Eur J Nutr

DOI 10.1007/s00394-016-1309-7

ORIGINAL CONTRIBUTION

Glyceamic and insulinaemic response to mashed potato alone,

or with broccoli, broccoli fibre or cellulose in healthy adults
Simon Ballance1 · Svein Halvor Knutsen1 · Øivind Winther Fosvold2 ·
Martin Wickham3 · Carmen Díaz‑Toledo Trenado3 · John Monro4 

Received: 15 January 2016 / Accepted: 2 September 2016

© Springer-Verlag Berlin Heidelberg 2016

Abstract  other iAUCs for GR and IR. For the potato meals contain-
Purpose To examine the role of realistic serving sizes of ing added broccoli fibre or cellulose, no significant differ-
broccoli, broccoli fibre and cellulose co-consumed with ences in GR or IR were observed when compared with the
mash potato, or mashed potato eaten alone, on glycaemic potato eaten alone.
and insulinaemic responses (GR and IR) in healthy adults. Conclusion Co-consumption of cooked broccoli with
Method A non-blind randomized crossover trial was con- mashed potato has a significant effect on glycaemic and
ducted with thirteen healthy subjects consuming four dif- insulinaemic responses compared to potato eaten alone.
ferent meals. Capillary blood samples between 0 and Our study suggests broccoli eaten with potato improves
180 min were analysed for glucose and insulin. The incre- glucose homeostasis and therefore indicates a general ben-
mental area under the fasting blood glucose and insulin eficial nutritional role for broccoli when eaten with a carbo-
curves (iAUC) was calculated for different time increments. hydrate staple.
Differences in GR and IR between meals were assessed by
repeated measures analysis of variance. Keywords  Blood sugar · Vegetable · Starch · Meal ·
Results The immediate GR and IR to one serving of Carbohydrate staple
mashed potato eaten with two servings of broccoli were
significantly lower than mashed potato eaten alone. The
peak, incremental peak and iAUC0–30min for GR and Introduction
iAUC0–30min for IR were all significantly lower for the broc-
coli–potato meal. This meal also takes longer to return Sustained postprandial hyperglycaemia and hyperinsuli-
to fasting baseline with a time-delayed lag in IR and GR naemia in healthy adults are strongly associated with an
compared to the potato only meal. The iAUC60–120min for increase in the risk of developing type 2 diabetes and/or
IR was significantly greater for the broccoli–potato meal metabolic syndrome. The postprandial glycaemic (GR) and
compared to the other meals. Yet there was no correspond- insulinaemic responses (IR) of starch rich-carbohydrate
ing significant difference between the broccoli–potato meal staples, such as bread, rice, pasta and potato, and the fac-
and the other meals for peak, incremental peak IR or any tors that might modulate them, have therefore received
extensive attention [1]. This includes understanding the
effect of adding additional ingredients to carbohydrate sta-
* Simon Ballance
simon.ballance@nofima.no ples and the effect of eating them as part of a mixed meal.
Other important variables include the amount and type
1
Nofima AS, Norwegian Institute of Food, Fisheries of glycaemic carbohydrate consumed, and the regime of
and Aquaculture Research, Osloveien 1, 1430 Ås, Norway
processing during food preparation. The general glycae-
2
Fjordland AS, Brynsengveien, Oslo, Norway mic and insulinaemic effects in healthy adults of protein,
3
Leatherhead Food Research, Leatherhead, UK fat, dietary fibre and organic acids additions to carbohy-
4
The New Zealand Institute for Plant & Food Research drate staples are all widely investigated [1]. Many studies
Limited, Palmerston North, New Zealand have assessed additions of either purified or cereal-derived

