Particle size of wheat, maize, and oat test meals: Effects on plasma glucose and insulin responses and on the rate of starch digestion in vitro

Abstract

When normal volunteers ate isocaloric wheat-based meals, their plasma insulin responses (peak concentration and area under curve) increased stepwise: whole grains less than cracked grains less than coarse flour less than fine flour. Insulin responses were also greater with fine maizemeal than with whole or cracked maize grains but were similar with whole groats, rolled oats, and fine oatmeal. The peak-to-nadir swing of plasma glucose was greater with wheat flour than with cracked or whole grains. In vitro starch hydrolysis by pancreatic amylase was faster with decreasing particle size with all three cereals. Correlation with the in vivo data was imperfect. Oat-based meals evoked smaller glucose and insulin responses than wheat- or maize-based meals. Particle size influences the digestion rate and consequent metabolic effects of wheat and maize but not oats. The increased insulin response to finely ground flour may be relevant to the etiology of diseases associated with hyperinsulinemia and to the management of diabetes.

Full text

Particle size of wheat, maize, and oat test meals: effects on
Am J Gun Nuir 1988;47:675-82. Printed in USA. © 1988 American Society for Clinical Nutrition 675
plasma glucose and insulin responses and on the rate of
starch digestion in vitro13
Kenneth WHeaton, MD; SamuelNMarcus, MD; Pauline M Emmett, BSc; and Cohn H Bolton, PhD
ABSTRACT When normal volunteers ate isocaloric wheat-based meals, their plasma in-
sulin responses (peak concentration and area under curve) increased stepwise: whole grains
<cracked grains <coarse flour <fine flour. Insulin responses were also greater with fine
maizemeal than with whole or cracked maize grains but were similar with whole groats, rolled
oats, and fine oatmeal. The peak-to-nadir swing ofplasma glucose was greater with wheat flour
than with cracked or whole grains. In vitro starch hydrolysis by pancreatic amylase was faster
with decreasing particle size with all three cereals. Correlation with the in vivo data was imper-
fect. Oat-based meals evoked smaller glucose and insulin responses than wheat- or maize-based
meals. Particle size influences the digestion rate and consequent metabolic effects ofwheat and
maize but not oats. The increased insulin response to finely ground flour may be relevant to
the etiology of diseases associated with hyperinsulinemia and to the management of
diabetes. AmfClinNutr l988;47:675-82.
KEY WORDS Particle size, starch digestion, plasma glucose, plasma insulin, wheat,
maize, oats
Introduction
It might reasonably be expected that within the ali-
mentary tract the more intact the structure of a cereal
grain is, the more slowly the starch within will be di-
gested. Larger food particles have a lower surface-to-vol-
ume ratio and this must reduce the access ofenzymes to
the interior ofthe particle as must the presence of intact
cell walls. However, data to support these ideas are
scanty. A meal of raw wheat flakes evoked a lower
plasma glucose peak than one of whole-meal wheaten
bread (1) but this may have been because raw starch is
poorly digested (2, 3). When rice was ground into flour
it evoked more glycemia and insulinemia than did the
whole grains (4). This phenomenon has not been re-
ported with any other cereal nor have comparisons been
made of coarsely ground and finely milled flour. How-
ever, the in vitro studies ofO’Dea et al (5, 6) suggest that
particle size may be important with wheat, oats, and rye
as well as rice.
The present study was designed to test the hypothesis
stated above. A particular aim was to compare coarsely
milled and finely milled wheat flour because bread and
other wheat-flour products have long been staple foods
but finely milled flour is a relatively recent invention
(7, 8).
Subjects and methods
Test foods
The cereal products used for making the four wheat-based
meals, the three maize-based meals, and the three oat-based
meals were derived from the same batches of English wheat
(Triticum vulgare, variety Avalon), mixed imported maize
(Zea mays), and British oats (Avena sativa). They were milled
at the Flour Milling and Baking Research Association (Chor-
leywood, Herts, UK) in October and November, 1984 and
shipped to Bristol where they were stored at 4 #{176}Cuntil the test
meals were prepared early in 1985.
The four grades of wheat were as follows: whole grains,
cracked grains or kibbled wheat in whjch each grain was broken
into six pieces on average, coarsely milled flour, and finely
milled flour. Both flours were whole meal (ie, unsifted). The
difference in particle size between the coarse and fine flour was
considerable as shown by sieving analysis (Table 1). The fine
flour was similar to the whole-meal flour sold for home baking
in Britain. The coarse flour is not commercially available in
IFrom the University Department of Medicine, Bristol Royal In-
firmary, Bristol, UK.
