The Association of Taste with Change in
Adiposity-Related Health Measures
Mary E. Fischer, PhD; Karen J. Cruickshanks, PhD; Carla R. Schubert, MS; Alex Pinto, MS; Guan-Hua Huang, PhD; Barbara E. K. Klein, MD;
Ronald Klein, MD; James S. Pankow, PhD
ARTICLE INFORMATION Article history:
Accepted 10 April 2014 Available online 29 May 2014
Keywords: Taste intensity Adiposity Cluster analysis Longitudinal Food hedonics
Copyrightª 2014 by the Academy of Nutrition and Dietetics.
2212-2672/$36.00
http://dx.doi.org/10.1016/j.jand.2014.04.013
ABSTRACT
The relationship between taste-intensity patterns and 5-year change in adiposity-related health measures was determined. Participants were members of the Beaver Dam Offspring Study, a study of the adult children of participants in the population-based Epidemiology of Hearing Loss Study. There were 1,918 participants (mean base-line age¼48.8 years; range¼22 to 84 years) with baseline taste (2005 to 2008) and follow-up (2010 to 2013) data. Outcomes included 5-year change in body mass index, waist circumference, blood pressure, nonehigh-density lipoprotein cholesterol, and glycosylated hemoglobin A1c, and hedonic ratings of specific foods. Cluster analysis with Ward’s minimum variance method identified the following 5 patterns of the suprathreshold taste intensities of salt, sweet, sour, and bitter: salt and sweet intensities slightly above population averages, average sour and bitter intensities; salt, sour, and bitter intensities above population average, average sweet intensity; salt, sour, and bitter intensities above population average, sweet intensity substantially above average; all intensities below population averages; and all intensities close to population average. The General Linear Model procedure was used for testing cluster differences in the outcomes. With covariate adjustment, the group with all intensities close to population averages had a significantly lower average increase in body mass index compared with the group with above-average intensities for salt, sour, and bitter (þ0.4 vs þ0.9), and in glycosylated hemoglobin A1c compared with the group with above-average intensities for all tastes (þ0.20% vs þ0.34%). Clusters differed in the hedonics of foods representing sweetness and saltiness. The study’s findings provide evidence that perceived taste intensity might be related to changes in adiposity-related health.
J Acad Nutr Diet. 2014;114:1195-1202.
F
OOD CHOICE PLAYS A ROLE IN TOTAL CALORIC intake and in the maintenance of health. The rela-tionship between dietary choices and adiposity has been of particular interest because of the association of obesity and central adiposity with chronic diseases, such as diabetes, cardiovascular disease, and cancer.1-3Previous cross-sectional studies have reported associations between specific dietary patterns and body mass index (BMI) and body fat distribution.4-6Prospectively, a link between food-choice patterns and change in BMI and waist circumference was observed in the Baltimore Longitudinal Study of Aging7andin the Framingham Offspring Cohort, subjects with higher Mediterranean-style dietary pattern scores were found to
have significantly smaller waist circumferences after approx-imately 7 years of follow-up.8
Many factors are involved in food choice and consumption, including taste, food preference, familiarity with food items, level of education, cultural habits, cooking habits, health at-titudes, weight concerns and dietary restraint, genetics, cost, availability, and advertising.9-13 The relative importance of
each of these factors in influencing food choice can vary be-tween individuals. However, taste has been found to be one of the strongest general influences,9,14 and research has
suggested that taste perception plays a role in the reinforcing value of food.15
In the studies of factors related to food choice or con-sumption, food preferences, and the broad concept offlavor, a combination of taste, olfaction, and somatosensation were evaluated,16,17and response to any of the specific basic tastes, namely salt, sweet, sour, and bitter, was generally not measured. Work has been done, primarily in small, select study populations, investigating the relationship of food preferences and consumption with perception of 6-n-propylthiouracil (PROP), a bitter thiourea compound,18-27and
To take the Continuing Professional Education quiz for this article, log in to
www.eatright.org, click the“myAcademy” link under your name at the top of the homepage, select“Journal Quiz” from the menu on your myAcademy page, click“Journal Article Quiz” on the next page, and then click the “Additional Journal CPE Articles” button to view a list of available quizzes, from which you may select the quiz for this article.
with the TAS2R38 taste receptor gene, which plays a role in PROP taster status.27-33 Studies have also evaluated the relationship of adiposity with PROP phenotype or genotype with inconsistent results.23,25,30,34-36
Because taste has been implicated as an important in flu-ence on dietary choices,9,14 and dietary patterns have been
found to be related to BMI and body fat distribution,4-8it is
possible that taste is associated with changes in adiposity over time. The purpose of the present study was to evaluate the association between perceived intensity of the basic tastes of salt, sweet, sour, and bitter presented at supra-threshold concentrations and longitudinal change in adiposity-related health measures. Patterns of taste in-tensities were identified and the relationship between these patterns and changes in the health measures was assessed. In addition, differences in hedonic ratings for various food items across the taste-intensity patterns were evaluated.
