- Research Article
- Open Access
Serum Cholinesterase Activities Distinguish between Stroke Patients and Controls and Predict 12-Month Mortality
- Einor Ben Assayag†1,
- Shani Shenhar-Tsarfaty†1, 3,
- Keren Ofek†2,
- Lilach Soreq†4,
- Irena Bova1,
- Ludmila Shopin1,
- Ronan M. G. Berg5,
- Shlomo Berliner1, 3,
- Itzhak Shapira1, 3,
- Natan M. Bornstein†1, 3 and
- Hermona Soreq†2Email author
© The Feinstein Institute for Medical Research 2010
- Received: 10 February 2010
- Accepted: 13 March 2010
- Published: 14 April 2010
To date there is no diagnostic biomarker for mild stroke, although elevation of inflammatory biomarkers has been reported at early stages. Previous studies implicated acetylcholinesterase (AChE) involvement in stroke, and circulating AChE activity reflects inflammatory response, since acetylcholine suppresses inflammation. Therefore, carriers of polymorphisms that modify cholinergic activity should be particularly susceptible to inflammatory damage. Our study sought diagnostic values of AChE and Cholinergic Status (CS, the total capacity for acetylcholine hydrolysis) in suspected stroke patients. For this purpose, serum cholinesterase activities, butyrylcholinesterase-K genotype and inflammatory biomarkers were determined in 264 ischemic stroke patients and matched controls during the acute phase. AChE activities were lower (P < 0.001), and butyrylcholinesterase activities were higher in patients than in controls (P = 0.004). When normalized to sampling time from stroke occurrence, both cholinergic parameters were correlated with multiple inflammatory biomarkers, including fibrinogen, interleukin-6 and C-reactive protein (r = 0.713, r = 0.607; r = 0.421, r = 0.341; r = 0.276, r = 0.255; respectively; all Pvalues < 0.001). Furthermore, very low AChE activities predicted subsequent nonsurvival (P = 0.036). Also, carriers of the unstable butyrylcholinesterase-K variant were more abundant among patients than controls, and showed reduced activity (P < 0.001). Importantly, a cholinergic score combining the two cholinesterase activities discriminated between 94.3% matched pairs of patients and controls, compared with only 75% for inflammatory measures. Our findings present the power of circulation cholinesterase measurements as useful early diagnostic tools for the occurrence of stroke. Importantly, these were considerably more distinctive than the inflammatory biomarkers, albeit closely associated with them, which may open new venues for stroke diagnosis and treatment.
Stroke cases span 780,000 patients annually, accounting for about 1 of every 17 deaths in the United States, with many of the surviving patients displaying severe disabilities (1). Unlike acute coronary syndromes and congestive heart failure with the myocardial isoform of creatine phosphokinase, troponin and B-type natriuretic peptide, there is no single biochemical diagnostic marker available for ischemic stroke that is sufficiently sensitive and specific (2).
Several markers, including basal National Institutes of Health Stroke Scale (NIHSS), infarct volume, early neurological deterioration, immuno-inflammatory profile, growth factors and stem and progenitor cells (2,3) have been explored in efforts to diagnose stroke patients and predict postevent survival. Nevertheless, none of these markers offer a gold standard parameter, especially for the milder stroke cases. One possible link involves the increased poststroke production of proinflammatory cytokines (4,5). Inflammatory markers correlate with brain infarction volume, stroke severity and clinical outcome (6–8), but the underlying neurotransmission pathway(s) and the putative bridges between inflammatory reactions and clinical outcome await further validations.
Recent studies demonstrate causal links between inflammatory pathways and cholinergic signaling (9,10). Specifically, the so-called cholinergic antiinflammatory pathway inhibits cytokine synthesis and release through activation of the ± 7 nicotinic acetylcholine (ACh) receptor (11–13). In cortical cells and hippocampal slice cultures, such activation prevents cell damage and cell death during experimental ischemia (14). Correspondingly, acetylcholinesterase (AChE) inhibition attenuates cerebral infarction volume in rats during experimental stroke (15).
