Low-glycemic

index diet in hyperlipidemia:

traditional starchy foods13

use of

David JA Jenkins,
MD; Thomas
MS Wolever,
MD; Janet Kalmusky,
1W; Silvia Guidici,
Connie Giordano,
RD; Robert
Patten, MD; Gerald S Wong, MD; Josephine
N Bird, MD;
Murray
Hall, MD; Gloria Buckley,
MSc; Adele Csima,
PhD; and J Alick Little, MD

RD;

Nutr

1987;46:66-7l.

KEY

WORDS

Hyperlipidemia,

serum

triglyceride,

Introduction

A number
in the

ofdietary

management

changes

have

been recommended
(1). They
include
to achieve
ideal body weight,

of hyperlipidemia

reduction
in caloric intake
reduction
in total fat and cholesterol
intake
with an increase in polyunsaturated:saturated
(P:S) ratio of fatty
acids especially
from o-3 fatty acids (2, 3), consumption
of vegetable
rather than animal
protein
(4, 5), and also
the

use

of certain

(6, 7) but

not

all (8)

forms

low

blood

lipids

and

comparative

freedom

from

cardiovascular
disease. A feature common
to a large proportion
of these societies
is their relatively
high carbohydrate
intake.
The carbohydrate
foods
(eg, legumes,
pasta, parboiled
cereal grains, etc) differ from those popular in western
industrialized
societies
and often produce
lower blood glucose
responses
(16). Because
it has been
suggested
that the accompanying
reduction
in insulin 1evels may be advantageous
in reducing
the stimulus
to hepatic lipogenesis
(17), we have looked
at the effects of
such foods in hyperlipidemic
patients.
In a recent study (16) we confirmed
that a selection
of
66

Am J C/in Nuir

starchy

foods

traditional
starchy
foods with low glycemic
indices
(GI)
appeared
to modify
blood-lipid
profiles
favorably
when
incorporated
into the diets of 12 hyperlipidemic
patients.
The aim of this study is to extend
these observations
to
determine

abnormalities

might

combination

of foods

gives

Thirty hyperlipidemic
patients
who had been followed
Lipid Clinic over the previous
4.7 0.6 yr were invited
ticipate
in a 3-mo study approved
by the University

in the
to parEthics

to this
the

best

what

types

maneuver

and

oflipid

what

respond

result.

Methods

of dietary

fiber. Modification
ofdietary
carbohydrate
has also been
recommended
but studies
of these diets have given less
clear results (9-15).
Many of the clinical
studies on which current
recommendations
are based were stimulated
by the findings
of
epidemioiogical
surveys
involving
societies
with traditionally

traditional

1987;46:66-7

Patients

I From the Departments

ofNutritional
Sciences (DJAJ,
SO, CO) and Biostatistics
(AC), Faculty of Medicine
and
of Endocrinology
and Metabolism
(DJAJ, TMSW,
JK,
OSw, JNB, MH, JAL) and the Department
of Clinical
(RP, GB), St Michaels Hospital, University
of Toronto,
tario, Canada.

TMSW, JK,
the Division
SO, CO, RP,
Biochemistry
Toronto,
On-

2 Supported
by grants from the Natural Sciences and Engineering
Research Council and the Quaker Oats Company.
3 Address
reprint requests to David JA Jenkins, MD, Department
of

Nutritional
Sciences,
Faculty of Medicine,
University
ronto, Ontario
M5S lA8, Canada.
Received
March 3, 1986.
Accepted
for publication
August
19, 1986.
1 . Printed

in USA.

C 1987 American

Society

of Toronto,

for Clinical

To-

Nutrition

Downloaded from www.ajcn.org by guest on October 25, 2011

ABSTRACT
To define those patients most likely to benefit from the hypolipidemic
effect
of low-glycemic-index
(GI) traditional
starchy
foods, 30 hyperlipidemic
patients
were studied
for 3 mo. During the middle month,
low-GI foods were substituted
for those with a higher 01
with minimal
change in dietary macronutnent
and fiber content.
Only in the group (24 patients)
with raised triglyceride
levels (types lib, III, and IV) were significant
lipid reductions
seen: total
cholesterol 8.8 1 .5% (p < 0.001), LDL cholesterol
9. 1 2.4% (p < 0.001), and serum triglyceride
19.3 3.2% (p < 0.001)
with no change
in HDL cholesterol.
The percentage
reduction
in
serum triglyceride
related to the initial triglyceride
levels (r
0.56, p < 0.01). The small weight
loss (0.4 kg) on the low-GI diet did not relate to the lipid changes.
Low-GI
diets may be of use
in the management
oflipid
abnormalities
associated
with hypertriglyceridemia.
Am J Clin