METHODS
Study Population
The study population was comprised of participants in the Beaver Dam Offspring Study, a longitudinal cohort study of the adult children of participants in the population-based Epidemiology of Hearing Loss Study (1993 to present).37-39 The baseline examination took place from 2005 through 2008 and there were 3,285 participants (ages 21 to 84 years, predominately non-Hispanic white).40 Of these, 2,374
par-ticipants completed the taste test.41 Taste testing was
per-formed in the baseline examination in response to a request from the National Institute on Deafness and Other Commu-nication Disorders to develop and test methods for assessing taste function in observational investigations.
The 5-year follow-up examination was conducted in 2010 through 2013. There were 1,918 participants with baseline taste-intensity measures and follow-up health information. Approval for this research was obtained from the Health Sciences Institutional Review Board of the University of Wisconsin and informed consent was obtained from all par-ticipants before each examination. Standardized protocols were followed by trained and certified examiners at each study phase.
Measurements
Taste Intensity.
Filter-paper disks, 3 cm in diameter, impregnated with suprathreshold concentrations of 1.0 mol/L sodium chloride (salt), 1.8 mol/L sucrose (sweet), 0.1 mol/L citric acid (sour), and 0.001 mol/L quinine (bitter), along with disks containing 1.2 to 1.6 mg PROP were used for the whole-mouth taste testing during the baseline examination. An outside laboratory provided the disks (L. M. Bartoshuk, Uni-versity of Florida). To minimize context effects, the tastes were presented in the standard order of salt, sweet, sour, bitter, and PROP. Each participant was asked to place each disk in his or her mouth and to move the disk around to moisten it with saliva. After approximately 10 seconds, the participant removed the taste disk and identified the tastant and estimated the intensity of the taste. Water was sipped between each tastant.A general labeled magnitude scale was used for rating the perceived taste intensity.42 The general labeled magnitude scale was anchored at one end with 0 labeled as “No
sensation” and at the other end with 100 labeled as “Strong-est imaginable sensation of any kind.” Training was con-ducted in the use of the scale and only those participants who successfully completed the training by rating a standard set of sensations in the proper order took part in the taste testing. Additional details of the taste testing have been published.43
Health Measures.
A number of health-related measures were obtained at baseline and at follow-up. Height and weight were measured using a Detecto 758C digital scale and height bar with the participants wearing clothing with pockets emptied and no shoes. BMI was calculated as weight in kilograms/(height in meters)2. Waist circumference, at the umbilicus with the participant standing, was obtained using a tape measure (Gullick II, Country Technology, Inc) with a tensioning device ensuring constant tension across partici-pants. Three sets of seated systolic and diastolic blood pres-sures were obtained with an automated blood pressure machine (Dinamap, GE Healthcare) after the participant had been sitting for 5 minutes; the third measurement was used in analyses. Blood samples were drawn and measurements of glycosylated hemoglobin A1c (HbA1c) using affinity chro-matography (Isolab) and serum total and high-density lipoprotein (HDL) cholesterol using reflectance spectropho-tometry were performed at the Collaborative Studies Clinical Laboratory, Fairview-University Medical Center, Minneapolis, MN. Non-HDL cholesterol was calculated as the difference between the total and the HDL cholesterol levels. The health measures from the baseline examination were subtracted from the follow-up measures to calculate the 5-year change. For a sensitivity analysis, the subset of participants with a history of diabetes, defined as a report of having been diagnosed by a doctor or a measured HbA1c 6.5% were excluded.Hedonic Ratings.
A hedonic general labeled magnitude scale44was used for rating the intensity of liking or disliking 10 food/drink items. The scale had a range of100 (strongest imaginable disliking of any kind) toþ100 (strongest imag-inable liking of any kind). The items rated included mayon-naise, whole milk, black coffee, dark chocolate, salted pretzels, grapefruit juice, sweets, strawberries, sausage, and milk chocolate.43The data were analyzed as continuous.Covariates.
Baseline factors found to be related to taste intensity41 were considered as possible covariates inthe modeling of the association between taste-intensity cluster and change in health. The demographic variables included age, sex, and education (college graduate [16þ years of education] yes or no). The lifestyle factors evaluated were current smoking, any alcohol consumption in the past year, and frequency of dieting (never, rarely, sometimes, often, or always). Olfactory impairment was determined using the San Diego Odor Identification Test45-47and was considered
pre-sent if less than six of the eight odorants were correctly identified. Participants also completed questions asking for the number of servings of vegetables and fruit consumed in a normal week. Response choices ranged from<1 per week to 4þ per day.
For participants aged 45 years and older, DNA was extracted from whole blood and genotyping was performed
using the Illumina IBC chip.48 The PLINK tool set, which
makes haplotype predictions using a standard E-M algo-rithm,49,50was used to construct TAS2R38 haplotypes. There
are three common nonsynonymous single nucleotide poly-morphisms within TAS2R38 (rs713598, rs1726866, and rs10246939) and the common amino acid substitutions at these sites are alanine for proline, valine for alanine, and isoleucine for valine. The analyses of the TAS2R38 haplotype only included participants with the common haplotypes of PAV and AVI.
