Fibroblast Growth Factor 21 Levels Are Associated With Perception and Neural Responses to Sweetness in Type 2 Diabetes Mellitus
Article information
, Ying Zhang1,2,*, Dian Zeng3,*, Dan Liu1, Rui Han1, Yuwei Lu1,2, Di Cheng1,2, Qinyi Wang1,2, Silin Liu1,2, Liang Wu1, Qian Wu1,2, Shujie Yu1,2, Anran Chen1,2, Jingyi Guo4, Wenli Ge1, Jiacheng Ni1, Jingyi Yang1, Xiaomeng Wu1, Lifei Ma1, Weiping Jia1, Qichen Fang1, Yuehua Li,5, Huating Li,1
Abstract
Background
The relationship between fibroblast growth factor 21 (FGF21) and sweet taste perception and preference in type 2 diabetes mellitus (T2DM) remains unclear. This study aims to investigate this relationship and examine the neural responses of T2DM patients to high-calorie sweet (HCS) food pictures, further exploring its correlation with FGF21 levels.
Methods
We assessed sweet taste perception and preference in 40 T2DM patients and 41 controls using classical scales. Subsequently, the neural responses of 11 T2DM patients and 11 controls to HCS pictures were examined using functional magnetic resonance imaging. FGF21 levels were measured using chemiluminescent immunoassay, and the correlations with taste perception and neural responses were analyzed.
Results
Increased FGF21 levels were associated with decreased sweet perception and increased sweet taste preference in T2DM patients. Compared to control, T2DM patients exhibited greater neural activations in the orbitofrontal cortex, anterior cingulate cortex (ACC), thalamus, and hippocampus (HCS vs. non-food) as well as the putamen (HCS vs. low-calorie food). Notable differences were observed in the parahippocampal gyrus, insula, ACC, and hippocampus in T2DM patients (HCS vs. high-calorie non-sweet). Additionally, FGF21 accounted for 30.39% and 32.4% of the associations between T2DM and ACC, and parahippocampal gyrus, respectively.
Conclusion
FGF21 levels were independently associated with changes in sweet taste perception and preference in T2DM patients and were significantly associated with activation in reward-related brain regions. This study reveals the potential role of FGF21 in regulating responses to sweet foods in T2DM and provides insight to develop new therapeutic strategies for diabetes.
Highlights
• T2D patients had reduced sweet taste perception but increased sweet preference.
• FGF21 levels were closely related to sweet taste perception and preference in T2D.
• T2D patients showed enhanced reward-related brain activation in response to sweet.
• FGF21 partially mediated the link between T2D and neural responses to sweetness
INTRODUCTION
Fibroblast growth factor 21 (FGF21) is an atypical member of the FGF family that functions as an endocrine hormone which regulates energy balance and glucose and lipid homeostasis through a heterodimeric receptor complex comprising FGF receptor 1 (FGFR1) and β-klotho (KLB) [1-3]. It is noteworthy that FGF21 can cross the blood-brain barrier, indicating the brain as a significant target organ for this factor [4]. This hormone plays a pivotal role in the regulation of macronutrient intake. Notably, FGF21 has been shown to suppress sweet and alcohol preferences in mice and sweet preference in monkeys by reducing dopamine levels in the nucleus accumbens, a key neurotransmitter in reward pathways in the brain [5]. Moreover, it has been demonstrated that a high carbohydrate diet could increase FGF21 levels in the liver of mice by activating the transcription factors carbohydrate-responsive element-binding protein (ChREBP) and peroxisome proliferator-activated receptor α (PPARα). FGF21, in turn, exerts a negative feedback effect on the central nervous system, curbing the intake of sweet foods [6,7]. Further research has delineated that FGF21 could signal to glutamatergic neurons in the ventromedial hypothalamus, thereby inhibiting both sugar consumption and the preference for sweet tastes in mice [8].