Recently, we demonstrated tight interrelationships between inflammatory reactions and the total circulating capacity for ACh hydrolysis (10,12). This suggested that stress-induced increases in this hydrolytic capacity (16) can accentuate the inflammatory response to stroke by counteracting the cholinergic antiinflammatory reaction. Specifically, the cholinergic antiinflammatory pathway may be mitigated by AChE and the closely related enzyme butyrylcholinesterase (BChE), both of which hydrolyze and inactivate ACh and are modulated under inflammatory insults. BChE is the major ACh hydrolyzing enzyme in the circulatory system (17). In addition to the “usual” (BChE-U) form, the Kalow variant (BChE-K) exists in allelic frequencies of 0.13–0.21 (18). BChE-K includes a single nucleotide polymorphism at position 1699 (SNP ID: rs1803274; Alleles: A/G). This leads to an alanine-to-threonine substitution at position 539, 36 residues upstream to the C-terminus of BChE. BChE-K hydrolyzes butyrylthiocholine and acetylthiocholine 30% and 20% less efficiently than BChE-U, respectively. BChE-K carriers would likely sustain improved cholinergic control over inflammation compared with noncarriers; therefore, we hypothesized that BChE-K carriers should be relatively protected from the consequences of ischemic stroke if the poststroke inflammatory reaction is solely damaging, but should inversely be more vulnerable to the occurrence of stroke and to the consequent damages if this inflammatory reaction also entails a certain neuroprotective value.
Elevated inflammatory cytokines can by themselves cause neurological decline (8); consequently, we hypothesized that serum AChE activity would reflect the intensity of the neuroinflammatory response in stroke patients, that the BChE-K variant could affect the risk to experience stroke and that both enzyme activities would predict the risk of poststroke mortality more distinctively than inflammatory markers. To challenge our working hypothesis, we searched for cholinergic predictors of the risk for poststroke survival.
Stroke patients. For this study, we specifically recruited patients with mild to moderate ischemic stroke or TIA, excluding severe strokes and patients who were unable to provide an independent informed consent. The proportion of lacunar strokes among all hospitals in Israel in 2007 was 33% of all ischemic strokes, with much higher prevalence among mild strokes (data from the National Acute Stroke Israeli Survey, NASIS 2007, Ministry of Health, Israel, http://www.israel-neurology.co.il/nasis1207.asp). All participants signed an informed consent when entering the study; therefore, patients suffering from global aphasia were automatically excluded, as were patients who were not interested in participating. All other stroke patients admitted to the Medical Center who met the inclusion criteria were enrolled.
Two hundred and sixty-four patients over 18 years of age with a documented mild to moderate acute ischemic stroke or TIA (neurological deficit lasting less than 24 h) and with suspected symptom onset within the preceding 24 h, were enrolled into the study: 139 with lacunar stroke, 56 with large-artery atherosclerotic stroke, 15 with cardioembolic stroke, 4 with stroke of undetermined etiology, 1 with a stroke of another determined etiology, and 49 with TIA. All patients enrolled were admitted to the Department of Emergency Medicine at the Tel-Aviv Sourasky Medical Center between July 2000 to July 2002 and January 2004 to January 2006. Exclusion criteria included patients with stroke resulting from trauma or an invasive procedure, severe stroke, cerebral hemorrhage, history of malignant tumor, chronic inflammatory disease, autoimmune disease, coagulation disorders, signs and symptoms of acute chronic infection or treatment with antiinflammatory agents, except for acetylsalicylic acid (≤325 mg/day) or inability to provide informed consent. None of the included patients received thrombolytic therapy. A signed informed consent was obtained from all patients enrolled.
Stroke subtyping followed the TOAST classification (19). Survival records for all patients were obtained at the Ministry of Health database. The patients’ NIHSS neurological state was assessed by a certified senior vascular neurologist. Table 1 presents the characteristics of patients and matched controls.
Controls. To each patient we matched one neurologically intact control by gender, age (±3 y) and body mass index (BMI). Controls were in routine followup in the various outpatient clinics between September 2000 and September 2004. Exclusion criteria included a history of cerebral or cardiac event during the previous 12 months, known inflammatory diseases, history of acute febrile disease or infection during the previous 3 months, known malignancy, pregnancy, steroidal or nonsteroidal treatment (except for acetylsalicylic acid at a dose of δ325 mg/dL), and invasive procedures (surgery, catheterization, and so on) during the previous 6 months. A written informed consent approved by the local Ethics Committee was obtained from all participants.
