Cerebral Ischemic Events After Diagnosis of Mitral Valve Prolapse
A Community-Based Study of Incidence and Predictive Factors
Background and Purpose— Association of mitral valve prolapse (MVP) with ischemic neurological events (INEs) is uncertain.
Methods— In the community of Olmsted County (Minn), we identified all MVP diagnosed (1989 to 1998) in patients in sinus rhythm with no prior history of INE. We measured INE rates and compared them with expected rates in our community to define the excess risk of INE.
Results— Among 777 eligible subjects (age, 49±20 years; 66% female; follow-up, 5.5±3.0 years), 30 patients had at least 1 INE during follow-up (at 10 years, 7±1%). Compared with expected INEs in the same community, subjects with MVP showed excess risk of lifetime INE (relative risk [RR], 2.2; 95% CI, 1.5 to 3.2; P<0.001) and during follow-up under purely medical management (RR, 1.8; 95% CI, 1.1 to 2.8; P=0.009). Independent determinants of INE were older age (RR, 1.08 per year; 95% CI, 1.04 to 1.11; P<0.001), mitral thickening (RR, 3.2; 95% CI, 1.4 to 7.4; P=0.008), atrial fibrillation (AFib) during follow-up (RR, 4.3; 95% CI, 1.9 to 10.0; P<0.001), and need for cardiac surgery (RR, 2.5; 95% CI, 1.1 to 5.8; P=0.03). INE 10-year rates were low in patients <50 years of age (0.4±0.4%, P=0.60 versus expected) but were excessive in patients >50 years of age (16±3%, P<0.001 versus expected) or with thickened leaflets (7±2%, P<0.001 versus expected). Predictors of follow-up AFib were age, mitral regurgitation, and left atrium diameter (all P<0.01).
Conclusions— In the community, subjects with MVP display a lifetime excess rate of INE compared with expected. Clinical (older age) and echocardiographic (leaflets thickening) characteristics define patients with MVP at high risk for INE, and subsequent AFib or need for cardiac surgery, both related to the degree of mitral regurgitation, increase the risk of INE.
Mitral valve prolapse (MVP) has been considered a potential source of ischemic neurological events (INEs) since thrombi adherent to the prolapsing mitral leaflets were anatomically described.1,2⇓ However, the relationship between MVP and INEs remains controversial. Reports of high INE risk among patients with MVP,3 mainly among young patients,4 contrast with other reports of intermediate5 and low rates.6–9⇓⇓⇓ Recent publications cast doubt on MVP-associated INE risk,10 particularly in young patients.11 The discordances between these data and current guidelines12 confuse the management of patients with MVP. Estimating the risk of INEs in MVP is particularly challenging because it has been inferred primarily from case-control studies4,11,13,14⇓⇓⇓ in which inhomogeneous cases and controls likely explain wide discordances.15 Changes in MVP echocardiographic diagnostic criteria16 confounded the risk analysis and
See Editorial Comment, page 1345
question the relevance of older reports.3,4,6,7,17,18⇓⇓⇓⇓⇓ Furthermore, analysis of past complications before MVP diagnosis,2,3,5,10⇓⇓⇓ as opposed to postdiagnosis events, prevented determination of rates and predictors of INE. However, 3 major limitations have prohibited a more suitable approach to this problem. First, low INE rates19 require very large populations for appropriate statistical power. Second, studies of patients referred to tertiary-care centers2–9,11,13,14⇓⇓⇓⇓⇓⇓⇓⇓⇓⇓ may lead to selection bias.20,21⇓ Conversely, despite relatively high MVP prevalence,10 population systematic echocardiographic studies will be underpowered because 500 MVP occurrences would require ≈25 000 echocardiographic screenings. Third, background population INE rates, necessary to determine excess risk attached to MVP, are usually unknown.18 Thus, rates and predictors of INEs associated with MVP remain uncertain.
