Bài báo khoa học: Valve repair improves the outcome of surgery for mitral regurgitation: a multivariate analysis

Mitral valve repair has been suggested as providing a better postoperative outcome than valve replacement for mitral regurgitation, but this impression has been obscured by differences in baseline characteristics and has not been confirmed in multivariate analyses.

Valve Repair Improves the Outcome of

Surgery for Mitral Regurgitation

A Multivariate Analysis

Maurice Enriquez-Sarano, Hartzell V Schaff, Thomas A Orszulak, A Jamil Tajik, Kent R Bailey, Robert L Frye

https://doi.org/10.1161/01.CIR.91.4.1022

Circulation 1995;91:1022-1028

Originally published February 15, 1995

 Article

 Figures & Tables

 Info & Metrics

 eLetters

Abstract

Background Mitral valve repair has been suggested as providing a better

postoperative outcome than valve replacement for mitral regurgitation, but this

impression has been obscured by differences in baseline characteristics and has not been confirmed in multivariate analyses

Methods and Results The outcomes in 195 patients with valve repair and 214 with replacement for organic mitral regurgitation were compared using multivariate

analysis All patients had preoperative echocardiographic assessment of left

ventricular function Before surgery, patients with valve repair were less symptomatic than those with replacement (42% in New York Heart Association functional class I

or II versus 24%, respectively; P=.001), had less atrial fibrillation (41% versus

53%; P=.017), and had a better ejection fraction (63±9% versus 60±12%, P=.016) After valve repair, compared with valve replacement, overall survival at 10 years was 68±6% versus 52±4% (P=.0004), overall operative mortality was 2.6% versus 10.3% (P=.002), operative mortality in patients under age 75 was 1.3% versus 5.7%

(P=.036), and late survival (in operative survivors) at 10 years was 69±6% versus 58±5% (P=.018) Late survival after valve repair was not different from expected survival After surgery, ejection fraction decreased significantly in both groups but was higher after valve repair (P=.001) Multivariate analysis indicated an

independent beneficial effect of valve repair on overall survival (hazard ratio,

0.39; P=.00001), operative mortality (odds ratio, 0.27; P=.026), late survival (hazard ratio, 0.44; P=.001), and postoperative ejection fraction (P=.001)

Conclusions Valve repair significantly improves postoperative outcome in patients with mitral regurgitation and should be the preferred mode of surgical correction The

low operative mortality is an incentive for early surgery before ventricular dysfunction occurs

 mitral valve

 prognosis

 ventricles

 myocardium

 surgery

Since the early days of cardiac surgery,1 2 3 4 the repair of regurgitant mitral valves has been pioneered with the perception that maintaining the normal architecture of the mitral valve is beneficial to the patient.5 However, the clinical suggestion that valve repair, compared with valve replacement, improves postoperative

survival6 7 8 9 10 11 and left ventricular function1213 14 has been obscured by

differences in baseline characteristics among the patients treated by the two

methods14 15 16 17 and has not been confirmed by multivariate analysis.15 16Moreover, comparison between repair and replacement has been hindered by the limited

number of repairs involved15 16 and by the failure to consider important predictors of outcome such as preoperative left ventricular function.18 19 20 Thus, the clinical

impact of valve repair has remained uncertain as has, therefore, its potential as the preferred surgical method of correction of mitral regurgitation Accordingly, we

examined the outcome after valve repair and valve replacement in patients with mitral regurgitation in whom preoperative left ventricular function could be assessed echocardiographically, having hypothesized that mitral valve repair improves

operative mortality, late survival, and postoperative residual left ventricular function compared with mitral valve replacement

Methods

The present study was based on a review of our experience with valve repair and valve replacement for the surgical correction of mitral regurgitation

Included in the study were patients who had repair or replacement of the mitral valve performed between January 1, 1980, and December 31, 1989; who had acquired organic mitral regurgitation as defined by echocardiographic and surgical

assessment; and who had preoperative (within 6 months before surgery)

echocardiography allowing at least the assessment of the left ventricular ejection fraction

Excluded were patients who had previous mitral regurgitation surgery, who had had previous or associated aortic or tricuspid valve replacement (patients with tricuspid valve repair were not excluded), and who had mitral regurgitation of ischemic or functional cause However, patients with associated coronary artery bypass graft surgery were not excluded

During that period, 2183 patients had mitral valve surgery at our institution Of these,

654 had had isolated mitral regurgitation, and 409 of that group had had organic mitral regurgitation and a preoperative echocardiogram Of the 409, 195 had valve repair and 214 had valve replacement The clinical follow-up was 98% complete up

to 1992 or death Complications were defined as previously described.21

Echocardiographic Examination

An echocardiographic examination was performed at a mean of 24±31 days before surgery and was analyzed as described previously.22 The left ventricular diameters and wall thicknesses were measured, and the ratio of diameter to wall thickness was calculated at end diastole and systole Ejection fraction was estimated by the

consensus of two observers using all parasternal and apical views of the left

ventricle23 in all cases This estimation was combined with calculations24 using left ventricular diameters in 322 patients and was used in isolation in 87 patients The values of ejection fraction used in the present analysis were electronically

transferred, as noted in the original report, without alterations Left atrial diameter was measured in systole With the use of noninvasive blood pressure

determinations, we estimated the end-systolic pressure,25 and the end-systolic wall stress was calculated as previously described.26

Left Ventricular Angiography

Left ventricular ejection fraction was measured by angiography in 219 patients Correlations with echocardiographic values were acceptable for routine

measurements (r=.61 overall, r=.70 in patients without coronary disease)

Surgical Procedure

Surgical repair of the mitral valve involved subvalvular, valvular (mostly resection or plication), and annular interventions as previously described.10 27 28 The lesions repaired compared with those replaced involved degenerative changes in 170 versus

141 patients and were endocarditic in 11 versus 29 patients, rheumatic in 8 versus

34 patients, and miscellaneous lesions in 6 versus 10 patients, respectively

(P<.0001) A valvular prolapse was repaired in 183 patients, involving the posterior leaflet in 128, the anterior leaflet in 20, and both leaflets in 35 Ruptured chordae were present in 145 of the 195 repair procedures The prostheses used in valve replacement were Starr-Edwards (84), disk prostheses (18), St Jude (8), Ionescu-Shiley (28), Carpentier-Edwards (72), and Hancock (4) Coronary artery bypass grafting was performed in 99 patients—57% of 195 patients (29%) with valve repair and 42 of 214 patients (20%) with valve replacement (P=.024)

Statistical Analysis

Group statistics were expressed as mean±1 SD Group comparisons were

performed with a standard ttest or χ2

test when appropriate The cumulative probability of survival was estimated by the Kaplan-Meier method The survival curves of the patients were compared in each group with the expected survival

based on age- and sex-matched actuarial data from the 1980 US white population

and tested with the one-sample log-rank test Unadjusted group survival

comparisons were based on the two-sample log-rank test Multivariate analysis for each end point was performed stepwise for clinical and echocardiographic

parameters Year of surgery was included to allow for the possibility of confounding treatment strategy with temporal improvements Then, the operative variables

(method of correction of the mitral regurgitation, expressed as repair or replacement

of the valve, and the associated coronary artery bypass graft surgery procedures) were added to the models A similar stepwise procedure was used, incorporating the angiographic ejection fraction instead of its echocardiographic counterpart The end points were overall survival, operative mortality, late survival of operative survivors, and postoperative ejection fraction Adjusted group survival comparisons were done with the Cox proportional hazards models Operative mortality comparisons were adjusted in a multivariate logistic analysis Postoperative ejection fraction

comparisons were adjusted via multiple linear regression The variables included in the multivariate analyses are listed in Table 1⇓ The changes in ejection fraction from the preoperative to the postoperative period were compared with a paired t test A value of P<.05 was considered significant

 View inline

 View popup

Table 1.

Parameters Included in Multivariate Analysis

Results

The strategy of the analysis was to compare the valve repair group with the valve replacement group regarding baseline characteristics of the patients, overall survival, operative survival, late survival, postoperative ejection fraction, and other end points

Baseline Characteristics

The preoperative baseline characteristics of the two groups are summarized in Table 2⇓ The results show multiple significant differences, mostly consistent with a better prognosis in patients with valve repair

 View inline

 View popup

Table 2.

Baseline Patient Characteristics

Overall Survival

The overall survival after valve repair was significantly superior to that after valve replacement (P=.0004), with 5- and 10-year survival rates, respectively, of 83±3% and 68±6% (repair) compared with 69±3% and 52±4% (replacement) The survival curves of the two groups and the expected survival are presented in Fig 1⇓ At 10 years, the survival after valve repair represents 100% (P=.64) of the expected

survival, and after replacement it is 77% (P=.0001) of expected With multivariate analysis, valve repair was an independent favorable predictor of overall survival (P=.00001; hazard ratio, 0.39; 95% confidence interval, 0.26 to 0.60) (Table 3⇓) The survival after valve repair was better than after valve replacement also when

stratified in patients with (at 6 years, 74±6% and 34±8%; P=.0002) and without (at 5 years, 87±3% and 73±3%; at 10 years, 73±7% and 61±5%; P=.006) coronary artery bypass graft surgery (Fig 2⇓) The multivariate model using angiographic ejection fraction confirmed the independent favorable impact of valve repair (P=.0018) The other predictors were age (P=.0001), coronary artery bypass surgery (P=.0026), creatinine level (P=.03), and angiographic ejection fraction (P=.08)

 Download figure

 Open in new tab

 Download powerpoint

Figure 1.

Plot of overall survival compared for valve repair and valve replacement groups (P=.0004) The expected survival rate for the total of 409 patients is also

represented The numbers at the bottom indicate, for each interval, the number of patients at risk

 Download figure

 Open in new tab

 Download powerpoint

Figure 2.

Plots of overall survival compared for repair and replacement groups for patients who had (left) or did not have (right) associated coronary artery bypass graft surgery (CABG)

 View inline

 View popup

Table 3.

Multivariate Analysis End Points

Operative Mortality

Operative death was defined as a death occurring within 1 month of surgery or during the same hospitalization Operative death occurred in 5 of the 195 patients with valve repair (2.6%) compared with 22 of 214 patients with valve replacement (10.3%) (P=.002) (Table 4⇓) With multivariate analysis, valve repair was an

independent favorable predictor of operative mortality (P=.026; odds ratio, 0.27; 95% confidence interval, 0.09 to 0.86) (Table 3⇑) Operative mortality in relation to age is presented in Table 4⇓

 View inline

 View popup

Table 4.

Operative Mortality

Late Survival

Late survival was analyzed in 375 patients, excluding 27 operative deaths and 7 patients lost to follow-up The causes of late death in valve repair and replacement were, respectively, coronary disease in 4 and 10 patients, left ventricular dysfunction (through sudden death or congestive heart failure) in 9 and 27, valvular

complications in 8 and 17, noncardiac causes in 6 and 8, and unknown causes in 2 and 4 Late survival was significantly better after valve repair than after valve

replacement (P=.018) at 5 and 10 years: 85±3% and 69±6% for repair and 77±3% and 58±5% for replacement, respectively At 10 years, the late survival of the valve repair group represented 100% (P=.77) of the expected survival as compared with 83% (P<.0001) in the replacement group The late survival curves of the repair and replacement groups are presented in Fig 3⇓ With multivariate analysis, valve repair was an independent favorable determinant of late survival (P=.001; hazard ratio, 0.44; 95% confidence interval, 0.27 to 0.73) (Table 3⇑) The incidence of left

ventricular dysfunction-related deaths was lower in the valve repair than in the valve replacement group (log rank P=.036)

 Download figure

 Open in new tab

 Download powerpoint

Figure 3.

Plots of late survival (in operative survivors) of patients with valve replacement (left) and valve repair (right) compared with their expected survival Note that in patients

with valve repair, there is no difference from the expected survival, whereas in

patients with valve replacement, the survival is significantly lower than expected

Left Ventricular Function

The postoperative left ventricular ejection fraction was measured in 315 patients 2.7±2.9 years after surgery—177 with valve repair and 138 with valve replacement Preoperative characteristics were not different between patients with and those without postoperative echocardiograms, showing that those with postoperative

ejection fraction were representative of the overall population regardless of whether

a repair or replacement was performed Ejection fraction remained stable with time in patients with multiple postoperative echocardiograms Overall, there was a decrease

in ejection fraction after surgery, from 62±10% to 52±13% (P=.0001) A significant decrease in the ejection fraction was observed in both groups: for valve repair,

63±9% to 54±11% (P=.0001), and for valve replacement, 60±12% to 49±15%

(P=.0001) Ejection fraction was significantly higher in the valve repair group than in the valve replacement group both before (P=.016) and after surgery (P=.001) (Fig 4⇓) Although absolute differences were modest with multivariate analysis, valve repair was an independent predictor of higher postoperative ejection fraction

(P=.001; odds ratio of ejection fraction ≥50%, 2.72; 95% confidence interval, 1.43 to 5.16) (Table 3⇑)

 Download figure

 Open in new tab

 Download powerpoint

Figure 4.

Plot of changes in ejection fraction from the preoperative echocardiogram (Pre) to the postoperative echocardiogram (Post) in 177 patients with valve repair and 138 patients with valve replacement

Other End Points

There was no significant difference between valve repair and valve replacement groups regarding the need for reoperation (free of reoperation: at 5 years, 90±2.5% and 93±2%; at 10 years, 75±10% and 80±6%, respectively; P=.47) (Fig 5⇓), the incidence of thromboembolism (at 10 years, 68±8% and 70±4% were free of

thromboembolism, respectively; P=NS), and the incidence of bacterial endocarditis (at 10 years, 92±5% and 97±1% were free of endocarditis, respectively; P=NS) However, patients with valve repair had a lower incidence of coumadin treatment (n=64, or 33%, compared with n=141, or 66%, in valve replacement; P=.0001) and of significant hemorrhage (at 10 years, 88±3% in the valve repair group and 73±4% in the valve replacement group were free of significant hemorrhage; P=.002)

 Download figure

 Open in new tab

 Download powerpoint

Figure 5.

Plot of freedom from reoperation in valve repair and replacement groups No

significant difference is observed

Discussion

In the present series of 195 valve repairs and 214 valve replacements performed for organic mitral regurgitation, there are differences in baseline characteristics between the two groups, but multivariate analysis confirms that valve repair is an independent

determinant of improved operative and late survival and of improved left ventricular function

Beneficial Results of Valve Repair Compared With Valve Replacement

In patients with mitral stenosis, conservative surgery—that is, commissurotomy— provides both low operative risk and excellent late survival, with a low incidence of valve-related complications.28 These results are at variance with those obtained with valve replacement.29Thus, in this setting, conservative surgery, if feasible, has

always been the preferred option Conservative surgery (valve repair) has also been attempted for mitral regurgitation since the early days of cardiac surgery,1 2 3 4 but the results have been more inconsistent30 31 32 33because the lesions of mitral

regurgitation are more complex and difficult to correct.34 The techniques of repair of regurgitant lesions have progressively improved,35 36 and with the decrease in

rheumatic valve disease37 they are now widely applied in the large population with degenerative mitral valve disease.10 37 38 39 40 The durability of valve repair also has been well demonstrated.41 42 43

However, valve repair can be difficult,44 and its superiority to prosthetic replacement has not been firmly established15 16 44 45 because of small sample sizes45 and of differences in baseline characteristics of patients14 15 16 17 that suggest that surgery was performed at an earlier stage in valve repair than in valve replacement It has remained uncertain, therefore, whether the observed differences in outcome are related to the method of correction of the mitral regurgitation or to a more favorable clinical situation existing before surgery Matching identical patients who have had either valve repair or replacement has been attempted,19 45but this method is

selective and has its drawbacks Thus, the best way of addressing the issue of

nonuniformity of the groups is to perform multivariate analyses that include the most important prognostic indicators When this type of analysis was performed in other studies, however, it did not show a significant independent prognostic impact of valve repair.15 16 In contrast, the present study demonstrates that the method of correction is indeed an independent determinant of the observed differences in survival after valve repair in comparison with valve replacement The large number

of repairs, the homogeneity of the patient population, and the preoperative

assessment of left ventricular function probably contribute to the statistical power of the present series

Mechanism of Improved Outcome

The lower operative mortality after valve repair in the present study is consistent with previous observations,9 but the mechanism has not been fully understood However, since left ventricular dysfunction is the major cause of late death after mitral valve surgery,46 the lower mortality rate in patients with valve repair is readily

understandable: after valve repair, postoperative ejection fraction is significantly higher than after valve replacement, and the incidence of death due to left ventricular dysfunction is reduced Although this phenomenon is in part related to a better

preoperative function, it also is an intrinsic effect of valve repair Improvement in left ventricular function was previously suggested, mainly on the basis of small

series,13 14 but in the present study, postoperative echocardiograms were available

in 315 of the 409 patients to conclusively confirm this improvement