Measurements of Serum Cholinesterase Catalytic Activities
AChE and BChE activities, BChE-K or BChE-U genotype and the total serum capacity to hydrolyze acetylcholine were determined retrospectively in frozen serum samples detailed previously (12). All laboratory methods were performed by a blinded technician or student, each method by the same person for all measures. In an attempt to minimize intratest variations, exclude day-of-measurement effects and ascertain the authenticity of the observed differences, we used the same reagents throughout the series and remeasured selected samples each test day.
Statistical analyses were conducted under the guidance of David M Steinberg (Tel Aviv). Collected data was summarized and displayed as mean ± standard deviation (SD) using SPSS/WIN (version 13.0, SPSS INC, Chicago, IL, USA). The different biomarkers for patients and controls were compared by Student t test for normally distributed variables and the Mann-Whitney U test for non-normally distributed variables. To assess associations among categorical variables, we used the χ2 test. AChE activity, CS and inflammatory biomarker values all were normalized to sampling time from stroke occurrence, and correlations between them determined by the two-tailed Spearman rank test. Linear discrimination analysis (LDA) adapted to paired data was used for discriminating stroke patients from controls based on their cholinergic and inflammatory parameters. The controls were each matched to individual patients by gender, age (±3 y) and BMI. For each data pair, differences were calculated for each variable, then inflammatory scores were derived by applying discriminate analysis (20,21) to the resulting differences. To conduct survival analyses, we used Partek Genomics Suite version 6.3β (6.06.1128) for multifactorial one-way analysis of variance (ANOVA), as instructed. To investigate the association between calculated BChE-K activity and the BChE K genotype, we used multivariate stepwise linear regression. A P value of <0.05 was considered statistically significant.
Venous blood was obtained from all stroke/TIA patients within 24 h from hospital admission (Time 1, average 26.5 + 15.9 h from symptoms onset). The biomarkers tested included the erythrocyte sedimentation rate (ESR) using Westergreen’s method (22), white blood cell count (WBCC) determined by the Coulter STKS electronic analyzer (Beckman Coulter, Nyon, Swiss) and concentrations of high sensitive C-reactive protein (hs-CRP) measured in the Behring BN II Nephelometer (DADE Behring, Marburg, Germany) (23). Plasma high sensitive IL-6 (hs-IL-6) levels were determined using a commercial enzyme-linked immunosorbent assay (ELISA) kit (R&D Systems, Minneapolis, MN, USA). The Clauss’ method (24) and a Sysmex 6000 analyzer (Sysmex Corporation, Hyaga, Japan) served to quantify fibrinogen.
All supplementary materials are available online at molmed.org .
Population characteristics and cholinergic parameters of stroke patients and matched controls.
Stroke patients at admission
Age, y (SD)
BMI, kg/m2 (SD)
Gender, males (%)
Current smokers (%)
hs-CRP, mg/L (SD)
ESR, mm/H (SD)
Fibrinogen, g/dL (SD)
WBCC, 103/µL (SD)
hs-IL-6, pg/mL (SD)
Acetylsalicylic acid, %
β Blockers, %
Angiotensin II Blockers, %
ACE Inhibitors, %
HMG-CoA reductase inhibitors, %
NIHSS, mean (SD)
mRs, mean (SD)
Cholinergic Status total activity, nmol substrate hydrolysed/min*ml (SD)
AChE activity, nmol substrate hydrolysed/min*ml (SD)
Cholinergic Parameters Discriminate between Stroke Patients and Matched Controls
Physiological parameters relevant for poststroke recovery should discriminate between patients and controls. Linear discriminate analysis demonstrated that the inflammatory score includes the following weights: 0.015 on ESR, 0.201 on the base 10 logarithm of hs-CRP, 0.176 on WBCC and 0.000598 on fibrinogen. Computing this score for each subject and subtracting the stroke patient score from that of a matched control yields a “delta” score for each pair. Figure 1A presents the inflammation score for the paired stroke patients against that for the matched controls.