Beyond its role in regulating the consumption of sweet foods in animals, FGF21 has been found to correlate with human sweet food preferences, potentially modulating human nutrient intake. An independent human genome‐wide association study found that the variations of FGF21 (rs838133 and rs838145) were related to the increase of relative carbohydrate and the decrease of protein and fat intake [9-11]. While the regulatory function of FGF21 in nutrient intake and energy balance is well-documented, the relationship between FGF21 levels in individuals with type 2 diabetes mellitus (T2DM) and their preference for sweet foods is not yet fully understood.
Functional magnetic resonance imaging (fMRI) has revolutionized our ability to examine neural responses to food stimuli, offering objective insights into the complex interplay between brain activity and food preferences [12,13]. Previous neuroimaging studies have identified key brain regions involved in food reward processing, including thalamus [14], basal ganglia (putamen, caudate, and nucleus accumbens) [15,16], orbitofrontal cortex (OFC) [17-19], anterior cingulate cortex (ACC) [19], insula [20,21], and hippocampus (HPC) [22-24]. In particular, sensory stimuli in foods that were considered enjoyable or had a high energy content trigger increased brain activity in these areas [19,20,25]. While these studies have primarily focused on high-calorie (HC) food, low-calorie (LC) food and non-food stimuli in healthy or obese populations, there is a critical gap in our understanding of the neural responses to sweet food stimuli in T2DM patients and how these responses relate to FGF21 levels.
This study aims to assess the correlation between FGF21 levels and the perception and preference for sweetness in patients with T2DM. Meanwhile, we utilize fMRI to elucidate the differences in neural activation patterns between T2DM patients and control when exposed to pictures of sweet foods, investigating the underlying associations between these neural responses and FGF21 levels. By integrating endocrine, behavioral, and neuroimaging approaches, this study seeks to provide novel insights into the complex interplay between FGF21, sweet food preference, and brain activity in T2DM.
METHODS
Study design and participant details
We enrolled 41 patients with T2DM and abdominal obesity at the Shanghai Sixth People’s Hospital Affiliated to Shanghai Jiao Tong University School of Medicine. Patients aged 35 to 65 years, sedentary, and on a stable anti-diabetic medication were included for eligibility assessment. Exclusion criteria included insulin usage, pregnancy, severe proliferative diabetic retinopathy, macroalbuminuria and renal dysfunction, history of cardiovascular events, cerebrovascular diseases, and severe osteoporosis. Additionally, we recruited 41 age- and sex-matched controls without T2DM and obesity. After recruitment, one participant with T2DM was excluded from the study due to an insufficient blood sample for the FGF21 measurement. This study was registered at the Chinese Clinical Trial Registry (ChiCTR2100046148) and received ethical approval from the Ethics Committee of the Shanghai Sixth People’s Hospital (2019-099) [26]. Written informed consent was obtained from all participants.
Sweet taste perception and preference trial
Participants rated sweet taste perception on the generalized labelled magnitude scale and preference on the labelled hedonic scale (LHS) [27-29]. LHS is a symmetric, bipolar scale with an arbitrary range from ‘most disliked sensation imaginable’ (–100) to ‘most liked sensation imaginable’ (+100), anchored in the center (0) by neutral [29]. This method has been shown to be useful in assessing sweet taste perception and preference [30]. A 50% glucose solution (China Otsuka Pharmaceutical Co. Ltd., Shanghai, China) was mixed with mineral water to obtain glucose concentrations of 0, 0.1, 0.18, 0.32, 0.56, and 1 mol/L. Each concentration was presented three times in a random order for a total of 18 presentations. Participants were asked to taste 5 mL of each without swallowing, then rate their taste perception and preference. Between each taste, participants rinsed their mouth with water and waited for 30 seconds. The area under the curve (AUC) was then calculated for each individual. The AUC serves as a measure of overall perceptual and preferential capability across the full concentration range tested [31].
