Effects of Prune (Dried Plum) Supplementation on Cardiometabolic Health in Postmenopausal Women: An Ancillary Analysis of a 12-Month Randomized Controlled Trial, The Prune Study

Abstract

Background

Estrogen withdrawal during menopause is associated with an unfavorable cardiometabolic profile. Prunes (dried plums) represent an emerging functional food and have been previously demonstrated to improve bone health. However, our understanding of the effects of daily prune intake on cardiometabolic risk factors in postmenopausal women is limited.

Objectives

We conducted an ancillary investigation of a randomized controlled trial (RCT), The Prune Study, to evaluate the effect of 12-mo prune supplementation on cardiometabolic health markers in postmenopausal women.

Methods

The Prune Study was a single-center, parallel-design, 12-mo RCT in which postmenopausal women were allocated to no-prune control, 50 g/d prune, or 100 g/d prune groups. Blood was collected at baseline, 6 mo, and 12 mo/post to measure markers of glycemic control and blood lipids. Body composition was assessed at baseline, 6 mo, and 12 mo/post using dual-energy X-ray absorptiometry. Linear mixed-effects models were used to evaluate the effect of time, treatment, and their interaction on cardiometabolic health markers, all quantified as exploratory outcomes.

Results

A total of 183 postmenopausal women (mean age, 62.1 ± 4.9 y) completed the entire 12-mo RCT: control (n = 70), 50 g/d prune (n = 67), and 100 g/d prune (n = 46). Prune supplementation at 50 g/d or 100 g/d did not alter markers of glycemic control and blood lipids after 12 mo compared with the control group (all P > 0.05). Furthermore, gynoid percent fat and visceral adipose tissue (VAT) indices did not significantly differ in women consuming 50 g/d or 100 g/d prunes compared with the control group after 12 mo of intervention. However, android total mass increased by 3.19% ± 5.5% from baseline in the control group, whereas the 100 g/d prune group experienced 0.02% ± 5.6% decrease in android total mass from baseline (P < 0.01).

Conclusions

Prune supplementation at 50 g/d or 100 g/d for 12 mo does not improve glycemic control and may prevent adverse changes in central adiposity in postmenopausal women.

This trial was registered at clinicaltrials.gov as NCT02822378.

Introduction

Cardiovascular diseases (CVDs) are consistently ranked as the leading cause of mortality worldwide, accounting for 18.6 million global deaths in 2019 [1]. In the United States alone, CVD currently affects 127.9 million adults [2], and this prevalence is estimated to rise to 131.2 million by 2035, representing nearly half of the United States population [3]. Between 2018 and 2019, the estimated direct and indirect costs of total CVD were $407.3 billion [2] and are projected to increase over 2-fold to $1.1 trillion by 2035 [3]. The prevalence of CVD is higher in men than in women for most age groups [4]. However, this disparity is reduced at midlife, which coincides with the onset of menopause. After 55 y of age, CVD accounts for greater all-cause mortality in women than in men [5]. Overall, the high prevalence, mortality, and annual economic costs underscore CVD as a significant public health burden in United States adults, particularly in postmenopausal women, who represent an important at-risk population for cardiometabolic diseases [6].

Epidemiological evidence from 1054 women in the Study of Women’s Health Across the Nation [7] demonstrates that several cardiometabolic risk factors, including blood apolipoprotein B, total cholesterol, and LDL cholesterol levels, are significantly elevated within the first 12 mo before and after the final menstrual period. Furthermore, menopause progression is associated with a higher prevalence of metabolic syndrome [8,9], which comprises a cluster of components, i.e., insulin resistance, dyslipidemia, hypertension, and central obesity [10]. Several studies leveraging data from the Study of Women’s Health Across the Nation report that the menopause transition is not only associated with total body fat mass gain and lean mass loss but also with increased regional body fat parameters, including android, gynoid, and visceral fat [11,12]. Taken together, these findings suggest that menopause is linked to an unfavorable cardiometabolic profile.

A decline in endogenous estradiol levels resulting from ovarian senescence is thought to be an important factor contributing to increased cardiometabolic risk in women advancing across the menopause transition [5,6]. The protective effects of estrogen on the cardiovascular system are mediated through several physiological pathways, such as inhibiting chronic inflammation and oxidative stress, upregulating the endothelial nitric oxide production and improving vasodilation, and regulating glucose and lipid metabolism [13,14], thus supporting the hypothesis that menopausal hormone therapy (MHT) may prevent CVD events. However, intent-to-treat analyses from randomized controlled trials (RCTs) of the Heart Estrogen/Progestin Replacement Study [15] and the Women’s Health Initiative [16] failed to demonstrate cardioprotective benefits from MHT. Post hoc findings revealed that MHT tended to confer reduced CVD risk in women only when initiated within the first menopausal decade but not in women beyond 10 y since menopause [17], suggesting that a therapeutic window exists for hormonal therapy during the menopause transition. Thus, there is a need to find suitable, nonpharmacological treatment approaches to reduce the progression of cardiometabolic disease, particularly in older postmenopausal women.

Diet is an important modifiable risk factor for chronic diseases [18], with long-standing evidence indicating that a higher fruit and vegetable intake is inversely associated with CVD risk [19]. In particular, dietary intake of flavonoid-rich foods, such as berries, is associated with reduced risk of CVD mortality [20], which may be attributed, in part, to polyphenol-mediated improvements in markers of dyslipidemia, glycemic control, oxidative stress, and inflammation [[21], [22], [23]]. Dietary guidelines recommend the consumption of fruits in whole-food, nutrient-dense forms, for example, dried fruits at a serving of a half cup/d [24]. Although consumer concerns exist around the sugar content in dried fruits, by definition, traditional dried fruits such as dried figs, dates, raisins, apricots, and prunes (dried plums), contain no added sugar but only naturally occurring simple sugars. For example, the total sugar composition of fresh plums includes glucose, fructose, and sucrose, whereas, during processing into prunes, the sucrose content is reduced due to hydrolysis into glucose and fructose [25]. Thus, dehydration and sucrose hydrolysis during processing increases the total glucose and fructose contents in dried plums compared to fresh plums. Despite their higher overall concentration of simple sugars, the glycemic index (GI) of prunes (GI = 29) is lower than that of their fresh counterparts (GI = 35), primarily due to higher fiber, including pectin and hemicellulose, as well as higher sorbitol contents [[25], [26], [27]]. Thus, the relatively lower GI of dried fruits may be partly attributed to the higher proportions of dietary fiber, fructose, and sorbitol, which delay the rate of gastric emptying and promote a lower postprandial release of blood glucose. In a recent systematic review and meta-analysis of 169 controlled-feeding trials (n = 10,357) evaluating the effect of various fructose-containing foods on adiposity, sugar-sweetened beverages consumed as excess energy increased adiposity in adults, whereas dried fruits consumed ≤50 g/d decreased adiposity [28], thus providing further evidence supporting dried fruits as a healthy, nutrient-dense snacking food.

Prunes (dried plums) have gained traction over the years as a functional food rich in vitamin K, fiber, potassium, boron, copper, and polyphenolic compounds, such as flavonoids, phenolic acids, and chlorogenic acids [25,29]. Numerous preclinical studies demonstrate a beneficial effect of prunes administered as either juices, concentrates, or prune-supplemented chow on cardiometabolic risk factors including hyperglycemia [30], triglycerides [30,31], and total cholesterol [32,33]. However, there are limited RCTs assessing the effect of prunes as a whole-food dietary intervention on CVD outcomes in postmenopausal women. In a crossover RCT of healthy postmenopausal women (n = 27), prune consumption at either 14 g/d or 42 g/d for 2 wk had no effects on systolic and diastolic blood pressure or microvascular function compared to baseline [34]. In a larger, 12-mo parallel-arm RCT of postmenopausal women with low bone mass (n = 160), 100 g/d prunes did not significantly alter body weight, total and LDL cholesterol, or atherogenic risk ratios compared to 75 g/d dried apples [35]. The Prune Study was a 12-mo parallel-design RCT [36] that aimed to investigate the effects of prunes at 2 doses, 50 g/d and 100 g/d, compared with a no-prune control group on bone mineral density (BMD), bone geometry, and bone strength in postmenopausal women. In our report of primary BMD endpoints [37], we observed a protective effect of 50 g/d prunes on total hip BMD in postmenopausal women. However, our understanding of the long-term effects of daily prune intake over the course of 12 mo on measures of fasting plasma glucose, insulin resistance, blood lipids, and central adiposity in this study population of postmenopausal women is limited.

Therefore, the purpose of the current investigation is to perform an ancillary analysis of the parent RCT evaluating the dose–response effects of daily prune intake for 12 mo on select cardiometabolic health markers and regional fat distribution in postmenopausal women who completed the entire trial. We hypothesized that both 50 g/d and 100 g/d prune supplementation for 12 mo would improve measures of glycemic control, blood lipids, and central adiposity in postmenopausal women.

