Accepted Manuscript

Serum uric acid concentrations and risk of intracerebral hemorrhage: A systematic

review and meta-analysis

Zhike Zhou, Yifan Liang, Jueying Lin, Xiaoqian Zhang, Huiling Qu, Junjie Xu,
Chuansheng Zhao, Mei Zhao

PII: S0021-9150(18)31198-5
DOI: 10.1016/j.atherosclerosis.2018.07.002
Reference: ATH 15609

To appear in: Atherosclerosis

Received Date: 10 March 2018

Revised Date: 1 June 2018
Accepted Date: 3 July 2018

Please cite this article as: Zhou Z, Liang Y, Lin J, Zhang X, Qu H, Xu J, Zhao C, Zhao M, Serum uric
acid concentrations and risk of intracerebral hemorrhage: A systematic review and meta-analysis,
Atherosclerosis (2018), doi: 10.1016/j.atherosclerosis.2018.07.002.

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to
our customers we are providing this early version of the manuscript. The manuscript will undergo
copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please
note that during the production process errors may be discovered which could affect the content, and all
legal disclaimers that apply to the journal pertain.
 ACCEPTED MANUSCRIPT

PT
RI
U SC
AN
M
D
TE
EP
C
AC
 ACCEPTED MANUSCRIPT

Serum uric acid concentrations and risk of intracerebral hemorrhage: A systematic review and

meta-analysis

Zhike Zhou1, Yifan Liang2, Jueying Lin3, Xiaoqian Zhang2, Huiling Qu4, Junjie Xu5, Chuansheng Zhao 2* & Mei

Zhao 6*

PT
1
Department of Geriatrics, The First Affiliated Hospital, China Medical University, Shenyang 110001, Liaoning,

RI
2
PR China; Department of Neurology, The First Affiliated Hospital, China Medical University, Shenyang

SC
3
110001, Liaoning, PR China; Department of Emergency, Zhongshan Hospital Xiamen University, Xiamen

4
361004, Fujian, PR China; Department of Neurology, People's Hospital of Liaoning Province, Shenyang

110016, Liaoning, PR China; 5

U
Department of Laboratory Medicine, The First Affiliated Hospital, China
AN
6
Medical University, Shenyang 110001, Liaoning, PR China; Department of Cardiology, The Shengjing
M

Affiliated Hospital, China Medical University, Shenyang 110004, Liaoning, PR China

D

*Corresponding author: Mei Zhao and Chuansheng Zhao

TE

E-mail addresses: zhaom1@sj-hospital.org or cszhao@mail.cmu.edu.cn

C EP
AC
 ACCEPTED MANUSCRIPT

1 Abstract

2 Background and aims: The relationship between serum uric acid (UA) and the risk of intracerebral hemorrhage

3 (ICH) remains controversial. The aim of our systematic review and meta-analysis was to ascertain the

PT
4 association between serum UA concentrations and the risk of ICH.

5 Methods: We systematically searched databases of Embase, Pubmed, Web of Science and Cochrane Library up

RI
6 to December 30, 2017, and additional papers were identified through a manual search. Mean difference (MD)

SC
7 for serum UA levels with 95% confidence intervals (CI) was calculated. Six studies, including 345 ICH patients,

8 574 ischemic stroke patients and 535 healthy controls, were identified for meta-analysis.

U
AN
9 Results: Our results revealed no statistically significant differences in the comparison of UA between ICH and

10 healthy controls (95% CI=-9.04-15.61); UA levels in patients with ischemic stroke were significantly higher
M

11 than those in healthy controls (95% CI=3.91-56.32); further subgroup analysis of age showed higher UA levels
D

12 in ICH patients over 65 years than healthy controls (age≥65: 95% CI=1.44-35.96). Subgroup of ethnicity
TE

13 (Asians: CI=-9.06-21.00; Caucasians: 95% CI=-68.43-8.43), gender (Men: 95% CI=-56.08-4.73; Women: 95%

14 CI=-27.19-35.91) and sample size (large samples: 95% CI=-20.54-41.05; small samples: 95% CI=-25.41-13.78)
EP

15 with respect to UA levels between ICH and healthy controls did not change these results.
C

16 Conclusions: This meta-analysis showed that serum UA levels did not increase the risk of ICH probably because
AC

17 of the dual roles of UA, i.e. pro-oxidant and antioxidant, in the progression of atherosclerosis. However, serum

18 UA may be a potential risk factor for ICH in the elderly. There were no race-specific differences in UA levels

19 between Asians and Caucasians as well as gender-related differences between men and women of the risk of

20 ICH.

21 Keywords Serum uric acid; Intracerebral hemorrhage; Meta-analysis

1
 ACCEPTED MANUSCRIPT

1 Introduction

2 Intracerebral hemorrhage (ICH) is a life-threatening type of stroke caused by the rupture of blood vessels into

3 the brain parenchyma; it accounts for 10-15% of all strokes [1]. ICH is more serious and disabling than ischemic