Anthropometric and biochemical assessments
After overnight fasting for at least 10 hours, anthropometric test was conducted in the morning. Anthropometric test included height, weight, body mass index (BMI), waist circumference, hip circumference, thigh circumference, systolic blood pressure, diastolic blood pressure, and heart rate. Body fat mass and percentage were measured using bioelectrical impedance analysis (DBA-210 software version 3.5, Donghuayuan Medical, Yanji, China). Blood samples were collected in fasting state for measuring fasting plasma glucose (FPG), insulin, glycosylated hemoglobin (HbA1c), total cholesterol, triglyceride, high-density lipoprotein cholesterol (HDL-C), and low-density lipoprotein cholesterol. HbA1c was measured using high-performance liquid chromatography (VARIANT II, Bio-Rad Laboratories Inc., Hercules, CA, USA). Blood glucose concentrations were measured via glucose oxidase method. Homeostasis model assessment of insulin resistance (HOMA-IR) and islet β-cell function (HOMA-β) were evaluated by the homeostasis model assessment: HOMA-IR=fasting insulin (FINS; mU/L)×FPG (mmol/L)/22.5, and HOMA-β=[FINS (mU/L)×6−3.33]/[FPG (mmol/L)−3.5] [32].
Serum FGF21 levels measurements
Serum FGF21 levels were measured using a chemiluminescent immunoassay (CLIA) (Guangdong Uniten Biotechnology Co. Ltd., Dongguan, China). In brief, a standard or serum sample was added to a cuvette and mixed with the reaction buffer containing salt and magnetic micro-particles pre-coated with antihuman FGF21 antibodies. After a short incubation, the magnetic beads were washed with phosphate buffered saline with tween (PBST) for three times to remove the non-special bindings, and beads were recovered under magnet adsorption. Afterwards, an acridinium ester conjugated secondary anti-human FGF21 antibodies were added to cuvette and mixed. The beads were further washed with PBST as previously described, followed by the adding of H2O2 and NaOH to stimulate the luminescence. The luminescence signals were captured by the detector and measured as absorbance, which are proportional to the concentrations of human FGF21 in the testing samples. The assay range of human FGF21 CLIA kit was 30 to 1,920 pg/mL. The inter-assay variation in FGF21 measurement was 2.90%.
Food pictures fMRI task
The experimental procedure was conducted utilizing a well-established food pictures fMRI task, as previously described by Avery et al. [33]. Eleven control participants and 11 individuals with T2DM underwent the food pictures fMRI task session following 5 to 6 hours of fasting. The visual stimuli consisted of 56 food and 28 non-food pictures, all of which are considered common in everyday life, with consistent viewing angle, resolution, luminance, and balanced energy levels. The 56 food pictures were divided into 28 LC and 28 HC pictures. All HC pictures had an energy density greater than 3.5 kcal/g and all LC pictures had an energy density less than 1 kcal/g [34]. The energy densities were confirmed by a registered dietitian. The HC pictures were further categorized into high-calorie sweet (HCS) and high-calorie non-sweet (HNS) foods. During the scanning process, blocks of HCS, HNS, LC, and non-food stimuli were presented in a random order using an event-related design. Each block presented for 28 seconds and featured seven pictures. Between each block, a central fixation cross appeared for 12 seconds (Fig. 1A). The total duration of each scanning session was 552 seconds.
(A) One of the two runs used in the food pictures functional magnetic resonance imaging task. Within each run, six blocks of visual pictures, two each of non-food (NF), low-calorie (LC), and high-calorie foods (one each of sweet, high-calorie sweet [HCS] and high-calorie non-sweet [HNS]), alternated with a fixation cross. Order of blocks was randomized from run to run with the constraint that a given stimulus category was not followed by the same category. (B, C) Regions with higher activation in type 2 diabetes mellitus (T2DM) patients (n=11) than healthy people (n=11) in the contrast of HCS vs. NF stimuli. (D, F) Regions with higher activation in T2DM patients (n=11) than control (n=11) in the contrast of HCS vs. LC foods. (E, G) Regions with higher activation in T2DM patients (n=11) than control (n=11) in the contrast of HCS vs. HNS foods. All threshold of P<0.05, cluster size ≥405 mm3, Alfa SIM corrected. Yellow or white clusters and red circles show larger responses in participants with diabetes compared with control for HCS vs. NF stimuli. Color bar indicates t-values. HPC, hippocampus; L, left; R, right; OFC, orbitofrontal cortex; ACC, anterior cingulate cortex.
