- Email: [email protected]
Cholesterol as a Risk Factor for Dementia and Cognitive Decline: A Systematic Review of Prospective Studies With Meta-Analysis
CLINICAL REVIEW
Cholesterol as a Risk Factor for Dementia and Cognitive Decline: A Systematic Review of Prospective Studies With Meta-Analysis Kaarin J. Anstey, Ph.D., Darren M. Lipnicki, Ph.D., Lee-Fay Low, Ph.D.
The relationships between total serum cholesterol (TC) and dementia and between TC and cognitive decline were investigated in a systematic review of 18 prospective studies. Follow-ups ranged from 3 to 29 years, and included a total of 14,331 participants evaluated for Alzheimer disease (AD), 9,458 participants evaluated for Vascular dementia (VaD), 1,893 participants evaluated for cognitive decline, and 4,793 participants evaluated for cognitive impairment. Compatible results were pooled using meta-analysis. Consistent associations between high midlife TC and increased risk of AD, and high midlife TC and increased risk of any dementia were found. There was no evidence supporting an association between late-life TC and AD, or between late-life TC and any dementia. No study reported a significant association between TC (measured in midlife or late-life) and VaD. An association between high midlife TC and cognitive impairment was found but there was only weak evidence for an association between TC and cognitive decline. Two of seven studies reporting data on the interaction between TC and apolipoprotein e4-allele had significant effects. Results suggest the effect of TC on dementia risk occurs in midlife but not late-life, and that there may be different cardiovascular risk factor profiles for AD and VaD. Results from additional studies involving long-term follow-up of midlife samples will allow for clarification of the association between age, TC and risk of specific types of dementia. These data are required to inform recommendations of modulation of cholesterol to reduce or delay dementia risk. (Am J Geriatr Psychiatry 2008; 16:343–354) Key Words: Alzheimer disease, vascular dementia, mild cognitive impairment, lipoprotein, APOE, systematic review
T
here is growing recognition that cognitive decline and dementia in late-life must be viewed
within a lifespan perspective. Prevention of dementia at the population level requires identification of
Received June 14, 2007; revised January 10, 2008; accepted January 16, 2008. From the Ageing Research Unit, Centre for Mental Health Research, Australian National University, Canberra ACT, Australia (KJA, DML); and Primary Dementia Collaborative Research Centre, University of New South Wales, Sydney NSW, Australia (L-FL). Send correspondence and reprint requests to Kaarin J. Anstey, Ph.D., Ageing Research Unit, Centre for Mental Health Research, Australian National University, Canberra ACT 0200, Australia. e-mail: [email protected] © 2008 American Association for Geriatric Psychiatry
Am J Geriatr Psychiatry 16:5, May 2008
343
TC as a Risk Factor for Dementia and Cognitive Decline modifiable risk factors, and of critical periods at which exposures may influence the course of cognitive development.1 Total serum cholesterol (TC) is implicated as one of several vascular risk factors for dementia.2 It is unclear whether associations between TC and age are similar for dementias with different aetiologies, for different rates of cognitive decline, or for cognitive impairment that do not meet diagnostic criteria.3 Also of interest is whether there are associations between the risk of dementia, cognitive decline or cognitive impairment, and serum lipids other than TC, and whether effects occur within specific subsamples, such as carriers of the apolipoprotein e4-allele (APOE*4).3 Methodological issues relating to study design are particularly pertinent when serum cholesterol is examined as a risk factor for dementia. Different levels of cholesterol between cohorts, and inclusion of subjects with comorbid conditions that lower cholesterol, may lead to inconsistencies in results.4 It also seems that the association between TC and cognitive decline may have a nonlinear association with age, with authors suggesting that high TC in midlife and low TC in late-life are associated with dementia and cognitive decline.5 This means that observations based on cross-sectional studies may be confounded by age differences within samples,6 and that prospective studies examining within-person change are optimal for uncovering the true nature of the association between serum cholesterol and cognition and dementia. If the association between serum cholesterol and dementia is nonlinear, with high levels in midlife increasing risk but low levels in late-life being a prodrome of dementia or Alzheimer disease (AD), as is the case with blood pressure,7 then participants with dementia at baseline need to be excluded from prospective studies. This criterion is reinforced by it having been established that the pathology underlying AD occurs decades before the onset of disease.8 Previous reviews on serum cholesterol have not excluded studies that did not screen for dementia at baseline, nor reported detailed search strategies, including how many articles were excluded on methodological grounds or for lacking data.3,5 In addressing these issues, we report here a systematic review designed to crystallize current knowledge concerning the role of serum lipids as risk factors for dementia,
344
cognitive decline, and cognitive impairment. Metaanalysis was also performed on studies that were comparable.
METHODS Literature Search The online databases PubMed (1950 to January 2007), PsycINFO (1872 to January 2007), and Cochrane CENTRAL (1800 to January 2007) were searched using “cholesterol” and a combination of keywords for cognition: “cognit*, memory, attention, reaction time, speed of processing, crystallized ability, crystallized intelligence, fluid ability, fluid intelligence, general mental ability, intelligence, executive function, neuropsychological testing, mini mental state examination, dementia, mild cognitive impairment (MCI), Alzheimer*” (where * indicates truncation). The search was limited to articles in English that reported data from humans. Inclusion and Exclusion Criteria Inclusion and exclusion criteria were similar to those previously used in a meta-analysis of smoking as a risk factor for dementia and cognitive decline.9 Studies were included if they 1) were prospective with at least 12 months follow-up; 2) analyzed a blood measure of cholesterol as the main variable of interest or as a covariate; 3) analyzed cognition or dementia as the dependent variable where cognition or dementia were assessed clinically (i.e., subjective assessments of cognition were not acceptable for inclusion, nor were dementia diagnoses obtained from medical records). Studies were excluded if they recruited a clinical sample or family members of participants with dementia, or did not assess cognition at baseline when the outcome was cognitive performance. Studies were excluded if they did not screen for dementia at baseline and participants’ age exceeded 60 years at baseline. For studies commencing in midlife where the upper age-range was not reported, an additional criteria was applied. Studies were excluded if their sample mean age plus two standard deviations of the sample mean age was greater than 60. Selection of relevant publications was conducted independently by the authors (K.A., D.L., L.F.L.) and
Am J Geriatr Psychiatry 16:5, May 2008
Anstey et al. any disagreements were resolved through discussions. Reference lists of selected publications were also searched for additional relevant materials. For multiple articles identified from a single study, preference was given to the publication with the longest follow-up period or the most comprehensive reporting of relevant data.
tiles, or normal versus high. Where multiple statistical models were reported, the result adjusted for the most covariates or with the smallest SE of estimates was selected. Studies that were compatible in terms of cholesterol measures, age-ranges and the intervals between cholesterol measurement and outcome assessment were included in meta-analyses.
Data Extraction
Data Analysis
Articles were de-identified (blinded title, author(s), year of publication, and journal name) before data extraction. The following information was extracted from each article and cross-checked by a second researcher; length of follow-up, description of cholesterol measures, description of cognitive or dementia measures, average age of participants, percentage of women, average years of education, and the relevant statistics describing the relationship between cholesterol and cognition or cholesterol and dementia. Authors were contacted for any missing information.
Three studies reporting data on AD, two studies reporting data on VaD and three studies reporting data on any dementia could be pooled in meta-analyses. For this review, relative risks, hazard ratios, and odds ratios were treated the same and are referred to as relative risks.9,14 The data points for the meta-analysis were the logarithms of the relative risks and their standard errors. The pooled effects are weighted such that the weight is inversely proportional to the SE of estimate of each study. Therefore, sample size is accounted for in the pooled effects since typically larger studies produce estimates with lower standard errors (higher precision) and are hence given larger weight. Fixed effect meta-analysis (inverse variance method) was used in all analyses as heterogeneity was not present.15
Data Compilation for Analysis Three potential dementia outcomes were addressed in the review including AD;10,11 Vascular dementia (VaD);12 and any dementia that potentially included AD, VaD, dementia with Lewy Bodies, frontotemporal dementia, Picks disease, alcohol related dementia, mixed and other dementias. The cognitive outcomes included a) cognitive performance at follow-up adjusted for baseline cognitive performance; b) cognitive performance change, which was the difference in cognitive performance between baseline and follow-up; c) cognitive decline where the sample was dichotomized into cognitive decliners and nondecliners; d) cognitive impairment, where the sample at follow-up was dichotomized into cognitively impaired and intact according to a cutoff; and e) MCI which was a clinical diagnosis at follow-up based on the Mayo Clinic Alzheimer Disease Research Centre Diagnosis.13 Midlife was defined as age 40 –59 and late-life as age 60 and older. Serum lipid measurements of TC, high density lipoprotein (HDL), low density lipoprotein (LDL), very low LDL, and triglycerides were reported in studies as a continuous variable, in terms of quar-
Am J Geriatr Psychiatry 16:5, May 2008
RESULTS Figure 1 shows the stages in obtaining studies for inclusion in the review. From 2,805 citations identified in the database search, 26 articles met our inclusion criteria. Of these, six did not report relevant data16 –21 and these data were not available from the authors. A further two articles22,23 were removed for representing multiple publications from a single study, leaving 18 articles with data that were available and appropriate for inclusion in the review. Characteristics of the 18 studies included in the review are shown in Table 1. Total Cholesterol and AD Eight studies were included that investigated an association between TC and risk of AD. These studies had a total of 14,331 participants, with follow-up length ranging from 4.80 to 29 (mean approx 13.34) years. Results are shown in Table 2. Of
345
TC as a Risk Factor for Dementia and Cognitive Decline
FIGURE 1.
Flowchart Describing the Process of Study Selection 2805 unique English language citations were identified by the search strategy 2706 citations were excluded as being irrelevant or for the abstract indicating failure to meet the inclusion criteria 99 articles were obtained for assessment against the inclusion criteria or to screen for additional potentially relevant citations 2 articles cited by others (and not identified by the search strategy) were obtained for assessment against the inclusion criteria 75 articles did not meet the inclusion criteria 26 articles met the inclusion criteria 6 articles did not report relevant data (which authors could not or did not subsequently supply); each of another 2 articles was displaced by an article that reported similar data from the same cohort with at least one of: a longer follow-up period, more dementia subtypes, analyses that adjusted for a greater number of covariates 18 articles were included in the review
four studies that recruited midlife participants at baseline, one (N ⫽ 1,026) found that TC averaged across mid- and late-life had no association with risk of AD.24 Two studies with samples of 1,291 and 444 found high cholesterol in midlife was associated with large increased risk of AD in late-life.25,26 The fourth study, with a sample of 1,027, found that a decrease in TC from midlife to late-life was associated with increased risk of AD in late-life.27 Five large studies with sample sizes ranging from 828 to 6,435 measured baseline TC in late-life. Three of these studies (pooled n: 599 AD and 10,195 Controls) were compatible for meta-analysis,28 –30 which revealed no association between TC measured in late-life, and AD. There was no difference in risk of AD between the first (lowest) quartile of TC and the
346
second (RR of 0.85, 95% CI: 0.67–1.10, z ⫽ 1.24, N ⫽ 5,544, p ⫽ 0.21), third (RR of 1.03, 95% CI: 0.79 –1.35, z ⫽ 0.25, N ⫽ 5,524, p ⫽ 0.80), or fourth (RR of 0.85, 95% CI: 0.65–1,12, z ⫽ 1.17, N ⫽ 5,526, p ⫽ 0.24) quartiles. Total Cholesterol and VaD Four studies were included that investigated an association between TC and risk of VaD. These studies had a total of 9,458 participants, with a follow-up length ranging from 4.80 to 29 (mean: 11.65) years. Results are shown in Table 3. The only study to include participants from midlife reported that changes in TC over time were unrelated to VaD risk.27 The other three studies similarly reported no
Am J Geriatr Psychiatry 16:5, May 2008
Source
Am J Geriatr Psychiatry 16:5, May 2008 5.6 (2.3)
Population based
Community-based HMO members Population based
Medicare recipients aged 65⫹ Medicare recipients aged 65⫹
North Karelia and FINMONICA
Adult Changes in Thought
Göteborg
Washington HeightsInwood Columbia Aging Project Northern Manhattan
Solfrizzi et al., Italian Longitudinal 200434 Study on Aging (1,445) Stewart et al., Honolulu-Asia Aging 200727 Study (1,027) Swan et al., National Heart, Lung, 199243 (12) and Blood Institute Twin Study Twins (war veterans)
Japanese American men
Electoral roll
Population based
4.8 (2.9)
Population based
North Karelia and FINMONICA
Rotterdam Study
7.1 (approx.)
