1 Jean Mayer US Department of Agriculture Human Nutrition Research Center on Aging at Tufts University, Boston, MA 02111, USA 2University of Georgia-Athens, Athens, GA 30602, USA 3Temple University, Philadelphia, PA 19122, USA 4Harvard University, Boston, MA 02115, USA 5Emory University, Atlanta, GA 30322, USA 6Wayne State University, Detroit, MI 48202, USA 7University of Kentucky, Lexington, KY 40536, USA 8Yonsei University, Seoul 120-749, Republic of Korea 9DSM Nutritional Products, CH-4002 Basel, Switzerland

Correspondence should be addressed to Elizabeth J. Johnson; [email protected] Received 4 January 2013; Revised 5 April 2013; Accepted 28 April 2013 Academic Editor: Paula Bickford

Copyright © 2013 Elizabeth J. Johnson et al. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

Oxidative stress is involved in age-related cognitive decline. The dietary antioxidants, carotenoids, tocopherols, and vitamin A may play a role in the prevention or delay in cognitive decline. In this study, sera were obtained from 78 octogenarians and 220 centenarians from the Georgia Centenarian Study. Brain tissues were obtained from 47 centenarian decedents. Samples were analyzed for carotenoids, 𝛼-tocopherol, and retinol using HPLC. Analyte concentrations were compared with cognitive tests designed to evaluate global cognition, dementia, depression and cognitive domains (memory, processing speed, attention, and executive functioning). Serum lutein, zeaxanthin, and 𝛽-carotene concentrations were most consistently related to better cognition (𝑃 < 0.05) in the whole population and in the centenarians. Only serum lutein was significantly related to better cognition in the octogenarians. In brain, lutein and 𝛽-carotene were related to cognition with lutein being consistently associated with a range of measures. There were fewer significant relationships for 𝛼-tocopherol and a negative relationship between brain retinol concentrations and delayed recognition. These findings suggest that the status of certain carotenoids in the old may reflect their cognitive function. The protective effect may not be related to an antioxidant effect given that 𝛼-tocopherol was less related to cognition than these carotenoids.

Cognitive decline in the elderly is a significant public health issue. It has been estimated that the incidence of mild

cognitive impairment (MCI) is approximately 19% in those younger than 75 years and 29% in those older than 85 years [1]. Further, 13% of people aged 65 years and older are afflicted with Alzheimer’s disease. Studies in centenarians

have reported considerable dementia, ranging from 42 to 100% [2, 3]. The number of individuals so affected is likely to increase given that the number of people over 65 years is rising. As with most age-related diseases, the most cost effective way to combat disease is through prevention. One possible strategy is nutrition intervention [4].

Fruit and vegetable intake has been associated with cognitive function [5–7]. For example, in a study of 13,388 women, it was found that total vegetable intake was significantly associated with reduced cognitive decline [8]. The strongest association was with greater intake of green leafy and cruciferous vegetables. Fruits and vegetables are major dietary sources of carotenoids. Carotenoids are a class of naturally occurring pigments that are synthesized by plants and produce the red, orange, and yellow colors of fruits and vegetables. Carotenoids are comprised of two subclasses: xanthophylls (lutein, zeaxanthin, and 𝛽-cryptoxanthin) and carotenes (𝛼-carotene, 𝛽-carotene, and lycopene). In the Nurses’ Health Study, among nonsupplement users, women in the highest quartile of plasma carotenoids had better cognitive performance than those in the lowest quartile [9]. Research has shown that patients with MCI had decreased plasma levels of antioxidants, including carotenoids [10]. Given that dietary carotenoids function as both antioxidants and anti-inflammatory agents and that oxidative stress and inflammation are believed to be involved in the pathogenesis of cognitive decline [11–20], intake of these dietary components may hold promise in cognitive health for the elderly.

The first objective of this study was to evaluate the relationship between serum concentrations of carotenoids and cognitivefunctioninsubjectsfromtheGeorgianCentenarian Study, a population-based multidisciplinary study of octogenarians and centenarians conducted in Georgia (USA) [21].

Given that an intervention with lutein was reported to improve cognitive function in the elderly [22] and that compared to carotenes, xanthophylls are preferentially taken up into brain tissue [23], a second objective of this research wastoevaluatethecross-sectionalrelationshipbetweenbrain carotenoids and premortem measures of cognitive function in a subgroup of the centenarian participants. For both serum andbraintissues,𝛼-tocopherolwasmeasuredforcomparison since it crosses the blood brain barrier and has antioxidant properties. Serum and brain retinol concentrations were measured because of the provitamin A activity of certain carotenoids.

This study provides a unique advantage of being able to assess the relationship between serum and brain carotenoids. If indeed serum concentrations of individual carotenoids reflect their levels in the brain and these brain carotenoids are related to better cognitive performance, serum carotenoid measures could be a useful tool for evaluating the benefits of dietary carotenoids to age-related cognitive health.

2009, was designed to identify and isolate longevity genes, neuropathology, functional capacity, and adaptational characteristics of centenarians [17]. Living status was classified as community dwelling or institutionalized where community dwelling included those living in private residences and institutionalized included individuals living in a skilled nursing facility or personal care home. The study involving serum analyses included 244 centenarians (defined in this study

1. as age 98yrs and older) and 80 octogenarians. The study involving brain tissue analyses included 47 centenarians who volunteered to donate their brain upon death. Subjects were recruited from the community, personal care homes, and skilled nursing facilities. The sample procedures and data collection methods have been described elsewhere [21].

