Almond Consumption Improves Inflammatory Profiles Independent of Weight Change: A 6-Week Randomized Controlled Trial in Adults with Obesity

Abstract

Background: Obesity is characterized by chronic low-grade systemic inflammation that contributes to metabolic dysfunction. Diet is a modifiable factor that can help reduce this inflammation. Nuts such as almonds are rich in unsaturated fats, and antioxidant and anti-inflammatory micronutrients, which may work synergistically to attenuate obesity-related inflammation. Hence, the objective of this study was to investigate whether daily almond consumption improves systemic inflammatory and immune markers in adults with obesity. Methods: In this randomized controlled parallel-arm trial (ClinicalTrials.gov ID NCT05530499), 69 adults (age 30–45 years) with obesity (BMI 30–45 kg/m²) were assigned to consume either 57 g/day of almonds (n = 38) or an isocaloric snack (cookie; n = 31) for six weeks. Fasting serum inflammatory cytokines, innate immune cell counts, body weight, serum glucose, insulin, lipid profile, and alpha-tocopherol were measured at baseline and week six. Dietary intake, compliance, palatability, acceptance, and appetite ratings were also assessed. Primary outcomes were analyzed using linear mixed models and baseline-adjusted linear models. Results: Subjective compliance was high in both groups, with greater acceptance of almonds (p < 0.05); however, serum alpha-tocopherol did not change. Almond consumption significantly decreased serum IL-6, TNF-α, and IFN-γ over 6 weeks compared with the cookie group (p < 0.05). No significant group differences were observed for innate immune cell counts, body weight, appetite ratings, blood pressure, or serum fasting glucose, insulin, total cholesterol (C), LDL-C, and triglycerides over six weeks. The almond group also increased intakes of monounsaturated fat, fiber, alpha-tocopherol, magnesium, zinc, and manganese, and improved diet quality indices relative to the cookie group (p < 0.05). Conclusions: Daily almond consumption for six weeks improved inflammatory cytokine profiles in adults with obesity, without changes in body weight under free-living conditions. These findings support recommending almonds as part of healthy dietary patterns to help attenuate obesity-related inflammation.

1. Introduction

Obesity is a state of chronic low-grade systemic inflammation, characterized by elevated circulating concentrations of pro-inflammatory cytokines and measurable shifts in innate immune cell populations [1,2]. Adipose tissue expansion in obesity amplifies inflammatory signaling contributing to immune cell activation which eventually exacerbates systemic inflammation [1,2,3]. This persistent inflammatory response is a key biological mechanism linking obesity to increased risk of insulin resistance, type 2 diabetes, and cardiovascular complications [4,5,6,7].

Diet is a modifiable factor that can help reduce the chronic low-grade systemic inflammation observed in obesity. Among various healthy eating patterns, the Mediterranean (MED)-style dietary pattern has consistently demonstrated beneficial effects on inflammatory and immune markers. Large epidemiological studies, including the Nurses’ Health Study and the ATTICA study, report that higher adherence to a MED-style dietary pattern is associated with lower concentrations of pro-inflammatory cytokines and immune cell counts [8,9]. For example, findings from the ATTICA study showed that participants with the highest adherence had, on average, 17% lower IL-6 concentrations and 14% lower white blood cell counts compared to those with the lowest adherence. Systematic reviews and meta-analyses of randomized controlled trials (RCTs) further support that MED-style dietary interventions reduce circulating inflammatory markers such as IL-6, IL-1β, and C-reactive protein (CRP) across diverse populations [10,11]. Another systematic review on plant-based diets shows a reduction in IL-6 but not TNF-alpha for individuals with overweight and obesity with diverse health status such as metabolic syndrome, CVD, or diabetes [12].

Nuts are a core component of high-quality dietary patterns [13] and may contribute substantially to the anti-inflammatory effects associated with these dietary patterns. Almonds, in particular, are among the most widely consumed tree nuts globally [14] and provide a variety of nutrients and bioactive compounds, including monounsaturated fats, vitamin E, zinc, magnesium, polyphenols, which are important for regulating inflammation, immune function, and oxidative stress [15,16,17]. To date, most evidence on almonds and inflammation has been synthesized from systematic reviews and meta-analyses that include studies conducted across diverse BMI categories, age ranges, and health statuses [18,19,20]. While some trials have enrolled individuals with obesity, these analyses typically combine individuals with overweight and obesity, include mixed nut interventions in which the independent effects of almonds cannot be isolated, or encompass wide age ranges [21,22,23,24], thereby increasing heterogeneity. Such heterogeneity reduces precision and limits obesity-specific inferences for inflammatory markers. This is particularly relevant given that inflammation varies with adiposity and age, and that in populations with elevated baseline inflammation, such as middle-aged adults with obesity [25,26,27,28,29], excess variability may obscure detectable almond intervention effects. Addressing these gaps could strengthen dietary recommendations aimed at lowering obesity-related inflammation and its metabolic consequences. Therefore, the primary objective of this randomized, controlled, parallel-arm clinical trial was to investigate the effects of six weeks of almond consumption on systemic immune and inflammatory markers in middle-aged adults with obesity.

2. Materials and Methods

The procedures involving human participants in this study were approved by the Institutional Review Board of the University of Missouri–Columbia (Protocol number 2090290). The study is registered on ClinicalTrials.gov (NCT05530499).

2.1. Participant Characteristics

Participants were recruited via public advertisements. Inclusion criteria included the following: (a) age 30–45 years old, (b) BMI of 30–45 kg/m², (c) willingness to maintain consistent diet and activity patterns, (d) willingness to consume study foods, (e) non-smokers over the past year, (e) and weight stable (no greater than 5 kg change over the last 3 months). Exclusion criteria included diagnosed diabetes (fasting blood glucose >125 mg/dL), pregnant or lactating individuals, regular consumer of nuts (>3 times/week), planning on getting viral vaccines during study, allergies to study foods, illicit drug use, current use of medication that interfere with immune measures, recent start of medications that affect metabolism or appetite and drug therapy for coronary artery disease, peripheral artery disease, congestive heart failure or dyslipidemia. A total of 90 individuals were screened to determine eligibility for participation in this study, and 69 participants started the study. The participant flow is depicted in Figure 1.