Predicting a significant contribution of BChE to the outcome of stroke, we further used the paired linear discriminate analysis to compute a new cholinergic score that includes the following weights: −0.0116 on AChE and 0.00301 on Cholinergic Status (12) (Figure 1B). The average score for this new measure was −2.75 with a standard deviation of 1.66, whereas the average score for the inflammatory measures was −0.50 with SD of 0.70. Correspondingly, the new cholinergic score yielded a sharper discrimination with a negative score for 94.3% of the pairs, as compared with only 75% for the inflammatory measures. Thus the post-stroke modulations in cholinesterase activities, both the decrease in AChE and the increase in BChE activity, emerged as valuable biomarkers of stroke. Next, we subjected our findings to a blinded independent validation test by employing the cholinergic status and AChE activity values to blindly assign each subject to a specific group with the highest conditional probability using linear discriminative analysis. This yielded discrimination with specificity of 97% and sensitivity of 80% for the employed tests (Figure 1C).
Total Cholinesterase Activities as Predictors of 12-Month Survival
Upon hospital admission, all stroke patients exhibited lower AChE activity levels than controls. Twenty-one patients died within the study period. To explore differences between survivors and non-survivors, we performed a multifactorial one-way ANOVA analysis that included cardiovascular risk factors, inflammatory biomarkers and the cholinergic parameters. Total cholinesterase activities, age, hyperlipidemia, and smoking emerged as the only significant parameters (P < 0.05) to distinguish between stroke survivors and nonsurvivors. Among these variables, age and hyperlipidemia were the most prominent factors (P < 0.01). Of note, 12-month nonsurvivors also presented lower calculated BChE activity during the acute phase of stroke compared with 12-month survivors (1271.6 ± 343 versus 1426.5 ± 315.4, P = 0.031, Figure 1D). When matched controls were included in the analysis, AChE activity was the most significant parameter to distinguish between groups (P = 5.5*10–10) with hs-CRP, WBCC and diabetes history following far behind (P = 0.012, P = 0.014 and P = 0.029, respectively). Next, we explored the possible tissue and cell type origins of AChE. Affymetrix exon array and GeneCard microarray tests demonstrated that the AChE gene, known to be primarily expressed in nerve and muscle also is expressed in high levels in nucleated blood cells (Figure 1E), suggesting that serum AChE levels reflect both central and peripheral contributions.
Cholinesterase Associations with the Inflammatory Response
The BChE K Variant Associates with Higher Stroke Risks
Multivariate stepwise linear regression model for calculated BChE activity
Reduction in serum AChE activity of the stroke patients likely reflects stroke-induced cholinergic hyperexcitation, which potentially can serve as protection from uncontrolled inflammatory reactions (10). However, compatible with the potentially lethal condition defined as cholinergic crisis (26), patients with lowest AChE values included most of the nonsurvivors. These findings suggest that mild stroke-induced decreases in AChE activity associate with mitigated inflammatory reactions. Indeed, moderate AChE inhibition improves recovery from cerebral infarction in rats with experimental stroke (15). Future studies will be required to explore the possible involvement in this decline of the AChE mRNA-targeted microRNA-132 (10).
Stroke patients show anxiety feelings (27), and the inverse association between circulating AChE activity and anxiety scores (25) highlight AChE activity as a putative link also following stroke. At the translational research level, these findings indicate that diagnostic measurements of serum AChE activity in patients potentially could be of value for improving the recovery prospects of stroke patients with low AChE levels.
BChE activity increased in poststroke patients, and patients with the unstable, less active BChE-K variant were somewhat more prevalent in stroke patients, suggesting that higher BChE activities reflected a better recovery process. BChE-K increases the risk for early onset cardiac disease (28), suggesting that other, rare BChE mutations with yet lower hydrolytic activities (for example, “atypical” D70G BChE ; or null BChE mutations ) may entail yet poorer recovery. Thanks to the slower hydrolytic pace of BChE compared with AChE, this increase can protect from the lethal consequences of cholinergic crisis due to AChE decline while preventing the debilitating outcome of failed cholinergic neurotransmission.
When studied alone, average values of specific inflammatory biomarkers showed statistically significant differences between stroke patients and carefully matched controls. However, no such differences were observed when the various inflammatory biomarkers of each subject were grouped together. This implies that the composite inflammatory status of stroke patients is both complex and variable in nature (30). In contrast, the cholinergic biomarkers provide clear segregation between stroke patients and controls, possibly due to their limited number.
