Higher fasting plasma FGF21 concentration is associated with lower ad libitum soda consumption in humans

ABSTRACT

Background

The hepatokine fibroblast growth factor 21 (FGF21) influences eating behavior and sugar consumption in rodent models. However, whether circulating FGF21 concentration is associated with food and soda intake in humans is still unclear.

Objective

We investigated whether fasting plasma FGF21 concentration is associated with objective measures of ad libitum food intake and soda consumption.

Methods

Healthy individuals [n = 109; 69 men, aged 34 ± 10 y; BMI (kg/m²): 30.4 ± 7.7; body fat by DXA: 30.5% ± 8.9%] with available plasma for hormonal measurements participated in an inpatient cohort study to objectively quantify ad libitum food and soda intake for 3 d using an automated and reproducible vending machine paradigm. Fasting plasma FGF21 concentration was measured by ELISA prior to ad libitum feeding.

Results

Fasting FGF21 concentration was inversely associated with daily soda intake (R = −0.22, P = 0.02 adjusted for demographics and anthropometrics), such that an interindividual difference of 200 pg/mL was associated with an average lower soda consumption by 68 kcal/d. Conversely, no associations were observed with total daily energy intake or macronutrient intake (all P > 0.17).

Conclusions

Higher plasma fasting FGF21 concentration is associated with lower ad libitum soda intake. Although this inverse correlation does not imply causation, the present results support the putative role of FGF21 in the reward pathways regulating sugar consumption in humans. This trial was registered at www.clinicaltrials.gov as NCT00342732.

Introduction

Chronic positive energy balance is responsible for weight gain, leading to obesity and its associated comorbidities. The underlying biological mechanisms that confer susceptibility to weight gain include hormones that are involved in daily energy homeostasis. Recently, our research group demonstrated that the extent of secretion of fibroblast growth factor 21 (FGF21), a liver-derived hormone (1), in response to acute low-protein overfeeding diets may characterize the human metabolic phenotype (thrifty vs. spendthrift) (2, 3) and identify individuals who are prone to gain weight over time (4–8).

Besides its effect on energy metabolism, FGF21 plays a role in regulating appetite and food preference (9, 10). Specifically, FGF21 suppression of simple sugar consumption may help reduce hepatoxicity and other metabolic consequences induced by excessive simple sugar consumption (11, 12). Fgf21-knockout mice consume greater amounts of sucrose (11). However, acute administration or overexpression of FGF21 suppresses the intake of both sugar and noncaloric sweeteners (9, 11), with no change in other nutrient consumption or total intake (13). In humans, FGF21 secretion increases following dietary interventions such as 3-d high-carbohydrate overfeeding (14)—in particular, fructose-rich diets (15). Human genetic variations in/near FGF21 are associated with increased self-reported carbohydrate and sugar intake (16, 17), with fasting FGF21 concentration being elevated in sweet-disliking individuals (10). Yet, most human studies (10, 16, 17) investigating the role of FGF21 on energy intake were based on self-reported dietary recalls or food-frequency questionnaires and did not objectively assess actual food and soda intake. In the current study, we investigated the relations between fasting plasma FGF21 concentrations and ad libitum food intake and soda consumption as objectively quantified by a reproducible vending machine paradigm in a large, ethnically diverse cohort of healthy individuals.

Methods

Participants and study design

Volunteers included in this cross-sectional analysis were part of a larger, ongoing, inpatient cohort study investigating the determinants of ad libitum energy intake assessed by a computerized vending machine paradigm (clinicaltrials.gov identifier: NCT00342732) (18, 19) conducted between 2003 and 2017 at the Clinical Research Unit of the NIH in Phoenix, Arizona. Of 250 participants who completed the study (Supplemental Figure 1), 109 had 1) available fasting plasma samples (collected since 2009) for the measurement of FGF21 and 2) valid measurements of ad libitum total, macronutrient, and soda intake over 3 d by vending machine systems during the inpatient admission. All participants were nonsmokers, not taking any type of medication, and were confirmed to be healthy based on medical history and laboratory testing on the day of admission. Prior to admission, all participants signed written and informed consent. The clinical protocol was approved by the Institutional Review Board of the National Institute of Diabetes and Digestive and Kidney Diseases. Upon admission, self-identified race was recorded using the following categories: American Indian/Alaska Native, Asian, Native Hawaiian or Other Pacific Islander, Black or African American, White, more than 1 race, unknown. Hispanic or non-Hispanic ethnicity was also recorded.

