Cigarette smoking and cerebral blood flow in a cohort of middle-aged adults

Abstract

Cigarette smoking is often associated with dementia. This association is thought to be mediated by hypoperfusion; however, how smoking behavior relates to cerebral blood flow (CBF) remains unclear. Using data from the Coronary Artery Risk Development in Young Adults (CARDIA) cohort (mean age = 50; n = 522), we examined the association between smoking behavior (status, cumulative pack-years, age at smoking initiation, and years since cessation) and CBF (arterial spin labeling) in brain lobes and regions linked to dementia. We used adjusted linear regression models and tested whether associations differed between current and former-smokers. Compared to never-smokers, former-smokers had lower CBF in the parietal and occipital lobes, cuneus, precuneus, putamen, and insula; in contrast, current-smokers did not have lower CBF. The relationship between pack-years and CBF was different between current and former-smokers (p for interaction < 0.05): Among current-smokers, higher pack-years were associated with higher occipital, temporal, cuneus, putamen, insula, hippocampus, and caudate CBF; former-smokers had lower caudate CBF with increasing pack-years. Results show links between smoking and CBF at middle-age in regions implicated in cognitive and compulsive/addictive processes. Differences between current and former smoking suggest that distinct pathological and/or compensatory mechanisms may be involved depending on the timing and history of smoking exposure.

Introduction

Cigarette smoking is among the most prevalent risk factors for cardiac disease, cancer, and mortality. Cigarette smoking has also been associated with higher risk of dementia and cognitive decline.^1,2 This association is hypothesized to reflect the adverse effects of smoking on vascular systems leading to hypoperfusion, oxidative damage, inflammation, and deficits in oxygen and nutrient delivery in the brain, and subsequently affecting cognitive functioning.^3–5 This process is accentuated as people age and develop vascular and respiratory-related diseases.^2,5

However, the relationship between chronic smoking and cerebral blood flow (CBF) has not been well studied.⁶ It is particularly important to understand the smoking-CBF relationship at younger age, before the burdens of aging and accumulated vascular and cerebrovascular disease become substantial.^7,8 This approach has been supported by numerous studies showing mid-life exposures are better predictors of future cognitive disease than exposures measured concurrently with cognition.^8,9

Prior investigations on smoking and changes in CBF have been conducted in small samples (<100 subjects) and have predominantly focused on acute effects of smoking (measuring CBF immediately after smoking or after short periods of withdrawal).^10,11 Studies incorporating data on chronic smoking are few and generally report an association between smoking and reduced CBF, but these studies are based on smaller^12,13 and/or older samples,^13,14 and did not explore regional CBF.^13,14 Findings from structural and functional imaging studies suggest that smoking is differentially related to changes in certain brain regions that have been linked to cognitive functioning,^15,16 motivating the search for regional differences in CBF between smokers and non-smokers.¹²

In the present study, we examined the relationship between smoking and CBF in a community-dwelling cohort of middle-aged adults. We assessed several aspects of smoking behavior (current smoking status, cumulative pack-years, age at start of smoking, and years since smoking cessation) and focused on regional CBF in brain lobes and candidate regions of dementia.

Methods

Subjects

The Coronary Artery Risk Development in Young Adults (CARDIA) study is a community-based longitudinal study of the determinants of cardiovascular disease in young adults.¹⁷ At baseline (1985–1986), the study recruited 5115 participants aged 18–30 years randomly selected by telephone numbers from census tracts in four US cities (Birmingham, Alabama; Chicago, Illinois; Minneapolis, Minnesota; and Oakland, California); the sample was balanced by race (black, white), sex, and education (high school degree or less, higher than high school). Participants have been followed since study inception, with examinations at years 2, 5, 7, 10, 15, 20, and 25 of follow-up; 72% (n = 3498) of the total baseline cohort participated in the Year 25 follow-up study visit. A sub-sample of these participants from three CARDIA field centers (Birmingham, Minneapolis, and Oakland) participated in the brain MRI examinations. Each field center had a target sample size and aimed to achieve a balanced sample within sex-race groups; enrollment was ended when that target was reached. Exclusion criteria were a contra-indication to MRI or a body size too large for the MRI tube bore. In total, 719 subjects participated in the CARDIA brain MRI sub-study. In this present study, we include the participants from the two study sites (Minneapolis and Oakland) where CBF measurements were obtained (details below). The CARDIA brain MRI sub-study was approved by the Institutional Review Boards at each participating site (the Institutional Review Board of the University of Minnesota, the Kaiser Permanente Northern California Institutional Review Board, University of Pennsylvania Institutional Review Board, The University of Alabama Birmingham Institutional Review Board, and the NIH Office of Human Subjects Research Protection for the Intramural Research Program, National Institute on Aging). Institutional Review Boards in the United States adhere to the ethical principles and guidelines for the protection of human subjects in research enumerated in the Belmont Report (http://www.hhs.gov/ohrp/humansubjects/guidance/belmont.html), produced by the National Commission for the Protection of Human Subjects of Biomedical and Behavioral Research (April 1979).

