Clinical Biochemistry 46 (2013) 716–721

Contents lists available at SciVerse ScienceDirect

Clinical Biochemistry

j ourna l homepage: www.e lsev ie r .com/ locate /c l inb iochem

Lipid profile components and incident cerebrovascular events versuscoronary heart disease; the result of 9 years follow-up in Tehran Lipidand Glucose Study

Maryam Tohidi, Reza Mohebi, Leila Cheraghi, Farhad Hajsheikholeslami, Saeed Aref, Sara Nouri,Fereidoun Azizi, Farzad Hadaegh ⁎Prevention of Metabolic Disorders Research Center, Research Institute for Endocrine Sciences, Shahid Beheshti University of Medical Sciences, Tehran, Iran

⁎ Corresponding author at: P.O. Box 19395-4763, TehraE-mail address: fzhadaegh@endocrine.ac.ir (F. Hada

0009-9120/$ – see front matter © 2013 The Canadian Shttp://dx.doi.org/10.1016/j.clinbiochem.2013.03.012

a b s t r a c t

a r t i c l e i n f o

Article history:

Received 1 November 2012Received in revised form 5 March 2013Accepted 13 March 2013Available online 24 March 2013

Keywords:StrokeCoronary heart diseaseLipidRisk factor

Objective: To assess the effects of lipid component total cholesterol (TC), triglycerides (TG), low densitylipoprotein cholesterol (LDL-C), high density lipoprotein cholesterol (HDL-C) and non-HDL-C on risk ofstroke events versus coronary heart disease (CHD).

Methods: The study was conducted on 2620 Iranians, aged ≥ 50 years, free from cardiovascular events atbaseline (1999–2001). The multivariable hazard ratios (HRs) for stroke/CHD were calculated for 1 mmol/Lchange in lipid components, using Cox proportional hazard regression.

Results: During 9.1 years of follow-up, 73 and 358 cases of stroke and CHD occurred. We found significantinteractions between TC and non-HDL-C with gender in risk prediction of stroke. Among women, multivariateadjusted HRs of ischemic stroke were 1.40 (1.08–1.82), 1.66 (0.71–3.86), 2.27 (0.58–8.91), 1.51 (1.06–2.15)and 1.36 (1.024–1.78) for TC, Ln TG, HDL-C, LDL-C and non-HDL-C respectively, while corresponding HRs of

ischemic stroke for men were 0.78 (0.55–1.11), 0.71 (0.33–1.51), 1.04 (0.24–4.47), 0.82 (0.56–1.22), 0.78(0.55–1.11), respectively. We found no interaction between gender and any of the lipids in risk prediction of in-cident CHD (p > 0.3). All lipid components were independently associated with CHD in whole population.

Conclusion: The associations of lipid components on ischemic stroke were modified by gender. Only amongfemale population, TC, LDL-C and non-HDL-C were independently associated with increased risk of ischemicstroke. Regarding CHD events, all lipid components were significant predictors.

© 2013 The Canadian Society of Clinical Chemists. Published by Elsevier Inc. All rights reserved.

1. Introduction

Cardiovascular disease ranked as the leading cause of mortalityhas become a major public health concern [1]. Over recent decades,epidemiologic studies show a divergent, statistically significanttrend in stroke incidence rates, with the rate being half in developedcountries and having a greater than twofold increase in developingcountries [2]. Recently Azarpazhooh et al. [3] have shown that theincidence of stroke in Iran was greater than those of most developedcountries. With these trends in mind, preventive strategies needto be urgently prioritized. Hypertension, diabetes, atrial fibrillation,physical inactivity and obesity are well established and modifiableindependent risk factors for stroke events [4,5].

The association between lipids and coronary heart disease (CHD)has been well established [6–8], but the predictive roles of lipidprofile components (serum total cholesterol (TC), triglycerides (TG),high density lipoprotein (HDL-C), low density lipoprotein (LDL-C),non-HDL-C)) on stroke events are still a matter of controversy [9].

n, Iran. Fax: +98 2122402463.egh).

ociety of Clinical Chemists. Publishe

Although many randomized control trials have shown that treatmentwith statin is associated with a significant reduction in incidence ofstroke [10], epidemiologic findings on the association between lipidcomponents and stroke are not consistent [11]. The ProspectiveStudies Collaboration in a meta-analysis from 61 prospective studiesfound a weak association between lipid profile components andstroke events [7]. Recently we showed that high TG and low HDL asthe component of metabolic syndrome did not increase the risk ofstroke events among an adult Tehranian population [5].

The aim of the present study was to evaluate the independentassociation of lipid profile components with stroke and CHD events.Furthermore, we investigated whether gender is an importantmodifier for different lipid components in prediction of incidentcerebrovascular and CHD among Iranian population.

2. Materials and methods

2.1. Study population

Subjects in this study were selected from among participants of theTehran Lipid and Glucose Study (TLGS), a prospective study conducted

d by Elsevier Inc. All rights reserved.

