We conducted a randomized prospective parallel-group clinical trial at the Loma Linda University Medical Center Diabetes Treatment Center, Loma Linda, California. The study duration for each participant was 24 weeks. The present study was designed to test the hypothesis that a peanut-enriched ADA meal plan would be more effective than a nut-free ADA meal plan on improving the nutrient profile of the total diet, blood lipids and glycemic control in adults with T2D.
Eligibility criteria
Adults with a medical diagnosis of T2D for at least 6 months and HbA1c less than 9.0% were recruited through advertisements on the Loma Linda University campus and surrounding communities. Individuals less than 20 years of age, that smoked, had nut allergies or a history of irritable bowel disease or diverticulitis that could be exacerbated by daily peanut intake, were excluded. Potential subjects that were habitual peanut or tree nut consumers must have been willing to discontinue the intake of all peanut and/or tree nuts for 6 weeks prior to their first scheduled clinic visit. Patients with liver disease, renal disease and/or severe dyslipidemia (TG >4.52 mmol/l or TC >7.77 mmol/l) were also excluded. Use of long-acting insulin and statins were permitted if the dose was stable for at least 3 months. Sixty subjects were enrolled and 57 completed the study. This study was approved by the Loma Linda University Institutional Review Board and has therefore been performed in accordance with the ethical standards laid down in the 1964 Declaration of Helsinki and its later amendments. Written informed consent was obtained from all study subjects.
Study protocol
During the week 0 visit, the daily resting energy expenditure (REE) for each participant was computed using the Harris Benedict equation. For participants with a BMI <25 kg/m2 (10% of participants), an individualized ADA meal plan was developed according to the REE results. An activity factor of 1.3 was utilized for all participants. Participants were prescribed a milk-free meal pattern if they stated they were lactose intolerant, and their traditional number of meals and snacks were honored to maximize dietary adherence. For participants with a BMI > 25 kg/m2 (90% of participants), the REE for each participant was computed using the Harris-Benedict equation after adjustment of body weight for overweight status, and they were prescribed energy intake deficits of 500 kcal in accordance with the ADA’s guidelines to facilitate modest weight loss [8]. Daily energy levels prescribed ranged between 1000 kcal to 2400 kcal.
Participants were randomly allocated to consume either an ADA meal plan with ~20% of energy from peanuts and to avoid other tree nuts (peanut group), or to consume an ADA meal plan without peanuts and tree nuts (control group). The amount of peanuts was determined based on previously published data reporting favorable changes in glucose and blood lipid levels in subjects with impaired glucose tolerance consuming a diet containing 20% of energy from another MUFA-rich nut [9]. The prescribed ADA meal plans in this study contained 35% total fat (15% MUFA), 45% carbohydrate and 20% protein.
At week 0, each participant met with the study dietitian for a 1-hour counseling session to receive their individualized ADA meal plan. The peanut group participants received dietary instruction on how to select 80% of their remaining energy needs using the ADA Food Exchange System. A supply of commercially available peanuts and/or peanut butter was provided to participants assigned to the peanut group at clinic visits. Participants were allowed to determine if they preferred pre-packaged single-serving peanuts (salted) only, pre-portioned peanut butter (with added salt and oil) only, or a combination of both to maximize dietary adherence. The peanuts were consumed as part of the participant’s customary meals and snacks and were the primary food source of MUFA (40% by energy and 52% total fat by weight) in the peanut group. Due to the 20% energy contribution from the peanuts, control group participants were advised to consume compensatory servings from the meat/meat substitutes (i.e. beef, fish, eggs, cheese, plant-based proteins) and fat (i.e. butter, margarine, mayonnaise, avocado, oil) exchange lists. Both groups were prescribed an equivalent number of carbohydrate (e.g. milk, fruit, bread/cereal) and vegetable exchanges. However, a 1000 kcal ADA meal plan would contain one additional meat/meat substitute exchange and three additional fat exchanges.
