Research

Association between atherogenic dyslipidemia and recurrent stroke risk in

patients with different subtypes of ischemic stroke
Lu Zhao1†, Ruihao Wang1,2†, Bo Song1, Song Tan3, Yuan Gao1, Hui Fang1, Jie Lu4, and
Yuming Xu1*
Background The association between atherogenic dyslipi- Key words: atherogenic dyslipidemia, ischemic stroke, recurrent event
demia and stroke recurrence remains unclear, and may be
influenced by different subtypes of ischemic stroke.
Aims We aimed to investigate whether atherogenic dyslipi- Survivors of acute ischemic stroke (IS) are at high risk of stroke
demia contributed to stroke recurrence in ischemic stroke
recurrence (1,2). Unfortunately, despite best medical treatment,
patients and in those with certain subtypes of ischemic stroke.
Methods We conducted a prospective hospital-based study there is still a high residual risk of cardiovascular diseases (3,4).
enrolling patients with acute ischemic stroke. Atherogenic dys- Atherogenic dyslipidemia (AD), which is defined as low high-
lipidemia was defined as high-density lipoprotein cholesterol density lipoprotein cholesterol (HDL-C) and high triglycerides,
<40 mg/dl and triglycerides ≥200 mg/dl. Ischemic stroke sub- may be one explanation for the high residual risk of stroke (5).
types were classified according to the Trial of Org 10172 in
AD is common in patients with established cardiovascular
Acute Stroke Treatment criteria. The patients were followed
up at 3, 6, 12 and 24 months after stroke onset. The association disease, type 2 diabetes mellitus and metabolic syndrome (5).
between atherogenic dyslipidemia and stroke recurrence was Patients with AD presented a wide range of abnormalities, includ-
analyzed by using multivariable Cox regression model. ing increased plasma concentrations of remnants, denser LDL,
Results In the 510 ischemic stroke patients, 64 patients (12·5%) more atherogenic HDL particles, and insulin resistant status (6).
had atherogenic dyslipidemia, and 66 patients (12·9%) experi-
Thus, AD may characterize stroke patients with higher stroke
enced stroke recurrence events within 24 months. Kaplan–
Meier analysis revealed that stroke recurrence rate was recurrence risk.
significantly higher in patients with atherogenic dyslipidemia However, data on the association between AD and stroke recur-
than those without in all the stroke patients (20·3% vs. 11·9%; rence in patients with stroke or transient ischemic attack (TIA)
P = 0·048), and more evident in those of large-artery atheroscle- are limited (7,8). And IS is a heterogenous disease with variable
rosis subtype (31·0% vs. 14·1%; P = 0·014), but not in the other
pathogenesis (9). AD therefore may not contribute to subsequent
subtypes. Multivariable Cox regression analysis revealed that
atherogenic dyslipidemia was associated with higher stroke strokes in patients with different stroke subtypes equally. To our
recurrence risk among stroke patients of large-artery athero- knowledge, there is no data on the influence of AD on stroke
sclerosis subtype (hazard ratio, 2·79; 95% confidence interval, recurrence in patients with different subtypes of IS. Thus we
1·24–6·28), but not significant in all the stroke patients (hazard carried out this prospective hospital-based study to explore the
ratio, 1·69; 95% confidence interval, 0·85–3·37).
prevalence of AD and its contribution to stroke recurrence among
Conclusions Atherogenic dyslipidemia is associated with
higher risk of stroke recurrence in ischemic stroke patients. a cohort of patients with acute IS.
Such association might be more pronounced in large-artery
atherosclerosis subtype and needs further investigation to
establish such relationship.
Patients and methods

Correspondence: Yuming Xu*, Department of Neurology, the First Patients included in this study were from the database of the
Affiliated Hospital of Zhengzhou University, No. 1 East Jianshe Road, Henan Province Stroke Registry which prospectively collected
Zhengzhou, Henan 450052, China. data on consecutive stroke patients within 14 days after stroke
E-mail: [email protected]
1
Department of Neurology, The First Affiliated Hospital of Zhengzhou
onset in the Department of Neurology of the First Affiliated Hos-
University, Zhengzhou, Henan, China pital of Zhengzhou University (Henan, China) and followed up
2
Department of Neurology, University of Erlangen-Nuremberg, Erlangen, for long periods (10). This study was approved by the Ethics
Germany Committee of the First Affiliated Hospital of Zhengzhou Univer-
3
Department of Neurology, Sichuan Academy of Medical Sciences and sity, and all patients or their designated relatives signed written
Sichuan Provincial People’s Hospital, Chengdu, Sichuan, China
4
Department of Medical Statistics, College of Public Health, Zhengzhou
informed consent. The present study involved patients enrolled in
University, Zhengzhou, Henan, China the registry between January 2010 and July 2011.

