- Email: [email protected]
Predictors of ischemic stroke in non-rheumatic atrial fibrillation
International
Journal
of
cardiology EISEVIER
International Journal of Cardiology 56 (1996) 6 l-70
Predictors of ischemic stroke in non-rheumatic atria1 fibrillation Michiaki Yoshida, Yasuyuki Nakamura”, Masahito Higashikawa, Masahiko Kinoshita First Department of Internal Medicine, Shiga University of Medical Science, Tsukinowa, Seta, Otsu, Shiga 520-21, Japan
Received 17 October 1995; revised 13 February 1996; accepted 21 May 1996
Abstract We retrospectively analyzed the clinical features of patients with non-rheumatic atria1 fibrillation to identify risk factors of ischemic stroke. Non-rheumatic atrial fibrillation is associated with an increased risk of ischemic stroke. However, the predictors of ischemic stroke in non-rheumatic atria1 fibrillation are unclear. The study population consisted of 122 patients with non-rheumatic atrial fibrillation who had no previous clinical cerebral strokes at the start of the follow-up. Patients with cardiomyopathy and paroxysmal or intermittent atria1 fibrillation were excluded from the study. The mean age was 61.7-C 12.8 years. We defined two endpoints; namely, occurrence of ischemic stroke (endpoint l), and ischemic stroke or cardiac death (endpoint 2). During the follow-up, 18 patients had ischemic stroke and 6 patients experienced cardiac death. The 5-year event-free rates for endpoints 1 and 2 were 87.4% and 85.0%, respectively. A Cox analysis revealed that endpoint 1 was significantly associated with age (risk ratio (RR)= 1.106, P=O.O052), end-diastolic left ventricular dimension (RR=O.882, P=O.O393), end-systolic left ventricular dimension (RR=1.149, P=O.O323) and the thickness of the interventricular septum (RR=1.493, P=O.Olll). Endpoint 2 was associated with age (RR=1.122, P=O.O004), left atria1 dimension (RR= 1.057, P=O.O666), end-diastolic left ventricular dimension (RR=O.935, P=O.O426), fractional shortening (RR=O.880, P=O.OOOl) and the thickness of the left ventricular posterior wall (RR= 1.644, P=O.O004). The present results suggest that, in addition to left ventricular dimensions and left atrial dimension, left ventricular hypertrophy may be associated with ischemic stroke. Keywords:
Non-rheumatic atria1 fibrillation;
Ischemic stroke; Predictors
1. Introduction Atrial fibrillation associated with rheumatic heart disease and mitral stenosis is well known to predispose the patient to systemic arterial embolism [l]. In addition, non-rheumatic atria1 fibrillation is associated with au increased risk for ischemic stroke [2,3]. *Corresponding author, Tel.: +8I 775 48 2213; fax: +81 775 43 5839.
In a population-based study [2,4], the risk for ischemic stroke among patients with non-rheumatic atrial fibrillation was estimated to be 5 times greater than that for comparable patients in sinus rhythm. Most of these strokes have the clinical characteristics of cerebral embolism. Recent randomized trials involving antitbrombotic aspirin in non-rheumatic
ported that antithrombotic therapy substantially reduces the occurrence of ischemic strokes and sys-
0167-5273/96/$15.00 0 1996 Elsevier Science Ireland Ltd. All rights reserved PII
SO167-5273(96)02726-X
therapy with warfarin and atrial fibrillation have re-
62
M. Yoshida et al. I International
temic embolism in patients with atria1 fibrillation [5-lo]. However, the predictors of ischemic stroke in non-rheumatic atria1 fibrillation are still unclear. We retrospectively analyzed the clinical features of patients with non-rheumatic atria1 fibrillation to identify risk factors for ischemic stroke.
2. Materials and methods We analyzed 122 patients with non-rheumatic atria1 fibrillation who had no previous clinical cerebral strokes as of the start of the follow-up. Adults with chronic sustained atria1 fibrillation with no evidence of mitral stenosis on two-dimensional echocardiography were eligible. Patients with hycardiomyopathy pertrophic and dilated cardiomyopathy were excluded from the study. Patients were also excluded if they had a finding of asymmetric septal hypertrophy with a ratio of the thickness of the interventricular septum to that of the left ventricular posterior wall of 1.30 or greater by echocardiography. This study did not include patients with paroxysmal or intermittent atria1 fibrillation. Other reasons for exclusion were echocardiographic evidence of an intracardiac thrombus, a left ventricular aneurysm, the presence of severe congestive heart failure, or prosthetic heart valves. The follow-up was terminated on 30 June 1994. The mean age of the 122 patients (93 males, 29 females) was 61.75 12.8 years (range 24-90). The underlying heart diseases consisted of 23 hypertension without any detectable heart disease, 16 ischemic heart diseases (8 old myocardial infarction and 8 angina pectoris), 10 sick sinus syndrome, 9 non-rheumatic valvular diseases (4 mitral valve prolapse, 4 mitral regurgitation and 1 aortic regurgitation), 8 congenital heart diseases, 7 hyperthyroidism, 2 chronic obstructive pulmonary diseases, 1 polymyositis, 1 constrictive pericarditis and 45 lone atria1 fibrillation. In this study, we defined lone atrial fibrillation as atria1 fibrillation without any detectable underlying heart disease. The mean follow-up period was 2476+1376 days. We defined two endpoints; namely, occurrence of ischemic stroke (endpoint l), and ischemic stroke or cardiac death (endpoint 2). The criteria for ischemic stroke were clinical signs or a medically confirmed history of an acute onset of ischemia in the cerebral arteries. Ischemic stroke was
Journal of Cardiology .56 (1996) 6/- 70
distinguished from intracerebral hemorrhage by computed tomographic brain scan. Since it was impossible to completely distinguish between thrombotic and embolic stroke, all ischemic strokes in patients with non-rheumatic atria1 fibrillation were recorded without respect to the probable stroke mechanism. Reports of stroke not witnessed by the attending physician or confirmed by a subsequent hospital admission were not accepted. Clinical, laboratory and echocardiographic features were obtained at the start of the follow-up and statistically analyzed. The variables consisted of age, sex, the level of total cholesterol, triglyceride and high density lipoprotein-cholesterol, diabetes mellitus, history of hypertension, systolic blood pressure, diastolic blood pressure, hematocrit, hemoglobin, platelet count, prothrombin time, the left atria1 dimension, left ventricular end-systolic dimension, left ventricular end-diastolic dimension, fractional shortening, ejection fraction, the thickness of the interventricular septum, the thickness of the left ventricular posterior wall, the ratio of the thickness of the interventricular septum to that of the left ventricular posterior wall by echocardiography and the administration of therapeutic drugs (warfarin, aspirin, other antiplatelets, calcium antagonists, pblockers and diuretics). All data were expressed as mean t 1 S.D. Baseline variables were compared by the unpaired t-test for continuous data and the xsquare test for categorical data. To identify which independent variables affected the prognosis, multivariate analysis was performed with fifteen prespecified clinical variables (age, sex, triglyceride level, diabetes mellitus, history of hypertension, systolic blood pressure, diastolic blood pressure, hematocrit, platelet count, left atria1 dimension, left ventricular end-diastolic dimension, left ventricular end-systolic dimension, fractional shortening, the thickness of the interventricular septum and the thickness of the left ventricular posterior wall) using the Cox proportional hazard model for endpoints 1 and 2 during the long-term follow-up period. Backward selection was used with a P value of 0.10 or greater required to remove a variable from the model. All pairwise interactions among the retained variables were evaluated simultaneously. The six therapeutic drugs were then added, one at a time, to the final model to determine whether they provided any additional statistically significant prediction of
M. Yoshida et al. I International
Journal of Cardiology
risk beyond that accounted for by the clinical variables. The Kaplan-Meier method was used to estimate event-free curves. The curves were compared using the log-rank test, and a P value of less than 0.05 was considered to indicate statistical significance.
