The values (mean ± SD) of anthropometric parameters of participants with normal weight and obesity are shown in Table 1.Table 1

Anthropometric parameters of participants with normal weight (NW) or with obesity (OB).

The mean values ± SEM of BMI in NW and OB according to PROP taster status are shown in Fig. 1. An ANOVA revealed a significant two-way interaction of the Taster group × NW/OB status for the BMI values (F _[2,104] = 4.556; P = 0.013). Post hoc comparisons showed that the BMI of ST OB was significantly higher than that of MT and NT OB (P = 0.013 and P = 0.023; Newman-Keuls test). We also considered whether the waist/hip ratio changed according to PROP status within NW or OB, and we did not find any significant difference. In addition, we also investigated whether there was a higher incidence of type 2 diabetes in the OB NT cohort. We found that although the difference was not statistically significant (chi squared = 1.5; P = 0.47), 25% of ST, 44% of MT and 46% of NT of OB subjects were affected by type 2 diabetes.Figure 1

Mean values ± SEM of body mass index (BMI) of supertaster (N = 9), medium taster (N = 28) and non-taster (N = 23) participants with normal weight (NW) and supertaster (N = 12), medium taster (N = 25) and non-taster (N = 13) participants with obesity (OB). A two-way analysis of variance (ANOVA) was used to compare BMIs across taster groups in participants with normal weight or with obesity (P = 0.013). The different letters indicate significant differences (P < 0.023; Newman-Keuls test). ST, supertaster participants; MT, medium-taster participants; and NT, non-taster participants.

The scores of the 3-Factor Eating Questionnaire by Stunkard & Messick³³ (TFEQ) for the assessment of the general eating attitudes of NW and OB according to their PROP taster status are shown in Table 2. The scores for disinhibition were higher for OB than NW for all three taster groups (P < 0.049; Newman-Keuls test subsequent two-way ANOVA), whereas no significant differences in the disinhibition scores were found according to the PROP taster status. No changes were found in the restraint or hunger scores between OB and NW or among the various PROP tasters (data not shown).Table 2

Scores for dietary restraint, disinhibition, and perceived hunger of supertasters (ST), medium tasters (MT) and non-tasters (NT) for both participants with normal weight (NW) or with obesity (OB).

Table 3 shows the erythrocyte fatty acid profile expressed as mol%, in NW and OB. We found a 13% increase in SAFA with a major contribution of PA at the expense of polyunsaturated fatty acids (PUFA) and mainly linoleic acid LA in OB compared with NW. In fact, the ratio of PA/LA was 28% higher in OB than in NW. This ratio has been used as a biomarker of DNL³⁰. The only SAFA measured that did not differ between NW and OB was pentadecanoic acid, which was previously suggested to be a biomarker of dietary dairy products³⁴. To confirm these data, we also measured other typical biomarkers of dairy product intake, such as c9, t11 conjugated linoleic acid and vaccenic acid³⁵, and typical biomarkers of omega-3 intake such as EPA and DHA levels³⁶, but we did not find any significant differences between OB and NW (data not shown).Table 3

Erythrocyte fatty acid profile and plasma beta-carotene level of participants with normal weight (NW) or with obesity (OB).

The plasma levels of the endocannabinoids AEA and 2-AG, and the AEA congener PEA (mean values ± SEM) were measured in NW and OB according to the PROP taster status and are shown in Fig. 2. An ANOVA revealed a significant two-way interaction of the Taster group × NW/OB status for the AEA or 2-AG values (F _[2,104] = 12.248; P < 0.0000 or F _[2,104] = 4.914; P = 0.009) (Fig. 2a and b). Post hoc comparisons showed that AEA plasma levels of ST OB were significantly lower than those of MT and NT OB (P = 0.003 and P = 0.0002; Newman-Keuls test), whereas those of ST NW were significantly higher than those of NT NW (P = 0.019; Newman-Keuls test) (Fig. 2a). The AEA plasma levels of MT and NT NW were significantly lower than those of the corresponding OB (P = 0.0043 and P = 0.00012; Newman-Keuls test) (Fig. 2a). The 2-AG plasma levels of ST and MT OB were significantly lower than those of NT OB (P = 0.044; Newman-Keuls test) who had levels higher than those of NT NW (P = 0.0024; Newman-Keuls test) (Fig. 2b). The associations between PROP taster group and AEA and 2AG on NW and OB remained after controlling by BMI, or after controlling for waist to hip ratio as tested by two-way ANCOVA analysis (data not shown). A pairwise comparison also showed that the PEA plasma levels of NT OB were significantly higher than those of MT and ST OB (P = 0.0207 and P = 0.0192; Newman-Keuls test) and of NT NW (P = 0.0179; Newman-Keuls test) (Fig. 2c). In addition, no changes were found in plasma oleoylethanolamide (OEA) levels (data not shown).Figure 2

