You are on page 1of 12

Food Quality and Preference 55 (2017) 7990

Contents lists available at ScienceDirect

Food Quality and Preference


journal homepage: www.elsevier.com/locate/foodqual

The effect of genotypical and phenotypical variation in taste sensitivity


on liking of ice cream and dietary fat intake
Yuchi Shen a, Orla B. Kennedy b, Lisa Methven a,
a
b

Sensory Science Centre, Department of Food and Nutritional Sciences, University of Reading, Whiteknights, Reading RG6 6AP, UK
Hugh Sinclair Human Nutrition Unit, Department of Food and Nutritional Sciences, University of Reading, Whiteknights, Reading RG6 6AP, UK

a r t i c l e

i n f o

Article history:
Received 18 May 2016
Received in revised form 16 August 2016
Accepted 22 August 2016
Available online 24 August 2016
Keywords:
CD36
CA6
PROP taster status
Ice cream
Fat intake

a b s t r a c t
Emerging evidence suggests fat can be perceived as a taste. G-protein coupled receptors as well as CD36, a
fatty acid translocase, have been proposed to be involved in fat perception. Therefore, differences in number of receptors and genotype of CD36 have both been proposed to influence inter-individual fat taste
perception. Fungiform papillae density (FPD) and PROP taster status are phenotypes related to receptor
number. Previous authors have proposed an association between such phenotypes and CA6 (gustin) genotype, because the latter influences receptor cell proliferation. The effect of these factors on fat perception,
preference and intake, requires further investigation. Therefore, the main aim of this study was to investigate the effects of taste sensitivity, including both genotypic and phenotypic variation, on liking of ice
cream and dietary fat intake. Participants (n = 136) age 1855 years were recruited in the UK. Hedonic
liking results demonstrated that liking for ice cream was significantly affected by the fat content of the
sample, and by demographic factors (gender, ethnicity, age) but not by the consumers CD36 rs1761667
or CA6 rs2274333 genotype, PROP taster status nor FPD. However, categorising taste sensitivity from participant responses to salt taste alone (rather than to salt and PROP) found significant differences, with low
salt perceivers liking the high fat (20%) ice cream substantially more than medium- and high salt perceivers. This indicated that increased taste sensitivity reduced liking of high fat. Cluster analysis highlighted that one group of consumers (18%) liked higher fat ice cream, whereas another (30%) liked
lower fat ice cream compared to the 52% of consumers that liked ice cream regardless of fat content.
There was a significant association between these groups and salt taste sensitivity. Concerning recorded
dietary intake, the high-fat liker group were found to have substantially higher dairy product consumption compared to high-fat dislikers. Fat intake as a percentage of total energy intake was significantly
related to CA6 genotype, however the minor allele frequency at rs2274333 is too low to draw firm conclusions within this study population.
2016 Elsevier Ltd. All rights reserved.

1. Introduction
Overconsumption of energy dense foods, particularly high fat
foods, and specifically foods high in saturated fats and free sugars,
is a major cause of the obesity epidemic (Nasser, 2001; Mela, 2001;
WHO, 2013, 2015). Sensory perception and preferences for these
nutrients may contribute to such overconsumption and, hence, it
is important to understand and quantify the extent of interindividual differences.
Fat perception is multi-sensorial, comprising texture, aroma
and taste (Schiffman, Graham, Sattely-Miller, & Warwick, 1998;
Le Calv et al., 2015). The textural aspects of fat are perceived as
Corresponding author at: Department of Food and Nutritional Sciences, The
University of Reading, PO Box 226, Whiteknights, Reading RG6 6AP, UK.
E-mail address: l.methven@reading.ac.uk (L. Methven).
http://dx.doi.org/10.1016/j.foodqual.2016.08.010
0950-3293/ 2016 Elsevier Ltd. All rights reserved.

mouthfeel sensations by trigeminal nerves wrapped around papillae (Whitehead, Beeman, & Kinsella, 1985; Whitehead & Kachele,
1994; Essick, Chopra, Guest, & McGlone, 2003). Greater fungiform
papillae density (FPD) can increase fat texture sensitivity and those
with lower FPD may be less sensitive to the texture of fat (Essick
et al., 2003; Nachtsheim & Schlich, 2013). Hayes and Duffy
(2007) also suggested that subjects with a higher FP count gave
higher creaminess rating to milk-cream mixtures. Additionally,
Hayes, Sullivan, and Duffy (2010) suggested that subjects with
higher FPD had lower preference for high-fat, salty foods.
Nachtsheim and Schlich (2014) demonstrated that subjects with
lower fungiform papillae counts had higher consumption of high
fat milk and spreads than high fungiform papillae counts subjects.
Recently, results from Running, Craig, and Mattes (2015)
demonstrated that medium- and long chain non-esterified fatty
acid (NEFA) were perceived as a unique taste sensation, separate

80

Y. Shen et al. / Food Quality and Preference 55 (2017) 7990

from the other five basic tastes (sweet, bitter, sour, salty and
umami).
PROP taster status is widely used as a marker of taste acuity.
Individuals are commonly classified into three different PROP
taster status groups: supertasters, medium-tasters and nontasters,
according to their relative sensitivity to PROP and sodium chloride
(Tepper, Christensen, & Cao, 2001). A number of studies suggest
that PROP tasters (supertaster and medium-tasters) are not only
more sensitive to PROP solutions, but also perceive higher bitter
intensity from various foods, such as wine (Pickering, Simunkova,
& DiBattista, 2004), caffeinated coffee (Drewnowski, Henderson,
Levine, & Hann, 1999) and grapefruit juice (Drewnowski,
Henderson, & Shore, 1997), as well as to differences in the perception of sweeteners in soft drinks (Zhao & Tepper, 2007). In addition,
authors such as Tepper (2008) have reviewed the role of PROP sensitivity in food choice and nutritional intake. Studies have also suggested that PROP sensitivity is positively associated with fat
perception (Tepper & Nurse 1997, 1998; Hayes & Duffy, 2007;
Nachtsheim & Schlich, 2013). Hayes and Duffy (2007) found subjects with higher intensity ratings towards the bitter-tasting substance PROP gave higher creaminess ratings to high fat cream. It
also has been reported that preference for food with varied fat contents is influenced by PROP taster status. Anliker, Bartoshuk, Ferris,
and Hooks (1991) found that PROP tasters preferred whole milk
significantly more than non-tasters, however, tasters had a significantly lower preference for cheese. Tepper and Nurse (1997)
demonstrated that oiliness and fat ratings were both significantly
higher in PROP taster groups compared to nontasters for salad
dressings of either 10% or 40% fat. Moreover, they found that nontasters preferred the higher fat sample. However, Yackinous and
Guinard (2001) found that PROP taster status was not related to
fat perception in a range of foods (crisps, chocolate drink, milk
pudding and mashed potatoes).
More recent studies have revealed that fat can be detected as a
taste when both mouthfeel and aroma are masked (Mattes, 2009).
Animal (Gilbertson, Liu, Kim, Burks, & Hansen, 2005) and human
(Mattes, 2009; Chale-Rush, Burgess, & Mattes, 2007; Newman &
Keast, 2013; Haryono, Sprajcer, & Keast, 2014; Keast & Costanzo,
2015; Running et al., 2015) studies have suggested that fat might
be the sixth basic taste sensation and some researchers have found
that genotype can influence perception of fat taste. CD36 is responsible for transporting fatty acids and the presence of different single nucleotide polymorphisms (SNPs) of CD36 may influence fat
perception (Laugerette et al., 2005; Abumrad, 2005). Keller
(2012) found that individuals with CD36 A/A genotype (at
rs1761667) rated salad dressing as creamier than either those with
A/G or G/G genotype. Recently, Mrizak et al. (2015) also reported
that obese Tunisian women with CD36 A/A genotype had significantly higher oleic acid detection thresholds. Regarding CD36
genotype and fat preference, Keller (2012) found that the A/A subjects showed higher preference for added fats, oils and spreads,
using data from participants self-report food questionnaires.
Gustin (Carbonic anhydrase VI (CA6)) has been reported to
affect the maintenance and development of taste buds (Henkin,
Martin, & Agarwal, 1999). More recently, studies found that the
polymorphism rs2274333 strongly associated with fungiform
papillae density, where individuals with the G/G genotype had
lower FPD (Melis et al., 2013; Barbarossa et al., 2015). Moreover,
Melis et al. (2013) found that individuals with A/A genotype were
more likely to be PROP supertasters, whereas those with G/G were
more frequently PROP nontasters. As discussed before, fat perception has been reported to associate with papillae density and PROP
taster status. Moreover, considering the role of gustin in fungiform
papillae development and its close association with PROP sensitivity (Padiglia et al., 2010; Calo et al., 2011; Melis et al., 2013), gustin
may plausibly play a role in fat perception.