fMRI data acquisition
The MRI was obtained using a 3.0-Tesla Siemens Prisma MRI system (Siemens, Munich, Germany) at the Shanghai Sixth People’s Hospital Affiliated to Shanghai Jiao Tong University School of Medicine. During the fMRI scan, participants lay in the supine position on the scanner bed to view the stimuli back-projected onto a screen through a mirror attached to a 64-channel head coil, and head movements were minimized using foam pillows.
Stimulus presentation and timing of all stimuli were achieved via a windows computer running E-prime 3.0 software (Psychology Software Tools, Sharpsburg, VA, USA). After a 3-plane localizer, a high-resolution three-dimensional (3D) volume was collected using a T1-weighted Magnetization Prepared Rapid Gradient Echo (MPRAGE) sequence (192 contiguous sagittal slices, repetition time [TR]=2,000 ms, echo time [TE]=2.98 ms, flip angle [FA]=9°, 1×1×1 mm voxels, field of view [FOV]=256×256 mm², phase-encoding=anterior-posterior). Thereafter, whole-brain blood oxygen level-dependent (BOLD) fMRI volumes were acquired using an Echo Planar Imaging (EPI) sequence (40 contiguous axial slices, TR=2,000 ms, TE=30 ms, FA=90°, voxel size=3×3×4 mm, FOV=240×240 mm², phase-encoding direction=posterior-anterior).
fMRI data analyses
The data preprocessing was performed using SPM12 (Wellcome Dept. Imaging Neuroscience, London, UK, http://www.fil.ion.ucl.ac.uk/spm), including slice timing, head motion correction (a least squares approach and a 6-parameter spatial transformation), and spatial normalization to the Montreal Neurological Institute (MNI) template (resampling voxel size=3×3×3 mm3). Subsequent data preprocessing included removal of linear trends and spatial smoothing (full width at half maximum=6 mm Gaussian kernel), followed by a band pass filtered by 0.01 to 0.1 Hz. The 3D T1 images were also spatially normalized to the MNI template. The signal from cerebrospinal fluid and white matter was regressed out before functional connectivity analysis.
A statistical parametric map was created for each participant by using a general linear model to calculate the contrast of HCS vs. non-food, HCS vs. LC, and HCS vs. HNS at each voxel. In accordance with previous work on food-related stimuli, our regions of interest (ROIs) were the thalamus, basal ganglia (putamen, caudate, nucleus accumbens), OFC, ACC, insula, and HPC. The ROIs were registered to structural images and to the Automated Anatomical Labeling atlas version 3 (AAL3) atlas (https://www.gin.cnrs.fr/en/tools/aal) [35]. We reported significant clusters within each ROI that initially reached a threshold of P<0.05, with a minimum cluster size of 405 mm3, and were Alpha SIM corrected for multiple comparisons.
Statistical analysis
Statistical analyses were performed using SPSS version 25.0 software (IBM Co., Armonk, NY, USA) and R version 4.3.0 software (R Foundation for Statistical Computing, Vienna, Austria). Kolmogorov-Smirnov test was used to determine the normal distribution of data. Data were expressed as mean±standard deviation, median (interquartile range), and percentage appropriately. Data with skewed distribution were natural logarithmically transformed (loge-transformed) before analysis. Characteristics were compared between groups by Student’s unpaired t-test for continuous data and chi-squared test for categorical data, respectively. The comparisons of sweetness and liking levels at different concentrations between the two groups were conducted using two-way analysis of variance, with Bonferroni adjustment for multiple comparisons. Pearson and partial correlation analyses, along with mediation analyses were appropriately utilized. Moreover, stepwise multivariable linear regression analysis was conducted to explore the factors influencing sweet taste perception and preference. All statistical tests were two-tailed, with a 0.05 significance level.