1. 2.1. Study Population. The Georgia Centenarian Study (GCS) [21], a population-based multidisciplinary study conducted in 44 counties in northern Georgia (USA) from 2001 to

(WAIS-III) Similarities Subtest [32], Behavioral Dyscontrol Scale [33], and Controlled Oral Word Association Test (COWAT) [34].

For participants in the separate brain donation component, additional cognitive tests were administered every six months until mortality. These tests included the Consortium toEstablishaRegistryforAlzheimer’sDisease(CERAD)neuropsychological battery, which is composed of five subtests derived from previously established cognitive tests (verbal fluency, Boston Naming Test, MMSE, Constructional Praxis, and Word List Memory) [35, 36]. These subtests have been found to be valid and reliable measures of cognition in normal aging and in Alzheimer’s disease [37]. Further, no differences were found between participants in the brain donation component of the GCS and the rest of the centenarian participants [38]. All of these tests or versions of them have been used and validated in aging research settings or have demonstrated sensitivity to health variables in epidemiological studies [39–42].

1. 2.5. Covariates and Predictors. In the analyses involving serum, covariates and predictors included age (80–89 or ≥98y), gender, and race (white or African American, by design). The proportion of participants from each age group recruited from skilled nursing facilities was based on estimates of the “institutionalized” population of the study area according to the 2000 US Census figures [21]. Residence statuswasnotconsideredasacovariatebecauseitwasapotential suppressor variable. Fifteen percent of the octogenarians and 62% of the centenarians resided in skilled nursing facilities. The remaining “community dwelling” participants resided in private residences and personal care homes. In the analyses involvingbrain tissue, which involvedcentenarian decedents, 31% of the decedents had resided in skilled nursing facilities.
2. 2.6. Statistical Analyses. Results are expressed as geometric means ± SDs. Relationships between trans and cis isomers of individual carotenoids and cognition did not differ appreciably. Therefore, the total (𝑡𝑟𝑎𝑛𝑠 + 𝑐𝑖𝑠) was used in the analysis. Given that the purpose of our analyses was to increase the precision with which an association could be estimated following adjustment for variables associated with our criterion, but not our predictor, we chose to analyze relationships between carotenoids, 𝛼-tocopherol, and retinol with cognitive measures using partial correlations. Thus, the partial correlation can provide an estimate of substantive interest but has the added advantage in that it does so in a standardized and easily interpretable metric. Statistical significance was set at 𝑃 < 0.05. All statistical analyses were performed using SAS version 9.0 (serum) and SPSS version 19.0 (brain).

1. 2.6.1. Serum Analyses. Data were verified for normality (Shapiro-Wilk test) and, when necessary, were logtransformed for normal distribution before further statistical analysis. Chi-square test and Student’s 𝑡-test were used to compare subject characteristics, serum carotenoid levels, and cognitive values between groups. Pearson’s correlations were performed to identify associations between cognitive

indices with age, sex, anthropometric variables, and other possible confounders. The associations between cognitive indices and serum carotenoids were determined by calculating partial Pearson’s correlation coefficients adjusted for age, sex, body mass index (BMI), smoking, alcohol, diagnosed hypertension, and diabetes. For the centenarians, diagnosis of hypertension and diabetes was drawn from proxy, family, staff, or charts.

1. 2.6.2. Brain Analyses. Datawereanalyzedforall47decedents together and also separately for decedents based on their premortem GDRS scores. The purpose was to determine differences between decedents who had intact cognitive function (GDRS = 1), mild memory loss (GDRS = 2), mild cognitive impairment (GDRS = 3), and dementia (GDRS >
2. 3. Results 3.1. Serum Analytes and Cognition

3.1.1. Subject Characteristics. The characteristics of the octogenarians and centenarians who provided serum are given in Table 1. A significantly greater proportion of the centenarians

Table 1: Subject characteristics.

Community dwelling (𝑛 = 150)

≥80 to ≤89y (𝑛 = 78)

≥98y (𝑛 = 220) 𝑃 value∗

Total (𝑛 = 298)

Institutionalized (𝑛 = 148) 𝑃 value∗

Age, yrs (mean ± SD) 84.2 ± 2.7 100.4±1.9 93.3 ± 8.3 99.7 ± 4.6 <0.0001 96.6±7.4 Male:female 27:51 36:184 0.0007 40:110 23:125 0.0187 63:235 Community dwelling (𝑛):institutionalized (𝑛)

66:12 84:136 <0.0001 — — 150:148

Education, yrs (mean ± SD) 13.0 ± 3.5 10.6 ± 3.8 <0.0001 12.1 ± 3.7 10.3 ± 3.7 <0.0001 11.2±3.8 Body mass index (kg/m2) (mean ± SD) 25.5 ± 4.8 22.6 ± 4.7 <0.0001 24.2 ± 4.7 22.5 ± 4.8 0.0059 23.3±4.8 Smoking (𝑛) <0.0001 0.035

None 33 156 88 101 189 Ex-smoker 40 54 53 41 94 Current 5 7 9 3 12 No information 0 3 0 3 3

Alcohol use (𝑛) <0.0001 <0.0001

None 22 134 67 89 156 Ex-drinker 27 45 32 40 72 Current 29 39 51 17 68

1. No information 0 2 0 2 2
2. No information 1 3 1 3 4

∗Significant difference between groups. Means were compared using Student’s 𝑡-test. Fisher’s exact test was applied for categorical variables.

were women, institutionalized, and nonsmokers compared to the octogenarians (𝑃 < 0.001). Furthermore, the centenarians had significantly less education years, alcohol use, and BMI (𝑃 < 0.001). There was no difference in the prevalence of hypertension in these two age groups.

