Impact of 18-month soy protein supplementation on steroid hormones and serum biomarkers of angiogenesis, apoptosis, and the growth hormone/IGF-1 axis: results of a randomized, placebo-controlled trial in males following prostatectomy

Abstract

Many studies have addressed the effects of dietary supplementation with soy protein on cancer risk and mortality, but there are only few randomized studies with soy in males. We used serum samples from a two year trial of soy protein isolate supplementation in middle aged to older males at risk of recurrence of prostate cancer after radical prostatectomy to determine soy effects on steroid hormones involved in prostate cancer (testosterone, SHBG, estradiol) and explore the effects on biomarkers of the growth hormone/IGF-1 axis, apoptosis, and angiogenesis. Compared with a casein-based placebo, 18 months of consumption of 19.2 g/day of whole soy protein isolate containing 24 mg genistein reduced circulating testosterone and SHBG, but not free testosterone, and did not affect serum concentrations of estradiol, VEGF, IGF-1, IGFBP-3, IGF-1/IGFBP-3 ratio, soluble Fas, Fas-ligand, and sFas/Fas-ligand ratio. Thus, soy protein supplementation for 18 months affected the androgen axis, but the effects on other cancer biomarkers remain to be more definitively determined. The study was registered at clinicaltrials.gov (NCT00765479).

Introduction

Soy consumption has been found to be associated with reduced mortality of cancer in one meta-analysis of epidemiological studies, particularly for female breast cancer (1). However, another study found no such association (2). Other meta-analyses have shown that consumption of soy and/or phytoestrogens, including soy isoflavones, reduced the risk of developing certain cancers, including prostate (3–7), breast (8), endometrial (9,10), and gastric cancer (11,12), as well as colon and colorectal cancer (11,13), although the evidence for the latter malignancy is inconsistent (14). However, these meta-analyses are of uneven quality (15).

Randomized clinical trials of the effect of soy consumption on cancer risk and mortality are needed, but only a few have been reported. Miyanaga et al. (16) randomized men without high grade prostatic intraepithelial neoplasia (HGPIN) at biopsy to isolated soy isoflavones in tablets and found a decreased incidence of study biopsy-detected prostate cancer after 12 months compared to men on placebo tablets, but only in subjects ≥ 65 years old and only in men who did not produce equol. There was no difference in HGPIN incidence between the two groups in this study. It should be noted that the sample sizes in this study were quite small and the duration of intervention was relatively short. In another study, Quaas et al. (17) randomized postmenopausal women to soy protein isolate or milk protein for three years and found no difference between the study arms in the incidence of endometrial hyperplasia and cancer at end-of-study biopsy. Previously, we found no effect of soy protein isolate supplementation for two years on prostate cancer recurrence following radical prostatectomy compared to a milk protein placebo (18). Although there are no randomized trials of soy with breast cancer as endpoint, soy and soy isoflavones did not affect mammographic breast density, an established breast cancer risk factor, in randomized studies (19–23) and did not affect breast epithelial cell proliferation obtained by fine-needle aspiration (24) or biopsy (25), suggesting a lack of protective effect. Some studies with cancer as endpoint combined soy or isoflavones with other agents precluding evaluation of a specific soy or isoflavone effect on prostate cancer risk (26) or cancer in women (27).

Many clinical studies have been devoted to the identification of soy effects on various biomarkers that are associated with cancer and cancer risk, such as prostate-specific antigen (PSA) and prostate cancer (5), but these did not have a cancer endpoint. Yet, there is substantial evidence from laboratory studies that soy isoflavones can affect essential components of cancer development, including inhibition of cell proliferation and apoptosis and stimulation of angiogenesis, as well as hormonal mechanisms involved in breast and prostate cancer (28–30). We conducted the above mentioned two-year randomized trial with soy protein isolate compared to a milk protein placebo (18) to assess soy effects on risk of recurrence following radical prostatectomy in middle aged-older subjects. Surplus serum samples from subjects that were cancer-free were available for the present study. We used these samples to determine soy effects on steroid hormones involved in prostate cancer (testosterone, sex hormone-binding globulin (SHBG), estradiol) and explored the effects on the following biomarkers of cancer development: cell growth – insulin-like growth factor-1 (IGF-1) and its binding protein IGFBP-3 (31–33); apoptosis – soluble Fas and Fas Ligand (34–36); and angiogenesis – vascular endothelial growth factor (VEGF) and basic fibroblast growth factor (bFGF) (37,38).