RESULTS
Clinical characteristics of study participants
A total of 40 patients with T2DM and 41 controls completed the sweet taste perception and preference tests. The characteristics of the all participants are shown in Table 1. In patients with T2DM, the BMI, waist circumference, fat mass, and HOMA-IR were significantly higher compared to the control group. Additionally, the T2DM group exhibited significantly lower levels of HOMA-β and HDL-C (all P<0.05).
Characteristics of participants between control and T2DM patients
Altered sweet taste perception and preference in patients with T2DM
T2DM patients demonstrated impaired sweet taste perception and increased sweet preference compare to control (Fig. 2). Sweet taste perception was significantly lower in T2DM patients at glucose concentrations of 0.32 mol/L (P=0.011), 0.56 mol/L (P<0.001), and 1 mol/L (P=0.004), with a significant interaction between glucose concentration and group (P<0.001). Meanwhile, the AUC of sweet taste perception was also significantly lower in participants with T2DM (P<0.001). Conversely, T2DM patients showed an increased preference for sweet tastes at a glucose concentration of 1 mol/L (P=0.022), as well as an elevated AUC of preferences (P<0.001).
Reduced sweet taste perception and increased preference in patients with type 2 diabetes mellitus (T2DM). (A, C) Using two-way analysis of variance (ANOVA) and Bonferroni’s multiple comparison test. Values were mean±standard error of the mean. (B, D) Area under the curve (AUC) of sweet taste perception and preference across the full range of concentrations tested in the two groups, respectively, using Student’s unpaired t-test: Blue lines and symbols, control (n=41); red lines and symbols, patients with T2DM (n=40). M, mol/L. aP<0.05, bP<0.01, and cP<0.001 indicate a difference between patients with T2DM and control.
FGF21 levels correlate with sweet taste perception and preference in patients with T2DM
Our findings indicated that FGF21 levels were significantly elevated in individuals with T2DM compared to control (P=0.025). Then, we conducted an analysis to explore the intricate associations in sweet taste perception, preference, FGF21 levels, and clinical characteristics in patients with T2DM (Supplementary Table 1). In T2DM patients, FGF21 levels negatively correlated with sweet taste perception at 0.32 mol/L (r=–0.35, P=0.027) and positively correlated with the AUC of sweet taste preference (r=0.32, P=0.046). Also, there was a negative correlation between HOMA-β and sweet taste perception at 0.56 mol/L (r=–0.32, P=0.045). We further investigated the influence factors of sweet taste perception and preference, using stepwise multiple linear regression (Table 2). The results showed that FGF21 levels (β=–0.37, P=0.008), fat mass (β=1.26, P<0.001), and BMI (β=–1.02, P=0.003) were found to be independently associated with sweet taste perception. Additionally, FGF21 levels were independently associated with sweet taste preference (β=0.32, P=0.046). No significant correlation was observed between FGF21 levels and sweet taste perception or preference in the control group (Supplementary Table 2). These findings revealed an association between FGF21 levels and sweet taste perception and preference in patients with T2DM.
Stepwise multivariable linear regression for sweet taste perception and preference in T2DM patients
Enhanced neural responses to high-calorie sweet foods in patients with T2DM
We further used fMRI to explore the brain neural responses to HCS foods between T2DM patients and control (Supplementary Table 3). Our results demonstrated that in the control group, HCS foods stimuli, in contrast to non-food stimuli, effectively activated the left insula. Notably, LC foods provoked a more substantial neural response within the right HPC and putamen compared to HCS foods. Compared with HCS foods, HNS foods induced a higher activation in the right putamen, ACC and bilateral HPC (Supplementary Table 4). In patients with T2DM, HCS foods stimuli significantly enhanced the activity in the right insula, left HPC, putamen and thalamus compared to non-food stimuli. In contrast to LC foods, HCS foods induced a stronger activation in the right putamen. Furthermore, HCS foods evoked a more significant neural response in the left HPC and caudate compared to HNS foods (Supplementary Table 4).
Compared to control, T2DM patients exhibited significant differences in neural responses to the disparity between HCS foods and other foods or non-food stimuli (Table 3). When contrasting HCS foods with non-food stimuli, patients with T2DM exhibited more neural activation in the right OFC, ACC, left thalamus, putamen and bilateral HPC compared to control (Fig. 1B and C). In the comparison of HCS to LC food pictures, the right putamen in T2DM patients showed more activation than in the control group (Fig. 1D and F). Further analysis revealed that in the right parahippocampal gyrus, insula, left ACC and bilateral HPC, the neural activation differences in response to HCS versus HNS foods were markedly pronounced in T2DM patients compared to the control group (Fig. 1E and G). Additionally, in the contrast between HCS foods and other foods or non-food stimuli, no significant increase in activation within the ROIs was observed in the control group relative to patients with T2DM. These findings suggested that the neural activation patterns within the brain’s reward circuitry in response to HCS foods stimuli in T2DM patients differed from those in control, indicating a unique neural mechanism for processing sweet signals in reward-related brain regions.
Neural response comparisons of high-calorie sweet foods between T2DM patients and control across conditions
FGF21 levels partially mediate the relationship between T2DM status and neural responses to HCS food
To examine the relationship between FGF21 levels and neural response, we conducted correlation analyses across all participants (both T2DM and control) for the contrasts HCS vs. nonfood, HCS vs. LC and HCS vs. HNS stimuli and serum FGF21 levels (Supplementary Table 5). Higher serum FGF21 levels significantly correlated with greater neural response to HCS relative to non-food stimuli in the right ACC. For the HCS vs. HNS contrast, FGF21 levels were strongly associated with increased activation in the left ACC, right parahippocampal gyrus, insula, and bilateral HPC. These associations remained significant after adjusting for age and gender (Supplementary Table 5). Furthermore, we conducted a mediation analysis to investigate whether serum FGF21 levels mediated the relationship between T2DM status and neural responses (Fig. 3). The results revealed that the mediating effect of FGF21 accounted for 30.39% (P=0.034) of the association between T2DM status and the left ACC, and 32.43% (P=0.046) between T2DM status and the right parahippocampal gyrus. These findings suggested that FGF21 may partially mediate the relationship between T2DM and neural responses.
The mediation role of serum fibroblast growth factor 21 (FGF21) levels in the correlation between (A) type 2 diabetes mellitus (T2DM) status and left anterior cingulate cortex (ACC) (high-calorie sweet [HCS]>high-calorie non-sweet [HNS]), and (B) T2DM status and right parahippocampal gyrus (HCS>HNS). ACME, average causal mediation effect; CI, confidence interval; ADE, average direct effect; L, left; R, right. aP<0.05 and bP<0.01, a quasi-Bayesian Monte Carlo simulation method (number of simulations=1,000) was used to calculate the significance of the mediation proportion, cProportion of the mediation: ACME/total effect.
DISCUSSION
This study revealed novel insights into sweet perception, preference, and neural responses in individuals with T2DM, as well as the potential role of FGF21 in these processes. Our findings demonstrated that T2DM patients exhibited decreased sweet taste perception but increased sweet preference compared to the control. Importantly, we found that FGF21 levels were independently associated with sweet taste perception and preference in patients with T2DM. fMRI analysis revealed heightened activation in reward-related brain regions in T2DM patients in response to HCS vs. non-food, HCS vs. LC, as well as HCS vs. HNS pictures, in contrast to healthy individuals. Notably, FGF21 levels may partially mediate the relationship between T2DM and neural responses in both the left ACC and the right parahippocampal gyrus.