On the day of admission, participants were placed on a standard weight-maintaining diet (WMD; 50% carbohydrate, 30% fat, and 20% protein) for 3 d prior to any metabolic testing. The individual weight-maintaining energy needs (WMENs) of the WMD were initially calculated based on weight and gender and then adjusted, if needed, to maintaining a stable weight. Glucose tolerance was assessed by a 75-g oral-glucose-tolerance test and only individuals without diabetes according to the American Diabetes Association criteria were included in the analysis. Body composition [percentage of body fat, fat mass (FM), fat-free mass (FFM)] was assessed using DXA (DPX-1; Lunar Radiation Corporation), with FM and FFM indexes (FMI and FFMI) calculated as respective body masses divided by height squared.

Ad libitum energy intake assessment

After at least 4 d following the WMD, ad libitum food and soda intakes were assessed for 3 d using a highly reproducible, computerized vending machine paradigm (20, 21).

During this 3-d period, participants were given free access for 23.5 h to a dedicated vending machine, which was stocked with 40 different food items that had been rated by the volunteer with an intermediate/high score (score: 4–8) on a 9-point Likert scale (1 = dislike extremely, 5 = neutral, 9 = like extremely) on the basis of a food-selection questionnaire administered on the day of admission. Participants were also asked to report their preferred type of soda. Water, coffee, 2% milk, orange juice, apple juice, and six 12-ounce cans of the subject's soda choice were also available each day in the vending machine. The vending machine required the use of a security code to access the shelves and each shelf could be accessed only twice per meal, once to retrieve an item and a second time to return the wrappings and leftovers for weighing and exact determination of caloric intake by the metabolic kitchen. Each volunteer was asked to follow his/her typical eating pattern as closely as possible when accessing the vending machine room, which was equipped with a table, chair, a microwave oven, and a toaster. Volunteers were not permitted to watch television or read during meals.

All food was weighed prior to placement into the vending machine and returned food leftovers were weighed by the metabolic kitchen staff to determine actual energy intake. The CBORD Professional Diet Analyzer Program (CBORD, Inc.) and the Food Processor database (ESHA version 10.0.0; ESHA Research) were used to calculate the daily total and macronutrient kilocalories consumed. Soda kilocalories were also calculated and included in the total carbohydrate intake by summing the carbohydrate content of foods consumed. Each food was categorized based on the content of simple sugars, and the kilocalories of food high in simple sugars (>30% of total kcal) was calculated.

FGF21 measurement

Fasting blood sample was drawn at 05:30 h following a day on a eucaloric diet and prior to starting the 3-d ad libitum period. Plasma was frozen at −70°C for later measurements of FGF21 concentration using the human FGF-21 quantikine ELISA kits from R&D Systems (intra-assay CV = 2.5%, interassay CV = 5.2%). Storage time did not affect FGF21 measurement.

Analytic measures

The glucose oxidase method (Glucose analyzer GM9, Analox Instruments; Lunenburg, MA, USA) and the automated immunoenzymometric assay (Tosoh Bioscience, Inc.) were used to measure plasma glucose and insulin concentrations, respectively.

Statistical analysis

Data analyses were performed using SAS software (version 9.3; SAS Institute). Data are expressed as means ± SDs. Plasma FGF21 concentrations were log₁₀-transformed before analyses to meet the assumptions of parametric tests (i.e., homoscedasticity and normal data distribution) and the regression coefficients were exponentiated to express the effect size as a multiplier. The differences in FGF21 concentration between sexes and ethnicities were assessed by unpaired t test. Linear regression analysis was used to adjust measurements of ad libitum energy intake (outcome measurements) for its known determinants including age, sex, ethnicity, FFMI, and FMI (22). Specifically, adjusted values were calculated by adding the average value calculated in the whole cohort to the residual values obtained via regression analysis including the aforementioned covariates (23). The Pearson correlation index (R) was used to quantify associations between FGF21 concentration (“exposure”) and adjusted energy intake measurements (“outcomes”). The intraclass correlation coefficient (ICC) was calculated for soda intake measurements across the 3 d via a linear mixed-model analysis accounting for repeated measurements using an autoregressive (1) covariance structure to quantify within-subject consistency.

Results

Clinical characteristics of the study cohort are reported in Table 1. The average BMI (in kg/m²) was >30 (class I obesity) and the average daily ad libitum energy intake was 3791 kcal/d (SD = 1413 kcal/d), equivalent to 137% of WMENs. Daily ad libitum soda intake was consistent within a subject across the 3 d (ICC = 0.77; 95% CI: 0.70, 0.83), and differed by sex (mean difference = 137 kcal/d, P = 0.002), such that soda consumption was, on average, greater in men compared with women by 67% (95% CI: 57%, 217%). Soda intake was negatively correlated with age (R = −0.25, P = 0.007), whereas no associations were observed with anthropometrics (all P > 0.3) (Supplemental Table 1).

