Rice: a high or low glycemic index food?1'2 - American Journal of ...

Rice: a high or low glycemic index food?1’2
.Jaiwite Brand )fj//j Edna Pang, and Lindsai’ Brainall
ABSTRACT We determined the glycemic (GI) and insulinindex
(II) values for 12 rice products, using eight healthy subjects.
The products were brown and white versions ofthree commercial
varieties ofrice [two varieties with normal amylose content (20%)
and the other with 28 amylose]. a waxy rice (0-2% amylose),
a converted rice, a quick-cooking brown rice, puffed rice cakes,
rice pasta. and rice bran. The GI of the rices ranged from 64
± 9 to 93 ± 1 1. where glucose = 100. The high amylose rice
gave a lower GI and II (P < 0.0 1 ) than did the normal-amylose
and waxy-rice varieties. The converted rice and most other rice
products gave a high GI. Insulin indices correlated positively
with GI (r = 0.75. P < 0.05). although they were lower than
expected. These results indicate that many varieties of rice,
whether white, brown, or parboiled, should be classified as high
GI foods. Only high-amylose varieties are potentially useful in
low-GI diets. Am J C/ui Nutr 1992:56:1034-6.
KEY WORDS Glycemic index, diabetes, rice
Introduction
Rice has given a wide range of results in glycemic index (GI)
studies around the world. The GI ofwhite rice has ranged from
as low as 54 to as high as 121 when bread (GI = 100) is used as
the reference food (1-3). This makes it difficult to classify ri’e
as a high- or low-GI food. and advice to individuals with diabetes
may be incorrect if the product has not been specifically tested
first. Oats, barley, and rye are all low-GI foods when consumed
as intact grains and it might be expected that rice would be
similar ( 1). Medium-term studies have shown that low-GI diets
improve glycemic control in patients with insulin-dependent diabetes
mellitus (IDDM) (4. 5) and non-IDDM (6). In some of
these studies rice was classified as a low-GI food whereas others
assigned it to the high-GI category.
It is likely that much of the variation in the GI of rice is due
to differences in the proportion ofstarch present as amylose. ie.
amylose: amylopectin ratio. Most rices contain 20% amylose
but varieties that contain a higher proportion of amylose (eg,
28%) have been shown to have a slower rate of digestion and
produce lower glycemic and insulin responses (7). The classification
of rice as a high- or low-GI food may therefore depend
on the amylose content of commercial varieties, but the consumer
has no way ofdetermining this from the food label. Parboiled
(converted) rice may also produce different responses than
normal rice, although this aspect has not been specifically investigated.
To clarify the responses to rices grown in Australia, we studied
the glycemic and insulin responses ofhealthy subjects to 12 rice
products: brown and white versions ofthree commercial varieties
ofrice (two varieties contained 20% amylose and the other 28%
amylose), a waxy rice with 0-2% amylose, a parboiled rice (Sungold;
Rice Growers Cooperative Ltd. Leeton, NSW), a quickcooking
brown rice. rice cakes (a puffed product), rice pasta,
and rice bran. We measured the responses to rolled oats, wheat
pasta. and rolled barley for comparison.
Materials and methods
[‘ODdS
Twelve rice products were studied together with wheat pasta,
rolled oats (porridge), and rolled barley. A 50-g carbohydrate
portion ofwhite bread was used as the reference food, although
the final result was expressed on a scale where glucose = 100.
The nutrient composition of the foods, their preparation, and
the amount given in the test meals, are shown in Table 1. The
foods were prepared according to the manufacturers’ instructions
and served with tinned tomatoes (100 g) with the exception of
rolled oats. which were given with milk (40 mL), and rice bran,
which was served as a pur#{233}eby adding boiling water and nonnutritive
flavoring to it.
Studi
c/L’sign
Eight healthy volunteers with normal glucose tolerance took
part in the study. Their ages ranged from 19 to 36 y (1 ± SD 30
± 10) and their body mass index (BMI; in kg/m2) from 18 to
25 (22 ± 3). The subjects took 50 g available carbohydrate portions
of each food except rice bran (25 g) in random order on
separate mornings after a 10-h overnight fast. For individual
subjects the tests were given 1 wk apart. Foods were consumed
over 1 2 mm with sufficient tea (made by using one tea bag brewed
for 1 mm) and 30 mL milk to bring the total meal volume to
600 mL.