Methods

Study design

The Prune Study (Clinical Trial Number NCT02822378) was a single-center, multi-arm, parallel-design, 12-mo RCT conducted at The Pennsylvania State University between June 2016 and February 2021. This RCT was designed to examine the efficacy of prunes at 2 doses, 50 g/d and 100 g/d, as a whole-food, nutritional intervention compared with a “no-prune” control group on bone outcomes in postmenopausal women. Full study protocols [36] and the results of the primary BMD endpoints [37] are reported elsewhere. The current investigation is an ancillary analysis of the parent RCT that aimed to evaluate the effects of prune intake (50 g/d and 100 g/d) compared with no-prune control for 12 mo on cardiometabolic health markers, collected as prespecified exploratory outcomes, in postmenopausal women who completed the entire trial. Briefly, participants were recruited using fliers/advertisements and completed preliminary screening via telephone interview. Subjects meeting initial eligibility criteria were scheduled for an in-person screening visit at The Women’s Health and Exercise Laboratory. After providing written informed consent, participants underwent several screening procedures, including a physical examination, dual x-ray absorptiometry (DXA) scan for BMD assessment; completed medical and health history questionnaires; and provided a fasting blood draw (for comprehensive metabolic and lipid panel), and results from these screening tests were reviewed to ascertain eligibility. Participants enrolled in the study at baseline were randomly allocated (1:1:1) to 1 of the 3 treatment groups, no-prune control, 50 g/d prune, or 100 g/d prune groups, using a computer-generated list maintained by study staff. Participants and clinic staff were not blinded to the treatment allocation, but all outcome collectors and assessors were blinded to the allocation. Cardiometabolic health outcomes, including glycemic control markers, blood lipid profile, body composition, and regional fat distribution, assessed in the current ancillary study were measured at baseline, 6 mo, and 12 mo (post); details of the overall RCT design are outlined in the study protocol [36]. This study was conducted in accordance with the guidelines of the Declaration of Helsinki and approved by the Pennsylvania State University Institutional Review Board (STUDY00004252). This study complies with the CONSORT reporting guidelines [38].

Participants and eligibility

We recruited postmenopausal women aged 55 to 75 y, with ≥12 consecutive months of amenorrhea (determined by self-report questionnaire); without severe obesity [BMI (kg/m²) <40]; healthy (determined by a screening questionnaire and complete metabolic panel); willing to include prunes in their daily diet; not taking any natural dietary supplement containing phenolics or <1 cup/d of blueberries/apples for ≥2 mo prior to study entry; nonsmoking and ambulatory; and had eligible areal BMD values from DXA assessments at the lumbar spine, total hip, and/or femoral neck corresponding to T-scores between 0.0 and −3.0. Exclusion criteria have been described in detail elsewhere [36]. Briefly, women were excluded if they 1) were regular consumers of prunes, dried apples, prune juice, or blueberries (≥1 cup/d); 2) had a history of vertebral fracture or fragility fractures after 50 y of age; 3) had untreated or current hyper- or hypoparathyroidism; 4) had a history of heart attack, stroke, thromboembolism, kidney disease, malabsorption syndrome, or seizure disorders; or 5) were taking any medications (hormonal, osteoporosis, or other) within 12 mo of study participation that would interfere with primary BMD outcomes.

Nutritional intervention

Eligible participants were randomly allocated to 1 of 3 groups: no-prune control, 50 g/d prune, or 100 g/d prune groups. Detailed descriptions of prune interventions and run-in plans to achieve the desired dose are described in the study protocol [36]. Briefly, participants assigned to either of the prune interventions (Prunus domestica L.) were administered with prune packages provided by The California Prune Board every 3 mo. The nutrient profiles of the 50 g/d prune and 100 g/d prune interventions are outlined elsewhere [37]. The 50-g dose represents a serving size of about 4 to 6 prunes, whereas the higher, 100-g dose is equivalent to about 10 to 12 prunes. The control group did not receive prune supplementation (0 g/d). These 2 prune doses were selected based on previous RCTs ranging from 3 to 12 mo that demonstrated positive findings on bone turnover markers and/or BMD in postmenopausal women after supplementation with either 50 g/d or 100 g/d prunes [[39], [40], [41], [42]]. Participants were counseled to gradually incorporate prunes into their habitual diets during a 2-wk “run-in” period, after which they were instructed to consume the prescribed prune dose at any time of the day provided that the entire amount was consumed daily during the 12-mo intervention period. Participants were also instructed to maintain their background diet throughout the study duration. To monitor compliance, participants recorded the days and number of prunes consumed on a daily supplement log as well as document any adverse events, including bloating, cramping, and gas. Participants were considered compliant to the intervention if they consumed >80% of prescribed prune dose during the 12-mo RCT. All participants received 1200 mg calcium and 800 IU vitamin D₃ daily from diet plus supplements (Nature Made Pharmavite LLC) as standard-of-care to meet the recommended dietary allowance of calcium and vitamin D₃ in women aged over 50 y [43].

Demographics and anthropometrics

Demographic, medical, and reproductive histories were evaluated using in-house questionnaires, which included questions regarding use of supplements, medications, menstrual history, and physical activity [36]. Total body weight was measured to the nearest 0.5 kg on a physician’s scale (Seca, Model 770). Height was measured to the nearest 0.1 cm using a calibrated stadiometer. BMI was calculated as total body weight divided by height squared.

Dietary intake assessment

Dietary intake was assessed by implementing 3-d diet recall diaries administered over 2 nonconsecutive weekdays and 1 weekend day at baseline and at the end of the 12-mo intervention (post). Participants were instructed to measure and record in detail all food and beverages consumed using standard measurement tools at various intervals throughout the day, i.e., breakfast, morning snack, lunch, afternoon snack, and dinner. The Nutritionist Pro Diet Analysis software (Axxya Systems) was used to code and analyze nutrient data from the 3-d diet logs at baseline and 12 mo for daily intake of total kilocalories, macronutrients, vitamins, and minerals.

Blood sample collection and storage

Blood samples were collected at baseline, 6-mo, and 12-mo study visits. All participants were instructed to fast overnight for ≥12 h prior to the blood draw. Antecubital venous blood (10 mL) was drawn into different blood collection vacutainers (BD Biosciences); details are described in the study protocol [36]. Briefly, serum was collected using blood drawn into sterile clot activator/separator gel-coated blood tubes, allowed to clot for 30 min at room temperature (20°C–24°C), and then centrifuged (Eppendorf 5804R) for 15 min at 4°C. Plasma was collected using blood drawn into sterile sodium heparin blood tubes that were centrifuged immediately at 1800 × g for 15 min at room temperature. Serum and plasma samples were dispensed into 2-mL polyethylene storage tubes, and all aliquots were frozen at −80°C until analysis.

Blood biochemistry assessments

Frozen plasma and serum aliquots collected at baseline, 6 mo, and 12 mo were sent to a commercial laboratory (Quest Diagnostics) for a comprehensive metabolic and lipid panel. Heparinized plasma was analyzed for fasting blood glucose (mg/dL) using spectrophotometry (Quest Diagnostics). Fasting plasma insulin (μIU/mL) was quantified using a chemiluminescent immunoassay [Insulin Kit SMN 10381429; intra-assay coefficient of variation (CV) ≤8.0%] on an IMMULITE 1000 Immunoassay System (Siemens Healthcare Diagnostics). Each assay was performed in duplicate. Insulin resistance was estimated using the HOMA-IR equation [44]: HOMA-IR = [fasting plasma glucose (mg/dL) × fasting plasma insulin (μIU/mL)]/405. Serum triglycerides (mg/dL), total cholesterol (mg/dL), and HDL cholesterol (mg/dL) were quantified using spectrophotometry (Quest Diagnostics), whereas LDL cholesterol (mg/dL) was calculated from the standard lipid profile using the Martin–Hopkins equation [45]. Serum concentrations (IU/L) of liver function enzymes, alkaline phosphatase (ALP), aspartate aminotransferase (AST), and alanine aminotransferase (ALT), were quantified using spectrophotometry (Quest Diagnostics).

Body composition

Body composition was assessed at baseline, 6 mo, and 12 mo using DXA, which is a useful tool to measure central adiposity [46]. All participants underwent total body DXA scans on a Hologic QDR4500 system performed by an International Society for Clinical Densitometry certified bone densitometry technologist (laboratory precision of ≤1.1% CV for body composition). Participants were requested to remove all metal objects to reduce interference with the scan and were positioned supine in the center of the DXA scanning platform aligned with the long axis of the scanner. Several DXA-acquired body composition variables were analyzed: fat mass (kg); lean body mass (kg); fat-free mass (kg); regional fat distribution, i.e., total mass (kg), percent fat (%), and fat mass (kg), at the android and gynoid regions of interest; estimated VAT mass (kg), VAT volume (cm³), and VAT area (cm²); android/gynoid percent fat ratio; fat mass index (kg/m²); and lean mass index (kg/m²).