Subjects and Methods

Study Design

This study used stored serum samples from a double-blind, randomized, casein-based placebo-controlled intervention trial with soy protein isolate (registered at clinicaltrials.gov; identifier: NCT00765479) previously reported in detail (18). Radical prostatectomy patients who were at elevated risk of recurrence because of unfavorable pathology were enrolled between July 1997 and November 2005 at the Manhattan VA Medical Center (VA) or May 2009 at New York University School of Medicine (NYU) as previously detailed (18). All participants from whom we used samples for the present study were enrolled at NYU, except one who was enrolled at the VA. A total of 159 eligible males were successfully enrolled, randomized to soy protein (n = 81) or placebo (n = 78), and were ultimately evaluable for the previously reported study (18). Dietary supplement use, customary soy consumption, and allergies to milk protein or soy were exclusionary criteria (18).

The intervention agent was a soy protein isolate-based beverage powder and the placebo was a similar caseinate-based product, produced for this clinical trial by Solae LLC. Subjects were instructed to daily consume a serving of beverage powder (47 g) containing either soy protein isolate (19.2 g as analyzed) or calcium caseinate (19.8 g). The beverage powders were sweetened with a mixture of sucrose and fructose to improve palatability and artificial strawberry flavoring was added to mask the taste difference between the two powders. The soy protein isolate powder contained per serving 70.5 mg of all forms of isoflavones and in aglycone equivalents, 41 mg total isoflavones, 23.8 mg of genistein, and 15.0 mg daidzein as previously detailed along with the nutrient composition of the powders (18). Subjects were instructed to incorporate the beverage consumption in their daily routine without changing their dietary habits otherwise.

Subjects were asked about their medical history and use of medications at baseline and at follow-up visits every 2 months in year one of the study and every three months in year two. At each study visit, blood samples were taken. Measurements for this study were conducted on samples collected at baseline, 2, 4, 8, and 18 months.

The randomized subjects from whom we used samples for the present study were selected on the basis of whether samples were available at all time points of interest and the time during the study at which samples were collected in order to limit the time period over which samples were stored. We did not attempt to match subjects on soy and placebo because of the limited number of subjects from whom samples were available. We included only samples from subjects who were compliant. Adherence was assessed by the number of beverage powder packets consumed/the number of packets supplied for self-reported compliance (18) and by measuring serum concentrations of genistein in subjects of the soy arm at various time points (data not shown).

Serum Assays

The samples used in this study were collected between September 2003 and July 2007 and aliquots were stored at −80°C from 2003-early 2007 and in liquid nitrogen vapor after that time. The albumin assays were carried out on fresh samples at the time they were collected, while all other assays were carried out between March 2008 and September 2009.

Serum concentrations of testosterone, SHBG, and estradiol were measured using automated immunoenzymometric assays (Tosoh AIA-600, Tosoh Bioscience, San Francisco, CA). We determined that for testosterone and estradiol the limit of quantitation was 10 ng/dL and 25 pg/mL, respectively with a within-run CV of 3.3% and 6.1%, respectively, and a between-run CV of 7.4% and 9.1%, respectively. For SHBG, Tosoh documented a limit of quantitation of 0.2 nmol/L with a within-run CV of 2.6% and a between-run CV of 2.2%. Testosterone measurements were assayed in duplicates. Serum albumin concentrations at baseline and at 12 and 24 months were determined by the NYU Tisch Hospital Clinical Pathology Laboratory using their routine assay. We used the mean of these three albumin measurements to calculate free testosterone concentrations at each time point from testosterone and SHBG concentrations using a calculator developed at the Hormonology Department, University Hospital of Ghent, Belgium (http://www.issam.ch/freetesto.htm).

