Valvular Strands and Cerebral Ischemia
Effect of Demographics and Strand Characteristics
Background and Purpose Valvular strands, thin filamentous material attached to the mitral or aortic valve, are seen during transesophageal echocardiography and have been associated with stroke. Little is known about this association in different age, sex, and race-ethnic subgroups and the effect of various strand characteristics on this association.
Methods From patients referred for transesophageal echocardiography, 73 patients with recent ischemic stroke (68) or transient ischemic attack (5) were age matched to 73 stroke- and transient ischemic attack–free control subjects. The association between valvular strands and cerebral ischemia was evaluated for the overall group and demographic subgroups. The effect of strand location, length, number, and valve thickness was also determined.
Results An association between cerebral ischemia and valvular strands was observed (odds ratio [OR]=4.4; 95% confidence interval [CI]=2.0 to 9.6). The association was found for both men and women and among all three race-ethnic groups. The OR was greater in those who were younger (12.5 [95% CI=2.4 to 64.5] for age <60, 4.8 [95% CI=1.3 to 18.2] for age 60 to 69, and 1.8 [95% CI=0.5 to 6.4] for age ≥70 years). Strands on both the mitral (OR=3.5; 95% CI=1.5 to 7.9) and aortic (OR=3.7; 95% CI=1.1 to 11.9) valve were associated with cerebral ischemia, whereas the number and length of strands were not. The effect of strands was independent of mitral or aortic valve thickness.
Conclusions Valvular strands, whether mitral or aortic, are associated with ischemic stroke, especially among younger persons.
Despite an extensive diagnostic workup, a cause for nearly 40% of ischemic strokes is not easily identified.1 TEE has proven useful in the evaluation of stroke, particularly those labeled cryptogenic. Potential cardiac sources of emboli better identified on TEE than on TTE include left atrial thrombus, patent foramen ovale, atrial septal aneurysm, spontaneous left atrial contrast, valvular vegetations, and aortic arch atheromas.2 More recently, TEE has identified mobile, filamentous strands on the cardiac valves that have been found to be associated with stroke and systemic embolization.3 4 5 6
The reports linking valvular strands and ischemic stroke, however, have been hampered by the lack of a control group or nonblinded interpretation of the TEE. In addition, little is known about the association in different demographic subgroups and how various characteristics of strands might affect the association. The aim of this study was to investigate the association between native valvular strands and cerebral ischemia, especially in different sex, age, and race-ethnic (black, Hispanic, white) subgroups, and the effect of strand location (mitral or aortic), number, length, and valve thickness on the association.
Subjects and Methods
Case and control subjects were selected from a group of 453 consecutive patients referred as part of their clinical evaluations to the echocardiography laboratory for TEE over a 9-month period. A subject was included in this study as a case if he/she had first ischemic stroke or TIA within the 3 months before TEE. In addition, enough clinical data (history, examination, brain imaging, TTE, extracranial duplex Doppler, and when available, transcranial Doppler, MRI, MR angiography, and angiography) must have been available to classify the cerebrovascular event as TIA or one of the subtypes of ischemic stroke (atherothrombotic, cardioembolic, lacunar, and cryptogenic) by criteria adapted from the Stroke Data Bank.7 Classification was performed blinded to the results of the TEE. Subjects were excluded from the study if they had a previous history of peripheral embolism, clinical suspicion of endocarditis, prosthetic cardiac valve, or a congenital heart defect. All subjects meeting the above criteria were included. Stroke risk factor data (hypertension, diabetes mellitus, cardiac disease, hypercholesterolemia, current cigarette smoking, and alcohol abuse) were collected from the history and were coded as present if noted by the physician. Cardiac disease included a history of myocardial infarction, angina, congestive heart failure, or atrial fibrillation.
A subject was included as a control if he/she had no history of stroke or TIA. Control subjects were frequency matched by age to the case subjects but were not matched by sex, race-ethnicity, or stroke risk factors. The same exclusion criteria applied to the case subjects were also applied to the control subjects. The first controls meeting all the criteria were selected.
TEE was performed with Hewlett-Packard Sonos 1000 or 1500 equipment with either a biplane or omniplane probe equipped with a 5-MHz transducer. Images were enlarged at the time of the study with the use of conventional zoom features, recorded on videotape, and reviewed at an off-line echo workstation where video frames of interest were digitized for analysis. The images were magnified as necessary (in most cases ×3 to ×5) and measured by computer graphics software.
Strands were defined as thin, mobile, filamentous projections attached to the valvular leaflets (Fig 1⇓). The number of strands was counted on any single frame exhibiting the largest number and classified as none, one, or more than one. Similarly, the maximum length was measured from the valve surface on any single video frame exhibiting the maximum length. All images were analyzed by a single echocardiographer (I.O.) blinded to the clinical indication for the study and the status of the subject as a stroke/TIA case or a control subject.
All data are expressed as mean±SD for continuous variables and as proportions for categorical variables. ORs were calculated along with 95% CIs and probability values. The Breslow-Day test for homogeneity of ORs was used to assess for differences among the demographic subgroups. The χ2 test was used to compare the frequency of risk factors among the stroke/TIA cases with and without strands. Data analysis was performed with SAS 6.10 software (SAS Institute).