Received: 1 September 2014; Accepted: 24 November 2014; Published Baseline data collection
online 28 Apirl 2015 Baseline data of the patients were recorded in paper-based case
†These authors contributed equally to this manuscript. report forms. Acute IS was diagnosed according to the World
Health Organization Criteria (11). The demographics included
Conflict of interest: None declared.
age, gender, body weight, height and body mass index (BMI).
Funding: This study was funded by the Youth Foundation of the First The stroke risk factors included history of stroke (defined as
Affiliated Hospital of Zhengzhou University and Science and Technology medical record confirmed history of ischemic, hemorrhagic
Agency of Henan Province.
stroke or subarachnoid hemorrhage), hypertension (defined as
DOI: 10.1111/ijs.12471 history of hypertension and newly diagnosed hypertension),

752 Vol 10, July 2015, 752–758 © 2015 World Stroke Organization
 L. Zhao et al. Research
diabetes mellitus (including history of diabetes mellitus and adjusted hazard ratios (HRs) and 95% confidence intervals (95%
newly diagnosed diabetes), history of coronary heart disease CIs) were calculated. Statistical tests were conducted, with a
(including myocardial infarction and angina pectoris), atrial 2-tailed α-level of 0·05 considered significant. Data were analyzed
fibrillation (confirmed by at least one electrocardiogram or using SPSS version 17·0 (SPSS Inc., Chicago, IL).
present during hospitalization). The personal history mainly con-
sisted of current or previous smoking, and moderate or heavy Results
drinking (defined as ≥ two standard alcohol consumption per
day). The severity of neurologic impairment was evaluated A total of 566 patients with acute IS were consecutively registered.
according to the National Institutes of Health Stroke Scale 551 patients had complete baseline information. Among them, 41
(NIHSS) within 24 hours after admission by trained research patients (7·4%) were lost to follow-up, leaving 510 patients for the
investigators through direct interview with the patient at bedside final analysis. There was no significant difference of baseline clini-
(12). Stroke subtypes were classified according to the Trial of ORG cal and laboratory characteristics between the patients with com-
10172 in Acute Stroke Treatment (TOAST) criteria by two trained plete follow-up data (n = 510) and those lost to follow-up
study neurologists. IS subtypes included large-artery atheroscle- (n = 41) (Online-only Data Supplemental Table S1). In the 510
rosis (LAA), cardioembolism (CE), small-artery occlusion (SAO), patients with follow-up data, the average age was 60·69 ± 13·90
stroke of other determined causes (OC) and stroke of undeter- years, and 319 cases (62·5%) were women. According to the
mined causes (UC) (13). We combined the stroke of OC and UC TOAST classification, there were 234 cases (45·9%) with LAA, 192
types as ‘other subtypes’ group. Blood samples were collected in cases (37·6%) with SAO, 23 cases (4·5%) with CE, 16 cases (3·1%)
fasting conditions for evaluation of the lipid profiles, fasting with other determined causes, and 45 cases (8·8%) of undeter-
plasma glucose (FPG) and other routine tests, the lipid profiles mined causes.
included total cholesterol (TC), triglycerides (TG), high-density AD was observed in 64 patients (12·5%). Table 1 showed char-
lipoprotein cholesterol (HDL-C) and low-density lipoprotein acteristics in patients with and without AD. Compared with
cholesterol (LDL-C). Atherogenic dyslipidemia was defined as low patients without AD, patients with AD had significantly higher
HDL-C and high triglycerides in consistent with previous studies BMI (25·97 ± 3·35 vs. 24·43 ± 4·55; P = 0·012), TC (5·07 ± 1·21 vs.
(6,7,14,15). The cutoff values were selected at HDL-C < 40 mg/dl 4·64 ± 1·18; P = 0·006) and FPG (7·64 ± 3·13 vs. 6·13 ± 2·81;
(1·04 mmol/l) and triglyceride ≥200 mg/dl(2·26 mmol/l) in both P < 0·001), and were more likely to have diabetes mellitus (40·6%
men and women (16). vs. 22·2%, P = 0·001). The remaining baseline data did not differ
between the two groups. The prevalence of AD did not differ
Follow-up of stroke recurrence
between different TOAST subtypes of stroke (P = 0·754; Table 2).
At 3, 6, 12 and 24 months after stroke onset, all patients or their
During the 2-year-follow-up period, stroke recurrence events
relatives were followed through telephone and asked whether the
happened in 66 patients (12·9%). Kaplan–Meier analysis revealed
patients had new symptoms and experienced re-hospitalization
that the stroke recurrence rate in patients with AD was signifi-
with a diagnosis of ischemic or hemorrhagic stroke. A recurrent
cantly higher in all the stroke patients (20·3% vs. 11·9%;
stroke was defined as a newly diagnosed stroke presenting as a
P = 0·048) and stroke patients of LAA subtype (31·0% vs. 14·1%;
new neurological deficit or a deterioration of the previous deficit
P = 0·014) (Table 2; Fig. 1), insignificantly lower in the SAO
(11), not considered to be because of edema, hemorrhagic trans-
subtype (4·3% vs. 10·1%; P = 0·393), and insignificantly higher in
formation, or concurrent illness. Hospitals that admitted patients
CE subtype (50·0% vs. 14·3%; P = 0·268) or other subtypes (con-
with recurrent strokes were contacted to verify the diagnosis.
sisting of other determined causes and undetermined causes)
Follow-up was performed by trained and qualified telephone
(20·0% vs. 7·8%; P = 0·198) (Table 2). Unadjusted Cox propor-
interviews who were not involved in the registry and blinded to
tional regression analysis showed that AD was borderlinely asso-
the baseline data.
ciated with higher stroke recurrence risk in all the stroke patients
Statistical analysis (HR, 1·82; 95% CI, 0·99–3·35; P = 0·052) and was significantly
For descriptive analysis, data were expressed as mean (±standard associated with an increased stroke recurrence risk among stroke
deviation) for normally distributed continuous variables and patients of LAA subtype (HR, 2·47; 95% CI, 1·17–5·23; P = 0·018)
median (interquartile range) for non-normally distributed con- (Table 3). Online-only Data Supplemental Table S2 displayed
tinuous variables, and counts (percentages) for categorical vari- potential variables associated with stroke recurrence. After adjust-
ables. Bivariate comparisons were conducted using the χ2test for ing for age, hypertension, diabetes mellitus, NIHSS score at
categorical variables, and Student t test or Mann-Whitney test for admission, BMI, and total cholesterol, the association was attenu-
normally distributed or non-normally distributed continuous ated in all the stroke patients (HR, 1·69; 95% CI, 0·85–3·37;
variables. Kaplan–Meier analysis was used to compare the stroke P = 0·135) but still significant in stroke patients of LAA subtype
recurrence rate between patients with and without atherogenic (HR, 2·79; 95% CI, 1·24–6·28; P = 0·013) (Table 3; Fig. 2).
dyslipidemia. The association between atherogenic dyslipidemia
and stroke recurrence was analyzed by multivariable Cox regres- Discussion
sion models by adjusting for confounders which were associated
with AD in the bivariate comparisions and/or stroke recurrence in In this prospective hospital-based study, we found that the preva-
univariable Cox regression models (P < 0·1). Unadjusted and lence of AD among patients with acute IS was 12·5%, patients

© 2015 World Stroke Organization Vol 10, July 2015, 752–758 753
 Research L. Zhao et al.