Figure 1.

Consort flow diagram of participants in the almond and cookie groups from enrollment to study completion.

2.2. Study Design

The study was a 6-week randomized, controlled, parallel-arm intervention. Participants were randomly assigned into one of two study arms: whole natural dry-roasted almonds (n = 38; 57 g/day (2 oz); 322 kcal) or isocaloric comparator of cookies (n = 31; 66 g (6 cookies); 325 kcal). The cookie was selected as the comparator because it represents a typical processed snack consumed in North America and was intended to reflect existing dietary exposure in this target population [30,31,32]. The nutrient composition of the study foods is shown in Table S1. Participants incorporated the assigned foods into their usual diet according to personal preference and maintained their customary dietary and physical activity habits. No dietary counseling was provided in this free-living trial. Study foods were dispensed at baseline and again at the week 3 check-in visit. Subjective compliance with study food consumption was monitored via daily text messages asking participants whether they consumed their food and the time of consumption (morning, afternoon, and evening). The baseline characteristics of the participants are shown in Table 1.

Table 1.

Baseline characteristics of participants in the almond and cookie groups in the 6-week randomized controlled trial in adults with obesity.

Characteristic	Almond	Cookie
Sample size, n	38	31
Sex, n (%)
Female	31 (82%)	20 (65%)
Male	7 (18%)	11 (35%)
Age §, years	37.8 ± 4.3	36.4 ± 4.4
BMI §, kg/m2	35.9 ± 4.3	34.8 ± 3.4
Race, n (%)
White	30 (79%)	27 (87%)
Black or African American	8 (21%)	2 (6.5%)
Asian	-	2 (6.5%)
Ethnicity, n (%)
Hispanic or Latino	2 (5%)	3 (10%)
Not Hispanic or Latino	36 (95%)	28 (90%)

^§, mean ± SD.

2.3. Sample Size Calculation and Group Randomizations

Sample size calculations for this study were informed by preliminary data from a previous intervention [33] involving a subset (n = 20) of participants per group (almond vs. cracker) in young adults with a mixed BMI status. In that population, we observed Cohen’s d effect sizes ranging from small to moderate (d = 0.19–0.60) for 8-wk changes in key markers such as IL-6, IL-10, and TNF-alpha. Given that the current study targets a population of middle-aged adults with obesity, who typically exhibit greater baseline inflammation [25,26,27,28,29], we anticipated at least equal or larger effect sizes than those observed in the earlier cohort. Assuming a moderate effect size (Cohen’s d = 0.6), a two-tailed alpha of 0.05, and 80% power, a sample size of approximately 25–30 participants per group was estimated to be sufficient to detect differences over 6 weeks in most markers. To account for anticipated attrition (~10–20%), we aimed to enroll approximately 34–38 participants per group to achieve ~30 completers per group. Randomization to the groups was performed using a restricted 1:1 allocation sequence generated in Microsoft Excel. A list containing 42 almond and 42 cookie assignments was randomly permuted using the RAND() function, and participants were assigned sequentially. Enrollment was stopped once 30 participants per group completed the study. This was a single-blinded trial, where clinical and laboratory analysts were blinded to group allocation. Blinding was achieved by coding samples with ID numbers.

2.4. Outcomes

Primary outcomes for this study were immune and inflammatory markers. Secondary outcomes for this analysis were anthropometric, cardiovascular, dietary, and appetite ratings. All outcomes were assessed before and after the 6-week study at minimum.

2.4.1. Blood Immune and Inflammatory Markers

Blood samples were obtained at each visit using butterfly needle venipuncture after at least 8 h of fasting. Blood samples for white blood cells (WBC) assessments were collected in EDTA tubes. Differential WBC count, i.e., absolute and proportions for lymphocytes, neutrophils, monocytes, eosinophils, and basophils were assessed using counting and microscopy methodologies by Quest Diagnostics–Lenexa in Kansas.

Serum samples were obtained from blood collected in serum separator tubes (SSTs) which were allowed to sit for 1 h before centrifugation for 10 min at 1.3 RCF. Serum samples were stored at −80 °C until analysis. Fasting serum concentrations of inflammatory markers TNF-alpha (kit no. KAC1751), IFN-gamma (KAC1231), IL-6 (KAC1261), and IL-10 (KAC1321) were quantified in duplicate using commercially available ELISA kits from Invitrogen (Carlsbad, CA, USA). According to the manufacturer’s documentation, intra-assay CVs ranged from 3.0 to 5.0% and the inter-assay CVs ranged from 5.2 to 8.8% for all cytokines measured. All samples were analyzed using the same batch and randomized across plates to reduce systematic bias.

2.4.2. Blood Metabolic and Other Markers

Fasting serum glucose concentrations were measured using an enzymatic amperometric glucose oxidase method by the YSI 2300 STAT Plus device (YSI, Yellow Springs, OH, USA). Fasting serum triglycerides (TG; E-BC-K238), total cholesterol (TC; E-BC-K109-S), and high-density lipoprotein cholesterol (HDL-C; E-BC-K221-M) concentrations were measured using commercially available enzymatic colorimetric assays (Elabscience^®, Houston, TX, USA). Low-density lipoprotein cholesterol (LDL-C) was calculated using Friedwald equation LDL-C = [(TC − HDL-C) − (TG/2.2)] [34]. Values were subsequently converted to mg/dL for reporting. The fasting serum concentrations of insulin were quantified using commercially available ELISA kit from Millipore (EZHI-14k, Burlington, MA, USA). Insulin sensitivity and resistance indices were calculated as we have done previously [35]. Quantitative insulin-sensitivity check index (QUICKI), an estimate of fasting insulin sensitivity, was computed using [1/[log(fasting insulin μU/mL) + log(fasting glucose mg/dL)]. Homeostatic model assessment for insulin resistance (HOMA-IR), an estimate of fasting insulin resistance, was computed using [fasting insulin (µU/mL) × fasting glucose (mg/dL)/405].

