Randomized, Placebo-Controlled Six-Month Intervention Study of Soy Protein Isolate in Men with Biochemical Recurrence after Radical Prostatectomy: A Pilot Study

Abstract

There is evidence to suggest that soy may be beneficial for prostate cancer patients, but few randomized trials have addressed this. We examined the effect of 6–8 month soy protein supplementation on prostate specific antigen (PSA) serum levels in men who recurred (PSA > 0.1 ng/ml) within three years of prostatectomy. Sixteen men were randomized to 20 g soy protein (~24–26/day genistein; ~40–43/day total isoflavones) or casein placebo. PSA was measured at base line and at 1, 2, 4, and 6–8 months. Serum genistein levels greatly increased from baseline and cholesterol decreased in the soy group. In both treatment arms PSA increased similarly and PSA doubling times were not different over the 6–8 months study duration. Two subjects in each group had stable PSA. A literature search for clinical studies of soy, isoflavones, and PSA revealed that supplementation with soy or isoflavones did not affect PSA in virtually all clinical studies identified. Although this study is too small to draw a definitive conclusion on the effect of soy protein on PSA in men with biochemical failure, the null finding in this study is consistent with the results of virtually all reports of soy and soy isoflavones in the literature.

Introduction

Prostate-specific antigen (PSA) is a serine protease member of the kallikrein-related peptidase family. It is produced by human prostatic epithelial cells and is secreted in the semen. It can leak into the blood stream in disease conditions of the prostate, prostatitis, benign prostate hypertrophy (BPH), and prostate cancer. It is measured in the blood for early detection of prostate cancer and to monitor efficacy of treatments for this malignancy. Treatments of early stage prostate cancer, radical retropubic prostatectomy and various radiation modalities, typically have curative intent. When local recurrence occurs or distant metastases have developed, PSA levels in the blood will start rising from the low to undetectable levels immediately following the initial treatment. Various approaches to deprive the cancer cells from androgenic stimuli and/or chemotherapy or salvage radiation treatment are then used as further therapeutic approaches, all with more or less severe side effects.

To delay these medical interventions or as adjuvant-type approaches, non-toxic treatments with dietary supplements and lifestyle changes have been proposed and to some degree tested (1,2). For example, Ornish and colleagues (3,4) have proposed and tested a lifestyle and dietary change approach showing an increase in relative telomere length in prostate biopsies after five years of intervention as well as a reduction in serum PSA. One of the most often proposed dietary supplements for prostate cancer is soy (1,2). Two systematic reviews and one meta-analysis that differed in scope have addressed this, but taken together yielded inconclusive result, in large part because of the lack of studies and their variable quality (5,6). We examined whether a soy protein supplement could delay or prevent recurrence after surgery, but found that this did not work as hypothesized (7).

We started a randomized placebo-controlled trial with soy protein isolate and a casein-based placebo in men with rising PSA after radical prostatectomy to determine whether soy could slow down, inhibit, or reverse the rise in PSA. Because of unanticipated changes in clinical practice after 1999, with radiation treatment or hormonal therapy being applied very early on after PSA failure impeding accrual, we were forced to close this study. We decided to report the data in this paper because we discovered that no comparable randomized, placebo-controlled studies have been reported when we conducted a literature review about soy and prostate cancer with a wider scope than the systematic reviews mentioned above (5,6). The results of this literature survey are included in this paper.

Materials and Methods

The study was approved by the Institutional Review Board of New York University School of Medicine, but it was not registered at clinicaltrials.gov which was not standard at the time.

Subjects

Seventeen subjects, ages 54–76, were enrolled between June 1997 and December 1999 at the Department of Urology of NYU School of Medicine (NYU) and the Manhattan Veterans Administration Medical Center (VA). Four (25%) of these men were African American, two (12.5%) were Hispanic, while the remaining 10 men (62%) were white (Table 1). One of these subjects dropped out before randomization, while 16 were evaluable, 15 of whom completed the 6 months study and one withdrew for further treatment after 4 months.

TABLE 1.

