Developing and Evaluating a High Resistant Starch Menu and Glycemic Response in Adults with Prediabetes: A Non-Randomized Pilot Study

The nondigestible ber, resistant starch (RS), is found in beans, peas, and legumes, bread, cereals and grains, plantains, and potatoes. Here, we determined the feasibility of developing, preparing, and distributing a 7-day menu with a weekly average of ~30 g RS/d, as well as the likability of the foods. First, a was formulated using a database, then prepared, stored, and delivered to mimic a future clinical trial. The and and were to determine changes in weight, BMI, glucose, and insulin before and after


1) What uncertainties existed regarding the feasibility?
To date, a 2,000 kcal/d menu containing a weekly average of ~30 g of resistant starch (RS)/d has not been developed, nor prepared and tested for quality and acceptability in adults with prediabetes. It was unknown if a menu could be formulated to sustain ~30 g RS/d at consumption. The feasibility of preparing and distributing the menu to study participants, as well as whether the menu would be acceptable for consumption in adults with prediabetes, was also unknown.
2) What are the key feasibility ndings?
It is feasible to formulate a 7-day menu where a weekly average of ~30 g RS (veri ed by a commercial assay) and the recommended macronutrient ranges (veri ed by nutrient analyses) are provided daily. We also determined that it is feasible to prepare, store, and distribute the menu items to participants, which simulated a protocol that would be used in a larger clinical trial of longer duration. The likeability of the menu items was a rmed based on the overall likeability scores and amount consumed by the participants.
3) What are the implications of the feasibility ndings for the design of the main study?
Clinical trials using RS as an intervention have historically used a supplement form of RS. Here, through purposeful development, testing, and planning, it is feasible to develop a menu (breakfast, lunch, snack, and dinner) that contains similar amounts of RS (~30 g/d) known to improve glucose homeostasis and insulin sensitivity. The present study found that in just 7 days, the high RS menu improved body weight, glucose, and insulin in adults with prediabetes; however, future clinical trials should be powered to test these outcomes. Using naturally occurring RS in foods will likely have an advantage over RS supplements. The high RS foods contain other nutrients such as ber and protein that may provide combined positive health effects.