STARCHY

IN

FOODS

Committee

and involving the use of low-GI foods during the

month.
Data on 12 ofthese
patients
have been reported
previously
(16). All patients
had been
classified
according
to
Frederickson
(18, 19) by their lipoprotein
pattern
on agarosegel electrophoresis,
both in whole plasma
and in supernatant
and infranatant
fractions,
after ultracentrifugation
at density
1 .006 g/mL (20). Patients classed as Type II were further divided
into Type II a or b depending
on whether their mean fasting
triglyceride
levels were below or above 2 mmol/L
(177 mg/i#{174}
mL) respectively (Table 1). The volunteers selected for the study
middle

were those

whose

mo or more

lipid results

had been stable over the past 2

and who had already

with a standard

Lipid

Clinic

shown

reasonable

compliance

diet. Their

diets were low in fat

of 208 23 mg/d and a mono- and

with a cholesterol
content
disaccharide
content
of2O 1.3% oftotal
calories.
Two-thirds
of the patients
took either no alcohol or < 2% of calorie
intake
as alcohol.
In
cases alcohol intake was kept at the same level
throughout
the study.

67

HYPERLIPIDEMIA

individually.
An exchange list was used to indicate
which foods in the control diet were to be substituted
for low01 foods. The foods listed in both the control and low-GI lists
were given in quantities
ofequal
carbohydrate
and caloric content to ensure similar composition
of the diets in both control
and test periods.
In addition,
problems
regarding the incorporation ofthe low-GI foods into diets were discussed at the 2-wk
visits. This was done in conjunction
with the body-weight
mensurement to ensure that weight changes were minimal.
The GIs ascribed to the foods used for exchange purposes
have been described
previously
(16). The food types for which
exchanges
were made included breads, breakfast
cereals,
root
vegetables, cereal grains and products, and legumes. Where possible foods commonly
found in the control diet were exchanged
developed

for foods with a lower 01 that

meal pattern
(eg, pumpernickel

could substitute
for wholemeal

into the same

bread; barley,

and spaghetti for potato and unconverted

rice; etc). Where
there was no low-GI equivalent,
as with root vegetables,
the
exchange was made with legumes. The dietary content of other
foods was held constant
throughout
the study.

bulgar,

Analyses

by mouth.

Fasting blood samples

terol (23) and triglycerides

Study

were analyzed
for high-density
lipoprotein
cholesterol
(HDL)
after heparin-manganese
precipitation
(25, 26). In a number
of

protocol

The aim was to reduce

the glycemic
effect ofthe diet by 20%
month during a 3-mo study while not
or fiber content
of the diet. This was

over the middle (low-GI)

altering the macronutrient

to be achieved by substituting
low-GI starchy foods into the diet
to replace more routinely
used starchy foods. During the 1st
and 3rd (control)
months,
patients consumed their normal Lipid
diet under close supervision.
Throughout
the entire period
patients
were seen in the clinic at 2-wk intervals
when they gave
fasting blood samples
after 14-h overnight
fasts. At the same
time body weight was measured
and diet history recorded over
the previous
week was collected.
This was used to calculate flutrient intakes (22). In addition,
for the last 13 patients
to enter
the study, 24-h urine collections
were made on the final day of

given

compliance,

dietary

to participants.

Sample

individual

instruction

packets

oflow-GI
foods
period to familiarize

to subjects

was

were given

at the start ofthe test

them with
the new foods. Cooking
instructions
and recipes for dishes using
low-GI were supplied and, where necessary, sample menus were
TABLE I
Details of patients

at commencement

Lipid

Sex

type

ofthe

this technique
did not allow the satisfactory
estimation
cholesterol
due to turbidity
that could not be cleared
by centzifugation.
Low-density
lipoprotein
(LDL) cholesterol was

of HDL

calculated

with

LDLcholesterol

the formula
=

total cholesterol

2M

-(HDL

cholesterol

+ [triglycende/2.2])

The formula was only applied where serum

triglyceride
levels
were < 4.5 mmol/L since above this figure the formula is inappropriate
(28). Serum
fructosamine
was determined
chemically by a method
analyzer,
Multistat

(29)

proteins

mean

adapted

for use on a microcentrifugal

III (Instrumentation
Laboratory
Ltd, Lexington, MA). This test was carried out on sera from the last six
patients to enter the study to measure
glycosylation
of plasma
and hence

blood

glucose

experimental
period.
Urinary
insulin
were estimated by radioimmunoassay
indicators
of 24-h insulin secretion.