Statistical Analyses
All analyses were performed using the Statistical Analysis Software (version 9.2, 2008, SAS Institute, Inc). To identify groups of participants with similar patterns of taste in-tensities, the data for participants with complete information (intensity ratings for salt, sweet, sour, and bitter; n¼2,146) were standardized to mean zero and variance one to achieve equal weighting of the four tastes in the clustering. A clus-tering procedure (PROC FASTCLUS) that utilizes Euclidean distances was used to explore solutions ranging from 4 to 20 clusters. PROC TREE produced results of hierarchal clustering
as a tree structure to further elucidate patterns of intensities. A grouping structure offive clusters best matched the data and participants were grouped into the clusters using the Ward’s minimum variance method.
To test for differences in baseline characteristics between clusters, the
c
2 test was used when the characteristic wascategorical and PROC GLM was used when the characteristic was continuous. PROC GLM was also used to estimate least-square mean baseline adiposity-related health measures, 5-year changes in the measures, and hedonic ratings for each cluster after adjustment for significant covariates. The ObsMargins adjustment was applied to allow for estimates proportional to the margins observed in our population. For pairwise comparisons, no adjustment was made for multiple comparisons.
RESULTS AND DISCUSSION
Five distinct clusters of taste-intensity ratings were identified and explained 67.4% of the total variation in the intensity data (Figure). Cluster 1 was characterized as having mean in-tensities slightly above average for salt and sweet and close to average for sour and bitter. Clusters 2 and 3 demonstrated
above-average mean intensities for salt, sour, and bitter; Cluster 2 had an average mean sweet intensity; and Cluster 3 had a very high mean intensity for sweet. Mean intensities for all four tastes were below average in Cluster 4 and were average in Cluster 5.
The clusters were significantly different with respect to age, sex, education, alcohol consumption in the past year, and frequency of dieting (Table 1). Participants in Cluster 4 were younger and more likely to be male and have a college degree than participants in the other clusters. Clusters 2 and 3 had the lowest percentages of males and college graduates. Cluster 4 also had the highest percentage of participants consuming any alcohol in the past year and the lowest percentage dieting often or always. Clusters did not differ significantly with respect to olfaction impairment, or TAS2R38 diplotype. There were also no significant differ-ences between clusters for baseline adiposity-related
health measures after adjustment for age, sex, college de-gree, alcohol consumption, and frequency of dieting. The observed baseline similarities and differences between clusters were consistent withfindings from previous work evaluating the factors related to taste intensity in this cohort.41
Significant differences were observed between clusters for the 5-year change in the adiposity-related health measures (Table 2) even after adjustment for covariates. Participants in the clusters with above-average intensities had greater increases in BMI, waist circumference, and HbA1c than participants in the cluster with average intensities. With multivariable adjustment, the mean increase in BMI was significantly (P¼0.02) greater in Cluster 2 (þ0.9) than in Cluster 5 (þ0.4), and the average increase in waist circum-ference was significantly (P¼0.045) greater in Cluster 1 (þ3.0 cm) than in Cluster 5 (þ2.0 cm). Cluster 5 also
Table 1. Baseline (2005-2008) characteristics and health measures, overall and by cluster: Beaver Dam Offspring Study
Characteristic or health measure Overall
Clustera 1 2 3 4 5 P value n 1,918 326 222 115 651 604 — meanstandard deviation! Age (y) 48.89.7 50.110.3 49.810.1 49.78.9 47.79.4 48.69.5 <0.01 %! Male sex 45.2 45.4 29.7 35.7 54.7 42.4 <0.001 College graduate 36.4 37.6 25.2 23.5 41.2 37.2 <0.001 Smoking—current 15.2 15.0 19.8 13.9 14.3 14.7 0.36 Alcohol—past year 90.5 89.3 88.3 85.2 93.2 90.1 0.02 Olfaction impairment 3.8 6.1 2.7 6.1 2.9 3.5 0.07 Diet frequency 0.02 Never or rarely 61.5 61.3 54.5 52.2 66.7 60.3 Sometimes 22.8 21.8 27.9 27.0 20.3 23.5 Often or always 15.7 16.9 17.6 20.9 13.1 16.2 TAS2R38 diplotypeb 0.72 PAV/PAV (taster) 17.3 14.1 18.0 21.5 17.1 18.4 PAV/AVI (heterozygote) 45.6 44.3 46.1 36.9 46.4 46.9 AVI/AVI (nontaster) 37.1 41.6 35.9 41.5 36.5 34.7
least square mean (standard error)c!
