Acute and second-meal effects of almond form in impaired glucose tolerant adults: a randomized crossover trial

The 2025 worldwide projection of IGT is 418 million (8.1% of the adult population) [1]. Lifestyle modification, including nutrition is the cornerstone of its management. Macro- and micronutrients, fiber content, and other components of the diet modulate meal-induced insulin secretion through changes in gastrointestinal transit time and nutrient absorption rates. Additionally, the content of one meal has the potential to affect insulin sensitivity at a second meal by altering circulating NEFA concentrations and daylong insulin demands [2].

Almonds are a low-glycemic index (GI) food, with high fiber, unsaturated fat and low carbohydrate content. There is an inverse relationship between nut consumption and risk of developing type 2 diabetes [3]. In addition, almond consumption increases satiety, reduces cardiovascular disease risk, decreases postprandial glycemia and moderates oxidative damage [4]. The component(s) of almonds responsible for these effects have not been determined. Almonds contain phytates and phenolics, that confer antioxidant, anti-inflammatory and lipid-lowering properties and inhibit trypsin and amylase activity [5]. A decreased rate of nutrient digestion may explain reported increases in satiety and blunted blood glucose response with almond consumption. Stimulation of the incretin and ileal-brake hormone, GLP-1, may also contribute. Consequently, almond consumption may be an effective dietary management tool in insulin resistant individuals who would benefit from replacement of saturated fat with unsaturated fat [6].

Inclusion of 60 g of almonds in meals of healthy individuals decreases glycemia, insulinemia, and postprandial oxidative damage as measured by increased protein thiol concentration [7]. Adding almonds (30-90 g) to a high-GI meal results in a dose-responsive decrease in 2-hour postprandial blood glucose AUCI [8]. However, consumption of almond oil with defatted almond flour, to mimic a bioaccessible almond form, significantly decreased 3-hour blood glucose AUCI with no difference in insulin response compared to when small almond particles were consumed [9]. Similarly, increased and sustained concentrations of cholecystokinin (CCK) and augmented hunger were reported with bioavailable almond oil compared to whole almonds [10]. This suggests the bioavailability of the lipid fraction may be responsible for decreased postprandial glycemia.

The present study evaluated the effects of whole almonds, almond oil, defatted almond flour, and almond butter on acute and second-meal postprandial blood glucose, insulin, NEFA, and GLP-1 concentrations, as well as satiety sensations, in IGT adults.

Eligibility criteria included: age 18-60 years; not taking medications known to affect glycemia, sleep, or appetite; weight stable (3 month fluctuation of <5 kg); regular breakfast consumer (≥100 kcal ingested within 2 hours of waking on ≥4 d/wk) o blood donation in the previous 3 months; no nut or relevant food allergy; at least one of the following risk factors: A) self-reported family history of type 2 diabetes; B) blood pressure ≥130/85 mmHg; C) fasting blood glucose >5.6 mmol/l; or D) waist circumference (men ≥102 cm; women ≥88 cm); and a 2-hour blood glucose value of 7.8 and <11.1 mmol/l (i.e., IGT) [11]. Height, weight, and body composition were measured using a wall-mounted stadiometer, a clinical scale, and bioelectrical impedence, respectively. A 2-hour, 75-gram oral glucose tolerance test (OGTT) was conducted at a second visit with participants in an 8-10 hour fasted state. The research was approved by the University Institutional Review Board.

One hundred-seventy individuals completed the first screening visit, of which 132 were eligible for and completed the second screening visit. Fourteen participants met all screening criteria and completed the full study protocol. Calculation of power indicated that 13 individuals were necessary to detect a change in blood glucose of 0.35 mmol/l (α = 0.05; Power = 0.80, SD = 0.3)[12]. Participant characteristics are shown in Table 1.

The study utilized a randomized, 5-arm, crossover, single-blinded design. Overnight fasted (8-10 hours) participants reported to the laboratory on 5 occasions separated by at least one week. Menstruating female participants completed test days within the follicular phase of their menstrual cycle. Individuals were requested to maintain their normal exercise, eating, and sleeping patterns and, if taking vitamins or medications, to take them at the same time every day before reporting for testing. Participants were also requested to consume the same meal each evening before reporting to the laboratory at their customary breakfast time.

Upon arrival to the laboratory, participants were weighed and body composition was determined. An indwelling catheter was placed and a baseline blood sample collected. Appetite ratings were scored on a 100 mm visual analogue scale (VAS) anchored with descriptors of "not at all" and "extremely." Next, the participant consumed a test breakfast within 10 minutes that consisted of orange juice and farina [prepared Cream of Wheat^®, B&G Foods, Inc.] containing: nothing V (vehicle), whole almonds (WA), almond butter (AB), defatted almond flour (AF) or almond oil (AO) in randomized order. Almonds were non-pareil, dry-roasted and added to the farina whole [provided by the Almond Board of California (Modesto, CA)]. Almonds and their processed forms were from the same almond harvest. Test breakfasts were matched on available carbohydrate and sweetness (nutrient composition shown in Table 2). The amount of almonds added to the cereal was 42.5 grams (~33 almonds) in accord with the Food and Drug Administration (FDA) qualified health claim regarding daily nut intake [13]. After completion of the breakfast meal, palatability of the foods was rated on a VAS (mean palatability scores are shown in Table 3).