Population based
MacArthur Studies of Successful Aging
5
29
3.5
5.8 (1.7)
18
5.6 (1.8)
21 (4.9)
21 (4.9)
3
4
36
Population based
Civil servants and municipal employees Population based
Observ. Period, Years (SD)
Zutphen Elderly Study
Iowa 65⫹ Rural Health Study
Israeli Ischemic Heart Disease Study
Study Name or Location
Descriptive Information for Included Studies
Beeri et al., 200431 (892) Hyman et al., 199638 (1,899) Kalmijn et al., 199636 (353) Karlamangla et al., 200452 (267) Kivepelto et al., 200153 (1,449) Kivipelto et al., 200225 (1,291) Li et al., 200528 (2,112) Mielke et al., 200532 (382) Reitz et al., 200429 (1,168) Reitz et al., 200537 (1,147) Slooter et al., 200030 (6,435)
Study (N)
TABLE 1.
Total; triglycerides
Total
Total; HDL
Total
Total; HDL
Total; HDL; Non-HDL; triglycerides LDL
Total; triglycerides
Total; HDL
Total
Total
Non-HDL
Total; HDL
Total
Total; HDL
Cholesterol Measure (s) Outcome and Outcome Measure (s)
71.2 (4) 61.8 8.68 AD (n ⫽ 48); VaD (n ⫽ 4); DSM-IV, NINCDS71.2 years (4) ADRDA 61.8 8.68 Dementia (n ⫽ 268); AD (n ⫽ 149); DSM-IV, 74.9 (5.9) NINCDS-ADRDA 59.5 13.7 (3) Dementia (n ⫽ 93); DSM-III-R 70 NA NA AD (n ⫽ 119); VaD (n ⫽ 54); NINCDS-ADRDA; 78.4 (6.2) dementia associated with stroke (VaD) 68.3 7.5 76.3 (5.8) Cognitive performance change; Memory, 68.4 cognitive/visuospatial and language factor 8.6 (4.6) scores 69.5a (9.1a) Dementia (n ⫽ 334); AD (n ⫽ 223); VaD (n ⫽ 48); Neurologic and neuropsychological 59.9a testing, neuroimaging, informant interviews, 24.1 with no more than physician reports primary levela 71.8 (5) MCI (n ⫽ 105); BSRT score in lowest 10th 56.4 percentile (with no dementia, normal MMSE 7.1 (4.7) score and ADL independence) Dementia (n ⫽ 56); AD (n ⫽ 31); VaD (n ⫽ 15); 80.2 (4.2) DSM-III-R, NINCDS-ADRDA, CADTC 0 10.8 (3.1) Cognitive decline (n ⫽ 6); Drop in digit symbol 55.2 test score of ⱖ6 points 0 12.9 (Continued)
MCI (⫽82); MCADRC
44.5 0% NA 79.1 (SEM ⫽ 0.1) Cognitive impairment; drop of 20 percentile 64.8 points (over 4 years), or score in lowest 10th 10.8 (SEM ⫽ 0.1) percentile on delayed recall task Cognitive decline (n ⫽ 51); drop in MMSE score 74.6 (4.2) ⬎2 points 0 24% with ⬎12 years 74 Cognitive decline (n ⫽ 37); drop in SPMSQ score to ⱕ6 between 1st (28 months) and 2nd 58.4 (⫹57 months) follow-up NA
Dementia (n ⫽ 309); TICS-m; DSM-IV
Mean Age, Years (SD) Female (%) Mean Years of Education
Anstey et al.
347
348
Total; Triglyceride; HDL; LDL; VLDL 7 Yoshitake et al., 199535 (828)
Japan
Population based
Total 3 Population based
Notes: Observ: observation; AD: Alzheimer disease; ADL: activities of daily living scale; BSRT: Babcock story recall Test; CADTC: California Alzheimer Disease and Treatment criteria; DSM-III-R: Diagnostic and Statistical Manual of Mental Disorders – Third Edition Revised; DSM-IV: Diagnostic and Statistical Manual of Mental Disorders – Fourth Edition; HDL: high-density lipoprotein fraction of cholesterol; HMO: health maintenance organization; LDL: low-density lipoprotein fraction of cholesterol; M: mean; MCADRC: Mayo Clinic Alzheimer Disease Research Center; MCI: mild cognitive impairment; MMSE: Mini-Mental State Examination; NA: not available; NINCDS-ADRDA: National Institute of Neurological and Communicative Diseases and Stroke/Alzheimer Disease and Related Disorders Association criteria; NINDS-AIREN: Neuroepidemiology Branch of the National Institute of Neurological Disorders and Stroke/Association Internationale pour la Recherche et l’Enseignement en Neurosciences criteria; SD: standard deviation; SEM: standard error of the mean; SPMSQ: Short Portable Mental Status Questionnaire; TICS-m: Modified Telephone Interview for Cognitive Status; VaD: Vascular Dementia; VLDL: very low-density lipoprotein fraction of cholesterol. a Values are for the cohort analyzed in a separate Rotterdam study article by Ott et al. (1998), of which Slooter et al. (2002) report data for 91% of participants.
76.1 (5.3) 63 NA Cognitive performance change; Change in MMSE 79.4 (3.4) score 57.6 NA 73.6 Dementia (n ⫽ 103); AD (n ⫽ 42); VaD 59.7 (n ⫽ 50); DSM-III-R, 26% low level (6 years) NINCDS-ADRDA, NINDS-AIREN
AD (n ⫽ 77); NINCDS-ADRDA Total; HDL 11⫹ Population based
Tan et al., Framingham 200324 (1,026) Wada et al., Japan 199742 (93)
Study Name or Location Study (N)
TABLE 1.
Continued
Source
Observ. Period, Years (SD)
Cholesterol Measure (s)
Outcome and Outcome Measure (s)
Mean Age, Years (SD) Female (%) Mean Years of Education
TC as a Risk Factor for Dementia and Cognitive Decline significant associations between VaD risk and TC. Meta-analyses of the two compatible studies (pooled n: 167 VaD and 8,327 Controls)29,30 did not reveal any significant associations between TC and VaD. There was no difference in risk of VaD between the first (lowest) quartile of TC and the second (RR of 1.19, 95% CI: 0.65, 2.17, z ⫽ 0.57, N ⫽ 4,251, p ⫽ 0.57), third (RR of 1.68, 95% CI: 0.88, 3.23, z ⫽ 1.57, N ⫽ 4,228, p ⫽ 0.12), or fourth (RR of 1.48, 95% CI: 0.79, 2.76, z ⫽ 1.23, N ⫽ 4,246, p ⫽ 0.22) quartiles. Total Cholesterol and any Dementia Six studies were included that investigated an association between TC and risk of any dementia. These studies had a total of 14,286 participants, with follow-up ranging from 5.6 to 36 (mean: 19.23) years. Results are shown in Table 3. Three studies took baseline measurements in midlife: one of these reported no association with risk of any dementia,31 whereas the other two reported that a decrease in TC from midlife was associated with risk of any dementia.25,27 Three studies with baseline measurements in late-life were compatible for meta-analysis (pooled n: 696 any dementia and 8,234 Controls).28,30,32 There was no difference in risk of any dementia between the first (lowest) quartile of TC and the second (RR ⫽ 1.00, 95% CI: 0.81, 1.24, z ⫽ 0.00, N ⫽ 4,407, p ⫽ 1.00), third (RR ⫽ 1.11, 95% CI: 0.88, 1.41, z ⫽ 0.87, N ⫽ 4,407, p ⫽ 0.38), or fourth (RR ⫽ 0.97, 95% CI: 0.77, 1.23, z ⫽ 0.22, N ⫽ 4,407, p ⫽ 0.82) quartiles. Total Cholesterol and Cognitive Decline or Impairment Table 4 shows the results of studies that investigated an association between TC and either cognitive decline or cognitive impairment. Five cognitive decline studies were included, comprising a total of 1,893 participants, with a follow-up ranging from 3 to 5.6 years. Two of these studies, with samples sizes of 93 and 12, found higher TC to be associated with a reduced risk of cognitive decline. The other three cognitive decline studies, with sample sizes of 1,147, 353, and 267, found no association between this outcome and TC. One study reported data on cognitive impairment (N ⫽ 1,899) on participants followed for 4 years. This study found that higher baseline TC in late-life was
Am J Geriatr Psychiatry 16:5, May 2008
Anstey et al.
TABLE 2.
Results for Total Cholesterol as a Risk Factor for Alzheimer Disease (AD) With Baseline Assessment in Mid-life and Late-life Follow-up (Years)
Study (N for AD/Control) Baseline midlife Kivipelto et al., 200225 (48/1270) Noktola et al. 199826 (27/397l) Tan et al., 200324 (77/959)
Stewart et al., 200727 (31/1027) Baseline late-life Li et al., 200528 (152/1868)
Ave. Age
Baseline Year/Age Mean age 50
71
Period
Effect
21
Hazard, Risk or Odds Ratio (95% CI), or Beta (p value)
Yes
OR: ⱖ6.5 vs ⬍6.5 mmol/l ⫽ 2.8 (1.2–6.7)a,b
NA
5 to 30
Yes
OR: ⱖ6.5 vs ⬍6.5 mmol/l ⫽ 3.1 h(1.2–8.5)a
Mean of first 15 measurements (1948–1978)
76.1
1988–2000
No
Recruited between 1965–1968, Mean age 50
80.2
1994–1996
Yes
HR: midlife: each 10 mg/dL increase ⫽ 0.95 (0.87–1.04)c HR: late life: each 10 mg/dL increase ⫽ 0.97 (0.90–1.05)c HR: change: 1.01 (0.92–1.11)c  ⫽ ⫺0.33, p ⫽ 0.03 (change in cholesterol over time,d fully adjusted for covariates, as a predictor of incident AD)e
Aged 65⫹
76.35
5.6
No No No Yes No Yes No No No No No
Age range 40–59
Reitz et al., 200429 (224/2226)
Mean age 78.4
NA
4.8
Slooter et al., 200030 (223/6101)
Mean age 69.5f
NA
5.8
Yoshitake et al., 199535 (42/725)
Mean age 73.6
NA
7
HR: Q2 vs Q1: 0.82 (0.51–1.31)c HR: Q3 vs Q1: 0.94 (0.58–1.50)c HR: Q4 vs Q1: 1.00 (0.61–1.62)c HR: Q2 vs Q1: 0.58 (0.34–0.97)c HR: Q3 vs Q1: 0.82 (0.48–1.41)c HR: Q4 vs Q1: 0.48 (0.26–0.86)c RR: Q2 vs Q1: 1.04 (0.73–1.47)c RR: Q3 vs Q1: 1.27 (0.84–1.90)c RR: Q4 vs Q1: 0.98 (0.66–1.45)c RR: Increase of 1 mmol/L: 0.99 (0.89–1.10)c RR: Increase of 1 SD: 1.10 (0.80–1.51)c
Notes: NA: data not available. OR: odds ratio; HR: hazard ratio; RR: relative risk; : the coefficient derived from linear random effects model. Statistical analyses: alogistic regression. b Multinomial regression. c Cox proportional hazards regression. d There was a greater decline in cholesterol level from baseline in subjects with AD at follow-up. e Random effects model. f Value is for the cohort analyzed in a separate Rotterdam study article by Ott et al. (1998), of which Slooter et al. (2002) report data for 91% of participants.
associated with reduced risk of cognitive impairment. Finally, two studies reported data on MCI. One reported that midlife TC as a risk factor for MCI, finding that high TC was associated with increased odds of MCI33 and the other showed that higher baseline TC in late-life was associated with reduced risk of MCI.34
Low Density Lipoprotein, Dementia, and Cognitive Decline The included studies reported no association between LDL and either AD29,35 or VaD.29,35 Data concerning an association between LDL and any dementia, cognitive decline or cognitive impairment, was not available.
