Effects of Exercise in Addition to a Family-Based Lifestyle Intervention Program on Hepatic Fat in Children With Overweight

Children with overweight/obesity are at greater risk of the development of cardiovascular diseases (CVD), type 2 diabetes, and nonalcoholic fatty liver disease (NAFLD) in adulthood (1,2) and even in childhood (3). Pediatric NAFLD is affecting nearly 34% of children with obesity (4). It is estimated that by 2025, 38 million children will have hepatic steatosis, the earliest manifestation of NAFLD (5).

Pediatric hepatic steatosis is an independent risk factor for type 2 diabetes (6,7). Excessive body mass gains during childhood may eventually induce the histological features of adult NAFLD (8). The treatment of overweight/obesity before puberty, however, can reduce the risk of the development of type 2 diabetes in adulthood (9). The early treatment of childhood overweight/obesity and its comorbidities is therefore vital.

Currently, there are no pharmacological options for NAFLD in children (10). Lifestyle modification is therefore the primary option (10). A systematic review of lifestyle interventions for the treatment of overweight/obesity in children 6–11 years of age observed that multicomponent behavior-changing programs achieved small, short-term reductions in BMI (11). However, no studies have examined the effect of a nonhypocaloric diet–based lifestyle intervention program on the percentage of hepatic fat (HF) in children.

Exercise training seems to be effective in reducing HF in adolescents (12) and in reducing the risk of the development of CVD and type 2 diabetes (13) in children. In a systematic review (12), it was reported that supervised exercise, designed following the international recommendations of physical activity, reduced HF in youths. However, there has been no such study exclusively involving children under 12 years of age, and certainly none that has used imaging methods to assess HF.

Given the above, lifestyle and behavioral modifications plus exercise training might be expected to have greater effects on HF (as well as on children’s health and psychological well-being) than interventions based on lifestyle modifications alone. The aim of the present work was to determine whether a multicomponent intervention program designed according to current evidence and guidelines, and including a family-based lifestyle and psycho-education program plus a supervised exercise intervention, is more effective at reducing the percentage of HF than the lifestyle program alone in children with overweight/obesity who were 8–12 years of age. The effects of these interventions on fat mass, CVD, and type 2 diabetes risk factors, physical fitness, dietary habits, physical activity, and psychological well-being are also discussed.

The EFIGRO project is a nonrandomized two-arm, parallel-design controlled trial (clinical trial reg. no. NCT02258126, ClinicalTrials.gov) (14). The study was conducted at the University of the Basque Country in Vitoria-Gasteiz (northern Spain), from September 2014 to June 2017. The Euskadi Clinical Research Ethics Committee approved the study protocol (PI2014045), which complies with the ethical guidelines of the Declaration of Helsinki (2013 revision). All parents/legal guardians gave their informed, written consent for their children to be included in the study; the children also gave their assent before enrolment.

The EFIGRO project compared differences in changes in the percentage of HF and cardiometabolic and diabetes risk factors between two groups: one group received a family-based lifestyle and psycho-educational program, hereafter referred to as the control intervention (CInt), and another group received the same intervention plus an exercise program, hereafter referred to as the intensive intervention (II).

Children and their families were recruited at the Pediatric Endocrinology Unit of the University Hospital of Araba, and at Primary Care Clinics in the city of Vitoria-Gasteiz. The main entry criterion was having overweight/obesity, as defined by the cutoffs for sex and age (by month) established by the World Obesity Federation (15). All children had to be 8–12 years of age. Subjects whose medical condition or medication limited their physical activity, or might affect the results obtained, were excluded. None of the children had diabetes or any other endocrine disorder, and all were nonsmokers.

Children were allocated to the CInt or II groups after baseline measurements. The design of the current project was conceived as a randomized controlled trial; however, this assignment was not completely random. A number of children/families (N = 11) did not have time to attend the exercise sessions and were thus allocated to the CInt group. This decision was made since the children/families had been encouraged to participate in the study by their pediatricians, citing probable health benefits.