Demographic and Baseline Data

		Soy Protein	Placebo	P value
Number of evaluable subjects		8	8
Age, years, mean (95% CI)		65.8 (60.2, 71.3)	63.8 (58.5, 69.0)	0.564 b
Body Weight, kg, mean (95% CI)		91.5 (75.8, 107.2)	80.5 (70.4, 90.7)	0.115 c
Body Mass Index, mean (95% CI)		29.4 (25.3, 33.5)	26.0 (23.7, 28.3)	0.083 c
Race a	White (Caucasian)	6 (75%)	4 (50%)	0.608 d
	African American	1 (12.5%)	3 (37.5%)
	Hispanic	1 (12.5%)	1 (12.5%)
Hospital	Tisch Hospital	6 (75%)	5 (62.5%)	1.000 c
	Manhattan VA	2 (25%)	3 (37.5%)

^aSelf-reported

^bTwo-sided t-test for difference between soy protein and placebo

^cTwo-sided Mann-Whitney test for difference between soy protein and placebo

^dTwo-sided Fisher exact test for difference between White versus African American and Hispanic

Study Design

After obtaining informed consent and confirmation of the base-line PSA level, the 16 men were randomized to intervention or placebo treatment, stratified by hospital (NYU versus VA) and race (white versus non-white). Blood samples were taken at base-line and at 1, 2, 4, 6, and 8 months after the start of the treatment. Serum was prepared and either stored at −80°C until analyzed for PSA and isoflavone concentration or submitted for a clinical chemistry screen including total cholesterol carried out by the NYU School of Medicine Faculty Practice out-patient laboratory.

Eligibility Criteria were as follows: Subjects had to have undergone radical prostatectomy for clinically localized prostate cancer less than 3 years ago; had to present with a detectable PSA level (≥ 0.1 ng/ml) at least 3 consecutive occasions no less than one week apart; and have a baseline PSA value between ≥ 0.1 ng/ml and ≤ 2.0 ng/ml.

Exclusion Criteria included any current adjuvant therapy for prostate cancer; any clinical evidence of local recurrence or metastases; a significant base-line intake of soy including; a recent history of anemia or iron deficiency problems or sub-clinical iron deficiency at baseline; a recent history of or current diabetes, the need for a sodium-free diet, a substantive tendency to be constipated, a history of thyroid disease, or another medical problem precluding the consumption of the soy and casein-containing beverage powders, including allergies against soy or milk protein; any current major disease, including major mental disease and major substance abuse problems (which often interfere with retention) or medication with significant side-effects.

The intervention agent, soy protein isolate (from non-GM sources), or placebo, casein, was administered contained in a highly palatable beverage powder specifically developed for this type of chemoprevention clinical trials and manufactured and generously donated by Solae LLC (St. Louis, MO). A single 47 g. package of beverage powder was administered daily, which provided subjects in the treatment arm with approximately 23–26 mg of genistein and 40–43 mg of total isoflavones from their 20 g/day soy protein intake. This amount of soy intake is similar to that of infants and children consuming soy as nutritional supplement because of allergy to cow’s milk protein or lactose intolerance, and it is similar to soy intake of a traditional Japanese diet and about 1.5 to 2 times the current average daily intake in older Japanese in Japan.⁸ The beverage powders were supplemented specifically for this trial in such a way that they are not distinguishable in macro- and micronutrient composition and only differ in the type of protein: soy protein isolate or calcium caseinate. Details of the nutrient composition of these materials have been published previously (5).

End-Points

The primary endpoint was the change in serum PSA from baseline over six months of intervention as expressed by the PSA doubling time calculated from slope of the change in PSA using the formula ln 2 (0.693)/ln slope (9). The concentrations of total cholesterol and genistein were secondary endpoints.

Laboratory Assays

The ultra-sensitive Tosoh PSA assay (Tosoh Medics, Foster City, CA) was used for all PSA assays in this project. This fully automated immunoenzymometric method utilizes the two standard Tandem Hybritech monoclonal antibodies. The biological detection limit of this assay is 0.01 ng/ml and the assay variation is 7% for PSA 0.1–1.0 ng/ml and 12% for PSA ≤ 7 ng/ml. These assays were conducted in the NYU Urology Research Laboratory which was CLIA and NY State-certified. All serum samples were frozen (−80°C) prior to analysis and assayed for PSA within a week.