A greater proportion of the institutionalized subjects were women and nonsmokers compared to the community dwelling subjects (𝑃 < 0.02 and 0.035, resp., Table 1). They were also significantly older than the community dwelling subjects (𝑃 < 0.001). The institutionalized subjects had significantly less education years, alcohol use, and BMI (𝑃 < 0.001). There was no difference in the prevalence of hypertension in these two groups.

1. 3.1.2. Serum Carotenoids, 𝛼-Tocopherol, and Retinol Concentrations. Serum concentrations of individual carotenoids, 𝛼tocopherol and retinol are given in Table 2. Compared to the octogenarians, the centenarians had lower mean values for all carotenoids and 𝛼-tocopherol, which was significantly different (𝑃 < 0.05) for all analytes except cis zeaxanthin and marginally significant for trans lutein (𝑃 < 0.075) and 𝛽carotene (𝑃 < 0.084). Similarly, the institutionalized subjects had lower mean values for all carotenoids and 𝛼-tocopherol which was significantly different (𝑃 < 0.05) for all analytes except for cryptoxanthin, 𝛼-carotene, and 𝛽-carotene and marginally significant for trans zeaxanthin (𝑃 < 0.052).

Serum retinol values were neither significantly different between octogenarians and centenarians nor between community dwelling and institutionalized subjects.

1. 3.1.3. Cognitive Function in Octogenarians and Centenarians. The cognitive function status of subjects from the Georgia Centenarian Study is found in Table S2 of the Supplementary Material. For all measures of cognitive function, mean values were significantly lower in the centenarians than in octogenarians (𝑃 < 0.0001) except for delayed recognition which was not different between the two groups. For all measures of cognitive function, the institutionalized subjects had significantly lower values than the community dwelling subjects (𝑃 < 0.0005).
2. 3.1.4. Relationships between Serum Carotenoids, 𝛼-Tocopherol, and Retinol and Cognitive Performance. In the total study population,serumluteinandzeaxanthinconcentrationswere most consistently related to better cognitive performance, with a significant correlation observed (𝑃 < 0.05) (Table 3) for all cognitive measures except delayed recognition. It should be noted that none of the serum analytes were correlated with delayed recognition. Serum concentrations of 𝛽-carotene were also significantly correlated to most measures of cognitive function (𝑃 < 0.05) with the exception

Table 2: Serum carotenoid, 𝛼-tocopherol, and retinol concentrations in subjects from the Georgian Centenarian Study (mean ± SD).

Community dwelling (𝑛 = 150)

nmol/L (carotenoids) 𝜇mol/L (𝛼-tocopherol, retinol)

≥80 to ≤89y (𝑛 = 78)

≥98y (𝑛 = 220) 𝑃 value∗

Total (𝑛 = 298)

Institutionalized (𝑛 = 148) 𝑃 value∗

Lutein, trans 213 ± 162 199 ± 177 0.0753 223 ± 199 181 ± 141 0.0072 204 ± 174 Lutein, cis 58 ± 58 53 ± 65 0.0006 63 ± 77 46 ± 46 0.0162 56 ± 63 Lutein, total 316 ± 255 293 ± 285 0.0156 334 ± 329 265 ± 207 0.0116 300 ± 276 Zeaxanthin, trans 47 ± 23 44 ± 28 0.0194 49 ± 30 42 ± 23 0.0516 46 ± 26 Zeaxanthin, cis 5 ± 16 5 ± 11 0.6388 7 ± 14 4 ± 9 0.0764 5 ± 12 Zeaxanthin, total 53 ± 32 49 ± 37 0.0355 56 ± 40 46 ± 28 0.0195 51 ± 35 Cryptoxanthin 159 ± 105 148 ± 125 0.0362 166 ± 137 134 ± 96 0.3174 150 ± 119 𝛼-Carotene 80 ± 110 63 ± 61 0.0122 71 ± 93 63 ± 54 0.6972 67 ± 76 𝛽-Carotene, trans 547 ± 836 443 ± 419 0.2200 542 ± 670 398 ± 406 0.0056 471 ± 558 𝛽-Carotene, cis 20 ± 26 17 ± 22 0.6538 19 ± 26 17 ± 20 0.5322 19 ± 22 𝛽-Carotene, total 568 ± 855 460 ± 432 0.0839 560 ± 69 415 ± 419 0.0039 488 ± 573 Lycopene, trans 240 ± 181 138 ± 119 <0.0001 199 ± 166 130 ± 112 <0.0001 164 ± 145 Lycopene, cis 389 ± 277 231 ± 227 <0.0001 324 ± 276 220 ± 212 0.0003 272 ± 251 Lycopene, total 629 ± 443 369 ± 337 <0.0001 523 ± 430 348 ± 309 0.0003 436 ± 384

𝛼-Tocopherol 31.9 ± 16.8 25.5±12.9 0.0011 29.6 ± 15.0 24.8 ± 13.2 0.0016 29.2 ± 14.3 Retinol 2.14 ± 0.07 1.84±0.62 0.449 2.00 ± 0.64 1.83 ± 0.59 0.247 1.92 ± 0.62

∗Significant difference between groups. Means were compared using Student’s 𝑡-test. Chi-square analysis was applied for categorical variables.