Our observation of increased sweet taste perception thresholds in T2DM patients aligns with previous study [36]. This impairment in sweet taste recognition may be more pronounced in poorly-controlled diabetes or in patients with complications [37]. Additionally, the heightened preference for sweet taste in T2DM patients observed in our study corroborates findings by Fernandez-Carrion et al. [38]. However, Al-Ghurayr et al. [39] reported no significant difference in sweet taste preference between T2DM patients and healthy controls. These discrepancies may stem from heterogeneity in patient populations, glycemic control, presence of chronic complications, medication use, and other potential confounding factors.
While previous fMRI studies have demonstrated increased brain responses in the thalamus, insula, putamen, and OFC towards HC foods compared to LC foods in healthy individuals [14,19,20], our study uniquely examines sweet food responses in T2DM patients. The OFC, through its interactions with the amygdala and ACC, plays a crucial role in assigning reward values to stimuli [17]. Additionally, HPC activation has been associated with food craving and observation of reward-predicting images [23,24]. Despite numerous fMRI studies focusing on HC and LC food preferences in healthy or obese populations, research on sweet food responses in T2DM patients remains limited. Notably, Stice et al. [40] examined the relative role of fat and sugar in the activation of reward brain regions in lean adolescents, and found that sugar caused greater activation in the putamen and gustatory regions than fat. Similarly, Stoeckel et al. [19] reported that activations in the ACC, insula, HPC, caudate, and putamen were stronger in response to HCS foods compared to savory HC foods in obese individuals. Interestingly, we found that these brain regions also play a role in response to HCS foods stimuli in individuals with T2DM, exhibiting significantly heightened neural activity compared to control stimuli.
Our research demonstrated that patients with T2DM showed substantially higher activations in response to HCS vs. HNS foods stimuli in the right parahippocampal gyrus, insula, left ACC, and bilateral HPC compared to control. Furthermore, when comparing HCS foods to LC food pictures, the right putamen in T2DM patients also showed elevated activation levels. These results indicated that T2DM patients had a stronger desire for HCS foods relative to the control group. Of note, the insula, OFC, and ACC are recognized as primary, secondary, and tertiary taste cortical areas, respectively [41]. In our study, these specific brain regions were more responsive to HCS food pictures in participants with T2DM in contrast to the control group, which may reflect changes in the calibration of signals related to sweet taste perception and food calories.
We found significantly higher levels of FGF21 in individuals with T2DM, and the current data correlates increased endogenous FGF21 levels with lower sweet taste perception, higher sweet taste preference, and greater neural activation in the ACC and parahippocampal gyrus. Taste epithelial cells do not express FGFR1 or KLB, so the effect of FGF21 on sweet food preference does not seem to be achieved by affecting taste [42]. In fact, circulating FGF21 is mainly generated in the liver and can be delivered to the central nervous system to regulate macronutrient preference via activating FGFR1 and KLB [4]. Many regions of the brain, such as ventromedial hypothalamus, basal ganglia, and ventral tegmental area, can express FGF21 receptor complexes [43-45]. FGF21 signaling to glutamatergic neurons in the ventromedial hypothalamus mediated sugar suppression and sweet taste preference [8]. However, long-term obesity and chronically high FGF21 can reduce the expression of FGF21 receptor complexes and cause FGF21 resistance [46,47]. The resistance of central FGF21 may result in a reduction in the signaling of HCS food pictures, thereby maintaining brain signaling related to HCS foods motivation. The putamen can receive connections from the OFC and ACC [48]. In the present study, an increase in FGF21 levels was observed to be positively correlated with neural activation in the ACC, and FGF21 may serve as a partial mediator between T2DM status and ACC activation. Although our findings should be interpreted cautiously, it is plausible to speculate that FGF21 resistance or altered FGF21 signaling may reduce HCS foods signaling in these brain regions of T2DM patients, potentially leading to hyperactivation of reward-motivation regions and affecting their perception and preference for sweetness.
