European Journal of Clinical Nutrition (2009) 63, 238245

& 2009 Macmillan Publishers Limited All rights reserved 0954-3007/09 $32.00
www.nature.com/ejcn

ORIGINAL ARTICLE

Dietary combination of soy with a probiotic or

prebiotic food significantly reduces total and LDL
cholesterol in mildly hypercholesterolaemic subjects
TA Larkin1,2, LB Astheimer1,2 and WE Price2,3
1
Department of Biomedical Science, University of Wollongong, Wollongong, New South Wales, Australia; 2ARC Smart Foods Key
Centre and 3Department of Chemistry, University of Wollongong, Northfields Ave, Wollongong, New South Wales, Australia

Objective: We hypothesized that a dietary combination of soy with either a probiotic (yoghurt) or a prebiotic (resistant starch)
would result in enhanced lipid-lowering effects compared with a control soy diet, possibly via improvements in isoflavone
bioavailability.
Subjects: Mildly hypercholesterolaemic subjects (men and post-menopausal women) older than 45 years were recruited via the
local media. Thirty-six subjects commenced the study; five withdrew.
Results: Soy probiotic significantly decreased total cholesterol (4.772.0%; P 0.038) and soy prebiotic significantly
decreased total and low-density lipoprotein cholesterol (5.571.6%; P 0.003 and 7.372.2%; P 0.005, respectively). The
bioavailabilities of daidzein, genistein or equol were not affected by probiotic or prebiotic consumption or associated with lipid
changes.
Conclusion: Dietary combination of soy with either a probiotic or a prebiotic resulted in significant lipid lowering, not related to
isoflavone bioavailability.

European Journal of Clinical Nutrition (2009) 63, 238245; doi:10.1038/sj.ejcn.1602910; published online 17 October 2007
Keywords: isoflavones; probiotic; prebiotic; lipids; bioavailability; equol

Introduction
Since publication of Anderson et al. (1995) meta-analysis
that reported significant lipid reductions from soy protein
intake, much research has been conducted in an attempt to
elucidate the specific roles of isoflavones and soy protein
in this hypocholesterolaemic effect. A more recent metaanalysis was inconclusive (Yeung and Yu, 2003) and
discrepancies remain in relation to the lipid-lowering effects
of these soy constituents (Lichtenstein, 1998). Isoflavone
supplements alone generally do not reduce lipids (Nestel
et al., 1997; Hodgson et al., 1998); however, in combination
with soy protein, appear necessary and result in dosedependent lipid lowering (Crouse et al., 1999; Merz-Demlow
et al., 2000; Wangen et al., 2001). Thus, for lipid improvements,
Correspondence: Dr TA Larkin, Department of Biomedical Science, University
of Wollongong, Northfields Ave, Wollongong, New South Wales 2522,
Australia.
E-mail: tlarkin@uow.edu.au
Received 17 August 2006; revised 4 August 2007; accepted 6 August 2007;
published online 17 October 2007

the soy protein matrix containing isoflavones and intact soy

protein appears to be more beneficial than either component
alone (Potter, 1998; Steinberg et al., 2003).
Following soy intake, the resulting plasma and urinary
concentrations of isoflavones and their metabolites vary
widely between individuals (Winter and Bokkenheuser,
1987; Rowland et al., 1999). Gut microflora plays an essential
role in isoflavone bioavailability because of the necessity of
bacterial enzymes for isoflavone absorption and metabolism
(Winter and Bokkenheuser, 1987; Adlercreutz, 1998; Rowland et al., 1999). If isoflavones are necessary for the
hypocholesterolaemic effect of soy, differences between
individuals in gut microflora and isoflavone bioavailability
may contribute to the variable lipid responses observed.
Adult gastrointestinal microfloral composition is relatively
stable; however, diet can substantially modify the metabolic
activity of certain bacterial populations (Hentges, 1980;
Parodi, 1999). Of human gut microflora, lactobacilli and
bifidobacteria are exclusively beneficial (Gibson, 1998) and
thus termed probiotic bacteria (OSullivan, 2001). These
bacteria have high b-glucosidase activity (Gibson, 1998;

Lipid effects of soy with a probiotic or prebiotic

TA Larkin et al

239
OSullivan, 2001; Turner et al., 2003), an enzyme necessary
for the initial hydrolysis of the soy isoflavone glucosides and
consequent absorption of the bioactive aglycones (Turner
et al., 2003). These probiotic bacteria can be introduced to
the gastrointestinal tract as live cultures contained in food
matrices, such as yoghurt (Cummings et al., 1997) or
alternatively, prebiotic dietary components can selectively
stimulate the growth and/or activity of specific probiotic
bacteria already resident in the gut (Gibson, 1999). Resistant
starch is a prebiotic carbohydrate and in the large intestine is
specifically fermented by bifidobacteria (Brown et al., 1999),
stimulating its growth and increasing b-glucosidase activity
(Silvi et al., 1999).
It is well established that gut microfloral balance affects
cholesterol levels (Jenkins et al., 1999) and lipid reductions
have been reported with both probiotic intake (AgerholmLarsen et al., 2000a) and prebiotic consumption (Taylor and
Williams, 1998; Topping and Clifton, 2001). Thus, intake of
Lactobacillus and Bifidobacterium or prebiotics that selectively
stimulate their growth in the gastrointestinal tract may
positively affect both isoflavone and lipid metabolism. We
hypothesized that intake of probiotic cultures or a prebiotic
dietary component would alter soy isoflavone bioavailability
and produce synergistic effects in relation to lipid lowering.
This study aimed to compare the lipid effects of soy intake
with those of a dietary combination of soy with either a
probiotic yoghurt or a prebiotic (resistant starch-enriched
bread) and also to determine whether any lipid effects
observed were related to changes in isoflavone bioavailability
or metabolism.