4 stroke and its mortality rate can be as high as 68% [2]. Though advanced therapies, including intravenous

PT
5 thrombolysis, arterial embolectomy, carotid stenting and carotid endarterectomy, are used in the therapy of

RI
6 ischemic stroke, at present, few improvement treatments have been identified for ICH. Therefore, emphasis has

7 been placed on secondary prevention aiming at reducing common risk factors of ICH, mainly hypertension,

SC
8 hyperlipidemia, diabetes, alcohol consumption and smoking [3-5]. In addition, several studies have focused on

U
9 other biomarkers e.g. serum uric acid (UA), for which it has not been ascertained whether it associated with ICH
AN
10 [6,7].

11 Uric acid is the end product of purine metabolism; it is catalyzed by xanthine oxidoreductase and eventually
M

12 eliminated by the kidney in the human body [8,9]. There is some evidence for an underlying mechanism linking
D

13 high levels of serum UA to elevated risks of incidence or mortality of atherosclerotic diseases, such as ischemic
TE

14 stroke, carotid artery disease and cardiovascular disease [10-13]. However, the relationship between serum UA

15 levels and the risk of ICH remains controversial. Previous studies reported that patients with ICH exhibited
EP

16 higher concentrations of UA when compared with healthy controls [14-16]. In contrast, several studies detected
C

17 lower concentrations of UA in ICH patients in comparison with healthy controls [17-20]. The conflicting
AC

18 outcomes mentioned above may originate from a single study with a limited sample size. Furthermore,

19 comprehensive data from studies included in previous meta-analyses were mainly limited to ischemic stroke or

20 stroke rather than hemorrhagic stroke [21-23]. Therefore, we performed a systematic overview of studies to

21 identify the relationship between serum UA level and the risk of ICH.

22

2
 ACCEPTED MANUSCRIPT

1 Materials and methods

2 Inclusion and exclusion criteria

3 Published studies reporting results on assessment of serum UA levels and ICH were identified. Studies were

4 eligible if they fulfilled the following criteria: (1) studies were case-control design; (2) studies investigated the

PT
5 correlation between UA and ICH patients; (3) the cases were confirmed ICH patients, and the controls were

RI
6 healthy people; (4) serum UA levels were provided in the case and control group. There were the following

7 exclusion criteria: (1) did not meet the inclusion criteria; (2) duplicated publications or studies with overlapping

SC
8 data; (3) abstracts, reviews, proceedings, letters, case report, or meta-analysis.

U
9
AN
10 Literature search

11 Databases of Embase, Pubmed, Web of Science and Cochrane Library were comprehensively searched for
M

12 articles up to December 30, 2017. The terms searched were as follows: (“uric acid” OR urate OR hyperuricemia)
D

13 AND (“intracerebral hemorrhage” OR “cerebral hemorrhage” OR “brain hemorrhage” OR “hemorrhagic stroke”

TE

14 OR “intracranial hemorrhage”). Additional articles were identified through a manual search. The reference lists

15 of all identified relevant reviews were also examined.

EP

16
C

17 Data extraction and quality assessment

AC

18 Data extraction from the selected trials was conducted independently by two reviewers according to a standard

19 protocol. Discrepancies were resolved by discussion or by consulting the original report. Data on the

20 characteristics of eligible studies was collected as follows: first author, publication year, country, detection

21 method for serum UA, other information including proportion of men, number of participants, serum UA levels

22 and mean age in the groups of ICH, healthy control and ischemic stroke. This study was performed by

3
 ACCEPTED MANUSCRIPT

1 guidelines of preferred reporting items for systematic reviews and meta-analyses (PRISMA) [24] (protocol

2 number: PROSPERO CRD 98640). The Newcastle-Ottawa Scale (NOS) was conducted to assess the quality of

3 studies [25]. Each study scored up to 9 points: 4 points for selection of the study, 2 points for comparability of

4 the groups, and 3 points for ascertainment of outcomes. Studies with 6 to 9 points were considered as high

PT
5 quality, while those with 0 to 6 points were considered to be of suboptimal quality.