Background
Resistant starch (RS) is not broken down by digestive enzymes and thus enters the large intestine where it undergoes fermentation by resident microbiota. Hence because of this lack of digestibility RS is considered an insoluble ber. Five types of RS have been identi ed based on the nature and properties of the RS granule: RS1, granules are resistant because they are trapped in the cell wall or matrix making it inaccessible to digestive enzymes and are found in foods such as seeds and whole grains; RS2, granules are resistant because of its native granule conformation where a tightly packed amylose structure forms a crystallized molecule often seen in raw potatoes and green bananas, as well as legumes and highamylose corn; RS3 is formed when the starch granules are gelatinized then cooled which causes the starch to crystalize, a process known as retrogradation; RS4 is a chemically modi ed RS that is not naturally occurring in foods and is formed through processes like esteri cation, crosslinking or transglycosylation; RS5 is heat-stable and formed when lipids bind to amylose in the starch granule to prevent expansion of the granule, which is necessary for digestive enzyme hydrolysis [2].
Determination of RS in food has evolved considerably since it was initially identi ed by Englyst and colleagues, who recognized RS fractions in both plant foods (in vitro) and human ileostomy e uent samples (in vivo) [3]. Presently, RS is quanti ed using the Association of O cial Agricultural Chemists (AOAC) method 2002.02/ American Association of Cereal Chemists (AACC) method 32-40.01, and factors, such as sample preparation, sample pretreatment, enzymes used, and incubation conditions determine the amount of RS quanti ed. Commercial kits are available to estimate RS in foods [4].
Adequate consumption of RS improves overall metabolic and physiological health due to its role as a prebiotic that supports the growth, viability, and function of certain gut microbes [5]. In addition to improvements in metabolic health, RS can impact quality-of-life measures in diseased populations due to microbiome modulation. Non-motor and depressive symptoms improved and fecal butyrate, an indicator of RS fermentation, increased in adults with Parkinson's disease after consuming 10 g RS3/d for 8 weeks [6].
Strong clinical evidence shows RS improves glycemic control which can attenuate the development of several diseases, including type 2 diabetes [7]. US adults consume ~2.0 g RS/1,000 kcal/d from foods [8] which is much lower than what has been observed in trials that resulted in improved glucose homeostasis and/or insulin sensitivity [9][10][11][12]. This highlights the need for higher intake of dietary RS in adults with prediabetes -before type 2 diabetes develops. Because over 88 million adults in the US population were estimated to have prediabetes in 2018 [13] and prediabetes is prevalent in nearly 1 in 5 adolescents aged 12-18 years and 1 in 4 young adults aged 19-34 years [14] it is important address lifestyle and dietary changes when symptoms rst arise.
Almost all clinical trials have used the supplement form of RS as an intervention due to ease of administration. Very few trials have examined foods with naturally occurring RS on metabolic outcomes, and none have studied a complete weekly menu high in RS. In a recent randomized cross over study, we reported that consuming baked then chilled potatoes (high RS) resulted in postprandial glucose and insulin reductions at 15 and 30 minutes and lower total area under curve (tAUC 0−120 min ) insulin than a boiled potato (low RS) in females with elevated fasting glucose [15]. However, only one type of food, a potato, not a full menu was tested in the study. A recent systematic review and meta-analysis of randomized crossover trials (RCT) found that signi cant decreases in postprandial glucose and insulin were reported to be due to consumption of starchy foods with high amylose content characterized by low gelatinization, increased degree of retrogradation, and large starch particle size [16]. However, the majority of the studies identi ed in this review included RS as a supplement or as food prepared with RS forti ed ingredients thereby noting the need for studies using naturally occurring RS food sources. In addition, a speci c dietary plan has not been identi ed to reduce the progression from prediabetes to type 2 diabetes. The Diabetes Prevention Programs show that following a low-calorie, low-fat diet with exercise can attenuate the development of type 2 diabetes by 58% [17]. However, a speci c dietary menu plan has not been identi ed.
The focus on consuming nutrients through food, not supplements, prompted the idea of formulating a menu containing foods with similar amounts of RS that have shown to improve glycemic homeostasis in clinical trials. Thus, there are two main phases of the study that occurred in time sequence. The rst phase ( rst objectives) involved the formulation of a 7-day menu where a weekly average of ~30g RS/d would be provided. Because RS can change during storage conditions, we assessed the feasibility of preparing and storing the food items in the menu so the RS content would be sustained upon consumption. The second phase (second objectives) used a non-randomized study design to determine the feasibility of distributing the 7-day menu and the likeability of each food item in the menu in adults with prediabetes. Changes in body weight, BMI, and glycemic response after consuming the 7-day menu were assessed. The data from both phases will serve as pilot and feasibility data for a future 8-week, randomized, cross-over, clinical trial where the physiological outcomes of consuming the RS menu compared to habitual dietary intake will be determined.