3. Results One-hundred-and-twenty-two patients were included in this analysis. The patients’ backgrounds Table I Baseline characteristics of 122 Non-rheumatic atria1 fibrillation
56 (1996) 61-70
63
are shown in Table 1. During the follow-up period, 18 patients had ischemic stroke and 6 patients experienced cardiac death (4 acute myocardial infarction, 2 congestive heart failure). The underlying heart diseases in the patients who reached endpoint 1 were 6 lone atria1 fibrillation, 6 hypertension without any detectable heart disease, 2 sick sinus syndrome, 2 mitral regurgitation, 1 old myocardial infarction and 1 atrial septal defect and the 6 cases of cardiac death consisted of 4 old myocardial infarction, 1 hypertension without any detectable heart disease and 1 lone atria1 fibrillation. There was no systemic
patients and the univariate analysis regarding endpoint 1
Variables
All patients (n = 122)
IS(-) (n=104)
IS(+) (n=18)
P value
Age (yea=)
61.7’128 93:29 (76.2) 24765 1376 175241 119288 42.4-c 13.2 19 (15.6) 35 (28.7) 133+23 77215 42.455.1 14.021.8 19.325.9 11.751.5
60.3% 13.0 80:24 (76.9) 242621368 174i41 1121-58 42.42 13.9 15 (14.4) 27 (26.0) 131222 76215 42.225.2 14.02 1.8 19.556.2 11.721.5
69.628.5 13:5 (72.2) 2765?1426 180?41 1662 185 42.2k9.8 4 (22.2) 8 (44.4) 143226 83212 43.123.8 14.22 1.3 18.324.0 11.4z1.2
0.004* 0.894 0.337 0.563 0.017’ 0.953 0.624 0.187 0.045’ 0.053 0.511 0.586 0.433 0.482
43.059.3 51.858.1 35.857.3 31.1k7.7 63.7k12.1 9.31- 1.6 9.22 1.6 1.02?0.13
42.8k9.7 51.627.7 35.5k7.1 31.457.8 64.0t12.1 9.22 1.5 9.1?1.6 1.02t0.13
44.126.8 52.8-c 10.2 37.228.3 29.627.0 62.1211.9 10.2+1.7 9.8k1.5 1.04?0.07
0.575 0.552 0.357 0.373 0.55 I 0.014’ 0.057 0.6X9
58 25 18 42 12 48
48 (46.2) 24 (23.1) 14 (13.5) 36 (34.6) 9 (8.7) 38 (36.5)
10 (58.8) 1 (5.9) 4 (23.5) 6 (35.3) 3 (17.6) 10 (58.8)
0.479 0.194 0.475 0.999 0.476 0.156
Sex (M:F, % male) F/U days (day) TC (mg/dl) TG (mgldl) HDL-C (mg/dl) DM (no., %) HTN (no., %) BPS (mm&z) BPD (mmHg) Ht (%) Hb (gldl) PLT (/mm’) m- w Echocardiographic findings LAD (mm) LVDd (mm) LVDs (mm) FS (%) EF (%) IVST (mm) LVPWT (mm) ASH Therapy Warfarin (no., %) Aspirin (no., %) Other antiplatelets (no., W) Ca antagonists (no., %) P-Blockers (no., %) Diuretics (no., %)
(47.9) (20.7) (14.9) (34.7) (9.9) (39.7)
Abbreviations: ASH, asymmetric septal hypertrophy; BPD, diastolic blood pressure; BPS, systolic blood pressure; Ca, calcium; DM, diabetes mellitus; HTN, hypertension; EF, ejection fraction; FS, fractional shortening; F/U, follow-up; Hb, hemoglobin; HDL-C, HDL-cholesterol; Ht, hematocrit; IS(-), no ischemic stroke group; IS( +), ischemic stroke group; IVST, the thickness of the interventricular septum; LAD, left atrial dimension; LVDd, end-diastolic left ventricular dimension; LVDs, end-systolic left ventricular dimension; LVPWT. the thickness of the left ventricular posterior wall; M:F, male:female; PLT, platelet count: PT. prothrombin time: TC, total cholesterol; TG. triglyceride. *P
64
M. Yoshida et al. I International
Journal of Cardiology
arterial thromboembolism outside of the cerebral arteries during the follow-up period. The 5-year event-free rates for endpoints I and 2 were 87.4 and 85.0%, respectively (Fig. 1). Table 1 and Table 2 summarize the statistical analyses of all of the variables for endpoints 1 and 2, respectively. Univariate analysis showed that the patients who reached endpoint 1 were significantly older (mean age 69.6k8.5 vs. 60.3213.0; P
0.70
iI
0.60
x 2
(146?25 vs. 13Ok-21; P
------I--)
0.80 8
56 (1996) hl- 70
c, ---------3
I L------ee-w-----. -
ww
--------
IS or cD(FE)
0.50
0.00 !
t
0
1000
24loo
122
98
63
moo
4000
moo
18
2
&o
F/U Days No. al Risk
36
Fig. 1. Event-free rates for ischemic stroke (endpoint 1) and ischemic stroke or cardiac death (endpoint 2) in non-rheumaticatrial fibrillation. The event-free rate was determined by the Kaplan-Meier method. The mean duration of follow-up was 242651376 days. The 5-year event-free rates were 87.4 and 85.08, respectively. Eighteen patients had ischemic stroke and 6 patients experienced cardiac death during the follow-up. Abbreviations: IS, ischemic stroke; CD, cardiac death; EPl, endpoint 1; EP2, endpoint 2; F/U, follow-up.
M. Yoshida et al. I International
Journal of Cardiology
56 (1996) 61-70
65
Table 2 Univariate analysis regarding endpoint 2 Variables
IS/CD(-)
Age (year) Sex (M:F, % male) F/U days (day) TC (mg/dl) TG (mgldl) HDL-C (mg/dl) DM (no., %) HTN (no., %) BPS (mmHg) BPD (mmHg) Ht (%) Hb (g/dl) PLT (/mm’) PT 6) Echocardiographic findings LAD (mm) LVDd (mm) LVDs (mm) FS (%) EF (%) IVST (mm) LVPWT (mm) ASH Therapy Warfarin (no., %) Aspirin (no., %) Other antiplatelets (no., %) Ca antagonists (no., %) P-Blockers (no., %) Diuretics (no., %)
59.8t13.1 1622 (77.6) 247621370 175241 111257 43.2213.9 12 (12.2) 23 (23.5) 130221 75215 42.325.1 14.0tl.8 19.5k6.4 11.7+1.6
69.4k8.2 17:7 (70.8) 2474% 1430 176240 154-cl63 39.42 10.2 7 (29.2) 12 (50.0) 146225 84214 42.525.1 14.02 1.6 18.623.7 11.5+1.1
0.001* 0.67 I 0.944 0.941 0.036 0.238 0.0x3 0.020 0.002” O.OOC+ 0.866 0.9x I 0.528 0.559
42.6e9.7 51.7k7.7 35.3r7.1 31.927.7 64.8211.7 9.121.5 9.Ok1.5 1.02?0.13
44.527.4 52.3k9.5 37.5k7.9 28.157.2 59.22 12.9 10.121.6 9.92 1.I 1.02t0.09
0.37 I 0.754 0.192 0.03 I ’ 0.040 0.008* 0.010’ 0.950
46 (46.9) 22 (22.4) 14 (14.3) 31 (31.6) 8 (8.2) 33 (33.7)
12 (52.2) 3 (13.0) 4 (17.4) 11 (47.8) 4 (17.4) 15 (65.2)
0.826 0.474 0.959 0.221 0.345 0.006*
Abbreviations:
IS/CD(-),
(n=98)
Risk ratio
Endpoint 1: IS Age 1.106 LVDd LVDs IVST
0.882 1.149 1.493
Endpoint 2: IS or CD Age 1.122 LAD LVDd FS (%) LVPWT Abbreviations:
1.057 0.935 0.880 1.644
(n=24)
P value
neither ischemic stroke nor cardiac death group; IS/CD( +), either ischemic stroke or cardiac death group.