Several limitations of our study need to be addressed. First, our current findings are limited to patients with mild to moderate ischemic stroke or TIA. This introduced a certain selection bias, although every effort was made to include all of the patients admitted to the Medical Center within the recruitment period. Second, the inflammatory biomarkers selected were few, and we did not perform serial measurements of inflammatory or cholinergic biomarkers and define the kinetics of changes in these values. In our current study we measured the cholinergic parameters within the same time points as the inflammatory markers. Since both are interconnected and regulate each other, our findings emphasize the importance of serial measurements of both markers simultaneously, while normalizing the data to sampling time from onset of stroke symptoms.
While considering the inclusion bias and differences in baseline parameters between stroke and control volunteers, our findings are compatible with the hypothesis that cholinergic modulations monitor the individual prospects of survival from ischemic stroke. In-depth understanding of the association between cholinergic hyperarousal and inflammatory reactions, and delineating in experimental animals their causal effects on the neurological state in acute stroke may foreshadow previously nonperceived strategies for early stroke diagnosis and treatment. Further studies also are needed to validate these measurements as a diagnostic tool, and extensive genetic analysis studies should be performed to test the possibility that BChE-K is a risk factor for stroke.
The authors declare that they have no competing interests as defined by Molecular Medicine, or other interests that might be perceived to influence the results and discussion reported in this paper.
The professional and dedicated statistical consultancy and advice of David M Steinberg, Department of Statistics and Operations Research at Tel Aviv University’s School of Mathematical Sciences are greatly appreciated. This study was supported by The Israel Science Foundation Morasha Grant #1876/08, and The German Israel Project, DIP-G 3.2 (to H Soreq). This work was performed in partial fulfillment of the requirements for a PhD degree of S Shenhar-Tsarfaty, Sackler Faculty of Medicine, Tel Aviv University.
- Rosamond W, et al. (2007) Heart disease and stroke statistics—2007 update: a report from the American Heart Association Statistics Committee and Stroke Statistics Subcommittee. Circulation. 115: e69–171.View ArticleGoogle Scholar
- Whiteley W, Tseng MC, Sandercock P. (2008) Blood biomarkers in the diagnosis of ischemic stroke: a systematic review. Stroke. 39:2902–9.View ArticleGoogle Scholar
- McColl BW, Allan SM, Rothwell NJ. (2009) Systemic infection, inflammation and acute ischemic stroke. Neuroscience. 158:1049–61.View ArticleGoogle Scholar
- Beamer NB, Coull BM, Clark WM, Hazel JS, Silberger JR. (1995) Interleukin-6 and interleukin-1 receptor antagonist in acute stroke. Ann. Neurol. 37:800–5.View ArticleGoogle Scholar
- Fassbender K, et al. (1994) Proinflammatory cytokines in serum of patients with acute cerebral ischemia: kinetics of secretion and relation to the extent of brain damage and outcome of disease. J. Neurol. Sci. 122:135–9.View ArticleGoogle Scholar
- Di Napoli M, Papa F. (2002) Inflammation, hemostatic markers, and antithrombotic agents in relation to long-term risk of new cardiovascular events in first-ever ischemic stroke patients. Stroke. 33:1763–71.View ArticleGoogle Scholar
- Elkind MS, Cheng J, Rundek T, Boden-Albala B, Sacco RL. (2004) Leukocyte count predicts outcome after ischemic stroke: the Northern Manhattan Stroke Study. J. Stroke Cerebrovasc. Dis. 13:220–7.View ArticleGoogle Scholar
- Vila N, Castillo J, Davalos A, Chamorro A. (2000) Proinflammatory cytokines and early neurological worsening in ischemic stroke. Stroke. 31:2325–9.View ArticleGoogle Scholar
- Metz CN, Tracey KJ. (2005) It takes nerve to dampen inflammation. Nat. Immunol. 6:756–7.View ArticleGoogle Scholar