Neither the staff delivering the intervention nor the study subjects were blinded to the arm assignment. However, the researcher in charge of analyzing the outcomes was thus blinded.

Both the CInt and the II groups participated in the lifestyle and psycho-educational program. Parents/caregivers and children separately attended the lifestyle (45 min) and psycho-educational (45 min) programs once every 2 weeks (i.e., 11 sessions in total), as has been explained in detail previously (14). The focus of the program was 1) to increase the parents’ and children’s knowledge about healthier dietary habits, 2) to increase their physical activity level, and 3) to promote sleep hygiene. The aim of the psycho-educational program was to provide skills to the parents/caregivers in order to optimize the family environment for making positive lifestyle changes and to learn assertive communication skills. The program also provided skills to children for managing their emotions and feelings and for improving their self-esteem and psychological well-being.

Only the subjects in the II group took part in the exercise program. The full design of the program is available elsewhere (14). Children attended three sessions per week. Briefly, sessions were designed and supervised by exercise specialists and consisted of 5 min of instruction time, 10 min of warm-up, 60 min of game-based cardiovascular endurance, 10 min of muscle strength exercises, and 5 min of cooling down and stretching exercises (overall, 90 min/session). Subjects were encouraged through motivation and game strategies to spend as much time as possible during the session in vigorous, high-intensity physical activity (16). The maximum heart rate was defined as the highest value obtained during the cardiopulmonary exercise test in the laboratory or in the 20 m shuttle run test (msrt). Children wore a Polar RS300X heart rate monitor during the sessions for recording exercise intensity (16).

The attendance of both the children and their parents/caregivers at the lifestyle and psycho-educational sessions was recorded. In the exercise program, children were marked “absent” if they did not attend a session or refused to participate in the proposed games or activities. Regardless of their adherence record, all subjects were encouraged to return for postintervention data collection.

Primary and secondary outcomes were assessed at baseline and after 22 weeks of intervention by the same trained researchers.

The percentage of HF was measured by MRI using a MAGNETOM Avanto 1.5-T System (Siemens Healthcare, Erlangen, Germany) equipped with a phased-array surface coil and a spine array coil and running Siemens Medical System software v.syngo.MR B17A, following the manufacturer instructions (17). Children with ≥5.5% HF were deemed to have hepatic steatosis (18).

Body mass and height were measured in duplicate. The cut points established by the World Obesity Federation (http://www.worldobesity.org/) by sex and for each month of age were used to define overweight and obesity (15). Pubertal stage was recorded by a pediatrician (19). Systolic and diastolic blood pressure were measured following recommendations for children (20). Participants were deemed hypertensive when their systolic or diastolic blood pressure was >90th age-, sex-, and height-specific percentile (20).

Total and abdominal fat, as well as lean mass, were measured by DEXA using a Hologic QDR 4500W device. The fat mass index (FMI) was then determined as fat mass [kg]/stature² [m²] and lean mass index (LMI) as lean mass [kg]/stature² [m²].

Cardiorespiratory fitness (CRF) and muscle strength were determined following validated protocols for children (21). Briefly, CRF was assessed by two tests: 1) an incremental CRF treadmill protocol with respiratory gas analysis until exhaustion, and 2) the 20msrt. Upper and lower body muscular strength were assessed by the handgrip and the standing broad jump tests, respectively.

Plasma triglycerides (TGs), HDL cholesterol (HDLc), LDL cholesterol (LDLc), insulin, glucose, and γ-glutamyl transferase (GGT) concentrations were determined as reported previously (14). HOMA of insulin resistance and the TG/HDLc ratio were then calculated. Insulin resistance was defined according to the age- and sex-specific cut points for HOMA values (22). Subjects with TG/HDLc ratio values of ≥2.0 were deemed to be at increased cardiometabolic risk (23).

Physical activity was determined by accelerometry (wGT3X-BT; Actigraph, Pensacola, FL), and dietary habits were assessed using two nonconsecutive 24-h recall records and food frequency questionnaires, as detailed previously (14).