Table 3 Results of multivariate analysis of clinical predictors for endpoints 1 and 2 Risk factors
IS/CD(+)
95% Confidence limits
P value
1.038, 0.783, 1.012, 1.096,
1.188 0.994 1.305 2.034
0.0052 0.0393 0.0323 0.0111
1.052, 0.996, 0.876, 0.824, 1.248,
1.196 1.121 0.998 0.939 2.167
0.0004 0.0666 0.0426 0.0001 0.0004
CD, cardiac death; IS, ischemic stroke.
less than 9-mm thick (16-86 (18.6%) vs. 2-35 (5.7%)), there were no significant differences between the event-free curves for the thickness of the interventricular septum in endpoint 1 (Fig. 5). In contrast, the event-free curve for the thickness of the left ventricular posterior wall in endpoint 2 did show a significant difference (P
M. Yoshida et cd. I International
66
1.00 --=--I)--L0.90 -
1
L-,i
0.80 0)
0.70 -
2
0.60 -
zi 4
0.50 -
2
0.40 -
p' w
0.30 -
Journal of Cardiology 56 (1996) 61- 70
x--,
-------I
--7
5 L ________ ~ L-----.
Age<62(N=51) --------
&.&q+71)
p=O.o09O(Log-Rank
Test)
0.20 0.10 0.00 0
1000
I 2000
-
I 3000
-
I 4000
.
, 5000
.
I 6000
F/U Days
No. at Risk Age>62
51
44
29
21
14
Aged2
71
54
34
1.5
4
2
Fig. 2. Relationship between age and the event-free rate for endpoint 1. Patients who were 62 years or older more frequently experienced ischemic stroke than patients who were younger than 62 years. Abbreviations: Age <62, younger than 62 years; Age ?62,62 years or older.
of cerebral stroke increases with age [2-5,121. Our present study also revealed that older patients (62 years or more) more frequently experienced cerebral strokes than younger patients (<62 years). In a re-evaluation of data from the Framingham Study [14], atrial fibrillation by itself, systolic blood pressure,age, the presenceof ischemic heart disease and the presence of congestive heart failure were
4. Discussion The incidence of non-rheumatic atria1 fibrillation increases with age, and the arrhythmia is found in 1% of persons more than 60 years old [4,11,12]. Non-rheumatic atria1 fibrillation is the most common cardiac disease associated with cerebral embolism [ 133.Previous studies have shown that the incidence
0.80 -
-L----, --.I !; I.-------__, _______
d i:: z 0.50 $ 0.40 3 w’ 0.30 -
-
Age<62&51)
--------
&&4.&7,)
p=0.0026@.+Rank
Test)
i;Ij , , , I , , , , , . ( 0
1000
mO0
3000
4000
5000
2
6000
F/U Days
No. at Risk Age>62
51
44
29
21
14
Age*62
71
54
34
15
4
Fig. 3. Relationship between age and the event-free rate for endpoint 2. Patients who were 62 years or older more frequently experienced ischemic stroke or cardiac death than patients who were younger than 62 years.
M. Yoshida et al. I International
B
0.70 -
dP
0.60 -
& Jn
0.40 -
wif
0.30 -
67
Journal of Cardiology 56 (1996) 61-70
‘i
-
0.50 -
%FS<30(N=48)
--------
%FSdO(N=73)
p=O.O109 (Log-Rank
Test)
0.20 0.10 0.00 0
1000
2ooo
48 73
36 61
18 44
I 4000
3ooo
.
I 5000
.
1 6000
F/U Days No. at Risk %FS<30 %FS,30
9 28
4 14
1 1
Fig. 4. Relationship between fractional shortening and the event-free rate for endpoint 2. Patients with fractional shortening of 30% or more less frequently experienced endpoint 2 than patients with fractional shortening of less than 30%. Abbreviations: %FS, fractional shortening.
sex, heart failure, hypertension, diabetes, thyrotoxicosis, smoking nor chest pain were significant predictors [5,16]. In the SPAF study, recent congestive heart failure, a history of hypertension and previous arterial thromboembolism were each significantly and independently associated with a substantial risk for thromboembolism [17]. Although congestive heart failure and hypertension have been
independent risk factors for stroke. A comparison of stroke patients with and without atria1 fibrillation found that hypertension and ischemic heart disease were non-significant and congestive heart failure was a significant risk factor for stroke [ 151. The AFASAK study suggested that previous myocardial infarction was more prevalent in patients with thromboembolic events. On the other hand, neither age,
0.80 2
0.70 -
2
0.60 -
z 4 P t’
W
. _________, L-------------.
-
IVST+N=35)
0.50 -
--------
WST&3@J=%)
0.40 -
p=O.O793 (L@‘.mk
Test)
0.30 0.20 0.10 0.00 0
I. 1OOa
I 2000
.
I MOO
,
, 4000
I %loo
H 6000
F/U Days
No. at Risk IVST<9
35
30
20
11
4
WST,9
86
67
42
25
14
2
Fig. 5. Relationship between the thickness of the interventricular septum and the event-free rate for endpoint I. Although patients with an interventricular septum at least 9 mm thick had ischemic stroke more frequently than those in whom this was less than 9 mm thick, there were no significant differences in their event-free curves. Abbreviations: IVST, the thickness of the interventricular septum.
M. Yoshida et al. I International
of Cardiology 56 (1996) 61-70
Journal
1.00 7 0.90 0.80 d3
k-
x-Lm+ -L-- I _____,-_ 7
0.70 0.60 -
i, - - - - _ _ _ - -, L------e-----.
t 2.:
0.50 -
I
0.40
w’
0.30 -
-
-
LvPwr<9(N=42)
--------
LVPwrqN=79)
p=O.O368 (Log-Rank
Test)
0.20 0.10 0.00 0
1000
7mO
3000
4ooo
.5000
I 6000
F/U Days
No. at Risk LVPWT<9
42
38
24
13
5
LvPwb9
79
59
38
23
13
2
Fig. 6. Relationship between the thickness of the left ventricular posterior wall and the event-free rate for endpoint 2. Patients with a left ventricular posterior wall at least 9 mm thick had ischemic stroke more frequently than those in whom this was less than 9 mm thick. Abbreviations: LVPWT, the thickness of the left ventricular posterior wall.
