点击显示 收起
From the Colorado State University, Fort Collins, CO (HJT, SS, AH, AD, BM, PW, ZZ, and WJ), and the AMC Cancer Research Center, Denver, CO (JH and CO)
2 Supported by award DAMD17-99-1-9062 from the Department of Defense, USAMRMC Breast Cancer Research Program. 3 Address reprint requests to HJ Thompson, Cancer Prevention Laboratory, Colorado State University, 1173 Campus Delivery, Fort Collins, CO 80523. E-mail: henry.thompson{at}colostate.edu.
ABSTRACT
Background: Health benefits associated with diets rich in vegetables and fruit (VF) are often attributed to the antioxidant activity of their constituent phytochemicals. However, in vivo evidence that VF actually reduce markers of oxidative stress is limited.
Objective: An 8-wk dietary intervention was conducted to test the hypothesis that increased VF consumption decreases oxidative stress. Urinary excretion of 8-isoprostane F2 (8-iso-PGF2) was used as an index of whole-body lipid peroxidation.
Design: The diets evaluated had comparable amounts of all macronutrients but varied in their content of VF. After a 2-wk low-VF (3.0 servings/d) run-in diet, 246 women were randomly assigned to receive either 3.6 (low) or 9.2 (high) servings VF/d. The low-VF group was switched to the high-VF diet during the final 2 wk of the study. Blood and first-void urine specimens were obtained at baseline and at 2-wk intervals thereafter.
Results: The run-in diet reduced 8-iso-PGF2 concentrations by 33% (P < 0.0001). The excretion of 8-iso-PGF2 with the low-VF diet remained the same as that with the run-in diet, whereas urinary concentrations of 8-iso-PGF2 were further reduced (P < 0.01) by the high-VF diet, either fed throughout the study or when the diet was switched from low to high VF (P = 0.05). The greatest reductions in 8-iso-PGF2 were observed in subjects in the highest quartile of baseline concentrations of 8-iso-PGF2.
Conclusions: A significant reduction in the excretion of 8-iso-PGF2 was induced by the run-in diet and the high-VF diet. The degree of reduction was related to the subject's baseline urinary concentration of 8-iso-PGF2.
Key Words: Vegetable • fruit • lipid peroxidation • dietary intervention • 8-isoprostane F2
INTRODUCTION
Health benefits associated with diets rich in vegetables and fruit (VF) are frequently attributed to the antioxidant activity of their phytochemical constituents (1, 2). However, the in vivo evidence that VF actually reduce markers of oxidative cellular damage is limited (3), and the results of published studies are contradictory and have frequently been null (4–7). Although null findings may simply reflect the lack of activity in vivo of many antioxidant phytochemicals because of poor absorption, rapid clearance, or chemical modification to antioxidant-inactive analytes during metabolism, other factors that may contribute to contradictory findings include the quality of the study design, compliance of research participants, technical problems inherent in the measurement of oxidized cellular molecules, and the artifactual formation of oxidized species during the analysis of tissues or body fluids for concentrations of oxidative markers.
Another possible explanation for the contradictory findings about the antioxidant activity of various phytochemical preparations is that there is an essential level of cellular oxidation that is the outcome of aerobic metabolism, and that it is not possible to reduce oxidation below some basal constitutive level. If this is the case, then it would be expected that dietary antioxidants would not be observed to reduce cellular oxidation in individuals who enter a study with low constitutive levels of oxidation. In such individuals, the effects of dietary antioxidants would be null, which suggests that the null effect is conditional on the oxidative status of an individual.
In light of the controversial data surrounding the purported role of the antioxidant activity of VF phytochemicals in vivo, and of the various explanations for such effects that were summarized above, a clinical intervention study was undertaken to determine 1) whether 2 whole-food defined diets that differed in VF content would affect the excretion of a biomarker of lipid peroxidation, and 2) whether a uniform response to dietary antioxidant phytochemicals would be observed or whether greater effects would occur in individuals with high baseline concentrations of the lipid peroxidation biomarker.