Anxiety was assessed using the State-Trait Anxiety Inventory for Children; stress was determined using the Children’s Daily Stressors Inventory; self-concept was assessed using the Self-Concept Form-5 Questionnaire; and depression was examined using the translated version of the Children’s Depression Inventory (14).

Since no information on the effect of exercise on the percentage of HF in children was available, the required sample size was calculated using information available for closely associated secondary outcome variables. The sample size actually obtained was N = 116. This figure was expected to provide at least 80% power for detecting an effect size (Cohen d) between the two experimental groups of 0.7 for insulin resistance and total body fat (minimum number required in each group = 34 for 80% power and α = 0.05).

The baseline characteristics of the subjects’ in the two intervention groups were compared using either the Student t test (for continuous variables) or the χ² test (for categorical variables). Differences between the intervention groups in terms of postintervention values for the primary and secondary outcomes were also examined using the Student t test. Within-group differences (pre- vs. postintervention values) in primary and secondary outcomes were examined using the Student paired t test following per protocol and intention-to-treat principles. For the intention-to-treat analyses, missing values at follow-up were obtained by multiple imputation using the following predictor variables: preintervention values, postintervention values, age, sex, and intervention group. Imputations were performed as four blocks: body composition variables (HF, body mass, BMI, FMI, LMI, abdominal fat), blood pressure variables (systolic and diastolic blood pressure), cardiometabolic risk variables (TG, HDLc, LDLc, TG/HDLc ratio, insulin, glucose, HOMA, GGT), and fitness variables (VO_2peak, end-time treadmill test, 20msrt result, handgrip strength, and standing broad jump).

In both the per protocol and the intention-to-treat analyses, differences between the CInt and II groups in terms of changes in primary and secondary outcomes were examined by ANCOVA adjusting for baseline values. Cohen d was used to estimate the effect size and 95% CI. Differences between the CInt and II groups in terms of changes in the percentage of HF were examined in extended models: model 1, unadjusted; model 2, adjusted by the corresponding baseline value and changes in height; model 3, adjusted by baseline values and changes in the FMI; and model 4, adjusted by baseline values and changes in abdominal fat.

Finally, the relationship between the changes (preintervention to postintervention values) in physical fitness variables and in the percentage of HF in each intervention group (CInt and II) were examined by regression analysis 1) using an unadjusted model, and 2) adjusting for baseline physical fitness and percentage of HF.

Significance was set at P < 0.05. All calculations were made using the Statistical Package for Social Sciences version 24.0 for Windows (SPSS Inc., Chicago, IL).

Figure 1 shows the flowchart for the trial, plus subject inclusions and exclusions. Table 1 shows the participants’ baseline characteristics. Overall, there were no significant differences in the socioeconomic, sociodemographic, anthropometric, or clinical characteristics between the CInt and II groups.

A total of 102 of the original 116 subjects (87.9%; N = 53 in the CInt group; N = 49 in the II group) successfully completed the trial, attending at least 50% of the sessions. Their data were included in the per protocol analyses. The data for those children that discontinued the intervention (N = 4 in the CInt group; N = 9 in the II group), or who did not attend at least 50% of the educational program sessions (N = 1 in the II group), were also included in the intention-to-treat analyses.

No significant difference was seen between the CInt and II groups in terms of attendance at the lifestyle and psycho-education program sessions, either for the parents/caregivers (86.4 ± 12.9% vs. 80.6 ± 15.3%; P = 0.334) or the children (87.2 ± 12.0% vs. 82.5 ± 14.6%; P = 0.496). The mean attendance rate for the II subjects with respect to the exercise program was 72.0 ± 16.1% of sessions. Exercise-related adverse events included knee and ankle pain (N = 2); no adverse events were recorded for the lifestyle and psycho-education programs.