Body mass index 30.0 (0.1) 29.8 (0.3) 29.7 (0.4) 30.0 (0.6) 29.9 (0.2) 30.2 (0.2) 0.80 Waist circumference (cm) 99.1 (0.4) 99.1 (0.8) 99.1 (1.0) 98.6 (1.3) 98.5 (0.6) 99.8 (0.6) 0.59 Systolic blood pressure (mm Hg) 125.4 (0.4) 126.0 (0.9) 126.9 (1.1) 126.7 (1.5) 124.7 (0.6) 125.1 (0.7) 0.36 Diastolic blood pressure (mm Hg) 73.6 (0.2) 73.1 (0.5) 74.5 (0.6) 74.1 (0.9) 73.1 (0.4) 74.0 (0.4) 0.18 Non-HDLdcholesterol (mg/dLe) 153.4 (0.9) 150.7 (2.1) 152.9 (2.6) 154.1 (3.5) 154.7 (1.5) 153.3 (1.5) 0.67
Hemoglobin A1c (%) 5.37 (0.01) 5.39 (0.03) 5.39 (0.04) 5.37 (0.05) 5.34 (0.02) 5.38 (0.02) 0.68
a
Cluster 1: slightly above average for salt and sweet and close to average for sour and bitter; Cluster 2: above average for salt, sour, and bitter, and average for sweet; Cluster 3: above average for salt, sour, and bitter, and very high for sweet; Cluster 4: below average for salt, sweet, sour, and bitter; Cluster 5: average for salt, sweet, sour, and bitter.
bAvailable for 1,027 participants 45 years of age and over at baseline with follow-up information (Cluster 1: n¼185; Cluster 2: n¼117; Cluster 3: n¼65; Cluster 4: n¼334; Cluster 5: n¼326). c
Cluster means adjusted for age, sex, college graduate, any alcohol consumption in past year, and frequency of dieting.
dHDL¼high-density lipoprotein. e
demonstrated a significantly lower increase in HbA1c than Clusters 2, 3, and 4. After participants with a history of diabetes at baseline were removed from the analyses, a significant (P¼0.02) difference between Cluster 3 (þ0.34%) and Cluster 5 (þ0.20%) remained. There were no significant differences in change in blood pressure or non-HDL cholesterol across clusters.
The taste clusters differed significantly with respect to the liking of several food items (Table 3). The differences, adjusted for covariates, were particularly strong for the items representing sweetness (sweets, strawberries, and milk chocolate) and saltiness (pretzels, sausage). Generally, par-ticipants in Cluster 4 (below-average taste intensities) dis-played the lowest mean hedonic score for these items followed by participants in Cluster 5 (average taste in-tensities), Cluster 1 (average or slightly above average taste intensities), and Clusters 2 and 3 (above-average taste in-tensities). There was little difference between clusters for the remaining food items, namely mayonnaise, whole milk, black coffee, dark chocolate, and grapefruit juice, which served as examples of creaminess, fat, bitterness, and sourness. There was also no significant difference between clusters with respect to frequency of vegetable and fruit consumption (
c
2¼4.73; P¼0.79).Previous work investigating the relationship between taste preferences and weight change suggested that higher hedonic ratings for sweet and creaminess were associated
with greater weight gain during an average of 5 years of follow-up.51 These results are compatible with ourfinding
that the clusters with the highest mean hedonic ratings for the sweet and salty food items demonstrated greater mean increases in BMI, waist circumference, and HbA1c. But many factors besides preference are involved in food choice and consumption.9-13 Overall, without consideration of taste
cluster, there were no significant associations between change in the adiposity-related health measures and hedonic ratings except for a relationship between liking whole milk and an increase in systolic blood pressure (Badjusted¼þ0.23 mm Hg per þ10 units on hedonic scale;
P¼0.02) and between liking milk chocolate and a decrease in systolic blood pressure (Badjusted¼0.27 mm Hg per þ10
units on hedonic scale; P¼0.049) (data not shown). There-fore, although some taste clusters differed with respect to liking certain food items and with respect to changes in adiposity-related health measures, evidence of a direct as-sociation between the liking of the food items and the adiposity changes was not found. The lack of evidence is not surprising given the limited number of food items evaluated and the number of factors involved in the path of going from food liking to purchase to consumption and, finally, to adiposity-related change.
This study is likely the first investigation of taste and adiposity to evaluate all four basic tastes and to use cluster analysis to distinguish groups of people based on
Table 2. Five-year change in health measure,aleast square mean change (standard error) overall, and by cluster: Beaver Dam Offspring Studyb
Health measure Overall
Clusterc
1 2 3 4 5
Significant pairwise comparison (P£0.05)
nd 1,918 326 222 115 651 604
least square mean change (standard error)!