TABLE 1.

Clinical characteristics of the study cohort¹

	Whole cohort (n = 109)	Men (n = 69)	Women (n = 40)
Ethnicity, n	10 BLK, 37 WHT, 46 NAM, 12 HIS, 4 O	5 BLK, 24 WHT, 28 NAM, 9 HIS, 3 O	5 BLK, 13 WHT, 18 NAM, 3 HIS,1 O
Sex (F/M), n	40/69	69	40
Age, y	34.3 ± 10.1	34.5 ± 10.5	33.8 ± 9.5
Body weight, kg	87.4 ± 22.6	88.9 ± 19.6	84.9 ± 27.1
Height, m	1.69 ± 0.09	1.74 ± 0.06*	1.61 ± 0.07
BMI, kg/m2	30.4 ± 7.7	29 ± 6*	32.6 ± 9.8
FFM, kg	59.8 ± 13	64.7 ± 10.6*	51.3 ± 12.6
FM, kg	27.6 ± 13.2	24.2 ± 10.4*	33.6 ± 15.4
Body fat, %	30.5 ± 8.9	26.1 ± 6.7*	38.1 ± 7
Fasting glucose, mg/dL	92.8 ± 6.7	93.3 ± 7.3	91.8 ± 5.4
2-h OGTT glucose, mg/dL	124.8 ± 28	122 ± 29.7	129.7 ± 24.6
Fasting insulin, µIU/mL	12.8 ± 16.3	13.1 ± 19.8	12.5 ± 7.4
2-h OGTT insulin, µIU/mL	90.1 ± 72.0	88.5 ± 74.6	92.5 ± 75.6
Fasting FGF21,2 pg/mL	185.5 ± 119.2	189.4 ± 121.7	178.8 ± 116.1
Ad libitum energy intake measurements3
Total energy intake, kcal/d	3791 ± 1413	4178 ± 1375*	3123 ± 1229
WMENs, kcal/d	2744 ± 250	2833 ± 205*	2590 ± 248
Total energy intake, % of WMENs	137 ± 45	147 ± 43*	120 ± 45
Soda intake, kcal/d	292 ± 229	342 ± 243*	205 ± 174
Carbohydrate intake, kcal/d	1953 ± 705	2165 ± 677*	1586 ± 598
Carbohydrate intake excluding soda, kcal/d	1661 ± 579	1823 ± 558*	1381 ± 510
Fat intake, kcal/d	1444 ± 657	1585 ± 663*	1202 ± 582
Protein intake, kcal/d	460 ± 172	496 ± 167*	399 ± 164
Energy intake from high-simple-sugar food, kcal/d	1364 ± 716	1503 ± 647*	1124 ± 772

¹Values are means ± SDs. *P < 0.05 compared with women by unpaired Student's t test. BLK, Black; FFM, fat-free mass; FGF21, fibroblast growth factor 21; FM, fat mass; HIS, Hispanic; NAM, Native American; O, other race/ethnicity; OGTT, oral-glucose-tolerance test; WHT, White; WMEN, weight-maintaining energy need.

²Plasma FGF21 concentration was measured in fasting state (05:30 h) prior to the 3-d ad libitum period using the vending machine system.

³Ad libitum energy intake measurements by the vending machine paradigm are reported as the average of 3 d.

The fasting FGF21 concentration positively correlated with percentage body fat (R = 0.29, P = 0.002) and BMI (R = 0.27, P = 0.005) and was, on average, 60% lower in Blacks compared with other ethnicities (P = 0.003), with no difference between sexes (P = 0.59) (Supplemental Table 1). On average, Blacks also had lower daily ad libitum energy intake (mean difference vs. non-Black individuals = −993 kcal/d, P = 0.03), carbohydrate intake (mean difference = −149 kcal/d, P = 0.01), and protein intake (mean difference = −34 kcal/d, P = 0.01), whereas no differences were observed in fat or soda intake (both P > 0.07). Fasting FGF21 concentration was inversely associated with daily soda intake (partial R = −0.22; P = 0.02, adjusted for age, sex, ethnicity, FMI, and FFMI; Figure 1), such that an interindividual difference of 200 pg/mL in fasting FGF21 concentration was associated with an average lower soda consumption by 68 kcal/d. No associations were observed between FGF21 and total calories, macronutrient intakes, carbohydrate intake excluding soda kilocalories, and total intake from food with high simple sugar content (all P > 0.17) (Supplemental Table 2). There was no correlation between BMI and soda intake in our cohort including overweight individuals or individuals with obesity (P = 0.9).