717M. Tohidi et al. / Clinical Biochemistry 46 (2013) 716–721

to determine the risk factors and outcomes for non-communicablediseases [12]. To summarize, 15005 people, aged 3 years and over, res-idents of district-13 of Tehran, underwent baseline assessment betweenFebruary 1999 and August 2001. In the current study, from this group,3394 subjects, aged ≥ 50 years were examined in a cross-sectionalphase of TLGS (1999 to 2001) and those with history of cardiovasculardisease (n = 435) and missing data on lipid profile (n = 54) were ex-cluded, leaving 2905 subjects, of whom 2620 subjects (1414 womenand 1206 men) were monitored until March 2009 for a medianfollow-up of 9.1 years (Fig. 1). Written informed consent was obtainedfrom all subjects and the ethical committee of the Research Institute forEndocrine Sciences approved this study.

2.2. Clinical and laboratory measurements

Baseline information including demographic data, were collectedby a trained interviewer, using a pretested questionnaire. Details ofanthropometric and blood pressure measurements were reportedelsewhere [12]. A blood sample was drawn between 7:00 and9:00 AM from all study participants, after 12–14 h overnight fasting.All the blood analyses were done at the TLGS research laboratory onthe day of blood collection. Plasma glucose was measured using anenzymatic colorimetric method with glucose oxidase. TC was assayedusing enzymatic colorimetric method with cholesterol esterase andcholesterol oxidase. HDL-C was measured after precipitation ofthe Apolipoprotein B containing lipoproteins with phosphotungsticacid; a precipitation method which has been shown to have goodcorrelation with direct enzymatic and HPLC methods [13–15]. TGwas assayed using an enzymatic colorimetric method with glycerolphosphate oxidase. In subjects, whose TG concentration was below4.52 mmol/L, LDL-C was calculated from serum TC, TG and HDL-Cconcentrations, using the Friedewald formula [16]. Non-HDL-C wascalculated by subtracting HDL-C from TC; TC/HDL-C was calculatedby dividing TC by HDL-C. These biochemical analyses were performedusing commercial kits (Pars Azmon Inc., Tehran, Iran) and a Selectra 2auto-analyzer (Vital Scientific, Spankeren, The Netherlands). The

15010 subjects aged 3 years and ovparticipated (February 1999 to August

3394 participants aged ≥ 50 years

Excluded 435Participant whit history of

CVD

2959 participat b

Excluded 54Participant missing data on

lipid profile

285 participants lofollow up

Fig. 1. Study p

quality of assays was monitored using assayed serum controls intwo different concentrations; TruLab N and TruLab P as well as anunassayed pooled serum. The intra- and inter-assay coefficients ofvariation (CV) were both 2.2% for glucose. For both total andHDL-cholesterol, intra and inter-assay CVs were 0.5 and 2% respec-tively. Intra and inter-assay CVs were 0.6 and 1.6% for TG respectively.

2.3. Definition of stroke/CHD outcome

To summarize, each participant was followed up annually for anymedical event leading to hospitalization by phone calls (at least 1 tothe maximum of 4 calls with intervals of 1 week). A trained nurseasked them regarding any medical condition and then a trainedphysician collected complementary data about that event, during ahome visit and by acquisition of data frommedical files. The collecteddata were then evaluated by an outcome committee consisting of aninternist, an endocrinologist, a cardiologist, an epidemiologist andother experts (a neurologist in the current study), when needed, toassign a specific outcome for each event. In the current study, CHD(ICD10 rubric I20–I25) and CVA (ICD10 rubric I60–I69, G45) codeswere used to define outcomes. Stroke was defined according to theWHO definition as, “rapidly developing clinical signs of or global dis-turbance of cerebral function, lasting >24 h or leading to death withno apparent cause other than that of vascular origin” [17]. Additional-ly, we used “imaging of an acute clinically relevant brain injury in pa-tients with fast vanishing symptoms” as another criterion of definitestroke. Possible stroke was defined as onset of an acute focalneurological deficit with absent imaging that is indicative of strokebut for which there is insufficient data to establish whether thesymptoms and the period fully qualify for the WHO definition fordefinite stroke [3]. Patients with Transient Ischemic Attack (TIA)may meet the above definition of possible stroke with the additionthat their symptoms resolve within 24 h. All cases, regardless ofdefinite or possible stroke or TIA description were included in thestroke definition. We classified strokes as ischemic or hemorrhagicaccording to all available information from informant interviews,

er2001)

ant free of CVDaseline

2905 participant withcomplete data on lipid

profile

st to 2620 participants (1414women) with a median

follow up of 9.1 years

opulation.

Table 1Baseline characteristics of study subjects.

