Association between genetic variant on chromosome 12p13 and stroke survival and recurrence: a one year prospective study in Taiwan
- Yi-Chen Hsieh1,
- Sudha Seshadri2, 3,
- Wen-Ting Chung4, 5,
- Fang-I Hsieh1, 6,
- Yi-Hsiang Hsu7, 8,
- Huey-Juan Lin9,
- Hung-Pin Tseng10,
- Li-Ming Lien11,
- Chyi-Huey Bai1, 12,
- Chaur-Jong Hu13,
- Jiann-Shing Jeng14,
- Sung-Chun Tang14,
- Chin-I Chen4,
- Chia-Chen Yu1,
- Hung-Yi Chiou1, 6Email author and
- the Formosa Stroke Genetic Consortium (FSGC)
© Hsieh et al; licensee BioMed Central Ltd. 2012
Received: 24 October 2011
Accepted: 3 January 2012
Published: 3 January 2012
The association between ischemic stroke and 2 single nucleotide polymorphisms (SNPs) on chromosome 12p13, rs12425791 and rs11833579 appears inconsistent across different samples. These SNPs are close to the ninjurin2 gene which may alter the risk of stroke by affecting brain response to ischemic injury. The purpose of this study was to investigate the association between these two SNPs and ischemic stroke risk, as well as prognostic outcomes in a Taiwanese sample.
We examined the relations of these two SNPs to the odds of new-onset ischemic stroke, ischemic stroke subtypes, and to the one year risk of stroke-related death or recurrent stroke following initial stroke in a case-control study. A total of 765 consecutive patients who had first-ever ischemic stroke were compared to 977 stroke-free, age-matched controls. SNPs were genotyped by Taqman fluorescent allelic discrimination assay. The association between ischemic stroke and SNPs were analyzed by multivariate logistic regression. Cox proportional hazard model was used to assess the effect of individual SNPs on stroke-related mortality or recurrent stroke.
There was no significant association between SNP rs12425791 and rs11833579 and ischemic stroke after multiple testing corrections. However, the marginal significant association was observed between SNP rs12425791 and large artery atherosclerosis under recessive model (OR, 2.30; 95%CI, 1.22-4.34; q-value = 0.062). Among the 765 ischemic stroke patients, 59 died or developed a recurrent stroke. After adjustment for age, sex, vascular risk factors and baseline stroke severity, Cox proportional hazard analysis indicated that the hazard ratios were 2.76 (95%CI, 1.34-5.68; q-value, 0.02) and 2.15 (95%CI, 1.15-4.02; q-value, 0.03) for individuals with homozygous variant allele of rs12425791 and rs11833579, respectively.
This is a precedent study that found genetic variants of rs12425791 and rs11833579 on chromosome 12p13 are independent predictors of stroke-related mortality or stroke recurrence in patients with incident ischemic stroke in Taiwan. Further study is needed to explore the details of the physiological function and the molecular mechanisms underlying the association of this genetic locus with ischemic stroke.
Keywordsstroke single nucleotide polymorphisms survival
A genome-wide association study (GWAS) conducted by the Cohorts for Heart and Aging Research Consortium in Genetic Epidemiology (CHARGE) reported a novel association between ischemic stroke and 2 single nucleotide polymorphisms (SNPs) on chromosome 12p13, rs12425791 and rs11833579, in a Caucasian sample . The researchers further examined the association between the two implicated SNPs and ischemic stroke in two independent replication samples, an African-American community-based cohort and a Dutch case-control. From both samples, they observed that rs12425791 was significantly associated with the incidence of ischemic stroke, particular in atherothrombotic stroke . Nevertheless, the results of subsequent validated studies have been inconsistent. The International Stroke Genetics Consortium (ISGC) and the Wellcome Trust Case-Control Consortium 2 (WTCCC2) performed a meta-analysis combining case-control data among European ancestry and a population-based genomewide cohort study; however, the results were controversial . In addition, an Italian and a Swedish case-control study also failed to show any associations between the two genetic variants on chromosome 12p13 and ischemic stroke [3, 4]. Conversely, a large population-based case-control study found a significant association between atherothrombotic stroke and rs12425791 in Japan .
