Randomized nutrient bar supplementation improves exercise-associated changes in plasma metabolome in adolescents and adult family members at cardiometabolic risk.

BACKGROUND: Poor diets contribute to metabolic complications of obesity, insulin resistance and dyslipidemia. Metabolomic biomarkers may serve as early nutrition-sensitive health indicators. This family-based lifestyle change program compared metabolic outcomes in an intervention group (INT) that consumed 2 nutrient bars daily for 2-months and a control group (CONT).
METHODS: Overweight, predominantly minority and female adolescent (Teen)/parent adult caretaker (PAC) family units were recruited from a pediatric obesity clinic. CONT (8 Teen, 8 PAC) and INT (10 Teen, 10 PAC) groups randomized to nutrient bar supplementation attended weekly classes that included group nutrition counseling and supervised exercise. Pre-post physical and behavioral parameters, fasting traditional biomarkers, plasma sphingolipids and amino acid metabolites were measured.
RESULTS: In the full cohort, a baseline sphingolipid ceramide principal component composite score correlated with adiponectin, triglycerides, triglyceride-rich very low density lipoproteins, and atherogenic small low density lipoprotein (LDL) sublasses. Inverse associations were seen between a sphingomyelin composite score and C-reactive protein, a dihydroceramide composite score and diastolic blood pressure, and the final principal component that included glutathionone with fasting insulin and the homeostatic model of insulin resistance. In CONT, plasma ceramides, sphinganine, sphingosine and amino acid metabolites increased, presumably due to increased physical activity. Nutrient bar supplementation (INT) blunted this rise and significantly decreased ureagenic, aromatic and gluconeogenic amino acid metabolites. Metabolomic changes were positively correlated with improvements in clinical biomarkers of dyslipidemia.
CONCLUSION: Nutrient bar supplementation with increased physical activity in obese Teens and PAC elicits favorable metabolomic changes that correlate with improved dyslipidemia. The trial from which the analyses reported upon herein was part of a series of nutrient bar clinical trials registered at clinicaltrials.gov as NCT02239198.

RCT Entities:   Population Interventions Outcomes

Introduction

Cardiovascular disease (CVD) remains the leading cause of mortality in the United States [1], and after decades of decline, is rising coincident with the increase in obesity, insulin resistance, and diabetes that characterize cardiometabolic risk (CMR) [2]. Notwithstanding hereditary predisposition, reduction in identified, modifiable lifestyle risk factors can reverse CMR and CVD [3]. It is estimated that 45.4% of all cardiometabolic deaths in the United States due to heart disease, stroke, and diabetes are associated with suboptimal intakes (too much or too little) of 10 dietary factors [4].

Fewer than 1% of American children and adolescents meet full recommended metrics of heart healthy nutrition, falling especially short of recommended intake in the categories of fruits, vegetables, fiber and essential fatty acids [5, 6]. Intensive pediatric lifestyle interventions for obesity are effective in achieving significant reduction in body mass index (BMI) but do not elicit stable changes in nutrition habits in children and adolescents [7]. These studies suggest a critical need for developing innovative tools to improve diet quality in youth.

We have previously shown that twice daily consumption for two weeks of a whole food based nutrient bar composed of a blueberry, dark chocolate, red grape, and walnut matrix, soluble and insoluble fiber, with supplemental vitamins, minerals and essential long chain fatty acids, significantly increased high density lipoprotein cholesterol (HDL), due primarily to a 28% increase in large HDL particles, in generally healthy and insulin sensitive lean and overweight adults [8]. In a subsequent 2 mo study of the effects of the nutrient bar on CMR markers in individuals across a range of BMIs, only those with low inflammation at baseline as assessed by high sensitivity C-reactive protein (CRP) < 14.3 nmol/L (<1.5 mg/L) responded comparably to those in the earlier trial, with not only increased HDL cholesterol and large HDL particles but also a trend toward increased high molecular weight adiponectin and a decrease in other CMR factors at 2 weeks, sustained at 2 months (9). In particular, a shift in low density lipoprotein (LDL) particle subfractions toward a less atherogenic profile was evident in the noninflamed group (decreased small and medium LDL and increased large LDL). Although the participants with overweight or obesity and CRP > 14.3 nmol/L did not show this response, they did experience an upward trend in adiponectin by 2 months [9]. These results suggest that there may be a continuum of metabolic responsiveness to this nutrient supplement that is slowed in the face of the chronic low-level inflammation commonly observed with obesity and insulin resistance. A 6 month study of this nutrient bar in obese adolescents with non-eosinophilic asthma showed improved lung function at 2 months, but favorable movement in cardiometabolic biomarkers only began to emerge at 6 months [10].

It is not known whether more sensitive biomarkers of early metabolic change may be capable of detecting short-term effects of the nutrient bar supplementation in persons at CMR. Past metabolomics studies in obese adolescents and adults have identified strong positive associations between baseline levels of branched chain, aromatic, sulfur, and gluconeogenic amino acid metabolites and parameters of inflammation and insulin resistance [11, 12]. Similarly, elevated levels of specific ceramide species have been shown to associate with inflammation, dyslipidemia and insulin resistance in both adult and adolescent obesity [13]. The relative sensitivity of these biomarkers to reflect moderate changes in dietary intake of polyphenols, essential lipids, fiber and vitamin/minerals, remains incompletely understood.

In the present randomized, controlled, non-blinded trial, a two month intervention with exercise and nutrition counseling alone (CONT) or with nutrient bar supplementation (INT) was performed in a high CMR cohort of adolescent (Teen)/parent adult caretaker (PAC) family units to determine 1) cross-sectional relationships in both adolescents and adults between traditional CMR biomarkers and amino acid and ceramide metabolites and 2) longitudinal changes within groups in the same CMR biomarkers following the lifestyle +/- nutrient bar intervention.

Methods

Study population

This study was approved by the University of California (UC) Benioff Children's Hospital Oakland Human Subjects Review Board, under approval number 2011022 entitled: The Impact of a Nutritional Supplement on Weight and Metabolic Health in a Parent-Child Intervention (S1 Protocol). All participating adolescents signed an assent document and all parent /adult legal guardian participants signed a written consent for their own participation and a separate consent for their participating adolescent(s). The study cohort was recruited from the UC San Francisco Benioff Children’s Hospital Oakland weight management program (Fig 1). Teen participants were eligible for inclusion if they had a BMI greater than the 95th percentile, were between 14 and 18 years of age, fluent in English (as group lifestyle counseling was conducted in English), and willing to eat the nutrient bar twice daily, after having tasted a sample, with one or both PACs also willing to participate. PAC inclusion criteria were the same except that there was no adult weight threshold. Exclusion criteria for both Teens and PACs included the diagnosis of diabetes mellitus (type 1 or 2) or hypertension (>140/90 for adults, >95th percentile + 5 mm Hg for Teens), and use of glucocorticoid, weight loss, insulin-sensitizing, lipid-lowering, or anti-hypertensive medication.

Fig 1

Trial enrollment profile.

Abbreviation: PAC, Parent Adult Caretaker.

With Institutional Review Board approval, PAC consent for self and adolescent, and adolescent assent, 17 dyads and 1 triad (19 PAC, 18 Teens) were enrolled (Fig 1 and S1 Flowchart) in May and June, 2011. One INT PAC dropped out before study measures were initiated but the rest of the family unit continued participation. All family units were seen within two weeks of the baseline group counseling session and within 2 weeks of the final session on the Clinical and Translational Research unit for pre-post blood testing. Participants were randomized at the baseline visit 1.25:1 to INT:CONT groups that met on separate days in the late afternoon after school and work for eight identical lifestyle counseling sessions in July and August 2011; 30 minutes of weekly group nutrition counseling followed by 30 minutes of supervised group exercise. In addition, the INT group was given a one week supply of nutrient bars at each visit through the seventh week and advised to return wrappers at the following session. Bar composition has been previously described (Bar2, in Table 1, reference [9]), and was designed to supplement prevalent nutrient deficiencies in the American diet up to daily recommended intake (RDI), including 515 mg polyphenols, 4 gm whey protein, 9 gm of total fiber (soluble and insoluble) and 200 mg of long chain omega 3 fatty acid docosahexaenoic acid (DHA). Most vitamins and minerals added to the whole food matrix of the bar are in amounts representing 10% to 50% of their corresponding recommended daily intake, with the exceptions of vitamin C (105 mg) and vitamin D (2300 IU) both added above RDI. Additional vitamin C serves as an antioxidant preservative for other bar ingredients, notably the omega 3 fatty acid DHA. Additional vitamin D addresses the prevalent deficiency in persons at high CMR, and the relatively low blood levels in the study cohort (Table 1).

Table 1

Baseline participant profiles.

	CONTROL (n = 16; 81% Female)19% HW, 19%HAI, 6% HB,6%NHW, 50% NHB			INTERVENTION (n = 20; 75% Female)40% HW, 40% HAI, 10% HB,5% NHW, 5% NHB
	PAC(N = 8)	Teen(N = 8)	CONTCombined(N = 16)	PAC(N = 10)	Teen(N = 10)	INTCombined(N = 20)
Age (yrs)	41.9 (12.2)	15.3* (1.3)	28.6 (16.1)	43.0 (3.8)	15.6* (1.4)	29.3 (14,3)
Weight **(Kg)	103.7 (14.9)	102.9 (27.9)	103.3 (21.7)	83.0 (18.6)	85.3 (8.7)	84.1(14.2)
Height (m)	1.64 (0.08)	1.69 (0.06)	1.66 (0.07)	1.61 (0.07)	1.63 (0.09)	1.62 (0.08)
Waist Circ. (cm)	116.0 (9.1)	106.9 (19.3)	111.4 (15.3)	102.0 (16.9)	103.0 (10.3)	102.5 (13.7)
BMI ** (Kg/m2)	38.6 (5.2)	35.9 (8.1)	37.2(6.7)	31.8 (5.9)	32.2 (4.6)	32.0(5.2)
WHRatio	70.7 (5.8)	63.3 (10.6)	67.0 (9.1)	63.2 (9.7)	63.4 (9.1)	63.4 (9.1)
TC (mmol/L)	5.1 (1.1)	4.2 (0.8)	4.7 (1.0)	5.2 (0.5)	4.5 (1.1)	4.9 (0.9)
TG (mmol/L)	1.3 (0.6)	1.4 (0.9)	1.3 (0.7)	2.0 (1.8)	1.2 (0.4)	1.6 (1.4)
HDL (mmol/L)	1.3 (0.3)	1.1 (0.2)	1.2 (0.2)	1.4 (0.3)	1.2 (0.3)	1.3 (0.3)
LDL (mmol/L)	3.2 (1.0)	2.5 (0.6)	2.8 (0.8)	3.0 (0.4)	2.8 (0.9)	2.9 (0.7)
TGHDLR	2.4 (1.7)	2.9 (2.1)	2.7 (1.9)	4.3 (6.3)	2.5 (1.1)	3.4 (4.5)
Vit D (nmol/L)	37.4 (10.0)	49.7 (21.2)	44.0 (16.9)	48.0 (9.0)	53.4 (16.2)	50.7 (15.2)
Glucose (mmol/L)	5.5 (0.2)	5.4 (0.3)	5.4 (0.3)	5.4 (0.4)	5.1 (0.3)	5.3 (0.4)
Insulin (pmol/L)	126.0 (81.)	179.3 (127.3)	154.8 (107.9)	76.8 (27.8)	154.1 * (77.0)	115.8 (69.1)
HOMA	5.25 (3.6)	7.2 (5.5)	6.3 (4.5)	3.1 (1.3)	5.9 * (3.0)	4.5 (2.7)
SBP(mm Hg)	127 (15)	112 (8) *	120 (14)	120 (11)	117 (8)	119 (9)
DBP(mm Hg)	77 (9)	66 (7)*	72 (10)	79 (10)	65 (9)*	72 (12)
HR	67 (13)	75 (7)	71 (11)	72 (9)	70 (10)	71 (9)
CRP ** (nmol/L)	43.8 (25.7)	57.1 (30.5)	50.5 (27.6)	31.4 (16.2)	28.6 (22.8)	30.5(20.0)
Adipo (ng/ml)	1556 (1389)	1313 (588)	1427 (1007)	2507 (1654)	1296 (587)	1933 (1381)

Values are mean (standard deviation).