Body mass index 0.6 (0.1) 0.7 (0.2) 0.9 (0.2) 0.8 (0.3) 0.5 (0.1) 0.4 (0.1) 2 vs 5 Waist circumference (cm) 2.5 (0.2) 3.0 (0.4) 2.9 (0.5) 2.7 (0.7) 2.7 (0.3) 2.0 (0.3) 1 vs 5 Systolic blood pressure (mm Hg) 1.9 (0.4) 1.4 (1.0) 2.6 (1.2) 2.2 (1.6) 2.6 (0.7) 1.2 (0.7) — Diastolic blood pressure (mm Hg) 1.9 (0.2) 2.0 (0.5) 2.3 (0.6) 1.7 (0.9) 2.3 (0.4) 1.4 (0.4) — Non-HDLe cholesterol (mg/dLf) 7.7 (0.8) 7.9 (2.0) 6.2 (2.5) 2.7 (3.3) 8.1 (1.4) 8.3 (1.5) — Hemoglobin A1c (%) 0.26 (0.01) 0.24 (0.03) 0.29 (0.04) 0.34 (0.05) 0.28 (0.02) 0.20 (0.02) 2 vs 5 3 vs 5 4 vs 5
aChange¼Examination 2 value (2010-2013)eExamination 1 value (2005-2008). b
Cluster means adjusted for age, sex, college graduate, any alcohol consumption in past year, and frequency of dieting.
c
Cluster 1: slightly above average for salt and sweet and close to average for sour and bitter; Cluster 2: above average for salt, sour, and bitter, and average for sweet; Cluster 3: above average for salt, sour, and bitter, and very high for sweet; Cluster 4: below average for salt, sweet, sour, and bitter; Cluster 5: average for salt, sweet, sour, and bitter.
d
Number for study population; number of participants included in the analyses of health measure changes ranged from 1,604 for change in hemoglobin A1c to 1,747 for changes in systolic and diastolic blood pressures.
e
HDL¼high-density lipoprotein.
similarities in the taste-intensity ratings. In a recent report, cluster analysis was used to group food items according to taste intensities of the foods.52 Past studies of taste and
adiposity have primarily concentrated on the association of adiposity with the phenotype or genotype for PROP taster status.34,36 When particular basic tastes were considered with adiposity in past work, the investigations emphasized taste preference rather than intensity and were generally focused on one or two tastes, usually sweetness and bitterness.44,53 Findings from these studies were not
consistent.
Strengths
Strengths of this study included having perceived intensity data for the four basic tastes, along with information on a number of taste and adiposity-related covariates, including frequency of dieting. Data were from partici-pants in the Beaver Dam Offspring Study, a large cohort investigation of the offspring of participants in the population-based Epidemiology of Hearing Loss Study. Adiposity-related health measures were available at two time points, with an approximate 5-year intervening period, which provided the opportunity to evaluate lon-gitudinal change in the measures. Previous investigations of adiposity and taste assessed cross-sectional relation-ships and not longitudinal change in adiposity. Standard-ized protocols for obtaining the measures were followed by trained examiners at each time point so that the observed changes were likely not a consequence of a sys-tematic measurement change. Whole-mouth taste testing provided an approximation of daily taste experience and the taste-testing protocol involved introducing the taste disks in a standard order to minimize context effects. The
general labeled magnitude scale, which has been shown to be valid for across-group comparisons,42 was used for rating intensity.
Limitations
Limiting this study is the fact that no formal dietary intake assessment information was available, the number of items included in the hedonic rating was small, and hedonic ratings were not based on presented foods. Previous work has found a relationship of diet and dietary patterns with adiposity,4-8
and the sensation of taste has been identified as one of the most important factors in dietary choice and intake.9,14But it was not possible in the present study to evaluate the rela-tionship between the observed taste-intensity clusters, food hedonics, and patterns of food consumption. A second consideration is that cluster analysis is dependent on the particular set of data being used and on the investigator’s interpretation. Different taste clusters may be found in other study populations. However, using clusters tofind common patterns of taste perception might be more useful for assessing the dietary and health consequences of taste than evaluating specific tastes. A third concern was that geno-typing was performed only on participants 45 years of age and older and, consequently, the relationship between TAS2R38 diplotype and taste cluster was not assessed for participants younger than 45 years. However, no significant difference in the TAS2R38 diplotypeetaste cluster relation-ship by age group was observed within the 45þ years sub-group. Finally, given the relatively young age of the population and only 5 years of follow-up, there was not adequate power to detect differences between the clusters in the incidence of disease outcomes, such as diabetes and cardiovascular disease.
Table 3. Food hedonicsaoverall and by cluster, least square mean (standard error): Beaver Dam Offspring Study, 2005-2008b
Food item Overall
Clusterc
1 2 3 4 5 P value
nd 1,918 326 222 115 651 604
least square mean (standard error)!