HPLC analysis with electrochemical detection (ESA) was used for the measurement of serum isoflavones using the procedure of Gamache et al. (10) with slight modifications according to Franke et al (11,12). After hydrolytic treatment with glucoronidase/sulfatase of 0.5 ml serum samples and trichloroactetic acid precipitation, the samples were extracted with ethyl acetate (3x) and evaporated under nitrogen to near dryness. The samples were re-dissolved in 0.1 ml methanol and then an equal amount of water will be added. This mixture was injected into our ESA HPLC system with a C18 MD-150 ESA column with isocratic elution with water-methanol-acetonitrile (68:25:7, v/v/v) mobile phase containing 0.2 mol/L sodium acetate buffer, pH 4.8, flow rate 0.6 ml/minute at 42°C column temperature. Genistein was quantitated using estradiol-glucuronide as the internal standard. Extraction recovery values were >90%, the detection limit of this assay was 200 fmoles injected on column, translating into 0.5 ng/mL serum, and its coefficient of variation was 7%.

Statistical Analysis

We computed the slope of PSA change (derived from regression analysis) and PSA doubling time, as well as the change in total cholesterol and genistein values from baseline to the 6 month time point. We compared these endpoints between the two treatment groups. The difference between the two groups in PSA doubling time and PSA at baseline was analyzed using a 2-sided Mann-Whitney test. A natural log transformation of PSA doubling data was applied prior to this analysis. For the change in cholesterol from baseline to 6 months within each treatment arm we used 2-sided Wilcoxon matched-pairs signed-ranks tests. The difference in cholesterol between the two treatment groups and interactions between treatment and time were analyzed using two-way repeated measures ANOVA. For comparison between the two treatment groups of PSA and cholesterol at baseline and for comparison between the two groups of the change in cholesterol at the 6 month time point we used a 2-sided Mann-Whitney test. Prizm software (version 4.03; GraphPad, San Diego, CA) was used for these analyses.

Literature Search Strategy

We did not attempt to conduct a formal systematic review, but searched PubMed over all available years on 08/22/2020 with the following search terms: (soy OR isoflavones) AND (PSA), yielding 93 hits. We then reviewed all titles, eliminated all laboratory studies and non-English publications, examined the abstracts of all remaining papers and eliminated studies that did not address PSA, general reviews without data, and epidemiological reports, and identified two systematic reviews and one meta-analysis (5,6). The remaining papers (n = 28) were all examined in detail which, eliminated 3 irrelevant papers of 25 papers that were reviewed in depth. To make sure we did not miss anything we then compared this initial list with a search using these terms: (soy OR isoflavones OR genistein) AND ((PSA OR (prostate cancer)) yielding 938 hits, which we reduced to 61 hits by applying a filter for clinical trials, 27 of which were relevant. Although this search largely overlapped the initial search, it yielded another 5 papers for a total of 30 papers that were deemed relevant and provided sufficient information for conclusion. These were reviewed in detail. We also searched www.clinicaltrials.gov for clinical trials with prostate cancer and soy (35 hits, one trial not identified on PubMed), isoflavones (17 hits, all overlapping with soy), or genistein (17 hits, 2 in addition to those found with soy as search term), two of the additional trials, both with mixtures, had results but neither reported on effects on PSA.

Results

Clinical Trial

Sixteen subjects completed 4 months of intervention, 15 completed 6 months, and 14 completed 8 months. At base-line their PSA ranged from 0.12–1.34 ng/ml. Serum levels of genistein at baseline were close to or below the 0.5 ng/mL limit of detection and we did not find differences between the soy group (median: 0.5 ng/mL; range: 0.5–2.5 ng/mL) and the placebo group (median: 0.5 ng/mL; range: 0.5–7.8 ng/mL). There was a marked and statistically significant increase in serum levels of genistein from base-line to four months (median: 168 ng/mL; range 57–448 ng/mL) in the group receiving the soy protein intervention, indicating that all subjects were compliant and that the isoflavones in the intervention material was highly bioavailable. Genistein levels in men on placebo remained at the limit of detection. The subjects in either group did not gain or lose weight over the six months of intervention (data not shown), but soy subjects were slightly heavier at baseline than those on placebo (Table 1).

