progressive dilatation of the left ventricle and mitral annulus caused by inflammatory pancarditis, which also frequently alters left ventricular contractility. In some cases, during the early stage of rheumatic valve involvement, mild or moderate degrees of symmetrically restricted anterior and posterior leaflet motion very often undetected on echocardiography can create hemodynamically significant degrees of mitral regurgitation if suddenly associated with a certain degree of annular dilatation, secondary to a new episode of rheumatic pancarditis. Because the factor triggering mitral regurgitation in this condition is the sudden onset of annular dilatation, these cases are usually considered as type I dysfunction. If fibrosis progressively creates greater restricted motion of both mitral leaflets, the mechanism responsible for mitral regurgitation will be type III anterior and posterior dysfunction. If the restricted leaflet motion predominantly affects the posterior leaflet with little or no interference with the motion of the anterior leaflet, the responsible mechanism will be a combined type pseudoii anterior/type III posterior (type pseudoiia/iiip) dysfunction. Type IIa: Anterior Leaflet Prolapse One hundred forty-two patients (33%) had type II anterior leaflet (a) dysfunction. On echocardiography, in true prolapse, the anterior leaflet overrides the mitral annulus plane during systole. Causes of true anterior leaflet prolapse are elongation of anterior primary chordae tendinae or papillary muscles and rupture of anterior primary chordae tendinae. 4 The paramedian anterior chordae are most frequently affected by rupture, especially during the active inflammatory phase. In some cases, rupture of primary anterior chordae can also be caused by bacterial endocarditis, the onset of which is favored by the presence of a dysfunctional rheumatic valve. Bacterial endocarditis on a rheumatic valve can remain clinically silent, and may be mistaken for an episode of rheumatic fever and hence improperly treated, particularly because of poor health conditions in developing countries. Significant posterior leaflet prolapse was only detected in one of our cases on echocardiography and during macroscopic evaluation of the per-operative lesions, as main mechanism responsible for mitral insufficiency. Type II Anterior Leaflet Pseudoprolapse/III Posterior One hundred forty-five children (34%) had mixed-type pseudoiia (anterior leaflet pseudoprolapse and IIIp (restricted posterior leaflet motion) dysfunction. On echocardiography, contrary to pure type IIa dysfunction, in the mixed type pseudoiia/iiip, the anterior leaflet does not override the mitral annulus plane during systole, just as in true prolapse. Mitral regurgitation in this mixed functional type is caused by lack of coaptation, resulting mainly from significant restricted motion of the posterior leaflet, allowing for the anterior leaflet to move up to the annular plane during systole without overriding it. In all such cases, there is a varying degree of associated annular dilatation depending on the time interval between the onset of significant mitral regurgitation and the time of surgery. Type III: Restricted Leaflet Motion One hundred eleven cases (26%) had type III (restricted anterior and/or posterior leaflet) dysfunction. In 98 cases, the restricted motion mainly affected the closure of the leaflets during systole, resulting in lack of leaflet coaptation generating mixed stenosis and regurgitation, either predominantly regurgitant (mitral orifice area 1.5 cm 2 on echocardiography), or stenotic (mitral orifice area 1.5 cm 2 ). In the majority of such cases, associated annular dilatation contributes to the degree of mitral regurgitation by aggravating the lack of coaptation between both mitral leaflets, as in the mixed type pseudoiia/iiip dysfunction. When restricted motion mainly affects leaflet opening during diastole because of fusion of both commissures and retraction of commissural chordae tendinae, or fusion of both papillary muscles to the ventricular surface of the commissures, pure mitral valve stenosis will be the predominant dysfunction. In our series, 13 cases (3%) had pure mitral stenosis. Type III patients were older than type II and type pseudoiia/iiip patients (13 1.2 years vs 9 1.1 years, respectively). Repair Techniques Correcting Annular Dilatation Mitral Annuloplasty Annular dilatation and deformity are corrected by ring annuloplasty, using a ring size based on the surface of the anterior leaflet or on the inter-trigonal distance. The major disadvantage of traditional rings is the lack of pediatric sizes under 24, because small sizes are potentially stenotic in a growing child. Technical alternatives are either the enlargement of the retracted posterior or anterior leaflet using a pericardial patch, or the insertion of a pediatric biodegradable ring. 5 Leaflet enlargement may preclude the need for a ring or may allow for insertion of a larger ring. Enlargement of the posterior leaflet is commonly performed in rheumatic mitral valve disease, while indications for enlargement of the anterior leaflet are exceptional. In our retro-