Measurement of nutrient profile of the total diet and cardiometabolic parameters
To ensure dietary adherence and to assess the nutrient profile of the total diet during the study, six unannounced 24-hour recalls (4 weekdays and 2 weekend days) were conducted by phone, one approximately every 4 weeks. In addition, the dietitian reviewed the prescribed number of ADA food exchanges and provided reinforcement at follow-up clinic visits. The six 24-hour recalls were analyzed using Nutrition Data System for Research software version 2006, developed by the Nutrition Coordinating Center, University of Minnesota, Minneapolis, MN, USA.
Cardiometabolic outcomes included weight, BMI, WC, blood lipids (TC, LDL-C, HDL-C, TG), blood lipid ratios (TC:HDL-C, LDL-C:HDL-C), FBG and HbA1c. Height was measured to the nearest centimeter using a stadiometer at week 0. Weight was measured using an internally calibrated segmental body composition scale/analyzer (model TBF-300A, Tanita®, Arlington Heights, IL, USA). BMI was calculated as weight(kg)/height(m2). WC was measured to the nearest 0.1 cm, midway between the last rib and the ileac crest.
Venous blood samples were collected at Quest Diagnostics Patient Care Centers after a 12 hour overnight fast at weeks 0, 12 and 24, and all testing was performed at Quest Diagnostics Laboratory, West Hills, CA, USA. Spectrophotometry [10] was used to determine serum glucose, TC, HDL-C, TG and LDL-C (with immunoseparation [11]), and HbA1c was measured using immunoturbidimetry [12].
Statistical methods
Sample size, power calculations, simple randomization scheme and statistical analysis were performed utilizing SAS version 9.2 (SAS Institute, Cary, NC, USA). The primary outcome measure for performing the power calculation was HDL-C. Using a mean difference of 0.20 mmol/L and SD of 0.24 mmol/L that was obtained from a walnut intervention study conducted in adults with type 2 diabetes, we had 80% power testing at an alpha of 0.05 to detect a difference of at least a 15% change in HDL-C with 46 subjects [13]. Bivariate statistical analysis using the chi-square test for differences in proportions and two-sided independent t-tests were performed on baseline characteristics using a probability value of 0.05.
An intent-to-treat analysis was performed and all percent change values presented are calculated from least-squares means estimated from mixed models. Week 0, 12 and 24 measurements were included in the analysis, with the exception of weight, BMI, and WC that included additional measurements from weeks 4, 8, 16 and 20. For each dependent variable the most appropriate covariance structure was chosen using likelihood ratio tests, and an unstructured, compound symmetry, heterogeneous compound symmetry, autoregressive, or heterogeneous autoregressive covariance structure was applied. The assumption used in the intent-to-treat model with regard to missing data and unmeasured end points for the dropouts was that they were missing at random.
To assess the significance of changes in the anthropometric and metabolic variables, a mixed-model repeated-measures analysis of covariance was used with diet, week, and diet × week interaction as fixed effects, adjusting for baseline measurements of the outcome variable. The mixed model included the treatment effect (control vs. peanut diet: Diet), time effect as a categorical variable (Time), and the interaction term between the two (Diet × Time). Change in weight from baseline (Δ Weight, in kg) was added to the model in light of the influence of weight change on WC and biological measurements. Stratified analysis were conducted for age (≤55y and >55y), gender, baseline BMI (≤30 kg/m2 and >30 kg/m2) and statin use (taking a statin and no statin use) to assess for potential 3-way interactions with the treatment and time effects. A natural log transformation was performed on outcome variables for the modeling analysis when indicated to improve normality. A histogram and residual plots were used to verify normality after the transformation. The Kenward-Roger method was employed to estimate denominator degrees of freedom for tests of fixed effects and Tukey-Kramer Honestly Significantly Different tests were performed to detect significant pair-wise differences among the two treatments.