Previous research has investigated the relationship between


different dietary patterns and risk of obesity. Work since the
1980s (Drewnowski, 1985; Miller, Lindeman, Wallace, &
Niederpruem, 1990; Salbe, DelParigi, Pratley, Drewnowski, &
Tataranni, 2004; Deglaire et al., 2015) has shown that obese people
tend to prefer high-fat food and have higher fat consumption.
However, it is also known that cognitive control of eating behaviour plays an important role in dietary intake and BMI (Tepper
& Ullrich, 2002; Harden, Corfe, Richardson, Dettmar, & Paxman,
2009) and this should be accounted for in any study of diet. A number of questionnaires are used to assess eating behaviour traits. The
Three-Factor Eating Questionnaire (TFEQ) (Stunkard & Messick,
1985) is the most widely used to measure cognitive restraint, disinhibition and hunger in adults. The results from previous research
suggested that women with low restraint and high disinhibition
scores tend to have higher BMI (Dykes, Brunner, Martikainen, &
Wardle, 2004).
In summary, fat preference and intake are not only influenced
by taste perceptions and preferences, but also by demographic factors, such as gender, age and eating behaviour. However, it remains
unclear to what degree genetic and phenotypic measurements of
fat taste perception influence the liking of fat in foods and dietary
fat intake. Therefore, the main aims of this study were to investigate how variations in taste sensitivity influenced the liking of
ice cream samples with a range of fat levels, and to investigate
any association between taste sensitivity and dietary fat intake.
2. Materials and methods
2.1. Subjects
Healthy adults aged 1855 years, were recruited by advertisement within and around the University of Reading, UK. The subjects included in this study are the same cohort used in a study
of vegetable perception and liking previously reported (Shen,
Kennedy, & Methven, 2016). Exclusion criteria were: smokers,
pregnancy, relevant food allergies, major medical conditions and
medication used that could impact on taste perception. The study
was given a favourable opinion to proceed by the University of
Reading Ethics committee (study number 12/04).
2.2. Study design
Participants attended two separate study days (visit 1 and visit
2), which were up to two weeks apart, with each visit lasting one
hour. In the first visit, participants were asked to complete the
study consent form and pre-study questionnaires, and a buccal cell
swab was collected. Anthropometric measurements and FPD measurements were also completed in this visit. Between visit 1 and 2,
participants were asked to complete a Food Frequency Questionnaire (FFQ), 3-day food diary and Three Factor Eating Questionnaire (TFEQ). During visit 2, they rated their liking for a range of
ice cream samples, and completed a suprathreshold PROP sensitivity test. All tasting sections were performed in sensory booths in
University of Reading with room temperature controlled to
22 1 C.
2.3. Phenotypical measurements
2.3.1. Suprathreshold test
A suprathreshold test was used to classify PROP taster status of
the participants. Using the method described by Tepper et al.
(2001) in which both the perceived intensity of bitterness from
PROP (0.000032 M, 0.00032 M, 0.0032 M) and saltiness from
sodium chloride (0.01 M, 0.1 M, 1 M) are rated on a Labelled

81

Y. Shen et al. / Food Quality and Preference 55 (2017) 7990

Magnitude Scale (LMS), participants were classified within supertaster, medium-taster and nontaster groups. Data was collected
using Compusense 5 software (Ontario, Canada), for further details
see Shen et al. (2016).
In addition the salt intensity data collected from the
suprathreshold method was used to group participants according
to their rated perception of salt. For each individual their total
score for the three salt solutions was calculated and quartile analysis was carried out on these total scores. Individuals were categorized as low salt perceivers if their total rating was below the value
of the first quartile, medium perceivers if their rating was between
the first and third quartile, and high perceivers if their rating was
equal to or above the third quartile.
2.3.2. Fungiform papillae density
Tongue anterior was coloured using food-grade blue dye (Dr.
Oetker, UK) and two digital photographs were taken from each
participant using a digital microscope (400x magnification, Maplin
Gadget Electronics, UK). A 1 cm2 frame was attached to the end of
the microscope in order that the number of papillae in this area
(number/cm2) could be quantified by two researchers independently. As described by Shen et al. (2016), participants were classified into three groups (low-, medium- and high-FPD).
2.4. Genotypical measurements
Buccal cell samples were collected from participants using sterile Omni swabs (Whatman, UK) and stored in tubes at 20 C prior
to analysis. DNA was extracted using Omega Bio-Tek E.Z.N.A.
Forensic DNA Kit (Norcross, GA, USA) according to the manufacturers instructions. A NanoDrop ND1000 spectrophotometer (Nanodrop Technologies Inc., Rockland, USA) was used to assess DNA
quantity. Concentration of DNA was approximately 15 ng per well.
CD36 (rs1761667) and CA6 (rs2274333) polymorphisms were analysed using the Fluorogenic 50 nuclease assay (TaqMan SNP Genotyping Assay from Applied Biosystems, Foster City, USA).
Genotypes were determined in a 25 lL reaction mix that contained
5 lL DNA sample and 20 lL master mix solution, (6.25 lL DNAse/
RNAse free water, 0.625 lL Taq-Man SNP Genotyping Assay, and
12.5 lL TaqMan Genotyping Master mix (Applied Biosystems Inc.).
The parameters of PCR were as following: 0 min at 55 C followed
by 95 C for 10 min, 92 C for 15 s and 60 C for 1 min (GeneAmp
PCR system 2700, Applied Biosystems). The fluorescence profile
of each well was measured in an Applied Biosystems 7300 System
(Applied Biosystems Inc., Foster City, CA, USA).
2.5. Ice cream formulation and assessment
Four different fat content (6%, 10%, 15% and 20% fat content of
unaerated mix) ice cream samples were produced using the formulations shown in Table 1. All ingredients were homogenised and
pasteurised to 80 C (Boil/5 ice cream pasteuriser, Carpigiani,
Bologna, Italy) before addition of the vanilla flavour and were

refrigerated overnight at 2 C. The mix was then whipped and frozen in a horizontal batch freezer (T. Giusti Ltd, Burton on Trent, UK)
to an overrun of approximately 100% (50 ml air/100 ml finished
product) before depositing into 4 oz (114 ml) paper cups (Bunzl
Catering Supplies, Epsom, UK) and stored at 18 C.
To assess the sensory profile of the ice cream a trained sensory
panel (n = 11) with between 6 months and 4 years experience carried out a quantitative descriptive analysis. A consensus vocabulary of 39 attributes was first reached before scoring each sample
individually, in isolated sensory booths, in duplicate.
To assess the liking of ice cream samples with the study participants, a 9-point hedonic category scale was used for overall liking, liking of taste and liking of mouthfeel. In both the
profiling and liking sessions, sample pots were labelled with 3digit random codes and stored in a 18 C freezer during the sensory session, before serving monadically in a balanced presentation
order. Participants were asked to eat plain crackers and drink
water to remove the creamy aftereffect during one-minute break
between samples.
2.6. Anthropometrics and diet assessment
Body mass index was measured by Tanita body composition
analyser (Tanita BC-418 MA body composition analyser, USA). To
ensure consistency of measurement, all anthropometric measurements, including weight, height, waist and hip circumference, were
carried out by the same researcher. Height was measured in metres
(m) using a wall-mounted stadiometer. Body fat percentage and
weight were measured by Tanita (BC-418 MA body composition
analyser, USA).
In order to assess participants diet, all participants were asked
to complete 133-item Food Frequency Questionnaire (FFQ) (Oxford
food-frequency questionnaire used in European Prospective Investigation of Cancer (EPIC, UK)) based on their last years food consumption. In addition, they completed a 3-day food diary which
was analysed by Diet plan software (Dietplan 6, window version,
Forestfield software, UK) in order to obtain a measure of individual
dietary intake. In order to categorise eating behaviour the 51-item
Three Factor Eating Questionnaire (TFEQ) (Stunkard & Messick,
1985) was used. For restraint, scores of 010 were classified as
low restraint, 1113 high restraint, and 1421 clinical range of
restraint. For disinhibition, scores of 08 were classed as low disinhibition, 911 high disinhibition, and 1216 clinical range of disinhibition. For hunger, scores of 07 were classed as low
susceptibility to hunger, 810 high susceptibility to hunger, and
1114 clinical range of susceptibility to hunger (Harden et al.,
2009).
2.7. Data analysis
All statistical analyses of the study participant data were performed using XLSTAT (version 2015.6.01, France). Data from the
suprathreshold tests was collected on the LMS scale which has

Table 1
The ingredients used to make four different fat levels of ice cream.
Ingredients (g)

Double Cream (total fat 48%)


Sucrose
Dextrose monohydrate
Stabiliser Dairylux 442 (Branwell brand)
Skim milk powder (SMP)
Water
Vanilla flavouring (added after heat treatment)

Dairy fat levels in final products


6%

10%

15%

20%

625
430
200
37.5
600.1
3107.4
7.5

1041.5
430
200
37.5
570.3
2720.7
7.5

1565
430
200
37.5
532.9
2234.6
7.5

2083.5
430
200
37.5
495.9
1753.1
7.5

Number (N) in Bold was the number of participants in each category. Percentage in Bold was the proportion in the total population (N = 136). n represents the number within each category, and % represents the proportion. BMI
value = Mean SD (standard deviation).

