Plant-Based No Added Fat or American Heart Association Diets, Impact on Cardiovascular Risk in Obese Hypercholesterolemic Children and Their Parents

Objective To perform a randomized trial to determine if there is cardiovascular disease (CVD) risk reduction from a plant-based no added fat diet (PB) and the American Heart Association Diet (AHA) in children. Study design Four-week (4/20/2013-5/18/2013) prospective randomized trial in a large Midwestern hospital system’s predominantly middle class outpatient pediatric practices. Thirty children (9–18 years old) parent pairs with a last recorded child BMI >95th percentile and child cholesterol >169 mg/dL were randomized to PB or AHA with weekly 2-hour classes of nutrition education. Results Children on PB had nine and children on AHA had four statistically significant (P<0.05) beneficial changes from baseline (mean decreases): BMI Z-scorePB (−0.14), systolic blood pressurePB (−6.43 mm Hg), total cholesterolPB (−22.5 mg/dL), low density lipoproteinPB (−13.14 mg/dL), hsCRPPB (−2.09 mg/L), insulinPB (−5.42uU/ml), myeloperoxidasePB/AHA (−75.34/69.23 pmol/L), mid-arm circumferencePB/AHA (−2.02/−1.55 cm), weightPB/AHA (−3.05/ −1.14kg) and waist circumferenceAHA (−2.96 cm). Adults on PB and AHA had seven and two respectively statistically significant (P<0.05) beneficial changes. The significant change favoring AHA was a 1% difference in children’s waist circumference. Difficulty shopping for food for the PB was the only statistically significant acceptability barrier. Conclusions PB and the AHA in both children and adults demonstrated potentially beneficial changes from baseline in risk factors for CVD. Future larger, long-term randomized trials with easily accessible PB foods will further define the role of the PB in preventing CVD.

or guardian also participated in the study and was assigned to follow the same diet that was given to his/her child to help with dietary compliance. Pregnant women were excluded from the study.
A computerized search of Cleveland Clinic medical records identified 1,278 potential participants ( Figure 1; available at www.jpeds.com). Eligible patients were invited by letter to participate in the study. Those interested contacted the principal investigator (PI) and were enrolled on a first come, first served basis. Informed consent was obtained from the participants who were eighteen years of age and older. Participants younger than eighteen years provided assent with parent/guardian approval. There was a time gap between the last recorded measurements, obtaining informed consent, and the start of the study. During this time gap before the start of the study 6 previously obese (BMI >95%) children had become overweight (BMI 85%-95%), and one hypercholesterolemic (>169 mg/dL) child's cholesterol had decreased to 169 mg/dL. Each child and parent pair received a fifty dollar stipend for each of the four weeks of the study.
Participants assigned to the PB were instructed to avoid all animal products and added fat, and to limit intake of nuts and avocado. 15 The AHA group was allowed 30% of calories from total fat, 7% of calories from saturated fat, less than 300 mg cholesterol and less than 1,500 mg of sodium daily. 17 All participants received standardized teaching at the time of consent to learn how to record a 24-hour dietary history. Participants completed two 3-day dietary histories consisting of two weekdays and one weekend day; one before the start of the study and one during the study. During the study, participants attended a total of 4 weekly 2-hour classes specific to their assigned diet consisting of one hour of nutrition education and one hour of cooking lessons with recipes provided.
Classes were led by acknowledged study collaborators. Weeks one and two focused on reading labels, where to buy food, food preparation, and how to stay on the assigned diets when eating away from home. Weeks three and four reviewed healthy food choices, the effects of diet on health, discussions of what worked and what did not work for the study participants. At the fifth and final study session, after all laboratory samples and measurements were obtained, participants had the option to attend an introductory class on the diet they were not assigned.
All analyses were performed in the Preventive Research Laboratory and Lab Diagnostic Core, Cleveland Clinic.
Measurements (height, weight, mid-arm circumference, waist circumference, and blood pressure) were also obtained at the start of the trial. BMI was calculated by dividing weight in kilograms by height in meters squared. Measurements of the physical activity of the children and adolescents were self-reported using the Physical Activity Questionnaire. 20 The PAQ consists of 9 questions which ask subjects to rate their physical activity for the previous 7 days, at different times of day and days of week, and how often they engaged in specific activities. All items are presented on a 5-point scale where 1 is low activity and 5 is high activity; the overall PAQ score is a mean of the 9 questions. All measurements were repeated at the completion of the study for comparison with baseline. Race/ethnicity was self-reported to help determine the comparability of the study groups. At the conclusion of the four-week trial, all participants completed a validated Food Acceptability Questionnaire, 21 which subjectively rated the ease of following their assigned diets and their general like or dislike of the diet.
The sample size of 15 adults and children per group was calculated to substantially exceed, even with a 20% drop-out rate, the 6-7 patients per group required to provide a power of 90% at a significance level of 0.05 to detect the within-group changes from baseline in total cholesterol described previously (mean ± standard deviation decrease of 60±26 mg/dl) 14 versus a null hypothesis mean decrease of 25±26mg/dl. We did not power our study to demonstrate statistically significant differences between two effective dietary interventions. Families were randomized to the two study groups in a 1:1 ratio in blocks of four families, stratified by the child's age group (age strata 9-13 years vs. 14-18 years). The randomization was performed by Ms. Worley using an SAS computer program between the end of enrollment and the first weekly session.
Demographics, comorbidities, and body measurements were collected in a REDCap database, 22 using double data entry. Laboratory values were provided in an Excel sheet. Diet journals were entered into and analyzed using Nutrition Data System for Research (NDSR) software.

STATISTICAL ANALYSES
Mean daily nutrients were computed for each subject within the pre-study and during-study periods. The BMI of the children was converted to age-and sex-adjusted percentiles and their corresponding z-scores; statistical analysis was performed on the z-scores. Parent and child subjects were analyzed separately because their outcomes were likely to be correlated, given genetic and environmental similarities. For the primary analysis, within-group changes from baseline to week 4 were computed and their means estimated with 95% confidence intervals; log-transformations of baseline and week 4 values of variables were performed as needed. For the secondary analysis, the PB and AHA groups were compared at the end of the trial, adjusting for baseline values, using analysis of covariance (ANCOVA) models. Where needed to meet model assumptions, both the baseline and week 4 values of variables were log-transformed. Study groups were compared on responses to each question on the food acceptability questionnaire using Fisher's exact tests and Cochran-Armitage trend tests. Sample sizes for individual variables reflect missing data. All analyses were performed on a complete-case basis. All tests were two-tailed and performed at a significance level of 0.05. SAS 9.2 software (SAS Institute, Cary, NC) was used for all analyses and R 3.0.0 (The R Foundation for Statistical Computing) was used for plots.

RESULTS
Sixteen families were randomized to the PB and 14 families to the AHA. Two families, both in the PB group were lost to follow-up. One discontinued after the first week, and the other after the third week. Both families were excluded from the analysis because no end of study data was available. The final study cohort consisted of 28 families, 14 in each group ( Figure  1). There were no significant between group differences in baseline demographic, nutrient, and clinical outcomes (Tables I-III; Tables II and III available at www.jpeds.com).
The total energy intake and the intake of almost all measured nutrients significantly decreased in children and adults in both groups, and dietary fiber intake significantly increased only in PB diet group (both children and adults) based on dietary histories completed during the study compared with those completed at baseline ( Figure 2 and Table  II). When comparing the PB and AHA groups during the study, children and adults in the PB group had a significantly lower intake of total protein, animal protein, cholesterol, total saturated fat, vitamin D, vitamin B12, percent of calories from fat and percent of calories from saturated fat. Children and adults in the PB group also had a significantly higher intake of total carbohydrates and dietary fiber than children in the AHA group. During the study, children and adults of both groups significantly reduced and increased intakes of the same nutrients, except for a trans-fat decrease only in adults on PB. Within-study energy intake and total fat intake were not significantly different between the two study groups, in children or adults.
Two goals for the children on PB were to consume no animal products and add no fat. During the study the mean (standard deviation) daily reported animal protein intake decreased from 42.32 (13.21) g to 2.24 (4.45) g (P <0.001) and the % of calories from fat and saturated fat was 18.04% (8.56%) and 3.59% (2.17%) respectively. The goals for the AHA children's group were to consume <30% of total calories from fat, <7% of calories from saturated fat, <1,500 mg sodium, and <300 mg cholesterol. The respective mean (standard deviation) reported values during the study were 25.38% (6.12%), 7.59% (2.38%), 1,699 (897.71) mg and 144 (105.57) mg. Adults in both groups reported similar changes (Table II). These results suggest good but not perfect compliance with the assigned diets.
On The Food Acceptability Questionnaire 18 using a seven-point response scale both children and parents in the PB group reported more difficulty purchasing the necessary food for their diet than the children and parents in the AHA group. Median difficulty ratings were 3 ("Slightly difficult") in the PB group and 5 ("Fairly easy") in the AHA group for both parent and child subject. Mean PB vs. AHA ratings were 3.7 vs. 5.1 for children and 3.5 vs. 5.1 for parents. There were no other statistically significant differences between the groups on how well they liked these foods, liked the taste, appearance appeal, how boring, ease of preparation, ease of maintaining diet at restaurants, effort to stay on diet, effect on cost of food purchases, satisfaction felt after meals, and overall satisfaction.
The only statistically significant differences between the PB and AHA groups after the intervention were that the children in the PB group had significantly lower week 4 BMI Zscores and hsCRP levels. Parents in the PB group had significantly lower total cholesterol, LDL, and Hgb A1C than parents in the AHA group. The only significant change favoring AHA was a 1% difference in children's waist circumference.
The primary analysis of our study was of evaluable cases.
In this study of PB in children we were most interested in determining if adhering to a PB would improve cardiovascular risk. A secondary intent-to-treat analysis was performed including the two child/parent pairs on PB who failed to complete the study. For those pairs, we assumed that there was no change from baseline in outcomes where baseline measures were available, and for measures with unavailable baseline data (insulin, IL-6, and myeloperoxidase), we assigned the median baseline value of the measure, computed separately for all parent and all child subjects, to both the baseline and end of study values. There were no differences between the evaluable-case analysis and intent-to-treat analysis in the statistical significance of within-group changes in outcome. The only differences between the evaluable-case analysis and intent-to-treat analysis in the statistical significance of the PB and AHA group comparisons were that children in the PB group no longer had significantly lower week 4 hsCRP or vitamin D intake, and parents in the PB no longer had significantly lower total protein or percent of calories from fat (data not shown).

DISCUSSION
We believe our inability to demonstrate more than a few significant differences between our intervention groups, versus many significant differences from baseline values, most likely reflects the fact our study was powered to detect changes from baseline values in the PB group with 4 weeks intervention. 14 Statistically significant differences that occurred between baseline and week 4 that might be of possible concern for increased cardiovascular risk include the elevation of Hgb A1c in both the plant-based and AHA children groups. However, the only statistically significant change in HOMA-IR in our study was in the PB children's group with a mean ratio (95% CI) of −1.25 (−2.01, −0.99), P-value 0.004, which suggests significantly decreased insulin resistance. In the PB children's group FIRI (fasting immunoreactive insulin) and in the adult PB group HgbA1c decreased significantly. All these values suggest, if anything, a decreased risk of diabetes in the PB group. Other studies have documented beneficial effects of PB on diabetes. 6,7,24 The statistically significant decreases of HDL in both the adult and children AHA group and the adult PB group are also of some concern. The decrease in serum HDL cholesterol after 4 weeks was most likely associated with early weight loss. 25 Vegan diets have been previously reported to be associated with decreased HDL, but vegan diets are also associated with a decreased risk of heart disease. 8,15,26 However, we are unaware of similar reports of lowering HDL on the AHA diet. There is heterogeneity described in the composition and function of HDL, and further characterization of HDL after exposure to these diets, might prove helpful in establishing the significance of these findings of decreased HDL. 27, 28 The statistically significant elevation of AST in the adults on the AHA diet is of unknown significance. Mild, transient increases in ALT and AST values have been observed in weight loss studies, but usually returned to below baseline levels after substantial weight loss. 29,30 Statistically significant reported differences in the PB and AHA diet in adults and children of decreased cholesterol, fat and calorie intake and increased fiber on the PB diet most likely were of benefit to the obese hypercholesterolemic children studied. 9,24 Previous reports demonstrate that dietary records, especially in individuals with higher BMI, commonly underestimate intake by nearly 50%, so the actual intake is difficult to know for certain. The reported decrease of protein intake, especially in the context of probable under-reporting of intake, is not of concern as PB diets have been shown to provide adequate protein and most other nutrients in adults and children. [31][32][33] The decreased intake of Vitamins B12 and D reported in the adult and child PB groups found in the current study have been noted previously. Vitamin B 12 definitely and Vitamin D probably warrant supplements for those on long-term PB diets. Intake of key nutrients is generally adequate in a balanced vegan diet, but it is still essential to monitor consumption of protein, n-3 fatty acids, iron, zinc, iodine, and calcium in long-term vegans. 31-33 Dietary intake of n-3 long-chain polyunsaturated fatty acids (PUFAs) eicosapentaenoic acid and docosahexaenoic acid, is low in vegans compared with omnivores. 34 Therefore, n-3 fatty acids, particularly in a no added fat vegan diet, should be especially carefully monitored and may also require supplementation with algae-derived n-3 PUFAs. The only significant described problem in our middle class study population for diet acceptance was the difficulty purchasing food. Cost may be a barrier to a PB in lower socioeconomic populations. In another study the only identified barrier to adherence was the effort required. 35 If the PB diet is to achieve ever increasing adaptation, barriers to easy, affordable access to plant-based, no added fat foods will need to be reduced.
The major limitations of our study are that this was a small study conducted for a short period of time in a select group of middle class patients with less than completely reliable measures of compliance and with no direct health outcomes measures. Also, the AHA is considered a standard of care and was used as a comparison group-there was no placebocontrolled group. There is also concern that long-term compliance with the PB could be problematic. This is especially true given the difficulties expressed by parents and children in finding food to purchase for the diet in our study, and in the effort required to maintain a PB in a previous study. 21 However, there are studies describing good acceptability and compliance with a PB. 15,16,35,36 Plant-based diets are generally recognized as safe for children and adolescents as long as the intake of key nutrients is monitored and appropriate supplements are provided. The results of our study suggest that the documented benefits of PB in adults, including, but not limited to, decreased overweight and obesity and decreased cardiovascular risk, most likely would be seen in children. These benefits, especially given the known onset of CVD in childhood, could improve the lifetime health of those populations who choose to eat a PB beginning in childhood.

CVD
Cardiovascular disease

AHA
American Heart Association Diet (also encourages fruits, vegetables, whole grains and low sodium intake but permits non-whole grains, low-fat dairy, selected plant oils, and lean meat and fish in moderation.)  Table 1 Demographics • Asian 1(7) 0(0)