Lipid peroxidation can result in chain reactions that generate a sustained increase in reactive compounds that can oxidize cellular macromolecules, and chain-breaking antioxidants are known to inhibit such events. We measured the effect of VF intake on the excretion of 8-isoprostane F2 (8-iso-PGF2), a marker that is now widely accepted as a stable and reliable index of overall lipid peroxidation (8, 9), which is not prone to introduction of oxidation artifacts during its measurement. F2-isoprostanes are products of free radical–catalyzed lipid peroxidation of arachidonic acid (8). They are formed in situ, esterified to phospholipids, and released by phospholipases into the plasma. F2-isoprostanes are subsequently removed from plasma via the kidney and are excreted in the urine. Elevated F2-isoprostane concentrations in plasma and urine have been found in patients with an array of disease conditions in which oxidative injury is a component (8, 9).
SUBJECTS AND METHODS
Subjects
The subjects were recruited from a group of women interested in women's health issues. To be eligible for participation, the women had to be 21 y of age, not be pregnant or lactating, have no known food allergies, not be consuming a special diet prescribed by a physician, had to be familiar with cooking and willing to prepare prescribed meals for 8 wk, had to be willing to keep a record of the foods eaten during the study, had to regularly consume 2 alcoholic beverages/d, not be using tobacco products of any type, not be taking multivitamin supplements or be willing to stop taking vitamin supplements while participating in the study, not be a competitive athlete, have maintained approximately the same body weight for the past 6 mo, and not be regularly taking medications, other than birth control or estrogen replacements. The failure of interested individuals to meet the eligibility criteria formed the basis for exclusion from the study. The clinical protocol was approved by the Institutional Committee for the Protection of Human Subjects. Two hundred forty-six women enrolled in the study and gave informed consent.
Study design
The study design is summarized in Figure 1. For the initial 2 wk (run-in phase), all subjects consumed a menu that was balanced in nutrients and designed to provide 30% of calories from fat (10% saturated fat, 10% monounsaturated fat, and 10% polyunsaturated fat) and 2–4 servings VF/d, depending on caloric needs. For the next 4 wk, the participants were assigned to 1 of 2 diet groups that differed in VF content. For the last 2 wk of the intervention, the low-VF group was switched to the high-VF diet. At the initiation of the study and at 2-wk intervals thereafter, samples of blood and 3 consecutive daily first-void urine specimens were obtained. The procedure for, and timing of, the sample collection was based on previous work reported by our laboratory that showed significant changes in plant food intake biomarkers and markers of oxidative damage within 2 wk of initiating a dietary change (10).
FIGURE 1.. Study design. The effects of a diet low in vegetables and fruit (VF) or a high-VF diet were studied during an 8-wk intervention. Blood and urine were obtained at baseline, after the run-in phase, and at 2-wk intervals thereafter. Participants assigned to the low-VF diet were switched to the high-VF diet during the final 2 wk of the intervention.
Dietary intervention
The study participants were given a cookbook containing daily menus and recipes that prescribed the pattern of the foods that were to be consumed during the study. To provide for different tastes and flexibility, the participants were also permitted some choice within specific botanical families of VF and within the meat, dairy, and grain food groups. Daily food records were maintained to document everything that was consumed. All 3 diets—the run-in (3.0 servings VF), the low-VF (3.6 servings), and the high-VF (9.2 servings)—met the recommended dietary allowances of the Dietary Reference Intakes; provided similar amounts of protein, fat, and carbohydrate (Table 1); and were balanced in nutrients from the various food groups. To ensure variety, plant foods were provided from a specified list of botanical families based on our previously published work (10); the distribution of actual consumption of VF by botanical family is shown in Table 2. Focus groups conducted before the main intervention was initiated identified convenience as a significant potential barrier to retention. Therefore, approximately one-third of all lunch or dinner entrees were prepared by a retail delicatessen from study recipes and were provided to the participants. The participants were gradually weaned from supplements before the diets began so that no vitamin supplements were consumed during the study.
View this table:
TABLE 1. Intake of macronutrients based on recorded food intake1
View this table:
TABLE 2. Average daily number of servings of vegetables and fruit by botanical family1
Laboratory measurements
Urinary 8-iso-PGF2
Analysis of 8-iso-PGF2 in urine provides a time-averaged index of lipid peroxidation over a relatively prolonged interval that offers greater utility than the more transitory information provided by analysis of blood. Moreover, urine contains very little arachidonic acid, thereby limiting artifactual oxidation and isoprostane production after sample collection. Because we found intraindividual day-to-day excretion of urinary 8-iso-PGF2 to vary considerably in human subjects, we elected to measure 8-iso-PGF2 abundance in pooled urine samples generated from multiple specimens. The first void of the morning urine was collected without preservative in plastic vessels on 3 consecutive d. The decision to use first voids rather than 24-h collections was based on data indicating that 24-h averages expressed per mg creatinine were not statistically different from values obtained from first voids and on our experience that collecting reliable 24-h urine samples from free-living subjects is problematic.
As noted previously (11), early studies of isoprostanes were limited in scope by the technical demands of sample processing for gas chromatography–mass spectrometry (GC-MS). However, as reviewed by Schwedhelm and Boger (12), reliable enzyme-linked immunosorbent assays (ELISAs) for specific isoprostanes, such as the one reported by Proudfoot et al (13), have become available. These assays give results that parallel those obtained via GC-MS. In this study, urine samples from each 3-consecutive-day collection period were pooled and subjected to solid-phase extraction. Isolute solid-phase extraction columns (catalog no. 221-0020-B; International Sorbent Technology Ltd, from Jones Chromatography, Lakewood, CO) containing 200-mg end-capped C18 sorbent in a 3-mL reservoir were used for the extraction. The extraction method of Wubert et al (14) was modified to enhance retention of 8-iso-PGF2, and recovery >95% was verified during method development by monitoring analyte extraction with tritiated 8-iso-PGF2 tracer. After extraction, the 8-iso-PGF2 concentration was determined with the use of an ELISA kit from Cayman Chemical (Ann Arbor, MI). The dried solid-phase extraction eluate containing 8-iso-PGF2 was reconstituted with assay buffer (tris buffered saline), and the reconstituted samples were further diluted such that the 8-iso-PGF2 concentration was in the most reliable range of the ELISA; that dilution was first estimated based on creatinine concentration. All time points from a subject were included within an analytic run, and both treatment groups were represented approximately equally within a run as well. This approach minimized the chances of having interassay variability masquerade as or obscure treatment effects. Identical samples of control urine were extracted and analyzed with every run to monitor assay performance. On the basis of this control sample, the interassay precision for 31 analytic runs over 54 d was 16.0%. To keep the analysts blind to treatment group, the project biostatistician provided pairs of subject accession numbers such that each pair contained one subject from each treatment group, and the analytic runs were configured from those pairs. The 8-iso-PGF2 values were expressed per mg creatinine. Creatinine concentration was measured with a kit (catalog no. 558-A; Sigma-Aldrich, St Louis, MO) based on visible absorbance of the Janovski complex chromophore generated by derivitization with picrate. The colorimetric creatinine assay is very consistent; interassay precision of identical control urine samples included with every analysis was 2.4% for 32 analytic runs over 58 d.
Plasma carotenoids
The concentration in plasma of 5 carotenoids—-carotene, ß-carotene, lycopene, lutein, and ß-cryptoxanthin—was determined to provide a biochemical index of compliance with the intervention diets. Plasma was separated from nonfasting blood, collected by using potassium EDTA as an anticoagulant, and stored at –80 °C for the subsequent measurement of carotenoids. Carotenoid concentrations were measured by using a reversed-phase HPLC method based on those of Peng and Peng (15) and Hess et al (16). Briefly, plasma was extracted with hexane, which was removed under reduced pressure. The extract was reconstituted with mobile phase and separated by isocratic reversed-phase chromatography. Photodiode array detection facilitated quantitation of 5 analytes at 3 different wavelengths from a single injection and enabled use of spectral data for confirmation of peak identity. Standard curves for each analyte generated with each analytic run proved to be remarkably stable over time. For 34 analytic runs over 69 d, the CV for the slopes of the curves ranged from a high of 8.4% for -carotene to a low of 2.0% for ß-cryptoxanthin. Working standard solutions containing all 5 analytes were prepared from a stock carotenoid solution, the concentration of which was regularly monitored by using absorbance photometry. Stock solutions of carotenoids in hexane and chloroform were stable at –20 °C when preserved with butylated hydroxytoluene. A human plasma control sample was extracted and analyzed with each run to monitor assay precision; the CV ranged from a high of 8.6% for lutein to a low of 3.7% for ß-cryptoxanthin. Analyses were configured such that all time points from a subject were analyzed together, and each run included approximately equal numbers of subjects from both treatment groups. Analysts were blind to treatment group.
Other measurements
Participants completed printed surveys at baseline and after intervention. The baseline surveys provided data on demographics, VF consumption, and levels of physical activity. The postintervention surveys provided self-reported compliance and feedback on the intervention. Self-reported compliance with the prescribed dietary intervention was measured by using the question, "On a scale from 1 to 10, how precisely did you follow the prescribed diet?," where 1 was "followed the diet infrequently" and 10 was "followed the diet precisely."
Statistical analyses
Differences in categorical variables at baseline across randomization groups were evaluated by using a chi-square test for independence of proportions. Continuous data were tested for differences in means by using a 2-group, two-sided t test or analysis of variance. Distributions for the continuous data were also checked to determine whether transformations were appropriate. Statistical results for 8-iso-PGF2 data were based on the log transformation of the raw data.
Maximum likelihood estimation of a multivariate repeated-measures mixed-effects model using all available data were performed for the analysis of urinary 8-iso-PGF2 and plasma carotenoids. This approach is conceptually identical to multivariate analysis of variance, but avoids the case-wise deletion of subjects with missing assessments and relaxes the assumption that missing data are missing completely at random (17). The maximum likelihood method provides unbiased estimates under the less restrictive assumption that missing data are missing at random; that is, missing values of the outcome may be related to observed values of the outcome, an unlikely event given the nature of this experiment. The regression model was parameterized to fit the experimental design; estimates were based on a piecewise linear model with knots at 2, 4, and 6 wk to allow the slopes to change whenever the diets changed: at the end of the run-in period and at crossover. The knot at 4 wk was placed to allow for the possibility that the full effect of the high-VF diet would be evident after 2 wk. The fully parameterized model with 3 knots is as follows:
RESULTS
Characteristics of the study population
Two hundred forty-six subjects enrolled in the study and gave informed consent; 208 completed the study. Thus, the dropout rate was 15.4%. The participants were predominantly white (95%), were well-educated (73% had 4 y of college), and had a median age of 49 y. Nearly 40% of the subjects were overweight [body mass index (BMI; in kg/m2) between 25 and 30], and 22% were obese (BMI 30); the median BMI was 26.5. The self-reported average number of daily servings of VF was not significantly different between groups: 4.3 compared with 4.0 (P = 0.14). The baseline characteristics of the subjects by diet group are shown in Table 3, which indicate that there were no significant differences in characteristics or outcome measures between the 2 intervention groups.
View this table:
TABLE 3. Baseline characteristics of the women by group1
Attrition
There were no significant differences in loss-to-follow-up between the randomization groups. The most common reason for dropping out of the study was inconvenience. The rate of attrition was stable throughout the study; 10 subjects dropped out after each 2-wk interval. Characteristics of the 15% of subjects who did not complete the study were compared with those of completers (data not shown). The subjects were not significantly different by randomization group, age, BMI, self-reported number of servings of VF, or education. Although the baseline 8-iso-PGF2 concentration was lower in the subjects with complete data than in those who dropped out (0.75 compared with 0.91 pg/mg creatinine), the difference was not statistically significant (P = 0.07). This is evidence in support of our assumption for statistical analyses that missing data were missing at random. Mean self-reported compliance was 85% and did not vary by intervention group.
Plasma carotenoids
Plasma carotenoid concentrations are shown in Table 4. At baseline, only ß-carotene was significantly different between diet groups (P = 0.02), an effect that was not evident by the 2-wk run-in diet. The concentration of each of the carotenoids was greater in the high-VF group than in the low-VF group after 2 wk and 4 wk of the respective diets. During the last 2 wk of the study, the low-VF group switched to the high-VF diet; at the end of this period, plasma concentrations of -carotene and lycopene were no longer significant and the other 3 analytes were converging as expected.
View this table:
TABLE 4. Effect of the dietary interventions on the plasma carotenoids of the women1
Urinary 8-iso-PGF2
Baseline concentrations of 8-iso-PGF2 ranged from 192 to 4873 pg/mg creatinine. The run-in diet was expected to reduce the between-subject variability in 8-iso-PGF2, which it did. However, mean 8-iso-PGF2 decreased by 33% (P < 0.0001) as well (Table 5). No further reductions in 8-iso-PGF2 were observed in the participants randomly assigned to the low-VF diet; whereas, after 2 wk of the high-VF diet, participants experienced another 14% reduction in the mean concentrations of this analyte. Concentrations of 8-iso-PGF2 remained constant in the high-VF group for the remainder of the study. All subjects consumed the high-VF diet for the last 2 wk of the study, and 8-iso-PGF2 declined significantly in the group switched from the low-VF to the high-VF diet (P = 0.05).
View this table:
TABLE 5. Effect of the dietary interventions on the urinary excretion of 8-isoprostane F2 in the women1
We hypothesized that subjects with high urinary 8-iso-PGF2 concentrations would be most responsive to dietary intervention and that there exists a low, constitutive concentration of lipid peroxidation below which a subject is unlikely to drop. To test this hypothesis, the study population was divided into quartiles by baseline 8-iso-PGF2, and the response of women in each quartile to the dietary interventions was assessed (Figure 2). A dramatic difference in response was observed depending on the baseline quartile of 8-iso-PGF2. The greatest reductions in urinary 8-iso-PGF2 were observed in the subjects with the highest baseline concentration of this analyte. After 2 wk of the run-in diet, the mean 8-iso-PGF2 concentration decreased by 45% (P < 0.001) in the highest quartile and by 18% (P 0.001) in the lowest 3 quartiles. After 2 wk of the study diets, a 20% difference (180 pg/mg) in mean 8-iso-PGF2 was observed between diet groups in the highest quartile (P = 0.01). P values are based on maximum likelihood estimates from the repeated-measures model described above, with an additional indicator variable for the baseline quartiles. No significant change in the urinary excretion of 8-iso-PGF2 was observed in subjects in the lowest quartile of baseline 8-iso-PGF2 (down 3% after 2 wk of the run-in diet and down by 12% after both the low- and high-VF diets after 2 wk of the study diets). The effect of the high-VF diet on isoprostane excretion seen after 2 wk was maintained thereafter. Interestingly, the subjects in the highest baseline quartile of 8-iso-PGF2, despite experiencing dramatic reductions in the urinary excretion of this analyte, still had higher 8-iso-PGF2 excretion than did the subjects in the lower 3 quartiles, irrespective of the diets to which they were assigned throughout the 8-wk intervention.
FIGURE 2.. Mean (±SEM) urinary excretion of 8-isoprostane F2 in women after the consumption of a diet low in vegetables and fruit (VF) or a high-VF diet. The data are displayed by quartiles of excretion of 8-isoprostane F2: 1st, <440 pg/mg creatinine; 2nd, 440 to <640 pg/mg creatinine; 3rd, 640 to <940 pg/mg creatinine; and 4th, 940 pg/mg creatinine. The number of subjects varied over time: baseline (n = 246), after the run-in diet (n = 232), after 2 wk of the intervention diets (n = 221), after 4 wk of the intervention diets (n = 208), and at the last visit (n = 201). Statistical significance was determined on the basis of maximum likelihood estimates from a repeated-measures mixed-effects model that used all available data. The hypotheses were tested by setting up appropriate contrasts (linear combinations of model parameters) within the mixed model; the P values associated with the resulting t statistics are reported. The percentage reduction in 8-isoprostane F2 attributed to the run-in diet was directly proportional to baseline concentrations of 8-isoprostane F2 excretion, ie, the reduction was small (6%; P = 0.05) in the lowest quartile and became progressively larger with increasing quartiles [15.1%, 29.9%, and 45.2% (P < 0.001) in the 2nd, 3rd, and 4th quartiles, respectively]. The effects of the dietary intervention were somewhat smaller than those observed during the run-in diet; in the lower quartiles, no significant difference in 8-isoprostane F2 excretion were observed between the diets. However, in the highest quartile, the difference was 180 pg/mg creatinine (P = 0.01) after 2 wk.
