|Home | About | Journals | Submit | Contact Us | Français|
The goal of this study was to determine whether obese women exhibit altered umami and sweet taste perception compared to normal-weight women. A total of 57 subjects (23 obese and 34 normal weight) participated in a 2-day study separated by 1 week. Half of the women in each group were evaluated using monosodium glutamate (MSG; prototypical umami stimulus) on the first test day and sucrose on the second test day; the order was reversed for the remaining women. We used two-alternative forced-choice staircase procedures to measure taste detection thresholds, forced-choice tracking technique to measure preferences, the general Labeled Magnitude Scale (gLMS) to measure perceived intensity of suprathreshold concentrations, and a triangle test to measure discrimination between 29 mmol/l MSG and 29 mmol/l NaCl. Obese women required higher MSG concentrations to detect a taste and preferred significantly higher MSG concentrations in a soup-like vehicle. However, their perception of MSG at suprathreshold concentrations, their ability to discriminate MSG from salt, and their preference for sucrose were similar to that observed in normal-weight women. Regardless of their body weight category, 28% of the women did not discriminate 29 mmol/l MSG from 29 mmol/l NaCl (nondiscriminators). Surprisingly, we found that, relative to discriminators, nondiscriminators perceived less savoriness when tasting suprathreshold MSG concentrations and less sweetness from suprathreshold sucrose concentrations but had similar MSG and sucrose detection thresholds. Taken together, these data suggest that body weight is related to some components of umami taste and that different mechanisms are involved in the perception of threshold and suprathreshold MSG concentrations.
Obesity rates are increasing worldwide at an alarming rate for reasons that remain obscure. One likely factor involved in the obesity epidemic is the response to the taste of foods, an important determinant of palatability and intake. To this end, many studies have investigated whether obese individuals have heightened sweet preferences (1), hypothesizing that sweet taste is a biological signal for calories and that the heightened preferences for this basic taste lead to or reinforce excessive energy intake. The results of these studies are mixed, and it has recently been argued (1) that this is due, in part, to the use of inappropriate sensory methods.
Although sweet taste serves to signal calorie sources, the savory sensation elicited by the amino acid glutamate (which is called umami (delicious) taste) serves to signal amino acids and protein (2), but this has been questioned (3). To our knowledge, few published human psychophysical studies have investigated the relationships among umami taste perception, hedonics, and body weight (but see ref. 4). Despite the paucity of research, there is some suggestion that such a relationship exists. For example, it is generally agreed that because protein is more satiating than fat or carbohydrate on a calorie-for-calorie basis (5), umami-rich foods may be especially satiating. Supporting this notion, studies in human infants revealed that protein hydrolysate formulas (which are rich in free glutamate) are more satiating than milk-based formulas (which are low in glutamate) (6), and studies in animal models revealed that rats with ad libitum access to monosodium glutamate (MSG) solution from preweaning to adulthood weigh less as adults than do rats without such access (7). Further, people who consume diets rich in umami substances such as the Japanese tend to have lower obesity rates than those who do not consume such diets (e.g., Americans) (8).
Inconsistent with this theory is a recent report indicating that the daily amount of MSG added to foods was positively associated with BMI in Chinese adults (9). The researchers argued that, as observed in animal models, chronic MSG intake may intoxicate arcuate nucleus neurons and disrupt the hypothalamic signaling cascade of leptin action, causing leptin resistance related to overweight/obesity (9). However, the dramatic increases in plasma glutamate levels that are needed to induce hypothalamic damage in animal model studies (51-fold above baseline (10)) are not evident in humans after MSG intake (11). Equally plausible is that if a positive association between BMI and MSG intake exists, it may be due to changes in the sensory perception of MSG as a consequence, rather than as a cause, of elevated body weight.
In light of the unknown relationship between umami taste and obesity, the uncertain relationship between sweet taste and obesity, and the overlap of these two tastes at mechanistic level (e.g., both employ the taste receptor T1R3 in the dimer that is responsible for recognizing (in part) sweeteners (T1R2+T1R3) and umami (T1R1+T1R3) compounds (12)), this study, using validated psychophysical methodology, examined the perception and preferences of these two taste qualities in obese and normal-weight women.
