New Pooled Cohort Risk Equations and Presence of Asymptomatic Brain Infarction

Transcription

1 New Pooled Cohort Risk Equations and Presence of Asymptomatic Brain Infarction Jong-Ho Park, MD, PhD; Jin Ho Park, MD, MPH, PhD; Bruce Ovbiagele, MD, MSc, MAS; Hyung-Min Kwon, MD, PhD; Jae-Sung Lim, MD, MSc; Jun Yup Kim, MD; BeLong Cho, MD, MPH, PhD; Jae Moon Yun, MD, MPH; Hyejin Lee, MD, MPH Background and Purpose The new pooled cohort risk (PCR) equations is sex- and race-specific estimates of the 10-year risk of atherosclerotic cardiovascular events among disease-free adults. Little is known about the association between the PCR model and presence of silent brain infarction (SBI). Methods We conducted a cross-sectional study of 1603 neurologically asymptomatic Korean people (mean age, 56.6±8.3; 838 men), who underwent brain MRI. We explored the association of PCR with SBI by race. SBI was divided into deep subcortical and hemispheric (hs-sbi). Results One-hundred seventy-five (10.9%) subjects had SBI. The PCR as white was independently related to the presence of SBI (odds ratio, 1.06; 95% confidence interval, ), multiple ( 2) SBIs (1.09; ), deep subcortical SBI (1.06; ), and hs-sbi (1.07; ). Compared with the lowest PCR category (<5%), dose response relationships were observed between increasing category (5% to <7.5%, 7.5% to <10%, and 10%) and the presence of SBI, respectively (1.85, ; 2.41, ; and 3.76, ), multiple SBIs (0.88, ; 8.44, ; and 8.47, ), deep subcortical SBI (1.92, ; 2.46, ; and 3.77, ), and hs-sbi (1.20, ; 5.59, ; and 5.96, ). C-statistic of PCR category for SBI was 0.63 ( ); multiple SBIs, 0.71 ( ); deep subcortical SBI, 0.62 ( ); and hs-sbi, 0.71 ( ). Calibration as black showed similar pattern to findings from white model. Conclusions Discrimination was fairly compatible with each race model. The PCR might serve as a simple clinical tool for identifying people at high risk for the untoward consequences of SBI, particularly multiple SBIs and hs-sbi. (Stroke. 2014;45: ) Key Words: cerebral infarction magnetic resonance imaging risk assessments Silent brain infarction (SBI), which is frequently seen on MRI, is common in older people without a clinical history of stroke. 1 Because SBI is frequently accompanied by white matter hyperintensities 2,3 and associated with an increased risk of cognitive decline 4 6 and clinical stroke, 7 timely identification and optimal implementation of strategies to mitigate future stroke and vascular cognitive impairment could be of great use to physicians. Advanced age and hypertension are the most widely accepted risk factors for SBI. 8 For Framingham risk score, increasing levels of the aggregate stroke risk score and cardiovascular risk score were associated with significantly related to prevalent risk of SBI (odds ratio, 1.27) 9 and cognitive decline, 10 respectively. However, studies on risk score for predicting SBI are few. Framingham 10-year prediction model provides a sex-specific absolute risk of vascular events, but with little validation in multiethnic populations. 11 To raise validation of the risk tool in an external population, the American College of Cardiology and the American Heart Association provided a guideline on the assessment of cardiovascular risk and issued new sex- and race-specific estimates of the 10-year risk for hard atherosclerotic cardiovascular disease (ASCVD) events for black and white men and women in 2013 as the American College of Cardiology/American Heart Association PCR equations. 12 This model has been validated to show a good discrimination of incident ASCVD risk in a population without ASCVD at baseline. 13 In this study, we evaluated the predictive ability and discriminative capacity of the PCR model for MRI-defined SBI among otherwise neurologically healthy individuals. Received August 1, 2014; final revision received September 30, 2014; accepted October 2, From the Department of Neurology, Myongji Hospital, Goyang, South Korea (J.