RI
6

7 Statistical analysis

SC
8 Statistical analysis was performed using the software Review Manager 5.2 and STATA version 14.0 in this

U
9 meta-analysis. A random effects model or a fixed effects model was used to calculated mean difference (MD)
AN
10 with 95% confidence intervals (CI) to measure the differences in serum UA level between the case group and

11 control group, and the Z test was utilized to assess the pooled effect size. I2 test (25, 50, and 75% represented
M

12 low, moderate, and high degrees of heterogeneity) was calculated to investigate the heterogeneity among studies
D

13 [26,27]. Fixed effects models were utilized for the evidence of no significant heterogeneity (I2 test exhibited
TE

14 ≤50%, and p>0.1); otherwise the random effects model was applied [26,28]. We also performed subgroup

15 analyses on age (<65 or older), gender, ethnicity and sample size of participants (≤50 or more). Sensitivity
EP

16 analysis was carried out by omitting a single study at each turn to examine its influence on the pooled risk
C

17 estimates. The risk of publication bias was evaluated by visual inspection of a funnel plot, Begg's test and
AC

18 Egger's test.

19

20 Results

21 Study selection and characteristics

22 A total of 209 relevant abstracts were identified initially. After screening the titles or abstracts, 177 irrelevant or

4
 ACCEPTED MANUSCRIPT

1 duplicated articles were excluded and thus 32 potentially eligible studies were assessed. After reviewing the full

2 text of the remaining studies, 26 studies were further excluded, mainly because 7 of the studies had irrelevant

3 outcomes whereas the other 19 studies lacked usable data. Thus, 6 eligible studies met our inclusion criteria and

4 these were finally included into the meta-analysis [14-20,29-31] (Fig. 1). These six studies were categorized as

PT
5 high quality according to the NOS criteria (Table 1). One of the six studies was divided into two subgroups i.e.

RI
6 the ischemic stroke group was divided into cerebral infarction (CI) group and transient ischemic attack (TIA)

7 group in this study [19]; this explains why the key characteristics summarized in Table 1 consist of 7 subgroups.

SC
8 The studies had examined a total of 7062 participants; of these, 345 were ICH patients, 574 were patients with

U
9 ischemic stroke and 535 were healthy controls. The data of UA levels in ICH patients and healthy controls were
AN
10 extracted from six studies [14,15,17-20] (Fig. 2A); five subgroups containing UA levels in groups of ischemic

11 stroke and healthy controls were identified [15,18-20] (Fig. 2B). Subgroup analysis based on age presents four
M

12 studies with younger patients (age<65) and two studies with older patients (age≥65) (Fig. 3A). In five of the six
D

13 trials which had investigated the relationship between serum UA levels in ICH and healthy controls, the study
TE

14 subjects were Asians, whereas the sixth trial had been performed in Caucasians (Fig. 3B). In the subgroup

15 analysis of gender, two studies had assayed serum UA levels in males and females respectively (Fig. 4A). The
EP

16 subgroup analysis of sample sizes indicated that there were two studies with large samples (n>50) and four
C

17 studies with small samples (n≤50) investigating UA levels in ICH and healthy controls (Fig. 4B). The flow
AC

18 diagram of the study selection process and the screening steps are presented in Figure 1.

19

20 Meta-analysis

21 The results indicated that there was low heterogeneity (I2=40%) among those six studies reporting differences of

22 serum UA levels in the comparisons of groups of ICH and healthy controls, but in contrast, high heterogeneity

5
 ACCEPTED MANUSCRIPT

1 (I2=81%) among those four studies reporting differences of serum UA levels between groups of ischemic stroke

2 and healthy controls. Consequently, a fixed-effects model was used in the groups of ICH and healthy controls to

3 pool data, and a random-effects model was used in the groups of ischemic stroke and healthy controls. No

4 significant differences were found in the comparisons of serum UA levels between the ICH group and controls

PT
5 (MD=3.28, 95% CI=-9.04-15.61, p=0.60) (Fig. 2A). Patients with ischemic stroke had a higher serum UA level

RI
6 than those in the healthy controls (MD=30.12, 95% CI=3.91-56.32, p=0.02) (Fig. 2B). Subgroup analysis of age

7 demonstrated that the level of UA was higher in ICH patients over 65 years old than that of healthy controls

SC
8 (age≥65: MD=18.70, 95% CI=1.44-35.96, p=0.03, I2=0%), while no statistical differences of UA levels were

U
9 found between ICH patients less than 65 years old and healthy controls (age<65: MD=-12.77, 95%
AN
10 CI=-30.38-4.84, p=0.16, I2=0%); and I2 was 84% in the test for subgroup differences (Fig. 3A). The results of

11 subgroup analysis on ethnicity indicated that in both Asians and Caucasians, there were no statistically
M

12 significant differences with respect to the serum UA levels of ICH patients and healthy controls (Asians:
D

13 MD=5.97, 95% CI=-9.06-21.00, p=0.44; Caucasians: MD=-30.00, 95% CI=-68.43-8.43, p=0.13) (Fig. 3B). A
TE

14 subgroup analysis based on gender revealed that neither men nor women displayed differences in relation to

15 serum UA levels in ICH patients compared to healthy controls (Men: MD=-25.67, 95% CI=-56.08-4.73, p=0.10,
EP

16 I2=0%; Women: MD=4.36, 95% CI=-27.19-35.91, p=0.79, I2=0%); and no statistical differences were present
C

17 between those two groups (p=0.18, I2=44.6%) (Fig. 4A). A further subgroup analysis on sample size revealed no
AC

18 statistical differences in relation to serum UA levels between ICH patients and healthy controls (large samples

19 size: MD=10.25, 95% CI=-20.54-41.05, p=0.51; small samples: MD=-5.82, 95% CI=-25.41-13.78, p=0.56) (Fig.