2Supported by a grant from the Bristol and Weston Health Author-
ity(project #454).
3Reprints not available.
Received February 27, 1987.
Accepted for publication June 16, 1987.
by guest on July 11, 2011www.ajcn.orgDownloaded from

676 HEATON ET AL
TABLE 1
Sieving analysis ofthe flours and fine meals used in this study and, for comparison, a typical coarse Irish whole-wheat flour
Aperture
Percent retained on sic ye ofgiven aperture afte r sequential sieving
Wheat flour,
fine Wheat flour,
coarse Wheat flour,
Irish Maize meal,
fine Oatmeal,
fine
m
2200
1350
1050
710
390
250
140
-
-
-
-4.2
3.8
13.6
16.9
15.6
7.2
12.1
20.1
5.0
7.5
16.8
15.8
6.6
12.5
16.4
5.7
4.0
-
-
-
0.2
6.8’
13.4
21.8
-
-
-
0.2
9.6’
6.8
13.4
Percent not retained
onanysieve 78.4 15.6 22.1 57.8 70.0
*With maize and oats, Sieves ofaperture 425 and 355 m were used; the percentages retained on the 425.tm sieve were 2.8 and 2.2, respectively,
and on the 355-pm sieve were 4.0 and 7.4, respectively.
England but closely resembles the flour traditionally used for
home baking in Ireland and still widely available there.
The three grades ofmaize were whole kernels, cracked grains
(ie, kernels coarsely chopped up into 23 pieces on average), and
flour (fine maize meal). The three grades of oats were whole
groats; steamed, rolled oats (each flake weighing on average 6.7
mg compared with 17.8 mg for an average groat); and flour
(fine oatmeal). For sieving analysis ofthe maize meal and oat-
meal see Table 1.
The test meals were prepared in batches and frozen at -20
‘C. Spare meals were prepared for use in the in vitro digestion
study.
All test meals were designed to contain 50 g carbohydrate
according to tables of food composition (9, 10). Thus, wheat-
based meals contained 76 g wheat product, maize-based meals
68 g maize product, and oat-based 69 g oat product. The four
batches offlour (coarse and fine wheat flour, fine oatmeal, and
fine maize meal) were each made into scones by forming a
dough with water, salt, baking powder and, to aid palatability,
aspartame (Canderel, GD Searle, High Wycombe, UK) and
baking at 230 #{176}Cfor 10-15 mm. Each of the other six grain
products was weighed into individual tin-foil containers, coy-
ered with 300 mL water, and baked at 180 #{176}Cuntil it was judged
to be cooked. The baking time was 4-5 h for whole grains and
2-3 h for cracked grains and rolled oats.
In vivo plasma glucose and insulin responses
Different but overlapping groups of 10 healthy volunteers
(Table 2) ate in random order and at intervals of at least 48 h
TABLE 2
Sex, age, and body mass index (kg/m2) ofthe groups ofvolunteers who
ate the three cereal-based meals
Cereal eaten Subjects
(M/F)
n
Mean age
(range)
y
Body mass index
(mean ±SEM)
Wheat 5/5 24 (21-37) 21.9 ± 0.63
Oats 9/1 24(21-27) 23.0±0.59
Maize 7/3 24 (2 1-40) 22.2 ±0.48
test breakfasts made from wheat (all four grades), maize (three
grades), or oats (three grades). Studies in women were done
during the first week ofthe menstrual cycle. The duration of the
overnight fast was standardized for each subject. Meals were
thawed overnight and warmed to 40-50 #{176}Cbefore being served.
The meals were weighed so that their water content could be
estimated by subtracting the weight ofthe raw ingredients. The
water content of the breakfast was made up to 350 mL with
decaffeinated coffee or weak tea. Meals were served with a
sprinkling ofaspartame, a few drops oflemonjuice, powdered
nutmeg, or powdered cinnamon as desired to increase palat-
ability. All breakfasts commenced at the same time for a given
subject and were consumed steadily under supervision over a
20-mm period. No attempt was made to standardize the rate
or amount ofmastication. Subjects were comfortably seated in
a quiet room throughout the study period.
Samples of venous blood were taken via an indwelling can-
nula during fasting and at 10, 20, 30, 40, 50, 60, 75, 90, 105,
120, 150, and 180 mm from the beginning ofthe meal. Plasma
samples were immediately separated by centrifugation and fro-
zen until analyzed. Insulin was measured by a standard radio-
immunoassay (1 1), all the samples from each subject’s set of
test meals being analyzed together. Blood glucose was mea-
sured on an autoanalyzer (AAI System, Technicon, Basing-
stoke, UK)by the glucose-oxidase method (12). The area under
the plasma glucose and insulin curves was taken as the area
above the zero-time or 10-mm value, whichever was the lower.