-H.P.); Department of Family Medicine, Seoul National University Hospital, Seoul, South Korea (J.H.P., B.C., J.M.Y., H.L.); Department of Neurosciences, Medical University of South Carolina, Charleston (B.O.); and Department of Neurology, Seoul Metropolitan Government-Seoul National University Boramae Medical Center, Seoul, South Korea (H.-M.K., J.-S.L., J.Y.K.). Guest Editor for this article was Tatjana Rundek, MD, PhD. The online-only Data Supplement is available with this article at /-/DC1. Correspondence to Hyung-Min Kwon, MD, PhD, Department of Neurology, Seoul Metropolitan Government-Seoul National University Boramae Medical Center, 20 Boramae-ro 5-gil, Dongjak-gu, Seoul , South Korea. hmkwon@snu.ac.kr 2014 American Heart Association, Inc. Stroke is available at DOI: /STROKEAHA

2 3522 Stroke December 2014 Methods Study Population Because the PCR equations was originally designed for adults aged 40 to 79 years, 12 we reviewed a consecutive series of 1736 neurologically healthy consecutive subjects aged from 40 to 79 years who visited Seoul National University Hospital Healthcare System for routine health check-ups from January 2006 through December 2011 and who underwent brain MRI. Neurologically healthy subjects were defined as those who had not experienced a stroke or transient ischemic attack and had no neurological symptoms or signs. Of 1736 participants, 133 participants had missing smoking status or lipid component of PCR model and were excluded from the final analysis, yielding a total of 1603 (92.3%) subjects. All subjects provided informed consent, and the study was approved by the institutional review board at Seoul National University Hospital ( ). Baseline Data Collection Clinical information including age, sex, prior diagnosed comorbid conditions, smoking status and use of antihypertensive, antidiabetic and lipid-lowering medications were obtained and a physical examination was performed on each subject by a trained physician. The PCR score was calculated using the coefficients for the equations for calculating the estimate of an individual s 10-year ASCVD risk, 12,13 which is shown in Table I in the online-only Data Supplement. Because the PCR model is mainly applicable to black and non-hispanic whites, it is recommended that other ethnic groups use of the equations for non-hispanic whites. 12 Given that there are no data available concerning the PCR in Asian populations, we also calibrated the study participants as black to compare discrimination capacity between white and black models. Blood pressure was measured 2 times following a standardized protocol and averaged for analysis. All serum samples of participants were obtained in the fasting state for 12 hours. Diabetes mellitus was defined as a fasting glucose level of 126 mg/dl or a hemoglobin A1c 6.5%, 14 or self-report of a prior diagnosis of diabetes mellitus with current use of insulin or oral hypoglycemic medications. Definition of SBI SBI was defined as a round or ovoid ischemic lesion of diameter between 3 and 15 mm in diameter and well-demarcated hyperintensity on T2-weighted image and central hypointensity with surrounding hyperintensity on fluid-attenuated inversion recovery image. 15 To avoid any ambiguous enrollment of participants with SBI, we precluded ill-defined lesions, diffuse white matter changes and hemorrhagic lesions. SBIs were divided into deep subcortical (ds-sbi) and hemispheric SBI (hs-sbi) according to their lesion location because hs-sbi was more related to dementia than ds-sbi. 5 SBIs in the corona radiata, basal ganglia, internal capsule, thalamus, brain stem or cerebellum were categorized as ds-sbis, and those in the cortical gray matter or white matter adjacent to cortex were categorized as hs-sbis. 5 hs-sbi was included without limitation of maximum size. Participants having 2 SBI lesions were defined as multiple SBIs. MRI Protocol MRI examinations were performed at field strengths of 1.5 T (Signa, GE Healthcare, Milwaukee, WI or Magnetom SONATA, Siemens, Munich, Germany). The imaging protocol used consisted of: T2-weighted fast spin-echo (repetition