The study population consisted of 57 women between 21 and 40 years of age, not taking any medication with exception of birth control pills, and reported having a BMI <25.0 kg/m2 (normal-weight group; N = 34) or >29.9 kg/m2 (obese group; N = 23). To corroborate self-reports, women were weighed and measured; one woman in the obese group had a BMI of 28.8. Exclusion criteria included diabetes, pregnancy, or lactation and history of chronic rhinitis or food allergies. All procedures were approved by the Office of Regulatory Affairs at the University of Pennsylvania, and each woman gave informed written consent.
For detection threshold testing, sucrose (Sigma-Aldrich, St Louis, MO) and MSG (USB, Cleveland, OH) concentrations ranging from 1 to 5.6 × 10−5 mol/l were prepared in quarter-log dilution steps. To test suprathreshold intensity perception, we used 0.00, 0.09, 0.36, and 1.05 mol/l sucrose solutions and 0.00, 0.02, 0.05, and 0.18 mol/l MSG solutions. All solutions were prepared using deionized water and presented at room temperature (22 °C).
For preference testing, we used 0.09, 0.18, 0.35, 0.70, and 1.05 mol/l sucrose and 0.005, 0.011, 0.021, 0.037, and 0.064 mol/l MSG. The sucrose stimuli were prepared using deionized water, and the MSG solutions were prepared in a vegetable consommé broth with glutamic acid removed (Ajinomoto, Kanagawa, Japan), as the addition of MSG generally decreases palatability to room-temperature water solutions but increases palatability in foods such as soup (13). Soup solutions were warmed to 40 °C and kept in thermal bottles until testing.
Subjects were tested on 2 days separated by at least 5 days. In randomized order, 29 of the women in each group were psychophysically tested with MSG on the first test day and sucrose on the second; the order was reversed for the remaining women. Because previous research revealed that phase of the menstrual cycle (14) and having a family history of alcoholism (15) may affect taste perception, women were queried about the regularity, length, and last date of their menses. The Family Interview for Genetic Studies was administered to make diagnoses of alcoholism according to the Diagnostic and Statistical Manual of Mental Disorders III criteria for family members up to second-degree relatives (16). To standardize testing procedures and the level of hunger/satiation at the time of testing, subjects were asked to abstain from smoking (if smokers, N = 26) and from eating for 12 h before testing. To assess compliance, a capillary blood sample was taken by finger prick (OneTouch; LifeScan, Milpitas, CA) to ensure that they had fasted, and carbon monoxide levels were measured using a Vitalograph-BreathCO monitor (Vitalograph, Lenexa, KS) to ensure that smokers had abstained from smoking. Upon arrival to the center (8–10 am), participants were provided with and consumed a standard breakfast, consisting of a protein bar (180 cal) and 14 ounces of orange juice (190 cal). Thirty minutes after eating, taste detection thresholds were measured and then the other psychophysical tests were performed, as described below.
Sucrose and MSG detection thresholds were assessed (for all but one subject) via a two-alternative, forced-choice staircase procedure (15,17). Testing began at 0.0032 mol/l for sucrose and at 0.0010 mol/l for MSG for all participants. On each trial, subjects were presented with pairs of solutions, one of which was the solution containing the taste and the other of which was deionized water. During each trial, subjects were asked to determine which of the pair contained a taste, after which they rinsed their mouths with distilled water. The concentration of the solution presented increased after a single incorrect response and decreased after two consecutive correct responses. A reversal was considered to have occurred when the concentration sequence changed directions. The procedure was terminated when four reversals meeting the following two criteria had occurred. First, there were no more than two dilution steps between the two successive reversals. Second, the series of reversals cannot form an ascending pattern (i.e., one in which positive and negative reversals are achieved at successively higher concentrations). These additional criteria ensure a more stable measure of the threshold attained (17). The threshold concentration was then calculated as the mean of the log values of the last four reversals.