Phase 1 Formulation of a 7-day menu with RS
Developing the menu Phase 1 includes the methods that address the primary objectives ( Figure 1). A 7-day high RS menu was developed using a database that quanti ed the amount of RS in foods commonly consumed by US adults [1]. In brief, the database was a compilation of 94 peer-reviewed research articles published between 1982 and 2018 that measured the amount of RS in commonly consumed foods in the US. The RS amount in each food in the database was reported with moisture included, not on a dry weight basis, which is how the food would be consumed. The high RS menu was formulated to have a weekly average of ~30g of RS/d, which is based on other clinical trials using RS as a supplement where improvements in glucose homeostasis and/or insulin sensitivity were observed [9,11,18].
The energy and nutrient content of the recipes and food items in the high RS menu were analyzed using the Nutrition Data System for Research (NDSR; version 2019, Nutrition Coordinating Center, University of Minnesota, Minneapolis, MN, USA). Recipes and food items were adjusted so the daily energy intake was 2,000 kcal and the macronutrients fell within the Acceptable Macronutrient Distribution Ranges (45-65% of energy from carbohydrate, 10-35% of energy from protein, and 20-35% of energy from fat) [19].
Preparing the food items Raw ingredients to prepare the high RS menu were purchased from a single local grocery store using the same brand names (when applicable) throughout the study period. All raw ingredients or products were purchased fresh and stored in the food lab to maintain quality and shelf life. The research team prepared all food items individually using recipes where each ingredient was weighed to the nearest gram. All food items were packaged in plastic food containers according to serving size weight, were labeled (e.g., food ID and date), and stored in a manner that increased RS while maintaining safe food handling practices based on industry recommendations [20].
The components each meal is shown in Table 1. The meals were prepared on Monday, Wednesday, and Friday which would be consumed by participants on Tuesday/Wednesday; Thursday/Friday; and Saturday/Sunday/Monday, consecutively.
Analyzing RS content in the menu RS in foods can be altered due to several factors such as cooking method, storage temperature and duration, and serving temperature [2]. There is a positive correlation between RS formation and duration of cold storage in some cooked, starchy foods [1]. However, in Phase 2 of this study the participants consumed each food item after a minimum of 1 day of storage. Hence RS was analyzed after the food items were stored for 1 day.
Each food item in the menu was prepared twice on separate occasions, and each were analyzed for RS in duplicate. Dry samples or samples with less than 25% moisture such as corn ake cereal, were nely milled for RS analysis. High moisture foods (≥25%) were removed after at least 1 day of storage at 4°C then prepared as would be consumed. For example, rolled oats in milk, plantains, potato salads, sandwiches, and all snack items would be consumed cold; therefore, these items were not heated prior to RS analyses. However, the bean and barley soup, lentil burger, and frozen meals were heated according to instructions and immediately prepared for RS analyses. The rationale for consuming certain items cold but heating other items was to maximize the RS amount in each food item while maintaining food safety and quality. The feasibility of developing, preparing, and storing the 7-day high RS menu was acceptable if the target weekly average of ~30 g RS/d was maintained after one day of storage.
The prepared food items were homogenized, then added to a glass ask lling 1/3 of the container. The asks were weighed in grams. The food samples were shell-frozen at -80°C for a minimum of 4 hours.
Once frozen, the samples were lyophilized for 48-72 hours depending on food sample type. Following lyophilization, the asks were weighed then moisture content was determined using the following formula: (wet weight (g) -dry weight (g) / wet weight (g))*100.
The dried food samples were nely milled in preparation for RS analysis. All the food samples were analyzed for RS in the prepared form that was "as eaten" according to the Resistant Starch Assay Kit (Rapid) instructions (Megazyme Ltd., Bray, Ireland). The RS measured using this kit employs the AOAC method 2002.02/AACC method 32-40.01 [21]. The calculations were completed using the Megazyme MegaCalc™ Excel® based calculator provided as supporting documents with the RS analysis kit.
After Initial RS analysis, the recipes and meals were once again slightly adjusted to ensure a weekly RS average of ~30g/d based on RS analysis data and the energy and macronutrient distribution were within the target ranges (Nutrient Database Software for Research, version 2019, Nutrition Coordinating Center, University of Minnesota, MN). Any ready to eat meals or foods that reported low RS was replaced with a food containing high RS (e.g., Chex cereal was replaced by corn akes). Once the menus were nalized, the RS analyzed in duplicates and mean values were reported as shown in Table 2 Phase 2: Feasibility assessment of menu distribution and evaluation, and changes in anthropometrics and glycemic response Feasibility of distributing the menu items The feasibility of distributing the menu to participants was assessed by preparation start and nish times. Feasibility would be accepted if 100% of the food items were prepared, packaged, and distributed to the participants on the same day (second objectives) as shown in the schedule under "Preparing the food items."

Participants
All methods associated with phase 2 addressed the second objectives ( Figure 1). First, adults with prediabetes evaluated each menu item (breakfast, lunch, snack, dinner) in the menu containing a weekly average of ~30 g RS/d ( rst objectives) using a non-randomized dietary intervention. Participants were recruited from the Houston, TX metroplex using yers, social media, and targeted emails. Interested participants contacted the investigators and were screened for inclusion criteria: adults with a self-report prediabetes diagnosis (e.g., fasting blood 100 -125 mg/dL, oral glucose tolerance test ≥ 140 -<200 mg/dL, or HgA1c 5.7% -6.4% mg/dL) [22] from a healthcare provider, between 20 and 65 years of age, able to speak and read English, a BMI ≥27 kg/m 2 , not following a speci c diet or have food allergies or intolerances to any of the study foods, able to safely prepare and consume all study foods. Females were excluded if they were pregnant or lactating. All participants provided written informed consent prior to data collection. This pilot study was approved by the Texas Woman's University, Houston, TX Institutional Review Board.