time/echo time=5000/127 ms), T1-weighted spin-echo (repetition time/echo time=500/11 ms), and fluid-attenuated inversion recovery (repetition time/echo time=8800/127 ms; inversion time=2250 ms) imaging. Images were obtained as 26 transaxial slices per scan. The slice thickness was 5 mm, with 1-mm interslice gap. Statistics Data were summarized as mean±sd or number of subjects (percentage), as appropriate. Comparisons across the groups were examined using the χ 2 test for categorical variables and Student t test for continuous variables. Participants were categorized into 4 groups according to their 10-year predicted ASCVD risk: <5%, 5% to <7.5%, 7.5% to <10%, and 10%. 13 The PCR was also assessed as continuous variable. The relationships between PCR category and SBI, ds-sbi, or hs-sbi were evaluated using a χ 2 linear-by-linear association test in univariate analysis. Multivariable logistic regression analyses were performed to determine whether the PCR model was an independent predictor of SBI, multiple SBIs, ds-sbi, or hs-sbi after adjusting for potential confounders. The lowest risk level of PCR (<5%) was the referent group for purposes of comparison. In addition to the baseline variables having P<0.10 for the presence of SBI/ds-SBI/hs-SBI in univariate analyses (Table 1), antiplatelet use and lipid modifier use were selected for entry into the multivariable models. Results are given as odds ratios and 95% confidence intervals. A linear trend of adjusted odds ratios across a severity of PCR category was examined using a likelihood ratio test. Model fit was tested using modifications of methods by Lemeshow and Hosmer. 16 Above analyses were conducted using IBM SPSS Version 22.0 (IBM SPSS Inc, Chicago, IL). Accuracies of the PCR model was assessed by calculating c-statistics (areas under the receiver operating characteristic curves) and equations by each race were compared using MedCalc software version 5.0 (Mariakerke, Belgium). A P value of <0.05 was considered statistically significant. Results Total Subjects A total of 1603 subjects (mean age, 56.6±8.3 years; men, 52.3%) were included in this study from 1736 consecutive participants after excluding 133 subjects with missing values for PCR component: 132 with missing smoking and 1 with serum lipid value. Of the 1603 study populations, 175 subjects (10.9%) harbored SBIs (1 SBI in 123, 2 SBIs in 39, and 3 SBIs in 13 subjects). Twenty-five percent received antihypertensive medication, 7.8% lipid modifier, and 9.5% antiplatelet use at baseline visits. Demographics and clinical features of subjects with and without SBI are provided in Table 1. Compared with subjects without SBI, those with SBI were older, had higher levels of systolic blood pressure, waist circumference, glucose, and HbA1c as well as higher frequencies of treatment for high blood pressure and diabetes mellitus, whereas total cholesterol and low-density lipoprotein cholesterol levels were more likely to be lower. PCR Model and the Presence of SBI/ds-SBI/hs-SBI The PCR scores between the SBI(+) subjects and SBI( ) subjects were significantly different (all P<0.001 for white and black equations; Table 1). Among the 175 SBI(+) subjects, 152 (9.5%) had ds-sbi and 32 (2.0%) hs-sbi. The Figure shows a dose response relationship between the 4 PCR score category by white (<5%, 5% to <7.5%, 7.5% to <10%, and 10%) and SBI severity (1, 2, or 3) or prevalence of SBI, ds-sbi, and hs-sbi. Subjects with higher PCR category were more likely to have a higher numbers of SBI (Figure [A]; P<0.001) and have greater incidence of SBI, ds-sbi, or hs-sbi (Figure [B]; P<0.001), which was a similar pattern to findings seen in black model (see Figure I in the online-only Data Supplement). Findings of univariate analyses to test the association between PCR model and presence of SBI/ds-SBI/hs-SBI are provided in Table II in the online-only Data Supplement. The highest ( 10%) PCR category (versus lowest PCR [<5%]) was associated with an increased risk of presence of SBI, ds-sbi, and hs- SBI (odds ratio, 2.97; 95% confidence interval, ; 2.86, ; and 6.15, , respectively for white and