Mean values ± SEM of endocannabinoids (AEA and 2-AG) and the palmitoylethanolamide (PEA) congener plasma levels measured in supertaster (N = 9), medium taster (N = 28) and non-taster (N = 23) participants with normal weight (NW) and supertaster (N = 12), medium taster (N = 25) and non-taster (N = 13) participants with obesity (OB). (a) Anandamide (AEA) values. (b) 2-arachidonoylglycerol (2-AG) values. (c) PEA values. A two-way analysis of variance (ANOVA) was used to compare the differences across taster groups in participants with normal weight or with obesity (AEA and 2-AG: P < 0.009). The different letters indicate significant differences (AEA: P < 0.019; 2-AG: P = 0.044; and PEA: P = 0.020; Newman-Keuls test). *A significant difference between participants with normal weight or with obesity (P < 0.017; Newman-Keuls test). ST, supertaster participants; MT, medium taster participants; and NT, non-taster participants.

The levels of erythrocytes SAFA (mean values ± SEM), PA/LA ratio, plasma FFA and retinol measured in NW and OB according to the PROP taster status are shown in Fig. 3. A significant two-way interaction of Taster group × NW/OB status on SAFA values or PA/LA or FFA values was found (F _[2,104] = 36.901, P = 0.0000; F _[2,104] = 7.8328, P = 0.00068 and F _[2,104] = 3.8502, P = 0.0243; two-way ANOVA). Post hoc comparisons showed that the SAFA values (Fig. 3a) and the PA/LA ratio (Fig. 3b) of ST NW were significantly lower than those of MT NW (P = 0.00011 and P = 0.0015, respectively; Newman-Keuls test), which also had levels lower than those of NT NW (P = 0.0069 and P = 0.0214, respectively; Newman-Keuls test). The SAFA values and the PA/LA ratio of ST and MT NW were significantly lower than those of the corresponding OB (SAFA values: P = 0.00012 and P = 0.0161, respectively; PL/LA: P = 0.00012 and P = 0.0032, respectively; Newman-Keuls test). Post hoc comparisons also showed that the FFA plasma levels of NT OB were significantly higher than those of NT NW (P = 0.0082; Newman-Keuls test) (Fig. 3c). The plasma retinol level of ST OB was lower than those of NT OB (P = 0.0333; Newman-Keuls test), which had levels higher than those of NT NW (P = 0.0265; Newman-Keuls test) (Fig. 3d).Figure 3

Mean values ± SEM of erythrocyte saturated fatty acid (SAFA) levels and the palmitic/linoleic acids ratio (PA/LA), plasma levels of free fatty acid (FFA) and retinol measured in supertaster (N = 9), medium taster (N = 28) and non-taster (N = 23) participants with normal weight (NW) and supertaster (N = 12), medium tasters (N = 25) and non-taster (N = 13) participants with obesity (OB). (a) Erythrocyte SAFA values. (b) Erythrocyte palmitic/linoleic acids ratio (PA/LA). (c) Plasma free fatty acid (FFA) values. (d) Plasma retinol values. A two-way analysis of variance (ANOVA) was used to compare the differences across taster groups in participants with normal weight or with obesity (SAFA, PA/LA: P < 0.00068 and FFA: P = 0.0243). The different letters indicate significant differences (SAFA: P < 0.0069; PA/LA: P < 0.0214; retinol: P = 0.0333; Newman-Keuls test). *A significant difference between participants with normal weight or with obesity (P < 0.0265; Newman-Keuls test). ST, supertaster participants; MT, medium taster participants; and NT, non-taster participants.