Trigiyceride

Cholesterol

yr
lIa

(27)

concentrations
over the
(30) and c-peptide
(31)
on the last 13 patients as
Urinary

creatinine

was

study
% Ideal
body weightt

Age

for total serum cholesofserum stored at -20#{176}

instances

Clinic

each period.
To maximize

were analyzed
(24). Aliquots

mg/dL

mg/dL

Glucose

2hpc

Follow up

mg/dL

mg/dL

yr

475

974

32115

13318

955

1312.3

3.81.5

514

1004

29412

25727

974

11918

6.21.8

4F
lib

6M

lF
III
IV
S

1 M

57

99

l3M
552
1063
3F
Two Ila and one lib patient were taking cholestyramine

taking other hypolipidemic

t From reference 21.

:1:Serum

and blood

values

302

337

2248

34627

before the study; their doses were maintained

agents.
obtained

before

commencement

ofthe

study

period.

92

1069

124

14011

over the study period.

5.3

4.10.8
No patients

were

Downloaded from www.ajcn.org by guest on October 25, 2011

At their initial attendance

at the Lipid Clinic, all patients
were screened
to exclude biochemical
evidence ofthyroid,
liver,
and renal disease. Carbohydrate
tolerance was assessed by measurement
ofblood
glucose fasting and 2 h aftertaking
75 gglucose

68

ET AL

JENKINS

TABLE 2
Comparison

of control

and experimental
Control

Mean

Measurement

diet

83.8

0.6

48.9

1 .268

18.6

0.881

Starch*
Protein*
Fat*

30.4
19.4
28.6

P:S ratio

0.6

Cholesterol

Oat + barley
Beans
Wheat + corn

1.4

0.7

0.2

2.6

0.5

Leafy vegetables

Percent

0.7
3.4
17 15

68.8

0.8

<

p
p

<
<

p <

5.9

1 .4

p <

2.5

0.4

p <

4.0

1.0

p <

-0.2 0.7
-0.1 0.2
0.03 0.25
-83 30
-0.2 0.1

p
p

<
<

0.0001
NS
NS
0.05
0.0001
NS
0.05
0.001
0.0001
0.001
NS
NS
NS
0.05
0.05

of daily calories.

Blood
measured
to check
tions (32).

for completeness

of the 24-h

urine

included

difficulty

in

obtaining

certain

of

the

more useful foods, such as pumpernickel

bread, on a regular basis; the need for a more expanded
list of low-GI
foods;
and, to a more variable
extent,
a small increase
in
flatulence.
None ofthese
prevented
completion
ofthe low01 month
with the exception
of one patient
who completed only 2 wk due to difficulty
in obtaining
the foods.
In general,
even where significant,
the differences
in
macronutrient
proportions
between
the high-GI
and control diets were small. Fat calories
were reduced
by 2.7%
in the low-GI
period with an accompanying
increase
in
carbohydrate
(Table
2). The P:S ratio remained
unchanged,
cholesterol
intake fell from 224 to 187 mg/d and
fiber consumption
increased
by 5.9 g/d (Table 2). In addition there was a small overall mean reduction
in body
weight
of 0.23 kg in the low-GI
period
by comparison
with the control
period,
which represented
a 0.42 kg fall
across the low-GI month.

p < 0.05

0.5
1.3 0.8
0.7 0.3
-2.7 0.6
-0.01 0.03
-36 14

0.8
0.2
0.8
80
1.7

patients

low-GI diets since the

commented
on by the

collec-

lipids

During

the low-GI
month
serum
lipids
were
sigreduced
for triglyceride,
15.3 3.0% (n = 30,
p < 0.001);
total cholesterol,
7.7 1.3% (n = 30, p
< 0.001);
and LDL cholesterol,
8.5 2.0% (n = 27, p
< 0.001)
(Table 3). There was no change
in HDL cholesterol. No significant
differences
in serum cholesterol
or
triglyceride
were seen between
the mean pretest
control
and posttest
control
period.
The mean values for serum
total cholesterol
were pretest control
period,
week 2, 267
12 mg/dL,
and week 4, 259 8 mg/dL;
low-GI period,
week 2, 244 12 mg/dL,
and week 4, 248 12 mg/dL;
and posttest
control
period, week 2, 267 12 mg/dL,
and
week 4, 267 12 mg/dL.
The corresponding
values for
serum
triglyceride
over these periods
were 337 44 mg/
nificantly

Results were expressed as mean SEM. Two- and 4-wk values

of each of the three periods
(pretest
control,
test, and posttest
control)
for each individual
were combined
to give a single value
for each period.
The average
of the two control period means
was compared
with the mean ofthe low-GI period
samples using
Students
t test for paired data. Similar
comparisons
were made
between the study mean and pre- and posttest control
means
separately
for all variables
under study. In addition,
simple and
multiple
regression
analysis was used to identify
the characteristics ofpatients
or diets that related best to the percent reduction
in their blood lipids. Absolute
and percent changes in lipids and
weight
were related to the absolute
levels and changes in dietary
factors.
Results

dL and 284 18 mg/dL; 239 27 mg/dL

and 257 18
mg/dL;
and 3 19 35 mg/dL
and 292 35 mg/dL.
No

Within the constraints

ofLipid
Clinic diets as currently
recommended,
our patients
were able to achieve
a mean
GI reduction
of 1 1. 1 0.8 units (p < 0.001).
Such diets
were considered
acceptable
for longer-term
use and 18

significant
differences
levels.
The effect of type
amined
in subgroups

TABLE 3
Mean changes
(values

in blood lipids on the low-glycemic

index diet in all patients

and in subgroups

were

noted

in fasting

of hyperlipidemia
of patients
(Table

with the same lipoprotein

blood-glucose

and
3).

diet

was ex-

phenotype

in mg/dL)*
Total

cholesterol

LD L cholesterol

Test
Control
Patientgroups

% change

MeanSEM

HDL

cholesterol

Test

Control

SEM

mg/dL

% change

MeanSEM

Thglyceride

Test

Control
p

SEM

mg/dL

MeanSEM

Test

% change
SEM

% change

Control
p

n*gJdL

McanSEM

SEM

n*g/dL

All patients
I =

30; for

LDL
n

cholesterol
27

263

Typellan=6
Type

lIb n - 7

Type

111

Type

IV n S

317

,i

Classified

304

-7.7

12

-3.32.4

15

-7.0

294

16

228
acco iding

.8

to Fredrickson

1.3

<

1.7

0.001

166

NS

236

<0.01

19.9
-8.9

2.0

et al (19).

12

-8.5

15

-6.33.1

197 15

-9.5

221
<

0.001

128 .8

2.0

3.8

0.001

43 4

NS

310

NS

548

7.55.9

NS

124

<

0.05

50 8

1.2 4.6

NS

284

<

0.05

39 .2

NS

390

<

-28.7
-7.4

3.1

-0.6

35

2.2

-1
-4.9

1.1
2.6

27
18
18

-15.3
-17.5

310

3.0

<

0.001

5.9

<

0.05

4.1

<0.001

-0.53.5

NS

-28.7
35

-19.6

Downloaded from www.ajcn.org by guest on October 25, 2011

Alcohol (g/d)
Calories
Body wt/kg

Significance
of difference

0.5

224

10.2

diet
SEM

1.9 0.8

2 1 .8

Change
-1 1.1

1.335
0.436
1.192
0.05
20

(mg/d)

Fiber (gJd)

Experimental

SEM

01
Carbohydrate
Sugars*

patients
have returned
to modified
end of the study. The problems

diets

STARCHY

FOODS

IN

Normal
Triglycerides

Raised

Triglycerides

the

diet

was

associated

low-GI

Cl)

a.
1

=
a,

cholesterol
Effect

Effect

oflipid

Division
phenotypic

lipids

(Table

3).

were

calculated

including

only

those

value

on the

low-GI

in diet

and

ofchanges

diet.
body

weight

on lipid

levels

of the patients
classification

glyceride

types

levels

TABLE

into subgroups
indicated
that

are

raised,

in blood

lipids

lIb,

III,

and

reducing

whose
mean dietary
cholesterol
intakes
(+13 6 mg/d,
ie, < 10 mg/d reduction
in cholesterol
intake,
n = 14) (Table 4). When fiber intakes
increased
by > 10 g/d (mean
17.3 3.0 g/d, n = 7, p
< 0.01)
the mean
lipid changes
were also similar
to those

IV,

in which

glycemic

on the low-glycemic

index

tn-

index

of

Mean

changes

Total

diet in all patients

cholesterol
Test

groups

Patient
n
No weight

Mean

SEM

and in specific

subgroups

LD L cholesterol

HDL

Control

SEM

Mean

SEM

in mg/dL)
Triglyceride

Test

% change
SEM

Mean

SEM%

Test

% change

Control

mg/dL

mg/dL

(values

cholesterol

Test

% change

Control

SEM

% change

Control
p

msJdL

SEM%

Mean

SEM

mg/dL

loss

P1

those

of the group
as a whole (Table
4). In addition
in the
subgroup
which showed
no decrease
in body weight over
the low-GI month
(+0.34
0. 10 kg, n = 1 2, p < 0.01)
significant
falls in both serum triglyceride
and total cholesterol were still recorded
(Table 4). In addition
if the 1mo period
of maximum
weight
loss over the previous

based on their
in patients
with

the

as did

were not reduced

class

dislipoproteinemia

267

15

-7.4

232

15

-7.0

263

15

-8.1

2.6

<

0.05

170

<

0.001

128

<

0.01

23

-8.6

3.0

<

0.05

42

-1 1.4

13.2

<

0.05

38

159 15

-9.4

3.3

<

0.05

43

190

-7.23.0

-2.0

4.2

NS

346

62

-14.5

5.8

<

0.05

2.2

NS

416

35

-13.5

3.5

<0.001

2.8

NS

337

44

-19.2

16.3

<

NS

32853

0.05

310

53

-13.2

5.1

<

0.05

354

27

-19.3

3.2

<

0.01

lOgincressein
dietary fiber
n

5% total
reduction
7

.4

.8

calorie
in fat

= 9

1% total
reduction

2.3

-0.5

0.01

calorie
in fat

n=9
<

lipids

27515

-6.72.2

<0.02

15

<0.05

508

-5.02.7

-15.04.7

<0.01

10 mg reduction
in dietary
cholesterol
n

14

Mean

267

IS

-5.9

2.2

<

0.02

178 19

-8.3

3.0

<

0.01

43

-7.3

252

12

-8.8

1.5

<

0.01

147 12

-9.1

2.4

<

0.01

42 4

-2.7

2.2

3.1

<

control

triglyceride
>

2.0

mg/dL
24

NS

Downloaded from www.ajcn.org by guest on October 25, 2011

FIG 1. Mean serum lipids, body weight, diet glycemic

index (01),
and daily dietary-fiber
intakes (DF) of the 24 patients
with mixed tnglyceridelevels
and the six patientswith
normal tniglycenidelevels.
These
are shown for the two control (C) and the low 01 (T) months.

<

blood

As mentioned,
the dietary
changes
even where statistically significant
were relatively
small. Ifthe lipid changes
in those who decreased
their fat intake by 5% or more of
total calories (n = 9) are compared
with the values for the
group as a whole then the results can be seen to be cornparable
(Table 4). Similarly,
those whose fat intakes
remaimed unchanged
(0.04 0.22%, ie, < 1% reduction
in
fat intake, n = 12) showed
significant
reductions
in blood

>

diet

lower

changes were seen in patients

with
where serum cholesterol
levels
3). Thus when the lipid reductions

with raised triglyceride

levels (> 2 mmol/L
or 177 mgI
100 mL), even larger mean reductions
were seen: total
serum cholesterol,
8.8 1 .5% (p < 0.001); serum triglyceride, 19.3 3.2% (p < 0.001);
and LDL cholesterol,
9.1
2.4%
(p < 0.0 1) (Table
3). The mean lipid values for
each of the 3 mo of the study in those with and without
raised serum-triglyceride
levels are shown
in Figure
1.
Again no significant
difference
was seen in the mean HDL-

a
E
E

>

with

However,
no consistent
type
Ila dislipoproteinemia
alone are raised (Table
on the

.=

69

HYPERLIPIDEMIA

JENKINS

70

was taken for each patient

then no corresponding
significant
reduction
in serum cholesterol
was noted and
the fall in triglyceride
was small. The mean
maximum
weight loss in any 1-mo period was -1.2 0.1 kg, which
was more than twice that seen in the present
study, while
total serum
cholesterol
fell only by 3 5 rng/dL
and Uiglyceride
by 28 13 mg/dL
(p < 0.05).
year

Regression

analysis

Urinary

insulin

fructosomine

and c-peptide

excretion

and

serum

levels

The decrease
24 h from

the

in urinary
mean

insulin

control

24-h

levels

ofO.8

urinary

1. 1 mU!

insulin

output

of 20. 1 1.3 mU was not significant

in the 13 patients
studied.
Similarly
24-h urinary
C-peptide
levels decreased
from the control
level of 69.3 ng by 3.6 8.1 ng (n = 10)
and plasma fructosamine
levels decreased
from 7.3 0.2%
by 0.1 0.1% (n = 6).
Discussion
This study has confirmed
that a dietary
manipulation
aimed primarily
at reducing
the glycemic
impact
of the
diet through
use of traditional
starchy
foods
instead
of
through
major alterations
in macronutrient
or fiber content may have beneficial
effects on the blood lipids. The
present
results suggest that this approach
may be most
appropriate
in the dietary
management
of patients
with
hypertriglyceridemia
with or without
raised cholesterol
levels rather than in patients
with hypercholesterolemia
alone (Type Ila).
The lack of relationship
between
the change
in blood
lipids and the GI is probably
a reflection
ofthe
similarity
of the diets eaten by the patients.
The change
in GI was
only half the level we hoped to achieve.
The rationale
for this approach
came from the studies
ofAlbrink

et al (17)

who

suggested

that

hepatic

lipogenesis,

especially

with respect to triglyceride

synthesis,
could be
reduced
by minimizing
postprandial-glucose
and -insulin
rises. Metabolic
studies
where emphasis
was placed
on
raising dietary-fiber
intake have demonstrated
that such
diets were associated
with lower blood-lipid
profIles (15,
17, 33,

GI study

34). These
diets were also rich in traditional
foods,
such as legumes,
in addition
to being

low-

high

in fiber.

Where metabolic
studies of high fiber (including
fiber) were undertaken
with no apparent
reduction
in dietary
glycemic
index,
no lowering
effect on serum
lipids was noted (8, 35). Furthermore
in relation
to lowGI foods it has been shown
that diets containing
large
amounts
ofwhole
legumes
reduce 24-h urinary
c-peptide
excretion
in normal
and diabetic
volunteers
indicating
that 24-h insulin
secretion
may be reduced
on such diets
(36). Nevertheless,
in comparison
with these reports
the
GI change induced
in the present studies was modest
and
may explain
why the reductions
in urinary
c-peptide
and
insulin
excretion
did not reach significance.
The fact that such a dietary modification
appeared
most
effective
only in the presence
ofa raised triglyceride
level
was ofinterest
especially
since under these circumstances
the diet reduced
triglyceride,
total-, and LDL-cholesterol
levels. On the other hand, the lack of significant
effect in
the Ila patients
may be due to the operation
of other
mechanisms
such
as reduced
LDL-receptor
function
in
some being primarily
responsible
for the hyperlipidemia
(37) and therefore
unresponsive
to small alterations
in
postprandial
glucose
and insulin
levels.
The dietary exchanges
used in this study involved
substitution
ofsuch
foods as pumpernickel
for wheat breads,
spaghetti
for potato,
and parboiled
for regular
rice and
an increased
consumption
of a range of other low-GI
starchy
foods. In general
these alterations
were achieved
without
large changes
in macronutrient
and fiber contents
ofthe diet. However,
even in subgroups
where significant
nutrient
differences
were observed,
the changes
in blood
lipids in response
to the low-GI diet were similar to those
seen in the rest of the group. Part of the reason for this
may be because
the changes
were relatively
small with
the exception
of the subgroup
who showed
a substantial
mean increase
in their dietary fiber intake. However
even
in this group the increased
fiber intake appeared
to have
come largely from wheat fiber, which has little effect on
blood lipids or glucose tolerance.
Nevertheless,
other viscous fibers are known
to lower serum lipids and
seems
likely that the uniform
increase
in consumption
of these
(eg, oats, barley,
and legumes)
in the group as a whole
did indeed contribute
to the decrease
in cholesterol
in the
low-GI period (15).
Many ofthe patients
with raised triglyceride
levels also
had abnormal
glucose tolerance
(Table 1). This association
has long been recognized
(38, 39). No significant
decrease
was seen in fasting blood glucose over the low-GI period.
However,
it was hoped that if major reductions
in postprandial
glucose
responses
occurred
these would
be reflected in lower levels of glycosylated
blood
proteins
and
reduced
urinary
output
of insulin
and c-peptide.
Fructosamine,
as a measure
of glycosylated
plasma
proteins,
was chosen
because
its half-life
is shorter
than the more
conventionally
used ThAIC,
where 6-8 wk periods
may
be required
to detect significant
changes.
The lack of effect
legume

seen

may

be due

to the relatively

low

mean

fasting

blood-

glucose
levels observed
throughout
the study: only five
patients
had fasting glucose levels > 6.0 mrnol/L
(108 mg/
100 mL) and only a relatively
small percentage
change
in GI was attained.
As might be expected
from this, the

Downloaded from www.ajcn.org by guest on October 25, 2011

Contrary
to expectations,
changes
in GI from control
to test period did not correlate
with changes
in any of the
serum-lipid
values although
the correlation
with percentage change in total-cholesterol
levels was almost significant
(r = 0.35, p = 0.056). Only P:S ratio was related to changes
in both triglyceride
and cholesterol
(r = 0.47, p < 0.01
and r = 0.42, p < 0.05, respectively).
Weight
loss and
differences
in fat and cholesterol
intakes did not correlate
with lipid changes
on the low-GI
diet. Relating
changes
in each of the lipids to several
combinations
of dietary
factors, including
fat and cholesterol
intakes,
in multiple
regression
equations
failed to demonstrate
the consistent
significance
ofany one factor as additional
variables
were
included
in the calculation.

ET AL

STARCHY

FOODS

levels
measured
were
also all in the normal
range and a reduction
from already low levels might have
been difficult
to achieve.
We conclude
that this study suggests that the inclusion
oftraditional
starchy foods with low-GIs
in the lipid-lowering diets ofhypertriglyceridemic
patients
may be a useful
aid to therapy.
It may also be appropriate
in view of the
glucose intolerance
commonly
observed
in such individuals. However,
the exact mechanism
by which blood lipids
were lowered
and their relation
to possible
changes
in
postprandial
nutrient
and endocrine
fluxes remain
to be
be defined.
U
fructosamine

References
1.

Gotto AM, Bierman

EL, Connor

WE, et al. Recommendations

for

8.

9.

Raymond TL, Connon WE, Lin DS, Warner S. Fry MM, Connon
SL. The interaction
ofdietary
fibers and cholesterol
upon the plasma
lipids and lipoproteins,
sterol balance and bowel function
in human
subjects. J Clin Invest l977;60: 1429-37.
Grande F, Anderson
JT, Keys A. Effect ofcarbohydrates
of leguminous seeds, wheat and potatoes
on serum cholesterol
in man. J Nutr
l965;86:3l3-7.

10.

1 1.

12.

13.

14.

15.

16.

Grande F, Anderson
JT, Keys A. Sucrose and various carbohydratecontaining
foods and serum lipids in man. Am J Clin Nutr l974;27:
1043-51.
Ahrens EH, Hirsch J, Oete K, Farquhen
JW, Stein Y. Carbohydrate
induced
and fat induced
lipemia.
Trans Assoc Am Phys 196 l;74:
134-46.
Coulston
AM, Liu GC, Reaven GM. Plasma glucose, insulin and
lipid responses
to high carbohydrate
low fat diets in normal humans.
Metabolism
l983;32:52-6.
Antan MA, Little JA, Lucas C, Buckley
GC, Csima A. Interrelationships
between
dietary carbohydrate
and fat in hyperlipidemic
patients.
Part 3: synergistic effect ofsucrose
and animal fat on serum
lipids. Atherosclerosis
1970;l 1:191-201.
Anderson
JW, Chen WL. Plant fiber carbohydrate
and lipid metabolism.
Am J Can Nutr l979;32:;346-63.
Anderson
JW, Story L, Siding B, Chen WL. Hypocholesterolemic
effects of high-fibre
diets rich in water-soluble
plant fibres. J Can
Diet Assoc l984;45:l40-9.
Jenkins DJA, Woleven TMS, Kalmusky
J, et al. Low glycemic index
carbohydrate
foods in the management
of hyperlipidemia.
Am J
Oin Nutn l985;42:604-17.

17. Albrink MJ, Newman

T, Davidson
diets on plasma lipids and insulin.

PC. Effect ofhigh- and low-fiber

Am J Clin Nutr l979;32:l486-

91.

18. Fredrickson
DS, Lees RS. A system for phenotyping
hyperlipoproteinemia.
Circulation
l965;3l:l-9.
19. Fredrickson
DS, Levy RI, Lees RI. Fat transport
in lipoproteins:
an
integrated approach
to mechanisms
and disorders.
N Engl J Med
l967;276:34-44, 94-103, 148-56, 215-25, 273-81.
20. Nobel RP. Electrophoretic
separation
ofplasma
lipoproteins
in ajarose gel. J Lipid Res l968;9:693-700.
21. Diem K, Lentner
C. Documenta
Geigy scientific tables. SA, Basic:
22.

JRoeigy,
1970.
Paul AA, Southgate

DAT.

McCance

and Widdowsons
1978. (Medical

position offoods.
4th ed. London:
HMSO,
Council special report series no 297.)
23. Allain CC, Poon LS, Chan CS, Richmond
determination
of total serum
cholesterol.

the comResearch

W, Fu PC. Enzymatic
Gin Chem
l974;20:

470-5.
24. Bucolo 0, David H. Quantitative
determination
ofserum
triglyceride
by the use ofenzymes.
Can Chem l973;l9:476-82.
25. US Department
of Health, Education
and Welfare.
Lipid Research
Clinics program:
manual oflaboratory
operations.
Vol 1 Lipid and
lipoprotein
analysis. Washington,
DC: US Government
Printing Office, 1974. (NIH publication
number
75-678.)
26. Lopes-Virella
MF, Stone P. Ellis 5, Colwell JA. Cholesterol
determination
in high-density
lipoproteins
separated
by three different
methods.
Gin Chem l977;23:882-4.
27. Friedwald
WT, Levy RT, Fredrickson
DS. Estimation
ofthe
concentration
oflow density lipoprotein
cholesterol
in plasma without
the use ofthe preparative
ultracentnifuge.
Can Chem 1972;l8:499502.
28. Probstfield
JL, Gotto AM. Lipoproteins
in health and disease: diagnosis and management.
In: Baylor College ofMedicine:
Cardiology
series l982;5:6-27.
29. Lloyd D, Marples J. Simple colonimetry ofglycated
serum protein
in a centrifugal
analyser.
Clin Chem 1984;30:l686-8.
30. Herbert
V, Lau K-S, Gottheb CW, Bleicher SJ. Coated charcoal
immunoassay
ofinsulin.
J Endocninol Metab l965;25:l375-84.
31. Kuzya T, Matsuda
A, Saito T, YOshida S. Human c-peptide immunoreactivity
(CPR) in blood and urine: evaluation
ofa radioimmunoassay
method and its clinical applications.
Diabetologia
1976:
l2;5l 1-8.
32. Faulkner
damentals
1976.
33.

WR, King JW. Renal

of clinical chemistry.

function.
In: Tietz NW, ed. FunPhiladelphia,
PA: WB Saunders,

Rivellese
A, Riccardi
0, Giacco A, Ct al. Effect of dietary
glucose control and serum lipoproteins
in diabetic patients.
l980;2:447-50.

fibre on
Lancet

34. Anderson
JW, Story L, Siding B, Chan W-JL, Petro MS, Story J.
Hypocholesterolemic
effects of oat-bran
or bean intake for hypocholesterolemic
men. Am J Clin Nutn l984;40:l
146-55.
35. Ullnich IH, Albnink MJ. Lack of effect of dietary
lipids, glucose, and insulin in healthy
young men
diets. Am J Clin Nutr 1982;36:4-9.

fiber on serum
fed high starch

Burke BJ, Hartog

M, Heaton KW, Hooper S. Assessment
of the
metabolic
effects of dietary carbohydrate
and fibre by measuring
urinary excretion
ofC-peptide.
Hum Nutr Clin Nutr l982;36C:37380.
37. Brown MS, Kovanen
PT, Goldstein
JL. Regulation
ofplasma
cholesterol by lipoprotein
receptors.
Science 198 l;2l2:628-35.
36.

38. Dunn FL, Bulheimer

DW, Grundy
SM, Raskin P. Pathogenesis
of
hypertniglyceridemia
in diabetes
mellitus:
the effects of improved
glucose regulation.
Clin Res l98028:39l(abstr).
39.

Nikkila

EA, Kekki

abetes mellitus.

M. Plasma

Metabolism

triglyceride

l973;22:l-22.

kinetics

transport

in di-

Downloaded from www.ajcn.org by guest on October 25, 2011

treatment
of hypenlipidemia
in adults. AHA special report. Circulation 1984;69: l065A-90A.
2. Kromhout
DK, Bosschieter
EB, Coulander CDeL. The inverse relation between
fish consumption
and 20-year mortality
from coronary heart disease. N Engl J Med l985;3l2:
1205-9.
3. Phillipson
BE, Rothrock
DW, Connor
WE, Harris WS, lllingwonth
DR. Reduction of plasma lipids, lipoproteins
and apoproteins
by
dietary fish oils in patients with hypentniglycenidemia.
N Engl J Med
1985;3l2:l2lO-6.
4. Sirtori CR, Agrandi
E, Conti F, Mantero 0, Oath E. Soybean-pmtein
diet in the treatment
oftype II hyperlipopnoteinaemia.
Lancet 1977;l:
257-77.
5. Knitchevsky
D, Tepper SA, Story JA. Influence
of soy protein and
casein on atherosclerosis
in rabbits. Fed Proc l978;37:747(abstr).
6. Miettinen
TA, Tarpila S. Effect ofpectin
on serum cholesterol
fecal
bile acids and biliary lipids in normolipidemic
and hyperlipidemic
individuals. Can Chim Acta l977;79:47l-7.
7. Jenkins DJA, Reynolds
D, Slavin B, Leeds AR, Jenkins AL, Jepson
EM. Dietary fiberand blood lipids: treatment of hypercholesterolemia
with guar cnispbread.
Am J Clin Nutr l980;33:575-8
1.

IN HYPERLIPIDEMIA