Original Research
Almond Consumption and Cardiovascular Risk Factors in
Adults with Prediabetes
Michelle Wien, DrPH, David Bleich, MD, Maya Raghuwanshi, MD, Susan Gould-Forgerite, PhD, Jacqueline Gomes, MBA,
Lynn Monahan-Couch, MPH, Keiji Oda, MPH
University of Medicine and Dentistry of New Jersey, Newark, New Jersey (M.W., D.B., M.R., S.G.-F., J.G.), West Chester University,
West Chester, Pennsylvania (L.M.-C.), Loma Linda University, Loma Linda, California (K.O.)
Objective: The authors tested the hypothesis that in adults with prediabetes, an almond-enriched American
Diabetes Association (ADA) diet improves measures of insulin sensitivity and other cardiovascular risk factors
compared with an ADA nut-free diet.
Methods: Design: Randomized parallel-group trial. Setting: Outpatient dietary counseling and blood
analysis. Subjects: Sixty-five adult participants with prediabetes. Intervention: Sixteen weeks of dietary
modification featuring an ADA diet containing 20% of energy from almonds (approximately 2 oz per day).
Measures of Outcome: Outcomes included fasting glucose, insulin, total cholesterol (TC), low-density
lipoprotein cholesterol (LDL-C), high-density lipoprotein cholesterol (HDL-C), triglycerides, TC:HDL-C, and
HbA1c, which were measured at weeks 0, 8, and 16. Body weight, body mass index (BMI), waist circumference,
blood pressure, and nutrient intake were measured at weeks 0, 4, 8, 12, and 16.
Results: The almond-enriched intervention group exhibited greater reductions in insulin (21.78 mU/ml vs.
+1.47 mU/ml, p 5 0.002), homeostasis model analysis for insulin resistance (20.48 vs. +0.30, p 5 0.007), and
homeostasis model analysis for beta-cell function (213.2 vs. +22.3, p 5 0.001) compared with the nut-free
control group. Clinically significant declines in LDL-C were found in the almond-enriched intervention group
(212.4 mg/dl vs. 20.4 mg/dl) as compared with the nut-free control group. No changes were observed in BMI
(20.4 vs. 20.7 kg/m2, p 5 0.191), systolic blood pressure (24.4 mm Hg vs. 23.5 mm Hg, p 5 0.773), or for
the other measured cardiovascular risk factors.
Conclusions: An ADA diet consisting of 20% of calories as almonds over a 16-week period is effective in
improving markers of insulin sensitivity and yields clinically significant improvements in LDL-C in adults with
prediabetes.
INTRODUCTION
Prediabetes currently affects up to 13% of the U.S. adult
population and 16% of U.S. teens [1], which translates to 54
million individuals [2]. Prediabetes is a precursor to type 2
diabetes mellitus (T2DM), which is present in approximately
8% of the U.S. population. Fifty percent of patients with
prediabetes will progress to T2DM unless there is aggressive
intervention (i.e., diet modification and exercise to facilitate
moderate weight loss or use of medications that improve
glucotoxicity without elevating endogenous insulin) [3].
The American Diabetes Association (ADA) guidelines
address the need for medical nutrition therapy and public
health interventions for patients with prediabetes to decrease
the risk of developing T2DM and cardiovascular disease
(CVD) [4]. Population-based prospective cohort studies [5,6]
have demonstrated that lifestyle modification including diet
can reduce the progression of prediabetes to T2DM, which
has stimulated enthusiasm for evaluating novel nutrition
approaches among patients with prediabetes. Such approach-
es have the potential to also improve the cluster of
diabetogenic and atherogenic abnormalities including insulin
resistance, dyslipidemia, and hypertension. Thus, health care
professionals are seeking feasible and innovative patient-
oriented strategies in the context of beneficial nutritional
therapies.