3 Park et al Vascular Risk Model and Silent Brain Infarction 3523 Table 1. Baseline Characteristics of Study Participants With and Without Silent Brain Infarction All (n=1603) Absent (n=1428) 3.10, ; 2.92, ; and 7.49, , respectively, for black; all P<0.01). PCR as a continuous variable was associated with the presence of SBI, ds-sbi, and hs-sbi (1.06, SBI Present (n=175)* P Value PCR components Age, y 56.6± ± ±8.4 <0.001 Male 838 (52.3) 743 (52.0) 95 (54.3) Ethnicity (white/black) 1603 (100) Total cholesterol, mg/dl 199.9± ± ± HDL-C, mg/dl 54.5± ± ± Systolic BP, mm Hg 128.2± ± ± Treatment for high BP 404 (25.2) 339 (23.7) 65 (37.1) <0.001 Diabetes mellitus 223 (13.9) 184 (12.9) 39 (22.3) Current smoker 263 (16.4) 239 (16.7) 24 (13.7) PCR score By white 8.1± ± ±11.7 <0.001 By black 9.5± ± ±11.7 <0.001 Body mass index, kg/m ± ± ± Waist circumference, cm 86.4± ± ± LDL-C, mg/dl 128.0± ± ± Triglycerides, mg/dl 122.5± ± ± Glucose, mg/dl 95.4± ± ± HbA1c, % 6.0± ± ± hs-crp, mg/dl 0.19± ± ± Creatinine, mg/dl 0.9± ± ± Alcohol use 765 (50.7) 692 (51.3) 73 (45.3) Antiplatelet use 153 (9.5) 131 (9.2) 22 (12.6) Lipid modifier use 125 (7.8) 107 (7.5) 18 (10.3) Values provided are number (%) or mean±sd, as appropriate, otherwise stated. BP indicates blood pressure; HbA1c, hemoglobin A1c; HDL-C, highdensity lipoprotein cholesterol; hs-crp, high sensitivity C-reactive protein; LDL-C, low-density lipoprotein cholesterol; PCR, pooled cohort risk; and SBI, silent brain infarction. *Of them, 152 (9.5%) had deep subcortical SBI and 32 (2.0%) hemispheric SBI ; ; 1.06, ; and 1.08, , respectively, for white; 1.06, ; 1.06, ; and 1.08, , respectively, for black; all P<0.001). Multivariable analyses for the presence of SBI and multiple SBIs by white is shown in Table 2 (see Table III in the online-only Data Supplement for black). Compared with the lowest PCR category (<5%), dose response relationships were observed between increasing PCR category (5% to <7.5%, 7.5% to <10%, and 10%) and presence of SBI, respectively (1.85, ; 2.41, ; and 3.76, for white and 1.96, ; 2.11, ; and 4.48, for black; all P<0.001 for a linear trend test). Multiple SBIs also had a dose response relationship with increasing PCR category (0.88, ; 8.44, ; and 8.47, for white and 1.30, ; 1.52, ; and 8.23, for black; all P<0.001 for a linear trend test). PCR as a continuous variable was significantly associated with the presence of SBI for white and black (all 1.06, ) and multiple SBIs (1.09, for white and 1.09, for black). Table 3 provides multivariable analyses for the presence of ds-sbi and hs-sbi by white (see Table IV in the online-only Data Supplement for black). Compared with the lowest PCR category, dose response relationships were observed between increasing PCR category and presence of ds-sbi (1.92, ; 2.46, ; and 3.77, for white and 2.12, ; 2.26, ; and 4.59, for black; all P<0.001 for a linear trend test) and hs-sbi (1.20, ; 5.59, ; and 5.96, for white and 0.97, ; 2.31, ; and 6.31, for black; P=0.003 and 0.004, respectively, for a linear trend test). PCR as a continuous variable was significantly associated with the presence of ds-sbi (1.06, for white and 1.07, for black) and hs-sbi (1.07, for white and 1.07, for black). Comparison of Model Discrimination by Race C-statistics of the PCR model with the presence of SBI, multiple SBIs, ds-sbi, and hs-sbi are given in Tables 2 and 3 for white and in Tables III and IV in the online-only Data Supplement for black. For PCR category, c-statistic for the presence of SBI was 0.63 ( ) for white and 0.63 ( ) for black; multiple SBIs, 0.71 ( ) for white and 0.68 ( ) Figure. Dose response relationship between distribution of pooled cohort risk (PCR) score by white model and frequency (%) of silent brain infarction (SBI) by increasing number (0 to 3; A) and frequency (%) of SBI, deep subcortical SBI and hemispheric SBI (B). Values are percentages of study subjects. All P<0.001 by linear-by-linear association tests.