13
 Eur J Nutr

dietary fibre. Far fewer studies have examined the effects of of potential importance. This includes, among others, type
co-consumed vegetables on GR/IR in healthy adults even of vegetable and how its preparation, vegetable-to-carbo-
though the many meals containing carbohydrate staples, hydrate staple ratio, and quantity and composition of other
such as cooked potatoes and rice, include them. food components and nutrients. It is clear further work is
Studies on co-consumed vegetable–staple carbohydrate required to understand the role of vegetables and the mech-
combinations would enable one to assess whether vegeta- anisms responsible for their potential glycaemic and insuli-
ble has any direct effect on the GR/IR of a carbohydrate naemic health effects. Clinical in vivo feeding experiments
staple. Yet there is surprisingly very little research on this with humans should play a key part of this work.
topic. A 120 g serving of fresh salad (cucumber, tomato In our current study, we have selected mashed potato as
and lettuce) consumed together with just over two portions the carbohydrate staple to be consumed by healthy human
of warm mashed potato (363 g) had no significant effect subjects at a typical and representative serving size of
(P > 0.05) on GR/IR in healthy subjects compared to that 150 g. To this, we have assessed the effect of adding two
for an equivalent serving of mashed potatoes eaten alone servings of boiled broccoli florets on postprandial GR and
[2]. In another study, 120 g boiled Chinese cabbage (bok IR. Minimally processed broccoli fibre was also added to
choy) and 200 g boiled rice were eaten by healthy subjects mashed potato to investigate a possible role of dietary fibre
[3]. The GR, as assessed by the incremental area above the and cell wall structure. This portion of broccoli fibre had an
fasting blood glucose curve over 120 min (iAUC0–120min) equivalent content of total fibre to that found in two serv-
and peak glucose, was moderately, but still significantly ings of broccoli. An equivalent amount of cellulose to the
(P < 0.05) less, compared to 200 g of rice eaten alone [3]. broccoli fibre preparation was also added to mashed potato
There was also a significant difference in peak insulin as a third meal combination to act as an insoluble dietary
response but not for iAUC0–120min for IR between these two fibre-rich negative control. Cellulose was chosen because
meals [3]. previous clinical studies have validated it as a suitable neg-
There are some further studies with mixed meals. ative control when investigating the GR lowering effect of
Healthy adults served a typical Swedish lunch (creamed soluble fibres [10].
potatoes, bread, meatballs, lingonberry jam and light beer) Our experimental design takes inspiration from the stud-
with or without carrots, spinach, peas or Brussels sprouts, ies discussed above that show cooked cruciferous vegeta-
containing 4.4 g of dietary fibre, all had similar GR or IR bles, such as broccoli, to have potential positive effects on
(P > 0.05), except for the meal with added spinach [4]. For lowering GR/IR of a co-consumed carbohydrate staple.
the spinach meal, a significantly lower (P < 0.05) post- Secondly, that the physical bulk properties of the vegeta-
prandial IR (peak and iAUC0–120min) was observed [4]. In ble matter in the small intestine might, in part, explain this
a follow-up study, where the dose of spinach was increased effect. The meals have also been prepared and processed
to 250 g, containing 7.3 g dietary fibre, postprandial GR, in a way that is representative of the situation for a typical
in addition, was also significantly less [5] than the control ‘ready-to-eat’ meal available from the supermarket apart
meal without spinach. At a more realistic serving size of from the test meals we have made were subjected to two,
vegetables, the postprandial GR was lower when 60 g each rather than the usual one, cooking–cold storage (>2 days)
of boiled broccoli and spinach containing 4.4 g dietary fibre cycle prior to re-heating and immediate consumption.
were added to a complex Japanese mixed meal, compared
with the same meal without vegetables [6].
In most of the above-mentioned studies, a thorough dis- Materials and methods
cussion of the possible mechanisms behind a response in
GR/IR is often lacking. Quite often discussion of the GR/ Study meals
IR of vegetable on carbohydrate rich staples in mixed
meals is overshadowed by discussion of the effects of other The meals were prepared in a professional kitchen used for
protein and fat-rich meal components [2, 3]. product development of ready meals (Fjordland AS, Oslo,
In a rat study with minimally processed broccoli fibre Norway). Each contained 150 g mashed potato. To ensure
added to corn oil, it was shown that physical properties homogeneity, the mashed potato was prepared in one large
of broccoli fibre in the small intestine may reduce entero- batch. Commercial chilled–stored (>2 days) sous-vide
hepatic bile acid recycling and intestinal lipid absorption cooked potatoes that were pre-blanched (Hoff AS, Jæren,
with a consequent increase in faecal bile acid excretion Norway) of the variety Faxe were pressed through a potato
[7]. Indeed, the physical properties of digested plant mat- ricer and mixed by hand. Broccoli var., Iron Man, was
ter and its role in glycaemic health has recently received obtained via Bama AS from Skallerød farm, Jeløy, Nor-
renewed attention and possible mechanisms are currently way and cold–stored at 2 °C in the dark until use. Floret
under renewed debate [8, 9]. There are also other factors bouquets were cut from the broccoli heads, washed in tap

13
Eur J Nutr

water, cooked in a pan of boiling water until tender, and

Energy
(kcal)
then cooled in cold water. Broccoli florets of 180 g portions

124
127
160
111
were immediately vacuum packed. Equivalent portions
(150 g) of potato alone, potato mixed with 8 g milled broc-

Waterb
coli fibre or 8 g milled cellulose were also identically vac-

292
294
285
300
uum packed at the same time. All packed meals received
heat treatment in a Convotherm combi-steamer for 30 min

Watera
at 98 °C. They were then cooled in running cold water for

120
122
285
120
20 min before chilled overnight transport to Leatherhead,
UK. Prior to consumption, each meal was re-heated in its

Ash

0.9
1.1
1.9
0.9
vacuum bag for 6 min in boiling water apart from the broc-
coli which was re-heated for 11 min. The ‘ready-to-eat’
core-meal temperature was 70–72 °C. Apart from the broc-

Protein
coli–potato combination, a measured glass of water was

2.7
4.6
7.8
2.7
supplied for consumption with the meal to ensure a total
consumed meal mass of 330 g throughout the study.

Fat

0.6
0.6
1.8
0.6
The nutrient composition of the meals was analysed as
follows. Protein was estimated (N × 6.25) from the anal-

Total
fibre
ysis of N by the method of Kjeldahl [11]. Fat was deter-

7.8
8.1
2.9

11
mined gravimetrically following acid hydrolysis, extraction
into diethyl ether and petroleum ether and evaporation [12].
Total dietary fibre was determined gravimetrically accord- Sugars

0.8
0.8
2.5
0.8
ing to AOAC 985.28 [13]. Total non-starch polysaccha-
ride in dried minimally processed broccoli fibre and in the
cooked broccoli was determined by the method of Englyst
Resistant
starch

[14] via GC-FID analysis of alditol acetates. Sugars were

1.7
1.7
1.7
1.7

determined as the sum of sucrose, glucose and fructose

after extraction in 50 % water:methanol by anion-exchange
chromatography with pulsed amperometric detection [15].
Digestible

Total and resistant starch ‘as eaten’ was determined by

starch

19.5
19.5
19.5
19.5

AOAC 2002.02 within 1 h of re-heating the potato. Avail-

able CHO was subsequently calculated as described by
Brouns et al. [16]. Water content was determined gravi-
ACHO

22.2
22.2
24.0
22.2

metrically following drying at 103 °C to constant weight

[17]. Ash was determined as the inorganic residue remain-
ing after removal of all water and organic matter by heating
Meal sizeb

at 550 °C [18]. Total energy content was calculated accord-

ing to EU Council Directive 1169/2011 [19]. The nutrient
330
330
330
330

  Not including glass of water consumed with the meal

composition of the test foods is shown in Table 1.

  Including glass of water consumed with the meal
Meal sizea
Table 1  Nutrient composition of the test meals

Preparation of broccoli fibre

158
158
330
150

A similar milling and cold-water extraction scheme

described by Mandimika et al. [7] for stalk-derived broc-
Mashed potato and broccoli fibre

coli fibre was employed. Broccoli (33 kg, 10–15 cm

Mashed potato and cellulose

stalk and with head) were washed in tap water. There-

Mashed potato and broccoli

after, the stalk was cutoff from the head and discarded to
yield 23.3 kg floret bouquets. These were then cut up in a
Garant MTK 661 tabletop bowl rotary chopper and yielded
Mashed potato

21 kg of about 2- to 5-mm broccoli floret particles. These

(g/portion)

were further reduced in size, water added, and the result-

ant solid–liquid suspension fed as a slurry into a Fryma
MZ-80 toothed colloid mill with a rotor–stator setting of
b
a

13
 Eur J Nutr

0.5 mm. A nylon 250 µm sieve mesh (2 × 1 m) was used allotted 15 min and were thus excluded from further analy-
to enclose the slurry of broccoli floret fragments, and these sis. Thirteen subjects completed the study of which 12 were
were washed in running tap water (approx. 8 °C) overnight. female. The mean age of these subjects was 47.6 (SEM
Using a 20 L Speidel hydropress (1–2 bar pressure), water 3.29) years with a mean BMI of 22.6 (SEM 0.53) kg/m2.
held in the washed floret fragments was pressed-out and The study was conducted according to guidelines laid down
discarded. The resultant fibrous cake was thinly spread on in the Declaration of Helsinki and study design approved
baking paper, placed on a tray, and dried to constant weight by the West Kent Research Ethics Committee, Aylesford,
in an oven at 65 °C. The yield was 1 kg. About 300 g of UK. Written informed consent was obtained from all sub-
this material was then dry milled into a fine powder with a jects. All clinical testing was conducted at Leatherhead
Retsch hammer mill fitted with a 1 mm sieve. Particle size Food Research, UK in October 2014.
distribution was determined with a Sympatec HELOS laser
diffraction sensor with ROSOS dry dispersing unit fitted Study protocol
with a R6 lens (0.5–1750 µm).
The night before the test the subjects were instructed to
Preparation of cellulose avoid strenuous physical activity, refrain from smoking or
consuming alcohol the evening before a test or during the
Cellulose sheets (28.5 × 17.5 cm) manufactured as a sub- day of the test. The subjects were instructed to consume
strate to make food grade chemically modified cellulose a similar carbohydrate-based evening meal before each
were cut into 10 × 2 mm pieces with scissors. These pieces test session. Subjects were also instructed to fast from
were dry milled into a fine powder with a Retsch hammer 20:00 the night before a test. Water consumption was not
mill fitted with a 1-mm sieve. Particle size distribution was restricted. Subjects should not have had a similar test for
determined with a Sympatec HELOS laser diffraction sen- the last 48 h (washout time). On each test day, the volun-
sor with ROSOS dry dispersing unit fitted with a R6 lens teers arrived at the Human Nutrition Unit, having fasted
(0.5–1750 µm). for at least 12 h prior to commencement, and they were
seated and asked to remain so for the duration of the test.
Scanning electron microscopy Upon arrival their blood glucose levels were checked using
a hand-held glucometer to ensure they had fasted cor-
The sample was mounted on an aluminium stub using dou- rectly and were suitable to take part. Once each subject
ble-sided tape coated with carbon. The sample was then was relaxed and comfortable, they were asked to provide a
coated with gold/palladium using a SC7640 auto/manual baseline glucose measurement for that day, against which
high-resolution sputter coater (Quorum Technologies, Ash- all of that day’s subsequent assessments were measured.
ford, UK). An EVO-50-EP environmental scanning elec- The subjects were given the four different potato-based
tron microscope (Zeiss, Cambridge, UK) was used to study meals in a non-blind randomized order on separate days
the sample at a magnification of x2000. (crossover) with a least 48-h washout between testing.
Each subject presented with a study meal including a glass
Subjects of water (were appropriate) was instructed to consume
the whole amount within a 15-min period. The first blood
Volunteers were pre-screened and asked initial recruitment sample was collected exactly 15 min after the first bite
questions in order to determine their suitability to take part of the sample food. After this point, blood samples were
in the study. The nature of the study and their involvement taken at 15-min intervals for the first hour, 30-min inter-
and responsibilities were described to them. Eligible vol- vals for the second hour and then after a 1-h interval for
unteers that were willing to participate were presented with the third hour. Samples were collected at 0, 15, 30, 45, 60,
an information sheet, containing study details, along with 90, 120, 180 min.
a written consent form at least 3 days before starting the Capillary blood samples were collected into small tubes
study. The inclusion criteria were age: 18–65 years, gen- containing lithium–heparin following a finger-prick, and
der: male or female, BMI 18–27 kg/m2, self-diagnosis as centrifuged at 3000 rpm for 10 min to separate the plasma.
healthy at the time of recruitment confirmed by medical The plasma samples were then analysed for glucose by
questionnaire. Fasting blood glucose: 4–6 mmol/L. Sub- an YSI 2300 Stat Plus Glucose and Lactate analyser. The
jects were excluded from the study if they had any history sensitivity of the analyser is 0–50 mmol/L, and the margin
of diabetes or had consumed anything apart from water of error is ±2 % or 0.2 mmol/L. Insulin was analysed in
12 h prior to starting the test. plasma using a sandwich ELISA (Mercodia, Uppsala, Swe-
Fifteen healthy subjects were recruited for one single den) according to manufactures instructions. Prior to insu-
cohort. Two subjects did not consume all test meals in the lin analyses, all plasma samples were stored at −80 °C.