Address reprint requests to: Michelle Wien, DrPH, Loma Linda University, Department of Nutrition, School of Public Health, 24951 N. Circle Drive, Nichol Hall 1107,
Loma Linda, CA 92350. E-mail: mwien@llu.edu
A subset of the lipid data was presented at a poster session at the 2008 Experimental Biology Meeting in San Diego, California, on April 6, 2008. This research was
supported in part by a grant from the Almond Board of California.
Journal of the American College of Nutrition, Vol. 29, No. 3, 189–197 (2010)
Published by the American College of Nutrition
189
Almonds contain high levels of fiber, arginine, magnesium,
polyphenolic compounds, vitamin E, and monounsaturated
fatty acids (MUFA), specifically oleic acid. Population-based
prospective cohort studies have shown an association between
frequent nut consumption and reduced risk of T2DM and CVD
[7–9]. These findings have generated proposed mechanisms for
these associations including improved insulin sensitivity,
increased antioxidant activity, and reduced concentrations of
total cholesterol (TC) and low-density lipoprotein cholesterol
(LDL-C). Prior in vivo studies have shown that MUFA
enhances the intestinal secretion of glucagon-like peptide-1
(GLP-1) [10–12], an incretin hormone that improves the
regulation of postprandial glucose disposal and insulin
secretion [13]. In addition, Wang et al. [14] have shown that
lipid infusions containing primarily polyunsaturated fats
(PUFA) trigger a gut-brain-liver axis that increases insulin
sensitivity in the liver. Intermediary roles for gut hormones,
including incretins, remain possible in this circuit and offer a
conceptual basis for antidiabetes diets.
We sought to explore a possible role for daily almond
consumption in persons with prediabetes as a simple, cost-
effective nutrition intervention to preserve beta-cell function
and improve insulin sensitivity and other CVD risk factors. We
therefore conducted a randomized trial of an almond-enriched
diet in the context of ADA diet guidelines to test the
hypothesis that in individuals with prediabetes, an almond-
enriched ADA diet improves measures of insulin sensitivity
and other CVD risk factors compared with an ADA nut-free
diet.
MATERIALS AND METHODS
We conducted a parallel-group randomized trial at the
University of Medicine and Dentistry of New Jersey
(UMDNJ), Newark, New Jersey, testing the hypothesis that
an almond-enriched ADA diet would be more effective than a
nut-free ADA diet on improving measures of insulin sensitivity
in individuals with prediabetes. The total study time period for
each participant was 16 weeks.
Eligibility Criteria
Research participants were recruited from a pool of adult
employees at UMDNJ, from patients attending the UMDNJ-
Newark–based Endocrine and Diabetes Clinics, and from
community-based health fairs. Inclusion criteria included the
presence of prediabetes according to the 2005 ADA diagnostic
guidelines (fasting blood glucose between 100 and 125 mg/dl
or casual blood glucose $ 140–199 mg/dl), body mass index
(BMI) 20–35 kg/m2, and willingness to discontinue vitamin E
supplement usage. Persons with a self-reported allergy to
almonds, history of irritable bowel disease or diverticulitis, use
of corticosteroids or immunosuppressant medications, or
presence of liver disease, renal disease, and/or severe
dyslipidemia (triglyceride [TG] . 400 mg/dl or TC .
300 mg/dl) were excluded. The study protocol was approved
by the UMDNJ-Newark campus Institutional Review Board,
and all participants gave written informed consent.
Participants were randomized without stratification using
computer-generated random integer generator software (www.
random.org) to consume almonds (almond enriched, interven-
tion) or to avoid nuts (nut-free, control). The principal
investigator enrolled the participants and generated the
allocation sequence, which was concealed until the interven-
tions were assigned.
Intervention Design
During the week 0 visit, daily energy needs for ADA diet
meal plans were computed based on resting energy expenditure
(REE) measurements obtained from a handheld self-calibrating
indirect calorimeter device (MedGem, model 100, Microlife
USA, Inc., Dunedin, FL). The 510K class II medical device
measures oxygen consumption (VO2) and assesses REE in 5 to
10 minutes using the Weir equation [15] and a constant RQ
value of 0.85. All but 5 participants were evaluated for REE
after an overnight fast (12–14 hours) but were tested at least
4 hours after a meal. Participants sat in a quiet room for a 15-
minute rest period and during testing. The study dietitian
prescribed an individualized ADA diet according to the REE
results and the participant’s self-reported activity level.
Participants with a BMI . 25 kg/m2 (all but 14 participants)
were prescribed energy intake deficits of 250–500 kcal in
accordance with the ADA’s guidelines to achieve modest
weight loss in persons with prediabetes [4]. No meals were
provided, and alcoholic beverages were limited to 2 per day for
men and 1 per day for women.
Participants consumed either an ADA diet with 20% of
energy from almonds and avoided other tree nuts and peanuts
(intervention) or consumed an ADA diet without tree nuts and
peanuts (control). The amount of almonds was determined
based on published data reporting favorable changes in insulin,
glucose, and lipid levels in subjects with impaired glucose
tolerance consuming a diet containing 20% of energy from
MUFA [16]. The prescribed ADA diets contained 15%–20%
protein, ,10% saturated fat, 60%–70% carbohydrate and
MUFA, and cholesterol , 300 mg/d.
A 3-day food and activity record (2 weekdays and 1
weekend day) was requested from participants 1 week prior to
the start of the study. At week 0, each participant met with the
study dietitian for a 1-hour counseling session to receive their
individualized ADA diet. The intervention group received
instruction on how to select 80% of their energy needs using
the ADA Food Exchange System. Monthly supplies of
prepackaged raw or dry roasted almonds were provided at
Almond Consumption and Cardiovascular Risk Factors
190 VOL. 29, NO. 3
clinic visits for the intervention participants. The clinic
supplied the entire almond portion of the diet to the
intervention participants, who were instructed to use only the
prepackaged study almonds. Both groups received instruction
to consume the same number of servings of carbohydrate
exchanges for a given calorie level. In light of the 20% energy
contribution from the almonds, control group participants were
prescribed compensatory servings from the meat and fat
exchange lists. In addition, each participant received 20-
minute counseling sessions for reinforcement of their ADA
diet at weeks 4, 8, and 12 (total of 120 minutes per participant).
Dietary Adherence
To evaluate adherence, a 3-day food/activity record was
completed by each participant at weeks 4, 8, 12, and 16. The study
dietitian reviewed the records according to the prescribed number
of ADA food exchanges and provided reinforcement. The records
were analyzed using the U.S. Department of Agricul-
ture’s database Web site (http://www.mypyramidtracker.gov/).
Dietary adherence was operationally defined as consuming
within 75% of the prescribed diet, which is the approximate
midpoint of the within- and between-subject variation in energy
expenditure as measured by doubly-labeled water versus self-
reported dietary intake [17]. Poor dietary adherence prompted
additional individualized follow-up phone calls for reinforce-
ment.
Outcomes
Outcomes included fasting glucose, insulin, TC, LDL-C,
high-density lipoprotein cholesterol (HDL-C), TG, TC:HDL-C,
and HbA1c. We also examined change in body weight, BMI,
waist circumference (WC), and blood pressure (BP). Plasma a-
tocopherol concentrations were evaluated as a biological marker
to assess almond consumption compliance. In addition, self-
reported dietary intake was examined in the context of changes
from baseline to the second 8 weeks of the study.
Anthropometrics and Laboratory Assessment
Height was measured to the nearest centimeter using a
stadiometer at week 0. Weight and BP were obtained at each
clinic visit. Weight was measured using an internally
calibrated segmental body composition scale/analyzer (model
BC-418 MA, Tanita, Arlington Heights, IL) and recorded to
the nearest 0.01 lb. BMI was calculated as weight (kg)/height
(m2). BP was measured using a calibrated automated digital
monitor (Omron HEM-711). WC was measured to the nearest
0.1 cm, midway between the last rib and the ileac crest.
Venous blood samples were collected at the New Jersey
Medical School General Clinical Research Center after a 12- to
14-hour fast at weeks 0, 8, and 16. Blood was disregarded in 5
participants at a single time point and for 1 participant at 2
time points due to blood draw protocol violations and/or acute
medical conditions known to affect biological measures.
Serum glucose, insulin, TC, LDL-C, HDL-C, TG, and HbA1c
were measured by the UMDNJ University Hospital Clinical
Laboratory according to Clinical Laboratory Improvement
Amendment methods and standardized enzymatic procedures.
Serum insulin levels were measured using direct enzyme-
linked immunoassay methods by LabCorp, Inc. and ICAM
Research Laboratory, UMDNJ. Plasma a-tocopherol concen-
trations were measured by LabCorp, Inc. using high-perfor-
mance liquid chromatography with fluorometric detection.
Insulin resistance was assessed using homeostasis model
analysis (HOMA) based on fasting glucose and insulin levels
[18]. HOMA for insulin resistance (HOMA-IR) and beta-cell
function (HOMA-B) were calculated using the formula
[insulin(pM) 3 glucose(mM)]/22.5 and [20 3 insulin(pM)]/
[glucose(mM) 2 3.5], respectively.
Sample Size
To achieve 80% power using a 5% significance level to
detect a 20% difference in HOMA-IR, a total of 44 participants
was required. Eighty-two individuals met inclusion criteria and
17 declined to participate; thus, 65 participants were enrolled
into the trial.
Analysis
Sample size and power calculations were performed using
SAS version 9.1 (SAS Institute, Cary, NC). Data were entered
into an SPSS version 12.0 (SPSS Inc., Chicago, IL) database,
and statistical analysis was performed using SPSS and SAS.
Bivariate statistical analysis using the chi-square test for
differences in proportions and 2-sided independent t tests were
performed on baseline characteristics using a probability value
of 0.05. To assess the significance of changes in anthropo-
metric and metabolic variables, a mixed-model repeated-
measures analysis of covariance was used with diet, week, and
diet 3 week interaction as fixed effects, adjusting for baseline
measurements of the outcome variable. A natural log
transformation was performed on outcome variables for the
modeling analysis when indicated to improve normality, and
the results were exponentiated for reporting purposes. An
appropriate within-subject covariance structure was deter-
mined for each of the outcome variables, and an unstructured,
compound symmetric or first-order autoregressive covariance
structure was applied. Additional analyses for glucose, insulin,
HOMA-IR, and HOMA-B models were performed and
adjusted for weight by adding the baseline weight as a
covariate. An intent-to-treat analysis was performed, and all
percentage change values presented are calculated from least-
squares means (LSM) estimated from mixed models. Week 0,
8, and 16 measurements were included in the analysis, with the
Almond Consumption and Cardiovascular Risk Factors
JOURNAL OF THE AMERICAN COLLEGE OF NUTRITION 191
exception of weight, BMI, WC, and BP that included
additional measurements from weeks 4 and 12. The assump-
tion used in the intent-to-treat model with regard to the
aforementioned disregarded samples and missing data and
unmeasured endpoints for the dropouts was that they were
missing at random.
RESULTS
Study recruitment occurred from January 2006 to January
2007, and the last participant completed the study in April
2007.
Sixty-five individuals met all inclusion criteria and enrolled
(intervention arm, n 5 32; control arm, n 5 33). During the
study, 11 participants withdrew (intervention arm, n 5 7;
control arm, n 5 4), primarily due to work and personal
schedule conflicts.
Participants randomized into the intervention and control
arms were similar in terms of baseline characteristics
(Table 1).
The intent-to-treat analyses are presented based on
percentage change in LSM (Table 2). Both groups experienced
declines in weight, BMI, and WC during the study, but there
were no significant differences between groups in these
measurements at any time point (p 5 0.191 to p 5 0.557).
Fasting blood glucose levels decreased ,2 mg/dl (,2%) in
both groups (p 5 0.978). However, almond consumption was
associated with a greater reduction in fasting insulin
(21.78 mU/ml [223.3%] vs. +1.47 mU/ml [+19.2%], p 5
0.002), HOMA-IR (20.48 [224.9%] vs. +0.30 [+15.5%], p 5
0.007) and HOMA-B (213.2 [217.8%] vs. +22.3 [+30.0%], p
5 0.001). The magnitudes of change in insulin, HOMA-IR,
and HOMA-B were virtually unaffected after adjusting for
weight. The intervention group experienced a 2.11-mU/ml
(27.6%) reduction in fasting insulin by week 8 and overall
reduction of 1.78 mU/ml (23.3%) at week 16, in contrast to the
control group, who showed a 0.06-mU/ml (0.8%) increase by
week 8 and overall increase of 1.47 mU/ml (19.2%) at week 16
(Fig. 1).
Two participants in each group were taking lipid-lowering
medications. There was no significant change in TC, HDL-C,
or TG between the almond-enriched intervention and the nut-
free control group. While failing to meet the prespecified
cutoff for statistical significance, a clinically significant
decline in LDL-C was found in the almond-enriched
intervention group (212.4 mg/dl vs. 20.4 mg/dl, p 5 0.052)
as compared with the nut-free control group. No significant
changes were observed in HbA1c, systolic BP, or diastolic BP
between the almond-enriched intervention and the nut-free
control group.
The mean intake of almonds for participants in the
intervention group was 60 g per day. There was a 0.27-mg/L
(2%) decrease in the mean plasma a-tocopherol level in the
nut-free control group (p 5 0.65) at week 8 in contrast to a
1.74-mg/L (17%) increase observed in the almond-enriched
intervention (p , 0.01).
Approximately 80% of participants met the operational
definition of dietary adherence in both groups. There was no
difference in self-reported mean dietary intakes from week 4 to
8 (first 8 weeks) and week 12 to 16 (second 8 weeks);
therefore, the 2 sets of 3-day food records were collapsed
(Table 3). Using paired data to evaluate within-group changes
from baseline to the second 8 weeks of the study, a 5%
decrease in energy from carbohydrate was observed in the
intervention group in the context of a 5% increase in total fat,
5% increase in MUFA, 1% increase in PUFA, 5-g/d increase in
fiber, and 10-mg a-tocopherol equivalent (TE) increase in
Table 1. Baseline Characteristics by Intervention and
Control Armsa
Characteristic
Intervention
(n 5 32)
Control
(n 5 33)
Age (y) 53 6 9 54 6 11
Gender
Female 22 (69) 26 (79)
Male 10 (31) 7 (21)
Race
Caucasian 12 (38) 13 (40)
Hispanic 4 (12) 5 (15)
African American 14 (44) 9 (27)
Asian 2 (6) 6 (18)
Weight (kg) 82.9 6 14.4 80.5 6 14.4
Body mass index (kg/m2) 30 6 5 29 6 5
Waist circumference (cm) 95 6 13 96 6 12
Resting energy expenditure
(kcal) 1708 6 364 1635 6 375
Plasma lipids, mg/dlb
Total cholesterol 202 6 36 199 6 42
LDL cholesterol 117 6 32 118 6 38
HDL cholesterol 63 6 16 59 6 12
Triglycerides 113 6 58 124 6 75
Total cholesterol:HDL
cholesterol 3.40 6 0.93 3.49 6 1.10
HbA1c (%) 5.8 6 0.6 6.1 6 0.5
Fasting blood glucose, mg/dlb 101 6 13 104 6 14
Fasting insulin, mU/ml 11.4 6 9.4 9.0 6 5.6
HOMA-IR 2.9 6 2.5 2.4 6 1.7
HOMA-B 112 6 82 83 6 60
Systolic blo