Food consumption and evaluation survey
Participants prepared and consumed the food items for each meal and were instructed to avoid consuming additional foods other than non-caloric beverages. Additionally, they received all pre-labeled surveys upon study enrollment. Each breakfast, lunch, snack, and dinner food items were evaluated each day, even if they were consumed more than once during the study. The surveys consisted of six questions that assessed the appearance, texture, smell, consistency or mouthfeel, taste, and overall likeability using a 9-point Likert scale where 1=dislike extremely and 9=like extremely. The survey also contained two open-ended questions regarding what was liked and disliked about the food item and how much they consumed (0%, 25%, 50%, 75%, or 100%). The benchmark established to indicate acceptability for this pilot study was if the participants consumed >75% [23] of the food items in the menu.
The completed surveys were returned to the investigators at the end of the study. To assess the mean response for the question "overall likeability" must be ≥7. Values at or above 7 indicate the food represents sensory characteristics that are acceptable for consumers [23].

Blood analyses
Blood was collected in vacutainers for glucose and insulin analysis. Blood was drawn at fasting (≥8 hours; water allowed) and 15, 30, 60, and 120 min after drinking 360 mL reduced-fat chocolate milk (210 kcal; 19.5g carbohydrate, 6.75g fat, and 6.75g protein) within 10 minutes at baseline (day before beginning RS rich menu evaluation) and the day after consuming the menu. The vacutainers were centrifuged at 4,000 rpm for 10 min and plasma was aliquoted in 1.5 mL tubes and frozen at -80°C until analysis. Glucose was determined using the colorimetric method (Stanbio Laboratory, Boerne, Texas, USA) and insulin by enzyme-linked immunoassay (Alpco, Salem, New Hampshire, USA) according to manufacturer instructions. All samples were analyzed in duplicate and repeated if the intra-coe cient of variation (CV) and inter-CV were ≥10% and 15%, respectively.

Anthropometrics
Height was measured at baseline using a stadiometer. Body weight (kg) was determined at baseline and the day after consuming the 7-day menu using the scale associated with the air-displacement plethysmograph machine (BOD POD, COSMED USA, Inc., Concord, California, USA). Body mass index (BMI; kg/m 2 ) was calculated at each time point. The participants wore the same clothing at both measurements.

Sample size
This pilot study rst aimed to evaluate the feasibility of preparing and distributing 7-day menu providing a weekly average of ~30g RS/d in order to investigate potential mechanisms of e cacy for this new intervention. The second objectives tested the likeability of a RS rich menu by human participants. Due to the lack of prior data needed to determine sample size, and the general guidelines for pilot food investigations a sample size of 12 is considered appropriate [24]. Considering this rule of thumb in conjunction with a medium standardized effect size of 0.3 ≤ δ < 0.7 for two time points and one-sided tail with a power of 80%, 15 subjects with prediabetes were considered appropriate to address the second objectives [25].

Statistical analysis
Data were analyzed using Statistical Package for Social Sciences (version 25, IBM SPSS, Armonk, New York, USA). Descriptive data are presented as mean ± SD. All data were screened for normality using Shapiro-Wilk tests and box plots. The tAUC (0-120 min) for glucose and insulin were calculated using the trapezoid formula. Wilcoxon signed rank tests were used to report mean differences and 95% CI for tAUC (0-120 min) glucose and insulin.