Statistical analyses

The parent RCT was designed to achieve 80% power for detecting the minimum relevant difference (i.e., minimum detectable effect size) in the primary endpoint, percent change from baseline in areal BMD at the lumbar spine and/or total hip, with a 2-sided type 1 error rate of 5% and anticipated 20% attrition rate (details summarized elsewhere [36]). The current study is an ancillary analysis of prespecified exploratory outcomes focused on select cardiometabolic health markers and regional body fat distribution in postmenopausal women and was thus conducted in all randomized participants who completed the 12-mo intervention, had data for prespecified exploratory cardiometabolic health outcomes, and were included in the analysis regardless of compliance to intervention.

Statistical analyses were performed using SAS (Statistical Analysis System, version 9.4). Descriptive univariate statistics (PROC UNIVARIATE) and statistical assumptions for parametric tests were conducted on all outcomes. Data distribution was assessed for normality using the Shapiro–Wilk test. Data points that were >3 SDs from the mean were considered as outliers and removed. Homoskedasticity was confirmed by assessing residual-versus-predicted value plots, and normality of the residuals were visually inspected using normal probability (Q-Q) plots. Natural logarithm transformation was applied, when necessary, and if transformations were unsuccessful at normalizing data distribution, nonparametric tests were used.

Primary analyses

Primary analyses of prespecified exploratory cardiometabolic health outcomes were conducted in all randomized participants who completed the 12-mo intervention, had data for prespecified exploratory cardiometabolic health outcomes, and were included in the analysis regardless of compliance to intervention. To compare the effects of the intervention on all cardiometabolic health outcomes, a linear mixed-effects (LME) model with repeated measures design (PROC MIXED) was used to fit the longitudinal observations at 3 time points (baseline, 6 mo, and 12 mo) during the study with random subject-level intercept and fixed effects of time, treatment, and their interaction (treatment × time). Mann–Whitney U tests were used to assess whether there were baseline differences between participants with and without missing data for the outcomes. Although some of the variables appeared to be different, they did not reach statistical significance (Supplemental Table 1), and thus, the missing data were unlikely to bias the treatment effect. Therefore, all available observations were included in this analysis without imputation for missing observations. Change in android adiposity was included as a covariate in mixed models assessing markers of glycemic control and blood lipid profiles. As a sensitivity analysis, for participants who completed the entire 12-mo intervention (i.e., completers only), percent change from baseline to 12 mo was calculated by subtracting measurements taken at baseline from postintervention values, then dividing this difference by the baseline value, and multiplying the result by 100. An LME model was used to test the effect of treatment on percent change endpoints, wherein participants were included as a random effect and baseline values for each endpoint were included as a covariate. The effect of treatment on nonnormally distributed percent change endpoints was analyzed using Kruskal–Wallis test (PROC NPAR1WAY). The analysis of longitudinal data for body mass, fat mass, and body fat percentage have been previously reported in the primary endpoint article [37], and thus, in the current ancillary study, for these 3 variables, only their percent change from baseline endpoints were analyzed. Furthermore, diet intake data for only total energy, protein, carbohydrate, and fat have been previously reported in the primary endpoint article [37]; this ancillary study extends these diet analyses by including additional macronutrient and micronutrient variables to provide a more comprehensive assessment of nutritional intake.

Subgroup analyses

Four additional sets of planned analyses were conducted in a subset of participants: 1) stratified by baseline glycemic status, i.e., participants who were normoglycemic at baseline (fasting plasma glucose <100 mg/dL), representing a majority (over 75%) of the study population, and participants who had elevated fasting plasma glucose at baseline (fasting plasma glucose of ≥100 mg/dL). LME models were repeated in this subset to determine whether the prune intervention differentially affected blood glucose in postmenopausal women who had normoglycemia compared with elevated fasting plasma glucose at baseline; 2) stratified by baseline BMI, i.e., participants who were normal weight at baseline (BMI: 18.5–24.9), representing nearly half of the study population (48.1%), and participants with overweight/obesity (BMI >24.9). LME models were repeated in this subset to determine whether the prune intervention differentially affected BMI in postmenopausal women who had normal weight BMI compared with overweight/obesity at baseline; 3) stratified by baseline total dietary fiber intake: participants who were not meeting their daily adequate intake [24,47] at baseline were grouped in the low-fiber category, i.e., having total dietary fiber intake <22 g/d at baseline. LME models were repeated in this subset to determine whether the prune intervention differentially affected total dietary fiber intake in women who were below the recommended daily adequate intake at baseline; 4) stratified by baseline cardiometabolic risk categories: elevated cardiometabolic risk: participants at baseline with overweight and/or obesity (BMI >24.9) and ≥1 additional elevated cardiometabolic risk factor: fasting plasma glucose ≥100 mg/dL, total cholesterol ≥200 mg/dL, triglycerides ≥150 mg/dL, LDL cholesterol ≥100 mg/dL, and HDL cholesterol <50 mg/dL. LME models were repeated in this subset to determine whether the prune intervention differentially affected markers of glycemic control and blood lipid profiles in postmenopausal women who had low risk compared with elevated risk at baseline.

Model covariance structures for all LME models were selected based on optimizing fit statistics evaluated as the lowest Bayesian information criterion. Whenever main effects of time, treatment, or interaction were detected, appropriate post hoc tests, such as the Tukey–Kramer’s adjustment, were performed to correct for multiple comparisons. Graphs were plotted using GraphPad PRISM (version 9.4.1, GraphPad Software). Descriptive data of baseline characteristics across all 3 groups are reported as mean ± SD or proportion (%). LME-based group-specific longitudinal mean estimates in the figures and tables are reported as mean ± SEM. All tests were 2-sided, and for all endpoints, P < 0.05 was considered statistically significant.

Results

Participant flow and baseline characteristics

A detailed description of participant flow throughout the parent RCT are summarized in the primary endpoint article [37]. Briefly, among 638 women who were screened for eligibility, 235 were enrolled at baseline and randomly assigned to 1 of the 3 intervention arms: control (n = 78), 50 g/d prune (n = 79), or 100 g/d prune (n = 78) (Figure 1). Overall, the parent RCT had a 22% attrition rate, which was notably highest (41%) in the 100 g/d prune group, primarily due to poor tolerance to the intervention, time commitment, or lost to follow-up [37]. A total of 183 participants completed the entire 12-mo intervention: control (n = 70), 50 g/d prune (n = 67), and 100 g/d prune (n = 46). A majority of these participants who completed the trial had >90% compliance to the assigned intervention in all treatment groups [37].

FIGURE 1.

CONSORT flow diagram of participants included in the ancillary analysis.

Due to the exploratory nature of the analyses included in the current ancillary investigation of The Prune Study, only a subset of participants was analyzed for fasting plasma insulin at the 6-mo study visit, and thus the sample sizes for insulin are 181 participants at baseline [control (n = 69), 50 g/d prune (n = 67), and 100 g/d prune (n = 45)], 124 participants at 6 mo [control (n = 47), 50 g/d prune (n = 43), and 100 g/d prune (n = 34)], and 181 participants at 12-mo/post [control (n = 69), 50 g/d prune (n = 67), and 100 g/d prune (n = 45)]. The remaining exploratory cardiometabolic health outcomes included in this ancillary analysis were completed in 235 participants at baseline [control (n = 78), 50 g/d prune (n = 79), and 100 g/d prune (n = 78)], 186 participants at 6 mo [control (n = 67), 50 g/d prune (n = 65), and 100 g/d prune (n = 54)], and 183 participants at 12-mo/post [control (n = 70), 50 g/d prune (n = 67), and 100 g/d prune (n = 46)].

Descriptive baseline characteristics of the demographics in 235 postmenopausal women enrolled in the parent RCT are reported elsewhere [37]. For the current ancillary study, descriptive baseline characteristics of blood biochemistry and body composition parameters among 183 randomized participants who completed the 12-mo RCT were mostly balanced across the 3 randomized groups (Table 1). The majority (79.8%) of participants (mean age, 62.2 ± 4.9 y; BMI, 25.7 ± 4.1) were in their late postmenopausal stage, i.e., >8 y since their last menstrual period [48], thus representing a study population at high risk of CVD [6]. Most participants enrolled at baseline were healthy based on cardiometabolic markers, characterized as having normoglycemia (79.8%) and optimal levels of serum HDL cholesterol (86.3%) and triglycerides (87.8%) (Supplemental Table 2); however, mean total cholesterol and LDL cholesterol levels at baseline were elevated.

