Plasma Lycopene, Other Carotenoids, and the Risk of Type 2 Diabetes in Women

Abstract

The authors conducted a nested case-control study from 1992 to 2003 among US women aged 45 years or older and free from cardiovascular disease and cancer to examine the prospective association among plasma lycopene, other carotenoids, and the risk of developing type 2 diabetes. During 10 years of follow-up, 470 cases of incident type 2 diabetes were selected and individually matched on age (±1 year) and follow-up time to 470 nondiabetic controls. Baseline plasma levels of lycopene, α-carotene, β-carotene, β-cryptoxanthin, and lutein/zeaxanthin were similar in cases and controls (all p > 0.05). A possible crude inverse association between plasma lycopene and risk of type 2 diabetes was attenuated upon multivariate adjustment. After control for plasma total cholesterol and known diabetes risk factors, the multivariate odds ratios of type 2 diabetes in the highest versus the lowest quartile of plasma carotenoids were 1.13 (95% confidence interval (CI): 0.60, 2.13) for lycopene, 1.27 (95% CI: 0.63, 2.57) for α-carotene, 1.10 (95% CI: 0.57, 2.13) for β-carotene, 0.91 (95% CI: 0.46, 1.81) for β-cryptoxanthin, and 1.35 (95% CI: 0.68, 2.69) for lutein/zeaxanthin. There was no prospective association between baseline plasma carotenoids and the risk of type 2 diabetes in middle-aged and older women.

Accumulating evidence indicates that oxidative stress, a condition of excessive reactive oxygen species, may play a role in the etiology of type 2 diabetes by inducing insulin resistance in the peripheral tissues and impairing insulin secretion from pancreatic β-cells (1–3). Carotenoids, a group of fat-soluble pigments present in many foods, particularly fruits and vegetables, have abundant conjugated double bonds to interrupt the chain reaction of lipid oxidation and to quench peroxyl radicals (4). This potent antioxidant capacity of carotenoids may provide protection against the development of type 2 diabetes. Among the major carotenoids detected in human tissues, lycopene has shown the most powerful antioxidant properties (5).

Previous cross-sectional studies (6–16) showed an inverse association between dietary or plasma carotenoids and type 2 diabetes or related metabolic indices, such as fasting blood glucose, glucose tolerance, and glycosylated hemoglobin. Prospective studies of carotenoids and the risk of type 2 diabetes have been limited. One nested case-control study (17) found that the baseline concentrations of serum β-carotene were lower in incident cases of type 2 diabetes than in controls, but the inverse association disappeared after adjustment for other risk factors. A prospective cohort study (18) reported a significantly reduced type 2 diabetes risk in association with higher dietary intake of β-cryptoxanthin but not with intakes of other major carotenoids. The randomized trial that directly evaluated the effect of long-term β-carotene supplementation in the primary prevention of type 2 diabetes did not find significant benefits of β-carotene on reducing the risk of type 2 diabetes (19). To further address the question, we examine the prospective association of baseline plasma lycopene and other carotenoids with the risk of type 2 diabetes in middle-aged and older women from the Women's Health Study (WHS).

MATERIALS AND METHODS

Study population

The WHS is a recently completed, randomized, double-blind, placebo-controlled clinical trial of low-dose aspirin and vitamin E in the primary prevention of cardiovascular disease and cancer in women (20–22). A β-carotene component was terminated in 1996 after a median duration of 2.1 years' treatment (23). In 1992, a total of 39,876 female US health professionals, aged 45 years or older and free from self-reported cardiovascular disease and cancer (except nonmelanoma skin cancer), were randomized into the WHS. Fasting baseline blood samples in chilled packages were collected from 28,345 (71 percent) participants via overnight courier, centrifuged immediately, aliquotted, and then stored in liquid nitrogen freezers at −140°C for 10 years until analyzed. Exposure to light and temperature variations were therefore minimized for these blood samples.

A prospective, nested, case-control design was used to identify 470 case-control pairs of WHS participants with baseline blood samples. Cases were women who were free from baseline diabetes and reported the development of type 2 diabetes in questionnaires during follow-up of the cohort over 10 years (maximum: 10.9 years). The validity of self-reported type 2 diabetes in the WHS has been confirmed using the American Diabetes Association diagnostic criteria (24). As described previously (25), the self-reported diagnosis of diabetes was verified in 406 (91 percent) of 446 women who provided baseline blood samples and responded to a telephone interview attempt. In parallel to the telephone interview, self-reported diabetes was also verified in 124 (91 percent) of 136 women who responded to a mailed supplemental diabetes questionnaire. In addition, 89 (99 percent) of 90 women whose primary care physician provided adequate information to apply American Diabetes Association criteria were confirmed to have type 2 diabetes on the basis of the combined information from the supplemental questionnaire and physician information. For each type 2 diabetes case, one control subject was chosen from women who had provided a baseline blood sample and remained free from diabetes throughout the follow-up. Each control was matched with a case on age (±1 year) and follow-up time. Because more than 85–90 percent of all WHS participants reported a recent fasting blood glucose screening on their annual follow-up questionnaires, the probability that diabetes was present but not detected would be low in this cohort of health professionals. The study protocol was approved by the Brigham and Women's Hospital institutional review board. Written, informed consent was obtained from all participants.

Blood assays

All investigators and laboratory personnel were blinded to the subjects' case-control status. All baseline blood samples were handled identically throughout the processes of blood collection, long-term storage, sample retrieval, and assays. Blood samples were thawed and assayed for carotenoids at Our Lady of Mercy Medical Center, Bronx, New York. Carotenoids, including α-carotene, β-carotene, β-cryptoxanthin, lycopene, and lutein/zeaxanthin, were quantified by reversed-phase high-performance liquid chromatography after extraction and concentration with standard methods. Internal standards (echinenone) were used to correct for recoveries of all samples analyzed. To control for the differences in concentrations of lipoprotein as nonspecific carrier for all carotenoids in plasma, we assayed total cholesterol by enzymatic, end-point spectroscopy using commercially available diagnostic kits (Sigma-Aldrich Chemical Co., St Louis, Missouri) and conventional methods. 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 percent, and the day-to-day and within-day precision (coefficient of variance) for these analytes was 5 percent. A subgroup of plasma samples (86 cases and 90 controls) also has been assayed earlier for specific insulin, by use of double-antibody systems (Linco Research, St Louis, Missouri) with less than 0.2 percent cross-reactivity between insulin and its precursors. Hemoglobin A1c was measured for all WHS participants by the Tina-Quant turbidimetric inhibition immunoassay (Hitachi model 911 analyzer; Roche Diagnostics, Indianapolis, Indiana) using packed red blood cells. The laboratory has participated in the US Quality Assurance Program to ensure methodology consistent with that of other laboratories.

Other baseline covariates

On the baseline questionnaire, women provided self-reports of age (years); weight and height (represented as body mass index in kg/m²); smoking status (never, former, current); 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); family history of diabetes in a first-degree relative (no, yes); menopausal status (yes, no, uncertain); postmenopausal hormone use (never, former, current); and multivitamin use (never, former, current). Hypertension (no, yes) was defined as having a physician diagnosis, a self-reported systolic blood pressure of 140 mmHg or higher or diastolic blood pressure of 90 mmHg or higher, or current treatment for high blood pressure. Hypercholesterolemia (no, yes) was defined as having a physician diagnosis, a self-reported cholesterol level of 240 mg/dl (6.22 mmol/liter) or higher, or current treatment for high cholesterol. Women also provided detailed dietary information at baseline by completing a 131-item validated semiquantitative food frequency questionnaire. 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 (26). All individual nutrients were adjusted for total energy intake by use of the residual method (27). The semiquantitative food frequency questionnaire used in the WHS has demonstrated reasonable validity as a measure of long-term average dietary intakes in populations of health professionals (28–30).

Statistical analysis

Statistical analysis was performed with SAS, version 8, software (SAS Institute, Inc., Cary, North Carolina). Cases were first compared with controls on the mean values or proportions of major risk factors and plasma biomarkers, by use of paired t tests for means and McNemar's tests for proportions. The plasma concentrations of carotenoids were then divided into quartiles based on the distribution among 470 controls. Diabetes risk factors were compared according to quartiles of plasma carotenoids among controls to assess potential confounding, by use of analysis of variance for means and chi-square tests for proportions. Spearman's correlation coefficients of plasma carotenoids with hemoglobin A1c and insulin were also estimated among controls.

Conditional logistic regression was performed to calculate odds ratios and 95 percent confidence intervals for incident diabetes according to quartiles of plasma carotenoids' concentrations, with the lowest quartile serving as the referent. Linear trends across increasing quartiles were tested by use of the median value of plasma carotenoids in each quartile as an ordinal variable. Crude models were matched for age and adjusted for plasma total cholesterol and randomized treatment assignment (aspirin, vitamin E, β-carotene, or placebo). Multivariate models further adjusted for lifestyle factors including smoking, alcohol use, vigorous exercise, menopausal status, postmenopausal hormone use, multivitamin use, and family history of diabetes; clinical factors including body mass index, history of hypertension, and hypercholesterolemia; and dietary factors including total energy intake, intake of energy-adjusted saturated fat, fiber, and glycemic load (all continuous). Because abnormalities in insulin sensitivity and glucose metabolism may precede overt diabetes by many years, analyses were repeated for nonobese women (body mass index of <30 kg/m²) and women who had favorable hemoglobin A1c levels (<6.5 percent) at baseline using unconditional logistic regression. Only 86 of 470 controls were obese at baseline, and no controls had a baseline hemoglobin A1c level of 6.5 percent or greater; therefore, analyses were not repeated for those women with unfavorable baseline characteristics.

RESULTS

Compared with women who remained free from diabetes, cases developing type 2 diabetes had considerably higher baseline body mass indexes and were less likely to exercise, drink alcohol, or currently use postmenopausal hormones, while they were more likely to have a family history of diabetes and personal history of hypertension or hypercholesterolemia at baseline (table 1). Women with incident type 2 diabetes also had higher total energy intake, energy-adjusted saturated fat intake, and lower fiber intake. Dietary glycemic load and intakes of fruits and vegetables did not differ between cases and controls. Among blood assays, baseline hemoglobin A1c and insulin were significantly higher in women who developed type 2 diabetes during follow-up than in those who did not. The baseline plasma concentrations of carotenoids, including lycopene, α-carotene, β-carotene, β-cryptoxanthin, and lutein/zeaxanthin, were all similar in cases and controls (all p > 0.05).

The distributions of major diabetes risk factors, including age and body mass index, were generally comparable according to quartiles of plasma carotenoids among controls. As shown in table 2, although a few lifestyle factors such as family history of diabetes and history of hypertension differed modestly between controls in the highest quartile of plasma lycopene and those in the lowest quartile, these differences were not statistically significant. The associations for other plasma carotenoids showed a pattern similar to that for plasma lycopene (data not shown). None of the plasma carotenoids was significantly correlated with baseline levels of hemoglobin A1c (Spearman's r ranging from −0.08 to −0.02) or insulin (Spearman's r ranging from −0.17 to 0.05) in controls.

In crude models that matched for age and adjusted for plasma total cholesterol and randomized treatment, the odds ratios and 95 percent confidence intervals of type 2 diabetes across increasing quartiles of plasma lycopene were 1.00 (referent), 0.77 (95 percent confidence interval (CI): 0.52, 1.13), 0.64 (95 percent CI: 0.44, 0.93), and 0.84 (95 percent CI: 0.58, 1.20) (linear p_trend = 0.30) (table 3). The reduced risk of type 2 diabetes limited to the third quartile of plasma lycopene was no longer significant after adjustment for the other diabetes risk factors in the multivariate model. The multivariate-adjusted models also did not reveal any significant associations between other plasma carotenoids and the risk of type 2 diabetes. After adjustment for lifestyle and clinical and dietary risk factors, the odds ratios of type 2 diabetes across the increasing quartiles of plasma carotenoids were 1.00, 0.89, 1.11, and 1.27 (p_trend = 0.38) for α-carotene; 1.00, 1.33, 1.07, and 1.10 (p_trend = 0.97) for β-carotene; 1.00, 1.44, 1.34, and 0.91 (p_trend = 0.54) for β-cryptoxanthin; and 1.00, 1.13, 1.37, and 1.35 (p_trend = 0.32) for lutein/zeaxanthin.

Subgroup analyses obtained results similar to the overall association for all cases and controls (table 4). Among women with a body mass index of less than 30 kg/m² at baseline (189 cases and 384 controls), the multivariate-adjusted odds ratios and 95 percent confidence intervals of type 2 diabetes in the highest quartile versus the lowest quartile of plasma carotenoids were 0.87 (95 percent CI: 0.49, 1.53) for lycopene, 0.98 (95 percent CI: 0.55, 1.75) for α-carotene, 0.86 (95 percent CI: 0.48, 1.55) for β-carotene, 0.83 (95 percent CI: 0.46, 1.50) for β-cryptoxanthin, and 0.89 (95 percent CI: 0.48, 1.62) for lutein/zeaxanthin. The corresponding odds ratios among women with a hemoglobin A1c level of less than 6.5 percent at baseline (420 cases and 469 controls) were 1.15 (95 percent CI: 0.71, 1.85) for lycopene, 1.03 (95 percent CI: 0.64, 1.67) for α-carotene, 1.02 (95 percent CI: 0.63, 1.67) for β-carotene, 1.00 (95 percent CI: 0.61, 1.62) for β-cryptoxanthin, and 1.14 (95 percent CI: 0.70, 1.87) for lutein/zeaxanthin. Among women with an additional lower level of hemoglobin A1c (<6.0 percent) at baseline, there remained no association between plasma carotenoids and the risk of type 2 diabetes (data not shown).

DISCUSSION

In this nested case-control study, we found no prospective association between baseline plasma lycopene and other carotenoids with the risk of type 2 diabetes in middle-aged and older women. A possible crude inverse association for lycopene was attenuated upon adjustment for the other risk factors for diabetes.

The condition of oxidative stress, due to free radical overproduction and/or antioxidant maintenance inadequacy, has been implicated in the pathogenesis of type 2 diabetes (1–3). Increased free radical activities impair insulin action and glucose disposal in the peripheral tissues (2, 3, 31, 32). Free radical-mediated tissue damages also contribute to pancreatic β-cell dysfunction and blunt insulin secretion (1, 3, 32–34). Furthermore, evidence is accumulating that oxidative stress may be actively involved in the pathogenesis of chronic inflammation, a possible common pathway underlying insulin resistance, diabetes, and cardiovascular disease (3). Antioxidants with scavenging ability serve as a defense system against the oxidative stress and therefore may play a protective role in development of type 2 diabetes (8), a hypothesis supported by findings from animal studies (35, 36). The extended double bonds of carotenoids make them powerful antioxidants (37). In population studies, higher consumption of fruits and vegetables, which are the main source of carotenoids, has been linked with lower risk of type 2 diabetes (38–42), suggesting a potential role of carotenoids in preventing the development of type 2 diabetes. However, since fruits and vegetables are also rich in other nutrients that are possibly related to glucose metabolism, epidemiologic evidence on the specific and definitive effects of carotenoids in the primary prevention of type 2 diabetes remains insufficient.