4 3524 Stroke December 2014 Table 2. as White Multivariable Regression Analyses for the Presence of SBI and Multiple SBIs ( 2), Calibrated Presence of SBI Presence of Multiple SBIs ( 2) n (%) OR (95% CI) P Value n (%) OR (95% CI) P Value Model I* PCR category <5% 55 (7.1) 1 (Referent)... 9 (1.2) 1 (Referent)... 5% to <7.5% 20 (8.8) 1.85 ( ) (1.3) 0.88 ( ) % to <10% 16 (10.5) 2.41 ( ) (5.9) 8.44 ( ) % 84 (18.6) 3.76 ( ) < (6.9) 8.47 ( ) <0.001 C-statistic (95% CI) ( ) ( ) Sensitivity Specificity Model II* PCR (continuous variable) 175 (10.9) 1.06 ( ) < (3.2) 1.09 ( ) <0.001 C-statistic (95% CI) ( ) ( ) Sensitivity Specificity CI indicates confidence interval; HbA1c, hemoglobin A1c; OR, odds ratio; PCR, pooled cohort risk; and SBI, silent brain infarction. *Adjusted for waist circumference, low-density lipoprotein cholesterol, HbA1c, antiplatelet use, and lipid modifier use. All P<0.001 for higher prevalence of SBI and multiple SBIs ( 2) with a linear trend test of adjusted ORs across PCR category. Comparison of c-statistic with black. for black; ds-sbi, 0.62 ( ) for white and 0.63 ( ) for black; and hs-sbi, 0.71 ( ) for white and 0.71 ( ) for black. For PCR as a continuous variable, c-statistic for SBI was 0.66 ( ) for white and black; multiple SBIs, 0.74 ( ) for white and 0.72 ( ) for black; ds-sbi, 0.65 ( ) for white and black; and Table 3. hs-sbi, 0.75 ( ) for white and 0.74 ( ) for black. Discrimination for the presence of SBI including multiple SBIs, ds-sbi, and hs-sbi was not significantly different between white model and black model (Tables 2 and 3). Overall, the PCR seemed to have a higher sensitivity for multiple SBIs and hs-sbi than for SBI or ds-sbi for white or black. Multivariable Regression Analyses for the Presence of ds-sbi and hs-sbi, Calibrated as White Presence of ds-sbi Presence of hs-sbi n (%) OR (95% CI) P Value n (%) OR (95% CI) P Value Model I* PCR category <5% 49 (6.4) 1 (Referent)... 6 (0.8) 1 (Referent)... 5% to <7.5% 17 (7.6) 1.92 ( ) (1.4) 1.20 ( ) % to <10% 14 (9.3) 2.46 ( ) (2.8) 5.59 ( ) % 72 (16.4) 3.77 ( ) < (4.9) 5.96 ( ) C-statistic (95% CI) ( ) ( ) Sensitivity Specificity Model II* PCR (continuous variable) 152 (9.6) 1.06 ( ) < (2.2) 1.07 ( ) C-statistic (95% CI) ( ) ( ) Sensitivity Specificity CI indicates confidence interval; ds-sbi, deep subcortical silent brain infarction; HbA1c, hemoglobin A1c; hs-sbi, hemispheric silent brain infarction; OR, odds ratio; and PCR, pooled cohort risk. *Adjusted for waist circumference, low-density lipoprotein cholesterol, HbA1c, antiplatelet use, and lipid modifier use. P<0.001 and P=0.003 for higher prevalence of ds-sbi and hs-sbi, respectively, with a linear trend test of adjusted ORs across PCR category. Comparison of c-statistic with black.

5 Park et al Vascular Risk Model and Silent Brain Infarction 3525 Discussion In this analysis of neurologically healthy subjects aged from 40 to 79 years, higher PCR score was significantly associated with a greater prevalence of SBI. PCR as a continuous variable showed significantly increased risk of presence of SBI (1.06, ), which is a similar pattern to findings from the Framingham Offspring Study (1.27, ). 9 The risk of SBI increased monotonically with a higher PCR category, and it was significantly increased near 2.5-fold (2-fold for black) higher risk of SBI for the third category (7.5% to <10%) and about 4-fold (4.5-fold for black) for the highest one ( 10%). The risk of multiple SBIs was robustly increased >8- fold higher for the highest categories in white and black. The risk of ds-sbi and hs-sbi also increased substantially with a higher PCR category, and it was significantly increased 2.5- and 6-fold higher risk of ds-sbi and hs-sbi, respectively, for the third category and 4- and 6-fold risk of ds-sbi and hs- SBI, respectively, for the highest one (similar dose response pattern to findings from black). When compared between 2 race applications, c-statistics by the receiver operating characteristic curve were fairly compatible with white model and black model for the presence of SBI including multiple SBIs, ds-sbi, and hs-sbi. However, our findings need to be strictly proven through prospective study design. In this study, the 10-year PCR model did not show high discriminative capacity for the presence of SBI (including ds-sbi) given relatively low c-statistics (<0.70 for either race model), thereby diminishing its potential value. 17 This may not be too surprising because the PCR model was not primarily designed to reflect an index measure of contemporary ischemic burden, but rather to help guide the decision to initiate statin therapy for primary prevention in adults without clinical ASCVD or diabetes mellitus, and with low-density lipoprotein cholesterol levels between 70 and 189 mg/dl. 18 Another explanation for PCR s modest discriminative capacity in this study is that all the participants were Asians. Indeed, the PCR model may have overestimated SBI risk for East Asian ancestry such as Koreans. 12 However, discrimination of the PCR for the presence of multiple SBIs and hs-pbi (by SBI location) was better (all >0.70 for white and black) than the ds-sbi suggesting modest value. 17 Likewise, the PCR was more sensitive in identifying multiple SBIs and hs-sbi than SBI or ds-sbi in both race models. Multiple SBIs correlate advanced vascular disease pattern (versus single SBI). 19 ds-sbi patterns are predominantly lacunes and are mainly associated with small-vessel disease pathology, 8 while hs-sbi (territorial and pial artery mediated) is more prevalent in extracranial large artery disease. 20 These findings conform with recent studies indicating that the PCR was designed to predict hard ASCVD risk, 12 in which large artery atherosclerotic stroke should be considered as high-risk disease of further coronary events. 21 Because the prevalence of hs-sbi (versus ds-sbi) was low (2.0% versus 9.5%), our findings certainly need to be validated with imaging method through larger scale prospective design in future studies. Furthermore, Framingham cardiovascular stroke risk score was significantly linked to cognitive decline, 10 which was related to hs-sbi more than ds-sbi in a longitudinal autopsy study, 5 thus assessing the role of the PCR as a potential predictive tool for assessing risk of dementia and targeting of vascular risk factors is also warranted. This study has limitations. We explored the relationship between PCR model and SBI including multiple SBIs, ds- SBI, and hs-sbi with a cross-sectional design in a retrospective manner, thus our results cannot confirm a predictive relationship between them. Although SBI and white matter hyperintensities are frequently observed and are the main MRI-defined small-vessel disease, 2,3 we could not adjust for white matter hyperintensities. Thus, our results simply suggest that the PCR could be an indicator of having multiple SBIs or hs-sbi, not mean to provide as a therapeutic/preventive direction for dementia or stroke prevention. Vascular status through angiographic imaging was not available in this database, otherwise could have yielded more concrete results about hs-sbi. Study subjects were neurologically healthy people but information on history of cardiac disease including coronary heart disease, heart failure, or atrial fibrillation was not obtained thereby there being a possibility of unmeasured confounding. Finally, we cannot exclude the possibility of selection bias because of hospital-based setting and age restriction between 40 and 79 years, but our participants were recruited as a large sample size from general populations who visited for routine health check-ups. Furthermore, the prevalence of SBI in our study was 10.9%, a finding that is in accord with that (10.7%) seen in the Framingham Offspring Study. 9 In conclusion, discrimination of the PCR was fairly compatible with white and black for the presence of SBI, including multiple SBIs, ds-sbi, and hs-sbi among stroke free healthy Korean (or Asian) populations. The new PCR model modestly discriminates presence of SBI, but nonetheless might serve as a simple clinical tool for identifying high-risk subjects predisposed to having SBI, particularly multiple SBIs and hs-sbi. None. Disclosures References 1. Price TR, Manolio TA, Kronmal RA, Kittner SJ, Yue NC, Robbins J, et al. Silent brain infarction on MRI and neurological abnormalities in community-dwelling older adults. The Cardiovascular Health Study. CHS Collaborative Research Group. Stroke. 1997;28: Longstreth WT Jr, Bernick C, Manolio TA, Bryan N, Jungreis CA, Price TR. Lacunar infarcts defined by MRI of 3660 elderly people: the Cardiovascular Health Study. Arch Neurol. 