Metabolic effects of a 13-weeks lifestyle intervention in older adults: The Growing Old Together Study

Longevity family members and controls are largely similar on baseline

Of the 164 individuals who started the intervention study, one dropped out prior to completion of the study (Fig. 1). A selection of the clinical baseline characteristics of the participants according to familial background, i.e. longevity family member or control, is depicted in Table 1 (complete clinical baseline characteristics are provided in Supplementary Table 1A). Baseline characteristics of ¹H-NMR metabolites are shown in Supplementary Table 1B. Gender differences were observed for many parameters. In contrast to the Leiden Longevity Study as a whole, in which many metabolic parameters differ significantly between longevity family members and controls [16,18,19], we observed only few significant differences in the parameters at baseline between the small groups included in the GOTO study. Therefore, we studied the effects of the intervention in both groups combined.

Figure 1. Flow chart of participants in the trial

Intervention improves body composition and metabolic health

The effect of the 13-weeks lifestyle change on clinical parameters is depicted in Table 2 and Supplementary Table 2A. For the primary outcome, i.e. fasting insulin, a minor mean (SE) decrease of 0.30 mU/L (3.21 mU/L) and a considerable heterogeneity (range −11.5-10.5 mU/L) was observed (Supplementary Fig. 1). Measures of body composition generally improved, as shown by a mean weight loss of 3.3 kg (0.18 kg), i.e. 4.2% (2.8%) of baseline weight (Fig. 2), a body fat mass decrease of 11.7% (8.9%), and a fat free mass decrease of 0.7 kg (0.1 kg). Measures of health and functioning showed a significant decrease of 4.3 mmHg (1.0 mmHg) in systolic and 1.7 mmHg (0.6 mmHg) in diastolic blood pressure. We noted that the changes in weight and systolic blood pressure (Fig. 3), but not in insulin levels, were dependent on baseline levels.

Figure 2. Effect of the intervention on weight by gender

Figure 3. Effect of the intervention on systolic blood pressure by baseline systolic blood pressure

Resting energy expenditure (REE) significantly decreased with 49.2 kcal/day (8.0 kcal/day). Hand grip strength was not changed by the lifestyle change, but physical functioning, mental QoL in women, and the Framingham Risk Score improved. The diagnostic measures showed a significant decrease in total and LDL cholesterol, HDL cholesterol in women, fT3, and leptin levels and an increase in adiponectin levels in men.

Plasma metabolite profile changes due to 3 months lifestyle intervention

To determine the overall effect of the intervention on the ¹H-NMR metabolites, we performed Principle Component Analysis. This analysis indicated that a major part of the variation in the metabolites (PC1, explaining 32.4% of the total variance) could be attributed to the effects of the lifestyle intervention (Fig. 4). As shown in Supplementary Fig. 2, the intervention effect, as represented by PC1, coincides with many of the measured ¹H-NMR metabolites. The effects of the intervention on the single ¹H-NMR metabolites in fasting blood are depicted in Fig. 5 and Supplementary Table 2B. Multiple amino acids levels changed significantly and include a decrease of the branched-chain amino acid leucine and the aromatic amino acid tyrosine. In addition, the levels of multiple glycolysis-related metabolites, ketone bodies, fatty acids, metabolites involved in fluid balance and inflammation, apolipoproteins, lipid concentrations, and lipoprotein particle sizes showed a significant decrease. Citrate levels, large HDL cholesterol concentrations, and HDL particle size, on the other hand, increased after the lifestyle change. For several HDL-related metabolites we observed an opposite effect of the intervention in men and women.

Figure 4. Effect of age, gender, and intervention on ¹H-NMR metabolite-based PC's. The colour of the blocks represents the magnitude of the effect, while the P-value is mentioned between brackets.

Figure 5. Effects of the intervention on ¹H-NMR metabolites. Effect sizes are per 1-SD log-transformed metabolite concentration and adjusted for age and gender. Squares indicate mean and error bars denote 95% confidence intervals. Blue squares indicate males, red squares indicate females. Individuals using lipid-lowering agents were removed before analysing fatty acids, fatty acids ratios, apolipoproteins, lipids, total lipid concentrations in lipoprotein subclasses, lipoprotein particle size and cholesterol. VLDL, very low density lipoprotein; IDL, intermediate density lipoprotein; LDL, low density lipoprotein; HDL, high density lipoprotein.

Effects of lifestyle intervention at old age (partly) independent of weight loss

To investigate whether the observed response mainly coincides with the change in weight, we adjusted for weight loss. For most of the parameters of health and functioning, diagnostic measurements, and ¹H-NMR metabolites, adjustment for weight loss reduced the effects of the intervention. However, the changes in fT3, total, VLDL, LDL, and IDL cholesterol, as well as those in phosphoglycerides, cholines, sphingomyelines, and some glycolysis intermediates, remained largely unchanged after this adjustment (Table 2, Supplementary Table 2A and Supplementary Table 2B).