1. Table 3: Partial correlation coefficients between cognition indices and serum carotenoids, 𝛼-tocopherol, and retinol in octogenarians and centenarians from the Georgia Centenarian Study (adjusted w/age, sex, education years, BMI, smoking, alcohol, hypertension, and diabetes, 𝑛 = 298).

Behavioral Dyscontrol Scale Total Score Lutein, (trans+cis) 0.114 −0.350a 0.177b 0.009 0.114 0.148c 0.244a 0.180b Zeaxanthin, (trans+cis) 0.128c 0.236a 0.134c 0.069 0.146c 0.169b 0.184b 0.196b Cryptoxanthin 0.026 −0.096 0.054 0.002 0.035 0.031 0.129c 0.045 𝛼-Carotene 0.014 0.068 0.017 0.029 0.009 0.015 0.074 0.029 𝛽-Carotene, (trans+cis) 0.072 −0.244a 0.173b 0.082 0.168b 0.186b 0.144c 0.227a Lycopene, (trans+cis) 0.014 0.142c 0.017 0.033 0.014 0.017 0.108 0.044 𝛼-Tocopherol 0.118 −0.194b 0.097 0.097 0.145c 0.106 0.121c 0.200b Retinol 0.089 0.078 0.113 0.022 0.105 0.006 0.045 0.116

Controlled Oral Word Association Test

FOME1 delayed recall

MiniMental State Exam

FOME delayed recognition

Global Deterioration Rating Scale

FOME delayed retention

WAIS-III Similarities Subtest2

1FOME: Fuld Object Memory Evaluation; 2WAIS-III Similarities Subtest: Wechsler Adult Intelligence Scale-III Similarities Subtest. Significantly related (𝑃 <): a0.001; b0.01, c0.05.

of the MMSE and delayed recognition (Table 3). Serum 𝛼tocopherol concentrations were inversely related to dementia severity (Geriatric Deterioration Scale) (𝑃 < 0.01) and positively related to delayed retention, abstract reasoning, and the Behavioral Dyscontrol Scale (𝑃 < 0.05). Higher serum lycopene concentrations were only related to a lower dementia severity (𝑃 < 0.01). Serum cryptoxanthin was only related to the WAIS-III test (𝑃 < 0.05, Table 5). Serum retinol concentrations were not related to any of the cognitive function measures.

In the octogenarians, serum lutein concentrations were significantly related to measures of global cognition, lower dementia severity, and executive function (𝑃 < 0.05) (Table

S3 of the Supplementary Material). In this age group, serum cryptoxanthin was inversely related to delayed recall (𝑃 < 0.05). There were no other significant relationships. In the centenarians, none of the serum carotenoids or 𝛼-tocopherol were related to global cognition or delayed recognition. Higher concentrations of lutein, zeaxanthin, 𝛽-carotene, and 𝛼-tocopherol were significantly related to lower dementia severity (𝑃 < 0.05). Additional significant relationships were found between lutein and abstract reasoning, between 𝛽carotene and verbal fluency (Controlled Oral Word Association Test), WAIS-III and executive function, between 𝛼tocopherol and executive function, and between retinol and delayed recall (𝑃 < 0.05).

1. Table 4: Mean (±SEM) concentrations of carotenoids, 𝛼-tocopherol, and retinol in the cerebellum, frontal, occipital, and temporal cortices (𝑛 = 47).
2. Table 5: Mean (±SEM) concentrations of carotenoids, 𝛼-tocopherol, and retinol in the brain (average of cerebellum, frontal, occipital, and temporal cortices) based on premortem GDRS scores in decedents with normal cognitive function (GDRS = 1), subjective mild memory loss (GDRS = 2), and mild cognitive impairment (GDRS = 3).

Global deterioration scale (GDRS)a 1 (𝑛 = 5) 2 (𝑛 = 7) 3 (𝑛 = 11) Lutein, trans 133 ± 21b 124 ± 17b,c 67 ± 14c Total lutein (cis + trans) 145 ± 22b 127 ± 18b,c 68 ± 15c Zeaxanthin, trans 45.0 ± 7.5 43.1 ± 6.3 25.9 ± 5.0 Cryptoxanthin 90.1 ± 22.4 63.2 ± 19.0 57.2 ± 15.1 𝛽-Carotene, trans 77.6 ± 10.5b 48.0 ± 8.9b,c 39.5 ± 7.1c Lycopene, trans 37.0 ± 9.0 26.1 ± 7.6 16.4 ± 6.1 𝛾-Tocopherol 1609 ± 370 2129 ± 313 1518 ± 249 𝛼-Tocopherol 67408 ± 5295 63205 ± 4475 60028 ± 3570 Retinol 572 ± 109 530 ± 92 691 ± 74

pmol/g

aGDRS = 1: no subjective complaints or objective evidence of memory deficits; GDRS = 2: subject complaints, but no objective evidence of memory deficits; GDRS = 3: mild cognitive impairment. Means not sharing a common superscript in the same row are significantly different at 𝑃 < 0.05 (univariate ANOVA with Bonferroni adjustment for multiple comparisons).