The 2 SNPs on chromosome 12p13, rs12425791 and rs11833579 are in close proximity to the gene NINJ2, which encodes ninjurin2, a member of the 'ninjurin' or nerve-injury-induced protein family . NINJ2 plays a role in nerve regeneration and may increase the risk of stroke by altering brain response to ischemic injury. If this were true, the minor allele that increases stroke risk would also be associated with greater stroke severity and disability. In this case-control study, we investigated the association between the two SNPs, rs12425791 and rs11833579 and the risk of ischemic stroke among Taiwanese population who had first-ever ischemic strokes. Furthermore, the study also examined whether the associations between these two SNPs and end-point of stroke-related mortality or recurrent stroke were present.
Subject recruitment and study design
The study was conducted in the Formosa Stroke Genetic Consortium (FSGC), a platform for hospital collaborations on studies related to the molecular biology of cerebrovascular diseases. A standard FSGC operation manual was developed after a series of consensus conferences joining by an expert panel (5 stroke neurologists and 3 epidemiologists) and revised after a pilot study completed in three hospitals (2 medical centers and 1 regional hospital). All the research assistants and study nurses from the participating hospitals were trained on the standard procedure of case enrollment, including structured questionnaire and blood sample collection. The hospitals of FSGC have successfully established the Taiwan Stroke Registry (TSR), which provided a comprehensive insight into stroke management . Ten hospitals joined the FSGC between 2005 to 2010 (Additional file 1, section 1) and 1732 stroke patients (84.3% ischemic, 5.5% transient ischemic attack[TIA] and 9.6% hemorrhagic) were recruited from FSGC hospitals and admitted to the Department of Neurology. Among 1460 ischemic stroke patients, there were 1103(75.5%) first-ever stroke and 357 (24.5) recurrent stroke. The criteria for the diagnosis of stroke have been detailed in previous study . Ischemic stroke is an onset of focal neurological deficit with signs or symptoms persisted longer than 24 hours with or without acute ischemic lesion(s) on brain CT or with acute ischemic diffusion-weighted imaging lesion(s) on MRI that corresponded to the clinical presentations. TIA is defined as a transient focal neurologic deficit of ischemic causes that resolved within 24 hours. The subtypes of ischemic stroke were classified according to the Trial of Org 10172 in Acute Stroke Treatment (TOAST) criteria . In the present case-control study, the subjects were composed of 765 first-ever ischemic stroke patients, who entered FSGC between 2005 and 2009. A total of 2736 healthy subjects were recruited as possible controls from a community-based prospective study of the nutrition health education program in Taipei City  and subjects who underwent physical examinations at TMUH during 2008-2009. Among these participants, 53 subjects with prevalent stroke were excluded. 977 age-matched controls were randomly selected in the remaining 2683 candidates. The study was approved by the ethics committees of the participating hospitals and Taipei Medical University on the understanding that all data would be coded and patient anonymity guaranteed. Informed consent was obtained from all participants and/or their relatives.
The design of the present study constitutes a two-step approach. In the primary analysis, we have conducted a case-control study with 765 Taiwanese patients who had first-ever ischemic stroke and 977 age-matched stroke-free controls for the two SNPs, rs12425791 and rs11833579 on chromosome 12p13. In the second step, the analysis on the association between these two SNPs and stroke-related mortality as well as stroke recurrent events were restricted only to ischemic stroke patients.
Data were collected from multiple centers as part of the TSR and applied to the FSGC for the improvement of sample quality and quantity . FSGC-trained neurologists and study nurses collected preadmission data, inpatient elements, discharge information, and follow-up at 1, 3, 6, and 12 months, an English translation of a sample form found in Additional file 1, section 2. Investigators who were in charge of assigning a National Institutes of Health Stroke Scale (NIHSS) score had certified by the FSGC and trained to complete data entry through a web-based FSGC database system. In healthy controls, data on age, sex, and presence of major vascular risk factors were compiled by trained investigators using a structured questionnaire. Subsequently, the data were extracted from medical charts or self-reported during interview under trained research assistants/nurses. The definitions of risk factors were described in Additional file 1, section 3.
Outcome data acquirement
Mortality rates were ascertained based on death-certificate data recorded at the Department of Health, Executive Yuan of Taiwan. Data of recurrent stroke were obtained in the history and examinations completed at 1, 3, 6, and 12 months follow-up in the FSGC data bank.