In the exercise program, the mean heart rate per session was 146 ± 16 bpm. High-intensity exercise was maintained for 49 ± 23% of the time, and moderate-intensity exercise for 32 ± 15% of the time.

No significant difference in percentage of HF was seen between the CInt and II groups at baseline (Table 2 and Supplementary Table 1, for the per protocol and intention-to-treat analyses, respectively). Although there were no significant differences in postintervention values of the percentage of HF between the two groups (Table 2 and Supplementary Table 1), only children in the intensive group reduced the percentage of HF after the intervention (Table 2 and Supplementary Table 1, for the per protocol and intention-to-treat analyses, respectively). The difference in the change in percentage of HF between the two groups (per protocol analysis) was significant (P < 0.02) (Fig. 2 and Supplementary Fig. 1); these results persisted after adjustment for baseline values for percentage of HF and changes in height (−1.20 ± 0.31% vs. 0.04 ± 0.30% for the II and CInt groups, respectively; P = 0.006), and when changes in FMI (−1.13 ± 0.27% vs. −0.02 ± 0.27% for the II and CInt groups, respectively; P = 0.004) and abdominal fat (−1.19 ± 0.26% vs. −0.03 ± 0.28% for the II and CInt groups, respectively; P = 0.004) were adjusted for instead of changes in height (Supplementary Fig. 1). Further adjustment for baseline VO_2peak (measured via the treadmill test) did not substantially alter the results (P < 0.05). The intention-to-treat analyses for differences between groups in terms of the change in percentage of HF returned similar results (Supplementary Fig. 2).

No significant differences were seen between the CInt and II groups in terms of any of the studied cardiometabolic or type 2 diabetes risk factors at baseline (Table 2 and Supplementary Table 1). Both groups returned significantly reduced BMI, FMI, and abdominal fat values after the intervention (P < 0.01), while no significant changes were seen in LMI (Table 2 and Supplementary Table 1 for the per protocol and intention-to-treat analyses, respectively). No significant differences were seen between the groups postintervention in any adiposity estimate (Table 2 and Supplementary Table 1). The reduction in BMI was significantly greater in the II group (P < 0.01), but no significant differences were seen between the groups in terms of the change in FMI or abdominal fat (Fig. 2). These results did not substantially differ after adjusting for baseline values (Supplementary Fig. 2). The intention-to-treat analyses returned similar results (Supplementary Fig. 3).

Per protocol analysis showed the insulin concentration and HOMA value to be significantly smaller (P < 0.01 and P < 0.05, respectively) in the CInt group after the intervention (Table 2). These reductions were not significant, however, in the intention-to-treat analysis (P < 0.09) (Supplementary Table 1). In the II group, the HDLc (P < 0.01), LDLc (P < 0.001), insulin (P < 0.05), and GGT (P < 0.05) concentrations all decreased significantly by the end of the intervention program (per protocol analysis). Diastolic blood pressure increased significantly in the II group, whereas no significant change was observed in the control group (Table 2). The intention-to-treat analysis returned similar results, but only the changes in TG (P < 0.05), insulin (P < 0.001), and GGT (P < 0.01) remained significant (Supplementary Table 1).

No significant differences were seen in terms of postintervention cardiometabolic and diabetes risk factors between the two groups (Table 2 and Supplementary Table 1), except with respect to the LDLc level (P < 0.05; both per protocol and intention to treat). However, greater reductions in LDLc level were seen in the II group than in the CInt group after the intervention; the per protocol and the intention-to-treat analyses returned similar results (Fig. 2 and Supplementary Figs. 2 and 3).

In the per protocol analysis, significant difference was seen between the groups in terms of the effect of the interventions on TG (P < 0.05) (Table 2), but this was attenuated in the intention-to-treat analyses (P < 0.07) (Supplementary Table 1). However, the differences in the change in TG levels were not consistent between the per protocol (Fig. 2 and Supplementary Fig. 2) and intention-to-treat analyses (Supplementary Fig. 3). Similarly, analysis-inconsistent differences were seen between groups in terms of the effect of the interventions on GGT levels (Fig. 2 and Supplementary Figs. 2 and 3).

At baseline, VO_2peak as measured by the treadmill test was lower in the II group than in the CInt group (P < 0.05), but no significant differences were seen in the rest of the fitness variables (Table 2 and Supplementary Table 1). Significant increases were seen in both groups for all physical fitness variables after the intervention (P ≤ 0.001) (Table 2 and Supplementary Table 1), but no significant differences were seen between them in terms of the changes in these values (Fig. 2), even after adjustment for baseline values (Supplementary Fig. 2). The intention-to-treat analysis returned similar results (Supplementary Fig. 3).

The increase in CRF measured in the 20msrt was significantly associated with the reduction in percentage of HF content in the II group subjects—at least in the per protocol analysis (Supplementary Table 2). This relationship was attenuated and became nonsignificant (P < 0.06) in the intention-to-treat analysis (Supplementary Table 3).

All comparisons were per protocol (Supplementary Table 4). No significant differences in dietary and physical activity variables were seen between the two groups at baseline. By the end of the intervention, both groups had significantly reduced their energy and fat intake and increased their intake of fruits and vegetables. No significant changes were seen, however, in the consumption of sugar and sugar-sweetened beverages. Neither moderate-to-vigorous physical activity nor sleep time changed after the intervention in either group, although sedentary time was significantly reduced in the II group (P < 0.05). No significant differences in dietary or physical activity variables were seen between the two groups at the end of the intervention.

No significant differences were seen in psychological variables between the two groups at baseline (Supplementary Table 4). By the end of the intervention, the subjects of both groups had experienced a significant increase in terms of emotional self-concept (P < 0.05). Improvements were also seen in physical self-concept, total depression, dysphoria, and anxiety, although they were only significant in the CInt group (P < 0.05 and P < 0.09, for the CInt and II intervention groups, respectively). No significant differences in the improvement in psychological well-being were seen between the two groups at the end of the intervention.

The present results reveal the II subjects to have experienced a clinically important reduction (nearly 20%) in percentage of HF; no such reduction was seen in the CInt group. Both the CInt and II subjects experienced a reduction in total and abdominal fat and insulin resistance. The II subjects also enjoyed a significant reduction in LDLc. Given the importance of treating pediatric hepatic steatosis early, not only to prevent future liver damage, but also to prevent type 2 diabetes, these findings should be taken into account in pediatric obesity management programs.

Family-based lifestyle education programs accompanied with psychological support are recommended for the treatment and prevention of pediatric obesity and related comorbidities (24). However, the current study shows that while the educational program followed was able to effectively reduce adiposity and insulin resistance, it was unable to reduce the percentage of HF in the children. HF was only reduced in those who participated in the supervised exercise program in addition to the lifestyle education program. These results agree with those of a study involving obese adolescents (25), in which a 4-month intervention focused on changing the quality of carbohydrate intake resulted in significant improvements in insulin sensitivity, but only a combined lifestyle, psycho-education, and resistance exercise training program resulted in a reduction in the percentage of HF.

As far as we are aware, the present work is the first to report the additional effect of supervised exercise on HF as measured by MRI in children. Other studies have reported significant reductions in HF after aerobic or resistance exercise training interventions, but in adolescents (13,26–28). In a recent systematic review and meta-analysis, it was reported that exercise training at moderate-to-vigorous or vigorous intensity, in sessions of at least 60 min and a frequency ≥3 sessions per week, are effective in reducing HF content in youth (12). The present findings are in agreement with these recommendations.

In adults and adolescents with overweight or obesity, lifestyle interventions including exercise, or not, have been reported to significantly reduce HF (29,30). However, these programs were based on hypocaloric diets that caused large body mass losses (30,31). In the current study, which involved preadolescent children, the lifestyle education program was not designed to achieve body weight loss in the short term, and certainly the CInt subjects lost no body mass or lean mass.

A systematic review and meta-analysis comparing the effect of exercise training and a hypocaloric diet on visceral adiposity (32) reported the latter to result in a larger body mass loss, whereas exercise training tended to induce larger reductions in visceral adiposity. Interestingly, in the present work, the effects of exercise on the percentage of HF were independent of the change in total or abdominal fat, suggesting a direct metabolic effect of exercise training on HF metabolism.

Family-based, structured lifestyle modification programs combined with behavioral strategies for treating obesity have been associated with adiposity reductions in children (33). In the current study, the children’s dietary habits were improved with respect to the development of obesity after the 22-week intervention. Improvements were also seen in several components of self-concept, depression, stress, and anxiety in children. These improvements in both dietary habits and psychological health are quite notable.

Physical activity was not increased at the end of the intervention in either the CInt or II group; indeed, in the CInt group there was even a trend toward a reduction in total physical activity. It should also be noted that several families declined to participate in the II arm given the relatively large number of sessions involved. A possible solution might be to reduce the number of sessions per week or the duration of each session. Future studies examining the effectiveness of less intensive exercise programs on the percentage of HF and cardiometabolic risk in overweight/obese children are warranted.

One of the strengths of this study is the use of MRI for measuring percentage of HF. Moreover, the sample size was also larger, and the duration of the intervention program was longer than in previous studies (13,26,27). Of note, however, is that several families did not agree to participate in the II due to the relatively elevated number of sessions, which may compromise the feasibility of the intensive program. There are, however, some limitations that should be mentioned. The most important limitation of the study is its not entirely strict randomization (14). This may limit the validity of the results. However, the participants in both groups were comparable at baseline, and adjustments for potential baseline differences between groups were made in analyses. Finally, the results were not adjusted for multiple comparisons.

A family-based, multicomponent intervention program including supervised exercise training and lifestyle and psycho-education, designed following international recommendations for obesity prevention and health promotion in children, was shown to have the potential to reduce HF, total and abdominal fat, and insulin resistance, and to improve dietary habits and psychological well-being in preadolescent children with overweight/obesity. These findings highlight the importance of promoting such programs as part of pediatric obesity treatment: improvements may be achieved not only in total and abdominal fat and insulin resistance but also in hepatic steatosis, a metabolic abnormality that increases the risk for type 2 diabetes and CVD.

Acknowledgments. The authors thank the pediatricians who took part in the recruitment of subjects and all those research assistants involved in assessment duties. The authors also thank Siemens Medical Systems for supplying the software for quantifying the percentage of HF and the assistants who helped run the exercise training sessions and educational program.

Funding. This project was funded by the Spanish Ministry of Health “Fondos de Investigación Sanitaria del Instituto de Salud Carlos III” (PI13/01335), the Spanish Ministry of Industry and Competitiveness (DEP2016-78377-R), and by EU Fondos Estructurales de la Unión Europea (FEDER) funds (“Una manera de hacer Europa”). Support was also provided by grants from the Spanish Ministry of Education, Culture and Sports (FPU14/03329), the Education Department of the Government of the Basque Country (PRE_2016_1_0057, PRE_2017_2_0224, PRE_2018_2_0057), the University of Granada Plan Propio de Investigación 2016-Excellence actions: Unit of Excellence on Exercise and Health (UCEES), and Junta de Andalucía, Consejería de Conocimiento, Investigación y Universidades (FEDER, ref. SOMM17/6107/UGR).

Funding sources were not involved in the collection, analysis, or interpretation of any data; in the writing of the report; or in the decision to submit the article for publication.

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. I.L. designed the study, analyzed the data, and drafted the manuscript. I.L., V.M.-V., J.R.R., and F.B.O. participated in the interpretation of the results. M.M., L.A., E.M., and M.O. collected the data and critically revised the manuscript. I.L., M.M., L.A., E.M., M.O., V.M.-V., J.R.R., and F.B.O. critically revised the manuscript for its intellectual content and approved the final version. I.L. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.