To measure serum concentrations of markers for angiogenesis, apoptosis, and the growth hormone/IGF-1 axis, we used Quantikine ELISA assay kits from R&D Systems (Minneapolis, MN), using a BioTek spectrophotometer plate reader (Winooski, VT) for the following markers: human bFGF (cat. no. DFB50; sensitivity: 3 pg/ml; intra-assay CV 9.7%; inter-assay CV 9.1%); human VEGF (cat. no. DVE00; sensitivity: 9 pg/ml; intra-assay CV 6.4%; inter-assay CV 8.8%); human soluble Fas (cat. no. DSF00; sensitivity: 20 pg/ml; intra-assay CV 4.6%; inter-assay CV 2.9%); human Fas ligand (cat. no. DFL00B; sensitivity: 5.1 pg/ml; intra-assay CV 3.6%; inter-assay CV 4.5%); IGF-1 (cat. no. DG100B; sensitivity: 0.022 ng/ml; intra-assay CV 4.0%; inter-assay CV 5.7%); and IGFBP-3 (cat. no. DGB300; sensitivity: 0.14 ng/ml; intra-assay CV 2.3%; inter-assay CV 2.3%). The precision information shown was provided by R& D Systems. Samples were diluted 1:8 for Fas, 1:32 for IGF-1, and 1:64 for IGFBP-3 before analysis and all ELISA assays were run in duplicate.

Statistical Methods

We computed changes in biomarker values from baseline to various time points (based on the measurement time line for each marker), examined changes from baseline to the various time points at which this marker was measured, and compared these changes between the two treatment groups. These analyses were conducted within each treatment arm using repeated measures one way analysis of variance (ANOVA) with Bonferroni correction for multiple comparisons and a test for linear trend. Differences between the two treatment groups (soy versus placebo) and interactions between treatment and time were analyzed using two-way repeated measures ANOVA. A log transformation of all biomarker data was applied prior to these analyses. For comparisons between the two treatment groups at baseline we used a t-test or Mann-Whitney test. MedCalc software (Ostend, Belgium) was used for the one way repeated measures ANOVA and Prizm software (GraphPad, San Diego, CA) was used for repeated measures two-way ANOVA and t- and Mann-Whitney tests. P values were 2-sided. All data are presented as means and 95% CI unless indicated otherwise.

Ethics

The study was approved by the Institutional Review Boards of New York University School of Medicine and the University of Illinois at Chicago and registered on October 3, 2008 at clinicaltrials.gov (NCT00765479).

Results

Baseline Characteristics

As detailed in Table 1, we used five sets of samples for the measurements on which we report here from subjects who were between the ages of 47 and 74 at baseline and were overwhelmingly Caucasian. The subjects from whom we used samples to measure serum concentrations of VEGF were different from those from whom we used samples to measure serum concentrations of sFas, Fas Ligand, IGF-1 and IGFBP-3, the concentrations of which were all measured in the same samples. The samples used to measure testosterone and estradiol were largely the same. In a subset of the testosterone samples SHBG was measured and the results were used to calculate free testosterone. There were no differences in mean age between subjects on placebo and subjects on soy protein, but subjects on placebo were heavier at baseline than those on soy protein (Table 1).

TABLE 1.