Mayonnaise 16.1 (0.7) 17.0 (1.5) 19.6 (1.9) 18.2 (2.6) 14.3 (1.1) 15.8 (1.1) 0.13 Whole milk 4.6 (0.9) 7.3 (2.1) 0.9 (2.6) 6.6 (3.6) 4.6 (1.5) 4.2 (1.6) 0.41 Black coffee 2.3 (1.2) 2.1 (2.9) 3.2 (3.6) 3.8 (4.9) 1.4 (2.1) 2.9 (2.2) 0.98 Dark chocolate 30.7 (0.9) 34.3 (2.2) 30.9 (2.7) 33.2 (3.8) 28.1 (1.6) 31.4 (1.6) 0.22 Salted pretzels 29.2 (0.5) 30.7 (1.3) 33.5 (1.6) 34.0 (2.2) 26.1 (0.9) 29.4 (1.0) <0.001 Grapefruit juice 6.0 (0.8) 7.0 (2.0) 1.4 (2.4) 7.4 (3.3) 7.5 (1.4) 5.7 (1.4) 0.26 Sweets 45.2 (0.6) 48.8 (1.5) 53.6 (1.8) 53.9 (2.5) 39.7 (1.1) 44.6 (1.1) <0.001 Strawberries 47.0 (0.6) 48.5 (1.5) 53.8 (1.8) 61.6 (2.5) 41.6 (1.1) 46.6 (1.1) <0.001 Sausage 30.3 (0.7) 33.4 (1.6) 33.0 (1.9) 40.2 (2.6) 26.6 (1.1) 29.8 (1.1) <0.001 Milk chocolate 46.0 (0.7) 50.0 (1.6) 53.7 (1.9) 53.5 (2.7) 41.0 (1.1) 45.0 (1.2) <0.001 a
Measured on a hedonic general labeled magnitude scale ranging from100 (strongest imaginable disliking of any kind) to þ100 (strongest imaginable liking of any kind).
b
Adjusted for age, sex, college graduate, any alcohol consumption in past year, and frequency of dieting.
c
Cluster 1: slightly above average for salt and sweet and close to average for sour and bitter; Cluster 2: above average for salt, sour, and bitter, and average for sweet; Cluster 3: above average for salt, sour, and bitter, and very high for sweet; Cluster 4: below average for salt, sweet, sour, and bitter; Cluster 5: average for salt, sweet, sour, and bitter.
d
CONCLUSIONS
Distinct patterns of response to suprathreshold concentrations of the basic tastes were observed in a large population. These clusters were found to be related to 5-year changes in adiposity-related measures, in particular BMI, waist circum-ference, and HbA1c, and to the hedonic ratings of some food items representing sweetness or saltiness. Given the reported associations of obesity and central adiposity with chronic disease,1-3 finding factors related to adiposity change is of
potential use in future health initiatives. Additional follow-up time is needed to evaluate the direct relationship between patterns of taste intensity and disease incidence.
References
1. Burke GL, Bertoni AG, Shea S, et al. The impact of obesity on car-diovascular disease risk factors and subclinical vascular disease. The Multi-Ethnic Study of Atherosclerosis. Arch Intern Med. 2008;168(9): 928-935.
2. Wolin KY, Carson K, Colditz GA. Obesity and cancer. Oncologist. 2010;15(6):556-565.
3. Preis SR, Pencina MJ, Mann DM, D’Agostino RB Sr, Savage PJ, Fox CS. Early-adulthood cardiovascular disease risk factor profiles among individuals with and without diabetes in The Framingham Heart Study. Diabetes Care. 2013;36(6):1590-1596.
4. Maskarinec G, Novotny R, Tasaki K. Dietary patterns are associated with body mass index in multiethnic women. J Nutr. 2000;130(12): 3068-3072.
5. Wirfält E, Hedblad B, Gullberg B, et al. Food patterns and compo-nents of the metabolic syndrome in men and women: A cross-sectional study within the Malmö Diet and Cancer cohort. Am J Epidemiol. 2001;154(12):1150-1159.
6. Liu E, McKeown NM, Newby PK, et al. Cross-sectional association of dietary patterns with insulin-resistant phenotypes among adults without diabetes in the Framingham Offspring Study. Br J Nutr. 2009;102(4):576-583.
7. Newby PK, Muller D, Hallfrisch J, Qiao N, Andres R, Tucker KL. Dietary patterns and changes in body mass index and waist circumference in adults. Am J Clin Nutr. 2003;77(6):1417-1425.
8. Rumawas ME, Meigs JB, Dwyer JT, McKeown NM, Jacques PF. Med-iterranean-style dietary pattern, reduced risk of metabolic syndrome traits, and incidence in the Framingham Offspring Cohort. Am J Clin Nutr. 2009;90(6):1608-1614.
9. Glanz K, Basil M, Maibach E, Goldberg J, Snyder D. Why Americans eat what they do: Taste, nutrition, cost, convenience, and weight control concerns as influences on food consumption. J Am Diet Assoc. 1998;98(10):1118-1126.
10. van den Bree MB, Eaves LJ, Dwyer JT. Genetic and environmental influences on eating patterns of twins aged 50 y. Am J Clin Nutr. 1999;70(4):456-465.
11. Mela DJ. Determinants of food choice: Relationships with obesity and weight control. Obes Res. 2001;9(suppl 4):249S-255S. 12. Turrell G, Kavanagh AM. Socio-economic pathways to diet: Modeling
the association between socio-economic position and food pur-chasing behavior. Public Health Nutr. 2006;9(3):375-383.