82 A. Kalangos spective study comparing the rigid Carpentier Edwards (CE) (Edwards Life Sciences, Irvine, CA) to the 3-dimensionally flexible biodegradable ring (Parvulus SA, Lonay, Switzerland), both used in pediatric rheumatic mitral valve repair, the transmitral gradient at discharge was significantly lower in cases with a biodegradable ring than those with a CE ring. There was a significant increase in the mean transmitral gradient during the first postoperative year of implantation in cases with a CE ring, whereas no significant increase of this gradient was observed in cases with a biodegradable ring. Lower gradients were present in cases with a biodegradable ring, be it in cases of pure mitral regurgitation or combined mitral stenosis and regurgitation. 6 The biodegradable ring group also had better left ventricular shortening fraction at 1 week postoperatively, as compared with the CE group. Except in patients with pure mitral stenosis, mitral annulus dilatation and deformity are present in the majority of rheumatic mitral valve disease patients, and concomitant mitral annuloplasty is mandatory for correction of annulus dilatation, native mitral annulus remodeling, and for long-term durability of repair. In cases of pure mitral stenosis with no obvious annular dilatation, concomitant mitral annuloplasty prevents leaks through split commissures by improving the coaptation surface between both mitral leaflets. Correcting Anterior Leaflet Prolapse Anterior leaflet prolapse caused by primary chordae tendinae elongation or rupture and elongation of papillary muscle heads can be corrected by chordal shortening, chordal transfer techniques, or the use of artificial cords. Chordal shortening and transfer techniques are chosen according to the degree of thickness and retraction of primary anterior chordae tendineae. If the thickness of the elongated primary chordae is mildly or moderately increased with no significant loss of flexibility, any of the following techniques might be appropriate: the split & tuck-in technique (Fig. 1A), chordal shortening at the free edge of the corresponding anterior leaflet segment (Fig. 1B), sliding shortening plasty of the elongated papillary muscle head (Fig. 1C), secondary chordal transfer (in the absence of retraction) (Fig. 1D), and/or the use of artificial cords (Fig. 1E). In the active carditis, we try to avoid chordal shortening and transfer techniques because of tissue fragility induced by the inflammatory process, to avoid compromising durability of repair. In such cases, use of artificial cords reinforced with small pericardial pledgets at both implantation sites (papillary muscle and free edge of the anterior leaflet) can potentially decrease the rate of early repair failure. If the anterior chordae tendinae are thick, stiff, and retracted in length, and the anterior leaflet prolapse mainly caused by elongation of the heads of the anterior or posterior papillary muscle on which thick, short primary anterior chordae tendinae are inserted, either dislocating the elongated head by suture fixation onto an adjacent site (Fig. 1F) or doing a sliding plasty of the papillary muscle head can be effective in correcting prolapse. In this situation, the split & tuck-in technique, chordal shortening at the free edge of the prolapsing anterior leaflet segment, or use of short artificial cords are not appropriate because they can convert prolapse into significant restricted motion by further reducing the mobility of the corresponding segment of the anterior leaflet. In cases where secondary chordae tendineae are more retracted than the primary chordae and result in true prolapse of the free edge of the anterior leaflet in a V deformity, resection of these secondary anterior chordae tendineae can correct the true anterior prolapse as we have already reported. 4 Correcting Anterior Leaflet Pseudoprolapse and Restricted Posterior Leaflet Motion In type pseudoiia/iiip cases, mitral valve repair can be performed using the below-described techniques either individually or in combination. Repair techniques should be aimed at increasing mobility of the restricted posterior leaflet segment by both resecting secondary and sometimes retracted primary chordae (Fig. 1G) and shaving the thickened posterior leaflet segment (Fig. 1H). Papillary muscle splitting at several sites to provide greater mobility to the retracted primary chordae tendineae is also recommended. In case of partial commissural fusion, commissurotomy combined with papillary muscle splitting can help increase mobility of the anterior and posterior commissural areas. Commissural splitting should respect the slight upward orientation of the physiologic commissural line, leave enough commissural primary chordae tendineae on both sides of the commissurotomy, and stop 1 to 2 mm before the insertion of the commissural valve tissue to the native annulus (Fig. 1I). Other repair techniques aimed at correcting the pseudoprolapse of the anterior leaflet segment can also improve the coaptation surface between the restricted posterior leaflet segment and the contra-lateral anterior one. Retracted posterior leaflet height increase. Detaching the retracted posterior leaflet segment from the native annulus and plicating the detached annular and leaflet segments longitudinally with interrupted stitches increase the height of the retracted posterior leaflet by partially reducing the dilated mitral orifice (Fig. 1J). 7 In the majority of type pseudoiia/iiip cases, retraction involves the P2 and P3 segments of the posterior leaflet, creating a V deformity between them. In case of extensive retraction between the two commissures, enlargement of a retracted posterior leaflet using a pericardial patch can also be useful in enhancing the coaptation surface between the two leaflets (Fig. 1K). 8 Lower anterior leaflet coaptation point. Lowering the coaptation point of the free edge of the anterior leaflet segment to the level of the corresponding coaptation point of the retracted posterior leaflet can be performed using chordal shortening techniques or artificial cords. The disadvantage of this technique is conversion from pseudoprolapse to restriction of the anterior segment, the degree of which depends on the severity of the restricted motion of the posterior leaflet. In case of severely restricted motion of the posterior segment, resection of retracted primary and secondary posterior chordae and replacement with artificial chords seems to re-establish a physiologic coaptation level without any intervention on the corresponding opposite anterior leaflet segment. Posterior leaflet free edge suspension. Suspending the free edge of the retracted posterior segment to the opposite anterior annulus or mitral annuloplasty ring is done by bringing the free
Rheumatic mitral valve and repair techniques 83 Figure 1 (A) The split & tuck-in technique. (B) Chordal shortening at the free edge of the prolapsing anterior leaflet segment. (C) Sliding shortening plasty of the elongated chordae tendinae. (Figure continues.) edge of the posterior leaflet up to the coaptation point of the corresponding opposite anterior leaflet segment (Fig. 1L). 9,10 Correcting Anterior and Posterior Leaflets Restricted Motion In type III dysfunction resulting in pure stenosis, anterior and posterior commissurotomy combined with anterior and posterior papillary muscle splitting should be performed. The above explained technical principles concerning commissurotomy should be taken into account. In papillary muscle splitting, two thirds of the thickness of the split papillary muscle should be left on the anterior side and one third on the posterior side; the trench should be extended down toward the base of the papillary muscle. In type III a p cases, commissurotomy, papillary muscle splitting, resection of anterior and posterior secondary chordae tendineae, and shaving or peeling the thickened leaflet
84 A. Kalangos Figure 1 (D) Secondary chrodal transfer to the free edge of the prolapsing anterior leaflet segment. (E) Use of artificial cord. (F) Dislocation of the elongated head by suture fixation into an adjacent site of the papillary muscle. (G) Resection of retracted posterior and anterior leaflet secondary cordae tendinae. (Figure continues.) tissue (Fig. 1M) can considerably increase mobility. Posterior and/or anterior leaflet extension using a pericardial patch according to the site and degree of retraction can allow for implantation of a bigger annuloplasty ring and thereby minimize the degree of residual stenosis. 8 Causes for Reoperation Early and Midterm Reoperation The main reasons for early reoperation (within the first 6 postoperative months) are usually errors in judgment with attempts to repair an irreparable valve, suboptimal repair at the time of initial surgery because of inadequate analysis of different valvular and subvalvular lesions, and inappropriate selection or application of certain repair techniques for varying degrees of annular dilatation, leaflet prolapse, or restricted leaflet motion. Failure to perform concomitant annuloplasty at the time of initial surgery has been reported to be a predictive factor for early reoperation. 11 Decisional mistakes concerning rheumatic valve repair at the time of initial surgery especially occur in the presence of extensive mixed fibrotic lesions, whether predominantly regurgitant or stenotic (type III). In some of these cases, even if the surgeon uses the appropriate techniques to effectively mobilize the