This study was approved by the Bristol and Weston District
Ethical Committee.
In vitro rate ofdigestion ofstarch bypancreatic amylase
Portions of food were incubated with pancreatin (dried pig
pancreas) for differing periods of time to generate differing
amounts of the digestion products glucose, maltose, maltotri-
ose, and a-limit dextrins (H Ghafari, unpublished observa-
tions, 1983). In a second incubation with the enzyme a-glucosi-
dase, the maltose and maltotriose were hydrolyzed to glucose
(as was any starch that was solubilized). The glucose generated
in the two incubations was measured and related to the esti-
mated amount ofstarch in the original food to give a figure for
percent digestion ofstarch at each time. In preliminary experi-
ments with soluble potato starch, conditions for the in vitro
by guest on July 11, 2011www.ajcn.orgDownloaded from

glucose
mM/I
1.0
6.0
5.0
&#{149}0
______fine flow
- - - coarse flour
crackedgrains 120
110
100
mg/dl
90
I0
70
60
PARTICLE SIZE AND STARCH DIGESTION 677
digestion study were established so that there was maximal
yield ofglucose. In these experiments the mean maximal yield
ofglucose was 68.5 mg/lOO mg starch (68.5%).
Portions of thawed test meal were oven dried at 85 #{176}Cto
constant weight to determine moisture content. Additional
portions were then weighed out to contain ‘360 mg dry solids
(that is, ‘-250 mg starch assuming starch content of ‘-70%)
(9). They were not warmed as the meals were before being
eaten. These portions were mixed thoroughly with 5 mL 0.2
mmol/L phosphate buffer pH 6.5 containing 10 mmol/L so-
dium chloride (digest medium). At time zero, 20 mg grade VI
pancreatin (Sigma Chemical Company, Poole, UK) in 5 mL
digest medium was added to a series of tubes containing the
food suspensions, and all but one of the tubes were incubated
at 37 #{176}C.After 5, 10, 15, 20, 30, 40, and 60 mm incubation (in
the case of the oat-based meals, 15, 30, 45, 60, 90, 120, and
180 mm) one tube was placed on ice to stop the digestion and
centrifuged at 3500 X g for 15 mm at 4 #{176}C.The unincubated
tube was kept on ice and used as a blank. Enzyme-free blanks
were used initially but were never found to generate glucose
and were omitted thereafter. The supernatant was decanted
and diluted 1 to 20 with digest medium containing 20 mmol/
L disodium EDTA. One milliliter of this solution was incu-
bated with two units a-glucosidase (Sigma) in 1 mL digest me-
dium at 25 #{176}Cfor 2 h. Glucose produced was measured by the
standard glucose-oxidase method with o-dianisidine (Sigma).
The mean between-assay coefficients of variations were as
follows: fine-wheat flour, 1 1.0%; coarse-wheat flour, 10.5%;
cracked wheat, 13.8%; whole-wheat grains, 34.9%; maize flour,
18. 1%; cracked maize, 24.8%; whole maize, 1 13.8%; oat flour,
1 1.5%; rolled oats, 22.2%; and whole groats, 18.6%. (At each of
the seven times, n= 5 for wheat products and n=6 for all other
products.) The large mean coefficients with whole grains of
wheat and maize are due to the very low extent of digestion
of these whole grains, which were clearly intact despite being
thoroughly cooked.
Statistical methods
One-way analysis of variance and Student’s ttest for paired
values were used to compare individual values for plasma glu-
cose and insulin, the swing or descent of plasma glucose from
peak to nadir, areas under plasma concentration curves, and
glucose released by in vitro digestion.
Results
In vivo plasma glucose and insulin responses
With the wheat-based meals, plasma glucose tended to
rise higher and fall lower after flour than after cracked or
whole grains (Fig 1). Consequently, the mean swing of
plasma glucose (descent from peak to nadir) was signifi-
cantly greater after flour (Table 3). The area under the
glucose curve tended to be higher after the two flours
than after whole or cracked grains (Table 3) but the
differences were not significant.
Plasma insulin rose significantly higher as the particle
size ofwheat decreased (Fig 2). This was reflected in step-
wise increases in the peak plasma concentration and in
the area under the curve (Table 4, Fig 3). The biggest
difference was between coarse and fine flour, the area un-
der the insulin curve being 38% higher after the fine flour
(p =0.0063).