To our knowledge, this is the first study to establish an association between serum FGF21 levels and sweet food preference in patients with T2DM. Our findings also suggest that FGF21 levels may serve as a partial mediator in the relationship between T2DM status and neural responses to HCS food pictures. These results provide preliminary insights into the appetitive regulatory mechanisms in T2DM. Limitations of our study include a relatively small sample size for the fMRI analysis. Furthermore, although we reported the interactions between FGF21 levels and sweet taste perception, preference, and neural responses in T2DM patients, these preliminary findings require further validation through prospective studies. Additionally, as Bae et al. [49] reported disparities in brain responses to HC and LC food stimuli between lean and obese individuals with T2DM, future research should investigate potential differences in sweet food preference between lean and obese T2DM patients.
FGF21 plays a metabolic regulatory role under conditions of nutritional excess and insulin resistance [50]. Our study demonstrated that elevated FGF21 levels were associated with decreased sweet perception and increased sweet taste preference in T2DM patients. We also observed increased activation in reward-related brain areas in response to HCS food stimuli in T2DM patients, which may be partially mediated by FGF21 levels. These findings suggest that FGF21 resistance and its impaired action in patients with T2DM may lead to abnormal activation of reward-related brain regions, thus affecting the perception and preference for sweetness in these individuals. Our results provide new perspectives on the complex interplay between FGF21, sweet taste perception, and brain reward circuitry in T2DM, opening up novel research directions for potential FGF21-based interventions in managing dietary preferences and metabolic control in T2DM patients.
SUPPLEMENTARY MATERIALS
Supplementary materials related to this article can be found online at https://doi.org/10.4093/dmj.2024.0390.
Correlations of sweet taste perception and preference in T2DM patients with clinical characteristics
The correlation between serum FGF21 levels and sweet taste perception or preference in the control
Characteristics of participants between control and T2DM patients underwent functional magnetic resonance imaging
Within-group comparisons for the control group and T2DM group contrasting differences in the highcalorie sweet food and other food or non-food conditions
The correlation between neural activation (β-value) and serum FGF21 in all participants
Notes
CONFLICTS OF INTEREST
No potential conflict of interest relevant to this article was reported.
AUTHOR CONTRIBUTIONS
Conception or design: Y.L., H.L.
Acquisition, analysis, or interpretation of data: P.K., Y.Z., D.Z., S.L., L.W., Q.W., S.Y., A.C., J.G., W.G., J.N., J.Y., X.W., L.M., W.J., Q.F.
Drafting the work or revising: P.K., D.L., R.H., Y.L., D.C., Q.W., S.L., L.W., Q.W., S.Y., A.C., J.G., W.G., J.N., J.Y., X.W., L.M., W.J., Q.F.
Final approval of the manuscript: all authors.
FUNDING
This study was supported by the National Key Research and Development Program of China (2022YFA1004804) to Huating Li and Weiping Jia; the Excellent Young Scientists Fund of NSFC (82022012), General Fund of NSFC (82270907), Major Program of NSFC (92357305), Innovative Research Team of High-level Local Universities in Shanghai (SHSMU-ZDCX 20212700) to Huating Li; the Shanghai Municipal Key Clinical Specialty (2017ZZ01013), Shanghai Research Center for Endocrine and Metabolic Diseases (2022ZZ01002) to Weiping Jia; the National Natural Science Foundation of China (82100879), Shanghai Pujiang Program (2020PJD044), and Exploration Fund Grant of Shanghai Sixth People’s Hospital (ynts202003) to Liang Wu.
ACKNOWLEDGMENTS
We would like to thank all the participants and investigators for their contributions to this study.