□ ORIGINAL ARTICLE □

Clinical Features of Ischemic Stroke during Treatment with

Dabigatran: An Association between Decreased Severity and
a Favorable Prognosis

Takeshi Hayashi, Yuji Kato, Takuya Fukuoka, Ichiro Deguchi, Hajime Maruyama,
Yohsuke Horiuchi, Hiroyasu Sano, Yuito Nagamine, Satoko Mizuno and Norio Tanahashi

Abstract
Objective Anticoagulation therapy with warfarin is associated with a favorable prognosis in ischemic
stroke. Dabigatran, a new oral anticoagulant, is widely used to prevent ischemic stroke in non-valvular atrial
fibrillation (NVAF) patients. However, its association with decreased severity and a favorable prognosis once
ischemic stroke has occurred remains unknown.
Methods We retrospectively reviewed all the patients with NVAF-associated ischemic stroke admitted to
our hospital from April 2011 to December 2014 and included those who received dabigatran therapy. We as-
sessed whether the patients were under regular use of the drug or discontinuance and classified them into 2
groups, the treatment and discontinuation groups. Clinical data, including the age, sex, ASCOD stroke pheno-
type, NVAF type, prescribed drug dose, comorbidities, CHADS2 score, renal function, National Institute of
Health Stroke Scale (NIHSS) score on admission, modified Rankin scale (mRS) score at discharge, D-dimer,
and brain natriuretic peptide, were investigated and compared between the groups.
Results Nine patients were under regular dabigatran therapy, and 6 were under discontinuance of the drug.
The age, sex, ASCOD stroke phenotype, NVAF type, comorbidities, renal function, and CHADS2 scores did
not differ between the 2 groups; however, the NIHSS scores were significantly lower in the treatment group.
The mRS scores at discharge were additionally decreased in the treatment group. Moreover, the D-dimer
scores were lower in the treatment group, thus suggesting a possible role in the decreased stroke severity.
Conclusion Dabigatran may therefore decrease the severity of ischemic stroke, even if ischemic stroke oc-
curs.

Key words: cardioembolic stroke, new oral anticoagulant, non-valvular atrial fibrillation

(Intern Med 54: 2433-2437, 2015)

(DOI: 10.2169/internalmedicine.54.4948)

embolic stroke in NVAF patients. Besides its preventive ef-

Introduction fect, this drug is beneficial for reducing disease severity
even when ischemic stroke occurs. Previous reports showed
Cardioembolic stroke is the most severe subtype of that anticoagulation therapy with warfarin was associated
ischemic stroke. Approximately half of all patients present- with less severe neurological deficits and a favorable prog-
ing with cardiogenic cerebral embolism die within 1 year of nosis (3-5).
its occurrence (1). Atrial fibrillation (AF), in particular non- Recently, new oral anticoagulants (NOACs) have become
valvular AF (NVAF), is the most common cause of cardio- available and are now widely used. Dabigatran is the first
embolic stroke (2). Therefore, the prevention of cerebral em- NOAC approved for the prevention of cardiogenic embolism
bolism in NVAF patients is of great importance. in the patients with NVAF (6). A RE-LY trial revealed that
For a long time, warfarin has been used to prevent cardio- dabigatran therapy was effective for preventing ischemic

Department of Neurology and Cerebrovascular Medicine, Saitama Medical University International Medical Center, Japan
Received for publication January 13, 2015; Accepted for publication February 15, 2015
Correspondence to Dr. Takeshi Hayashi, [email protected]

2433
 Intern Med 54: 2433-2437, 2015 DOI: 10.2169/internalmedicine.54.4948

stroke in the patients with NVAF compared with warfarin treatment with anti-diabetic medication (8). The initial mea-
therapy (7). Conversely, it remains unknown whether this surements of the D-dimer and BNP levels were used in the
drug has any beneficial effect in reducing the disease sever- present study. The D-dimer level was measured within 24
ity after the occurrence of ischemic stroke. hours and BNP was measured within 72 hours after admis-
We herein investigate the clinical and laboratory features sion in all the patients. Furthermore, because stroke etiology
of patients with ischemic stroke who were taking dabigatran. other than cardioembolism may be implicated even in the
Through this study, we aim to elucidate whether NOACs patients with NVAF, we investigated the ASCOD stroke
have a protective effect against ischemic stroke, even after phenotypes in all patients, because such phenotyping takes
its occurrence. all the possible stroke etiologies into consideration and is
highly recommended in stroke studies (9). The phenotypes
Materials and Methods were compared between the treatment group and discontinu-
ation group.
This study was approved by the Institutional Review Statistical analysis was performed using the PASW Statis-
Board at Saitama Medical University International Medical tics software program (version 18, SPSS, Chicago, USA).
Center. The age, D-dimer, and BNP differences between the groups
When a patient with stroke is admitted to our hospital, we were analyzed using the Wilcoxon test; differences in other
typically review the medication that the patient has been variables were assessed using the chi-square test. For each
prescribed. When patients or family members do not bring analysis, a p value of less than 0.05 was considered to be
their medication records, we contact the primary physician statistically significant.
and obtain the drug information of the patient (current and
past drug histories). Thus, in our hospital, the history of pre- Results
scriptions for all patients with stroke is available. We retro-
spectively reviewed all the patients with ischemic stroke as- There were 15 patients with NVAF-associated ischemic
sociated with NVAF admitted to our hospital from April stroke, to whom dabigatran had been prescribed. There were
2011 to December 2014. Among these, 15 patients were no patients with transient ischemic attack; all the patients
prescribed dabigatran, and they were thus included in this had complete strokes. Among them, 9 patients were taking
study. dabigatran regularly, indicating that they had been taking the
Through patient interviews or assessments of the drug re- drug within 12 hours. Six patients, on the other hand, had
cords, we confirmed whether a patient was under regular discontinued the drug; 2 of these had confidentially discon-
use of dabigatran or discontinuation. The reason of the dis- tinued the drug by themselves, the other 2 had suspended its
continuation varied. We divided the patients into 2 groups: a use because their stools appeared blackish, 1 had tentatively
treatment group, consisting of the patients with a regular stopped its use due to a scheduled colon polypectomy, and
drug intake with the last medication administered within 12 another had discontinued its use following the doctor’s rec-
hours, and a discontinuation group, consisting of the patients ommendation. The discontinuation period varied from 5
who were discontinuing the drug treatment. Discontinuation days to months (Table 1). Due to the short half-life of
was defined as those who suspended the drug for more than dabigatran, all the patients in the discontinuation group did
24 hours, however, all the patients in this group had stopped not have residual effects of this drug.
the drug for more than 5 days (discussed later). None of the The demographic data, ASCOD subtype, AF type, pre-
patients used warfarin or other NOACs simultaneously. The scribed drug dose, comorbidities, CHADS2 scores, cre-
age, sex, prescribed drug dose, discontinued period (only for atinine clearance, NIHSS scores on admission, mRS scores
the discontinuation group), NVAF type (i.e., paroxysmal or at discharge, and the laboratory data (D-dimer and BNP) of
persistent), comorbidities (e.g., heart failure, hypertension, all the patients are listed in Table 1. We suspected that pa-
diabetes mellitus, and past history of stroke), CHADS2 tients under regular dabigatran therapy may have had etiolo-
score, creatinine clearance (calculated by the Cockcroft- gies other than cardioembolism, and therefore performed
Gault formula), National Institute of Health Stroke Scale ASCOD subtyping in all patients. However, the implication
(NIHSS) score on admission, modified Rankin Scale (mRS) of atherosclerosis (A), small vessel disease (S), and other
score at discharge, serum D-dimer levels, and serum brain etiologies (O) did not differ among the two groups. With re-
natriuretic peptide (BNP) were retrospectively reviewed. The gard to dissection (D), 1 patient in the treatment group had
patients with clinical symptom of heart failure of New York a proven carotid artery dissection (case 3). The type of AF
Heart Association II or worse, and/or an ejection fraction (i.e., paroxysmal or persistent) was also similar between the
less than 50% by echocardiography, were defined as having two groups; 4/9 in the treatment group and 2/4 in the dis-
heart failure. The patients with a blood pressure of !140/90 continuance group had paroxysmal AF.
mmHg and/or those receiving antihypertensive medication As shown in Table 2, each data set was compared be-
were defined as having hypertension (8). Diabetes mellitus tween the two groups. There were no differences in the age
was defined as a blood glucose level of ! 200 mg/dL and a and sex between the groups. Comorbidities, such as heart
glycated hemoglobin level of !6.5% on admission and/or failure, hypertension, diabetes mellitus, and a past history of

2434
 Intern Med 54: 2433-2437, 2015 DOI: 10.2169/internalmedicine.54.4948

(pg/mL)
stroke (all of these are components of the CHADS2 score),

3,549.3
297.9
722.7
166.8

215.5
160.4
131.4
192.8

172.6

575.5
BNP

41.7
35.7
53.4

75.3
NE
as well as the renal function, did not significantly differ be-
D-dimer tween the two groups. Additionally, the CHADS2 scores did
Ǎg/mL)
not differ between the two groups. The NIHSS scores, how-

0.71

2.25

1.61
0.64

4.59

2.81
4.08
NE
NE

0.5
0.5

0.6
0.5
0.5
0.5
ever, showed a significant difference; the median score in

Creatinine clearance was calculated by Cockcroft-Gault formula. BNP: brain natriuretic peptide, mRS: modified Rankin Scale, NIHSS: National Institute of Health Stroke Scale
the treatment group was 1, whereas that in the discontinu-
(mL/min) admission discharge

ation group was 12.5 (p<0.05). The mRS scores at discharge

score at
mRS

2
1

1
3
0

5
0
1
1

4
3
2
2

4
was also significantly lower in the treatment group (median,
1) than in the discontinuation group (median, 4) (p=0.027).
The results of the NIHSS and mRS scores are expressed as
score on
NIHSS

15
10

19
16
box plots in Figure, respectively. With regard to the labora-

4
1

1
3
1

5
0
4
2
1
1

tory findings, the BNP concentration did not differ between

the two groups; however, the D-dimer level was markedly
Creatinine
history of score before clearance

100.3

lower in the treatment group, thus indicating that the

70.6
95.4

58.2

53.2
50.8

56.2
75.8
31.2

90.7
47.1
60.4
74.5

42.6
109

thrombi tended to be larger in the discontinuation group.

Discussion
admission
CHADS2

2
2
1
3
0
2
0
3
4

3
3
3
4
3
3

The patients with NVAF still have a risk of stroke, even if

they are under anticoagulation therapy (10). In a RE-LY
trial, the annual occurrence of ischemic stroke or systemic
stroke
Past

ï
+
ï
ï
+
ï
ï
ï
ï

+
ï
+
+
+
+

embolism was 1.71% in the warfarin group, 1.54% in the

dabigatran 110-mg b.i.d. group, and 1.11% in the dabigatran
Heart Hyper- Diabetes
failure tension mellitus

150-mg b.i.d. group (7). However, which patients are at the

ï

ï
ï
ï
ï
ï
+
ï
+
ï
ï
ï

ï
+

highest risk of having ischemic strokes while under antico-

agulation therapy remains unclear. For the patients under
warfarin therapy, insufficient anticoagulation therapy is asso-
+
ï
+
+

ï
ï
+
ï
+
+
+
ï

+
ï

ciated with a higher stroke occurrence rate (11); thus, the

status of anticoagulation should be monitored by measuring
ï
ï
ï
ï

ï
ï
ï
ï
ï
ï
ï
ï

ï
+

the prothrombin time-internationalized ratio (10, 12). Con-

versely, for those under dabigatran therapy, the relative dif-
1 month < Paroxysmal
Paroxysmal

Paroxysmal
Paroxysmal
Paroxysmal

1 month < Paroxysmal

fibrillation

Persistent
Persistent

Persistent
Persistent
Persistent
Persistent

Persistent
Persistent

Persistent
Type of
atrial

ferences in the risks have not yet been identified. From the
present study, we observed the discontinuation of dabigatran
may be one of the risks. Six of the 15 (40%) patients with
discontinua-

ischemic stroke under dabigatran therapy had discontinued

tion period

1 month <
1 month <

4 weeks
5 days

the drug. However, because we do not know the frequency

ï

ï
ï
ï
ï
ï
ï
ï
ï

of the tentative suspension of dabigatran in the patients

Table 1. Cases of Ischemic Stroke Prescribed Dabigatran.

without stroke, we were unable to evaluate the risk ade-

(mg/day)

quately. Nevertheless, we speculate that 40% of the cases of

drug
dose

220
220
220
220

300
220
220

220
220
220
220
220

300

220
220

drug discontinuation are far worse than general adherence.

As described above, 3/6 patients discontinued the drug with-
Sex ASCOD subtype

A0S0C1O0D0
A1S0C1O0D0
A3S0C1O0D0

A0S0C1O0D0
A1S0C1O0D0

A0S0C1O0D0
A1S0C1O0D0

A1S1C1O0D0
A0S0C1O0D0
A9S0C1O0D9
A0S0C1O0D1
A0S0C1O0D0

A0S0C1O0D0

A0S0C1O0D0
A0S0C1O0D0

out a specific reason, and 2 stopped because their stools ap-

peared blackish. However, when we investigated the human
hemoglobin levels in these patients’ stool samples, no gas-
trointestinal bleeding was observed. Therefore, NOACs
should be discontinued only after careful consideration. A
M
M
M
M

M
M
M
M

M
M
M
M
M

previous report revealed that the reluctance to reinitiate anti-

coagulation therapy after gastrointestinal bleedings was asso-
Drug intake Age

12 Discontinued 65
13 Discontinued 71
10 Discontinued 84
11 Discontinued 82
70

59
73
66
83

15 Discontinued 83
83
67
75
89

14 Discontinued 68

ciated with a poor outcome (13).

Even if a stroke has occurred, the regular use of dabiga-
Regular

Regular
Regular
Regular
Regular
Regular
Regular
Regular
Regular

tran made it less severe (Table 2, Figure). This is similar to

the previously confirmed findings of warfarin treat-
ment (3-5). Anticoagulation therapy, either with warfarin or
Case

dabigatran, may be useful not only for preventing ischemic

No.
1

6
7
8
9
2
3
4
5

stroke, but also for reducing the stroke severity when it oc-
curs. As shown in Table 1, the patients in the discontinu-

2435
 Intern Med 54: 2433-2437, 2015 DOI: 10.2169/internalmedicine.54.4948

Table 2. Comparison of Demographics, Risk Factors, Laboratory Data, CHADS2 Score, NIHSS

Score, and mRS in Stroke Patients with and without Regular Dabigatran Use.

Treatment group (n =9) Discontinuation group (n =6) p value

Age (mean ± SD) 73.9 ± 9.6 75.5 ± 8.5 0.739
Male sex 9 (100%) 4 (66.7%) 0.063
Heart failure 0 (0%) 2 (33.3%) 0.063
Hypertension 5 (55.5%) 4 (66.7%) 0.667
Diabetes mellitus 2 (22.2%) 2 (33.3%) 0.634
Past history of stroke 5 (55.5%) 2 (33.3%) 0.398
Creatinine clearance (mL/min) 74.39 ± 22.75 57.75 ± 21.01 0.173
D-dimer (Ǎg/mL) 0.56 ± 0.08 3.07 ± 1.25 0.010
BNP (pg/mL) 227.0 ± 206.1 762.6 ± 1,381.1 0.388
CHADS2 before admission (median,range) 3 (0–4) 2.5 (1–4) 0.563
NIHSS on admission (median, range) 1 (1–5) 12.5 (1–19) 0.019
mRS at discharge (median, range) 1 (0–3) 4 (1–5) 0.027
Creatinine clearance was calculated by Cockcroft-Gault formula. BNP: brain natriuretic peptide, mRS:
modified Rankin Scale, NIHSS: National Institute of Health Stroke Scale

Figure. Box plots of the National Institute of Health Stroke Scale (NIHSS) scores on admission (A)
and modified Rankin Scale (mRS) scores at discharge (B) in the stroke patients with NVAF. The
NIHSS scores were significantly lower in the treatment group than in the discontinuation group
(p<0.05). Additionally, the mRS scores were lower in the treatment group than in the discontinuation
group (p=0.027).

ation group had not taken dabigatran for 5 days or more. tran therapy (16). However, ASCOD subtyping revealed that
Due to the short half-life of this drug, we can presume that the potential involvement of etiologies other than cardioem-
the patients in the discontinuation group had no residual ef- bolism was similar in both groups, and thus the stroke
fects of this drug. severity-reducing effect of dabigatran was not a result of the
With regard to the stroke severity-reducing effect of war- differences in stroke etiology.
farin, several mechanisms have been postulated. Warfarin A previous study showed that intracranial hemorrhage in
may make the thrombi smaller or more fragile (14). The in- the patients under dabigatran therapy tended not to be se-
hibition of the thrombotic system may result in predomi- vere (17). In the present study, we demonstrated that
nance of the fibrinolytic system (15). In the present study, ischemic stroke in the dabigatran-treated patients may be
the treatment group showed a significantly lower level of se- less severe than in those without regular use of the drug.
rum D-dimer than the discontinuance group (Table 2). This There are several limitations associated with this study; the
suggests that dabigatran made the causative thrombi smaller, present study investigated a small number of patients and
thereby reducing the severity of the stroke. was retrospective in nature. Therefore, a prospective study
There are other possibilities which may have affected the with a larger number of patients would be required to con-
stroke severity and clinical outcome. Although they were not firm the present findings. Nonetheless, we speculate that the
statistically significant, heart failure and decreased creatinine beneficial effect of anticoagulation therapy with dabigatran
clearance tended to be more common in the discontinuation would be more significant than expected by the number of
group (Table 1, 2). We additionally hypothesized that the ischemic stroke occurrences according to the present find-
stroke etiology was entirely different in the treatment group. ings.
For instance, ischemic stroke may be caused by carotid
atherothrombosis in the patients with NVAF under dabiga- The authors state that they have no Conflict of Interest (COI).

2436
 Intern Med 54: 2433-2437, 2015 DOI: 10.2169/internalmedicine.54.4948

Acknowledgement ME, Hennerici MG. The ASCOD phenotyping of ischemic stroke

We are grateful to Dr. Masaki Takao for his valuable support (Updated ASCO phenotyping). Cerebrovasc Dis 36: 1-5, 2013.
with the manuscript preparation. 10. Albertsen IE, Rasmussen LH, Overvad TF, Graungaard T, Larsen
TB, Lip GY. Risk of stroke or systemic embolism in atrial fibrilla-
References tion patients treated with warfarin: a systematic review and meta-
analysis. Stroke 44: 1329-1336, 2013.
1. Kubo M, Kiyohara Y, Ninomiya T, et al. Decreasing incidence of 11. Hart RG, Pearce LA, Aguilar MI. Meta-analysis: antithrombotic
lacunar vs other types of cerebral infarction in a Japanese popula- therapy to prevent stroke in patients who have nonvalvular atrial
tion. Neurology 66: 1539-1544, 2006. fibrillation. Ann Intern Med 146: 857-867, 2007.
2. Tanizaki Y, Kiyohara Y, Kato I, et al. Incidence and risk factors 12. Gallagher AM, Setakis E, Plumb JM, Clemens A, van Staa TP.
for subtypes of cerebral infarction in a general population: the Hi- Risks of stroke and mortality associated with suboptimal antico-
sayama study. Stroke 31: 2616-2622, 2000. agulation in atrial fibrillation patients. Thromb Haemost 106: 968-
3. Hylek EM, Go AS, Chang Y, et al. Effect of intensity of oral anti- 977, 2011.
coagulation on stroke severity and mortality in atrial fibrillation. N 13. Witt DM, Delate T, Garcia DA, et al. Risk of thromboembolism,
Engl J Med 349: 1019-1026, 2003. recurrent hemorrhage, and death after warfarin therapy interrup-
4. Nakamura A, Ago T, Kamouchi M, et al. Intensity of anticoagula- tion for gastrointestinal tract bleeding. Arch Intern Med 172:
tion and clinical outcomes in acute cardioembolic stroke: the 1484-1491, 2012.
Fukuoka Stroke Registry. Stroke 44: 3239-3242, 2013. 14. van Ramshorst B, van Bemmelen PS, Hoeneveld H, Faber JA,
5. Kim BJ, Kim HJ, Do Y, et al. The impact of prior antithrombotic Eikelboom BC. Thrombus regression in deep venous thrombosis.
status on cerebral infarction in patients with atrial fibrillation. J Quantification of spontaneous thrombolysis with duplex scanning.
Stroke Cerebrovasc Dis 23: 2054-2059, 2014. Circulation 86: 414-419, 1992.
6. Hughes B. First oral warfarin alternative approved in the US. Nat 15. Yasaka M, Yamaguchi T, Miyashita T, Tsuchiya T. Regression of
Rev Drug Discov 9: 903-906, 2010. intracardiac thrombus after embolic stroke. Stroke 21: 1540-1544,
7. Connolly SJ, Ezekowitz MD, Yusuf S, et al. Dabigatran versus 1990.
warfarin in patients with atrial fibrillation. N Engl J Med 361: 16. Cha MJ, Kim YD, Nam HS, Kim J, Lee DH, Heo JH. Stroke
1139-1151, 2009. mechanism in patients with non-valvular atrial fibrillation accord-
8. Hayashi T, Seahara Y, Kato Y, et al. Clinical characteristics of car- ing to the CHADS2 and CHA2DS2-VASc scores. Eur J Neurol
dioembolic transient ischemic attack: comparison with noncar- 19: 473-479, 2012.
dioembolic transient ischemic attack. J Stroke Cerebrovasc Dis 23: 17. Komori M, Yasaka M, Kokuba K, et al. Intracranial hemorrhage
2169-2173, 2014. during dabigatran treatment. Circ J 78: 1335-1341, 2014.
9. Amarenco P, Bogousslavsky J, Caplan LR, Donnan GA, Wolf

Ⓒ 2015 The Japanese Society of Internal Medicine

http://www.naika.or.jp/imonline/index.html

2437