Carotenoids in Young and Elderly Healthy Humans: Dietary Intakes, Biochemical Status and Diet-Plasma Relationships

Carotenoids in young and elderly healthy humans: dietary intakes, biochemical status and diet-plasma relationships

YL Carroll1, BM Corridan1 and PA Morrissey1*

1Nutritional Sciences, Department of Food Science and Technology, University College Cork, Cork, Ireland

Objective: To determine dietary carotenoid concentrations using an established and newly developed food frequency questionnaire (FFQ) method, to determine plasma carotenoid concentrations and to determine the relationship between these dietary and plasma variables in 24 ± 45 y and 65y groups. Design: Descriptive assessment of (FFQ), 7 ± d estimated records, and plasma carotenoids and their relationships in 24 ± 45y and 65y groups. Setting: Free living urban adults in Ireland. Subjects: Sixty-four volunteers aged 24 ± 45y and 54 volunteers aged 65y. Results: b-carotene was the predominant plasma carotenoid, but older groups had lower plasma concentrations of several carotenoids compared to younger groups (P < 0.005). b-carotene and lycopene were the major dietary carotenoids reported by estimated records and FFQ. Several estimated record and plasma carotenoid concentrations were positively associated in younger groups but not in older groups. FFQ overestimated dietary carotenoids relative to estimated records (P 0.05), generally did not re¯ect estimated record carotenoid concentrations and showed positive associations with plasma carotenoids only in older men. Neither of the dietary methods revealed a positive association between plasma and dietary b-carotene concentrations, whereas b-cryptoxanthin was strongly associated. Conclusions: Dietary and plasma concentrations of individual carotenoids are documented in young and elderly groups of a European country. Estimated record data reveals positive associations between diet and plasma carotenoids in younger, but not elderly groups. Further work examining diet-plasma relationship in older groups and developing a common FFQ suitable for use in several European countries is required. Sponsorship: Commission of the European Communities: AAIR Project (AIR2-CT93-0888). Descriptors: carotenoids; dietary assessment; elderly; biomarkers

Introduction tissue b-carotene concentrations and risk of myocardial infarction (Kardinaal et al, 1993). High fruit and vegetable consumption is associated with Data from several studies support a role for carotenoids reduced incidence of several chronic diseases. Evidence for in protection against chronic disease in elderly as well as a protective effect of greater vegetable consumption is middle-aged groups. A prospective study of residents of a consistent for cancers of the stomach, oesophagus, lung, retirement community, reported reduced risks of colon and oral cavity and pharynx, endometrium, pancreas and colon all sites combined cancers, with increasing intake of (Steinmetz & Potter, 1996). Mortality from coronary heart vegetables and fruits in elderly women (Shibata et al, disease, has been reported by several prospective studies, to 1992). Another prospective study on cardiovascular disease be inversely associated with consumption of fruit and mortality, reported that the bene®cial effects of increased vegetables (Knekt et al, 1994; Key et al, 1996). Biological fruit and vegetable consumption are apparent in the elderly concentrations of antioxidants derived from fruits and (Gaziano et al, 1995). Age related macular degeneration is vegetables are also inversely associated with risk of coro- another condition which has been shown to be inversely nary heart disease. An inverse association between serum associated with consumption of foods rich in carotenoids carotenoids and risk of coronary heart disease, was reported (Seddon et al, 1994). in a prospective study (Morris et al, 1994), while a nested In view of the inverse relationships between carotenoids case-control study, concluded that low serum levels of and disease, both dietary and biochemical status of carote- carotenoids, were associated with an increased risk of noids are individually worthy of further study. Little subsequent myocardial infarction among smokers (Street published data is available on tissue carotenoid concentra- et al, 1994). A cross-sectional, multicentre study, also tions in Irish population groups. Research studies generally reported an inverse association between high adipose tend to exclude those over the age of 65y and studies addressing this age group are warranted. Dietary data on consumption of carotenoids were, in the past, usually expressed as b-carotene, b-carotene equivalents or retinol equivalents. More recently, an extensive carotenoid food *Correspondence: Prof PA Morrissey, Nutritional Sciences, Department of composition database listing values for the ®ve major Food Science and Technology, University College Cork, Cork, Ireland. Received 7 October 1997; revised 18 February 1999; accepted carotenoids occurring in fruits, vegetables and multi-com- 26 February 1999 ponent foods, has been compiled in the USA (Mangels et Carotenoids in diet and plasma YL Carroll et al 645 al, 1993; Chug-Ahuja et al, 1993). In Europe, analysis of volunteers. All procedures were approved by the Clinical dietary carotenoids, other than b-carotene, is dif®cult Research Ethics Committee at University College, Cork. because many national food composition databases only Following screening, eligible subjects completed a 7d provide limited values for individual carotenoids (Souci et estimated record. The FFQ was administered on the day al, 1987; Holland et al, 1991b). However, analysis of foods following completion of the estimated records. Fasting in Finland, Spain, the Netherlands and UK has generated an blood samples were obtained at screening and on the day extensive body of published data on the carotenoid com- after completing the 7d estimated record. position of foods consumed in Europe (Heinonen et al, 1988, 1989; Ollilainen et al, 1988, 1989; Granado et al, Biochemical analysis of plasma carotenoids 1992; Olmedilla et al, 1993; Vollebregt & Feskens, 1993; Samples were protected from natural light. Plasma was Hart & Scott, 1995). As part in a multi-centre European immediately separated and stored at 7 70C until ana- study, a common carotenoid food composition database, lysed. Plasma carotenoids, tocopherols and retinol were based mainly on data from the above sources was agreed extracted from 0.2mL plasma with 0.2mL 10mmol SDS between participants. This database, providing data on solution, 0.4mL ethanol and triplicate extractions with lycopene, a-carotene, b-carotene, b-cryptoxanthin and 0.4mL n-hexane (0.05% w=v butylated hydroxy toluene luteinzeaxanthin concentrations in 106 food descriptors, (BHT)) (Burton et al, 1985). The hexane extracts were was used in the present study. dried under nitrogen and reconstituted in 20mL dichloro- Biomarkers of dietary intakes of nutrients are becoming methane followed by 180 mL acetonitrile-methanol (75:20 increasingly popular in nutrition research. The ability of by volume). A 50mL sample was injected onto a tempera- plasma to act as a biomarker of dietary intakes of individual ture controlled (25C) reverse phase HPLC system (Scott & carotenoids is of interest. Several intervention studies have Hart, 1993). The column system included Spherisorb ODS- shown that plasma carotenoids are responsive to increased 2 Guard Cartridges (Alltech, Lancashire, UK) in line with and reduced intakes of fruits and vegetables (Brown et al, Spherisorb ODS-2, 5 mm, 150mm64.5mm pre-column 1989; Micozzi et al, 1992; Rock et al, 1992; Fuller et al, (Alltech, Lancashire, UK) and Vydac 201TP54 1993; Yeum et al, 1996). However, data from the USA 250mm64.5mm analytical column (Separations Group, indicate that under usual dietary conditions, the associations California, USA). The mobile phase was acetonitrile- between dietary and plasma carotenoid concentrations are methanol-dichloromethane (75:20:5 by volume) containing moderate and generally do not exceed correlation coef®- 0.05% v=v triethylamine and using 0.05 mol ammonium cients more than r 0.5 (Ascherio et al, 1992; Forman et al, acetate in the methanol component of the mobile phase. 1993). As far as we are aware, in Europe, only one Detection involved two on-line UV=VIS detectors (Shi- published study has examined the association between matzu, Japan), with all carotenoids detected at 450nm, individual carotenoids in diet and serum and was limited while the UV detector was programmed to change from to a group of women aged 50 ± 65y (Scott et al, 1996). wavelength 325 ± 295nm during the analysis, in order to As participants in a multi-centre study, leading on to an detect retinol, g-tocopherol and a-tocopherol. The data intervention phase, volunteers were required to complete an were processed using Millennium 2.1 software data proces- especially developed, but not previously evaluated FFQ. sing package (Waters Corporation, Milford, USA). Six The FFQ which was devised by reference to existing FFQ carotenoids including lutein, zeaxanthin, b-cryptoxanthin, in use at participating centres, necessitated inclusion of a all-trans lycopene, a-carotene and all-trans b-carotene suf®ciently wide range of foods to facilitate its use in each were quanti®ed by reference to ®ve point calibration of the ®ve participating European countries. curves. Total plasma carotenoids were computed by sum- The primary aim of this present study was to assess ming the ®ve individual carotenoids. For each subject, dietary concentrations of individual carotenoids using an triplicate plasma samples from screening and from the established dietary method, to assess plasma concentrations day following completion of the estimated records were of individual carotenoids and to examine the associations run on the same day. Plasma b-carotene concentrations in between these dietary and plasma carotenoid concentrations samples from the Fat soluble Vitamin Quality Assurance in younger and older groups of Irish adults. A secondary Program deviated in the range 3 ± 8% from the mean of the aim of this present study was to evaluate the performance of analyte concentration, in the programme conducted by the the FFQ, by reference to an established dietary method and National Institute of Standards and Technology (NIST, by reference to plasma carotenoid concentrations, in Gaithesburg, USA). By the criteria of this programme, younger and older groups of Irish volunteers. our laboratories performance is evaluated as acceptable relative to the current state of the practice for measurement of b-carotene. Methods As part of a multi-centre European study, 69 healthy Dietary analysis volunteers aged between 25 ± 45y and 57 healthy volun- Volunteers completed seven consecutive days of estimated teers 65y were recruited for this study and two follow-on records. Detailed written and verbal instructions on how to supplementation studies. The studies on the younger and record the amount of food and drink consumed during the older groups were carried out ten months apart. All volun- 7d were given to volunteers. Volunteers were visited four teers were screened by a medical history, physical exam- times during the recording period and any discrepancies ination, and biochemical and haematological pro®le. corrected. The weights of all foods were calculated (g=d), Subjects had normal lipid metabolism as indicated by with the aid of a photographic atlas and standard portion fasting serum cholesterol and triacylglycerol concentra- sizes (Ministry of Agriculture, Fisheries and Food, 1993). tions. Body Mass Index (BMI) ranged from 19 ± FFQ relating to eating habits over the previous three 31kg=m2. Subjects were not adhering to any special diets months was also completed by all volunteers. The FFQ was and were non-smokers. Informed consent was given by all comprised of a list of 110 food items and was divided into Carotenoids in diet and plasma YL Carroll et al 646 seven sections: green, red-orange, white-yellow coloured carotenoid intakes expressed as nutrient densities. The vegetables, fruits, processed foods, dairy products, other Mann ± Whitney U test was also used to examine differ- foods. Eleven options for frequency of consumption includ- ences in plasma carotenoid concentrations in younger and ing; 1, 2, 3, 4, 5, 6, 7 (times per week), fortnightly, older groups. Differences between male and female esti- monthly, seldom and never, were given in all sections mated record carotenoid intakes (expressed as nutrient except the dairy and other food sections. Foods were densities) and differences in plasma carotenoids were quanti®ed in the most appropriate units, for example, examined by the Mann ± Whitney U test. Spearman's slices of cucumber, tablespoons of baked beans, with rank correlation coef®cients were calculated to investigate reference to standard portion sizes (Ministry of Agriculture, the associations between dietary and plasma data and Fisheries and Food, 1993). Intake of foods from the FFQ between the two dietary methods. Plasma carotenoids was calculated in g=d. were also adjusted for BMI, plasma triglycerides and The estimated records and FFQ from both studies were plasma cholesterol by General Linear Models. Adjusting coded by one person. Dietary carotenoids were quanti®ed for these variables did not improve correlations and the by reference to a comprehensive database which was crude correlations are presented. The ability of the FFQ to incorporated into a computerised dietary analysis pro- correctly classify volunteers into the highest and lowest gramme, Comp-Eat (Nutrition Systems, London, UK). tertiles of the estimated record carotenoid distribution was Compiled as part of the larger multi-centre study, this examined. The ability of each of the dietary methods to carotenoid food composition database was based on pub- correctly classify volunteers into the highest and lowest lished values, many of which were established in labora- tertiles of the plasma carotenoid distribution was also tories of participants in this multi-centre study (Granado et examined. The Cochran Q test was used to determine al, 1992; Olmedilla et al, 1993; Hart & Scott, 1995). Most whether the percentage of volunteers correctly classi®ed of the data was derived from Finnish, Dutch, Spanish and by estimated records into the same tertile of the plasma UK sources (Heinonen et al, 1988, 1989; Ollilainen et al, carotenoid distributions were signi®cantly different from 1988, 1989; Granado et al, 1992; Olmedilla et al, 1993; the percentage of volunteers correctly classi®ed by FFQ. Vollebregt & Feskens 1993; Hart & Scott, 1995), with The extent of misclassi®cation into opposite tertiles, was reference to data from other sources limited to a small also recorded and examined by the Cochran Q test. The number of food items that had not been analysed in the within- and between-person coef®cient of variation of above publications (Holland et al, 1991; Tonucci et al, individual carotenoids in plasma and estimated records 1995; Burlingame, 1993; Mangels et al, 1993). Values for were calculated by reference to values of within- and ®ve categories of carotenoids including luteinzeaxanthin, between person variances, obtained from repeated mea- b-carotene, a-carotene, lycopene and b-cryptoxanthin were sures ANOVA. included in the database. Results Statistical analysis Sixty four subjects in the younger and 54 subjects in the Characteristics of the study populations are shown in Table older groups, completed the studies. Five subjects in the 1. The average age of volunteers was 31y in the younger younger groups were excluded from analysis because FFQ groups and 70y in the older groups. All volunteers were in were not completed. Three subjects in the older group were good health and had biochemical and haematological excluded because they did not complete either the diet values within the reference range. record or the FFQ. Statistical analysis was performed with Mean dietary carotenoid intakes of males and females, Datadesk 4, 2 statistical software package (Data Descrip- assessed by estimated records and FFQ, are presented in tion Inc., New York, USA) and SPSS (SPSS Inc, Chicago, Table 2. In both age groups, the predominant dietary USA) statistical software package (Norusis SPSS Inc.). The carotenoids reported by both estimated records and FFQ effects of age group and dietary methods on dietary were b-carotene and lycopene. There was no signi®cant carotenoid intakes were examined by ANOVA with two- difference between males and females in absolute carote- way interaction. Differences between estimated record and noid intakes, assessed by either dietary method, in any of FFQ carotenoid intakes in younger and older males and in the age groups. The effects of age group and dietary younger and older females were examined by post-hoc method on dietary carotenoid intake in male and females Bonferroni tests. The Mann ± Whitney U test was used to were evaluated by ANOVA with two-way interactions. In examine differences between older and younger groups males, dietary method had a signi®cant effect on the dietary

Table 1 Characteristics of groups aged 24 ± 45y and 65y

24 ± 45y 65y

Males (n 32) Females (n 32) Males (n 25) Females (n 29)

Mean s.d. Mean s.d. Mean s.d. Mean s.d.