reported to be risk factors in several studies, there have been conflicting results. A previous study reported that the hematocrit level was significantly higher in patients with cerebral infarction [1X]. In contrast, age was the only significant risk factor among variables other than echocardiographic findings and therapeutic drugs in our study. In general, the incidence of strokes in atria1 fibrillation is affected by the underlying heart disease, which includes coronary artery disease, cardiomyopathy and hypertensive heart disease,as well as rheumatic heart disease and valvular prosthesis [12,13]. In the Framingham Study [4], eight out of 30 patients with chronic lone atrial fibrillation experienced cerebrovascularevents, suggestinga rather high rate of embolism for patients with chronic lone atrial fibrillation. However, the overall risk of systemic embolism is lower in patients with chronic lone atrial fibrillation than in those With an underlying cardiac condition [ 19,201.In our study, 6 of the 18 cerebral stroke were lone atrial fibrillation and 8 of the 18 cerebral stroke (6 hypertension without any detectable heart disease and 2 sick sinus syndrome) had a history of hypertension, but it was not a significant risk factor in endpoint 1.by multivariate analysis. Although previous studies have found a
high risk of thromboembolic complications in patients with atrial fibrillation due to hyperthyroidism [16], none of the patients with this condition experienced cerebral thromboembolism during. the follow-up period in this study. This may be because patients with hyperthyroidism in our study were relatively younger (mean age 54.92 14.9 years) than those in other studies. Studies which have attempted to relate the left atrial dimension to stroke risk have yielded conflicting results. It is generally accepted that the left atrium is dilated in persons with atria1 fibrillation, but it is unclear whether left atria1 enlargement is a cause or a consequenceof the arrhythmia. In our study, left atial dimension was a significant risk factor in endpoint 2 by multivariate analysis, but was not a significant risk factor in endpoint 1. Some previous reports showed that large atria were more likely to embolize than small atria. However, the duration of atria1fibrillation was not defined, and the retrospective nature of these studies has been criticized [16,21-231. Other reports have suggestedthat left atria1dimension in patients with atrial fibrillation is not a reliable predictor of stroke [24,25]. This issue can be resolved through prospective studies of left atrial dimension prior to stroke. There have been
M. Yoshida et al. I International
Joumul of Cardiology
a few previous reports regarding echocardiographic findings in non-rheumatic atria1 fibrillation and stroke. In the SPAF study [26], left ventricular dysfunction in two-dimensional echocardiogramsand the size of the left atrium in M-mode echocardiograms were the strongest independent predictors of later thromboembolism. In the AFASAK study [5,27], there were no significant echocardiographic parameters that predicted thromboembolism. In our study, the thickness of the interventricular septum, end-diastolic left ventricular dimension and endsystolic left ventricular dimension were significant predictors in a multivariate analysis. The thickness of the left ventricular posterior wall, fractional shortening, end-diastolic left ventricular dimension and left atrial dimension were significant predictors when cardiac death was also considered an endpoint of this analysis (endpoint 2). Fractional shortening and enddiastolic left ventricular dimension were negative risk factors. Though it has been previously demonstrated from some studies that there is an association between ischemic strokes and the extent of left ventricular dysfunction [26,28], the significance of end-diastolic left ventricular dimension in our study is unclear. The thickness of the interventricular septum and that of the left ventricular posterior wall have never been recognized as predictors of stroke in previous studies.We excluded casesof hypertrophic cardiomyopathy based on the ratio of the thickness of the interventricular septum to that of the left ventricular posterior wall, even if they had not been previously diagnosed as hypertrophic cardiomyopathy. Based on our results, the thickness of the left ventricular wall and the left ventricular dimension affected the prognosis. However, the mechanismthrough which the left ventricular dimension is a risk factor for ischemic stroke is unclear. Although neither warfarin nor aspirin had a significant impact on stroke prevention in atrial fibrillation patients in our retrospective study, recent randomized trials involving antithrombotic therapy with warfarin and aspirin have reported conclusive results [5-lo]. These reports revealed that antithrombotic therapy with warfarin substantially reduced the occurrence of ischemic strokes and systemic embolism in patients with atrial fibrillation. However, it is unclear whether aspirin is effective in preventing stroke in atria1 fibrillation patients. We could not
56 (1996) 61-70
69
determine the efficacy of these drugs becauseof the small number of atrial fibrillation patients and the retrospective nature of our study. The present study revealed that the thickness of the left ventricular wall and the left ventricular dimension by echocardiography were significant independent predictors of stroke in non-rheumatic atrial fibrillation patients. Previous studies have not identified the thickness of the left ventricular wall as a risk factor for stroke in patients with non-rheumatic atrial fibrillation. Non-rheumatic atrial fibrillation patients with these risk factors should receive prophylactic antithrombotic therapy to reduce the risk of stroke.
References HI Askey JM, Bernstein S. The management of rheumatic heart disease in relation to systemic arterial embolism. Prog Cardiovasc Dis 1960, 3: 220-232. I21Wolf PA, Dawber TR, Thomas HE Jr, Kannel WB. Epidemiologic assessment of chronic atrial fibrillation and risk of stroke: The Framingham Study. Neurology 1978; 28: 973-977. [31 Wolf PA, Abbott RD, Kannel WB. Atria1 fibrillation: a major contributor to stroke in the elderly: The Framingham Study. Arch Intern Med 1987; 147: 1561-1564. 141 Kannel WB, Abbott RD, Savage DD, McNamara PM. Epidemiologic features of chronic atrial fibrillation: The Framingham Study. N Engl J Med 1982; 306: 1018-1022. [51 Petersen P, Boysen G, Godtfredsen J, Andersen ED, Andersen B. Placebo-controlled, randomized trial of warfarin and aspirin for prevention of thrombotic complications in chronic atria1 fibrillation: The Copenhagen AFASAK Study. Lancet 1989; 1: 175-179. WI The Boston Area Anticoagulation Trial for Atrial Fibrillation Investigators. The effect of low-dose warfarin on the risk of stroke in patients with non-rheumatic atria1 fibrillation. N Engl J Med 1990; 323: 1505-1511. 171 Stroke Prevention in Atria1 Fibrillation Investigators. Stroke Prevention in Atria1 Fibrillation Study: final results. Circulation 1991; 84: 527-539. PI Connolly SJ, Laupacis A, Gent M, Roberts RS, Cairns JA, Joyner C. Canadian Atrial Fibrillation Anticoagulation (CAFA) Study. J Am Co11 Cardiol 1991; 18: 349-355. [91 Ezekowitz MD, Bridgers SL, James KE, et al. Warfarin in the prevention of stroke associated with non-rheumatic atria1 fibrillation. N Engl J Med 1992; 327: 1406-1412. [lOI Stroke Prevention in Atrial Fibrillation Investigators. Warfarin versus aspirin for prevention of thromboembolism in atria1 fibrillation: Stroke Prevention in Atria1 Fibrillation II Study. Lancet 1994; 343: 687-691.
70
M. Yoshida et al. I International
(I I] Kuramoto K, Matsushita, S, Yamanouchi H. Atria1 fibrillation as a cause of myocardial and cerebral infarctions. Jpn Circ J 1984; 48: 67-74. 1121 Zipes DP. Atrial fibrillation. In: Braunwald E, ed., Heart Disease: A Textbook of Cardiovascular Medicine. 4th ed, Philadelphia: W.B. Saunders Company, 1992: 682-684. [13] Sherman DG, Dyken ML, Fisher M, Harrison MJG, Hart RG. Antithrombotic therapy for cerebrovascular disorders. Chest 1989; 95(Suppl.l): 14OS-15%. [ 141 Wolf PA, Kannel WB, McGee DL, Meeks SL, Bharucha NE, McNamara PM. Duration of atrial fibrillation and imminence of stroke: The Framingham study. Stroke 1983; 14: 664667. [ 151 Britton M, Gustafsson C. Non-rheumatic atria1 fibrillation as a risk factor for stroke. Stroke 1985; 16: 182-188. [16] Peterson P. Thromboembolic complications in atria1 fibrillation. Stroke 1990; 21: 4-13. [ 171 The Stroke Prevention in Atria1 Fibrillation Investigators. Predictors of thromboembolism in atria1 fibrillation: I. Clinical features of patients at risk. Ann Intern Med 1992; 116: l-5. [ 181 Naito S, Imataka K, Seko Y, Fujii J. The predictive factors for cerebral infarction in patients with non-rheumatic atria1 fibrillation. J Cardiol 1990; 20: 385-390 (in Japanese). [19] Brand FN, Abbott RD, Kennel WB, Wolf PA. Characteristics and prognosis of lone atrial fibrillation: The Framingham Study. J Am Med Assoc 1985; 254: 3449-3453. [20] Kopecky SL, Gersh BJ, McGoon MD, et al. The natural history of lone atria1 fibrillation. N Engl J Med 1987; 317: 669-674.
Journal of Cardiolog)
Xi (1996) 61-70
[21] Moss AJ. Atria1 fibrillation and cerebral embolism (Edilorial). Arch Neurol 1984; 41: 707. [22] Caplan LR, D’Cruz I, Hier DB, Reddy H, Shah S. Atrial size, atria1 fibrillation and stroke. Ann Neurol 1986: 19: 158-161. [23] Cabin HS, Clubb KS, Hall C, Perlmutter RA, Feinstein AR. Risk for systemic embolization of atrial fibrillation without mitral stenosis. Am J Cardiol 1990; 65: 1112-I 116. 1241 Tegeler CH, Hart RG, Sherman DG, Quiroga ES, Easton JD. Atrial fibrillation-stroke study (abstr). Stroke 1986; 17: 131. [25] Wiener I. Clinical and echocardiographic correlates of systemic embolization in non-rheumatic atria1 fibrillation. Am J Cardiol 1987; 59: 177. [26] The Stroke Prevention in Atria1 Fibrillation Investigators. Predictors of thromboembolism in atrial fibrillation: II. Echocardiographic features of patients at risk. Ann Intern Med 1992; 116: 6-12. [27] Peterson P, Kastrup J, Helweg-Larson S, Boysen G, Godtfredsen J. Risk factors for thromboembolic complications in chronic atria1 fibrillation: The Copenhagen AFASAK study. Arch Intern Med 1990; 150: 819-821. [28] Dunkman WB, Johnson GR, Carson PE, Bhat G, Farrell L, Cohn JN, for the V-Hem VA Co-operative Studies Group. Incidence of Thromboembolic Events in Congestive Heart Failure. Circulation 1993; 87(Suppl. VI): V194-101.