All Men Women p-value

VariablesAge, (years) 60.3 (7.5) 61.5 (7.8) 59.3 (7.0) b0.001Systolic bloodpressure, (mmHg)

133.1 (22.0) 131.4 (21.5) 134.5 (22.3) b0.001

Diastolic bloodpressure, (mmHg)

81.7 (11.8) 80.6 (12.2) 82.6 (11.3) b0.001

Fasting plasmaglucose, (mmol/L)

6.11 (2.53) 5.99 (2.29) 6.22 (2.73) 0.018

Waist to hip ratio 0.93 (0.08) 0.95 (0.07) 0.90 (0.08) b0.001Total cholesterol,(mmol/L)

5.98 (1.23) 5.53 (1.08) 6.38 (1.22) b0.001

Ln triglycerides,(mmol/L)

0.65 (0.52) 0.59 (0.54) 0.71 (0.51) b0.001

High densitylipoprotein-C, (mmol/L)

1.11 (0.29) 1.02 (0.26) 1.20 (0.29) b0.001

Low densitylipoprotein-C, (mmol/L)

3.89 (1.01) 3.58 (0.90) 4.16 (1.03) b0.001

Non-high densitylipoprotein-C, (mmol/L)

4.87 (1.21) 4.51 (1.07) 5.18 (1.23) b0.001

TC/HDL 5.67 (1.69) 5.73 (1.69) 5.63 (1.69) 0.123Risk factors

Diabetes, (%) 458 (17.5) 190 (15.8) 268 (19.0) 0.034Hypertension, (%) 1158 (45.2) 461 (38.9) 697 (50.1) b0.001Obesity, (%) 755 (29.3) 197 (16.6) 558 (40.2) b0.001Smoking, b0.001Current smoker, (%) 276 (10.5) 241 (20) 35 (2.5)Past smoker, (%) 286 (10.9) 242 (20.1) 44 (3.1)

Lipid lowering drug, (%) 176 (6.7) 39 (3.2) 137 (9.7) b0.001Anti-hypertensiondrug, (%)

445 (17) 117 (9.7) 328 (23.2) b0.001

Mean (SD) are shown for continuous variables and % for categorical variables.Comparisons between men and women were done with independent T-test, Fisher'sexact test and Chi-square test as appropriate.

718 M. Tohidi et al. / Clinical Biochemistry 46 (2013) 716–721

medical records, and brain imaging studies (Computed Tomographyand Magnetic Resonance Imaging) [18].

To avoid possible misclassification of censoring nonvasculardeaths, we used the following methods; 1. verbal autopsy, i.e. methodof obtaining as much information as possible about a deceased personby asking questions of the immediate family and others who candefine the mode of death and situations preceding death; thismethod is used especially in developing countries and in settingsand situations in which postmortem pathologic examination is notfeasible. 2. These individuals are referred for autopsy to the ForensicMedicine organization and all related information is obtained andrechecked by the outcome committee. 3. The only misclassificationthat could occur, may be a case found dead in the street with noidentification or of one claiming a body, with an unknown cause ofdeath; we had no such case [19].

2.4. Statistical analyses

Cox regression analysis was developed to estimate the hazardratio of stroke events for a 1 mmol/L change in TC, LDL-C, Ln TG,HDL-C, non-HDL-C and one unit change in TC/HDL-C. Follow-up dura-tion was defined as the period between entrance to study and the endpoint in each analysis. End points were considered as the first stroke/CHD events and censoring was defined as leaving the residence area,non-stroke/CHD death, lost to follow-up or until the end of follow-up.

To explore the shape and strength of the associations between lipidsand incident stroke/CHD instead of using arbitrary predeterminedcut-points; we used restricted cubic spline functions of lipid measuresto represent their continuous relationship with the stroke risk so thatthe relationships were meaningfully in accordance with substantivebackground knowledge; multivariate restricted cubic splines wereused with 4 knots defined at the 5th, 25th, 75th, and 95th percentiles[20,21].Whenwe associated lipid levels with the risk of incident stroke,we examined performances of the models with respect to discrimina-tion, calibration and deviance. We observed that as compared to thecubic spline model, the linear ones were more parsimonious, and thevalues for performances were not significantly different, indicatingthat non-linear models did not fit the data better than their linearcounterparts. We therefore reported only the linear models (dataavailable on request).

Each candidate covariate (age as a time dependent variable, sex,SBP, diabetes, smoking status, waist to hip ratio (WHR), lipidlowering drug use, anti-hypertensive drugs) were entered in themultivariable analysis. Interactions between sex–lipid profiles withincident stroke/CHD were tested by log–likelihood ratio test, inmultivariate analysis. As there was a significant interaction betweensex and serum cholesterol in risk prediction of ischemic stroke(p = 0.027), we stratified our analysis by sex. Furthermore tocompare our findings with other studies, we did our analysis in apooled sample, as well. The proportional hazard assumption in theCox model was checked graphically and with the Schoenfeld residualtest. Since, the proportionality assumption was not found for age,extended Cox analysis was applied in the total population as well asin men for stroke events (including ischemic ones); for other modelsthe proportionality assumptions were generally appropriate, hence,we used Cox proportional regression analysis. Statistical analyseswere performed using STATA version 11.0 (StataCorp LP, CollegeStation, Texas).

