Reductions in C-reactive protein in older adults with type 2 diabetes are related to improvements in body composition following a randomized controlled trial of resistance training

Training Protocol

The PRT group trained 3 days per week under supervision using pneumatic resistance equipment (Keiser Sports Health Ltd., Fresno, CA), at two sites. A version of PRT known as power training was employed, in which the concentric contraction (lifting) was completed as quickly as possible, whilst the eccentric contraction (lowering) was completed over 4 s. The exercises targeted large symmetrical muscle groups of the arms, legs, and trunk: seated row, chest press, leg press, knee extension, hip flexion, hip extension, and hip abduction. For each exercise, participants performed 3 sets of 8 repetitions (2 sets of 8 on each leg for hip flexion, hip extension, and hip abduction). The intensity was set at 80 % of the most recently determined 1 repetition maximum (1RM), which was re-assessed every 4 weeks. When 1RM testing was not feasible, resistances were increased by targeting a Borg scale rating of perceived exertion between 15 and 18.

The sham-exercise group trained on the same equipment, 3 times a week, under supervision from the same trainers at different times of the day so as to remain blinded to the investigators’ hypotheses, with both interventions offered as potentially beneficial. The resistance was set as low as possible and not progressed, and participants were instructed to perform concentric and eccentric contractions slowly.

Assessments

Blinded outcome assessments were conducted at 0 and 12 months at the University of Sydney. Participants were required to withhold food, liquids, and medication for 12 h prior to metabolic testing. A 24-h food recall was performed on the day of the baseline assessment, and participants were asked to follow the same diet prior to subsequent assessments at 12 months. Assessments at 12 months were performed 72 h after each participant’s final training session.

Demographics and health status

Participants were asked routine questions to obtain demographic information, as well as their current health status relating to prescribed medications, smoking status (current or past), and the presence of other chronic diseases.

Measure of systemic inflammation

High-sensitivity CRP was measured in duplicate by enzyme-linked immunosorbent assay (ELISA; eBIOSCIENCE, Camarillo, CA), with the average value used in statistical analyses. The lowest detectable concentration was 0.01 mg/L, and the intra-assay coefficient of variation was 6.3 %. To remove the potential confounding of acute inflammation, participants with concentrations of CRP >10 mg/L were excluded from analyses [4].

Anthropometric measurements

Morning fasting stretch stature (wall-mounted Holtain stadiometer, Holtain Limited, Crymych Pembs., UK) and naked weight [weight in gown (kg) − weight of gown (kg)] were measured in triplicate to the nearest 0.1 cm and 0.01 kg, respectively.

Measures of body composition

SMM and total fat mass (FM) were determined using bioelectrical impedance analysis (BIA; RJL Systems, Inc., Clinton, MI, USA) [8, 9]. All participants were fasting and in a state of euhydration, and BIA was performed in the morning, at a similar time of day for all participants. BIA was not available in 3 subjects due to presence of a pacemaker. SMM was determined using the equation from Janssen et al [8]. Skeletal muscle mass was determined using the following equation [8]:

with height in cm, BIA resistance in ohms (average of 3 measures), sex coded 1 for men and 0 for women, and age in years.

FM was determined using BIA, by subtracting lean body mass from total mass to determine FM. Lean body mass was determined using the equation from Lukaski et al [9].

with height (Ht) in cm, BIA resistance in ohms (average of 3 measures), naked body weight (BW) in kg, Xc reactance in ohms, and sex coded 1 for men and 0 for women

Computed tomography (CT; GE High Speed CTI Scanner, MIL, USA at the Royal Prince Alfred Hospital, Sydney) was used to quantify VAT (cm²), mid-thigh muscle cross-sectional area (cm²; CSA), and mid-thigh muscle attenuation.

For CT scans of the abdomen, a 1-mm-thick slice was taken at the mid-point between the iliac crest and lowest rib (determined with the participant supine). This scan location is concordant with the waist circumference measurement site used by the International Diabetes Federation and is a criterion used by the IDF to classify metabolic syndrome [10]. Settings were kV = 100 and mA = 170 with a displayed field of view (DFOV) 45–48, depending on subject size.