Finally, the beneficial effects on physical functioning and mental QoL in women were also largely independent of weight loss (Table 2 and Supplementary Table 2A). Hence, the effect of the lifestyle change on many of the metabolic parameters and well-being in this study occurs partly or fully independent of the observed loss in weight.

Study population

Participants for the GOTO study were recruited from the Leiden Longevity Study (LLS), a longitudinal cohort consisting of 421 families of long-lived Caucasian siblings, together with their offspring and the partners thereof [31]. For the current intervention couples consisting of offspring from the long-lived siblings and their current partners were included. In case one of the two was not eligible to participate, single offspring or controls were included to obtain the required sample size. Individuals of ages between 46 and 75 years and having a BMI ≥23 ≤35 kg/m² were recruited between February and October 2012. Potential participants underwent a telephonic screening and a screenings home visit. Exclusion criteria were: type I or type II diabetes (on diabetic medication); fasting blood glucose level ≥7.0 mmol/L; weight change ≥3 kg over the past 6 months; engagement in heavy/intensive physical activity (top sport or physically heavy work); any disease or condition that seriously affects body weight and/or body composition including active types of cancer; heart failure (NYHA III/VI), COPD (GOLD III/VI); recent (<3 months prior to intervention) immobilisation for >1 week; psychiatric or behavioural problems; use of thyroid medication, immunosuppressive drugs (e.g. prednisone, methotrexate, biologicals (TNF-alpha antagonists); concurrent participation in any other intervention study or weight management program, or not having a general practitioner. The Medical Ethical Committee of the Leiden University Medical Center approved the study and all participants signed written informed consent. All experiments were performed in accordance with relevant and approved guidelines and regulations. This trial was registered at the Dutch Trial Register (http://www.trialregister.nl) as NTR3499.

Intervention

The intervention comprised 13 weeks of 25% lowered energy balance by 12.5% reduction in energy intake and 12.5% increase in physical activity. Baseline energy intake and expenditure were assessed by an online version of a 150-item food frequency questionnaire (FFQ) [32] and by the International Physical Activity Questionnaire- Short Form (IPAQ-SF) [33]. The IPAQ-SF collects information on time spent walking, in moderate and vigorous physical activity, and sitting over the last seven days to estimate the total metabolic equivalent (MET) in minutes per week.

Individual guidelines were prescribed by respectively a dietician and physiotherapist in consultation with the participant to match the subjects' preferences and physical capabilities. The dietary guidelines were as much as possible according to the ‘Dutch Guidelines for a healthy diet’ [34]. Participants were advised to increase the amount of physical activity in such a way it fitted in their daily life pattern, as couple or alone, by walking, cycling, adjusted activities in and around the house and participation in local sport activities and facilities.

During the intervention participants had weekly contact with the dietician and physiotherapist by phone, email or at the participants home (alternating schedule) to check and stimulate adherence to the intervention and to discuss practical problems and solutions. To optimally guide the participants, both dietician and physiotherapist combined elements from the Attitude, Social influence and self-Efficacy model (ASE) [35], the Stages of Change Model [36] and Motivational Interviewing [37].

Participants daily recorded their eating behaviour and physical activity in a diary. To quantitatively assess dietary intake two telephonic 24-h recalls were performed during the first month and two during the last month of the intervention. Days of the recall were unannounced to the participant and randomized to obtain a good distribution of the different days of the week, including weekend days. During the monthly home visits body weight and body composition were measured. To quantitatively determine physical activity prior to the intervention and at the end of the intervention accelerometers worn at home during seven days on wrist and ankle (GENEActiv, Activinsights, Kimbolton, UK) were used.

Anthropometrics

Height and weight were measured to the nearest 0.1 cm and 0.1 kg, respectively (Seca Clara 803, Seca Deutschland, Hamburg, Germany), with the person dressed in light clothing and without shoes. Waist circumference was measured to the nearest cm at the midpoint between the lowest rib and the top of the iliac crest with a non-elastic tape in standing position without shoes. Fat free mass and fat mass were measured using the In-Body 720 body composition analyser (Biospace, Cerritos, CA, USA).

Blood pressure

Trained staff members measured blood pressure in sitting position after a 10-minute rest on the dominant arm using a validated blood pressure device (Maxi-Stabil 3, Welch Allyn, Leiden, the Netherlands). Blood pressure was measured 4 times, twice in the first part of the afternoon and twice at the end of the afternoon. Systolic and diastolic blood pressures were calculated as the average of the four measurements.

Energy metabolism

Resting metabolic rate was measured 65 minutes after a standardized meal by indirect calorimetry, using a ventilated hood system (Care Fusion Canopy Jaeger Oxycon Pro, CareFusion Germany, Hoechberg, Germany). The standardized meal was 500 kcal and consisted of one raisin bun, one whole-wheat bun with margarine and 20 grams of Gouda cheese and a cup of tea without milk or sugar. Participants were lying on a bed under the ventilated hood in a quiet, temperature controlled room for 30 minutes. The initial 5 minutes of the measurement were not used for the analysis. VO2 and VCO2 were measured every minute. Resting energy expenditure (REE) and respiratory quotient (RQ) were calculated using the formulas: $\text{REE}=3.91\text{ VO}2+1.10\text{ VCO}2-1.93\text{N}.$$\text{RQ}=\text{VCO}2/\text{VO}2$

To exclude outliers in energy expenditure, the degree of variation based on the coefficient of variation (CV) was examined based on the mean and the SD of five data points that were used. Data per 5 consecutive minutes was included if the RQ data over these 5 minutes had a CV <5%.