7-day menu
The high RS 7-day menu can be found in Table 1. The breakfast items consisted of oats, plantains, corn ake cereal, and tacos. Lunch and snack items included soups, sandwiches, and salads with foods containing RS such as potatoes, beans and legumes, pastas, and bread. The dinners consisted of frozen meals and the addition of other side items (e.g., whole wheat bread) if additional kcal were needed.
The energy, nutrient, and RS content of the meals are provided in Table 2. The daily mean caloric value of the meals was 2,082 ± 82 kcal where 56.4% of energy came from carbohydrates, 25.2% from fat, and 18.1% from protein. The mean amount of ber (44.1 ± 9.5/d) was above the recommended daily amounts of 38 g for men and 25 g for women between 18 and 50 years old and 30 g for men and 21 g for women > 50 years old [19].
The feasibility of maintaining a weekly average of ~30g RS/d The total RS varied where the lowest amount was provided on Monday (15.6 g) and the highest on Wednesday (39.7 g); however, the weekly average equated to 30.3 ± 7.9/d ( Table 2), which is the daily amount that has resulted in positive health bene ts. According to the RS quanti cation of each menu item after one day of storage, the results indicate that it was feasible to formulate and prepare the 7-day menu items to contain a weekly average of ~30 g RS/d.
Baseline participant characteristics are present in Table 3. The majority were female (73.3%), had a mean age of 37.9 ± 14 years and were obese (BMI 33.4 ± 7 kg/m 2 ). The most prevalent racial category was African American (40%) and Hispanic (33.3%). The duration of prediabetes was 34 ± 38.3 months and fasting glucose was the most frequently used method to diagnose prediabetes.

Feasibility of preparing, packaging, and distributing the menu items
The objective in preparing and distributing the high RS menu items was to ensure the food was prepared then stored at 4°C for a minimum of one day to increase RS content. Meal preparation began between 9:00 and 10:00 AM and ended no later than 3:00 PM on Monday, Wednesday, and Friday. We observed 100% compliance with this schedule. With regards to participant pick up and consumption pattern we observed 100% compliance with this schedule. Therefore, it is feasible to prepare the 7-day menu the same day as participant pick-up in the mid-afternoon. This food preparation and pick-up schedule will be implemented in the larger randomized, cross-over feeding trial for 8-weeks.
Results from the food evaluation survey can be found in Table 4. The frozen meal category had highest mean likeability for the question "Considering the appearance, aroma, taste, and texture, how much do you like the product OVERALL" followed by sandwiches, breakfast items, salads, then soups. Of all the food items, the lowest likeability scores included lima bean soup, succotash, and chickpea pasta salad, while the most liked food items included the steak portobello frozen meal, shrimp alfredo frozen meal, and hummus.

Common themes from two open-ended questions on the survey provided insight into the overall likeability
scores. The lima bean soup was disliked because of the cold temperature (n = 3), bland avor (n = 3), texture (n = 3), and appearance (n = 1). However, some participants thought the soup was refreshing (n = 2) and smelled good (n = 1). The bean succotash was disliked because of the overall texture (n = 1), gassy nature (n = 1), and addition of lima beans (n = 1) although some thought the dish was very avorful (n = 4). The chickpea pasta salad was disliked because of a "sour or weird avor," (n = 2) but in contrast was a great snack option (n = 2). For the top liked food items, the steak portobello frozen meal had a good sauce (n = 2) even though small portion size (n = 1) and broccoli and mushroom additions (n = 2) were disliked by some. The shrimp alfredo frozen meal had good ingredient combinations and good avor (n = 4); no dislike comments were made for this meal. The participants reported that the hummus was a good snack especially with the carrots (n = 5) even though it was grainy (n = 1).

Anthropometrics and biomarkers
Data on anthropometrics and biomarkers (n = 11) were collected at baseline and after consuming the 7day high RS menu. Interestingly, in such a short time frame both weight (mean -1.6 kg; 95% CI -2.6 to -0.7) and BMI (mean -0.5 kg/m 2 ; 95% CI -2.8 to 8) changed (data not shown). This suggests consuming the high RS menu may improve body weight in adults with prediabetes.
The biomarker results are found in Table 5. Three participants were excluded from the blood analyses. Two were outliers based on normality tests and one was due to the inability of obtaining postprandial blood. As expected, mean baseline fasting glucose met the criteria for prediabetes diagnosis, further validating that the participants indeed had prediabetes. Following the 7-day menu, glucose decreased at all time points with the largest reductions at 30 and 60 minutes. Insulin also decreased at all time points except at 15 minutes. Similar to glucose, the largest reductions were observed at 30 and 60 minutes.
These results indicate RS can improve glucose homeostasis in a short period of time.