Baseline PSA levels were higher in the soy group than in the placebo group (Tables 2 and 3). After 6 months treatment, PSA in the soy group was increased by 34.1% (95% CI: 7.3%, 60.8%), while in placebo PSA increased by 53.6% (95% CI: 11.1%, 118.3%). One subject in the soy group and two in the placebo group had minimally negative slopes and PSA doubling times and one subject in the soy group had a doubling time of 71.44 months (6 years); PSA levels of these subjects were considered “stable” (Table 2). Regardless of whether or not the data of these stable subjects were included in the statistical analysis, PSA doubling time was not different between the two groups as assessed by the Mann Whitney test (Table 3).

TABLE 2.

Individual data of baseline and end-of-study PSA values (ng/ml), slope of change in PSA (ln ng/ml/month), and PSA doubling times (months)

Soy Protein					Placebo
Subject	PSA		Ln Slope	PSA Doubling Time a	Subject	PSA		Ln Slope	PSA Doubling Time a
	Baseline	End				Baseline	End
1	0.390	0.670 d	0.1213	5.71	9	0.215	0.640 c	0.1823	3.80
2	0.275	0.470 c	0.0980	7.07	10	0.275	0.735 b	0.1407	4.93
3	1.335	2.545 b	0.0767	9.04	11	0.810	1.490 b	0.0956	7.25
4	0.420	0.670 b	0.0617	11.23	12	0.130	0.195 b	0.0473	14.65
5	0.815	1.085 b	0.0599	11.57	13	0.195	0.260 b	0.0378	18.33
6	0.570	0.955 b	0.0552	12.55	14	0.135	0.185 b	0.0356	19.47
7	0.660	0.700 b	0.0097	71.44 e	15	0.230	0.220 b	−0.0152	−45.59 e
8	0.120	0.130 b	−0.0063	−110.00 e	16	0.138	0.140 b	−0.0121	−57.27 e

^aDoubling time = 0.693/ln slope,⁹ calculated from PSA data over the entire period of supplementation (subject 1 completed 4 months of intervention, subject 1 and 9 completed 6 months, and all others completed 8 months)

^b8 months after baseline

^c6 months after baseline

^d4 months after baseline

^eConsidered stable

TABLE 3.

Baseline and end-of-study PSA values (ng/ml) and PSA doubling times (months)

		Soy	Placebo	P value a
All subjects
N		8	8
Baseline PSA	Median	0.495	0.205	0.059
	Range	0.120 to 1.335	0.130 to 0.810
End-of-study PSA	Median	0.685	0.240	0.195
	Range	0.130 to 2.545	0.140 to 1.490
PSA Doubling time	Median	10.14	6.09	0.721
	Range	71.44 to −110.0	19.47 to −57.27
Subjects with rising PSA only
N		6	6
Baseline PSA	Median	0.495	0.205	0.037
	Range	0.275 to 1.335	0.130 to 0.810
End-of-study PSA	Median	0.813	0.450	0.180
	Range	0.470 to 2.545	0.185 to 1.490
PSA Doubling time	Median	10.14	10.95	0.818
	Range	5.71 to 12.55	3.80 to 19.47

^aTwo-sided Mann-Whitney test for difference between soy protein and placebo

Serum levels of total cholesterol decreased from base-line to four months by 10% in the group receiving the soy protein intervention, but increased 3% in the placebo group (Table 4). There was a significant interaction between time and treatment.

TABLE 4.

Concentrations of Total Cholesterol (mg/dL) at Baseline and Change from Baseline at 1 – 6 Months on Study

Group		Baseline	Δ (%) 6 months	P change a	P difference b	P interaction b
Soy	Mean	205.4 c	−9.6 d	0.016	0.695	0.014 e
	95% CI	185.1, 225.7	−15.9, 3.2
	N	8	7
Placebo	Mean	197.6 c	4.0 d	0.461
	95% CI	176.6, 218.7	−7.2, 15.15
	N	8	8

^aTwo-sided Wilcoxon matched-pairs signed-ranks test for change in cholesterol between baseline and 6 months

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

^cP = 0.433

^dP = 0.040 for difference between soy and placebo (2-sided Mann-Whitney test)

^eBecause the interaction is significant, the P value for the difference is unreliable

There were no adverse events, self-reported compliance was excellent, and one subject withdrew after 4 months because of starting radiation treatment.