* Denotes statistically significant (p <0.05) difference by unpaired student’s t-test between PAC (Parent Adult Caretakers) and Teen subgroups within Control (CONT) and/or Intervention (INT) groups.

** Denotes statistically significant (p <0.05) differences for the designated variable by unpaired student’s t-test between combined CONT and INT groups

Abbreviations: n–number; HW–Hispanic White; HAI–Hispanic American Indian; HB–Hispanic Black; NHW–Nonhipanic White; NHB–Nonhispanic Black; yrs–years; Kg- kilograms; m–meters; mmol/L–millimoles per liter; nmol/L–nanomoles per liter; mm Hg–millimeters of mercury; ng/ml–nanograms per milliliter; Waist Circ–Waist Circumference; TGHDLR–Triglyceride to HDL ratio; SBP–Systolic blood pressure, DBP–Diastolic blood pressure, TC- Total Cholesterol; TG–Triglyceride; LDL-Low Density Lipoprotein cholesterol; HDL–High Density Lipoprotein cholesterol; Vit D– 25 hydroxy vitamin D; HOMA–Homeostatic model assessment of insulin resistance; CRP–high sensitive C-reactive protein; HR–resting heart rate; Adipo- High Molecular Weight Adiponectin.

Consumption of 2 bars each day was advised, with the first to be eaten before noon and the second in either the afternoon or evening, together with a minimum of 8 ounces of water with each bar due to the high fiber content. Compliance with eating the supplement (INT group) and home lifestyle adherence (both INT and CONT groups) were evaluated with phone call or text communication to each participating family every other day. Pre-post change in 25 hydroxy vitamin D levels served as an objective measure of compliance with the nutrient bars.

Procedures and assays

Assessment of physical (weight, height, waist circumference, blood pressure, resting heart rate), behavioral (self-report diet by Block Food Frequency Questionnaire (Nutritionquest, Berkeley CA) [14] and activity (8 questions of usual weekly activity) [15], and metabolic (CMR biomarkers listed below) status was conducted at baseline and study completion on all Teen and PAC participants.

Anthropometric and clinical evaluation: All anthropometric measures were performed in duplicate in the clinical research center and if not within 10% agreement, were repeated a third time. The reported measure is an average of the two closest numbers. Height was measured with a stationary stadiometer. Weight was measured using a digital electronic scale and the BMI and waist to height ratio (WHRatio) were calculated. Waist circumference (Waist Circ) was measured at end expiration to the nearest mm with a Gulick II Plus tape midway between the lowest border of the rib cage and the upper border of the iliac crest.

Blood pressure and Resting Heart Rate: Each was measured in triplicate after 5 minutes sitting quietly with readings taken at least one minute apart. An automatic digital blood pressure monitor was used with cuff size adjusted for arm size.

Traditional CMR biomarkers: Fasting blood samples were drawn and processed in the UCSF Benioff Children’s Hospital Oakland Clinical Research Center and sent to ARUP Diagnostic Laboratories for: standard lipid profile [total triglyceride, total cholesterol, and cholesterol within HDL and LDL (calculated by the Friedwald formula with reflex to direct LDL), glucose, insulin, 25 hydroxy Vitamin D level and CRP. TG to HDL ratio and non-HDL were calculated. Fasting insulin and glucose were used to calculate the Homeostasis Model Assessment of Insulin Resistance Index (HOMA) according to the formula: fasting insulin (microU/L) x fasting glucose (nmol/L)/22.5. Lipoprotein particle subclasses were analyzed by an ion mobility procedure that sensitively and directly measures concentrations of lipoprotein particle subfractions [16]. High molecular weight adiponectin was measured by solid-phase sandwich ELISA (Quantikine; R&D Systems, Minneapolis, MN).

Metabolomic analyses: 1) Targeted analyses of 42 amine-containing metabolites consisting of 20 major amino acids, and secondary metabolites of arginine and cysteine whose levels are sensitive to inflammation and oxidative stress were performed on stored samples preserved at -70o Fahrenheit. Briefly, plasma was acidified with 5% perchloric acid containing 8 stable isotope internal standards. Acid-soluble supernatant was used for strong-cation exchange solid-phase extraction to capture cationic amine- containing metabolites. Extracted metabolites were further derivatized with isopropylchloroformate. Derivatives of metabolites were resolved using Agilent 1260 ultra-high pressure liquid chromatography and eluted with a gradient of water and isopropanol (65% v/v). An Agilent 6490 triple quadrupole mass spectrometer was used to detect resolved analytes and quantify them using authenticated external and internal standards. 2) Sphingolipidomics by electrospray tandem mass-spectrometry by validated techniques [17] was used to identify sphingolipid metabolites, including ceramides.

Statistical analysis

De-identified baseline and study completion data points, paired by participant study ID, were entered into SPSS (IBM Statistics, Seattle, WA). Descriptive analyses of the study cohort were summarized and results for Teens and PACs, as well as for combined CONT and INT groups, were compared by unpaired Student t-test. Continuous physical and metabolic variables were tested for normality by examining the skewness, kurtosis and the Shapiro Wilk tests and transformed as necessary before analysis. Most of the variables in our data were normally distributed. Log transformations were conducted for the continuous physical and metabolic variables that were skewed to make them as normal as possible. The Shapiro-Wilk tests show (with p-value > 0.05) that all of the transformed variables except two (i.e. small and medium LDL particle concentrations) are approximately normal. The measures of metabolite concentrations used for principal component analysis (PCA) were Z-transformed to render them normally distributed on the same scale with mean of zero and standard deviation of one. Pre-post change in absolute metabolite levels were compared by repeated samples paired t-test. Baseline z-scores of metabolites were subjected to principal component analysis (PCA) without rotation. PCA is an unsupervised analysis that aims to decrease the complexity of data by reducing variables to a smaller number of principal components (PC). Each of the PCs were vectors of metabolite contributions. A direct oblimin rotation was used and 6 factors before the bend in the scree plot (S1 Fig), and eigenvalues >1 were retained. Component scores for each participant were calculated with a standardized scoring coefficient. A PCA model with oblique rotation was tested to examine the factor correlation matrix. Since none of the factor correlations were over ± 0.32, indicating factor correlations are not driven by the data, we estimated a model with orthogonal rotation to reveal a simplified structure with interpretable factor loadings.

To identify metabolite patterns of interest, bivariate Pearson-correlation analysis was performed between percent changes in metabolites and clinical biomarkers of CMR (dyslipidemia and insulin resistance) and with lipoprotein particle subfraction distributions separately. A generalized estimating equation procedure determined the significance of longitudinal changes in PC scores and their contributing metabolites using age and gender as co-variates. Wald Chi-square tests determined the significance of pairwise differences in treatment responses. Statistical analyses were performed using IBM SPSS Statistics version 25 (IBM, Armonk, NY, USA) and Rstudio version 1.1.456, and p-values <0.05 were considered statistically significant for the differences in traditional biomarkers. A Bonferroni correction for multiple comparisons adjusted the p-value set as significant to <0.002 for the correlational analyses with amino acid and ceramide metabolites.

Results

Cohort description

The study cohort was predominantly female, 61% Hispanic (half who identify as White and half as American Indian), 25% Nonhispanic Black, 8% Hispanic Black. and 6% Nonhispanic White, with more Hispanic Not Black than Nonhispanic Black participants randomized by chance to the INT group (Table 1). Full assessment of physical, behavioral, metabolic, and metabolomic status was conducted at baseline and study completion on the cohort of 36 participants (S1 Data).

There was excellent attendance in both INT and CONT groups with all Teens and PAC participating in more than 78% of group sessions and 100% of baseline and follow-up assessment visits. There was considerable prevalence of obesity, dyslipidemia, inflammation and insulin resistance in all participants and mild hypertension in adult participants at baseline (Table 1). Mean baseline BMI was over 30 mg/m2 in all participants, did not differ by sex, but was significantly higher in the CONT group relative to INT. The mean CRP was over 28.6 nmol/L (3 mg/L) in all participants but higher in Teen participants assigned to the CONT group than to the Teen INT group, and between the combined CONT and INT groups. Despite these random differences in BMI and CRP, central adiposity (WHtRatio), blood pressure, lipids, insulin resistance and inflammation were similar in both groups, suggesting closely matched CMR profiles. Although in population studies, Waist Circ is greater in males as compared with females, only one of the 18 PAC was male; while his Waist Circ fell below the adult female PAC mean, it was not the lowest recorded among PAC. The average Waist Circ for Teen males (106.3 ± 19.2 cm) vs female (103.8 ± 11.8 cm) Teens did not differ significantly (p = 0.72) so Waist Circ data are not reported by sex. Compliance with nutrient bar intake by self-report was 85.8 ± 11.1% and 86.7 ± 13.8% among INT group adults and Teens respectively.

Clinical changes in CVD risk factors following intervention

The quality of self-report diets assessed by Block food frequency questionnaire was poor in all participants at baseline (S2 Data). Modest but significant improvement in CONT PAC was evident in categories of saturated fat intake and added sugar (less sugared beverage) but trends in the same direction did not reach significance in INT PAC or in Teens from either group. Average daily servings of dietary fruit (1.4 for PAC, 1.1 for Teens), vegetables, not including potatoes (1.5 for PAC, 0.9 for Teens), and fiber (10 grams for PAC, 8.5 grams for Teens) were comparably low in both CONT and INT groups and did not change in any age subgroup (full nutritional data are in S1 and S2 Data). Self-report activity increased in all participants in both groups (Table 2). Weight was stable, even in the INT group despite the addition of 220 kcal in two daily nutrient bars. Good compliance with nutrient bar intake was suggested by a significant increase in plasma 25 hydroxy Vitamin D (Vit D) in the INT group PAC and Teens, but neither CONT subgroup. There were no other consistent effects of nutrient bar supplementation evident in traditional CMR biomarkers, nor were there significant within- or between- group changes in any anthropometric measures in combined CONT and INT groups.