3. Results

The study sample consisted of 2620 individuals (1414 womenand 1206 men) aged over 50 years, with mean age of 60.3 years(SD = 7.5). A comparison between those who were followed upand who were not showed that non-participants were more likelyto be women (66% vs. 54%) and had reported higher rate of

consumption of lipid lowering drugs (11.9% vs. 6.7%) (SupplementaryTable 1). During 9.1 years of follow-up, 73 (42 women, 31 men) and358 (161 women, 197 men) cases of first stroke and first CHD oc-curred, considering 7 patients who had CHD events subsequently de-veloped stroke events. The incidence rate of stroke events was 326(CI: 259–410) per 100 000 person-years and the corresponding ratefor CHD was 1663 (CI: 1499–1844) per 100 000 person-years.Table 1 highlights the baseline characteristics of the participants.Among both gender those who experience stroke/CHD were older,had higher systolic and diastolic blood pressure, TC, Ln TG, LDL-Cand used more hypertensive drugs compared to individuals with nostroke/CHD.

Hazard ratios (HR) of stroke for lipid profiles are expressed inTable 2. In the age adjusted model for pooled population, none ofthe lipid profile components were associated with increased risk ofstroke. Among women, the HRs of stroke were 1.30 (1.02–1.67),1.89 (0.94–3.78), 1.23 (0.36–4.21), 1.39 (1.01–1.93) and 1.31 (1.02–1.68) for a 1 mmol/L change in TC, Ln TG, HDL-C, LDL-C andnon-HDL-C respectively. But after further adjustment for potentialrisk factors (age, SBP, diabetes, WHR, antihypertensive drugs, lipidlowering drugs and smoking), TC, non-HDL-C and LDL lost their asso-ciation with stroke events. Among men, in an age adjusted model, thecorresponding HRs of stroke were 0.84 (0.62–1.13), 0.85 (.0.48–1.51),0.83 (0.25–2.71), 0.86 (0.61–1.22), 0.85 (0.63–1.15) for a 1 mmol/Lchange in TC, Ln TG, HDL-C, LDL-C and non-HDL-C respectively.After excluding 15 cases of hemorrhagic stroke from our outcome,among women with ischemic stroke (n = 25), the multivariate HRsof ischemic stroke for TC (1.40 (1.08–1.82)), non-HDL (1.36 (1.04–1.78)) and LDL-C (1.51 (1.06–2.15)) became significant. Amongmen with ischemic stroke (n = 33), however, none of the lipidcomponents were predictors. Data are shown in Table 3.

Table 4 highlights the HRs of CHD for lipid profile components. Wefound no interaction between sex and any of the lipid profile

Table 2Hazard ratios and 95% confidence interval of stroke events for lipid profile components.

Women Men All

Age adjusted p-v Multivariatea p-v Age adjusted p-v Multivariatea p-v Age adjusted p-v Multivariatea p-v p-vb

TC 1.30 (1.02–1.67) .035 1.25 (0.96–1.62) .100 0.84 (0.62–1.13) .252 0.76 (0.56–1.04) .085 1.01 (0.84–1.23) .886 0.98 (0.80–1.21) .879 .038Ln TG 1.89 (0.94–3.78) .073 1.67 (0.77–3.62) .193 0.85 (0.48–1.51) .584 0.57 (0.29–1.12) .102 1.13 (0.73–1.75) .576 0.89 (0.53–1.45) .611 .105HDL 1.23 (0.36–4.21) .744 1.22 (0.31–4.69) .776 0.83 (0.25–2.71) .751 1.15 (0.33–4.03) .831 0.83 (0.37–1.90) .662 1.11 (0.44–2.78) .823 .840LDL 1.39 (1.01–1.93) .044 1.36 (0.97–1.91) .073 0.86 (0.61–1.22) .399 0.81 (0.57–1.14) .229 1.03 (0.81–1.30) .812 1.04 (0.81–1.33) .777 .057Non-HDL 1.31 (1.02–1.68) .033 1.28 (0.98–1.67) .071 0.85 (0.63–1.15) .291 0.75 (0.55–1.03) .075 1.03 (0.85–1.25) .792 0.99 (0.80–1.23) .948 .037TC/HDL 1.16 (0.95–1.41) .159 1.15 (0.93–1.43) .190 0.95 (0.79–1.15) .593 0.86 (0.70–1.06) .168 1.04 (0.91–1.19) .605 0.99 (0.85–1.15) .897 .143

TC; total cholesterol, TG; triglycerides, HDL; high density lipoprotein, LDL; low density lipoprotein, p-v; p-value.a Model was adjusted for; age, (sex for All), systolic blood pressure, waist to hip ratio, anti-hypertension drug, diabetes, smoking and lipid lowering drug.b The p-value for interaction between each lipid profile components and gender in the multivariate model.

Table 3Hazard ratios and 95% confidence interval of ischemic stroke for lipid profile components.

