Potential Effects of Prolonged Water-Only Fasting Followed by a Whole-Plant-Food Diet on Salty and Sweet Taste Sensitivity and Perceived Intensity, Food Liking, and Dietary Intake

The overconsumption of calorie-dense foods high in added salt, sugar, and fat is a major contributor to current rates of obesity, and methods to reduce consumption are needed. Prolonged water-only fasting followed by an exclusively whole-plant-food diet free of added salt, oil, and sugar may reduce the consumption of these hyper-palatable foods, but such effects have not been quantified. Therefore, we conducted a preliminary study to estimate the effects of this intervention on salty and sweet taste detection and recognition thresholds and perceived taste intensity after at least five days of fasting and at refeed day three. We also assessed the effects on sweet, salty, and fatty food preference and overall dietary consumption 30 days after the day three refeed visit. Based on this data, we estimated that 10 days after the start of the fasting, salty taste recognition, sweet taste detection, and sweet taste recognition thresholds decreased significantly, salty taste intensity ratings increased significantly, and sweet taste intensity ratings decreased significantly. We also have preliminary data that prolonged water-only fasting followed by refeeding on an exclusively whole-food-plant diet may reduce salty/fatty and sweet/fatty food liking, reduce sugar intake, and increase vegetable intake. These results support further research into the effects of fasting and diet on taste function and food likability and consumption.


Introduction
Global obesity has more than tripled in the last 30 years, affecting more than 70 million adults and 13 million children in the United States alone. Obesity is correlated with all-cause mortality and is among the leading preventable causes of death worldwide [1]. One explanation for rising obesity rates is the overconsumption of calorie-dense, nutrient-poor foods high in salt, sugar, and/or fat [2]. Sugar, salty, and fatty foods are also hyper-palatable [3], which may contribute to their overconsumption [4].
Human taste sensation begins when molecules in food and drink bind to taste bud receptors on the tongue (e.g., sucrose binds to the heterodimer T1R2-T1R3) [5], activating downstream neurological processes enabling humans to distinguish sweet, salty, sour, bitter, and umami tastes as well as sense and respond to nutrient intake [2]. Data regarding the effects of taste sensitivity, perceived taste intensity, and/or taste liking on food consumption or vice versa is largely inconclusive but studies suggest that taste may be malleable in response to dietary change [6]. Data also suggest that palatability may be a better indicator of food consumption than taste sensitivity or perceived intensity [7].
Prolonged water-only fasting, defined as the complete abstinence of food and liquids except for pure water for >48 hours, is an established, well-tolerated method of therapeutic fasting that has been practiced for millennia to treat a variety of conditions and has been found to be safe under medical supervision [8]. Clinical observation led to the hypothesis that prolonged water-only fasting followed by an exclusively whole-plant-food diet results in decreased salty and sweet food palatability along with reduced obesogenic food and increased whole-plant food consumption [9]. As an initial step in testing this hypothesis, we conducted a preliminary study to estimate the effects of the intervention on salty and sweet taste sensitivity and perceived taste intensity as well as the longer term effects on food preference and dietary consumption. This study was approved by the Institutional Review Board at the National University for Natural Medicine (RB61917, 10/2017). The research was conducted as described in the approved protocol and complied with the standards of the Declaration of Helsinki.

Study participants
Participants in this study were recruited from incoming patients at the TrueNorth Health Center. Between October 2017 and November 2018, we enrolled 30 consenting participants ages 18-75 years who were elected and were approved by a non-research clinician to undergo a water-only fast lasting five to 14 days. Patients were ineligible if they had been to the health center or smoked tobacco or marijuana products within the last 12 months, ate a diet free of added salt, oil, or sugar, had a self-reported taste or smell disorder, had active cancer, diabetes, or hypertension diagnosis, were pregnant or lactating within the last three months, or had a known allergy to potential testing compounds (sodium chloride, sucrose, citric acid, or caffeine). One of 30 participants did not complete the dietary screener questionnaire (DSQ) for unknown reasons. Six of 30 participants were unable to complete the study requirements due to a change in treatment plan length unrelated to fasting. One out of 30 participants withdrew from the study due to nausea. Of participants completing the refeed day 3 visit (n=23), 20 completed the web-based DSQ and 13 completed the web-based food-liking questionnaire at the remote, follow-up visit.