ST NW showed plasma levels of beta-carotene that were lower than those of MT NW (P = 0.0409; Newman-Keuls test subsequent to a two-way ANOVA), which had levels higher than those of MT OB (P = 0.0044; Newman-Keuls test subsequent to a two-way ANOVA) (Fig. 4).Figure 4

Mean values ± SEM of the plasma levels of beta-carotene measured in supertaster (N = 9), medium taster (N = 28) and non-taster (N = 23) participants with normal weight (NW) and supertaster (N = 12), medium taster (N = 25) and non-taster (N = 13) participants with obesity (OB). ST, supertaster participants; MT, medium taster participants; and NT, non-taster participants. The different letters indicate significant differences (P = 0.0409; Newman-Keuls test subsequent to the two-way ANOVA). *A significant difference between participants with normal weight or with obesity (P = 0.0044; Newman-Keuls test subsequent to the two-way ANOVA).

The results of the data analysis are referred to both genders, however, when limited to women, did not change significantly (data not shown).

Table 4 shows the effect size and power calculated for each dependent variable after conducting two-way ANOVA with PROP taste and NW /OB status as independent variables.Table 4

Effect size and Power calculated for each dependent variable (DV) after conducting two-way ANOVA with PROP taste and NW /OB status as independent variables.

The effect size was very large for PROP taste and NW /OB status and AEA and SAFA, medium in the case of 2AG, PA/LA and FFA and small in case of BMI, PEA, retinol and beta-carotene in according to Cohen’s guidelines (1988). However, the same analyses in the case of PEA, retinol and beta-carotene resulted underpowered, at the limit of significance in case of BMI and FFA, and had sufficient power in the case of AEA, 2AG, SAFA and PA/LA.

One hundred and ten Caucasian volunteers (34M and 76F) were recruited in the area of Cagliari, Italy. They were divided into two groups based on their BMI, resulting in random classification by age and gender: NW (BMI from 18–25 kg/m²) (N = 60; 19M, 41F) and OB (BMI from 30 to 50 kg/m²) (N = 50; 15M, 35F).

The exclusion criteria included major diseases (e.g., diabetes and kidney disease), pregnancy or lactation, food allergies, and the use of medications that interfere with taste or smell (e.g., steroids, antihistamines, and certain antidepressants). All participants were verbally informed regarding the procedure and the aim of the study. They reviewed and signed an informed consent form. The present study was conducted according to the guidelines of the Declaration of Helsinki of 1975 (revised in 1983), and all procedures involving human participants were approved by the ethical committee of the Brotzu Institution. The study has been registered at ClinicalTrials.gov with Identifier: NCT02729584. Written informed consent was obtained from all participants.

All participants were requested to abstain from eating, drinking and using oral care products or chewing gums for at least 8 h prior to the sample withdrawal and the screening test for PROP taste sensitivity. All participants were required to be in the test room 15 min before the beginning of the trials, and their weight (kg) and height (m) were recorded to calculate the body mass index (BMI) (kg/m²). A 4 ml sample of blood was collected (at 8.00 AM) from each participant. Samples were immediately centrifuged and stored at −80 °C until the analyses were completed as described below. Participants completed the TFEQ³³ for the assessment of general eating attitudes. The questionnaire estimates three aspects of cognitive control of eating behaviour: dietary restraint, disinhibition and perceived hunger.

The participants were screened and classified according to PROP taster status through the impregnated paper screening test, which was previously tested for validity and reliability^58–60, and strongly correlate with the taste receptor cell depolarization⁶¹. The intensity ratings for PROP or NaCl were collected using the Labelled Magnitude Scale (LMS), a 100-mm scale anchored with the phrases “barely detectable” to “strongest imaginable”⁶². After tasting each solution, the participants placed a mark on the scale corresponding to his/her perception of the stimulus. Participants were categorized as non-tasters if they rated the PROP disk lower than 13 mm on the LMS; they were categorized as supertasters if they rated the PROP disk higher than 67 mm on the LMS. All other participants were classified as medium tasters⁵⁸. Based on their taster group assignments, 38.33% of the NW were NT (N = 23; 8M, 15F), 46.67% were MT (N = 28; 10M, 18F), and 15% were ST (N = 9; 1M, 8F), whereas 26% of the OB were NT (N = 13; 7M, 6F), 50% were MT (N = 25; 7M, 18F), and 24% were ST (N = 12; 2M, 10F).