13
Eur J Nutr

Calculations and statistical analysis

The incremental area under the glucose and insulin

response curves (iAUC) above baseline was calculated
for specified time intervals between 0–180 min using the
standard trapezoid geometric method [16]. Minitab version
17 was used for all statistical analysis. The Kolmogorov–
Smirnov test revealed all data were normally distributed.
Statistical differences between meals (fixed factor) were
therefore assessed for subjects (random factor) by repeated
measures ANOVA. The criterion for significance was a
two-tailed P < 0.05. Comparison between meals was made
with the post hoc Bonferroni pairwise test at a confidence
interval of 95 %.

Fig. 1  Scanning electron microscope picture of the broccoli fibre.

Magnification ×2000. Scale bar 10 µm
Results

All test meals were rich in available carbohydrate (22.2– GR when compared to the other meals that returned to fast-
24 g) almost exclusively originating from the potato. An ing baseline by 120 min (Fig. 2). This delayed response is
extra 1.8 g available carbohydrate in the broccoli meal nicely illustrated for iAUC60–120min for IR which was sig-
came from sugars in the broccoli (Table 1). All meals were nificantly greater (P = 0.004) compared to all other meals
low in protein and fat but quite high (7.3–11 g) in dietary (Table 3B; Fig. 2b). Nevertheless, there was no correspond-
fibre, apart from the potato only based meal that only had ing significant difference between the broccoli–potato meal
2.7 g dietary fibre (Table 1). All meals had a low-energy and the other meals for peak, incremental peak IR or any
content (Table 1). They also had an equal mass when co- other iAUCs for GR or IR (Table 3; Fig. 2).
consumed with a glass of water (Table 1). The ‘liquid frac- For mashed potato mixed and eaten with broccoli fibre,
tion’ (glass of water + moisture in the meal food compo- there is also no significant difference in GR or IR for peak,
nents) was therefore similar in all meals (285–300 g). incremental peak or iAUC0–30min, compared to the potato
Scanning electron microscopy revealed the broccoli fibre only and potato–cellulose meals (Table 3; Fig. 2). Apart
to comprise mostly collapsed plant cell walls (Fig. 1) par- from iAUC0–30 min, there was also no significant difference
ticularly rich in polysaccharide but also protein (Fig. 1; in these same parameters between the broccoli fibre–potato
Table 1). The dietary fibre and NSP content and NSP com- meal and the broccoli–potato meal (Table 3B). For the
position in the broccoli and in the minimally processed potato–cellulose meal, there was no significant difference
broccoli fibre were identical when expressed in terms of in incremental peak glucose or iAUC0–30min for IR, com-
amounts in the respective meals (Table 2). The cumulative pared to the broccoli–potato meal (Table 3). Otherwise, the
size distribution of 10–90 % of dehydrated broccoli fibre cellulose–potato meal had a quite similar GR and IR to the
particles was 45–430 µm. The size distribution of the cel- potato only meal (Table 3; Fig. 2).
lulose was similar at 25–375 µm.
Table  3 and Fig. 2 show the postprandial GR and IR
for all test meals. For each corresponding meal, both these Discussion
physiological responses closely mirror one another (Fig. 2).
Peak responses for all meals occurred at 30 min. For the Co-consumption of two servings of cooked broccoli with
broccoli–mashed potato meal, however, peak (P = 0.009), one serving of mashed potato appears to lower the imme-
incremental peak (P = 0.026) and iAUC0–30min (P = 0.003) diate glycaemic impact and significantly reduce the result-
for GR were all significantly less compared to either potato ing acute insulin demand in healthy subjects. Broccoli con-
consumed alone or potato mixed with cellulose (Fig. 2a; sumed with potato seems it might significantly suppress
Table 3A). A similar trend was observed for the iAUC0–30 the rate of glucose loading and the release of stored insu-
min IR which was significantly lower (P  = 0.003) for the lin in response to the glucose challenge. Eaten with potato,
broccoli–potato meal than the potato meal eaten alone. broccoli serves to consequently delay and then elevate and
Another difference is that the broccoli–potato meal takes extend the duration of the IR post-peak over fasting base-
180 min to return to fasting baseline for both IR and GR line. Nonetheless, while the GR, following broccoli addi-
(Fig. 2). There appears to be a time-delayed lag in IR and tion to potato, was also delayed and extended post-peak,

13
 Eur J Nutr

Table 2  Dietary fibre and non-starch polysaccharides in broccoli and broccoli fibre

(g/meal portion) Total fibre Total NSP Uronic acid Rhamnose Fucose Arabinose Xylose Mannose Galactose Glucose

Broccoli 5.2 4.1 1.0 0.1 0.03 0.6 0.2 0.1 0.5 1.4
Broccoli fibreA 4.9 4.0 1.0 0.1 0.03 0.6 0.2 0.1 0.4 1.4

Values for carbohydrates all expressed as polysaccharide equivalents

A
  Values are for material prior to mixing with potato, subsequent packaging and heat treatment

Table 3  Fasting, peak, incremental peak and incremental areas under the curves above baseline for different periods between 0 and 180 min of
capillary blood glucose (A) and insulin (B) in healthy subjects following the study meals (mean of n = 13, ±SE)
Mashed potato Mashed potato Mashed potato Mashed potato
and broccoli and broccoli fibre and cellulose

(A)
Concentration (mmol/l)
 Fasting 5.0 ± 0.1A 5.1 ± 0.1A 5.1 ± 0.1A 5.1 ± 0.1A
 Peak 8.3 ± 0.3A 7.5 ± 0.3B 7.9 ± 0.3AB 8.2 ± 0.3A
 Incremental peak 3.2 ± 0.3A 2.5 ± 0.3B 2.8 ± 0.3AB 3.1 ± 0.3AB
iAUC (mmol × min/L)
 0–30 min 55.3 ± 4.5A 40.8 ± 3.8B 47.8 ± 3.4AB 54.9 ± 3.6A
 0–60 min 109.5 ± 10.4A 98.0 ± 9.9A 102.6 ± 9.3A 110.6 ± 11.7A
 0–120 min 126.8 ± 16.6A 129.9 ± 13.7A 122.8 ± 13.8A 128.2 ± 17.4A
 60–120 min 17.4 ± 7.5A 31.9 ± 6.4A 18.8 ± 5.7A 17.6 ± 7.1A
 120–180 min 1.68 ± 1.3A 4.8 ± 2.0A 1.8 ± 0.7A 1.7 ± 1.2A
(B)
Concentration (mU/l)
 Fasting 3.8 ± 0.6A 5.3 ± 1.0A 3.7 ± 0.5A 5.1 ± 0.7A
 Peak 35.6 ± 3.6A 31.2 ± 3.2A 35.4 ± 2.8A 32.3 ± 2.5A
 Incremental peak 31.8 ± 3.4A 25.9 ± 3.1A 31.7 ± 2.8A 27.3 ± 2.6A
iAUC (mU × min/L)
 0–30 min 605.5 ± 72.8.A 374.3 ± 50.9B 543.7 ± 49.6A 525.3 ± 51.55AB
 0–60 min 1109.8 ± 112.6A 932.7 ± 123.4A 1138.4 ± 89.0A 1028.6 ± 102.2A
 0–120 min 1354.2 ± 134.3A 1358.4 ± 179.5A 1380.8 ± 114.3A 1215.3 ± 129.5A
 60–120 min 244.4 ± 70.8A 425.7 ± 86.3B 242.5 ± 54.4A 186.7 ± 64.5A
 120–180 min 50.4 ± 21.4A 56.4 ± 26.6A 34.1 ± 13.0A 39.0 ± 21.2A