For CT scans of the mid-thigh, a 1-mm slice was performed at the mid-point between the inguinal crease and the proximal pole of the patella measured with the subject supine and knee flexed. The non-dominant leg was used. Settings were kV = 100 and mA = 170 with a (DFOV) 25. Scan images were analysed according to optical density by a blinded investigator using NIH Image software (Version 1.63, National Institutes of Health) programmed with specific macros to quantify cross-sectional adipose tissue. To determine VAT, macros were programmed to select the outer perimeter extending from the paraspinal muscles to the anterior abdominal muscles. The program calculated this measure by summing the area within the selected perimeter occupied by pixels with optical density in the range of 140 to 240. Thigh muscle density was calculated according to a specific optical density range (10–113) chosen to best discriminate muscle from fat and bone. Co-efficient of repeatability (CR), determined using a Bland–Altman plot on a subset of 10 scans, was found to be excellent at 0.49 for VAT, 0.44 and for mid-thigh CSA [11].

Statistical analyses

All data were assessed for normality using histograms and descriptive statistics. Normally distributed data are presented as mean ± SD; otherwise, data are presented as median (range) or frequencies as appropriate. CRP data required square root transformation to achieve normal distribution for use with parametric statistics. Comparisons of variables between groups were performed using a one-way ANOVA. Linear regressions were performed to determine the relationship between continuous variables at baseline. A linear mixed-effects model with repeated measures was used to determine changes over time as this method allows for all available data to be used without imputation for missing values. CRP was entered as the dependent variable with group, time, and group × time interaction entered as fixed effects and an unstructured covariance matrix. To determine if there was any association between group assignment and changes in measures of body composition on reductions in CRP, sequential linear mixed models were constructed with CRP as the dependent variable and full factorial interaction between group, time and the change in the body composition variable of interest. Any variables found to be associated with CRP using mixed-effects models (p < 0.10) were then explored using linear regression models stratified by group allocation. A p value <0.05 indicated statistical significance as all hypotheses were specified a priori. IBM SPSS version 21 (SPSS Inc. Chicago, IL, USA) was used for all analyses.

Associations between body composition and CRP at baseline

Results are presented in Table 2. As hypothesized, CRP was directly associated with body weight (r = 0.53, p < 0.0001) and BMI (r = 0.52, p < 0.0001) and FM (r = 0.59, p < 0.0001); but unexpectedly, only a trend was observed with VAT (r = 0.21, p = 0.10). SMM was directly associated with CRP (r = 0.33, p < 0.01); however, mid-thigh CSA was not associated with CRP (r = 0.03, p = 0.84).

A stepwise multiple regression model was then constructed with CRP entered as the dependent variable, and SMM and BMI entered as the independent variables, since both variables were related to CRP, and did not display colinearity (r = 0.51). In this model, only BMI was associated with CRP (r = 0.50, p < 0.0001), while SMM no longer showed any independent association (r = 0.08, p = 0.53). Body weight was unable to be used as this displayed colinearity with SMM (r > 0.7). Similarly, FM displayed colinearity with body weight and BMI (r > 0.7), and thus precluded the use of this variable in the model.

Changes in body composition

Data for the full cohort of participants has been previously reported here [12]. In line with our previously reported data, there was also no group × time interaction for changes in whole body SMM or FM within this subset with CRP data (data not shown). However, in contrast to the whole cohort, a significant group × time interaction was found within this subset for changes in mid-thigh CSA (β = 5.49, p = 0.03) and VAT (β = −19.14, p = 0.03) in favor of the PRT group.

Changes in CRP

Results are presented in Table 2. Using a linear mixed-effects model, a trend for a group × time interaction was present, with greater reductions in CRP in the PRT group relative to sham-exercise, as hypothesized (β = −0.25, p = 0.087).