Analysis of MVP in Olmsted County (Minn) takes advantage of numerous subjects with long-term follow-up after diagnosis in a community in which background INE rates22 and subsequent occurrence of atrial fibrillation (AFib) are defined. We hypothesized that patients diagnosed with MVP in the community incur an excess lifetime INE risk but that INEs are confined to high-risk subsets predictable by clinical and echocardiographic variables and that complex interactions exist between mitral regurgitation (MR) caused by MVP and subsequent occurrence of AFib and INEs.
Materials and Methods
Medical care in Olmsted County is largely community contained; only 1 laboratory provides echocardiography, and all care providers are linked through the Rochester Epidemiology Project.22 It was therefore possible to identify all Olmsted County residents with diagnosis of MVP between 1989 and 1998 confirmed by echocardiography.23 Of those, patients with no past history of INE, no documented AFib, and no mitral surgery before diagnosis were included. Residency in Olmsted County before diagnosis was required to avoid temporary residency motivated by medical care.
Cardiovascular history (including history of hypertension, myocardial infarction, congestive heart failure, and endocarditis), baseline clinical status, and comorbidity were noted. Follow-up care was similarly provided to the general and MVP populations, and events were determined by review of medical records, by questionnaires, or through telephone interviews. AFib, paroxysmal or permanent, was considered present if recorded on ECG. INEs included both cerebral infarction and transient ischemic attack (TIA). Cerebral infarction was defined as the persistence for >24 hours of a focal neurological deficit caused by altered circulation of the cerebral hemispheres, brain stem, or cerebellum with or without CT or MRI documentation.24 TIA was defined as focal neurological symptoms of sudden occurrence and rapid resolution (<24 hours) related to altered circulation of the brain.25 Neurological diagnoses were confirmed by a neurologist from our group for all INEs among the general and MVP populations.
All patients underwent comprehensive Doppler echocardiography during routine examination. MVP diagnosis used current criteria10,23,26⇓⇓ based on systolic leaflet displacement beyond the mitral annulus >2 mm in long-axis views. Diagnoses of flail or thickened leaflets10 were based on previously recommended criteria. Left ventricular diameters, left ventricular ejection fraction, and left atrial (LA) diameter measurements were guided by 2-dimensional echocardiography. MR severity was assessed semiquantitatively in 3 categories: absent or trivial, slight, and moderate or worse. All Doppler echocardiographic data were used as originally recorded without alteration.
Baseline characteristics are presented as mean±SD for continuous variables. Comparison between groups used the 2-sample t test, χ2, or analysis of variance. Event rate estimations used the Kaplan-Meier method overall and stratified by age, for lifetime risk (censoring at death or last follow-up), and under conservative management (censoring at cardiac surgery). INE rates were also determined during follow-up in patients in sinus rhythm only by censoring at occurrence of AFib and during follow-up in patients sinus rhythm and without surgery. Observed INE rates were compared with expected rates by use of the 1-sample log-rank test. Expected INE hazards were calculated from incidence rates of both cerebral infarctions24 and TIAs25 from the Olmsted County population with sex, age, rhythm, and surgical history similar to those of MVP subjects. Comparison was performed overall to define excess lifetime risk of INE and censored at specific follow-up AFib or cardiac surgery to define excess risk in sinus rhythm, under conservative management, and both. Baseline clinical and echocardiographic variables were tested as potential predictors of INEs with Cox proportional-hazards modeling. Follow-up AFib and cardiac surgery were used as time-dependent variables. Variables with P<0.10 by univariate analysis were incorporated into multivariate models, and P<0.05 was considered significant.
Among 19 370 screened, 777 Olmsted County residents diagnosed with MVP during the study period presented in sinus rhythm with no documented AFib or INE history before diagnosis. The diagnostic echocardiography was suggested by auscultatory abnormalities (systolic murmur or click) in 542 patients. MVP was silent and diagnosis was incidental in 235 patients in whom echocardiography was motivated by minor cardiorespiratory symptoms (n=62), family history of murmur (n=7), possible vascular symptoms (n=30), or noncardiovascular concerns (n=34). In 102 patients, echocardiography was ordered for general medical examination and considered systematic. Table 1 details baseline clinical and echocardiographic characteristics. The population was young (median age, 45 years) with normal blood pressure and with rare history of hypertension or myocardial infarction. Left ventricular ejection fraction was 62±7%; in only 31 patients (4%), it was <50%. The MR was moderate or worse in 124 patients (16%), and the LA was ≥40 mm in 192 (25%). In 70 randomly selected echocardiograms, agreement of 93% for leaflets thickening >5 mm (κ=0.85; thickness, 5.6±1.1 versus 3.5±1.0 mm; P<0.001) and κ=0.89 for degree of MR were found between reinterpretation and initial report.