TABLE 1.

Baseline characteristics of postmenopausal women who completed the trial and by group randomization

Characteristic1	Completers (n = 183)	Control group (n = 70)	50 g/d Prune (n = 67)	100 g/d Prune (n = 46)
Demographics
Age (y)	62.2 ± 4.9	62.0 ± 4.8	62.3 ± 4.6	61.9 ± 5.5
Age at menopause (y)	50.3 ± 4.9	50.2 ± 4.9	50.7 ± 4.8	50.0 ± 4.9
Time since menopause (y)	11.7 ± 6.8	11.8 ± 6.9	11.4 ± 6.3	11.8 ± 7.5
Height (cm)	162.9 ± 5.8	163.6 ± 5.7	162.0 ± 5.8	162.9 ± 5.7
Body mass (kg)	68.2 ± 10.9	67.0 ± 10.9	69.3 ± 11.1	68.3 ± 10.7
BMI (kg/m2)	25.7 ± 4.1	25.0 ± 3.8	26.4 ± 4.5	25.6 ± 3.6
BMI category2
Normal weight	48.1%	52.8%	44.7%	45.6%
Overweight	36.6%	32.8%	35.8%	43.4%
Obesity	15.3%	14.4%	19.5%	11%
Blood biochemistry
Fasting plasma glucose (mg/dL)	94.3 ± 14.3	92.8 ± 9.3	93.9 ± 11.0	97.2 ± 22.4
Glycemic status3
Normoglycemia	79.8%	81.4%	80.6%	76.1%
Prediabetes	18.0%	18.6%	16.4%	19.6%
Type II diabetes mellitus	2.2%	0%	3.0%	4.3%
Fasting plasma insulin (μIU/mL)	8.3 ± 7.0	6.9 ± 4.8	9.4 ± 7.0	8.8 ± 9.3
HOMA-IR	2.0 ± 1.8	1.6 ± 1.3	2.3 ± 2.0	2.2 ± 2.2
Total cholesterol (mg/dL)	204.3 ± 33.7	206.1 ± 27.5	206.1 ± 37.1	199.1 ± 37.0
HDL cholesterol (mg/dL)	66.8 ± 16.7	67.1 ± 15.5	66.9 ± 18.3	66.3 ± 16.4
LDL cholesterol (mg/dL)	117.6 ± 28.7	119.8 ± 25.4	117.3 ± 28.8	114.7 ± 33.3
Triglycerides (mg/dL)	98.3 ± 52.7	94.2 ± 39.9	107.6 ± 69.4	90.9 ± 38.8
Body composition
Total body fat mass (kg)	27.7 ± 7.7	26.8 ± 8.3	28.8 ± 7.6	27.5 ± 6.7
Total body lean mass (kg)	37.2 ± 4.2	36.9 ± 3.8	37.2 ± 4.1	37.4 ± 4.5
Total body fat-free mass (kg)	38.9 ± 4.1	39.0 ± 4.0	39.1 ± 4.3	38.5 ± 4.1
Percent body fat (%)	40.8 ± 5.8	39.9 ± 6.7	41.6 ± 5.2	40.6 ± 4.5
Android total mass (kg)	4.9 ± 1.1	4.7 ± 1.1	5.3 ± 1.2	4.7 ± 1.1
Gynoid total mass (kg)	10.9 ± 1.8	10.6 ± 1.7	11.3 ± 1.9	10.7 ± 1.9
A:G ratio	0.89 ± 0.16	0.86 ± 0.15	0.92 ± 0.16	0.87 ± 0.16
VAT mass (kg)	0.58 ± 0.29	0.52 ± 0.26	0.68 ± 0.32	0.53 ± 0.24
Fat mass index (kg/m2)	10.4 ± 3.0	9.6 ± 2.8	11.6 ± 3.3	9.9 ± 2.3
Lean mass index (kg/m2)	13.9 ± 1.3	13.7 ± 1.2	14.2 ± 1.4	13.9 ± 1.2

Abbreviations: A:G ratio, android-to-gynoid percent fat ratio; HDL, high-density lipoprotein; HOMA-IR, homeostatic model assessment for insulin resistance; LDL, low-density lipoprotein; SD, standard deviation; VAT, visceral adipose tissue.

¹Data are presented as mean ± SD or proportion (%).

²BMI classification (kg/m²): normal weight 18.5 ≤ BMI < 25; overweight 25 ≤ BMI < 30; obesity BMI ≥30.

³Glycemic status (mg/dL): normoglycemia: <100; prediabetes: 100–125; type II diabetes mellitus: ≥126.

Dietary intake

Average daily intake of total energy, macro- and micronutrients from 3-d dietary recall logs for baseline and 12-mo/post visits are summarized in Table 2. No significant treatment × time interactions were detected for total energy, carbohydrate, protein, and fat intakes. However, carbohydrate and potassium intake increased in all groups (main effect of time, P < 0.05), whereas the intake of monounsaturated fatty acids and vitamin E decreased in all groups (main effect of time, P < 0.05). Total dietary fiber intake (g/d) increased from baseline (21.53 ± 0.75) to 12-mo/post (23.12 ± 1.05) in all 3 groups (Table 2; main effect of time, F_(1,336) = 4.36, P = 0.03). Protein intake among the groups appeared to be different at baseline although statistical significance was not achieved. Furthermore, the subset of participants in the low-fiber category at baseline experienced a substantial increase in total dietary fiber intake from baseline (15.78 ± 0.36) to 12-mo/post (19.64 ± 0.47) (Supplemental Figure 1; main effect of time, F_(1,211) = 19.06, P < 0.001).

TABLE 2.

Daily nutrient intake calculated from 3-d diet records at baseline and 12-mo post intervention

Daily nutrient intake1	Control group		50 g/d Prune		100 g/d Prune		P values
	(mean ± SD)		(mean ± SD)		(mean ± SD)
	Baseline (n = 67)	Post (n = 63)	Baseline (n = 62)	Post (n = 61)	Baseline (n = 45)	Post (n = 44)	Treatment	Time	Treatment × Time
Total energy (kcal)	1838 ± 429	1829 ± 432	1767 ± 547	1759 ± 523	1697 ± 402	1775 ± 471	0.475	0.533	0.480
Macronutrients
Protein (g)	78.3 ± 19.6	76.8 ± 20.0	71.7 ± 18.6	70.3 ± 19.3	70.2 ± 21.1	68.2 ± 18.4	0.026	0.350	0.919
Carbohydrate (g)	206.4 ± 69.9	211.8 ± 61.9	193.5 ± 56.7	209.0 ± 68.8	195.5 ± 57.2	232.5 ± 95.9	0.555	<0.001	0.080
Dietary fiber (g)	20.4 ± 7.1	22.4 ± 8.0	21.1 ± 10.6	21.7 ± 8.6	23.0 ± 13.9	25.2 ± 13.5	0.319	0.037	0.672
Total fat (g)	75.4 ± 23.4	73.8 ± 23.0	75.9 ± 36.0	70.4 ± 28.4	68.8 ± 19.9	65.5 ± 20.5	0.206	0.080	0.636
Saturated fat (g)	24.4 ± 9.5	23.5 ± 9.3	26.4 ± 19.1	22.4 ± 10.1	21.1 ± 8.1	19.9 ± 7.2	0.069	0.051	0.327
MUFA (g)	23.9 ± 9.3	22.2 ± 8.8	22.9 ± 11.3	21.7 ± 10.9	20.7 ± 7.2	19.3 ± 8.4	0.161	0.047	0.970
PUFA (g)	13.1 ± 5.5	13.3 ± 5.8	12.7 ± 7.7	12.3 ± 6.7	12.6 ± 5.5	11.3 ± 5.4	0.495	0.358	0.509
Total cholesterol (mg)	261.8 ± 96.0	242.8 ± 117.5	250.5 ± 127.7	242.7 ± 128.1	237.9 ± 131.1	218.3 ± 106.7	0.481	0.167	0.985
Vitamins
A (RE)	1081.0 ± 653.3	1154.0 ± 772.8	965.9 ± 864.7	1133.0 ± 980.9	1185.0 ± 878.8	990.0 ± 698.9	0.868	0.788	0.071
C (mg)	102.5 ± 60.0	99.4 ± 60.5	92.6 ± 57.6	92.6 ± 56.5	139.2 ± 149.6	108.2 ± 171.0	0.227	0.051	0.067
D (μg)	4.1 ± 4.2	4.3 ± 2.8	3.2 ± 1.9	3.2 ± 3.9	4.1 ± 4.6	3.7 ± 3.6	0.172	0.964	0.527
E (mg)	2.1 ± 4.1	1.7 ± 3.0	1.7 ± 3.7	1.1 ± 2.7	3.1 ± 5.6	2.1 ± 3.6	0.290	0.029	0.647
K (μg)	151.4 ± 158.8	163.9 ± 185.2	181.4 ± 296.5	192.8 ± 239.8	209.4 ± 208.5	211.7 ± 193.4	0.403	0.255	0.607
Minerals (mg)
Calcium	856.5 ± 325.8	837.8 ± 301.6	778.0 ± 283.1	774.1 ± 335.8	897.3 ± 399.9	787.4 ± 331.4	0.321	0.120	0.216
Phosphorus	1062.0 ± 341.1	1067.0 ± 311.4	1011.0 ± 347.9	998.1 ± 316.6	1023.0 ± 358.6	976.1 ± 361.2	0.434	0.561	0.597
Iron	12.4 ± 4.5	12.8 ± 4.4	11.9 ± 5.6	11.5 ± 5.1	12.7 ± 5.7	12.1 ± 6.1	0.588	0.612	0.581
Potassium	2563.0 ± 773.1	2584.0 ± 699.0	2455.0 ± 844.7	2647.0 ± 971.4	2593.0 ± 1035.0	2877.0 ± 1330.0	0.579	0.019	0.395
Magnesium	261.0 ± 101.8	279.5 ± 95.5	261.0 ± 104.5	254.3 ± 118.2	278.5 ± 136.1	274.4 ± 133.1	0.779	0.718	0.276
Zinc	8.4 ± 3.0	8.0 ± 2.4	8.2 ± 3.4	7.9 ± 4.1	8.6 ± 3.7	8.1 ± 4.5	0.937	0.199	0.903