Because carotenoids are not closely regulated by homeostatic mechanisms, blood concentrations of carotenoids are potentially good estimates of habitual dietary intake (28). Plasma carotenoids are inversely associated with type 2 diabetes in a number of cross-sectional studies (6–14). However, the low concentration of plasma carotenoids in patients with diabetes or impaired glucose tolerance may have occurred as a consequence of the altered metabolism of carotenoids (6) and/or the exhausted antioxidant system as a result of diabetes or prediabetes (43). There is evidence that carotenoids, especially β-carotene, transiently decrease during acute illness and return to normal with resolution of the illness, accompanied by simultaneous change in systemic markers of inflammation (44, 45). These findings suggest further that low concentrations of carotenoids in patients with disease might reflect not only the dietary intake but also the other physiologic processes related to the disease (46); therefore, the cross-sectional associations could be explained by reverse causation. To our knowledge, there are only two prospective studies of the relation between baseline dietary or plasma carotenoids and incident type 2 diabetes (17, 18). In a nested case-control study of middle-aged Finnish men and women (17), the risk of type 2 diabetes was lower for those in the highest tertile of baseline serum β-carotene compared with those in the lowest tertile, but this association did not persist upon control for other risk factors. No other major carotenoids were examined in this study. In another Finnish cohort of 2,285 men and 2,019 women aged 40–69 years (18), baseline intakes of α-, β-, and γ-carotene, lycopene, or lutein plus zeaxanthin, estimated by use of a dietary history interview, were not associated with risk of type 2 diabetes during 23 years of follow-up. However, higher dietary β-cryptoxanthin was significantly associated with reduced risk of type 2 diabetes in this cohort. The results from our study in middle-aged and older US female health professionals are otherwise consistent with the previous findings.

Our current study provides novel prospective data for several plasma carotenoids in relation to the subsequent development of type 2 diabetes. For example, lycopene has the strongest singlet oxygen-quenching capacity among the major antioxidant carotenoids (5). High concentrations of dietary lycopene are limited to a small number of red-colored plant foods in the US diet, mainly tomato products such as tomatoes, tomato juice, tomato sauce, and ketchup (47). For these reasons, lycopene has attracted broad interest for its potential role in the prevention of oxidative stress-related chronic diseases such as cancer (48, 49) and cardiovascular disease (49, 50). The association of lycopene with risk of type 2 diabetes is much less studied. Despite a wide array of measurement methods, our study in combination with other prospective studies does not support an important role of lycopene and other carotenoids in the primary prevention of type 2 diabetes. In line with these observations, clinical trial data on the efficacy of carotenoids in preventing type 2 diabetes are limited to one report, in which β-carotene supplementation for 12 years had no effect on the risk of type 2 diabetes in 21,468 US male physicians (19).

Several methodological issues must be considered as potential limitations of the current study. First, although plasma carotenoids are thought to reflect dietary intake, the correlations between plasma concentrations of carotenoids and dietary intake assessed by use of a semiquantitative food frequency questionnaire were low in this nested case-control sample (Spearman's r ranging from −0.05 to 0.01; all p > 0.05). The weak correlation indicates that other physiologic factors might have significant influence on plasma carotenoids beyond the intakes. The single measurement of plasma carotenoids could also introduce nondifferential misclassification and bias the association toward the null hypothesis. Second, the stability of plasma carotenoids stored at −140°C in liquid nitrogen-chilled freezers was not directly evaluated in our study. However, results from other studies support the reasonable long-term stability of these biochemical markers (51, 52). Third, plasma concentrations of carotenoids may reflect other behavioral or dietary patterns that were associated with glucose metabolism. Although comprehensive adjustments were made for multiple risk factors, residual confounding from unknown risk factors and other risk factors either not measured or measured with errors cannot be completely ruled out in observational studies. Finally, these results apply to middle-aged and older women who were generally healthy and willing to participate in a clinical trial. The levels of plasma carotenoids in the present study were generally comparable to those in previously published cross-sectional studies; however, our findings of no prospective association with risk of type 2 diabetes still need to be confirmed or refuted in other populations.

In conclusion, our study found little evidence for an association between baseline plasma lycopene and other carotenoids with the risk of type 2 diabetes after adjustment for multiple risk factors. Direct evidence is needed on whether or not increased plasma carotenoids will improve insulin sensitivity and glucose metabolism. Further research is needed to elucidate the specific mechanisms of dietary carotenoid absorption and metabolism, the effect of plasma carotenoids on free radicals, and the role of these biologic activities in the pathogenesis of type 2 diabetes.

Supported by research grants CA-047988 and H-080467 from the National Institutes of Health, Bethesda, Maryland, and by a grant from Roche Vitamins, Inc., Parsippany, New Jersey.

The authors are indebted to the entire staff of the Women's Health Study for their dedicated and conscientious collaboration.

Conflict of interest: none declared.

References