Association between changes in body composition and changes in CRP

Results are presented in Table 3. Changes in body composition variables were entered as covariates, with a full factorial interaction with group and time to compare the effects of changes in body composition on CRP between the PRT and sham groups. In these models, increases in SMM in the PRT group were significantly associated with reductions in CRP compared to increases in SMM in the sham group (β = −0.23, p = 0.04). Similarly, reductions in FM tended to be significantly associated with reductions in CRP in the PRT group compared to reductions in FM in the sham group (β = 0.09, p = 0.07). No differences between the PRT and sham groups in terms of the relationship with CRP were observed with changes in body weight (β > −0.01, p = 0.94), BMI (f < 0.01, β = 0.01, p = 0.95), VAT (β < 0.01, p = 0.56), mid-thigh CSA (β = 0.01, p = 0.70), or mid-thigh muscle attenuation (β = −0.03, p = 0.73).

Due to the significant interaction noted above of group, time, and body composition change, we stratified by group assignment to explore the relationships within each group. Using univariate regression models, stratified by group allocation, a significant inverse association was found between changes in CRP and changes in SMM in the PRT group (r = −0.48, p = 0.01), but this was not seen in the sham-exercise group (r = 0.03, p = 0.87; Fig. 2a). Similarly, changes in CRP were directly related to changes in FM within the PRT group (r = 0.45, p = 0.02), but not in the sham group (r = −0.18, p = 0.32; Fig. 2b).

Association between skeletal muscle and CRP

The direct association between SMM and CRP at baseline within our cohort is in agreement with previous investigations reporting the cross-sectional relationship between systemic inflammation and SMM or lean body mass in other cohorts [17, 18]. The authors in prior studies concluded that lean tissue might in fact contribute to systemic inflammation. However, our analyses show that after adjusting for BMI, SMM is no longer associated with CRP. This is likely due to the direct relationship between SMM and BMI; with over nutrition and excess adiposity resulting in a higher quantity of SMM. Furthermore, our longitudinal data, in agreement with some but not all previous investigations [13] indicate that anabolic exercise such as PRT can reduce systemic inflammation. Furthermore, our linear mixed-effects models and linear regression models indicate that increases in SMM following PRT were associated with reductions in CRP, indicating that increasing lean tissue with anabolic exercise may potentially mediate reductions in systemic inflammation.

Given that CRP has been shown to predict the loss of SMM in older adults [7], the observed association between reductions in CRP with increases in SMM is of great significance. Previous data within a similar cohort showed no relationship between changes in lean body mass and changes in CRP following 4 months of PRT [19]. To our knowledge, ours is the only other study to investigate the relationship between changes in markers of systemic inflammation, and changes in lean tissue in a cohort of individuals with type 2 diabetes. Therefore, our results provide a significant and novel advance in knowledge in this field. A notable difference between our study and that of Brooks et al [19] is that in the latter, this relationship was explored between changes in CRP and changes in quadriceps muscle type I fiber cross-sectional area only, as opposed to whole body SMM within our investigation. It is possible that changes in whole body SMM are more reflective of changes in systemic inflammation. In agreement with this, increases in regional indices of SMM (CT scan) within our cohort showed no relationship with reductions in CRP. Further studies utilizing more precise whole body composition techniques such as dual-energy X-ray absorptiometry or multi-slice MRI scanning are required to further explore this relationship.

Association between adiposity and CRP

Figure 2b illustrates that reductions in CRP following PRT were related to reductions in fat mass. This is important, given the predictive value of CRP in CVD. Until recently, CRP was believed to be produced exclusively by hepatocytes, namely in response to circulating interleukin-6 (IL-6) [20]. However, production of CRP within subcutaneous and visceral adipose tissue has also been reported [21–23]. Thus, reductions in circulating CRP can be influenced both directly, and indirectly, by reductions in adipose tissue. The finding that CRP is produced by both subcutaneous and visceral adipose tissue supports our finding that reductions in FM (inclusive of both depots) predicted reductions in CRP, rather than VAT in isolation. In addition to adipose tissue, macrophages and smooth muscle cells from atherosclerotic plaques [24–26] and inflamed renal epithelial cells [27] have been shown to contribute to circulating levels of CRP. Adipose tissue is a major contributor to circulating cytokines, which could initiate or mediate local inflammatory responses (including CRP production) within these tissues [28]. Thus, reduced adiposity after PRT could have both directly and indirectly reduced circulating CRP.