Abbreviation: RE, retinol activity equivalents.

¹Analyses do not include calcium and vitamin D from the supplements used by participants; P < 0.05 considered significantly different.

Glycemic control markers

No significant differences were observed in fasting plasma glucose, insulin, and HOMA-IR in postmenopausal women among the control, 50 g/d prune, and 100 g/d prune groups throughout the trial (Table 3). In addition, percent change from baseline for these measures of glycemic control did not differ between groups (Supplemental Table 3). Fasting plasma glucose (mg/dL) slightly increased from baseline (89.49 ± 0.20) to 12-mo/post (92.05 ± 0.63) among participants who were normoglycemic at baseline (Supplemental Figure 2A; main effect of time, F_(2,466) = 6.81, P < 0.01). In contrast, fasting plasma glucose decreased from baseline (107.11 ± 0.55) to 12-mo/post (103.19 ± 2.36) in a small subset of participants (all groups) who had elevated fasting plasma glucose at baseline (Supplemental Figure 2B; main effect of time, F_(2,111)=8.62, P < 0.001). Among subgroup analyses by baseline cardiometabolic risk, the intervention effects on percent change from baseline for glycemic control markers did not differ between subgroups (Supplemental Table 4).

TABLE 3.

Glycemic control markers and blood lipids over time in postmenopausal women following consumption of control, 50 g/d prunes, and 100 g/d prunes

Outcomes1	Control group			50 g/d Prune			100 g/d Prune			P values
Fasting plasma glucose (mg/dL)	Baseline (n = 78)	6 mo (n = 67)	12 mo (n = 68)	Baseline (n = 77)	6 mo (n = 65)	12 mo (n = 67)	Baseline (n = 77)	6 mo (n = 54)	12 mo (n = 42)	Treatment	Time	Treatment × Time
	93.17 ± 1.06	92.46 ± 0.98	93.07 ± 1.03	91.96 ± 1.06	93.23 ± 1.34	96.06 ± 1.29	94.16 ± 1.20	93.40 ± 1.40	93.64 ± 1.40	0.81	0.10	0.18
Fasting plasma insulin (μIU/mL)	Baseline (n = 69)	6 mo (n = 47)	12 mo (n = 69)	Baseline (n = 67)	6 mo (n = 43)	12 mo (n = 67)	Baseline (n = 45)	6 mo (n = 34)	12 mo (n = 45)	Treatment	Time	Treatment × Time
	6.93 ± 0.58	7.24 ± 0.67	8.01 ± 0.76	9.43 ± 0.85	9.46 ± 1.19	9.59 ± 1.05	8.79 ± 1.38	7.54 ± 0.83	7.82 ± 0.73	0.89	0.08	0.13
HOMA-IR	Baseline (n = 69)	6 mo (n = 46)	12 mo (n = 67)	Baseline (n = 67)	6 mo (n = 41)	12 mo (n = 67)	Baseline (n = 45)	6 mo (n = 34)	12 mo (n = 42)	Treatment	Time	Treatment × Time
	1.63 ± 0.15	1.61 ± 0.16	1.91 ± 0.21	2.26 ± 0.23	2.29 ± 0.34	2.41 ± 0.31	2.15 ± 0.33	1.81 ± 0.23	1.92 ± 0.21	0.80	0.96	0.21
Serum total cholesterol (mg/dL)	Baseline (n = 78)	6 mo (n = 67)	12 mo (n = 68)	Baseline (n = 79)	6 mo (n = 65)	12 mo (n = 67)	Baseline (n = 78)	6 mo (n = 55)	12 mo (n = 43)	Treatment	Time	Treatment × Time
	206.09 ± 3.0	209.00 ± 3.9	210.76 ± 3.6	205.70 ± 4.0	201.50 ± 4.5	208.64 ± 4.5	203.83 ± 4.1	202.78 ± 4.9	201.62 ± 6.1	0.55	0.31	0.56
Serum HDL cholesterol (mg/dL)	Baseline (n = 78)	6 mo (n = 67)	12 mo (n = 68)	Baseline (n = 79)	6 mo (n = 65)	12 mo (n = 67)	Baseline (n = 78)	6 mo (n = 55)	12 mo (n = 43)	Treatment	Time	Treatment × Time
	66.78 ± 1.7	69.04 ± 1.6	69.02 ± 1.7	66.78 ± 2.0	65.98 ± 1.9	66.61 ± 2.0	65.82 ± 1.9	64.89 ± 2.1	64.20 ± 2.4	0.47	0.91	0.13
Serum LDL cholesterol (mg/dL)	Baseline (n = 78)	6 mo (n = 67)	12 mo (n = 68)	Baseline (n = 78)	6 mo (n = 65)	12 mo (n = 67)	Baseline (n = 77)	6 mo (n = 54)	12 mo (n = 42)	Treatment	Time	Treatment × Time
	120.30 ± 2.8	120.80 ± 3.7	122.35 ± 3.2	117.59 ± 3.2	115.10 ± 3.8	120.01 ± 3.7	117.64 ± 3.3	116.40 ± 4.3	118.73 ± 5.2	0.50	0.04	0.96
Serum triglyceride (mg/dL)	Baseline (n = 77)	6 mo (n = 67)	12 mo (n = 66)	Baseline (n = 78)	6 mo (n = 66)	12 mo (n = 64)	Baseline (n = 75)	6 mo (n = 53)	12 mo (n = 42)	Treatment	Time	Treatment × Time
	92.77 ± 4.3	92.32 ± 4.2	92.48 ± 5.0	104.39 ± 7.4	105.65 ± 5.5	114.04 ± 7.1	94.57 ± 5.1	99.50 ± 5.6	95.19 ± 5.8	0.65	0.04	0.58

Abbreviations: HDL, high-density lipoprotein; HOMA-IR, homeostatic model assessment for insulin resistance; LDL, low-density lipoprotein.

¹Data are mean ± SEM; for all outcomes, data were natural-log transformed for analysis using the linear mixed-effect model (PROC MIXED) to determine the effect of time-by-treatment and adjusted for change in android adiposity. Tukey–Kramer post hoc test was used to correct P values for multiple comparisons. P < 0.05 considered significantly different.

Blood lipid and liver function profile

No significant treatment × time interactions were observed for serum triglycerides, total cholesterol, LDL cholesterol (Table 3), or liver function enzymes, ALP, AST, and ALT (data not shown). Although there was a main effect of time for serum LDL cholesterol and triglycerides (Table 3; P = 0.04), no differences were detected post hoc after correcting for multiple comparisons. Percent change from baseline for blood lipids and liver function indices (data not shown) did not differ between groups either in the completers-only analyses (Supplemental Table 3) or by baseline cardiometabolic risk subgroups (Supplemental Table 4).