Subjects and methods

Subjects
Male and female subjects of at least 45 years and good
general health were recruited via the local media. Inclusion
Pro-1

S + YC

Pro-2

S + YP

Washout

Week -2

Pre-1

SC

Pre-2

S + RS

Week 0

criteria were mild hypercholesterolaemia (total cholesterol

45.5 mmol l1 and low-density lipoprotein (LDL) cholesterol
43.0 mmol l1) and for women to have been post-menopausal for more than 12 months. Exclusion criteria included
elevated triglycerides (45 mmol l1), BMI o20 or 435, the
use of hormone-replacement therapy in the previous 6
months, the use of cholesterol-lowering medication, antibiotics or probiotic supplements in the previous 2 months,
alcohol intake (42 standard drinks per day) and cigarette
smoking.

Study design
The study consisted of two separate cohorts for comparison
of each of probiotic and prebiotic treatments with a
corresponding control (Figure 1). For each cohort, a
randomized crossover design was used, with every subject
acting as their own control. Initially, a general health
questionnaire was administered via telephone. Eligible
subjects were instructed, and routinely reminded, to maintain their usual diet and level of physical activity throughout
the study. After a 2-week wash-in period during which soy
foods, probiotics and prebiotics were excluded from the diet,
subjects commenced the study, which involved two 5-week
dietary periods separated by a 4-week washout. All subjects
consumed soy milk and soy cereal daily during both dietary
periods plus probiotic or prebiotic foods during the corresponding treatments and received advice from a dietician on
how to substitute these foods into their usual diets so as to
minimize changes in their macronutrient and caloric intake.
A test soy meal was consumed prior to and at the end of both
5-week dietary periods to determine the effects on isoflavone
bioavailability following a single, controlled soy intake.
Subjects were allocated to one of four groups (Pro-1, Pro-2,
Pre-1 or Pre-2), which determined the intervention (probiotic
or prebiotic) and the order of crossover (Figure 1). The
control and active dietary treatments for the probiotic cohort
S + YC

S + YP
Washout &
crossover

SC

S + RS

Week 5

Week 9

Week 14

Test (soy) meal

Figure 1

Study design.

European Journal of Clinical Nutrition

Lipid effects of soy with a probiotic or prebiotic

TA Larkin et al

240
were soy control yoghurt (S YC) and soy probiotic
yoghurt (S YP), respectively, and for the prebiotic cohort
were soy control (SC) and soy resistant starch bread (S RS),
respectively. To assist with compliance, subjects with a
preference for either yoghurt or bread were allocated accordingly; otherwise, allocation to the probiotic or prebiotic
intervention was random as was the order of the crossover.
For each test meal and daily during both 5-week dietary
periods, all subjects consumed 250 ml soy milk (So Natural,
UHT Calciforte, Taren Point, New South Wales, Australia)
and 45 g soy cereal (Specialty Cereals, Mt Kuring-gai, New
South Wales, Australia), providing a total of 38 mg daidzein
and 68 mg genistein and 4.5 mg glycitein (aglycone equivalents). Sufficient quantities of each food were provided to
subjects at the beginning of each dietary period; soy milk was
given in 1 l cartons, from which subjects were asked to
measure and consume 250 ml daily and soy cereal was prepackaged as individual 45 g serves.
For the probiotic intervention, subjects consumed 100 ml
yoghurt per day (Vaalia, Pauls Dairy, Brisbane, Qld, Australia)
either with (probiotic) or without (control) live bacterial
cultures. The probiotic yoghurt contained 108 colony
forming units (CFU) of each of Lactobacillus acidophilus,
Bifidobacterium bifidus and Lactobacillus GG per 100 g daily
serve. There were no flavour or texture differences between
the two yoghurts, both were vanilla flavoured of low fat
(1.4% content) and the containers were unmarked except for
a numerical code. Individual 100 ml serves of yoghurt were
provided fresh to subjects every 3 weeks and stored under
refrigeration. For the prebiotic intervention, bread was baked
using 70% resistant starch flour (Penford Australia Ltd,
Sydney, New South Wales, Australia) by a local bakery (K & M
Bakery, Woonona, New South Wales, Australia) to provide
approximately 4 g resistant starch per slice. Bread was baked
every 23 weeks and either delivered to subjects on the day
of baking or frozen in the clinic area for pick-up. Subjects
consumed 45 slices of bread per day (to provide a total of
1620 g RS).
At the start of weeks 0, 5, 9 and 14 (Figure 1), subjects
attended the clinic at the University of Wollongong on two
consecutive days. On the first morning, height and weight
were recorded, a fasted blood sample was taken, then a test
soy meal of 250 ml soy milk and 45 g soy cereal was
consumed. After this, a food frequency questionnaire was
administered and subjects commenced a 24 h urine collection. Subjects did not consume any other soy foods,
probiotics or prebiotics for 48 h post-meal but otherwise
continued their usual activities and diet. Blood was again
collected 8 and 24 h after the test meal and a second 24 h
urine sample was collected between 24 and 48 h post-meal.
Urine samples were collected in 2.5 l collection bottles
containing 1 g ascorbic acid and after completion, the total
volume of each 24 h collection was recorded and duplicate
5 ml aliquots were stored at 80 1C until analysis. Blood was
collected by a registered nurse into ethylenediaminetetraacetic acid tubes (Sarstedt) which were inverted and then
European Journal of Clinical Nutrition

placed on ice until they were centrifuged at 4 1C for 10 min at

2100 g. Plasma was harvested and stored in duplicate aliquots
at 80 1C until analysis.