Journal
of
cardiology EISEVIER
International Journal of Cardiology 56 (1996) 6 l-70
Predictors of ischemic stroke in non-rheumatic atria1 fibrillation Michiaki Yoshida, Yasuyuki Nakamura”, Masahito Higashikawa, Masahiko Kinoshita First Department of Internal Medicine, Shiga University of Medical Science, Tsukinowa, Seta, Otsu, Shiga 520-21, Japan
Received 17 October 1995; revised 13 February 1996; accepted 21 May 1996
Abstract We retrospectively analyzed the clinical features of patients with non-rheumatic atria1 fibrillation to identify risk factors of ischemic stroke. Non-rheumatic atrial fibrillation is associated with an increased risk of ischemic stroke. However, the predictors of ischemic stroke in non-rheumatic atria1 fibrillation are unclear. The study population consisted of 122 patients with non-rheumatic atrial fibrillation who had no previous clinical cerebral strokes at the start of the follow-up. Patients with cardiomyopathy and paroxysmal or intermittent atria1 fibrillation were excluded from the study. The mean age was 61.7-C 12.8 years. We defined two endpoints; namely, occurrence of ischemic stroke (endpoint l), and ischemic stroke or cardiac death (endpoint 2). During the follow-up, 18 patients had ischemic stroke and 6 patients experienced cardiac death. The 5-year event-free rates for endpoints 1 and 2 were 87.4% and 85.0%, respectively. A Cox analysis revealed that endpoint 1 was significantly associated with age (risk ratio (RR)= 1.106, P=O.O052), end-diastolic left ventricular dimension (RR=O.882, P=O.O393), end-systolic left ventricular dimension (RR=1.149, P=O.O323) and the thickness of the interventricular septum (RR=1.493, P=O.Olll). Endpoint 2 was associated with age (RR=1.122, P=O.O004), left atria1 dimension (RR= 1.057, P=O.O666), end-diastolic left ventricular dimension (RR=O.935, P=O.O426), fractional shortening (RR=O.880, P=O.OOOl) and the thickness of the left ventricular posterior wall (RR= 1.644, P=O.O004). The present results suggest that, in addition to left ventricular dimensions and left atrial dimension, left ventricular hypertrophy may be associated with ischemic stroke. Keywords:
Non-rheumatic atria1 fibrillation;
Ischemic stroke; Predictors
1. Introduction Atrial fibrillation associated with rheumatic heart disease and mitral stenosis is well known to predispose the patient to systemic arterial embolism [l]. In addition, non-rheumatic atria1 fibrillation is associated with au increased risk for ischemic stroke [2,3]. *Corresponding author, Tel.: +8I 775 48 2213; fax: +81 775 43 5839.
In a population-based study [2,4], the risk for ischemic stroke among patients with non-rheumatic atrial fibrillation was estimated to be 5 times greater than that for comparable patients in sinus rhythm. Most of these strokes have the clinical characteristics of cerebral embolism. Recent randomized trials involving antitbrombotic aspirin in non-rheumatic
ported that antithrombotic therapy substantially reduces the occurrence of ischemic strokes and sys-
0167-5273/96/$15.00 0 1996 Elsevier Science Ireland Ltd. All rights reserved PII
SO167-5273(96)02726-X
therapy with warfarin and atrial fibrillation have re-
62
M. Yoshida et al. I International
temic embolism in patients with atria1 fibrillation [5-lo]. However, the predictors of ischemic stroke in non-rheumatic atria1 fibrillation are still unclear. We retrospectively analyzed the clinical features of patients with non-rheumatic atria1 fibrillation to identify risk factors for ischemic stroke.
2. Materials and methods We analyzed 122 patients with non-rheumatic atria1 fibrillation who had no previous clinical cerebral strokes as of the start of the follow-up. Adults with chronic sustained atria1 fibrillation with no evidence of mitral stenosis on two-dimensional echocardiography were eligible. Patients with hycardiomyopathy pertrophic and dilated cardiomyopathy were excluded from the study. Patients were also excluded if they had a finding of asymmetric septal hypertrophy with a ratio of the thickness of the interventricular septum to that of the left ventricular posterior wall of 1.30 or greater by echocardiography. This study did not include patients with paroxysmal or intermittent atria1 fibrillation. Other reasons for exclusion were echocardiographic evidence of an intracardiac thrombus, a left ventricular aneurysm, the presence of severe congestive heart failure, or prosthetic heart valves. The follow-up was terminated on 30 June 1994. The mean age of the 122 patients (93 males, 29 females) was 61.75 12.8 years (range 24-90). The underlying heart diseases consisted of 23 hypertension without any detectable heart disease, 16 ischemic heart diseases (8 old myocardial infarction and 8 angina pectoris), 10 sick sinus syndrome, 9 non-rheumatic valvular diseases (4 mitral valve prolapse, 4 mitral regurgitation and 1 aortic regurgitation), 8 congenital heart diseases, 7 hyperthyroidism, 2 chronic obstructive pulmonary diseases, 1 polymyositis, 1 constrictive pericarditis and 45 lone atria1 fibrillation. In this study, we defined lone atrial fibrillation as atria1 fibrillation without any detectable underlying heart disease. The mean follow-up period was 2476+1376 days. We defined two endpoints; namely, occurrence of ischemic stroke (endpoint l), and ischemic stroke or cardiac death (endpoint 2). The criteria for ischemic stroke were clinical signs or a medically confirmed history of an acute onset of ischemia in the cerebral arteries. Ischemic stroke was
Journal of Cardiology .56 (1996) 6/- 70
distinguished from intracerebral hemorrhage by computed tomographic brain scan. Since it was impossible to completely distinguish between thrombotic and embolic stroke, all ischemic strokes in patients with non-rheumatic atria1 fibrillation were recorded without respect to the probable stroke mechanism. Reports of stroke not witnessed by the attending physician or confirmed by a subsequent hospital admission were not accepted. Clinical, laboratory and echocardiographic features were obtained at the start of the follow-up and statistically analyzed. The variables consisted of age, sex, the level of total cholesterol, triglyceride and high density lipoprotein-cholesterol, diabetes mellitus, history of hypertension, systolic blood pressure, diastolic blood pressure, hematocrit, hemoglobin, platelet count, prothrombin time, the left atria1 dimension, left ventricular end-systolic dimension, left ventricular end-diastolic dimension, fractional shortening, ejection fraction, the thickness of the interventricular septum, the thickness of the left ventricular posterior wall, the ratio of the thickness of the interventricular septum to that of the left ventricular posterior wall by echocardiography and the administration of therapeutic drugs (warfarin, aspirin, other antiplatelets, calcium antagonists, pblockers and diuretics). All data were expressed as mean t 1 S.D. Baseline variables were compared by the unpaired t-test for continuous data and the xsquare test for categorical data. To identify which independent variables affected the prognosis, multivariate analysis was performed with fifteen prespecified clinical variables (age, sex, triglyceride level, diabetes mellitus, history of hypertension, systolic blood pressure, diastolic blood pressure, hematocrit, platelet count, left atria1 dimension, left ventricular end-diastolic dimension, left ventricular end-systolic dimension, fractional shortening, the thickness of the interventricular septum and the thickness of the left ventricular posterior wall) using the Cox proportional hazard model for endpoints 1 and 2 during the long-term follow-up period. Backward selection was used with a P value of 0.10 or greater required to remove a variable from the model. All pairwise interactions among the retained variables were evaluated simultaneously. The six therapeutic drugs were then added, one at a time, to the final model to determine whether they provided any additional statistically significant prediction of
M. Yoshida et al. I International
Journal of Cardiology
risk beyond that accounted for by the clinical variables. The Kaplan-Meier method was used to estimate event-free curves. The curves were compared using the log-rank test, and a P value of less than 0.05 was considered to indicate statistical significance.