Meals that share a letter are not significantly different

it was not elevated over that of the other meals subjected The available carbohydrate load in our study was half
to significantly greater acute insulin responses. The broc- that of most previous studies, and the vegetable portion was
coli evens out the homeostatic response of the carbohydrate about one third more. The meal composition of two serv-
load in the potato with the associated potential beneficial ings of broccoli to one serving of mashed potato would
health implications. therefore maximize any contribution; the vegetable may
A similar shift of the blood glucose profile was not have on the GR/IR of the mashed potato-based meal. Such
observed when cooked bok choy or fresh salad was, respec- a carbohydrate staple-to-vegetable ratio may represent a
tively, eaten with rice or mashed potato compared to the threshold where the effects of co-ingested vegetable start
carbohydrate staples eaten alone [2, 3]. Peculiarly, for the to exert its effects on GR/IR. This reinforces the conclu-
bok choy–rice study, the peak glucose response for rice sions of other studies that high doses of vegetable, prefer-
eaten alone occurred at 60 min [3]. This unusual result is at ably cooked, are required to have an effect on GR [5]. It
odds with similar studies, even with Asian subjects, where seems the vegetable-to-carbohydrate serving size ratio is
peak postprandial GR for rice and other carbohydrate sta- also very important, particularly, in a meal with low fat
ples occurs at around 30 min [6]. and protein content. Selection of the type of vegetable and

13
Eur J Nutr

(A) 9 ileostomy study of cooked pureed carrot, the gross tissue

structure of intact cells remained unchanged [24]. How-
Capillary blood glucose (mM/L)

ever, the material had swelled and pectin in the cell wall
8
was thought to have been solubilized and then released
during stomach–small intestine digestion, which facili-
7 tated subsequent movement of water into cell wall voids
[24, 25]. A subsequent reduction in mass transfer through
6 reduced diffusion and digesta mixing and decreased free
water would ultimately result in delayed glucose uptake.
5
A similar active role was proposed for swollen insoluble
undigesible plant cell walls in recent in vitro studies with
kiwifruit [26] and in rat studies with broccoli fibre [7] and
4
0 15 30 45 60 75 90 105 120 135 150 165 180 tossa jute fibre [27]. Any fibre solubilized during digestion
Time (min) is undoubtedly also important, and it may contribute to any
of increased viscosity, reduced mixing, and masking of
(B) 40 starch and/or amylase inhibition, to name just a few of the
35
hypothesized mechanisms.
Capillary blood insulin (mU/L)

Substitution of broccoli with minimally processed broc-

30 coli fibre had only a very weak effect on GR. Since the
25 NSP content and composition in broccoli and minimally
processed broccoli fibre in the meal were identical, fac-
20
tors other than simply the NSP content and composition
15 would seem responsible for differences in GR/IR response.
Release and subsequent loss of pectins may have been
10
greater for cooked broccoli than in the minimally processed
5 broccoli fibre. The latter was also evenly mixed into the
0
potato, as a dehydrated powder before re-heating and con-
0 15 30 45 60 75 90 105 120 135 150 165 180 sumption, compared to the broccoli, which was re-heated
Time (min) separately and eaten together with the potato. Larger parti-
cles of intact vegetable cell matter from the broccoli meal
Fig. 2  Mean changes in plasma glucose (a) and insulin (b) in healthy may in some way impact GR and IR as suggested above
subjects (n = 13 ± SEM) after the consumption of the study meals. via delayed gastric emptying. Another reason for the dif-
Mashed potato (filled circle), mashed potato with broccoli (open cir- ference could have been the dose of broccoli fibre was not
cle), mashed potato with broccoli fibre (upside-down filled triangle)
and mashed potato with cellulose (open triangle) high enough (ca. 5 % w/w of the meal) or it did not fully
rehydrate to the same state as the cooked broccoli.
At very best in our work, cellulose had a very weak
its nutritional characteristics when eaten is also of likely effect, if any, on GR/IR. Previous studies in rats and pigs
importance. A vegetable consumed with a carbohydrate have shown that inclusion of solid particles, including crys-
staple ought not to contribute an additional high load of talline cellulose, may reduce GR by increasing viscosity
available carbohydrates. in the lumen [28, 29]. This effect is considered both par-
The actual mechanisms responsible whereby veg- ticle size and dose dependent. Other in vitro studies, have
etable, such as broccoli, may have an impact on carbohy- recently shown insoluble dietary fibre additions, including
drate staple GR/IR are open to debate. The solid bulk of wood fibre, to pre-gelatinized starch suspensions may act
the hydrated broccoli may have delayed gastric emptying, by sequestration or inhibition of amylase and that lumen
as could have its slightly higher energy load, which in turn viscosity may not fully explain how insoluble fibres might
delays the absorption of glucose into the blood. Broccoli impact GR [30].
is rich in polyphenols and phenolic acids [20]. Flavonol In a study with mashed potato mixed meals combined
glycosides, hydroxycinnamic acids, caffeic acid and ferulic addition of rapeseed oil and salad attenuated, the strong
acid are abundant [21] and may have a potential to inhibit increase in IR in healthy subjects induced by protein,
digestive enzymes [22], such as pancreatic α-amylase, and/ from chicken breast, added alone [2]. A similar effect was
or inhibit active glucose uptake [23]. Other mechanisms observed when bok choy and oil where eaten together with
related to the physical state of the broccoli in the small chicken and rice [3]. However, in both these studies, fat and
intestine may also be important. In an in vitro digestion and vegetable were added to protein and carbohydrate staple