Isoflavone analysis
Extraction of isoflavones from plasma and urine was based
on the methods reported by Gamache and Acworth (1998)
with modifications and optimizations (Larkin, 2005). All
reagents and standards were from Sigma-Aldrich (Castle Hill,
New South Wales, Australia) and solvents were high
performance liquid chromatography grade (Crown Scientific,
Moorebank, New South Wales, Australia). In the purified,
extracted samples, isoflavones were separated via reversephase high performance liquid chromatography (Shimadzu
auto-injector) with an isocratic mobile phase of 50 mM
sodium acetate buffer pH 4.8 with acetic acid/methanol
(45:55) and quantified by electrochemical detection at
750 mV (VT-03, Antec Leyden) using optimized methods
developed in our laboratory (Larkin, 2005). Mixed standards
of genistein (Sigma, G6649), daidzein (Sigma, D7802), equol
(Sigma, 45405), O-desmethylangolensin (ODMA; Plantech,
UK) at concentrations between 0.2 and 2 mM for plasma, and
0.2 and 4 mM for urine, were included in each high
performance liquid chromatography run at least in triplicate.
Limits of quantification for plasma and urine, respectively
were 5 ng ml1 and 1 mg ml1, for daidzein and genistein,
15 ng ml1 and 2 mg ml1, for equol 60 ng ml1 and for
ODMA 6 mg ml1(Larkin, 2005). Within-run coefficients of
variation for peak area of standards (% standard deviation/
mean) were less than 5% for daidzein and genistein, less than
10% for equol and less than 20% for ODMA. Recoveries were
greater than 70% and calculated values were not adjusted for
recoveries.

Lipid analysis
Total cholesterol and triglycerides were measured in whole
plasma. High-density lipoprotein (HDL) was isolated by
precipitation using the dextran sulphate/magnesium chloride
method (Warnick et al., 1982). Briefly, plasma samples
were mixed with a solution of dextran sulphate and
magnesium chloride, centrifuged and then the clear supernatant was aliquoted for HDL measurement. Total cholesterol,
HDL cholesterol and triglycerides were determined using a
commercially available enzymatic colorimetric test kit
(Roche Diagnostics, Castle Hill, New South Wales, Australia)
and analysed in triplicate using an auto-analyser (Cobas Mira
Plus; Roche Diagnostics). Plasma LDL cholesterol was
calculated using the Friedewald et al.s (1972) calculation.
The coefficient of variation was 0.8% for cholesterol and
1.5% for triglycerides.

Ethics
Ethics approval for this study was granted by the University
of Wollongong Human Research Ethics Committee (Ethics

Lipid effects of soy with a probiotic or prebiotic

TA Larkin et al

241
number 02/248) and all procedures complied with National
Health and Medical Research Council standards required in
Australia. Written, informed consent was obtained from all
subjects prior to commencement of the study.

Statistics
Initially, the normality of data sets was examined (SPSS 11.5,
SPSS Inc. Chicago, IL, USA). Analysis of variance (ANOVA)
with repeated measures and Bonferroni post hoc analysis
(SPSS 11.5, SPSS Inc.) were used and reported with F values,
degrees of freedom (d.f.; hypothesis d.f., error d.f.) and
P-values (for between groups analyses or Bonferroni post hoc).
Factors for two-way ANOVA were always treatment and
week. Students t-tests were two-tailed (Microsoft Excel/SPSS
11.5, SPSS Inc.) and paired, as only within-subject comparisons were made. All values are reported as mean7s.e.m. To
check for an intervention order effect, two-way ANOVA with
repeated measures (factors of treatment and week) and
between groups analysis were performed. To determine the
effects of the dietary periods on plasma and urinary
isoflavone concentrations, three-way ANOVA with repeated
measures (factors of treatment, week and sample time) were
conducted, but reported in detail elsewhere (Larkin et al.,
submitted manuscript). Any relation between plasma or
urinary isoflavone concentrations or changes and lipid levels
or changes, were tested by bivariate correlations with
Pearsons correlation coefficient (SPSS 11.5, SPSS Inc.). For
determination of any overall effects of the 14-week period,
the data of all subjects were included in one-way ANOVA
with repeated measures and Bonferroni post hoc analysis.