Table 2

Effects of interventions on anthropometric and clinical measures.

	CONTROL				INTERVENTION
	PAC		Teen		PAC		Teen
	Baseline	Post	Baseline	Post	Baseline	Post	Baseline	Post
Activity Score	8.6 (6.9)	14.8 * (2.8)	12.1 (4.9)	14.3 * (4.0)	8.2 (3.6)	10.6 * (3.5)	10.3 (4.2)	14.3* (3.7)
Vit. D (nmol/L)**	37.4 (10.0)	40.6 (8.5)	49.7 (21.2)	57.2 (18.4)	47.9 (14.2)	74.4 (26.7)*	53.4 (16.2)	72.6 (24.7)*
Weight (Kg)	103.7 (14.9)	103.6 (15.4)	102.9 (27.9)	103.9 (28.6)	82.9 (18.6)	83.2 (18.7)	85.3 (8.7)	86.1 (8.8)
Waist Circ (cm)	115.9 (9.1)	110.5 * (10.7)	106.9 (19.3)	106.6 (18.3)	102.0 (16.9)	101.5 (13.7)	103.0 (10.3)	103.1 (9.60
WHRatio	0.71 (0.06)	0.67* (0.07)	0.63 (0.10)	0.63 (0.08)	0.63 (0.11)	0.63 (0.10)	0.64 (0.09)	0.63 (0.08)
BMI (kg/m2)	38.6 (5.2)	38.5 (5.1)	35.9 (8.1)	36.1 (8.5)	31.8 (5.9)	31.9 (6.0)	32.2 (4.6)	32.5 (4.8)
Adipo (ng/ml)	1556 (1389)	1933 (1738)	1313 (588)	1212 (433)	2506 (1654)	3134 (2584)	1296 (587)	1193 (451)
CRP (nmol/L)	43.8 (25.7)	59.1 (51.4)	57.1 (30.5)	34.3 (25.7)	31.4 (16.2)	41.0 (29.5)	27.6 (22.8)	34.3 (23.8)
Glucose (nmol/L)	5.5 (0.2)	5.6 (0.5)	5.3 (0.3)	5.4 (0.5)	5.4 (0.4)	5.4 (0.4)	5.1 (0.3)	5.2 (0.5)
Insulin (pmol/L)	21.1 (13.5)	18.4 (8.8)	29.9 (21.2)	29.4 (15.9)	12.8 (4.6)	12.5 (7.4)	25.8 (12.8)	29.5 (21.9)
HOMA	5.3 (3.6)	4.6 (2.1)	7.2 (5.5)	7.2 (4.1)	3.1 (1.3)	3.1 (2.0)	5.9 (3.0)	6.8 (5.2)
TC (mmol/L)	5.1 (1.1)	5.0 (1.)	4.2 (0.8)	4.1 (0.9)	5.2 (0.5)	5.4 (0.6)	4.5 (1.0)	4.6 (1.0)
TG (mmol/L)	1.3 (0.6)	1.3 (0.6)	1.4 (0.9)	1.5 (1.1)	2.0 (1.8)	1.7 (0.9)	1.2 (0.4)	1.3 (0.6)
LDL (mmol/L)	3.2 (1.0)	3.1 (1.3)	2.5 (0.6)	2.3 (0.6)	3.0 (0.4)	3.1 (0.4)	2.8 (0.9)	2.7 (0.7)
HDL (mmol/L)	1.3 (0.3)	1.3 (0.3)	1.1 (0.2)	1.1 (0.2)	1.4 (0.3)	1.4 (0.2)	1.2 (0.3)	1.2 (0.)3
TGHDLR	2.4 (1.9)	2.5 (1.7)	2.9 (2.1)	3.5 (2.9)	4.3 (6.3)	2.9 (2.2)	2.5 (1.1)	2.6 (1.1)
SBP **T (mm Hg)	127 (15)	124 (15)	112 (8)	117 (9)	120 (11)	119 (14)	117 (8)	109* (10)
DBP (mm Hg)	77 (9)	80 (9)	66 (7)	65 (7)	79 (10)	76 (8)	65 (9)	65 (8)
HR	67 (13)	69 (10)	75 (7)	80 (8)	72 (9)	64 (8)	70 (10)	73 (11)

Values are mean ± SD (standard deviation).

* Denotes statistically significant differences (p < 0.05) by paired student’s t-test within a group.

** Denotes statistically significant (p < 0.05) time by group interactions for the designated variable for both PAC and Teens (Vitamin D levels) unless specified (T for Teens only, SBP).

Abbreviations as listed with Table 1.

Subgroup analysis of PAC and Teens separately showed some favorable changes in both CONT and INT groups. PAC CONT participants decreased Waist Circ (Table 2). Among Teen participants, CRP trended lower in CONT but did not reach significance (p = 0.1). Among Teens, systolic blood pressure decreased significantly by 7% (p = 0.03) in INT only, trending upward in CONT Teens (p = 0.08) with a significant between-group difference.

Baseline correlations between metabolomic profiles and clinical biomarkers

Previous bar supplementation trials showed that chronic inflammation (CRP > 14.3 nmol/L or 1.5) was associated with slower response in metabolic improvement. All participants, both Teens and PACs in this study, met criteria for obesity and had baseline CRP > 14.3 nmol/L. To test the hypothesis that subtle shifts in metabolism may precede changes in traditional biomarkers, a targeted analysis of plasma ceramides and amino acids was performed.

Baseline metabolite data were z-transformed and subjected to PCA analysis. The top six PCs extracted, each composed of distinct sets of linked metabolites, explained 59.3% of the total variance in the data set (Table 3). PC1, which explained 16.7% of the variance, was composed exclusively of ceramides and ceramide-1-phophates; PC2 of amino acid metabolites; PC3 of sphingomyelins; PC4 of dihydroceramide species, glucosylceramide, and one specific ceramide subspecies Cer 18:1; PC5 of biomarkers of meat consumption (3-methylhistidine, 1-methylhistidine, and beta-alanine [18]) and transsulfuration and polyamine metabolites (cystathionine, putrescine); and PC6 of amino acids in antioxidant defense (notably glutathione).

Table 3

Pattern matrix of baseline principal component variables.

Metabolites	PC1	Metabolites	PC2	Metabolites	PC3	Metabolites	PC4	Metabolites	PC5	Metabolites	PC6
	(16.7)		(15.3)		(10.4)		(8.5)		(4.9)		(3.5)
Cer C22:0	0.88	Met	0.76	T. SPM	0.98	DHCer C22:0	0.89	Cysth	0.96	Sar	0.90
Cer C24:1	0.86	Tyr	0.65	SPM C22:0	0.97	DHCer C26:0	0.87	3-MH	0.95	Asp	0.83
Cer C20:0	0.84	Val	0.61	SPM C14:0	0.94	DHCer C24:0	0.86	Put	0.90	GSH	0.62
Cer C16:0	0.83	Trp	0.57	SPM C16:0	0.91	DHCer C16:0	0.80	1-MH	0.74
Cer C22:1	0.82			SPM C24:1	0.87	Cer C18:1	0.79	Beta-Ala	0.63
T. Cer	0.82			SPM C22:1	0.86	DHCer C20:0	0.76
Cer C14:0	0.80			SPM C24:0	0.85	DHCer C14:0	0.76
C1P C18:0	0.79			SPM C18:1	0.78	DHCer C18:0	0.75
Cer C24:0	0.72			SPM C20:0	0.68
Cer C26:1	0.59

Numbers in () denote % of total variance explained by each PCs.

Numbers below PC denotes correlation between metabolites and PC noted on each column heading.

Abbreviations: PC: Principal Component, Cer: Ceramide, T. Cer: Total Ceramide, C1P: Ceramide-1-phosphate, T. SPM: Total Sphinogomyelin, SPM: Sphinogomyelin, DHCer: Dihydroceramide, GlcCer: Glucosylceramide, Met: methionine, Tyr: tyrosine, Val: valine, Trp: tryptophan, Cysth: cystathionine, 3-MH: 3-methylhistidine, Put: putrescine, 1-MH: 1-methylhistidine, Beta-Ala: beta-alanine, Sar: sarcosine, Asp: aspartate, and GSH: glutathione.

Fig 2A shows bivariate correlational analysis between baseline PC1-PC6 scores with parameters of obesity, dyslipidemia, insulin resistance and inflammation. Within the full cohort, no significant correlations were observed between any PC scores and anthropometric measures of weight, BMI, WC, or waist-height ratio. In contrast, highly significant associations between specific PC scores and metabolic parameters of dyslipidemia, insulin resistance and inflammation were present. PC1 (ceramide-related metabolites) was positively correlated with fasting total cholesterol, triglycerides, and HDL (r = 0.41, 0.38, and 0.35 respectively, all at p ≤ 0.04), and adiponectin (r = 0.52, p = 0.002). PC 3 (sphingomyelin metabolites) was inversely associated with CRP (r = -0.52, p = 0.002). PC4 (dihydroceramide compounds) was inversely associated with diastolic blood pressure (DBP, r = -0.48, p = 0.004) and PC6 (including antioxidant glutathione) showed an inverse association with fasting insulin (r = -0.34 p = 0.04) and HOMA (r = -0.35, p < 0.05).

Fig 2

Correlograms for Principal Component (PC) scores with A. traditional biomarkers of cardiometabolic risk and B. Lipoprotein subspecies. Color-coded r values indicated on the y axes to the right. Positive correlations are depicted in blue and negative correlations in red. The strength of correlation is represented by the size as well as the intensity of the color of the spheres plotted. Distinct associations are shown between PC1 (sphingolipid ceramide) metabolites with parameters of dyslipidemia (total cholesterol, triglyceride and HDL) in 2A (shown if p < 0.05) and small LDL IVc and IVa particles in 2B (shown if p < 0.01). PC3 and PC4 showed non-overlapping associations with CRP and diastolic blood pressure (DBP), respectively. Abbreviations: SBP–Systolic blood pressure, DBP–Diastolic blood pressure, TC- Total Cholesterol; TG–Triglyceride; LDL-Low Density Lipoprotein cholesterol; HDL–High Density Lipoprotein cholesterol; TGHDLR–Triglyceride to HDL-cholesterol ratio, Adipo- High Molecular Weight Adiponectin, FGB–Fasting blood glucose, ISN–Insulin, HOMA-IR–Homeostatic model assessment of insulin resistance; CRP–high sensitive C-reactive protein; VitD– 25 hydroxy vitamin D; PC–principal component. Lipoprotein sub-species include HDL3-2a (small) and HDL2b (large); LDLIVc, IVb, IVa, IIIb (Very Small); LDLIIIa (Small); LDL IIb (Medium); and LDLIIa and I (Large); intermediate density lipoprotein–IDL, IDLII (small) and IDLI (large); very low density lipoprotein–vLDL, vLDLsm, Int, Lg (small, Intermediate and Large); Diam—Diameter.