Genomic DNA was extracted using the phenol/chloroform method and then stored at -80°C until use. SNPs were genotyped by Taqman fluorescent allelic discrimination assay with the ABI Prism 7900HT sequence detection system (Applied Biosystems, Foster City, CA, USA) Specific Taqman probes and primers were obtained from Applied Biosystems Assay-by-DesignTM service for SNP genotyping. The genotyping success rate was greater than 98% for all SNPs. To assure data quality, we sequenced 5% of samples using the ABI 3100 DNA sequencer (Applied Biosystems). The genotype concordance rate between duplicate samples was 100%.
Student's t-test was used to compare continuous variables between cases and controls. The chi-square test was applied to test differences in categorical variables. Hardy-Weinberg equilibrium test for each SNP in healthy controls and cases were also assessed by chi-square analysis. To estimate the odds ratio (OR) and 95% confidence interval (CI), a multivariate logistic regression was applied under the adjustment of potential confounders. For each SNP, we tested for association under additive, co-dominant dominant and recessive models. The Cox proportional hazard model was used to assess the effect of individual SNPs on survival, the time of the stroke to the date of stroke-related death or recurrence-free survival time, the time from initial stroke occurrence to stroke recurrence. Hazard ratios (HRs) and 95% CI were estimated by fitting the Cox model while adjusting for stroke-related risk factors, including age, sex, hypertension, diabetes, cigarette smoking and alcohol drinking status, obesity and NIHSS score at beginning. Kaplan-Meier curves and log-rank tests were used to assess the differences in overall survival and recurrence-free survival by individual polymorphisms. We performed random-effect meta-analysis to pooled OR/HR and 95% CI across previous published studies which explored the association between SNPs on chromosome 12p13 and ischemic stroke [1–5]. All statistical analyses were performed with SAS Genetics software (version 9.1; SAS Institute, Cary, NC) or STATA . Two-tailed p values were calculated and statistical significance was set at p-values < 0.05 nominally. To account for multiple comparison, we calculated the false discovery rate (FDR) using the Benjamini-Hochberg procedure and expressed using q-values .
Characteristics of patients with first-ever ischemic stroke and healthy control subjects
Cases (n = 765)
Controls (n = 977)
Age, mean (SD), y
62.8 ± 12.2
63.4 ± 11.9
Diabetes mellitus, %
Heart disease, %
Ever smoking, %
Ever drinking, %
Body mass index, mean (SD), kg/m2
25.3 ± 3.9
24.7 ± 3.3
Fasting glucose, means (SD), mmol/L
6.9 ± 2.7
5.8 ± 1.7
HDL-C, mean (SD), mmol/L
1.1 ± 0.4
1.3 ± 0.3
LDL-C, mean (SD), mmol/L
3.3 ± 1.0
3.2 ± 0.8
Triglyceride, mean (SD), mmol/L
1.6 ± 1.1
1.5 ± 1.1
Cholesterol, mean (SD), mmol/L
5.0 ± 1.2
5.2 ± 0.9
NIHSS at baseline, median (IQR)
Large artery atherosclerosis
Small vessel occlusion
The odds ratios (ORs) of rs12425791 and rs11833579 genotypes for ischemic stroke
Total ischemic stroke
Survival analysis for rs12425791 and rs11833579 and risk of death or stroke recurrence among first-ever ischemic stroke patients
Neither death nor
observed (n = 706)
(n = 59)
HR* (95% CI)
The present study investigated the relationship between the SNP rs12425791 and rs11833579 on chromosome 12 and ischemic stroke and poor prognosis. The results showed there was a marginal significant association between SNP rs12425791 and LAA under recessive model. Previous studies concerning the relationship between SNP rs12425791 and rs11833579 and ischemic stroke were controversial. The GWAS finding and a large Japanese case control study presents an association between NINJ2 SNPs and ischemic stroke [1, 5]. Although we observed that the SNP rs12425791 and rs11833579 on chromosome 12 were not significantly associated with ischemic stroke, the increased ischemic stroke risk trend were found among subjects carried the homozygous variant allele of these two SNPs. On the contrary, the results from a meta-analysis performed by ISGC and WTCCC2 including samples of European ancestry, a small young Italian case-control sample, and a recent report on 3 Swedish case-control samples did not find an association [2–4]. The possible reasons for the difference might be ethnicity (European Caucasian versus Asian) or study design (incident/first-ever ischemic stroke versus prevalent stroke). However, different ethnicity is unlikely the concern for different results since the chromosome 12p13 SNPs were significantly associated with ischemic stroke in both European Caucasians and Asian [1, 5]. In addition, the MAF of SNP rs12425791 in our control subjects (23.5%) was similar with Han Chinese in Beijing, China (CHB) from the International HapMap project (24%), and the samples of European ancestry in the CHARGE (19%) , ISGC group (18-22%)  and the Swedish population (16-19%). For SNP rs11833579, the A allele frequency in the present study (32.9%) was 10% higher than in the CHARGE and ISGC group (23%)[1, 2], but was similar to the CHB (30%).