All participants provided written informed consent for the brain MRI sub-study and for all other CARDIA exams.¹⁸

Imaging data

Structural (T1, T2, and FLAIR), and arterial spin labeling (ASL) imaging sequences were acquired as part of the brain MRI examinations.¹⁸ The scan duration was 40 min. CBF (volume of flow per unit brain mass per unit time (mL/100 g/min) was measured with non-background suppressed pseudo-continuous ASL (PCASL) data acquired at the Minneapolis and Oakland study sites using gradient-echo echoplanar imaging (EPI) with the following parameters: TR/TE: 4 s/11 ms, voxel size: 3.4 × 3.4 × 5 mm³, matrix = 64 ×64, flip angle = 90°, FOV = 220 mm, bandwidth =3004 Hz/pixel, echo spacing = 0.44 ms and EPI factor = 64. 20 slices with a distance factor of 20% were acquired from inferior to superior in a sequential order. PCASL labeling was performed at 9 cm below the center of the imaging volume with labeling duration =1.48 s, post labeling delay (PLD) = 1.5 s, RF gap =0.36 ms, RF duration = 0.5 ms and mean G_z = 0.6 mT/m; 40 label and control pairs were acquired.

ASL data were processed using custom MATLAB scripts (The Mathworks Inc., Natick, Massachusetts, USA) using functions from SPM8 (Wellcome Department of Imaging Neuroscience, London, UK, http://www.fil.ion.ucl.ac.uk/spm/software/spm8/) and ASL toolbox.¹⁹ The processing consists of aligning the raw EPI images using a two-step procedure,²⁰ first by estimating a 6-parameter rigid body motion spatial transformation, and subsequently regressing out the spurious motion component caused by systematic label-control alteration before applying the transformation to the images. Thereafter, the motion parameters were regressed out from the EPI time series.²⁰ Subsequently, the volumes were smoothed with an isotropic Gaussian kernel with a full-width-half-max = 5 mm. Finally, a CBF time series was obtained with each CBF volume computed using a single compartment model according to Alsop et al.²¹

$$\mathit{CBF}(\text{ml}/100\text{g}/\text{min})=\frac{60\times 100\lambda \Delta {\mathit{Me}}^{\omega /T_{1,\mathit{blood}}}}{2\alpha T_{1,\mathit{blood}}{M}_0(1-e^{-\tau /T_{1,\mathit{blood}}})}$$

where $\Delta M$ is control-label difference, λ (=0.9 ml/g) is the brain/blood partition coefficient, ω (=1.5 s) is the post labeling delay, ${\text{T}}_{1,\text{blood}}$ (1.64 s) is the ${\text{T}}_1$ of blood, α (=0.85) is the labeling (tagging) efficiency, ${\text{M}}_0$ is the equilibrium magnetization of the brain and was made equal to the corresponding control images for each CBF volume in the time series, and τ (=1.48 s) is the labeling duration. The final CBF map was obtained by averaging the CBF time series. Quality control was performed by visual inspection of the slices and CBF maps with extensive negative CBF voxels and extreme values at the edge of the brain (suggestive of motion artifacts) were discarded from the analysis (n = 3). The CBF maps were subsequently registered to the T1-weighted images of the same subject, for which anatomical ROIs were segmented using a multi-atlas label fusion method²² and mean CBF values were calculated in each anatomical region of interest.

We report on CBF in the GM as it found to be more reliably measured than WM-CBF.^23,24 The outcomes of interest were selected a priori and included the regional values of GM CBF in brain lobes and in brain structures that are widely linked to Alzheimer’s disease: hippocampus, parahippocampal gyrus, entorhinal cortex, cuneus, and precuneus, and in structures that have been linked to various forms of dementia and cognitive functioning: caudate, insula, putamen and thalamus.^25–29

Smoking data

Information on cigarette smoking was collected at baseline and each follow-up study visit (years 2, 5, 7, 10, 15, 20, and 25) through standardized interviewer-based questionnaires. Participants reported whether they have ever smoked regularly (regular smoking was defined as smoking at least 5 cigarettes per week, almost every week). If they answered yes, participants were then asked whether they were currently smoking regularly for more than three months and the number of cigarettes they smoke on average per day; participants also reported the number of years that they were smokers and the age of smoking initiation.