Blood was drawn and appetite was rated 15, 45, 60, 90, 120, 180, and 240 minutes after completion of the test breakfast. At minute 240, participants consumed a standard lunch within 10 minutes that consisted of a plain white bagel, grape or strawberry jelly, and tap water (250 ml). Palatability of the lunch was rated on a VAS. Afternoon blood sampling and appetite scoring occurred using the same time intervals as the morning.

Three milliliters of blood were collected in a red top vacutainer at each draw. After clotting and centrifugation, serum was aliquoted and stored at -80°C for later analysis of insulin, glucose, and NEFA. Four ml of blood were collected in ice-cooled EDTA-plasma vacutainer, 40 μl DPP-IV inhibiter (Millipore, St. Charles, MO) was added, and samples were handled according to manufacturer's directions for analysis of GLP-1. All samples were analyzed in duplicate. Insulin and glucose were measured by electrochemiluminescence and the hexokinase method, respectively. Sensitivity of the insulin immunoassay was 1.39 pmol/l (within-run coefficient of variation (CV) of 1.9%). Hexokinase sensitivity was 0.12 mmol/l (within-run CV of 0.41%). NEFA were analyzed with an enzymatic colorimetric method on an automated analyzer with a sensitivity of 0.00014 mEq/L (within-run CV of 0.75%). GLP-1 was assessed by radioimmunoassay. Sensitivity of the assay was 3 pmol/l (within-run CV of 30.3%). Lipid panel assessment was conducted by MidAmerica Clinical Laboratories, Indianapolis, IN.

Nutrient data were analyzed with the Nutrition Data Systems for Research 2008 (University of Minnesota, Minneapolis, MN). Statistical testing was conducted with SPSS, Version 17.0 (SPSS Inc., Chicago, IL). Repeated measures analysis of variance (ANOVA) was used to explore main effects and, when appropriate, post hoc analyses were conducted with Bonferroni adjustment. Significance was set at p < 0.05. Data are represented as Mean ± SEM. Area under the Curve (AUC) with respect to increase (I) was computed to measure concentration change over time [14] (Formula 1) and quantitative insulin sensitivity check index (QUICKI) was calculated [15].

Fasting concentrations served as baseline for daylong (0-490 minutes) and morning responses (time 0-240 minutes) and the blood sample taken at 240 minutes served as the baseline for afternoon responses (time 240-490 minutes).

Formula 1. Incremental Area Under the Curve

Where n equals the total amount of measurements, m_i equals the individual measurements, and t_i equals the time between measurements [14].

The aims of this study included confirmation that nut consumption improves the metabolic profile with respect to diabetes risk; determination of the relative contributions of different almond fractions on these indices and whether acute post-prandial benefits translate to improved insulin sensitivity at a subsequent eating event (second-meal effect). To enhance the ecological validity of the work, whole almonds were included to explore effects with natural mastication and the quantity of almonds included in the test meal corresponded to the recommended intake level in the FDA approved qualified health claim for nuts [13].

In the current study, significantly greater fasting insulin concentrations with WA and AO led to lower QUICKI values and therefore less calculated insulin sensitivity. Due to its correlation with the hyperinsulinemic euglycemic clamp [15], QUICKI is the preferred method for quantifying insulin sensitivity in populations with perturbed insulin sensitivity. Despite higher baseline insulin concentrations, consumption of WA and AO decreased morning blood glucose AUCI compared to V. Postprandial breakfast insulin and NEFA AUCI were not greater after consumption of WA and AO suggestive of greater insulin sensitivity (e.g. the decreased blood glucose AUCI was not determined by a concurrent increased insulin response). Similarly, consumption of 60 g of almonds with white bread decreased 2-hour blood glucose and insulin AUCI in healthy individuals compared to a control meal [7]. Moreover, in healthy men, bioaccessible almond composition produced a lower 3-hour blood glucose response with no significant difference in the insulin or NEFA response [9]. Larger almond particles did not produce the same effect. Previous data from our laboratory did not find a clear relationship between amount of almond chewing (predefined number of chews) and changes in glucose and insulin concentrations in a group of healthy participants [16]. Discrepancies between studies may be due to differences in almond particle sizes (e.g., naturally masticated versus predefined) which could alter nutrient bioaccessibility.

In contrast to lipid-containing treatments, the treatments with little fat (V and AF) produced the largest immediate postprandial glucose responses. The role of fat in decelerating gastric emptying may be partly responsible [17]. Although the AF treatment contained polyphenolic compounds, there was no evidence of impairment of starch digestion in the current study as has been previously reported [18]. NEFA concentrations after consumption of AF were lower than V in the morning postprandial period without differences in insulin concentrations, indicating a slight improvement in NEFA suppression. In comparison, no difference in NEFA concentrations between the combination of AF and AO, large almond particles, and control sunflower oil [9] suggests minimal benefit to the presence of the defatted flour fraction on metabolic risk outcomes.