Physical performance

Physical performance was assessed by the short physical performance battery (SPPB), which consisted of three components: balance, gait speed, and chair rise ability [38]. Handgrip strength was determined by three consecutive measures using a hand dynamometer (Jamar, Lafayette Instrument, Lafayette, IN, USA) at both hands.

Quality of life

Quality of life was assessed using the Short Form Health Survey-12 (SF-12). This questionnaire [39] distinguishes physical and mental health, each assessed by six items.

Framingham risk score

The Framingham risk score (FRS), which estimates the 10-year risk for developing coronary heart disease, was calculated using the criteria proposed by the Expert Panel on Detection, Evaluation, and Treatment of High Blood Cholesterol in Adults [40]. The score is based on the age, gender, total and HDL cholesterol serum level, smoking status, and systolic blood pressure of an individual.

Diagnostic measurements

All measurements were performed in fasted serum collected by venipuncture. Cholesterol, free thyroxine (fT4), glucose, high-density lipoprotein (HDL) cholesterol, triglycerides, high-sensitivity C-reactive protein (hsCRP), and thyroid stimulating hormone (TSH) were measured on the Roche/Hitachi Modular P800 analyzer (Roche Diagnostics, Almere, The Netherlands). Dehydro-epiandrosterone sulfate (DHEAS), insulin, insulin-like growth factor 1 (IGF-1), and insulin-like growth factor 1 binding protein 3 (IGF-BP3) were assessed on the Immulite 2000 XPi (Siemens, Eschborn, Germany). Adiponectin and leptin were determined using Human Adiponectin and Leptin RIA kits (EMD Millipore Corporation, Billerica, MA, USA). Free triiodothyronine (fT3) was determined using the ARCHITECT Free T3 assays (Abbott Laboratories, Abbott Park, IL, USA) on the Hitachi Modular E170 analyzer (Roche Diagnostics). Coefficients of variation for all laboratory analyses were <8%. Low-density lipoprotein (LDL) cholesterol was calculated using the Friedewald formula [41], while the homeostasis model assessment-estimated insulin resistance (HOMA-IR) was calculated using the publically available HOMA calculator (https://www.dtu.ox.ac.uk/homacalculator) [42].

Hydrogen-1 Nuclear Magnetic Resonance metabolites

¹H-NMR metabolites were measured using a previously described platform [43]. For our analysis we used the total lipid concentrations, fatty acid composition, and low-molecular-weight metabolites, including amino acids, glycolysis-related metabolites, ketone bodies and metabolites involved in fluid balance and immunity. All metabolite concentrations were natural log-transformed and scaled to standard deviation units before analysis.

Statistical analysis

Baseline differences between longevity family members and controls were calculated using a linear mixed model adjusted for age, gender (fixed effects), and household (random effects). The effects of the intervention were determined using a linear mixed model adjusted for age, gender, status (longevity family member or control) (fixed effects), household, and individual (random effects). Parameters were analyzed separately in men and women if there was a significant gender-difference at baseline. For additional analyses, weight was added to the model to determine weight loss-independent effects.

Principle Component Analyses and the following association analyses with age, gender and intervention were performed in R [44]. Principle Components (PCs) were computed using the function prcomp of the stats package [44]. Association analyses with the obtained PCs were performed using mixed linear models, function lmer, of package lmerTest [45]. Heatmaps were drawn to visualize the magnitude of the statistics from the association analyses using labeledHeatmap of the WGCNA package [46].

Sample size calculation was based on fasting insulin as primary end point, whereby we assumed a decrease of 21% (9.75 to 7.69 μIU/mL), since this was observed after 3 months in the calorie restriction with exercise group of the CALERIE study [6], which is comparable to our study. As the mean fasting insulin levels (μIU/mL) in longevity family members and controls in the complete LLS cohort were 6.93 (SD=4.3) and 8.70 (SD=6.6), respectively, and a correlation of 0.6 between repeated insulin measurements was assumed [47], we based our power calculation on an expected mean decrease of 1.65 μIU/mL, SD 4.9. With a power of 80% and an α of 5% this translates in a required sample size of 72 individuals. In a recent large meta-analysis, genetic background did not influence the association of healthy diet with fasting glucose or insulin [48]. Nevertheless, we doubled the sample size of our study to account for potential differences between longevity family members and controls in response to the intervention. Taking into account a dropout rate of 10%, we aimed to include 80 couples in the intervention.

All statistical analyses were performed with STATA/SE 11.2 (StataCorp LP, College Station, TX, USA) and SPSS Statistics v20 (IBM Corp, Armonk, NY, USA) and a P≤0.05 was considered significant.