Study protocol and data management
Enrolled participants attended study visits at baseline (before the start of fasting), after five days of fasting (FD5), the final day of fasting in participants fasting between 10 and 14 days, and after three days of refeeding (RD3) on an exclusively whole-plant-food diet free of added salt, oil, and sugar as previously described [8]. At each visit, participants had weight and blood pressure measured and completed taste detection and recognition threshold (DT and RT) and intensity rating assays as described below. At baseline, participants also reported demographics, had their height measured, and completed the DSQ and foodliking questionnaire (FLQ) as described below. At a remote, follow-up visit 30-days after RD3, participants completed the same independently accessed, web-based DSQ and FLQ. Study data were collected, stored, and managed using REDCap electronic data capture tools hosted at the TrueNorth Health Foundation.

Water-only fasting and refeeding
Water-only fasting and refeeding on an exclusively whole-plant-food diet free of added salt, oil, and sugar were medically supervised by an independent clinician specializing in water-only fasting and refeeding at the TrueNorth Health Center (TNHC). Research personnel had no input into the participants' treatment plans. The water-only fasting protocol reportedly used at TNHC has been previously described and found to be tolerable and low risk [8].

DT and RT assays
DT is the lowest measurable concentration that a stimulus is detected as something other than water. RT is the lowest measurable concentration that a stimulus is recognized as having the correct characteristic taste (e.g., sodium chloride is salty and sucrose is sweet). DT and RT assays were conducted in a private room that minimized outside auditory, olfactory, and visual stimuli. Participants were instructed to refrain from eating, drinking (except for room temperature, distilled water), brushing teeth, and chewing gum up to one hour prior to testing. Test administrators and participants were blinded to testing solutions.
A standard three-alternative-forced-choice (3AFC) method was used to determine sweet and salty taste DT and RTs as previously described [10][11][12]. Briefly, the assay consisted of presenting participants with three 10mL samples labeled with a five-digit numbering system. The first group contained three samples of distilled water and the following groups contained two samples of distilled water and one sample of the test stimuli in ascending concentration. Sample order within groups was randomized. For each trial, participants were asked to identify which of the three samples differed from the other two and whether or not they recognized the stimulus as sweet, salty, sour, bitter, or umami. There was not an option for participants to indicate that they could not tell a difference between samples, thus they were required to indicate a sample that was different from the others. Participants were instructed to follow the sip-and-spit procedure, in which they swish a solution for five seconds, spit it out, record results, and then rinse with distilled water. Trials were separated by at least 15 seconds.

Suprathreshold intensity rating assay
Intensity ratings are commonly used to assess perceived taste intensity [10]. Assays were conducted in the same room and directly following the DT and RT assays. Participants were similarly asked to refrain from eating, drinking (except for room temperature, distilled water), brushing teeth, and chewing gum up to one hour prior to testing. Test administrators and participants were both blinded to testing solutions. Perceived intensity was assessed using four different concentrations of sodium chloride and sucrose [10]. For each stimuli, the samples were presented randomly. Participants followed the same swish-and-spit procedure and then rated the intensity of each concentration using a visual analog scale (VAS) graded from 0 to 100 where 0 = no detectable taste and 100 = extremely intense taste as previously described [13].

Food-liking and DSQs
Participants' preference for standard foods was assessed using a modified version of the PrefQuest food liking questionnaire (see Appendix Table 6 for full questionnaire) [14]. The questionnaire was translated from French to English and French foods uncommonly eaten in the USA were exchanged for comparable American foods. The modified version was beta tested on 40 people and adapted further to correct for participant feedback. Participants' dietary consumption was assessed using the web-based, selfadministered DSQ from the National Health and Nutrition Examination Survey (NHANES) 2009-2010 series as described [15].