Before suprathreshold intensity testing, each subject was first familiarized with use of the general Labeled Magnitude Scale (gLMS) (18) with the top of the scale described as the “strongest imaginable” sensation of any kind (1) by rating the intensity of 10 oral and nonoral sensations (e.g., the tingling of a carbonated beverage, the warmth of lukewarm water) (19). The gLMS is a computerized psychophysical tool that requires subjects to rate the perceived intensity along a vertical axis lined with adjectives that are spaced semilogarithmically, based upon experimentally determined intervals to yield ratio-quality data (18). Then, participants were trained to identify each of the five taste qualities by presenting them with standard specimens (20). Salty taste was identified as the predominant taste quality from 150 mmol/l sodium chloride, bitterness as the predominant quality from 0.05 mmol/l quinine hydrochloride, sweetness as the predominant quality from 300 mmol/l sucrose, sourness as the predominant quality from 3 mmol/l citric acid, and umami as the predominant quality from 100 mmol/l MSG.
For suprathreshold testing of sucrose and MSG, solutions were randomly and separately presented in two blocks of four concentrations. After tasting each solution, subjects used the gLMS to rate intensity of the particular tastant. One minute separated the presentation of each sucrose stimulus, 2 min separated the presentation of each MSG stimulus (a longer interval was needed due to aftertaste sensations), and 3 min separated each of the two blocks. The mean of the intensity at each concentration (for each taste quality) during the two-block series provided the estimate of subjects’ taste intensity perception. All women completed the assessment of sucrose taste intensity, and all but three (normal-weight women) completed the assessment of MSG taste intensity.
In order to ensure that group differences in perception of taste were indeed specific to taste sensations and not differences in how the scales were used, each subject also judged intensities of weights using the same gLMS scales used to judge taste intensities (1,20). Subjects were asked to use the gLMS to rate the heaviness of six opaque, sand-filled jars (225, 380, 558, 713, 870, and 999 g).
For every subject, the intensity ratings were averaged across levels for each stimulus set (i.e., six weights and four concentrations for each taste). Significant correlations were found between sucrose sweetness intensity and heaviness of the weights (r = 0.45; P < 0.001) and MSG savoriness and heaviness of the weights (r = 0.33; P = 0.01). Because these variables should be unrelated, this indicated that the gLMS ratings required standardization across subjects. To determine a standardization factor, each subject’s average intensity for heaviness was divided by the grand mean for heaviness across weight levels and subjects. Each individual’s taste intensity ratings for the four concentrations of MSG and the four concentrations of sucrose was then divided by her personal standardization factor to eliminate scale-use bias (20).
Subjects were presented with pairs of solutions that differed in concentration of the stimuli being assessed (sucrose or MSG, depending on the taste condition). They were first presented with a pair of samples chosen from the middle range (0.18 vs. 0.70 mol/l for sucrose or 0.011 vs. 0.037 mol/l for MSG) and asked to taste each without swallowing and then point to which of the pair they liked better. Each subsequent pair was then determined by the subject’s preceding preference choice. The procedure continued until the subject had either chosen a given concentration of sucrose when it was paired with both a higher and lower concentration or had chosen the highest or lowest solution two consecutive times. The entire task was then repeated with stimulus pairs presented in reverse order. Subjects rinsed their mouths with water after each sample, and a 1-min interval separated each pair. The geometric mean of the concentrations chosen during the two trial series provided the estimate of sucrose or MSG preferences.
To determine whether subjects could discriminate the taste quality of MSG (a mixture of umami and salty taste) from NaCl (salty taste only), triangle tests were conducted on both days of testing (21,22). Participants were presented with three cups; two contained the same stimulus (either 29 mmol/l NaCl or 29 mmol/l MSG), and one contained the other stimulus. They were asked to taste all three and choose the one that is different from the other two. On each testing day, they received 12 presentations with two randomized presentations of the six possible combinations. A correct response for 8 of 12 trials was significantly different from chance at the 5% level, and the subject was considered to have discriminated between the two stimuli. Based on data from the two triangle tests, subjects were then grouped by ability to discriminate MSG from NaCl during both tests (discriminating (D) subjects), during only one of the two tests (semidiscriminating (SD) subjects), or during neither of the tests (nondiscriminating (ND) subjects).