Table 1 Patient characteristics in ischemic stroke patients with and without atherogenic dyslipidemia

Atherogenic dyslipidemia*

Variable Yes (N = 64) No (N = 446) P value

Demographics
Age, y 58·77 ± 13·71 60·97 ± 13·92 0·236
Women 39 (60·9%) 280 (62·8%) 0·776
NIHSS Score at Admission 4 (1–7) 4 (2–8) 0·459
BMI, kg/m2 25·97 ± 3·35 24·43 ± 4·55 0·012
Risk Factors
Hypertension 43 (67·2%) 257 (57·6%) 0·146
DM 26 (40·6%) 99 (22·2%) 0·001
Coronary Heart Disease 5 (7·8%) 57 (12·8%) 0·255
Atrial Fibrillation 3 (4·7%) 17 (3·8%) 0·716
Previous Stroke History 28 (43·8%) 116 (26·0%) 0·533
Current or Previous Smoking 23 (35·9%) 163 (36·5%) 0·925
Moderate or Heavy Drinking 16 (25·0%) 115 (25·8%) 0·893
Blood Biochemical Tests
Hemoglobin, g/l 139·76 ± 24·33 134·73 ± 18·66 0·120
Fibrinogen, g/l 3·06 ± 0·88 3·34 ± 1·89 0·257
Homocysteine, mmol/l 21·10 ± 11·51 20·50 ± 10·96 0·718
TC, mmol/l 5·07 ± 1·21 4·64 ± 1·18 0·006
LDL-C, mmol/l 3·00 ± 0·87 3·03 ± 0·96 0·814
FPG, mmol/l 7·64 ± 3·13 6·13 ± 2·81 <0·001
TOAST Classification 0·754
LAA Subtype 29 (45·3%) 205 (46·0%)
SAO Subtype 23 (35·9%) 169 (37·9%)
CE Subtype 2 (3·1%) 21 (4·7%)
Other Subtypes† 10 (15·6%) 51 (11·4%)
Secondary Prevention Medications Application
Antithrombotic Drugs 46 (71·9%) 309 (69·3%) 0·590
Lipid-lowering Drugs 20 (31·3%) 164 (36·8%) 0·394
Antihypertensive Drugs 35 (54·7%) 211 (47·3%) 0·241
Hypoglycemic Drugs 18 (28·1%) 87 (19·5%) 0·104

*Defines as HDL-C < 40 mg/dl (1·04 mmol/l) and triglyceride ≥200 mg/dl (2·26 mmol/l).
†
Other Subtypes consisted of stroke subtypes other determined causes and undetermined causes.
NIHSS indicates National Institutes of Health Stroke Scale; BMI, Body Mass Index; DM, Diabetes mellitus; TC, total cholesterol; LDL-C, low-density
lipoprotein cholesterol; FPG, fasting plasma glucose; TOAST, Trial of ORG 10172 in Acute Stroke Treatment; LAA, large-artery atherosclerosis; SAO,
small-artery occlusion; CE indicated cardiac embolism.

Table 2 Prevalence of atherogenic dyslipidemia and its association with stroke recurrence among different subtypes of stroke

Stroke recurrence rate

Stroke subtypes AD prevalence* All AD (+) AD (−) P value

All Subtypes (n = 510) 64 (12·5%) 66 (12·9%) 13 (20·3%) 53 (11·9%) 0·048

LAA Subtype (n = 234) 29 (12·4%) 38 (16·2%) 9 (31·0%) 29 (14·1%) 0·014
SAO Subtype (n = 192) 23 (12·0%) 18 (9·4%) 1 (4·3%) 17 (10·1%) 0·393
CE Subtype (n = 23) 2 (8·7%) 4 (17·4%) 1 (50·0%) 3 (14·3%) 0·268
Other Subtypes (n = 61)† 10 (16·4%) 6 (9·8%) 2 (20%) 4 (7·8%) 0·198

AD defines as HDL-C < 40 mg/dl (1·04 mmol/l) and triglyceride ≥200 mg/dl (2·26 mmol/l).
*No statistically significant difference according to Chi-Square test, P = 0·754.
†
Other Subtypes consisted of stroke subtypes other determined causes and undetermined causes.
AD indicates atherogenic dyslipidemia; LAA, large-artery atherosclerosis; SAO, small-artery occlusion; CE, cardiac embolism.

with AD had a significantly higher stroke recurrence risk than Our findings in the prevalence and clinical characteristics of
those without, and the association between AD and increased AD were consistent with the previous studies (7,8). The preva-
stroke recurrence risk appeared more pronounced among stroke lence of AD in SOS-TIA study was 11·1% among TIA patients
patients of LAA subtype. with a cerebral ischemic lesion (7), and a recently published post

754 Vol 10, July 2015, 752–758 © 2015 World Stroke Organization
 L. Zhao et al. Research

Fig. 1 Kaplan–Meier curves of stroke recurrence rate in patients with ischemic stroke of all subtypes (a) and those of large-artery atherosclerosis subtype
(b) with and without AD.

Table 3 Cox proportional regression analysis of atherogenic dyslipidemia and stroke recurrence

All subtypes of stroke LAA subtype Non-LAA subtypes*

Unadjusted Adjusted Unadjusted Adjusted Unadjusted Adjusted

AD
HR 1·82 1·69 2·47 2·79 1·18 0·49
95% CI 0·99–3·35 0·85–3·37 1·17–5·23 1·24–6·28 0·41–3·39 0·11–2·19
P Value 0·052 0·135 0·018 0·013 0·762 0·353
Age
HR 1·02 1·02 1·02 1·02 1·04 1·04
95% CI 1·00–1·04 1·00–1·04 1·00–1·04 0·99–1·04 1·01–1·07 1·00–1·07
P Value 0·016 0·049 0·202 0·298 0·023 0·037
Hypertension
HR 2·11 2·18 2·38 2·59 1·81 1·64
95% CI 1·21–3·66 1·21–3·94 1·13–5·03 1·14–5·88 0·80–4·12 0·68–3·94
P Value 0·008 0·010 0·023 0·023 0·155 0·272
Diabetes Mellitus
HR 1·83 1·54 1·63 1·37 2·24 2·60
95% CI 1·11–3·02 0·89–2·66 0·82–3·23 0·67–2·82 1·06–4·73 1·10–6·14
P Value 0·018 0·124 0·162 0·391 0·035 0·029
NIHSS score
HR 1·04 1·04 1·00 1·00 1·07 1·08
95% CI 1·00–1·08 1·00–1·08 0·94–1·06 0·93–1·06 1·02–1·13 1·02–1·13
P Value 0·065 0·062 0·864 0·887 0·005 0·005
BMI
HR 0·97 0·96 1·00 0·97 0·94 0·89
95% CI 0·93–1·02 0·91–1·00 0·94–1·1 0·91–1·03 0·87–1·03 0·79–1·01
P Value 0·262 0·066 0·899 0·309 0·168 0·069
Total Cholesterol
HR 1·04 0·96 1·08 0·89 1·02 1·16
95% CI 0·86–1·26 0·76–1·20 0·83–1·39 0·67–1·18 0·76–1·38 0·77–1·76
P Value 0·674 0·709 0·587 0·414 0·888 0·474

*Non-LAA subtypes included stroke subtypes of small-artery occlusion, cardiac embolism, other determined caused and undetermined causes.
LAA indicates large-artery atherosclerosis; HR, hazard ratio; CI, confidence ratio; NIHSS, National Institutes of Health Stroke Scale; BMI, Body Mass
Index.

© 2015 World Stroke Organization Vol 10, July 2015, 752–758 755
 Research L. Zhao et al.

Fig. 2 Adjusted survival curves derived from Cox proportional model of stroke recurrence rate in patients with ischemic stroke of all subtypes (a) and those
of large-artery atherosclerosis subtype (b) with and without AD. Age, hypertension, diabetes mellitus, NIHSS score at admission, BMI, and total cholesterol
were adjusted.