3. Results One-hundred-and-twenty-two patients were included in this analysis. The patients’ backgrounds Table I Baseline characteristics of 122 Non-rheumatic atria1 fibrillation
56 (1996) 61-70
63
are shown in Table 1. During the follow-up period, 18 patients had ischemic stroke and 6 patients experienced cardiac death (4 acute myocardial infarction, 2 congestive heart failure). The underlying heart diseases in the patients who reached endpoint 1 were 6 lone atria1 fibrillation, 6 hypertension without any detectable heart disease, 2 sick sinus syndrome, 2 mitral regurgitation, 1 old myocardial infarction and 1 atrial septal defect and the 6 cases of cardiac death consisted of 4 old myocardial infarction, 1 hypertension without any detectable heart disease and 1 lone atria1 fibrillation. There was no systemic
patients and the univariate analysis regarding endpoint 1
Variables
All patients (n = 122)
IS(-) (n=104)
IS(+) (n=18)
P value
Age (yea=)
61.7’128 93:29 (76.2) 24765 1376 175241 119288 42.4-c 13.2 19 (15.6) 35 (28.7) 133+23 77215 42.455.1 14.021.8 19.325.9 11.751.5
60.3% 13.0 80:24 (76.9) 242621368 174i41 1121-58 42.42 13.9 15 (14.4) 27 (26.0) 131222 76215 42.225.2 14.02 1.8 19.556.2 11.721.5
69.628.5 13:5 (72.2) 2765?1426 180?41 1662 185 42.2k9.8 4 (22.2) 8 (44.4) 143226 83212 43.123.8 14.22 1.3 18.324.0 11.4z1.2
0.004* 0.894 0.337 0.563 0.017’ 0.953 0.624 0.187 0.045’ 0.053 0.511 0.586 0.433 0.482
43.059.3 51.858.1 35.857.3 31.1k7.7 63.7k12.1 9.31- 1.6 9.22 1.6 1.02?0.13
42.8k9.7 51.627.7 35.5k7.1 31.457.8 64.0t12.1 9.22 1.5 9.1?1.6 1.02t0.13
44.126.8 52.8-c 10.2 37.228.3 29.627.0 62.1211.9 10.2+1.7 9.8k1.5 1.04?0.07
0.575 0.552 0.357 0.373 0.55 I 0.014’ 0.057 0.6X9
58 25 18 42 12 48
48 (46.2) 24 (23.1) 14 (13.5) 36 (34.6) 9 (8.7) 38 (36.5)
10 (58.8) 1 (5.9) 4 (23.5) 6 (35.3) 3 (17.6) 10 (58.8)
0.479 0.194 0.475 0.999 0.476 0.156
Sex (M:F, % male) F/U days (day) TC (mg/dl) TG (mgldl) HDL-C (mg/dl) DM (no., %) HTN (no., %) BPS (mm&z) BPD (mmHg) Ht (%) Hb (gldl) PLT (/mm’) m- w Echocardiographic findings LAD (mm) LVDd (mm) LVDs (mm) FS (%) EF (%) IVST (mm) LVPWT (mm) ASH Therapy Warfarin (no., %) Aspirin (no., %) Other antiplatelets (no., W) Ca antagonists (no., %) P-Blockers (no., %) Diuretics (no., %)
(47.9) (20.7) (14.9) (34.7) (9.9) (39.7)
Abbreviations: ASH, asymmetric septal hypertrophy; BPD, diastolic blood pressure; BPS, systolic blood pressure; Ca, calcium; DM, diabetes mellitus; HTN, hypertension; EF, ejection fraction; FS, fractional shortening; F/U, follow-up; Hb, hemoglobin; HDL-C, HDL-cholesterol; Ht, hematocrit; IS(-), no ischemic stroke group; IS( +), ischemic stroke group; IVST, the thickness of the interventricular septum; LAD, left atrial dimension; LVDd, end-diastolic left ventricular dimension; LVDs, end-systolic left ventricular dimension; LVPWT. the thickness of the left ventricular posterior wall; M:F, male:female; PLT, platelet count: PT. prothrombin time: TC, total cholesterol; TG. triglyceride. *P
64
M. Yoshida et al. I International
Journal of Cardiology
arterial thromboembolism outside of the cerebral arteries during the follow-up period. The 5-year event-free rates for endpoints I and 2 were 87.4 and 85.0%, respectively (Fig. 1). Table 1 and Table 2 summarize the statistical analyses of all of the variables for endpoints 1 and 2, respectively. Univariate analysis showed that the patients who reached endpoint 1 were significantly older (mean age 69.6k8.5 vs. 60.3213.0; P
0.70
iI
0.60
x 2
(146?25 vs. 13Ok-21; P
------I--)
0.80 8
56 (1996) hl- 70
c, ---------3
I L------ee-w-----. -
ww
--------
IS or cD(FE)
0.50
0.00 !
t
0
1000
24loo
122
98
63
moo
4000
moo
18
2
&o
F/U Days No. al Risk
36
Fig. 1. Event-free rates for ischemic stroke (endpoint 1) and ischemic stroke or cardiac death (endpoint 2) in non-rheumaticatrial fibrillation. The event-free rate was determined by the Kaplan-Meier method. The mean duration of follow-up was 242651376 days. The 5-year event-free rates were 87.4 and 85.08, respectively. Eighteen patients had ischemic stroke and 6 patients experienced cardiac death during the follow-up. Abbreviations: IS, ischemic stroke; CD, cardiac death; EPl, endpoint 1; EP2, endpoint 2; F/U, follow-up.
M. Yoshida et al. I International
Journal of Cardiology
56 (1996) 61-70
65
Table 2 Univariate analysis regarding endpoint 2 Variables
IS/CD(-)
Age (year) Sex (M:F, % male) F/U days (day) TC (mg/dl) TG (mgldl) HDL-C (mg/dl) DM (no., %) HTN (no., %) BPS (mmHg) BPD (mmHg) Ht (%) Hb (g/dl) PLT (/mm’) PT 6) Echocardiographic findings LAD (mm) LVDd (mm) LVDs (mm) FS (%) EF (%) IVST (mm) LVPWT (mm) ASH Therapy Warfarin (no., %) Aspirin (no., %) Other antiplatelets (no., %) Ca antagonists (no., %) P-Blockers (no., %) Diuretics (no., %)
59.8t13.1 1622 (77.6) 247621370 175241 111257 43.2213.9 12 (12.2) 23 (23.5) 130221 75215 42.325.1 14.0tl.8 19.5k6.4 11.7+1.6
69.4k8.2 17:7 (70.8) 2474% 1430 176240 154-cl63 39.42 10.2 7 (29.2) 12 (50.0) 146225 84214 42.525.1 14.02 1.6 18.623.7 11.5+1.1
0.001* 0.67 I 0.944 0.941 0.036 0.238 0.0x3 0.020 0.002” O.OOC+ 0.866 0.9x I 0.528 0.559
42.6e9.7 51.7k7.7 35.3r7.1 31.927.7 64.8211.7 9.121.5 9.Ok1.5 1.02?0.13
44.527.4 52.3k9.5 37.5k7.9 28.157.2 59.22 12.9 10.121.6 9.92 1.I 1.02t0.09
0.37 I 0.754 0.192 0.03 I ’ 0.040 0.008* 0.010’ 0.950
46 (46.9) 22 (22.4) 14 (14.3) 31 (31.6) 8 (8.2) 33 (33.7)
12 (52.2) 3 (13.0) 4 (17.4) 11 (47.8) 4 (17.4) 15 (65.2)
0.826 0.474 0.959 0.221 0.345 0.006*
Abbreviations:
IS/CD(-),
(n=98)
Risk ratio
Endpoint 1: IS Age 1.106 LVDd LVDs IVST
0.882 1.149 1.493
Endpoint 2: IS or CD Age 1.122 LAD LVDd FS (%) LVPWT Abbreviations:
1.057 0.935 0.880 1.644
(n=24)
P value
neither ischemic stroke nor cardiac death group; IS/CD( +), either ischemic stroke or cardiac death group.