Blood samples for tocopherol measurements were collected in light-protected tubes and transported to a Quest Diagnostics^® laboratory for analysis at the end of the participant visit. Tocopherol concentrations were assessed using chromatography methodology by SLI Quest Diagnostics–Nichols Valencia in California.

2.4.3. Anthropometric and Blood Pressure Outcomes

Height was measured to the nearest 1 mm using a stadiometer (SECA 216 Height Measuring Rod, SECA, Hamburg, Germany). Body weight was assessed in lightweight clothes to the nearest 100 g using standard calibrated electronic scales. Waist and hip circumferences were recorded at the narrowest part of the torso and the widest part of the hip–gluteal region respectively using a standard measuring tape. Resting systolic and diastolic blood pressure (BP) was assessed using an OMRON automatic BP device (model BP:725, Kyoto, Japan) on the non-dominant arm after participants’ five minutes of sitting. Height, weight, waist and hip circumferences, and blood pressure measurements were conducted twice and averaged for analysis. Body composition indices (fat free mass (FFM), fat mass, and their percentages) were assessed using bioelectrical impedance analysis (BIA, RJL-Systems Quantum—V Segmental, Ref: Q5S, Clinton Township, MI, USA) in a reclining position.

2.4.4. Dietary Outcomes

Dietary data were collected using the NCI’s semi-quantitative past month version of the Diet History Questionnaire III (DHQ III) before and after the 6-week intervention [36]. Total and detailed DHQ nutrient compositions were analyzed and used to calculate Healthy Eating Index (HEI), Dietary Approaches to Stop Hypertension (DASH), and MED dietary pattern indices with the dietary index package in R (v4.5.2) [37] to provide important contextual information regarding dietary patterns. Self-administered single 24 h dietary recalls using ASA24^® Dietary Assessment Tool (2022 version) [38] were also conducted before and after the intervention.

2.4.5. Free-Living Appetite Ratings

The intensity of hunger, fullness, desire to eat, and prospective consumption were measured on electronic versions of 100 mm visual analog scales, with scale endpoints ranging from “not at all” to “extremely” [35]. Participants recorded their responses every hour during the waking hours in a 24 h period. The time points between 8 am and 8 pm had the most complete data and were considered for further analysis. Intermittent missing time points within an assessment day were imputed within participant using neighboring observed values (carried forward or backward or averaged when both were available) to enable calculation of within-day AUCs. Morning, afternoon, evening, and total 12 h period area under the curve (AUC) indices were computed using the trapezoidal rule.

2.4.6. Acceptance and Palatability Ratings

A 9-point food action rating scale assessed the weekly acceptability of almond and cookies (1 = “I would eat this if I were forced to” and 9 = “I would eat this every opportunity I had”) [39]. Additionally, the palatability of food items was evaluated weekly by a hedonic general labeled magnitude scale (gLMS) (100 = “Extremely unpalatable” and 100 = “Extremely palatable”) [40].

2.5. Statistical Analyses

Statistical analyses were conducted in R (version 4.5.2). Linear mixed model analyses were performed using the lmerTest package [41] with group (Almond and Cookie) as the between-subject fixed effect, week (Baseline and Week 6) as the within-subject fixed effect, the group × week interaction, and participant included as a random intercept to account for repeated measures. For serum markers assessed via ELISA, the plate number was included as a covariate to control for variations according to plates. For dietary outcomes, energy intake was included as a covariate. Anova type III tests were conducted on the fitted model using the car package. When a significant week × group interaction was observed, estimated marginal means and planned contrasts were calculated using the emmeans package [42] to perform the following pairwise comparisons: (1) almond vs. cookie at baseline, (2) almond vs. cookie at week 6, (3) baseline vs. week 6 within the almond group, (4) baseline vs. week 6 within the cookie group, and (5) change from baseline to week 6 in the almond group vs. change from baseline to week 6 in the cookie group. p-values were adjusted for multiple comparisons using the multivariate t distribution (mvt) adjustment [43]. Baseline-adjusted linear regression models were also fit using the lm() function in R to evaluate group differences at week 6 while adjusting for baseline values of the outcome.

Missing outcome data were not imputed for linear mixed models, as maximum likelihood estimation inherently accommodates incomplete repeated-measures data under the missing-at-random assumption [44]. For baseline adjusted analyses, missing values were addressed using multiple imputation by chained equations implemented with the mice package using predictive mean matching [45]. Imputation was performed separately within each treatment group and included sex, race, age, baseline BMI, and the respective baseline value of each outcome as predictors as has been done by others [46]. The number of imputations was determined using the von Hippel rule based on the maximum observed proportion of missing data [47]. Imputed datasets were generated with 50 iterations per chain, analyses were conducted within each imputed dataset, and fixed-effect estimates were pooled using Rubin’s rules [48]. Estimated marginal means, contrasts, and joint tests were subsequently computed from the pooled model estimates.

Model assumptions were evaluated using residual diagnostics, including skewness and kurtosis, Levene’s tests, and graphical inspection of residual distributions. Data not meeting normality assumptions were transformed using the Johnson’s family of transformations [49] in JMP Pro (version 17.0); however, only the non-transformed data are presented for interpretation purposes unless otherwise indicated.