In the community dwelling subjects, serum zeaxanthin had significant relationships with most measures of cognitive function (Supplementary Material, Table S4), with higher concentrations being significantly related to global cognitive performance, lower dementia severity, delayed recall and retention, verbal fluency, and concept formation/abstraction (𝑃 < 0.05). Higher serum lutein was significantly related to global cognitive function, lower dementia severity, and delayed recall and retention (𝑃 < 0.05). Other significant relationships were found between higher 𝛽-carotene concentrations and lower dementia, delayed recall and retention, and verbal fluency and executive function (𝑃 < 0.05). In the community dwelling subjects, 𝛼-tocopherol was only significantly related to executive function (𝑃 < 0.05). There were no significant relationships found between cryptoxanthin, 𝛼carotene and retinol and cognitive measures.

Fewer significant relationships were found between serum analytes and cognitive measures in the institutionalized subjects. In this group, significant relationships were found between lower dementia severity and serum concentrations of lutein, zeaxanthin, 𝛽-carotene, and 𝛼-tocopherol.

Serum lutein, zeaxanthin, cryptoxanthin, and 𝛼-carotene were significantly related to concept formation/abstraction (𝑃 < 0.05). Lutein, zeaxanthin, 𝛽-carotene, 𝛼-tocopherol and retinol were significantly related to executive function (𝑃 < 0.05).

3.2. Brain Analytes and Cognition

3.2.1. Subject Characteristics. The characteristics of the cententarian population for which brain tissues were available are described in Table S5 of the Supplementary Material. Of the 47 decedents, five had normal cognitive function

1. (GDRS = 1), seven had subjective mild memory loss
2. (GDRS = 2), and eleven had MCI (GDRS = 3). There were 24 decedents who had different stages of dementia (GDRS 4 to 7). Subjects with dementia were slightly older than those with intact cognitive function. However, the differences were not significant. Eighty-nine percent of the decedents were females, and 89% were Caucasians. Race/ethnicity, sex, education, BMI, smoking status, and alcohol use did not differ by GDRS status. A greater proportion of the

Brain

∗

120

concentrations (pmol/g)

99.9 ± 6.0

100

Brain carotenoid

74.8 ± 7.1

80

58.7 ± 3.2

60

34.3 ± 1.7

40

22.2 ± 2.1

20

0.0 ± 0.0

0

Lutein

Zeaxanthin

Lycopene

Cryptoxanthin

-carotene𝛼

-carotene𝛽

concentrations (nmol/L)

Serum carotenoid

Serum ∗

450

384.6 ± 38.5

400

350

300

220.2 ± 23.6

250

184.0 ± 26.5

200

138.3 ± 18.8

150

100

69.2 ± 8.2

50.0 ± 4.9

50

0

Lutein

Zeaxanthin

Lycopene

Cryptoxanthin

-carotene𝛼

-carotene𝛽

∗Significantly greater than all other carotenoids,

𝑃 < 0.02 (one-way ANOVA)

∗Significantly greater than all other carotenoids, 𝑃 < 0.0001 (one-way ANOVA)

(a)

(b)

Figure 1: Mean (± SEM) concentrations of carotenoids (trans isomers) in the brain (average of cerebellum frontal serum, occipital, and temporal cortices) of decedents from the Georgia Centenarian Study (𝑛 = 42). Dark gray bars indicate xanthophyll carotenoids. Light gray bars indicate carotenes.

decedents were institutionalized; however, the differences between institutionalized and community dwelling subjects were not statistically significant. The prevalence of diabetes and hypertension did not differ between these two groups.

1. 3.2.2. Brain Carotenoid, 𝛼-Tocopherol, and Retinol Concentrations. The mean concentration of individual carotenoids, 𝛼tocopherol, and retinol in the cerebellum, frontal, occipital, and temporal cortices from the 47 decedents is shown in Table 4. The range of values for carotenoids was 0–661pmol/g. For 𝛼-tocopherol, the range was 22,979– 137,576pmol/g and for retinol 202–2,233pmol/g. Mean lutein and zeaxanthin concentrations were significantly greater in the cerebellum compared to the frontal, occipital, and temporal cortices. Concentrations of cryptoxanthin and 𝛽carotene the were highest in the occipital cortex and were significantly different from the frontal and temporal cortices. 𝛼-Carotene was not detected in these brain tissues. Contrary to lutein and zeaxanthin, concentrations of 𝛼-tocopherol and retinol were the lowest in the cerebellum and significantly different from all three cortical regions.

The proportion of cis to trans isomers was much lower in the brain than in the serum. The ratio of cis to trans lutein and 𝛽-carotene was ∼0.25 in the serum while in the brain it was only ∼0.04. Although present in the serum, cis isomers of zeaxanthin and lycopene were not detected in any ofthebrainregionsanalyzed.Ofnoteisthatbraincarotenoids were significantly related to their concentrations in serum (𝑃 < 0.01 for all, except cis lutein: 𝑃 < 0.05). 𝛼-Tocopherol concentrations in the cortices were also significantly related to serum concentrations (𝑃 < 0.01). However, 𝛼-tocopherol in the cerebellum was not. Retinol concentrations in all brain regions were not related to serum retinol concentrations.