Women Men All

Age adjusted p-v Multivariatea p-v Age adjusted p-v Multivariatea p-v Age adjusted p-v Multivariatea p-v p-vb

TC 1.42 (1.12–1.81) .004 1.40 (1.08–1.82) .011 0.89 (0.64–1.25) .495 0.78 (0.55–1.11) .175 1.11 (0.90–1.37) .316 1.09 (0.87–1.37) .467 .027Ln TG 1.75 (0.81–3.78) .156 1.66 (0.71–3.86) .238 1.06 (0.57–2.00) .845 0.71 (0.33–1.51) .372 1.24 (0.77–2.02) .376 0.99 (0.57–1.71) .960 .324HDL 2.10 (0.59–7.48) .254 2.27 (0.58–8.91) .241 0.65 (0.16–2.57) .538 1.04 (0.24–4.47) .954 1.03 (0.42–2.52) .944 1.44 (0.54–3.84) .471 .332LDL 1.53 (1.10–2.12) .011 1.51 (1.06–2.15) .022 0.91 (0.61–.1.34) .628 0.82 (0.56–1.22) .329 1.14 (0.88–1.47) .323 1.13 (0.86–1.48) .383 .051Non-HDL 1.39 (1.08–1.78) .010 1.36 (1.04–1.78) .023 0.92 (0.66–1.28) .605 0.78 (0.55–1.11) .169 1.12 (0.90–1.37) .310 1.07 (0.85–1.34) .584 .045TC/HDL 1.14 (0.91–1.42) .240 1.13 (0.89–1.43) .311 0.99 (0.81–1.22) .916 0.87 (0.69–1.09) .231 1.05 (0.91–1.22) .512 .99 (0.84–1.17) .916 .262

TC; total cholesterol, TG; triglycerides, HDL; high density lipoprotein, LDL; low density lipoprotein, p-v; p-value.a Model was adjusted for; age, (sex for All), systolic blood pressure, waist to hip ratio, anti-hypertension drug, diabetes, smoking and lipid lowering drug.b The p-value for interaction between each lipid profile components and gender in the multivariate model.

719M. Tohidi et al. / Clinical Biochemistry 46 (2013) 716–721

components in risk prediction of incident CHD (all p > 0.3). In pooledpopulation, all the lipid profile components were independently asso-ciated with CHD. In multivariate analysis, the HRs of CHD were 1.16(1.06–1.27), 1.27 (1.02–1.57), 0.59 (0.38–0.91), 1.23 (1.10–1.38),1.19 (1.09–1.30), 1.14 (1.07–1.21) for TC, Ln TG, HDL-C, LDL-C andnon-HDL-C and TC/HDL-C, respectively.

4. Discussion

In the current study during more than 9 years of follow up of aMiddle Eastern population, we demonstrated that none of the lipidprofile components were independently predictors of stroke events;however after excluding the hemorrhagic cases from our outcome,we found a statistically significant difference in the association oflipid profile components with ischemic stroke events for differentsexes. Among women, TC, non-HDL-C and LDL-C independent ofpotential risk factors, were associated with increased risk of develop-ing ischemic stroke events; however among men, none of the lipidprofile components increased the risk of stroke events. Also, in sex

Table 4Hazard ratios and 95% confidence interval of coronary heart disease for lipid profile compo

Women Men

Age adjusted p-v Multivariatea p-v Age adjusted p-v

TC 1.17 (1.04–1.32) .008 1.12 (0.99–1.26) .074 1.26 (1.12–1.43) b .001Ln TG 1.84 (1.37–2.47) b .001 1.34 (0.97–1.86) .076 1.58 (1.23–2.03) b .001HDL 0.37 (0.21–0.67) .001 0.49 (0.27–0.90) .020 0.49 (0.27–0.88) .018LDL 1.24 (1.07–1.44) .005 1.21 (1.04–1.41) .016 1.26 (1.08–1.48) .004Non-HDL 1.22 (1.09–1.36) .001 1.15 (1.02–1.30) .021 1.32 (1.17–1.49) b .001TC/HDL 1.21 (1.12–1.31) b .001 1.15 (1.06–1.26) .001 1.18 (1.10–1.27) b .001

TC; total cholesterol, TG; triglycerides, HDL; high density lipoprotein, LDL; low density lipoa Model was adjusted for; age, (sex for All), systolic blood pressure, waist to hip ratio, anb The p-value for interaction between each lipid profile components and gender in the m

and traditional risk factor adjusted analyses, all the lipid profilecomponents were associated with increased risk of CHD.

The independent association between lipid profile componentsand CHD events in our study is consistent with widely accepted con-tribution between lipids and CHD [6–8] but the roles of lipids onstroke events are still a matter of controversy [11]. In line with ourfindings, the results of some previous studies have indicated thatthe association between lipids and risk of stroke may vary by sex[22,23]. In a study among rural Chinese, Zhang et al found that TC,non-HDL and TC/HDL ratio were related to ischemic stroke as riskfactors only among women [22]. Recently, Kim, NS et al. haveexamined whether sex modifies the genetic effect of C-399T onsusceptibility to ischemic stroke. They concluded that C-399T NPYpromoter polymorphism should be considered a genetic risk factorfor ischemic stroke in the older adult and female Korean populations[24]. But as claimed by E. Cuadrado-Godia [25], the reasons fordifferences according to sex may need further research.

In the Women Health Study [26], conducted on 27 000 women,aged ≥45 years, only TC, LDL-C and non-HDL-C were independentlyassociated with increased risk of ischemic stroke. HDL-C lost its

nents.