hoc study of two large clinical trials found that approximately 1 in in the study of PERFORM trial, the association was attenuated
10 patients with TIA or stroke had AD (8). We found that stroke after multivariable adjustment (HR 1·23; 95% CI, 1·03–1·48).
patients with AD were more likely to have diabetes, higher BMI However, the analyses were not performed in the different stroke
and FPG value, and this might be explained by insulin resistance, subtypes (8). We hypothesize that AD may not equally contribute
which was considered to play an important role in the pathogenic to subsequent stroke events among stroke patients of LAA and
basis of atherogenic dyslipidemia (15). SAO subtypes equally. IS is a heterogeneous disease, the etiology
Our data showed that stroke patients with AD had significantly of different TOAST subtypes varies (19), and the pathogenesis of
increased stroke recurrence risk than those without AD. Some stroke recurrence also differs (20). Kumral et al found a significant
published studies have addressed the prognostic value of the indi- association between baseline dyslipidemia and stroke recurrence
vidual components of AD in patients with cerebrovascular dis- in the large-artery disease subtype, but not in the other stroke
eases (17,18). A post hoc analysis of the Stroke Prevention by subtypes (21). Mounting evidence support the strong association
Aggressive Reduction in Cholesterol Levels (SPARCL) trail found between lipid disorder and large artery atherosclerosis stroke (22–
that in patients with recent stroke or TIA, lower baseline HDL-C 24). Warfarin-Asprin Symptomatic Intracranial Disease (WASID)
predicted the risk of recurrent stroke, and high triglyceride level study revealed that lipid disorder has the most strongest associa-
was associated with major cardiovascular event (18). Fukuoka tion of the severity of the severity of intracranial large artery
Stroke Registry study revealed low HDL-C level was an indepen- stenosis (23). Bang OY et al found that elevated levels of serum
dent risk factor for a stroke recurrence within one year after onset triglycerides and non-high-density lipoprotein, but not low-
(17). To our knowledge, only two researches addressed the impact density lipoprotein (LDL), were associated with large atheroscle-
of AD on the stroke recurrence in patients with stroke or TIA. The rotic stroke (22). Chinese Intracranial Atherosclerosis(CICAS)
deleterious role of AD in cerebrovascular diseases was first studied study revealed that low HDL-C level is associated with the devel-
in the SOS-TIA study, which found that AD might be associated opment of intracranial artery stenosis in Chinese patients with
with intracranial artery stenosis and higher risk of early recurrent acute IS (24). In contrast, the pathogenesis of small-artery occlu-
stroke among patients with TIA (7). A recently published post hoc sion stroke has not been fully clarified, but is considered to be due
study of SPARCL and PERFORM (Prevention of Cerebrovascular to nonatherosclerotic arteriopathy, which is mainly caused by
and Cardiovascular Events of Ischemic Origin With Terutroban in hypertension and diabetes (25). Atherosclerosis Risk in Commu-
Patients With a History of Ischemic Stroke or Transient Ischemic nities (ARIC) study found that lacunar infarcts caused by the
Attack) trail found that among subjects with stroke or TIA while occlusion of small arteries were associated with diabetes or LDL
receiving statin and otherwise best medical therapy, those with cholesterol (26). These findings support that AD may cardinally
AD had a higher residual cardiovascular risk than those without contribute to stroke recurrence in stroke patients of LAA, which
AD (8). could explain our results. Our study indicates stroke patients of
The most important finding of this study was that AD was LAA subtype may benefit from the combination of fibrate and
independently associated with increased recurrent stroke risk statin, which was not yet supported by the current evidence from
among stroke patients of LAA subtype. The post hoc study of the Action to Control Cardiovascular Risk in Diabetes
SPARCL trail showed the association between AD and major car- (ACCORD) study conducted in the stroke patients which were
diovascular events in patients with IS or TIA became insignificant not classified by stroke etiology, however the subgroup analysis of
after multivariable adjustment (HR 1·24; 95% CI, 0·93–1·65), and the ACCORD study showed individuals with AD seemed to