Results
Thirty-six subjects (15 females and 21 males) commenced
the study, however 5 people withdrew for a variety of
personal reasons; consequently, 31 subjects (12 females and
19 males) completed the study. All females were postmenopausal for at least 12 months, 10 had never used
hormone-replacement therapy and the other 2 had not in
the previous 6 months. Distribution of subjects and mean
ages for the four study groups are presented in Table 1.
Subject age was not significantly different between the
groups (F 1.020, P 0.399, one-way ANOVA). There were
no significant changes in body weight during the study and
reported food intake did not differ significantly between the
two dietary periods and the washout (data not shown). Mean
intake of daidzein and genistein for each test soy meal and
for the daily soy intake during dietary periods was 0.4870.01
and 0.8470.02 mg kg1 body mass, respectively.
Mean lipid levels (range in parentheses) for all subjects
at baseline were TCh: 6.6570.17 mmol l1 (5.559.19), HDL:
1.2670.06 mmol l1 (0.661.81), LDL: 4.4870.18 mmol l1
(3.157.65), TG: 1.9970.14 mmol l1 (1.024.38) with no
statistical differences between any of the groups for week 0

Table 1

Subject characteristics of study groups

Group

Dietary intervention

n Females Males Age at start (years)

Weeks 05 Weeks 914

Pro-1
Pro-2
Pre-1
Pre-2
Total

S YC
S YP
SC
S RS

S YP
S YC
S RS
SC

10
6
9
6

4
2
3
3

6
4
6
3

60.372.2
55.571.3
56.771.8
61.874.8

31

12

19

58.671.4

Abbreviations: n number of subjects; S, soy; S YC, soy control yoghurt;

S YP, soy probiotic yoghurt; SC, soy control; S RS, soy resistant starch;
RS, resistant starch bread.
Pro-1 and Pro-2: crossover groups of probiotic cohort; Pre-1 and Pre-2:
crossover groups of prebiotic cohort. Mean7s.e.m.

levels (TCh: F 1.233, P 0.317; HDL: F 0.413, P 0.745;

LDL: F 1.376, P 0.271; TG: F 1.007, P 1.405, one-way
ANOVA). Within each group, distribution was normal for
each of the four lipid measures.
There were no significant differences between the two
probiotic groups for lipid changes (TCh: F 1.930, P 0.186;
HDL: F 0.588, P 0.456; LDL: F 1.115, P 0.309; TG:
F 2.684, P 0.125, two-way ANOVA with repeated
measures), thus groups were combined for further analysis.
However, there were significant differences between the two
prebiotic groups for total and LDL cholesterol (TCh:
F 5.534, P 0.035; HDL: F 0.015, P 0.905; LDL:
F 5.224, P 0.040; TG: F 0.277, P 0.608, two-way
ANOVA with repeated measures). Pre-1 showed a significant
increase in LDL (Po0.01) and a trend for an increase in TCh
(P 0.069) during the washout period (weeks 59), whereas
Pre-2 had a mean decrease in both of these measure over the
same time (Table 2). Interestingly, this latter group followed
the S RS dietary period first (Figure 1), which may indicate
an order effect. The prebiotic groups were initially analysed
separately, then combined for further analyses since these
significant differences were not effective for the actual
dietary treatments.
Total cholesterol was significantly reduced with 5 weeks of
S YP and S RS treatments and LDL cholesterol was
significantly reduced after the S RS and SC dietary periods
(Table 3, Students paired t-tests, comparing weeks 0 and 5).
There were no significant differences between either of the
probiotic or prebiotic treatments and their respective controls for any of the four lipid measures (two-way ANOVA
with repeated measures). There was an overall temporal
effect for both TCh and LDL (F3,27 7.055, P 0.001 and
F3,27 5.027, P 0.007, respectively, one-way ANOVA with
repeated measures), but none for either of HDL or TG
(F3,27 1.900 P 0. 531 and F3,27 2.182, P 0.133, respectively, one-way ANOVA with repeated measures). TCh was
significantly reduced between weeks 0 and 14 (P 0.009)
and weeks 9 and 14 (P 0.026) and LDL was significantly
reduced between weeks 0 and 5, weeks 0 and 14, and weeks 9
European Journal of Clinical Nutrition

Lipid effects of soy with a probiotic or prebiotic

TA Larkin et al

242
Table 2 Lipid changes during study for each group
Lipid level (mmol l1)

ANOVAa

Weeks 05

Weeks 914

Week 0

Week 5

Week 9

Week 14

d.f.