The role of inflammation in adaptations to PRT

Given the known catabolic effects of systemic inflammation, and its role in anabolic resistance [29], our data suggest that reductions in CRP may allow more robust increases in SMM in response to PRT. However, our study does not prove that reductions in inflammation secondary to reduced adiposity preceded the skeletal muscle hypertrophy, and thus further investigations are needed to delineate the time course of these adaptations to PRT. An alternative pathway to CRP reduction after PRT is also possible. Skeletal muscle is known to secrete IL-6 in response to muscle contraction, which can bring about an anti-inflammatory environment by inducing the production of the anti-inflammatory cytokines IL-1ra and IL-10 [30]. Importantly, the magnitude of IL-6 production is dependent on the amount of skeletal muscle recruited [30, 31], which would be optimized by training with high speeds/volumes/intensities/numbers of muscle groups (such as our power training group). These circulating anti-inflammatory myokines may also potentially mediate the effects of PRT by suppressing CRP production within adipose and non-adipose tissues, thus contributing to reduced inflammation. These pathways to lower CRP (reduced fat mass and increased muscle contractile activity) may operate simultaneously and require further study to determine their relevance and importance in type 2 diabetes and other cohorts characterized by chronic systemic inflammation.

Anti-inflammatory efficacy of body composition changes may depend on how they are achieved

Reductions in FM following PRT were significantly associated with reductions in CRP, while reductions in FM following sham-exercise were not (Fig. 2b), despite similar reductions in FM between groups, as previously reported [12]. This suggests that not only is reduction in adipose tissue necessary for reduction in systemic inflammation, but the intervention chosen to mediate this body composition shift may play an important role. Other strategies to reduce adipose tissue, such as dietary changes, would not incorporate any potential therapeutic effect of skeletal muscle contraction. In addition, dietary restriction by itself may cause losses of muscle tissue [32] as well as adipose tissue. Thus, the benefits of fat loss could be attenuated or eliminated by the adverse effects of muscle loss and reduced contractile activity on overall inflammatory milieu.

It is also possible that combined strategies to reduce inflammation may be even more beneficial than any single intervention such as anabolic exercise. While our data suggest that reductions in CRP can be achieved through PRT-mediated reductions in adiposity, other interventions to reduce systemic inflammation, such as non-steroidal anti-inflammatory drugs [33], poly-unsaturated fatty acids [34, 35], or antioxidants could potentially enhance the anabolic response to PRT in older adults with type 2 diabetes.

Future directions, limitations, and conclusions

Most importantly, further prospective studies are needed to determine the nature of the relationship between CRP and SMM, which is potentially bi-directional. Elevated CRP may attenuate a robust response to anabolic stimuli, thus both leading to sarcopenia and impeding recovery from it. Alternatively, sarcopenia/dynapenia and the reduced contractile activity accompanying this condition could reduce anti-inflammatory myokine release into the circulation and thus contribute to an overall enhancement of systemic inflammatory profile. Understanding of these relationships may improve targeting of individuals likely to require augmentation of standard PRT regimens to enhance body composition and metabolic outcomes in clinical cohorts.

In conclusion, we have demonstrated that 12 months of PRT reduced systemic inflammation in older adults with type 2 diabetes, mediated through improvements in body composition, relationships which were not seen with sham exercise. The significant associations between changes in CRP and changes in FM and SMM after PRT indicate that both body composition compartments should be targeted in evidence-based lifestyle interventions aimed at maximally reducing systemic inflammation in older adults with type 2 diabetes. It is time that the insufficient goal of “weight loss” be replaced with “muscle gain and fat loss” in overweight/obese cohorts if optimal metabolic health is to be achieved.