756 Vol 10, July 2015, 752–758 © 2015 World Stroke Organization
 L. Zhao et al. Research
benefit from the combination (27). A further meta-analysis also 5 Fruchart JC, Sacks F, Hermans MP et al. The Residual Risk Reduction
supported that fibrate treatment directed at markers of AD sub- Initiative: a call to action to reduce residual vascular risk in patients
with dyslipidemia. Am J Cardiol 2008; 102:1K–34K.
stantially reduced subsequent vascular event risk (14).
6 Arca M, Montali A, Valiante S et al. Usefulness of atherogenic dyslipi-
Our study had limitations. First, it was a hospital-based study demia for predicting cardiovascular risk in patients with angiographi-
conducted only in one hospital, the sample size was small, and cally defined coronary artery disease. Am J Cardiol 2007; 100:1511–6.
hospital selective bias was inevitable (28). As the sample sizes in 7 Sirimarco G, Deplanque D, Lavallee PC et al. Atherogenic dyslipi-
CE subtype and other subtypes of IS were small, type II error demia in patients with transient ischemic attack. Stroke 2011;
42:2131–7.
(false negative association between AD and stroke recurrence)
8 Sirimarco G, Labreuche J, Bruckert E et al. Atherogenic dyslipidemia
might happen due to a lack of statistical power, we still could not and residual cardiovascular risk in statin-treated patients. Stroke 2014;
determine the relationship between AD and stroke recurrence in 45:1429–36.
CE subtype and other subtypes or conclude whether AD contrib- 9 Saeed M. Editorial comment – unraveling the pagodian architecture of
uted to the recurrence of different stroke subtypes. Second, as stroke as a complex disorder. Stroke 2004; 35:824–5.
10 Song B, Fang H, Zhao L et al. Validation of the ABCD3-I score to
Peduzzi et al suggested that the number of outcome events per
predict stroke risk after transient ischemic attack. Stroke 2013;
variable (EPV) = 10 to be most prudent in a proportional hazards 44:1244–8.
regression analysis (29), the value of EPV in our multivariable 11 Stroke – 1989. Recommendations on stroke prevention, diagnosis, and
Cox proportional regression analysis seemed not high enough to therapy. Report of the WHO Task Force on Stroke and other Cerebro-
exclude overfitting (overestimation of the true effect), thus a con- vascular Disorders. Stroke 1989; 20:1407–31.
12 Brott T, Adams HP Jr, Olinger CP et al. Measurements of acute
founding bias related to covariables might still exist despite
cerebral infarction: a clinical examination scale. Stroke 1989; 20:864–
adjustment. Third, lipid profiles during follow-up period which 70.
may influence prognosis were not examined. Fourth, the 13 Adams HP Jr, Bendixen BH, Kappelle LJ et al. Classification of subtype
follow-up was obtained via telephone interview, which could of acute ischemic stroke. Definitions for use in a multicenter clinical
bring inaccuracy comparing to a face-to-face clinic visit. Never- trial. TOAST. Trial of Org 10172 in Acute Stroke Treatment. Stroke
1993; 24:35–41.
theless, we tried to minimize such bias with telephone interviews
14 Lee M, Saver JL, Towfighi A, Chow J, Ovbiagele B. Efficacy of fibrates
conducted by trained and qualified telephone interviews, and for cardiovascular risk reduction in persons with atherogenic dyslipi-
hospitals that admitted patients with recurrent strokes were con- demia: a meta-analysis. Atherosclerosis 2011; 217:492–8.
tacted to verify the diagnosis. Finally, although we classified the 15 Cabre A, Babio N, Lazaro I et al. FABP4 predicts atherogenic dyslipi-
