|Home | About | Journals | Submit | Contact Us | Français|
Cardiovascular disease (CVD) risk factors may potentially influence plasma levels of carotenoids. However, data on the association of plasma carotenoids with CVD related biomarkers are only limited.
We examined the cross-sectional association of plasma carotenoids with blood lipids, hemoglobin A1c (HbA1c), and C-reactive protein (CRP) in middle-aged and older women initially free of CVD and cancer.
Participants from three nested case-control studies in Women's Health Study were pooled. Baseline plasma carotenoids, including α-carotene, β-carotene, β-cryptoxanthin, lycopene, and lutein/zeaxanthin, blood lipids, HbA1c, and CRP were available for 2895 women.
Women who were current smokers or obese had lower plasma levels of most carotenoids expect for lycopene. After adjusting for age, race, lifestyle factors, clinical factors, plasma total cholesterol, and dietary carotenoids, an increase of 30 mg/dl in LDL-cholesterol was associated with 17% increase in α-carotene, 16% increase in β-carotene, and 8.5% increase in lycopene; an increase of 10 mg/dl in HDL-cholesterol was associated with 5.3% decrease in lycopene; an increase of 0.3% in HbA1c was associated with 1.4% increase in lycopene; and an increase of 2 mg/L in CRP was associated with 1.3% decrease in β-carotene (all p<0.01).
In middle-aged and older women free of CVD and cancer, plasma carotenoids were associated with smoking, obesity, LDL-cholesterol, HDL-cholesterol, HbA1c, and CRP. The associations differ among individual carotenoids, possibly reflect metabolic effects of lifestyle and physiologic factors on plasma carotenoids, and may partially explain the inverse association of plasma carotenoids with CVD outcomes observed in epidemiologic studies.
Fruits and vegetables are rich in carotenoids, a group of plant-derived, fat-soluble pigments. With abundant conjugated double bonds, carotenoids have shown potent antioxidant properties.(1, 2) Plasma or adipose carotenoids have been potentially associated with a reduced risk of cardiovascular disease (CVD) in population-based observational studies.(3-5) However, large-scale clinical trials on β-carotene supplementation either showed no benefit or increased the risk of CVD outcomes.(6-9) Inherent confounding in epidemiologic studies may partially explain the discrepancy. In addition to dietary intake, other lifestyle and physiologic factors that are involved in CVD development could have influenced the concentrations of carotenoids in plasma and body tissues.
Smoking and obesity, both established CVD risk factors, have been associated with lower serum β-carotene levels in a number of studies.(10-17) Relatively fewer studies have examined the association of plasma carotenoids with biomarkers related to CVD risk, such as blood lipids subtype, glycosylated hemoglobin A1c (HbA1c), and C-reactive protein (CRP). Most previous investigations on plasma carotenoids and CVD related biomarkers have been either relatively small, or focused on only selected carotenoids, predominantly β-carotene. We therefore conducted a cross-sectional study to examine the association of five major plasma carotenoids, including α-carotene, β-carotene, β-cryptoxanthin, lycopene, and lutein/zeaxanthin, with risk factors and biomarkers related to CVD in a subsample of middle-aged and older women free of CVD and cancer.
The Women's Health Study (WHS) is a recently completed, randomized, double-blind, placebo-controlled clinical trial of low-dose aspirin and vitamin E in the primary prevention of CVD and cancer.(18-20) The β-carotene component was terminated in 1996 after a median treatment duration of 2.1 years.(9) In 1992, 39,876 female US health professionals, aged ≥45 years and free from self-reported CVD and cancer (except non-melanoma skin cancer), were randomized into the WHS. Baseline blood samples were collected from 28,345 (71%) participants in chilled package via overnight courier, centrifuged immediately, aliquotted and stored in liquid nitrogen freezers at -140°C for 10 years until analyzed. The study protocol was approved by the Brigham and Women's Hospital institutional review board. Written informed consent was obtained from all participants.