4.2
3.9
3.8

22.9
22.6
23.5
37
37
26
3
3
2
33
59
8
4
7
1
22
52
26
26
60
30
25
50
25
3
6
3
31
63
6
5
10
1
23
50
27
23
51
28
33
50
17
2
3
1
17
43
40
8
21
19
28
56
16
25
49
14
20
44
36
8
18
15
26
55
19
25
52
18
24
52
24
High-FPD
Medium-FPD
Low-FPD
FPD groups

33
70
33

3.5
3.3
5.3

23.5
22.3
23.5
38
25
38
3
2
3
42
0
8
5
0
7
22
28
50
26
32
58
25
25
50
3
3
6
25
69
6
4
11
1
24
51
26
24
52
26
50
33
17
3
2
1
15
52
33
7
25
16
31
49
21
27
43
18
27
54
20
11
22
8
24
48
27
23
46
26
25
50
25
Low perceiver
Medium perceiver
High perceiver
Salt intensity group

34
68
34

5.0
2.8
4.5

23.4
22.3
23.7
26
37
37
2
3
3
33
59
8
4
7
1
28
49
23
32
57
27
25
50
25
3
6
3
38
56
6
6
9
1
28
47
25
28
48
26
17
66
17
1
4
1
31
42
27
15
20
13
26
53
21
23
47
18
15
56
29
6
23
12
34
46
20
32
44
19
28
49
23
Supertaster
Medium-taster
Nontaster
PROP Taster Status

38
67
31

3.6
4.2
3.9

22.4
23.2
23.2
25
50
25
2
4
2
33
59
8
4
7
1
41
54
5
47
63
6
25
67
8
3
8
1
13
81
6
2
13
1
41
52
7
42
53
7
100
0
0
6
0
0
25
62
13
12
30
6
47
50
3
41
44
3
37
51
12
15
21
5
40
56
4
38
53
4
39
54
7
A/A
A/G
G/G
CA6 genotypes

53
74
9

22.9
23.0
22.7
%

38
50
12
3
4
1

n
%

33
33
33
4
4
4

n
%

22
50
28
26
58
32

n
%

8
67
25
1
8
3

n
%

13
31
56
2
5
9

n
%

27
50
23
28
51
23

n
%

33
33
33
2
2
2

n
%

29
42
29
14
20
14

n
%

22
52
26
19
46
23

n
%

17
46
37
7
19
15

27
49
23

%
n

26
47
22
24
49
27

8 (6%)
12 (9%)
116 (85%)
12 (9%)
16 (12%)
102 (75%)
6 (4%)
48 (35%)
88 (65%)
41 (30%)

AGE-2
(3055)

African

Caucasian

East
Asian

West
Asian

Group 1
(12)

Group 2
(34)

Group 3
(58)

(Mean SD)

BMI (kg/m2)
Socio-economic groups
Ethnicity

Male

95 (70%)

Total
(n = 136)
N
%

33
66
37
A/A
A/G
G/G
CD36 genotypes

3.1.1. Phenotype and genotype characteristics


Table 2 summarises the demographic, taste phenotype and
taste genotype profiles of participants in the study. It demonstrates
the frequencies of CD36 and CA6 genotypes measured within the
study population. For CD36 (rs1761667), 24% of participants
were homozygous A/A, almost half of participants (49%) were

AGE-1
(1829)

One hundred and thirty-six healthy adults participated in the


study. As showed in Table 2, the majority of participants were
female (70% females and 30% males). Sixty-five percent of participants were classified as AGE-1 group (age 1829 years), and
remaining 35% as AGE-2 group (age 3055 years). In this study,
seventy-five percent of participants were Caucasian, and others
were from African and Asian ethnic groups. Most of participants
(85%) were classified as high-SEG, with 9% in the medium-SEG
and only 6% were from the low-SEG. The BMI range in current
study was 1743.5 kg/m2 with a mean of 22.9 0.34 kg/m2; 74%
of participants were in the normal weight range (18.525 kg/m2).

Female

3.1. Subject characteristics

Age

3. Results

Gender

semantic labels spaced in a quasi-logarithmic manner, the data


output was in log10 units which were antilogged before applying
any further analysis.
Participants were self-classified into four ethnic groups, including Caucasian, African, West Asian, and East Asian. Two age groups
were classified as AG-1 (aged from 18 to 29) and AG-2 (aged from
30 to 55). Social economic group (SEG) was defined according to
the 2010 National Statistics Socio-economic Classification Guidelines (Rose and Pevalin, 2010). In this study, classes 1 and 2 were
grouped together as the high-SEG, classes 3 and 4 as the
medium-SEG and classes 58 as the low-SEG.
Chi-square test was used to analyse the associations between
category data. Correlation between age and FPD was tested by
Pearsons correlation test. If two factors were significantly associated from Chi-square test, those factors could not be considered
as independent factors together within a subsequent analysis of
variance (ANOVA) or analysis of covariance (ANCOVA). Data collected from the 3-day food diaries was checked for outliers using
Grubbs test resulting in removal of data from 5 volunteers. FFQ
data were analysed by FETA software (University of Cambridge,
UK, FFQ entry and processing program). Correlation between data
from the diet diaries and the FFQ was tested by Pearsons correlation test. ANCOVA was conducted using Type II sum of square (SS)
to investigate which factors (ice cream samples; phenotypical and
genotypical measures of taste sensitivity; demographic factors)
had a significant effect on rated ice cream liking and dietary intake.
Interactions between taste sensitivity factors were tested and, for
ice cream liking, between ice cream sample samples and other factors. BMI data was found to follow a normal distribution and used
as the covariant throughout the ANCOVA data analysis. ANOVA
was used to test for any effect of factors (phenotypic and genotypic
measures of taste sensitivity; demographic factors; TFEQ factor
groups) on BMI. Tukeys HSD test was selected for post hoc comparisons of means and p < 0.05 was considered significant in all
analysis. Agglomerative Hierarchical Clustering (AHC) analysis
(proximity type was dissimilarities by Euclidean distance, agglomeration by Ward method) was used to group consumers based
upon their overall liking ratings of the ice cream samples. Sensory
profiling data of the ice cream samples was analysed using a mixed
model ANOVA in Senpaq version 4.1 (QI Statistics, Reading, UK)
where main effects were tested against the sample by assessor
interaction.

3.0
3.9
4.8

Y. Shen et al. / Food Quality and Preference 55 (2017) 7990

Table 2
Demographic, taste phenotype and taste genotype profiles of participants in the study.

82

83

Y. Shen et al. / Food Quality and Preference 55 (2017) 7990

heterozygous A/G and the remaining 27% were G/G. For gustin
(rs2274333), 39% of the study population had the putatively functional homozygous A/A genotype, more than half of participants
(54%) were heterozygous A/G and the remaining 7% participants
carried the G/G genotype.
According to the intensity rating from PROP suprathreshold
test, 28% of participants were supertasters, 49% were classified as
PROP medium-tasters and the remaining 23% participants were
nontasters. From quartile analysis of the salt intensity data, participants were grouped into perception groups according to their
overall ratings the three salt solutions: high salt perceivers (HSP,
total rating 129233), medium (MSP, total rating 58128) and
low (LSP, total rating 1857). Participants were also classified into
three FPD groups. Twenty-four percent participants were in the
high-FPD group (FPD range from 76 to 132 papillae/cm2), over half
of participants (52%) were in the medium-FPD group (FPD range
from 43 to 73 papillae/cm2) and 24% participants were the lowFPD group (FPD range from 12 to 42 papillae/cm2).

3.1.2. Eating behaviour categorisations


As shown in Table 3, participants were classified as three different groups in each factor from low to clinical range. A significant association was found between disinhibition groups and
susceptibility to hunger groups (p = 0.006), in which the majority
of participants (n = 94) fell within both the low susceptibility
group to hunger and low disinhibition groups. However, no significant association was found between the restraint groups and the
other two groups.

3.2. Relationship between sensory sensitivity (phenotype and


genotype) and demographic characteristics
There was no significant association found between the two
genotypes (CD36 and CA6; P = 0.31) in this study. Moreover, no
association was found between CD36 and PROP taster status
(p = 0.91), salt intensity group (p = 0.93), nor between CD36 and
FPD groups (p = 0.72). There was also no significant association
between CD36 and any demographic factor in this study (CD36 versus gender: p = 0.20; CD36 versus ethnicity groups: p = 0.095).
There were a larger proportion of females (27%) with A/A CD36
genotype compared to males A/A participants (17%) (Table 2),
however this observation is not conclusive due to the lower proportion of males in the study.
As reported in Shen et al. (2016), a significant association was
found between PROP taster status and FPD groups (p = 0.029)
and individuals perceiving greater bitterness from PROP had higher
fungiform papillae density. However, no significant association
was found between CA6 genotypes and PROP taster status
(p = 0.81), nor between CA6 and FPD groups (p = 0.6). There was
no significant association was found between salt perception group
and FPD group (p = 0.35), nor between CA6 genotypes and salt
group (p = 0.52). Additionally, no associations were found between
either PROP taster status or salt perception groups and any demographic factors. A significant association was found between ethnicity and CA6 genotypes (p = 0.016), but only a very small
proportion of participants had G/G genotype (n = 9). Despite a
greater proportion of males in the low-FPD (Female: 19%; Male:
37%), there was no significant association between gender and

Table 3
Categorization of subjects from Three Factor Eating Questionnaire (TFEQ).
Categories

Score range

Mean SD

Low restraint
High restraint
Clinical range of restraint

87
21
28

64
15
21

010
1113
1421

63
12 1
16 2

Low disinhibition
High disinhibition
Clinical range of disinhibition

110
16
10

81
12
7

08
911
1215

52
10 1
13 1

Low susceptibility to hunger


High susceptibility to hunger
Clinical range of susceptibility to hunger

109
21
6

80
15
5

07
810
1111

42
91
11 0

n presented the number of participants in each category, and % represents the proportion in the total population (N = 136).