- - 0 10 20 30 O 50 50 15 90 105 120 150 110
Time (mm)
FIG 1.Mean plasma glucose concentration in 10 normal subjects
after four isocaloric whole-wheat meals ofdifferent particle size.
With the maize-based meals there were no differences
in plasma glucose responses. However, plasma insulin
tended to rise more after maize flour than after cracked
grains or whole maize (Fig 4) and the area under the in-
sulin concentration curve was 60% higher after the flour
than after cracked grains and 89% higher than after
whole maize (Table 4).
With the oat-based meals, plasma glucose tended to
rise higher with decreasing particle size but the areas un-
der the glucose concentration curves were not signifi-
cantly different (Table 3). Similarly, insulin values
showed no relation to particle size (Table 4). Indeed, the
peak insulin concentration and area under the curve
were higher after rolled oats than after fine oatmeal flour
although this difference did not reach statistical signifi-
cance.
When the responses to all three oat-based test meals
were combined and compared with the responses to all
the maize-based and wheat-based meals, the rise in
plasma glucose was substantially less with oats (mean
maximum rise for all oat-based meals, 1.6 ± SEM 0.1
mmol/L vs 2.7 ± 0. 1 for maize and 2.6 ±0. 1 for wheat,
both p<0.001 from unpaired ttests). The area under
the glucose curve was substantially lower after the oat-
based meals (55.2 ±6.4 mmol/L mm vs 94.4 ±7.0 with
maize and 95.0 ±6.3 with wheat, p <0.001 from un-
paired ttests). Furthermore, the area under the insulin
curve after oat flour was only 57% of that after maize
flour and 56% of that after fine wheat flour (p <0.05)
although the particle size profile was similar for all three
flours (Table 1). Different groups ofsubjects ate the three
groups of meals but the subjects were similar in age, sex,
and body mass index (except for a preponderance of
males in the oats group) and had similar fasting values
for plasma glucose (Table 3).
In vitro digestion of starch
With all three cereals the rate of starch digestion was
inversely related to particle size. Thus with wheat the or-
by guest on July 11, 2011www.ajcn.orgDownloaded from

#{163}0
30
-fine flour
---coarse flour
.cracked rains
whole grains
0 1020 30 050 60 75 90 105 120 150 160
Time (mm)
678 HEATON ET AL
TABLE 3
Plasma glucose responses to meals ofwheat, maize, and oats ofdifferent particle sizes (mean ± SEM)
Plasma glucose concentration Area under
concentration
Fasting Peak Nadir Swing curve
mmol/L mmol/L mmol/L mmol/L mmol/L mm
Wheat
Whole grains 3.8 ±0. 15.9 ±0.2 3.3 ±0.2 2.7 ±0.2 80.5 ±9.7
Cracked grains 3.9 ±0. 1 6.2 ±0.2 3.3 ±0.2 2.9 ±0.2 86.3 ±I 7.8
Coarse flour 3.8 ±0.2 6.5 ±0.2 3. 1±0.2 3.4 ± 0.2 108.2 ± 2.85
Fine flour’ 3.9 ±0. 1 6.6 ±0.2 3.2 ± 0. 1 3.4 ±0.3 106.2 ±1 1.5
Significant differences None None None 4 >lt
3> l,2t None
Maize
Wholegrains 4.0 ±0.1 6.7 ± 0.3 3.2 ± 0.1 3.5 ± 0.4 102.6 ± 15.3
Cracked grains 4. 1 ±0. 1 6.4 ±0.2 3.2 ± 0. 1 3.2 ±0.2 8 1 .4 ±10.4
Flour 3.9 ±0. 1 6.5 ±0.2 3.0 ± 0.2 3.4 ± 0.2 99.3 ± 10.0
Significant differences None None None None None
Oats
Whole grains 3.7 ±0. 1 5.0 ±0.2 3.0 ±0.2 2.0 ±0.2 4 1.0 ± 7.9
Rolled oats 3.7 ± 0. 1 5.3 ± 0.2 3. 1 ± 0.2 2.2 ±0.2 52.9 ±7.8
Flour 3.7 ±0. 1 5.7 ±0.3 3. 1 ±0.2 2.6 ±0.4 7 1.6 ±15.0
Significant differences None None None None None
SOne subject’s data excluded because oftechnical problems in assay ofplasma glucose after this meal.
tp= 0.047 using the unpaired ttest.
 p = 0.022 and 0.040, respectively, using the unpaired Itest.
der of digestion rates was whole grains <cracked grains
<coarse flour <fine flour (Fig 5). There were obvious
differences between whole and cracked grains and be-
tween cracked grains and coarse flour and a significant
further increase in digestion rate with fine vs coarse flour.
