Abstract
The association of smoking with preclinical cognitive decline remains unclear and may be modified by the APOE ɛ4 genotype. In 5,705 participants (mean age: 63.9±9.1 years; 57.4% women) from the population-based Rotterdam Study, we investigated the relationship between smoking and cognitive decline over a 5.5-year period and examined potential effect modification by APOE ɛ4 genotype. We found that current smoking was related to decline in global cognition [difference compared to never smoking: –0.06 (95% C.I.–0.10;–0.01)], as well as decline on specific cognitive tests including the Letter Digit Substitution Task, the 15-Word Learning Test, and the Purdue Pegboard. We found no evidence for effect modification by APOE ɛ4 genotype on this relation.
INTRODUCTION
Vascular disease plays an important role in the etiology of dementia. Smoking is one of the most important risk factors of vascular disease, and has received considerable attention as a potentially modifiable risk factor for dementia [1–3]. Moreover, smoking has even been suggested to interact with one of the strongest risk factors for dementia, the APOE ɛ4 genotype, such that smoking increases the risk of dementia predominantly in individuals who do not carry the APOE ɛ4 allele [4]. However, some studies suggest smoking to be particularly detrimental to white matter integrity, memory, and learning in APOE ɛ4 carriers [5, 6]. These seemingly inconsistent findings may partly relate to the fact that compared to individuals without an APOE ɛ4 allele, carriers of this allele are also at substantially greater risk of mortality [7, 8], precluding a diagnosis of dementia. One strategy to bypass this potential bias is to investigate the early cognitive changes occurring in the long preclinical phase of dementia [9, 10]. We therefore investigated the effect of smoking, including potential effect modification of APOE ɛ4 genotype, on cognitive decline over a 5.5-year period in a large sample of community-dwelling, dementia-free elderly.
MATERIALS AND METHODS
Setting
This project was embedded within the Rotterdam Study, a prospective population-based cohort study in middle-aged and elderly persons that is ongoing since 1990 in Ommoord, a district of Rotterdam [11]. In 1990, 7,983 participants were enrolled. In 2000 and in 2006, the cohort was expanded by enrolling another 3,011 and 3,932 participants. Every 4 years, all participants are re-invited to undergo follow-up examinations at the Rotterdam Study research center. The Rotterdam Study has been approved by the Medical Ethics Committee of the Erasmus MC and by the Dutch Ministry of Health, Welfare and Sports, implementing the “Wet Bevolkings Onderzoek: ERGO (Population Screening Act: Rotterdam Study)”. All participants provided written informed consent to participate in the study.
Population for analyses
From 2002 onwards, an extensive cognitive test battery was implemented in the Rotterdam Study [11]. For this study, the first examination round for each participant that included this broad cognitive testing (2002–2008) was set as baseline. Of the 9,950 persons participating between 2002 and 2008, 8,041 persons that were free of dementia had complete information on smoking habits, APOE ɛ4 genotype, and cognitive performance. From these, 5,705 underwent another cognitive examination during the follow-up examinations between 2009 and 2013 (mean follow-up time: 5.6±0.6 years), comprising the population for our analyses.
Assessment of smoking
At baseline all participants were asked about their past and current smoking habits, including smoking of cigarettes, cigars, and pipes [12]. Smoking status was categorized into “current”, “former”, or “never”.
Assessment of APOE ɛ4 genotype
APOE genotype was determined using polymerase chain reaction on coded DNA samples in the baseline cohort and with a bi-allelic Tacqman assay (rs7412 and rs429358) in the two extensions of the Rotterdam Study [11]. Participants were coded as APOE ɛ4 positive (carrier of one or two APOE ɛ4 alleles) or as negative (non-carrier).
Assessment of cognition
At baseline and at the subsequent follow-up visit, participants underwent a cognitive examination that included the Letter Digit Substitution Task (LDST), Stroop test, Word Fluency Test (WFT), 15-Word Learning Test (15-WLT), and the Purdue Pegboard (PPB) [13]. Briefly, with the LDST we assessed processing speed and executive function. The Stroop test was used to measure the speed of reading (subtask 1), the speed of color naming (subtask 2), andinterference of automated processing and attention (subtask 3). The WFT was used to test efficiency of searching in long-term memory. The 15-WLT measured verbal learning (the immediate recall task), retrieval from verbal memory (the delayed recall task) and recognition of verbal stimuli (the recognition task). The PPB measured dexterity and fine motor skills. As a marker for general cognition, we calculated the g-factor according to a previously described method [13]. All test scores were converted into z-scores. Z-scores for the Stroop test were inverted because higher scores indicate worse performance, whereas higher scores on the other tests indicate better performance.
Other covariates
At baseline, information on relevant covariates was obtained by interview, physical examination, blood sampling, and medical records [11]. The covariates that we assessed were: level of education, body mass index, systolic and diastolic blood pressure, serum total cholesterol, serum high-density lipoprotein cholesterol, diabetes mellitus, and use of blood pressure lowering and lipid-lowering medication.
Statistical analysis
We investigated the relationship of smoking with global cognitive decline, and decline on the separate cognitive tests using multivariable linear regression analyses. We compared former smokers and current smokers with never smokers in all analyses. We adjusted all analyses for age, sex, educational level, inter-test interval, test score at baseline, cardiovascular risk factors, and APOE ɛ4 genotype. Next, we repeated these analyses while stratifying for APOE ɛ4 genotype to examine potential effect modification, and formally investigated statistical interaction of smoking with APOE ɛ4 genotype. Finally, we performed a sensitivity analysis in which we excluded all participants that became demented during the follow-up period (n = 43) to verify that the results were not driven by the participants that became demented.