20 4B).

21

22

6
 ACCEPTED MANUSCRIPT

1 Sensitivity analysis and publication bias

2 Sensitivity analysis performed by omitting one individual study at a time revealed that no single study exerted

3 any obvious influence on the pooled mean difference (Supplementary Fig. 1). Not only the shape of funnel plot

4 for overall analysis seemed symmetrical (Supplementary Fig. 2), both of the Begg’s test (p=0.26)

PT
5 (Supplementary Fig. 3) and Egger’s test (p=0.189) (Supplementary Fig. 4) suggested that there was no obvious

RI
6 risk of publication bias in these studies.

SC
8 Discussion

U
9 This systematic review involved a search of the literature to explore whether elevated levels of serum UA would
AN
10 be associated with the risk of ICH. A total of six eligible studies with approximately 345 ICH patients were

11 included in present meta-analysis [14,15,17-20]. As far as we are aware, there is no relevant meta-analysis to
M

12 investigate the relationship between serum UA levels and ICH; it is possible to undertake a more precise
D

13 assessment and estimate of the impact of serum UA on ICH if one conducts a meta-analysis than can be
TE

14 acquired from a single study. The results emerging from comprehensive meta-analysis confirmed that there was

15 no significant difference in UA levels between ICH and controls, while patients with ischemic stroke exhibited
EP

16 prominently higher UA levels than healthy controls.

C

17 The inconsistent results of our meta-analysis between ICH and ischemic stroke are supported for the
AC

18 speculation that the pathogenesis of atherosclerosis likely involves complex changes related to serum UA levels

19 in different conditions. Numerous studies have revealed that UA, as a pro-oxidant, is positively associated with

20 high risk of ischemic stroke and cardiovascular disease [19,29]. Evidence emerging from a meta-analysis

21 revealed that higher serum UA levels may contribute to cerebral infarction in diabetic patients [22]. To illustrate

22 this point, UA has been found to stimulate the synthesis in rats of monocyte chemoattractant protein-1 by

7
 ACCEPTED MANUSCRIPT

1 vascular smooth muscle cells, which is known to have a key role in stimulating macrophage infiltration in

2 atherosclerotic vessels [30]. In patients with antihypertensive medications, high UA level appears to be

3 associated with inadequate control of blood pressure, both of which would accelerate the progression of arterial

4 stiffness [31]. Another clinical trial demonstrated that serum UA is positively related to pulse wave velocity,

PT
5 which is considered to be an early predictor of atherosclerosis in subjects with normal or mildly reduced renal

RI
6 function [32]. Although UA has pro-oxidant properties that may be neurotoxic promoting the progression of

7 atherosclerosis, in some chemical microenvironments, it can also act as an antioxidant i.e. it may be

SC
8 neuroprotective [33]. It has been demonstrated that an elevated UA level tends to be injurious rather than

U
9 protective in subjects with ischemic stroke as well as those with cardiovascular disease [34,35]; and there is one
AN
10 report that the high prevalence of hyperuricemia in black Africans was associated with a poor functional

11 outcome after stroke [36]. In the present study, higher levels of serum UA were found in patients of ischemic
M

12 stroke than those of healthy controls, which implied that hyperuricemia might be a risk factor for ischemic
D

13 stroke. However, no significant difference in the UA level was revealed between ICH and healthy controls,
TE

14 which suggested that the double-edged properties i.e. pro-oxidant and antioxidant, may reduce the impact of UA

15 on the risk of ICH.

EP

16 Furthermore, there may also be other causes affecting the levels of UA in patients with ICH. There is
C

17 convincing evidence that serum UA levels were clearly decreased in patients with cerebral amyloid angiopathy
AC

18 (CAA) which is considered to be an important pathogeny of ICH in older people [37]. One distinct feature of

19 CAA pathology is the deposition of Aβ in brain vessels, leading to injury of arteries due to Aβ-mediated

20 oxidative stress, and thus to an increased predisposition towards ICH [38]. It has been reported that UA is an

21 important endogenic antioxidant; it accounts for approximately one half of the antioxidant capacity of serum,

22 and this can reduce oxidative stress through its actions as a scavenger of free radicals and iron chelator [39].

8
 ACCEPTED MANUSCRIPT

1 Due to the consumption of UA in the disease-related free radical production, the concentrations of UA decrease

2 markedly in patients with CAA [40]. It can be argued that there were probably CAA-associated ICH patients

3 recruited in the studies examined in the current meta-analysis, which may contribute to our results that lower

4 concentrations of UA were exhibited in ICH patients than those of ischemic stroke patients.