In Fig 2B positive association patterns are illustrated between PC1 scores and atherogenic lipoprotein particles, specifically very small LDLIVc (r = 0.53 p = 0.001), LDLIVa (r = 0.44 p = 0.009) and all three subclasses of VLDL (large, r = 0.51 p = 0.002; medium, r = 0.48 p = 0.004, and small, r = 0.45 p = 0.006). The PC1 score was also associated with very small LDL IVb (r = 0.42, p = 0.11) and inversely with LDL diameter (r = -0.35, p = 0.04), both above p value threshold set at 0.01. It is also of note that the baseline relationship between PC1 Cers and HDL was driven primarily by association with the small HDL32a subclass (r = 0.33, p < 0.05) not large HDL2b.

Metabolomic and lipidomic changes observed with intervention

We next determined how increased physical activity by itself or with nutrient bar supplementation modulated these baseline metabolomic parameters. Analysis of PC change was performed with adjustments of potential covariates including age, gender, and baseline CRP. While there was no change in the plasma total ceramide pool, results showed significant divergence in several key ceramide species within PC1 between CONT and INT.

Results show significant increases in C14:0 (15%), C20:0 (20%), C22:0 (14%) and C24:1 (17%) ceramides and a nearly significant increase in C16:0 (14%, p = 0.055) in CONT. In contrast, C24:1 ceramide significantly decreased in the INT group by 18%; all other ceramides trended lower or remained unchanged (Table 4). Analysis of group by time interactions showed significant between group differences in all of these C14:0, C16:0, C20:0, C22:0 and C24:1 ceramide species.

Table 4

Intervention effects on fasting plasma ceramides.

Measure			Mean	SD	p values
					Within group	Time x group
Cer C14:0	CONT	Pre	0.06	0.020	0.050	0.007
		Post	0.069	0.020
	INT	Pre	0.069	0.022	0.070
		Post	0.062	0.018
Cer C16:0	CONT	Pre	0.115	0.036	0.055	0.024
		Post	0.131	0.036
	INT	Pre	0.122	0.031	0.233
		Post	0.115	0.031
Cer C18:0	CONT	Pre	0.212	0.072	0.727	0.192
		Post	0.216	0.068
	INT	Pre	0.217	0.072	0.179
		Post	0.195	0.067
Cer C18:1	CONT	Pre	0.033	0.008	0.211	0.058
		Post	0.035	0.008
	INT	Pre	0.033	0.009	0.152
		Post	0.031	0.009
Cer C20:0	CONT	Pre	1.094	0.412	0.019	0.045
		Post	1.318	0.480
	INT	Pre	1.184	0.398	0.745
		Post	1.158	0.461
Cer C20:1	CONT	Pre	0.008	0.004	0.279	0.325
		Post	0.009	0.004
	INT	Pre	0.009	0.004	0.745
		Post	0.009	0.004
Cer C22:0	CONT	Pre	1.635	0.688	0.104	0.040
		Post	1.864	0.688
	INT	Pre	1.748	0.662	0.206
		Post	1.593	0.666
Cer C22:1	CONT	Pre	0.019	0.008	0.501	0.117
		Post	0.02	0.008
	INT	Pre	0.021	0.009	0.083
		Post	0.019	0.009
Cer C24:0	CONT	Pre	7.317	2.876	0.534	0.356
		Post	7.945	3.112
	INT	Pre	6.918	2.773	0.476
		Post	6.585	3.005
Cer C24:1	CONT	Pre	0.814	0.372	0.139	0.014
		Post	0.956	0.372
	INT	Pre	1.00	0.358	0.042
		Post	0.82	0.358
T. Cer	CONT	Pre	11.496	4.292	0.289	0.144
		Post	12.781	4.524
	INT	Pre	11.534	4.141	0.288
		Post	10.775	4.365

Pre-post changes for key ceramide (Cer) species are shown for CONT (Control) and INT (Intervention) groups with p values for both within group and between group (Time x group) differences. Metabolites with significant between group differences are shaded in light gray.

CI—confidence interval of within group changes; SD–standard deviation; T. Cer- Total Ceramide pool.

Sphinganine and sphingosine are essential substrates for ceramide synthesis through de novo synthesis and salvage pathways. In CONT only, levels of sphinganine and sphingosine increased significantly by 59% and 50%, respectively resulting in significant pairwise differences between CONT and INT groups (Table 5).

Table 5

Intervention effects on plasma sphingolipid bases.

	CONT			INT
Sphingoid Bases	Baseline (nmol/L)	Pairwise Diff (nmol/L)	95% CI (nmol/L)	Baseline (nmol/L)	Pairwise Diff (nmol/L)	95% CI (nmol/L)	Time x group p values
Sphingosine	38.2 ± 4.7	+19.4*	+1.0 - +37.8	241.9 ± 82.7	-10.2	-95.1 - +74.7	0.02
Dihydro-sphingosine-1-phosphate	5.2 ± 0.7	+2.4	0.0 - + 4.7	4.2 ± 0.4	+1.3	-0.3 - + 3.0	0.05
Sphingosine -1-phosphate	282.9 ± 25.1	+88.8*	+0.4 - +177.1	256.8 ± 22.7	+130.5*	+38.4 - +222.5	0.007

*denotes significant within group changes (bolded) before and after intervention at p ≤ 0.002.

Time x group denotes significance of between group changes.

CI- confidence interval of within group changes.

Sphingosine-1-phosphate (S1P) is a terminal breakdown product of ceramide and an important anti-inflammatory and vascular signaling lipid mediator associated with lipoproteins, particularly HDL, in plasma. As shown in Table 5, S1P levels increased differentially in both CONT (31%; p = 0.05) and more so INT (51%; p = 0.03) participants. The S1P increase in the INT group differed from the degree of change in CONT at p = 0.007.

Table 6 lists amino acid metabolites with significant within group and between group changes. Results show that among CONT, fasting levels of proline, ornithine, lysine, alanine and threonine trended upward. In contrast, the nutrient bar INT significantly altered amino acid metabolism, such that concentrations of several gluconeogenic, sulfur redox and urea cycle intermediates decreased, notably fasting levels of serine, proline, aspartate, cystathionine, alanine, glutamine methionine, and citrulline. A small but significant increase in histidine concentration associated with favorable nitrogen balance, improved bioavailability of arginine, and an increase in the Fischer ratio the molar ratio of BCAA (leucine, valine, isoleucine) and aromatic amino acids (AAA, phenylalanine and tyrosine), indicative of favorable vascular and liver functions respectively, were observed in the INT group only.

Table 6

Differential changes in plasma amino acid metabolites following CONT and INT interventions.

	Control			INT
Amino Acids	Baseline (μmol/L)	Pairwise Diff (μmol/L)	95% CI (μmol/L)	Baseline (μmol/L)	Pairwise Diff (μmol/L)	95% CI (μmol/L)	Time x Group
Serine	127.9 ± 9.0	+7.4	-13.5 - +28.4	157.0 ± 10.0	-33.4*	-50.5 - -16.3	0.0001
Proline	186.5 ± 9.3	+23.9*	+2.1 - +45.8	249.8 ± 12.7	-50.6*	-73.0 - -28.3	0.0001
Aspartate	25.1 ± 2.1	+2.2	-3.1 - +7.4	38.6 ± 2.5	-10.9*	-16.4 - -5.3	0.0001
Cystathionine	1.0 ± 0.2	-0.04	-0.2 - +0.1	0.8 ± 0.04	-0.3*	-0.4 - -0.2	0.0001
Sarcosine	19.2 ± 2.1	-1.6	-7.0 - +3.8	29.6 ± 1.7	-10.1	-14.1 - -6.2	0.0001
Ornithine	84.8 ± 8.3	+41.8*	+9.9 - +73.6	122.1 ± 8.0	+8.3	-35.9 - +52.7	0.001
Arg Bioavail Ratio	0.6 ± 0.06	-0.2	-0.3 - -0.05	0.4 ± 0.03	+0.3*	+0.05 - +0.5	0.001
Lysine	245.6 ± 11.7	+38.1*	+0.9 - +75.4	300.3 ± 13.8	+0.7	-26.7 - +28.1	0.002
Alanine	221.8 ± 13.5	+40.1*	+11.4 - +68.8	284.1 ± 18.3	-40.7*	-72.8 - -8.7	0.003
Glutamine	537.3 ± 29.5	-17.2	-85.2 - +50.8	594.1 ± 26.7	-89.4*	-140.7 - -38.1	0.007
Threonine	176.4 ± 14.8	+30.7*	+0.9 - +60.4	177.8 ± 10.2	-9.5	-34.4 - +15.4	0.008
Methionine	25.0 ± 1.4	-0.3	-2.8 - +2.2	27.9 ± 1.0	-3.0*	-4.9 - -1.3	0.01
Fischer Ratio	1.9 ± 0.8	+0.1	-0.01 - +0.4	1.7 ± 0.06	+0.2*	+0.05 - +0.3	0.01
Citrulline	60.5 ± 5.2	+11.5	-6.8 - +29.7	89.4 ± 9.7	-33.1*	-58.2 - -8.0	0.02
Histidine	20.8 ± 1.1	+1.5	-0.3 - +3.3	18.9 ± 1.1	+3.0*	+0.7 - +5.3	0.02
Tryptophan	39.1 ± 3.1	+5.8	-0.6–12.2	48.3 ± 2.2	+2.4	-2.3 - +7.1	0.02
Leucine	65.4 ±3.8	+9.5*	+3.7 - +15.4	71.7 ± 2.8	+1.5	-5.5 - +8.5	0.02
Phenylalanine	29.2 ± 1.9	+3.2	-2.6 - +9.0	35.8 ± 1.3	-3.7	-11.5 - +4.0	0.03

*denotes significant within group changes before and after intervention at p ≤ 0.002.

Time x Group denotes significance of between group changes.

CI- confidence interval of within group changes, Fischer Ratio—ratio of sum of branched chain amino acids (valine, leucine, isoleucine) over sum of aromatic amino acids (phenylalanine, tyrosine). Arg Bioavail Raio–ratio of arginine to sum of citrulline and ornithine.

To establish the clinical significance of differences in metabolism detected among INT participants, bivariate correlations between percent changes in ceramide and amino acid metabolites and clinical biomarkers of dyslipidemia and insulin resistance were considered (Fig 3). Positive correlations were observed between changes in TG and changes in ceramides composed of C14:0 (r = 0.6; p = 0.005), C16:0 (r = 0.7; p = 0.001), C24:1 (r = 0.7; p = 0.0004) and C26:1 (r = 0.6; p = 0.007) acyl chain lengths (Fig 3A). Specific associations were evident between percent changes in C18:1 ceramide and changes in very small LDL IVc (r = 0.61; p = 0.004) and IVb (r = 0.60; p = 0.005) particles (Fig 3B). In addition, changes in fasting blood glucose (FBG; Fig 3C) were positively associated with percent changes in citrulline (Cit; r = 0.61, p = 0.003) and cystathionine (Cysth; r = 0.57, p = 0.009), and the arginine bioavailability index (ArgBioavail; r = 0.70, p < 0.001), assessed by ratio of arginine to sum of citrulline + ornithine. Cysth change also correlated with change in insulin (r = 0.57, p = 0.009), and accordingly with HOMA (r = 0.63, p = 0.003). A strong positive correlation emerged for ornithine alone with change in CRP (r = 0.63, p < 0.001). Consistent with the baseline analysis, changes in amino acids did not associate significantly with lipid parameters.

Fig 3

Correlograms for specific changes in ceramides with A. Traditional biomarkers of cardiometabolic risk (CMR) and with B. Lipoprotein subspecies. The correlogram for amino acid metabolites with traditional CMR biomarkers is shown in C. Bivariate correlations between percent changes in ceramide and amino acid metabolites with percent changes in lipid and insulin resistance parameters were performed using pearson correlation analysis. Correlations with p values less than 0.01 are shown. Right colored bar indicates scale range of pearson correlation coefficients depicted. Dark red denotes negative, whereas blue indicates positive correlations observed. Arginine bioavailability ratio (ArgBioavail) was calculated using molar ratios of arginine and sum of citrulline and ornithine. Fischer ratio (FischerR) was calculated using the ratios of molar sum of branched chain amino acids (valine, leucine, and isoleucine and aromatic amino acids (phenylalanine, and tyrosine). Abbreviations: SBP–systolic blood pressure, DBP–diastolic blood pressure, TC- total cholesterol; TG–triglyceride; LDL-low density lipoprotein cholesterol; HDL–high density lipoprotein cholesterol; TGHDLR–TG to HDL-cholesterol ratio, FGB–fasting blood glucose, Ins–insulin, HOMA–homeostatic model assessment of insulin resistance; CRP–high sensitive C-reactive protein; LDL_IVb and IV_a (very small LDL); Cer—Ceramide, T. Cer—Total Ceramide, SPM—Sphinogomyelin, Gln—glutamine, Cit–citruilline, Ser–serine, Ala–alanine, Orn–ornithine, Met—Methionine, Pro–Proline, His–histidine, Cys–cysteine, Asp–aspartic acid, Cysth—cystathionine, Sar–sarcosine, DHSIP-Dihydrosphingosine-1-phosphate, SIP—Sphingosine-1-phosphate.

Discussion

Improved nutrition represents a lifestyle modification target to lower the growing global cardiometabolic disease burden [19], especially among youth and their families struggling with socioeconomic disparity and the many associated barriers to a healthy diet [20]. Even without economic barriers, lifestyle adherence is challenging [21]. Furthermore, strict dietary counseling and adherence for the management of hyperlipidemia has been associated with increased anxiety and depression in youth [22]. The nutrient bar strategy utilized in this study was developed to enable research on the impact of nutrition on CMR, by facilitating daily delivery of nutrients from foods missing from a standard American diet, essential fibers, minerals, vitamins, omega-3 fatty acids, and polyphenols, in a readily consumed, low-calorie, palatable nutrient bar [8-10]. Results herein show that a 2 month nutrient bar intervention together with increased physical activity improved systolic blood pressure in Teens only without other significant changes observed in traditional biomarkers, but was associated with metabolomic shifts in the full INT PAC and Teen group that correlate with favorable changes in cardiometabolic disease risk biomarkers.

Baseline metabolomic analysis identified ceramides as unique biomarkers of atherogenic dyslipidemia in both Teens and adults at heightened CMR (Fig 2). Although the metabolomic panel included BCAA and AAA previously implicated in CMR [23], only PC1, composed primarily of ceramide metabolites, associated with baseline TG levels and TG-rich VLDL subspecies, suggesting that ceramides may contribute to TG homeostasis. Fatty-acyl CoA is a substrate for both ceramide and TG biosynthesis and thus, excess ceramide and TG synthesis may occur under conditions of abundant fatty acid availability through both high fat diets or high refined carbohydrate diets and de novo lipogenesis.

In contrast to TGs that serve a passive fatty acid storage function, ceramides have unique signaling roles in mediating insulin resistance and mitochondrial bioenergetics [24]. Ceramides may therefore act as cellular energy sensors to regulate tissue fuel uptake and storage via insulin signaling. In obesity with chronic inflammation, excess ceramide, particularly C16:0 Cer, is associated with mitochondrial dysfunction, impaired sphingolipid cell signaling and insulin resistance [25]. Decreasing tissue ceramide through pharmacological or genetic manipulations can reverse or attenuate insulin resistance [26], providing support for a direct role of ceramides as critical part of fuel sensing circuitry in the body [27]. Ceramides may represent a sensitive biomarker of metabolic dysregulation when energetic supply chronically exceeds storage capacity. The current findings suggest that this tipping point may be influenced not only by absolute caloric intake but also by the dietary nutrient value.

Intensive lifestyle modification has been shown to improve CMR biomarkers in the obese [28]. In this trial, despite excellent participation in weekly exercise sessions, self-report of increased exercise between sessions in all subjects and both subjective and objective evidence of compliance with nutrient bar intake in the INT group, there were minimal changes in traditional parameters of dyslipidemia, inflammation, and insulin resistance (Table 2). Changes in the plasma metabolome were however able to signify highly significant differences between CONT and INT (Tables 4, 5 and 6).

The nutrient supplement INT modified effects of physical activity on plasma ceramide metabolism. Contrary to the expectation that physical activity would lead to lower ceramide formation by improving lipid oxidation capacity, significant increases in C14:0, C16:0, C20:0 and C22:0 and overall increasing trends with all ceramide species were observed in CONT participants. A study by Bergman and co-workers similarly found that acute exercise transiently increases serum ceramides in obese untrained subjects [29]. Significant increases in sphinganine (Table 5) in CONT may reflect increasesd de novo synthesis driven in part by a higher rate of lipolysis following exercise. Exercise increases lipolysis [30] but given obesity related decreases in muscle fatty acid oxidation (FAO) [31], a rise in free palmitate following exercise may, in the short-term, favor increased ceramide synthesis. The observed rise in ceramides in CONT may therefore reflect transient increases in fatty acid mobilization. Further training which would result in an increased FAO capacity would be expected to normalize these parameters.

Effects of exercise on ceramides were largely blunted when subjects consumed daily nutrient bars, suggesting enhanced muscle FAO by the provision of nutrient substrates missing from a typical American diet. The nutrient bar is enriched in polyphenols and omega-3 fatty acids, known to increase muscle peroxisome proliferator activated receptor coactivator 1 alpha activity and mitochondrial biogenesis [32], processes which would be expected to improve amino acid and fatty acid catabolism. In support of this mechanism, we also observed significant lowering of non-essential amino acids (NEAA), serine, proline, aspartate, glutamine, and alanine (Table 6). Lower NEAA’s may reflect improved substrate utilization and clearance of biomarkers associated with what has been termed the “metabolic gridlock” associated with obesity [33]. Furthermore, because serine is a substrate for synthesis of sphingolipids, its decrease may also have contributed to the lack of a rise in ceramide detected among INT participants. Alternatively, serine mobilization from other sites into plasma may be lessened when ceramide synthesis is not triggered.

Although clinical changes in TG, FBG, insulin and lipoproteins were not significant, we observed several significant correlations between changes in the plasma lipidome with changes in TG and physiologically interrelated pro-atherogenic small dense LDL particles (Fig 3). Plasma contains upwards of 200 distinct SPL species distributed across HDL, apoB-containing lipoproteins and albumin [34], but differential association with the most atherogenic lipoprotein subspecies has not been described. As with baseline correlations, the change to change correlations following the nutrient bar intervention were highly granular with specific ceramide species showing associations with changes in TG. Ceramides with C14:0, C16:0, C24:1 and C26:1 showed positive associations with changes in TG. C14:0 ceramide has been previously been implicated in nonalcoholic fatty liver disease (NAFLD) in adolescent children [35]. C16:0 ceramide inhibits mitochondrial FAO, oxidant production, and impairs Complex IV activity [25]. C24:1 ceramide significantly decreased in the INT group. Inhibition of synthesis of very long chain ceramides (≥C22) by deletion of fatty acid elongase 6 protects against hepatic steatosis in high fat diet induced obesity [36]. Very small LDL particles showed a highly specific correlation with C18:1 ceramide (Fig 3B). Deletion of ceramide synthase 1 responsible for C18 ceramide synthesis has also been shown to increase mitochondrial FAO and protect against high fat diet induced NAFLD [37]. Interaction between specific ceramide species and small LDL may be involved in the progression from NAFLD to nonalcoholic steatohepatitis (NASH). NASH is not only characterized by high concentrations of small LDL and a low mean LDL diameter but post hoc analyses from the Pioglitazone vs Vitamin E vs Placebo for the Treatment of Nondiabetic Patients with NASH (PIVENS) trial demonstrated that resolution of NASH was associated with an increase in the average LDL particle diameter and decreased small and very small LDL particles [38]. These lipoprotein subspecies parameters worsened in the participants who did not resolve their NASH. Larger LDL particles have improved affinity for LDL receptors and decreased retention in the subendothelial glycoprotein matrix, so would be expected to lead to decreased LDL and reduced atherogenesis. These findings collectively suggest that early changes in ceramide species among INT participants signal favorable shifts in lipid metabolism that may precede improvement in traditional clinical biomarkers.

In both CONT and INT participants, sphingosine-1-phosphate (S1P) levels increased, but with S1P levels rising higher among INT participants. A previous study showed that exercise increases plasma S1P levels and decreases erythrocyte ceramide levels and this effect was more evident in untrained subjects [39]. S1P is degraded by sphingosine-1-phosphate lyase (SPL). SPL dependent degradation of S1P is terminal step in sphinogolipid degradation critical for regulating the pool size of all sphingolipids. In blood, ~50–70% of S1P is bound to apolipoprotein M within the HDL complex and is important for its reverse cholesterol transport [40]. Decreased S1P bound to HDL is associated with higher risk for coronary artery disease [41]. Although there were no absolute changes in HDL subspecies, quantification of HDL-bound S1P may provide further insights as to whether increases in plasma S1P in our study participants contribute to improved antiatherogenic HDL functions.

Correlation analysis with changes in amino acids showed highly specific associations between changes in citrulline, arginine bioavailability (ratio of arginine to sum of citrulline + ornithine) and cystathionine and changes in measures of insulin resistance. Obesity and insulin resistance are associated with increased CRP, increased citrulline and decreased arginine bioavailability in Teens [42]. This may reflect increased nitric oxide demand associated with the inflammatory metabolic stress of insulin resistance. Interestingly in this cohort with high baseline CRP, only change in ornithine correlated with change in CRP. Arginase expression and increased Ornithine/Arginine ratio has been shown to associate with vasculopathies in humans and in animals fed a high fat, high sugar Western diet [43]. The favorable pairwise difference in arginine bioavailabity in the INT vs CONT groups may have contributed to sub-group specific improvement in systolic BP among INT Teens. Insulin sensitizing treatment with metformin plus pioglitazone for 3 months also improves arginine bioavailability in obese subjects with impaired glucose tolerance [44]. Change in Cystathionine (Cysth) was positively associated with change in glucose, insulin, and HOMA. Cysth is part of an alternate redox pathway to NOS, synthesized by cystathionine-beta-synthase (CBS) via the transsulfuration pathway. The expression of CBS in the liver observed in mice on a high fat diet has been proposed to be a defense mechanism against increased oxidative stress [45]. Decreased Cysth among INT participants may therefore reflect a lower burden of oxidative stress that could lead to improved insulin regulation.

Our study is limited by its relatively short observation period and the small sample size that precluded detailed analysis of data stratified by age. However, covariate analysis showed no significant effects of age with either the baseline correlations or with metabolic responses to the bar suggesting the metabolic impact of the nutrient bar intervention is conserved between the age groupings considered. The sample size was insufficient to consider hereditary factors. The findings in this predominantly minority study cohort may therefore not be generalizable. Furthermore, race and ethnicity, BMI and CRP differed by chance between INT and CONT groups although both groups exhibited comparable and considerable baseline CMR risk. The fact that lower total plasma Cers and specifically lower Cer 24:1 are described in persons of African American ancestry [46] who by chance represented a larger proportion of the CONT group, cannot explain either the differential rise in total Cers among controls, nor the significant lowering of the Cer24:1 species in the INT group only. A larger study will be required to stratify analyses by race, ethnicity and other hereditary factors that may affect sphingolipid metabolism to determine if this may have influenced results.

In summary, our study has uncovered several granular and unique associations between plasma ceramides and amino acid metabolites with clinical parameters of dyslipidemia, inflammation and insulin resistance. These findings suggest that remediation of essential nutrients typically lacking in western style diets should be considered an essential component of preventive interventions directed towards the alleviation of CMR. In this regard, the metabolomic changes as monitored herein are more sensitive indicators of favorable, but subtle shifts in metabolism. A lack of response in traditional clinical CMR biomarkers does not necessarily reflect a lack of compliance with behavioral interventions. The positive associations between changes in established traditional risk factors and changes in metabolomic biomarkers support the hypothesis that metabolomic biomarkers are sensitive prognostic indicators at the leading edge of response to lifestyle therapy, possibly mediated at the level of mitochondrial function [47]. Further larger and longer studies are needed to validate whether the changes observed in these analyses will predict and precede future favorable changes in traditional CMR biomarkers and the benefits to weight regulation that can be anticipated with improved metabolism [48].

Institutional Review Board approved protocol.

(PDF)

Click here for additional data file.

Flowchart detailing the study stages.

(DOCX)

Click here for additional data file.

Scree plot that justifies retention of six principal components in the study analyses.

(TIFF)

Click here for additional data file.

Full de-identified data set of study measures.

(XLS)

Click here for additional data file.

Block Food Frequency dietary data.

(XLS)

Click here for additional data file.

17 Apr 2020

PONE-D-19-34911

Randomized Nutrient bar supplementation improves exercise-associated changes in plasma metabolome in adolescents and adult family members at cardiometabolic risk

PLOS ONE

Dear Prof. Michele Mietus-Snyder,

Thank you for submitting your manuscript to PLOS ONE. After careful consideration, we feel that it has merit but does not fully meet PLOS ONE’s publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process.

We would appreciate receiving your revised manuscript by 30th July. When you are ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter.

To enhance the reproducibility of your results, we recommend that if applicable you deposit your laboratory protocols in protocols.io, where a protocol can be assigned its own identifier (DOI) such that it can be cited independently in the future. For instructions see: http://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols

Please include the following items when submitting your revised manuscript:

A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). This letter should be uploaded as separate file and labeled 'Response to Reviewers'.

A marked-up copy of your manuscript that highlights changes made to the original version. This file should be uploaded as separate file and labeled 'Revised Manuscript with Track Changes'.

An unmarked version of your revised paper without tracked changes. This file should be uploaded as separate file and labeled 'Manuscript'.

Please note while forming your response, if your article is accepted, you may have the opportunity to make the peer review history publicly available. The record will include editor decision letters (with reviews) and your responses to reviewer comments. If eligible, we will contact you to opt in or out.

We look forward to receiving your revised manuscript.

Kind regards,

Massimiliano Ruscica, Ph.D.

Academic Editor

PLOS ONE

Journal requirements:

When submitting your revision, we need you to address these additional requirements:

1.    Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming. The PLOS ONE style templates can be found at http://www.plosone.org/attachments/PLOSOne_formatting_sample_main_body.pdf and http://www.plosone.org/attachments/PLOSOne_formatting_sample_title_authors_affiliations.pdf

2. We note that you have included the phrase “data not shown” in your manuscript. Unfortunately, this does not meet our data sharing requirements. PLOS does not permit references to inaccessible data. We require that authors provide all relevant data within the paper, Supporting Information files, or in an acceptable, public repository. Please add a citation to support this phrase or upload the data that corresponds with these findings to a stable repository (such as Figshare or Dryad) and provide and URLs, DOIs, or accession numbers that may be used to access these data. Or, if the data are not a core part of the research being presented in your study, we ask that you remove the phrase that refers to these data.

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Yes

Reviewer #2: Yes

Reviewer #3: Partly

**********

2. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: Yes

Reviewer #2: I Don't Know

Reviewer #3: No

**********

3. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

Reviewer #2: Yes

Reviewer #3: Yes

**********

4. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #2: Yes

Reviewer #3: Yes

**********

5. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: Dear Editor,

I carefully read the article by Mietus-Snyder et al., which is overall original and interesting. The manuscript is well-written and balanced in its parts and the abstract is informative.

Some comments for the Authors:

- The limitations of the study (e.g. the small sample size and the short observation period) should be addressed in the discussion

- In the discussion, Authors might refer to the article by Cicero et al. doi: 10.1016/j.ijcard.2019.05.069

- In Table 1, waist circumference values (and related parameters) should be reported separately for male and female

- Figure 4 should be improved because it is now unreadable

Reviewer #2: Major revision

The Authors compared cardiometabolic risk factors change after consumption of 2 nutrient bars daily for 2 months in overweight subjects to determine 1) cross-sectional relationships in both adolescents and adults between traditional CMR biomarkers and amino acid and ceramide metabolites and 2) longitudinal changes within groups in the same CMR biomarkers following the lifestyle +/- nutrient bar intervention.

- Study design: why the Authors chose a 2-month intervention when in the background they stated that, with the same bar, change in cardiometabolic biomarkers in obese adolescents can be seen only after 6 months?

- It is not completely clear which is the added value of the PAC group, even more so that they are not randomly assigned in CONT or INT group and that the inclusion criteria are different.

- how the group randomization occurred? CRP and BMI are very different, with INT group being less inflamed and less obese. The Author stated that, despite these differences, other parameters are similar thus suggesting close CMR profile. However, studies that you cited in the introduction support the efficacy differences according to CRP and BMI levels.

- Table 1. Why PAIRED t-test to compare Parent vs Teen in the same group? Unpaired t-test is more appropriate.

- In general, statistics notes in tables are not clear. It is not clear which groups the Authors are comparing. Line 126 is not clear, and it is not clear the meaning of each symbol used to denote statistically significant. b letter to denote significant differences is not defined in figure legend

- Please, use correct symbols for metric unit. Meter is not in capital letter, while liter and kilograms are.

- Line 223-224-225 the quality of self-report diets assessed by food frequency … however there is no data reported. It could be useful to add a questionnaire analysis (as supplemental)

- it is not clear why table 1 and table 2 are differently organized. In table 1 CONT vs INT are reported, while in table 2 TEEN and PAC.

Minor points:

- Please revise all the abbreviations along the paper. They should be defined the first time they appear (e.g. line 53 BMI). Please be consistent with abbreviations: Line 126 – what does “C” mean? Is it CONT group? Line 153 WHtR/WHRatio in table 1 and many others.

- Please revise typos (e.g. line 87 “adolecents”; line 89 “nutritient”)

Reviewer #3: The manuscript addresses an interesting topic. Several statistical methods are combined to achieve the goal and answer to the well-posed research questions. The results are also rather interesting, but several model's assumptions are overlooked. My overall feeling is that the data are of interest, but the work requires a deep statistical revision, possibly by an expert.

1. Rotation is more usually used in factor analysis than in PCA. PCA can be used as an ad hoc approximation to factor analysis. In factor analysis, the objective is to find a small number underlying common factors which generate the observed variables. It is assumed that each observed variable is generated as a linear function of the underlying factors, plus a unique random error term. The underlying factors are assumed to be orthonormal. The solution is a set of factor loadings (if there are m factors and n variables this will be an n by m matrix) and a set of n unique variances. The maximum likelihood solution is undefined down to an orthogonal transformation (i.e. rotations) of the factor loadings. That is, any rotatoion of the factor loadings is an equally good solutions.

So it makes sense to rotate these loadings in a way which maximises ease of interpretation. Different rotation algorithms are based on different approaches to what is easiest to interpret. If you rotate the solution of a PCA it is no longer a PCA! Please, clarify this point, as this is crucial to understand the procedure you are considering. According to the introduction of the GEE approach, I am more incline to believe that you considered a factor model. There are several methods for estimating the factor loadings and communalities, including the principal component method, principal factor method, the iterated principal factor method and maximum likelihood estimation. The principal component method is one of the most common approaches to estimation and will be employed on the rootstock data seen in previous posts. The principal component method is rather misleading in its naming that no principal components are calculated. Please, mention all the drawbacks in using oblique rotations. The results obtained by an oblique rotation will be less likely to be replicated by future studies (this occurs because of sampling error).

2. A graphical representation of the dimensionality reduction methods should be provided. Please, provide the scree plot and info on the proportion of variance explained by the components. Similarly, please provide the individual and variable graphs in the lower dimensional space. Maybe, I also miss something, and how the scores are computed is rather unclear. In principle, you have one score per-component; here, one common score is considered. Please, clarify.

3. The sample size is rather small. Please, provide evidence that the assumptions behind any performed statistical tests are fulfilled. This is also true for the GEE model. For the GEE approach, it is rather unclear which working matrix is considered and why. Both are crucial points. The authors must provide evidence that all the model's assumptions are fulfilled. Otherwise, inferential results are not reliable.

4. I cast some doubts on the computations provided in Table 1. The standard deviations (what does SEM mean?) for the "combined" data are rather unplausible, according the group-specific summary info.

**********

6. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: No

Reviewer #2: No

Reviewer #3: No

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files to be viewed.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email us at figures@plos.org. Please note that Supporting Information files do not need this step.

4 Aug 2020

A full letter with these comments has been uploaded but they are copied again here.