The group effect on total compliance % was analyzed by standard least squares regression. A GEE-based multinomial logistic regression was conducted in R using the multgee package [50] to evaluate whether the timing of consumption (morning, afternoon, and evening) differed by group, accounting for repeated days of data per participant. FACT and palatability data were analyzed using non-parametric analysis of week (W1, W2, W3, W4, W5, and W6) and group interaction effects using the nparLD package in R [51]. For non-parametric analyses, post hoc tests for significant effects were conducted using Mann–Whitney analyses for between-subject factors, and nparLD for within-subject factors, with adjustments across contrasts (comparions) using Benjamini–Hochberg correction.

3. Results

3.1. Participants Anthropometric, Cardiovascular, and Biochemical Characteristics

There were no group differences in body weight, waist and hip circumferences, blood pressure, serum glucose, insulin, insulin sensitivity and resistance indices, tocopherol (alpha and beta-gamma) concentrations, and serum lipid profile (total cholesterol, LDL-C, and triglycerides) over the 6-week intervention (Table 2). Although there were statistically significant group × week interaction effects for total fat mass %, total fat-free mass %, none of the post hoc comparisons were significant. HDL-C concentrations were higher in the almond group than in the cookie group at W6 (BL-adjusted group effect, p < 0.05); however, this difference was largely driven by a tendency for HDL-C to decline in the cookie group (Group × Week Effect, p ≤ 0.1).

Table 2.

Anthropometric, body composition, and clinical data for almond and cookie groups at baseline and week 6 in adults with obesity.

	Baseline		Week 6		BL-Adjusted Model p-Values	Linear Mixed Effect Model p-Values
	Almond	Cookie	Almond	Cookie	BL-Adjusted Group	Group	Week	Group × Week
Weight (kg)	100.15 ± 14.91	99.96 ± 18.66	98.67 ± 14.6	98.82 ± 16.58	0.894	0.964	0.155	0.947
BMI (kg/m2)	36.15 ± 4.37	34.54 ± 3.18	35.74 ± 4.47	35.09 ± 3.49	0.694	0.222	0.191	0.630
Waist Circumference (cm)	102.31 ± 11.2	101.8 ± 9.93	101.87 ± 10.69	101.06 ± 9.74	0.409	0.740	0.542	0.406
Waist–Hip Ratio	0.85 ± 0.09	0.87 ± 0.08	0.85 ± 0.08	0.86 ± 0.08	0.911	0.314	0.331	0.612
BIA Fat%	44.02 ± 9.02	45.9 ± 4.32	46.91 ± 6.46	45.52 ± 5.64	0.147	0.802	0.226	0.040
BIA FFM%	55.98 ± 9.02	54.1 ± 4.32	53.09 ± 6.46	54.48 ± 5.64	0.128	0.802	0.226	0.040
Systolic BP (mmHg)	123.74 ± 11.55	123.63 ± 12.03	121.18 ± 9.26	121.68 ± 11.08	0.748	0.948	0.078	0.816
Diastolic BP (mmHg)	82.66 ± 7.02	80.77 ± 8.13	82.49 ± 6.71	80.17 ± 7.63	0.479	0.220	0.754	0.846
TG (mg/dL)	109.57 ± 62.67	82.78 ± 49.34	91.17 ± 41.91	90.48 ± 67.66	0.351	0.213	0.933	0.139
TC (mg/dL)	236.76 ± 42.84	221.43 ± 35.7	230.48 ± 50.77	232.33 ± 44.98	0.391	0.448	0.355	0.228
LDL (mg/dL)	159.92 ± 43.37	150.76 ± 31.18	157.25 ± 50.07	162.45 ± 38.25	0.365	0.813	0.306	0.243
HDL (mg/dL)	55.66 ± 7.22	52.36 ± 9.11	56.33 ± 8.37	48.74 ± 6.69	0.002	<0.001	0.187	0.074
Glucose (mg/dL)	105.68 ± 11.6	102.43 ± 9.63	106.86 ± 13.52	108.32 ± 12.47	0.459	0.717	0.033	0.147
Insulin (μU/mL)	20.01 ± 13.69	14.48 ± 9.56	16.87 ± 14.37	14.38 ± 8.5	0.611	0.071	0.406	0.476
HOMA-IR ǂ	5.25 ± 3.83	3.72 ± 2.57	4.4 ± 3.58	3.87 ± 2.37	0.731	0.211	0.413	0.262
QUICKI ǂ	0.31 ± 0.03	0.33 ± 0.03	0.32 ± 0.03	0.33 ± 0.05	0.726	0.201	0.378	0.320
α-tocopherol (mg/L)	12.13 ± 4.36	11 ± 2.71	11.62 ± 3.67	11.26 ± 2.98	0.997	0.200	0.684	0.861
β-γ tocopherol (mg/L)	1.48 ± 0.61	1.69 ± 0.7	1.77 ± 0.92	1.7 ± 0.66	0.554	0.466	0.569	0.657

Non-transformed data presented as mean ± SD. ^ǂ Indicates p-values reported for JN transformed variable. Repeated measures were analyzed using linear mixed-effects models (fixed effects: group, week, group × week; random intercept: participant) with kit plate number included as a covariate for serum markers and derived indices. The Group, Week, and Group × Week columns report type III tests of fixed effects from these models; pairwise comparisons used multivariate-t (mvt) adjustment. Baseline-adjusted group reports the group effect at week 6 from a separate baseline-adjusted linear regression. Missing data were handled by maximum likelihood (mixed models) or multiple imputation (regression models). QUICKI, quantitative insulin-sensitivity check index; HOMA-IR, homeostatic model assessment for insulin resistance; FFMI, fat-free mass index; TC, total cholesterol; HDL, high density lipoproteins; LDL, low density lipoproteins; TG, triglycerides; BP, blood pressure; WHR, Waist-to-hip ratio.

3.2. Study Food Compliance, Acceptance, and Palatability

Mean subjective compliance percentage was high for both groups (almond: 93.9% and cookie: 92.5%) and no statistically significant group differences were observed. There were no group differences in time of consumption as well.