13
 Eur J Nutr

together. It was, therefore, not possible to separate the indi- Ethical standards All human studies have been approved by the
vidual effect on IR or GR of vegetable addition to a com- appropriate ethics committee and therefore have been performed in
accordance with the ethical standards laid down in the 1964 Declara-
plex meal. Separate addition of spinach, but not carrots, tion of Helsinki and its later amendments.
Brussel sprouts or peas to a complex potato-based meal
elicited a significantly lower IR in healthy subjects than Informed consent  All persons participating in the clinical study gave
a mixed meal without spinach [4]. The same was found informed consent prior to their inclusion.
for GR where spinach was added at a higher dose [5].
This may point to a role for specific vegetable types in
suppressing GR and IR when eaten in combination with
References
other meal ingredients such as moderate amounts of pro-
tein and fat in a mixed meal setting. It has also been shown 1. Wolever TMS (2006) The Glycaemic Index: a physiological
that addition of vegetable side dishes to a meal with a classification of dietary carbohydrates. CABI, Wallingford
moderate-fat content may also decrease the postprandial 2. Hätönen KA, Virtamo J, Eriksson JG, Sinkko HK, Sundvall JE,
GR in healthy subjects without a disproportionate increase Valsta LM (2011) Protein and fat modify the glycaemic and
insulinaemic responses to a mashed potato-based meal. Br J Nutr
in IR [6]. 106:248–253
3. Sun L, Ranawana DV, Leow MK-S, Henry CJ (2014) Effect
of chicken, fat and vegetable on glycaemia and insulinae-
Conclusion mia to a white rice-based meal in healthy adults. Eur J Nutr
53:1719–1726
4. Gustafsson K, Asp N, Hagander B, Nyman M (1993) Effects of
The current work demonstrates that the addition of cooked different vegetables in mixed meals on glucose homeostasis and
broccoli to mashed potato has a significant effect on GR satiety. Eur J Clin Nutr 47:192–200
and IR in healthy subjects compared with potato eaten 5. Gustafsson K, Asp N-G, Hagander B, Nyman M (1995) Satiety
effects of spinach in mixed meals: comparison with other vegeta-
alone. Both peak, incremental peak blood glucose and the bles. Int J Food Sci Nutr 46:327–334
immediate glycemic response were significantly reduced 6. Kameyama N, Maruyama C, Matsui S, Araki R, Yamada
as was the immediate demand for insulin. The mechanisms Y, Maruyama T (2014) Effects of consumption of main and
responsible for this effect and their individual significance side dishes with white rice on postprandial glucose, insulin,
glucose-dependent insulinotropic polypeptide and glucagon-
are open to discussion. The results of our study suggest that like peptide-1 responses in healthy Japanese men. Br J Nutr
addition of broccoli to potato improves glucose homeosta- 111:1632–1640
sis and therefore indicates a physiologically significant role 7. Mandimika T, Paturi G, De Guzman CE, Butts CA, Nones K,
for vegetables when eaten together with a carbohydrate sta- Monro JA, Butler RC, Joyce NI, Mishra S, Ansell J (2012)
Effects of dietary broccoli fibre and corn oil on serum lipids, fae-
ple. The ratio of vegetable to carbohydrate staple is impor- cal bile acid excretion and hepatic gene expression in rats. Food
tant, as is probably the type of vegetable and the way it is Chem 131:1272–1278
prepared prior to consumption. Since potato has a favour- 8. Gidley MJ (2013) Hydrocolloids in the digestive tract and related
able content of nutrients over rice and pasta, its consump- health implications. Curr Opin Colloid Interface Sci 18:371–378
9. Dhital S, Dolan G, Stokes JR, Gidley MJ (2014) Enzymatic
tion as a carbohydrate staple corner stone of a mixed meal hydrolysis of starch in the presence of cereal soluble fibre poly-
should be encouraged. Addition of a large portion of veg- saccharides. Food Funct 5:579–586
etables, such as broccoli, as part of a meal, which in turn 10. Maki KC, Carson ML, Miller MP, Turowski M, Bell M, Wilder
is part of a balanced diet, is likely to confer both long- and DM, Rains TM, Reeves MS (2008) Hydroxypropylmethylcel-
lulose and methylcellulose consumption reduce postprandial
short-term beneficial effects on health. insulinemia in overweight and obese men and women. J Nutr
138:292–296
Acknowledgments The authors would like to acknowledge the 11. NMKL (2003) Nitrogen. Determination in foods and feeds