Demographic and Baseline Data of Subjects in Tables 2 and 3

	Subjects in Table 2				Subjects in Table 3
	Testosterone		SHBG & Free Testosterone		Estradiol		VEGF		Fas/Fas Ligand & IGF-1 - IGFBP-3
	Soy	Placebo	Soy	Placebo	Soy	Placebo	Soy	Placebo	Soy	Placebo
Number of subjects	19	23	10	14	16	17	9	9	7	8
Age, years (mean) 95% CI	58.3 (55.7, 61.0)	61.0 (57.7, 64.3)	58.7 (54.4, 63.1)	61.1 (56.8, 65.5)	57.9 (55.1, 60.7)	60.1 (56.0, 64.1)	60.2 (54.9, 65.5)	61.8 (56.2, 67.4)	57.9 (52.4, 63.3)	59.3 (52.7, 65.8)
Body Weight, kg (mean) 95% CI	86.9 (79.9, 93.9)	93.4 (85.9, 100.9)	88.3 (78.6, 98.0)	97.4 (87.3, 108.1)	86.5 (78.7, 94.2)	93.5 (85.6, 101.4)	89.7 (82.0, 97.5)	92.3 (83.8, 100.9)	87.9 c (76.0, 100.0)	101.2 c (89.7, 112.6)
BMI (mean) na 95% CI	29.0 15 (24.7, 29.2)	28.6 19 (26.3, 31.0)	27.4 7 (24.6, 30.3)	30.3 11 (26.8, 33.8)	26.7 13 (24.1, 29.3)	28.5 13 (26.0, 31.1)	26.6 8 (22.5, 30.7)	26.2 8 (24.5, 27.9)	26.9 d 5 (22.4, 31.4)	32.1 d 5 (28.4, 35.9)
Raceb
White	18	22	10	13	15	17	9	8	7	8
African American	1	1	0	1	1	0	0	1	0	0

^aBMI could not be calculated for some subjects because of missing height data.

^bSelf-reported.

^cP = 0.077 for difference between the soy and placebo groups (2-sided t-test).

^dP = 0.039 for difference between the soy and placebo groups (2-sided t-test).

Effects on Steroid Hormones and SHBG

Serum concentrations of testosterone and SHBG were reduced in the soy supplementation group, but not the placebo over the 18 months of observation (Table 2). There was an interaction between time and treatment (soy vs. placebo) in two-way repeated measure analysis of variance which made the P values for difference between the two groups (0.494 for soy and 0.064 for placebo) unreliable. Calculated free testosterone concentrations were not affected by soy supplementation and were not different between the two groups (Table 2). The mean albumin concentrations used to calculate free testosterone were not different between the soy and placebo groups (P = 0.234; 2-sided t-test - Table 2) and albumin concentrations did not change from baseline at 12 and 24 months in either group (data not shown). Estradiol concentrations were not affected by soy supplementation and were not different between the two groups, while the testosterone:estradiol ratio was unsurprisingly reduced (Table 2). There was an interaction between time and treatment (soy vs. placebo) in two-way repeated measure analysis of variance of this ratio which made the P value for the difference between the two groups (0.139) unreliable. Body weight increased slightly over 18 months by 0.7-1.2% in the soy sub-groups and by 0.8-1.9% in the placebo sub-groups without a difference between the soy and placebo groups (P = 0.259-0.983; 2-sided t-tests; data not shown).

TABLE 2.

Steroid Hormone and SHBG Serum Concentrations at Baseline and Change from Baseline at 2 - 18 Months on Study ^a