The study included 73 case and 73 control subjects. The case subjects consisted of 68 patients with ischemic stroke and 5 with TIA. The mean ages of the case (63±13 years) and control (64±13 years) subjects were similar. Forty-nine percent of the case subjects and 51% of the control subjects were women. Race-ethnic distribution was approximately evenly divided among the case subjects and included 24 blacks (33%), 26 Hispanics (36%), and 21 whites (29%), while the control subjects consisted of more whites (39, 53%) and fewer blacks (13, 18%) and Hispanics (20, 27%).
Overall, strands were seen in 34 of 73 case subjects (47%) and in 12 of 73 control subjects (16%) (OR=4.4; 95% CI=2.0 to 9.6). The association was seen in both men and women, Hispanics, whites, and also, although it was not statistically significant, in blacks (Table 1⇓). The ORs were not significantly different between men and women (P=.36) or the different race-ethnic groups (P=.40). Although the association between strands and cerebral ischemia was greater in younger patients (OR=12.5 for age <60, OR=4.8 for age 60 to 69, and OR=1.8 for age ≥70; Fig 2⇓), this did not reach statistical significance (P=.17).
Table 2⇓ shows the association between strands and cerebral ischemia for different strand characteristics. Strands were more common in case subjects than in control subjects at both the mitral (OR=3.5) and the aortic (OR=3.7) positions. The presence of more than one strand did not increase the association. Longer strands also did not significantly increase the association, although mitral valve strands longer than 7.5 mm showed a nonsignificant 3.5-fold increase in the OR compared with shorter mitral valve strands. Mitral valve thickness and aortic valve thickness did not differ between those with (mitral, 2.9±1.0 mm; aortic, 2.9±1.0 mm) and without (mitral, 2.9±1.9 mm; aortic, 2.7±1.1 mm) strands.
Valvular strands were identified in 15 of 30 cryptogenic (50%), 8 of 19 atherothrombotic (42%), 5 of 8 lacunar (62%), and 3 of 11 cardioembolic strokes (27%) and in 3 of 5 TIA cases (60%). Overall, strands were not significantly more frequent in cryptogenic stroke patients (50%) than in noncryptogenic stroke patients (42%).
In the stroke/TIA cases, the frequency of traditional stroke risk factors (hypertension, diabetes mellitus, cardiac disease, hypercholesterolemia, current cigarette smoking, and alcohol abuse) did not significantly differ between patients with and without strands among the 69 patients in whom this information was available (Table 3⇓).
Valvular strands may represent small, endothelial-covered, fibrinous protuberances known pathologically as Lambl’s excrescences, which are thought to result from mechanical trauma.8 9 In one pathologically examined case of a patient with TEE-visualized strands, the mitral valve revealed Lambl’s excrescences.3
Valvular strands were common in our patients (47% of stroke patients and 16% of control subjects) and were associated with cerebral ischemia with an OR of 4.4. In four other studies, strands have been found in 6.3% to 22.5% of case subjects with cerebral ischemia and in 0.3% to 12% of control subjects, with ORs ranging from 2.1 to 21.8.3 4 5 6 Although the prevalence of strands varies, all investigations have found an association between strands and cerebral ischemia. There are several possible explanations for the disparate frequencies. In our control subjects the prevalence of strands increased with age, and the different studies examined patients of different ages. Selection bias for performance of TEE at different institutions may also contribute to the discrepancies. Moreover, standardized definitions of strands have not yet been developed, and methodological differences, including the use of magnification, may affect strand detection and frequency. However, this should equally affect case and control subjects, and although it may affect the frequency of strands, it should not affect the associations.
In our study the association between strands and cerebral ischemia was seen in both men and women and among all race-ethnic groups, although it did not reach statistical significance in blacks. There were relatively few black patients, and the OR for this group may have been affected by the particularly high prevalence of strands in the control subjects. The association was greatest in the younger patients. In those aged <60 years, the OR was 12.5, decreasing to 4.8 in those 60 to 69 years and 1.8 in those aged ≥70 years. Tice et al6 also commented that mitral strands might be more important in younger patients and reported that they were particularly prevalent in patients aged <50 years who were thought to have a cardioembolic source. They did not, however, report on the ages of the two patients with strands without cerebrovascular disease, and therefore an increased association in younger patients could not be determined.
The association with cerebral ischemia was present for strands on both the mitral (OR=3.5) and aortic (OR=3.7) valves. Previous reports have focused on mitral strands, although Freedberg et al4 included patients with both mitral and aortic strands but did not determine the association for each valve separately. To our knowledge, no one has investigated the effect of strand number or length on the association with cerebral ischemia. Although we did not find a significant association between strand number or length and cerebral ischemia, there was a nonstatistically significant 3.5-fold increase in the OR with mitral strands longer than 7.5 mm compared with shorter mitral strands.