Based on these data, we defined mid-life smoking status as self-reported status (never, former, current smoker) at the CARDIA Year 25 follow-up examination (at the time of the MRI, mean age = 50) which we further verified against data from the seven earlier study exams; generally, self-reported smoking was consistent across study waves; we re-coded 39 participants (7%) who reported being never-smokers at Year 25 as former smokers based on reporting they were former or current smokers in previous exams.

For ever-smokers, we computed cumulative pack-years at Year 25 as follows: at each study visit, we estimated for each subject the number of pack-years by averaging the number of cigarettes smoked per day at that visit and the previous visit and dividing by the interval in years between visits. We summed all consecutive pack-years to obtain cumulative pack-years at the last follow-up (Year 25). Each pack-year represents exposure to 7300 cigarettes (1 year × 365 days × 1 pack/day × 20 cigarettes/pack). Age at smoking initiation was defined as the age at the first study wave when a participant reported being a former or current smoker. Years since smoking cessation were calculated for former smokers at Year 25 by counting the years since the last exam on which they last reported being regular smokers.

Covariates

All analyses included adjustment for a priori selected potential important demographic, vascular, respiratory, and psychological/behavioral factors that have been linked to brain health and smoking. Demographic variables included age, sex, race, educational attainment (defined as low – high school education or lower – versus high), and study center. Vascular risk factors included body mass index and physical activity (measured as composite scores computed based on self-reported frequency and intensity of physical activities)³⁰ assessed at the time of the MRI (Year 25) as well as history of vascular conditions: history of hypertension (defined as having diastolic blood pressure ≥ 90 and/or systolic blood pressure ≥ 140 and/or having reported taking antihypertensive medication at the Year 25 and/or prior study visits), history of cerebro/vascular disorders (self-reported occurrence of: heart attack, angina, heart failure, rheumatic heart disease, mitral valve prolapse, stroke and peripheral vascular disease), history of hypercholesterolemia (current or prior total cholesterol ≥ 240 mg/DL and/or use of cholesterol lowering medication), and history of diabetes mellitus (defined following ADA criteria for levels of fasting, non-fasting or postprandial oral glucose tolerance test results, Hemoglobin A1c percent, or use of anti-diabetes medication at the Year 25 and/or prior study visits). Analyses also included adjustment for history of respiratory illness (which was self-reported at Year 25 and included present or previous occurrence of asthma, tuberculosis, chronic bronchitis, chronic obstructive pulmonary disease, emphysema, or use of medication for asthma or for breathing problems). Behavioral/psychological covariates were assessed at Year 25 and included alcohol consumption (classified as abstinent, light-risk drinking, and high-risk drinking, based on sex-specific weekly maximum drinking limits published by the National Institute on Alcohol Abuse and Alcoholism³¹) and depressive symptoms (Center for Epidemiologic Studies-Depression CES-D scores³²).

Statistical analyses

There were 522 subjects with complete data on GM-CBF, smoking behavior, and covariates of interest. We assessed the relationships of mid-life smoking status with lobar and candidate regions’ GM-CBF values, using multivariable linear regression models. In secondary analyses restricted to former and current smokers, we investigated the separate relationships of each of cumulative pack-years, age at start of cigarette smoking, and years since cessation and each of the regional GM-CBF. We first investigated these characteristics in ever-smokers (combining former and current smokers) and then separately for current and former smokers as these two sub-groups may be different. To test the significance of this potential difference, we added an interaction term between the smoking characteristic of interest and current smoking status in linear regression analyses based on the ever-smokers sample. We assessed the shape of the associations of regional and CBF and continuous measures of smoking behavior (pack-years, years since cessation, and age at start of smoking) in former and current smokers using penalized splines³³ which helped us check the linearity of the associations. In sensitivity analyses, we adjusted the analyses for regional GM volume to control for potential influence of local GM atrophy.

All analyses tested two-tailed hypotheses with a significance level of 0.05. Analyses were completed using SAS version 9.3 (SAS Institute, Cary, NC).

Results

Compared to the other groups, current smokers at Year 25 had proportionally more black participants, participants with lower educational attainment, and participants with hypertension, respiratory illnesses, and alcohol use; they were also younger and had more depressive symptoms and lower physical activity levels (Table 1). Mean pack-years was 18.50 for current smokers and 7.24 for former smokers; participants started smoking in their late teens (mean age at smoking initiation = 16.43 years and 18.37 years, in smokers and former smokers, respectively). Former smokers quit on average 17.5 years prior to the MRI. The regional CBF measures in the total sample and across smoking groups are described in Supplemental Table S1.