Because ratings were first collected after one week of food consumption, baseline (pre-intervention) values are not considered. Overall, almonds had higher acceptance than cookies (group effect, p < 0.05) but similar palatability. Acceptance and palatability decreased over the subsequent weeks of the intervention (week effect, p < 0.05, Table 3), with no group × week interaction, indicating that changes over time were similar between groups.

Table 3.

Acceptance and palatability ratings of study foods across weeks 1–6 in adults with obesity.

	Week	Almond Median (IQR)	Cookie Median (IQR)	p-Values
FACT Scale	Week 1 ^,**	6 (6–7)	5 (5–6)	Group: 0.015 Week: <0.001 Group × Week: 0.520
	Week 2 ^,**	6 (5–7)	5 (5–6)
	Week 3 **	6 (5–7)	5 (5–6)
	Week 4 *,**	6 (4.25–7)	6 (5–6)
	Week 5 **	6 (3.25–7)	5 (4–6)
	Week 6	5.5 (4.75–7)	5 (3.25–5.75)
Palatability (/100 mm)	Week 1	58.62 (41.93–64.53)	58.14 (41.18–73.27)	Group: 0.503 Week: 0.034 Group × Week: 0.398
	Week 2 ^,^^	58.62 (50.3–63.75)	57.22 (41.38–70.78)
	Week 3	57.28 (35.67–65.15)	56.82 (44.71–72.09)
	Week 4 ^,^^	58.28 (41.23–66.57)	58.62 (50.46–68.08)
	Week 5	51.16 (41.28–64.12)	57.47 (44.22–60.35)
	Week 6	54.95 (41.38–65.2)	56.62 (46.65–60)

Non-transformed data presented as median (IQR). * BH-adjusted pairwise p-value ≤ 0.05 for timepoint vs. W5; ** BH-adjusted pairwise p-value ≤ 0.05 for timepoint vs. W6; ^{^} BH-adjusted pairwise p-value ≤ 0.1 for timepoint vs. W5; ^{^^} BH-adjusted pairwise p-value ≤ 0.1 for timepoint vs. W6. FACT and palatability outcomes were analyzed using non-parametric repeated-measures models (week, group, week × group; nparLD, R). Post hoc Mann–Whitney (between-group) and nparLD (within-subject) contrasts were performed with Benjamini–Hochberg (BH) correction.

3.3. Appetite Ratings

No group differences in morning, afternoon, evening, or 12 h AUCs were observed for hunger, desire to eat, or prospective consumption ratings over the 6-week intervention (Table 4). Although there were statistically significant group × week interaction effects for 12 h fullness AUC, none of the post hoc comparisons were significant.

Table 4.

Subjective appetite ratings for almond and cookie groups at baseline and week 6 in adults with obesity.

	Baseline		Week 6		BL-Adjusted Model p-Values	Linear Mixed Effect Model p-Values
	Almond	Cookie	Almond	Cookie	BL-Adjusted Group	Group	Week	Group × Week
Hunger
Morning AUC	70.4 ± 32.09	62.45 ± 35.72	68.84 ± 33.99	82.04 ± 45.88	0.217	0.729	0.202	0.120
Afternoon AUC	85.11 ± 37.2	84.34 ± 48.55	75.27 ± 38.63	73.83 ± 39.25	0.781	0.919	0.124	0.957
Evening AUC	44.04 ± 25.51	33.5 ± 27.28	39.58 ± 19.3	38.77 ± 26.32	0.957	0.274	0.939	0.267
12-Hour AUC	199.55 ± 56.74	180.28 ± 80.14	183.69 ± 77.75	194.64 ± 88.87	0.535	0.820	0.767	0.247
Fullness
Morning AUC	118.46 ± 47.21	130.28 ± 52.48	121.52 ± 39.78	108.06 ± 50.34	0.179	0.988	0.313	0.079
Afternoon AUC	159.39 ± 54.04	160.96 ± 67.36	167.65 ± 49.06	155.28 ± 73.19	0.689	0.710	0.537	0.308
Evening AUC	112.27 ± 33.69	116.7 ± 37.4	107.11 ± 31.25	98.81 ± 39.04	0.260	0.808	0.032	0.180
12-Hour AUC	390.12 ± 113.47	407.94 ± 138.92	396.27 ± 101.23	362.14 ± 146.39	0.209	0.826	0.425	0.040
Desire to Eat
Morning AUC	73.02 ± 28.58	69.06 ± 40.21	68.13 ± 35.76	78.35 ± 49.38	0.351	0.681	0.825	0.297
Afternoon AUC	81.64 ± 41.46	88.41 ± 50.11	66.89 ± 31.2	73.23 ± 43.94	0.671	0.462	0.023	0.992
Evening AUC	43.46 ± 27.91	39.82 ± 29.99	35.27 ± 19.09	40.31 ± 27.37	0.389	0.918	0.394	0.357
12-Hour AUC	198.12 ± 59.39	197.29 ± 86.99	170.29 ± 68.08	191.89 ± 95.87	0.436	0.540	0.075	0.335
Prospective Consumption
Morning AUC	74.55 ± 29.78	72.09 ± 37.4	70.04 ± 29.85	81.88 ± 49.59	0.418	0.560	0.796	0.268
Afternoon AUC	84.77 ± 36.18	103.19 ± 53.7	76.87 ± 33.02	78.03 ± 41.84	0.563	0.315	0.004	0.185
Evening AUC	44.62 ± 24.77	41.54 ± 32.65	41.29 ± 21.1	42.41 ± 24.13	0.683	0.898	0.820	0.574
12-Hour AUC	203.94 ± 62.42	216.83 ± 92.81	188.21 ± 67.03	202.31 ± 99.74	0.994	0.469	0.088	0.933

Non-transformed data presented as mean ± SD. Values are AUC × 10². Repeated measures were analyzed using linear mixed-effects models (fixed effects: group, week, group × week; random intercept: participant). The Group, Week, and Group × Week columns report type III tests of fixed effects from these models; pairwise comparisons used multivariate-t (mvt) adjustment. Baseline-adjusted group reports the group effect at week 6 from a separate baseline-adjusted linear regression. Missing data for analyses were handled by maximum likelihood (mixed models) or multiple imputation (regression models).