%D

%D

Pro-1
TCh
HDL
LDL
TG

7.0770.48
1.2970.13
5.0070.49
1.7170.18

6.9870.45
1.3270.12
4.9070.44
1.9170.25

6.9570.48
1.3270.10
4.8170.46
1.8170.21

6.5270.51
1.2070.09
4.6370.50
1.6570.17

5.199
1.186
2.497
0.533

3.6
3.6
3.6
3.6

0.042
0.391
0.157
0.676

0.0970.16
0.0370.03
0.1070.10
0.2070.19

0.972.2
3.673.4
1.572.2
13.6711.3

0.4370.21
0.1270.08
0.1870.15
0.1670.18

6.372.9
7.775.2
4.073.4
4.378.3

Pro-2
TCh
HDL
LDL
TG

6.3170.22
1.1270.15
4.0970.28
2.4170.48

6.1770.29
1.0670.16
3.9570.35
2.7370.61

6.2570.25
1.1570.15
4.0970.29
2.4070.76

6.0970.23
1.0770.15
3.8970.28
2.6370.36

1.134
0.852
0.577
1.191

3.3
3.3
3.3
3.3

0.460
0.551
0.669
0.445

0.1470.14
0.0670.06
.1470.14
0.3270.20

2.472.3
4.876.4
3.873.6
11.477.5

0.1670.18
0.0770.04
0.2070.22
0.2370.48

2.473.0
6.174.2
4.175.9
35.3720.9

Pre-1
TCh
HDL
LDL
TG

6.7270.20
1.3070.09
4.5070.26
2.0270.26

6.5170.16
1.2870.09
4.2770.20a
2.1070.30

6.8870.13
1.3070.09
4.6870.17b
1.9970.22

6.3770.27
1.2470.07
4.3070.27
1.8470.12

7.357
0.582
7.801
0.295

3.6
3.6
3.6
3.6

0.020
0.648
0.017
0.828

0.2270.10
0.0270.03
0.2470.12
0.0870.20

3.171.4
1.372.6
4.572.8
6.7710.0

0.5070.17
0.0570.05
0.3870.16
0.1570.20

7.672.6
3.073.2
8.673.7
0.5711.2

Pre-2
TCh
HDL
LDL
TG

6.2970.25
1.2770.11
4.1370.29
1.9670.22

6.1170.21
1.2470.09
3.8870.25
2.1770.28

5.8870.20
1.2670.10
3.8170.22
1.8070.17

5.9870.15
1.2570.15
3.6370.12
2.4170.41

2.227
0.242
1.782
1.387

3.3
3.3
3.3
3.3

0.264
0.863
0.323
0.397

0.1870.07
0.0370.05
0.2570.08
0.2170.20

2.771.0
1.074.0
5.771.8
11.6710.7

0.1070.08
0.0170.07
0.1770.13
0.6170.36

1.871.5
2.275.3
3.673.5
33.1719.4

Abbreviations: ANOVA, analysis of variance; HDL, high-density lipoprotein; LDL, low-density lipoprotein; TCh, total cholesterol; TG, triglycerides.
Pro-1 and Pro-2: two groups of probiotic cohort; Pre-1 and Pre-2: two groups of prebiotic cohort. D change over a 5-week period. Mean7s.e.m.
Values in a row with different letters are significantly different. Po0.01, Bonferroni post hoc analysis.
a
One-way ANOVA with repeated measures.

and 14 (P 0.021, P 0.049 and P 0.035, respectively).

While there was a trend for a mean increase in TG with both
soy control dietary periods, there were no increases with
S YP and S RS (Table 3).
At the end of each dietary period, plasma and urinary
concentrations of daidzein and genistein were highly
variable between subjects (Table 4). There were no significant
correlations between lipid levels and plasma or urinary
daidzein or genistein concentrations and during the same
dietary period (data not shown). Equol occurrence in plasma
and urine also showed high inter-individual variability with
three distinctly different patterns of equol production
evident among subjects. Throughout the study, 12 subjects
consistently had plasma and/or urine equol, 10 subjects had
equol in a single plasma or urine sample and 9 subjects never
produced equol. There were no significant differences
between these subgroups in baseline lipids, nor any of the
lipid effects of any of the dietary periods (data not shown).

Discussion
In this study, 5 weeks of soy plus probiotic intake
significantly decreased total cholesterol by 4.772.0%
(0.3270.14 mmol l1). This is similar to other studies of
moderately and hypercholesterolaemic subjects, reporting
European Journal of Clinical Nutrition

significant reductions in total and LDL cholesterol of up to 6

and 10%, respectively with intake of probiotic milk products
(Agerbaek et al., 1995; Richelson et al., 1996; Schaafsma et al.,
1998; Agerholm-Larsen et al., 2000b; Xiao et al., 2003), as
well as the results of a small meta-analysis (Agerholm-Larsen
et al., 2000a). Greany et al. (2004) also examined the effects
of combined soy and probiotic intake on lipids; however in
their study, intake of soy protein isolate over 6 weeks
significantly reduced total and LDL cholesterol in hypercholesterolaemic subjects by 5 and 6%, respectively, while
concurrent intake of 109 CFU L. acidophilus and Bifidobacterium bifidum had no additional benefit. It is possible that the
significant hypocholesterolaemic effect of soy protein isolate
was maximal and that combination of soy and probiotic
showed no further improvement.
In vitro, probiotic bacteria including L. acidophilus and
B. bifidus can assimilate cholesterol (Gomes and Malcata,
1999) and deconjugate bile salts (Pereira and Gibson, 2002).
Probiotic bacteria may also assimilate dietary cholesterol,
reducing its absorption (Pereira and Gibson, 2002; Xiao et al.,
2003) and increase bile acid excretion, further reducing serum
cholesterol as it is converted to replace the excreted bile acids
(Sanders, 2000; Pereira and Gibson, 2002). However, these
effects are dependent on the ingested bacterial strains
colonizing the small intestine where most cholesterol absorption takes place (Pereira and Gibson, 2002).

Lipid effects of soy with a probiotic or prebiotic

TA Larkin et al

243
Table 3 Lipid effects of each dietary period
Lipid level (mmol l1)
P

Change in lipid
(mmol l1)

(%)

0.324
0.038
0.250
0.003

0.1270.12
0.3270.14
0.0970.08
0.3670.10

1.571.7
4.772.0
1.171.8
5.571.6

4.4970.31
4.3670.33
4.0170.15
4.1170.19

0.192
0.133
0.030
0.005

0.1470.10
0.1770.10
0.2170.09
0.3270.10

2.572.6
3.972.4
4.172.1
7.372.2

1.2370.10
1.2470.09
1.2970.07
1.2970.07

1.2270.10
1.1570.08
1.2770.08
1.2470.05

0.663
0.084
0.582
0.200

0.0170.03
0.0970.05
0.0270.03
0.0470.03

0.372.8
6.573.9
1.772.5
2.172.4

1.9970.32
2.0570.23
1.9370.17
1.9770.15

2.2070.22
2.0870.29
2.2270.24
1.9870.14

0.334
0.834
0.153
0.946

0.2170.21
0.0370.15
0.2970.19
0.0170.15

Week 0

Week 5

TCh
S YC
S YP
SC
S RS

6.7470.32
6.7070.30
6.3970.18
6.6270.15

6.6270.30
6.3870.32
6.2970.13
6.2670.18

LDL
S YC
S YP
SC
S RS

4.6470.33
4.5270.30
4.2270.19
4.4470.17

HDL
S YC
S YP
SC
S RS
TG
S YC
S YP
SC
S RS

22.3710.7
2.076.0
17.2710.0
4.877.7

Abbreviations: Prebiotic cohort (n 15): SC, soy control; S RS, soy resistant starch; Probiotic cohort (n 15): S YC, soy control yoghurt; S YP,
soy probiotic yoghurt; TCh, total cholesterol; TG, triglycerides.
Mean7s.e.m.
a
Paired t-tests comparing weeks 0 and 5 of each treatment.