Table 1.

Characteristics of the CARDA brain MRI sample with CBF data by smoking status at Year 25.

	Mean (SD) or n (%)
	Never smokers n = 267		Former smokers n = 171		Current smokers n = 84
	n (%)
Gender (Female)	137	(51.3%)	98	(57.3%)	41	(48.8%)
Race (Black)a	89	(33.3%)	51	(30%)	50	(59.5%)
Educational attainment (≤high school)a	27	(10.1%)	35	(20.5%)	38	(45.2%)
History of vascular diseaseb	25	(9.4%)	20	(11.7%)	9	(10.7%)
History of Hypertensiona	60	(22.5%)	53	(31%)	40	(47.6%)
History of diabetes	22	(8.2%)	14	(8.2%)	8	(9.5%)
History of high cholesterol	72	(27.0%)	44	(25.7%)	24	(28.6%)
History of respiratory illnessc	40	(15.0%)	35	(20.5%)	22	(26.2%)
Alcohol consumption
Non-user	100	(37.5%)	68	(39.8%)	24	(28.6%)
Low-risk drinking	133	(49.8%)	70	(40.4%)	31	(36.90%)
High-risk drinkinga	34	(12.7%)	33	(19.3%)	29	(34.5%)
	Mean (SD)
Age at MRI, yearsa	50.23	(3.58)	50.99	(3.16)	49.73	(3.39)
Body mass index, kg/m2	28.11	(5.16)	28.45	(5.67)	29.15	(5.96)
Physical activity, composite score of frequency and intensitya	415.47	(284.15)	405.87	(273.36)	307.24	(250.14)
Depressive symptoms, CES-Da	8.51	(6.48)	8.75	(6.81)	11.49	(8.63)
Smoking characteristics
Cumulative pack-yearsa	–	–	7.24	(9.48)	18.50	(11.68)
Age of smoking initiation, yearsa	–	–	18.37 d	(5.21)	16.43	(3.70)
Years since cessation, years	–	–	17.52	(11.03)	–	–

^aCharacteristics that are significantly different across the smoking categories (p < 0.05); p-values obtained using Chi-square tests for categorical variables and ANOVA tests for continuous variables.

^bHistory of vascular disease: heart attack, heart failure, angina, rheumatic heart disease, mitral valve prolapse, stroke, or peripheral vascular disease.

^cHistory of respiratory illness: asthma, tuberculosis, chronic bronchitis, chronic obstructive pulmonary disease, emphysema, or use of medication for asthma or for breathing problems.

^dn = 8 missing.

Compared to never-smokers, former smokers had significantly lower CBF in the parietal and occipital lobes (Coefficients = −3.76 and −3.02 mL/100 g/min, respectively) (Table 2). In contrast, no trend for a lower CBF was observed in current smokers and this group was not statistically different than never-smokers. A similar pattern was observed for the candidate regions’ analyses (Table 3): only former smokers had lower CBF, compared to never-smokers, with significantly reduced CBF in the cuneus, precuneus, putamen, and insula.

Table 2.

Smoking status at mid-life and lobar cerebral blood flow.

	Betaa	95% CI	p	Meanb	SE
Frontal CBF
Never-smoker	Ref.			57.82	0.71
Former smoker	−1.83	−4.05, 0.38	0.105	55.99	0.87
Current smoker	0.07	−3.03, 3.17	0.963	57.89	1.34
Temporal CBF
Never-smoker	Ref.			54.87	0.68
Former smoker	−1.48	−3.60, 0.65	0.172	53.40	0.83
Current smoker	1.11	−1.86, 4.07	0.463	55.98	1.28
Parietal CBF
Never-smoker	Ref.			59.39	0.84
Former smoker	−3.76	−6.39, −1.13	0.005	55.63	1.03
Current smoker	−0.40	−4.07, 3.28	0.830	59.00	1.59
Occipital CBF
Never-smoker	Ref.			59.71	0.81
Former smoker	−3.02	−5.55, −0.49	0.020	56.69	0.99
Current smoker	2.39	−1.15, 5.92	0.185	62.09	1.53

^aCoefficients adjusted for age, sex, race, educational attainment, study center, body mass index, history of high cholesterol, history of hypertension, history of diabetes, history of cerebro- and cardio-vascular events, history of respiratory disease, alcohol use, physical activity, and depressive symptoms; n = 522.

^bAdjusted means and standard errors of CBF values across smoking groups obtained from the multivariable regression models.

CBF: cerebral blood flow.

Table 3.

