Self-report of fruit and vegetable intake that meets the 5 a day recommendation is associated with reduced levels of oxidative stress biomarkers and increased levels of antioxidant defense in premenopausal women.

Free radicals and reactive oxygen species are produced as by-products of normal cellular oxidative metabolism. Oxidative stress refers to a condition where overproduction of these byproducts causes damage to components of the body, particularly DNA, lipids, and proteins 1 . Oxidative stress can be balanced through multiple antioxidant mechanisms, which can stabilize or quench reactive oxygen species. Oxidative stress has been implicated in infertility 2 , impaired follicular growth 3 , endometriosis 4 , spontaneous abortion 5 , 6 , and increased risk of delivering a low birth weight infant 7 as well as several chronic degenerative diseases 8 – 10 . Greater understanding of the potential mediators of oxidative stress, given the role it may play in reproductive health, is an important issue for reproductive-age women. Dietary strategies have been advocated to reduce the risks associated with oxidative stress, as a myriad of antioxidants can be obtained from the diet. Fruits and vegetables are rich sources of antioxidants, including vitamins C and E, carotenoids and flavonoids, as well as numerous other phytochemicals 11 . Some intervention trials have shown decreased levels of oxidative stress markers, such as F 2 -isoprostanes, with a fruit and vegetable-rich diet 12 – 17 . However, other studies of fruit and vegetable intake did not find an association with measures of oxidative stress, perhaps due to small numbers (< 15 adults) in the intervention groups 18 , 19 or because the intervention was based on a supplement extracted from foods 19 . Additionally, supplementation with purified antioxidants has not proven to be protective against or chronic disease, and has even resulted in increased risk of disease in some rare cases 20 – 24 . With such conflicting results, more evidence is clearly needed to understand the impact of fruit and vegetable intake on oxidative stress levels. The 2010 Dietary Guidelines for Americans specify a need to increase intake of fruits and vegetables to provide important nutrients, decrease chronic disease risk, and provide a low calorie food choice to maintain a healthy weight 25 . The USDA 5 A Day campaign recommends 2.5 cups (five serving equivalents) of any combination of fruits and vegetables per day for a typical American adult diet; however, the average American meets only 59% and 42% of the goal for vegetable and fruit intake, respectively 26 , 27 . Previous studies of fruit and vegetable intake or antioxidant supplementation on oxidative stress markers have been mostly limited to diseased populations, or populations with particular risk for oxidative stress (i.e. smokers). Previous work from the BioCycle study found that adherence to a Mediterranean diet was associated with decreased lipid peroxidation 28 and that oxidative stress varies across the menstrual cycle 29 . The 5 A Day recommendation may be a simple, easy-to-remember guide to help improve diet and decrease oxidative stress but no prior studies have examined typical daily consumption of the recommended servings of fruits and vegetables in healthy, young women in relation to oxidative stress while taking menstrual cycle variability into account. This study aims to fill these important data gaps by investigating the association between usual fruit and vegetable intake and biomarkers for oxidative stress and antioxidant status in healthy, premenopausal women.

The BioCycle Study is a prospective cohort study of oxidative stress and hormone variation conducted between 2005 and 2007 in 259 healthy, regularly menstruating women aged 18–44y. Participants were recruited using a variety of community-based approaches (clinics, newspaper advertisements, fliers, etc.) from across western New York and were followed for 1 ( n = 9) or 2 ( n = 250) menstrual cycles. Most women (71%) were employed and 58% were full-time students during their participation. Exclusion criteria included current use of oral contraceptives or for the past 3 months, regular intake of vitamin and mineral supplements or certain prescription medications; pregnant or breastfeeding in the past 6 months; and diagnosis of chronic medical conditions, including metabolic disorders and gastrointestinal diseases associated with malabsorption. At the initial telephone screening, women with a self-reported height and weight resulting in a body mass index (BMI, kg/m 2 ) 35 and those with current or planned dietary restrictions for weight loss or medical reasons were excluded. One participant who reported daily multivitamin use in her study diary was excluded, leaving 258 women in this analysis. Details of this study have been previously described 30 . The University at Buffalo Health Sciences Institutional Review Board (IRB) approved the study, and served as the IRB designated by the National Institutes of Health for this study under a reliance agreement. All participants provided written informed consent. Participants were followed for up to two menstrual cycles, with up to eight clinic visits per cycle, timed to cycle phase using fertility monitors to correspond to menses, mid-follicular phase, late-follicular phase, luteinizing hormone (LH)/follicle-stimulating hormone (FSH) surge, predicted ovulation, and early luteal, mid-luteal, and late luteal phases 30 , 31 . These visits correspond to approximately days 2, 7, 12, 13, 14, 18, 22, and 27 of a standardized 28 day cycle. Collection and handling protocols were designed to minimize variability in preanalytic factors, as previously described 32 . The study population was highly compliant, with 94% of women completing ≥7 clinic visits/cycle and 100% completing at least five visits/cycle, with fewer visits typically due to shorter cycles. Nutrient data was collected using a food frequency questionnaire (FFQ) developed and validated by the Nutrition Assessment Shared Resource (NASR) of the Fred Hutchinson Cancer Research Center (FHCRC). The semi-quantitative FFQ was administered three times, once at baseline to determine usual intake over the past 6 months and once at the end of each of two cycles to determine usual intake in the month of the previous cycle. The FFQ was administered at the appointment occurring in the late luteal phase of the menstrual cycle and was reviewed by staff to ensure completion of the questionnaire. At least one cycle-specific FFQ was available for 97% of participants. Additionally, 24-hour dietary recalls were conducted up to four times per cycle (menses, mid-follicular phase, ovulation, and mid-luteal phase), on days corresponding with blood sample collection, for a total of up to eight recalls over two cycles. All participants completed ≥2 dietary recalls per cycle, and the majority (87%) completed 4 recalls per cycle. For both the FFQ and 24-hour recalls, average daily fruit and vegetable servings were calculated using the 5 A Day method, summing servings across fruit and vegetable items 33 , 34 . A vegetable serving was defined as 1 cup (250 mL) of raw leafy vegetables, ½ cup (125 mL) of other cooked or raw vegetables, or ½ cup (125 mL) of vegetable juice. A serving of fruit was defined as 1 medium fruit, ½ cup (125 mL) of chopped, cooked, or canned fruit, ¼ cup (62.5 mL) of dried fruit, or ¾ cup (187.5 mL) of 100% fruit juice. All dietary data were analyzed using the Nutrition Data System for Research software version 2005, developed by the Nutrition Coordinating Center, University of Minnesota, Minneapolis, MN. Collection of fasting blood samples was scheduled to occur between 0700 and 0830 hours at each cycle visit. All lipid peroxidation measurements were performed using EDTA anticoagulated blood plasma. Free F 2 -isoprostanes were the primary marker of lipid peroxidation, as it is considered to be the gold standard 35 . The performance characteristics of the analytical laboratory methods have been previously described in detail 36 , 37 . F 2 -isoprostanes were measured in plasma with a gas chromatography-mass spectrometry-based method by the Molecular Epidemiology and Biomarker Research Laboratory (University of Minnesota, Minneapolis, MN) (9.4% Coefficient of Variation [CV]). 9-hydroxyoctadecadieneoic acid (9-HODE, 9.0% CV) and 13-hydroxyoctadecadienoic acid (13-HODE, 9.2% CV) were determined by HLPC with diode array detection at 234 nanometers (nm) 38 . Fat-soluble antioxidant vitamins and micronutrients were measured in duplicate at the University at Buffalo (Buffalo, NY). Serum retinol (6.2% CV), carotenoids (β-carotene, β-cryptoxanthin, lycopene and lutein; 5.3%, 6.0%, 7.6%, and 6.6% CVs, respectively), and tocopherols (α- and γ-tocopherol; 5.0% and 4.6% CVs, respectively) were measured simultaneously by HPLC 39 . The dinitrophenylhydrazine method was used to determine total ascorbic acid (9.6% CV) in heparin plasma which was stabilized immediately at collection in 6% meta-phosphoric acid 40 . Antioxidant enzymes were measured using kinetic enzyme assays adapted to the Cobas Fara II autoanalyzer (Roche Diagnostic Systems, Inc., Basel, Switzerland) 41 . Erythrocyte superoxide dismutase (SOD) activity was determined by the inhibition of the oxidation of cytochrome C by xanthine/xanthine oxidase (4.6% CV). One unit of SOD activity was defined as the amount of enzyme needed for 50% inhibition of the reaction. Erythrocyte glutathione peroxidase (GPx; 5.0% CV) and erythrocyte glutathione reductase (GSHR; 3.7% CV) were performed using OxiTek reagent kits from ZeptoMetrix (Buffalo, NY). Erythrocyte enzyme activities were normalized per gram of hemoglobin (Hb) 37 . At baseline, height (m) and weight (kg) were measured using standard protocols and used to calculate BMI. Participants completed questionnaires regarding reproductive health history, lifestyle, family, and physical activity. Standard International Physical Activity Questionnaire cut points 42 were used to create high, moderate, and low physical activity categories. In daily diaries, women recorded intake of medications, supplements, and/or vitamins. Descriptive statistics for continuous and categorical covariates were compared between women meeting and not meeting the 5 A Day recommendation based on the average consumption from all 24-hour recalls (≤8 per woman) using ANOVA or Fisher's exact test, as appropriate. Total energy intake was averaged across all 24-hour recalls in the descriptive tables. Mean levels of lipid peroxidation, antioxidant enzymes, and antioxidant vitamins by menstrual cycle phase were compared between women meeting and not meeting the 5 A Day recommendation based on the average consumption per cycle (≤4 recalls per cycle) using repeated measures ANOVA. Descriptive results for FFQ measures were similar and are not presented. Oxidative stress and antioxidant measures (up to 8 measures per woman for each cycle) were analyzed in association with fruit and vegetable servings reported in the same cycle using linear mixed models that accounted for repeated measures and the correlation between and within participants. All analyses compared meeting the 5 A Day recommendation to not meeting it. For analyses based on the FFQ, the fruit and vegetable intake for each cycle was used to categorize women for that cycle. For analyses based on 24-hour recalls: dietary recall during menses was paired with oxidative stress measures during menses; dietary recall during the mid-follicular phase with mid-follicular oxidative stress measures, dietary recall on the day of predicted ovulation with the oxidative stress measures during the three peri-ovulatory period visits (late-follicular, LH/FSH surge, predicted ovulation) and dietary recall on the mid-luteal visit with early, mid, and late luteal phase oxidative stress measures. Women could change categories by cycle (FFQ and 24-hour recall analyses) and by menstrual cycle phase (24-hour recall analyses). Lipid peroxidation markers, antioxidant enzymes, and antioxidant vitamins were log transformed to improve fit in multivariable models. Models based on 24-hour recall are menstrual-cycle phase specific while those based on FFQ relate the typical diet for each cycle to multiple measures of oxidative stress in that cycle. Covariates considered for inclusion in multivariate models were determined a priori after a review of prior literature and included: age (continuous), energy intake (continuous), clinically-measured BMI (continuous), race (Caucasian, African-American, Asian, other), smoking status (never, current/past), physical activity (low/moderate, high), marital status (married, not married), parity (0, ≥1), income (<$19,999; $20,000–39,999; $40,000–74,999; ≥$75,000), education (completed high school, did not complete high school), and past oral contraceptive use (yes, no). Variables were included in the final adjusted model if they also changed the exposure coefficient by >15% and were significant at p =0.10 in either the FFQ or 24-hour cycle specific model. The final adjusted model controlled for age, energy intake, race, income, marital status, parity, and former oral contraceptive use. SAS version 9.2 (SAS Institute, Cary, NC) was used for all statistical analyses. All testing was based on a priori hypotheses and no adjustments were made for multiple comparisons.