PT
5 In the subgroup analysis of age, serum UA was positively associated with the risk of ICH in patients over 65

RI
6 years old, whereas the levels of UA in ICH patients younger than 65 years old were comparable to those of

7 healthy controls, these results of which suggested serum UA might be a potential risk factor for ICH in the

SC
8 elderly. Though there was no statistically significant difference between ICH patients and healthy controls,

U
9 serum UA level detected in men was higher than women, which was in agreement with previous study of
AN
10 cardiovascular patients [41]. The inconsistency of UA levels in different age and gender may be associated with

11 factors such as lifestyles, diet, purine metabolism, renal function, estrogen levels and genotyped SLC2A9 single
M

12 nucleotide polymorphisms [42,43]. In addition, the subgroup analysis based on either ethnicity, gender or
D

13 sample size in relation to serum UA levels between ICH and healthy controls detected no statistically significant
TE

14 differences, which implied that there were no race-specific differences between Asians and Caucasians,

15 gender-related differences between men and women, or sample size-related differences between large and small
EP

16 on the results of our meta-analysis. Although low statistical heterogeneity of correlation was existed in the
C

17 studies of evaluating the hazard ratio of high UA level for ICH, there may be clinical heterogeneity and
AC

18 methodological heterogeneity existed in the meta-analysis. Present results of subgroup analyses suggested that

19 potential heterogeneity may be derived from differences in age (I2=84%) or ethnicity (I2=65.7%), rather than

20 diversities in gender or sample size. Other possible sources of heterogeneity may include different diagnose

21 methods for ICH, various assays for UA measurements, diverse cut-off values of UA, baseline health status of

22 different subjects, and varying covariates adjusted in each individual study.

9
 ACCEPTED MANUSCRIPT

2 Limitations

3 Several limitations of this meta-analysis should be recognized. Though low statistical heterogeneity was found

4 in certain comparisons, there might be clinical heterogeneity or methodological heterogeneity in the

PT
5 meta-analysis. Furthermore, the different methods of detection for serum UA levels may have an impact on

RI
6 sensitivity and reliability of the underlying data in the results. As all of the blood samples of serum UA were

7 obtained after ICH rather than before its occurrence, the levels of serum UA may have been altered by ICH

SC
8 event. There were probably some of the patients included in the pooled analysis which may have been treated

U
9 for hyperuricemia. The proportion of CAA-associated ICH patients, patients of different ages and ethnicity
AN
10 recruited in the study may also affect the results of meta-analysis. As the number of included studies is less than

11 ten in the meta-analysis, effect estimates for the risk of publication bias assessed by funnel plot, Begge’s test and
M

12 Egger’s test should be interpreted with caution.

D

13
TE

14 Conclusions

15 Although the main finding of the study is that serum UA does not increase the risk of ICH, the meta-analysis
EP

16 provides support that the dual roles of UA i.e. pro-oxidant and antioxidant have potential mechanisms linking
C

17 the pathogenesis of ICH. In addition, serum UA may be a potential risk factor for ICH in the elderly. There is no
AC

18 evidence for any race-specific differences between Asians and Caucasians as well as gender-related differences

19 between men and women on the risk of ICH. Further clinical investigations with comprehensive data,

20 CAA-associated ICH patients recruited or not, large sample sizes and well-designed studies will be necessary to

21 validate our results.

22

10
 ACCEPTED MANUSCRIPT

3 Conflict of interest

4 The authors declared they do not have anything to disclose regarding conflict of interest with respect to this

PT
5 manuscript.

RI
6

7 Financial support

SC
8 This work was supported by National Natural Science Foundation of China (No. 81372104, Chuansheng Zhao);

U
9 The Shenyang Population and Health Technical Critical Special Project (No. F16-206-9-01, Chuansheng Zhao);
AN
10 The Program of the Distinguished Professor of Liaoning Province, Neurology (Chuansheng Zhao).

11
M

12 Author contributions
D

13 Zhou, Z. K., Zhao, M. and Zhao, C. S. conceived the study. Zhou, Z. K. and Liang, Y. F. selected reports and
TE

14 extracted the data. Zhou, Z. K., Lin, J. Y., Zhang, X. Q., and Qu, H. L. analysed and interpreted the data. Zhou,

15 Z. K., Xu, J. J., and Zhao, C. S. wrote the first draft of the manuscript. All authors reviewed the manuscript and
EP

16 approved the final version.