Mean Liking (1-9)

5
Overall Liking
6%

10%

Liking of taste
15%

Liking of Mouthfeel

20% Fat Content (% w/w unaerated mix)

Fig. 1. Participant mean liking of four ice cream samples varying in fat content. Error bars represent standard error of the mean (N = 136).

84

Y. Shen et al. / Food Quality and Preference 55 (2017) 7990

mouthfeel substantially lower for the 20% fat ice cream compared
to their rating for other three samples (shown in Supplementary
Tables S1S3). Similarly, no effect of salt intensity group was found
in overall liking (p = 0.19), liking of taste (p = 0.32) and liking of
mouthfeel (p = 0.81). However, there was a significant interaction
between ice cream samples and salt intensity group for overall liking (p = 0.03), liking of taste (p = 0.02) and liking of mouthfeel
(p = 0.03). The LSP group rated their overall liking, liking of taste
and mouthfeel substantially higher for the 20% fat ice cream compared to the MSP (p = 0.017, p = 0.047 and p = 0.55 respectively)
and HSP (p = 0.12, p = 0.068 and p = 0.32 respectively) groups
(Fig. 3) (detail in Supplementary Tables S1S3).
For FPD group, no effect was found in overall liking (p = 0.063),
liking of taste (p = 0.11) and liking of mouthfeel (p = 0.11). There
were no significant interactions between ice cream samples and
FPD group, however there was a trend for the Low-FPD group to
rate their overall liking, liking of taste and mouthfeel lower for
the 20% fat sample (shown in Supplementary Tables S1S3).
Comparing the different CD36 genotypes, there was also no significant effect on overall liking (p = 0.24), liking of taste (p = 0.37)
nor liking of mouthfeel (p = 0.60) of the ice cream. There were no

FPD groups (p = 0.088) and the lower proportion of males in the


study population limit this finding. In this study, there was a weak
but significant negative correlation between age was FPD (correlation 0.19; p = 0.029).
3.3. Ice cream liking
As shown in Fig. 1, the percentage of fat in the ice cream significantly affected consumer liking (p < 0.0001). The 20% fat ice cream
was significantly less liked, but there was no difference between
the other three levels (6, 10 and 15% fat). A similar trend was also
found in liking of taste (p < 0.0001) and liking of mouthfeel
(p = 0.001).

Aribute Intensity (0-100)

3.3.1. The influence of sensory sensitivity (genotypes and phenotypes)


on liking of ice cream
In this study, PROP taster status did not show any significant
impact on overall liking (p = 0.99), liking of taste (p = 0.96) and liking of mouthfeel (p = 0.25). There were no significant interactions
between ice cream samples and PROP taster status, however, the
nontaster group rated their overall liking, liking of taste and

80.0
70.0
60.0
50.0
40.0
30.0
20.0
10.0
0.0

6%

10%

15%

20% Fat Content (% w/w unaerated mix)

Fig. 2a. Appearance and mouthfeel attributes found to differ significantly between ice cream samples formulated to different fat contents. Error bars represent +/ half the
Tukey HSD values. AE = aftereffect, evaluated 30 s post swallowing. Appearance attributes not found to differ significantly between samples were: glossy, sticky, density in
pot. Mouthfeel attributes not found to differ significantly between samples were: mouthdrying and mouthdrying aftereffect.

Aribute Intensity (0-100)

50.0
45.0
40.0
35.0
30.0
25.0
20.0
15.0
10.0
5.0
0.0
Dairy Fat
Odour
6%

Cardboard
Odour
10%

Acid (sour)
Taste

15%

20%

Dairyfat
(creamy)
Flavour

O Flavour
(Plasc or
Chemical)

Bier AE

Fat Content (% w/w unaerated mix)

Fig. 2b. Odour, taste and flavour attributes found to differ significantly between ice cream samples formulated to different fat contents. Error bars represent +/ half the
Tukey HSD values. AE = aftereffect, evaluated 30 s post swallowing. Odour attributes not found to differ significantly between samples were: vanilla, sweet, caramel, milk
powder, sour milk, overall odour intensity and off odour. Taste and flavour attributes not found to differ significantly between samples were: sweet, bitter, vanilla, caramel
and cooked milk, sweet aftereffect and sour aftereffect.

85

Y. Shen et al. / Food Quality and Preference 55 (2017) 7990

(i)
8.0

and liking of mouthfeel (p = 0.015). Mean liking ratings for high


salt perceivers with CD36 A/G genotype (n = 17) were higher than
those for MSP of the same genotype (n = 8) (overall liking 7.1 versus 6.3, p = 0.051; liking of taste 7.1 versus 6.2, p = 0.065). Within
HSP, the mean ratings from the CD36 A/G genotype (n = 17) were
higher than from the CD36 A/A genotype (n = 7) (overall liking
7.1 versus 5.9, p = 0.074; liking of taste 7.1 versus 5.8, p = 0.078;
liking of mouthfeel 6.8 versus 5.4, p = 0.027). Additionally within
liking of mouthfeel, the mean ratings for the HSP group with
CD36 A/A genotype (n = 7) were significantly lower than those for
the MSP of the same genotype (n = 18) (5.4 versus 6.9, p = 0.006).
There was some interaction between PROP taster status with
CD36 genotype for overall liking (p = 0.056), liking of taste
(p = 0.062) and liking of mouthfeel (p = 0.013). Mean liking ratings
for PROP supertaster participants with CD36 A/A genotype (n = 8)
were higher than those for PROP non-tasters with CD36 A/A genotype (n = 9) (overall liking 7.2 versus 6.4, p = 0.64; liking of taste
7.0 versus 6.4, p = 0.91; liking of mouthfeel 7.1 versus 6.3,
p = 0.68). This difference was not the case for PROP supertaster versus non-taster participants with CD36 G/G genotype (n = 12, n = 8
respectively). However, some of the pairwise comparisons are far
from significant and the number of participants in each of these
groups would need to be substantially higher to test whether such
interactions between CD36 genotype and taste sensitivity (PROP
taster status or salt intensity perception) are real effects.

Low Salt Percepon Group


Medium Salt Percepon Group
High Salt Percepon Group

Overall Liking (scale 1-9)

7.5
7.0
6.5
6.0
5.5
5.0
4.5
4.0
3.5
3.0
6%

10%

15%

20%

Fat Content of Ice Cream (%w/v)

(ii)
8.0

Low Salt Percepon Group


Medium Salt Percepon Group
High Salt Percepon Group

Liking of Taste (scale 1-9)

7.5
7.0
6.5
6.0
5.5
5.0
4.5
4.0
3.5
3.0
6%

10%

15%

20%

Fat Content of Ice Cream (%w/v)


Fig. 3. Mean liking of four ice cream samples varying in fat content where
participants were grouped by their salt perception ratings. (i) Overall liking of ice
cream; (ii) Liking of taste of ice cream. Error bars represent standard error of the
mean (N = 136). Salt perception groupings determined by quartile analysis of total
salt intensity ratings from the Supertaster test.

significant interactions between ice cream samples and CD36


group, however there was a consistency across the different
CD36 genotypes where all groups showed increased likings (overall
liking, liking of taste and mouthfeel) from 6% to 15% fat ice cream,
but the liking ratings dropped substantially for the 20% fat content
sample (shown in Supplementary Tables S1S3).
There was a significant difference between CA6 genotype on
overall liking (p = 0.031) but no significant difference between
any specific SNP pair. There was a significant difference in liking
of taste (p = 0.004), where the subjects with the G/G genotype at
SNP rs2274333 had a significantly lower liking of taste (overall
means of G/G: 5.9; A/G: 6.8; A/A: 6.5). Similarly, there a significant
difference in liking of mouthfeel (p = 0.028), but no significant difference between any specific SNP pair. However, these differences
must be treated with caution as the CA6 SNP rs2274333 G/G genotype groups was extremely small (n = 9).
There was only one significant interaction between taste genotype and taste phenotype which was between CD36 and salt intensity group, for overall liking (p = 0.028), liking of taste (p = 0.035)

3.3.2. The influence of demographic factors on liking of ice cream


Liking of ice cream was significantly different between males
and females (overall liking p = 0.001, liking of taste p = 0.002, liking
of mouthfeel p = 0.017), with males always rating higher. There
was a significant difference between ethnic groups for all liking
modalities (p < 0.0001, p < 0.0001 and p = 0.015 respectively);
however although the West Asian group gave higher ratings, the
numbers of participants in all non-Caucasian groups was too small
to draw conclusions. There were no significant differences in liking
of ice cream associated with age or SEG.
3.3.3. Liking clusters
Although the highest fat (20%) ice cream had the lowest mean
liking, not all consumers disliked it. Cluster analysis (AHC) identified three groups of participants with different liking patterns
(Table 4). Individuals in cluster 1 (52%) were non-discriminators
who gave consistently high liking ratings regardless of fat content.
However, those in cluster 2 (30%) were high-fat dislikers, where
mean liking decreased substantially once the fat content in the ice
cream samples increased above 15%. Cluster 3 (18%) could be
described as high-fat likers as their liking ratings were substantially lower at 6 and 10% fat content. A significant association was
found between overall liking clusters and salt intensity groups
(p = 0.02). The majority of high-fat dislikers were classified as
HSP and MSP, only 7% (n = 3) were LSP. Most of high-fat likers
were MSP (50%) and LSP (38%) and only 13% (n = 3) were HSP. From
chi-square tests, no significant associations were found between
the liking clusters and other taste genotypes, nor between the clusters and taste phenotypes.

