Effects of elastic band resistance training and nutritional supplementation on muscle quality and circulating muscle growth and degradation factors of institutionalized elderly women: the Vienna Active Ageing Study (VAAS)

VAAS was a randomized, controlled intervention study which was designed to test whether 6 months of a supervised, progressive resistance exercise training using elastic bands with and without nutrient supplementation were able to influence physical fitness of institutionalized elderly men and women (Oesen et al. 2015). Eligible participants were randomly, but stratified by gender, assigned to one of the three intervention groups: resistance exercise training (RT), RT plus nutritional supplementation (RTS), or cognitive training (CT), the latter serving as control group. The interventions started immediately following the baseline tests (T1), which were repeated after 3 (T2) and 6 months (T3).

Inclusion and exclusion criteria have been described previously (Oesen et al. 2015). Briefly, the participants were untrained, over 65 years old and free of any medical condition which would impair their participation in a resistance training class. Mini-mental state scores were ≥22 (Folstein et al. 1975). Written informed consent was obtained from all participants. The present study was conducted in accordance to the Austrian laws (including Doctors Act, CISA, Data Protection Act), the Declaration of Helsinki (as revised in Edinburgh 2000), and in analogous accordance with ICH-GCP guidelines. The study has been approved by the ethics committee of the City of Vienna (EK-11-151-0811) and registered at ClinicalTrials.gov (NCT01775111).

From a total of 117 participants of the VAAS, only those were initially included in the current study where data on MQ and serum markers were available (n = 104). Due to the low sample size of men (n = 13) these were additionally excluded from the presented analyses. Blood samples were available from all women (n = 91) at baseline (T1), from 76 women after 3 months (T2), and from 70 women after 6 months (T3). Additionally, MQ of both upper and lower extremities could be assessed from 77 women at T1, from 68 women at T2, and from 59 women at T3 (Fig. 1).

All intervention groups met twice a week for about 60 min in small groups of not more than 10 people. The groups were supervised by experienced sports scientists. The resistance exercise programme was designed to train all major muscle groups based on guidelines provided by the American College of Sports Medicine (ACSM) for older adults (Nelson et al. 2007). Following an adaptation phase of 4 weeks using low external resistance (1 set of 15 repetitions per exercise, yellow Thera-Band^® (The Hygenic Corporation, Akron, OH, USA), exercise intensity was progressively increased by adapting the resistance of the elastic band (based on the Thera-Band^® force–elongation table) (Page and Ellenbecker 2011) from yellow to red and further to black. The exercise volume was further enhanced by increasing the number of sets from one to two. The RTS group followed the same training protocol as the RT group. Additionally, a nutrient supplement drink (FortiFit, NUTRICIA GmbH, Vienna, Austria) was provided every morning after breakfast, as well as immediately after each training session. Each drink supplied a total energy of 150 kcal and contained 20.7 g protein (3 g leucine, >10 g essential amino acids), 9.3 g carbohydrates, 3 g fat, vitamins (800 IU vitamin D, 2.9 mg vitamin B6, 3 μg vitamin B12) and minerals. A research dietician distributed the supplements and monitored adherence. Participants were instructed to maintain their regular food intake. The CT group served as control participating in activities including cognitive tasks (memory training) and coordinative tasks (such as manual dexterity) twice weekly to provide a timely effort which was equal to those of the RT and RTS group (Oesen et al. 2015).

Standing height was measured without shoes to the nearest 0.5 cm using a commercial stadiometer (Seca, Hamburg, Germany) with the shoulders kept in a relaxed position and arms allowed to hang freely. Body mass was evaluated with a digital scale (BWB 700, Tanita, Amsterdam, Netherlands) to the nearest 0.1 kg with subjects lightly dressed and barefoot. Body mass index (BMI) was calculated as the ratio between the weight (kg) and height squared (m). To determine body composition bioelectric impedance analysis (BIA) was used, which has been shown to provide reliable data of body composition in comparison to dual-energy X-ray absorptiometry (DXA) (Roubenoff et al. 1997). BIA was performed in the morning after an overnight fast using a BIA Analyzer 2000-S (Data-Input GmbH, Darmstadt, Germany). Skeletal muscle mass was determined using the equation of Janssen et al. (2000): Skeletal muscle mass (kg) = [height²/R × 0.401) + (gender × 3.825)–(age × 0.071)] + 5.102, where height is measured in centimeters; R is BIA resistance in ohms; for gender, men = 1 and women = 0; and age is in years.