C

17
AC

18 Acknowledgments

19 We would like to acknowledge all the authors of studies included in this paper.

20

21 References

22 1. Biffi A, Devan WJ, Anderson CD, Ayres AM, Schwab K et al (2011) Statin use and outcome after

11
 ACCEPTED MANUSCRIPT

1 intracerebral hemorrhage: case-control study and meta-analysis. Neurology 76(18):1581-8

2 2. Koton S, Schneider AL, Rosamond WD, Shahar E, Sang Y et al (2014) Stroke incidence and mortality trends

3 in US communities, 1987 to 2011. JAMA 312(3):259-68

4 3. Wang X, Dong Y, Qi X, Huang C, Hou L (2013) Cholesterol levels and risk of hemorrhagic stroke: a

PT
5 systematic review and meta-analysis. Stroke 44(7):1833-9

RI
6 4. Tsai CF, Anderson N, Thomas B, Sudlow CL (2016) Comparing Risk Factor Profiles between Intracerebral

7 Hemorrhage and Ischemic Stroke in Chinese and White Populations: Systematic Review and Meta-Analysis.

SC
8 PLoS One 11(3):e0151743

U
9 5. Boulouis G, Morotti A, Goldstein JN, Charidimou A (2017) Intensive blood pressure lowering in patients with
AN
10 acute intracerebral haemorrhage: clinical outcomes and haemorrhage expansion. Systematic review and

11 meta-analysis of randomised trials. J Neurol Neurosurg Psychiatry 88(4):339-345

M

12 6. Wu H, Jia Q, Liu G, Liu L, Pu Y et al (2014) Decreased uric acid levels correlate with poor outcomes in acute
D

13 ischemic stroke patients, but not in cerebral hemorrhage patients. J Stroke Cerebrovasc Dis 23(3):469-75
TE

14 7. Karagiannis A, Mikhailidis DP, Tziomalos K, Sileli M, Savvatianos S et al (2007) Serum uric acid as an

15 independent predictor of early death after acute stroke. Circ J 71(7):1120-7

EP

16 8. Cherubini A, Polidori MC, Bregnocchi M, Pezzuto S, Cecchetti R et al (2000) Antioxidant profile and early
C

17 outcome in stroke patients. Stroke 31(10):2295-300

AC

18 9. So A, Thorens B (2010) Uric acid transport and disease. J Clin Invest 120(6):1791-9

19 10. Weir CJ, Muir SW, Walters MR, Lees KR (2003) Serum urate as an independent predictor of poor outcome

20 and future vascular events after acute stroke. Stroke 34(8):1951-6

21 11. Hozawa A, Folsom AR, Ibrahim H, Nieto FJ, Rosamond WD et al (2006) Serum uric acid and risk of

22 ischemic stroke: the ARIC study. Atherosclerosis 187(2):401-7

12
 ACCEPTED MANUSCRIPT

1 12. Feig DI, Kang DH, Johnson RJ (2008) Uric acid and cardiovascular risk. N Engl J Med 359(17):1811-21

2 13. Kumral E, Karaman B, Orman M, Kabaroglu C (2014) Association of uric acid and carotid artery disease in

3 patients with ischemic stroke. Acta Neurol Scand 130(1):11-7

4 14. Guo Q, Li H (2015) The relationship between serum ferritin, uric acid and prognosis of patients with

PT
5 intracerebral hemorrhage. Chin J Neuorsurg Dis Res 6:510-3

RI
6 15. Pan M, Shao D, Liu Y, Li W, Zhao R et al (2015) A comparison study on risk factors in patients with

7 ischemic stroke versus hemorrhagic stroke. Chin J Dis Control Prev 9:883-6

SC
8 16. Prasad C, Dwivedi NC, Gupta P, Shukla SK, Shukla R et al (2016) Serum uric acid level in patients of acute

U
9 stroke. Int J Adv Med. 3:393-7
AN
10 17. Plildori M, Mecocci P, Frei B (2001) Plasma vitamin C levels are decreased and correlated with brain

11 damage in patients with intracranial hemorrhage or head trauma. Stroke 32(4):898-902

M

12 18. Chen L, Wang S, Zhang X (2008) The relationship between acute cerebral infarction and intracerebral
D

13 hemorrhage with serum lipid, uric acid and blood coagulation indexes. Chinese Journal of Stroke 3:394-7
TE

14 19. Peng F, Zhang B, Zhong X, Li J, Xu G et al (2008) Serum uric acid levels of patients with multiple sclerosis

15 and other neurological diseases. Mult Scler 14(2):188-96

EP

16 20. Zhu R, Zhou N (2012) Gender difference of serum bilirubin in stroke patients. Chinese General Practice
C

17 8:2533-5
AC

18 21. Chao TF, Liu CJ, Chen SJ, Wang KL, Lin YJ et al (2014) Hyperuricemia and the risk of ischemic stroke in

19 patients with atrial fibrillation -Could it refine clinical risk stratification in AF? Int J Cardiol 170(3):344-9