13. van den Berg L, Henneman P, Willems van Dijk K, et al. Heritability of dietary food intake patterns. Acta Diabetol. 2013;50(5):721-726. 14. Dressler H, Smith C. Food choice, eating behavior, and food liking
differs between lean/normal and overweight/obese, low-income women. Appetite. 2013;65:145-152.
15. Panek-Scarborough LM, Dewey AM, Temple JL. Sensation and perception of sucrose and fat stimuli predict the reinforcing value of food. Physiol Behav. 2012;105(5):1242-1249.
16. Duffy VB. Variation in oral sensation: Implications for diet and health. Curr Opin Gastroenterol. 2007;23(2):171-177.
17. Prescott J. Chemosensory learning andflavor: Perception, preference and intake. Physiol Behav. 2012;107(4):553-559.
18. Drewnowski A, Henderson SA, Shore AB, Barratt-Fornell A. Nontasters, tasters, and supertasters of 6-n-propylthiouracil (PROP) and hedonic response to sweet. Physiol Behav. 1997;62(3):649-655.
19. Tepper BJ, Nurse RJ. Fat perception is related to PROP taster status. Physiol Behav. 1997;61(6):949-954.
20. Duffy VB, Bartoshuk LM. Food acceptance and genetic variation in taste. J Am Diet Assoc. 2000;100(6):647-655.
21. Kaminski LC, Henderson SA, Drewnowski A. Young women’s food preferences and taste responsiveness to 6-n-propylthiouracil (PROP). Physiol Behav. 2000;68(5):691-697.
22. Yackinous C, Guinard JX. Relation between PROP taster status and fat perception, touch, and olfaction. Physiol Behav. 2001;72(3): 427-437.
23. Yackinous CA, Guinard JX. Relation between PROP (6-n-propylthiouracil) taster status, taste anatomy and dietary intake measures for young men and women. Appetite. 2002;38(3):201-209. 24. Dinehart ME, Hayes JE, Bartoshuk LM, Lanier SL, Duffy VB. Bitter taste markers explain variability in vegetable sweetness, bitterness, and intake. Physiol Behav. 2006;87(2):304-313.
25. Drewnowski A, Henderson SA, Cockroft JE. Genetic sensitivity to 6-n-propylthiouracil has no influence on dietary patterns, body mass indexes, or plasma lipid profiles of women. J Am Diet Assoc. 2007;107(8):1340-1348.
26. Lim J, Urban L, Green BG. Measures of individual differences in taste and creaminess perception. Chem Senses. 2008;33(6):493-501. 27. Duffy VB, Hayes JE, Davidson AC, Kidd JR, Kidd KK, Bartoshuk LM.
Vegetable intake in college-aged adults is explained by oral sensory phenotypes and TAS2R38 genotype. Chemosens Percept. 2010;3(3-4): 137-148.
28. Duffy VB, Davidson AC, Kidd JR, et al. Bitter receptor gene (TAS2R38), 6-n-Propylthiouracil (PROP) bitterness and alcohol intake. Alcohol Clin Exp Res. 2004;28(11):1629-1637.
29. Mennella JA, Pepino MY, Reed DR. Genetic and environmental de-terminants of bitter perception and sweet preferences. Pediatrics. 2005;115(2):e216-e222.
30. Timpson NJ, Christensen M, Lawlor DA, et al. TAS2R38 (phenylthio-carbamide) haplotypes, coronary heart disease traits, and eating behavior in the British Women’s Heart and Health Study. Am J Clin Nutr. 2005;81(5):1005-1011.
31. Sandell MA, Breslin PA. Variability in a taste-receptor gene de-termines whether we taste toxins in food. Curr Biol. 2006;16(18): R792-R794.
32. Sacerdote C, Guarrera S, Smith GD, et al. Lactase persistence and bitter taste response: Instrumental variables and Mendelian randomization in epidemiologic studies of dietary factors and cancer risk. Am J Epidemiol. 2007;166(5):576-581.
33. Hayes JE, Wallace MR, Knopik VS, Herbstman DM, Bartoshuk LM, Duffy VB. Allelic variation in TAS2R bitter receptor genes associ-ates with variation in sensations from and ingestive behaviors toward common bitter beverages in adults. Chem Senses. 2011;36(3):311-319.
34. Tepper BJ, Ullrich NV. Influence of genetic taste sensitivity to 6-n-propylthiouracil (PROP), dietary restraint and disinhibition on body mass index in middle-aged women. Physiol Behav. 2002;75(3):305-312.
35. Duffy VB. Associations between oral sensation, dietary behaviors and risk of cardiovascular disease (CVD). Appetite. 2004;43(1):5-9. 36. Goldstein GL, Daun H, Tepper BJ. Adiposity in middle-aged women is
associated with genetic taste blindness to 6-n-propylthiouracil. Obes Res. 2005;13(6):1017-1023.