Because half of the women were smokers, and smoking has been shown to affect taste perception (15,23,24), preliminary data analyses included both smoking status and body weight groupings as factors. However, because we found no significant effect of smoking status on any of the outcomes, all subsequent analyses focused on body weight grouping only. To determine whether obese women perceived the intensity of suprathreshold concentrations of sucrose and MSG differently than did normal-weight women, we conducted separate two-way ANOVAs with group as the between-subjects factor and the four concentrations of the solutions as the within-subject factor. Taste detection thresholds and perceived intensity of suprathreshold concentration data were positively skewed. Sucrose detection thresholds and perceived taste intensity of sucrose and MSG required square root transformations, and MSG detection thresholds required logarithmic transformation to approximate a normal distribution. When ANOVAs revealed significant effects, post hoc Fisher least significant difference analyses were conducted.
In addition, a χ2 test was conducted to determine whether obese women were less likely than normal-weight women to discriminate 29 mmol/l MSG from 29 mmol/l NaCl. Pearson correlation coefficients were used to assess relationships among taste detection thresholds, taste intensity of suprathreshold concentrations, preferences, and BMI. Data in the tables and figures are presented as means ±s.e.m. or medians with semi-interquartile range (SIQR = [75th – 25th percentile]/2) for skewed data sets. All analyses were performed with Statistica 8.0 (StatSoft, Tulsa, OK), and criterion for statistical significance was P < 0.05.
Normal-weight and obese women did not significantly differ in age, race, years of education, income level, fasting glucose levels, phase of the menstrual cycle at testing, percentage with familial history of alcoholism, or percentage of smokers (Table 1).
Obese women had significantly higher detection thresholds for MSG (i.e., lower MSG taste sensitivity; F(1,54) = 4.90; P = 0.03; Figure 1a), but not for sucrose (P = 0.84; Figure 1b), than did normal-weight women. MSG and sucrose detection thresholds were not correlated (r(56df) = 0.19; P = 0.17).
For both taste stimuli, we found a main effect of concentration (MSG savoriness: F(3,156) = 37.72; P < 0.0001; MSG saltiness: F(3,156) = 17.01; P < 0.0001; sucrose sweetness: F(3,165) = 349.99; P < 0.0001) but not of body weight category, and no interactions between body weight category and concentration. Figure 2a–c depicts the mean intensity ratings of savoriness and saltiness for MSG and of sweetness for sucrose at each concentration level, obtained from normal-weight and obese women. Savoriness and saltiness perception increased with ascending concentrations of MSG, and sweetness perception increased with ascending concentrations of sucrose. We found no significant correlations between taste detection thresholds and the perceived intensity of any of the suprathreshold concentrations tested.
We found no statistical differences between the groups in the intensity of sucrose most preferred (P = 0.43), but obese women tended to prefer soups with higher MSG concentrations than did normal-weight women (F(1,55) = 3.28; P = 0.08). When MSG detection thresholds were included as a covariate in the analysis, this difference became statistically significant (F(1,53) = 5.44; P = 0.02; Table 2).
We found no differences in the proportions of obese and normal-weight women who were categorized as D, SD, or ND subjects (P = 0.70; Table 2). For both body weight groups, we found a significant relationship in how subjects performed in the discrimination test between day 1 and day 2 (χ2(df = 1) = 10.35; P < 0.005). We found that 74% of the women who discriminated MSG from NaCl on day 1 also discriminated it on day 2. Similarly, 70% of the women who failed to discriminate MSG from NaCl on day 1 also failed on day 2.