20 22. Du L, Ma J, Zhang X (2017) Higher Serum Uric Acid May Contribute to Cerebral Infarction in Patients with

21 Type 2 Diabetes Mellitus: a Meta-Analysis. J Mol Neurosci 61(1):25-31

22 23. Li M, Hou W, Zhang X, Hu L, Tang Z (2014) Hyperuricemia and risk of stroke: A systematic review and

13
 ACCEPTED MANUSCRIPT

1 meta-analysis of prospective studies. Atherosclerosis 232(2):265-70

2 24. Moher D, Liberati A, Tetzlaff J, Altman DG; PRISMA Group (2009) Preferred reporting items for

3 systematic reviews and meta-analyses: the PRISMA statement. PLoS Med 6(7):e1000097

4 25. Stang A (2010) Critical evaluation of the Newcastle-Ottawa scale for the assessment of the quality of

PT
5 nonrandomized studies in meta-analyses. Eur J Epidemiol 25(9):603-5

RI
6 26. Zintzaras E, Ioannidis JP (2005) HEGESMA: genome search meta-analysis and heterogeneity testing.

7 Bioinformatics 21(18):3672-3

SC
8 27. Higgins JP, Thompson SG, Deeks JJ, Altman DG (2003) Measuring inconsistency in meta-analyses. BMJ

U
9 327(7414):557-60
AN
10 28. Higgins JP (2002) Quantifying heterogeneity in a meta-analysis. Stat Med 21:1539-58

11 29. Wang Z, Lin Y, Liu Y, Chen Y, Wang B et al (2016) Serum uric acid levels and outcomes after acute
M

12 ischemic stroke. Mol Neurobiol 53(3):1753-9

D

13 30. Kanellis J, Watanabe S, Li JH, Kang DH, Li P t al (2003) Uric acid stimulates monocyte chemoattractant
TE

14 protein-1 production in vascular smooth muscle cells via mitogen-activated protein kinase and

15 cyclooxygenase-2. Hypertension 41(6):1287-93

EP

16 31. Cicero AF, Rosticci M, Fogacci F, Grandi E, D'Addato S et al (2017) High serum uric acid is associated to
C

17 poorly controlled blood pressure and higher arterial stiffness in hypertensive subjects. Eur J Intern Med
AC

18 37:38-42

19 32. Cicero AF, Kuwabara M, Johnson R, Bove M, Fogacci F et al (2018) LDL-oxidation, serum uric acid,

20 kidney function and pulse-wave velocity: Data from the Brisighella Heart Study cohort. Int J Cardiol

21 261:204-8

22 33. Proctor P H (2008) Uric acid: neuroprotective or neurotoxic? Stroke 39(5):e88

14
 ACCEPTED MANUSCRIPT

1 34. Maxwell AJ, Bruinsma KA (2001) Uric acid is closely linked to vascular nitric oxide activity. Evidence for

2 mechanism of association with cardiovascular disease. J Am Coll Cardiol 38(7):1850-8

3 35. Kanellis J, Johnson RJ (2003) Elevated uric acid and ischemic stroke: Accumulating evidence that it is

4 injurious and not neuroprotective. Stroke. 34(8):1956-7

PT
5 36. Mapoure YN, Ayeah CM, Doualla MS, Ba H, Ngahane HBM et al (2017) Serum Uric Acid Is Associated

RI
6 with Poor Outcome in Black Africans in the Acute Phase of Stroke. Stroke Res Treat 2017:1935136

7 37. Hu Q, Liu A, Huang M, Cheng L, Kang H et al (2014) Lower serum uric acid levels in cerebral amyloid

SC
8 angiopathy: a pilot study. Neurol Sci 35(7):1035-9

U
9 38. Perry G, Smith MA, McCann CE, Siedlak SL, Jones PK et al (1998) Cerebrovascular muscle atrophy is a
AN
10 feature of Alzheimer’s disease. Brain Res 791(1-2):63-6

11 39. Davies KJ, Sevanian A, Muakkassah-Kelly SF, Hochstein P (1986) Uric acid-iron ion complexes. A new
M

12 aspect of the anti-oxidant functions of uric acid. Biochem J 235(3):747-54

D

13 40. Brouns R, Wauters A, Van De Vijver G, De Surgeloose D, Sheorajpanday R et al (2010) Decrease in uric
TE

14 acid in acute ischemic stroke correlates with stroke severity, evolution and outcome. Clin Chem Lab Med

15 48(3):383-90
EP

16 41. Fang J, Alderman MH (2000) Serum uric acid and cardiovascular mortality the NHANES I epidemiologic
C

17 follow-up study, 1971-1992. National Health and Nutrition Examination Survey. JAMA 283(18):2404-10
AC

18 42. Yang XL, Kim Y, Kim TJ, Jung S, Kim CK et al (2016) Association of serum uric acid and cardioembolic