All

Multivariatea p-v Age adjusted p-v Multivariatea p-v p-vb

1.22 (1.07–1.39) .003 1.13 (1.04–1.23) .003 1.16 (1.06–1.27) .001 .4041.23 (0.92–1.64) .163 1.63 (1.35–1.98) b .001 1.27 (1.02–1.57) .031 .5450.71 (0.38–1.33) .288 0.37 (0.25–0.56) b .001 0.59 (0.38–0.91) .017 .3301.26 (1.07–1.48) .005 1.17 (1.05–1.30) .003 1.23 (1.10–1.38) b .001 .6911.27 (1.11–1.45) b .001 1.19 (1.10–1.29) b .001 1.19 (1.09–1.30) b .001 .3811.14 (1.05–1.23) .002 1.20 (1.14–1.26) b .001 1.14 (1.07–1.21) b .001 .602

protein, p-v; p-value.ti-hypertension drug, diabetes, smoking and lipid lowering drug.ultivariate model.

720 M. Tohidi et al. / Clinical Biochemistry 46 (2013) 716–721

association after adjusting for other conventional risk factors. In theNorthern Manhattan Study [9], conducted among participants aged>40 years, none of the baseline lipid profile components (TC,HDL-C, LDL-C TG and non-HDL-C) were associated with risk ofdeveloping ischemic stroke even in unadjusted analysis; howeverafter excluding those who took lipid lowering medications, theyfound a trend toward an increase in risk of ischemic stroke with anLDL-C level greater than 130 mg/dL. In the Emerging Risk FactorsCollaboration study [27] among individuals in Europe and NorthAmerica, the sex and risk factor adjusted HRs of ischemic stroke for1 SD increase in TG, HDL-C and non-HDL-C were 1.02 (0.94−1.11),0.93 (0.84−1.02) and 1.12 (1.04−1.20), respectively.

In the present study, TC was associated with ischemic strokeamong women but not men, after controlling for age, systolic bloodpressure, waist to hip ratio, anti-hypertension drugs, diabetes,smoking and lipid lowering drugs. In line with our results, in theWomen Pooling Project, a prospective study of 24 343 US women,aged >30, higher TC levels were associated with risk of developingstroke events [28]. However, contrast to our findings among men,Iso et al. [29] showed that the hazard ratio of non-hemorrhagic strokeincreased for each level of cholesterol among men. Recently, in ameta-analysis of 61 prospective studies, the Prospective StudiesCollaboration [7] highlighted that the association between bloodcholesterol and stroke mortality was modified by age and bloodpressure levels.

Findings of epidemiologic studies on the association betweentriglycerides and stroke are inconsistent [30]. In our study, fasting tri-glycerides did not increase the risk of stroke in either sex. Leonards etal. [31] have raised concern about the necessity to run large studies todetect the effect of TG as they exhibit a relatively weak influence onstroke which might lead to a type 2 error in statistics. In line withour findings, in the ARIC study, fasting TG levels did not predict ische-mic stroke [32]. Recently some studies suggested that non-fastingtriglyceride level may be superior to fasting levels for determiningcardiovascular risk. Bansal et al. highlighted that non-fasting TG wasindependently associated with ischemic stroke, but found no clearassociation between ischemic stroke and fasting triglycerides [33].

Some studies have raised questions about the role of HDL-C onatherothrombotic and atherosclerotic pathology [34]. In this study,there was no association between HDL-C and ischemic stroke inboth sexes. The absence of association of HDL-C with stroke in ourstudy may contribute to lack of variation on HDL-C distribution dueto the high prevalence of low HDL-C among the Tehranian population[35]. Sacco et al. [36] in the Northern Manhattan Stroke Study showedinverse associations between HDL-C and risk of ischemic stroke in theelderly and among different racial or ethnic groups.

In our dataset, we found out significant independent associationsbetween non-HDL-C and LDL-C with stroke events among women.Recently, some US studies have suggested that non-HDL-C might bea better predictor for coronary heart disease [37] but data on thepredictive ability of non-HDL on stroke is sparse. In the third NationalHealth and Nutrition Examination Survey in the US [38], a 10 mg/dLincrease in LDL-C and HDL-C did not increase the risk of developingstroke in both univariate and multivariate analyses; after categoriza-tion of the participants according to non-HDL-C levels, the HR ofstroke for those with non-HDL-C 190–430 mg/dL was 5.81 (1.96–17.27) compared to those with non-HDL-C 35–129 mg/dL.

Some of our study limitations however need to be mentioned.First, due to the limited number of our stroke events we were unableto separately analyze and assess the risk factors for hemorrhagicstroke. Wu et al. [39] had concluded that the risk factors of strokesubtypes differ as they have different etiopathologies. Second, ablood sample was taken at baseline and we did not consider lipid pro-file changes during follow up. Third, our study was conducted amongMiddle East Caucasian individuals, aged over 50 years; hence wecannot extrapolate our findings to other populations.

In conclusion, we demonstrated that the associations of lipidprofile components on ischemic stroke events were modified by sex.From the clinical stand point, the present study provides evidencebased support, only in women for the American Stroke Associationrecommendation that recognizes TC, LDL-C as well as non-HDL-C asrisk factors for ischemic stroke events. Hence, therapeutic life stylemodification including adherence to dietary guidelines (i.e. reductionin saturated fat intake) and increase in physical activity as well astreatment with HMG-Co inhibitor medications should be consideredamong Iranian women for primary prevention of ischemic stroke.

Supplementary data to this article can be found online at http://dx.doi.org/10.1016/j.clinbiochem.2013.03.012.

Conflict of interest

The authors declare that they have no competing interests.

Acknowledgments

This study was supported by grant no.121 from the National Re-searchCouncil of the Islamic Republic of Iran.Wewould like to acknowl-edge Ms. Nilufar Shiva for the language editing of the manuscript.

References

[1] Lopez AD, Mathers CD, Ezzati M, Jamison DT, Murray CJ. Global and regionalburden of disease and risk factors, 2001: systematic analysis of population healthdata. Lancet 2006;367:1747–57.

[2] Feigin VL, Lawes CM, Bennett DA, Barker-Collo SL, Parag V. Worldwide strokeincidence and early case fatality reported in 56 population-based studies: asystematic review. Lancet Neurol 2009;8:355–69.

[3] Azarpazhooh MR, Etemadi MM, Donnan GA, Mokhber N, Majdi MR, Ghayour-MobarhanM, et al. Excessive incidence of stroke in Iran: evidence from theMashhadStroke Incidence Study (MSIS), a population-based study of stroke in the MiddleEast. Stroke 2010;41:e3-10.

[4] O'Donnell MJ, Xavier D, Liu L, Zhang H, Chin SL, Rao-Melacini P, et al. Risk factorsfor ischaemic and intracerebral haemorrhagic stroke in 22 countries (theINTERSTROKE study): a case-control study. Lancet 2010;376:112–23.

[5] Hadaegh F, Mohebi R, Cheraghi L, Tohidi M, Moghaddam NB, Bozorogmanesh M,et al. Do different metabolic syndrome definitions predict cerebrovascular eventsand coronary heart disease independent of their components?: 9 years follow-upof the tehran lipid and glucose study. Stroke: J Cereb Circ 2012;43:1669–71.

[6] Sharrett AR, Ballantyne CM, Coady SA, Heiss G, Sorlie PD, Catellier D, et al. Coronaryheart disease prediction from lipoprotein cholesterol levels, triglycerides,lipoprotein(a), apolipoproteins A–I and B, and HDL density subfractions: theAtherosclerosis Risk in Communities (ARIC) study. Circulation 2001;104:1108–13.

[7] Lewington S, Whitlock G, Clarke R, Sherliker P, Emberson J, Halsey J, et al. Bloodcholesterol and vascular mortality by age, sex, and blood pressure: a meta-analysisof individual data from 61 prospective studies with 55,000 vascular deaths. Lancet2007;370:1829–39.

[8] Canoui-Poitrine F, Luc G, Bard JM, Ferrieres J, Yarnell J, Arveiler D, et al. Relativecontribution of lipids and apolipoproteins to incident coronary heart diseaseand ischemic stroke: the PRIME Study. Cerebrovasc Dis 2010;30:252–9.

[9] Willey JZ, Xu Q, Boden-Albala B, Paik MC, Moon YP, Sacco RL, et al. Lipid profilecomponents and risk of ischemic stroke: the Northern Manhattan Study(NOMAS). Arch Neurol 2009;66:1400–6.

[10] Baigent C, Keech A, Kearney PM, Blackwell L, Buck G, Pollicino C, et al. Efficacyand safety of cholesterol-lowering treatment: prospective meta-analysis ofdata from 90,056 participants in 14 randomised trials of statins. Lancet2005;366:1267–78.

[11] Goldstein LB, Bushnell CD, Adams RJ, Appel LJ, Braun LT, Chaturvedi S, et al. Guide-lines for the primary prevention of stroke: a guideline for healthcare professionalsfrom the American Heart Association/American Stroke Association. Stroke2011;42:517–84.

[12] Azizi F, Ghanbarian A, Momenan AA, Hadaegh F, Mirmiran P, Hedayati M, et al.Prevention of non-communicable disease in a population in nutrition transition:Tehran Lipid and Glucose Study phase II. Trials 2009;10:5.

[13] Jabbar J, Siddiqui I, Raza Q. Comparison of twomethods (precipitationmanual andfully automated enzymatic) for the analysis of HDL and LDL cholesterol. J Pak MedAssoc 2006;56:59–61.

[14] Okazaki M, Sasamoto K, Muramatsu T, Hosaki S. Evaluation of precipitationand direct methods for HDL-cholesterol assay by HPLC. Clin Chem 1997;43:1885–90.

[15] Hedayati MDM, Azimzadeh I, Arbabi SI. Comparison of direct method withprecipitation method of HDL-cholesterol assay. J Fac Med 2007;31:285–7.

[16] Friedewald WT, Levy RI, Fredrickson DS. Estimation of the concentration oflow-density lipoprotein cholesterol in plasma, without use of the preparativeultracentrifuge. Clin Chem 1972;18:499–502.

721M. Tohidi et al. / Clinical Biochemistry 46 (2013) 716–721

[17] Organization WH. Cerebrovascular disorders: a clinical and research classification;1978.