Table 4 Mean plasma and urinary isoflavones at the end of each

5-week dietary period
Daidzein
Plasma (ng ml1)
S YC
S YP
SC
S RS

45.0716.8
66.2717.8
43.1712.9
45.7710.4

Urine (mg)
S YC
S YP
SC
S RS

17.872.1
19.072.1
18.173.2
18.573.6

(0260)
(4.1237)
(0162)
(0133)

(4.038.1)
(5.634.5)
(2.057.9)
(5.961.3)

Genistein

158738.8
201745.5
197747.4
153727.7

10.872.2
9.971.5
8.871.6
9.771.5

(0514)
(46.1682)
(58.3598)
(0355)

(2.738.3)
(0.8123.9)
(0.6022.9)
(2.621.9)

Abbreviations: S, soy; S YC, soy control yoghurt; S YP, soy probiotic

yoghurt; SC, soy control; S RS, soy resistant starch; RS, resistant starch
bread.
Plasma: fasted morning sample; urinary: 48 h sample following test soy meal.
Mean7s.e.m., range in parentheses.

Our finding that 5 weeks intake of soy plus resistant

starch significantly reduced total and LDL cholesterol
by 5.571.6% (0.3670.10 mmol l1) and 7.372.2%
(0.3270.10 mmol l1), respectively, is of particularly interest. There are few human studies in this area; however,
similar findings indicate a potential benefit of combining
prebiotic and soy intake. Total and LDL-cholesterol levels

were also significantly reduced after daily intake of soy

protein and 18 g of resistant starch for 3 weeks (Muir et al.,
1998) and 33 g soy protein and 8 g soluble fibre for 1 month
(Jenkins et al., 1999). Furthermore, 2 weeks daily intake of
21.5 g high amylose corn starch, similar to resistant starch,
was associated with lower total and LDL cholesterol (Jenkins
et al., 1999) and animal studies demonstrate significantly
reduced lipid levels with prebiotic intake (Taylor and
Williams, 1998). Prebiotic intake directly affects the carbohydrate fermentation activity of the microflora, which itself
has also been suggested to be related to lipid profiles (Jenkins
et al., 1999). Production of propionate may decrease serum
cholesterol (Jenkins et al., 1999) and this is one of the major
short chain fatty acid products of intestinal fermentation of
resistant starch (Brown et al., 1998; Rowland, 1999; Topping
and Clifton, 2001).
In the current study, total cholesterol remained lower 4
weeks after soy plus resistant starch intake ceased. When soy
plus resistant starch was consumed first, there was a mean
reduction in total cholesterol (0.1870.07 mmol l1) during
this 5-week dietary period and then a further mean reduction
(0.1570.10 mmol l1) during the 4-week washout. In contrast, no other groups showed a mean decrease in total
cholesterol during the washout period, suggesting a longlasting effect of resistant starch intake in relation to lipid
lowering. The convenient incorporation of resistant starch
into bread (Brown, 1996; Hoebler et al., 1999) allows for
potential functional food developments with soy in relation
to hypercholesterolaemia.
The only significant effect of the soy control diet was a
reduction in LDL of 4.172.1% (0.2170.09 mmol l1) in the
prebiotic cohort. During all dietary periods, the soy foods
provided 38 mg daidzein, 68 mg genistein, 4.5 mg glycitein
(aglycone equivalents) and 10 g soy protein per day. The lack
of significant lipid reductions may support the suggestion
that a minimum amount of soy protein is necessary for lipidlowering effects. The 1999 health claim for the hypocholesterolaemic effect of soy protein (FDA, USA) was based on a
daily intake of 25 g soy protein, and three meta-analyses
reporting reductions in total cholesterol of approximately
4% or more were based on soy protein intakes of between
22.5 and 47 g per day (Anderson et al., 1995; Weggemans and
Trautwein, 2003; Harland and Carr, 2004). The significant
lipid-lowering effects of the combination of soy with a
probiotic or a prebiotic were not associated with changes in
isoflavone bioavailability or equol production. Results of
probiotic and prebiotic effects on isoflavone bioavailability
are reported elsewhere (TA Larkin, submitted manuscript).
The lipid-lowering effects of soy intake may depend on
initial cholesterol levels. In the original meta-analysis
(Anderson et al., 1995), only subjects with total cholesterol
greater than 6.7 mmol l1 achieved significant lipid reductions with soy intake. In the present study, 10 subjects had
total cholesterol greater than 6.7 mmol l1 (mean
7.6370.30 mmol l1) at the beginning of their control soy
dietary period and this significantly decreased during the
European Journal of Clinical Nutrition