subtypes of initial IS, we could not classify the subtypes of recur- demia development. The PREDIMED study. Atherosclerosis 2012;
222:229–34.
rent stroke, as the follow-up information was obtained via tele-
16 National Cholesterol Education Program (NCEP) Expert Panel on
phone interview and was not available in the registry. Detection, Evaluation, and Treatment of High Blood Cholesterol in
In conclusion, our study supports that atherogenic dyslipi- Adults (Adult Treatment Panel III). Third Report of the National
demia is associated with higher risk of stroke recurrence in Cholesterol Education Program (NCEP) Expert Panel on Detection,
patients with ischemic stroke. Such association might be more Evaluation, and Treatment of High Blood Cholesterol in Adults (Adult
Treatment Panel III) final report. Circulation 2002; 106:3143–421.
pronounced in the LAA subtype, but requires further investiga-
17 Kuwashiro T, Sugimori H, Ago T, Kamouchi M, Kitazono T. Risk
tion to establish this relationship. Nevertheless, treatment mea- factors predisposing to stroke recurrence within one year of non-
sures to stroke patients with AD, especially those of LAA subtype, cardioembolic stroke onset: the Fukuoka Stroke Registry. Cerebrovasc
might help reduce stroke recurrence rate. Dis 2012; 33:141–9.
18 Amarenco P, Goldstein LB, Callahan A 3rd et al. Baseline blood pres-
sure, low- and high-density lipoproteins, and triglycerides and the risk
Acknowledgements of vascular events in the Stroke Prevention by Aggressive Reduction in
Cholesterol Levels (SPARCL) trial. Atherosclerosis 2009; 204:515–20.
The authors would like to thank all the patients and their relatives 19 Hillen T, Coshall C, Tilling K, Rudd AG, McGovern R, Wolfe CD.
Cause of stroke recurrence is multifactorial: patterns, risk factors, and
who participated in this study. We also thank Dr. Juan Gui (from
outcomes of stroke recurrence in the South London Stroke Register.
Tongji Hospital, Huazhong University of Science and Technology, Stroke 2003; 34:1457–63.
China) and Dr. Xinwu Cui (from Caritas hospital, Bad Mergen- 20 Shin DH, Lee PH, Bang OY. Mechanisms of recurrence in subtypes of
theim, Germany) for their assistance in writing the manuscript. ischemic stroke: a hospital-based follow-up study. Arch Neurol 2005;
62:1232–7.
21 Kumral E, Evyapan D, Gokcay F, Karaman B, Orman M. Association of
References baseline dyslipidemia with stroke recurrence within five-years after
1 Donnan GA, Fisher M, Macleod M, Davis SM. Stroke. Lancet 2008; ischemic stroke. Int J Stroke 2014; 9(Suppl. A100):119–26.
371:1612–23. 22 Bang OY, Saver JL, Liebeskind DS, Pineda S, Ovbiagele B. Association
2 Davis SM, Donnan GA. Clinical practice. Secondary prevention after of serum lipid indices with large artery atherosclerotic stroke. Neurol-
ischemic stroke or transient ischemic attack. N Engl J Med 2012; ogy 2008; 70:841–7.
366:1914–22. 23 Turan TN, Makki AA, Tsappidi S et al. Risk factors associated with
3 Hackam DG, Spence JD. Combining multiple approaches for the sec- severity and location of intracranial arterial stenosis. Stroke 2010;
ondary prevention of vascular events after stroke: a quantitative mod- 41:1636–40.
eling study. Stroke 2007; 38:1881–5. 24 Qian Y, Pu Y, Liu L et al. Low HDL-C level is associated with the
4 Amarenco P, Labreuche J. Lipid management in the prevention of development of intracranial artery stenosis: analysis from the Chinese
stroke: review and updated meta-analysis of statins for stroke preven- IntraCranial AtheroSclerosis (CICAS) study. PLoS ONE 2013;
tion. Lancet Neurol 2009; 8:453–63. 8:e64395.