With maize and oats (Figs 6 and 7) there was also an
inverse relationship between particle size and rate of di-
insulin
mU/I
FIG 2. Mean plasma insulin concentration in 10 normal subjects
afterfourisocaloric whole-wheat mealsofdifferent particle size. (1 mU!
L = 7.175 pmol/L.)
gestion. With maize the difference between whole and
cracked grains was greater than that between cracked
grains and fine maize meal (flour), at least from 20 mm
onwards.
As fine flour all three cereals were digested at a similar
rate; at 30 mm the conversion of starch to simple sugars
was 46% for wheat, 42% for maize, and 41% for oats.
However, as whole grains, oats were digested much faster
than wheat or maize; at 30 mm, the conversion of starch
to sugars was 18% for oats vs 5% for wheat and 3% for
maize (both p<0.005).
Correlation between in vitro and in vivo measurements
Among the wheat-based meals there was a significant
correlation between in vitro digestibility of a meal (cx-
pressed as the mean percent ofstarch converted to sugars
after 30 mm incubation) and the mean rise in plasma
glucose and between in vitro digestibility and the mean
swing of plasma glucose after the same meal (r =0.96
and 0.97, respectively, p <0.05 for both). The correla-
tion was not significant with the area under the glucose
concentration curve, with the peak plasma insulin, or
with the area under the insulin concentration curve.
Small numbers precluded looking for correlations
among the oat-based and maize-based meals.
Discussion
When wheat and maize are milled before cooking,
their contained starch is digested more rapidly in vitro
by guest on July 11, 2011www.ajcn.orgDownloaded from

insuLin
mU/I mm
3000
2000
1000
0 whole cracked coarse fine
grains grains flour flour
FIG 3. Area under plasma insulin concentration curve (mean
±SEM)after four isocaloric whole-wheat meals. Significant differences
as follows: fine flour >all others, coarse flour >whole grains.
PARTICLE SIZE AND STARCH DIGESTION 679
TABLE 4
Plasma insulin responses to meals ofwheat, maize, and oats ofdifferent particle sizes (mean ± SEM)
Plasma insuli n concentration
Area under curve
Fasting Peak
mUlL (pmol/L) mUlL (pmol/L) mUlL mm (pmol/L mm)
Wheat
Wholegrains 7.2±1.1(52±9) 31.4±2.8(225±20) 1211±165(8689±1184)
Crackedgrains 7.5± 1.6(54± 11) 34.2±3.8(245±27) 1403±267(10067± 1916)
Coarse flour 6.3 ±1.2(45 ± 9) 37.7 ± 3.0(270 ± 22) 1718 ±160(12327 ± 1148)
Fine flour 8.3 ±1.6(60 ±11) 46.9 ±4.3(337 ±31) 2366 ±279(16976 ±2002)
Significant differences None 4 >1, 2, 3’
3>lt 4>1, 2, 3t
3>l
Maize
Wholegrains 9.3 ±1.5 (67 ±1 1) 41.8 ± 5.2 (300 ±37) 1238 ±212(8883 ±1521)
Crackedgrains 6.9± 1.3(50±9) 42.3±5.5(304±39) 1461 ±243(10483± 1744)
Flour 8.0± 1.7(57± 12) 48.7±7.5(349±54) 2341±459(16797±3293)
Significant differences None None 3 >1, 211
Oats
Wholegrains 8.0 ± 1.0(57 ±7) 36.4 ±3.9(261 ±28) 1 1 1 1 ±169(7971 ±1212)
Rolled oats 7.7 ±1.6(55 ±1 1) 41 .4 ±4.8 (297 ±34) 1701 ±2 1 1 (12204 ±1514)
Flour 7.9± 1.6(57± 11) 36.4±5.3(261 ±38) 1323± 182(9492± 1306)
Significant differences None None None
*p<0.001, 0.028, and 0.003, respectively.
t p <0.001 ,0.019, and 0.017, respectively.
p=0.006.
§p= 0.030.
IIp=0.010 and 0.012, respectively.
and evokes a greater plasma insulin response, indicating
that the starch is digested and absorbed more rapidly in
vivo as well as in vitro. Finely milled wheat flour is di-
gested in vitro significantly faster than its coarsely milled
equivalent and evokes a substantially greater insulin re-
sponse.
Cooking conditions were not identical for all grades of
cereal; all test meals were fully cooked in the everyday
sense, but it is possible that some starch remained ungel-
atinized in the scones, which had briefbaking times. This
being so, our data underestimate the increase in digest-
ibility that occurs when wheat and maize are ground into
flour. Freezing the meals may have caused retrograda-
tion ofsome starch. However, the freezing and reheating
were identical for all meals so there should not have been
any major differences in their content ofresistant starch
(13, 14).
Thus, with wheat and maize the findings were much as
predicted. They indicate that grinding these grains into
flour, especially fine flour, affects their digestibility pro-
foundly, but simply breaking them into several small
pieces has little or no effect.
These conclusions are essentially from the insulin re-
sponses. The glucose responses were not significantly
different with the various grades of wheat and maize cx-
cept that there were differences between the wheat-based
meals for the swing of plasma glucose. Thus, it appears
that measurements of plasma glucose are less sensitive
than measurements of plasma insulin in detecting
differences in the digestibility ofstarch (at least in nondi-
abetic subjects); our previous work suggests the same for
the absorption of sugars from sugar-rich meals (1 1). On
the other hand, both studies indicate that differences in
the responses to foods that would otherwise be missed
can be detected if the swing of plasma glucose is calcu-
lated.
The idea that the structural integrity of starchy foods
determines their rate ofdigestion has been foreshadowed
by studies showing that the plasma glucose and insulin
response to rice and corn is less than that to potato (15-
19) because when rice and corn are cooked, they retain
their solidity and presumably their structure more than
does potato. The traditional, coarsely milled wheat and
rye products bulgur and pumpernickel were reported to
evoke less glycemia (in diabetic subjects) than did ordi-
by guest on July 11, 2011www.ajcn.orgDownloaded from

_______fine flour
cracked grains
. . whole graiiw
60
50
50
40
30
20
10
insulin
mU/I
mg glucose
per 100mg
starch
10
0 5 10 15 20
I #{149}#{149}#{149}  whole grains
30 tO 50 60
Time (mm)
Time (miii)
mg glucose
per 100mg
starch
50
&0
30
20
10
0
mg glucose
per 100mg
starch
- - -- - #{149}i_- - - rolled oats
--- -I- -    .:whole grains
0 15 30 5 60 90 120
Time (mm) 180
680 HEATON ET AL
FIG 4. Mean plasma insulin concentration in 10 normal subjects
after three isocaloric maize meals ofdifferent particle size. (1 mU/L
= 7.175 pmol/L.)
nary wheat and rye bread (20), but in these studies the
source of wheat and rye varied. When normal subjects
swallowed rice, maize, or lumps ofpotato without chew-
ing, the glycemic response was greatly reduced (21); un-
chewed potato can even be recovered from the feces (22).
O’Dea et al (4, 5) compared the glucose and insulin re-
sponses ofnormal subjects to batches ofwhite and brown
rice in two forms, as whole grains and milled into flour,
and found increased glycemia and insulinemia with the
milled materials. They also found starch digestion by
FIG 6. In vitro digestion ofstarch in three maize products of differ-
ent particle size. Glucose and the glucose equivalent of maltose and
maltotriose released at seven time points (mean ±SEM ofsix different
estimations). At every time point the three values are significantly
different from each other(p <0.05). (1 mg glucose = 5.551 mol.)
amylase in vitro to be much faster with the ground rice
and there were close correlations between the percent of
starch digested in 30 mm and the peak plasma glucose
and insulin concentrations.
Correlations between in vitro and in vivo measures of
wheat starch digestion could be demonstrated in this
study but there were anomalies. Correlations existed
with indices ofglucose but not ofinsulin response, which
is paradoxical in view ofthe greater sensitivity of insulin
measurements. The difference between coarse and fine
wheat flour was substantial in vivo (with indices of insu-
lin release) but less impressive in vitro. The striking
difference between whole and cracked maize in vitro had
no counterpart at all in vivo. Ofcourse, the in vitro incu-
bation system is a very crude approximation of gastroin-
testinal events. In particular, it involves nothing equiva-
0 5 10 15 20 ( 40 50 60
Time 1mm)
FIG 5. In vitro digestion ofstarch in four whole-wheat products of
different particle size. Glucose and the glucose equivalent of maltose
and maltotriose released at seven time points (mean ±SEM of five
different estimations). With 38 out ofthe 42 possible comparisons the
mean values are significantly different from each other (p <0.05). (1
mgglucose = 5.551 mol.)
FIG 7. In vitro digestion ofstarch in three oats products of different
particle size. Glucose and the glucose equivalent ofmaltose and malto-
triose released at seven time points (mean ±SEM of six different esti-
mations). With 17 ofthe 21 possible comparisons the mean values are
significantly different from each other(p <0.05). (1 mgglucose = 5.551
zmol.)
by guest on July 11, 2011www.ajcn.orgDownloaded from

PARTICLE SIZE AND STARCH DIGESTION 681
lent to the chewing of food which, with whole grains es-
pecially, must reduce particle size and facilitate access of
digestive enzymes. Also, there is no preliminary peptic
digestion, which may leave some starch (at least in the
unmilled and coarsely milled products) encapsulated
within a protein matrix (23). Finally, gastric emptying is
a determinant ofthe rate ofdigestion in vivo.
The more rapid digestion of rice, wheat, and maize
when they are cracked or milled into flour is presumably
due to the easier access of amylase to starch when the
surface-to-volume ratio offood particles is increased and
cell walls are mechanically disrupted. Loss of histological
integrity has also been invoked to explain why the starch
in beans, peas, and lentils is digested faster in vitro if the
pulses are dry milled before cooking (24). An additional
factor in vivo may be faster gastric emptying ofsmall par-
tides. However, with maize a 23-fold reduction in parti-
dc size that should have led to faster gastric emptying
made no difference to the plasma insulin response.
With oats the in vitro findings were out of line with
the in vivo findings. In vitro digestion rate increased as
particle size decreased but in vivo the plasma glucose and
insulin responses were not significantly different between
the three oat-based meals. This implies that with oats
there is a factor that strongly influences digestion or ab-
sorption in vivo but that was inoperative in the in vitro
system. This factor could be the viscous properties of the
soluble fl-D-glucan that is the main component of dietary
fiber in oats (25). This gummy material might create a
viscous microclimate in the intestinal lumen and so im-
pede the diffusion ofamylase to starch or the diffusion of
glucose and maltose towards the mucosa. These effects
would have been minimized in the in vitro incubation
because this involved constant mechanical shaking.
Another anomaly with oats was that the meals evoked
relatively small glucose and insulin responses compared
with the other two grains although the starch was di-
gested in vitro equally fast; indeed, whole groats were
more digestible than whole wheat and maize. Again, this
can be explained by oats containing an unusually high
proportion of soluble viscous fiber that limits the rate of
digestion or absorption in vivo. Our findings are consis-
tent with those of Jenkins et al (18) who reported the
glycemic index after oat flakes to be only 49 ± 8 com-
pared with 75 ± 10 after wheat flakes and 80 ± 6 after
corn flakes. A caveat is that the starch content ofour oats
meals may not have been identical with that ofour maize
and wheat meals because we used food-table values for
starch and did not measure it directly.
In conclusion, this study demonstrates that wheat and
maize, two ofthe major staple foods ofmankind, are di-
gested faster the more they are reduced in particle size.
Finely milled wheat flour evokes substantially greater in-
sulin responses than coarsely milled flour. Oat products
do not behave in this way, possibly because ofthe differ-
ent physical properties ofoat fiber. Our finding of greater
insulin secretion after finely milled flour raises the possi-
bility that in susceptible genotypes regular consumption
ofsuch flour increases the risk ofdiseases in which hyper-
insulinemia has a possible etiological role, namely, obe-
sity (26), diabetes (27), gallstones (28), and atherosclero-
sis (29). There may also be a practical consequence. Be-
cause the management of diabetes mellitus is aided by
incorporating into the diet slow-release carbohydrates
such as legume seeds (30), it may also be aided by the
replacement of flour with whole or cracked grains or by
the use ofa coarsely milled flour. 13
The in vitro studies were expertly performed by Miss A Moffau, who
also helped with statistical analysis. Plasma glucose and insulin assays
were carried out by the Chemical Pathology Department ofthe Bristol
Royal Infirmary. Dr SCW Hook kindly provided the samples of whole
and milled wheat, maize, and oats and carried out the sieving analyses
at FMBRA, Chorleywood, Herts, UK. Dr C Oettl#{233}helped with the col-
lection of samples.
References
1. Thomas B, Elchazly M. The physiological effects and changes of
the dietary-fibre of wheat in the digestive tract. Qualitas Plant-
arum-Plant Foods Hum Nutr l976;26:2l 1-26.
2. BeazellJM, Schmidt CR, Ivy AC. On the digestibility ofraw potato
starch in man. J Nutr 1939; 17:77-83.
3. Collings P, Williams C, MacDonald I. Effects ofcooking on serum
glucose and insulin responses to starch. Br Med J 198 l;282:l032.
4. O’Dea K, Nestel PJ, AntonoffL. Physical factors influencing post-
prandial glucose and insulin responses to starch. Am J Clin Nutr
l980;33:760-5.
5. O’Dea K, Snow P, Nestel P. Rate ofstarch hydrolysis in vitro as a
predictor ofmetabolic responses to complex carbohydrate in vivo.
AmiClinNutr l981;34:l991-3.
6. Snow P, O’Dea K. Factors affecting the rate ofhydrolysis of starch
in food. Am J Gin Nuts 198l;34:272l-7.
7. McCance RA, Widdowson EM. Breads white and brown, their
place in thought and social history. London: Pitman, 1956.
8. Kent NL. Technology ofcereals. 3rd ed. Oxford: Pergamon, 1984.
9. Paul AA, Southgate DAT. McCance and Widdowson’sthe compo-
sition offoods. 4th ed. London: HMSO, 1978.
10. Platt BS. Representative values offoodscommonly used in tropical
countries. London: HMSO, 1962.
11. Oettl#{233}GJ, Emmett PM, Heaton KW. Glucose and insulin re-
sponses to manufactured and whole-food snacks. Am J Clin Nutr
l987;45:86-9l.
12. Trinder P. Determination ofglucose in blood usingglucose oxidase
with an alternative oxygen acceptor. Ann am Biochem 1969;6:
24-7.
13. Englyst HN, Cummings JH. Digestion of the polysaccharides of
some cereal foods in the human small intestine. Am J Gin Nutr
l985;42:778-87.
14. Englyst HN, CummingsJH. Digestion of polysaccharides of potato
in the small intestine ofman. Am J Gin Nutr l987;45:423-3l.
15. Crapo PA, Reaven G, Olefsky J. Plasma glucose and insulin re-
sponses to orally administered simple and complex carbohydrates.
Diabetes l976;25:74l-7.
16. Crapo P, Insel J, Sperling M, Kolterman 0G. Comparison of se-
rum glucose, insulin, and glucagon responses to different types of
complex carbohydrate in non-insulin-dependent diabetic patients.
AmJClinNutr 198l;34:184-90.
17. Vaaler S, Hanssen KF, Aagenaes 0. Plasma glucose and insulin
responsesto orallyadministered carbohydrate rich foodstuffs. Nutr
Metab 1980;24: 168-75.
18. Jenkins DJA, Wolever TMS, Taylor RH, et al. Glycemic index of
by guest on July 11, 2011www.ajcn.orgDownloaded from

682 HEATON ET AL
foods: a physiological basis for carbohydrate exchange. Am J Gin
Nutr 198 l;34:362-6.
19. Coulston A, Oreenfield MS. Kraemer FB, Tobey TA, Reaven GM.
Effect ofdifferences in source ofdietary carbohydrate on plasma
glucose and insulin responses to meals in patients with impaired
carbohydrate tolerance. Am J Clin Nutr l98l;34:27 16-20.
20. Jenkins DJA, Wolever TMS, Jenkins AL, et al. Low glycemic re-
sponses to traditionally processed wheat and rye products: bulgur
and pumpernickel bread. Am J Clin NuIr l986;43:5 16-20.
21. Read NW, Welch IMcL, Austen CI, et al. Swallowingfood without
chewing a simple way to reduce postprandial glycaemia. BrJ Nutr
l986;55:43-7.
22. Farrell JH. The effect of mastication on the digestion of food. Br
DentJ 1956; 100:149-55.
23. Holm J, Bj#{246}rckI, Asp N-G, Sj#{246}bergLB, Lundquist I. Starch avail-
ability in vitro and in vvo after flaking, steam-cooking and pop-
pingofwhcat. J Cereal Sd l985;3:l93-206.
24. Wursch P, DelVedovo S, Koellreutter B. Cell structure and starch
nature as key determinantsofthe digestion rate ofstarch in legume.
Am J Gin Nutr l986;43:25-9.
25. Englyst H, Anderson V, Cummings JH. Starch and non-starch
polysaccharides in some cereal foods. JSci Food Agric 1983;34:
1434-40.
26. Assimacopoulos-Jeannet F, Jeanrenaud B. The hormonal and
metabolic basis ofobesity. Gin Endocrinol Metab l976;5:337-65.
27. Jackson WPU, van Meighem W, Keller P. Insulin excess as the
initial lesion in diabetes. Lancet 1972; 1:1040-4.
28. Scragg RKR, McMichael AJ, Baghurst PA. Diet, alcohol, and rela-
tive weight in gallstone disease: a case-control study. Br Med J
l984;288:l 113-9.
29. Stout RW. Insulin and atheroma-an update. Lancet 1987; 1:
1077-9.
30. Simpson HCR, Simpson RW, Lousley 5, etal. A high carbohydrate
leguminous fibre diet improves all aspects ofdiabetic control. Lan-
cet l981;l:l-5.
by guest on July 11, 2011www.ajcn.orgDownloaded from