The finding that strands are associated with ischemic stroke does not necessarily mean that strands are an actual embolic source. If so, it might be expected that strands would be most frequent among those with cryptogenic stroke. In our study strands were seen in only slightly higher proportions of patients with cryptogenic (50%) compared with noncryptogenic stroke (42%). Tice et al,6 however, did suggest that strands embolize because they are more often found in young patients with probable cardioembolic strokes. In a case of a patient with a leg embolism thought to be of cardiac origin, pathological examination of the embolus suggested that it was composed of Lambl’s excrescences.10 Alternatively, strands may be associated with another disorder or risk factor associated with stroke. Two previous reports have attempted to control for some factors. In a substudy, Freedberg et al4 analyzed 41 patients with strands and no other embolic sources and matched them by age, sex, hypertension, and smoking to 41 patients without strands or other embolic sources.4 The patients with strands had a much higher frequency of emboli with an OR of 10.0. Cohen et al5 used multiple logistic regression to adjust for confounding factors and still found that mitral valve strands were associated with brain infarcts (OR=2.06).5 The same group has also reported that strands are not associated with an increased risk of recurrent cerebral ischemia, and therefore the association may not be causal.11
In our study we did not control for other stroke risk factors because these data were unavailable for the control subjects. Moreover, since our control subjects were selected from patients referred for TEE, who usually have some type of cardiac disease, the frequency of vascular risk factors is likely to be inflated and would result in an inappropriate adjustment of the OR. In addition, the frequency of strands in these control subjects might be different than in a community-based control group. A cohort of nonhospitalized, nonreferred volunteers for TEE would be needed to overcome these difficulties.
To investigate the effect of other stroke risk factors, we examined the frequency of hypertension, diabetes mellitus, cardiac disease, hypercholesterolemia, current cigarette smoking, and alcohol abuse in our stroke/TIA case subjects and found that it did not significantly differ among those with and without strands. The presence of these risk factors was slightly more common in the patients without strands, especially for diabetes mellitus, cardiac disease, and current cigarette smoking. This suggests that these risk factors are not strongly associated with strands.
Another potential limitation of this and other studies is that case subjects are usually drawn from patients who were referred for TEE for clinical reasons and therefore do not necessarily represent the overall stroke/TIA population. There may be more cryptogenic patients since these patients are more likely to be referred for TEE.
The appropriate therapy for patients with strands is undetermined. Patients in our institution have received aspirin, ticlopidine, or warfarin. In the case-control substudy by Freedberg et al,4 there was no significant difference in the use of antiplatelet or anticoagulant therapy in patients with and without strands, suggesting that this type of therapy may not influence the presence of strands. However, there are reports of strands resolving after institution of warfarin or the addition of dipyridamole to warfarin.12 13 Informed and rational decisions regarding therapy will depend on future studies further defining the pathophysiology of strands and their role in cerebral ischemia.
Selected Abbreviations and Acronyms
|TIA||=||transient ischemic attack|
This study was supported by grants from the National Institute of Neurological Disorders and Stroke (R01 NS 27517, R01 NS 29993, R01 NS 33248, R01 NS 32525, and TS NS 07153) and the Mellam Family Foundation.
Presented in part in abstract form at the 22nd International Joint Conference on Stroke and Cerebral Circulation, Anaheim, Calif, February 6-8, 1997.
- Received April 25, 1997.
- Revision received July 8, 1997.
- Accepted July 25, 1997.
- Copyright © 1997 by American Heart Association
DeRook FA, Comess KA, Albers GW, Popp RL. Transesophageal echocardiography in the evaluation of stroke. Ann Intern Med. 1992;117:922-932.
Lee RJ, Bartzokis T, Teoh TK, Grogin HR, Choi D, Schnittger I. Enhanced detection of intracardiac sources of cerebral emboli by transesophageal echocardiography. Stroke. 1991;22:734-739.
Cohen A, Crassard I, Tzourio C, Amarenco P, on behalf of the FAPS Investigators. Mitral valve strands and the risk of brain infarcts: a case-control study. Stroke. 1996;27:16. Abstract.
Tice FD, Slivka AP, Walz ET, Orsinelli DA, Pearson AC. Mitral valve strands in patients with focal cerebral ischemia. Stroke. 1996;27:1183-1186.
Foulkes MA, Wolf PA, Price TR, Mohr JP, Hier DB. The Stroke Data Bank: design, methods, and baseline characteristics. Stroke. 1988;19:547-554.
Lambl VA. Papillare exkreszenzen an der semilunar-klappe der aorta. Wien Med Wochenschr. 1856;6:244-247.
Cohen AA, Tzourio C, Chauvel C, Bertrand B, Crassarrd I, Bousser M, Amarenco P, on behalf of the FAPS Investigators. Strands on native mitral valves do not increase the risk of recurrent brain infarction: follow-up study of a cohort. Circulation. 1996;93:I-217. Abstract.
Schnittger I. Valvular strands. In: Daniel WG, Kronzon I, Mugge A, eds. Cardiac Embolism. Baltimore, Md: Williams & Wilkins; 1996.