[18] Fahimfar N, Khalili D, Mohebi R, Azizi F, Hadaegh F. Risk factors for ischemicstroke; results from 9 years of follow-up in a population based cohort of Iran.BMC Neurol 2012;12:117.

[19] Hadaegh F, Harati H, Ghanbarian A, Azizi F. Prevalence of coronary heart diseaseamong tehran adults: Tehran lipid and glucose study. Eastern Mediterraneanhealth journal = La revue de sante de la Mediterranee orientale = al-Majallahal-sihhiyah li-sharq al-mutawassit 2009;15:157–66.

[20] Harrell FE. Regression modeling strategies. New York: Springer; 2001.[21] Harrell Jr FE, Lee KL, Mark DB. Multivariable prognostic models: issues in

developing models, evaluating assumptions and adequacy, and measuring andreducing errors. Stat Med 1996;15:361–87.

[22] Zhang X, Sun Z, Zheng L, Liu S, Xu C, Li J, et al. Gender differences in blood lipidsand the risk of ischemic stroke among the hypertensive adults in rural China.Neurol India 2007;55:338–42.

[23] Cui R, Iso H, Yamagishi K, Saito I, Kokubo Y, Inoue M, et al. High serum total cho-lesterol levels is a risk factor of ischemic stroke for general Japanese population:the JPHC study. Atherosclerosis 2012;221:565–9.

[24] Kim NS, Ko MM, Cha MH, Oh SM, Bang OS. Age and sex dependent genetic effectsof neuropeptide Y promoter polymorphism on susceptibility to ischemic stroke inKoreans. Clin Chim Acta 2010;411:1243–7.

[25] Cuadrado-Godia E, Jimenez-Conde J, Ois A, Rodriguez-Campello A, Garcia-RamalloE, Roquer J. Sex differences in the prognostic value of the lipid profile after thefirst ischemic stroke. J Neurol 2009;256:989–95.

[26] Kurth T, Everett BM, Buring JE, Kase CS, Ridker PM, Gaziano JM. Lipid levels andthe risk of ischemic stroke in women. Neurology 2007;68:556–62.

[27] Di Angelantonio E, Sarwar N, Perry P, Kaptoge S, Ray KK, Thompson A, et al.Major lipids, apolipoproteins, and risk of vascular disease. JAMA 2009;302:1993–2000.

[28] Horenstein RB, Smith DE, Mosca L. Cholesterol predicts stroke mortality in theWomen's Pooling Project. Stroke 2002;33:1863–8.

[29] Iso H, Jacobs Jr DR, Wentworth D, Neaton JD, Cohen JD. Serum cholesterol levelsand six-year mortality from stroke in 350,977 men screened for the multiplerisk factor intervention trial. N Engl J Med 1989;320:904–10.

[30] Antonios N, Angiolillo DJ, Silliman S. Hypertriglyceridemia and ischemic stroke.Eur Neurol 2008;60:269–78.

[31] Leonards C, Ebinger M, Batluk J, Malzahn U, Heuschmann P, Endres M. The role offasting versus non-fasting triglycerides in ischemic stroke: a systematic review.Front Neurol 2010;1:133.

[32] Shahar E, Chambless LE, Rosamond WD, Boland LL, Ballantyne CM, McGovern PG,et al. Plasma lipid profile and incident ischemic stroke: the Atherosclerosis Risk inCommunities (ARIC) study. Stroke 2003;34:623–31.

[33] Bansal S, Buring JE, Rifai N, Mora S, Sacks FM, Ridker PM. Fasting compared withnonfasting triglycerides and risk of cardiovascular events in women. JAMA2007;298:309–16.

[34] Navab M, Berliner JA, Subbanagounder G, Hama S, Lusis AJ, Castellani LW, et al.HDL and the inflammatory response induced by LDL-derived oxidized phospho-lipids. Arterioscler Thromb Vasc Biol 2001;21:481–8.

[35] Zabetian A, Hadaegh F, Azizi F. Prevalence of metabolic syndrome in Iranian adultpopulation, concordance between the IDF with the ATPIII and the WHOdefinitions. Diabetes Res Clin Pract 2007;77:251–7.

[36] Sacco RL, Benson RT, Kargman DE, Boden-Albala B, Tuck C, Lin IF, et al.High-density lipoprotein cholesterol and ischemic stroke in the elderly: theNorthern Manhattan Stroke Study. JAMA 2001;285:2729–35.

[37] Liu J, Sempos CT, Donahue RP, Dorn J, Trevisan M, Grundy SM. Non-high-densitylipoprotein and very-low-density lipoprotein cholesterol and their risk predictivevalues in coronary heart disease. Am J Cardiol 2006;98:1363–8.

[38] Li C, Ford ES, Tsai J, ZhaoG, Balluz LS, Gidding SS. Serumnon-high-density lipoproteincholesterol concentration and risk of death from cardiovascular diseases among U.S.adults with diagnosed diabetes: the ThirdNational Health andNutrition ExaminationSurvey linked mortality study. Cardiovasc Diabetol 2011;10:46.

[39] Wu CY,Wu HM, Lee JD, Weng HH. Stroke risk factors and subtypes in different agegroups: a hospital-based study. Neurol India 2010;58:863–8.