Body composition and regional fat distribution

BMI did not differ in postmenopausal women among the control, 50 g/d prune, and 100 g/d prune groups throughout the trial (Table 4; treatment × time effect, P = 0.08). BMI slightly increased from baseline (22.48 ± 0.19) to 12-mo/post (22.67 ± 0.27) among participants who were normal weight at baseline (Supplemental Figure 3A; main effect of time, F_(2,283) = 3.94, P = 0.02). However, a decrease in BMI was observed from baseline (28.87 ± 0.29) to 12-mo/post (28.50 ± 0.37) among participants with overweight/obesity at baseline (Supplemental Figure 3B; main effect of time, F_(2,305) = 5.59, P < 0.01). Lean body mass and fat-free mass decreased from baseline to 12 mo/post in all participants (Table 4; main effect of time, P < 0.05). The 50 g/d prune group had higher android fat distribution [total mass (kg), percent fat (%), and fat mass (kg)]; android/gynoid percent fat ratio; VAT mass, volume, and area; and fat mass index than the control group and the 100 g/d group throughout the intervention (Table 4; main effect of group, all P < 0.05). A treatment × time interaction was observed for gynoid percent fat (%) (Table 4; P = 0.01), but no differences were detected post hoc after correcting for multiple comparisons. No treatment × time interactions were found for all other central adiposity measures in postmenopausal women among the control, 50 g/d prune, and 100 g/d prune groups throughout the trial (Table 4). Among completers of the entire 12-mo intervention, android total mass increased by 3.19% from baseline in the control group, whereas the 100 g/d prune group experienced 0.02% decrease in android total mass from baseline (Table 5; main effect of group, F_(2,167) = 4.92, P < 0.01, Tukey’s post hoc, P < 0.01). All remaining percent changes in body composition and regional fat distribution from baseline did not differ between groups (Table 5).

TABLE 4.

Body composition and regional fat distribution over time in postmenopausal women following consumption of control, 50 g/d prunes, and 100 g/d prunes

Outcomes1	Control group			50 g/d Prune			100 g/d Prune			P values
BMI2 kg/m2)	Baseline (n = 78)	6 mo (n = 68)	12 mo (n = 70)	Baseline (n = 79)	6 mo (n = 66)	12 mo (n = 67)	Baseline (n = 78)	6 mo (n = 54)	12 mo (n = 46)	Treatment	Time	Treatment × time
	25.10 ± 0.45	25.12 ± 0.49	24.98 ± 0.45	26.27 ± 0.50	26.45 ± 0.56	26.50 ± 0.53	25.92 ± 0.42	26.09 ± 0.53	25.65 ± 0.55	0.12	0.07	0.08
Lean body mass2 (kg)	Baseline (n = 78)	6 mo (n = 68)	12 mo (n = 70)	Baseline (n = 79)	6 mo (n = 66)	12 mo (n = 67)	Baseline (n = 78)	6 mo (n = 54)	12 mo (n = 46)	Treatment	Time	Treatment × time
	36.93 ± 0.45	36.98 ± 0.47	36.64 ± 0.44	37.27 ± 0.51	37.05 ± 0.52	37.12 ± 0.48	37.09 ± 0.48	37.10 ± 0.65	37.17 ± 0.74	0.90	0.04	0.81
Fat-free mass2 (kg)	Baseline (n = 77)	6 mo (n = 65)	12 mo (n = 66)	Baseline (n = 76)	6 mo (n = 62)	12 mo (n = 64)	Baseline (n = 71)	6 mo (n = 47)	12 mo (n = 40)	Treatment	Time	Treatment × time
	38.98 ± 0.47	39.12 ± 0.50	38.65 ± 0.46	39.22 ± 0.53	38.94 ± 0.56	38.90 ± 0.50	38.61 ± 0.47	38.51 ± 0.59	38.09 ± 0.65	0.62	<0.01	0.72
Android total mass3 (kg)	Baseline (n = 78)	6 mo (n = 62)	12 mo (n = 66)	Baseline (n = 77)	6 mo (n = 62)	12 mo (n = 62)	Baseline (n = 76)	6 mo (n = 52)	12 mo (n = 43)	Treatment	Time	Treatment × time
	4.77 ± 1.23	4.74 ± 1.42	4.86 ± 1.37	5.28 ± 1.34	5.25 ± 1.46	5.26 ± 1.54	4.86 ± 1.24	4.86 ± 1.46	4.77 ± 1.60	<0.001	<0.0001	0.87
Android percent fat3 (%)	Baseline (n = 78)	6 mo (n = 62)	12 mo (n = 66)	Baseline (n = 77)	6 mo (n = 62)	12 mo (n = 62)	Baseline (n = 76)	6 mo (n = 52)	12 mo (n = 43)	Treatment	Time	Treatment × time
	37.96 ± 1.00	36.26 ± 1.20	37.70 ± 1.16	41.30 ± 0.90	41.17 ± 1.12	41.64 ± 1.04	39.09 ± 0.86	39.43 ± 1.01	38.46 ± 1.21	<0.0001	<0.0001	0.90
Android fat mass3 (kg)	Baseline (n = 78)	6 mo (n = 62)	12 mo (n = 66)	Baseline (n = 77)	6 mo (n = 62)	12 mo (n = 62)	Baseline (n = 76)	6 mo (n = 52)	12 mo (n = 43)	Treatment	Time	Treatment × time
	1.88 ± 0.09	1.80 ± 0.11	1.92 ± 0.10	2.24 ± 0.09	2.23 ± 0.10	2.27 ± 0.11	1.96 ± 0.08	1.97 ± 0.09	1.89 ± 0.11	<0.0001	<0.0001	0.91
Gynoid total mass2 (kg)	Baseline (n = 78)	6 mo (n = 62)	12 mo (n = 66)	Baseline (n = 77)	6 mo (n = 62)	12 mo (n = 62)	Baseline (n = 76)	6 mo (n = 52)	12 mo (n = 43)	Treatment	Time	Treatment × time
	10.74 ± 0.19	10.61 ± 0.21	10.71 ± 0.20	11.26 ± 0.21	11.14 ± 0.24	11.05 ± 0.23	10.77 ± 0.20	10.81 ± 0.25	10.80 ± 0.29	0.20	0.60	0.25
Gynoid percent fat (%)	Baseline (n = 78)	6 mo (n = 62)	12 mo (n = 66)	Baseline (n = 77)	6 mo (n = 62)	12 mo (n = 62)	Baseline (n = 76)	6 mo (n = 52)	12 mo (n = 43)	Treatment	Time	Treatment × time
	43.31 ± 0.57	42.53 ± 0.62	43.31 ± 0.58	44.62 ± 0.50	44.99 ± 0.57	45.11 ± 0.54	44.03 ± 0.50	44.75 ± 0.61	43.74 ± 0.74	0.10	0.06	0.01
Gynoid fat mass (kg)	Baseline (n = 78)	6 mo (n = 62)	12 mo (n = 66)	Baseline (n = 77)	6 mo (n = 62)	12 mo (n = 62)	Baseline (n = 76)	6 mo (n = 52)	12 mo (n = 43)	Treatment	Time	Treatment × time
	4.71 ± 0.13	4.57 ± 0.14	4.69 ± 0.13	5.07 ± 0.13	5.06 ± 0.15	5.03 ± 0.15	4.77 ± 0.12	4.87 ± 0.15	4.77 ± 0.18	0.12	0.32	0.11
A:G ratio	Baseline (n = 78)	6 mo (n = 62)	12 mo (n = 66)	Baseline (n = 77)	6 mo (n = 62)	12 mo (n = 62)	Baseline (n = 76)	6 mo (n = 52)	12 mo (n = 43)	Treatment	Time	Treatment × time
	0.87 ± 0.01	0.84 ± 0.02	0.86 ± 0.02	0.92 ± 0.01	0.91 ± 0.02	0.92 ± 0.02	0.88 ± 0.01	0.87 ± 0.01	0.87 ± 0.02	0.04	0.08	0.75
VAT mass2 (kg)	Baseline (n = 78)	6 mo (n = 62)	12 mo (n = 66)	Baseline (n = 77)	6 mo (n = 62)	12 mo (n = 62)	Baseline (n = 76)	6 mo (n = 52)	12 mo (n = 43)	Treatment	Time	Treatment × time
	0.53 ± 0.02	0.51 ± 0.03	0.54 ± 0.03	0.67 ± 0.03	0.67 ± 0.03	0.68 ± 0.03	0.58 ± 0.03	0.58 ± 0.03	0.55 ± 0.03	<0.01	0.22	0.33
VAT volume2 (cm3)	Baseline (n = 78)	6 mo (n = 62)	12 mo (n = 66)	Baseline (n = 77)	6 mo (n = 62)	12 mo (n = 62)	Baseline (n = 76)	6 mo (n = 52)	12 mo (n = 43)	Treatment	Time	Treatment × time
	580.7 ± 31.6	560.7 ± 38.6	586.1 ± 37.5	728.4 ± 38.7	725.7 ± 41.1	744.5 ± 41.3	631.0 ± 32.8	627.5 ± 38.0	603.0 ± 41.7	<0.01	0.22	0.33
VAT area2 (cm2)	Baseline (n = 78)	6 mo (n = 62)	12 mo (n = 66)	Baseline (n = 77)	6 mo (n = 62)	12 mo (n = 62)	Baseline (n = 76)	6 mo (n = 52)	12 mo (n = 43)	Treatment	Time	Treatment × time
	111.42 ± 6.0	107.57 ± 7.4	112.44 ± 7.2	136.69 ± 7.3	139.28 ± 7.9	142.90 ± 7.9	121.03 ± 6.2	119.82 ± 7.4	115.66 ± 8.0	0.02	0.24	0.61
FMI2 (kg/m2)	Baseline (n = 78)	6 mo (n = 62)	12 mo (n = 66)	Baseline (n = 77)	6 mo (n = 60)	12 mo (n = 61)	Baseline (n = 76)	6 mo (n = 52)	12 mo (n = 43)	Treatment	Time	Treatment × time
	9.83 ± 0.32	9.58 ± 0.38	9.79 ± 0.36	11.48 ± 0.36	11.54 ± 0.40	11.44 ± 0.39	10.32 ± 0.28	10.43 ± 0.34	10.10 ± 0.36	<0.01	0.04	0.10
LMI2 (kg/m2)	Baseline (n = 78)	6 mo (n = 62)	12 mo (n = 66)	Baseline (n = 77)	6 mo (n = 60)	12 mo (n = 61)	Baseline (n = 76)	6 mo (n = 52)	12 mo (n = 43)	Treatment	Time	Treatment × time
	13.76 ± 0.14	13.79 ± 0.15	13.73 ± 0.14	14.17 ± 0.17	14.03 ± 0.17	14.03 ± 0.18	13.93 ± 0.15	13.82 ± 0.18	13.79 ± 0.17	0.41	0.07	0.30

