Paravalvular leakage after mitral valve replacement: improved long-term survival with aggressive surgery?

Background: Following mitral valve replacement, surgical closure of paraval vular leaks is usually advised in severely symptomatic patients and in thos e requiring blood transfusions for persisting haemolysis. However, the long -term prognosis of less symptomatic patients or those not needing blood tra nsfusions is unknown. Methods: Between 1987 and 1997, we observed 96 patien ts with mitral paravalvular leakage. A paraprosthetic leak was diagnosed af ter a median time of 119 days (range: 1 day-23 years) after primary mitral valve replacement. During an average follow-up of 5 years (range: 1-23 year s), 50/96 patients were referred for surgical closure. Results: Compared wi th patients who received conservative treatment, those referred for surgery had a significantly lower mean preoperative haematocrit (P = 0.002) with a higher proportion of patients being in the NYHA class III/IV (P = 0.03). A ge, gender, left ventricular function and number and size of leaks did not differ between the groups. The 30-day postoperative mortality for valve reo peration wets 6% (3/50); during follow-up three further patients died, resu lting in an overall mortality rate of 12%, In the group treated conservativ ely there was a mortality rate of 26% (12/46). Thus, the actuarial survival for patients referred for surgery was 98, 90 and 88% after 1, 5 and 10 yea rs, compared with 90, 75 and 68% For patients treated conservatively (long- rank P = 0.03). In addition, there was a significant increase in mean haema tocrit levels (P = 0.0001) and an improvement in NYHA class III/IV symptoms (P = 0.002), vertigo (P = 0.001) and fatigue (P = 0.001) after surgery. Co nclusions: Following mitral valve replacement. a more aggressive surgical t reatment is recommended for patients with paraprosthetic leaks. Surgery sho uld be offered to less symptomatic patients, as well as those not requiring blood transfusion. (C) 2000 Elsevier Science B.V. All rights reserved.