Statistical analysis
The sample size for this study was determined on a model-based simulation study of changes in suprathreshold intensity from day 0 to day 5 of fasting. The target of the simulation was estimating 90% confidence intervals (CI) instead of power. This approach for justifying pilot study sample size is recommended when little is known about the parameter(s) of interest over a hypothesis testing approach [16]. Based on having no knowledge of fasting effects on taste adaptation and limited knowledge of taste adaptation in general, we judge that a width of 1.5 or less is an acceptable precision for the range of plausible values (i.e., confidence intervals) for each effect size. Based on the simulation study, a sample size of 20 appears sufficient to create 90% confidence intervals with a width of less than 1.5 for each taste. To accommodate possible loss to follow-up, we recruited a sample size of 30 for this pilot study. The scripts and reports used to obtain sample size estimates are available upon request.
Descriptive statistics were calculated for demographics, clinical characteristics, fasting lengths, as well as mean health metrics at baseline, fasting day 5, and refeed day 3. DSQ data were analyzed to calculate predicted intakes of fiber (g), whole grains (oz equivalents), total added sugars (tsp equivalents), added sugars from sugar-sweetened beverages (SSBs) (tsp equivalents), vegetables (cup equivalents), fruit (cup equivalents) at baseline and 30-day post-RD3 visits. Two-tailed, paired t-tests were used to estimate mean differences in health metrics comparing RD3 vs. baseline and predicted intakes of foods and nutrients comparing 30 days after RD3 vs. baseline. DT and RT were summarized by mean and standard deviations for each assay and time point; moreover, generalized estimating equations (GEEs) were fit to estimate changes in DT and RT 10 days after starting water-only fasting. For analysis of perceived intensity ratings, we excluded the zero concentrations as only five of the 160 tests had a value other than zero at this concentration. The mean and standard deviation of perceived intensity responses were computed for each assay, concentration, and time point. GEEs were fit to estimate intensity ratings 10 days after starting the water-only fast. Note that 10 days after starting water-only fasting, some participants were still in the fasting stage, some were in the refeeding stage, and some had completed refeeding.
To estimate the marginal effect of days since start of fast on intensity ratings, DTs, and RTs, GEE [17] models with log(1+response) were fit with terms for days since start of fast, log10(concentration), and their interaction. Models were fit separately within each taste. Parameter estimates on the original scale and point-wise 90% CIs were obtained by the Delta method. For food liking and diet consumption, two tailed paired t-tests were used to estimate mean differences comparing survey responses collected at baseline to survey responses collected remotely 30 days after departure. SAS 9.4 (Cary, NC) was used to analyze demographic, health metric, and food and nutrient data. Taste DT and RT, suprathreshold, and food-liking data analyses were done in R version 3.6.1 (2019-07-05) [18] using the geepack [19], geex [20], and ggplot2 [21] packages. The R scripts used to run these models are available upon request.

Salty and sweet taste DT and RT
To assess the effect of water-only fasting and refeeding on salty and sweet taste sensitivity, we used the 3AFC method to measure mean differences in DT and RT of 10 different concentrations of sodium chloride and sucrose, respectively (see Materials and Methods). Mean DT concentrations for sodium chloride remained unchanged at FD5 (4.6mM) and at RD3 (4.2mM) compared to baseline (4.3mM; Table 3). The mean sodium chloride concentrations at which salty taste was recognized changed from 40.8mM at baseline to 31.0mM at FD5 and 26.4mM at RD3 ( Table 3). Mean DT concentrations for sucrose were also unchanged at FD5 (15.2mM) and RD3 (8.1mM) compared to baseline (11.6mM; Table 3). The mean sucrose concentrations at which sweet taste was recognized were unchanged from 25.7mM at baseline to 18.2mM at FD5 and 18.2mM at RD3 ( Table 3).

FIGURE 1: Changes in salty (NaCl) and sweet (sucrose) detection and recognition thresholds overtime
Actual (grey) and estimated (blue) NaCl detection (A) and recognition (B) and sucrose detection (C) and recognition (D) thresholds from days since start of fast. Each grey line is a single participant's trajectory. Colored lines show GEE model fit with pointwise 90% confidence intervals. Value is estimated change 10 days from start of fast with 90% confidence intervals. NaCl, sodium chloride; mM, millimolar.