Smoking status and cerebral blood flow in candidate brain regions.

	Betaa	95% CI	p	Meanb	SE
Hippocampus CBF
Never-smoker	Ref.			49.55	0.66
Former smoker	−0.87	−2.92, 1.18	0.404	48.68	0.80
Current smoker	−0.27	−3.13, 2.59	0.854	49.28	1.24
Parahippocampal gyrus CBF
Never-smoker	Ref.			48.35	0.76
Former smoker	−0.29	−2.65, 2.08	0.773	48.06	0.92
Current smoker	−0.03	−3.33, 3.27	0.986	48.32	1.43
Entorhinal cortex CBF
Never-smoker	Ref.			45.02	1.10
Former smoker	1.58	−1.87, 5.04	0.368	46.60	1.35
Current smoker	3.42	−1.41, 8.24	0.165	48.43	2.09
Cuneus CBF
Never-smoker	Ref.			58.14	0.78
Former smoker	−2.96	−5.41, −0.51	0.018	55.18	0.96
Current smoker	2.04	−1.38, 5.47	0.241	60.18	1.48
Precuneus CBF
Never-smoker	Ref.			63.12	0.93
Former smoker	−3.54	−6.45, −0.63	0.017	59.58	1.14
Current smoker	−0.08	-4.14, 3.98	0.968	63.04	1.76
Putamen CBF
Never-smoker	Ref.			48.67	0.59
Former smoker	−3.32	−5.16, −1.48	<0.001	45.35	0.72
Current smoker	0.77	−1.80, 3.34	0.557	49.44	1.11
Insula CBF
Never-smoker	Ref.			53.40	0.71
Former smoker	−2.74	−4.95, −0.53	0.015	50.66	0.86
Current smoker	−0.10	−3.18, 2.99	0.951	53.31	1.33
Caudate CBF
Never-smoker	Ref.			38.72	0.54
Former smoker	−0.65	−2.35, 1.05	0.453	38.06	0.66
Current smoker	0.39	−1.99, 2.76	0.750	39.10	1.03
Thalamus CBF
Never-smoker	Ref.			48.31	0.92
Former smoker	−0.92	−3.77, 1.93	0.528	47.40	1.11
Current smoker	0.50	−3.48, 4.48	0.804	48.82	1.72

^aCoefficients adjusted for age, sex, race, educational attainment, study center, body mass index, history of high cholesterol, history of hypertension, history of diabetes, history of cerebro- and cardio-vascular events, history of respiratory disease, alcohol use, physical activity, and depressive symptoms; n = 522.

^bAdjusted means and standard errors of CBF values across smoking groups obtained from the multivariable regression models.

CBF: cerebral blood flow.

Smoking characteristics in former and current smokers and CBF

Among ever-smokers, the relationship between pack-years and CBF differed in current compared to former smokers (interaction terms of pack-years ×current smoking < 0.05 for most regional analyses; Tables 4 and 5). For current smokers, higher pack-years were associated with significantly higher CBF (p < 0.05) in the occipital and temporal lobes, the cuneus, putamen, insula, and hippocampus, and showed a trend of higher caudate CBF (p = 0.05) and parietal CBF (p = 0.08). There was no positive association between pack-years and higher CBF in former smokers; this group showed an overall trend for lower CBF with higher pack-years which reached statistical significance only in the caudate. Overall, there were no associations between age of initiation of smoking and years since cessation and regional CBF.

Table 4.

Smoking characteristics and lobar cerebral blood flow among former and current smokers.

	Current smokers n = 84			Former smokers n = 171
	Betaa	95% CI	p	Betaa	95% CI	p
Frontal CBF
Pack-years	0.18	−0.10, 0.467	0.214	−0.09	−0.29, 0.19	0.319
Age of smoking initiationb	0.06	−0.75, 0.88	0.877	0.17	−0.17, 0.50	0.329
Years since cessation	–	–	–	−0.04	−0.20, 0.13	0.665
Temporal CBF
Pack-years	0.31	0.04, 0.59	0.026	−0.12	−0.29, 0.05	0.161
Age of smoking initiationb	0.34	−0.48, 1.15	0.413	0.16	−0.12, 0.45	0.263
Years since cessation	–	–	–	−0.02	−0.16, 0.12	0.770
Occipital CBF
Pack-years	0.38	0.08, 0.68	0.015	−0.01	−0.22, 0.19	0.900
Age of smoking initiationb	0.15	−0.75, 1.06	0.740	0.04	−0.32, 0.40	0.824
Years since cessation	–	–	–	0.08	−0.09, 0.25	0.373
Parietal CBF
Pack-years	0.28	−0.03, 0.60	0.077	−0.04	−0.29, 0.20	0.717
Age of smoking initiationb	−0.001	−0.93, 0.92	0.998	0.06	−0.35, 0.48	0.762
Years since cessation	–	–	–	0.05	−0.15, 0.24	0.656

^aCoefficients adjusted for age, sex, race, educational attainment, study center, body mass index, history of high cholesterol, history of hypertension, history of diabetes, history of cerebro- and cardio-vascular events, history of respiratory disease, alcohol use, physical activity, and depressive symptoms.