1998;55: de Leeuw FE, de Groot JC, Achten E, Oudkerk M, Ramos LM, Heijboer R, et al. Prevalence of cerebral white matter lesions in elderly people: a population based MRI study. The Rotterdam Scan Study. J Neurol Neurosurg Psychiatry. 2001;70: Vermeer SE, Prins ND, den Heijer T, Hofman A, Koudstaal PJ, Breteler MM. Silent brain infarcts and the risk of dementia and cognitive decline. N Engl J Med. 2003;348: Troncoso JC, Zonderman AB, Resnick SM, Crain B, Pletnikova O, O Brien RJ. Effect of infarcts on dementia in the Baltimore longitudinal study of aging. Ann Neurol. 2008;64: Blum S, Luchsinger JA, Manly JJ, Schupf N, Stern Y, Brown TR, et al. Memory after silent stroke: hippocampus and infarcts both matter. Neurology. 2012;78: Kobayashi S, Okada K, Koide H, Bokura H, Yamaguchi S. Subcortical silent brain infarction as a risk factor for clinical stroke. Stroke. 1997;28:

6 3526 Stroke December Vermeer SE, Longstreth WT Jr, Koudstaal PJ. Silent brain infarcts: a systematic review. Lancet Neurol. 2007;6: Das RR, Seshadri S, Beiser AS, Kelly-Hayes M, Au R, Himali JJ, et al. Prevalence and correlates of silent cerebral infarcts in the Framingham offspring study. Stroke. 2008;39: Kaffashian S, Dugravot A, Elbaz A, Shipley MJ, Sabia S, Kivimäki M, et al. Predicting cognitive decline: a dementia risk score versus the Framingham vascular risk scores. Neurology. 2013;80: D Agostino RB Sr, Vasan RS, Pencina MJ, Wolf PA, Cobain M, Massaro JM, et al. General cardiovascular risk profile for use in primary care: the Framingham Heart Study. Circulation. 2008;117: Goff DC Jr, Lloyd-Jones DM, Bennett G, Coady S, D Agostino RB, Gibbons R, et al ACC/AHA Guideline on the Assessment of Cardiovascular Risk: A Report of the American College of Cardiology/ American Heart Association Task Force on Practice Guidelines. Circulation. 2014;129(25 suppl 2):S49 S Muntner P, Colantonio LD, Cushman M, Goff DC Jr, Howard G, Howard VJ, et al. Validation of the atherosclerotic cardiovascular disease Pooled Cohort risk equations. JAMA. 2014;311: American Diabetes Association. Standards of medical care in diabetes Diabetes Care. 2013;36(suppl 1):S11 S Wardlaw JM, Smith EE, Biessels GJ, Cordonnier C, Fazekas F, Frayne R, et al; STandards for ReportIng Vascular changes on neuroimaging (STRIVE v1). Neuroimaging standards for research into small vessel disease and its contribution to ageing and neurodegeneration. Lancet Neurol. 2013;12: Lemeshow S, Hosmer DW Jr. A review of goodness of fit statistics for use in the development of logistic regression models. Am J Epidemiol. 1982;115: Ohman EM, Granger CB, Harrington RA, Lee KL. Risk stratification and therapeutic decision making in acute coronary syndromes. JAMA. 2000;284: Stone NJ, Robinson JG, Lichtenstein AH, Bairey Merz CN, Blum CB, Eckel RH, et al; American College of Cardiology/American Heart Association Task Force on Practice Guidelines ACC/AHA guideline on the treatment of blood cholesterol to reduce atherosclerotic cardiovascular risk in adults: a report of the American College of Cardiology/ American Heart Association Task Force on Practice Guidelines. J Am Coll Cardiol. 2014;63(25 pt B): Park JH, Kwon HM, Lee J, Kim DS, Ovbiagele B. Association of intracranial atherosclerotic stenosis with severity of white matter hyperintensities [published online ahead of print April 9, 2014]. Eur J Neurol won+hm%2c+lee+j%2c+kim+ds%2c+ovbiagele+b.+associati on+of+intracranial+atherosclerotic+stenosis+with+severity+of+white +matter+hyperintensities. Accessed April 15, Lee DK, Kim JS, Kwon SU, Yoo SH, Kang DW. Lesion patterns and stroke mechanism in atherosclerotic middle cerebral artery disease: early diffusion-weighted imaging study. Stroke. 2005;36: Lackland DT, Elkind MS, D Agostino R Sr, Dhamoon MS, Goff DC Jr, Higashida RT, et al; American Heart Association Stroke Council; Council on Epidemiology and Prevention; Council on Cardiovascular Radiology and Intervention; Council on Cardiovascular Nursing; Council on Peripheral Vascular Disease; Council on Quality of Care and Outcomes Research. Inclusion of stroke in cardiovascular risk prediction instruments: a statement for healthcare professionals from the American Heart Association/American Stroke Association. Stroke. 2012;43: