Effects of Tomato and Soy on Serum Adipokine Concentrations in Postmenopausal Women at Increased Breast Cancer Risk: A Cross-Over Dietary Intervention Trial

Abstract

Context:

Breast cancer risk among postmenopausal women increases as body mass index increases. Practical preventive methods to reduce risk of breast cancer are lacking. Few studies have investigated the effects of carotenoids and isoflavones on circulating adipokines in postmenopausal women.

Objective:

The aim was to examine the effects of lycopene- and isoflavone-rich diets on serum adipokines.

Design:

This was a 26-week, two-arm, longitudinal crossover trial.

Setting:

Participants were recruited from clinics at The Ohio State University Comprehensive Cancer Center.

Participants:

Seventy postmenopausal women at increased breast cancer risk participated in the study. The mean age and body mass index of participants was 57.2 years and 30.0 kg/m², respectively; the study was comprised of 81.4% whites.

Interventions:

The interventions included 10 weeks of consumption of a tomato-based diet (≥25 mg lycopene daily) and 10 weeks of consumption of a soy-based diet (≥40 g of soy protein daily), with a 2-week washout in between.

Main Outcome Measures:

Changes in serum adiponectin, leptin, and the adiponectin to leptin ratio were examined for each intervention through linear mixed models, with ratio estimates corresponding to postintervention adipokine concentrations relative to preintervention concentrations.

Results:

After the tomato intervention, among all women, adiponectin concentration increased (ratio 1.09, 95% confidence interval (CI) 1.00–1.18), with a stronger effect observed among nonobese women (ratio 1.13, 95% CI 1.02–1.25). After the soy intervention, adiponectin decreased overall (ratio 0.91, 95% CI 0.84–0.97), with a larger reduction observed among nonobese women (ratio 0.89, 95% CI 0.81–0.98). Overall, no significant changes in leptin or the adiponectin to leptin ratio were observed after either intervention.

Conclusions:

Increasing dietary consumption of tomato-based foods may beneficially increase serum adiponectin concentrations among postmenopausal women at increased breast cancer risk, especially those who are not obese. Additional studies are essential to confirm these effects and to elucidate the specific mechanisms that may make phytonutrients found in tomatoes practical as breast cancer chemopreventive agents.

Breast cancer risk among postmenopausal women increases as body mass index (BMI) increases (1–3). The relationship between obesity and breast cancer may be partially explained by the biological effects of the obesity-related adipokines, adiponectin and leptin, which are produced in and secreted from adipocytes including those in the breast (4). Studies have demonstrated an inverse association between circulating adiponectin and breast cancer risk overall (5–8) and particularly among postmenopausal women (6, 8). Other studies have demonstrated a positive association between circulating leptin and breast cancer risk among postmenopausal women (5, 9, 10).

Currently postmenopausal women, who are all at risk of developing breast cancer, lack practical preventive methods to reduce their risk. One potential strategy could be through dietary modifications. Studies have investigated specific dietary patterns, yet few have examined the effects of tomato- and soy-enriched diets on anthropometrics and circulating adipokine concentrations and if these dietary changes could translate into significant modalities for breast cancer prevention (11, 12).

Lycopene, the predominant carotenoid in tomatoes, is a potent antioxidant relative to other carotenoids (13). Experimental and animal studies have shown that lycopene inhibits the growth of breast cancer cells and mammary tumorigenesis (14, 15). Additionally, lycopene has been shown to suppress biomarkers of cellular proliferation, including some proinflammatory cytokines, and alter adipocyte function (16), which may be associated with postmenopausal breast cancer risk. Some epidemiological studies have indicated that higher lycopene consumption is associated with reduced cancer risk (17, 18), but data for breast cancer have been mixed. A meta-analysis (19) of 15 studies examining the association between lycopene and breast cancer demonstrated no association [relative risk (RR) 0.99, 95% confidence interval (CI) 0.93–1.06]. Although studies have found that lycopene positively modulates cytokine-driven pathways, lycopene may not be able to overcome the effects of certain stressors such as obesity (20).

Attention has also turned toward soy-based foods as chemopreventive agents. The typical Asian diet consists of fairly large amounts of soy, which is thought to be related to the lower breast cancer incidence among Asian women compared with their counterparts in the United States (21). It is hypothesized that the high content of isoflavones, particularly daidzein and genistein, contribute to the protective effects of soy. Isoflavones are polyphenolic phytonutrients capable of exerting both estrogenic and antiestrogenic effects (22), possibly contributing to increased and decreased risk of breast cancer, respectively. Results of epidemiological studies of the association between soy consumption and breast cancer have been mixed (23). A meta-analysis (23) of 14 studies examining the association between dietary isoflavone intake and breast cancer incidence demonstrated a modest reduction among high isoflavone consumers compared with those with the lowest consumption. Subgroup analysis revealed that risk reduction was significant in Asian, but not Western, women. This study also demonstrated that isoflavone intake was associated with lower breast cancer risk among postmenopausal but not premenopausal women (23). To our knowledge, no studies have examined the effect of isoflavone intake on adipokine concentrations among postmenopausal women who are at increased breast cancer risk.

Although the initial feasibility study was designed to determine whether women at increased breast cancer risk would be willing to adhere to the proposed dietary interventions, we also assessed the effects of these lycopene-rich and/or isoflavone-rich dietary modifications on circulating biomarkers of cell signaling in postmenopausal women at high risk of developing breast cancer (24). For the present study, we sought to examine the effects of these interventions on biomarkers of obesity, specifically serum adiponectin, leptin, and the adiponectin to leptin (A/L) ratio.

Materials and Methods

Study sample

As described elsewhere (24), postmenopausal women (ie, age >55 y and no menstruation for 12 months or ≤55 y, no menstruation for 12 months, and FSH concentration >30 mIU/mL) not taking hormone replacement therapy or a selective estrogen receptor modulator and at increased breast cancer risk were recruited for participation in this study between February 2003 and September 2004. Increased risk was defined as having a BMI of 25.0–42.0 kg/m² (inclusive) and/or having a first-degree relative (ie, mother, daughter, or sister) who was diagnosed with breast cancer. The study obtained institutional review board approval.

Interventions

Participants were enrolled in a 26-week, two-arm longitudinal nutrition intervention in which each woman served as her own control. The study consisted of three 2-week washout periods (at baseline, midpoint, and end of the study) and two 10-week dietary intervention periods. During each 2-week washout period, women were instructed to abstain from tomato and soy products. The tomato-based diet was implemented first because we hypothesized that it would be easier for participants to abstain from consuming soy, rather than tomato-based, products at the start of the study, thus increasing the likelihood of participants' adherence to the interventions for a longer period.

For the 10-week tomato intervention, women were instructed to consume two or more tomato products daily to equal a minimum of 25-mg servings of lycopene and to abstain from consuming soy. Women documented their daily intake of tomato products on a worksheet. To aid in the consumption of tomato products, women were provided tomato juice, tomato paste, and spaghetti sauce free of charge.

For the 10-week soy intervention, women were instructed to consume at least 40 g of soy protein daily. They were given a powdered soy product (DuPont Technologies International) with instructions to mix the powder with any liquid. Women documented their consumption by completing a daily soy calendar. During the soy intervention, women were asked to limit their consumption of tomato-based products to 5 mg of lycopene per day. Women were also given recipes to encourage consumption of the soy protein.

Given that many brands of multivitamins contain certain elements that could have affected the results of the study, women who reported taking a multivitamin at enrollment were asked to replace it with a standard vitamin supplement (Centrum; Pfizer) provided by the research staff. Women not already taking a multivitamin were not given the standard supplement.

Measures

At baseline, self-reported characteristics (eg, demographic information and alcohol/tobacco consumption) were collected; height and weight were measured by research staff. Dietary data were obtained through food frequency questionnaires at baseline and from 3-day food records at baseline as well as at the end of each intervention arm.

Biospecimen collection and laboratory analyses

Blood was collected at baseline and the end of each 10-week dietary intervention period; weight was also measured at each of these time points. Urine was collected to measure the effect of the dietary interventions on isoflavone concentrations, specifically the predominant soy isoflavones, genistein and daidzein, and their metabolic forms (dihydrogenistein and dihydrodaidzein, respectively). Adherence was assessed by data reported on the tomato worksheet, soy calendar, soy protein package counts, and blood and urine tests. Serum carotenoid concentrations (lycopene and β-carotene) served as markers of tomato consumption; urinary isoflavone concentrations served as markers of soy consumption.

Serum specimens were used to measure changes in adiponectin and leptin, which were assessed using the human adiponectin/Acrp30 quantikine and human leptin quantikine ELISA kits (R&D Systems). Samples were assayed blindly, in duplicate, random order. Each batch included replicates and blinded serum controls to assess laboratory variation. The coefficients of variation for the serum assays were 9.18% and 6.31% for leptin and adiponectin, respectively. Assay sensitivity was less than 7.8 pg/mL for leptin and 0.08 ng/mL for adiponectin. No samples were below the detection limits.

Statistical analyses

Means and frequencies were used to assess the distribution of study participants' characteristics and to categorize changes in biomarkers over time. Serum adipokines were measured before and after each 10-week intervention period. Treatment effects were estimated as the difference in pre- and postintervention biomarker concentrations after each washout period. Our analysis followed intent-to-treat principles: all available measurements from participants were included in the analysis, regardless of adherence level of the participant and whether or not she dropped out.

Serum adiponectin and leptin concentrations were not normally distributed and therefore were natural log transformed; back-transformed data (geometric means and 95% CIs) are presented for ease of interpretation. Change in adiponectin, leptin, and A/L ratio over the tomato and soy interventions were examined. We also examined the change in these measures between first study visit and the final study visit. Linear mixed models were used to test for changes over each time interval and taking into account the correlation within subjects. Results for adipokine concentrations are given as e^β and e^{β ± 1.96 (SE)}, representing the ratio of serum adipokine concentrations postintervention relative to preintervention concentrations and the corresponding 95% CI, respectively.

The models included fixed effects for time (weeks 2, 12, 14, 24, and 26) and an unstructured variance-covariance structure was assumed for the errors. The same model was used to estimate and test each intervention effect: the effect of the tomato intervention was obtained by taking the difference in the estimated means at weeks 12 and 2, and the effect of the soy intervention was obtained by taking the difference in the means at weeks 24 and 14. We also used our model to compute a combined intervention effect, which equaled the difference in week 26 and week 2 means. In secondary analyses, we extended our models to include fixed effects of baseline obesity status (BMI < 30 kg/m² vs ≥ 30 kg/m²) and obesity-by-time interactions to examine differences in adipokine change by obesity status. The Kenward-Roger method was used to calculate the degrees of freedom for our tests (25). These models were fit using PROC MIXED with a REPEATED statement in SAS (SAS Institute version 9.2). A two-sided significance level of α = .05 was used for all tests.

Results

Participant characteristics

Participants' baseline characteristics are shown in Table 1. As previously reported (24), the mean age of the 70 women enrolled in the study was 57.2 years. Most participants were white (n = 57; 81.4%) and married (n = 51; 72.9%). The average BMI was 30.0 kg/m², and almost half of the women were never-smokers (n = 33; 47.8%) and very few (n = 3; 4.4%) reported consuming two or more alcoholic beverages per day.

Table 1.

Baseline Characteristics of Study Participants (n = 70)

Characteristics	n (%)
Age, y
40–49	8 (11.4)
50–59	41 (58.6)
60–69	18 (25.7)
70–79	3 (4.3)
Race
White	57 (81.4)
Black	13 (18.6)
Marital Status
Married	51 (72.9)
Unmarried	19 (27.1)
Educationa
High school or less	8 (14.5)
Some postsecondary education	16 (29.1)
Associate's degree	4 (7.3)
Bachelor's degree	15 (27.3)
Graduate/professional degree	12 (21.8)
Annual incomea
≤$24 999	6 (10.9)
$25 000–$49 999	11 (20.0)
$50 000–$74 999	13 (23.6)
$75 000–$99 999	8 (14.5)
$100 000–$149 999	8 (14.5)
≥$150 000	9 (16.4)
BMI, kg/m2
<18.5	1 (1.4)
18.5–24.99	8 (11.4)
25.0–29.99	31 (44.3)
≥30.0	30 (42.9)
Smoking statusb
Never smoker	33 (47.8)
Former smoker	26 (37.7)
Current smoker	10 (14.5)
Alcohol consumption (average drinks per day)b
None	24 (34.8)
One or fewer	42 (60.9)
Two or more	3 (4.4)

Percentages may not sum to 100% due to rounding.

^aFifteen participants were missing data on education and annual income.

^bOne participant was missing data on smoking status and alcohol consumption.

Effect of tomato and soy interventions on serum adipokine concentrations

Although there were no significant pre- or postchanges in weight (81.86 kg vs 81.98 kg, P = .54), BMI (30.19 kg/m² vs 30.24 kg/m², P = .57), or waist (100.44 cm vs 99.65cm, P = .23) and hip (111.50 cm vs 111.91 cm, P = .17) circumferences after the tomato-based diet, significant effects on adipokine concentrations were observed. Geometric means (95% CIs) of the adipokine concentrations for the tomato arm at weeks 2 and 12, with stratification by obesity status (nonobese, BMI < 30.0 kg/m²; obese, BMI ≥ 30.0 kg/m²) are shown in Table 2. Five women were excluded from the tomato arm analyses due to insufficient serum samples for adipokine measurements. After the tomato arm, there was a 9% overall increase in serum adiponectin (ratio 1.09, 95% CI 1.00–1.18). This increase was slightly stronger among nonobese women (ratio 1.13, 95% CI 1.02–1.25). There was also a nonsignificant 11% reduction in serum leptin after the tomato arm (ratio 0.89, 95% CI 0.72–1.12), which was stronger among nonobese women (ratio 0.76, 95% CI 0.57–1.01). Additionally, there was a nonsignificant 20% increase in the A/L ratio (ratio 1.20, 95% CI 0.97–1.50), which was stronger and significant only among nonobese women (ratio 1.46, 95% CI 1.11–1.91). Analyses of adipokine change differences after the tomato arm comparing nonobese and obese women were significant for leptin and the A/L ratio (P = .04 and P = 0.01, respectively) but not for adiponectin (P = .21). As previously reported, women were adherent to the tomato-based dietary intervention (24), as evidenced by a daily consumption of 29.7 mg of lycopene and carotenoid biomarker concentrations over time, which returned to baseline by week 14.

Table 2.

Effect of 10-Week Tomato Arm of Intervention on Serum Adipokine Concentrations (n = 65)^a

Biomarker	Geometric Mean (95% CI)b		Ratio (95% CI)	PWeek 2 vs Week 12	PChange Difference by BMI
	Week 2	Week 12
Adiponectin, μg/mL
All women	12.98 (11.19–15.06)	14.12 (12.25–16.26)	1.09 (1.00–1.18)	.04	.21
BMI < 30 kg/m2	14.38 (11.78–17.55)	16.24 (13.52–19.50)	1.13 (1.02–1.25)	.02
BMI ≥ 30 kg/m2	11.50 (9.25–14.30)	11.73 (9.54–14.42)	1.02 (0.90–1.16)	.75
Leptin, ng/mL
All women	26.05 (20.70–32.62)	23.29 (17.92–30.27)	0.89 (0.72–1.12)	.32	.04
BMI < 30 kg/m2	19.45 (14.55–26.01)	14.83 (11.09–19.83)	0.76 (0.57–1.01)	.06
BMI ≥ 30 kg/m2	36.84 (26.79–50.68)	44.50 (31.49–62.90)	1.21 (0.86–1.70)	.27
A/L ratio
All women	0.50 (0.37–0.67)	0.60 (0.44–0.82)	1.20 (0.97–1.50)	.10	.01
BMI < 30 kg/m2	0.74 (0.51–1.07)	1.08 (0.77–1.51)	1.46 (1.11–1.91)	.008
BMI ≥ 30 kg/m2	0.31 (0.21–0.47)	0.27 (0.18–0.39)	0.85 (0.61–1.18)	.33

^aFive women were excluded from these analyses due to insufficient serum samples for completion of adipokine measurements.

^bAdiponectin and leptin were not normally distributed and therefore were natural log transformed for normality; back-transformed data (geometric means and 95% CIs) are presented for ease of interpretation.

Ten women were excluded from the soy arm analyses due to insufficient serum samples for the adipokine measurements. Similar to the findings from the tomato-based diet, no significant changes were observed for anthropometric measures (P > .05) after the soy-based diet. Geometric means (95% CIs) of serum adipokines for the soy intervention at weeks 14 and 24 and ratios (95% CIs) comparing pre- and postintervention biomarker concentrations, with stratification by obesity status are shown in Table 3. After the soy intervention, there was an overall 9% reduction in adiponectin (ratio 0.91, 95% CI 0.84–0.97), which was significant only among nonobese women (ratio 0.89, 95% CI 0.81–0.98). There was no change in serum leptin after the soy intervention overall (ratio 0.92, 95% CI 0.80–1.07), but among nonobese women we observed an 18% reduction (ratio 0.82, 95% CI 0.68–0.99). There was no change in the A/L ratio after the soy intervention overall or in either strata of obesity. Changes in adiponectin and A/L ratio did not differ by obesity status, but the change difference for leptin was marginally significant (P = .06). Significant increases in urinary soy isoflavones including daidzein, dihydrodaidzein, genistein, and dihydrogenistein were observed during the soy intervention (all values of P < .001), which returned to baseline levels by week 26, although women generally consumed three fourths (1.5 packets) of the recommended two packets of soy protein daily (24).

Table 3.

Effect of 10-Week Soy Arm of Intervention on Serum Adipokine Concentrations (n = 60)^a

Biomarker	Geometric Mean (95% CI)b		Ratio (95% CI)	PWeek 14 vs Week 24	PChange Difference by BMI
	Week 14	Week 24
Adiponectin, μg/mL
All women	14.38 (12.52–16.52)	13.04 (11.21–15.17)	0.91 (0.84–0.97)	.008	.51
BMI < 30 kg/m2	17.04 (14.31–20.29)	15.18 (12.49–18.46)	0.89 (0.81–0.98)	.01
BMI ≥ 30 kg/m2	11.42 (9.37–13.93)	10.68 (8.57–13.30)	0.93 (0.84–1.05)	.23
Leptin, ng/mL
All women	25.65 (20.89–31.48)	23.63 (18.67–29.90)	0.92 (0.80–1.07)	.27	.06
BMI < 30 kg/m2	18.32 (14.48–23.18)	15.03 (11.65–19.38)	0.82 (0.68–0.99)	.04
BMI ≥ 30 kg/m2	40.31 (30.51–53.27)	43.81 (32.78–58.56)	1.09 (0.86–1.37)	.47
A/L ratio
All women	0.56 (0.42–0.73)	0.56 (0.41–0.75)	1.00 (0.85–1.17)	>.99	.18
BMI < 30 kg/m2	0.92 (0.68–1.25)	1.01 (0.73–1.40)	1.09 (0.89–1.34)	.39
BMI ≥ 30 kg/m2	0.28 (0.20–0.40)	0.25 (0.17–0.36)	0.88 (0.68–1.13)	.31

^aTen women were excluded from these analyses due to insufficient serum samples for completion of adipokine measurements.

^bAdiponectin and leptin were not normally distributed and therefore were natural log transformed for normality; back-transformed data (geometric means and 95% CIs) are presented for ease of interpretation.

Table 4 contains geometric means (95% CIs) of serum adipokines at weeks 2 and 26, which demonstrates the combined effects of the two interventions. Five women were excluded from these analyses due to insufficient serum samples for adipokine measurements. Overall, there were no differences in anthropometrics, serum adiponectin or leptin at weeks 2 and 26. However, among nonobese women, there was a 21% reduction in serum leptin (ratio 0.79, 95% CI 0.63–0.99) and a 34% increase in the A/L ratio (ratio 1.34, 95% CI 1.04–1.74); these changes were different from those exhibited by obese women (P = .03 for leptin and P = .02 for A/L ratio), whose adipokines were similar at weeks 2 and 26.

Table 4.

Effect of the Combination of Tomato and Soy Dietary Interventions on Serum Adipokine Concentrations (n = 65)^a

Biomarker	Geometric Mean (95% CI)b		Ratio (95% CI)	PWeek 2 vs Week 26	PChange Difference by BMI
	Week 2	Week 26
Adiponectin, μg/mL
All women	12.98 (11.19–15.06)	13.30 (11.58–15.27)	1.02 (0.94–1.12)	.58	.20
BMI < 30 kg/m2	14.38 (11.78–17.55)	15.38 (12.88–18.38)	1.07 (0.95–1.20)	.24
BMI ≥ 30 kg/m2	11.50 (9.25–14.30)	10.98 (8.99–13.41)	0.95 (0.84–1.09)	.49
Leptin, ng/mL
All women	26.05 (20.70–32.62)	23.70 (18.86–29.78)	0.91 (0.77–1.08)	.28	.03
BMI < 30 kg/m2	19.45 (14.55–26.01)	15.40 (12.01–19.76)	0.79 (0.63–0.99)	.04
BMI ≥ 30 kg/m2	36.84 (26.49–50.68)	42.63 (32.05–56.71)	1.16 (0.89–1.50)	.25
A/L ratio
All women	0.50 (0.37–0.67)	0.56 (0.42–0.74)	1.12 (0.91–1.37)	.28	.02
BMI < 30 kg/m2	0.74 (0.51–1.07)	0.99 (0.74–1.34)	1.34 (1.04–1.74)	.03
BMI ≥ 30 kg/m2	0.31 (0.21–0.47)	0.26 (0.18–0.36)	0.83 (0.61–1.12)	.21

^aFive women were excluded from these analyses due to insufficient serum samples for completion of adipokine measurements.

^bAdiponectin and leptin were not normally distributed and therefore were natural log transformed for normality; back-transformed data (geometric means and 95% CIs) are presented for ease of interpretation.

Discussion

We hypothesized that dietary interventions that beneficially modify circulating adipokines (ie, increase serum adiponectin and reduce leptin) could be a practical approach for prevention of obesity-related breast cancer among high-risk postmenopausal women. Thus, we assessed the effects of tomato- and soy-rich dietary interventions on adiponectin, leptin, and the A/L ratio. Our findings demonstrate that serum adiponectin increased after the consumption of the lycopene-rich diet and decreased after the consumption of the isoflavone-rich diet. Interestingly, the individual effects of these interventions on adiponectin concentration were stronger among nonobese women. Additionally, there was a reduction in serum leptin after the soy intervention, which was significant only among nonobese women.

Several dietary intervention trials have evaluated the effects of either carotenoids or isoflavones on adiponectin and/or leptin concentrations in women (26–32), but few focused on postmenopausal women (26, 27, 29–31). Although studies have examined effects of carotenoids in general, ours is the first to examine the effects of a lycopene-rich diet among postmenopausal women. In our study, women consumed roughly 30 mg/d of lycopene during the 10-week tomato arm, which is 6–15 times the average intake in the United States (33). This increased consumption translated into a 9% increase in serum adiponectin, although participants did not lose weight. Another study that examined the effect of carotenoids on adiponectin among postmenopausal women (26) reported a 12% increase in plasma high molecular weight adiponectin after a 3-week intervention of β-cryptoxanthin among moderately obese Japanese women. Interestingly, in subgroup analyses, we found a 13% increase in total adiponectin among nonobese women, yet there was no significant change among obese women. Adiponectin plays a role in regulating glucose homeostasis and fatty acid metabolism. Thus, the increase in adiponectin observed among nonobese women during the tomato intervention suggests that lycopene may effectively improve insulin sensitivity in this subgroup, which could reduce their breast cancer risk (34, 35). Furthermore, this improvement in insulin sensitivity could lead to changes in adiponectin isoform distribution (eg, low, medium, or high molecular weight) (36), which may also be associated with breast cancer.

Our finding that the tomato arm increased the A/L ratio by 46% among nonobese women also supports the hypothesis that high lycopene consumption may be protective against breast cancer. This is consistent with the inverse association between serum A/L ratio and breast cancer risk overall and tumor aggressiveness observed by Chen et al (5). Lack of change in adiponectin or the A/L ratio after the tomato arm among obese women may indicate that a longer duration of tomato consumption is necessary to modify these biomarkers. Further studies involving longer-term lycopene interventions among postmenopausal women are necessary to clarify its potential as a breast cancer chemopreventive agent.

The soy arm resulted in a 9% decrease in adiponectin overall and an 18% decrease in leptin among nonobese women. Four studies have also examined the effect of dietary isoflavones on adipokines in postmenopausal women (27, 29–31), yielding varied results. Riesco et al (27) found that a 6-month mixed-training exercise intervention with or without the addition of 17.5 mg/d of isoflavones resulted in decreased leptin and no significant change in adiponectin among overweight to morbidly obese (BMI 28–40 kg/m²) women. In a smaller crossover study, Phipps et al (29) found that higher consumption (approximately 130 mg/d of isoflavones) for approximately 3 months had no significant effect on leptin in postmenopausal women. In a 2-month trial, Wu et al (31) showed that consumption of 50 mg/d of soy decreased leptin by 39.5%. Charles et al (30) showed that, after a 12-week randomized dietary soy intervention (160 mg isoflavones per day), there was a small increase in plasma adiponectin (∼2 μg/mL, P = .03) and no significant change in leptin concentration. Although we noted a decrease in adiponectin, Charles et al (30) reported a small increase in adiponectin after a 12-week soy intervention. Our leptin data were similar to the findings of Phipps et al (29), which showed no effect of isoflavones on leptin overall but differed from the findings of Wu et al (31), showing a substantial decrease in leptin after a high consumption of isoflavones. This difference could be due to differences in baseline urinary isoflavones [higher baseline concentrations in our study, >10 μmmol/L vs 1–2 μmmol/L in the study by Wu et al (31)] as well as the response to the soy interventions [significant weight loss after the soy intervention in the study by Wu et al (31), whereas none was observed in our study].

The decrease in leptin after the soy intervention supports its potential chemopreventive effect; however, this effect may be limited to nonobese women. Leptin regulates body fat storage and mediates long-term regulation of energy. It may be that isoflavones more readily modify leptin when concentrations are already low (ie, among nonobese women). Furthermore, this effect may be observable in nonobese women after a relatively short duration of soy consumption through the antiinflammatory effects of the isoflavones (37, 38), independent of weight loss. Moderate weight loss, particularly among obese adults, results in improvements in leptin concentration, thereby improving leptin receptor function and energy intake regulation (39). Although participants did not lose weight, the increased isoflavone intake may have contributed to lower leptin. Our previous study (40) found no correlation between plasma and breast tissue leptin. Thus, it is important to investigate the effects of isoflavone intake on breast tissues to clarify its potential for chemoprevention.

There are no clear explanations for the observed decrease in adiponectin after the soy intervention. It may be that the protective effects of soy against breast cancer are limited to certain subgroups (eg, Asian women, whose soy consumption typically begins in utero and who have relatively lower BMI). It is also possible that the mechanism(s) of protection does not involve adiponectin. Further studies are required to clarify these mechanisms.

There were many strengths of this study. First, this is the only prospective study to examine the effects of lycopene on adipokines and one of a few to examine the effects of isoflavones in postmenopausal women. Second, the randomized crossover design with appropriate washout periods provided valid results that could not be obtained through case-control studies of the association between these dietary patterns and adipokine concentrations. The use of highly reproducible immunoassays to assess our primary outcomes also strengthened our study. Additionally, the adherence to both interventions was confirmed through biomarker measures. Finally, our sample size was larger or comparable with other studies of carotenoid and/or isoflavone interventions (26, 27, 29–31), and there were relatively few losses to follow-up.

This study also had some limitations that should be considered in the interpretation of our results. First, the nonrandom allocation of the dietary interventions, with all participants completing the lycopene-rich intervention followed by the isoflavone-rich intervention, may have affected our results. However, the 2-week washout period between the interventions, resulting in participants' blood carotenoid concentrations returning to baseline levels, minimizes the concern of carryover bias; however, the washout period may not adequately account for other temporal effects (eg, environmental changes, effects of being in a research study for a prolonged period). Second, findings are generalizable only to postmenopausal women with increased breast cancer risk. Third, dietary information was ascertained only for tomato and soy consumption. Therefore, we cannot be certain that our findings are the result of these diets alone. Related to this, although tomatoes are high in nutrients, vitamins, minerals, and other bioactive phytochemicals, our focus on lycopene could be a limitation. Further study of the effects of additional compounds found in tomato-based foods would therefore be of great interest. And finally, our measurement of only total adiponectin, rather than that of other isoforms, which could have been affected by the dietary interventions, was also a limitation.

In summary, the present study suggests that increasing consumption of lycopene and isoflavones may have beneficial effects on adipokines (ie, increasing adiponectin and decreasing leptin, respectively). Lycopene consumption increased adiponectin overall, with a stronger effect among nonobese women, whereas isoflavone consumption reduced leptin among nonobese women. Further studies are essential to elucidate longer-term effects of dietary lycopene and isoflavones, separately as well as in combination with exercise interventions, and their practicality in breast cancer chemoprevention in postmenopausal women, particularly those at higher risk.