^bOne hundred sixty-three former smokers with data on age of smoking initiation.

CBF: cerebral blood flow.

Table 5.

Smoking characteristics and cerebral blood flow in candidate brain regions among former and current smokers.

	Current smokers n = 84			Former smokers n = 171
	Betaa	95% CI	P	Betaa	95% CI	p
Cuneus CBF
Pack-years	0.47	0.17, 0.78	0.003	−0.03	−0.23, 0.17	0.772
Age of smoking initiationb	0.06	−0.86, 0.99	0.890	0.03	−0.31, 0.37	0.854
Years since cessation	–	–	–	0.10	−0.06, 0.26	0.203
Precuneus CBF
Pack-years	0.30	−0.06, 0.65	0.101	−0.17	−0.43, 0.09	0.197
Age of smoking initiationb	0.07	−0.97, 1.11	0.889	0.21	−0.24, 0.66	0.361
Years since cessation	–	–	–	0.10	−0.12, 0.32	0.356
Putamen CBF
Pack-years	0.31	0.07, 0.54	0.012	−0.08	−0.23, 0.07	0.302
Age of smoking initiationb	0.01	−0.69, 0.72	0.972	0.10	−0.17, 0.36	0.473
Years since cessation	–	–	–	0.07	−0.05, 0.20	0.245
Insula CBF
Pack-years	0.40	0.12, 0.67	0.006	−0.12	−0.31, 0.07	0.203
Age of smoking initiationb	−0.04	−0.88, 0.80	0.932	0.21	−0.11, 0.53	0.204
Years since cessation	–	–	–	0.09	−0.06, 0.25	0.223
Caudate CBF
Pack-years	0.20	−0.001, 0.41	0.051	−0.16	−0.31, −0.01	0.038
Age of smoking initiationb	0.13	−0.48, 0.74	0.675	0.17	−0.10, 0.43	0.212
Years since cessation	–	–	–	−0.001	−0.13, 0.13	0.992
Hippocampus CBF
Pack-years	0.26	0.002, 0.51	0.049	−0.08	−0.25, 0.09	0.352
Age of smoking initiationb	0.38	−0.37, 1.13	0.313	0.18	−0.11, 0.47	0.215
Years since cessation	–	–	–	−0.02	−0.16, 0.12	0.780
Parahippocampal gyrus CBF
Pack-years	0.17	−0.15, 0.49	0.300	−0.17	−0.39, 0.04	0.118
Age of smoking initiationb	0.42	−0.49, 1.35	0.356	0.38	0.02, 0.75	0.037
Years since cessation	–	–	–	−0.08	−0.26, 0.10	0.385
Entorhinal cortex CBF
Pack-years	0.18	−0.32, 0.68	0.474	−0.04	−0.37, 0.28	0.788
Age of smoking initiationb	0.49	−0.95, 1.94	0.500	0.37	−0.20, 0.94	0.200
Years since cessation	–	–	–	−0.24	−0.51, 0.03	0.079
Thalamus CBF
Pack-years	0.13	−0.20, 0.47	0.436	−0.15	−0.43, 0.14	0.308
Age of smoking initiationb	0.47	−0.50, 1.43	0.336	0.11	−0.38, 0.61	0.646
Years since cessation	–	–	–	0.04	−0.20, 0.27	0.751

^aCoefficients adjusted for age, sex, race, educational attainment, study center, body mass index, history of high cholesterol, history of hypertension, history of diabetes, history of cerebro- and cardio-vascular events, history of respiratory disease, alcohol use, physical activity, and depressive symptoms.

^bOne hundred sixty-three former smokers with data on age of smoking initiation.

CBF: cerebral blood flow.

We repeated all analyses adjusting for regional GM volume (Supplementary Tables S2 and S3) and results remained unaltered.

Discussion

In a community-based cohort of middle-aged adults, we found that former smokers had different patterns of CBF than current smokers. Compared to never-smokers, former smokers had lower CBF in the occipital and parietal lobes as well as in several candidate structures linked to cognitive disorders, the cuneus, precuneus, putamen, and insula; in contrast, current smokers did not have lower CBF. Further, among current smokers, a higher cumulative exposure to cigarette smoking was associated with significantly higher CBF levels in the occipital and temporal lobes, in the hippocampus, cuneus, and in deep cortex structures, notably the dorsal striatum and the insula.

There is accumulating evidence that smoking negatively affects brain organization and function. Several studies on macrostructural brain outcomes show that smoking is associated with smaller volumes in several brain regions.^12,15 Volume differences reported by these studies may partly reflect our findings of perfusion differences by smoking status, both widely in GM, and more specifically in certain regions in the brain which have been implicated in cognition and psychiatrically described behaviors. We note, however, that adjusting for GM volume did not alter our conclusions. Further, this study suggests that the associations of smoking with CBF can be already detected at a younger age where there is some, but not extensive cardiovascular disease and/or aging-related physiologic changes that may mask such associations.

Nicotine and tobacco smoke have been shown to have both vasoconstrictive and vasodilatory effects on the cerebrovasculature.³⁴ In general, the few studies that have investigated smoking and CBF have found that chronic cigarette smoking was associated with reduced global CBF.^12–14,35 The exact mechanisms by which cigarette smoking (whether through nicotine, smoke or other components) change CBF are not clear. The relationship between chronic smoking and reduced CBF has been suggested to arise due to several related pathways that can be damaged by smoking and smoking-induced oxidative stress including impaired endothelial function, nitric oxide vasodilatation, and neurovascular coupling, hypocapnia, and structural changes and damage in blood vessels.³⁶ Other data regarding more acute nicotine administration and cigarette smoking show conflicting results and associations with both increases and/or decreases in regional CBF.^35–38 Some of the discrepancy in results may be due to factors such as small sample size, smoking history (e.g. duration and extent of smoking and individual differences in nicotine dependence), different timing of the CBF measurements after the nicotine or cigarette consumption, dose of cigarette smoke or nicotine, the effects of other substances found in cigarettes, and differential reactions depending on whether participants were in abstinent or satiated states.^36,39,40

Our results in a community-based sample suggest differences in CBF between former and current smoking, which has not previously been reported. There are several possible explanations for this finding. The differences may reflect acute responses in current smokers who would have had a more recent exposure to tobacco smoking, which has been related to increases in CBF.^10,35,37 Similarly, increasing CBF with increasing pack-years among current smokers may reflect CBF fluctuations related to craving and smoking deprivation symptoms which have also been linked to increased CBF.^36,39,41 We did not have information on when was the time of the last cigarette smoked before the ASL-MRI scan. Future studies may benefit from including, at the time of the scan, information about the last cigarette consumption (i.e. the last time a cigarette was smoked and how many cigarettes were smoked).

However, it is also likely that the findings reflect chronic influences of smoking behavior and intervals between starting and stopping smoking. While decreases in CBF are often thought to indicate compromised brain function, increases in CBF are thought to represent cellular and vascular compensatory responses to pathological changes.^42–44 From this perspective, the dose–response relationship between pack-years and higher CBF observed in current smokers suggests a system that may be stressed and actively compensating with increased CBF.⁴⁵ In parallel, the reduced CBF in former smokers may be representing the longer term results of having experienced and over used these compensatory mechanisms.⁴⁵ This inverse pattern of the association between pack-years and CBF in former smokers was observed across all regions of interest but reached statistical significance only with the caudate CBF. Overall, the results suggest the hypothesis that relation of CBF to smoking may depend on when CBF is measured in the trajectory of CBF changes modulated by smoking status. The findings thus suggest the age at which a sample is studied, taking into account secular changes in smoking habits, may provide different views of the pathologic process. For example, other studies that have included older persons show lower CBF in current smokers.^13,14,46 It is possible that different physiological responses and compensatory mechanisms as well as different health and vascular disease statuses modify the relationship of CBF and smoking at different ages. It should be noted that these previous studies in older adults were conducted in the 1980s and used different technologies to measure blood flow and hence some methodological differences might also explain the difference in results.^13,14,46

Regionally, smoking status and extent were associated with the temporal, occipital, and parietal lobes, which is consistent with other data linking these regions with smoking behavior. For instance, parietal and occipital CBF have been linked to smoking, nicotine withdrawal, and cigarette craving scores.^36,37 Other studies have shown smoking is associated with smaller GM in the parietal, occipital, and temporal lobes^15,16 and with functional changes in the occipital, parietal, and temporal lobes.⁴⁷ We also found relationships with specific candidate regions of Alzheimer’s disease and dementia spread across the temporal lobe (hippocampus), occipital lobe (cuneus), parietal lobe (precuneus), and deep brain regions (putamen, caudate, and insula). One important characteristic common to these structures is their involvement in motor processes, inhibitory control, reward systems, and the management of habits and repeated stimuli.⁴⁸ Indeed, in addition to their links to cognitive disorders, these structures have been linked in structural and functional MRI studies to addiction^49–51 and compulsive disorders.^48,52–56 This interesting link between smoking and changes in perfusion in brain regions at the intersection of compulsive, reward, and cognitive networks warrants further investigation in longitudinal studies.