Parameter	Group	Baseline	Δ, %				ANOVA b P	Trend b P	2-Way ANOVA c
			2 months	4 months	8 months	18 months			Difference P	Interaction P
Testosterome	Soy (n = 18)	441.5 ng/dL (352.6, 530.4)	−3.2 (−14.0, 7.6)	−6.7 (−18.8, 5.5)	−8.6 (−20.6, 3.4)	−13.8 (−23.4, −4.1)	0.045	0.004	0.494	0.013
	Placebo (n = 21)	350.5 ng/dL (296.4, 404.6)	8.7 (−4.2, 21.6)	6.0 (−6.3, 18.4)	4.7 (−8.5, 18.0)	12.8 (−1.9, 27.6)	0.299	0.202
SHBG	Soy (n = 10)	35.9 nmol/L (22.5, 49.3)	−6.5 (−17.0, 3.9)	−12.3 (−27.6, 3.0)	−19.6 (−29.8, −9.4)	−21.4 (−29.0, −13.4)	0.001	<0.001	0.064	0.020
	Placebo d (n = 13)	22.1 nmol/L (15.7, 28.5)	−6.8 (−21.2, 7.5)	−10.8 (−24.0, 2.3)	−6.1 (−22.1, 9.9)	7.7 (−13. 2, 28.6)	0.115	0.811
Albumin	Soy (n = 10)	4.46 g/dL (average of baseline, 12, and 24 months) (4.30, 4.63)
	Placebo (n = 13)	4.34 nmol/L (average of baseline, 12, and :24 months) (4.12, 4.56)
Free Testosterone (calculated)	Soy (n = 10)	8.00 ng/dL (6.72, 9.21)	−4.01 (−22.83, 14.81)	2.55 (−21.78, 26.87)	−2.87 (−25.35, 19.61)	−5.92 (−19.29, 7.45)	0.818	0.348	0.318	0.793
	Placebo (n = 13)	8.36 ng/dL (6.81, 9.91)	7.00 (−11.66, 25.65)	10.85 (−9.40, 31.11)	9.75 (−13.34, 32.84)	8.83 (−12.15, 29.82)	0.903	0.569
Estradiol	Soy (n = 11)	37.7 ng/mL (32.5, 43.0)	6.5 (−10.4, 23.3)	−3.2 (−13.6, 7.2)	12.1 (−10.5, 34.7)	12.5 (−8.3, 33.3)	0.356	0.305	0.106	0.161
	Placebo (n = 11)	38.5 ng/mL (35.0, 41.9)	−8.8 (−18.7, 1.2)	−3.3 (−19.1, 12.6)	−5.7 (−19.9, 8.6)	2.0 (−13.5, 17.6)	0.611	0.899
Ratio Testosteron / Estradiol	Soyee (n = 10)	8.9 (7.0, 10.7)	−1.6 (−23.1, 19.8)	−4.4 (−17.7, 8.9)	−11.6 (−35.9, 12.8)	−26.2 (−37.0, −15.5)	0.008	0.001	0.139	0.006
	Placebo (n = 11)	9.1 (6.9, 11.4)	25.8 (5.8, 45.7)	6.7 (−11.7, 25.0)	6.4 (−12.7, 25.5)	21.5 (−12.3, 55.2)	0.092	0.713

^aAll data were log transformed for analysis and are presented as means (95% CI).

^bOne-way repeated measurement ANOVA.

^cTwo-way repeated measurement ANOVA for difference between the two groups and for interaction between treatment and time.

^dOne outlier identified using Grubb’s test was removed from the analysis.

Effects on Biomarkers of Angiogenesis, Apoptosis, and the Growth Hormone/IGF-1 Axis

Serum concentrations of VEGF were not affected by the soy protein supplementation and remained stable over 18 months, as did VEGF concentrations in the placebo group (Table 3). There was no difference between the two groups at baseline, while there was a difference in change over time between the two groups, but no interaction between treatment and time (Table 3). We also attempted to measure serum concentrations of bFGF, but these appeared to be extremely variable within subjects and between time points and groups, impeding interpretation (data not shown). Body weight decreased slightly over 18 months by 0.2% in the soy sub-group and by 4.2% in the placebo sub-group (P = 0.076; 2-sided t-test; data not shown).

TABLE 3.

Serum Concentrations of VEGF, Fas, Fas-Ligand, IGF-1, and IGFBP-3 at Baseline and Change from Baseline at 2-18 Months ^a