Salty and sweet perceived intensity ratings
We assessed the effects of water-only fasting and refeeding on perceived salty and sweet taste suprathreshold intensities by comparing mean VAS ratings (0-100) for four concentrations of sodium chloride and sucrose, respectively. Mean salty taste VAS intensity ratings uniformly increased towards more intense at FD5 and RD3 compared to baseline for all non-zero sodium chloride concentrations ( Table 4). For example, at a sodium chloride concentration of 50mM, mean (SD) salty taste intensity ratings increased from 24.7 (14.21) points at baseline to 30.5 (23.94) points at FD5 and 34.8 (22.23) points at RD3 ( Table 4). Mean sweet taste VAS intensity ratings remained unchanged or decreased slightly towards less intense at FD5 and/or RD3 compared to baseline ( Table 4). For example, at a sucrose concentration of 100mM, mean (SD) sweet taste intensity decreased from 30.0 (13.14) points at baseline to 29.3 (17.01) points at FD5 (
GEE models were fit to estimate changes in perceived salty and sweet taste intensity ratings overtime. The GEE model estimates that 10 days after starting water-only fasting the mean (90% CI) salty taste VAS intensity rating would increase by 5.5 (0.7, 10.2), 5.1 (0.2, 9.9), and 0.9 (-4.9, 6.8) points for the 50, 100, and 200mM sodium chloride concentrations, respectively. The GEE model estimates that 10 days after starting of water-only fasting the mean (90% CI) sweet taste VAS intensity rating would decrease by -5.4 (-9.

Food liking and diet consumption
We assessed changes in food liking using a previously validated salty, sweet, and fatty food-liking questionnaire (FLQ) that was adapted to American English and administered on-site at baseline and remotely 30 days after RD3. Food liking is calculated on a scale of 1 to 4 with a higher number indicating increased liking. Of the 23 participants who completed the RD3 visit, only 13 completed the follow-up FLQ. Figures 3A-3D show that for salty taste the FLQ score dropped by -0.36 points (P=.121), for sweet taste the FLQ score dropped by -0.22 points (P=.072), for fatty and salty tastes the FLQ score dropped by -0.74 points (P=.001), and for fatty and sweet tastes the FLQ score dropped by -0.72 points (P=.011). We assessed changes in food consumption using the NHANES web-based DSQ that was administered onsite at baseline and remotely 30 days after RD3. Out of the 23 participants who completed the RD3 visit, 20 completed the DSQ at the follow-up visit. The DSQ measures food consumption over the 30 previous days. Compared to baseline, at the 30-day, follow-up visit, there was a significant increase in cups of vegetables (0.4; P=.0054) and a significant decrease in the grams of total added sugar (-3.9; P<.0001) and added sugar from sugar sweetened beverages consumed (-0.9; P=.0084) ( Table 5 and Figure 4).

Discussion
The complex interrelationship between human taste and nutrient intake is poorly understood. It is especially hard to draw conclusions because of heterogeneous research methodologies and a lack of knowledge regarding the clinical meaningfulness of taste function measures. Nevertheless, increasingly high rates of obesity and chronic disease necessitate the need for continued research into interventions that may naturally reduce obesogenic food consumption. To this end, we conducted the first observational study measuring the effects of water-only fasting followed by a whole-plant-food diet on salty and sweet taste function. These data were then used in GEE modeling to estimate effects 10 days after starting water-only fasting followed by an exclusively whole-plant-food diet. Food liking and intake were also assessed 30 days after RD3.
We observed a trend toward decreased mean sodium chloride and sucrose DT and RT concentrations after five days of water-only fasting that was further supported by GEE modeling estimates. This is in agreement with previous data showing that salty and sweet taste recognition sensitivity increased after only 13 hours of water-only fasting [22]. It is accepted that individuals with lower DT and RT have increased taste sensitivity [10], but standard threshold ranges, if/how one threshold affects the other, and if/how they are correlated to taste preference and/or dietary consumption (let alone health outcomes) has not been elucidated. Therefore, it is difficult to draw conclusions beyond that fasting and diet may modulate salty and sweet taste function toward increased sensitivity.
Perceived taste intensity represents the perceived concentration of a taste that is above the RT [10]. GEE models estimated that there was a uniform trend toward increased perceived salty taste intensity. Conversely, GEE models estimated that there was a uniform trend toward decreased perceived sweet taste intensity. Given the relatively small change in perceived salty and sweet taste intensities, these changes may not be biologically or clinically meaningful. However, previous studies have found that decreasing salt and sugar consumption reduced their respective perceived intensities [6,23], and perceived sucrose taste intensity (as well as palatability) may be a factor in overconsumption of sweet foods and beverages [24].
Indeed, we observed a trend toward decreased salty and sweet food liking and a significant decrease in salty and fatty and sweet and fatty food liking at the follow-up visit. Additionally, there was a decrease in predicted sugar and sugary beverage intake and an increase in vegetable and fiber intake at the follow-up visit. The DSQ does not report specifically on salt or fat intake, so we are unable to assess if salty (or fatty) food consumption also declined, but the data suggest vegetable consumption increased. Nonetheless, these preliminary results provide support for the hypothesis that water-only fasting followed by a whole-plantfood diet may reduce preference for salty, sweet, and fatty foods and alter dietary consumption accordingly.
Salty taste mechanisms occur through both taste buds and trigeminal fibers and remain poorly defined relative to other tastes. Interestingly, oral microbiota composition was found to impact taste function with an increased abundance of six specific taxa associated with lower salty taste thresholds [25]. It is unknown if fasting affects oral microbiota but there is evidence that a "vegan" diet affects oral microbiota diversity, structure, and relative abundance [26]. Natriuresis is a well-known consequence of fasting, particularly in obese or hypertensive individuals [27], but this effect is unlikely to be related to the increased salty taste sensitivity and perceived intensity observed after fasting as there is evidence that sodium balance is not regulated by taste function [28].
Sweet taste mechanisms are fairly well defined and appear to function differently than other tastes in that there are sweet taste receptors throughout the body, including on cells in the gastrointestinal tract and central nervous system [5]. Additionally, sweet taste sensitivity, but not salty, sour, bitter, or umami, varies throughout the day with lower RTs observed in the morning and higher RTs at night in a pattern that is positively correlated with leptin levels and negatively correlated with blood glucose levels [29]. Fasting also decreases leptin and blood glucose levels [30], which may be one potential way in which fasting affects sweet taste sensation and perception.
This study has notable limitations including that participant dropout successively increased over the course of the study with only 46% of participants completing the FLQ at the follow-up visit. This combined with the low sample size at baseline and high inter-individual variations in taste may have affected our ability to detect statistically significant changes in threshold concentrations. As such, future studies should utilize a priori sample size calculations based on the effect sizes found in our study and employ strategies to improve retention rates and refine inclusion-exclusion criteria to minimize population variability. Additionally, the FLQ, which was translated from French to English, has not been formally tested for reliability and validity and an independent study is needed. The questionnaire also consisted of over 80 questions, which may account for the low recidivism in this questionnaire. The DSQ also did not provide information on salt or fat consumption, which prevented the assessment of the effects of the intervention on the consumption of these foods. Therefore, it may be useful to use other/additional methods or questionnaires to assess taste function as well as food liking and dietary consumption in future studies.