Literature Review of Soy and PSA by Characteristics of Subjects

Twenty nine of 30 papers identified as potentially relevant described human studies with sufficient detail in methods and results on PSA to allow conclusions. At the outset, it should be noted that some of the reviewed studies had small sample sizes which limited their statistical power and several studies were not randomized or placebo-controlled limiting their impact. In Table 5, these published studies are listed by characteristic (intervention material, study design, type of study cohort, effects on PSA).

TABLE 5.

Characteristics of Studies on Soy and PSA Identified in the Literature

Study Characteristics	References:
Intervention Material:
Synthetic genistein	13
Isolated isoflavones	14–21
Isoflavone-rich soy or red-clover extracts	22–25
Supplement mixtures including soy isoflavones	27–29
Soy containing bread or cereal	30, 37
Soy products-rich diet	31, 34
Soy protein isolate	32, 33, 35, 36, 38, 41, 42
Soy milk or “soy beverage”	39, 40
Type of Study Cohort:
Healthy middle-aged to older men	30–32
Men with elevated PSA but negative biopsies	17
Men on watchful waiting or active surveillance	33–36
Men scheduled for radical retropubic prostatectomy	13–16, 37, 38
Men with rising PSA after prostatectomy or radiotherapy	19, 20, 39–41
Men on androgen deprivation or radiation therapy	18, 21, 42
Men with untreated stage B, C, or D prostate cancer	18
Type of Study:
Randomized, placebo controlled	13–17, 21, 25, 26, 33, 42
Randomized comparing two or three treatments (no placebo)	32, 34, 36, 37
Cross-over design	30, 31, 35
Single arm	18–20, 22–24, 27–29, 38–41
Effect of Intervention on PSA:
No effect on PSA	14–21, 25–28, 30–36, 38, 42
Inconsistent effects	22–24, 29, 37, 39–41
Statically significant reduction in PSA	13 (borderline), 37 (soy vs. wheat)

Nine studies examined effects of isolated isoflavones. Four of these studies randomized men to isoflavones or placebo for 2–6 weeks prior to radical prostatectomy (13–16). One study involved men with negative prostate biopsies randomized to isoflavones or placebo for 12 months (17). One study randomized men with prostate cancer to very high doses of isoflavones for 3 months (18). Two non-randomized studies examined effects of isoflavones on PSA in men who had failed surgical, radiation, or androgen-deprivation therapy (19,20). One randomized, placebo-controlled study examined effects of isoflavones in men while they were on radiation therapy for prostate cancer (21). In none of these studies with isolated soy isoflavones were effects in PSA or differences with placebo found. In one randomized placebo-controlled study prior to radical prostatectomy, synthetic genistein was tested and a borderline significant reduction of PSA compared to placebo was observed (13).

Four studies examined effects of soy or red-clover extracts rich in isoflavones. Three of these were not randomized and placebo controlled and found inconsistent effects on PSA (22–24). One study randomized men on active surveillance to soy extract or placebo for 6 months and found no effect on PSA (25).

Two studies randomized men with rising PSA after prior treatment to a supplement mixture including soy isoflavones or placebo for 6 months (26) or for 6–10 weeks (27,28) in trials with a cross-over design. None of these studies found an effect of the mixture on PSA compared to placebo. One not randomized study provided a similar cohort of men with supplement mixture including soy isoflavones for 6 months and found inconsistent effects on PSA (29).

Thirteen studies provided men with soy protein or high soy content foods (e.g., soy milk), six of which randomized and placebo controlled. These 13 studies are summarized below grouped by the characteristics of their study subjects.

Healthy Men –

Three studies reported the effects of soy on PSA in “healthy” men, although one study included 46 hyperlipidemic men one of whom had prostatic intraepithelial neoplasia (30). Two studies were of a randomized cross-over design with a low soy control group. One study randomized middle-aged men to soy milk and a soy-based cereal as sources of soy and dairy foods and egg substitutes as control for 3–4 weeks and in subgroup of 6 men on soy and 7 on control diet for a total of 12 weeks (30). The other study provided 23 middle-aged men with normal PSA and no evidence of prostate cancer with diets high or devoid of soy products for 3 months followed by a cross-over after a one month washout (31). The third study measured PSA in samples from a colon cancer screening study that randomized older men to soy protein isolate containing isoflavones (n = 34) or soy protein stripped of isoflavones (n = 47) for 12 months (32). None of these three studies observed soy effects on PSA or differences in PSA between the two study arms.