3.4. Dietary Assessments

All nutrients and indices were adjusted for energy intake in the analyses (Table 5 and Table S2). The energy-adjusted analyses of the past month DHQ data shows that total MUFA, oleic acid, total fiber, insoluble fiber, alpha-tocopherol, beta-tocopherol, calcium, magnesium, manganese, phosphorous, zinc, and dietary pattern scores for HEI, MED, and DASH were higher in the almond group, and delta-tocopherol, gamma-tocopherol, and refined grains intake were lower in the almond group compared to the cookie group at the end of the 6-week intervention (BL and energy-adjusted group effect, p < 0.05, Table 5). Although there were no BL-adjusted group differences in protein and potassium intake at W6, statistically significant group × week interaction effects indicate a decrease in protein and potassium intake in the cookie group over the 6-week intervention. The findings from the single ASA24 dietary recall data reflecting intake over the day preceding the visits at BL and W6 also reflected higher intake of MUFA, oleic acid, alpha-tocopherol, and phosphorous in the almond group at W6 (BL and energy-adjusted group effect, p < 0.05, Table S2) and a tendency for a decline in potassium intake in the cookie group (Group × Week Effect, p < 0.05, Table S2).

Table 5.

Self-reported dietary intake (energy-adjusted) from the Diet History Questionnaire III for the almond and cookie groups at baseline and week 6 in adults with obesity.

	Baseline		Week 6		BL-Adjusted Model p-Values	Linear Mixed Effect Model p-Values
	Almond	Cookie	Almond	Cookie	BL-Adjusted Group	Group	Week	Group × Week
Carbohydrate (g)	170.75 ± 5.35	169.27 ± 6	159.75 ± 5.55 *	173.43 ± 5.94	0.317	0.404	0.267	0.014
Total fat (g) ǂ	−0.03 ± 0.07	−0.12 ± 0.08	0.18 ± 0.07	−0.06 ± 0.08	0.241	0.069	0.001	0.076
Total MUFA (g)	22.47 ± 0.86	21.28 ± 0.97	27.56 ± 0.92 *,**	21.9 ± 0.96	0.026	0.002	<0.001	0.001
Oleic acid (g)	20.78 ± 0.81	19.54 ± 0.92	25.91 ± 0.87 *,**	20.33 ± 0.9	0.014	0.001	<0.001	0.001
Total PUFA (g)	13.43 ± 0.45	12.49 ± 0.5	14.36 ± 0.48	12.77 ± 0.5	0.471	0.022	0.106	0.384
Total protein (g)	63.96 ± 3.84	70.83 ± 4.28	63.46 ± 3.89	65.01 ± 4.26 **	0.652	0.445	0.019	0.049
Total dietary fiber (g)	15.11 ± 0.69	14.43 ± 0.78	15.75 ± 0.73 *	12.7 ± 0.77	0.027	0.032	0.324	0.032
Insoluble dietary fiber (g)	10.52 ± 0.5	9.67 ± 0.56	11.49 ± 0.53 *	8.63 ± 0.55	0.006	0.002	0.930	0.022
Soluble dietary fiber (g) ǂ	−0.37 ± 0.07	−0.44 ± 0.08	−0.46 ± 0.07	−0.5 ± 0.08	0.881	0.588	0.048	0.670
Total alpha-tocopherol (mg) ǂ	−0.1 ± 0.11	−0.21 ± 0.13	0.69 ± 0.12 *,**	−0.38 ± 0.13	0.000	<0.001	0.001	<0.001
Beta-tocopherol (mg)	0.35 ± 0.02	0.3 ± 0.03	0.44 ± 0.03 *,**	0.28 ± 0.03	0.009	0.001	0.088	0.012
Delta-tocopherol (mg)	1.97 ± 0.13	1.83 ± 0.14	1.88 ± 0.13 *	2.73 ± 0.14 **	0.000	0.015	0.001	<0.001
Gamma-tocopherol (mg)	10.61 ± 0.51	9.61 ± 0.58	9.19 ± 0.55	10.74 ± 0.57	0.040	0.653	0.751	0.006
Calcium (mg)	850.4 ± 50.54	878.43 ± 56.87	971.97 ± 53.01 *	753.72 ± 56.11	0.001	0.154	0.964	<0.001
Magnesium (mg)	270.74 ± 12.93	279.04 ± 14.6	334.71 ± 13.7 *,**	246.05 ± 14.36	0.000	0.015	0.121	<0.001
Manganese (mg) ǂ	0.02 ± 0.11	−0.1 ± 0.13	0.27 ± 0.12 *	−0.21 ± 0.12	0.021	0.040	0.356	0.013
Phosphorus (mg)	1064.74 ± 44.16	1158.97 ± 49.36	1130.12 ± 45.3	1034.8 ± 49.02 **	0.035	0.993	0.160	<0.001
Potassium (mg)	2203.59 ± 70.48	2215.49 ± 79.05	2226.37 ± 73.17	2003.72 ± 78.24 **	0.090	0.272	0.023	0.005
Zinc (mg)	8.78 ± 0.36	9.09 ± 0.4	9.38 ± 0.38	8.2 ± 0.4	0.043	0.366	0.545	0.002
Refined grain (oz)	3.05 ± 0.18	2.76 ± 0.21	2.75 ± 0.2 *	4.08 ± 0.2 **	<0.001	0.018	0.001	<0.001
Nuts, seeds, soy, legumes (oz) ǂ	−0.03 ± 0.13	−0.29 ± 0.15	0.81 ± 0.14 *,**	−0.54 ± 0.15	<0.001	<0.001	0.007	<0.001
HEI-2015 index	63.46 ± 1.5	63.96 ± 1.68	65.7 ± 1.57 *	56.41 ± 1.66	<0.001	0.024	0.013	<0.001
Mediterranean index	4.32 ± 0.27	3.93 ± 0.3	4.75 ± 0.28 *	3.53 ± 0.3	0.017	0.021	0.938	0.038
DASH index	24.04 ± 0.8	23.78 ± 0.89	25.2 ± 0.83	22.41 ± 0.88	0.003	0.159	0.818	0.007

Data are presented as model-adjusted least squares means ± SE. ^ǂ Indicates JN-transformed variable means ± SE and p-values. * Mvt-adjusted pairwise p-value < 0.05 for almond vs. cookie within the same timepoint, ** Mvt-adjusted pairwise p-value < 0.05 for BL vs. W6 within the same group. Repeated measures were analyzed using linear mixed-effects models (fixed effects: group, week, group × week; random intercept: participant) with energy intake included as a covariate. The Group, Week, and Group × Week columns report type III tests of fixed effects from these models; pairwise comparisons used multivariate-t (mvt) adjustment. Baseline-adjusted group reports the group effect at week 6 from a separate baseline-adjusted linear regression. Missing data were handled by maximum likelihood (mixed models) or multiple imputation (regression models).

3.5. Immune and Inflammatory Marker Assessments

Overall, there were no group differences in immune markers over the 6-week intervention (Figure 2). However, almond consumption for 6 weeks resulted in lower concentrations of IL-6, TNF-alpha and IFN-gamma compared with cookie (BL-adjusted group effect, p < 0.05 Figure 3, Table S3). No baseline-adjusted group differences were observed for the anti-inflammatory marker IL-10 at week 6; however, a statistically significant group × week interaction indicated that the increase in IL-10 over time was greater in the almond group compared with the cookie group. Analyses for IL-10 were conducted in a subset of participants due to quality control issues affecting two assay plates that rendered those data unusable, and results should therefore be interpreted cautiously due to reduced statistical power (Table S3).

Figure 2.

Immune markers in the almond and cookie groups at baseline and week 6 in adults with obesity. (A) Absolute basophil count. (B) Basophils (%). (C) Absolute eosinophil count. (D) Eosinophils (%). (E) Absolute monocyte count. (F) Monocytes (%). (G) Absolute neutrophil count. (H) Neutrophils (%). (I) Absolute lymphocyte count. (J) Lymphocytes (%). (K) White blood cell count. Non-transformed data are presented as mean ± SE. (Blue: Almond, Red: Cookie).

Figure 3.

Inflammatory markers in the almond and cookie groups at baseline and week 6 in adults with obesity. (A) TNF-alpha. (B) IFN-gamma. (C) IL-6. (D) IL-10. Non-transformed data are presented as mean ± SE. * BL-adjusted group effect p-value < 0.05. (Blue: Almond, Red: Cookie).

4. Discussion

Almond consumption for six weeks lowered serum concentrations of pro-inflammatory cytokines, including IL-6, TNF-α, and IFN-γ compared to an isocaloric ultra-processed snack of cookies. These improvements in inflammatory markers occurred under weight-maintenance conditions and were accompanied by improvements in overall diet quality and higher acceptance of almonds relative to cookies. Interestingly, other than the tendency of a decline in HDL-C with cookie consumption, this ultra-processed food did not impact other physiological or inflammatory markers. Since no changes in other cardiometabolic markers were observed over the six-week period, the pattern supports the interpretation that changes in low-grade inflammation may occur earlier and potentially precede detectable changes in cardiometabolic markers, which may require longer durations to manifest [52,53].

The observed reductions in IL-6, TNF-α, and IFN-γ suggest that daily almond intake may help shift the balance of systemic inflammatory signaling toward a less pro-inflammatory state in adults with obesity. Previous meta-analyses examining the effects of almond consumption on inflammatory markers have yielded mixed results. One meta-analysis found that almond intake significantly reduced circulating CRP concentrations in adults across varying BMI and health statuses, but did not report obesity-specific subgroup effects [18]. Another found significant reductions in IL-6 overall, but these effects were more prominent in participants without obesity, however, with substantial heterogeneity observed within that subgroup [20]. A third meta-analysis found that while CRP and IL-6 were significantly lowered overall, these changes were not significant for individuals with obesity, and no significant effect overall was observed for TNF-α [19]. While the present study did not assess CRP, the magnitude and consistency of changes across cytokines (29–47% greater improvements in the almond group) suggest a stronger effect than previously reported in smaller or more heterogeneous populations. These effects were observed under free-living, weight-stable conditions, reinforcing that the nutrient composition of almonds, rather than energy restriction, mediated the cytokine shifts. These shifts are biologically plausible given both the specific nutrient changes and the overall improvements in diet quality documented in this study. Participants who consumed almonds had higher intakes of monounsaturated fatty acids, primarily oleic acid, which are known to reduce pro-inflammatory lipid mediators when they displace dietary saturated fats [54,55,56].

Almond consumption also increased intakes of key minerals, including magnesium, zinc, and manganese which synergistically may help support healthy immune and antioxidant responses that counteract chronic low-grade inflammation [18,57,58]. The almond group also had higher total fiber intake, driven mainly by the naturally high insoluble fiber content of almonds (approximately 80% of their total fiber (16)). Current evidence on how insoluble fiber influences inflammation, however, remains limited [59]. Collectively, these nutrient shifts were reflected in significantly higher HEI, MED, and DASH dietary pattern scores for the almond group compared to the cookie group. These dietary improvements with almond consumption likely acted through multiple complementary and synergistic pathways to produce the observed pattern of improved inflammation.