TITLE: Randomized Nutrient bar supplementation improves exercise-associated changes in plasma metabolome in adolescents and adult family members at cardiometabolic risk

Dear Dr. Ruscica,

Thank you for the opportunity to respond to the constructive reviews provided to our submission to PLOS One. We appreciate your generous window through 7/30/20. We are grateful that reviewers 1 and 2 agree that this is a technically sound piece of research and hope we have satisfied reviewer 3’s concerns about the rigor of the statistical assessment. We appreciate that all three reviewers agree we have been transparent with provision of the data on which our conclusions are based and have presented the results in intelligible language. We have performed all of the statistical analyses again with oversight from a senior biostatistician and have addressed all reviewer comments in the revised manuscript now respectfully submitted as both marked (with tracked edits) and unmarked (edits accepted) versions. We have made every effort to comply with the PLOS One Style requirements in accord with journal templates.

Before addressing the specific reviewer comments point-by-point below, we would like to respond to your editorial request to avoid reporting any “data not shown”. We have removed this phrase from the manuscript in the single place it had been inappropriately used (line 284). The preceding statement stated the lack of significant correlation for specific relationships between variables listed, and are of course supported by the full data set already provided as a supplemental table. We are adding to the original supplemental data set, an additional data set with our Block FFQ nutrition data, and the PC analysis scree plot, per reviewer request.

Our full responses are embedded below in bold to each of the reviewer comments and referenced by page/line of the resubmission.

5. Review Comments to the Author

REVIEWER #1: Dear Editor,

I carefully read the article by Mietus-Snyder et al., which is overall original and interesting. The manuscript is well-written and balanced in its parts and the abstract is informative.

Some comments for the Authors:

- The limitations of the study (e.g. the small sample size and the short observation period) should be addressed in the discussion

Response: We thank the reviewer for the comment on the overall original and interesting nature of this work and agree that our small sample size and relatively short observation period pose significant limitations. We had recognized these things in the discussion of the paper, opening the paragraph on study limitations with the statement (p 26 line 22) “Our study is limited by its small sample size…” and we have added the limitation also posed by the short observation period. We had already concluded that a prospective study with a longer observation period will be necessary to validate the hypotheses generated by this work but have added in accord with the reviewer’s accurate critique that future validation studies will also need to be larger (p27, lines 22).

- In the discussion, Authors might refer to the article by Cicero et al. doi: 10.1016/j.ijcard.2019.05.069

Response: Thank you for making us aware of this interesting article. Although it refers to a familial hypercholesterolemia cohort rather than the mixed dyslipidemia under consideration in our study, it adds important evidence to the challenge of motivating healthy dietary behavioral change in children. We have added this reference (#22) at the opening to our Discussion (p 22, lines 1-3).

- In Table 1, waist circumference values (and related parameters) should be reported separately for male and female

Response: The reviewer’s point is well taken given average waist circumference based upon population data is greater in males as compared with females, but only one of the 18 Parent Adult Caretakers (PAC) was male. His WC of 98.0 cm actually fell below the adult female PAC average WC (108.8 + 15.6 cm) but well within the full range observed (81.0 to 136.7 cm). We did compare this metric for male (106.3 + 19.2 cm) vs. female (103.8 + 11.8 cm) TEENS and no significant difference was found (p = 0.72). Because the effort to present baseline data for both adult PAC and Teens in both INT and CONT groups already necessitates extensive baseline and follow-up tables 1 & 2, we respectfully did not add this additional categorization. We do however now mention our rationale – namely that WC did not differ by sex in this study cohort (p 12, lines 4-8)- and include that BMI did not differ by sex either (p 11, line 22).

- Figure 4 should be improved because it is now unreadable

Response: Figure 4 – now numbered as Figure 3 - has been regenerated with the new PC analyses at improved resolution. We have also reorganized Table 4 into Table 4a (for differential ceramide changes between Control and Intervention groups) and 4b (for changes in sphingolipid bases).

REVIEWER #2: Major revision

The Authors compared cardiometabolic risk factors change after consumption of 2 nutrient bars daily for 2 months in overweight subjects to determine 1) cross-sectional relationships in both adolescents and adults between traditional CMR biomarkers and amino acid and ceramide metabolites and 2) longitudinal changes within groups in the same CMR biomarkers following the lifestyle +/- nutrient bar intervention.

- Study design: why the Authors chose a 2-month intervention when in the background they stated that, with the same bar, change in cardiometabolic biomarkers in obese adolescents can be seen only after 6 months?

Response: This was the first trial of the nutrition bar in adolescents, temporally preceding the study of obesity asthma in adolescents referenced by the reviewer. We did not therefore know that traditional biomarkers would not respond within 2 months. Indeed, it was because of our lack of movement in traditional biomarkers at 2 months, that the subsequent trial in adolescent asthma was extended for a full 6 months, while we began exploring the post hoc analyses for change in sphingolipid ceramide biomarkers that may precede traditional biomarkers, as reported upon in this work. Unfortunately, neither metabolomics or lipidomics were performed on the longer adolescent trial.

- It is not completely clear which is the added value of the PAC group, even more so that they are not randomly assigned in CONT or INT group and that the inclusion criteria are different.

Response: Their inclusion was primarily to reinforce family level change for the adolescents – and we believe this study design contributed to the unusually good adherence and compliance observed.

- how the group randomization occurred? CRP and BMI are very different, with INT group being less inflamed and less obese. The Author stated that, despite these differences, other parameters are similar thus suggesting close CMR profile. However, studies that you cited in the introduction support the efficacy differences according to CRP and BMI levels.

Response: Although BMI and level of CRP certainly correlate, the differences in response to the nutrition bar previously described in adults were associated with the level of inflammation rather than BMI. It is important to note that any level of CRP elevation over 1.5 mg/L (SI units) was sufficient to blunt the early response to the nutrition bar without an obvious dose relationship. While the CRP, a correlate as noted of BMI, was significantly higher in the CONT vs INT cohorts in this study, both groups had a mean CRP exceeding 1.5 mg/L (= 14.3 nmol/L). As noted above, this was the first trial of this nutrient intervention in adolescents and it was not clear, a priori, whether or not younger age would negate the previously observed effect of inflammation on response. None-the-less we recognize and have noted in limitations that the imperfect randomization scheme in this small cohort. The lower end of the range of BMIs was equivalent in both cohorts (INT 25.2 to 46.5; CONT 25.1 to 50.7) but one dyad randomized to the CONT group with higher BMI’s of 48.9 and 50.7 for the adolescent and adult respectively, did skew the higher range limit and therefore the mean. If this one dyad is not included, the range of BMIs for the CONT group becomes 25.1 to 43.4 and there is no longer a significant difference between the two groups in BMI. It is a significant limitation as noted (p 27, lines 4-6), but we hope the reviewers agree, the findings remain provocative and worthy of further study.

- Table 1. Why PAIRED t-test to compare Parent vs Teen in the same group? Unpaired t-test is more appropriate.

Response: The Statistical Methods had been written with an emphasis on the pre-post analyses of changes in metabolites that were compared by paired t-tests. We have added clarification that the group comparisons in all initial descriptive analyses and any subsequent between group comparisons were done, as the reviewer correctly notes they should be done, with unpaired t-tests.

- In general, statistics notes in tables are not clear. It is not clear which groups the Authors are comparing. Line 126 is not clear, and it is not clear the meaning of each symbol used to denote statistically significant. b letter to denote significant differences is not defined in figure legend

Response: Thank you for a careful review of the tables and for identifying this lack of clarity. We have focused the statistical notes in the tables in efforts to describe significant differences more clearly. We no longer us letters a & b but simply one or two Asterix. The legend for Table 1 currently reads:

* Denotes statistically significant difference (p < 0.05) by unpaired student’s t-test between PAC (Parent Adult Caretakers) and Teen subgroups within Control (CONT) and/or Intervention (INT) groups.

** Denotes statistically significant difference (p <0.05) for the designated variable by unpaired student’s t-test between combined CONT and INT groups

And the legend for Table 2 currently reads:

* Denotes statistically significant difference (p < 0.05) by paired student’s t-test within a group.

- Please, use correct symbols for metric unit. Meter is not in capital letter, while liter and kilograms are.

Response: Thank you for this correction. These metric unit symbols have been corrected in both Tables and text.

- Line 223-224-225 the quality of self-report diets assessed by food frequency … however there is no data reported. It could be useful to add a questionnaire analysis (as supplemental)

Response: We have added additional descriptive data from the self-report diets to document the small number of significant changes and to better document the absence of substantive change in dietary behavior (p 12, lines 12-17). We also include the dietary data derived from the Block Food Frequency Questionnaires now as well as an additional supplementary data table of these analyses.

- it is not clear why table 1 and table 2 are differently organized. In table 1 CONT vs INT are reported, while in table 2 TEEN and PAC.

Response: We thank the reviewer for noting this unnecessary change in formatting a complex set of data. We have now organized both Table 1 and Table 2 in the same order.

Minor points:

- Please revise all the abbreviations along the paper. They should be defined the first time they appear (e.g. line 53 BMI). Please be consistent with abbreviations: Line 126 – what does “C” mean? Is it CONT group? Line 153 WHtR/WHRatio in table 1 and many others.

Response: We have carefully reviewed the full text and to ensure that all abbreviations are defined at first appearance and that they are used consistently in the text tables, and figures.

- Please revise typos (e.g. line 87 “adolecents”; line 89 “nutritient”)

Response: Both typos are corrected.

REVIEWER #3: The manuscript addresses an interesting topic. Several statistical methods are combined to achieve the goal and answer to the well-posed research questions. The results are also rather interesting, but several model's assumptions are overlooked. My overall feeling is that the data are of interest, but the work requires a deep statistical revision, possibly by an expert.

Response: The senior author, Dr. Jung Suh, who has an MPH in biostatistics, performed the original statistical analyses but in light of this reviewer concern, Dr. Suh repeated the full metabolomic and lipidomic analysis with consultation from a senior biostatistician in the first author’s institution, Dr. Jichuan Wang, whose expertise includes structural equation modeling, multilevel modeling, and principal components analyses for large data sets. His valuable contribution to this resubmission is now acknowledged, and the input reflected in the comments below.

1. Rotation is more usually used in factor analysis than in PCA. PCA can be used as an ad hoc approximation to factor analysis. In factor analysis, the objective is to find a small number underlying common factors which generate the observed variables. It is assumed that each observed variable is generated as a linear function of the underlying factors, plus a unique random error term. The underlying factors are assumed to be orthonormal. The solution is a set of factor loadings (if there are m factors and n variables this will be an n by m matrix) and a set of n unique variances. The maximum likelihood solution is undefined down to an orthogonal transformation (i.e. rotations) of the factor loadings. That is, any rotatoion of the factor loadings is an equally good solutions.

So it makes sense to rotate these loadings in a way which maximises ease of interpretation. Different rotation algorithms are based on different approaches to what is easiest to interpret. If you rotate the solution of a PCA it is no longer a PCA! Please, clarify this point, as this is crucial to understand the procedure you are considering.

According to the introduction of the GEE approach, I am more incline to believe that you considered a factor model. There are several methods for estimating the factor loadings and communalities, including the principal component method, principal factor method, the iterated principal factor method and maximum likelihood estimation. The principal component method is one of the most common approaches to estimation and will be employed on the rootstock data seen in previous posts. The principal component method is rather misleading in its naming that no principal components are calculated. Please, mention all the drawbacks in using oblique rotations. The results obtained by an oblique rotation will be less likely to be replicated by future studies (this occurs because of sampling error).

Response: In the revised manuscript, the PCA was implemented in the following way: First, a PCA model with no rotation was conducted to just extract the components. Second, the meaningful components were identified by checking eigenvalues and scree plot. Third, a PCA model with oblique rotation was tested to examine the factor correlation matrix. Since none of the factor correlations were over ± 0.32, indicating factor correlations are not driven by the data, we finally estimated a model with orthogonal rotation (that assumes the factors are not correlated) to reveal simple structure so that the factor loadings are easy to interpret.

2. A graphical representation of the dimensionality reduction methods should be provided. Please, provide the scree plot and info on the proportion of variance explained by the components. Similarly, please provide the individual and variable graphs in the lower dimensional space. Maybe, I also miss something, and how the scores are computed is rather unclear. In principle, you have one score per-component; here, one common score is considered. Please, clarify.

Response: In response to the reviewer’s comment, the scree plot is included with the supplemental materials of the revised manuscript. Because the elbow of the curve on this scree plot is closer to 6 rather than 5 principle components, we retained 6 components for further analysis. The Scree plot is included in the uploaded Response to Reviewers and is included with Supporting Data as part of this resubmission. It would not attach in this window.

The new PCA analysis changes Table 3 slightly (the full listing of the principal components), and therefore all analyses involving these components were redone, as were Figures 2 and 4 (the latter now becomes Figure 3 as noted below). All of the main findings of the original submission are confirmed. Table 3, due to increased width with a 6th PC included, is added as a screen shot to the manuscript with tracked edits. It has also been added for the Editors to the resubmission as a Word file under category of “other”.

3. The sample size is rather small. Please, provide evidence that the assumptions behind any performed statistical tests are fulfilled. This is also true for the GEE model. For the GEE approach, it is rather unclear which working matrix is considered and why. Both are crucial points. The authors must provide evidence that all the model's assumptions are fulfilled. Otherwise, inferential results are not reliable.

Response: Yes, the sample size is small. However, model estimation terminated normally. We had already explained in the manuscript that this was a post-hoc analysis undertaken when the anticipated change in traditional biomarkers was not observed. The findings described stand up to rigorous analysis. As noted above, we had acknowledged in the original manuscript the importance of validating the reported observations in a larger prospective study cohort. The small sample size is mentioned as a limitation of the study in the Discussion section.

The reviewer points out an important issue, i.e., working matrix specified for GEE model estimation. Various residual correlation structures (e.g., autoregressive, exchangeable, and unstructured) are available for GEE, and the model provides robust parameter estimates even though the correlation structure may be incorrectly specified. As a common practice, we specified an unstructured correlation structure in our GEE model.

We agree with the reviewer that oblique rotations have some drawbacks. As the reviewer points out, the results obtained by an oblique rotation will be less likely to be replicated by future studies. In addition, the definition for a factor loading is different depending on whether it is in a factor pattern matrix or in a factor structure matrix. Because of correlated factors, the factor loadings are not equivalent to the correlations between the observed variables and the factors. In the revised manuscript, orthogonal rather than oblique rotation was implemented (see responses to Comment 1).

4. I cast some doubts on the computations provided in Table 1. The standard deviations (what does SEM mean?) for the "combined" data are rather unplausible, according the group-specific summary info.

Response: The abbreviation SEM refers to Standard Error of the Mean and should have been defined. We agree with the reviewer that Standard Deviation (SD) provides a better indication of dispersion of data around the mean for normally distributed continuous variables, and now report the SD for all outcomes reported in Tables 1 and 2, as well as the absolute metabolite data reported in Table 4a.

We thank again all reviewers for their helpful critiques that have strengthened this research report.

Submitted filename: PLOS One_Response to Reviewers_cover letter_28JUL2020.pdf

Click here for additional data file.

15 Sep 2020

PONE-D-19-34911R1

Randomized Nutrient bar supplementation improves exercise-associated changes in plasma metabolome in adolescents and adult family members at cardiometabolic risk

PLOS ONE

Dear Dr. Mietus-Snyder,

Thank you for submitting your manuscript to PLOS ONE. After careful consideration, we feel that it has merit but does not fully meet PLOS ONE’s publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process.

Please submit your revised manuscript by 15 days. If you will need more time than this to complete your revisions, please reply to this message or contact the journal office at plosone@plos.org. When you're ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

Please include the following items when submitting your revised manuscript:

A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'.

A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'.

An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter. Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter.

If applicable, we recommend that you deposit your laboratory protocols in protocols.io to enhance the reproducibility of your results. Protocols.io assigns your protocol its own identifier (DOI) so that it can be cited independently in the future. For instructions see: http://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols

We look forward to receiving your revised manuscript.

Kind regards,

Massimiliano Ruscica, Ph.D.

Academic Editor

PLOS ONE

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. If the authors have adequately addressed your comments raised in a previous round of review and you feel that this manuscript is now acceptable for publication, you may indicate that here to bypass the “Comments to the Author” section, enter your conflict of interest statement in the “Confidential to Editor” section, and submit your "Accept" recommendation.

Reviewer #1: All comments have been addressed

Reviewer #3: (No Response)

**********

2. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Yes

Reviewer #3: Yes

**********

3. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: Yes

Reviewer #3: Yes

**********

4. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: No

Reviewer #3: Yes

**********

5. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #3: Yes

**********

6. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: Authors improved their manuscript following my suggestions. I have no more comment on it and suggest the warmly acceptation of the paper for publication.

Reviewer #3: I really appreciate the efforts to answer to my points. Detailed answered were provided, showing a deep knowledge on the statistical methods employed.

Only one minor, but substantial, point is still overlooked. The authors ackwonledge that the small sample size is a limitation of the study. However, all the employed tests are based on some assumptions (like the normality of the data, but not only) which must be checked to ensure a reliable inference. In the main text, I read "Continuous physical and metabolic variables were tested for normality by examining the skewness, kurtosis and the Shapiro Wilk tests and transformed as necessary before analysis". Please, provide evidence of the normality of the data and make clear which transformations are adopted and why.

**********

7. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: No

Reviewer #3: No

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email PLOS at figures@plos.org. Please note that Supporting Information files do not need this step.

24 Sep 2020

RESPONSE: In real-life datasets, variables do not always follow the normal distribution. All continuous physical and metabolic variables in our study were tested for normality. The results of the Shapiro-Wilk tests show that most of the variables in our data were normally distributed, but log transformations were conducted for the 11 of 31 continuous physical and metabolic variables that were skewed to make them as normal as possible. The results of the Shapiro-Wilk tests show (with p-value > 0.05) that all of the transformed variables except two (i.e. small and medium LDL particle concentrations) are approximately normal.

The measures of metabolite concentrations used for principal component analysis (PCA) were Z-transformed to render them normally distributed on the same scale with mean of zero and standard deviation of one. In addition, the generalized estimating equation (GEE) model used for our longitudinal data analysis is robust to data non-normality. The normality assumption of linear regression is about residuals [ε~ i.i.d. N(0, σ²)], and the GEE model does not necessarily assume normality of residuals.

We hope that this explanation will satisfy reviewer 3’s last remaining concern. We have added the following wording (p10, lines 12-18) to the current statement (in quotes here) to clarify this important point:

"Continuous physical and metabolic variables were tested for normality by examining the skewness, kurtosis and the Shapiro Wilk tests and transformed as necessary before analysis." Most of the variables in our data were normally distributed. Log transformations were conducted for the continuous physical and metabolic variables that were skewed to make them as normal as possible. The Shapiro-Wilk tests show (with p-value > 0.05) that all of the transformed variables except two (i.e. small and medium LDL particle concentrations) are approximately normal. The measures of metabolite concentrations used for principal component analysis (PCA) were Z-transformed to render them normally distributed on the same scale with mean of zero and standard deviation of one.

We thank you for your prompt consideration. This manuscript will support a pending grant review to further probe these findings if available for the study section to view when they meet mid-October.

Submitted filename: PLOS One_Resubmission2ResponseToReviewers_24SEP2020.docx

Click here for additional data file.

28 Sep 2020

Randomized Nutrient bar supplementation improves exercise-associated changes in plasma metabolome in adolescents and adult family members at cardiometabolic risk

PONE-D-19-34911R2

Dear Dr. Michele Mietus-Snyder,

We’re pleased to inform you that your manuscript has been judged scientifically suitable for publication and will be formally accepted for publication once it meets all outstanding technical requirements.

Within one week, you’ll receive an e-mail detailing the required amendments. When these have been addressed, you’ll receive a formal acceptance letter and your manuscript will be scheduled for publication.

An invoice for payment will follow shortly after the formal acceptance. To ensure an efficient process, please log into Editorial Manager at http://www.editorialmanager.com/pone/, click the 'Update My Information' link at the top of the page, and double check that your user information is up-to-date. If you have any billing related questions, please contact our Author Billing department directly at authorbilling@plos.org.

If your institution or institutions have a press office, please notify them about your upcoming paper to help maximize its impact. If they’ll be preparing press materials, please inform our press team as soon as possible -- no later than 48 hours after receiving the formal acceptance. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact onepress@plos.org.

Kind regards,

Prof. Massimiliano Ruscica

Academic Editor

PLOS ONE

Additional Editor Comments (optional):

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. If the authors have adequately addressed your comments raised in a previous round of review and you feel that this manuscript is now acceptable for publication, you may indicate that here to bypass the “Comments to the Author” section, enter your conflict of interest statement in the “Confidential to Editor” section, and submit your "Accept" recommendation.

Reviewer #3: All comments have been addressed

**********

2. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #3: (No Response)

**********

3. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #3: (No Response)

**********

4. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #3: (No Response)

**********

5. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #3: (No Response)

**********

6. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #3: (No Response)

**********

7. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #3: No

30 Sep 2020

PONE-D-19-34911R2

Randomized Nutrient bar supplementation improves exercise-associated changes in plasma metabolome in adolescents and adult family members at cardiometabolic risk

Dear Dr. Mietus-Snyder:

I'm pleased to inform you that your manuscript has been deemed suitable for publication in PLOS ONE. Congratulations! Your manuscript is now with our production department.

If your institution or institutions have a press office, please let them know about your upcoming paper now to help maximize its impact. If they'll be preparing press materials, please inform our press team within the next 48 hours. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information please contact onepress@plos.org.

If we can help with anything else, please email us at plosone@plos.org.

Thank you for submitting your work to PLOS ONE and supporting open access.

Kind regards,

PLOS ONE Editorial Office Staff

on behalf of

Dr. Massimiliano Ruscica

Academic Editor

PLOS ONE