Discussion
The main ndings of this non-randomized pilot study include the following: 1) it is feasible to formulate, prepare, and distribute a 7-day menu with a weekly average of ~30 g RS/d and 2) the food items in the menu had appropriate sensory characteristics that led to high adherence and acceptability in adults with prediabetes measured by the overall percentage of food items consumed (>75%) and a mean overall likeability score ≥7. In addition, following the short 7-day intervention, glucose, insulin, and body weight were reduced although this pilot study did not test for signi cance. Resistant starch may not be solely responsible for these changes as a reduction in energy intake and increased dietary ber likely contributed. These pilot data established preliminary results for future larger randomized, cross-over feeding trials sampling a larger number of participants for a longer duration. Appropriately powered interventions including menus providing ~30 g RS/d through whole foods could lead to changes in dietary recommendations that attenuate, or even prevent, the progression of prediabetes to type 2 diabetes.

High resistant starch (RS) menu
Although several caveats in developing high-RS menus exist, we were able to use a published database [1] to develop recipes that contained ~2,000 kcal/d, recommended macronutrient ranges, and a weekly average of ~30g RS/d. The caveats are associated with the many factors that may in uence RS content: plant breed, growing and storage conditions, preparation, storage, and consumption temperature [2]. These challenges have steered researchers away from whole-food RS dietary interventions; therefore, using RS supplements instead. However, we have now shown it is feasible to develop a high RS menu and maintain health-bene ting RS amounts in foods. To develop the menu, foods from a database [1] such as potatoes, lentils, white and lima beans, chickpeas, plantains, pasta, and barley were utilized as they provided the highest amount of RS per gram serving size. This pilot study is unique as it ensured the RS contribution of each menu item "as consumed" from the RS analyses in our lab. To the best of our knowledge, we are not aware of any studies estimating the naturally occurring RS content in test meals after production.
We have previously reported that the usual intake of RS in US adults is 1.9 g/1,000 kcal [8], and 2.1 g/1,000 kcal in adults with prediabetes [26], which is <15% of the RS contained in our menu. The most frequent contributors of RS in the US diet were French fries and other fried white potatoes, rice, and beans, legumes, and peas [8]. Our high-RS menu replaced the fried potatoes and white rice with more nutrient-dense options that more closely align with the Dietary Guidelines for Americans 2020-2025 [27].
This is the rst pilot study evaluating the adherence and likeability of a full menu high in RS in adults with prediabetes. Over ¾ of the high-RS menu items were consumed where frozen meals had the highest adherence and salads and soups had the lowest. Adherence to the frozen meals could be due to familiarity as they are commonly consumed foods in the US. The steak portabella and shrimp alfredo received the highest likeability score where a " avorful sauce" was mentioned. Of the salads, the chickpea pasta and succotash were consumed the least and had the lowest likeability scores. Some participants stated the chickpea pasta had a "weird avor" and "sour taste" indicating the seasoning will need to be adjusted for future studies. The "overall texture" and "lima beans" were comments indicating dislike for the succotash. The lima bean soup had the lowest adherence and likeability due to "cold consumption, bland, texture, appearance" mentions. Future randomized crossover trials will modify this recipe by pureeing the soup thoroughly, adding chives and paprika for color, and adding crackers or croutons for added texture.
Adherence to the 7-day high RS menu was higher than other dietary interventions of similar populations.
One study in adults with type 2 diabetes found 67% met adherence criteria to a vegan diet compared to only 44% in the American Diabetes Association group at 22 weeks with no difference diet likeability, or satisfactory rating, between groups [28]. The vegan diet included many high RS foods, such as legumes, beans, and peas, potatoes, bread, and whole grains. The higher adherence in this pilot study could be due to differences in study duration where adherence rates typically decline over time. Obtaining food preferences (e.g., dislike salmon but like tuna) will be necessary to maintain adherence in longer feeding trials. The seasonings could be adjusted in some recipes (e.g., replace lemon and herbs with curry in potato salad) to reduce monotony without altering the RS amount.