FIGURE 1.

Higher fasting plasma FGF21 concentration is associated with lower ad libitum soda consumption. Inverse relationship between plasma fasting FGF21 concentration and ad libitum soda consumption assessed over 3 d using the vending machines. Soda intake values were adjusted for age, sex, FMI, FFMI, and ethnicity via linear regression analysis. Partial correlation value (R) is shown along with its significance (P) and sample size (n). FFMI, fat-free mass index; FGF21, fibroblast growth factor 21; FMI, fat mass index.

Discussion

In the current study including 109 healthy individuals predominantly with obesity, we investigated whether fasting FGF21 concentration was associated with objective and accurate measures of ad libitum food and soda intake. We found that a greater fasting FGF21 concentration was associated with lower soda consumption but not with total ad libitum food intake.

It has been proposed that the liver regulates energy intake (“hepatostatic theory”) via secreting hepatokines that provide information centrally about energy intake and reserve from the periphery (24). The hepatokine FGF21 plays a role in the regulation of energy homeostasis (25) as it acts through the FGF21 receptor complex, which is expressed in multiple regions of the brain implicated in energy intake the including paraventricular nucleus of the hypothalamus (26). Fgf21-knockout mice had decreases in preference for sweet and alcohol (9), whereas Fgf21 heterozygous and knockout mice had increased preference for a high-sucrose diet compared with wild-type animals, but with no difference in the amount of total energy or nutrient consumption (11).

In line with findings in mouse models, we found that higher plasma fasting FGF21 concentration was associated with lower soda consumption but not with ad libitum food intake in healthy participants who self-selected food and beverages from a computer-operated vending machine for 3 d. Consistent with our current findings, human studies show through food-preference questionnaires that sweet-disliking participants have higher fasting FGF21 concentrations compared with sweet-liking participants (10). Furthermore, variations (rs838145 and rs838133) in the genomic region that encompasses the FGF21 gene are associated with increased carbohydrate and sugar intake (10, 16, 27). Taken together, these findings provide compelling evidence that FGF21 is implicated in a hormonal liver-to-brain feedback loop, which regulates sugar consumption (25, 26). Because our current observational study cannot infer about causality between FGF21 concentration and soda consumption, future interventional studies in larger and more diverse cohorts are warranted to test the effects of exogenous FGF21 administration on soda and food intake.

A strength of our study is the objective assessment both of food and soda consumption for 3 d using an accurate and reproducible ad libitum method (20). Furthermore, we measured plasma FGF21 concentration by ELISA to obtain accurate measurements in the postabsorptive state. Based on data from our previous study (4) including repeated FGF21 measurements on different days in the same conditions as in the current study, the within-subject CV of fasting plasma FGF21 measurements was equal to 10%. Despite these strengths, this was a cross-sectional analysis, which included mainly individuals with obesity, which may limit generalizability. As the current cross-sectional study is nested within a cohort study with different aims, there might have been a potential selection bias due to including only individuals with available frozen plasma samples for measurement of FGF21 concentration. Yet, all consecutive participants were included in the current analysis to minimize any selection bias. Furthermore, the recorded intake in an inpatient setting may not reflect energy intake in free-living conditions. However, self-reported questionnaires (10, 16, 27) on energy intake are often imprecise, whereas our method is able to measure calorie intake accurately.

In conclusion, higher plasma fasting FGF21 concentration is associated with lower ad libitum soda intake but not with food intake. Although the observed inverse correlation between FGF21 and soda intake does not imply causation, the present results support the putative role of FGF21 in the reward pathways regulating sugar and soda consumption in humans.

Notes

This study was supported by the Intramural Research Program of the National Institutes of Health, National Institute of Diabetes and Digestive and Kidney Diseases. PP was supported by the program “Rita Levi Montalcini for young researchers” from the Italian Minister of Education and Research (Ministero dell'Istruzione, dell'Università e della Ricerca).

Supplemental Figure 1 and Supplemental Tables 1 and 2 are available from the “Supplementary data” link in the online posting of the article and from the same link in the online table of contents at https://academic.oup.com/ajcn/.

Abbreviations used: FFM, fat free mass; FFMI, fat-free mass index; FGF21, fibroblast growth factor 21; FM, fat mass; FMI, fat mass index; ICC, intraclass correlation coefficient; WMD, weight-maintaining diet; WMEN, weight-maintaining energy need.

Data Availability

Data described in the manuscript will be made available upon request pending application and approval by the Institutional Review Board of the NIDDK.