Notably, the almond intervention did not affect systemic innate or adaptive immune cell counts. This aligns with current understanding that obesity-related chronic inflammation is primarily driven by local immune cell infiltration and activation within adipose tissue which contributes to systemic inflammation [60,61,62], but does not necessarily cause large fluctuations in total circulating immune cell numbers over short periods. Although dietary patterns and weight loss can modulate this local immune environment and shift adipose tissue macrophage phenotypes [61], detecting corresponding changes in circulating blood counts may be challenging when there is no significant weight reduction, acute inflammatory challenge, or overt disease. While there are limited data on immune cells from nut-based studies, a systematic review found that vegetarian dietary patterns are associated with lower leukocyte counts in observational studies, yet intervention trials show inconsistent effects [63]. In one eight-week DASH diet study, reduced lymphocyte counts were observed, but these occurred alongside weight and fat loss, suggesting that adiposity change may mediate the effect [64]. Similarly, in a crossover trial, 4-week consumption of soy pudding modestly altered regulatory T cell counts, and in a 12-week RCT of a Brazilian cardioprotective diet with extra virgin olive oil, modest reductions in total leukocytes and lymphocyte-to-monocyte ratios were noted [65,66]. Collectively, these studies highlight that weight loss and intervention duration strongly influence total leukocyte counts, whereas shifts in immune cell subsets can occur with diet changes even without weight loss. Subtle immunological shifts may be better captured via adipose tissue analyses, gene expression profiling of isolated innate and adaptive cell subsets, and functional immune assays.

Almond consumption increased intake of alpha-tocopherol (vitamin E), a lipid-soluble antioxidant [67]. However, no significant differences in serum alpha-tocopherol concentrations were observed between groups over the six-week period. This is consistent with previous findings suggesting that modest dietary increases may not rapidly alter circulating levels in individuals with normal baseline concentrations due to tight homeostatic regulation and a saturable alpha-tocopherol pool [33]. Moreover, several factors may contribute to variability in circulating responses to intake, including substantial inter-individual heterogeneity in absorption efficiency (estimated ~10–79%) [68], influences of baseline physiological and inflammatory status [69], and due to its fat-soluble nature, potential sequestration in liver fat [70], and greater partitioning into adipose tissue [71]. Collectively, these factors may attenuate detectable changes in serum concentration in individuals with obesity despite increased intake.

High compliance rates and higher acceptance scores for almonds compared to cookies underscore the feasibility of incorporating almonds into habitual eating patterns. Interestingly, cookies might have been expected to be more palatable due to their sweetness and ultra-processed characteristics, but there were no differences in palatability, and participants rated almonds as more acceptable overall. Appetite ratings remained stable with both almond and cookie consumption. Importantly, body weight remained stable in both groups, indicating effective energy compensation for both the ultra-processed cookie snack and almonds over 6 weeks. The findings are consistent with studies supporting that nuts do not contribute to positive energy balance when incorporated into a free-living diet [72,73,74].

5. Conclusions

The strengths of this study include its controlled design, high participant adherence, and stringent BMI and age criteria. However, the sample size, while adequate to detect changes in cytokine concentrations, may have limited power to detect changes in immune cell distributions. Other limitations pertained to quality control issues with the IL-10 assay, which precluded detection of true effects due to potential underpowering. The relatively short duration precludes conclusions about long-term sustainability of these effects. Moreover, participants were generally healthy aside from obesity, and effects in populations with more advanced metabolic disease may differ. As no a priori hypothesis regarding sex differences was specified, recruitment was not stratified by sex, limiting the ability to draw conclusions about sex-specific effects. In conclusion, this study provides new evidence that daily almond consumption can improve inflammatory profiles in middle-aged adults with obesity, independent of weight change and under free-living conditions. These results, together with the observed improvements in diet quality, support including almonds as a practical component of dietary strategies to help attenuate obesity-related inflammation. Longer-term studies with larger samples may be needed to determine whether the observed improvements in inflammation mediate improvements in cardiometabolic outcomes in obesity. Additionally, functional immunological analyses and immunophenotyping could provide further insight into the mechanisms underlying almond-induced improvements in inflammation. Future studies should also be specifically designed and powered to evaluate sex-based differences.

Abbreviations

The following abbreviations are used in this manuscript:

DASH	Dietary approaches to stop hypertension
HEI	Healthy eating index
IFN	Interferon gamma
IL	Interleukin
MED	Mediterranean
TNF	Tumor necrosis factor

Supplementary Materials

The following supporting information can be downloaded at https://www.mdpi.com/article/10.3390/nu18050875/s1, Table S1: Nutrient composition of almond and cookie study snacks in the 6-week randomized controlled trial in adults with obesity; Table S2. Self-reported dietary intake (energy-adjusted) from the ASA-24 dietary recalls for the almond and cookie groups at baseline and week 6 in adults with obesity; Table S3: Immune and inflammatory markers for the almond and cookie groups at baseline and week 6 in adults with obesity. Reference [75] is cited in Supplementary Materials.

Author Contributions

Conceptualization, J.D. and V.V.-P.; methodology, J.D. and V.V.-P.; investigation, J.D., P.B. and A.A.; formal analysis, J.D., A.A. and E.R.; data curation, J.D., A.A. and E.R.; writing—original draft preparation, J.D., A.A. and E.R.; writing—review and editing, all authors; supervision, J.D.; project administration, J.D.; funding acquisition, J.D. All authors have read and agreed to the published version of the manuscript.

Institutional Review Board Statement

The study was conducted in accordance with the Declaration of Helsinki and approved by the Institutional Review Board of University of Missouri–Columbia (protocol code 2090290 and date of approval: 11 May 2022).

Informed Consent Statement

Informed consent was obtained from all subjects involved in the study.

Data Availability Statement

The data presented in this study are available on request from the corresponding author due to ongoing analyses related to this research.

Conflicts of Interest

J.D. received funding from the Almond Board of California for this study. V.V.-P. is co-investigator on the grant. Other authors have no disclosures. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript; or in the decision to publish the results.

Funding Statement

This research was funded by Almond Board of California.