The acetonitrile (CH₃CN), methanol (CH₃OH), chloroform (CHCl₃), n-hexane (C₆H₁₄), ethanol (C₂H₅OH), and acetic acid (CH₃COOH) were HPLC grade and purchased from Sigma Chemicals Co. (St. Louis, MO, USA). All standards of fatty acid standards were purchased from the same company.

Ascorbic acid, potassium hydroxide (KOH), and hydrochloric acid (HCl) were purchased from Carlo Erba, Milano, Italy. Deferoxamine mesylate was purchased from CIBA-Geigy (Basel, Switzerland). Internal deuterated standards for the AEA, 2-AG, palmitoylethanolamide (PEA) and oleoylethanolamide (OEA) quantification by isotope dilution ([2H]8AEA, [2H]52AG, [2H]4 PEA, and [2H]4 OEA) were purchased from Cayman Chemicals (MI, USA).

The total lipids were extracted by the method of Folch⁶³. Briefly, samples of human plasma (1 ml) or erythrocytes were each homogenized into a 2:1 chloroform-methanol solution containing 2 μg of vitamin E and deuterated AEA (200 ng), 2-AG (300 ng), OEA (200 ng), and PEA (100 ng).

An aliquot of the plasma lipid extract was used for HPLC separation to determine the total free fatty acids as described previously⁶⁴. The total lipid quantification was performed by the method of Chiang⁶⁵.

An aliquot of the erythrocyte lipid fraction of each sample was mildly saponified, and the fatty acids were analysed by HPLC (an Agilent 1100 HPLC system with a diode array detector, Palo Alto, Calif., USA) as previously described⁶⁶. Spectra (195–315 nm) of the eluate were obtained every 1.28 s and were electronically stored. These spectra were acquired to confirm the identified HPLC peaks⁶⁷.

Because SAFA are transparent to UV, after derivatization, they were measured as fatty acid methyl esters using a gas chromatograph (Agilent, Model 6890, Palo Alto, CA) as described previously⁴⁹.

The results for other lipid parameters prompted us to analyse beta-carotene as a biomarker of vegetable and fruit intake. Therefore, due to the limited availability of the remaining samples, we were able to perform the analysis in only a subgroup of NW (N = 47) and OB (N = 38). An aliquot of the lipid fraction was analysed by HPLC with an array detector as previously described⁶⁸.

Aliquots of the organic phase (chloroform) containing extracted lipids of human plasma samples were evaporated to dryness under a vacuum and reconstituted with 0.3 ml of 100% methanol.

Quantification of AEA, 2-AG, PEA and OEA was conducted by liquid chromatography-atmospheric pressure chemical ionization-mass spectrometry (LC-APCI-MS) and using selected ion monitoring (SIM) at M + 1 values for the four compounds and their deuterated homologues as previously described⁶⁹.

The data are expressed as the mean ± SD or SEM as specified in the legends. The differences between two groups were assessed using an unpaired, two-tailed Student’s t-test. The data sets involving more than two groups were assessed by ANOVA, and specifically, a two-way ANOVA was used to evaluate the differences for all parameters related to the PROP status in NW and OB. To evaluate whether the possible differences were due to a different gender distribution, we ran the same data analysis while considering only women. A two-way ANCOVA confirmed the association between PROP status and endocannabinoids after controlling for BMI and waist/hip ratio. Post hoc comparisons were conducted with Newman-Keuls test. Data with different superscripted letters were significantly different according to the post hoc ANOVA analysis. Chi-squared was used to evaluate whether the frequency of participants with T2D, was different among PROP taster groups. Post hoc statistical power analysis was conducted by using SPSS software (SPSS Inc. Headquarters, Chicago, IL, USA) to evaluate the effect size and power for each dependent variable after conducting two-way ANOVA where PROP taste and NW /OB status were independent variables. The eta squared, as parameter showing the effect size, and power are reported for each dependent variable in Table 4.

The statistical analyses were conducted using STATISTICA for WINDOWS (version 6.0; StatSoft Inc., Tulsa, OK, USA). P values < 0.05 were considered significant.