Body weight
Following the consumption of the 7-day menu, a -1.6 kg and -0.5 kg/m 2 change in body weight and BMI, respectively, occurred. Several mechanisms may be associated with weight loss. First, the net energy content of RS is less than digestible starch (2.59 kcal/g vs 4 kcal/g) thus reducing the caloric contribution from the diet [29]. Next, increased fat and decreased carbohydrate oxidation have been observed following RS intake [30]. Lastly, RS fermentation by speci c gut microbiome taxa produce metabolites that lead to decreased adipogenesis, fat accumulation, and release of satiety-promoting hormones such as glucagon-like peptide and peptide YY [31] [32]. We did not measure satiety in the present pilot study; however, participants (n=2) stated "could not eat, too full" in the open-ended questions. Fiber also contributes to slower digestion rates and feelings of fullness and could work synergistically with RS to reduce body weight [33]. The 7-day menu provided 44.1 ± 9.5 of dietary ber per day, which is almost 2.5 times higher than normal ber intake among adults with prediabetes [34]. In addition, a reduction in total daily energy intake cannot be ruled out since the high RS menu provided 2,000 kcal/d, which may be lower than what the participants habitually consume. It is unclear if weight loss would be sustained since our intervention only lasted 7 days. One study found that high RS foods, speci cally dried beans and high-ber low-fat grains contributing to a high-carbohydrate and ber diet has been shown to predict the most weight loss at 1y in adults at high risk of type 2 diabetes [35] suggesting the high RS menu could have lasting health bene ts. This pilot study assessed the feasibility of preparing and delivering the 7-day menu; therefore, the impact of RS on body weight and glycemia will be examined in future research.
Changes in glucose and insulin may be associated with several mechanisms. Metabolites resulting from RS fermentation have been shown to improve insulin resistance and de novo glucose production through several signaling pathways [31]. In addition, meta-analyses have found that foods with higher amylose, retrograded starch, and starches with less ability to form gels improve postprandial glycemia These studies demonstrate the effect of RS structurally and functionally on glycemic response; however, are limited in addressing the gap in translating the scienti c rigor to real world practice. The majority of studies are based on the manipulation of RS particles by either storage, supplementation, or genetically altering amylose content. In comparison our study has utilized naturally occurring RS rich foods commonly available to US consumers. In addition, by providing all meals for 7 days, the likelihood of consuming external foods that could alter glycemia was small, even though these data were not collected. Results from this 7-day pilot test intervention are comparable with studies that observed RS effects on glycemic response as we observed similar tAUC (0-120 min) glucose and insulin reductions. It is also important to note that our study was conducted on adults with prediabetes with obesity and hence providing a unique insight to potential changes in biochemical responses after consuming a RS rich diet.
This is vital because a prospective study assessing dietary intake every 4 years found that adults consuming a high total starch and starch: low-ber diet, after adjusting for several dietary and lifestyle factors, had higher risk for developing type 2 diabetes [44]. These results can be misleading since the starchy foods are likely from highly re ned, low ber sources -many of which are also low in RS. In this pilot study, using a similar population, starchy foods rich in both RS and ber elicited positive health outcomes.

Limitations And Future Research
Some limitations of this non-randomized pilot and feasibility study should be noted. The sample size was small thus the level of signi cance for post RS meal consumption data was not assessed. The primary purpose of the pilot study was to develop a high RS menu and determine the likeability and acceptability in adults at risk for developing type 2 diabetes. Assessing changes in glycemic response and anthropometrics were not the primary goal of this study. However, the decreases observed in this study will help us determine the sample size for randomized trials that would improve glycemic control. We were not able to identify the mechanistic action of the test meals with high RS and its association with the metabolic outcomes, as well. For example, it is unknown if the RS, dietary ber, or reduced energy contributed more to the weight and glycemia changes or if they worked synergistically together. To reduce participant burden, we did not collect information on external foods consumed during the 7-day trial.
Based on these results, adjusting some of the menu items to improve taste, texture, and appearance will be necessary to improve adherence. A longer feeding trial with additional biomarker analysis such as satiety hormones will be bene cial to identify longer-term metabolic impacts. Fecal samples to test changes in microbiome structure and function would provide additional insight to the mechanisms associated with consuming a high RS diet provided through foods.

Conclusions
Developing, preparing, and distributing a menu with the same amount of RS known to improve glycemic health (~30 g/d) is feasible and liked overall among adults with prediabetes. Body weight and BMI decreased, as well as fasting, postprandial, and tAUC (0−120 min) glucose and insulin following the 7-day