Conclusions
Obesity is a global issue that undoubtedly stems from unprecedented access to hyper-palatable foods with unnatural amounts of salt, sugar, and/or fat. That taste plasticity may be altered to affect palatability in such a way as to reduce the preference for and consumption of these foods (and promote the consumption of healthy foods) is appealing. This study began a preliminary assessment into the effects of prolonged wateronly fasting and refeeding with an exclusively whole-plant-food diet free of added salt, oil, and sugar on changes in taste function as well as food liking and consumption. Overall, the data are encouraging but additional research is necessary to determine if prolonged water-only fasting and refeeding may increase salty and sweet taste sensitivity, decrease liking of sweet, salty, and fatty foods, decrease sugar consumption, and increase vegetable consumption. Future studies continuing this work may benefit by measuring additional tastes, especially fatty, as well as by focusing on taste hedonics using foods with a complete taste profile rather than isolated tastants.    Values presented are mean (standard deviation). FD5, fasting day 5; RD3, refeed day 3; BMI, body mass index; SBP, systolic blood pressure; DBP, diastolic blood pressure; cm, centimeter; kg, kilogram; m2, square meter; mm, millimeter; Hg, mercury; y, years.

Additional Information Disclosures
Human subjects: Consent was obtained or waived by all participants in this study. National University of Natural Medicine issued approval RB61917. The study was conducted in accordance with the Declaration of Helsinki. Animal subjects: All authors have confirmed that this study did not involve animal subjects or tissue. Conflicts of interest: In compliance with the ICMJE uniform disclosure form, all authors declare the following: Payment/services info: All authors have declared that no financial support was received from any organization for the submitted work. Financial relationships: Alan C. Goldhamer declare(s) nonfinancial support from TrueNorth Health Foundation. A.C.G. is the unpaid president of the Board of Directors of the TrueNorth Health Foundation. Alan C. Goldhamer declare(s) Owner TrueNorth Health Center from TrueNorth Health Center. A.C.G. is the owner of the TrueNorth Health Center. Other relationships: All authors have declared that there are no other relationships or activities that could appear to have influenced the submitted work.