Lipid effects of soy with a probiotic or prebiotic

TA Larkin et al

244
control soy period by 0.3270.13 mmol l1 (4.171.8%;
P 0.04, Students paired t-test). Thus, similar to Andersons
meta-analysis, soy intake resulted in significant lipid lowering in subjects with baseline total cholesterol greater than
6.7 mmol l1. In the current study, none of the treatments
significantly affected triglyceride or HDL-cholesterol levels,
consistent with literature (Agerbaek et al., 1995; Richelson
et al., 1996; Schaafsma et al., 1998).
There was also an overall temporal effect, of significantly
reduced total and LDL cholesterol over the 14-week study
duration. Considering the trend towards a reduction in total
and LDL cholesterol with the soy control and the lack of
significant difference between the treatment and control for
both cohorts, the significant effects of combination with the
probiotic or prebiotic appear to be additive rather than solely
attributable to either of these latter components. However,
subject numbers were small and as the study design did
not include control probiotic or prebiotic treatments, the
proportional effect of each component cannot be assessed.
Further examination of the effects reported here, particularly
long-term studies, would be beneficial, as would inclusion of
microbial analyses to determine whether these effects are
related to specific changes in gut microflora activity. Overall,
our results suggest limited lipid-lowering effects of soy
consumption and potential synergistic hypocholesterolaemic action between soy and probiotic bacteria.

Acknowledgements
We acknowledge financial support from The Australian
Research Council, Specialty Cereals Pty Ltd and the ARC
Smart Foods Key Centre at the University of Wollongong.
The following companies provided foods for the dietary
interventions: Specialty Cereals Pty Ltd, Mt Kuring-gai, New
South Wales, Australia; So Natural Foods, Taren Point, New
South Wales, Australia; Vaalia, Pauls Dairy, Brisbane, Qld,
Australia; Penford Australia Ltd, Sydney, New South Wales,
Australia.

References
Adlercreutz H (1998). Evolution, nutrition, intestinal microflora,
and prevention of cancer: a hypothesis. Proc Soc Exp Biol Med 217,
241246.
Agerbaek M, Gerdes L, Richelson B (1995). Hypocholesterolaemic
effect of a new fermented milk product in healthy middle-aged
men. Eur J Clin Nutr 49, 346352.
Agerholm-Larsen L, Bell M, Grunwald G, Astrup A (2000a). The effect
of a probiotic milk product on plasma cholesterol: a meta-analysis
of short-term intervention studies. Eur J Clin Nutr 54, 856860.
Agerholm-Larsen L, Raben A, Haulrik N, Hansen A, Manders M,
Astrup A (2000b). Effect of 8 week intake of probiotic milk
products on risk factors for cardiovascular diseases. Eur J Clin Nutr
54, 288297.
Anderson J, Johnstone B, Cook-Newell M (1995). Meta-analysis of
the effects of soy protein intake on serum lipids. N Engl J Med 333,
276282.

European Journal of Clinical Nutrition

Brown I (1996). Complex carbohydrates and resistant starch. Nutr Rev

54, S115S119.
Brown I, Wang X, Conway P (1999). Sticky and selective starches.
Microbiol Aust 9, 1819.
Brown I, Wang X, Topping D, Playne M, Conway P (1998). High
amylose maize starch as a versatile prebiotic for use with probiotic
bacteria. Food Aust 50, 602609.
Crouse J, Morgan T, Terry J, Ellis J, Vitolins M, Burke G (1999). A
randomized trial comparing the effect of casein with that of
soy protein containing varying amounts of isoflavones on
plasma concentration of lipids and lipoproteins. Arch Intern Med
59, 20702076.
Cummings J, Roberfroid M, Andersson H, Barth C, Ferro-Luzzi A,
Ghoos Y et al. (1997). A new look at dietary carbohydrate:
chemistry, physiology and health. Eur J Clin Nutr 51, 417423.
Friedewald W, Levy R, Fredickson D (1972). Estimation of the
concentration of LDL cholesterol in plasma without the use of the
preparative ultracentrifuge. Clin Chem 18, 499502.
Gamache PH, Acworth IN (1998). Analysis of phytoestrogens and
polyphenols in plasma, tissue, and urine using HPLC with
coulometric array detection. Proc Soc Exp Biol Med 217, 274280.
Gibson G (1998). Dietary modulation of the human gut microflora
using prebiotics. Br J Nutr 80, S209S212.
Gibson G (1999). Dietary modulation of the human gut microflora
using the prebiotics oligofructose and inulin. J Nutr 129, S1438
S1441.
Gomes A, Malcata F (1999). Bifidobacterium spp. and Lactobacillus
acidophilus: biological, biochemical, technological and therapeutical properties relevant for use as probiotics. Trends Food Sci Tech
10, 139157.
Greany K, Nettleton J, KE W, Thomas W, Kurzer M (2004). Probiotic
consumption does not enhance the cholesterol-lowering effect of
soy in postmenopausal women. J Nutr 134, 32773283.
Harland J, Carr T (2004). Does a practical daily intake of B25 g soy
protein significantly lower cholesterol? A meta-analysis of recent
studies. J Nutr 134, S1267.
Hentges D (1980). Does diet influence human fecal microflora
composition. Nutr Rev 38, 329336.
Hodgson J, Puddey I, Beilin L, Mori T, Croft K (1998). Supplementation with isoflavonoid phytoestrogens does not alter serum lipid
concentrations: a randomized controlled trial in humans. J Nutr
128, 728732.
Hoebler C, Karinthi A, Chiron H, Champ M, Barry J (1999).
Bioavailability of starch in bread rich in amylose: metabolic
responses in healthy subjects and starch structure. Eur J Clin Nutr
53, 360366.
Jenkins D, Vuksan V, Rao A, Vidgen E, Kendall C, Tariq N et al. (1999).
Colonic bacterial activity and serum lipid risk factors for
cardiovascular disease. Metabolism 48, 264268.
Larkin TA. Soy isoflavone bioavailability: effects of probiotic and
prebiotic consumption and oil supplementation. PhD Thesis,
University of Wollongong: Australia, 2005.
Lichtenstein A (1998). Soy protein, isoflavones and cardiovascular
disease risk. J Nutr 128, 15891592.
Merz-Demlow B, Duncan A, Wangen K, Xu X, Carr T, Phipps W et al.
(2000). Soy isoflavones improve plasma lipids in normoc
holesterolemic, premenopausal women. Am J Clin Nutr 71,
14621469.
Muir J, Walker K, Kaimakamis M, Cameron M, Govers M, Lu Z et al.
(1998). Modulation of fecal markers relevant to colon cancer risk:
a high-starch Chinese diet did not generate expected beneficial
changes relative to a Western-type diet. Am J Clin Nutr 68,
372379.
Nestel P, Yamashita T, Sasahara T, Pomeroy S, Dart A, Komesaroff P
et al. (1997). Soy isoflavones improve systemic arterial compliance
but not plasma lipids in menopausal and perimenopausal women.
Arterioscler Thromb Vasc Biol 17, 33923398.
OSullivan D (2001). Screening of intestinal microflora for effective
probiotic bacteria. J Agric Food Chem 49, 17511760.