19 stroke in patients with acute ischemic stroke. J Neurol Sci 370:57-62

20 43. Brandstätter A, Lamina C, Kiechl S, Hunt SC, Coassin S et al (2010) Sex and age interaction with genetic

21 association of atherogenic uric acid concentrations. Atherosclerosis 210(2):474-8

22 Figure Legends

15
 ACCEPTED MANUSCRIPT

1 Table 1 General characteristics of the included studies

2 Fig. 1 Flow chart of study selection in the meta-analysis

3 Fig. 2 Forest plots for the comparisons of serum UA levels between intracerebral hemorrhage (ICH) patients

4 and healthy controls (A), as well as between ischemic stroke and healthy controls (B). CI: confidence interval

PT
5 Fig. 3 Forest plots of the subgroup analyses on age (A) and ethnicity (B) in relation to serum UA levels between

RI
6 intracerebral hemorrhage (ICH) patients and healthy controls. CI: confidence interval

7 Fig. 4 Forest plots of the subgroup analyses on gender (A) and sample size (B) in relation to serum UA levels

SC
8 between intracerebral hemorrhage (ICH) patients and healthy controls. CI: confidence interval

U
9 Supplement Material for online publication only
AN
10 Fig. Forest plot of the sensitivity analysis in the present meta-analysis (individual names for each study have

11 been omitted). CI: confidence interval

M

12 Fig. Funnel plot for detecting the risk of publication bias in the meta-analysis. MD: mean difference
D

13 Fig. Begg’s plot for detecting the risk of publication bias in the meta-analysis. SMD: standard mean
TE

14 difference

15 Fig. Egger’s plot for detecting the risk of publication bias in the meta-analysis
C EP
AC

16
 ACCEPTED MANUSCRIPT

PT
Hemorrhagic Controls Ischemic
No. of UA, No. of UA, No. of UA,

RI
Detecting
First author Year Country Age, Age, Age, NOS
methods male µmol/L male µmol/L male µmol/L
Years Years Years
patients Mean±SD patients Mean±SD patients Mean±SD

SC
Polidori MC 2001 America HPLC 7/13 255.3±63.2 62.4±6.1 10/20 285.3±39.3 58.3±5.8 NR NR NR 8

U
Chen L 2008 China ELISA 28/50 189.31±127.91 64±11 31/50 210.28±70.59 64±11 36/62 268.8±85.75 66±10 7

AN
Peng F(CI) 2008 China ELISA 29/44 311.9±104.7 56.6 43/89 312.1±92.8 41.0 50/84 291.3±101.6 56.0 8

Peng F(TIA) 2008 China ELISA 29/44 311.9±104.7 56.6 43/89 312.1±92.8 41.0 26/38 344.6±130.6 64.1 8

M
64.07±12.
1

Zhu R 2012 China ELISA 59/102 289.17±125.23 62.75±12.53 57/129 295.85±92.42 61.55±13.63 68/136 306.33±84.33 8
53

D
Guo Q 2015 China NR 23/40 238.69±62.25 66.12±9.14 24/40 226.98±52.53 NR NR NR NR 7

TE
Pan M 2015 China NR 60/96 322.16±99.16 66.22±10.67 137/207 297.33±94.57 64.00±7.36 150/254 350.75±112.71 69.78±9.8 7

M male, n number, UA uric acid, NR not reported, ELISA enzyme-linked immunosorbent assay, HPLC high-performance liquid chromatography, CI cerebral infarction, TIA transient ischemic attack, NOS
EP
Newcastle-Ottawa Scale
C
AC
 ACCEPTED MANUSCRIPT

PT
RI
U SC
AN
M
D
TE
EP
C
AC
 ACCEPTED MANUSCRIPT

PT
RI
U SC
AN
M
D
TE
EP
C
AC
 ACCEPTED MANUSCRIPT

PT
RI
U SC
AN
M
D
TE
EP
C
AC
 ACCEPTED MANUSCRIPT

PT
RI
U SC
AN
M
D
TE
EP
C
AC
 ACCEPTED MANUSCRIPT
Serum uric acid did not increase the risk of intracerebral hemorrhage, the reason of which might be
that the dual roles of uric acid i.e. pro-oxidant and antioxidant have potential mechanisms linking the
pathogenesis of intracerebral hemorrhage.
Higher levels of serum uric acid were found in patients of ischemic stroke than those of healthy
controls, which implied that hyperuricemia might be a risk factor for ischemic stroke.
In the subgroup analysis of age, serum UA was positively associated with the risk of ICH in patients
over 65 years old, which suggested serum UA might be a potential risk factor for ICH in the elderly.
There were no race-specific differences in uric acid levels between Asians and Caucasians as well as

PT
gender-related differences between men and women on the risk of intracerebral hemorrhage.

RI
U SC
AN
M
D
TE
C EP
AC