Despite efforts to have the participants maintain their weight throughout the study, weight loss was observed. The average weight loss over the 8 wk was 2.9 kg (6.5 lb); the greatest effects were observed during the run-in phase and in the first 2 wk of the intervention diets. Changes in weight were the same in both intervention groups. Inclusion of weight change as a covariate in the statistical model used to assess the effects of diet on excretion of 8-iso-PGF2 did not change the conclusions; rather, statistical significance increased somewhat due to the reduction in variability when weight change was added to the regression.
DISCUSSION
In the present study, a significant reduction in the excretion of 8-iso-PGF2 was induced by the run-in and high-VF diets, and the degree of reduction observed was related to the concentration of 8-iso-PGF2 in an individual subject's baseline urine specimen. Factors that should be considered in the evaluation and interpretation of these findings are discussed in the following paragraphs.
Compliance
Self-reported compliance with the intervention diets was high—85%. Although it is well known that self-report can be biased to overestimate a measurement such as compliance, we also assessed plasma carotenoids as a biochemical marker for compliance. The data shown in Table 4 are consistent with a high level of compliance. Specifically, plasma concentrations of all measured carotenoids increased significantly when participants were switched from the run-in diet to the high-VF diet; however, plasma carotenoid concentrations in participants switched from the run-in diet to the low-VF diet either remained the same or decreased during the same time interval. Moreover, when participants in the low-VF diet group were switched to the high-VF diet, their plasma carotenoids increased to concentrations comparable with those of participants assigned to the high-VF diet group.
Run-in diet
As noted in the Subjects and Methods section, the run-in diet was designed to meet the recommended dietary allowances, but it was low in VF (3.0 servings/d as consumed). It was expected that the uniformity of dietary intake would reduce the variability in excretion of 8-iso-PGF2 among subjects. As shown in Table 5, the variability in concentrations of 8-iso-PGF2 observed at baseline was surprisingly large; SD: 534 pg/mg; range: 192-4873 pg/mg creatinine. The run-in diet reduced the SD by nearly 50%. In addition, the effect of the run-in diet on excretion of 8-iso-PGF2 was also striking; concentrations were reduced by 33% (Table 5). These data provide clear evidence that modification of dietary behaviors from those self-selected to a common set prescribed via a dietary intervention can dramatically reduce a marker of whole-body oxidative stress that characterizes peroxidation of lipids. Thus, we argue that cellular oxidation, at least that of lipids, can be influenced by lifestyle practices even when VF intake is relatively low. Although we cannot identify the basis for the effect of the run-in diet and the lifestyle changes that compliance with this dietary intervention necessitated, several observations are warranted. First, participants remarked that they were unaccustomed to the meal pattern prescribed: 3 meals and 2 snacks daily. They also observed that they were eating away from home less frequently than was typical before the intervention. We note that the intervention menus were balanced in an effort to meet the recommended dietary allowances and distribute the intake of protein, fat, and carbohydrate throughout the day. Another effect of the diet was to reduce overall intakes of simple carbohydrates, such as desserts and sodas. Also, as noted in the Results section, weight loss did occur in many subjects in both the low and high VF groups, but the effect of diet on 8-iso-PGF2 excretion was not diminished when weight loss was included in the statistical model.
Intervention diets
The run-in diet reduced the excretion of 8-iso-PGF2 by 33%, and this level of excretion served as the reference point for determining the effects of low compared with high intakes of VF. As shown in Table 5, the excretion of 8-iso-PGF2 was not affected by the transition from the run-in diet to the low-VF diet or over the 4 wk of the low-VF diet. The low-VF diet was similar to the run-in diet, with the exception of an emphasis on whole grains. On the other hand, consumption of the high-VF diet reduced 8-iso-PGF2 excretion an additional 14% within 2 wk of initiating the intervention, an effect that was maintained throughout the high-VF intervention. That a high VF diet can specifically reduce an oxidative stress biomarker for lipid peroxidation is further supported by the change in urinary excretion of 8-iso-PGF2 when the low-VF diet group was switched to the high-VF diet for the final 2 wk of the experiment.