Lipid effects of soy with a probiotic or prebiotic

TA Larkin et al

245
Parodi P (1999). The role of intestinal bacteria in the causation and
prevention of cancer: modulation by diet and probiotics. Aust J
Dairy Tech 54, 103121.
Pereira D, Gibson G (2002). Effects of consumption of probiotics and
prebiotics on serum lipid levels in humans. Crit Rev Biochem Mol
Biol 37, 259281.
Potter S (1998). Soy protein and cardiovascular disease: the impact of
bioactive components in soy. Nutr Rev 56, 231235.
Richelson B, Kristensen K, Pedersen S (1996). Long-term (6 months)
effect of a new fermented milk product on the level of plasma
lipoproteinsa placebo-controlled and double blind study. Eur J
Clin Nutr 50, 811815.
Rowland I (1999). Optimal nutrition: fibre and phytochemicals.
Proc Nutr Soc 58, 415419.
Rowland I, Wiseman H, Sanders T, Adlercreutz H, Bowey E (1999).
Metabolism of oestrogens and phytoestrogens: role of the gut
microflora. Biochem Soc Trans 27, 304308.
Sanders M (2000). Considerations for use of probiotic bacteria to
modulate human health. J Nutr 130, S384S390.
Schaafsma G, Meuling W, van Dokkum W, Bouley C (1998). Effect of
a milk product, fermented by Lactobacillus acidophilus and with
fructo-oligosaccharides added, on blood lipids in male volunteers.
Eur J Clin Nutr 52, 436440.
Silvi S, Cresci R, Rowland I (1999). Resistant starch modifies gut
microflora and microbial metabolism in human flora-associated
rats inoculated with faeces from Italian and UK donors. J Appl
Microbiol 86, 521530.
Steinberg F, Guthrie N, Villablanca A, Kumar K, Murray M (2003).
Soy protein with isoflavones has favorable effects on endothelial
function that are independent of lipid and antioxidant

effects in healthy postmenopausal women. Am J Clin Nutr 78,

123130.
Taylor G, Williams C (1998). Effects of probiotics and prebiotics on
blood lipids. Br J Nutr 80, S225S230.
Topping D, Clifton P (2001). Short-chain fatty acids and human
colonic function: roles of resistant starch and nonstarch polysaccharides. Physiol Rev 81, 10311064.
Turner N, Thomson B, Shaw I (2003). Bioactive isoflavones in
functional foods: the importance of gut microflora on bioavailability. Nutr Rev 61, 204213.
Wangen K, Duncan A, Xu X, Kurzer M (2001). Soy isoflavones
improve plasma lipids in normocholesterolemic and mildly
hypercholesterolemic postmenopausal women. Am J Clin Nutr
73, 225231.
Warnick G, Benderson J, Albers J (1982). Dextran sulfate-Mg2
precipitation procedure for quantitation of high-density lipoprotein cholesterol. Clin Chem 28, 13791388.
Weggemans R, Trautwein E (2003). Relation between soy-associated
isoflavones and LDL and HDL cholesterol concentrations in
humans: a meta-analysis. Eur J Clin Nutr 57, 940946.
Winter J, Bokkenheuser V (1987). Bacterial metabolism of natural
and synthetic sex hormones undergoing enterohepatic circulation. J Steroid Biochem 27, 11451149.
Xiao J, Kondo S, Takahashi N, Miyaki K, Oshida K, Hiramatsu A et al.
(2003). Effects of milk products fermented by Bifidobacterium
longum on blood lipids in rats and healthy adult male volunteers.
J Dairy Sci 86, 24522461.
Yeung J, Yu T (2003). Effects of isoflavones (soy phyto-estrogens)
on serum lipids: a meta-analysis of randomised controlled trials.
Nutr J 2, 1.

European Journal of Clinical Nutrition