MQ was determined as the ratio of muscle strength or power and muscle mass as suggested by Barbat-Artigas et al. (2012b). The MQ score of upper extremities was calculated based on handgrip strength measured by hand dynamometer (kg) divided by muscle mass calculated using the BIA equation of Janssen et al. (2000). To calculate power for MQ of lower extremities the equation for peak power (W) [−715.218 + 13.915 × body weight (kg) + 33.425 × stand in 20 s] of Smith et al. (2010) predicting lower-body muscle power in older adults using 30-s chair stand test was used. The calculated power was then divided by muscle mass.

Statistical analyses were performed with IBM SPSS Statistics 22 (IBM Corporation, New York, USA). The number of participants chosen for the VAAS was based on power estimation (G*Power 3.1.0), which estimated the sample size to be 86 using isokinetic peak torque as primary endpoint (Faul et al. 2007; Oesen et al. 2015). Although men were excluded for this secondary analysis, a total of 91 women participated in the current study still representing a higher sample size as anticipated.

For all secondary endpoints as included into the current analyses Shapiro–Wilk test was used to check for normal distribution which was violated for most of the blood-based parameters. Therefore, differences between groups at baseline were determined by Kruskal–Wallis test followed by Bonferroni-corrected Mann–Whitney tests for post hoc analyses. To analyse for time effects, Friedman tests were applied and if significant followed by Wilcoxon tests with Bonferroni corrections. Effect sizes (r) were calculated by dividing the standardized test statistics (z score) by the square root of the total observations. Potential correlations between functional parameters and blood-based biomarkers were assessed using Spearman’s rank correlation coefficient. Data are shown as median (minimum–maximum) and a p value of less than 0.05 was considered significant.

The women (n = 91) which were included in the current analysis were distributed nearly equally between intervention groups (CT: 33 %, RT: 36 %, RTS: 31 %). At baseline the intervention groups did not differ in age, weight, BMI, muscle mass, fat mass and MMST (p > 0.05). Comparisons between groups revealed differences for MQ of upper extremities [H(2) = 6.81, p = 0.033] and IGF-1 [H(2) = 6.71, p = 0.035] at baseline. However, pairwise comparisons with adjusted p values have confirmed a difference just for MQ of upper extremities which was slightly higher in the RT group as compared to the CT group (+24 %, p = 0.029, r = 0.326) (Table 1).

As partially shown previously (Hofmann et al. 2015), baseline levels of GDF-15 correlated negatively with fat mass (ρ = −0.258; p = 0.016), skeletal muscle mass (ρ = −0.234; p = 0.032), MQ of upper extremities (ρ = −0.259; p = 0.023), and handgrip strength (ρ = −0.381; p < 0.001). Activin A levels were negatively associated with MQ of lower extremities (ρ = −0.282; p = 0.009). IGF-1 correlated positively with muscle mass (ρ = 0.313; p = 0.004) and handgrip strength (ρ = 0.224; p = 0.046) but not with MQ of either upper or lower extremities.

Interestingly, lower levels of myostatin but higher levels of activin A as well as higher activin A-to-follistatin ratios at baseline T1 were associated with a smaller increase in muscle mass between T1 and T3 or even muscle loss (myostatin: ρ = 0.336; p = 0.010 Fig. 2a, activin A: ρ = −0.268 Fig. 2b; p = 0.042, activin A-to-follistatin ratio: ρ = −0.345; p = 0.008). On the other hand, performance in chair stand test at T1 correlated negatively with differences in GDF-15 between T and T3 (ρ = −0.243; p = 0.043) (Fig. 3).

As valuable blood-based biomarkers should be sensitive to changes in skeletal muscle mass or function, associations between the respective differences between T1 and T3 were calculated. In this respect, changes in follistatin correlated positively with changes in muscle mass (ρ = 0.367; p = 0.005) while changes in myostatin or activin A were not associated with skeletal muscle mass alterations (Fig. 2d–f). We did not detect any correlations between changes in blood-based biomarkers and changes in functional parameters or MQ.