One study suggested the NEFA concentration 4 hours after a test breakfast accounted for ~50% of the variability in the glycemic response to a standard lunch [19]. We found no significant difference at this time point and do not confirm that NEFA concentrations explain second-meal metabolic differences. However, AB resulted in the lowest overall degree of NEFA suppression in the morning period and was associated with the greatest blood glucose response to the standard lunch. The overall NEFA response in the period before the meal may be a stronger determinant of the second-meal glycemic response than the concentration immediately preceding the second meal. Additionally, no differences were observed in glucose, insulin, or GLP-1 concentrations at 4 hours after the test breakfast, suggesting other determinants of second-meal effects. While the mechanism remains uncertain, this trial confirms the phenomenon. Prior work revealed that inclusion of slowly digestible carbohydrate (e.g., lentils) in a breakfast meal reduced the glucose response after lunch [20]. We show that inclusion of a high unsaturated fat load with breakfast is also effective. Together, these data support the efficacy of dietary moderation of carbohydrate absorption kinetics from a morning meal for extended glycemic control in populations at risk for or with type 2 diabetes.

The high unsaturated fatty acid composition of almonds may account for the blunted glucose concentrations noted in the postprandial period. Acute consumption of PUFA and MUFA decreases postprandial glucose AUCI without altering insulin concentrations [21] due to increased efficiency of insulin receptor signaling and increased glycemic control through stimulation of GLP-1 [6]. Although no significant treatment effects were detected in GLP-1 concentrations, WA and AO led to an overall greater and sustained GLP-1 response that may have contributed to blunted second-meal blood glucose concentrations [22] and modified satiety [23].

The role of almond lipid bioavailability in triggering the release of gut peptides and contributing to energy balance is complicated by differences in the metabolic profiles following WA and AO versus AB consumption. Previous research in healthy participants showed lower breakfast and increased afternoon blood glucose AUCI after consumption of a standard lunch when peanut butter or butter was consumed in a mixed breakfast meal [24]. Our data show this similar afternoon rebound with consumption of AB in the breakfast meal, which cannot be attributed

solely to the lipid component. Additionally, in vitro gastric and duodenal digestion modeling found greater duodenal lipid digestion in finely ground almond particles compared to defatted finely ground almonds with almond oil added back, suggesting that differential dispersion of the lipid (e.g. different surface areas of the lipid droplets) may determine bioaccessibility [25]. Altering the physical form of nuts may have unexpected metabolic effects that warrant further investigation.

Differences in fullness were not likely due to variations in the macronutrient content of the test foods. All provided 75 g of available carbohydrate and the treatments matched on protein, fat, total and soluble fiber, and energy led to variable satiety responses. The cognitive influence of the visual cue of WA and the increased orosensory stimulation from chewing may be responsible for satiety effects [26]. Lipid consumed in oil form previously induced a greater and sustained CCK response and greater satiety in women versus consumption of WA [10], a finding not confirmed in the current study.

An unavoidable limitation of the current study was that breakfast meals were not matched on energy, volume, or macronutrient composition. Due to the study design, available carbohydrate (the main determinant of GI) was standardized between all treatments and macronutrients were matched when possible. The subjective palatability of the treatments was not consistent, with AO considered significantly less desirable than all other treatments except for AF. However, post hoc covariate analysis did not reveal palatability significantly altered results. Additionally, participants were instructed to consume the same meal before reporting for each visit, although significant differences were found. Fewer calories were consumed the night before WA compared to V (~200 kcal) and a lower percent of calories from carbohydrate was consumed the night before AO compared to AB (~5%). The extent to which these differences can explain postprandial breakfast responses is unknown. Greatly altering the GI of a dinner may produce a varied response after a breakfast meal, but the absorptive characteristics between the evening meals consumed before test days in the current study were unlikely so drastically different as those that have previously been shown to produce carry-over effects to the breakfast meal [27]. The possibility exists that differences in dietary intake may also be an artifact of the difficulty of accurately assessing dietary intake and single meal reporting precludes the ability to employ calculations such as the Goldberg cut-off to determine plausibility of reported intake. Nonetheless, the macronutrient and energy intake data appear to fall within normal ranges (49-54% energy from carbohydrate, 28-33% energy from fat, 18-19% energy from protein, and ~30% of mean estimated daily energy requirements).

In summary, inclusion of almonds in the breakfast meal of IGT adults decreased blood glucose concentrations and increased satiety acutely and after a second meal. The lipid component of the almond appears to be largely responsible for the immediate post-ingestive response, although it cannot account for the second-meal response. Overall, daylong glucose, insulin and NEFA concentrations were attenuated in the WA and AO treatments, indicating an improved hormonal profile with their consumption. Importantly, the absolute magnitude of the blood glucose-lowering response equals that achieved with acute administration of acarbose in individuals with IGT [28], suggesting the physiological relevance and applicability of the current findings.