- Shaked I, et al. (2009). MicroRNA-132 potentiates cholinergic anti-inflammatory signaling by targeting acetylcholinesterase. Immunity. 31:965–73 cover. Accompanied by preview O’Neil, L.A.J. (2009) Boosting the brain’s ability to block inflammation via microRNA-132. Immunity. 31:854–5.View ArticleGoogle Scholar
- Borovikova LV, et al. (2000) Vagus nerve stimulation attenuates the systemic inflammatory response to endotoxin. Nature. 405:458–62.View ArticleGoogle Scholar
- Ofek K, et al. (2007) Cholinergic status modulations in human volunteers under acute inflammation. J. Mol. Med. 85:1239–51.View ArticleGoogle Scholar
- Wang H, et al. (2003) Nicotinic acetylcholine receptor alpha7 subunit is an essential regulator of inflammation. Nature. 421:384–8.View ArticleGoogle Scholar
- Hejmadi MV, Dajas-Bailador F, Barns SM, Jones B, Wonnacott S. (2003) Neuroprotection by nicotine against hypoxia-induced apoptosis in cortical cultures involves activation of multiple nicotinic acetylcholine receptor subtypes. Mol. Cell. Neurosci. 24:779–86.View ArticleGoogle Scholar
- Fujiki M, Kobayashi H, Uchida S, Inoue R, Ishii K. (2005) Neuroprotective effect of donepezil, a nicotinic acetylcholine-receptor activator, on cerebral infarction in rats. Brain Res. 1043:236–41.View ArticleGoogle Scholar
- Kaufer D, Friedman A, Seidman S, Soreq H. (1998) Acute stress facilitates long-lasting changes in cholinergic gene expression. Nature. 393:373–7.View ArticleGoogle Scholar
- Loewenstein-Lichtenstein Y, et al. (1995) Genetic predisposition to adverse consequences of anti-cholinesterases in ‘atypical’ BCHE carriers. Nat. Med. 1:1082–5.View ArticleGoogle Scholar
- Bartels CF, et al. (1992) DNA mutation associated with the human butyrylcholinesterase K-variant and its linkage to the atypical variant mutation and other polymorphic sites. Am. J. Hum. Genet. 50:1086–103.PubMedPubMed CentralGoogle Scholar
- Adams HP Jr, et al. (1993) 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. 24:35–41.View ArticleGoogle Scholar
- Abdi H. (2007) Discriminant correspondence analysis. In: Encyclopedia of Measurement and Statistics. Salkind NJ (ed.), Rasmussen K (managing ed.) SAGE Publications, Thousand Oaks, CA, pp. 270–5.Google Scholar
- McLachlan GJ. (2004) Discriminant Analysis and Statistical Pattern Recognition. Hoboken (NJ): Wiley-Interscience. 526 pp.Google Scholar
- Haematology ICfSi. (1977) Recommendation of measurement of erythrocyte sedimentation rate of human blood. Am. J. Clin. Pathol. 68:505–7.View ArticleGoogle Scholar
- Rifai N, Tracy RP, Ridker PM. (1999) Clinical efficacy of an automated high-sensitivity C-reactive protein assay. Clin. Chem. 45:2136–41.PubMedGoogle Scholar
- Clauss A. (1957) Rapid physiological coagulation method in determination of fibrinogen [in German]. Acta. Haematol. 17:237–46.View ArticleGoogle Scholar
- Sklan EH, et al. (2004) Acetylcholinesterase/paraoxonase genotype and expression predict anxiety scores in Health, Risk Factors, Exercise Training, and Genetics study. Proc. Natl. Acad. Sci. U. S. A. 101:5512–7.View ArticleGoogle Scholar
- Hetherington KA, Losek JD. (2005) Myasthenia gravis: myasthenia vs. cholinergic crisis. Pediatr. Emerg. Care. 21:546–8; quiz 549-51. Review.View ArticleGoogle Scholar
- Barker-Collo SL. (2007) Depression and anxiety 3 months post stroke: prevalence and correlates. Arch. Clin. Neuropsychol. 22:519–31.View ArticleGoogle Scholar
- Nassar BA, et al. (2002) Relation between butyrylcholinesterase K variant, paraoxonase 1 (PON1) Q and R and apolipoprotein E epsilon 4 genes in early-onset coronary artery disease. Clin. Biochem. 35:205–9.View ArticleGoogle Scholar
- Duysen EG, Li B, Lockridge O. (2009) The butyrylcholinesterase knockout mouse a research tool in the study of drug sensitivity, bio-distribution, obesity and Alzheimer’s disease. Expert Opin. Drug Metab. Toxicol. 5:523–8.View ArticleGoogle Scholar
- Jerrard-Dunne P, et al. (2004) Inflammatory gene load is associated with enhanced inflammation and early carotid atherosclerosis in smokers. Stroke. 35:2438–43.View ArticleGoogle Scholar