In this study the MQ of lower, but not that of upper extremities increased in both training groups (RT and RTS). As whole body skeletal muscle mass was used for the determination of both the MQ of upper and lower extremities, this observation might indirectly reflect the amelioration in chair stand test while handgrip strength was not affected by the strength training programme. Interestingly, the elastic band resistance training that we have used in our study seems to have more pronounced effects on functional performance than on strength as the arm lifting test and chair stand test showed improvements while handgrip strength and isokinetic peak torque measurements of knee extensors and flexors were not affected (Oesen et al. 2015). Furthermore, it is noteworthy that leg muscles suffer greater losses in strength and MQ than arm muscles with ageing (Frontera et al. 2000; Lynch et al. 1999). If we assume that our strength training programme was designed to challenge upper and lower extremities in a similar way, the adaptation capacity of the lower extremities to strength training could have been higher leading to more pronounced effects, although this is in contrast to another study showing that strength gains are higher for upper extremities as the muscles of the lower limbs are elicited more frequently and therefore, have a smaller potential to gain strength at older age (Sousa et al. 2011). Another factor that influences the determination of MQ is the lack of a uniform consensus which parameters to include in the calculation of MQ. While some studies including ours used whole body skeletal mass determined by BIA or DXA (Barbat-Artigas et al. 2012a; Schroeder et al. 2012), others determined skeletal muscle mass separately for upper and lower body by means of DXA or ultrasound (Kennis et al. 2014; Radaelli et al. 2014; Straight et al. 2015). In addition, several formulas are used to determine muscle mass by BIA, whereby it is recommended to use equations derived from a similar age group, which is what we ensured by using the formula of Janssen et al. (2000). This formula has been developed in a multiethnic sample of 388 men and women, aged 18–86 years and includes age and gender as parameters. Furthermore, we checked the quality of BIA measurements by a parallel assessment of muscle mass by DXA in a smaller subgroup of the participants (n = 45). Statistical analyses revealed a high correlation between DXA and BIA (ρ = 0.847, p < 0.001). The situation is even more complex for assessing strength or power as a variety of methods is available. Therefore, we decided to use chair stand test and handgrip test as suggested by Barbat-Artigas et al. (2012b), but future studies evaluating the impact of these different methods on the assessment of MQ and on its influence on clinically relevant outcomes are highly recommended.

Although performance parameters and MQ were increased in both training groups (RT and RTS), follistatin and the activin A-to-follistatin ratio were altered in the RT group only between T2 and T3 hinting to an adaptive delay in the response to training. Follistatin antagonizes myostatin and activin A and as such it is regarded as a positive regulator of muscle mass. Animal studies have shown that acute and chronic endurance exercises increase follistatin mRNA in the liver as well as in skeletal muscles (Hansen et al. 2011; Ziaaldini et al. 2015). Furthermore, serum levels of follistatin increased transiently during an ultra-marathon (Kerschan-Schindl et al. 2015). One of the rare studies investigating the influence of a comparable training setting (endurance or strength training) on follistatin levels in blood and muscle biopsies of young men has been published by Diel et al. (2010). Similar to our study, follistatin remained unchanged in blood and biopsy samples after 3 months (Diel et al. 2010), but the authors did not investigate long-term effects of training making a conclusion difficult. Interestingly, follistatin was not altered in the RTS group. As the strength training programme was the same in both groups, this hints to some direct or indirect effect of the nutritional supplement containing proteins and vitamins on serum follistatin levels. It has been shown that circulating follistatin levels increase in response to a fasting period while activin A levels decrease (Vamvini et al. 2011). We have collected the blood after an overnight fast which was the same in all groups. The nutritional intake was assessed at T1 and T3. In addition, the nutritional supplementation resulted in an increased uptake of vitamin D and folic acid but not in protein being between 0.8 and 1.0 g/kg/day in the RTS group (Franzke et al. 2015a). Furthermore, plasma levels of vitamin B12 and folic acid in erythrocytes were enhanced due to supplementation (Franzke et al. 2015b). There is an ongoing discussion of whether antioxidant supplementation may blunt an exercise-induced training effect (Peternelj and Coombes 2011). With respect to functional performance we did not observe any differences between the RT and RTS group while we did in follistatin levels. In vitro studies have shown that there could be a direct effect of vitamin D administration on follistatin levels as 1α,25-dihydroxyvitamin D3 decreased follistatin in osteoblasts (Woeckel et al. 2013) but increased follistatin in myoblasts (Garcia et al. 2011) showing the complex situation in different organ systems. As circulating levels of follistatin represent an overall measure of all follistatin-generating tissues further studies are needed to elucidate these complex interacting networks.

Follistatin regulates both, activin A and myostatin (Lee et al. 2010; Vamvini et al. 2013). Elevated expression of activins promotes muscle wasting and cachexia, whereas blocking of activin type II receptors induces strong skeletal muscle hypertrophy and protects from atrophy (Chen et al. 2014; Lach-Trifilieff et al. 2014). In addition, Chen et al. (2014) showed that increasing circulating activin A in mice not only promoted the reduction of body weight and muscle mass in a dose-dependent manner, but also reduced muscle function highlighting the therapeutic potential of activin A inhibitors. Therefore, we would have expected a decrease of activin A in the RT and RTS group but instead we detected a small decrease in activin A in CT. However, concerning age-related changes it is still not clear whether activin A levels are influenced by age itself. While some studies have revealed increased circulating activin A levels with age (Baccarelli et al. 2001; Loria et al. 1998), our working group and others did not find any differences between young and old women (Hofmann et al. 2015; Klein et al. 2004). Taking a closer look at the data we found that most of the activin A levels at baseline were below 1 ng/ml while only few subjects (n = 12) displayed higher levels up to 6 ng/ml. Interestingly, higher levels of activin A at baseline were associated with decreases of skeletal muscle mass between T1 and T3 irrespective of the intervention group hinting to a more catabolic situation in these individuals at the beginning of the intervention and potentially making it more difficult for them to increase muscle mass (Fig. 2b). As activin A function can be inhibited by binding of follistatin to activin A, the activin A-to-follistatin ratio was determined. We observed that the ratio was lowered in the RT group shifting the plasma environment to a more follistatin-dominated one while it was unaffected in the CT group despite lower levels of activin A in this group. This highlights the importance of observing networks of biomarkers, such as the follistatin/activin A/myostatin-axis rather than single ones.

Both representatives of the TGF-β superfamily, myostatin (also known as GDF-8) and GDF-15 were found to be unaffected by the intervention. Because of its distinct impact on fat and muscle mass, several studies have dealt with muscular myostatin expression in context with acute strength exercise (Hulmi et al. 2008; Jensky et al. 2010, 2007) and long-term training (Brooks et al. 2010; Diel et al. 2010; Suetta et al. 2013) in young as well as aged women and men. One interesting though unexpected finding of our study was that myostatin levels at baseline correlated positively with changes in muscle mass. Studies investigating circulating levels of myostatin in response to training interventions are still contradictory which weakens final statements on the role of circulating myostatin in adaptations to resistance training. In this respect it has been shown that 10 weeks of high-intensity resistance exercise in young healthy men leads to a decrease in circulating myostatin (Walker et al. 2004). On the other hand serum myostatin propeptide is not altered in young and healthy men performing strength training for 3 months (Diel et al. 2010), and serum myostatin even increases after a 6-month lifestyle intervention programme in obese children (Ehehalt et al. 2011). Differences in measurement methods, age, training loads and training durations may cause these conflicting results. Another hypothesis was provided by Willboughby (2004) who suggested that increases in serum levels of the follistatin-like related gene and the concomitant down-regulation of the activin IIb receptor would counteract even increases in myostatin observed after heavy resistance training. Having these aspects in mind we conclude that the beneficial effects of strength training observed for MQ and functional parameters in our study might be due to blocking of activin A and myostatin by enhanced levels of follistatin rather than by lower circulating levels of activin A and myostatin.

Similar to myostatin we could not find any changes in circulating GDF-15 neither in one of the training groups nor in CT over the time of intervention, but GDF-15 levels were negatively associated with MQ of upper extremities at baseline. Additionally, better performance in the chair stand test was negatively associated with changes in muscle mass between T1 and T3 confirming the negative effects of GDF-15 on skeletal muscle (Fig. 3). This is similar to a study of patients undergoing cardiac surgery in which elevated GDF-15 levels were associated with quadriceps muscle atrophy and were elevated after the surgery (Bloch et al. 2013). Traditionally, this biomarker is suggested to reflect the status of cardiac muscle (Sinning et al. 2015) or lung tissue (Mutlu et al. 2015) and we suggest further investigation is needed to elucidate its role in skeletal muscle.

IGF-1 is considered as an important positive regulator of muscle mass which did not change in response to resistance training with or without supplementation. Our data confirm a previous study in older men and women showing positive effects of a 12-week elastic band exercise programme on body composition and physical fitness without improving IGF-1 levels (So et al. 2013). Similarly, IGF-1 was unaffected by low-load resistance exercise with blood flow restriction in older men (Patterson et al. 2013). However, age seems to be a main determinant in the IGF-1 response to exercise training. We have shown that IGF-1 level differ between young and old women (Hofmann et al. 2015), and young men responded to a strength training programme with increased levels of IGF-1 (Takano et al. 2005). There is growing evidence that a chronic inflammatory state suppresses the GH/IGF1 axis (Andreassen et al. 2012; O’Connor et al. 2008; Pass et al. 2009; Strle et al. 2007). According to the ‘inflammageing theory’ elevated levels of proinflammatory cytokines are found with ageing (Schmidt et al. 2011). As we have shown that hs-CRP concentrations in our study population are higher than in young women (Halper et al. 2015), we hypothesize that increased levels of proinflammatory mediators could have blunted an increase in IGF-1 levels.