To further explore individual differences, we examined whether MSG and sucrose perception and preferences varied as a function of ability to discriminate 29 mmol/l MSG from 29 mmol/l NaCl. We found a significant interaction between discriminator groups and concentration in the perception of MSG savoriness (F(6,153) = 3.11; P < 0.01; Figure 3a). Only the D subjects rated the lowest MSG concentration (0.02 mol/l) as tasting more savory than water. They also perceived the highest MSG concentration as tasting significantly more savory than did the other two groups. Further, while D subjects and SD subjects rated both 0.05 mol/l and 0.18 mol/l MSG as more savory than water, ND subjects rated only 0.18 mol/l MSG as more savory than water.
We found a significant main effect of both discriminator group (F(2,51) = 5.67; P < 0.01) and concentration (F(3,153) = 17.08; P < 0.0001) on the saltiness perception of MSG. Figure 3b plots the interaction between group and concentration and shows that, although all women perceived a weak salty taste when tasting the two highest concentration of MSG, ND subjects perceived significantly more saltiness than the other two groups across all concentrations. ND and SD subjects also perceived less sweetness in some sucrose concentrations than did D subjects (F(6,162) = 2.59; P < 0.05; Figure 3c).
Interestingly, MSG and sucrose detection thresholds did not differ significantly among D, SD, and ND subjects (F(2,54) = 0.14; P = 0.87 and F(2,53) = 0.91; P = 0.41; respectively; Table 3), nor did the concentration of sucrose or MSG most preferred (Table 3).
We found that obese women have a significant, twofold increase in MSG detection thresholds and prefer a higher concentration of MSG in a food vehicle (i.e., soup) than do normal-weight women. We observed no such differences for sucrose detection thresholds or sucrose preferences. Importantly, the MSG detection thresholds of normal-weight women and the sucrose detection thresholds in both obese and normal-weight women were within the range previously reported (25,26). Apparently, the MSG detection threshold in obese women is elevated.
These data suggest that a recently reported positive association between MSG intake and obesity (9) may be due, in part, to differences in the sensory perception of and/or preference for MSG as a consequence, rather than as a cause, of elevated body weight. However, caution is needed in the interpretation of these results as causality cannot be inferred from these association studies. Because our study was not designed to assess possible mechanisms, the reason(s) for these differences in MSG taste detection thresholds and preferences is unknown, so those proposed below are necessarily speculative.
First, the simplest explanation for the finding that obese women prefer higher concentrations of MSG in food is because they have a higher taste detection threshold and thus are less sensitive to MSG taste. However, this was not supported by suprathreshold data. Perception of MSG taste intensity as well as the ability to discriminate MSG from an isomolar concentration of salt was similar between obese and normal-weight women.
Second, differences in the concentration of MSG preferred in food by obese and lean women could potentially be the consequence of differences in the perception of other, nontaste properties of the soup, such as smell, presuming they somehow feed back on judgments of umami (since umami stimuli were the other chemical variable that was manipulated). Because alteration in the sense of smell is attenuated in obese individuals (27,28), it is possible that an attenuated perception of the flavor of the food, due to reduced perception of aroma components, in obese women was compensated during preference tests by increasing preferred levels of MSG concentrations.
Third, differences in MSG detection thresholds and preferences may be due to differences in dietary habits as well as nutritional status. Regarding dietary habits, low-protein diets are reported to decrease taste sensitivity (higher detection thresholds) by retarding turnover time of gustatory cells in the taste buds and thus reducing taste receptor function in animal models (29). Conversely, in humans, preference for and presumably greater intake of high-protein food was associated with increased MSG taste sensitivity (lower detection thresholds) (30). Regarding nutritional status, protein-deficient individuals or individuals with low nutritional status prefer higher concentrations of MSG (31) or casein hydrolysate solutions (which are high in MSG) (13) than do well-nourished controls. Because we did not collect dietary records or micronutrient status of the women in our sample, we do not know whether differences we observed in MSG taste detection thresholds and preferences between obese and normal-weight women reflect differences in protein intake. However, there are reports that obesity is associated with a generally poor micronutrient status or malnutrition (32) and the prevalence of nutritional deficiencies, including significantly increased deficient albumin levels with increasing BMI in the obese (33).