Concerning to study design, we performed a meta-analysis combining results among all studies related to NINJ2 gene and ischemic stroke stratified by incident/first-ever ischemic stroke and prevalent stroke. The results found that a significant association between both SNPs on NINJ2 gene and ischemic stroke for (rs12425791, OR, 1.20; 95%CI, 1.04-1.38; for rs11833579, OR, 1.30; 95%CI, 1.10-1.54) among incident/first-ever ischemic stroke patients but not prevalent cases for (rs12425791, OR, 1.03; 95%CI, 0.99-1.07; for rs11833579, OR, 1.03; 95%CI, 0.99-1.08) in Additional file 1, section 5. Survival bias might be the reason for the lack of association between the SNPs and prevalent ischemic stroke. The result of meta-analysis might also correspond to our data which indicate that patients with the homozygous variant allele of these two SNPs had an increased risk of adverse outcomes following ischemic stroke. These two SNPs rs12425791 and rs11833579, located on chromosome 12p13, are in close proximity to the NINJ2 gene. Ninjurin 2 is a homophilic adhesion molecule expressed in many tissues including peripheral nerves, especially in Schwann cells and ganglion neurons [6, 12]. It has been suggested that ninjurin 2, a nerve injury-induced protein, is involved in neuronal growth and play a role in nerve regeneration since it is up regulation during axotomy or nerve damage, promoting neurite outgrowth [6, 13]. It is possible that the level of ninjurin 2 expression affects how the brain responds to cerebral ischemic insults. In an analogous manner, NINJ2 might be a candidate gene for stroke prognosis since it may influence the reaction to brain cells to injury. Our study found the initial NIHSS and mRS was not correlated with NINJ2 gene based upon genotype (data not shown), but further observed that first-ever ischemic stroke patients with the variant allele at these two SNPs had an increased risk of unfavorable outcome. Moreover, when we analyzed the relative change of modified Rankin Scale (mRS) during one year, we also found that 2.7% of subjects were regression (ΔmRS > 0) among patients with homozygous variant allele of SNP rs12425791 while only 0.7% of subjects with G allele became worse (data not shown). These evidence support the hypothesis of our study that the minor allele of NINJ2 gene might be related to greater stroke severity.
Several published studies following by the original research articles have encountered some limitations in sample size, multiple testing, population stratification, or publication bias [14, 15]. In this study, a post hoc power calculation was done based on the survival probability were 10% and 5.6% among the variant allele carriers and the subjects with wild allele, respectively. The present sample size can reach to more than 80% power to detect both SNPs with the concluded genotypic HRs at the nominal Type I error rate less than 0.05. Also, we applied the false discovery rate method as multiple testing to prevent the false-positive association. Population stratification and publication bias may also lead to false-positive association. However, the limited evidence supports that population stratification and public bias is a minor issue for genetic association studies [16, 17].
This study has several strengths. First, all ischemic stroke patients in our study had accurate diagnosis confirmed by CT/MRI. We also examined the association between SNPs and various subtypes of ischemic stroke based on the abundant ancillary information in the FSGC data bank. Moreover, the mean time interval between stroke onset and entry was remarkably short and completed in less than one day (mean+SD: 0.89 ± 1.70), thus reduced survivor bias, a significant problem in most stroke case series utilized for genetic analyses. Finally we could adjust our analyses for initial stroke severity since the NIHSS was administered in all cases prior to initiation of intravenous thrombolysis or other stroke therapies. However, there were some limitations in this study. Survival after stroke could depend on factors other than stroke severity, such as concomitant cardiac and respiratory disease. We have not examined the possible ranges of pleiotropic associations on the NINJ2 SNPs. In addition, the haplotype analysis of these two SNPs were not showed due to low linkage disequilibrium (r2 = 0.53). According to the basis of the Hapmap data, the correlation coefficients were similar among Japanese populations (r2 = 0.69). The identified SNP rs12425791 is more likely to be in linkage disequilibrium with the causal variants. Therefore, further detailed fine-mapping or re-sequencing in the region will be needed to identify the actual causal genetic variants.
