Human and Clinical nutrition
Correlations of Vitamin A and E Intakes with the Plasma Concentrations of Carotenoids and Tocopherols among American Men and Women1 ALBERTO ASCHERIO,*2 MEIR J. STAMPFER,*r GRAHAM ERIC B. RIMM,* USA LÃŒTIN*AND WALTER C WILLETT*"
A. COLDITZ,*'
*Department of Epidemiology, Harvard School of Public Health, Boston, MA; tChanning Laboratory, Department of Medicine, Brigham and Women's Hospital and Harvard Medical School, Boston, MA; and ^Department of Nutrition, Harvard School of Public Health, Boston, MA damage produced by singlet oxygen or certain organic free radicals (Burton and Ingold 1984, Krinsky and Deneke 1982, Willett 1990). Other carotenoids, which are present in relatively large amounts in the diet (Bieri and Evarts 1974, Micozzi et al. 1990) and in plasma (Kaplan et al. 1987, Parker 1989) may also be effective antioxidants (Burton 1989, Di Mascio et al. 1989) and possibly influence the risk of cancer. However, the currently available food composition tables used to estimate nutrient intakes in most epidemiological studies do not provide values for spe cific carotenoids and tocopherols. Furthermore, little is known about the determinants of their plasma concentrations. Knowledge of dietary and other factors that influence blood concentrations of specific carotenoids and tocopherols is essential when analyzing and interpreting data using these measure ments. In a few investigations, correlations between di etary variables and plasma concentrations of specific carotenoid fractions (Nierenberg et al. 1989, Roidt et al. 1988, Stryker et al. 1988) were reported. However, only one of these studies included lycopene (Roidt et al. 1988), and none examined lutein or zeaxanthin. Similarly, little is known about determinants of spe cific tocopherols in plasma. We therefore studied the associations of dietary and other factors with the plasma concentrations of ß-carotene, oc-carotene, lycopene, lutein, zeaxanthin, retinol, a-tocopherol and y-tocopherol among a sample of men and women participating in two large prospective cohorts.
ABSTRACT The authors investigated the association of diet and other factors with the plasma concentrations of carotenoids, retino! and tocopherols in a sample of 121 men and 186 women participating in two prospective investigations of dietary etiologies of chronic diseases. Lycopene (mean concentration, 0.82 nmol/L in men, 0.76 umol/L in women), ß-carotene(mean 0.46 umol/L in men, 0.58 umol/L in women) and lutein (mean 0.28 jimol/L in men, 0.27 |xmol/L in women) were the major circulating carotenoids. Among nonsmokers, di etary carotenoid, as typically calculated in epidemiologie studies, was significantly correlated with plasma ßcarotene (r = 0.34 in men, r = 0.30 in women), acarotene (r = 0.52 in men, r = 0.37 in women) and lutein (r = 0.36 in men, r = 0.19 in women), but not with plasma zeaxanthin (r = 0.11 and r = 0.02) or lycopene (r = 0.13 and r = 0.01) after adjusting for plasma choles terol and triglycérides,body mass index and energy intake. Total vitamin E intake was positively associated with plasma concentrations of a-tocopherol (r = 0.51 in men, r = 0.41 in women) and inversely associated with plasma concentrations of y-tocopherol (r = -0.51 in men r = -0.42 in women), but this was primarilydue to use of vitamin E supplements. Measurements of specific carotenoids can provide independent information beyond the usual calculation of carotene intake in epidemiologie studies. J. Nutr. 122: 1792-1801, 1992. INDEXING KEY WORDS:
•carotenoids •tocopherols •vitamin A •vitamin E •humans
Numerous epidemiological studies based on blood measurement and dietary intake (Peto et al. 1981, Wille« 1990, Ziegler 1988) support the hypothesis that high intake of ß-carotene may reduce the inci dence of cancer. A plausible biological mechanism is provided by ß-carotene's ability to block oxidative 0022-3166/92
$3.00 ©1992 American
Institute
of Nutrition.
Supported by Research Grant HL35464 from the National In stitute of Health. Graham Colditz was supported by a Faculty Research Award (FRA 398) from the American Cancer Society. 2To whom correspondence should be addressed.