In our judgment, these findings are consistent with the interpretation that a high intake of VF, as part of a nutrient-balanced diet, does exert a modest antioxidant effect in vivo. However, the effect of the high-VF diet was much smaller than the effect observed in the subjects who transitioned from a self-selected to a balanced, prescribed diet that had a VF content similar to that reported for the US population (3.4 servings/d). Thus, we do not find it surprising that contradictory data about the in vivo effects of antioxidant-rich diets or antioxidant supplements are being reported; there are many factors that are likely to be involved in determining the levels of oxidative stress to which individuals are exposed, and, this in turn, will affect their response to ingested antioxidants.
Analysis by baseline quartiles of 8-iso-PGF2
It is reasonable to assume that some basal level of oxidation accompanies normal cellular metabolism (18, 19). This assumption led to our second working hypothesis that a uniform response to dietary antioxidant phytochemicals would not be observed, but rather that greater effects would occur in the subjects with high baseline concentrations of 8-iso-PGF2 and that little or no response to increased dietary phytochemical antioxidants would be observed in the subjects with low levels of oxidative stress at baseline. To test this hypothesis, the effects of the dietary interventions on urinary excretion of 8-iso-PGF2 were analyzed by baseline quartiles of urinary excretion of this analyte (Figure 2). Consistent with our study hypothesis, the percentage reduction in 8-iso-PGF2 attributed to the run-in diet was directly proportional to baseline concentrations of 8-iso-PGF2 excretion. Specifically, the reduction was small (6%; P = 0.05) in the lowest quartile and was progressively larger with increasing quartiles [15.1%, 29.9%, and 45.2% (P < 0.001) in the 2nd, 3rd, and 4th quartiles, respectively].
An effect of the high-VF dietary intervention on 8-iso-PGF2 excretion, above that attributed to the run-in diet, was observed only in the subjects in the highest quartile of baseline 8-iso-PGF2 excretion. In those participants, 8-iso-PGF2 excretion was reduced by an additional 20% (P = 0.01) below concentrations observed after the run-in diet, whereas the low-VF diet had no further effect. This finding implies the value of efforts directed to defining a physiologic range of lipid peroxidation that is considered "normal" compared with those levels of peroxidation that may have pathophysiologic significance. The results also indicate that the ability to detect the effects of antioxidant regimes in vivo is conditional, ie, it depends on the oxidative status of the subject at the time the treatment is initiated. If this finding is confirmed, it indicates the need for appropriate caution in the interpretation of null findings when the antioxidant activities of diets and compounds are evaluated in vivo with the use of oxidative biomarkers as the endpoints for analysis.
Summary
To our knowledge, this is the first report of a possible relation between urinary concentrations of a marker of whole-body lipid peroxidation and the ability of this measure to predict subsequent responsiveness to a dietary intervention designed to increase ingestion of phytochemical antioxidants. As such, these findings must be regarded with appropriate caution in an area rife with contradictory findings. Nonetheless, these results are consistent with the interpretation that a high intake of VF does exert modest antioxidant effects in vivo. In addition, these findings may provide valuable insights concerning the reasons for the antioxidant conundrum as reviewed in the article by Seifried et al (20). Our findings also indicate that it may be possible, via urinary analysis, to identify individuals who are likely to respond to increases in phytochemical antioxidant intake.
ACKNOWLEDGMENTS
We thank the participants for their commitment to the dietary interventions and also thank John McGinley for his assistance in the preparation of this manuscript for submission.
HJT was the principal investigator, oversaw all aspects of the project, and conducted all of the intervention sessions, study design, data evaluation, laboratory analyses, and manuscript preparation. JH supervised the formulation of dietary interventions, participant contact during the intervention, and the nutritional analyses of diets and food records. SS provided clinical supervision for the project. AH, WJ, and ZZ were responsible for processing samples and for the laboratory analyses of all blood and urine specimens. AD, CO, and BM developed and tested the recipes, formulated the study diets, prepared all participant materials, performed all participant contacts, and entered all clinical and food record data for nutritional analyses. PW was the project biostatistician and assisted with the study design, power estimation, and statistical analyses of all reported data. None of the authors had appointments to any advisory board or financial or personal interests in any organization sponsoring this research at the time the research was done.
REFERENCES