Table 3 Results of multivariate analysis of clinical predictors for endpoints 1 and 2 Risk factors
IS/CD(+)
95% Confidence limits
P value
1.038, 0.783, 1.012, 1.096,
1.188 0.994 1.305 2.034
0.0052 0.0393 0.0323 0.0111
1.052, 0.996, 0.876, 0.824, 1.248,
1.196 1.121 0.998 0.939 2.167
0.0004 0.0666 0.0426 0.0001 0.0004
CD, cardiac death; IS, ischemic stroke.
less than 9-mm thick (16-86 (18.6%) vs. 2-35 (5.7%)), there were no significant differences between the event-free curves for the thickness of the interventricular septum in endpoint 1 (Fig. 5). In contrast, the event-free curve for the thickness of the left ventricular posterior wall in endpoint 2 did show a significant difference (P
M. Yoshida et cd. I International
66
1.00 --=--I)--L0.90 -
1
L-,i
0.80 0)
0.70 -
2
0.60 -
zi 4
0.50 -
2
0.40 -
p' w
0.30 -
Journal of Cardiology 56 (1996) 61- 70
x--,
-------I
--7
5 L ________ ~ L-----.
Age<62(N=51) --------
&.&q+71)
p=O.o09O(Log-Rank
Test)
0.20 0.10 0.00 0
1000
I 2000
-
I 3000
-
I 4000
.
, 5000
.
I 6000
F/U Days
No. at Risk Age>62
51
44
29
21
14
Aged2
71
54
34
1.5
4
2
Fig. 2. Relationship between age and the event-free rate for endpoint 1. Patients who were 62 years or older more frequently experienced ischemic stroke than patients who were younger than 62 years. Abbreviations: Age <62, younger than 62 years; Age ?62,62 years or older.
of cerebral stroke increases with age [2-5,121. Our present study also revealed that older patients (62 years or more) more frequently experienced cerebral strokes than younger patients (<62 years). In a re-evaluation of data from the Framingham Study [14], atrial fibrillation by itself, systolic blood pressure,age, the presenceof ischemic heart disease and the presence of congestive heart failure were
4. Discussion The incidence of non-rheumatic atria1 fibrillation increases with age, and the arrhythmia is found in 1% of persons more than 60 years old [4,11,12]. Non-rheumatic atria1 fibrillation is the most common cardiac disease associated with cerebral embolism [ 133.Previous studies have shown that the incidence
0.80 -
-L----, --.I !; I.-------__, _______
d i:: z 0.50 $ 0.40 3 w’ 0.30 -
-
Age<62&51)
--------
&&4.&7,)
p=0.0026@.+Rank
Test)
i;Ij , , , I , , , , , . ( 0
1000
mO0
3000
4000
5000
2
6000
F/U Days
No. at Risk Age>62
51
44
29
21
14
Age*62
71
54
34
15
4
Fig. 3. Relationship between age and the event-free rate for endpoint 2. Patients who were 62 years or older more frequently experienced ischemic stroke or cardiac death than patients who were younger than 62 years.
M. Yoshida et al. I International
B
0.70 -
dP
0.60 -
& Jn
0.40 -
wif
0.30 -
67
Journal of Cardiology 56 (1996) 61-70
‘i
-
0.50 -
%FS<30(N=48)
--------
%FSdO(N=73)
p=O.O109 (Log-Rank
Test)
0.20 0.10 0.00 0
1000
2ooo
48 73
36 61
18 44
I 4000
3ooo
.
I 5000
.
1 6000
F/U Days No. at Risk %FS<30 %FS,30
9 28
4 14
1 1
Fig. 4. Relationship between fractional shortening and the event-free rate for endpoint 2. Patients with fractional shortening of 30% or more less frequently experienced endpoint 2 than patients with fractional shortening of less than 30%. Abbreviations: %FS, fractional shortening.
sex, heart failure, hypertension, diabetes, thyrotoxicosis, smoking nor chest pain were significant predictors [5,16]. In the SPAF study, recent congestive heart failure, a history of hypertension and previous arterial thromboembolism were each significantly and independently associated with a substantial risk for thromboembolism [17]. Although congestive heart failure and hypertension have been
independent risk factors for stroke. A comparison of stroke patients with and without atria1 fibrillation found that hypertension and ischemic heart disease were non-significant and congestive heart failure was a significant risk factor for stroke [ 151. The AFASAK study suggested that previous myocardial infarction was more prevalent in patients with thromboembolic events. On the other hand, neither age,
0.80 2
0.70 -
2
0.60 -
z 4 P t’
W
. _________, L-------------.
-
IVST+N=35)
0.50 -
--------
WST&3@J=%)
0.40 -
p=O.O793 (L@‘.mk
Test)
0.30 0.20 0.10 0.00 0
I. 1OOa
I 2000
.
I MOO
,
, 4000
I %loo
H 6000
F/U Days
No. at Risk IVST<9
35
30
20
11
4
WST,9
86
67
42
25
14
2
Fig. 5. Relationship between the thickness of the interventricular septum and the event-free rate for endpoint I. Although patients with an interventricular septum at least 9 mm thick had ischemic stroke more frequently than those in whom this was less than 9 mm thick, there were no significant differences in their event-free curves. Abbreviations: IVST, the thickness of the interventricular septum.
M. Yoshida et al. I International
of Cardiology 56 (1996) 61-70
Journal
1.00 7 0.90 0.80 d3
k-
x-Lm+ -L-- I _____,-_ 7
0.70 0.60 -
i, - - - - _ _ _ - -, L------e-----.
t 2.:
0.50 -
I
0.40
w’
0.30 -
-
-
LvPwr<9(N=42)
--------
LVPwrqN=79)
p=O.O368 (Log-Rank
Test)
0.20 0.10 0.00 0
1000
7mO
3000
4ooo
.5000
I 6000
F/U Days
No. at Risk LVPWT<9
42
38
24
13
5
LvPwb9
79
59
38
23
13
2
Fig. 6. Relationship between the thickness of the left ventricular posterior wall and the event-free rate for endpoint 2. Patients with a left ventricular posterior wall at least 9 mm thick had ischemic stroke more frequently than those in whom this was less than 9 mm thick. Abbreviations: LVPWT, the thickness of the left ventricular posterior wall.
reported to be risk factors in several studies, there have been conflicting results. A previous study reported that the hematocrit level was significantly higher in patients with cerebral infarction [1X]. In contrast, age was the only significant risk factor among variables other than echocardiographic findings and therapeutic drugs in our study. In general, the incidence of strokes in atria1 fibrillation is affected by the underlying heart disease, which includes coronary artery disease, cardiomyopathy and hypertensive heart disease,as well as rheumatic heart disease and valvular prosthesis [12,13]. In the Framingham Study [4], eight out of 30 patients with chronic lone atrial fibrillation experienced cerebrovascularevents, suggestinga rather high rate of embolism for patients with chronic lone atrial fibrillation. However, the overall risk of systemic embolism is lower in patients with chronic lone atrial fibrillation than in those With an underlying cardiac condition [ 19,201.In our study, 6 of the 18 cerebral stroke were lone atrial fibrillation and 8 of the 18 cerebral stroke (6 hypertension without any detectable heart disease and 2 sick sinus syndrome) had a history of hypertension, but it was not a significant risk factor in endpoint 1.by multivariate analysis. Although previous studies have found a
high risk of thromboembolic complications in patients with atrial fibrillation due to hyperthyroidism [16], none of the patients with this condition experienced cerebral thromboembolism during. the follow-up period in this study. This may be because patients with hyperthyroidism in our study were relatively younger (mean age 54.92 14.9 years) than those in other studies. Studies which have attempted to relate the left atrial dimension to stroke risk have yielded conflicting results. It is generally accepted that the left atrium is dilated in persons with atria1 fibrillation, but it is unclear whether left atria1 enlargement is a cause or a consequenceof the arrhythmia. In our study, left atial dimension was a significant risk factor in endpoint 2 by multivariate analysis, but was not a significant risk factor in endpoint 1. Some previous reports showed that large atria were more likely to embolize than small atria. However, the duration of atria1fibrillation was not defined, and the retrospective nature of these studies has been criticized [16,21-231. Other reports have suggestedthat left atria1dimension in patients with atrial fibrillation is not a reliable predictor of stroke [24,25]. This issue can be resolved through prospective studies of left atrial dimension prior to stroke. There have been
M. Yoshida et al. I International
Joumul of Cardiology
a few previous reports regarding echocardiographic findings in non-rheumatic atria1 fibrillation and stroke. In the SPAF study [26], left ventricular dysfunction in two-dimensional echocardiogramsand the size of the left atrium in M-mode echocardiograms were the strongest independent predictors of later thromboembolism. In the AFASAK study [5,27], there were no significant echocardiographic parameters that predicted thromboembolism. In our study, the thickness of the interventricular septum, end-diastolic left ventricular dimension and endsystolic left ventricular dimension were significant predictors in a multivariate analysis. The thickness of the left ventricular posterior wall, fractional shortening, end-diastolic left ventricular dimension and left atrial dimension were significant predictors when cardiac death was also considered an endpoint of this analysis (endpoint 2). Fractional shortening and enddiastolic left ventricular dimension were negative risk factors. Though it has been previously demonstrated from some studies that there is an association between ischemic strokes and the extent of left ventricular dysfunction [26,28], the significance of end-diastolic left ventricular dimension in our study is unclear. The thickness of the interventricular septum and that of the left ventricular posterior wall have never been recognized as predictors of stroke in previous studies.We excluded casesof hypertrophic cardiomyopathy based on the ratio of the thickness of the interventricular septum to that of the left ventricular posterior wall, even if they had not been previously diagnosed as hypertrophic cardiomyopathy. Based on our results, the thickness of the left ventricular wall and the left ventricular dimension affected the prognosis. However, the mechanismthrough which the left ventricular dimension is a risk factor for ischemic stroke is unclear. Although neither warfarin nor aspirin had a significant impact on stroke prevention in atrial fibrillation patients in our retrospective study, recent randomized trials involving antithrombotic therapy with warfarin and aspirin have reported conclusive results [5-lo]. These reports revealed that antithrombotic therapy with warfarin substantially reduced the occurrence of ischemic strokes and systemic embolism in patients with atrial fibrillation. However, it is unclear whether aspirin is effective in preventing stroke in atria1 fibrillation patients. We could not
56 (1996) 61-70
69
determine the efficacy of these drugs becauseof the small number of atrial fibrillation patients and the retrospective nature of our study. The present study revealed that the thickness of the left ventricular wall and the left ventricular dimension by echocardiography were significant independent predictors of stroke in non-rheumatic atrial fibrillation patients. Previous studies have not identified the thickness of the left ventricular wall as a risk factor for stroke in patients with non-rheumatic atrial fibrillation. Non-rheumatic atrial fibrillation patients with these risk factors should receive prophylactic antithrombotic therapy to reduce the risk of stroke.