The findings that obese women required higher MSG concentrations than did normal-weight women to detect a taste but yet had normal perception of MSG at suprathreshold concentrations adds to the large body of evidence supporting the independence of taste thresholds and suprathreshold taste perception (1,20,21). These data are also consistent with the notion that there are multiple receptor mechanisms for umami stimuli (34–36). Several G-protein-coupled receptors, including truncated type 1 and 4 metabotropic glutamate receptors (MGluRs) (37) and the heterodimer T1R1+T1R3 (refs. 12,38), have been identified in taste bud cells and implicated in umami taste perception. Although sweet and umami compounds employ the T1R3 G-protein-coupled receptor in the receptor dimer responsible for taste perception, our data suggest that these compounds may activate different mechanisms at subthreshold levels. In other words, as speculated for bitterness (20), different receptor systems or transduction pathways may be involved in the perception of threshold and suprathreshold MSG concentrations (36). Supporting this notion is the finding that T1R3 knockout mice, despite reduced perception of suprathreshold MSG and sucrose concentrations, have detection thresholds for MSG and sucrose comparable to their wild-type counterparts (39).
The finding that smoking status did not significantly affect any of the outcomes studied may appear to be at odds with previous findings (23,24), including our own (15), that smokers have higher detection thresholds than nonsmokers. Differences in the procedure used as well as in sample characteristics may explain this discrepancy in the results. For example, in our previous study, women were asked not to eat for several hours before the test and sucrose detection thresholds were assessed after smokers smoked a cigarette whereas in this study, glucose levels were measured to enhance compliance with 12-h fasting, the level of hunger/satiation at the time of testing was controlled by providing a standard breakfast, and detection thresholds were assessed under smoking abstinence.
Regardless of their body weight category, 28% of the women did not discriminate 29 mmol/l MSG from 29 mmol/l NaCl. This is remarkably similar to the findings of Lugaz et al. (21), who found that 27% of 109 subjects showed no significant difference in sensitivity to MSG and NaCl solutions (“putative hypotasters”), and to the findings of Chen et al. (22), who reported that ~21% of 242 subjects were unable to discriminate 29 mmol/l MSG from 29 mmol/l NaCl significantly greater than chance. In our study, ND and SD subjects, did not reliably distinguish 29 mmol/l MSG from 29 mmol/l NaCl, perceived less sweetness from sucrose and less savoriness from MSG than did D subjects, who could reliably distinguish the compounds.
Although the inference of molecular mechanisms of taste perception from psychophysical data has major limitations, these data are consistent with the hypothesis that differences in suprathreshold taste perception among D, SD, and ND subjects are related to differences in T1R3, the receptor subunit common to detecting sweeteners and umami compounds (35). Surprisingly, we found that ND subjects have MSG detection threshold levels similar to those of D subjects. Because the NaCl detection threshold is around 6 mmol/l (ref. 26) and the MSG detection threshold of ND subjects was around 2 mol/l, low levels of MSG detection threshold in these participants are unlikely to be explained by perception of salt. Although we cannot rule out the possibility that MSG detection thresholds of ND subjects were due to a higher sensitivity to detect sodium because we did not measure NaCl detection thresholds, we interpret these data as further evidence for the idea that thresholds and suprathreshold intensity measures are mediated by different mechanisms. We hypothesize that MSG detection thresholds are determined by receptors of the MGluR family (36), whereas suprathreshold (intensity and preferences) measures are mediated by T1R receptors. Further research is required to test these speculative hypotheses.
Taken together, these findings suggest that body weight is related to some components of umami taste perception and that different mechanisms are involved in the perception of threshold and suprathreshold MSG concentrations. New insights into the contributions and consequences of taste in the etiology of obesity may suggest strategies to overcome diet-induced diseases.
This project was funded in part by an investigator-initiated research grant from Ajinomoto and a grant from the Pennsylvania Department of Health. The Pennsylvania Department of Health specifically disclaims responsibility for any analyses, interpretations, or conclusions. M.Y.P. is currently a fellow from an NIDAT32 DA07313 grant at the School of Medicine, Washington University, St Louis, MO, USA.
The authors declared no conflict of interest.