Men on Watchful Waiting or Active Surveillance –

Kumar et al. (33) randomized men on watchful waiting for 12 weeks to soy protein isolate (n = 39) or a milk protein placebo (n = 37). Aronson et al. (34) randomized 18 men on watchful waiting or active surveillance to a low fat diet supplemented with soy protein or a western diet for 4 weeks. There were no soy effect on PSA and no differences between soy and placebo in these two studies. Urban et al. (35) conducted a six week randomized cross over study with soy protein isolate containing or stripped of isoflavones in 34 men over 55 years of age who had a negative prostate biopsy or elevated PSA but no biopsy results and probably on watchful waiting. Hamilton-Reeves et al. (36) randomized men who had preneoplastic prostate lesions on biopsy or had prostate cancer and were on active surveillance to similar soy protein isolates containing isoflavones (n = 20) or stripped of isoflavones (n = 20) and milk protein (n = 18) for 6 months. No effects on PSA were found in the latter two studies of the presence or absence of isoflavones in the soy supplements and there was no difference in PSA between the soy groups and the milk protein group in the study by Hamilton-Reeves et al. (36).

Men Scheduled for Radical Retropubic Prostatectomy –

One study randomized men for 20–30 days to bread containing soy (n = 8), bread containing soy plus linseed (n = 10), or wheat bread (n = 8) (37). Mean PSA was reduced over baseline by 12.7% on soy, increased by 21.3% on soy plus linseed, and 40% on wheat bread (P = 0.021). One not randomized study examined the effects of soy protein supplementation for 2–11 weeks prior to radical prostatectomy and did not observe a reduction in PSA (38).

Men with Rising PSA after Radical Prostatectomy or Radiotherapy –

Three studies, all not randomized, provided men who had rising PSA after initial surgical or radiation therapy failed (n = 29) with a soy beverage for 6 months (39), soy milk (n = 20) for 12 months (40), or soy protein isolate (n = 35) for 2 months (41). Although PSA in some subjects in these studies stabilized or decreased, it increased in others and no consistent effect across these studies is apparent. There are no randomized studies in this population.

Men on Androgen Deprivation Therapy –

Sharma et al. (42) randomized 33 men to a soy protein beverage powder or a milk protein placebo for 12 weeks while they were on androgen-deprivation therapy and reported an absence of effect on PSA without showing the actual data.

Discussion

In this small randomized clinical trial, we did not find differences between the soy protein and placebo groups in PSA doubling time measured over 6–8 months. The soy supplementation did markedly increase serum levels of genistein, indicating that the intervention was effective on those indicators of soy effects, and it modestly lowered serum total cholesterol compared to placebo. We were unable to identify other randomized trials, registered or published, testing the effects of soy protein isolate on PSA in men with biochemical failure after initial surgical treatment with curative intent. The study was limited by its very small sample size, resulting in wide confidence intervals for changes in PSA over 6 month period in both arms, and was thus inconclusive on the potential of the intervention to be beneficial or detrimental. Also, the observed change in PSA over 6 months was much smaller than expected on the basis of clinical data used for the original sample size calculation (59 subjects in each arm) and therefore the power to detect differences was very limited. On the other hand, the number of subjects with stable PSA and the PSA doubling times of subjects with continuing rising PSA was not different between the two groups. The study had as major strength its randomized placebo-controlled design and documented soy effect on serum cholesterol and genistein levels. Our result are in line with those of published studies in men with biochemical failure with soy protein which were not randomized and not placebo controlled (39–41).

Our literature survey overall indicates that soy and soy isoflavones do not affect circulating levels of PSA. All randomized studies with soy protein or soy foods showed an absence of effects on serum PSA, regardless of whether this was studied in men with normal PSA levels, men with elevated PSA but no biopsy-detectable cancer, men with prostate cancer awaiting radical surgical treatment, or men after failure of initial treatment. The few studies that were not randomized yielded the same negative result or inconsistent findings. Only in one study comparing soy and wheat-containing bread was a reduction in PSA identified in the soy group (37). Our broad-scope literature review is thus in agreement with systematic reviews and a meta-analysis with a more narrow scope (5,6).