Overall, this cohort consisted of young women (mean age: 27.3 years, range 18–44) who were of healthy weight (mean BMI: 24.1 kg/m 2 ) and were non-smokers ( Table 1 ). Most had completed high school (87.2%), were not married (74.8%), and had no children (74.2%). The women in this cohort had low overall fruit and vegetable intake (median=2.71 servings/d). Women who met the 5 A Day recommendation (17.8% of all subjects) tended to be older, were more likely to be Caucasian and have higher income. They were also more likely to be married, have children, and to be prior oral contraceptive users. Meeting the 5 A Day recommendation was associated with greater energy intake, but no differences were observed with respect to BMI, physical activity, education, and smoking. F 2 -isoprostanes were observed to be significantly lower across the menstrual cycle among women who consume 5 A Day (p<0.05), but no differences were observed for 9-HODE or 13-HODE ( Table 2 ). SOD was also lower among those meeting the recommendation (p<0.005) with no significant differences in GPx or GSHR levels. With regard to anti-oxidant vitamins, most were significantly higher when 5 A Day was met including α-tocopherol, ascorbic acid, retinal, lutein, β-carotene and β-cryptoxanthin. An inverse association was observed for γ-tocopherol and lycopene was not related to 5 A Day consumption. In multivariable analyses ( Table 3 ), results were generally consistent for both cycle phase- specific models based on 24-hour recalls and the typical diet analysis based on cycle-specific FFQ. All effect estimates for lipid peroxidation markers were in the expected direction, with lower levels when 5 A Day was met, but only F 2 -isoprostanes were significantly lower after adjustment. This difference may reflect the fact that total HODEs and free F2 isoprostanes represent lipid peroxides derived from different parent lipid molecules (linoleic and arachidonic acids, respectively) as well as different lipid peroxidation processes which may vary in response to the same dietary influences. We also anticipated lower antioxidant enzyme status with 5 A Day and that pattern of results was somewhat stronger for the typical diet analysis based on FFQ where GSHR was significantly lower. Ascorbic acid, lutein, β-carotene and β-cryptoxanthin were all significantly increased in both cycle phase-specific (24-hour recall) and cycle-specific (FFQ) models when 5 A Day was met. Α-tocopherol was significantly increased in the phase-specific model only and no differences were observed for γ-tocopherol, retinol or lycopene.

Among healthy premenopausal women, we observed that meeting the 5 A Day recommendation was associated with lower concentrations of lipid peroxidation, as measured by F 2 -isoprostanes, with lower concentrations of erythrocyte antioxidant enzymes, specifically SOD (in unadjusted models),and GSHR (after adjustment), and higher concentrations of most plasma/serum antioxidant vitamins. These findings are particularly relevant for reproductive-age women, as oxidative stress has been implicated in infertility and poor birth outcomes in addition to a host of chronic diseases 2 – 7 , 9 , 10 . No prior studies have focused on young adult women or have looked at the efficacy of the 5 A Day recommendation in terms of oxidative stress, but other investigations have reported an inverse relationship between fruit and vegetable intake and biochemical oxidative stress measures 12 , 43 , 44 . A cross-sectional analysis of male and female adolescents (mean age: 15 y, n=285) found a substantial inverse association between total fruit and vegetable intake and F 2 -isoprostanes, primarily explained by vitamin C and folate intake (41). Studies in older adults report similar findings 9 , 44 . A randomized intervention showed that increased Brassica vegetable intake, but not a micronutrient supplement, decreased F 2 -isoprostane levels in men and postmenopausal women (mean age: 57 y, n=20), providing evidence that the vegetables impacted oxidative stress where supplementation did not 9 . The Study of Women's Health Across the Nation showed that in nonsmoking women (ages 42–52 y) greater daily vegetable intake was associated with a lower concentration of urinary F 2 -isoprostanes 44 Our findings support the notion that increased fruit and vegetable consumption is associated with lower F 2 -isoprostane levels and we observed a similar effect when 5 A Day was met within menstrual cycle phase as with typical diet. Fruits and vegetables are thought to ameliorate oxidative stress partially due to their high antioxidant content. As anticipated, we observed increased levels of α-tocopherol, lutein, β-carotene, β-cryptoxanthin, and ascorbic acid when the 5 A Day recommendation was met. In general, we observed lower levels of antioxidant enzyme activity among women who met the 5 A Day recommendation, particularly for typical diet, consistent with the explanation that more activity is required in the presence of higher oxidative stress 45 , 46 . Erythrocyte activity of SOD, GPx, and GSHR were generally inversely associated with servings of fruits and vegetables but only GSHR was significant in the adjusted model. This study, while it expands on previous research, has several limitations. The BioCycle cohort is has strict inclusion criteria which results in a healthy, normal weight, sample of women with regular menstrual cycles and no recent pregnancy or hormonal contraception. These inclusion criteria strengthen the internal validity of the study by limiting known sources of variability in the complex biologic measures under study, but do limit generalizability. However, young adult women are not well studied in previous investigations of dietary intake and oxidative stress and our analysis fills an important knowledge gap – whether or not the straightforward message to eat 5 A Day is associated with changes in biomarkers of oxidative stress and antioxidant defense in healthy young women. A particular strength of the BioCycle Study is the multiple measures of lipid peroxidation, antioxidant enzymes and vitamins which were measured up to 8 times over the course of each menstrual cycle. Our measures of fruit and vegetable consumption are all based on self-reported intake. We used FFQs to capture typical consumption for each cycle recognizing that this method might be subject to recall bias but comparison to 24-hour recalls in the same study period and cycle phase-specific analyses yielded similar findings. We also conducted analyses using fruit and vegetable servings as a continuous measure and categorized by servings up to 5 A Day, all with generally similar findings. Finally, oxidative stress measures are known to vary across the menstrual cycle, specifically increased F 2 -isoprostanes associated with estrogen levels 29 , and these analyses take this important source of variability into account. Our analyses considered both typical consumption and cycle phase-specific associations in evaluating the 5 A Day recommendation. Of note is that 5 A Day and the Dietary Guidelines for Americans, 2010, do not exclude starchy vegetables like potatoes from their recommendations to increase fruit and vegetable consumption. We recognize that white potatoes products can be an indicator of an unhealthy diet 47 but also note that potatoes are a good source of vitamin C 48 . Since potatoes are commonly eaten and they are part of the 5 A Day recommendation, we include them in the calculation of servings for our analyses. As such, our findings reinforce the simple message that increasing fruit and vegetable consumption and meeting 5 A Day is associated with lower oxidative stress. A potentially successful avenue for increasing fruit and vegetable consumption is to encourage more consumption of foods already being eaten. While ketchup and fried potatoes were eaten by 70–85% of women in our study, they were the most commonly eaten servings among the lowest fruit and vegetable consumers, possibly explaining the lack of association for lycopene since ketchup is a major source of lycopene 49 . Green salads were the 3 rd most common serving for the lowest consumers but they were the most common serving for all other women. Given the acceptability of green salads to this population, they may represent a key avenue for increasing fruit and vegetable consumption. Strategies that encourage adding volume and/or additional vegetables and fruits to green salads and mixed dishes is a potentially easy and acceptable way to change the eating habits of young adult women to include more servings of fruits and vegetables. In summary, we found that self-report of meeting the 5 A Day recommendation was associated with lower biomarkers of oxidative stress and improved antioxidant defense. For isoprostanes, benefits were consistently observed for both a menstrual cycle phase-specific analysis and an analysis of typical diet by cycle. Women who consume very low levels of fruits and vegetables may be encouraged to eat more fruits and vegetables in green salads and mixed dishes as a strategy to increase their average daily servings and potentially impact their oxidative stress measures.