© 2015 World Stroke Organization Vol 10, July 2015, 752–758 757
 Research L. Zhao et al.

25 Jackson CA, Hutchison A, Dennis MS et al. Differing risk factor pro- Supporting information
files of ischemic stroke subtypes: evidence for a distinct lacunar arte-
riopathy? Stroke 2010; 41:624–9. Additional Supporting Information may be found in the online
26 Bezerra DC, Sharrett AR, Matsushita K et al. Risk factors for lacune version of this article at the publisher’s web-site:
subtypes in the Atherosclerosis Risk in Communities (ARIC) Study.
Neurology 2012; 78:102–8.
Table S1. Comparisons of baseline characteristics of ischemic
27 Ginsberg HN, Elam MB, Lovato LC et al. Effects of combination lipid
therapy in type 2 diabetes mellitus. N Engl J Med 2010; 362:1563–74. stroke patients with and lost to follow-up.
28 Schulz UG, Rothwell PM. Differences in vascular risk factors between
etiological subtypes of ischemic stroke: importance of population- Table S2. Univariable Cox analysis for characteristics and recur-
based studies. Stroke 2003; 34:2050–9. rent stroke in 510 patients with ischemic stroke.
29 Peduzzi P, Concato J, Feinstein AR, Holford TR. Importance of events
per independent variable in proportional hazards regression analysis.
II. Accuracy and precision of regression estimates. J Clin Epidemiol
1995; 48:1503–10.

758 Vol 10, July 2015, 752–758 © 2015 World Stroke Organization