References HI Askey JM, Bernstein S. The management of rheumatic heart disease in relation to systemic arterial embolism. Prog Cardiovasc Dis 1960, 3: 220-232. I21Wolf PA, Dawber TR, Thomas HE Jr, Kannel WB. Epidemiologic assessment of chronic atrial fibrillation and risk of stroke: The Framingham Study. Neurology 1978; 28: 973-977. [31 Wolf PA, Abbott RD, Kannel WB. Atria1 fibrillation: a major contributor to stroke in the elderly: The Framingham Study. Arch Intern Med 1987; 147: 1561-1564. 141 Kannel WB, Abbott RD, Savage DD, McNamara PM. Epidemiologic features of chronic atrial fibrillation: The Framingham Study. N Engl J Med 1982; 306: 1018-1022. [51 Petersen P, Boysen G, Godtfredsen J, Andersen ED, Andersen B. Placebo-controlled, randomized trial of warfarin and aspirin for prevention of thrombotic complications in chronic atria1 fibrillation: The Copenhagen AFASAK Study. Lancet 1989; 1: 175-179. WI The Boston Area Anticoagulation Trial for Atrial Fibrillation Investigators. The effect of low-dose warfarin on the risk of stroke in patients with non-rheumatic atria1 fibrillation. N Engl J Med 1990; 323: 1505-1511. 171 Stroke Prevention in Atria1 Fibrillation Investigators. Stroke Prevention in Atria1 Fibrillation Study: final results. Circulation 1991; 84: 527-539. PI Connolly SJ, Laupacis A, Gent M, Roberts RS, Cairns JA, Joyner C. Canadian Atrial Fibrillation Anticoagulation (CAFA) Study. J Am Co11 Cardiol 1991; 18: 349-355. [91 Ezekowitz MD, Bridgers SL, James KE, et al. Warfarin in the prevention of stroke associated with non-rheumatic atria1 fibrillation. N Engl J Med 1992; 327: 1406-1412. [lOI Stroke Prevention in Atrial Fibrillation Investigators. Warfarin versus aspirin for prevention of thromboembolism in atria1 fibrillation: Stroke Prevention in Atria1 Fibrillation II Study. Lancet 1994; 343: 687-691.
70
M. Yoshida et al. I International
(I I] Kuramoto K, Matsushita, S, Yamanouchi H. Atria1 fibrillation as a cause of myocardial and cerebral infarctions. Jpn Circ J 1984; 48: 67-74. 1121 Zipes DP. Atrial fibrillation. In: Braunwald E, ed., Heart Disease: A Textbook of Cardiovascular Medicine. 4th ed, Philadelphia: W.B. Saunders Company, 1992: 682-684. [13] Sherman DG, Dyken ML, Fisher M, Harrison MJG, Hart RG. Antithrombotic therapy for cerebrovascular disorders. Chest 1989; 95(Suppl.l): 14OS-15%. [ 141 Wolf PA, Kannel WB, McGee DL, Meeks SL, Bharucha NE, McNamara PM. Duration of atrial fibrillation and imminence of stroke: The Framingham study. Stroke 1983; 14: 664667. [ 151 Britton M, Gustafsson C. Non-rheumatic atria1 fibrillation as a risk factor for stroke. Stroke 1985; 16: 182-188. [16] Peterson P. Thromboembolic complications in atria1 fibrillation. Stroke 1990; 21: 4-13. [ 171 The Stroke Prevention in Atria1 Fibrillation Investigators. Predictors of thromboembolism in atria1 fibrillation: I. Clinical features of patients at risk. Ann Intern Med 1992; 116: l-5. [ 181 Naito S, Imataka K, Seko Y, Fujii J. The predictive factors for cerebral infarction in patients with non-rheumatic atria1 fibrillation. J Cardiol 1990; 20: 385-390 (in Japanese). [19] Brand FN, Abbott RD, Kennel WB, Wolf PA. Characteristics and prognosis of lone atrial fibrillation: The Framingham Study. J Am Med Assoc 1985; 254: 3449-3453. [20] Kopecky SL, Gersh BJ, McGoon MD, et al. The natural history of lone atria1 fibrillation. N Engl J Med 1987; 317: 669-674.
Journal of Cardiolog)
Xi (1996) 61-70
[21] Moss AJ. Atria1 fibrillation and cerebral embolism (Edilorial). Arch Neurol 1984; 41: 707. [22] Caplan LR, D’Cruz I, Hier DB, Reddy H, Shah S. Atrial size, atria1 fibrillation and stroke. Ann Neurol 1986: 19: 158-161. [23] Cabin HS, Clubb KS, Hall C, Perlmutter RA, Feinstein AR. Risk for systemic embolization of atrial fibrillation without mitral stenosis. Am J Cardiol 1990; 65: 1112-I 116. 1241 Tegeler CH, Hart RG, Sherman DG, Quiroga ES, Easton JD. Atrial fibrillation-stroke study (abstr). Stroke 1986; 17: 131. [25] Wiener I. Clinical and echocardiographic correlates of systemic embolization in non-rheumatic atria1 fibrillation. Am J Cardiol 1987; 59: 177. [26] The Stroke Prevention in Atria1 Fibrillation Investigators. Predictors of thromboembolism in atrial fibrillation: II. Echocardiographic features of patients at risk. Ann Intern Med 1992; 116: 6-12. [27] Peterson P, Kastrup J, Helweg-Larson S, Boysen G, Godtfredsen J. Risk factors for thromboembolic complications in chronic atria1 fibrillation: The Copenhagen AFASAK study. Arch Intern Med 1990; 150: 819-821. [28] Dunkman WB, Johnson GR, Carson PE, Bhat G, Farrell L, Cohn JN, for the V-Hem VA Co-operative Studies Group. Incidence of Thromboembolic Events in Congestive Heart Failure. Circulation 1993; 87(Suppl. VI): V194-101.