Table 4
Mean of overall liking ratings of participants grouped by AHC.
Cluster

6% Fat

10% Fat

15% Fat

20% Fat

Cluster 1: nondiscriminator
Cluster 2: high-fat disliker
Cluster 3: high-fat liker

71
41
24

52%
30%
18%

7.6 0.1aA
6.5 0.2bcBC
4.5 0.4dD

7.7 0.1aA
6.1 0.3bcBC
5.6 0.4cdCD

7.5 0.2aA
6.9 0.3abAB
6.8 0.3abcABC

7.4 0.1aA
2.6 0.2eE
6.8 0.3abcABC

n = number of participants in each group, and % present the proportion in total number of participants (N = 136). Numbers in table represent the mean of overall liking rating
of each group for four ice cream samples. acMean values not sharing a common superscript lowercase letter in the same row differed significantly (p < 0.05) in same row. A
Mean values not sharing a common superscript uppercase letter in the same column differ significantly (p < 0.05).

86

Y. Shen et al. / Food Quality and Preference 55 (2017) 7990

3.4. Ice cream sensory profile

90

(i)

3.5. Dietary intake


The effects of various factors on dietary intake were evaluated
using data from the 3-day diet diaries (Sections 3.5.13.5.4). The
FFQ captures data from a longer time period than the diet dairies,
but due to its recall nature, is typically less accurate. Calculated
daily energy intake (kcal), fat intake (g) and fat as a percentage
of energy intake were all found to positively and significantly correlate between diet diary data and FFQ (correlation coefficients and
corresponding significance levels of 0.25 (p = 0.004), 0.23
(p = 0.009) and 0.25 (p = 0.01) respectively).
3.5.1. The influence of sensory sensitivity (genotypes and phenotypes)
on dietary intake
There were no significant relationships between any taste
genotype or phenotype on daily energy intake. There were significant interactions between the CA6 genotype with both PROP
taster status (p = 0.03) and salt intensity group (p = 0.02); however there were no significant pairwise comparisons in either
case, and the G/ G group was far too small (n = 9) to draw any
conclusions.
Regarding daily fat intake (g), there was a significant effect
of PROP taster status (p = 0.03) where supertasters had a higher
mean fat intake than medium and non-tasters (75.7 versus 62.6
and 62.9 g/day; p = 0.04, p = 0.1 respectively) (Fig. 4i). Although
there was no significant effect of genotype on fat intake (g),
there was a significant interaction between the CA6 genotype
and salt intensity group of (p = 0.02) but with no clear trend
and no significant pairwise comparison. Fat intake as a percentage of total energy intake was significantly related to CA6
genotype (p = 0.015) with the A/A genotype tending to have a
lower intake than the A/G genotype (31% compared to 34%,
p = 0.068).
There was no significant difference in BMI between taste phenotype or genotype groups, and indeed the mean BMI was very
similar for each group (Table 2); however as the representation
of underweight, overweight and obese participants in the study
was low, a firm conclusion cannot be made.

85
a
80

Fat Intake (g per day)

75
70

ab
b

65
60
55
50
45
40
Supertaster

Medium-taster

Liking Cluster 1 (Nondiscriminators, n=71)

Liking Cluster 2 (high fatdislikers, n=41)

Non-taster

(ii) 40
Fat Intake (% of Total Energy Intake)

Altering the fat content of the ice cream samples did alter the
sensory profile of the ice creams. Of 39 attributes assessed, 20 differed significantly between samples (Figs. 2a and 2b).
The sensory profile results conclude that the ice creams formulated to different fat contents differed in their sensory profile. The most striking differences in mouthfeel were, as
expected, that increasing fat content increased body (p = 0.01),
smoothness (p < 0.0001), mouthcoating (p = 0.01), creaminess
of mouthfeel (p < 0.0001), was slower to dissolve in mouth
(p = 0.0002) and led to greasiness on the lips (p = 0.04). The
6% fat formulation led significantly to more ice crystals being
felt in the mouth than compared to the 10% and 20% fat samples (sample effect p = 0.0005) and was perceived to be more
powdery than the 15% and 20% fat samples (p = 0.0001).
Increasing the cream in the ice cream formulation did lead to
an increase in dairy fat (creamy) flavour (p = 0.002) where the
6% and 10% fat ice creams were significantly lower than the
15% and 20% fat samples. The lower fat samples tended to be
slightly higher in off notes; the 6% fat sample being significantly higher in off-flavour than the 15% and 20% fat samples.
Fat appeared to mask acidic taste and bitter aftertaste in the
ice cream as these were significantly higher, although still
scored at low intensity, in the 6% fat sample compared to the
15% and 20% fat samples.

35

30

25

20
Liking Cluster 3 (high fatlikers, n=24)

Fig. 4. Mean fat intake (g) per day calculated from 3 day diet diaries; (i) differences
between PROP taster status groups (g/day), (ii) differences between ice cream liking
cluster groups (fat as % total energy intake/day). Error bars represent standard error
of the mean for each group. abBars with different letters indicate significant
difference at p < 0.05.

3.5.2. The influence of liking cluster groups on dietary intake


Liking cluster, resulting from ice cream tasting, had no significant relationship with total energy intake (p = 0.18) nor fat intake
(p = 0.21). However, the high-fat likers tended to have a substantially higher recorded fat intake (g per day) compared to the highfat dislikers and non-discriminator clusters (75 versus 66 and
63 g/day, p = 0.35, p = 0.13 respectively). The relationship between
liking cluster and fat intake as a percentage of energy consumed
was significant (p = 0.014), however in this case both the highfat likers and the high-fat dislikers tended to have a higher
intake compared to non-discriminators (34% and 35% versus
31%, p = 0.27 and p = 0.065 respectively) (Fig. 4ii).

3.5.3. The influence of eating behaviour on dietary intake and BMI


The three factors tested in the TFEQ (restraint, disinhibition and
susceptibility to hunger) did not show significant effects on selfreported total energy, nor on fat intake (g). However, there was a
tendency for restraint to relate to fat intake (p = 0.072) with the
high restraint group having the substantially lower intake than
the low restraint group (54.4 versus 69.0 g/day, p = 0.053). There
was a significant relationship between restraint and percent of

Y. Shen et al. / Food Quality and Preference 55 (2017) 7990

energy from fat (%) (p = 0.031), where the high restraint group
tended to have a substantially lower percentage fat intake than
the low restraint group (28% versus 33%, p = 0.056).
A significant difference in BMI was found between disinhibition
groups (p = 0.001), where the clinical range of disinhibition group
had significantly higher BMI than low disinhibition group (26.6
versus 22.5 kg/m2, p = 0.002)). There was no significant difference
in BMI between restraint groups (p = 0.51) nor susceptibility to
hunger groups (p = 0.18).

3.5.4. The influence of demographic factors on dietary intake, eating


behaviour and BMI
In this study, demographic factors did not showed any significant effects on total energy intake, nor on the percent of energy
from fat or dietary fat intake. However, there was a clear trend
for males to consume more energy than females (1937 versus
1812 kcal/day, p = 0.26) as would be expected. There was a significant association between gender and restraint groups (p = 0.005).
The majority of male participants (83%) belonged to the low
restraint group, the other 15% were in the clinical range of
restraint, and only 2% were high restraint. In the female group,
56% female were in the low restraint group, 21% were in the high
restraint groups, and the remaining 23% were in the clinical range
of restraint.
Males had significantly higher BMI compared to female participants (means 24.1 and 22.4 respectively) (p = 0.007) and the older
AGE-2 group had significantly higher BMI compared to the AGE-1
group (means 24.6 and 22.0 respectively) (p = <0.0001). The lowSEG group had significant higher BMI (mean 28.8) compared to
the other two groups (means 22.3 and 24.7, p < 0.0001 and
p = 0.032 respectively), however, this remains inconclusive due to
the lower number in the low-SEG (n = 8).

3.5.5. Dairy product consumption from the Food Frequency


Questionnaire (FFQ)
Daily dairy product consumption was calculated from FFQ.
There was no significant effect of any genotypes (CD36 (p = 0.78)
and CA6 (p = 0.48)), nor between any phenotypes (PROP taster status (p = 0.99) and FPD group (p = 0.92) or any demographic factors
(gender: p = 0.52, Ethnicity: p = 0.97, SEG: p = 0.72) on dairy product consumption. However, a significant effect of ice cream liking
cluster was found (p = 0.033) where the high-fat likers had substantially higher dairy product consumption compared to highfat dislikers (mean of high-fat likers versus high-fat dislikers: 391
g versus 272, p = 0.052) (Fig. 5).

Dairy product Consumption (g /


day)

500
450
400
350
300
250
200
150
100
50
0
Liking Cluster 1 (Non- Liking Cluster 2 (high Liking Cluster 3 (high
discriminators, n=71) fat-dislikers, n=41)
fat-likers, n=24)
Fig. 5. Mean intake of dairy products (g per day) calculated from food frequency
questionnaire. Error bars represent standard error of the mean for each group.
ab
Bars with different letters indicate significant difference at p < 0.05.

87

4. Discussion
In this study, both genotypes (CD36 (rs1761667) and CA6
(rs2274333)) and phenotypes (PROP taster status, salt sensitivity
groups and FPD group) of taste sensitivity measurements were
used to classify participants. Although PROP taster status is the
most common method to classify participants according to taste
sensitivity, it is a reflection on both overall taste sensitivity and
specific sensitivity to the thiourea group in PROP (Hayes & Keast,
2011), which varies with genotype for the bitter taste receptor
T2R38. As taste sensitivity to fat will not be effected by the specific
sensitivity to the thiourea group, we decided to separately take the
salt intensity measures from the PROP taster status test as a marker of general taste sensitivity (as described in Section 2.3). Considering the relationship between the genotype and phenotype of
taste sensitivity, the results concluded that there is no significant
association between CD36 and CA6. Moreover, no significant association was found between CD36 and either PROP taster status or
salt intensity group. Additionally, there was a lack of relationship
between CD36 and FPD groups in the current study. Gene expression results from animal (Laugerette et al., 2005; Fukuwatari
et al., 1997) and human (Simons, Kummer, Luiken, & Boon, 2011)
studies have found CD36 to be more associated with circumvallate
papillae than fungiform papillae, so the lack of relationship with
FPD was perhaps not surprising. Our results did not find any relationship between CA6 (gustin) genotype and FPD groups, nor
between CA6 and PROP taster status or salt taste intensity group.
Similar results were also found in Feeney and Hayes (2014) study,
as discussed in Shen et al. (2016).
The results from our study conclude that the taste genotypes
(CD36 and CA6) and phenotypes (PROP taster status and FPD
group) did not significantly affect liking ratings for ice cream samples varying in fat content. However, there was a trend for both
PROP non-tasters and the low-FPD group to show substantially
lower liking of the 20% fat content sample and, conversely, there
was a significant interaction between ice cream samples and salt
intensity group where the low salt perception (LSP) group rated
their liking (overall, taste and mouthfeel) substantially higher for
the 20% fat ice cream. From results of previous studies we predicted those of low taste sensitivity would like the higher fat sample more. A previous study by Tepper and Nurse (1997) found that
PROP supertasters (n = 25) and medium-tasters (n = 25) were able
to discriminate different in fat content 40% and 10% salad dressing,
but nontasters (n = 25) could not. Moreover, they also suggested
that nontasters showed higher hedonic preference for the 40% fat
salad dressing despite not being able to discriminate the difference
in fat content between the two samples (Tepper & Nurse, 1998).
Additionally, Keller, Steinmann, Nurse, and Tepper (2002) also
found PROP nontaster children preferred full-fat milk compared
to PROP tasters. Although this is in-line with the results from our
salt groupings, it contradicts those from our PROP and FPD groupings. However the sensory profiling results demonstrated that the
change in fat content led to numerous differences between the ice
cream samples. Although the higher fat samples were more
smooth, coating and creamy in mouthfeel, as well as higher in
dairy fat flavour, they were also less powdery and lower in ice crystals in the mouth. Therefore, the PROP supertaster and high FPD
group results, where mean liking was not significantly different
between the low and high fat content samples, are perhaps less
surprising. Higher sensitivity to the ice and powdery mouthfeel
in the low fat samples might have reduced liking for some of these
participants; but also their higher sensitivity to the coating and
creamy mouthfeel in the high fat samples might reduce liking for
other supertasters. Hence leading to no overall effect on the mean
liking results for these groups. Overall therefore, although difference in mouthfeel and taste perception have influenced the liking

88

Y. Shen et al. / Food Quality and Preference 55 (2017) 7990

of fat content in the ice cream, it is difficult to predict whether it is


the mouthfeel or taste differences that will have the greatest effect
on liking, and in a real food matrix such as ice cream, a change in
fat content does influence many mouthfeel and flavour attributes.
Keller et al. (2012) found CD36 genotype to affect preference for
fat. In their study, five different SNPs of CD36 gene were analysed,
including the one used in our study (rs1761667). They found that
participants with A/A genotype at rs1761667 perceived greater
creaminess from salad dressings compared to the other genotype
groups (G/G & A/G), but that this A/A group had higher preference
for added fats and oils. However, we need to consider that CD36
affects fatty acid transportation and that the level of free fatty acids
in the ice cream samples would be low. Therefore, it is perhaps
unsurprising that CD36 genotype did not directly influence liking
of fat level in the ice cream. In addition, the previous study
(Keller et al., 2012) recruited 317 African-Americans whereas the
current study recruited predominantly Caucasian-British and it is
not yet clear whether the phenotypical traits resulting from this
genotype are expressed within different ethnic groups. However,
there was one interesting significant interaction in our study
between CD36 genotype and salt perception group, whereby those
of high salt perception (ie more taste sensitive) liked ice cream less
overall if they were A/A genotype compared to A/G or G/G genotype. In fact for the 20% fat ice cream sample the mean liking rating
for the HSP A/A group was as low as 3.6 (compared to 6.5 and 5.4
for the HSP A/G and G/G groups respectively). This is partly in-line
with results by Keller et al. (2012) where the A/A genotype perceived greater creaminess, as in the present study it might be the
higher overall perception of the HSP A/A group that has reduced
their liking for ice cream.
In our study, fat content significantly influenced liking ratings,
and the highest fat content sample was liked the least overall.
The 20% fat sample may have exceeded a threshold of fat perception (taste and/or mouthfeel) resulting in a much stronger sensation compared with other samples, which resulted in lower liking
ratings. Indeed the sensory profiling panel rated the 20% fat sample
as most intense in dairy fat flavour, as well as smoothness, mouthcoating and creamy mouthfeel, and although the sample was not
significantly different from the 15% fat sample using the Tukey
HSD test, the difference in means for creamy mouthfeel was 10
units (0100 scale) higher. However, not everyone disliked the
20% fat samples. In the current study, 18% participants increased
their liking rating for ice cream as the fat level increased and rated
liking of the 20% fat ice cream sample significantly higher than the
6% sample, which cannot be explained by the genotypes or phenotypes of taste sensitivity. Differences in daily dairy consumption
might contribute to this, high-fat likers in our study had substantially higher dairy product intake compared to the high-fat dislikes.
Results from a recent study by Bakke, Shehan, and Hayes (2016)
showed a similar trend; participants who reported typically consuming whole milk in their diet preferred milks with a higher fat
content. We speculate that the results may be specific to the fat
type; the cluster that disliked high fat in ice cream (ie dairy fat)
consumed less dairy products (Fig. 5), however they did not consume less fat in their diet (Fig. 4ii).
In our current study, two questionnaires were used to investigate dietary intake, the FFQ and a 3-day diet diary. Although there
was a significant positive correlation between data from these two
measures, the relationship was not strong which might be
expected as they represent two different measures. The 3-day diaries are usually more accurate than the FFQ as they do not rely on
recall, but they represent only a snapshot time interval. We
decided to use the diet diary data to relate to taste sensitivity as
it was expected to provide the more accurate data set. Restrained
eaters were not excluded from the study and therefore, as expected
restrained eating resulted in lower fat intake. Before the study we

predicted that PROP non-tasters and participants with a low FPD


would have a higher fat intake. In a previous study Keller et al.
(2002) reported that PROP nontaster girls (aged 45 years) showed
higher preference for full-fat milk and consumed more discretionary fat. Studying FPD, Nachtsheim and Schlich (2014) found
that participants with low FPD consumed more high fat milk and
spread, although they found no significant influence of FPD on
energy intake recorded by diet-diary. In our study we found no
effect of FPD or salt taste intensity group on fat intake, and
although we found a significant relationship between PROP taster
status and intake (g fat/day), it was in the opposite direction to
expected. Supertasters had a higher mean fat intake than medium
and non-tasters. This is, however, in-line with results from
Yackinous and Guinard (2002) where female PROP supertasters
and medium-tasters were reported to have a higher percentage
of dietary energy from fat than female nontasters (although we
did not find an interaction with gender in our study). In contrast
again, Borazon, Villarino, Magbuhat, and Sabandal (2012) found
no relationship between PROP taster status and fat intake. The reason for such different results might be the different methodological
approaches taken to assess dietary intake. FFQ was used in
Yackinous and Guinard (2002) and Keller et al. (2002) study,
whereas Borazon et al. (2012) and the current study used 3-day
food diaries. In addition, different age range of the participants
might be also lead to differences in energy and fat intake, as
Keller et al. (2002) studied preschool children (aged 45 years),
Borazon et al. (2012) studied adolescents, Yackinous and Guinard
(2002) focused on college students (mean age 20 years) and the
current study investigated a wider age range (1855 years). Moreover, as discussed earlier, an increase in taste and mouthfeel sensitivity, is likely to influence liking of foods varying in fat content,
and hence choice and intake of such foods, in more than one direction as fat effects numerous attributes within food.
As previous studies have related CD36 rs1761667 genotype to
preference for fat in foods (Keller et al., 2012) and CA6 rs2274333
to taste sensitivity (Padiglia et al., 2010; Calo et al., 2011; Melis
et al., 2013), the current study investigated whether these genotypes influenced dietary fat intake. In our study, CD36 rs1761667
did not show a significant effect on total energy and fat intake.
However, fat intake as a percentage of total energy intake was significantly related to CA6 rs2274333 genotype with the A/A genotype tending to have a lower intake than the A/G genotype.
Although no previous studies have related CA6 rs2274333 genotype
to dietary fat intake, previous studies found the A/A genotype to be
associated with supertasters (Padiglia et al., 2010; Calo et al., 2011;
Melis et al., 2013). This led to the prediction that those with the A/
A genotype might consumer less fat, and indeed this is concluded
in our study results. However, our phenotype data did not support
this finding and it must be considered with caution as the minor
allele frequency in CA6 rs2274333 is very low.
5. Limitations
There are several limitations in present study. Firstly, ice cream
is as very popular dairy product in the west. This might bias liking
ratings, due to both its familiarity and its level of sweetness. Secondly, the overall all number of participants in this study
(n = 136) is small, particularly the number of participants with gustin CA6 G/G genotypes (n = 9), and as results of less representation
of results from this groups of participants. A low proportion of CA6
G/G participants was expected from previous research; however
the proportion of our study population with this genotype was
only than in the previous studies (7% compared to 13% in the
Melis et al., 2013 study; 16% in the Barbarossa et al., 2015 study).
Additionally, although the majority of the participants (n = 102)
were CaucasianBritish, other ethnic groups were included, but

Y. Shen et al. / Food Quality and Preference 55 (2017) 7990

not at representative group sizes, e.g. Asian (n = 28) African (n = 6),


and the functionality of the genes may vary between ethnic groups.
In this study we did not investigate for relationships between
taste genotypes and phenotypes on BMI, because the majority
(74%) of participants were in the normal weight BMI range
(18.525 kg/m2). A wider range of BMI is needed to investigate
such relationships as the current literature is sparse; for example
Yackinous and Guinard (2002) reported BMI did not relate to PROP
taster status, and Keller et al. (2012) reported BMI was not associated with CD36 rs1761667 genotype, however they did reported an
association with another CD36 SNP (rs3840546) in AmericanAfrican females (n = 317).
Furthermore, in the current study, DNA was extract from buccal
swab cell samples, which has been report to provide lower
amounts of DNA than saliva or blood samples (Satia-Abouta
et al., 2002; Mulot, Stucker, Clavel, Beaune, & Loriot, 2005;
Archer et al., 2016) and is, therefore, insufficient for further gene
expression analysis. There are many factors that contribute to food
choice, hence taste sensitivity and taste receptor genotype will
only provide part of a fundamental understanding of food preference and choice.
6. Conclusion
The findings of this study demonstrated that liking for ice cream
was significantly influenced by fat content, but not by FPD, PROP
taster status, nor by CD36 or CA6 genotype. However, categorising
taste sensitivity from participant responses to salt taste alone
(rather than to salt and PROP) found significant differences, with
low salt perceivers liking 20% fat ice cream substantially more than
medium- and high salt perceivers. There were groups of consumers
that liked higher fat and lower fat ice cream, compared to the
majority of consumers that liked ice cream regardless of fat content. These three groups could not be solely explained by taste
genotype or phenotype, however, there was a significant association between the groups and salt taste sensitivity. Liking of high
fat in ice cream was found to be associated with higher selfrecorded dairy product consumption. Within the cohort examined,
the only significant relationship between taste genotype related to
dietary intake was for CA6, with the rs2274333 A/A genotype tending to have a lower intake of fat as a percentage of energy intake
than the A/G genotype. However the minor allele frequency at
rs2274333 is too low to draw firm conclusions within this study
population. Incorporation of direct measures of fat perception into
future studies would help to understand the role of fat perception
in fat liking and intake.
Appendix A. Supplementary data
Supplementary data associated with this article can be found, in
the online version, at http://dx.doi.org/10.1016/j.foodqual.2016.08.
010.
References
Abumrad, N. A. (2005). CD36 may determine our desire for dietary fats. Journal of
Clinical Investigation, 115, 29652967.
Anliker, J. A., Bartoshuk, L., Ferris, A. M., & Hooks, L. D. (1991). Childrens food
preferences and genetic sensitivity to the bitter taste of 6-n-propylthiouracil
(PROP). American Journal of Clinical Nutrition, 54, 316320.
Archer, N. S., Liu, D., Shaw, J., Hannan, G., Duesing, K., & Keast, R. (2016). A
comparison of collection techniques for gene expression analysis of human oral
taste tissue. PLoS One, 11, e0152157.
Bakke, A. J., Shehan, C. V., & Hayes, J. E. (2016). Type of milk typically consumed, and
stated preference, but not health consciousness affect revealed preferences for
fat in milk. Food Quality and Preference, 49, 9299.
Barbarossa, I. T., Melis, M., Mattes, M. Z., Calo, C., Muroni, P., Crnjar, R., et al. (2015).
The gustin (CA6) gene polymorphism, rs2274333 (A/G), is associated with

89

fungiform papilla density, whereas PROP bitterness is mostly due to TAS2R38 in


an ethnically-mixed population. Physiology & Behavior, 138, 612.
Borazon, E. Q., Villarino, B. J., Magbuhat, R. M. T., & Sabandal, M. L. (2012).
Relationship of PROP (6-n-propylthiouracil) taster status with body mass index,
food preferences, and consumption of Filipino adolescents. Food Research
International, 47, 229235.
Calo, C., Padiglia, A., Zonza, A., Corrias, L., Contu, P., Tepper, B. J., et al. (2011).
Polymorphisms in TAS2R38 and the taste bud trophic factor, gustin gene cooperate in modulating PROP taste phenotype. Physiology & Behavior, 104,
10651071.
Chale-Rush, A., Burgess, J. R., & Mattes, R. D. (2007). Evidence for human orosensory
(taste?) sensitivity to free fatty acids. Chemical Senses, 32, 423431.
Deglaire, A., Mejean, C., Castetbon, K., Kesse-Guyot, E., Hercberg, S., & Schlich, P.
(2015). Associations between weight status and liking scores for sweet, salt and
fat according to the gender in adults (The Nutrinet-Sante study). European
Journal of Clinical Nutrition, 69, 4046.
Drewnowski, A. (1985). Food perceptions and preferences of obese adults: A
multidimensional approach. International Journal of Obesity, 9, 201212.
Drewnowski, A., Henderson, S. A., Levine, A., & Hann, C. (1999). Taste and food
preferences as predictors of dietary practices in young women. Public Health
Nutrition, 2, 513519.
Drewnowski, A., Henderson, S. A., & Shore, A. B. (1997). Taste responses to naringin,
a flavonoid, and the acceptance of grapefruit juice are related to genetic
sensitivity to 6-n-propylthiouracil. The American Journal of Clinical Nutrition, 66,
391397.
Dykes, J., Brunner, E. J., Martikainen, P. T., & Wardle, J. (2004). Socioeconomic
gradient in body size and obesity among women: The role of dietary restraint,
disinhibition and hunger in the Whitehall II study. International Journal of
Obesity and Related Metabolic Disorders, 28, 262268.
Essick, G., Chopra, A., Guest, S., & McGlone, F. (2003). Lingual tactile acuity, taste
perception, and the density and diameter of fungiform papillae in female
subjects. Physiology & Behavior, 80, 289302.
Feeney, E. L., & Hayes, J. E. (2014). Exploring associations between taste perception,
oral anatomy and polymorphisms in the carbonic anhydrase (gustin) gene CA6.
Physiology & Behavior, 128, 148154.
Fukuwatari, T., Kawada, T., Tsuruta, M., Hiraoka, T., Iwanaga, T., Sugimoto, E., et al.
(1997). Expression of the putative membrane fatty acid transporter (FAT) in
taste buds of the circumvallate papillae in rats. FEBS Letters, 414, 461464.
Gilbertson, T. A., Liu, L., Kim, I., Burks, C. A., & Hansen, D. R. (2005). Fatty acid
responses in taste cells from obesity-prone and -resistant rats. Physiology &
Behavior, 86, 681690.
Harden, C. J., Corfe, B. M., Richardson, J. C., Dettmar, P. W., & Paxman, J. R. (2009).
Body mass index and age affect three-factor eating questionnaire scores in male
subjects. Nutrition Research, 29, 379382.
Haryono, R. Y., Sprajcer, M. A., & Keast, R. S. (2014). Measuring oral fatty acid
thresholds, fat perception, fatty food liking, and papillae density in humans.
Journal of Visualized Experiments, 8, e51236.
Hayes, J. E., & Duffy, V. B. (2007). Revisiting sugar-fat mixtures: Sweetness and
creaminess vary with phenotypic markers of oral sensation. Chemical Senses, 32,
225236.
Hayes, J. E., & Keast, R. S. (2011). Two decades of supertasting: Where do we stand?
Physiology & Behavior, 104, 10721074.
Hayes, J. E., Sullivan, B. S., & Duffy, V. B. (2010). Explaining variability in sodium
intake through oral sensory phenotype, salt sensation and liking. Physiology &
Behavior, 100, 369380.
Henkin, R. I., Martin, B. M., & Agarwal, R. P. (1999). Efficacy of exogenous oral zinc in
treatment of patients with carbonic anhydrase VI deficiency. The American
Journal of the Medical Sciences, 318, 392405.
Keast, R. S. J., & Costanzo, A. (2015). Is fat the sixth taste primary? Evidence and
implications. Flavour, 4, 17.
Keller, K. L. (2012). Genetic influences on oral fat perception and preference:
Presented at the symposium The taste for fat: New discoveries on the role of
fat in sensory perception, metabolism, sensory pleasure and beyond held at the
Institute of Food Technologists 2011 Annual Meeting, New Orleans, LA, June 12,
2011. Journal of Food Science, 77, S143S147.
Keller, K. L., Liang, L. C., Sakimura, J., May, D., van Belle, C., Breen, C., ... Chung, W. K.
(2012). Common variants in the CD36 gene are associated with oral fat
perception, fat preferences, and obesity in African Americans. Obesity (Silver
Spring), 20, 10661073.
Keller, K. L., Steinmann, L., Nurse, R. J. J., & Tepper, B. (2002). Genetic taste sensitivity
to 6-n-propylthiouracil influences food preference and reported intake in
preschool children. Appetite, 38, 312.
Laugerette, F., Passilly-Degrace, P., Patris, B., Niot, I., Febbraio, M., Montmayeur, J. P.,
et al. (2005). CD36 involvement in orosensory detection of dietary lipids,
spontaneous fat preference, and digestive secretions. The Journal of Clinical
Investigation, 115, 31773184.
Le Calv, B., Saint-Lger, C., Babas, R., Gelin, J. L., Parker, A., Erni, P., et al. (2015). Fat
perception: How sensitive are we? Journal of Texture Studies, 46, 200211.
Mattes, R. D. (2009). Oral detection of short-, medium-, and long-chain free fatty
acids in humans. Chemical Senses, 34, 145150.
Mela, D. J. (2001). Determinants of food choice: Relationships with obesity and
weight control. Obesity Research, 9, 249S255S.
Melis, M., Atzori, E., Cabras, S., Zonza, A., Calo, C., Muroni, P., ... Tomassini
Barbarossa, I. (2013). The gustin (CA6) gene polymorphism, rs2274333 (A/G), as
a mechanistic link between PROP tasting and fungiform taste papilla density
and maintenance. PLoS One, 8, e74151.

90

Y. Shen et al. / Food Quality and Preference 55 (2017) 7990

Miller, W. C., Lindeman, A. K., Wallace, J., & Niederpruem, M. (1990). Diet
composition, energy intake, and exercise in relation to body fat in men and
women. American Journal of Clinical Nutrition, 52, 426430.
Mrizak, I., Sery, O., Plesnik, J., Arfa, A., Fekih, M., Bouslema, A., ... Khan, N. A. (2015).
The A allele of cluster of differentiation 36 (CD36) SNP 1761667 associates with
decreased lipid taste perception in obese Tunisian women. British Journal of
Nutrition, 113, 13301337.
Mulot, C., Stucker, I., Clavel, J., Beaune, P., & Loriot, M. A. (2005). Collection of human
genomic DNA from buccal cells for genetics studies: Comparison between
cytobrush, mouthwash, and treated card. Journal of Biomedicine and
Biotechnology, 2005, 291296.
Nachtsheim, R., & Schlich, E. (2013). The influence of 6-n-propylthiouracil
bitterness, fungiform papilla count and saliva flow on the perception of
pressure and fat. Food Quality and Preference, 29, 137145.
Nachtsheim, R., & Schlich, E. (2014). The influence of oral phenotypic markers and
fat perception on fat intake during a breakfast buffet and in a 4-day food record.
Food Quality and Preference, 32, 173183.
Nasser, J. (2001). Taste, food intake and obesity. Obesity Review, 2, 213218.
Newman, L. P., & Keast, R. S. J. (2013). The test-retest reliability of fatty acid taste
thresholds. Chemosensory Perception, 6, 7077.
Padiglia, A., Zonza, A., Atzori, E., Chillotti, C., Calo, C., Tepper, B. J., et al. (2010).
Sensitivity to 6-n-propylthiouracil is associated with gustin (carbonic
anhydrase VI) gene polymorphism, salivary zinc, and body mass index in
humans. The American Journal of Clinical Nutrition, 92, 539545.
Pickering, G. J., Simunkova, K., & DiBattista, D. (2004). Intensity of taste and
astringency sensations elicited by red wines is associated with sensitivity to
PROP (6-n-propylthiouracil). Food Quality and Preference, 15, 147154.
Rose, D., & Pevalin, D. (2010). The national statistics socio-economic classification:
(Rebased on the SOC2010) user manual (Vol. 3, pp. 170). Basingstoke: Palgrave
Macmillan.
Running, C. A., Craig, B. A., & Mattes, R. D. (2015). Oleogustus: The unique taste of
fat. Chemical Senses, 40, 507516.
Salbe, A. D., DelParigi, A., Pratley, R. E., Drewnowski, A., & Tataranni, P. A. (2004).
Taste preferences and body weight changes in an obesity-prone population. The
American Journal of Clinical Nutrition, 79, 372378.
Satia-Abouta King, I. B., Abouta, J. S., Thornquist, M. D., Bigler, J., Patterson, R. E.,
Kristal, A. R., Shattuck, A. L., Potter, J. D., & White, E. (2002). Buccal cell DNA
yield, quality, and collection costs: Comparison of methods for large-scale
studies. Cancer Epidemiology, Biomarkers & Prevention, 11, 11301133.
Schiffman, S. S., Graham, B. G., Sattely-Miller, E. A., & Warwick, Z. S. (1998).
Orosensory perception of dietary fat. Current Directions in Psychological Science,
7, 137143.

Shen, Y., Kennedy, O. B., & Methven, L. (2016). Exploring the effects of genotypical
and phenotypical variations in bitter taste sensitivity on perception, liking and
intake of brassica vegetables in the UK. Food Quality and Preference, 50, 7181.
Simons, P. J., Kummer, J. A., Luiken, J. J., & Boon, L. (2011). Apical CD36
immunolocalization in human and porcine taste buds from circumvallate and
foliate papillae. Acta Histochemica, 113, 839843.
Stunkard, A. J., & Messick, S. (1985). The three-factor eating questionnaire to
measure dietary restraint, disinhibition and hunger. Journal of Psychosomatic
Research, 29, 7183.
Tepper, B. J. (2008). Nutritional implications of genetic taste variation: The role of
PROP sensitivity and other taste phenotypes. Annual Review of Nutrition, 28,
367388.
Tepper, B. J., Christensen, C. M., & Cao, J. (2001). Development of brief methods to
classify individuals by PROP taster status. Physiology & Behavior, 73, 571577.
Tepper, B. J., & Nurse, R. J. (1997). Fat perception is related to PROP taster status.
Physiology & Behavior, 61, 949954.
Tepper, B., & Nurse, R. (1998). PROP taster status is related to fat perception and
preference. Annals of the New York Academy of Sciences, 855, 802804.
Tepper, B. J., & Ullrich, N. V. (2002). Influence of genetic taste sensitivity to 6-npropylthiouracil (PROP), dietary restraint and disinhibition on body mass index
in middle-aged women. Physiology & Behavior, 75, 305312.
Whitehead, M. C., Beeman, C. S., & Kinsella, B. A. (1985). Distribution of taste and
general sensory nerve endings in fungiform papillae of the hamster. American
Journal of Anatomy, 173, 185201.
Whitehead, M. C., & Kachele, D. L. (1994). Development of fungiform papillae, taste
buds, and their innervation in the hamster. The Journal of Comparative
Neurology, 340, 515530.
WHO (2013). World health authority, global action plan for the prevention of noncommunicable diseases, 20132020. Geneva, Switzerland: World Health
Organization.
WHO (2015). Guideline: Sugars intake for adults and children. Geneva, Switzerland:
World Health Organization.
Yackinous, C., & Guinard, J.-X. (2001). Relation between PROP taster status and fat
perception, touch, and olfaction. Physiology & Behavior, 72, 427437.
Yackinous, C. A., & Guinard, J.-X. (2002). Relation between PROP (6-npropylthiouracil) taster status, taste anatomy and dietary intake measures for
young men and women. Appetite, 38, 201209.
Zhao, L., & Tepper, B. J. (2007). Perception and acceptance of selected high-intensity
sweeteners and blends in model soft drinks by propylthiouracil (PROP) nontasters and super-tasters. Food Quality and Preference, 18, 531540.

You might also like