Finger-prick capillary blood samples were taken at 0 (fasting),
15, 30, 60, 90, and 120 mm after the meal began. Hands were
placed in a 45 #{176}C water bath for 2 mm before being punctured
with an Autolet device (Owen Mumford Ltd. Woodstock, UK)
with autoclix lancets (Boehringer Mannheim, Mannheim, FRG).
Blood (800 tL) was collected into l-mL tubes with heparin and
centrifuged at 12 500 X g for 1 mm at room temperature and
the plasma stored at -20 #{176}C before analysis. Glucose was assayed
I From the Human Nutrition Unit. Department of Biochemistry,
University of Sydney. Sydney.
2 No reprints available.
Received Ianuarv 21. 1992.
Accepted for publication July 8. 1992.
Downloaded from ajcn.nutrition.org by guest on August 9, 2014
1034 Am J C/in Nuir 1992:56:1034-6. Printed in USA. © 1992 American Society for Clinical Nutrition

GLYCEMIC INDEX OF RICE 1035
TABLE 1
The composition and weight ofthe foods
Foods
Protein
(nitrogen X 6.25) Fat
Carbohydrate
by difference
Weight of uncooked
50-g carbohydrate
portion
Preparation
% % % g
Calrose white 5.7 0.5 79.9 62.6 Boiled for 14 mm
Calrose brown 6.6 2.7 75.6 66.1 Boiled for 35 mm
Pelde white 6.4 0.5 79.6 62.8 Boiled for 14 mm
Pelde brown 7.2 2.9 75.4 66.3 Boiled for 30 mm
Pelde (parboiled)5 7.0 1.0 79.4 63.0 Boiled for 14 mm
Doongara white (high amylose) 8.0 0.6 77.5 64.5 Boiled for 14 mm
Doongara brown (high amylose) 8.7 2.5 74. 1 67.5 Boiled for 30 mm
Sunbrown Quick 8.3 3.0 77.4 64.6 Boiled for 16 mm
Waxy rice (0-2% amylose) 7.3 0.8 77.2 64.8 Boiled for 14 mm
Rice cakes 8.0 2.7 8 1.5 6 1.3 Puffed product
Rice bran 13.7 22.4 47.4 52.8t Pur#{233}ed with boiling
water and flavoring
Brown rice pasta 8.9 2.6 75.5 66.2 Boiled for 16 mm
Wheat pasta 12. 1 0.2 75.9 65.9 Boiled for 20 mm
Rolled oats 8.9 9.7 68.8 72.7 Boiled for 10 mm
Rolled barley 7.5 2.4 76.9 65.0 Boiled for 10 mm
S Sungold: Rice Growers Cooperative Ltd. Leeton, NSW.
t Weight of 25-g carbohydrate portion.
by using the glucose hexokinase method on a Cobas-Fara centifugal
analyzer (Roche Diagnostica, Basle, Switzerland) and insulin
by radioimmunoassay with the Bio-RIA ‘25I-Insulin Chromacode
Radioimmunoassay kit (Bio-Mega Diagnostica Inc.
Hamon,
Montreal).
The GI and insulin index (II) of each food was calculated as
described previously (3) by using a 50-g carbohydrate portion
of white bread as the reference food. For rice bran a 25-g carbohydrate
portion was consumed and the incremental area under
the curve was doubled. The GI was then multiplied by the factor
70/ 100 to express the final result on a scale where glucose = 100.
When glucose is used as the reference food the GI of bread is
70 (2). The mean ± SE ofeight subjects was calculated. In some
instances, one or two subjects gave a GI or II outside two SDs
from the mean of the group. Their results were excluded and
the mean of the remaining subjects reported. Comparisons between
the foods were made by using two-way analysis of variance
and the Bonferroni adjustment of the Student’s t test. Linearregression
analysis was used to test for an association between
glycemic and insulin responses. The protocol was approved by
the Medical Ethical Review Committee of the University of
Sydney and all subjects gave written, informed consent.
Results
The GI and II of the foods tested are shown in Table 2. The
GIs of the white rices ranged from 64 ± 9 to 93 ± 1 1 . The rice
with a higher amylose content (Doongara, 28% amylose) gave
a significantly lower GI and II (P < 0.01) than did the normalamylose
rice varieties (Calrose and Pelde, 20% amylose). The
waxy rice with 0-2% amylose had a GI similar to that of those
containing normal amounts of amylose.
The GIs of the brown rices were similar to their white counterparts
except for the Pelde variety. Pelde brown rice (GI = 76
± 1 1) gave a significantly lower GI than did white Pelde (GI
= 93 ± 1 1, P < 0.0 1). The parboiled (converted) Pelde rice gave
a high GI (87 ± 7). Brown rice pasta and rice cakes gave high
GIs (92 ± 8 and 82 ± 1 1, respectively) whereas rice bran gave
an extremely low GI ( 19 ± 3). The wheat pasta (GI = 58), rolled
oats (GI = 58), and rolled barley (GI = 66) gave results that
were similar to those reported by others (3).
TABLE 2
The glycemic and insulin indexes ofthe foods5
Foods
Glycemic index
(glucose = 100)
Insulin index
(glucose = 100)
Calrose white (n = 8) 83 ± 13 67 ± 15
Calrose brown (n = 8) 87 ± 8 51 ± 7
Pelde white (n = 7) 93 ± 1 1 67 ± 1 1
Pelde brown (n = 8) 76 ± 6 55 ± 10
Pelde (parboiled) (n = 8) 87 ± 7 57 ± 6
Doongara white (high
amylose) (n = 8) 64 ± 9 40 ± 10
Doongara brown (high
amylose) (n = 8) 66 ± 7 39 ± 6
Sunbrown Quick (n = 8) 80 ± 7 54 ± 6
Waxy rice
(0-2% amylose)
(11 = 7) 88 ± 11 89 ± 19
Rice cakes (n = 6) 82 ± 1 1 73 ± 12
Rice bran (n = 8) 19 ± 3 23 ± 4
Brownricepasta(n=6) 92±8 72±18
Wheat pasta (n = 6) 58 ± 7 52 ± 9
Rolled oats (n = 7) 58 ± 4 54 ± 12
Rolled barley (n = 8) 66 ± 5 64 ± 1 1
S SE. The rice with a higher amylose content (Doongara, 28%
amylose) gave a significantly lower glycemic index and insulin index (P
< 0.0 1) than did the normal-amylose rice varieties (Calrose and Pelde,
20% amylose).
Downloaded from ajcn.nutrition.org by guest on August 9, 2014

1036 MILLER ET AL
The II ofthe rices correlated positively with their GI (r = 0.75,
P < 0.05) but were relatively low compared with the GI. That
is. the regression equation predicts that a rice with a GI of 80
has an II of 60.
Discussion
These results indicate that most varieties of rice sold in
Australia, whether white, brown, or parboiled, should be classified
as high-GI foods. Only one variety with a higher amylose
content (Doongara) could be said to be potentially useful in low-
G1 diets. The white Doongara rice had a GI and II of 64 and
40, respectively. These values are similar to those ofwheat pasta
and porridge, which are recognized as low-GI foods. Rice bran,
which is rich in fiber and oil, had an extremely low GI and may
be a useful supplement in diabetic diets.
The relationship between amylose content and GI ofrice may
be exponential rather than linear. The GI of a rice with 20%
amylose is already very high: 93 ± 1 1 in the Pelde variety. It is
conceivable that this is approaching the maximum that is physiologically
possible (the glycemic response to a similar carbohydrate
load of glucose is not much higher). Hence, reducing
the amylose content to only 0-2% amylose appears to make
little difference. However. increasing amylose has a pronounced
effect that has been shown by others (7). There are now varieties
of rice with 35% amylose that may be predicted to have lower
GIs than that ofthe Doongara rice (28% amylose) studied here.
The parboiled rice in the present study gave a high GI (87
± 7), unlike the parboiled rice studied by Jenkins et al (1) (GI
= 54, glucose = 100). Parboiling ofrice involves steeping paddy
rice in water at 60 #{176}C for > 1 h followed by steaming, drying,
and milling. This process does not appear to alter the glycemic
response because the GI of this variety of rice (Pelde) was 93
± 1 1 before and 87 ± 7 after parboiling (P> 0.05). The parboiled
rice used by Jenkins et al may have a low GI because the variety
used has a high amylose content rather than because ofthe parboiling
process per se. This suggestion, however, needs to be
corroborated by GI testing and amylose:amylopectin analysis of
other parboiled rices.
An unexpected observation was the relationship between GI
and II: the II was usually lower on the relative scale than was
the GI of the foods. For example, the GI of Calrose brown rice
is 83 but its II was only 5 1 . We have no explanation for this
result unless bread, the reference food, is unique in its capacity
to stimulate insulin. The effect may be increasingly observed as
glycemic and insulin responses to real foods are measured. Collier
et al (8) reported that insulin responses to carbohydrate foods
were potentiated by coingestion oflarge amounts offat, ie, insulin
responses were higher than the relative glycemic response if fat
was added to the food. In the present study, fat content is negligible
and insulin secretion is decreased relative to the glycemic
response. The higher protein content of wheat pasta and rice
bran may explain the higher parity of GI and II for these two
foods. Coingestion of protein is known to increase the insulin
response to a carbohydrate load (9).
Findings such as this indicate that both GI and II should be
considered when determining the optimal carbohydrate foods
for individuals with diabetes. Insulin concentrations are a major
determinant of blood cholesterol and triglyceride concentrations
and influence the progression ofcoronary heart disease (10). We
cannot assume that a food with a high glycemic response will
have a high insulin response. It may be possible that many rice
varieties, despite having a high GI, can be included in diets that
improve carbohydrate and insulin metabolism in individuals
with diabetes. Both the Canadian and the French groups have
specifically included rice in their long-term GI diet studies (4,
5). In the Canadian studies, parboiled rice with a GI of54 (glucose
= 100) was specified and in the French studies a rice with a GI
of 56 (measured directly) was used.
All subjects received a 50-g carbohydrate portion of white
bread as the reference food in this study. However, we then
adjusted all the results down by a factor of 70/100 so that glucose
would theoretically have the highest score on the scale, ie, 100
[bread has a GI of 70 if glucose is the standard (2)]. We feel
strongly that an index where glucose is 100 is easier to understand
because, theoretically, nothing is likely to be higher than this.
Jenkins et al (1) have made this adjustment in the other direction
to many oftheir foods in order to express their early results with
a white-bread standard ( 1 ). Their decision to make white bread
- 100 has resulted in some confusion. We recommend that GI
testing and calculation be carried out as we have done, with the
final result expressed on a scale where glucose is 100. This means
that any food can be used as the standard as long as its relationship
to a 50-g glucose load is known.
These results emphasize the need for individual countries to
carry out their own GI testing, particularly of raw agricultural
products, which are more likely to vary from one geographical
location to another than do processed products, such as breakfast
cereals. The findings also raise questions about the value of the
GI alone without knowledge ofthe insulin response. In Australia,
long-term studies are needed of the clinical utility of low-GI
diets that specifically incorporate rice. C]
References
1. lenkins DIA. Wolever DIA, Jenkins AL. Starchy foods and glycemic
index. Diabetes Care 1988:11:149-59.
2. Jenkins DJA, Wolever TMS, Taylor RH. et al. Glycemic index of
foods: a physiological basis for carbohydrate exchange. Am I Clin
Nutr 1981:34:362-6.
3. Brand IC. Nicholson PL, Thorburn AW, Truswell AS. Food processing
and the glycemic index. Am I Clin Nutr 1985:42:1192-6.
4. Fontvieille AM, Acosta M, Rizkalla SW, et al. A moderate shift
from high to low glycemic index foods for 3 weeks improves the
metabolic control of Type I (IDDM) diabetic subjects. Diabetes
Nutr Metab 1988:1:139-43.
5. Collier GR. Giudici 5, Wolever TMS, et al. Low glycemic index
starchy foods improve glucose control and lower serum cholesterol
in diabetic children. Diabetes Nutr Metab 1988:1:11-18.
6. Brand IC. Colagiuri S. Crossman 5, Allen A, Roberts DCK, Truswell
AS. Low glycemic index foods improve long term glycemic control
in NIDDM. Diabetes Care 1991:14:95-101.
7. luliano BO. Goddard MS. Cause of varietal difference in insulin
and glucose responses to ingested rice. Plant Foods Hum Nutr
1986:36:35-41.
8. Collier G, McLean A, O’Dea K. Effect ofco-ingestion offat on the
metabolic responses to slowly and rapidly absorbed carbohydrates.
Diabetologia 1984:26:50-4.
9. Nuttall FQ. Mooradian AD. Gannon MC, Billington C, Krezowski
P. Effect ofprotein ingestion on the glucose and insulin response to
a standardized oral glucose load. Diabetes Care 1984:5:465-70.
10. Zavaroni I, Bonora E, Pagliara M, et aI. Risk factors for coronary
artery disease in healthy persons with hyperinsulinaemia and normal
glucose tolerance. N Engl I Med 1989:320:702-6.
Downloaded from ajcn.nutrition.org by guest on August 9, 2014