This cross-sectional analysis was conducted using the pre-existing baseline measurements of both cases and controls in three independent nested case-control studies within the WHS. The case-control study of diabetes (n=940) included 470 cases of incident type 2 diabetes and equal number of controls that were individually matched with cases on age (±1 y) and follow-up time (± 6 months). The case-control study of breast cancer (n=1016) included 508 cases of incident breast cancer and equal number of controls, individually matched on age, smoking status (never, former, and current smoker), and follow-up time. The case-control study of CVD (n=966) included 483 cases of incident CVD and equal number of controls, individually matched on age, smoking status, and follow-up time. For the current analysis, participants who reported a pre-randomization history of CVD or cancer after randomization (n=5) and duplicate participants who were included in more than one case-control studies (n=22) were excluded. A total of 2895 women free of CVD and cancer remained for analysis, among whom the prevalence of diabetes, hypercholesterolemia, and hypertension was 3.2%, 35.1%, and 34.6% respectively.
In each case-control study, blood samples were thawed and assayed for carotenoids at Our Lady of Mercy Medical Center, Bronx, NY. Carotenoids, including α-carotene, β-carotene, β-cryptoxanthin, lycopene, and lutein/zeaxanthin, were quantified by reversed-phase high-performance liquid chromatography (HPLC) after extraction and concentration using standard methodology.(21) All investigators and laboratory personnel were blinded to the subjects' case-control status. All blood samples were handled identically throughout the processes of blood collection, long-term storage, sample retrieval and assays. Internal standards (echinenone) were used to correct for recoveries of all samples analyzed. The laboratory prepared and assayed internal and external blinded quality control specimens in every run. From these control specimens, the accuracy for each measured carotenoid was within 7%, and the day-to-day and within-day precision (coefficient of variance) for these analytes was 5%. The laboratory has participated in the US Quality Assurance Program to ensure methodology consistent with other laboratories.
Plasma lipids, HbA1c and CRP were measured previously as part of a larger study in the WHS. Plasma lipids and CRP were assayed at a core laboratory facility and were available for 2878 of the 2895 (99.4%) participants. Total cholesterol and high-density lipoprotein (HDL) cholesterol were measured enzymatically using a Hitachi 911 autoanalyzer (Roche Diagnostics, Basel, Switzerland), and low-density lipoprotein (LDL) cholesterol was directly measured (Genzyme, Cambridge, Mass). High-sensitivity CRP was analyzed using a validated immunoturbidimetric method (Denka Seiken, Tokyo, Japan), and the levels were similar to expected values in a population of healthy middle-aged women.(22) The coefficients of variation varied from 2.2% to 6.1% for CRP concentrations ranging from 0.15 to 1.90 mg/L. HbA1c was measured by the Tina-Quant turbidimetric inhibition immunoassay (Hitachi 911 Analyzer; Roche Diagnostics, Indianapolis, IN) using packed red blood cells, and was available for 2869 of the 2895 (99.1%) participants. The coefficient of variation for quality control samples of HbA1c was 7.2%.
Women provided baseline self-reports of age (in years), weight and height [represented as body mass index (BMI) in kg/m2], smoking status (never, former, current <15 cigarettes/day, current ≥15 cigarettes/day), alcohol use (rarely/never, 1-3 drinks/month, 1-6 drinks/week, ≥1 drink/day), vigorous exercise (rarely/never, <1, 1-3, ≥4 times/week), menopausal status (yes, no, uncertain). Information on multivitamin supplement use (never, former, current), post-menopausal hormone use (never, former, current), diagnosis of diabetes, hypertension, and hypercholesterolemia (no, yes) were also collected in the baseline questionnaire. Hypertension was defined as having a physician diagnosis, or self-reported systolic blood pressure ≥140 mmHg or diastolic blood pressure ≥90 mmHg, or antihypertensive treatment. Hypercholesterolemia was defined as having a physician diagnosis, or self-reported cholesterol level ≥240 mg/dL (6.22 mmol/L), or anti-hypercholesterolemia treatment.
Women also completed a 131-item validated semi-quantitative food frequency questionnaire (SFFQ). The average daily intakes of individual food items were calculated by multiplying the intake frequency by the specified portion size of each item. Nutrient intakes were computed by multiplying the intake frequency of each unit of food by the nutrient content of the specified portion size according to food composition tables from the Harvard Food Composition Database.(23) All individual nutrients were adjusted for total energy intake using the residual method.(24) The SFFQ used in the WHS has demonstrated reasonable validity as a measure of long-term average dietary intakes in populations of health professionals.(25)
Statistical analyses were performed using SAS software (SAS Institute, Cary, NC, USA) version 9.1. Spearman correlations of plasma carotenoids with dietary carotenoids estimated from SFFQ were calculated. Because the distributions of plasma carotenoids were highly skewed, we applied logarithmic transformation to these variables. Plasma carotenoids were then compared across categories of demographic factors (age, race), lifestyle factors (smoking, alcohol use, vigorous exercise, menopausal status, postmenopausal hormone use, and multivitamin use) and clinical factors (BMI, history of diabetes, hypertension and hypercholesterolemia) using analysis of variance. The association of plasma carotenoids with CVD related biomarkers were evaluated by comparing plasma carotenoids across categories of each biomarker using analysis of covariance, first adjusting for age, race, and total cholesterol, then further adjusting for other lifestyle factors, clinical factors, and dietary intake of respective carotenoids. Following the same adjustment strategy, linear regression models were used to determine the difference in plasma carotenoids associated with unit change in biomarkers. Analyses on biomarkers were further stratified by lifestyle factors including smoking status (current vs. not current), BMI (<25 kg/m2 vs. ≥25 kg/m2), and dietary fat intake (above vs. below median). Interactions were tested using Wald Chi-square test. We used a lower value of P<0.01 as the threshold to indicate significance due to the inherent multiple number of comparisons in our study. In sensitivity analyses, we excluded subjects who had prevalent diabetes at baseline as well as those who developed incident diabetes or incident CVD during the cohort follow-up, the study results were similar and therefore not presented.
Plasma concentration and dietary intake were significantly correlated with one another for each carotenoid except for lycopene, though the magnitude of correlations was only moderate (Spearman r ranging from 0.034 for lycopene to 0.13 for β-cryptoxanthin). For all plasma carotenoids, the geometric means were similar across categories of age, race, menopausal status, postmenopausal hormone use, and multivitamin use (Table 1). Women who were current smokers and obese had lower plasma concentrations of all carotenoids except for lycopene. Women who vigorously exercised ≥1 time/week tended to have higher plasma β-carotene, β-cryptoxanthin, and lutein/zeaxanthin, and those who consumed alcohol 1-6 drinks/week had higher plasma lutein/zeaxanthin only. Women with hypercholesterolemia had higher plasma lycopene, and those with hypertension had lower plasma β-carotene.
For all plasma carotenoids and in both simple adjusted and multivariate adjusted model, the geometric means were similar across levels of total cholesterol (Table 2). After adjusting for age, race and total cholesterol, the geometric means of plasma α-carotene, β-carotene, and lycopene tended to increase with increasing levels of LDL cholesterol; the geometric means of all plasma carotenoids except for lycopene tended to increase with increasing levels of HDL cholesterol and decrease with increasing levels of HbA1c and CRP. In contrast to other plasma carotenoids, plasma lycopene was inversely associated with HDL cholesterol and positively associated with HbA1c and CRP in age, race, and total cholesterol adjusted model. Some, but not all, of these associations remained significant after further adjusting for lifestyle factors, clinical factors, and dietary intake of respective carotenoids (Table 2). In the multivariate adjusted model, women who had higher LDL cholesterol had significantly higher plasma concentration of β-carotene; those who had lower HDL cholesterol or higher HbA1c had higher plasma concentration of lycopene; and those who had higher CRP had lower concentration of α-carotene and β-carotene. β-cryptoxanthin and lutein/zeaxanthin were not significantly associated with any CVD related biomarkers in multivariate models.
Linear regression performed on log-transformed plasma carotenoids obtained generally consistent results (Table 3). After adjusting for age, race, total cholesterol, lifestyle factors, clinical factors, and dietary intake of respective carotenoids, an increase of 30 mg/dl in LDL cholesterol was associated with 17% (calculated as e0.16=1.17) increase in plasma α-carotene, 16% increase in plasma β-carotene, and 8.5% increase in plasma lycopene; an increase of 10 mg/dl in HDL cholesterol was associated with 5.3% decrease in plasma lycopene; an increase of 0.3% in HbA1c was associated with 1.4% increase in plasma lycopene; and an increase of 2 mg/L in CRP was associated with 1.3% decrease in plasma β-carotene. There was a marginally significant inverse association between CRP and plasma α-carotene (p=0.06). When these linear regression analyses were stratified by smoking status, BMI, and dietary fat intake, the regression coefficients were somewhat different across strata, but the interactions were not statistically significant (all P for interaction > 0.01).
In this cross-sectional sample of 2895 middle-aged and older women free of CVD and cancer, we confirmed findings from other studies that smokers and obese individuals had lower plasma concentrations of carotenoids. We also found that plasma α-carotene, β-carotene, and lycopene were positively associated with LDL cholesterol, plasma lycopene was positively associated with HbA1c and inversely associated with HDL cholesterol, and plasma β-carotene was inversely associated with CRP. These associations were independent of various lifestyle factors, clinical factors, and dietary intake of respective carotenoids.
Lower plasma concentrations of carotenoids among smokers (10, 12-16) and obese individuals (16, 17) have been previously reported in a number of studies. These associations may reflect both different intakes across subgroups of study subjects and metabolic effect of lifestyle and anthropometry factors on plasma carotenoids. Smoking is known to increase the production of oxygen-derived free radicals. Carotenoids, as potent antioxidants, retard the proliferation of free radicals and protect against free radical-mediated tissue damage.(26, 27) The interaction with free radicals results in the fragmentation and loss of carotenoid molecules.(28) Excessive oxidative stress and depletion of antioxidants are also present among obese individuals. An alternative explanation of the inverse association between plasma carotenoids and BMI is that compared to normal weight persons, overweight or obese individuals with greater body fat storage may have lower circulating carotenoids in plasma due to a high proportion of carotenoids, as lipid-soluble compounds, being stored in adipose tissue.(10)
The positive association between plasma carotenoids and serum cholesterol levels observed in previous studies (10, 12-17) reflect the fact that the lipophilic carotenoids are absorbed with dietary fats(29) and transported in lipoproteins.(30) In our study, though plasma carotenoids were unassociated with total cholesterol, α-carotene, β-carotene and lycopene were strongly and positively associated with LDL cholesterol, and all carotenoids except for lycopene were positively associated with HDL cholesterol after adjusting for total cholesterol. The positive association of plasma carotenoids with LDL cholesterol remained significant after multivariate adjustment. Although hypercholesterolemia is complicated with greater free radical production,(31) it appeared that the potentially increased utilization of carotenoids in hypercholesterolemia is less biologically important in determining plasma carotenoids compared to the role of cholesterol as a non-specific carrier.