Lifetime Risk of INEs
During a mean follow-up of 5.5±3.0 years (complete in 96%), 30 patients (14 men, 16 women) experienced a first INE (12 TIAs, 18 cerebral infarctions). In 7 patients, 8 recurrent INEs during follow-up resulted in a total of 38 INEs (17 TIAs, 21 cerebral infarctions). Diagnosis was confirmed by a neurologist in all cases and by either CT or MRI in 28 patients. Only 2 of 28 INE patients tested had concomitant cerebrovascular stenosis by Doppler ultrasound (both with cerebral infarction). One INE (TIA) only occurred in a patient <50 years of age at diagnosis. Regarding temporal relationship between cardiac surgery and INE, INEs occurred after cardiac surgery (9 mitral surgery for severe MR, 2 events occurring within the first postoperative month) in 10 patients (6 TIAs, 4 infarctions) and in 20 patients (7 TIAs, 13 infarctions) under medical management. Regarding temporal relationship between AFib and INE, in 16 patients, the first INE occurred after follow-up AFib (8 TIAs, 8 infarctions), and among these, 10 were postsurgery and 6 were under medical management. Finally, in 14 patients, the first INE occurred without detected AFib and under conservative management (5 TIAs, 9 infarctions).
Hence, 5- and 10-year first INE rates (Table 2) were 3±0.7% and 7±1%, respectively (linearized rate, 0.7% per year). Comparison of observed to community expected INE rates showed excess INE risk among MVP subjects (relative risk [RR], 2.2; 95% CI, 1.5 to 3.2; P<0.001). When analysis was restricted to cerebral infarctions, comparison with expected rates also showed excess risk in MVP (RR, 1.72; 95% CI, 1.02 to 2.73; P=0.02) (Figure 1).
Predictors of First INE
Differences in baseline characteristics between patients with and without INEs and Cox proportional-hazards analysis for time to first INE adjusted for age are presented in Table 1.
In multivariate analysis, independent baseline predictors of INEs were older age (RR, 1.08 per year; 95% CI, 1.05 to 1.12; P<0.001), mitral valve thickening (RR, 3.4; 95% CI, 1.3 to 10.5; P=0.01) with borderline association of LA diameter (RR, 1.05 per 1 mm; 95% CI, 0.99 to 1.10; P=0.083). These variables were also independent predictors of INEs, when analysis was restricted to medical management (older age, P<0.001; mitral valve thickening, P=0.01). When models included follow-up events as time-dependent variables, independent predictors of INEs were older age (RR, 1.08; 95% CI, 1.04 to 1.11; P<0.001), mitral valve thickening (RR, 3.2; 95% CI, 1.4 to 7.4; P=0.008), occurrence of follow-up AFib (RR, 4.3; 95% CI, 1.9 to 10.0; P<0.001), and need for cardiac surgery during follow-up (RR, 2.5; 95% CI, 1.1 to 5.8; P=0.03). When analysis was restricted to cerebral infarctions, independent predictors were older age (RR, 1.10; 95% CI, 1.05 to 1.15; P<0.001), mitral valve thickening (RR, 3.2; 95% CI, 1.1 to 9.3; P=0.03), and follow-up AFib (RR, 5.9; 95% CI, 2.1 to 16.1; P<0.001).
MR severity of moderate or worse was associated with higher INE risk (RR, 7.6), but adjusted for age, the trend toward higher risk did not reach significance (RR, 1.45; P=0.35). Heart failure, although overrepresented among INE patients because it more frequently required mitral surgery (P=0.006), did not independently predict INEs (P=0.47).
Table 2 presents INE rates (lifetime and under medical management) of patients subsets formed according to predictors defined, with comparison in each subset to expected rates.
Excess INE lifetime risk observed globally was marked in patients developing follow-up AFib but was also observed in patients remaining in sinus rhythm. Patients <50 years of age at diagnosis had very low INE risk (Figure 2), and most excess risk was observed in patients >50 years of age. Patients with thickened and nonthickened leaflets appeared to have very close INE rates, but patients with thickened leaflets were much younger (46±19 versus 52±21 years, P<0.001) and therefore had lower expected INE rates than patients with nonthickened leaflets and a much higher and significant risk ratio of INEs compared with expected. The pattern was similar with INEs occurring under conservative management (Table 2, middle), although statistical significance was lower because of lower event numbers. Remarkably, for patients remaining in sinus rhythm and never undergoing mitral surgery, excess INE risk was detectable with thickened leaflets (at 10 years, 4±1%; RR, 2.2; 95% CI, 1.1 to 4.1; P=0.009) (Figure 3).
AFib and Predictors
During follow-up, 63 patients experienced AFib (paroxysmal or permanent). In 18, AFib occurred after cardiac surgery (7 within the first postoperative month); in 45, AFib occurred under medical management. Five- and 10-year AFib rates were 7±1% and 13±2% and were 5±1% and 10±2% under medical management.
Independent predictors of AFib were older age (RR, 1.05; 95% CI, 1.04 to 1.08; P<0.001), LA ≥40 mm (RR, 1.9; 95% CI, 1.1 to 3.2; P=0.01), and higher MR severity complicating the MVP (P<0.001). The RR of AFib during follow-up compared with patients without MR was high for moderate or worse MR (RR, 8.1; 95% CI, 2.4 to 28.0; P<0.001) and slight MR (RR, 3.7; 95% CI, 1.1 to 12.6; P=0.03). Baseline LA ≥40 mm (adjusted RR, 1.8; 95% CI, 0.9 to 3.2; P=0.07) and moderate or worse MR (RR, 2.8; 95% CI, 1.5 to 5.3; P=0.002) remained predictive of AFib occurring under medical management.
The present study shows that patients routinely diagnosed in the community with MVP in sinus rhythm incur an excess lifetime INE risk. This excess risk of INEs is observed even under conservative management and is independently predicted by age and mitral valve thickening and in a more borderline manner by LA enlargement. The risk of INEs is increased during follow-up by the need for mitral surgery. AFib during follow-up is frequent, increases the INE risk, and is determined by direct MVP consequences, ie, a large degree of MR and LA enlargement.
INEs in Patients With MVP
Initial suspicion of an MVP-INE association27,28⇓ was supported by case reports29 and case-control studies4,14⇓ and by pathological description of thrombi adherent to MVP leaflets.1,2⇓ This led to researchers considering a larger role of MVP in INE occurrence, particularly in patients <45 years of age4 and without other stroke causes.3,4⇓ The MVP-stroke association encountered skepticism,17 particularly because of MVP overdiagnosis by criteria, which had to be revised in late 1980s.16,23,26⇓⇓ This raised legitimate questions about the relevance of studies based on these obsolete criteria.2–4,6,7,14,18,19,28–30⇓⇓⇓⇓⇓⇓⇓⇓⇓⇓ The MVP-INE association was recently denied in a high-profile case-control study,11 which could not confirm MVP overrepresentation4 among young stroke patients.
These discordant data underscore major methodological issues in assessing MVP-INE association. Case-control studies are biased in the choice of cases and particularly of supposedly representative controls, which is questionable in hospital-based studies.15 Most data emanate from tertiary-care centers,3–9,11,14⇓⇓⇓⇓⇓⇓⇓⇓ where the referred populations do not reflect the community.20,21⇓ Only 2 population-based studies reported an MVP-INE association.10,18⇓ One in our community before 1989 used older echocardiographic criteria.18 The other reported no more past neurological complications among 84 individuals with MVP than among those without MVP.10 This small sample underscores power issues of echocardiographic screening, which would require 30 000 to 130 000 examinations to detect an MVP sample similar to ours. The present study takes advantage of the Rochester Epidemiology Project to include all community members (a large sample) routinely diagnosed with MVP with appropriate criteria and followed up long term after diagnosis. In this community, hazards of INEs by age, sex, occurrence of AFib, and cardiac surgery have been defined,24,25⇓ allowing comparison of MVP-INE rates to rates expected with identical clinical characteristics in the same community. These unique advantages allow us to define INE hazard and most importantly excess lifetime INE risk.
Hence, we observed that MVP patients have more than twice the lifetime INE risk as similar community members, in part because of cardiac surgery required by MVP but also observed under conservative management. Excess INE risk is noticeable in view of minimal INE risk factors in this population and of the link between specific MVP characteristics with higher INE risk. Indeed, more MR resulting from MVP leads to LA enlargement and follow-up AFib, which is directly associated with INE occurrence. Hence, the excess INE risk is not incidental but is MVP related.
Risk of INEs in Specific Subsets of MVP
High- and low-risk subsets were predictable by clinical and echocardiographic parameters. Age is a major determinant of INEs in all populations.31 In MVP, age determines excess INE risk. In patients <50 years of age, INE rates were extremely low and not significantly different from the community. Whether the 1.7 RR truly represents excess INE risk in young MVP patients is beyond realistic analysis. With moderate MVP prevalence, the population-attributable INE risk resulting from MVP in patients <50 years of age appears low, giving little probability of positive results to low-power case-control series.11 After 50 years of age, the excess INE risk with MVP appears unrelated to vascular obstructive lesions and may be related to aging-related thromboembolic risk factors32 and to higher rates of AFib and valvular surgery.
Echocardiographic mitral valve thickening independently predicts higher INE rates. Its outcome predictive value has been variably reported, from a strong risk factor for complications6 to lack of association with stroke.5,18⇓ Our study demonstrates that leaflet thickening independently predicts subsequent INEs and even cerebral infarctions. Mitral thickening, possibly through higher risk of thrombus formation,1,2⇓ may account for excess INE risk among patients conservatively managed and in sinus rhythm.
AFib occurrence predicts subsequent INEs in patients with but also without MVP.33 AFib risk has been variably estimated in MVP,9,10⇓ but recent data showed that it is high with severe MR.34 AFib incidence for any given degree of MR and LA enlargement increases with aging,34 explaining in part the marked excess INE risk in older patients with MVP compared with community members of similar age and sex. Ten years after diagnosis of MVP, at least 13% of patients initially in sinus rhythm will have presented a diagnosed episode of AFib and will be at high risk for INEs. This incidence is much higher than that reported in the general population of similar age,31 even though a formal comparison is not possible.
The degree of MR determines the need for cardiac surgery, which is a known cause of INEs, perioperatively and through prosthetic thromboembolic complications. Hence, MR emerges as an indirect risk factor for INEs in patients with MVP through LA enlargement, occurrence of AFib, and need for cardiac surgery. This complex interaction between MR and subsequent occurrence of INEs in patients with MVP is at variance with previous reports of INE-protective effects in nonvalvular AFib.35 Hence, MVP puts patients at high risk for INEs through multiple ways.
Our study analyzed INEs in community members clinically diagnosed with MVP as opposed to persons systematically screened. Systematic screening would require massive numbers of prospective echocardiograms that no existent epidemiological program would qualify for, leaving community studies as the only possible source of cohorts of sufficient size. Our study was not designed to define the risk of INEs in silent MVP, but this goal is less relevant clinically. Indeed, no difference in INE risk was noted according to the indication for echocardiography (P=0.26) or in patients with and without auscultatory abnormality after adjustment for age (P=0.11). Furthermore, comparing observed INE rates in all community patients diagnosed with MVP to expected rates in the same community is the essential step in defining the excess risk of INEs associated with this condition.
Interventions to minimize INE risk in high-risk patients are important, but tracking all antithrombotic treatment and their timing was not possible. Careful analysis of their potentially harmful complications or preventive effect on INEs requires future clinical trials or risk-modeling studies.
Patients diagnosed with MVP in the community incur an overall excess lifetime risk of INEs compared with a similar community population. This excess risk is also observed under medical management during which it is confined to high-risk subsets of patients. Advancing age and thickened mitral leaflets at diagnosis are major baseline predictors of excess risk of INEs during follow-up. Subsequently, the need for cardiac surgery and occurrence of AFib in the course of the disease are independently associated with the risk of ischemic neurological complications. These time-dependent risk factors for excess rates of INEs are all related to the degree of MR resulting from MVP, which therefore plays an indirect role in the occurrence of INEs. High-risk subgroups that will require careful monitoring for the risk of INEs can be defined.
This study was supported in part by National Institutes of Health grants HL 64928 and AR 30582. We appreciate the support of J. Sicks and R. Mueller.
- Received September 4, 2002.
- Revision received December 16, 2002.
- Accepted December 24, 2002.
- ↵Pomerance A. Ballooning deformity (mucoid degeneration) of atrioventricular valves. Br Heart J. 1969; 31: 343–351.
- ↵Hanson MR, Conomy JP, Hodgman JR. Brain events associated with mitral valve prolapse. Stroke. 1980; 11: 499–506.
- ↵Sandok BA, Giuliani ER. Cerebral ischemic events in patients with mitral valve prolapse. Stroke. 1982; 13: 448–450.
- ↵Bonow R, Carabello B, DeLeon A, Edmunds L, Fedderly B, Freed M, Gaasch W, McKay C, Nishimura R, O’Gara P, O’Rourke R, Rahimtoola S. ACC/AHA guidelines for the management of patients with valvular heart disease. Circulation. 1998; 98: 1949–1984.
- ↵Jones H, Naggar C, Seljan M, Downing L. Mitral valve prolapse and cerebral ischemic events: a comparison between a neurology population with stroke and a cardiology population with mitral valve prolapse observed for five years. Stroke. 1982; 13: 451–453.
- ↵Barletta G, Gagliardi R, Benvenuti L, Fantini F. Cerebral ischemic attacks as a complication of aortic and mitral valve prolapse. Stroke. 1985; 16: 219–223.
- ↵Levine R, Triulzi M, Harrigan P, Weyman A. The relationship of mitral annular shape to the diagnosis of mitral valve prolapse. Circulation. 1987; 75: 756–767.
- ↵Boughner DR, Barnett HJ. The enigma of the risk of stroke in mitral valve prolapse. Stroke. 1985; 16: 175–177.
- ↵Orencia AJ, Petty GW, Khandheria BK, Annegers JF, Ballard DJ, Sicks JD, O’Fallon WM, Whisnant JP. Risk of stroke with mitral valve prolapse in population-based cohort study. Stroke. 1995; 26: 7–13.
- ↵Levine R, Stathogiannis E, Newell J, Harrigan P, Weyman A. Reconsideration of echocardiographic standards for mitral valve prolapse: lack of association between leaflet displacement isolated to the apical four chamber view and independent echocardiographic evidence of abnormality. J Am Coll Cardiol. 1988; 11: 1010–1019.
- ↵Brown RD, Jr, Petty GW, O’Fallon WM, Wiebers DO, Whisnant JP. Incidence of transient ischemic attack in Rochester, Minnesota, 1985–1989. Stroke. 1998; 29: 2109–2113.
- ↵Watson RT. TIA, stroke, and mitral valve prolapse. Neurology. 1979; 29: 886–889.
- ↵Orencia AJ, Petty GW, Khandheria BK, O’Fallon WM, Whisnant JP. Mitral valve prolapse and the risk of stroke after initial cerebral ischemia. Neurology. 1995; 45: 1083–1086.
- ↵Wolf P, Abott R, Kannel W. Atrial fibrillation as an independent risk factor for stroke: the Framingham study. Stroke. 1991; 22: 983–988.