Figure 1 shows the mean carotenoid (trans isomers) concentrations in the brain and matched serum for the

decedents that had both serum and brain tissue. In the brain, xanthophylls (lutein, zeaxanthin, and cryptoxanthin) accounted for 72% of total carotenoids, of which lutein accounted for 34% of the total and was significantly greater than all other carotenoids (𝑃 < 0.02). The proportion of carotenes (𝛼-carotene, 𝛽-carotene, and lycopene) was higher than xanthophylls in serum, accounting for 57% of the total carotenoids of which 𝛽-carotene accounted for 37% of the total and was significantly greater than all other carotenoids (𝑃 < 0.0001).

3.2.3. Brain Carotenoids, 𝛼-Tocopherol, and Retinol Concentrations in Decedents with Intact Cognitive Function, Mild Memory Loss, and MCI. The mean brain concentrations of carotenoids, 𝛼-tocopherol, and retinol for decedents with GDRS ≤3 and GDRS >3 were not significantly different (data not shown) with the exception of 9 cis 𝛽-carotene which was significantly higher in dementia (17 ± 3.1 versus 9,8 ± 1.7pmol/g).

Table 5 shows brain carotenoids, 𝛼-tocopherol, and retinol concentrations in decedents with normal cognitive function, mild memory loss, and MCI. Due to the advanced pathological changes in brain and significant reduction in brain volume associated with dementia, decedents with dementia were not included in this analysis [43]. Additionally, the neuropathological and neurobiological brain changes associated with MCI are quantitatively less than those associated with dementia [44]. Mean concentrations of all carotenoids were found to progressively decrease with increasing GDRS scores from 1 to 3. However, only the differences in lutein and 𝛽-carotene concentrations between subjects with normal cognitive function and MCI were statistically significant (𝑃 < 0.05). When data were adjusted for age, sex, education, diabetes, and hypertension, only the differences observed in lutein concentrations between the

1. Table 6: Cross-sectional relationship between concentrations of carotenoids, 𝛼-tocopherol, and retinol in the cortex (average concentrations of the frontal, occipital, and temporal cortices) and premortem measures of cognitive function in subjects with normal cognitive function, mild memory loss, and MCI who completed the cognitive function tests a year prior their death (𝑛 = 21).

MMSE FOME recognition SIB COWAT BDS CERAD verbal fluency CERAD Boston naming test

Lutein (trans+cis) 0.494a −0.297 0.076 0.142 0.459 0.455 0.572a Zeaxanthin 0.439 −0.281 0.139 0.177 0.286 0.461a 0.207 Cryptoxanthin 0.056 0.325 0.021 0.152 0.046 0.302 0.084 𝛽-Carotene, trans 0.265 −0.010 −0.014 0.269 0.137 0.265 0.008 Lycopene, trans 0.124 −0.003 0.132 0.067 −0.085 0.065 0.275 𝛼-Tocopherol 0.393 −0.295 0.568a 0.637a 0.434 0.328 0.165 Retinol 0.161 −0.456 0.168 0.400 0.067 0.108 −0.307

Values are partial correlation coefficients adjusted for sex, education, diabetes, and hypertension. a𝑃 ≤ 0.05. MMSE:Mini-MentalStateExamination;FOME:FuldObjectMemoryEvaluation;SIB:SevereImpairmentBattery;COWAT:ControlledOralWordAssociation Test; BDS: Behavioral Dyscontrol Scale; CERAD: Consortium to Establish a Registry for Alzheimer’s Disease.

two groups remained significant (𝑃 < 0.05). Mean brain retinol concentration was not significantly different between individuals with MCI and normal cognitive function. Similar results were obtained when concentrations of carotenoids, 𝛼tocopherol, and retinol in the cerebellum and cortex (average of frontal, temporal, and occipital) were analyzed separately. Datacorrectedforcovariatesarenotreportedduetothesmall sample size in each GDRS subgroup.

1. 3.2.4. Relationship between Brain Carotenoids, 𝛼-Tocopherol, and Retinol and Cognitive Function Measures. Of the 23 decedents with normal cognitive function, mild memory loss, and MCI, data for 21 subjects whose cognitive function tests were performed within one year (4.3 ± 2.8mo, range: 0.3–10.5mo) prior to their death were analyzed for associations with cognition. A significant positive correlation was observed between lutein concentrations in the cortex and the MMSE (global measure of cognitive function) and the Boston Naming Test (CERAD battery, a measure of language), while a negative correlation was observed with Geriatric Depression Scale (𝑃 < 0.05) (Table 6). The positive association of zeaxanthin in the cortex and verbal fluency was statistically significant (𝑃 < 0.05). 𝛼-Tocopherol was positively associated with SIB (a global measure of cognitive function) and COWAT (a measure of executive function) (𝑃 < 0.05). In the case of retinol, there was a negative relation with FOME-delayed recognition (𝑃 < 0.05). No associations were observed with the other listed measures of cognitive function.

Carotenoids in the cerebellum were not associated with any of the cognitive function measures with the exception of a negative association between lutein and the Geriatric Depression Scale (𝑟 = −0.63, 𝑃 = 0.005). 𝛼-tocopherol in the cerebellum was positively associated with MMSE (𝑟 = 0.535, 𝑃 = 0.027) and SIB (𝑟 = 0.602, 𝑃 = 0.01), both of which are measures of global cognition.

In this cross-sectional study involving octogenarians and centenarians, we found significant relationships between

serum and brain concentrations of dietary carotenoids and variousmeasuresofcognitivefunction.Nospecificdomainof the cognitive performance showed the strongest relationship with either serum or brain concentrations of carotenoids since significant relationships were observed with memory, executive function, and language. The fact that specific carotenoids were associated with more than one cognitive function and that these associations remained statistically significant after controlling for potential confounding factors supports a possible role for these phytonutrients in agerelated cognitive health. The present study is the first report on serum and brain carotenoids, 𝛼-tocopherol, and retinol concentrations and their relationship to cognitive function in the oldest of the old. This is of importance given the dramatic increase in the number of Americans surviving into their 80s and 90s and the increased prevalence of age-related cognitive diseases such as Alzheimer’s disease [4].

Others have also found relationships with dietary carotenoids and age-related cognitive function. In crosssectional and longitudinal analysis of 442 subjects aged 65–94 years, Perrig et al. [45] reported that a higher 𝛽carotene plasma level was associated with better memory performances (free recall, recognition, and vocabulary). In the Rotterdam Study of 5182 community participants aged 55–95yrs,cross-sectionalanalysisfoundthatalowerintakeof 𝛽-carotene was associated with impaired cognitive function asmeasuredbytheMMSE[46].Inbothofthesestudies,there were no significant relationships with vitamin E. Also, no other carotenoids were evaluated. Two studies have reported that supplementation with antioxidants including 𝛽-carotene [47, 48] reduces the risk of cognitive decline. However, given that both studies involved multivitamin/mineral supplementation, a specific effect of 𝛽-carotene is difficult to determine. In the EVA, a cross-sectional study using a variety of cognitive measures in 589 subjects (68–79yrs), it was found that those with the lowest cognitive functioning (<25th percentile) had a higher probability of having low plasma levels of lycopene and zeaxanthin, but not lutein or 𝛽-carotene [49]. However, in a prospective study of older adults (mean age 73y), Morris and colleagues found no association between 𝛽-carotene intake and risk of

Alzheimer’s disease [50]. Additionally, the Age-Related Eye Diseases Study Research Group [51] found no effect of a multivitamin/mineral supplementation which included 𝛽-carotene on MMSE score and a battery of cognitive measures in a population with a median age of 69yrs. Rinaldi et al. found that plasma levels of lutein, zeaxanthin, and 𝛼-carotene were lower in MCI and Alzheimer’s disease subjects compared to controls but no difference for lycopene and 𝛽-carotene [10]. 𝛽-Cryptoxanthin was also significantly lower than controls in subjects with Alzheimer’s disease but not in subjects with MCI. Others have reported lower levels of zeaxanthin, 𝛽-cryptoxanthin, lycopene, and 𝛽carotene but not lutein and 𝛼-carotene in Alzheimer’s patients than in controls [52]. Inconsistencies among studies may be due to limited sample size, cognitive tests used, method of carotenoid assessment, or characteristics of the subject population. The major difference between our present study and previous research is the age of the population. In our population, the average age was 97yrs compared to averages of 67–77yrs for the studies discussed above.

Whether a possible protective effect of carotenoids differs between the old and the oldest of the old remains to be tested. However, in our study, the most consistent relationships with cognition were observed for serum lutein, zeaxanthin, and 𝛽-carotene, reflecting diets rich in green leafy vegetables and orange and yellow vegetables such as carrots, sweet potatoes, and winter squash [53]. This remained to be true for the centenarians and with respect to living status, but only lutein remained significantly related to better measures of cognitive performance in the octogenarians. Furthermore, in brain tissue only concentrations of lutein and 𝛽-carotene were significantly lower in the cortex and cerebellum of subjects with MCI compared to those with normal cognitive function. Lutein and 𝛽-carotene may thus be important carotenoids for maintaining normal cognitive function in older adults. Consumption of vegetables, particularly the green leafy varieties that are rich sources of lutein and 𝛽carotene,wasassociatedwithslowerratesofcognitivedecline in two large cohort studies [5, 8]. Other evidence suggests that lutein supplementation, alone or in combination with docosahexaenoic acid, may be able to improve certain aspects of cognitive performance in healthy older women [22].

Although much recent work has focused on lutein and its role in ocular health [54], lutein was also the dominant carotenoid in various regions of the centenarian brains. On the contrary, carotenes (𝛼-carotene, 𝛽-carotene, and lycopene) were predominant in serum, which more closely reflects dietary intake. These findings suggest that although not predominant in the diet, there seems to be a preferential uptakeofluteinfromthecirculationintothebrain.Craftetal. reported similar preferential uptake of xanthophylls in brain [23]. Trans isomers of lutein, zeaxanthin, cryptoxanthin and 𝛽-carotene, 𝛼-tocopherol, and retinol were detected in all the brain tissues analyzed in this study. Only three cis isomers, two of lutein and one of 𝛽-carotene, were detected in the centenarian brain, which were not reported in the elderly braintissues analyzedby Craftet al. [23]. Also,theratiosofcis to trans isomers in the brain were much lower than in serum,

indicating a preferential uptake of trans isomers in the brain of these centenarians.

The majority of our findings find an association between lutein concentrations in serum or brain with age-related cognitive performance. However, there were significant associations for zeaxanthin and 𝛽-carotene as well. Therefore, the protective effect of carotenoids does not appear to be limited to the provitamin A carotenoids (𝛽-cryptoxanthin, 𝛼-carotene, and 𝛽-carotene) nor to a class of carotenoids (xanthophylls versus carotenes). In addition, brain retinol levels were positively related to very few of the measures of cognition and negatively related to delayed recognition.

A protective effect of carotenoids may, in part, to be related to an antioxidant effect given that an antioxidant function is common to all carotenoids [55–58]. Furthermore, 𝛼-tocopherol, a major dietary antioxidant, was found to be related to several measures of cognitive. Cortical carotenoids may be protective in nature and may also influence interneuronalcommunicationandfunctionviamultiplemechanisms. Other mechanisms by which certain carotenoids may function include modulation of functional properties of synaptic membranes along with changes in their physicochemical and structural features [59]. Carotenoids have also been shown to enhance gap junctional communication [60] which in the retina is important for light processing and may be important for the development of neural circuitry in the visual system. Lutein and zeaxanthin, as macular pigments in the retina, have been related to increased visual processing speed and to reduced scotopic noise (noise associated with vision under dim light conditions) [61–63]. Lutein may also have antiinflammatory action in the brain, lowering inflammatory markers, and preventing cognitive decline [64, 65]. Neuroinflammation is also one of the factors that contribute to the pathogenesis of MCI and AD with increased levels of inflammatory markers being correlated with cognitive impairment [66].

The significant progressive decline in brain lutein and 𝛽carotene with increased impairment from normal to MCI indicates that these carotenoids may play a role in preventing cognitive decline. MCI is thought to be the transitional stage between normal aging and the earliest symptoms of AD. There may be as much as a 50% likelihood of individuals with MCI developing AD within five years [67]. Our finding of a significant decrease in lutein and 𝛽-carotene in subjects with MCI indicates that these carotenoids may play a role in maintenance of cognitive health prior to a decline to MCI. Future clinical studies should focus on nutritional interventions with lutein and 𝛽-carotene in subjects with MCI. Thus, far supplementation studies with 𝛽-carotene have yielded mixed results; the Physician Health Study II showed a long-term beneficial effect [68], while some studies showed no effect on cognitive function with 𝛽-carotene or antioxidant supplementation [69, 70]. Effects of lutein supplementation on subjects with MCI have not been studied to date.

The strength of this study is that we were able to evaluate cognitive function in the elderly using a battery of cognitive tests that included the MMSE. This approach is more powerful than using only the less sensitive measure

of MMSE. With the other tests, the scores ranges are wider, allowing for a better ability to study cognitive impairment. We were also able to evaluate cognitive relationships with a variety of dietary carotenoids using measures of serum concentrations. Carotenoid assessment using serum concentrations may be preferred to food frequency questionnaires because the high interindividual variation in intestinal bioavailability of carotenoids [71] does not need to be considered. Whether such variability exists for, uptake of carotenoids into brain tissue is not known. What is known is that the macular pigment response to supplementation of lutein from diet or supplements varies widely among individuals [72, 73]. A variable response of uptake into the brainmayalsoexistgiventhesignificantrelationshipbetween concentrations of xanthophylls in the retina and brain [26].

Thus, another strength of this study is the measures in brain tissue that are associated with premortem measures of cognition. Additionally, the strength of these cross-sectional relationships lies in the high interindividual variability in brain carotenoid concentrations, similar to that observed in the serum of these centenarians. Another strength of the study is that associations with cognitive function were evaluatedwithcarotenoids,𝛼-tocopherol,andretinolactually embedded in brain tissue. However, the significant correlations between carotenoid concentrations in serum and brain tissue suggest that serum carotenoid measures could be a useful tool for evaluating the benefits of dietary carotenoids to age-related cognitive health.

One limitation to this study is that, in cross-sectional analysis, it is not possible to affirm whether these low levels of carotenoids preceded or were the consequence of cognitive impairment. Low carotenoid status, that is, poor food choices, may be a reflection of poor cognitive status. The significant relationships observed may not be due to specific effects of the individual carotenoids but, indication of overall healthier diets and lifestyles. Another limitation is that our analyses were performed in a sample drawn from a relatively small population. Therefore, we were unable to perform multivariate regression analysis that would have provided additional information regarding the independent contributions of the individual nutrient components.

In conclusion, this is the first study to evaluate the role of carotenoids, 𝛼-tocopherol, and retinol in cognitive function in the oldest of the old. To date, previous studies evaluated older populations with average ages of 20–30 years younger than those of this study. Evaluations of the role of diet to health in this age group are becoming increasingly important given the rise in both life expectancy and the segment of the population who are >80 years. While far from conclusive, the idea that certain carotenoids can influence cognitivefunctioniscertainlyfeasible.Thesignificanceofour findings requires further research using biological studies, longitudinal epidemiological studies, and clinical trials with carotenoid supplementation.