This study provides the first evidence that chromosome 12p13 are an independent predictor of mortality or stroke recurrence in patients with incident ischemic stroke in Taiwan. Further study is needed to explore the details of the physiological function and the molecular mechanisms underlying the association of this genetic locus with ischemic stroke.
List of abbreviation
single nucleotide polymorphism
genome-wide association study
Cohorts for Heart and Aging Research Consortium in Genetic Epidemiology
International Stroke Genetics Consortium
Wellcome Trust Case-Control Consortium 2
Formosa Stroke Genetic Consortium
Taiwan Stroke Registry
Trial of Org 10172 in Acute Stroke Treatment
National Institutes of Health Stroke Scale
false discovery rate
minor allele frequency
small vessel occlusion
large artery atherosclerosis
Han Chinese in Beijing
modified Rankin Scale
This work was supported by grants from the National Science Council of Taiwan (NSC97-2321-B-038-002, NSC98-2321-B-038-001 and NSC99-2321-B-038-001 and NSC99-2314-B-038-022-MY3). Additional supports were received from a Ministry of Education Topnotch Stroke Research Center grant and the Department of Health Center of Excellence for Clinical Trial and Research in Neuroscience (DOH99-TD-B-111-003, DOH100-TD-B-111-003, DOH101-TD-B-111-003), Dr. Chi-Chin Huang Stroke Research Center, and Taipei Medical University-Wan Fang Hospital (95TMU-WFH-07).
- Ikram MA, Seshadri S, Bis JC, Fornage M, DeStefano AL, Aulchenko YS, Debette S, Lumley T, Folsom AR, van den Herik EG, Bos MJ, Beiser A, Cushman M, Launer LJ, Shahar E, Struchalin M, Du Y, Glazer NL, Rosamond WD, Rivadeneira F, Kelly-Hayes M, Lopez OL, Coresh J, Hofman A, DeCarli C, Heckbert SR, Koudstaal PJ, Yang Q, Smith NL, Kase CS, Rice K, Haritunians T, Roks G, de Kort PLM, Taylor KD, de Lau LM, Oostra BA, Uitterlinden AG, Rotter JI, Boerwinkle E, Psaty BM, Mosley TH, van Duijn CM, Breteler MMB, Longstreth WT, Wolf PA: Genomewide association studies of stroke. N Engl J Med. 2009, 360: 1718-1728. 10.1056/NEJMoa0900094.PubMed CentralView ArticlePubMedGoogle Scholar
- International Stroke Genetics Consortium; Wellcome Trust Case-Control Consortium 2: Failure to validate association between 12p13 variants and ischemic stroke. N Engl J Med. 2010, 362: 1547-1550.View ArticleGoogle Scholar
- Lotta LA, Giusti B, Saracini C, Vestrini A, Volpe M, Rubattu S, Peyvandi F: No association between chromosome 12p13 single nucleotide polymorphisms and early-onset ischemic stroke. J Thromb Haemost. 2010, 8: 1858-1860. 10.1111/j.1538-7836.2010.03912.x.View ArticlePubMedGoogle Scholar
- Olsson S, Melander O, Jood K, Smith JG, Lövkvist H, Sjögren M, Engström G, Norrving B, Lindgren A, Jern C, International Stroke Genetics Consortium (ISGC): Genetic variant on chromosome 12p13 does not show association to ischemic stroke in 3 Swedish case-control studies. Stroke. 2010, 42: 214-216.View ArticlePubMedGoogle Scholar
- Matsushita T, Umeno J, Hirakawa Y, Yonemoto K, Ashikawa K, Amitani H, Ninomiya T, Hata J, Doi Y, Kitazono T, Iida M, Nakamura Y, Kiyohara Y, Kubo M: Association study of the polymorphisms on chromosome 12p13 with atherothrombotic stroke in the Japanese population. J Hum Genet. 2010, 55: 473-476. 10.1038/jhg.2010.45.View ArticlePubMedGoogle Scholar