Received 24 October 1991. Accepted 20 May 1992. 1792
Downloaded from https://academic.oup.com/jn/article-abstract/122/9/1792/4769444 by Washington University in St. Louis user on 25 March 2018
DETERMINANTS
OF PLASMA CAROTENOIDS
METHODS Study population. Men in this study were a subsample drawn from the Health Professionals Followup Study, a cohort of 51,529 male health professionals enrolled in a prospective study of dietary etiologies of heart disease and cancer (Grobbee et al. 1990). All participants completed a 131-item semiquantitative food frequency questionnaire (FFQ) in 1986. Of 323 cohort members living in the Boston area who were invited to participate, 157 agreed to take part in the validation study. Of these, 135 men completed two 1-wk diet records during a 1-y period followed by a second administration of the same dietary question naire (Rimm et al. 1992). The second questionnaire was completed 2 to 6 wk after the end of the diet recording period. One man was excluded for having more than 70 items left blank, and seven men for reporting a total energy intake outside the range of 3347 to 17,573 kj/d in either of the two question naires. Of the remaining 127 subjects, 121 provided a blood sample for the determination of plasma carotenoids and tocopherols and are included in this report. Women in this study were similarly selected from the Nurses' Health Study, a prospective study of 121,700 female registered nurses (Willett et al. 1987). Of 346 Boston-area cohort members invited to participate, 195 agreed to take part in the validation study, and 186 completed a second questionnaire and provided a blood sample. Data from the women's diet records are not included in this report. The research described in this report has been approved by the Committee on the Use of Human Subjects in Research of the Harvard School of Public Health. Dietary assessment. The 131-item questionnaire completed by men and the 126-item questionnaire completed by women are both refined and expanded versions of a previously validated semiquantitative FFQ (Willett et al. 1985) and have been previously described (Rimm et al. 1992). The food composition database used to calculate nutrient values is based primarily on U.S. Department of Agriculture publica tions (Consumer and Food Economics Institute 1976-1989) and is continually supplemented by other published sources and personal communications from laboratories and manufacturers. Nutrient intakes were calculated from the questionnaire by mul tiplying the frequency of use of each food by the nutrient composition of the portion size specified on the questionnaire, and summing across all foods to obtain a total nutrient intake for each individual. Calculations included specific information for brands and types of vitamin supplements, cold breakfast cereals, margarines, and types of fat or oil used for frying, cooking and baking. Diet records were analyzed using CBORD (1988, version 3.O.3., CBORD Group, Inc., Ithaca, NY), a nutrient software based on the ESHA nutrient Downloaded from https://academic.oup.com/jn/article-abstract/122/9/1792/4769444 by Washington University in St. Louis user on 25 March 2018
AND TOCOPHEROLS
1793
database (1988, ESHA Research Database, Salem, OR). For reported foods that were not in the database, recipes were obtained from the participants and ap propriate amounts of the component foods were en tered. Available food composition tables do not permit the calculation of specific carotenoids. The value for dietary carotene that we computed represents the estimated vitamin A activity from dietary carote noids, accounting for most of the ß-carotene plus about half the cc-carotene and a small fraction of lycopene (Stryker et al. 1988). Thus, all of the vitamin A activity from fruits and vegetables and one third of the vitamin A activity for dairy products in published food tables is reported as dietary carotene. Other data. Participants reported their age, smoking habits, height and weight. Body mass index (BMI), used as a measure of obesity, was computed as weight(kg)/height(m)2. Laboratory analyses. Blood specimens were col lected in EDTA-treated tubes between 0800 and 1200 h from nonfasting participants, at the end of the second week of diet recording. The tubes were imme diately covered with aluminum foil and stored in the dark on ice for up to 3 h until the plasma was sepa rated. Plasma was stored at -70°C for periods up to 15 mo until shipment on dry ice for laboratory analyses. Plasma retinol, carotenoids and tocopherols were measured by reverse-phase HPLC in the laboratory of Hoffman La Roche in Basel, Switzerland (Hess et al. 1991). Plasma cholesterol and triglycéride concentra tions were determined using Roche kits based on the methods of Richmond (1973) and Bucolo and David (1973), respectively. Statistical analyses. Pearson correlation coeffi cients were used to assess the relationship between plasma concentrations and other variables. To im prove normality, all variables, except age, were logg transformed before applying correlation and re gression procedures. Energy-adjusted nutrients were calculated as the residuals after regressing each spe cific nutrient on total energy intake (Willett 1990, Willett and Stampfer 1986). An analogous procedure was used to adjust plasma concentrations of carote noids, tocopherols and retinol for age, plasma choles terol, plasma triglycérides or BMI, as indicated in the Results section. Multivariate regression models were fitted using Proc Reg in SAS (SAS Institute, Cary, NC). Reported nutrient intakes were derived from the FFQ administered after completion of the diet records, unless otherwise specified. Results were con sidered significant for two-tailed P values lower than 0.05.
RESULTS Means and SD of plasma concentrations of caro tenes, retinol and tocopherols among men and
1794
ASCHERIO
ET AL.
TABLE 1 Intake and plasma concentration of carotenoids and vitamin E and characteristics of the study population
- 121)SD10.62.90.100.290.380.100.030.388.40.790.950.6625907.144.322.80283.57.215.8W - 186)SD7.24.90.100.370.320.110. VariableAge, yBody kg/mzPlasma mass index, concentrationsct-Carotene, \unol/Lß-Carotene, \\mollLLycopene, \unol/LLutein, \imol/LZeaxanthin, \unol/LRetinol, \unol/La-Tocopherol, \imol/Ly-Tocopherol, \imol/LCholesterol, mmol/LTriglycérides, mmol/LIntakesEnergy, kj/dCarotene,1 \imol/dRetinol, \imol/dRetinol, total, \imol/dVitamin food, \unol/dVitamin E,2 total, \imol/dAlcohol, E, food, g/dCurrent smokers, %MenMean55.724.90.110.460.820.280.072.1327.11.445.301.23838911.323.822.41108.421.412.39.0(a 'Micromoles 2Micromoles
of ß-carotene equivalent. of a-tocopherol equivalent.
women in the study are shown in Table 1. Men who were current cigarette smokers had a mean plasma ßcarotene concentration of 0.30 umol/L, compared with 0.48 umol/L in past smokers or men who had never smoked (P = 0.0009). Women had higher con centrations, but a similar difference was observed between smokers and nonsmokers (0.42 vs. 0.60 Hmol/L, P - 0.003). Because of the small number of current smokers, some analyses are reported only for current nonsmokers. Because smoking status is almost always adjusted for in epidemiological studies, correlations within smoking status strata are more relevant to the interpretation of associations between dietary and other factors with disease incidence. Among men, 54 subjects who had never smoked had a plasma ß-caroteneconcentration of 0.54 |imol/L and a dietary carotene intake of 12.66 nmol/d, both higher than the corresponding values among 50 exsmokers (0.43 (imol/L and 9.78 ^mol/d, P = 0.03 and P - 0.03, respectively). In women, no material dif ference in plasma ß-caroteneconcentrations was seen between those who had never smoked and past smokers. Nonsmoking women had 25% higher concentra tions of plasma ß-carotene than did nonsmoking men (P - 0.01), but no difference was observed after ad justing plasma concentrations of ß-carotene for plasma lipids and dietary carotene. Plasma retinol Downloaded from https://academic.oup.com/jn/article-abstract/122/9/1792/4769444 by Washington University in St. Louis user on 25 March 2018
was 10% higher in men than in women, a difference that remained significant after adjusting for age and plasma lipids. Plasma concentrations of other carote noids and of tocopherols were similar in men and women. Crude correlations of age, BMI and plasma lipids with plasma concentrations of carotenoids, retinol and tocopherols are reported in Table 2. The corre lation matrix between concentrations of plasma carotenoids, retinol and tocopherols is shown in Table 3; all these plasma values were adjusted for age and plasma lipids, and plasma concentrations of ßcarotene were also adjusted for BMI and alcohol intake. Dietary carotene estimated by the FFQ and by the two 1-wk diet records was associated with plasma concentrations of a- and ß-carotene and lutein, but not lycopene or zeaxanthin. The crude correlation coefficient between the FFQ estimate of carotene intake and the plasma concentration of ß-carotene was 0.24 in men and 0.27 in women, 0.29 in both sexes after energy-intake adjustment, and 0.34 and 0.30 after adjustment of the plasma concentrations for plasma lipids, age, BMI and alcohol intake. When current smokers were not excluded from the analysis, the adjusted correlation coefficients between dietary carotene and plasma concentrations of a-carotene were higher for men (0.38) and lower for women
DETERMINANTS
OF PLASMA CAROTENOIDS
AND TOCOPHEROLS
1795
(0.23). In men, plasma concentrations of oc-carotene, but not those of other carotenoids, had a stronger Wr* correlation with diet record estimates than with FFQ «ìs-,i" estimates of carotene intake (Table 4). The FFQ-estimated intake of preformed vitamin A was weakly correlated with plasma concentrations of retinol, whereas intake of vitamin E estimated by questionnaire was correlated with plasma a-to>-'ì"o¡ij2 copherol and inversely correlated with plasma y-tocopherol (Table 5). Excluding from the analysis 21 men and 59 women who reported the use of supple ments containing vitamin E, the correlations between OxI^OTt-irOCA^-t^^^-ieS^HCou^OOOOOOOOOOVCtUICSQOCSO__H_OcSOOeSoooooooo22 ,ufal(ArtE intake of vitamin E and plasma concentrations of tocopherols decreased substantially (r - 0.11 for aEo?o"-1J3.C1'S and r = 0.05 for y-tocopherol in men, i = 0.12 and r -0.15 in women).3 In multivariate analyses, current smoking status ~S£*£"o was significantly associated with lower plasma con centrations of ßand a-carotene in men and women, after adjusting for age, plasma lipids, alcohol intake, energy intake, BMI and carotene intake. A weaker oS
Correlations of Vitamin A and E Intakes with the Plasma Concentrations of Carotenoids and Tocopherols among American Men and Women1 ALBERTO ASCHERIO,*2 MEIR J. STAMPFER,*r GRAHAM ERIC B. RIMM,* USA LÃŒTIN*AND WALTER C WILLETT*"
A. COLDITZ,*'
*Department of Epidemiology, Harvard School of Public Health, Boston, MA; tChanning Laboratory, Department of Medicine, Brigham and Women's Hospital and Harvard Medical School, Boston, MA; and ^Department of Nutrition, Harvard School of Public Health, Boston, MA damage produced by singlet oxygen or certain organic free radicals (Burton and Ingold 1984, Krinsky and Deneke 1982, Willett 1990). Other carotenoids, which are present in relatively large amounts in the diet (Bieri and Evarts 1974, Micozzi et al. 1990) and in plasma (Kaplan et al. 1987, Parker 1989) may also be effective antioxidants (Burton 1989, Di Mascio et al. 1989) and possibly influence the risk of cancer. However, the currently available food composition tables used to estimate nutrient intakes in most epidemiological studies do not provide values for spe cific carotenoids and tocopherols. Furthermore, little is known about the determinants of their plasma concentrations. Knowledge of dietary and other factors that influence blood concentrations of specific carotenoids and tocopherols is essential when analyzing and interpreting data using these measure ments. In a few investigations, correlations between di etary variables and plasma concentrations of specific carotenoid fractions (Nierenberg et al. 1989, Roidt et al. 1988, Stryker et al. 1988) were reported. However, only one of these studies included lycopene (Roidt et al. 1988), and none examined lutein or zeaxanthin. Similarly, little is known about determinants of spe cific tocopherols in plasma. We therefore studied the associations of dietary and other factors with the plasma concentrations of ß-carotene, oc-carotene, lycopene, lutein, zeaxanthin, retinol, a-tocopherol and y-tocopherol among a sample of men and women participating in two large prospective cohorts.
ABSTRACT The authors investigated the association of diet and other factors with the plasma concentrations of carotenoids, retino! and tocopherols in a sample of 121 men and 186 women participating in two prospective investigations of dietary etiologies of chronic diseases. Lycopene (mean concentration, 0.82 nmol/L in men, 0.76 umol/L in women), ß-carotene(mean 0.46 umol/L in men, 0.58 umol/L in women) and lutein (mean 0.28 jimol/L in men, 0.27 |xmol/L in women) were the major circulating carotenoids. Among nonsmokers, di etary carotenoid, as typically calculated in epidemiologie studies, was significantly correlated with plasma ßcarotene (r = 0.34 in men, r = 0.30 in women), acarotene (r = 0.52 in men, r = 0.37 in women) and lutein (r = 0.36 in men, r = 0.19 in women), but not with plasma zeaxanthin (r = 0.11 and r = 0.02) or lycopene (r = 0.13 and r = 0.01) after adjusting for plasma choles terol and triglycérides,body mass index and energy intake. Total vitamin E intake was positively associated with plasma concentrations of a-tocopherol (r = 0.51 in men, r = 0.41 in women) and inversely associated with plasma concentrations of y-tocopherol (r = -0.51 in men r = -0.42 in women), but this was primarilydue to use of vitamin E supplements. Measurements of specific carotenoids can provide independent information beyond the usual calculation of carotene intake in epidemiologie studies. J. Nutr. 122: 1792-1801, 1992. INDEXING KEY WORDS:
•carotenoids •tocopherols •vitamin A •vitamin E •humans
Numerous epidemiological studies based on blood measurement and dietary intake (Peto et al. 1981, Wille« 1990, Ziegler 1988) support the hypothesis that high intake of ß-carotene may reduce the inci dence of cancer. A plausible biological mechanism is provided by ß-carotene's ability to block oxidative 0022-3166/92
$3.00 ©1992 American
Institute
of Nutrition.
Supported by Research Grant HL35464 from the National In stitute of Health. Graham Colditz was supported by a Faculty Research Award (FRA 398) from the American Cancer Society. 2To whom correspondence should be addressed.
Received 24 October 1991. Accepted 20 May 1992. 1792
Downloaded from https://academic.oup.com/jn/article-abstract/122/9/1792/4769444 by Washington University in St. Louis user on 25 March 2018
DETERMINANTS
OF PLASMA CAROTENOIDS
METHODS Study population. Men in this study were a subsample drawn from the Health Professionals Followup Study, a cohort of 51,529 male health professionals enrolled in a prospective study of dietary etiologies of heart disease and cancer (Grobbee et al. 1990). All participants completed a 131-item semiquantitative food frequency questionnaire (FFQ) in 1986. Of 323 cohort members living in the Boston area who were invited to participate, 157 agreed to take part in the validation study. Of these, 135 men completed two 1-wk diet records during a 1-y period followed by a second administration of the same dietary question naire (Rimm et al. 1992). The second questionnaire was completed 2 to 6 wk after the end of the diet recording period. One man was excluded for having more than 70 items left blank, and seven men for reporting a total energy intake outside the range of 3347 to 17,573 kj/d in either of the two question naires. Of the remaining 127 subjects, 121 provided a blood sample for the determination of plasma carotenoids and tocopherols and are included in this report. Women in this study were similarly selected from the Nurses' Health Study, a prospective study of 121,700 female registered nurses (Willett et al. 1987). Of 346 Boston-area cohort members invited to participate, 195 agreed to take part in the validation study, and 186 completed a second questionnaire and provided a blood sample. Data from the women's diet records are not included in this report. The research described in this report has been approved by the Committee on the Use of Human Subjects in Research of the Harvard School of Public Health. Dietary assessment. The 131-item questionnaire completed by men and the 126-item questionnaire completed by women are both refined and expanded versions of a previously validated semiquantitative FFQ (Willett et al. 1985) and have been previously described (Rimm et al. 1992). The food composition database used to calculate nutrient values is based primarily on U.S. Department of Agriculture publica tions (Consumer and Food Economics Institute 1976-1989) and is continually supplemented by other published sources and personal communications from laboratories and manufacturers. Nutrient intakes were calculated from the questionnaire by mul tiplying the frequency of use of each food by the nutrient composition of the portion size specified on the questionnaire, and summing across all foods to obtain a total nutrient intake for each individual. Calculations included specific information for brands and types of vitamin supplements, cold breakfast cereals, margarines, and types of fat or oil used for frying, cooking and baking. Diet records were analyzed using CBORD (1988, version 3.O.3., CBORD Group, Inc., Ithaca, NY), a nutrient software based on the ESHA nutrient Downloaded from https://academic.oup.com/jn/article-abstract/122/9/1792/4769444 by Washington University in St. Louis user on 25 March 2018
AND TOCOPHEROLS
1793
database (1988, ESHA Research Database, Salem, OR). For reported foods that were not in the database, recipes were obtained from the participants and ap propriate amounts of the component foods were en tered. Available food composition tables do not permit the calculation of specific carotenoids. The value for dietary carotene that we computed represents the estimated vitamin A activity from dietary carote noids, accounting for most of the ß-carotene plus about half the cc-carotene and a small fraction of lycopene (Stryker et al. 1988). Thus, all of the vitamin A activity from fruits and vegetables and one third of the vitamin A activity for dairy products in published food tables is reported as dietary carotene. Other data. Participants reported their age, smoking habits, height and weight. Body mass index (BMI), used as a measure of obesity, was computed as weight(kg)/height(m)2. Laboratory analyses. Blood specimens were col lected in EDTA-treated tubes between 0800 and 1200 h from nonfasting participants, at the end of the second week of diet recording. The tubes were imme diately covered with aluminum foil and stored in the dark on ice for up to 3 h until the plasma was sepa rated. Plasma was stored at -70°C for periods up to 15 mo until shipment on dry ice for laboratory analyses. Plasma retinol, carotenoids and tocopherols were measured by reverse-phase HPLC in the laboratory of Hoffman La Roche in Basel, Switzerland (Hess et al. 1991). Plasma cholesterol and triglycéride concentra tions were determined using Roche kits based on the methods of Richmond (1973) and Bucolo and David (1973), respectively. Statistical analyses. Pearson correlation coeffi cients were used to assess the relationship between plasma concentrations and other variables. To im prove normality, all variables, except age, were logg transformed before applying correlation and re gression procedures. Energy-adjusted nutrients were calculated as the residuals after regressing each spe cific nutrient on total energy intake (Willett 1990, Willett and Stampfer 1986). An analogous procedure was used to adjust plasma concentrations of carote noids, tocopherols and retinol for age, plasma choles terol, plasma triglycérides or BMI, as indicated in the Results section. Multivariate regression models were fitted using Proc Reg in SAS (SAS Institute, Cary, NC). Reported nutrient intakes were derived from the FFQ administered after completion of the diet records, unless otherwise specified. Results were con sidered significant for two-tailed P values lower than 0.05.
RESULTS Means and SD of plasma concentrations of caro tenes, retinol and tocopherols among men and
1794
ASCHERIO
ET AL.
TABLE 1 Intake and plasma concentration of carotenoids and vitamin E and characteristics of the study population
- 121)SD10.62.90.100.290.380.100.030.388.40.790.950.6625907.144.322.80283.57.215.8W - 186)SD7.24.90.100.370.320.110. VariableAge, yBody kg/mzPlasma mass index, concentrationsct-Carotene, \unol/Lß-Carotene, \\mollLLycopene, \unol/LLutein, \imol/LZeaxanthin, \unol/LRetinol, \unol/La-Tocopherol, \imol/Ly-Tocopherol, \imol/LCholesterol, mmol/LTriglycérides, mmol/LIntakesEnergy, kj/dCarotene,1 \imol/dRetinol, \imol/dRetinol, total, \imol/dVitamin food, \unol/dVitamin E,2 total, \imol/dAlcohol, E, food, g/dCurrent smokers, %MenMean55.724.90.110.460.820.280.072.1327.11.445.301.23838911.323.822.41108.421.412.39.0(a 'Micromoles 2Micromoles
of ß-carotene equivalent. of a-tocopherol equivalent.
women in the study are shown in Table 1. Men who were current cigarette smokers had a mean plasma ßcarotene concentration of 0.30 umol/L, compared with 0.48 umol/L in past smokers or men who had never smoked (P = 0.0009). Women had higher con centrations, but a similar difference was observed between smokers and nonsmokers (0.42 vs. 0.60 Hmol/L, P - 0.003). Because of the small number of current smokers, some analyses are reported only for current nonsmokers. Because smoking status is almost always adjusted for in epidemiological studies, correlations within smoking status strata are more relevant to the interpretation of associations between dietary and other factors with disease incidence. Among men, 54 subjects who had never smoked had a plasma ß-caroteneconcentration of 0.54 |imol/L and a dietary carotene intake of 12.66 nmol/d, both higher than the corresponding values among 50 exsmokers (0.43 (imol/L and 9.78 ^mol/d, P = 0.03 and P - 0.03, respectively). In women, no material dif ference in plasma ß-caroteneconcentrations was seen between those who had never smoked and past smokers. Nonsmoking women had 25% higher concentra tions of plasma ß-carotene than did nonsmoking men (P - 0.01), but no difference was observed after ad justing plasma concentrations of ß-carotene for plasma lipids and dietary carotene. Plasma retinol Downloaded from https://academic.oup.com/jn/article-abstract/122/9/1792/4769444 by Washington University in St. Louis user on 25 March 2018
was 10% higher in men than in women, a difference that remained significant after adjusting for age and plasma lipids. Plasma concentrations of other carote noids and of tocopherols were similar in men and women. Crude correlations of age, BMI and plasma lipids with plasma concentrations of carotenoids, retinol and tocopherols are reported in Table 2. The corre lation matrix between concentrations of plasma carotenoids, retinol and tocopherols is shown in Table 3; all these plasma values were adjusted for age and plasma lipids, and plasma concentrations of ßcarotene were also adjusted for BMI and alcohol intake. Dietary carotene estimated by the FFQ and by the two 1-wk diet records was associated with plasma concentrations of a- and ß-carotene and lutein, but not lycopene or zeaxanthin. The crude correlation coefficient between the FFQ estimate of carotene intake and the plasma concentration of ß-carotene was 0.24 in men and 0.27 in women, 0.29 in both sexes after energy-intake adjustment, and 0.34 and 0.30 after adjustment of the plasma concentrations for plasma lipids, age, BMI and alcohol intake. When current smokers were not excluded from the analysis, the adjusted correlation coefficients between dietary carotene and plasma concentrations of a-carotene were higher for men (0.38) and lower for women
DETERMINANTS
OF PLASMA CAROTENOIDS
AND TOCOPHEROLS
1795
(0.23). In men, plasma concentrations of oc-carotene, but not those of other carotenoids, had a stronger Wr* correlation with diet record estimates than with FFQ «ìs-,i" estimates of carotene intake (Table 4). The FFQ-estimated intake of preformed vitamin A was weakly correlated with plasma concentrations of retinol, whereas intake of vitamin E estimated by questionnaire was correlated with plasma a-to>-'ì"o¡ij2 copherol and inversely correlated with plasma y-tocopherol (Table 5). Excluding from the analysis 21 men and 59 women who reported the use of supple ments containing vitamin E, the correlations between OxI^OTt-irOCA^-t^^^-ieS^HCou^OOOOOOOOOOVCtUICSQOCSO__H_OcSOOeSoooooooo22 ,ufal(ArtE intake of vitamin E and plasma concentrations of tocopherols decreased substantially (r - 0.11 for aEo?o"-1J3.C1'S and r = 0.05 for y-tocopherol in men, i = 0.12 and r -0.15 in women).3 In multivariate analyses, current smoking status ~S£*£"o was significantly associated with lower plasma con centrations of ßand a-carotene in men and women, after adjusting for age, plasma lipids, alcohol intake, energy intake, BMI and carotene intake. A weaker oS