Rheumatic mitral valve and repair techniques 85 Figure 1 (H) Shaving of the thickened posterior leaflet. (I) Anterior and posterior commissural splitting. (J) Retracted posterior leaflet height increases at P2-3 segments as described by Izumoto et al. 7 (K) Enlargement of the retracted posterior leaflet. (L) Suspension of the retracted free edge of the posterior leaflet segment to the opposite anterior mitral annulus. (M) Peeling of the thickened anterior leaflet. anterior and posterior leaflets and subvalvular apparatus as much as possible, repair can result in significantly high postrepair mean transmitral gradients ( 8 mmhg), in part because of the reductive effect of concomitant annuloplasty despite appropriate ring size. In our experience, this specific subset of patients who are older than the type II, type pseudoiia/iiip patients, are more predisposed to undergo reoperation over the 10 postoperative years, freedom from reoperation at 10 years being 72.9% for type III compared with 94% and 91% for type II and pseudo IIa/IIIp patients, respectively. Furthermore, ongoing rheumatic activity at the time of surgery is another significant predictive factor for early and midterm reoperation. 12-15 Routine antibiotic prophylaxis after surgery can limit the progression of structural deterioration of the mitral valve by preventing the new episodes of leaflet inflammation. Technical failures caused by the fragility of inflammatory valvular and subvalvular structures in active rheumatic disease can result in the need for early reoperation. The most frequent mechanical causes for technical failure are dehiscence of the annuloplasty ring, rupture of shortened inflammatory chordae tendinae, suture dehiscence of transferred chordae, dislocated papillary heads, or elongated chordae buried into a papillary muscle trench, as well as valvular or subvalvular tissue tears at suture attachment sites. Transformation of inflammation into progressive fibrosis and excessive retraction can also cause post-repair valve failure at midterm (between the 6 th postoperative month and 5 years) by reducing leaflet mobility and coaptation surface. In our
86 A. Kalangos Figure 2 Freedom from reoperation according to the different types of mitral valve dysfunction. series, freedom from reoperation in patients operated during the active phase of rheumatic disease was 45% at 5 years. Eight patients were reoperated on within the first 6 postoperative months, and nine patients between 6 months and 5 years after surgery because of significant recurrent mitral regurgitation. Late Reoperation Progression of leaflet retraction after initial repair, complicated sometimes by valvular and subvalvular calcification over time, is the most important cause of late reoperation in the rheumatic pediatric population. In our series, the severity of fibrous retraction (more prominent in type III), mixed lesions resulting in a mitral orifice area under 1.5 cm 2 at the time of initial repair, and recurrent rheumatic fever episodes were risk factors for late reoperation. Patients with type III lesions were older than those with type I, II, and pseudoiia/ IIIp patients. Endocarditis of a repaired rheumatic mitral valve can lead to new lesions (usually chordal rupture), and thereby accelerate the ongoing destructive process of chronic rheumatic valve disease. Fifty-one out of 428 cases were reoperated on between the 5 th and 18 th postoperative years. In 48 out of 51 cases, the cause for reoperation was progression of fibrous retraction, and in the remaining three cases, the sudden onset of bacterial endocarditis. Long-Term Durability of Rheumatic Mitral Valve Repair Fibrosis, retraction, deformity, and the progressive deterioration of valvular and subvalvular lesions with time in rheumatic mitral valve disease increase the risk of reoperation, 15 as compared with degenerative mitral valve disease. Although correct analysis of lesions and appropriate repair techniques limit the risk of early and midterm reoperation, long-term results depend on the initial type of valvular dysfunction and the rate of progressive structural deterioration. In our series, mitral valve repair in type IIa and type pseudoiia/iiip had better long-term outcome as compared with type I and type III dysfunctional classes (P.05; Fig. 2). In the type I group, the majority of cases were operated on during the acute stage of rheumatic pancarditis, which explains the lower rate of freedom from reoperation as compared with the other types. Similar results have been reported by Chauvaud et al, 15 with freedom from reoperation at 20 years being 65% in type pseudoiia/iiip, 63% in type II, and 46% in type III. The difference in reoperation rate between their type II (1.9% patients/year) and type III (2.7% patients/ year) dysfunctional class was statistically significant. 15 In our series, survival was 97% and freedom from any valve-related event (including hospital death, reoperation, late death, and thromboembolic events) was 76% at 17 years. Thromboembolism rate was very low (0.2% patients/year), although the incidence of atrial fibrillation (2% patients/year) remained high over the follow-up period. Conclusion Despite the greater rate of reoperation as compared with degenerative mitral valve disease, rheumatic mitral valve repair provides stable actuarial survival with fewer thromboembolic complications in a pediatric population noncompliant to anticoagulation. Similarly, despite the greater rate of reoperation as compared with mitral valve replacement with mechanical prosthesis, rheumatic mitral valve repair as a nonthrombogenic surgical procedure remains an attractive alternative to mitral valve replacement whenever feasible.