Parameter	Group	Baseline	Δ, %				ANOVAb P	Trendb P	2-Way ANOVA c
			2 months	4 months	8 months	18 months			Difference P	Interaction P
VEGF	Soy d (n = 7)	38.7 pg/mL (24.8, 52.5)	0.8 (−18.7, 20.4)	7.0 d (−27.0, 41.0)	−5.2 (−26. 7, 16.2)	−2.0 (−29.4, 25.3)	0.986	0.822	0.022	0.185
	Placebo (n = 9)	46.4 pg/mL (36.0, 56.7)	16.0 (−7.4, 39.3)	−8.8 (−25.8, 8.2)	8.4 d (−13.8, 30.5)	−11.0 (−32.1, 10.1)	0.058	0.435
Soluble Fas	Soy (n = 7)	7.77 ng/mL (6.11, 9.39)	3.5 (−10.2, 17.2)	−4.1 (−13.3, 5.2)	2.8 (−7.7, 13.3)	−3. 3 (−11.6, 5.0)	0.314	0.975	0.434	0.211
	Placebo (n = 8)	7.28 ng/mL (5.47, 9.09)	6.6 (−15. 4, 28.7)	11.0 (−7.5, 29.5)	11.3 (−72.5, 29.7)	8.5 (−14.1, 31.2)	0.595	0.207
Fas-ligand	Soy (n = 7)	65.0 pg/dL (40.4, 89.6)	10.2 (5.0, 15.4)	10.0 (−5.7, 25.8)	7.2 (−6.8, 21.3)	11.4 (−10.7, 33.5)	0.389	0.372	0.074	0.481
	Placebo (n = 8)	85.5 pg/dL (65.6, 103.3)	5.0 (−12.3, 22.2)	21.9 (−1.7, 45.5)	10.7 (−10.7, 32.1)	10.3 (−8.5, 29.0)	0.944	0.874
Ratio Fas/ Fas-ligand	Soy (n = 7)	145.1 (64.3, 225.89)	−6.1 (−17.5, 5.3)	−11.2 (−26.1, 3.8)	−3.2 (−13.9, 7.6)	−10.8 (−28.4, 6.8)	0.117	0.392	0.082	0.724
	Placebo (n = 8)	90.9 (58.3, 123.6)	3.4 (−19.4, 26.2)	−5.9 (−32.6, 20.7)	5.0 (−20.6, 30.5)	1.0 (−26.6, 28.6)	0.717	0.641
IGF-1	Soy (n = 7)	93.6 ng/mL (72.8, 114.5)	14.1 (−12.7, 40.9)	−2.1 (−21.7, 17.6)	6.5 (−10.1, 23.0)	2.8 (−19.9, 25.5)	0.596	0.243	0.614	0.509
	Placebo (n = 7)	82.6 ng/mL (48.2, 117.0)	12.7 (−0.9, 26.4)	19.6 (−1.0, 40.1)	14.7 (2.7, 26.7)	22.8 (−4.8, 50.4)	0.057	0.660
IGFBP-3	Soy (n = 7)	1.93 μg/mL (1.36, 2.49)	−0.7 (−13.5, 12.1)	−7.7 (−28.1, 12.6)	5.2 (−10.3, 20.7)	−2.4 (−10.4, 5.7)	0.287	0.470	0.961	0.609
	Placebo (n = 8)	2.06 μg/mL (1.64, 2.49)	13.7 (0.7, 26.7)	−4.6 (−39.5, 30.3)	7.4 (−15.0, 29.8)	0.1 (−26.3, 26.4)	0.706	0.742
Ratio IGF-1/ IGFBP-3	Soy (n = 7)	0.053 (0.037, 0.069)	15.6 (−13.0, 44.3)	15.8 (−45.0, 76.5)	2.2 (−14.0, 18.4)	5.1 (−14.1, 24.3)	0.713	0.208	0.807	0.994
	Placebo (n = 8)	0.041 (P=0.219) (0.29, 0.090)	2.2 (−10.8, 15.3)	15.2 e (−16.4, 46.8)	15.4 (−27.4, 58.2)	14.3 (−18.1, 46.7)	0.398	0.087

^aAll data were log transformed for analysis and are presented as means (95% CI).

^bOne-way repeated measurement ANOVA.

^cTwo-way repeated measurement ANOVA for difference between the two groups and for interaction between treatment and time.

^dOne subject was removed from the analysis at this time-point because of one extreme outlying value.

^eTwo subjects were removed from the analysis at this time-point, one because of outlying values at all time points identified using Grubb’s test, and one because of one extreme outlying value.

Serum concentrations of soluble Fas and Fas-ligand and their ratio and concentrations of IGF-1 and IFGBP-3 and their ratio were not different between the two groups at baseline and remained stable over 18 months without an interaction between treatment and time (Table 3). Body weight increased slightly over 18 months by 1.8% in the soy sub-group and by 1.5% in the placebo sub-group (P = 0.852; 2-sided t-test; data not shown).