The significance of effects on serum PSA of dietary and other interventions is not easy to interpret, particularly in the context of localized prostate cancer, local recurrence, or distant metastases. It is too simple to equate increasing or decreasing PSA with an increased or decreased number of prostate cancer cells. Multiple effects of soy are likely and pleomorphic effects of soy isoflavones have been documented (43–45). Treatments that have a differentiating effect on these cells can increase their PSA secretion even if their proliferation is inhibited. For example, we showed that 9-cis-retinoic acid at concentrations that inhibit in vitro cell proliferation increase PSA secretion by LNCaP cells (46). Thus, the absence of a change in serum PSA in response to interventions such as with soy protein does not necessarily mean an absence of an effect on tumor growth or progression. However, we found that genistein increases apoptosis and reduces both proliferation and PSA protein expression of LAPC-4 prostate cancer cells that have wild-type androgen receptor (47). Of note, LNCaP cells represent prostate cancer after androgen-deprivation treatment and have a mutated androgen receptor that responds promiscuously to genistein because the isoflavone binds to and activates the mutated receptor (47). It is important to consider the concentrations used in in vitro studies with genistein, as well as the presence of androgen, and compare this with the prostatic genistein concentrations found in men consuming soy. Gardner et al. (48) have documented that prostatic levels of genistein in men consuming isoflavone tablets containing 32 mg genistein per day for two weeks prior to prostatectomy are approximately 2.23 μmol/kg tissue, 5 times higher than serum levels in the same subjects. We used soy protein providing 23–26 mg genistein per day resulting in a mean serum level of 168 ng/mL genistein (~0.62 μmol/L) which probably resulted in a genistein concentration in prostate tissue of approximately 1.6 μmol/kg genistein. When genistein is applied to LAPC-4 cells in vitro at 5–10 μmol/L in culture media containing stripped serum, there is about 5% reduction in cell proliferation, but there is no reduction in proliferation in response to genistein when supplementing the media with androgen at a physiological concentration (47). In fact, we did not see any reduction of proliferation of cells in androgen-supplemented culture until 25 μmol/L genistein and above. Thus, it should come as no surprise that only high isoflavone concentrations in vitro and pharmacological doses in vivo may have inhibitory effects on proliferation of prostate cancer cells that could lead to reduction in serum PSA.

Our study and most of the studies in the literature we reviewed deal with relatively early stage prostate cancers that have not been exposed to androgen-deprivation and almost certainly still have androgen receptors that are not mutated. Thus, finding that serum PSA is not affected by consumption of soy protein in our trial and virtually all other studies providing 25 to no more than 100 mg genistein per day likely indicates an absence of an inhibitory effect on growth and progression of early stage prostate cancer at those genistein dose levels. Consistent with this notion, we showed that soy protein supplementation does not reduce biochemical recurrence after radical prostatectomy in men with unfavorable pathology (7). By contrast, Hamilton-Reeves et al. (36) 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 in a 6 months randomized study. However, this study was limited by its small sample size (n = 18–20/group) and short follow-up from baseline. Miyanaga et al. (18) conducted a one-year study by in healthy Japanese men with a negative prostate biopsy at baseline who were randomized to soy isoflavone tablets or placebo and observed a reduced incidence of prostate cancer. However, this was only found for older men (≥ 65 years) who did not produce equol and not in younger men or in equol producers. Of note, this study, unlike most others, did not involve soy protein but isolated isoflavones, and it was of relatively short duration and depended on biopsy-detected cancer subject to inherent sampling error. These two studies are the only ones that have reported effects of soy consumption on the development and large randomized studies with long follow-up are lacking.

Conclusions

This small randomized study with protein isolate supplementation for six to eight months compared to placebo did not find a difference in PSA doubling time in men with biochemical failure after radical prostatectomy. Although the study was clearly limited in statistical power, the observed absence of effects of soy protein on PSA is consistent with results from virtually all other clinical studies with soy protein intervention and PSA and with the effects of physiological concentrations of genistein on prostate cancer cells with wild-type androgen receptor in laboratory studies.