Missing data in the covariates were handled using 5-fold multiple imputation, based on determinant, outcome and included covariates. All analyses were performed with IBM SPSS Statistics for Windows (IBM Corp, Armonk, NY, USA), Version 21.0.
RESULTS
Baseline characteristics of the study sample are summarized in Table 1. We found that current smoking was associated with accelerated decline in global cognition [difference in standardized g-factor score compared to never smoking: –0.06 (95% confidence interval (CI): –0.10;–0.01)] and cognitive decline on the LDST, 15-WLT immediate recall, 15-WLT delayed recall, 15-WLT recognition, and PPB, independent of cardiovascular risk factors and APOE ɛ4 genotype (Table 2). We found no associations between former smoking and cognition, except that former smokers tended to perform better on the PPB (Table 2).
Characteristics of the study sample
HDL, high-density lipoprotein. Values represent means (standard deviation) for continuous variables or percentages for categorical variables. Data represent original data without imputed values. Missing values were present for body mass index (1.4%), blood pressure (0.3%), total cholesterol (1.2%), HDL-cholesterol (1.2%), education level (0.9%), diabetes mellitus (0.5%), lipid lowering medication (0.4%), blood pressure lowering medication (0.4%). *Sample for analyses.
Influence of smoking on cognitive decline in the g-factor and in specific cognitive tests
LDST, Letter Digit Substitution Task; CWI, Color-Word Interference; WFT, Word Fluency Test; WLTimm, 15-Word Learning Test immediate recall; WLTdel, 15-Word Learning Test delayed recall; WLTrecog, 15-Word Learning Test recognition; PPB, Purdue Pegboard. Adjusted for age, sex, educational level, inter-test interval, test score at baseline, diabetes mellitus, body mass index, systolic blood pressure, diastolic blood pressure, serum total cholesterol, high-density lipoprotein cholesterol, use of blood pressure lowering medication, use of lipid-lowering medication, and APOE ɛ4 genotype.
After stratifying for APOE ɛ4 genotype, risk estimates were broadly similar for the associations between current smoking and the various cognitive tests (Fig. 1, Supplementary Table 1). We found no associations between former smoking and cognitive decline in either APOE ɛ4 carriers or non-carriers. When we repeated all analyses after exclusion ofparticipants that became demented during the follow-up period, the results did not change.
Influence of APOE ɛ4 genotype on the relationship between current smoking and cognitive decline. Associations between current smoking and cognitive decline stratified for APOE ɛ4 genotype. Adjusted for age, sex, educational level, inter-test interval, test score at baseline, diabetes mellitus, body mass index, systolic blood pressure, diastolic blood pressure, serum total cholesterol, high-density lipoprotein cholesterol, use of blood pressure lowering medication, and use of lipid-lowering medication. LDST, Letter Digit Substitution Task; CWI, Color-Word Interference; WFT, Word Fluency Test; WLTimm, 15-Word Learning Test immediate recall; WLTdel, 15-Word Learning Test delayed recall; WLTrecog; 15-Word Learning Test recognition; PPB, Purdue Pegboard.
DISCUSSION
In this large population-based study, we found that current smoking is associated with decline in global cognition and decline on specific cognitive tests including the LDST, the 15-WLT, and the PPB. We found no evidence for effect modification by APOE ɛ4 genotype on this relation.
The strengths of our study include the large study population, the longitudinal design and the neuropsychological assessment, which covered multiple cognitive domains. A potential limitation is the attrition of 29% between baseline testing and re-examination. On average, the persons included in the current analyses were younger, smoked less often, had a higher education and less often had diabetes than non-participants [14, 15].
We further established the detrimental effect of current smoking on the development of subclinical cognitive deterioration. Importantly, we also demonstrated that compared to never smoking, former smoking was not associated with a higher rate of cognitive decline. This supports a benefit of smoking cessation for prevention of cognitive decline, as was previously demonstrated for other diseases related to smoking, including lung cancer and cardiovascular disease [16, 17].
Previous evidence highlighted that the detrimental effect of smoking on the development of dementia was restricted to those individuals not carrying an APOE ɛ4 allele. However, when investigating cognitive decline, we found no evidence for a differential effect of smoking on the rate of cognitive decline according to APOE ɛ4 genotype. These findings suggesting that at least in the early preclinical phase of dementia, there is no interaction of smoking with APOE ɛ4 genotype.
Footnotes
ACKNOWLEDGMENTS
The dedication, commitment, and contribution of inhabitants, general practitioners, and pharmacists of the Ommoord district to the Rotterdam Study are gratefully acknowledged.
The Rotterdam Study is supported by the Erasmus MC and Erasmus University Rotterdam; the Netherlands Organisation for Scientific Research (NWO); the Netherlands Organisation for Health Research and Development (ZonMW); the Research Institute for Diseases in the Elderly (RIDE); the Netherlands Genomics Initiative; the Ministry of Education, Culture and Science; the Ministry of Health, Welfare and Sports; the European Commission (DG XII); and the Municipality of Rotterdam. None of the funders had any role in design and conduct of the study; collection, management, analysis, and interpretation of the data; and preparation, review, or approval of the manuscript.