Discussion

In this randomized intervention study with ~20 g/ day soy protein isolate providing ~70 mg/day total isoflavones and ~24 mg/day genistein versus a caseine-based placebo to males aged 47-74 years for up to 18 months, we found a progressive reduction in circulating testosterone and SHBG in subjects in the soy group, leaving free testosterone concentrations unchanged. By contrast, serum concentrations of estradiol and biomarkers of angiogenesis, apoptosis, and the growth hormone/IGF-1 axis were not affected by soy protein supplementation, but the small sample sizes and variation in some biomarker data preclude definitive conclusions.

Most randomized clinical trials with soy protein have not found effects on circulating testosterone and SHBG (16,39–46) and two meta-analyses confirmed this (5,47). All these studies had a duration between 3 weeks and 3 months. Only three studies reported the same reduction in testosterone and/or SHBG in men on soy that we observed in our 18-month study. MacKey et al. (48) found a reduction in SHBG, but not testosterone, in an 18-week study with soy, regardless of the isoflavone content. Hamilton-Reeves et al. (49) conducted a 6-month study with soy protein isolate at a dose twice that of the current study in men with prostate cancer or preneoplastic prostate lesions identified at biopsy. They also found a reduction in SHBG, but not testosterone or free testosterone; this occurred not only in the two groups on soy, one with isoflavones removed, but also in a milk protein placebo group. By contrast, we found no effect on SHBG (or testosterone) in our casein placebo group. Hamilton-Reeves et al. also found no effect on estradiol concentrations in the groups on milk protein and soy protein containing isoflavones, but observed an elevation in the groups on soy devoid of isoflavones. Pendleton et al. (50) reported a reduction is testosterone and free testosterone in a small one-year study with soy milk in men with rising PSA after prostatectomy and reported a reduction in serum testosterone and free testosterone. However, statistical analysis of the raw data presented in that paper does not substantiate that finding and indicates no change from baseline. In clinical studies with isolated isoflavones, no effects were found on circulating levels of testosterone or SHBG (51–53).

In the present study, the decrease in circulating testosterone and SHBG did not become substantial until 4 to 8 months into the study (Table 2). Thus, it is conceivable that the testosterone and SHBG reduction in males consuming soy does not take effect until after 4 months, which would be consistent with some of the findings of MacKey et al. (48) and Hamilton-Reeves et al. (49) in 18 week and 6 month studies, respectively. Soy isoflavones do not appear to cause this effect, as Miyanaga et al. (16) did not find a reduction in these hormonal biomarkers in men in a randomized one-year study with soy isoflavones in tablet form. The results of the latter study are consistent with those of MacKey et al. (48) and Hamilton-Reeves et al. (49) who also found that isoflavones are not responsible for soy protein-caused decrease of testosterone and SHBG. It should be noted, however, that the study by Miyanaga et al. (16) included Japanese subjects who may have been consuming soy habitually, which is much less likely in the Australian and US subjects in the studies by MacKey et al. (48) and Hamilton-Reeves et al. (49), respectively.

The absence we found of effects of soy supplementation on biomarkers of cancer-associated processes in males is consistent with the lack of a protective effect of soy on prostate cancer recurrence (18) and an overall lack of a soy effect on PSA identified in the meta-analysis by van Die et al. (5). Although Hamilton-Reeves et al. (54) found that fewer men with preneoplastic prostate lesions on soy protein, with or without isoflavones, developed biopsy-detectable prostate cancer than men on milk protein, their study was small (n = 18-20) and only had 6 months of follow-up from baseline. In the one-year study by Miyanaga et al. (16) conducted in healthy Japanese men with a negative prostate biopsy at baseline who were randomized to soy isoflavone tablets or placebo, a reduced incidence of prostate cancer was observed in older men (≥ 65 years) who did not produce equol, but not in younger men or in equol producers. This study, unlike most others, did not involve soy protein but isolated isoflavones, was of relatively short duration, and depended on biopsy-detected cancer subject to inherent sampling error. In studies with females, soy did not reduce risk of endometrial hyperplasia and cancer (17) or affect mammographic breast density suggesting lack of protection against breast cancer (19–23).