- Araki T, Milbrandt J: Ninjurin, a novel adhesion molecule, is induced by nerve injury and promotes axonal growth. Neuron. 1996, 17: 353-361. 10.1016/S0896-6273(00)80166-X.View ArticlePubMedGoogle Scholar
- Hsieh FI, Lien LM, Chen ST, Bai CH, Sun MC, Tseng HP, Chen YW, Chen CH, Jeng JS, Tsai SY, Lin HJ, Liu CH, Lo YK, Chen HJ, Chiu HC, Lai ML, Lin RT, Sun MH, Yip BS, Chiou HY, Hsu CY: AHA/ASA Get With The Guidelines - Stroke Performance Indicators: Surveillance of Stroke Care in the Taiwan Stroke Registry. Circulation. 2010, 122: 1116-1123. 10.1161/CIRCULATIONAHA.110.936526.View ArticlePubMedGoogle Scholar
- Adams HP, Bendixen BH, Kappelle LJ, Biller J, Love BB, Gordon DL, Marsh EE: Classification of subtype of acute ischemic stroke. Definitions for use in a multicenter clinical trial. TOAST. Trial of Org 10172 in Acute Stroke Treatment. Stroke. 1993, 24: 35-41. 10.1161/01.STR.24.1.35.View ArticlePubMedGoogle Scholar
- Hsieh YC, Hung CT, Lein LM, Bai CH, Chen WH, Yeh CY, Chen YH, Hsieh FI, Chiu HC, Chiou HY, Hsu CY: A significant decrease in blood pressure through a family-based nutrition health education programme among community residents in Taiwan. Public Health Nutr. 2009, 12: 570-577. 10.1017/S1368980008002875.View ArticlePubMedGoogle Scholar
- Stata Corporation: STATA 11.0 Statistics/Data Analysis. 2009, College Station, Texas, Stata corporation, [http://www.stata.com]Google Scholar
- Benjamini Y, Hochberg Y: Controlling the false discovery rate: a practical and powerful approach to multiple testing. JR Stat Soc Ser B. 1995, 57: 289-300.Google Scholar
- Araki T, Milbrandt J: Ninjurin2, a novel homophilic adhesion molecule, is expressed in mature sensory and enteric neurons and promotes neurite outgrowth. J Neurosci. 2002, 20: 187-195.Google Scholar
- Kubo T, Yamashita T, Yamaguchi A, Hosokawa K, Tohyama M: Analysis of genes induced in peripheral nerve after axotomy using cDNA microarrays. J Neurochem. 2002, 82: 1129-1136.View ArticlePubMedGoogle Scholar
- Pruissen DM, Kappelle LJ, Rosendaal FR, Algra A: Genetic association studies in ischaemic stroke: replication failure and prospects. Cerebrovasc Dis. 2009, 27: 290-294. 10.1159/000199467.View ArticlePubMedGoogle Scholar
- Domingues-Montanari S, Fernández-Cadenas I, del Rio-Espinola A, Corbeto N, Krug T, Manso H, Gouveia L, Sobral J, Mendioroz M, Fernández-Morales J, Alvarez-Sabin J, Ribó M, Rubiera M, Obach V, Martí-Fàbregas J, Freijo M, Serena J, Ferro JM, Vicente AM, Oliveira SA, Montaner J: Association of a Genetic Variant in the ALOX5AP with Higher Risk of Ischemic Stroke: A Case-Control, Meta-Analysis and Functional Study. Cerebrovasc Dis. 2010, 29: 528-537. 10.1159/000302738.View ArticlePubMedGoogle Scholar
- Cardon LR, Palmer LJ: Population stratification and spurious allelic association. Lancet. 2003, 361: 598-604. 10.1016/S0140-6736(03)12520-2.View ArticlePubMedGoogle Scholar
- Lohmueller KE, Pearce CL, Pike M, Lander ES, Hirschhorn JN: Meta-analysis of genetic association studies supports a contribution of common variants to susceptibility to common disease. Nat Genet. 2003, 33: 177-182. 10.1038/ng1071.View ArticlePubMedGoogle Scholar
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.