An inverse association between plasma carotenoids and blood glucose has been reported in previous cross-sectional studies,(32-34) suggesting a relation between impaired glucose metabolism, increased free radical activities, and reduced antioxidant concentrations. Fewer data are available regarding the association with elevated HbA1c, an indicator of chronic hyperglycemia. Our study found that higher levels of HbA1c were associated with lower plasma concentrations of α-carotene, β-carotene and β-cryptoxanthin, which is consistent with an earlier report,(35) but the inverse association in our study was attenuated and no longer significant after multivariate adjustment. Our finding of a positive association of plasma lycopene with HbA1c contrasted with some earlier studies.(35, 36)
An inverse association between plasma carotenoids and inflammatory markers was reported previously among elderly nuns(37) and lung cancer patients.(38) In adults with acute inflammatory conditions such as tuberculosis(39) and pancreatitis,(40) there were transient decrease in serum carotenoids and increase in CRP level, which normalize with resolution of the illness. In the Third National Health and Nutrition Examination Survey (NHANES III), serum β-carotene was strongly and inversely associated with CRP levels and white blood cell count after adjusting for carotene intake and other possible confounders.(41) Our study results agree with this earlier study in a representative population sample by showing that high CRP was associated with low plasma α- and β-carotene after multivariate adjustment. In many inflammatory disorders, the inflammatory responses induce the release of chemical mediators and activate peripheral blood mononuclear cells, which in turn produce excessive reactive oxygen species or oxygen free radicals,(42-45) and then increase the utilization of carotenoids.
In the current study, the observed associations with CVD risk factor and related biomarkers were not uniform for all carotenoids. The carotenoids differ in tissue localization and in antioxidant properties.(46) Lutein/zeaxanthin and β-cryptoxanthin were less-reactive antioxidant compared to other carotenoids in vitro and were not associated with any CVD related biomarkers in our study. Lycopene has the most powerful antioxidant properties among major carotenoids detected in human tissues.(47) However, our study, in consistency with others,(10, 34) noted that many associations observed for plasma lycopene were in directions opposite to other carotenoids, for reasons yet to be fully understood. Dietary lycopene is derived predominantly from consumption of tomatoes and tomato products.(48) It is possible that subjects who consume large amount of lycopene-rich foods have some distinctive but un-recognized characteristics. Another possible explanation could be the different features of cis and trans lycopene, which were not determined in the present study.
Several potential limitations of our study deserve consideration. First, we cannot establish any cause-effect association from this cross-sectional study. Second, although we do not anticipate major effects on the blood measurement due to processing and storage, single measurement of plasma carotenoids and CVD related biomarkers could introduce non-differential misclassification, which will bias the association towards the null. This may also explain the relatively low correlation between dietary carotenoids and plasma carotenoids in the current study compared to other studies.(10, 32, 49) In addition to variability in blood measurements, the diversity of carotenoid food sources also serves to reduce individual magnitudes of dietary-plasma carotenoids correlations. Of note, the Harvard Food Consumption database used USDA Agriculture Handbook no. 8 published in 1993 and had not included the most updated carotenoids values of foods. Third, although multiple factors were comprehensively adjusted for in our analyses, residual confounding from unknown or poorly measured determinants of plasma carotenoids cannot be completely ruled out. Finally, our study results apply to middle-aged and older, predominantly White women who were initially free of CVD and cancer. Studies in other populations are needed to confirm our findings.
In conclusion, we found in a cross-sectional sample of middle-aged and older women, that plasma carotenoids were associated with smoking, obesity, LDL cholesterol, HDL-cholesterol, HbA1c, and CRP levels. These associations were different among individual carotenoids, and may partially explain the observed inverse association of plasma carotenoids with CVD outcomes in previous population studies. These findings emphasize the importance of adequately controlling for confounders of plasma carotenoids in epidemiological studies of CVD.
We are indebted to the 39,876 participants in the Women's Health Study for their dedicated and conscientious collaboration, and to the entire staff of the Women's Health Study for their continuous efforts.
Supported by research grants CA-047988 and HL-043851 from the National Institutes of Health, Bethesda, MD.
Authorship LW was responsible for data analysis and manuscript preparation. JMG and JEB were responsible for study funding, data collection, and provided editorial review and comments about the manuscript. EPN were responsible for data collection, and contributed critical editorial input, significant advice and consultation on data analyses and result interpretation. HDS was responsible for the study design, data collection, and manuscript preparation.
Conflict of interest: None.