Our study has several strengths, including the community-based sample, the exploration of sub-group differences between former and current smokers, the investigation of different aspects of smoking behavior, and the assessment of several potential confounders (their association with GM CBF is described in Supplementary Table S4). Additionally, smoking status and cumulative smoking measures such as pack-years are prospectively collected data. However, there was only one MRI at the end of the follow-up and it is possible smokers and non-smokers have different brain characteristics preceding the MRI. Longitudinal studies with repeated CBF measures can confirm our findings of CBF differences between former and current smokers. Although our analyses adjusted for several important covariates, we cannot rule out the presence of residual or unmeasured confounding, such as the effects of other forms of nicotine use, stress, or earlier developmental factors, such as familial history of smoking and cardiovascular health. Because this was not a clinic based sample, we did not specifically ask about or control for some factors that have been shown to influence CBF measurements,⁵⁷ including blood gasses, carboxyhemoglobin, hematocrit levels, and food and caffeine restrictions. Some smokers could also have elevated hematocrit,⁵⁸ which lead to an underestimation of CBF. Our findings however do not show lower CBF in smokers, which is unlikely due to hematocrit effects but rather the extent of the CBF elevation in smokers could have been underestimated. Other ASL-related effects including labeling efficiency, motion differences, and arterial transit time could also influence CBF measurements. We found no systematic or large differences in labeling efficiency and motion parameters across smoking groups so these factors are not likely to generate a systematic bias in our results. CARDIA ASL MRI was part of a multimodal MRI protocol that did not allow for multiple post-labeling data to be acquired to enable measurement of arterial transit time; however, the post-labeling delay used in our study (1.5 s) is close to the arterial transit time for GM for these middle-aged subjects, and hence is unlikely to cause significant transit time artifacts.⁵⁹

In conclusion, in a community-based sample of middle-aged adults, cigarette smoking status and cumulative pack-years were associated with changes in CBF, an important early indicator of cerebrovascular pathology. These associations were observed in several brain regions linked to cognitive, compulsive, and addiction disorders, further strengthening the rationale for future research on changes in regional perfusion across the life-course as a mechanism underlying the relationship of smoking with dementia and brain aging. Our findings also suggest that past and ongoing smoking may be related differently to CBF. Longitudinal investigations are needed to assess whether higher CBF in smokers reflect a compensatory mechanism which may weaken in the future and to identify how and when the patterns of regional perfusion change with changes in smoking status throughout the lifespan.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The Coronary Artery Risk Development in Young Adults Study (CARDIA) is conducted and supported by the National Heart, Lung, and Blood Institute (NHLBI) in collaboration with the University of Alabama at Birmingham (HHSN268201300025C & HHSN268201300026C), Northwestern University (HHSN268201300027C), University of Minnesota (HHSN268201300028C), Kaiser Foundation Research Institute (HHSN268201300029C), and Johns Hopkins University School of Medicine (HHSN268200900041C). CARDIA is also partially supported by the Intramural Research Program of the National Institute on Aging (NIA) and an intra-agency agreement between NIA and NHLBI (AG0005). This manuscript has been reviewed by CARDIA for scientific content.

Declaration of conflicting interests

The author(s) declared the following potential conflicts of interest with respect to the research, authorship, and/or publication of this article: This work was initiated while DC Goff was at the Colorado School of Public Health. The views expressed in this article are those of the authors and do not necessarily represent the views of the National Heart, Lung, and Blood Institute; National Institutes of Health; or the United States Department of Health and Human Services. All other authors report no disclosures.

Authors’ contributions

ME and LL conceived and designed the study; ME, RA, SD, DJ, TH, DG, JD, CD, NB, and LL contributed to data acquisition, analysis, and interpretation; ME and LL drafted the manuscript; RA, SD, DJ, TH, DG, JD, CD, NB authors critically reviewed the manuscript for important intellectual content; ME, RA, SD, DJ, TH, DG, JD, CD, NB, and LL approved the manuscript for submission.

Supplementary material

Supplementary material for this paper can be found at the journal website: http://journals.sagepub.com/home/jcb