Age (y) 31 6 32 7 70 4 71 4 BMI (kg=m2) 24 2 24 3 26 2 26 3 Plasma lipids (mmol=L) Total cholesterol 4.7 0.9 4.8 0.7 5.3 0.9 5.9 0.7 HDL cholesterol 1.3 0.3 1.7 0.4 1.1 0.3 1.3 0.3 Triacylglycerols 1.1 0.4 0.9 0.3 1.2 0.6 1.2 0.4 Haemoglobin (g=dL) 15.4 0.7 13.3 0.6 14.7 0.8 13.3 0.8 Albumin (g=L) 46.9 2.3 44.9 2.2 42.8 2.2 42.6 1.9 Carotenoids in diet and plasma YL Carroll et al 647 Table 2 Dietary carotenoid intakes (mg=d) assessed by estimated records and FFQ in groups of men and women aged 24 ± 45y and 65y

Younger Males (n 32) Older Males (n 25) Younger Females (n 32) Older Females (n 29)

Mean s.d. Mean s.d. Mean s.d. Mean s.d.

Estimate records a-carotene 775 656 812 651 771 598 973 530 b-carotene 2921 1797 3099 2060 2850 1539 3358 1447 b-cryptoxanthin 189 288 116 153 165 157 108 160 Lycopene 3198 4129 2092 1809 2877 2066 2285 2367 Luteinzeaxanathin 1005 675 778 320 943 385 1015 463 Total carotenoids 8089 5002 6899 4624 7605 3007 7739 3245 Food frequency questionnaire a-carotene 2285 1470* 1217 1375 2426 1872* 1223 797 b-carotene 8077 4463* 5277 4607 8795 6022* 5496 3035* b-cryptoxanthin 471 466* 295 346 727 803* 317 338 Lycopene 7642 6262* 2026 2211 8045 9393* 4615 5368 Luteinzeaxanathin 2323 1436* 1877 1462* 2615 2120* 2120 1277* Total carotenoids 20800 9563* 10693 8475 22608 16368* 13773 8823*

*P 0.05; signi®cantly different from estimated records by post hoc Bonferroni tests. intake of all carotenoids examined. Age group had a other groups. The percentage of volunteers classi®ed into signi®cant effect on the dietary intake of the following: the same tertiles of the estimated record carotenoid dis- a-carotene (P 0.016), b-carotene (P 0.05), lycopene tributions also tended to be higher in younger men than in (P 0.007) and total carotenoids (P 0.001). There was other groups. A high proportion of volunteers (6 ± 36%) also a signi®cant interaction between age group and dietary were misclassi®ed by FFQ into opposite tertiles of the method for these carotenoids. In females, dietary method estimated records carotenoid distributions. In younger had a signi®cant effect on the dietary intake of all carote- women, there was more misclassi®cation by FFQ into noids examined. Age group had a signi®cant effect on the opposite (36%), than correct classi®cation (32%) into the dietary intake of the following: a-carotene (P 0.014), b- same tertiles of the estimated record b-carotene distribu- carotene (P 0.034), b-cryptoxanthin (P 0.006) and total tion. carotenoids (P 0.015). There was also a signi®cant inter- In younger volunteers, there was no signi®cant differ- action between age group and dietary method for these ence between men and women in estimated record carote- carotenoids. In the younger group, FFQ gave approximately noid intakes expressed in terms of nutrient density (Table 3-fold higher estimates of carotenoid intakes than estimated 4). In the older groups, the intakes of a-carotene (P 0.05), records which the post-hoc Bonferroni test found to be luteinzeaxanthin (P 0.01) and total carotenoids signi®cant (Table 2). In the older group, FFQ carotenoid (P 0.05), expressed as nutrient densities, were higher in intakes were signi®cantly higher than estimated records for women than in men (Table 5). The intakes of carotenoids, all carotenoids except a-carotene, b-cryptoxanthin and assessed by estimated records and expressed as nutrient lycopene. densities, did not differ signi®cantly between the older and The most consistent association between FFQ and esti- younger groups. mated records was observed for b-cryptoxanthin (Table 3). The plasma concentrations at screening and baseline Spearman correlation coef®cients ranged from r 0.55 ± were averaged and the mean plasma concentrations are 0.62 (P 0.001) and 59 ± 75% of volunteers were classi®ed shown in Table 6. Spearman correlation coef®cients in the into the same tertiles of the estimated record b-cryptox- range r 0.63 ± 0.91 were observed between repeat screen- anthin distribution. With the exception of younger males, ing and baseline plasma carotenoid concentrations in both there was little association between intakes of most car- age groups. In the younger groups, plasma carotenoids otenoids, assessed by estimated records and FFQ, in the were ranked in the order: b-carotene, lycopene and lutein in males, whereas in women the order was: b-carotene, b- Table 3 Effect of age and dietary method on dietary carotenoid intake in cryptoxanthin and lycopene. In the older groups, b-caro- groups of men and women aged 24 ± 45y and 65y tene, lutein and a-carotene were the predominant plasma carotenoids in men and women. There were no signi®cant Effect of Effect of dietary Age group6 differences observed between plasma carotenoid concen- age group method dietary method PP Ptrations in men and women in any of the age groups. With the exception of a-carotene and b-carotene, the younger Males men and women had higher plasma carotenoid concentra- a-carotene 0.016 0.001 0.010 tions than the older men and women (P < 0.005). Plasma a- b-carotene 0.048 0.001 0.025 b-cryptoxanthin 0.055 0.001 0.418 carotene concentrations were signi®cantly higher in older Lycopene 0.001 0.007 0.005 women and in the total older group than in younger women Luteinzeaxanthin 0.106 0.001 0.597 and the total younger group (P < 0.005). Total carotenoids 0.001 0.001 0.002 In both younger men and women, positive correlations Females a-carotene 0.014 0.001 0.001 between estimated record and plasma values were observed b-carotene 0.034 0.001 0.004 for several carotenoids, with the exception of b-carotene b-cryptoxanthin 0.006 0.001 0.036 (Tables 7 and 8). In the older groups, there was little Lycopene 0.054 0.001 0.172 association between estimated record and plasma carote- Luteinzeaxanthin 0.368 0.001 0.227 noid concentrations, except for b-cryptoxanthin in men Total carotenoids 0.015 0.001 0.012 (r 0.46, P 0.04) and total carotenoids in women Carotenoids in diet and plasma YL Carroll et al 648 Table 4 Spearman's rank order correlation coef®cients and percentage of subjects correctly classi®ed in same tertile and misclassi®ed in opposite tertile for dietary carotenoid concentrations assessed by estimated records and FFQ in groups of men and women aged 24 ± 45y and 65y

24 ± 45y 65y

Spearman % of classi®cation Spearman % of classi®cation correlation correlation coef®cients Same Opposite coef®cients Same Opposite r tertiles tertile r tertiles tertile

Males a-carotene 0.29 55 27 0.25 38 25 b-carotene 0.27* 41 27 0.23 38 19 b-cryptoxanthin 0.55*** 64 9 0.62*** 75 6 Lycopene 0.30*** 45 23 0.12 31 31 Luteinzeaxanathin 0.37** 50 23 0.40 44 13 Total carotenoids 0.39** 64 23 0.27 58 19 Females a-carotene 0.06 32 27 0.14 30 30 b-carotene 0.03 32 36 0.24 50 25 b-cryptoxanthin 0.61*** 59 14 0.48 60 10 Lycopene 0.32 45 23 0.28 45 20 Luteinzeaxanathin 0.32 41 18 0.44* 55 10 Total carotenoids 0.42 55 18 0.33 50 25

*P 0.05; **P 0.01; ***P 0.01 signi®cant correlation between estimated records and FFQ.

Table 5 Energy intakes (MJ=d) (excluding alcohol) and nutrient was evident in males (r 0.68, P < 0.0001) and females densities (mg=MJ) measured by estimated records in groups of men and (r 0.53, P 0.0025). In the group of older men, positive women aged 24 ± 45y and 65y correlations were observed for several carotenoids compar- Younger Younger Older Older ing FFQ and plasma carotenoid concentrations. In the males females males females group of older women, the only positive correlation (n 32) (n 32) (n 25) (n 29) between FFQ and plasma carotenoids occurred for b- Mean s.d. Mean s.d. Mean s.d. Mean s.d. cryptoxanthin (r 0.65, P 0.0001). More younger (36 ± 73%) than older volunteers (20 ± 60%) were correctly Energy 10.4 2.4 8.6 1.5 9.4 1.9 7.8 1.4 classi®ed by estimated records into plasma tertiles. a-carotene 85 73 101 82 87 63 129 80* Within the younger groups and within the older groups, b-carotene 320 205 370 222 335 203 441 216 b-cryptoxanthin 19 24 21 19 13 17 13 18 there was no signi®cant difference between estimated Lycopene 314 346 379 302 227 201 307 357 records and FFQ in their ability to correctly classify Lutein 111 85 119 49 86 34 135 68** volunteers into the same tertiles of the plasma carotenoid zeaxanathin distribution. Similarly, within the younger and within the Total carotenoids 852 486 989 480 749 475 1025 517* older groups, the percentage of volunteers misclassi®ed by *P 0.05; **P 0.01 signi®cant difference between males and females in estimated records into opposite tertiles of the plasma median carotenoid intakes expressed in mg=MJ energy (Mann Whitney U carotenoid distributions, did not differ signi®cantly from test). the percentage misclassi®ed by FFQ. In the younger groups, a mean of 46d elapsed between collection of screening and baseline plasma samples, with a minimum Table 6 Average plasma carotenoid concentrations (nmol=L) of screening and baseline plasma samples in groups of men and women interval of 13d and a maximum interval of 57d. 85% of aged 24 ± 45y and 65y screening and baseline samples were collected within 40 ± 57d of one another in the younger age groups, with a Younger Younger Older Older median interval of 51d. In the older groups, a mean of 38d males females males females (n 32) (n 32) (n 25) (n 29) elapsed between collection of screening and baseline plasma samples, with a minimum interval of 19d and a Mean s.d. Mean s.d. Mean s.d. Mean s.d. maximum interval of 56d. The median interval was 39d in the older groups. The within-person coef®cient of variation a-carotene 92 43 107 53* 122 81 166 94 b-carotene 393 194 462 186 472 222 553 254 was consistently lower than the between-person coef®cient b-cryptoxanthin 191 113* 296 226* 117 96 123 62 of variation in individual carotenoids, in both plasma and Lycopene 297 125** 253 110** 91 69 111 61 estimated records, as shown in Table 9. Within- and Lutein 207 70** 237 73** 140 62 170 56 between-person variability in dietary carotenoids assessed Zeaxanathin 97 40** 83 25** 49 25 57 19 Luteinzeaxanathin 304 99** 320 94** 187 73 223 73 by estimated records, exceeded the variability in plasma Total carotenoids 1693 473** 1813 567** 992 395 1177 436 carotenoid concentrations. In all groups, the ratio of within- person:between-person variability in estimated records, *P 0.01; **P 0.001 signi®cant difference between older and younger was lowest for b-cryptoxanthin. groups (Mann Whitney U test).

Discussion (r 0.27, P 0.05). In younger men and women, the only signi®cant association between FFQ and plasma carote- Data on the consumption and plasma concentrations of noids was observed for b-cryptoxanthin and this correlation individual carotenoids and the associations between these Carotenoids in diet and plasma YL Carroll et al 649 Table 7 Spearman's rank order correlation coef®cients and percentage of men correctly classi®ed in same tertile and misclassi®ed in opposite tertile between average plasma and estimated records or FFQ carotenoid concentrations in groups of men aged 24 ± 45y and 65y

Younger males Older males

Spearman % of classi®cation Spearman % of classi®cation correlation correlation coef®cients Same Opposite coef®cients Same Opposite r tertiles tertile r tertiles tertile

Estimated records a-carotene 0.60* 68 5 0.32 44 19 b-carotene 0.05 36 27 0.04 31 25 b-cryptoxanthin 0.66*** 59 14 0.46* 50 13 Lycopene 0.52* 55 14 0.14 32 38 Luteinzeaxanathin 0.44* 55 9 0.07 31 25 Total carotenoids 0.07 36 23 0.11 50 19 Food frequency questionnaire a-carotene 0.31 45 18 0.70*** 63 6 b-carotene 7 0.09 32 36 0.34 50 19 b-cryptoxanthin 0.68*** 68 5 0.54** 50 13 Lycopene 0.26 50 18 0.47* 50 13 Luteinzeaxanathin 0.27 45 23 0.39 50 13 Total carotenoids 7 0.16 23 36 0.55* 44 6

*P 0.05; **P 0.01; *** P 0.001 signi®cant correlation between carotenoid concentrations in plasma and diet assessment method. Table 8 Spearman's rank order correlation coef®cients and percentage of women correctly classi®ed in same tertile and misclassi®ed in opposite tertile between average plasma and estimated records or FFQ carotenoid concentrations in groups of men aged 24 ± 45y and 65y

Younger females Older females

Spearman % of classi®cation Spearman % of classi®cation correlation correlation coef®cients Same Opposite coef®cients Same Opposite r tertiles tertile r tertiles tertile

Estimated records a-carotene 0.42** 55 18 0.11 45 35 b-carotene 0.29 50 27 7 0.09 20 45 b-cryptoxanthin 0.74*** 73 5 0.52 60 10 Lycopene 0.43** 59 14 0.32 40 30 Luteinzeaxanathin 0.32 50 27 0.14 40 25 Total carotenoids 0.41* 45 23 0.27* 45 25 Food frequency questionnaire a-carotene 0.24 45 23 7 0.14 30 35 b-carotene 0.11 32 41 0.12 40 25 b-cryptoxanthin 0.53** 59 14 0.65*** 75 0 Lycopene 0.50 59 9 0.44 60 10 Luteinzeaxanathin 7 0.02 36 32 0.10 35 25 Total carotenoids 0.23 45 27 0.34 55 20

*P 0.05; **P 0.01; ***P 0.001 signi®cant correlation between carotenoid concentrations in plasma and diet assessment method. variables, are not well documented in European popula- that the discrepancy between Irish and UK data is smaller tions. The primary purpose of the present study was to than the discrepancies within UK data (Maisey et al, 1995; assess plasma carotenoid concentrations, to assess dietary Scott et al, 1996). These differences between the various intakes of individual carotenoids using an established diet- studies may be attributable to the use of different food ary method, and to examine the relationship between these composition databases, different dietary assessment meth- dietary and plasma variables in groups of younger and older ods and population differences. In this present study, a volunteers. modi®ed version of the carotenoid food composition data- In agreement with other studies, the predominant dietary base developed by Scott et al (1996) was used. Weighed carotenoids reported by estimated records in both the records, which are known to be associated with under- younger and older groups are b-carotene and lycopene reporting of food intakes (Black et al, 1993) and analysis of (Forman et al, 1993; Yong et al, 1994; Scott et al, 1996). only fruit and vegetable sources of carotenoids (Scott et al, The concentrations of total carotenoids reported by esti- 1996), may also account for these differences in carotenoid mated records in both age groups are similar to those intakes between Ireland and the UK. The estimated record reported in the USA (Forman et al, 1993; Yong et al, data in the present study shows that composite dishes, for 1994). However, the intake of b-carotene assessed by example, soups and pizza are signi®cant sources of carote- estimated records was 1.5 ± 4-times higher than the levels noids. In this present study, some carotenoids were also reported in older women in the UK, but it should be noted partly derived from animal foods with butter, dairy Carotenoids in diet and plasma YL Carroll et al 650 Table 9 Within-(CWw) and between-subject (CVb) coef®cient of variation in plasma and estimated records of men and women aged 24 ± 45y and 65y

Younger males (n 32) Younger females (n 32) Older males (n 25) Older females (n 29)

CVw CVb CVw CVb CVw CVb CVw CVb Plasma a-carotene 27.03 66.66 33.39 70.42 25.06 94.09 24.14 80.30 b-carotene 29.44 69.61 25.21 56.92 30.02 66.52 25.82 65.06 b-cryptoxanthin 31.02 83.69 46.66 107.88 74.16 115.85 29.24 72.18 Lycopene 28.02 59.35 25.99 61.51 33.24 106.32 32.82 77.97 Luteinzeaxanthin 16.52 45.88 14.50 41.58 19.26 55.55 11.24 46.18 Total carotenoids 17.63 39.49 15.74 44.20 26.18 56.44 18.03 52.40 Estimated Records a-carotene 162.47 224.21 151.57 204.98 118.33 212.05 121.87 144.44 b-carotene 114.26 162.77 104.78 142.89 84.44 175.90 90.27 114.02 b-cryptoxanthin 137.44 404.17 116.00 251.99 110.50 348.07 139.60 393.11 Lycopene 175.88 341.64 182.39 189.99 99.01 228.81 224.12 274.10 Luteinzeaxanathin 84.63 177.47 107.19 108.15 54.94 108.79 93.31 120.60 Total carotenoids 90.87 163.62 95.87 104.60 68.11 177.34 98.64 110.95

products and eggs and egg dishes identi®ed as signi®cant moderately stable, as shown by correlations in the range sources of lutein and b-carotene. This agrees with a report r 0.63 ± 0.91, between repeat plasma samples and within- from the USA, where dairy products contribute approxi- person coef®cients of variation generally less than 30% in mately 7% of the total pro-vitamin A carotenoid intake this present study. (Block, 1994). Similarly, data from Finland indicate that In the younger group, there were positive associations margarines, oils, butter, milk products and eggs provide between most plasma and estimated record concentrations 24 ± 25% of b-carotene and lutein intakes in men (Jarvinen, of carotenoids, except b-carotene. Although moderate, the 1995). In this present study, the inter- and intra-individual magnitude of the correlations between estimated records coef®cients of variation in estimated record carotenoid and individual plasma carotenoid concentrations observed intakes are high in all age groups, but variation in carote- in the younger groups compare well with those reported noid intakes was always greater between individuals than elsewhere, by estimated records, weighed intakes and diet within individuals. In estimated records, b-cryptoxanthin histories (Forman et al, 1993; van Staveren et al, 1994; tended to show that lowest ratio of within-person:between- Yong et al, 1994; Scott et al, 1996). In the present study, person variability in carotenoid intakes and this may good reproducibility of plasma carotenoids was observed as account for its stronger associations with plasma b-cryp- shown by correlations in the range r 0.65 ± 0.85, and toxanthin concentrations. coef®cients of variation generally ranging from 20 ± 30%. b-Carotene was the predominant plasma carotenoid in This level of reproducibility is similar to that observed by all groups in this present study. Plasma carotenoid con- others for plasma carotenoids (Campbell et al 1994, Yong centrations were ranked in the order: b-caro- et al, 1994, Apgar et al, 1996, Scott et al, 1996). Therefore, tene > lutein > a-carotene > b-cryptoxanthin > lycopene > the interval between screening and baseline plasma samples zeaxanthin in older groups. This contrasts with the pro®le is unlikely to account for the observation of only moderate of the younger groups in which plasma lycopene concen- associations between estimated record and plasma carote- trations were considerably higher and ranked second or noids. In the younger groups, the association between third to b-carotene concentrations. Others have shown that estimated record and plasma carotenoid concentrations lycopene and b-carotene were the major plasma carote- was strongest for b-cryptoxanthin, and was stronger than noids in younger and older subjects respectively (Yeum et that reported elsewhere (Forman et al, 1993; van Staveren al, 1996). In this present study, all individual plasma et al, 1994; Yong et al, 1994; Scott et al, 1996). No carotenoid concentrations, with the exception of a- and b- association between estimated record and plasma b-caro- carotene, were higher in younger groups than older groups. tene concentrations was evident in any of the age groups in Plasma a-carotene concentrations were higher in older than the present study. Others report signi®cant, but moderate, younger volunteers. It has been shown elsewhere that age is correlations for b-carotene in the range r 0.27 ± 0.52 (van inversely related to plasma lycopene concentrations (Brady Staveren et al, 1994; Yong et al, 1994; Scott et al, 1996), et al, 1996). Others have shown that age is directly but not all studies have observed signi®cant correlations associated with serum b-carotene concentrations, although between plasma and estimated record b-carotene concen- an age effect on plasma b-carotene has not always been trations (Forman et al, 1993). Within- and between-person reported (Hallfrisch et al, 1994; Brady et al, 1996; Santos et variation in estimated record and plasma b-carotene con- al, 1996). Whether these effects of age on plasma carote- centrations was similar to that observed for other carotenoid concentrations are due to physiological mechanisms, noids and may not explain the lack of association between or are merely attributable to different dietary patterns in estimated record and plasma b-carotene concentrations. A younger and older groups, has not yet been established. The possible explanation for the poor association between relative order of plasma carotenoid concentrations appears estimated record and plasma b-carotene concentrations, to be population speci®c. b-Cryptoxanthin is the predomi- may be the wide range of foods in which it is distributed. nant plasma carotenoid in Spanish women, whereas data Data from the present study indicate that approximately 10 from the US indicate that lycopene is the predominant foods account for 90% of b-carotene consumption com- plasma carotenoid (Micozzi et al, 1992; Olmedilla et al, pared to approximately 4 foods accounting for 90% of 1994). Plasma concentrations of individuals carotenoids are lycopene and b-cryptoxanthin intakes. Other studies also Carotenoids in diet and plasma YL Carroll et al 651 show that the number of foods contributing to b-carotene found in the intake of b-carotene (Scott et al, 1996). intake is considerably greater than that of lycopene and b- However, this UK study did note that the strength of the cryptoxanthin (Granado et al, 1996). This extended number association between dietary and plasma b-carotene concen- of sources of b-carotene has implications for absorption, trations varied according to season, with no signi®cant matrix composition and quanti®cation by the dietary assess- association observed in winter, compared to a signi®cant ment tool, and may thereby affect the relationship between association in spring (Scott et al, 1996). Dietary data were dietary and plasma b-carotene. The food matrix in which recorded in November in the younger groups and in January the carotenoids are provided may affect the absorption and in the older groups, and in the context of the ®ndings in the physical inaccessibility of carotenoids in plant tissues, for UK study, this may help to explain why no positive example in pigment protein, complexes may reduce their associations were observed for b-carotene with any of the bioavailability (Brown et al, 1989; de Pee et al, 1995). dietary methods in the present study. As part of a multi- By contrast with the many positive correlations observed centre study the carotenoid food composition database used in the younger groups, only b-cryptoxanthin in men and in the present study was compiled from published analysis total carotenoids in women showed signi®cant associations of foods sourced mainly in Europe. This could in¯uence the between plasma and estimated record carotenoid concen- diet-plasma relationships observed in the present study trations, in the older groups. The lack of association which was con®ned to Irish population groups. A carote- between estimated record and plasma carotenoid distribu- noid food composition database, based on analysis of foods tions in the older groups was also re¯ected in the poor sourced in Ireland, is not available. Many carotenoid food ability of estimated records to correctly classify individuals composition databases are compiled from published data into extremes of the plasma carotenoid distributions, with relating to foods sourced in several countries (Mangels et as many as 45% being misclassi®ed into opposite tertiles. al, 1993; West & Poortvliet, 1993). Comparison of carote- This low level of association between plasma and estimated noid intakes estimated using several food composition records in the older groups, does not agree with the trends databases has shown that the use of a single database may of signi®cant associations observed for several carotenoids give misleading carotenoid intake values (Granado et al, in the younger groups, in this present study and in other 1997). Analysis of dietary data by reference to two different studies (Ascherio et al, 1992; Forman et al, 1993; Yong et carotenoid composition databases, indicated that although al, 1994). The poor agreement between estimated record estimates of carotenoid intake differed signi®cantly, only and plasma carotenoid concentrations in the older group minor differences in carotenoid rankings and diet-serum was unanticipated and possible explanations may include correlations were observed using either data source (Van- physiological differences, differences in validity of esti- denlangenberg et al, 1996). mated records in younger and older volunteers, and seaso- A secondary purpose of this present study was to assess nal differences in consumption of carotenoids. An alteration the validity of a FFQ developed for assessing dietary in gastric acid secretion with age could be complicating the carotenoid intakes in several European countries. Validity relationship between dietary and plasma carotenoid con- was assessed by comparison of mean intakes with esti- centrations (Hartz et al, 1992; Tang et al, 1996). However, mated records, by examining correlations with estimated a study of patients positive for Helicobacter pylori,(a records, by observing the extent of correct classi®cation condition associated with atrophic gastritis and hypochlor- and gross misclassi®cation into tertiles of estimated records hydria) has shown that these patients do not have low carotenoid concentrations, by examining correlations with concentrations of a-carotene, b-carotene or lycopene (San- plasma and by observing the extent of correct classi®cation derson et al, 1997). Furthermore, in this present study, and gross misclassi®cation into tertiles of plasma carote- screening excluded volunteers with gastrointestinal disor- noid concentrations. The FFQ concentrations of dietary ders, so it is unlikely that this is an adequate explanation. carotenoids were approximately 2 ± 3-fold higher than esti- Female gender and age over 45y have been shown to be mated records in the present study. Furthermore, the FFQ predictors of underreporting of food consumption (Hirvo- total carotenoid and b-carotene intakes in both age groups, nen et al, 1997) and may be another reason why the in this present study, are at least 2-fold higher than those associations between estimated record and plasma carote- reported by FFQ in other studies (Forman et al, 1993; noids were poor in the older groups. Underreporting, as Mares-Perlman et al, 1993; Vandenlangenberg et al, 1996; de®ned by energy intakes less than 1.1 times the basal OckeÂ et al, 1997). Overestimation of carotenoid intakes by metabolic rate (Goldberg et al, 1991), was evident in the FFQ by 10 ± 30% of estimated records and by 38 ± 50% of present study. However, similar percentages of volunteers weighed record concentrations have been observed (Yong in both age groups (16% of men and 9 ± 10% of women in et al, 1994; Bingham et al, 1994). This is attributed to a both age groups) underreported energy intakes in the pre- greater reported frequency of consumption in the question- sent study and this degree of underreporting is not unusual naire methods than actually measured by the weighed in dietary studies (Black et al, 1991). If underreporting was records (Bingham et al, 1994). In this present study, distorting the diet-plasma relationship, this bias should also correlations between dietary intakes of most carotenoids have in¯uenced the associations between plasma and esti- assessed by estimated records and FFQ, were poor in most mated record carotenoid concentrations in the younger groups except younger men. In contrast to this present groups. Spanish data revealed that marked seasonality in study, modest, but signi®cant correlations between esti- some food items, caused differences in the dietary supply of mated record and FFQ have been observed for several b-cryptoxanthin and lycopene (Granado et al, 1996). This carotenoids in young men and women (Rimm et al, 1992; group also reported signi®cant seasonal variation in some, Forman et al, 1993; Yong et al, 1994). Misclassi®cation by but not all, serum carotenoid concentrations of Spanish men FFQ in opposite tertiles of the estimated record carotenoid and women (Olinedilla et al, 1994). It remains to be distributions ranged from 19 ± 36%, and is much greater determined whether seasonal effects are profound in the than that reported elsewhere (Bingham et al, 1994; Boni- Irish population. In the UK, no seasonal difference was facj et al, 1997). The only group in which several Carotenoids in diet and plasma YL Carroll et al 652 signi®cant associations between FFQ and plasma carote- References noids were observed was in older men. Other studies have Apgar J, Makdani D, Sowell AL, Gunter EW, Hegar A, Potts W, Rao D, reported signi®cant associations between several plasma Wilcox A & Smith JCD (1996): Serum carotenoid concentrations and and dietary carotenoids, estimated by FFQ (Coates et al, their reproducibility in Belize. Am. J. Clin. Nutr. 64, 726 ± 730. 1991; Ascheno et al, 1992; Vandenlangenberg et al, 1996). Ascherio A, Stampfer MJ, Colditz GA, Rimm EB, Litin L & Willett WC Conversely, several studies failed to note a positive rela- (1992): Correlations of vitamin A and E intakes with the plasma concentrations of carotenoids and tocopherols among American men tionship between plasma and lycopene or b-carotene con- and women. J. Nutr. 122, 1792 ± 1801. centrations assessed by FFQ (Coates et al, 1991; Ascherio Bingham SA, Gill C, Welch A, Day K, Cassidy A, Khaw KT, Sneyd MJ, et al, 1992; Forman et al, 1993; Kardinaal et al, 1995; OckeÂ Key TJA, Roe, L & Day NE (1994): Comparison of dietary assessment et al, 1997). b-Cryptoxanthin was the only carotenoid for methods in nutritional epidemiology: weighed records v. 24h recalls, which signi®cant correlations were observed between food-frequency questionnaires and estimated-diet records. Br. J. Nutr. 72, 619 ± 643. plasma and FFQ in both age groups. Other studies have Black AE, Goldberg GR, Jebb SA, Livingstone MBE, Cole TJ & Prentice also shown that correlations between plasma and FFQ were AM (1991): Critical evaluation of energy intake data using fundamental stronger for b-cryptoxanthin than other carotenoids principles of energy physiology: 2. Evaluating the results of published (Forman et al, 1993; Vandenlangenberg et al, 1996). surveys. Eur. J. Clin. Nutr. 45, 583 ± 599. There was no statistical evidence, in the present study, Black AE, Prentice AM, Goldberg GR, Jebb SA, Bingham SA, Living- stone MBE & Coward WA (1993): Measurements of total energy that FFQ were better than estimated records in correctly expenditure provide insights into the validity of dietary measurements classifying individuals into extremes of the plasma carote- of energy intake. J. Am. Diet. Assoc. 93, 572 ± 579. noid distribution. Block G (1994): Nutrient sources of provitamin A carotenoids in the The 110 item FFQ administered in these studies American diet. Am. J. Epidemiol. 139, 290 ± 293. Bonifacj C, Gerber M, Scali J & Daures JP (1997): Comparison of dietary included only foods high in carotenoids, and this emphasis assessment methods in a Southern French population: use of weighed combined with the lack of representation of other foods, records, estimated-diet records and a food frequency questionnaire. Eur. may have accounted for its poor performance. It has been J. Clin. Nutr. 51, 217 ± 231. shown in another study that adding a list of carotenoid-rich Brady WE, Mares-Perlman JA, Bowen P & Stacewicz-Sapuntzakis M foods to a standard FFQ, did not improve the validity of the (1996): Human serum carotenoid concentrations are related to physio- logic and lifestyle factors. J. Nutr. 126, 129 ± 137. questionnaire (Enger et al, 1995). The presence of irrele- Brown ED, Micozzi MS, Craft NE, Bieri JG, Beecher G, Edwards BK, vant nutrient sources on a questionnaire may contribute to Rose A, Taylor PR & Smith JC Jr (1989): Plasma carotenoids in normal increased misclassi®cation rather than to increased preci- men after a single ingestion of vegetable or puri®ed b-carotene. Am. J. sion (Block, 1994). It has also been shown that fruit and Clin. Nutr. 49, 1258 ± 1265. vegetables, which are perceived as healthy foods, are most Burlingame B (1993): Carotenoid content of New Zealand foods (personal communication). In: The Carotenoid Content of Foods with Special often overreported by FFQ, while meats and dairy products, Reference to Developing Countries, ed. CE West & EJ Poorvliet. which are considered to be less healthy are most often Vitamin A Field Support Project (VITAL), International Science and underreported (Feskanich et al, 1993). Further development Technology Institute, Inc. Arlington, Virginia 22209, USA. is required to establish a valid FFQ suitable for use in Burton GW, Webb A & Ingold KU (1985): A mild, rapid, and ef®cient method of lipid extraction for use in determining vitamin E=lipid ratios. several European countries. Lipids 20, 29 ± 39. Campbell DR, Gross MD, Martine MC, Grandits GA, Flavin JL Potter JD (1994): Plasma carotenoids as biomarkers of vegetable and fruit intake. Conclusions Cancer Epidemiol., Biomarkers & Prev, 2, 493 ± 500. Chug-Ahuja JK, Holden JM, Forman MR, Mangels AR, Beecher GR & Estimated records indicate that b-carotene and lycopene are Lanza E (1993): The development and application of a carotenoid the major dietary carotenoids. The pro®le of plasma car- database for fruits, vegetables, and selected multicomponent foods. J. Am. Diet. Assoc. 93, 318 ± 323. otenoids show that b-carotene is the major carotenoid in Coates RJ, Eley JW, Block G, Gunter EW, Sowell AL, Grossman C & both age groups. Younger groups have higher plasma Greenberg RS (1991): An evaluation of food frequency questionnaire concentrations of lycopene, b-cryptoxanthin, lutein zeax- for assessing dietary intake of speci®c carotenoids and vitamin E among anthin and total carotenoids than older groups. Moderate low-income black women. Am. J. Epidemiol. 134, 658 ± 671. positive associations, similar to those observed in other de Pee S, West CE, Muhilal, Karyadi D & Hautvast JGAJ (1995): Lack of improvement in vitamin A status with increased consumption of dark- countries, exist between several plasma and estimated green leafy vegetables. Lancet 346, 75 ± 81. record dietary carotenoid concentrations in younger, but Enger SM, Longnecker MP, Shikany JM, Swenseid ME, Chen M-J, Harper not older groups. Plasma b-cryptoxanthin concentrations JM & Haile RW (1995): Questionnaire assessment of intake of speci®c were strongly associated with both estimated record and carotenoids. Cancer Epidemiol., Biomarkers & Prev. 4, 201 ± 205. FFQ concentrations in both age groups. b-Carotene in Feskanich D, Rimm EB, Giovanucci EL, Colditz GA, Stampfer MJ, Litin LB & Willett WC (1993): Reproducibility and validity of food intake plasma is not associated with estimated record concentra- measurements from a semiquantitative food frequency questionnaire. tions in any age group. The data imply that plasma J. Am. Diet. Assoc. 93, 790 ± 796. carotenoid concentrations may be a useful biomarker of Forman MR, Lanza E, Yong L-C, Holden JM, Graubard BI, Beecher GR, several carotenoids, excluding b-carotene, in groups aged Melitz M, Brown ED & Smith JC (1993): The correlation between two dietary assessments of carotenoid intake and plasma carotenoid con- 24 ± 45y. Further work is required to determine if plasma centrations: application of a carotenoid food-composition database. Am. carotenoids can be a useful biomarker of dietary carotenoid J. Clin. Nutr. 58, 519 ± 524. concentrations in older populations. Fuller CJ, Parker RS, Spielman A & Roe DA (1993): Carotenoid-depletion diet for use in long-term studies. J. Am. Diet. Assoc. 93, 812 ± 814. Acknowledgements ÐThis work has been carried out with ®nancial support Gaziano JM, Manson JE, Branch LG, Colditz GA, Willett WC & Buring JE (1995): A prospective study of consumption of carotenoids in fruits from the Commission of the European Communities (AIR2 CT93-0888 and vegetables and decreased cardiovascular mortality in elderly. Ann. DG XII SSMA), `Increased fruit and vegetable consumption within the Epidemiol. 5, 255 ± 260. EC: potential health bene®ts'. b-Cryptoxanthin and zeaxanthin standards Goldberg GR, Black AE, Jebb SA, Cole TJ, Murgatroyd PR, Coward WA for HPLC analysis were provided free of charge by Hoffman La Roche, and Prentice AM (1991): Critical evaluation of energy intake data using Switzerland. The technical assistance of SineÂad McCarthy and SiobhaÂn fundamental principles of energy physiology: 1. Derivation of cut-off Higgins are gratefully acknowledged. limits to identify under-recording. Eur. J. Clin. Nutr. 45, 569 ± 581. Carotenoids in diet and plasma YL Carroll et al 653 Granado F, Olmedilla B, Blanco I & Rojas-Hidalgo E (1997): Variability Olmedilla B, Granado F, Blanco I & Rojas-Hildago E (1993): Quantitation in the Intercomparison of Food Carotenoid Content Data: A User's of provitamin-A and non-provitamin A carotenoids in the fruits most Point of View. Crit. Rev. Food Sci. Nutr. 37, 621 ± 633. commonly consumed in Spain. In Food and Cancer Prevention: Granado F, Olmedilla B, Blanco I & Rojas-Hidalgo E (1992): Carotenoid Chemical and Biological Aspects, eds KW Waldron, IT Johnson & composition in raw and cooked Spanish vegetables. J. Agric. Food GR Fenwick, pp 141 ± 145. Cambridge: Royal Society of Chemistry. Chem. 40, 2135 ± 2140. Olmedilla B, Granado F, Blanco I & Rojas-Hildago E (1994): Seasonal Granado F, Olmedilla B, Blanco I & Rojas-Hidalgo E (1996): Major fruit and sex-related variations in six serum carotenoids, retinol, & a- and vegetable contributors to the main serum carotenoids in the Spanish tocopherol. Am. J. Clin. Nutr. 60, 106 ± 110. Diet. Eur. J. Clin. Nutr. 50, 246 ± 250. Rimm EB, Giovanucci EL, Stampfer MJ, Colditz GA, Litin LB & Willett Hallfrisch J, Muller DC & Singh VN (1994): Vitamin A and E intakes and WC (1992): Reproducibility and validity of an expanded self-adminis- plasma concentrations of retinol, b-carotene, and a-tocopherol in men tered semiquantitative food frequency questionnaire among male health and women of the Baltimore Longitudinal Study of Aging. Am. J. Clin. professionals. Am. J. Epidemiol. 135, 1114 ± 1126. Nutr. 60, 176 ± 182. Rock CL, Swendseid ME, Jacob RA & Mc Kee RW (1992): Plasma Hart DJ & Scott KJ (1995): Development and evaluation of an HPLC carotenoid levels in human subjects fed a low carotenoid diet. J. Nutr. method for the analysis of carotenoids in foods, and the measurement of 122, 96 ± 100. the carotenoid content of vegetables and fruits commonly consumed in Sanderson MJ, White KLM, Drake IM & Schorah CJ (1997): Vitamin E the UK. Food Chem. 54, 101 ± 111. and carotenoids in gastric biopsies: the relation to plasma concentra- Hartz SC, Rosenberg IH & Russell RM (1992): Nutrition in the Elderly; tions in patients with and without Helicobacter pylori gastritis. Am. J. The Boston Nutritional Survey. London: Smith-Gordon & Co Ltd. Clin. Nutr. 65, 101 ± 106. Heinonen M, Ollilainen V, Linkola E, Varo P & Koivistoinen P (1988): Santos MS, Meydani SN, Leka L, Wu D, Fotouhi N, Meydani M, Carotenoids and retinoids in Finnish foods: dietary fats. J. Food Comp. Hennekens CH & Gaziano JM (1996): Natural killer cell activity in Anal. 1, 334 ± 340. elderly men is enhanced by b-carotene supplementation. Am. J. Clin. Heinonen MI, Ollilainen V, Linkola EK, Varo PT & Koivistoinen PE Nutr. 64, 772 ± 777. (1989): Carotenoids in Finnish foods: vegetables, fruits and berries. Scott KJ & Hart DJ (1993): Further observations on problems associated J. Agric. Food Chem. 37, 655 ± 659. with the analysis of carotenoids by HPLC-2 : Column Temperature. Hirvonen T, Mannisto S, Roos E & Pietinen P (1997): Increasing Food Chem. 47, 403 ± 405. prevalence of underreporting does not necessarily distort dietary sur- Scott KJ, Thurnham DI, Hart DJ, Bingham SA & Day K (1996): The veys. Eur. J. Clin. Nutr. 51, 297 ± 301. correlation between the intake of lutein, lycopene and b-carotene from Holland B, Welch AA, Unwin ID, Buss DH, Southgate PAA & Southgate vegetables and fruits, and blood plasma concentrations in a group of DAT (1991): McCance & Widdowson's The Composition of Foods, 5th women aged 50 ± 65 years in the UK. Br. J. Nutr. 75, 409 ± 418. edition. London: The Royal Society of Chemistry and Ministry of Seddon JM, Ajani UA, Sperduto RD, Hiller R, Blair N, Burton TC, Farber Agriculture, Fisheries and Food. MD, Gragoudas ES, Haller J, Miller DT, Yannuzzi LA & Willett W for Jarvinen R (1995): Carotenoids, retinoids, tocopherols and tocotrienols in the Eye Disease Case-Control Study Group (1994): Dietary carotenoids, the diet; the Finnish mobile clinic health examination survey. Int. J. Vit. vitamins A, C, and E, and advanced age-related macular degeneration. Nutr. Res. 65, 24 ± 30. JAMA 272, 1413 ± 1420. Kardinaal AFM, Kok FJ, Ringstad J, Gomez-Aracena J, Mazaev VP, Shibata A, Paganini-Hill A, Ross RK & Henderson BE (1992): Intake of Kohlmeier L, Martin BC, Aro A, Kark JD, Delgado-Rodriguez M, vegetables, fruits, beta-carotene, vitamin C and vitamin supplements Riemersma RA, van't Veer P, Huttenen JK & Martin-Moreno JM and cancer incidence among the elderly: a prospective study. Br. J. (1993): Antioxidants in adipose tissue and risk of myocardial infarction: Cancer. 66, 673 ± 679. the EURAMIC study. Lancet 342, 1379 ± 1384. Souci SW, Fachmann W & Kraut H (1987): Food Composition and Kardinaal AFM, van't Veer P, Brants HAM, van den Berg H, van Nutrition Tables. Wissenschaftliche Verlagsgesellschaft mbH, Stuttgart, Schoonhoven J Hermus RJJ (1995): Relations between antioxidant vitamins in adipose tissue, plasma and diet. Am. J. Epidemiol. 141, Germany. Steinmetz KA & Potter JD (1996): Vegetables, fruit and cancer preven- 440 ± 450. Key TJA, Thorogood M, Appleby PN & Burr ML (1996): Dietary habits tion: A review. J. Am. Diet. Assoc. 96, 1027 ± 1039. and mortality in 11,000 vegetarians and health conscious people: results Street DA, Comstock GW, Salkeld RM, Schuep W & Klag MJ (1994): of a 17 year follow up. Br. Med. J. 313, 775 ± 779. Serum antioxidants and myocardial infarction: are low levels of Knekt P, Reunanen A, Jarvinen R, Seppanen R, Heliovaara M & Aromaa carotenoids and a-tocopherol risk factors for myocardial infarction? A (1994): Antioxidant vitamin intake and coronary mortality in a Circulation 90, 1154 ± 1161. longitudinal population study. Am. J. Epidemiol. 139, 1180 ± 1189. Tang G, Serfaty-Lacrosniere C, Camilo EM & Russell RM (1996): Gastric Maisey S, Loughridge J, Southon S & Fulcher R (1995): Variation in food acidity in¯uences the blood response to a b-carotene dose in humans. group and nutrient intake with day of the week in an elderly population. Am. J. Clin. Nutr. 64, 622 ± 626. Br. J. Nutr. 73, 359 ± 373. Tonucci LH, Holden JM, Beecher GR, Khachik F, Davis CS & Mulokozi Mangels AR, Holden JM, Beecher GR, Forman MR & Lanza E (1993): G (1995): Carotenoid content of thermally processed tomato-based food Carotenoid content of fruit and vegetables: An evaluation of analytical products. J. Agric. Food Chem. 43, 579 ± 586. data. J. Am. Diet. Assoc. 93, 284 ± 296. van Staveren WA, de Groot LCPGM, Blauw YH & van der Wielen RPJ Mares-Perlman JA, Klein BEK, Klein R, Ritter LL, Freudenheim JL & (1994): Assessing diets of elderly people: problems and approaches. Luby MH (1993): Nutrient supplements contribute to the dietary intake Am. J. Clin. Nutr. 59, suppl, 211S ± 223S. of middle- and older-aged adult residents of Beaver Dam, Wisconsin. Vandenlangenberg GM, Brady WE, Nebeling LC, Block G, Forman M, J. Nutr. 123, 176 ± 188. Bowen PE, Stacewitz-Sapuntzakis M & Mares-Perlman JA (1996): Micozzi MS, Brown ED, Edwards BK, Bieri JG, Taylor PR, Khachik F, In¯uence of using different sources of carotenoid data in epidemiologic Beecher GR & Smith JC Jr (1992): Plasma carotenoid response to studies. J. Am. Diet. Assoc. 96, 1271 ± 1275. chronic intake of selected foods and b-carotene supplements in men. Vollebregt YCJ & Feskens EJM (1993): Samenstelling van voedingsmid- Am. J. Clin. Nutr. 55, 1120 ± 1125. delentabellen met gehalten aan retinol en b-caroteen, vitamine E en Ministry of Agriculture, Fisheries & Food (1993): Food Portion Sizes, 2nd pectine ten behoeve van o.a. Zutphen-studie. RIVM, rapport NR. edn. London: HMSO. 441111 002. Morris DL, Krichevsky SB & Davis CE (1994): Serum carotenoids and West CE & Poortvliet EJ (1993): The Carotenoid Content of Foods with coronary heart disease. JAMA 272, 1439 ± 1441. Special Reference to Developing Countries. Virginia: U.S.A.I.D., OckeÂ MC, Bueno-de-Mesquita HB, Pols MA, Smit, HA, van Staveren WA VITAL International Science and Technology Institute, Inc. & Kromhout D (1997): The Dutch EPIC food frequency questionnaire. Yeum K-J, Booth SL, Sadowski JA, Liu C, Tang G, Krinsky NI & Russell II. Relative validity and reproducibility for nutrients. Int. J. Epidemiol. RM (1996): Human plasma carotenoid response to the ingestion of 26, Suppl 1, s49 ± s57. controlled diets high in fruit and vegetables. Am. J. Clin. Nutr. 64, 594 ± Ollilainen V, Heinonen M, Linkola E, Varo P & Koivistoinen P (1988): 602. Carotenoids and retinoids in Finnish foods: meat and meat products. Yong L-C, Forman MR, Beecher GR, Graubard BI, Campbell WS, J. Food Comp. Anal. 1, 178 ± 188. Reichman ME, Taylor, PR, Lanza E, Holden JM & Judd JT (1994): Ollilainen V, Heinonen M, Linkola E, Varo P & Koivistoinen P (1989): Relationship between dietary intake and plasma concentrations of Carotenoids and retinoids in Finnish foods: dairy products and eggs. carotenoids in premenopausal women: application of the USDA-NCI J. Dairy Sci. 72, 2257 ± 2265. carotenoids food composition database. Am. J. Clin. Nutr. 60, 223 ± 230.