High Density Lipoprotein, Dementia, and Cognitive Decline
Interactions With APOE
None of the included studies found HDL to be associated with AD,24,28,29,35 VaD,29,35 or any dementia.28,31 Similarly, HDL was not associated with either cognitive decline,36,37 or cognitive impairment.34
Seven of the included studies reported data on APOE*4, cholesterol, and dementia risk. One of these found that participants with both APOE*4 and high TC had higher risk of cognitive decline than partici-
Am J Geriatr Psychiatry 16:5, May 2008
349
TC as a Risk Factor for Dementia and Cognitive Decline
TABLE 3.
Results for Total Cholesterol as a Risk Factor for Vascular Dementia (VaD) With Baseline Assessment in Midlife and Late-Life
Study (N for VaD/Control) VaD baseline mid-life Stewart et al., 200727 (15/971) VaD baseline late-life Reitz et al., 200429 (119/2226) Slooter et al., 200030 (48/6101)
Yoshitake et al., 199535 (50/828) Any dementia, baseline assessment in midlife Beeri et al., 200431 (309/1583) Kivipelto et al., 200225 (57/1270) Stewart et al., 200727 (56//971) Any dementia, baseline assessment in late life Li et al., 200528 (268/1844)
Baseline Year/ Age
Follow-up (Years)
Hazard, Risk or Odds Ratio (95% CI), or Beta (p value)
Ave. Age
Period
Effect
1965–1968
80.2
1994–1996
No
 ⫽ ⫺.37, p ⫽ .24 (change in cholesterol over time,a unadjusted, as a predictor of VaD)b
Mean age 78.4
NA
4.8
Mean age 69.5d
NA
5.8
Mean age 73.6
NA
7
No No No No No No No No
HR: Q2 vs Q1: 0.78 (0.32–1.94)c HR: Q3 vs Q1: 1.57 (0.65–3.79)c HR: Q4 vs Q1: 1.05 (0.42–2.60)c RR: Q2 vs Q1: 1.67 (0.75–3.74)c RR: Q3 vs Q1: 1.83 (0.70–4.80)c RR: Q4 vs Q1: 2.00 (0.85–4.70)c RR: Each 1 mmol/L increase: 1.13 (0.90–1.41)c RR: Increase of 1 SD: 0.89 (0.66–1.21)c
Age range 40–65
82
36
No
OR: 1.00 (0.99–1.01)e
Mean age 50
71
21
Yes
1965–1968
80.2
1994–1996
Yes
OR: ⱖ6.5 vs ⬍6.5 mmol/l ⫽ 2.8 (1.2–6.7)e,f Any dementia (n ⫽ 57), Control (n ⫽ 1,270)  ⫽ ⫺0.39, p ⫽ 0.001 (change in cholesterol over time,a fully adjusted, as a predictor of all incident dementia)b
Aged 65⫹
76.35
5.6
No No No No No No No No No No
HR: Q2 vs Q1: 0.96 (0.68–1.38)c HR: Q3 vs Q1: 1.01 (0.70–1.46)c HR: Q4 vs Q1: 1.16 (0.81–1.67)c HR: Q2 vs Q1: 1.05 (0.48–2.29)c HR: Q3 vs Q1: 0.97 (0.43–2.17)c HR: Q4 vs Q1: 0.31 (0.11–0.85)c RR: Q2 vs Q1: 1.02 (0.77–1.36)c RR: Q3 vs Q1: 1.23 (0.88–1.72)c RR: Q4 vs Q1: 0.95 (0.69–1.31)c RR: Each 1 mmol/L increase: 0.97 (0.89–1.06)c
Mielke et al., 200532 (93/289)
Cohort aged 70
Slooter et al., 200030 (334/6101)
Mean age 69.5d
75, 79, 81, 83, 85, and 88 NA
18
5.8
Notes: NA: data not available; OR: odds ratio; HR: hazard ratio; RR: relative risk; : the coefficient derived from linear random effects model. a There was a greater decline in cholesterol level from baseline in subjects with dementia at follow-up. b Random effects model. c Cox proportional hazards regression. d Value is for the cohort analyzed in a separate Rotterdam study article by Ott et al. (1998), of which Slooter et al. (2002) report data for 91% of participants. e Logistic regression. f Multinomial regression.
pants with only one of these risk factors.36 Another study found that the association between APOE*4 and AD was weakened after adjusting for TC.26 Five studies24,25,28,30,38 did not find any interaction between APOE*4 and TC in relation to dementia risk. Studies Not Meeting Criteria That Were Included in Recent Reviews Three studies reported in recent reviews of cholesterol and dementia3,5 did not meet the criteria for in-
350
clusion in our review. One study was excluded because dementia diagnoses (including dementia, memory impairment, AD, VaD, and dementia not otherwise specified) were obtained from medical records.39 The other two studies were excluded because their sample contained participants age 60 or older who were not screened for dementia at baseline.40,41 One of these studies had an older baseline sample40 but the other had a midlife sample at baseline (mean age: 52.7, SD: 4.7) and 26-year follow-up. A large proportion of this
Am J Geriatr Psychiatry 16:5, May 2008
Anstey et al.
TABLE 4.
Results for Total Cholesterol as a Risk Factor for Cognitive Performance Change, Cognitive Impairment and Mild Cognitive Impairment Follow-up (Years)
Study Cognitive performance change (continuous) Reitz et al., 200537 (1147)
Wada et al., 199742 (93)
Cognitive performance change (dichotomous outcome) (N Decline/No decline) Swan et al., 199243 (6/6) Kalmijn et al., 199636 (51/302) Karlamangla et al., 200452 (37/230) Cognitive impairment Hyman et al., 199638 (NA) Mild cognitive impairment (N MCI/control) Kivipelto et al, 200153 (82/1270) Solfrizzi et al., 200434 (139/2824)
Baseline Year or Age
Age
Period
Effect
Statistic
Mean age 76.3
NA
5.6
Mean age 79.4
NA
3
No No No Yes
Beta (SE)a Memory: ⫺0.04 (0.8), p ⫽ 0.6b Cog./visuo.: 0.2 (0.3), p ⫽ 0.4b Language: ⫺0.1 (0.1), p ⫽ 0.6b Repeated measures ANOVA group (high cholesterol vs low cholesterol) by time on MMSE: p ⫽ 0.0178c
Mean age 55.2
NA
6
Yes
Mean age 74.6
NA
3
No
Decliners: 5.20 mmol/L vs. non-decliners: 5.36 mmol/L (p ⬍0.01)d OR: High: 1.38 (95% CI: 0.75–2.55)e,f
Mean age 74
NA
2.5 and 4.5
No
OR: each 10 mg/dL increase: 0.86 (95% CI: 0.73–1.02)e
Age 65⫹
79.1
4–7
Yes
OR: 0.997 (95% CI: 0.993–0.999)e
Mean age 50
71
21
Yes
Mean age 71.8
NA
OR: ⱖ6.5 vs ⬍6.5 mmol/l: 1.9 (95% CI: 1.2–3.0)e,g RR: Q2 vs Q1: 0.58 (95% CI: 0.33–0.98)h RR: Q3 vs Q1: 0.51 (95% CI: 0.28–0.77)h RR: Q4 vs Q1: 0.50 (95% CI: 0.28–0.87)h
3.5
Yes Yes Yes
Notes: SE: standard error; NA: data not available; CI: confidence interval; OR: odds ratio; RR: relative risk; Q1, Q2, Q3, Q4: the lowest, second lowest, third lowest and highest quartile of the total cholesterol measure. Statistical analyses: aBetas reported for Time ⫻ total cholesterol adjusted for age, sex, education, ethnicity, and APOE. b Generalized estimating equations. c Analysis of variance. d t-Tests were calculated from descriptive statistics reported by Stewart et al., 2007. e Logistic regression. f Odds ratios were calculated from descriptive statistics reported by Kalmijn et al. (1996). g Multinomial regression. h Poisson regression models.
sample (N ⫽ 3,734) would have been under age 60 and hence this study needs to be considered to avoid potential bias. This study found no association between midlife TC and AD, VaD, or any dementia.40
CONCLUSIONS The studies included in this review reported few associations between serum lipid levels and the risk of developing dementia. Nevertheless, within the reported associations high TC in midlife predicting AD was a consistent finding. Congruent with this was the finding of a decrease in TC from mid- to
Am J Geriatr Psychiatry 16:5, May 2008
late-life being associated with an increased risk of either AD or any dementia. These latter findings cast doubt over the generalizability of null results reported by studies in which the measure of TC was an average taken across many years.24 Such an approach may not be sensitive to an association between TC and dementia risk that changes over time. Meta-analyses of late-life TC in relation to AD, VaD, and any dementia did not reveal any significant associations, suggesting the possibility that the effect of TC on dementia risk may be limited to midlife. Our review contained no studies that found an association between TC and VaD. This could be due to the small number of studies that included VaD as an
351
TC as a Risk Factor for Dementia and Cognitive Decline outcome, or the lower prevalence of VaD. For example, whereas the Honolulu, HI-Asia Aging Study27 had 56 cases of AD, it had only 15 cases of VaD, limiting the statistical power to detect a significant association between TC and VaD. Alternatively, it is possible that TC does not contribute to the pathological mechanisms associated with VaD. If this latter conclusion is the case, then the results indicate a need to develop different cardiovascular risk factor profiles for AD and VaD. Five of the eight studies reporting on cognitive decline, cognitive impairment, and MCI found a significant association between the outcome and TC. High TC in late-life was reported to be associated with a decreased risk of both cognitive decline42 and cognitive impairment.34,38 However, there was no effect of late-life TC on cognitive decline found in studies with larger samples, leading to the conclusion that there remains a lack of reliable evidence supporting any association between high TC in latelife and cognitive decline. One extremely small study reported that high midlife TC was associated with a decreased risk of cognitive decline43 and another larger study found that high midlife TC increased the risk of MCI in late-life. No significant associations were found in any study reporting results for HDL or LDL. However, the small number of studies reporting relevant data prevents strong conclusions being drawn and it would be beneficial if future studies reported all available indices of serum cholesterol. The seven studies reporting data on the interaction between APOE*4 and TC were mostly consistent in revealing a lack of interaction. It is possible that this was due to the small number of participants in each individual study who were both APOE*4 positive and had potentially high risk cholesterol levels. There is also evidence that APOE*4 in conjunction with hypercholesterolemia may affect subtypes of dementia differently44 and hence analyses of any dementia as an outcome may dilute effects. The effect of APOE*4 and other risk factors may amplify any adverse effect of TC at critical periods in the life-course that may not be captured in studies with infrequent occasions of measurement. Studies with multiple occasions of measurement from mid to late-life27,45 therefore allow for the most accurate modeling of the relationships among risk factors and cognitive outcomes.
352
Limitations of systematic reviews include potential publication bias in favor of positive findings. The fact that we identified studies that did not focus on cholesterol, but reported results incidentally31,38 reduces this potential publication bias. Further, our search strategy identified more nonsignificant results than significant. The review was limited by the small number of comparable studies in any one category, such that we were unable to investigate test for publication bias. As this review is based on observational studies, it is possible that health and life style factors associated with TC explain the associations between TC and dementia and TC and cognitive decline.46 For example, high TC in midlife may result from poor nutrition, poorer general health, and be associated with less physical activity or frailty47,48 or be associated with depression in late-life.49 Although many studies adjusted for these factors, there was inconsistency among studies in the choice of covariates. The review is also limited by the lack of information on participants taking cholesterol lowering medication, which may confound the results. There is evidence that statin use is associated with lower dementia risk independent of any cholesterol lowering effects.50 Cultural and ethnic differences in life style and genetic factors may also have affected the observed associations. The strengths of the review included an a priori search strategy, duplicate study selection and data extraction, a comprehensive literature review, and the reporting of study characteristics and excluded studies.51 If declining TC precedes AD then in any study there may be individuals at different stages of the trajectory that leads to AD but who do meet diagnostic criteria. The only situation where this could be overcome would be in a longitudinal study where all participants have been followed up until death, and retrospective analysis is conducted with full information on final dementia outcomes. Although we excluded studies including older participants that did not screen for dementia at baseline, it is possible that some of the selected studies included individuals with preclinical dementia. It is also possible, but less likely, that those with primarily midlife samples at baseline included a very small number of participants with preclinical dementia. This review points to a need for further data from longitudinal studies involving several occasions of measurement from midlife to late adulthood, and for
Am J Geriatr Psychiatry 16:5, May 2008
Anstey et al. studies reporting the association between TC and subtypes of dementia. This will allow for further quantitative meta-analyses of this important association that will yield firm conclusions concerning the associations between age, serum lipids, risk of subtypes of dementia, and cognitive decline. Such research is required to inform decisions regarding the
modulation of cholesterol to prevent or delay specific types of dementia. The authors thank Chwee von Sanden and Richard Burns for their assistance. This work was supported by the National Health and Medical Research Council of Australia.
References 1. Whalley LJ, Dick FD, McNeill G: A life-course approach to the aetiology of late-onset dementias. Lancet Neurol 2006; 5:87–96 2. Panza F, D’Introno A, Colacicco AM, et al: Cognitive frailty: predementia syndrome and vascular risk factors. Neurobiol Aging 2006; 27:933–940 3. Panza F, D’Introno A, Colacicco AM, et al: Lipid metabolism in cognitive decline and dementia. Brain Res Rev 2006; 51:275–292 4. Launer LJ: The epidemiologic study of dementia: a life-long quest? Neurobiol Aging 2005; 26:335–340 5. Kivipelto M, Solomon A: Cholesterol as a risk factor for Alzheimer’s disease— epidemiological evidence. Acta Neurol Scand Suppl 2006; 185:50 –57 6. Hofer SM, Sliwinski MJ: Understanding ageing. An evaluation of research designs for assessing the interdependence of ageingrelated changes. Gerontology 2001; 47:341–352 7. Qiu C, Winblad B, Fratiglioni L: The age-dependent relation of blood pressure to cognitive function and dementia. Lancet Neurol 2005; 4:487– 499 8. Braak H, Braak E: Neuropathological stageing of Alzheimer-related changes. Acta Neuropathol 1991; 82:239 –259 9. Anstey KJ, von Sanden C, Salim A, et al: Smoking as a risk factor for dementia and cognitive decline: a meta-analysis of prospective studies. Am J Epidemiol 2007; 166:367–378 10. American Psychiatric Association. Diagnostic and Statistical Manual of Mental Disorders. 3rd Revised. Ed. Washington, DC, American Psychiatric Association, 1987 11. McKhann G, Drachman D, Folstein M, et al: Clinical diagnosis of Alzheimer’s disease: report of the NINCDS-ADRDA Work Group under the auspices of Department of Health and Human Services Task Force on Alzheimer’s Disease. Neurology 1984; 34:939 –944 12. Roman GC, Tatemichi TK, Erkinjuntti T, et al: Vascular dementia: diagnostic criteria for research studies. Report of the NINDSAIREN International Workshop. Neurology 1993; 43:250 –260 13. Petersen RC, Smith GE, Ivnik RJ, et al: Apolipoprotein E status as a predictor of the development of Alzheimer’s disease in memoryimpaired individuals. JAMA 1995; 273:1274 –1278 14. de Craen AJ, Gussekloo J, Vrijsen B, et al: Meta-analysis of nonsteroidal antiinflammatory drug use and risk of dementia. Am J Epidemiol 2005; 161:114 –120 15. Cochrane Handbook for Systematic Reviews of Interventions 4.2.5 [updated May 2005]. Vol Issue 3. Chichester, UK, John Wiley & Sons, 2005 16. Bernick C, Katz R, Smith NL, et al: Statins and cognitive function in the elderly: the Cardiovascular Health Study. Neurology 2005; 65:1388 –1394 17. Cervilla JA, Prince M, Mann A: Smoking, drinking, and incident cognitive impairment: a cohort community based study included in the Gospel Oak project. J Neurol Neurosurg Psychiatry 2000; 68:622– 626 18. Kalmijn S, Launer LJ, Ott A, et al: Dietary fat intake and the risk of
Am J Geriatr Psychiatry 16:5, May 2008
incident dementia in the Rotterdam Study. Ann Neurol 1997; 42:776 –782 19. Prince M, Lovestone S, Cervilla J, et al: The association between APOE and dementia does not seem to be mediated by vascular factors. Neurology 2000; 54:397– 402 20. Rea TD, Breitner JC, Psaty BM, et al: Statin use and the risk of incident dementia: the Cardiovascular Health Study. Arch Neurol 2005; 62:1047–1051 21. Zimetbaum P, Frishman WH, Ooi WL, et al: Plasma lipids and lipoproteins and the incidence of cardiovascular disease in the very elderly. The Bronx Aging Study. Arterioscler Thromb 1992; 12:416 – 423 22. Moroney JT, Tang MX, Berglund L, et al: Low-density lipoprotein cholesterol and the risk of dementia with stroke. JAMA 1999; 282:254 –260 23. Romas SN, Tang MX, Berglund L, et al: APOE genotype, plasma lipids, lipoproteins, and AD in community elderly. Neurology 1999; 53:517–521 24. Tan ZS, Seshadri S, Beiser A, et al: Plasma total cholesterol level as a risk factor for Alzheimer disease: the Framingham Study. Arch Intern Med 2003; 163:1053–1057 25. Kivipelto M, Helkala EL, Laakso MP, et al: Apolipoprotein E epsilon4 allele, elevated midlife total cholesterol level, and high midlife systolic blood pressure are independent risk factors for late-life Alzheimer disease. Ann Intern Med 2002; 137:149 –155 26. Notkola IL, Sulkava R, Pekkanen J, et al: Serum total cholesterol, apolipoprotein E epsilon 4 allele, and Alzheimer’s disease. Neuroepidemiology 1998; 17:14 –20 27. Stewart R, White LR, Xue QL, et al: Twenty-six-year change in total cholesterol levels and incident dementia: the Honolulu-Asia Aging Study. Arch Neurol 2007; 64:103–107 28. Li G, Shofer JB, Kukull WA, et al: Serum cholesterol and risk of Alzheimer disease: a community-based cohort study. Neurology 2005; 65:1045–1050 29. Reitz C, Tang MX, Luchsinger J, et al: Relation of plasma lipids to Alzheimer disease and vascular dementia. Arch Neurol 2004; 61:705–714 30. Slooter AJC, Ruitenberg A, van Duijn CM, et al: The effect of apoE on dementia is not through atherosclerosis: The Rotterdan Study: Reply. Neurology 2000; 54:2357–2358 31. Beeri MS, Rapp M, Silverman JM, et al: Coronary artery disease is associated with Alzheimer disease neuropathology in APOE4 carriers. Neurology 2006; 66:1399 –1404 32. Mielke MM, Zandi PP, Sjogren M, et al: High total cholesterol levels in late life associated with a reduced risk of dementia. Neurology 2005; 64:1689 –1695 33. Kivipelto M, Helkala EL, Laakso MP, et al: Midlife vascular risk factors and Alzheimer’s disease in later life: longitudinal, population based study. BMJ 2001; 322:1447–1451 34. Solfrizzi V, Panza F, Colacicco AM, et al: Vascular risk factors,
353
TC as a Risk Factor for Dementia and Cognitive Decline incidence of MCI, and rates of progression to dementia. Neurology 2004; 63:1882–1891 35. Yoshitake T, Kiyohara Y, Iwamoto H, et al: Incidence and risk factors of vascular dementia and Alzheimer’s disease in a defined elderly Japanese population: The Hisayama Study. Neurology 1995; 45:1161–1168 36. Kalmijn S, Feskens EJ, Launer LJ, et al: Cerebrovascular disease, the apolipoprotein e4 allele, and cognitive decline in a community-based study of elderly men. Stroke 1996; 27:2230 – 2235 37. Reitz C, Luchsinger J, Tang MX, et al: Impact of plasma lipids and time on memory performance in healthy elderly without dementia. Neurology 2005; 64:1378 –1383 38. Hyman BT, Gomez-Isla T, Briggs M, et al: Apolipoprotein E and cognitive change in an elderly population. Ann Neurol 1996; 40:55– 66 39. Whitmer RA, Sidney S, Selby J, et al: Midlife cardiovascular risk factors and risk of dementia in late life. Neurology 2005; 64:277– 281 40. Kalmijn S, Foley D, White L, et al: Metabolic cardiovascular syndrome and risk of dementia in Japanese-American elderly men. The Honolulu-Asia aging study. Arterioscler Thromb Vasc Biol 2000; 20:2255–2260 41. Kuusisto J, Koivisto K, Mykkanen L, et al: Association between features of the insulin resistance syndrome and Alzheimer’s disease independently of apolipoprotein E4 phenotype: cross sectional population based study. BMJ 1997; 315:1045–1049 42. Wada T, Matsubayashi K, Okumiya K, et al: Lower serum cholesterol level and later decline in cognitive function in older people living in the community, Japan. J Am Geriatr Soc 1997; 45:1411–1412 43. Swan GE, LaRue A, Carmelli D, et al: Decline in cognitive performance in aging twins. Heritability and biobehavioral predictors from the National Heart, Lung, and Blood Institute Twin Study. Arch Neurol 1992; 49:476 – 481
354
44. Borroni B, Grassi M, Costanzi C, et al: APOE genotype and cholesterol levels in lewy body dementia and Alzheimer disease: investigating genotype-phenotype effect on disease risk. Am J Geriatr Psychiatry 2006; 14:1022–1031 45. Solomon A, Kareholt I, Ngandu T, et al: Serum cholesterol changes after midlife and late-life cognition: twenty-one-year follow-up study. Neurology 2007; 68:751–756 46. Chiang CJ, Yip PK, Wu SC, et al: Midlife risk factors for subtypes of dementia: A Nested Case-Control Study in Taiwan. Am J Geriatr Psychiatry 2007; 15:762–771 47. Monastero R, Palmer K, Qiu C, et al: Heterogeneity in risk factors for cognitive impairment, no dementia: population-based longitudinal study from the Kungsholmen Project. Am J Geriatr Psychiatry 2007; 15:60 – 69 48. Zuliani G, Volpatol S, Romagnoni F, et al: Combined measurement of serum albumin and high-density lipoprotein cholesterol strongly predicts mortality in frail older nursing-home residents. Aging Clin Exp Res 2004; 16:472– 475 49. Almeida OP, Flicker L, Norman P, et al: Association of cardiovascular risk factors and disease with depression in later life. Am J Geriatr Psychiatry 2007; 15:506 –513 50. Jick H, Zornberg GL, Jick SS, et al: Statins and the risk of dementia. Lancet 2000; 356:1627–1631 51. Shea BJ, Grimshaw JM, Wells GA, et al: Development of AMSTAR: a measurement tool to assess the methodological quality of systematic reviews. BMC Med Res Methodol 2007; 7:10 52. Karlamangla AS, Singer BH, Reuben DB, et al: Increases in serum non-high-density lipoprotein cholesterol may be beneficial in some high-functioning older adults: MacArthur studies of successful aging.[see comment]. J Am Geriatr Soc 2004; 52:487– 494 53. Kivipelto M, Helkala EL, Hanninen T, et al: Midlife vascular risk factors and late-life mild cognitive impairment: a populationbased study. Neurology 2001; 56:1683–1689
Am J Geriatr Psychiatry 16:5, May 2008
Cholesterol as a Risk Factor for Dementia and Cognitive Decline: A Systematic Review of Prospective Studies With Meta-Analysis Kaarin J. Anstey, Ph.D., Darren M. Lipnicki, Ph.D., Lee-Fay Low, Ph.D.
The relationships between total serum cholesterol (TC) and dementia and between TC and cognitive decline were investigated in a systematic review of 18 prospective studies. Follow-ups ranged from 3 to 29 years, and included a total of 14,331 participants evaluated for Alzheimer disease (AD), 9,458 participants evaluated for Vascular dementia (VaD), 1,893 participants evaluated for cognitive decline, and 4,793 participants evaluated for cognitive impairment. Compatible results were pooled using meta-analysis. Consistent associations between high midlife TC and increased risk of AD, and high midlife TC and increased risk of any dementia were found. There was no evidence supporting an association between late-life TC and AD, or between late-life TC and any dementia. No study reported a significant association between TC (measured in midlife or late-life) and VaD. An association between high midlife TC and cognitive impairment was found but there was only weak evidence for an association between TC and cognitive decline. Two of seven studies reporting data on the interaction between TC and apolipoprotein e4-allele had significant effects. Results suggest the effect of TC on dementia risk occurs in midlife but not late-life, and that there may be different cardiovascular risk factor profiles for AD and VaD. Results from additional studies involving long-term follow-up of midlife samples will allow for clarification of the association between age, TC and risk of specific types of dementia. These data are required to inform recommendations of modulation of cholesterol to reduce or delay dementia risk. (Am J Geriatr Psychiatry 2008; 16:343–354) Key Words: Alzheimer disease, vascular dementia, mild cognitive impairment, lipoprotein, APOE, systematic review
T
here is growing recognition that cognitive decline and dementia in late-life must be viewed
within a lifespan perspective. Prevention of dementia at the population level requires identification of
Received June 14, 2007; revised January 10, 2008; accepted January 16, 2008. From the Ageing Research Unit, Centre for Mental Health Research, Australian National University, Canberra ACT, Australia (KJA, DML); and Primary Dementia Collaborative Research Centre, University of New South Wales, Sydney NSW, Australia (L-FL). Send correspondence and reprint requests to Kaarin J. Anstey, Ph.D., Ageing Research Unit, Centre for Mental Health Research, Australian National University, Canberra ACT 0200, Australia. e-mail: [email protected] © 2008 American Association for Geriatric Psychiatry
Am J Geriatr Psychiatry 16:5, May 2008
343
TC as a Risk Factor for Dementia and Cognitive Decline modifiable risk factors, and of critical periods at which exposures may influence the course of cognitive development.1 Total serum cholesterol (TC) is implicated as one of several vascular risk factors for dementia.2 It is unclear whether associations between TC and age are similar for dementias with different aetiologies, for different rates of cognitive decline, or for cognitive impairment that do not meet diagnostic criteria.3 Also of interest is whether there are associations between the risk of dementia, cognitive decline or cognitive impairment, and serum lipids other than TC, and whether effects occur within specific subsamples, such as carriers of the apolipoprotein e4-allele (APOE*4).3 Methodological issues relating to study design are particularly pertinent when serum cholesterol is examined as a risk factor for dementia. Different levels of cholesterol between cohorts, and inclusion of subjects with comorbid conditions that lower cholesterol, may lead to inconsistencies in results.4 It also seems that the association between TC and cognitive decline may have a nonlinear association with age, with authors suggesting that high TC in midlife and low TC in late-life are associated with dementia and cognitive decline.5 This means that observations based on cross-sectional studies may be confounded by age differences within samples,6 and that prospective studies examining within-person change are optimal for uncovering the true nature of the association between serum cholesterol and cognition and dementia. If the association between serum cholesterol and dementia is nonlinear, with high levels in midlife increasing risk but low levels in late-life being a prodrome of dementia or Alzheimer disease (AD), as is the case with blood pressure,7 then participants with dementia at baseline need to be excluded from prospective studies. This criterion is reinforced by it having been established that the pathology underlying AD occurs decades before the onset of disease.8 Previous reviews on serum cholesterol have not excluded studies that did not screen for dementia at baseline, nor reported detailed search strategies, including how many articles were excluded on methodological grounds or for lacking data.3,5 In addressing these issues, we report here a systematic review designed to crystallize current knowledge concerning the role of serum lipids as risk factors for dementia,
344
cognitive decline, and cognitive impairment. Metaanalysis was also performed on studies that were comparable.
METHODS Literature Search The online databases PubMed (1950 to January 2007), PsycINFO (1872 to January 2007), and Cochrane CENTRAL (1800 to January 2007) were searched using “cholesterol” and a combination of keywords for cognition: “cognit*, memory, attention, reaction time, speed of processing, crystallized ability, crystallized intelligence, fluid ability, fluid intelligence, general mental ability, intelligence, executive function, neuropsychological testing, mini mental state examination, dementia, mild cognitive impairment (MCI), Alzheimer*” (where * indicates truncation). The search was limited to articles in English that reported data from humans. Inclusion and Exclusion Criteria Inclusion and exclusion criteria were similar to those previously used in a meta-analysis of smoking as a risk factor for dementia and cognitive decline.9 Studies were included if they 1) were prospective with at least 12 months follow-up; 2) analyzed a blood measure of cholesterol as the main variable of interest or as a covariate; 3) analyzed cognition or dementia as the dependent variable where cognition or dementia were assessed clinically (i.e., subjective assessments of cognition were not acceptable for inclusion, nor were dementia diagnoses obtained from medical records). Studies were excluded if they recruited a clinical sample or family members of participants with dementia, or did not assess cognition at baseline when the outcome was cognitive performance. Studies were excluded if they did not screen for dementia at baseline and participants’ age exceeded 60 years at baseline. For studies commencing in midlife where the upper age-range was not reported, an additional criteria was applied. Studies were excluded if their sample mean age plus two standard deviations of the sample mean age was greater than 60. Selection of relevant publications was conducted independently by the authors (K.A., D.L., L.F.L.) and
Am J Geriatr Psychiatry 16:5, May 2008
Anstey et al. any disagreements were resolved through discussions. Reference lists of selected publications were also searched for additional relevant materials. For multiple articles identified from a single study, preference was given to the publication with the longest follow-up period or the most comprehensive reporting of relevant data.
tiles, or normal versus high. Where multiple statistical models were reported, the result adjusted for the most covariates or with the smallest SE of estimates was selected. Studies that were compatible in terms of cholesterol measures, age-ranges and the intervals between cholesterol measurement and outcome assessment were included in meta-analyses.
Data Extraction
Data Analysis
Articles were de-identified (blinded title, author(s), year of publication, and journal name) before data extraction. The following information was extracted from each article and cross-checked by a second researcher; length of follow-up, description of cholesterol measures, description of cognitive or dementia measures, average age of participants, percentage of women, average years of education, and the relevant statistics describing the relationship between cholesterol and cognition or cholesterol and dementia. Authors were contacted for any missing information.
Three studies reporting data on AD, two studies reporting data on VaD and three studies reporting data on any dementia could be pooled in meta-analyses. For this review, relative risks, hazard ratios, and odds ratios were treated the same and are referred to as relative risks.9,14 The data points for the meta-analysis were the logarithms of the relative risks and their standard errors. The pooled effects are weighted such that the weight is inversely proportional to the SE of estimate of each study. Therefore, sample size is accounted for in the pooled effects since typically larger studies produce estimates with lower standard errors (higher precision) and are hence given larger weight. Fixed effect meta-analysis (inverse variance method) was used in all analyses as heterogeneity was not present.15
Data Compilation for Analysis Three potential dementia outcomes were addressed in the review including AD;10,11 Vascular dementia (VaD);12 and any dementia that potentially included AD, VaD, dementia with Lewy Bodies, frontotemporal dementia, Picks disease, alcohol related dementia, mixed and other dementias. The cognitive outcomes included a) cognitive performance at follow-up adjusted for baseline cognitive performance; b) cognitive performance change, which was the difference in cognitive performance between baseline and follow-up; c) cognitive decline where the sample was dichotomized into cognitive decliners and nondecliners; d) cognitive impairment, where the sample at follow-up was dichotomized into cognitively impaired and intact according to a cutoff; and e) MCI which was a clinical diagnosis at follow-up based on the Mayo Clinic Alzheimer Disease Research Centre Diagnosis.13 Midlife was defined as age 40 –59 and late-life as age 60 and older. Serum lipid measurements of TC, high density lipoprotein (HDL), low density lipoprotein (LDL), very low LDL, and triglycerides were reported in studies as a continuous variable, in terms of quar-
Am J Geriatr Psychiatry 16:5, May 2008
RESULTS Figure 1 shows the stages in obtaining studies for inclusion in the review. From 2,805 citations identified in the database search, 26 articles met our inclusion criteria. Of these, six did not report relevant data16 –21 and these data were not available from the authors. A further two articles22,23 were removed for representing multiple publications from a single study, leaving 18 articles with data that were available and appropriate for inclusion in the review. Characteristics of the 18 studies included in the review are shown in Table 1. Total Cholesterol and AD Eight studies were included that investigated an association between TC and risk of AD. These studies had a total of 14,331 participants, with follow-up length ranging from 4.80 to 29 (mean approx 13.34) years. Results are shown in Table 2. Of
345
TC as a Risk Factor for Dementia and Cognitive Decline
FIGURE 1.
Flowchart Describing the Process of Study Selection 2805 unique English language citations were identified by the search strategy 2706 citations were excluded as being irrelevant or for the abstract indicating failure to meet the inclusion criteria 99 articles were obtained for assessment against the inclusion criteria or to screen for additional potentially relevant citations 2 articles cited by others (and not identified by the search strategy) were obtained for assessment against the inclusion criteria 75 articles did not meet the inclusion criteria 26 articles met the inclusion criteria 6 articles did not report relevant data (which authors could not or did not subsequently supply); each of another 2 articles was displaced by an article that reported similar data from the same cohort with at least one of: a longer follow-up period, more dementia subtypes, analyses that adjusted for a greater number of covariates 18 articles were included in the review
four studies that recruited midlife participants at baseline, one (N ⫽ 1,026) found that TC averaged across mid- and late-life had no association with risk of AD.24 Two studies with samples of 1,291 and 444 found high cholesterol in midlife was associated with large increased risk of AD in late-life.25,26 The fourth study, with a sample of 1,027, found that a decrease in TC from midlife to late-life was associated with increased risk of AD in late-life.27 Five large studies with sample sizes ranging from 828 to 6,435 measured baseline TC in late-life. Three of these studies (pooled n: 599 AD and 10,195 Controls) were compatible for meta-analysis,28 –30 which revealed no association between TC measured in late-life, and AD. There was no difference in risk of AD between the first (lowest) quartile of TC and the
346
second (RR of 0.85, 95% CI: 0.67–1.10, z ⫽ 1.24, N ⫽ 5,544, p ⫽ 0.21), third (RR of 1.03, 95% CI: 0.79 –1.35, z ⫽ 0.25, N ⫽ 5,524, p ⫽ 0.80), or fourth (RR of 0.85, 95% CI: 0.65–1,12, z ⫽ 1.17, N ⫽ 5,526, p ⫽ 0.24) quartiles. Total Cholesterol and VaD Four studies were included that investigated an association between TC and risk of VaD. These studies had a total of 9,458 participants, with a follow-up length ranging from 4.80 to 29 (mean: 11.65) years. Results are shown in Table 3. The only study to include participants from midlife reported that changes in TC over time were unrelated to VaD risk.27 The other three studies similarly reported no
Am J Geriatr Psychiatry 16:5, May 2008
Source
Am J Geriatr Psychiatry 16:5, May 2008 5.6 (2.3)
Population based
Community-based HMO members Population based
Medicare recipients aged 65⫹ Medicare recipients aged 65⫹
North Karelia and FINMONICA
Adult Changes in Thought
Göteborg
Washington HeightsInwood Columbia Aging Project Northern Manhattan
Solfrizzi et al., Italian Longitudinal 200434 Study on Aging (1,445) Stewart et al., Honolulu-Asia Aging 200727 Study (1,027) Swan et al., National Heart, Lung, 199243 (12) and Blood Institute Twin Study Twins (war veterans)
Japanese American men
Electoral roll
Population based
4.8 (2.9)
Population based
North Karelia and FINMONICA
Rotterdam Study
7.1 (approx.)
Population based
MacArthur Studies of Successful Aging
5
29
3.5
5.8 (1.7)
18
5.6 (1.8)
21 (4.9)
21 (4.9)
3
4
36
Population based
Civil servants and municipal employees Population based
Observ. Period, Years (SD)
Zutphen Elderly Study
Iowa 65⫹ Rural Health Study
Israeli Ischemic Heart Disease Study
Study Name or Location
Descriptive Information for Included Studies
Beeri et al., 200431 (892) Hyman et al., 199638 (1,899) Kalmijn et al., 199636 (353) Karlamangla et al., 200452 (267) Kivepelto et al., 200153 (1,449) Kivipelto et al., 200225 (1,291) Li et al., 200528 (2,112) Mielke et al., 200532 (382) Reitz et al., 200429 (1,168) Reitz et al., 200537 (1,147) Slooter et al., 200030 (6,435)
Study (N)
TABLE 1.
Total; triglycerides
Total
Total; HDL
Total
Total; HDL
Total; HDL; Non-HDL; triglycerides LDL
Total; triglycerides
Total; HDL
Total
Total
Non-HDL
Total; HDL
Total
Total; HDL
Cholesterol Measure (s) Outcome and Outcome Measure (s)
71.2 (4) 61.8 8.68 AD (n ⫽ 48); VaD (n ⫽ 4); DSM-IV, NINCDS71.2 years (4) ADRDA 61.8 8.68 Dementia (n ⫽ 268); AD (n ⫽ 149); DSM-IV, 74.9 (5.9) NINCDS-ADRDA 59.5 13.7 (3) Dementia (n ⫽ 93); DSM-III-R 70 NA NA AD (n ⫽ 119); VaD (n ⫽ 54); NINCDS-ADRDA; 78.4 (6.2) dementia associated with stroke (VaD) 68.3 7.5 76.3 (5.8) Cognitive performance change; Memory, 68.4 cognitive/visuospatial and language factor 8.6 (4.6) scores 69.5a (9.1a) Dementia (n ⫽ 334); AD (n ⫽ 223); VaD (n ⫽ 48); Neurologic and neuropsychological 59.9a testing, neuroimaging, informant interviews, 24.1 with no more than physician reports primary levela 71.8 (5) MCI (n ⫽ 105); BSRT score in lowest 10th 56.4 percentile (with no dementia, normal MMSE 7.1 (4.7) score and ADL independence) Dementia (n ⫽ 56); AD (n ⫽ 31); VaD (n ⫽ 15); 80.2 (4.2) DSM-III-R, NINCDS-ADRDA, CADTC 0 10.8 (3.1) Cognitive decline (n ⫽ 6); Drop in digit symbol 55.2 test score of ⱖ6 points 0 12.9 (Continued)
MCI (⫽82); MCADRC
44.5 0% NA 79.1 (SEM ⫽ 0.1) Cognitive impairment; drop of 20 percentile 64.8 points (over 4 years), or score in lowest 10th 10.8 (SEM ⫽ 0.1) percentile on delayed recall task Cognitive decline (n ⫽ 51); drop in MMSE score 74.6 (4.2) ⬎2 points 0 24% with ⬎12 years 74 Cognitive decline (n ⫽ 37); drop in SPMSQ score to ⱕ6 between 1st (28 months) and 2nd 58.4 (⫹57 months) follow-up NA
Dementia (n ⫽ 309); TICS-m; DSM-IV
Mean Age, Years (SD) Female (%) Mean Years of Education
Anstey et al.
347
348
Total; Triglyceride; HDL; LDL; VLDL 7 Yoshitake et al., 199535 (828)
Japan
Population based
Total 3 Population based
Notes: Observ: observation; AD: Alzheimer disease; ADL: activities of daily living scale; BSRT: Babcock story recall Test; CADTC: California Alzheimer Disease and Treatment criteria; DSM-III-R: Diagnostic and Statistical Manual of Mental Disorders – Third Edition Revised; DSM-IV: Diagnostic and Statistical Manual of Mental Disorders – Fourth Edition; HDL: high-density lipoprotein fraction of cholesterol; HMO: health maintenance organization; LDL: low-density lipoprotein fraction of cholesterol; M: mean; MCADRC: Mayo Clinic Alzheimer Disease Research Center; MCI: mild cognitive impairment; MMSE: Mini-Mental State Examination; NA: not available; NINCDS-ADRDA: National Institute of Neurological and Communicative Diseases and Stroke/Alzheimer Disease and Related Disorders Association criteria; NINDS-AIREN: Neuroepidemiology Branch of the National Institute of Neurological Disorders and Stroke/Association Internationale pour la Recherche et l’Enseignement en Neurosciences criteria; SD: standard deviation; SEM: standard error of the mean; SPMSQ: Short Portable Mental Status Questionnaire; TICS-m: Modified Telephone Interview for Cognitive Status; VaD: Vascular Dementia; VLDL: very low-density lipoprotein fraction of cholesterol. a Values are for the cohort analyzed in a separate Rotterdam study article by Ott et al. (1998), of which Slooter et al. (2002) report data for 91% of participants.
76.1 (5.3) 63 NA Cognitive performance change; Change in MMSE 79.4 (3.4) score 57.6 NA 73.6 Dementia (n ⫽ 103); AD (n ⫽ 42); VaD 59.7 (n ⫽ 50); DSM-III-R, 26% low level (6 years) NINCDS-ADRDA, NINDS-AIREN
AD (n ⫽ 77); NINCDS-ADRDA Total; HDL 11⫹ Population based
Tan et al., Framingham 200324 (1,026) Wada et al., Japan 199742 (93)
Study Name or Location Study (N)
TABLE 1.
Continued
Source
Observ. Period, Years (SD)
Cholesterol Measure (s)
Outcome and Outcome Measure (s)
Mean Age, Years (SD) Female (%) Mean Years of Education
TC as a Risk Factor for Dementia and Cognitive Decline significant associations between VaD risk and TC. Meta-analyses of the two compatible studies (pooled n: 167 VaD and 8,327 Controls)29,30 did not reveal any significant associations between TC and VaD. There was no difference in risk of VaD between the first (lowest) quartile of TC and the second (RR of 1.19, 95% CI: 0.65, 2.17, z ⫽ 0.57, N ⫽ 4,251, p ⫽ 0.57), third (RR of 1.68, 95% CI: 0.88, 3.23, z ⫽ 1.57, N ⫽ 4,228, p ⫽ 0.12), or fourth (RR of 1.48, 95% CI: 0.79, 2.76, z ⫽ 1.23, N ⫽ 4,246, p ⫽ 0.22) quartiles. Total Cholesterol and any Dementia Six studies were included that investigated an association between TC and risk of any dementia. These studies had a total of 14,286 participants, with follow-up ranging from 5.6 to 36 (mean: 19.23) years. Results are shown in Table 3. Three studies took baseline measurements in midlife: one of these reported no association with risk of any dementia,31 whereas the other two reported that a decrease in TC from midlife was associated with risk of any dementia.25,27 Three studies with baseline measurements in late-life were compatible for meta-analysis (pooled n: 696 any dementia and 8,234 Controls).28,30,32 There was no difference in risk of any dementia between the first (lowest) quartile of TC and the second (RR ⫽ 1.00, 95% CI: 0.81, 1.24, z ⫽ 0.00, N ⫽ 4,407, p ⫽ 1.00), third (RR ⫽ 1.11, 95% CI: 0.88, 1.41, z ⫽ 0.87, N ⫽ 4,407, p ⫽ 0.38), or fourth (RR ⫽ 0.97, 95% CI: 0.77, 1.23, z ⫽ 0.22, N ⫽ 4,407, p ⫽ 0.82) quartiles. Total Cholesterol and Cognitive Decline or Impairment Table 4 shows the results of studies that investigated an association between TC and either cognitive decline or cognitive impairment. Five cognitive decline studies were included, comprising a total of 1,893 participants, with a follow-up ranging from 3 to 5.6 years. Two of these studies, with samples sizes of 93 and 12, found higher TC to be associated with a reduced risk of cognitive decline. The other three cognitive decline studies, with sample sizes of 1,147, 353, and 267, found no association between this outcome and TC. One study reported data on cognitive impairment (N ⫽ 1,899) on participants followed for 4 years. This study found that higher baseline TC in late-life was
Am J Geriatr Psychiatry 16:5, May 2008
Anstey et al.
TABLE 2.
Results for Total Cholesterol as a Risk Factor for Alzheimer Disease (AD) With Baseline Assessment in Mid-life and Late-life Follow-up (Years)
Study (N for AD/Control) Baseline midlife Kivipelto et al., 200225 (48/1270) Noktola et al. 199826 (27/397l) Tan et al., 200324 (77/959)
Stewart et al., 200727 (31/1027) Baseline late-life Li et al., 200528 (152/1868)
Ave. Age
Baseline Year/Age Mean age 50
71
Period
Effect
21
Hazard, Risk or Odds Ratio (95% CI), or Beta (p value)
Yes
OR: ⱖ6.5 vs ⬍6.5 mmol/l ⫽ 2.8 (1.2–6.7)a,b
NA
5 to 30
Yes
OR: ⱖ6.5 vs ⬍6.5 mmol/l ⫽ 3.1 h(1.2–8.5)a
Mean of first 15 measurements (1948–1978)
76.1
1988–2000
No
Recruited between 1965–1968, Mean age 50
80.2
1994–1996
Yes
HR: midlife: each 10 mg/dL increase ⫽ 0.95 (0.87–1.04)c HR: late life: each 10 mg/dL increase ⫽ 0.97 (0.90–1.05)c HR: change: 1.01 (0.92–1.11)c  ⫽ ⫺0.33, p ⫽ 0.03 (change in cholesterol over time,d fully adjusted for covariates, as a predictor of incident AD)e
Aged 65⫹
76.35
5.6
No No No Yes No Yes No No No No No
Age range 40–59
Reitz et al., 200429 (224/2226)
Mean age 78.4
NA
4.8
Slooter et al., 200030 (223/6101)
Mean age 69.5f
NA
5.8
Yoshitake et al., 199535 (42/725)
Mean age 73.6
NA
7
HR: Q2 vs Q1: 0.82 (0.51–1.31)c HR: Q3 vs Q1: 0.94 (0.58–1.50)c HR: Q4 vs Q1: 1.00 (0.61–1.62)c HR: Q2 vs Q1: 0.58 (0.34–0.97)c HR: Q3 vs Q1: 0.82 (0.48–1.41)c HR: Q4 vs Q1: 0.48 (0.26–0.86)c RR: Q2 vs Q1: 1.04 (0.73–1.47)c RR: Q3 vs Q1: 1.27 (0.84–1.90)c RR: Q4 vs Q1: 0.98 (0.66–1.45)c RR: Increase of 1 mmol/L: 0.99 (0.89–1.10)c RR: Increase of 1 SD: 1.10 (0.80–1.51)c
Notes: NA: data not available. OR: odds ratio; HR: hazard ratio; RR: relative risk; : the coefficient derived from linear random effects model. Statistical analyses: alogistic regression. b Multinomial regression. c Cox proportional hazards regression. d There was a greater decline in cholesterol level from baseline in subjects with AD at follow-up. e Random effects model. f Value is for the cohort analyzed in a separate Rotterdam study article by Ott et al. (1998), of which Slooter et al. (2002) report data for 91% of participants.
associated with reduced risk of cognitive impairment. Finally, two studies reported data on MCI. One reported that midlife TC as a risk factor for MCI, finding that high TC was associated with increased odds of MCI33 and the other showed that higher baseline TC in late-life was associated with reduced risk of MCI.34
Low Density Lipoprotein, Dementia, and Cognitive Decline The included studies reported no association between LDL and either AD29,35 or VaD.29,35 Data concerning an association between LDL and any dementia, cognitive decline or cognitive impairment, was not available.
High Density Lipoprotein, Dementia, and Cognitive Decline
Interactions With APOE
None of the included studies found HDL to be associated with AD,24,28,29,35 VaD,29,35 or any dementia.28,31 Similarly, HDL was not associated with either cognitive decline,36,37 or cognitive impairment.34
Seven of the included studies reported data on APOE*4, cholesterol, and dementia risk. One of these found that participants with both APOE*4 and high TC had higher risk of cognitive decline than partici-
Am J Geriatr Psychiatry 16:5, May 2008
349
TC as a Risk Factor for Dementia and Cognitive Decline
TABLE 3.
Results for Total Cholesterol as a Risk Factor for Vascular Dementia (VaD) With Baseline Assessment in Midlife and Late-Life
Study (N for VaD/Control) VaD baseline mid-life Stewart et al., 200727 (15/971) VaD baseline late-life Reitz et al., 200429 (119/2226) Slooter et al., 200030 (48/6101)
Yoshitake et al., 199535 (50/828) Any dementia, baseline assessment in midlife Beeri et al., 200431 (309/1583) Kivipelto et al., 200225 (57/1270) Stewart et al., 200727 (56//971) Any dementia, baseline assessment in late life Li et al., 200528 (268/1844)
Baseline Year/ Age
Follow-up (Years)
Hazard, Risk or Odds Ratio (95% CI), or Beta (p value)
Ave. Age
Period
Effect
1965–1968
80.2
1994–1996
No
 ⫽ ⫺.37, p ⫽ .24 (change in cholesterol over time,a unadjusted, as a predictor of VaD)b
Mean age 78.4
NA
4.8
Mean age 69.5d
NA
5.8
Mean age 73.6
NA
7
No No No No No No No No
HR: Q2 vs Q1: 0.78 (0.32–1.94)c HR: Q3 vs Q1: 1.57 (0.65–3.79)c HR: Q4 vs Q1: 1.05 (0.42–2.60)c RR: Q2 vs Q1: 1.67 (0.75–3.74)c RR: Q3 vs Q1: 1.83 (0.70–4.80)c RR: Q4 vs Q1: 2.00 (0.85–4.70)c RR: Each 1 mmol/L increase: 1.13 (0.90–1.41)c RR: Increase of 1 SD: 0.89 (0.66–1.21)c
Age range 40–65
82
36
No
OR: 1.00 (0.99–1.01)e
Mean age 50
71
21
Yes
1965–1968
80.2
1994–1996
Yes
OR: ⱖ6.5 vs ⬍6.5 mmol/l ⫽ 2.8 (1.2–6.7)e,f Any dementia (n ⫽ 57), Control (n ⫽ 1,270)  ⫽ ⫺0.39, p ⫽ 0.001 (change in cholesterol over time,a fully adjusted, as a predictor of all incident dementia)b
Aged 65⫹
76.35
5.6
No No No No No No No No No No
HR: Q2 vs Q1: 0.96 (0.68–1.38)c HR: Q3 vs Q1: 1.01 (0.70–1.46)c HR: Q4 vs Q1: 1.16 (0.81–1.67)c HR: Q2 vs Q1: 1.05 (0.48–2.29)c HR: Q3 vs Q1: 0.97 (0.43–2.17)c HR: Q4 vs Q1: 0.31 (0.11–0.85)c RR: Q2 vs Q1: 1.02 (0.77–1.36)c RR: Q3 vs Q1: 1.23 (0.88–1.72)c RR: Q4 vs Q1: 0.95 (0.69–1.31)c RR: Each 1 mmol/L increase: 0.97 (0.89–1.06)c
Mielke et al., 200532 (93/289)
Cohort aged 70
Slooter et al., 200030 (334/6101)
Mean age 69.5d
75, 79, 81, 83, 85, and 88 NA
18
5.8
Notes: NA: data not available; OR: odds ratio; HR: hazard ratio; RR: relative risk; : the coefficient derived from linear random effects model. a There was a greater decline in cholesterol level from baseline in subjects with dementia at follow-up. b Random effects model. c Cox proportional hazards regression. d Value is for the cohort analyzed in a separate Rotterdam study article by Ott et al. (1998), of which Slooter et al. (2002) report data for 91% of participants. e Logistic regression. f Multinomial regression.
pants with only one of these risk factors.36 Another study found that the association between APOE*4 and AD was weakened after adjusting for TC.26 Five studies24,25,28,30,38 did not find any interaction between APOE*4 and TC in relation to dementia risk. Studies Not Meeting Criteria That Were Included in Recent Reviews Three studies reported in recent reviews of cholesterol and dementia3,5 did not meet the criteria for in-
350
clusion in our review. One study was excluded because dementia diagnoses (including dementia, memory impairment, AD, VaD, and dementia not otherwise specified) were obtained from medical records.39 The other two studies were excluded because their sample contained participants age 60 or older who were not screened for dementia at baseline.40,41 One of these studies had an older baseline sample40 but the other had a midlife sample at baseline (mean age: 52.7, SD: 4.7) and 26-year follow-up. A large proportion of this
Am J Geriatr Psychiatry 16:5, May 2008
Anstey et al.
TABLE 4.
Results for Total Cholesterol as a Risk Factor for Cognitive Performance Change, Cognitive Impairment and Mild Cognitive Impairment Follow-up (Years)
Study Cognitive performance change (continuous) Reitz et al., 200537 (1147)
Wada et al., 199742 (93)
Cognitive performance change (dichotomous outcome) (N Decline/No decline) Swan et al., 199243 (6/6) Kalmijn et al., 199636 (51/302) Karlamangla et al., 200452 (37/230) Cognitive impairment Hyman et al., 199638 (NA) Mild cognitive impairment (N MCI/control) Kivipelto et al, 200153 (82/1270) Solfrizzi et al., 200434 (139/2824)
Baseline Year or Age
Age
Period
Effect
Statistic
Mean age 76.3
NA
5.6
Mean age 79.4
NA
3
No No No Yes
Beta (SE)a Memory: ⫺0.04 (0.8), p ⫽ 0.6b Cog./visuo.: 0.2 (0.3), p ⫽ 0.4b Language: ⫺0.1 (0.1), p ⫽ 0.6b Repeated measures ANOVA group (high cholesterol vs low cholesterol) by time on MMSE: p ⫽ 0.0178c
Mean age 55.2
NA
6
Yes
Mean age 74.6
NA
3
No
Decliners: 5.20 mmol/L vs. non-decliners: 5.36 mmol/L (p ⬍0.01)d OR: High: 1.38 (95% CI: 0.75–2.55)e,f
Mean age 74
NA
2.5 and 4.5
No
OR: each 10 mg/dL increase: 0.86 (95% CI: 0.73–1.02)e
Age 65⫹
79.1
4–7
Yes
OR: 0.997 (95% CI: 0.993–0.999)e
Mean age 50
71
21
Yes
Mean age 71.8
NA
OR: ⱖ6.5 vs ⬍6.5 mmol/l: 1.9 (95% CI: 1.2–3.0)e,g RR: Q2 vs Q1: 0.58 (95% CI: 0.33–0.98)h RR: Q3 vs Q1: 0.51 (95% CI: 0.28–0.77)h RR: Q4 vs Q1: 0.50 (95% CI: 0.28–0.87)h
3.5
Yes Yes Yes
Notes: SE: standard error; NA: data not available; CI: confidence interval; OR: odds ratio; RR: relative risk; Q1, Q2, Q3, Q4: the lowest, second lowest, third lowest and highest quartile of the total cholesterol measure. Statistical analyses: aBetas reported for Time ⫻ total cholesterol adjusted for age, sex, education, ethnicity, and APOE. b Generalized estimating equations. c Analysis of variance. d t-Tests were calculated from descriptive statistics reported by Stewart et al., 2007. e Logistic regression. f Odds ratios were calculated from descriptive statistics reported by Kalmijn et al. (1996). g Multinomial regression. h Poisson regression models.
sample (N ⫽ 3,734) would have been under age 60 and hence this study needs to be considered to avoid potential bias. This study found no association between midlife TC and AD, VaD, or any dementia.40
CONCLUSIONS The studies included in this review reported few associations between serum lipid levels and the risk of developing dementia. Nevertheless, within the reported associations high TC in midlife predicting AD was a consistent finding. Congruent with this was the finding of a decrease in TC from mid- to
Am J Geriatr Psychiatry 16:5, May 2008
late-life being associated with an increased risk of either AD or any dementia. These latter findings cast doubt over the generalizability of null results reported by studies in which the measure of TC was an average taken across many years.24 Such an approach may not be sensitive to an association between TC and dementia risk that changes over time. Meta-analyses of late-life TC in relation to AD, VaD, and any dementia did not reveal any significant associations, suggesting the possibility that the effect of TC on dementia risk may be limited to midlife. Our review contained no studies that found an association between TC and VaD. This could be due to the small number of studies that included VaD as an
351
TC as a Risk Factor for Dementia and Cognitive Decline outcome, or the lower prevalence of VaD. For example, whereas the Honolulu, HI-Asia Aging Study27 had 56 cases of AD, it had only 15 cases of VaD, limiting the statistical power to detect a significant association between TC and VaD. Alternatively, it is possible that TC does not contribute to the pathological mechanisms associated with VaD. If this latter conclusion is the case, then the results indicate a need to develop different cardiovascular risk factor profiles for AD and VaD. Five of the eight studies reporting on cognitive decline, cognitive impairment, and MCI found a significant association between the outcome and TC. High TC in late-life was reported to be associated with a decreased risk of both cognitive decline42 and cognitive impairment.34,38 However, there was no effect of late-life TC on cognitive decline found in studies with larger samples, leading to the conclusion that there remains a lack of reliable evidence supporting any association between high TC in latelife and cognitive decline. One extremely small study reported that high midlife TC was associated with a decreased risk of cognitive decline43 and another larger study found that high midlife TC increased the risk of MCI in late-life. No significant associations were found in any study reporting results for HDL or LDL. However, the small number of studies reporting relevant data prevents strong conclusions being drawn and it would be beneficial if future studies reported all available indices of serum cholesterol. The seven studies reporting data on the interaction between APOE*4 and TC were mostly consistent in revealing a lack of interaction. It is possible that this was due to the small number of participants in each individual study who were both APOE*4 positive and had potentially high risk cholesterol levels. There is also evidence that APOE*4 in conjunction with hypercholesterolemia may affect subtypes of dementia differently44 and hence analyses of any dementia as an outcome may dilute effects. The effect of APOE*4 and other risk factors may amplify any adverse effect of TC at critical periods in the life-course that may not be captured in studies with infrequent occasions of measurement. Studies with multiple occasions of measurement from mid to late-life27,45 therefore allow for the most accurate modeling of the relationships among risk factors and cognitive outcomes.