Rheumatic mitral valve and repair techniques 87 Acknowledgment The author would like to thank Drs D. Vala, S. Cherian, M. Cikirikcioglu, and N. Murith for their invaluable editorial assistance. References 1. Kaplan E. Recent epidemiology of group A streptococcal infections in North America and abroad: a review. Pediatrics 1996;97(suppl):945-948 2. Committee on rheumatic fever and bacterial endocarditis of the American Heart Association. Jones criteria (revised) for guidance in diagnosis of rheumatic fever. Circulation 1984;69:203A-208A 3. Carpentier A. Cardiac valve surgery - The French correction. J Thorac Cardiovasc Surg 1983;86:323-337 4. Kalangos A, Beghetti M, Vala D, et al. Anterior mitral leaflet prolapse as a primary cause of pure rheumatic mitral insufficiency. Ann Thorac Surg 2000;69:755-761 5. Kalangos A, Sierra J, Vala D, et al. Annuloplasty for valve repair with a new biodegradable ring; an experimental study. J Heart Valve Dis 2006; 15:783-790 6. Kalangos A, Christenson J, Beghetti M, et al. Mitral valve repair for rheumatic valve disease in children: midterm results and impact of the use of a biodegradable mitral ring. Ann Thorac Surg 2008;86:161-169 7. Izumoto H, Kiyoyuki E, Ohuchi S, et al. Mitral technique of posterior mitral annuloplasty. J Card Surg 1999;14:270-272 8. Chauvaud S, Jebara V, Chachques JC, et al. Valve extension with glutaraldehyde-preserved autologous pericardium. Results in mitral valve repair J Thorac Cardiovas Surg 1991;102:171-178 9. Sierra J, Christenson JT, Lahlaidi NH, et al. Mitral valve insufficiencysuspension of the posterior leaflet to the mitral annuloplasty ring. Ann Thorac Surg 2007;85:e6-e8 10. Myers P, Christenson J, Cikirikcioglu M, et al. Leaflet suspension to the contralateral annulus to address restriction or tethering-induced mitral and tricuspid regurgitation in children. Results of a case-control study. J Thorac Cardiovasc Surg 2010;140:1110-1116 11. Yau TM, El-Ghoneimi YA, Armstrong S, et al. Mitral valve repair and replacement for rheumatic disease. J Thorac Cardiovasc Surg 2000;119:53-60 12. Duran CMG, Gometza B, De Vol EB: Valve repair in rheumatic mitral disease. Circulation 1991;84(suppl III):III-125-III132 13. Duran CMG, Gometza B, Balasundaram S, et al. A feasibility study of valve repair in rheumatic mitral regurgitation. Eur Heart J 1991; 12(suppl B):34-38 14. Skoularigis J, Sinovich V, Joubert G, et al. Evaluation of the long-term results of mitral valve repair in 254 young patients with rheumatic mitral regurgitation. Circulation 1994;90(part II):II-167-II-174 15. Chauvaud S, Fuzellier JF, Berrebi A, et al. Long-term (29 years) results of reconstructive surgery in rheumatic mitral valve insufficiency. Circulation 2001;104(suppl I):I-12-I-15