skilful technical assistance of Aida Sainz Fernandez, Hanne Zobel, according to Kjeldahl, NMKL method No. 6, 4th edn. Nordic
Vibeke Høst and Ingunn Berget. This study is part of project no. Committee on Food Analysis
225148 in The Research Council of Norway with financial support by 12. Croon L, Fuchs G (1980) Fetthaltsbestämning i mjöl och mjöl-
the Research Funding for Agriculture and the Food Industry in Nor- produkter. Vår Föda 32:425–427
way (85 %) and Norwegian potato industry (15 %). Additional finan- 13. Prosky L, Asp N-G, Furda I, DeVries JW, Schweizer TF, Har-
cial support (25 % of total) is acknowledged from Project no. 225347/ land BF (1984) Determination of total dietary fiber in foods
F40 in The Research Council of Norway with funds from the Norwe- and food products: collaborative study. J Assoc Off Anal Chem
gian Levy on Agricultural Products. 68:677–679
14. Englyst HN, Quigley ME, Hudson GJ (1994) Determination

Compliance with ethical standards  of dietart fiber as non-starch polysaccharides with gas-liquid-
chromatographic, high-performance liquid-chromatographic or
Conflict of interest  The authors declare that they have no conflict of spectrophotometric measurement of constituent sugars. Analyst
interest. 119:1497–1509

13
Eur J Nutr

15. Helgerud T, Knutsen SH, Afseth NK, Stene KF, Rukke EO, uptake in rat intestinal brush border membrane vesicles. J Nutr
Ballance S (2016) Evaluation of hand-held instruments for rep- 119:1698–1704
resentative determination of glucose in potatoes. Potato Res 24. Tydeman EA, Parker ML, Faulks RM, Cross KL, Fillery-Travis
59:99–112 A, Gidley MJ, Rich GT, Waldron KW (2010) Effect of carrot
16. Brouns F, Bjorck I, Frayn K, Gibbs A, Lang V, Slama G,
(Daucus carota) microstructure on carotene bioaccessibility
Wolever T (2005) Glycaemic index methodology. Nutr Res Rev in the upper gastrointestinal tract. 2. In vivo digestions. J Agric
18:145–171 Food Chem 58:9855–9860
17. SS-ISO-6496 (2005) Animal feeding stuffs—determination
25. Tydeman EA, Parker ML, Wickham MS, Rich GT, Faulks RM,
of moisture and other volatile matter content (ISO 6496:1999, Gidley MJ, Fillery-Travis A, Waldron KW (2010) Effect of car-
IDT). Swedish Standards Institute, Stockholm rot (Daucus carota) microstructure on carotene bioaccessibilty in
18. NMKL (2005) ASH, gravimetric determination in foods. NMKL the upper gastrointestinal tract. 1. In vitro simulations of carrot
method No. 173, 2nd edn. Nordic Committee on Food Analysis digestion. J Agric Food Chem 58:9847–9854
19. EU (2011) Regulation (EU) No. 1169/2011 of the European Par- 26. Monro JA (2013) Kiwifruit, carbohydrate availability, and the
liament and of the Council on the provision of food information glycemic response. Adv Food Nutr Res 68:257–271
to consumers. Off J Eur Union L304:18–63 27. Takahashi T, Furuichi Y, Mizuno T, Kato M, Tabara A, Kawada
20. Vallejo F, Tomás-Barberán F, García-Viguera C (2003) Health- Y, Hirano Y, Ky Kubo, Onozuka M, Kurita O (2009) Water-hold-
promoting compounds in broccoli as influenced by refrigerated ing capacity of insoluble fibre decreases free water and elevates
transport and retail sale period. J Agric Food Chem 51:3029– digesta viscosity in the rat. J Sci Food Agric 89:245–250
3034. doi:10.1021/jf021065j 28. Takahashi T, Karita S, Ogawa N, Goto M (2005) Crystalline
21. Vallejo F, Tomás-Barberán FA, Ferreres F (2004) Characterisa- cellulose reduces plasma glucose concentrations and stimulates
tion of flavonols in broccoli (Brassica oleracea L. var. italica) by water absorption by increasing the digesta viscosity in rats. J
liquid chromatography–UV diode-array detection–electrospray Nutr 135:2405–2410
ionisation mass spectrometry. J Chromatogr 1054:181–193. 29. Takahashi T, Sakata T (2002) Large particles increase viscosity
doi:10.1016/j.chroma.2004.05.045 and yield stress of pig cecal contents without changing basic vis-
22. Griffiths DW (1981) The polyphenolic content and enzyme
coelastic properties. J Nutr 132:1026–1030
inhibitory activity of testas from bean (Vicia faba) and pea 30. Hardacre AK, Yap S-Y, Lentle RG, Monro JA (2015) The effect
(Pisum spp.) varieties. J Sci Food Agric 32:797–804 of fibre and gelatinised starch type on amylolysis and apparent
23. Welsch CA, Lachance PA, Wasserman BP (1989) Dietary phe- viscosity during in vitro digestion at a physiological shear rate.
nolic compounds: inhibition of Na+-dependent d-glucose Carbohydr Polym 123:80–88

13