352
Limitations of systematic reviews include potential publication bias in favor of positive findings. The fact that we identified studies that did not focus on cholesterol, but reported results incidentally31,38 reduces this potential publication bias. Further, our search strategy identified more nonsignificant results than significant. The review was limited by the small number of comparable studies in any one category, such that we were unable to investigate test for publication bias. As this review is based on observational studies, it is possible that health and life style factors associated with TC explain the associations between TC and dementia and TC and cognitive decline.46 For example, high TC in midlife may result from poor nutrition, poorer general health, and be associated with less physical activity or frailty47,48 or be associated with depression in late-life.49 Although many studies adjusted for these factors, there was inconsistency among studies in the choice of covariates. The review is also limited by the lack of information on participants taking cholesterol lowering medication, which may confound the results. There is evidence that statin use is associated with lower dementia risk independent of any cholesterol lowering effects.50 Cultural and ethnic differences in life style and genetic factors may also have affected the observed associations. The strengths of the review included an a priori search strategy, duplicate study selection and data extraction, a comprehensive literature review, and the reporting of study characteristics and excluded studies.51 If declining TC precedes AD then in any study there may be individuals at different stages of the trajectory that leads to AD but who do meet diagnostic criteria. The only situation where this could be overcome would be in a longitudinal study where all participants have been followed up until death, and retrospective analysis is conducted with full information on final dementia outcomes. Although we excluded studies including older participants that did not screen for dementia at baseline, it is possible that some of the selected studies included individuals with preclinical dementia. It is also possible, but less likely, that those with primarily midlife samples at baseline included a very small number of participants with preclinical dementia. This review points to a need for further data from longitudinal studies involving several occasions of measurement from midlife to late adulthood, and for
Am J Geriatr Psychiatry 16:5, May 2008
Anstey et al. studies reporting the association between TC and subtypes of dementia. This will allow for further quantitative meta-analyses of this important association that will yield firm conclusions concerning the associations between age, serum lipids, risk of subtypes of dementia, and cognitive decline. Such research is required to inform decisions regarding the
modulation of cholesterol to prevent or delay specific types of dementia. The authors thank Chwee von Sanden and Richard Burns for their assistance. This work was supported by the National Health and Medical Research Council of Australia.
References 1. Whalley LJ, Dick FD, McNeill G: A life-course approach to the aetiology of late-onset dementias. Lancet Neurol 2006; 5:87–96 2. Panza F, D’Introno A, Colacicco AM, et al: Cognitive frailty: predementia syndrome and vascular risk factors. Neurobiol Aging 2006; 27:933–940 3. Panza F, D’Introno A, Colacicco AM, et al: Lipid metabolism in cognitive decline and dementia. Brain Res Rev 2006; 51:275–292 4. Launer LJ: The epidemiologic study of dementia: a life-long quest? Neurobiol Aging 2005; 26:335–340 5. Kivipelto M, Solomon A: Cholesterol as a risk factor for Alzheimer’s disease— epidemiological evidence. Acta Neurol Scand Suppl 2006; 185:50 –57 6. Hofer SM, Sliwinski MJ: Understanding ageing. An evaluation of research designs for assessing the interdependence of ageingrelated changes. Gerontology 2001; 47:341–352 7. Qiu C, Winblad B, Fratiglioni L: The age-dependent relation of blood pressure to cognitive function and dementia. Lancet Neurol 2005; 4:487– 499 8. Braak H, Braak E: Neuropathological stageing of Alzheimer-related changes. Acta Neuropathol 1991; 82:239 –259 9. Anstey KJ, von Sanden C, Salim A, et al: Smoking as a risk factor for dementia and cognitive decline: a meta-analysis of prospective studies. Am J Epidemiol 2007; 166:367–378 10. American Psychiatric Association. Diagnostic and Statistical Manual of Mental Disorders. 3rd Revised. Ed. Washington, DC, American Psychiatric Association, 1987 11. McKhann G, Drachman D, Folstein M, et al: Clinical diagnosis of Alzheimer’s disease: report of the NINCDS-ADRDA Work Group under the auspices of Department of Health and Human Services Task Force on Alzheimer’s Disease. Neurology 1984; 34:939 –944 12. Roman GC, Tatemichi TK, Erkinjuntti T, et al: Vascular dementia: diagnostic criteria for research studies. Report of the NINDSAIREN International Workshop. Neurology 1993; 43:250 –260 13. Petersen RC, Smith GE, Ivnik RJ, et al: Apolipoprotein E status as a predictor of the development of Alzheimer’s disease in memoryimpaired individuals. JAMA 1995; 273:1274 –1278 14. de Craen AJ, Gussekloo J, Vrijsen B, et al: Meta-analysis of nonsteroidal antiinflammatory drug use and risk of dementia. Am J Epidemiol 2005; 161:114 –120 15. Cochrane Handbook for Systematic Reviews of Interventions 4.2.5 [updated May 2005]. Vol Issue 3. Chichester, UK, John Wiley & Sons, 2005 16. Bernick C, Katz R, Smith NL, et al: Statins and cognitive function in the elderly: the Cardiovascular Health Study. Neurology 2005; 65:1388 –1394 17. Cervilla JA, Prince M, Mann A: Smoking, drinking, and incident cognitive impairment: a cohort community based study included in the Gospel Oak project. J Neurol Neurosurg Psychiatry 2000; 68:622– 626 18. Kalmijn S, Launer LJ, Ott A, et al: Dietary fat intake and the risk of
Am J Geriatr Psychiatry 16:5, May 2008
incident dementia in the Rotterdam Study. Ann Neurol 1997; 42:776 –782 19. Prince M, Lovestone S, Cervilla J, et al: The association between APOE and dementia does not seem to be mediated by vascular factors. Neurology 2000; 54:397– 402 20. Rea TD, Breitner JC, Psaty BM, et al: Statin use and the risk of incident dementia: the Cardiovascular Health Study. Arch Neurol 2005; 62:1047–1051 21. Zimetbaum P, Frishman WH, Ooi WL, et al: Plasma lipids and lipoproteins and the incidence of cardiovascular disease in the very elderly. The Bronx Aging Study. Arterioscler Thromb 1992; 12:416 – 423 22. Moroney JT, Tang MX, Berglund L, et al: Low-density lipoprotein cholesterol and the risk of dementia with stroke. JAMA 1999; 282:254 –260 23. Romas SN, Tang MX, Berglund L, et al: APOE genotype, plasma lipids, lipoproteins, and AD in community elderly. Neurology 1999; 53:517–521 24. Tan ZS, Seshadri S, Beiser A, et al: Plasma total cholesterol level as a risk factor for Alzheimer disease: the Framingham Study. Arch Intern Med 2003; 163:1053–1057 25. Kivipelto M, Helkala EL, Laakso MP, et al: Apolipoprotein E epsilon4 allele, elevated midlife total cholesterol level, and high midlife systolic blood pressure are independent risk factors for late-life Alzheimer disease. Ann Intern Med 2002; 137:149 –155 26. Notkola IL, Sulkava R, Pekkanen J, et al: Serum total cholesterol, apolipoprotein E epsilon 4 allele, and Alzheimer’s disease. Neuroepidemiology 1998; 17:14 –20 27. Stewart R, White LR, Xue QL, et al: Twenty-six-year change in total cholesterol levels and incident dementia: the Honolulu-Asia Aging Study. Arch Neurol 2007; 64:103–107 28. Li G, Shofer JB, Kukull WA, et al: Serum cholesterol and risk of Alzheimer disease: a community-based cohort study. Neurology 2005; 65:1045–1050 29. Reitz C, Tang MX, Luchsinger J, et al: Relation of plasma lipids to Alzheimer disease and vascular dementia. Arch Neurol 2004; 61:705–714 30. Slooter AJC, Ruitenberg A, van Duijn CM, et al: The effect of apoE on dementia is not through atherosclerosis: The Rotterdan Study: Reply. Neurology 2000; 54:2357–2358 31. Beeri MS, Rapp M, Silverman JM, et al: Coronary artery disease is associated with Alzheimer disease neuropathology in APOE4 carriers. Neurology 2006; 66:1399 –1404 32. Mielke MM, Zandi PP, Sjogren M, et al: High total cholesterol levels in late life associated with a reduced risk of dementia. Neurology 2005; 64:1689 –1695 33. Kivipelto M, Helkala EL, Laakso MP, et al: Midlife vascular risk factors and Alzheimer’s disease in later life: longitudinal, population based study. BMJ 2001; 322:1447–1451 34. Solfrizzi V, Panza F, Colacicco AM, et al: Vascular risk factors,
353
TC as a Risk Factor for Dementia and Cognitive Decline incidence of MCI, and rates of progression to dementia. Neurology 2004; 63:1882–1891 35. Yoshitake T, Kiyohara Y, Iwamoto H, et al: Incidence and risk factors of vascular dementia and Alzheimer’s disease in a defined elderly Japanese population: The Hisayama Study. Neurology 1995; 45:1161–1168 36. Kalmijn S, Feskens EJ, Launer LJ, et al: Cerebrovascular disease, the apolipoprotein e4 allele, and cognitive decline in a community-based study of elderly men. Stroke 1996; 27:2230 – 2235 37. Reitz C, Luchsinger J, Tang MX, et al: Impact of plasma lipids and time on memory performance in healthy elderly without dementia. Neurology 2005; 64:1378 –1383 38. Hyman BT, Gomez-Isla T, Briggs M, et al: Apolipoprotein E and cognitive change in an elderly population. Ann Neurol 1996; 40:55– 66 39. Whitmer RA, Sidney S, Selby J, et al: Midlife cardiovascular risk factors and risk of dementia in late life. Neurology 2005; 64:277– 281 40. Kalmijn S, Foley D, White L, et al: Metabolic cardiovascular syndrome and risk of dementia in Japanese-American elderly men. The Honolulu-Asia aging study. Arterioscler Thromb Vasc Biol 2000; 20:2255–2260 41. Kuusisto J, Koivisto K, Mykkanen L, et al: Association between features of the insulin resistance syndrome and Alzheimer’s disease independently of apolipoprotein E4 phenotype: cross sectional population based study. BMJ 1997; 315:1045–1049 42. Wada T, Matsubayashi K, Okumiya K, et al: Lower serum cholesterol level and later decline in cognitive function in older people living in the community, Japan. J Am Geriatr Soc 1997; 45:1411–1412 43. Swan GE, LaRue A, Carmelli D, et al: Decline in cognitive performance in aging twins. Heritability and biobehavioral predictors from the National Heart, Lung, and Blood Institute Twin Study. Arch Neurol 1992; 49:476 – 481
354
44. Borroni B, Grassi M, Costanzi C, et al: APOE genotype and cholesterol levels in lewy body dementia and Alzheimer disease: investigating genotype-phenotype effect on disease risk. Am J Geriatr Psychiatry 2006; 14:1022–1031 45. Solomon A, Kareholt I, Ngandu T, et al: Serum cholesterol changes after midlife and late-life cognition: twenty-one-year follow-up study. Neurology 2007; 68:751–756 46. Chiang CJ, Yip PK, Wu SC, et al: Midlife risk factors for subtypes of dementia: A Nested Case-Control Study in Taiwan. Am J Geriatr Psychiatry 2007; 15:762–771 47. Monastero R, Palmer K, Qiu C, et al: Heterogeneity in risk factors for cognitive impairment, no dementia: population-based longitudinal study from the Kungsholmen Project. Am J Geriatr Psychiatry 2007; 15:60 – 69 48. Zuliani G, Volpatol S, Romagnoni F, et al: Combined measurement of serum albumin and high-density lipoprotein cholesterol strongly predicts mortality in frail older nursing-home residents. Aging Clin Exp Res 2004; 16:472– 475 49. Almeida OP, Flicker L, Norman P, et al: Association of cardiovascular risk factors and disease with depression in later life. Am J Geriatr Psychiatry 2007; 15:506 –513 50. Jick H, Zornberg GL, Jick SS, et al: Statins and the risk of dementia. Lancet 2000; 356:1627–1631 51. Shea BJ, Grimshaw JM, Wells GA, et al: Development of AMSTAR: a measurement tool to assess the methodological quality of systematic reviews. BMC Med Res Methodol 2007; 7:10 52. Karlamangla AS, Singer BH, Reuben DB, et al: Increases in serum non-high-density lipoprotein cholesterol may be beneficial in some high-functioning older adults: MacArthur studies of successful aging.[see comment]. J Am Geriatr Soc 2004; 52:487– 494 53. Kivipelto M, Helkala EL, Hanninen T, et al: Midlife vascular risk factors and late-life mild cognitive impairment: a populationbased study. Neurology 2001; 56:1683–1689
Am J Geriatr Psychiatry 16:5, May 2008