There is in vitro evidence of the effects of isoflavones on VEGF and bFGF in cell models (28–30,55). However, we did not find effects on serum concentrations of VEGF and there are no other published clinical studies of soy and angiogenesis biomarkers. There are also no other clinical studies of soy and apoptosis serum biomarkers and we are the first to report the lack of a soy effect on circulating Fas and Fas-ligand. There is, however, one randomized study comparing soy protein versus milk protein supplementation for 6 months that examined immunohistochemical expression of biomarkers in prostate biopsy tissue and found no effects on Bcl-2 or Bax compared to a baseline biopsy (54). This study also found no effect on PCNA expression, consistent with the reported absence of effects of isolated soy isoflavones on the cell proliferation marker Ki67 in prostate tissue (52,53). By contrast, in similar studies of 3-6 weeks treatment with isolated soy isoflavones, reduced prostatic RNA expression was found for Bcl-2, Bax, caspase-7, and anti-proliferative genes such as E2F4 (56). Thus, the possible effects of soy on apoptosis and cell proliferation remain undetermined.

By contrast, there are several clinical studies of soy and the IGF axis, most of them in females (57–61). In males, soy protein supplementation for 3 months has been reported to increase circulating IGF-1 (62,63). This may be a transient effect as we observed an elevation in IGF-1 after 2-month supplementation, but not at later time points (Table 3). Consistent with this notion, Adams et al. (64) did not find an effect on IGF-1, IGFBP-3, or their ratio in a 12-month study comparing soy protein containing isoflavones with soy protein from which isoflavones had been removed in healthy subjects, 85% of which were male. Studies with soy-containing bread for 18 weeks (65) or a whole diet/life style intervention including soy for one year (66) also did not find intervention effects on IGF-1 or IGFBP-3. By contrast, a one-year study in obese men and women combined found that soy protein (n = 25) compared to whey protein (n = 23) increased both IGF-1 and IGFBP-3, but did not report separately on the male results (67). In a short-term randomized study with isolated soy isoflavones in males prior to prostate surgery, there was no effect on circulating IGF-1, although IGFBP-3 was reduced compared to baseline but not compared to placebo controls (53).

The present study had a major strength in its long duration of intervention and the number of time points, but also has several limitations. Although the subjects were drawn from a randomized trial, the present study was not a randomized study and may therefore not be representative of the trial population. However, the baseline characteristics of the subjects on soy and those on placebo were not different, except for their body weight and BMI, with the placebo subjects being slightly heavier than those on soy. This difference was at the cusp of statistical significance for the subjects whose samples were examined for changes in concentrations of Fas and its ligand and IGF-1 and IGFBP-3 and may have influenced the observed lack of a soy effect on these parameters. This was also the case in the parent trial that did not apply baseline weight or BMI as stratification factors in the randomization. However, we did not find differences between the soy and placebo groups in the slight changes in body weight over the 18 months of the present study. The present study was not powered to evaluate effects on the endpoints included, the number of available samples was limited, and the blood sampling time points and most endpoint analyses were dictated by the protocol of the original clinical trial (18). The analyses of soy intervention effects on biomarkers of apoptosis, angiogenesis and the IGF axis were particularly limited by small sample sizes and should be considered exploratory. Circulating testosterone levels display a distinct circadian rhythm with variation over the day. We could, however, not precisely standardize the time of blood sampling because of the often-difficult logistics of study visit scheduling, although most samples were taken in the morning hours. The effects of sample storage duration were not determined for the endpoints studied. Whether equol production was a factor was not studied. The soy and milk protein doses were at the low end of doses studied in some but not all other clinical trials. These doses were selected to produce palatability that would limit loss of adherence which was a low 5% in the parent trial (18).

In conclusion, 18 months of consumption of 19.2 g/day of whole soy protein isolate containing 24 mg genistein by middle-aged to older males reduced circulating testosterone and SHBG, but not free testosterone SHBG compared with the casein-based placebo. The effects on other cancer biomarkers remain to be more definitively determined and future studies with larger sample sizes are needed to substantiate our findings.