37. Cruickshanks KJ, Wiley TL, Tweed TS, et al. Prevalence of hearing loss in older adults in Beaver Dam, Wisconsin. The Epidemiology of Hearing Loss Study. Am J Epidemiol. 1998;148(9):879-886. 38. Cruickshanks KJ, Tweed TS, Wiley TL, et al. The 5-year incidence and
progression of hearing loss: The epidemiology of hearing loss study. Arch Otolaryngol Head Neck Surg. 2003;129(10):1041-1046. 39. Cruickshanks KJ, Nondahl DM, Tweed TS, et al. Education,
occu-pation, noise exposure history and the 10-yr cumulative inci-dence of hearing impairment in older adults. Hear Res. 2010;264(1-2):3-9.
40. Zhan W, Cruickshanks KJ, Klein BEK, et al. Generational differences in the prevalence of hearing impairment in older adults. Am J Epidemiol. 2010;171(2):260-266.
41. Fischer ME, Cruickshanks KJ, Schubert CR, et al. Taste intensity in the Beaver Dam Offspring Study. Laryngoscope. 2013;123(6):1399-1404.
42. Bartoshuk LM, Duffy VB, Green BG, et al. Valid across-group com-parisons with labeled scales: The gLMS versus magnitude matching. Physiol Behav. 2004;82(1):109-114.
43. Cruickshanks KJ, Schubert CR, Snyder DJ, et al. Measuring taste impairment in epidemiologic studies. The Beaver Dam Offspring Study. Ann N Y Acad Sci. 2009;1170:543-552.
44. Duffy VB, Lanier SA, Hutchins HL, Pescatello LS, Johnson MK, Bartoshuk LM. Food preference questionnaire as a screening tool for assessing dietary risk of cardiovascular disease within health risk appraisals. J Am Diet Assoc. 2007;107(2):237-245.
45. Murphy C, Schubert CR, Cruickshanks KJ, Klein BE, Klein R, Nondahl DM. Prevalence of olfactory impairment in older adults. JAMA. 2002;288(18):2307-2312.
46. Raynor LA, Pankow JS, Cruickshanks KJ, et al. Familial aggregation of olfactory impairment and odor identification in older adults. Laryn-goscope. 2010;120(8):1614-1618.
47. Schubert CR, Cruickshanks KJ, Fischer ME, et al. Olfactory impair-ment in an adult population: The Beaver Dam Offspring Study. Chem Senses. 2012;37(4):325-334.
48. Keating BJ, Tischfield S, Murray SS, et al. Concept, design and implementation of a cardiovascular gene-centric 50 K SNP array for large-scale genomic association studies. PLoS ONE. 2008;3(10): e3583.
49. Purcell S, Neale B, Todd-Brown K, et al. PLINK: A tool set for whole-genome association and population-based linkage analyses. Am J Hum Genet. 2007;81(3):559-575.
50. plink.Whole genome association analysis toolset.http://pngu.mgh. harvard.edu/purcell/plink/. Accessed April 25, 2014.
51. Salbe AD, DelParigi A, Pratley RE, Drewnowski A, Tataranni PA. Taste preferences and body weight changes in an obesity-prone popula-tion. Am J Clin Nutr. 2004;79(3):372-378.
52. van Dongen MV, van den Berg MC, Vink N, Kok FJ, de Graaf C. Taste-nutrient relationships in commonly consumed foods. Br J Nutr. 2012;108(1):140-147.
53. Bartoshuk LM, Duffy VB, Hayes JE, Moskowitz HR, Snyder DJ. Psy-chophysics of sweet and fat perception in obesity: Problems, solu-tions and new perspectives. Philos Trans R Soc Lond B Biol Sci. 2006;361(1471):1137-1148.
AUTHOR INFORMATION
M. E. Fischer is an assistant scientist, C. R. Schubert is a researcher, A. Pinto is an associate researcher, and B. E. K. Klein and R. Klein are professors, Department of Ophthalmology and Visual Sciences, University of Wisconsin, Madison. K. J. Cruickshanks is a professor, Departments of Ophthalmology and Visual Sciences and Population Health Sciences, University of Wisconsin, Madison. G.-H. Huang is chairman and a professor, Institute of Statistics, National Chiao Tung University, Hsinchu, Taiwan. J. S. Pankow is a professor, Division of Epidemiology and Community Health, University of Minnesota, Minneapolis.
Address correspondence to: Mary E. Fischer, PhD, Department of Ophthalmology and Visual Sciences, 610 Walnut St, 10th Floor WARF, University of Wisconsin, Madison, WI 53726-2336. E-mail:fischer@episense.wisc.edu
STATEMENT OF POTENTIAL CONFLICT OF INTEREST
No potential conflict of interest was reported by the authors.
FUNDING/SUPPORT
The project described was supported by R01AG021917 from the National Institute on Aging, National Eye Institute, and National Institute on Deafness and Other Communication Disorders, and by unrestricted funds from Research to Prevent Blindness. The content is solely the re-sponsibility of the authors and does not necessarily reflect the official views of the National Institute on Aging or the National Institutes of Health.
ACKNOWLEDGEMENTS