Abbreviations: A:G ratio, android-gynoid percent fat ratio; FMI, fat mass index; LMI, lean mass index; VAT, visceral adipose tissue.

¹Data = mean ± SEM.

²Data were natural-log transformed for analysis using the linear mixed-effect model (PROC MIXED) to determine the effect of time-by-treatment.

³Friedman test (nonparametric) was used. Tukey–Kramer post hoc test was used to correct P values for multiple comparisons. P < 0.05 considered significantly different.

TABLE 5.

Completers-only analysis of percent change from baseline in body composition and regional fat distribution in postmenopausal women following consumption of control, 50 g/d prunes, and 100 g/d prunes

Percent change from baseline1	Control (n=70)	50 g/d Prune (n=67)	100 g/d Prune (n=43)	Treatment P value
Body weight (%)2	−0.05 ± 0.46	0.29 ± 0.45	−0.15 ± 0.49	0.75
BMI (%)2	0.01 ± 0.45	0.35 ± 0.45	−0.16 ± 0.49	0.74
Total body fat mass (%)	1.29 ± 0.81	0.99 ± 0.76	0.19 ± 0.78	0.651
Total body lean mass (%)	−0.66 ± 0.42	−0.30 ± 0.47	−0.72 ± 0.52	0.766
Fat-free mass (%)	−0.69 ± 0.41	−0.28 ± 0.45	−1.05 ± 0.48	0.418
Percent body fat (%)	1.20 ± 0.51	0.76 ± 0.51	0.49 ± 0.56	0.727
Android total mass (%)	3.19 ± 0.673	0.73 ± 0.683,4	−0.02 ± 0.854	0.008
Android percent fat (%)	0.80 ± 0.95	1.49 ± 0.94	−0.06 ± 0.92	0.471
Android fat mass (%)	4.12 ± 1.36	2.58 ± 1.32	−0.01 ± 1.39	0.130
Gynoid total mass (%)2	0.68 ± 0.45	−0.83 ± 0.61	−0.42 ± 0.53	0.28
Gynoid percent fat (%)	1.17 ± 0.59	1.07 ± 0.54	0.05 ± 0.52	0.317
Gynoid fat mass (%)	1.96 ± 0.89	0.27 ± 0.93	−0.35 ± 0.82	0.264
A:G ratio (%)	−0.28 ± 0.83	0.38 ± 0.70	−0.16 ± 0.75	0.721
VAT mass (%)2	3.35 ± 1.69	4.72 ± 2.52	1.16 ± 2.29	0.60
VAT volume (%)2	3.37 ± 1.69	4.74 ± 2.53	1.16 ± 2.29	0.59
VAT area (%)2	3.36 ± 1.70	4.75 ± 2.53	1.15 ± 2.29	0.60
FMI (%)	2.09 ± 0.75	0.54 ± 0.86	0.32 ± 0.81	0.356
LMI (%)	0.13 ± 0.36	−1.12 ± 0.51	−0.83 ± 0.54	0.236

Abbreviations: A:G ratio, android-gynoid percent fat ratio; FMI, fat mass index; LMI, lean mass index; VAT, visceral adipose tissue.

¹Data are percent change from baseline to 12 mo (%) mean ± SEM. The linear mixed-effects model (PROC MIXED) was used (unless indicated otherwise) to determine the effect of treatment on each outcome and adjusted for baseline values of the outcome. Tukey–Kramer post hoc test was used to correct P values for multiple comparisons.

²Nonparametric Kruskal–Wallis test (PROC NPAR1WAY).

³Labeled means without a common number significantly differ, P < 0.01.

⁴Labeled means without a common number significantly differ, P < 0.01.

Discussion

In the current ancillary investigation of the parent RCT, The Prune Study, we aimed to explore whether the addition of 50 g or 100 g prunes into the daily diet for 12 mo affected select cardiometabolic health outcomes, including blood glucose, insulin, HOMA-IR, total cholesterol, HDL and LDL cholesterol, triglycerides, liver function indices, and regional body fat distribution, in postmenopausal women. Prune supplementation at either 50 g/d or 100 g/d dose had no significant effects on markers of glycemic control, blood lipids, and liver enzymes after 12 mo compared with the no-prune (0 g/d) control group. Furthermore, BMI, lean body mass, fat-free mass, android and gynoid percent fat, and VAT indices were unaltered in women consuming either 50 g or 100 g prunes daily compared with the control group after 12 mo of intervention. However, we noted a significant reduction in android total mass from baseline in women supplemented with 100 g/d prunes compared with the control group, who experienced an increase in android total mass from baseline. Collectively, our findings suggest that 12-mo supplementation of prunes into the daily diet with a single serving (50 g/d) or 2 servings (100 g/d) did not impact markers of glycemic control and prevented adverse changes in central adiposity in postmenopausal women.

Increased central adiposity and insulin resistance associated with the menopause transition predisposes women to developing type II diabetes, which is an important CVD risk factor [49]. In our RCT, we observed no significant changes in fasting plasma glucose, insulin, and HOMA-IR after 12-mo prune supplementation. Results from our unadjusted analyses show that the 2.82% increase in blood glucose from baseline observed in the 50 g/d prune group was significantly higher than the −1.88% change in the 100 g/d group, but this effect was lost after adjusting for android adiposity (Supplemental Table 3). This suggests that body fat may be partially mediating these changes in blood glucose in postmenopausal women and that, while prunes at either dose did not improve glycemic control, they did not result in elevated fasting plasma glucose. Additionally, the lack of a prune effect on glycemic control markers might be partly due to a majority (79.8%) of our study population being normoglycemic at baseline. Indeed, the mean blood glucose concentrations in the control and prune groups were below the prediabetes threshold at all time points throughout the trial. In a small subset of women with elevated fasting plasma glucose at baseline, blood glucose decreased in all groups from baseline to post, although these changes were not clinically meaningful and did not reach the normal glycemic threshold. Prior work assessing mixed dried fruit interventions comprising raisins, dates, plums, and figs (28 g each) demonstrated significant but modest elevations in fasting plasma glucose in adult men and women (n = 55, aged 40.5 y) after 4 wk of intervention compared with controls consuming a carbohydrate-matched snack, which persisted even after excluding adults who exhibited substantial changes in body weight [50]. In contrast, others reported either null effects in short-term crossover RCTs of 2-wk prune interventions (∼42 g/d doses) [34,51] or reductions in 2-h postprandial glucose and insulin responses after supplementation of either 238 kcal prunes (∼100 g) in adult women (n = 19, aged 39.2 y) [52] or 41.8 g prunes in adult men and women (n = 49, aged 37.4 y) compared to isocaloric control snacks [53]. Disagreement in these findings might be partly attributed to differences in prune doses and duration, study population, and assessment of glycemic control. Overall, data on the long-term effects of prunes on blood glucose and insulin in postmenopausal women are inconclusive and need further investigation, particularly including glycated hemoglobin measurements, which may be more informative when monitoring prune effects on the kinetics of glycemic control in longitudinal RCTs.

The menopause transition is also linked to increased prevalence of dyslipidemia, which is characterized by elevated total cholesterol, LDL cholesterol, and triglycerides, and low HDL cholesterol [54]. In our study, serum concentrations of total cholesterol, LDL and HDL cholesterol, and triglycerides in postmenopausal women were unaffected after 12-mo supplementation of 50 or 100 g/d prunes compared to the control group. Prunes are rich in phenolics, polyols, and soluble and insoluble dietary fibers, including pectin, cellulose, and hemicellulose [26], which are suggested to exert positive effects on lipid metabolism through several biological mechanisms, such as delayed gastric emptying, increased satiety, and gut microbiota-mediated fermentation of dietary fibers into short-chain fatty acids, which inhibit cholesterol and lipid synthesis in the liver and increase hepatic conversion of cholesterol into bile acids, thus increasing the rate of fecal bile acid excretion [55]. However, we observed no prune effects on blood lipids in our cohort of postmenopausal women, despite half of the study population (53%) exhibiting elevated total cholesterol and a majority (70.5%) with elevated LDL cholesterol, thus characterizing an “at-risk” population that may benefit from the potential lipid-lowering effects of prunes. A plausible explanation for these null findings is that our RCT was conducted in a free-living study population that was instructed to incorporate prunes into their habitual diet, and in absence of a controlled-feeding clinical setting, it is likely that consumption of other background foods diluted the potential to observe favorable effects on blood lipids. Nevertheless, our findings are in agreement with previous studies reporting null prune effects on blood lipids [34,51,53], yet others demonstrate reductions in total cholesterol [56] and LDL cholesterol [35,57,58] or mild elevations in LDL cholesterol [50]. Altogether, findings from the effects of prunes on blood lipid and liver function indices are inconsistent, most likely due to varying study designs, participant characteristics, selection of comparative control, prune dosage, and duration.

Estrogen withdrawal during the menopause transition is concomitantly associated with increased central adiposity shifting to an android phenotype that predisposes women at risk of cardiometabolic disease [12,59]. Epidemiological evidence indicates that although alterations in body weight and BMI after midlife are attributed more to aging than menopause itself [60], changes in regional body fat distribution are strongly associated with menopause transition [61]. This is possibly due to the loss of estrogen’s regulation on adipose tissue distribution [62], with early research demonstrating that estrogen therapy prevents the accrual of central android fat after menopause in early postmenopausal women [63]. In our entire cohort of postmenopausal women, reductions in lean mass and fat-free mass and increments in fat mass and percent body fat observed over time in all groups were not surprising, as prior studies suggest that the menopause transition is directly related to lean mass loss and fat mass gain [11]. In our study, there were no changes in body weight, BMI, or DXA-derived regional fat indices in postmenopausal women supplemented with prunes compared to control, consistent with previous studies demonstrating null effects of prune supplementation on body weight, BMI, and percent body fat [35,51,53,56]. We further noted that neither of the prune interventions detrimentally affected BMI in women who already had overweight or obesity at baseline. The 50 g/d prune group had elevated android and visceral fat distribution and android/gynoid percent fat ratio compared to the other groups throughout the entire 12-mo intervention, which may have precluded our ability to detect significant differences between groups over time. However, we observed that android total mass increased from baseline in the control group, which was significantly different than the 100 g/d prune group, who experienced a modest decrease in android total mass from baseline. Results from our unadjusted analyses show that the percent decline in android total mass in the 50 g/d group was also significantly different than the percent increase in the control group (data not shown), but this effect was lost after controlling for android total mass at baseline. Our primary endpoint article reports that the minutes of weekly high-magnitude loading exercise decreased in the control group compared to the prune groups at the end of the study [37], thus we cannot exclude the possibility of other covariates influencing body fat distribution, such as energy expenditure and physical activity, that were not controlled for and may partly explain these findings.

Approximately 95% of the United States population does not meet adequate intake of dietary fiber [64], which is established as a nutrient of public health concern for underconsumption [24]. In addition, adequate fiber intake is associated with reduced cardiometabolic risk [65]. The adequate dietary reference intake for total dietary fiber is recommended at 14 g per 1000 kcal/d, which is ∼22 g/d for women aged 51 years and older [24,47]. Previous research highlights that United States adults consuming dried fruit had higher dietary fiber intake than those who did not [66]. In our RCT, the interventions provided ∼3.94 g fiber from the 50 g/d prune dose and 7.89 g fiber from the 100 g/d prune dose [37]. Thus, we aimed to explore whether 1 (50 g) or 2 servings (100 g) of prunes daily for 12 mo achieved adequate dietary fiber intake in women at the end of the study compared to the no-prune control group, with a corollary objective to assess fiber intake among a subset of women who were already below this reference intake (low-fiber category) at baseline. Interestingly, we noted that fiber intake increased from baseline to 12-mo/post in not only the 50 g/d and 100 g/d prune groups but also in control women, and this trend was also observed in the subset of participants in the low-fiber category at baseline, albeit not reaching the recommended adequate intake threshold. Dietary intake assessments employed in human studies are subjective measurements that are prone to recall and self-selection biases, and the record of nutrient intake data are relatively more accurate as participants progress throughout the 12-mo RCT, all of which are speculated to partially explain some of the trends observed in the diet data [67]. Notably, the incorporation of either 50 g or 100 g prunes into the daily diet did not alter total energy intake (kilocalories), which remained similar in all groups over time, suggesting that women in the prune intervention groups might have intentionally changed their habitual diet regimen to allow prunes to possibly replace the calories obtained from their otherwise usual food group selections.

A major strength of the current ancillary investigation is the assessment of select measures of cardiometabolic health in a large cohort of postmenopausal women. However, blood pressure and vascular function were not measured, as prior work demonstrated null effects on these outcomes after 2 to 8 wk of ∼42 g prune supplementation [34,53] but should be considered in future long-term investigations. In this ancillary study, potential limitations include the following: the analysis was conducted in all randomized participants who completed the 1-y intervention and had data for exploratory cardiometabolic health outcomes and was thus not intent-to-treat; potentially increased likelihood of type 1 error from analysis of multiple outcomes; and higher attrition in the 100 g/d prune group primarily due to poor tolerance to the intervention. Thus, these findings must be interpreted with caution and assessed in future studies with larger sample sizes. Additional limitations of this study include racial and ethnic homogeneity of our study participants (mostly Caucasian), thus limiting the generalizability of our results. The NIH policy for clinical research underlines the importance of inclusion of women across the lifespan [68]. The magnitude of unfavorable changes in the cardiometabolic profile is significant during the years surrounding the final menstrual period [59], thus future work is warranted in perimenopausal women across various stages of the menopause transition to determine whether prune supplementation can improve cardiometabolic profile within this therapeutic window.

In conclusion, the addition of 50 g to 100 g prunes to the daily diet for 12 mo had no adverse effects on cardiometabolic risk factors in postmenopausal women and had no further detrimental impacts on body fat composition and regional fat distribution. Postmenopausal women are at high risk of cardiometabolic disease, and current evidence does not recommend hormone therapy for CVD prevention in this age group [69]. Collectively, from our RCT of postmenopausal women from The Prune Study, findings from our primary BMD endpoint analyses [37] as well as the current ancillary analysis of exploratory cardiometabolic health markers suggest that prunes may represent a dietary nonpharmacological treatment that preserves bone health concurrently with an improvement in dietary fiber intake, without any adverse effects on cardiometabolic risk factors and central adiposity in postmenopausal women.

Acknowledgments

We thank the study participants and The California Prune Board for providing funding for this work. We also thank the research staff at The Women's Health and Exercise Lab, Ellen Bingham, Emily Ricker, Rebecca Mallinson, Heather Allaway, and Emily Lundstrom, for their assistance during the study.

Author contributions

The authors’ responsibilities were as follows – MJDS, NIW, CJR, CW: designed the research; NCAS, KJK, JJD: conducted research and collected data; JJD, CJR, HL, and MJDS: participated in data analysis and interpretation and wrote the article; JJD, CJR, HL, NCAS, KJK, NIW, CW, MGF, CHN, MJDS: revised manuscript; MJDS: had primary responsibility for the final content; and all authors: read and approved the final manuscript.

Conflict of interest

CW and CJR are members of the Nutritional Advisory Panel for The California Prune Board. All other authors report no conflicts of interest.

Funding

This work was supported by The California Prune Board Award Number: 180215 (to MJD and CJR). In addition, this study was supported by the National Center for Advancing Translational Sciences of the National Institutes of Health (NIH) Award Number TL1TR002016 (to JJD). The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH. The funding agencies had no role in the study design, data collection, analysis, interpretation, writing, decision to publish, or preparation of the manuscript.

Data availability

Data described in the manuscript, code book, and analytic code will be made available upon request pending application and approval.

Appendix A. Supplementary data

The following is the Supplementary data to this article: