Diabetes mellitus is a worldwide epidemic. It is reported by the International Diabetes Federation that at least 177 million people in the world have diabetes and World Health Organization figures estimate that this will rise to 300 million by 2025 1. Approximately 90–95% of people who are diagnosed with diabetes have type 2 diabetes, which results from insulin resistance combined with relative insulin deficiency 2. The beneficial effect of exercise on insulin sensitivity has been well established 3. Exercise alone, in the absence of body weight change, is able to enhance glucose homeostasis 4. Because skeletal muscle is the largest reservoir for glucose disposal 5, muscle wasting from aging, sarcopenia and/or inactivity exacerbates problems of peripheral glucose uptake due to insulin deficiency or insensitivity. Muscle weakness, decreased muscle mass and decreases in skeletal muscle fiber numbers and size are related to, and may precede, insulin resistance, glucose intolerance, and type 2 diabetes 6, 7. Although it has been shown that exercise can enhance insulin sensitivity and glycemic control 4, 8, the mechanisms underlying such benefits are not fully understood. Indeed there are recent reviews of the metabolic benefits of physical activity 9–12, but there is no systematic review of the effects of exercise on specific skeletal muscle adaptations in people with impaired glucose tolerance (IGT) or type 2 diabetes to our knowledge.
Therefore, the purpose of this investigation was to systemically review morphological and metabolic adaptations within skeletal muscle to exercise training in people with type 2 diabetes mellitus or IGT, for the purpose of identifying underlying mechanisms that may be targeted in future pharmacologic or lifestyle interventions to combat this global epidemic.
A systematic review of all published literature, regardless of study design, investigating adaptations within skeletal muscle to exercise training exposure in individuals with type 2 diabetes mellitus or IGT was conducted. One of the (YW) authors conducted the search; the other two authors (DS and MFS) reviewed all retrieved articles for eligibility, and accuracy of data extraction onto pre-designed forms. Disagreements were resolved by consensus. A quantitative meta-analysis was not undertaken due to the wide variety of interventions used and outcomes assessed, but individual effect sizes (ESs) were calculated within each trial whenever possible using published data.
A literature review was conducted from the year 1966 to 2008 using computerized databases including Medline (1967–2008), Premedline (most recently published), CINAHL (1982–2008), AMED (1985–2008), EMBASE (1966–2008) and SportDiscus (1967–2008), with the last search being conducted in March 2008. Firstly, three categorical searches were conducted using the following keywords: (1) ‘exercise’, ‘physical activity’, ‘training’, ‘aerobic’, ‘resistance’, ‘weight lifting’ or ‘strength’; (2) ‘diabet’, ‘insulin’ or ‘glucose’; and (3) ‘biopsy’ or ‘muscle’. Secondly, searches 1–3 were combined using ‘AND’ and duplicate results were removed. All titles were manually searched for potential inclusion. Reference lists of retrieved papers, authors' names, and review articles were hand-searched for additional relevant citations. Unpublished theses were not included in the literature search.
Articles were selected from the initial search on the basis of the following criteria: the study design was a randomized controlled trial, non-randomized controlled trial, or uncontrolled trial with the intervention of an aerobic or resistance or combined training; human subjects with type 2 diabetes or IGT were used; a muscle biopsy was conducted; the full-length article was published in a peer-reviewed journal. Aerobic exercise involves exercise performed for extended periods with large muscle activity involving hundreds of consecutive repetitions that challenge the delivery of oxygen to the active muscles. Resistance training involves weight training or the use of high-resistance machines with exercise that is limited to a few repetitions (generally less than 20) before exhaustion 13. Dietary co-interventions were allowed, only if the exercise effect could be isolated from the dietary effect.
Studies were excluded for the following reasons: only animal subjects were studied; subjects without diabetes or IGT were included; only subjects with type 1 diabetes were included; an acute exposure only to an exercise bout of any kind (<3 sessions); an exercise intervention meeting the above definitions was not included in the study design; no biopsy results were presented; they were non-English language papers or unpublished; the articles were reviews or abstracts.
Study quality was assessed based on a modification of Delphi List 14 for assessing the quality of randomized control trials and was extended to non-randomized controlled trials. Additional quality variables considered were supervision of exercise training, compliance, and dropout rate; all considered important in exercise studies.
Data were extracted for the description of methodology and outcomes of each trial. From this data, values reported as standard error (SE) of the mean were converted to standard deviation (SD). Study characteristics were summarized as mean (SD) or median (range) as appropriate for normal or non-normally distributed data, respectively. Relative ES (mean changeTreatment − mean changeControl) ÷ SDPooledbaseline was calculated for controlled trials if it was applicable 15. ES for each of the controlled trials was interpreted according to the method of Cohen 16.
The study screening process at each step is presented in Figure 1. The 26 papers 17–42 were retrieved from 18 studies including 3 randomized controlled trials reported in 9 publications 17–25 (2 trials reported in 8 publications with resistance training 17–24 and 1 with aerobic training 25), 2 non-randomized controlled trials 26, 27 (with aerobic training 26, 27) and 13 uncontrolled trials reported in 15 publications 28–42 (11 trials reported in 13 publications with aerobic training 28, 30–41, 1 with resistance training 42 and 1 with combined training 29).
Study quality assessment
An assessment of the study quality according to a modified Delphi list is presented in Table 1. The study quality was performed for three randomized controlled trials reported in nine publications 17–25 and two non-randomized controlled trials 26, 27. In general, the study quality was moderate, and none of the studies met all of the Delphi list quality criteria. No randomized controlled trials reported allocation concealment. All studies specified their inclusion criteria, and reported groups or subjects were similar at baseline regarding important prognostic valuables. Four studies published in ten papers reported dropouts with rates ranging from 0–9.1% 17–24, 26, 27. One study published in four papers reported exercise compliance 17–20. Only one study published in four papers reported blinding of outcome assessors 17–20. The participants from four studies reported in ten publications performed their training intervention under complete supervision 17–26; in the other studies, instructions were given on how to increase the physical activities and the exercise sessions were recorded 27. One study published in four papers employed an intention-to-treat (ITT) analysis with imputation of missing data 17–20 (60 out of 62 subjects completed the study, method of imputation not reported); one study published in four papers 21–24 planned an intention-to-treat analysis but had no dropouts.
Table 1. Study quality
Treatment allocation concealed?
Groups/subjects similar at baseline regarding important prognostic values?
Eligibility criteria specified?
Blinded outcome assessors?
Supervision of exercise sessions?
Did the analysis include an intention-to-treat analysis?
Were point estimates and measures of variability presented for the primary outcome measures?
Four resistance-type exercise machines targeting the upper body and leg press and leg extension
High week 1–5: 50% 1RM; week 6–10: 50–60% 1RM
Two sets of ten repetitions
NR (two biopsies: one before and one after the training)
In total, the 18 studies consisted of 291 subjects, ranging in size from 6 to 50, with median 10 and interquartile range 44.
In the randomized controlled trials and non-randomized controlled trials, all studies consisted of mixed gender cohorts 17–27; in the uncontrolled trials, five studies published in six papers were in male cohorts 31, 32, 34, 37, 38, 42, seven studies published in eight papers were in mixed cohorts 28–30, 33, 35, 36, 39, 41 and one study did not report gender 40.
Most subjects were middle-aged or older, with few subjects above 65. Across all cohorts, the age ranged from 36 to 66, with median 59 and interquartile range 30.
Body mass index (BMI) was reported in all but four uncontrolled studies 28, 29, 31, 40. Almost all subjects were overweight or obese with mean BMI 29.6 ± 3.1 kg/m2, except for one uncontrolled study 37 with lean subjects (BMI = 23.6 ± 2.5 kg/m2).
Most [61% (11/18)] trials 17–24,27–29,31,32,34–36,38,40,42 included only type 2 diabetes; 28% (5/18) of trials 25, 26, 30, 37, 41 included only IGT; 11% (2/18) of trials 33, 39 included both type 2 diabetes and IGT. Overall, 66% of the subjects studied had type 2 diabetes and 34% of the subjects had IGT.
Except for one uncontrolled trial 33, all other studies of type 2 diabetes 17–24,27–29,31,32,34–36,38–40,42 reported the subjects' diabetic medications. Among these studies, two trials reported in five papers included some or all subjects with insulin treatment 17–20, 42; subjects in the other 10 studies reported in 15 publications 21–24, 27–29, 31, 32, 34–36, 38–40 were treated only with oral agents or did not take diabetic medications. In studies including type 2 diabetes, 18% of subjects took insulin, 81% took oral agents, and 16% did not take diabetic medications. Duration of diabetes was reported in all but four studies published in five papers 29, 33, 35, 36, 39 and ranged from 2.5 to 12.1 years with median 6.0 years and interquartile range 9.6 years across these nine trials.
Only two studies published in five papers 17–20, 27 reported medications other than diabetic medications, and only one study published in four papers 17–20 reported chronic diseases other than type 2 diabetes, which included hypertension, cardiovascular disease, or exercise limitations.
Nine trials published in ten papers 25,26,29,30,33,, 35–38,42 reported habitual physical activity level as “sedentary”, although precise definitions were not given.
Table 2 provides an overview of the training interventions.
In the exercise training studies, the most common form of training prescribed (78% of studies) was aerobic training. Two randomized controlled trials reported in eight papers 17–24 and one uncontrolled trial 42 used resistance training, and one uncontrolled trial 29 used combined aerobic and resistance training.
Except for one study 27 of low intensity aerobic training, all exercise prescriptions consisted of either moderate or high intensity aerobic and/or resistance training. In the three resistance training studies reported in 9 publications, the intensity increased progressively from 50 to 80% of the one-repetition maximum (1RM) 17–24, 42, thus concluding training with high intensity progressive resistance training at the time of biopsy. In the aerobic training studies, the intensity was low in one study, moderate in six studies, and high in six studies. One study used moderate intensity in one group and high in the other 30. The combined training study utilized high intensity aerobic training (80–90% VO2max) plus light muscle power training 29.
In the three resistance training studies reported in nine publications, the volume was three sets of eight repetitions 17–20, three sets of ten repetitions 21–24, and two sets of ten repetitions 42. In the aerobic training and combined training studies, the volume varied from 30 31, 32, 38 to 60 34, 37, 41 min per session of continuous exercise, mean 48 ± 9 min.
Three sessions per week was the most common training prescription 17–24, 27, 28, 34, 37, 42, but it ranged from two 29, 40 to six sessions per week 31, 32.
Trial duration ranged from 4 to 52 weeks, median 12 weeks; the majority of studies were between 4 and 16 weeks 17–32, 34–39, 41, 42 and two studies were between 36 and 52 weeks 36, 40.
Time of biopsy
Ten trials published in 17 papers 17–26, 29–32, 34–36 reported time interval between the last exercise session and biopsy. It ranged from 16 to 96 h, with median 60 and interquartile range 80.
In general, morphological outcomes were improved after exercise training with moderate to large ESs, and were specific to the type of training (Tables 3 and 4). Muscle fiber size (muscle fiber diameter or cross-sectional area) of both Type I and Type II fibers significantly increased in three studies published in four papers 19, 20, 23, 29, but not in the fourth 42, which included resistance training. Of the three positive studies, two studies published in three papers 19, 20, 23 included only resistance training and one included combined training 29. In contrast, muscle fiber size did not increase in any of the three trials that included only aerobic training 30, 33, 37. However, capillary density increased after aerobic training 29, 37, but not after resistance training 23.
One randomized controlled trial including resistance training 20 reported that the improved muscle fiber cross-sectional area was accompanied by reduced whole body insulin resistance. The other randomized controlled trial included one-leg resistance training 23 and reported that the increased leg glucose clearance was more than explained by increases in muscle mass. However, four uncontrolled trials 29, 30, 37, 42 did not report any relationships between the changes in the morphological outcomes and whole body insulin sensitivity.
Glycogen and glycogen synthase
Increased glycogen and glycogen synthase levels were the most consistent positive adaptations observed (Tables 5 and 6), and in general, the ESs were very large. Four trials reported in five publications 17, 18, 25, 26, 30 (one with resistance training and three with aerobic training) of six studies reported in seven publications for glycogen 17, 18, 23, 25, 26, 30, 42, reported significant improvements in glycogen content, and one 23 of those six studies reported a trend for increased glycogen content after resistance training. The other study  reported that glycogen content inside the muscle fibres didn't change significantly after exercise. Three (one with resistance training and two with aerobic training) 23, 32, 35 out of four studies for glycogen synthase 23, 25, 32, 35 reported significant increases in this enzyme activity after training durations ranging from 6 to 9 weeks. In one study 25, the combination of a high-carbohydrate diet and exercise resulted in a significant decrease in glycogen synthase compared to the high-carbohydrate diet only group. Thus, both resistance and aerobic training have been shown to improve glycogen and glycogen synthase, and no data are available on direct comparisons of the efficacy of different exercise modalities for these outcomes.
Five 23, 30, 31, 35, 37 out of seven 18,23,27,30,31,, 35,37 studies reported that glucose facilitated transporter 4 (GLUT4) mRNA and/or protein expression increased after either resistance or aerobic exercise training (Tables 7 and 8). However, one randomized controlled trial with resistance training reported that GLUT4 mRNA did not change, but human sodium dependent D-glucose co-transporter 3 (hSGLT3) mRNA significantly increased 18.
Table 7. Metabolic outcomes—GLUT4 and insulin signaling (randomized controlled trials and non-randomized controlled trials)
Pre-exercise (Mean ± SD)
Post-exercise (Mean ± SD)
p value (EXE vs. CON)
RCT, randomized controlled trial; NRCT, non-randomized controlled trial; CON, control; EXE, exercise; NR, not reported; NS, not significant; IR, insulin receptor; IRS-1, insulin receptor substrate-1; PI-3K, phosphatidylinositol 3-kinase; PI-3K p85, the p85 subunit of PI-3K; PKB, protein kinase B; GLUT4, glucose facilitated transporter 4; hSGLT3, human sodium dependent d-glucose co-transporter 3; AMPK, 5′AMP-activated protein kinase; EX responders: subjects who showed a decrease in both systolic and diastolic blood pressure (an indicator of response to exercise).
As for other parameters in the insulin signalling pathway, one randomized controlled trial 23 showed that the protein content of insulin receptor (IR) increased. The serine/threonine kinase Akt, also known as protein kinase B (PKB), has recently come to the forefront as a potential downstream mediator of insulin signalling pathway 35. Notably, both aerobic training 35 and resistance training 23 have improved Akt/protein kinase B.
Overall, the improvements of GLUT4, insulin receptor and Akt/protein kinase B by both resistance and aerobic exercise could play a beneficial role in insulin sensitivity, if these preliminary findings are confirmed.
One randomized controlled trial of resistance training 18 reported that the change in hSGLT3 transcript levels was positively correlated with glucose uptake, as measured by the change in muscle glycogen stores (r = 0.53, p = 0.02). The other randomized controlled trial of one-leg resistance training 23 reported that the increased protein content of GLUT4, insulin receptor, protein kinase B are part of the mechanism behind the improvement in insulin action. Two uncontrolled trials 31, 37 did not report any relationships between the changes in the morphological outcomes and whole body insulin sensitivity. And, in another two uncontrolled trials, one 30 reported that the increase in GLUT4 was not significantly related to the changes in insulin-stimulated glucose disposal and the other 35 reported that the improvements in the phosphatidylinositol (PI) 3-kinase of insulin receptor signalling and GLUT4 expression were accompanied by improved insulin-stimulated glucose disposal and insulin-stimulated glucose storage.
Only few data were available (Tables 9 and 10). In the only resistance training trial reported in three publications 22–24 for hydroxyl-acyl-CoA dehydrogenase (HAD), which is involved in lipid oxidation, no significant changes were shown, whereas aerobic training 34, 37 significantly improved HAD. Citrate synthase, a measure of oxidative capacity, was also improved by this aerobic training intervention 34, but not by resistance training 22–24. One trial 34 reported that the enhanced insulin sensitivity observed after exercise was associated with a marked increase in citrate synthase and HAD. However, other studies did not report such relationships.
Two trials 19, 36 measured some inflammatory factors, but one 19 of them did not find any changes after exercise (Tables 11 and 12). The other trial of aerobic exercise 36 reported some beneficial changes, which included increases in inhibitor κB α (IκBα), inhibitor κB β (IκBβ), nuclear factor κB p50 (NFκB p50) and reduction in tumor necrosis factor α (TNFα) protein. It has been hypothesized that excessive activity of IκB/NFκB is the mechanism underlying skeletal muscle insulin resistance 43. This trial 36 also reported that subjects with type 2 diabetes had reduced IκB protein abundance in muscle compared to healthy subjects. It is noteworthy that abundance at the protein level of IκB is inversely correlated with the activity of the IκB/NFκB pathway (decreased IκB levels reflecting increased phosphorylation and proteolysis of the protein), thereby contributing to the activation of IκB/NFκB pathway and leading to insulin resistance. Therefore, in this study, the increases of inhibitor κB α, inhibitor κB β and NFκB p50 protein levels in response to training showed that this abnormality is potentially modifiable by exercise training 36. None of these studies presented any data investigating whether whole body insulin sensitivity was associated with changes in these inflammatory markers, however.
Some beneficial improvements were seen in lipid metabolism after exercise training (Tables 13 and 14). In the two aerobic training trials 34, 37 for intra-muscular triglyceride concentration (IMTG), it decreased significantly after exercise training. One of those two studies 34 also showed that the enhanced insulin sensitivity observed after exercise was associated with a marked decrease in intra-muscular triglyceride concentration. Two other aerobic training trials 27, 40 reported increases of uncoupling protein 3 (UCP3) after exercise training, which would theoretically protect mitochondria against lipid-induced mitochondrial damage 44. Increases in adiponectin receptor I and II were reported in one aerobic training trial 41, and these changes could play a part role in improved insulin sensitivity. However, no consistent pattern has been demonstrated for other lipid or adipokine outcomes.
One randomized controlled trial reported increases of Na, K-pump content 21, which has been demonstrated to be low in patients with type 2 diabetes compared to healthy adults 45, 46 and has been reported to increase both with endurance and sprint training in healthy individuals 47–52. The intracellularly produced lactate leaves the cell via simple diffusion and two proteins: monocarboxylate transporter (MCT) 1 and MCT4 53. The density of MCT1 and/or MCT4 in human skeletal muscle is elevated after endurance training 54–56 and resting blood lactate concentrations may be elevated in patients with type 2 diabetes 57–59. One randomized controlled trial 22 reported a significant increase of MCT after resistance exercise. One uncontrolled trial reported increases of metallothioneins I and II (MTs-I + II), which have antioxidant, anti-inflammatory, and antiapoptotic functions 38.
Quality and quantity of retrieved studies
A limited amount of data exists exploring the adaptation of skeletal muscle to exercise training in subjects with IGT or type 2 diabetes. Only 3 randomized controlled trials, 2 non-randomized controlled trials and 13 uncontrolled trials of exercise training adaptations were isolated after extensive searches of the published literature. The quality of the randomized controlled trials and non-randomized controlled trials was only moderate, and none of the studies met all of the Delphi list quality criteria. Most of these studies did not report exercise compliance and blinding of outcome assessors, and did not employ an intention-to-treat analysis. Most of the trials included small sample sizes (median 10), which could cause inaccuracy due to subjects' individual differences and type II error due to underpowering.
Of the four studies reported in five papers 19,20,23,, 29,42, which included resistance 19, 20, 23, 42 or combined modality 29, muscle fiber size was significantly increased by an average of 19.0% in three studies reported in four publications 19, 20, 23, 29, but not in any of the three aerobic training trials 30, 33, 37. The 19% increase in fiber area in diabetes is somewhat lower than the 28% increase in healthy older (60- to 72-year-old) subjects reported 60, 61. In contrast, capillary density increased after aerobic training 29, 37, but not after resistance training 23. These findings were, respectively, consistent with the known effects of resistance training or aerobic training in healthy adults. In healthy untrained adults, the principal adaptation to resistance training is the significant increase in muscle fiber size 62, 63. Eventually, the increases in size for a large number of fibers will result in a noticeable increase in the cross-sectional size of the muscle 63. However, there is usually no increase in the numbers of capillaries observed after resistance training. The increases in size of muscle fibers result in the muscle capillaries being moved apart, a situation that has been described as capillary dilution 63. One systematic review reported conflicting evidence regarding the effects of resistance training on muscle capillarization in the elderly subjects 64. An increase in muscle fiber size was accompanied by significant increases in capillary contacts per fiber, capillary to fiber ratio and capillary to fiber perimeter exchange index, but the growth of capillaries was proportional to fiber growth, resulting in no change to sharing factor, capillary density or capillary contacts per fiber area 64–66.
On the other hand, the major adaptation to aerobic exercise is muscle tissue capillarization 63. Harris also reported that aerobic exercise can result in positive structural and functional changes with respect to skeletal muscle capillarization in elderly subjects 64. Similarly, an eight-week aerobic exercise intervention increased capillary contacts, capillary-to-fiber ratio, and capillary density in both healthy young and elderly subjects, but did not change any of the parameters regarding muscle fiber size 67.
Thus, resistance training may improve muscle fiber size but not capillarization in subjects with IGT or type 2 diabetes, whereas aerobic exercise may improve muscle capillarization but not muscle fiber size in these subjects. These adaptations appear qualitatively and quantitatively similar to the adaptations observed in healthy young and older adults. However, there are no studies directly comparing the magnitude of such benefits in cohorts with and without diabetes.
Glycogen and glycogen synthase
Glycogen content increased in all but one 42 of six studies reported in seven papers 17, 18, 23, 25, 26, 30, 42; significantly in five studies published in four papers 17, 18, 25, 26, 30, and with a similar trend (15.8%, p < 0.1) after resistance training in the last 23. It is likely that limitation in study design (one legged training, short (6 weeks) duration may have attenuated the response in this study 23 and that small sample size (10) may have resulted in a type II error. The other study 42, which reported no significant change in glycogen (p = 0.662) after resistance training, was the only study that included long-standing (diabetes duration 12.1 ± 7.0 years), insulin-treated (exogenous insulin treatment for 7.0 ± 8.0 years) subjects. This study included subjects with more severe type 2 diabetes than any other retrieved studies. The long diabetes duration requiring insulin treatment may have blunted the ability to store glycogen in these subjects in response to exercise, and measurement of glycogen synthase would have been helpful to determine why glycogen storage could not be increased in these subjects.
Glycogen synthase increased after both resistance and aerobic training 23, 32, 35, but not when aerobic exercise was combined with a high CHO diet compared to diet alone 25. This may be explained by the fact that participants in the diet plus exercise group had greater muscle glycogen content than did the diet only group, and this was associated with a lower proportion of active muscle glycogen synthase relative to inactive enzyme 25 due to down-regulation of the enzyme. A lower active to inactive glycogen synthase activity ratio limits glycogen storage 68 and thus the major site of insulin-stimulated glucose disposal during a euglycemic-hyperinsulinemic clamp 69. Thus, by chronic stimulation of glucose uptake in muscle, exercise plus a high-carbohydrate diet result in a greatly increased muscle glycogen content with a concomitant decrease in active glycogen synthase, limiting any further effect of exercise on glucose disposal during laboratory testing, as was seen in this study 25.
GLUT4 mRNA and/or protein expression increased in five 23, 30, 31, 35, 37 of seven trials 18, 23, 27, 30, 31, 35, 37 after exercise training. In one negative randomized controlled trial of resistance training 18, hSGLT3, an insulin-independent glucose transporter (but not GLUT4 mRNA), markedly increased in the exercise group 18. GLUT4 allows transport of glucose down its concentration gradient, while hSGLT3 transports glucose against its concentration gradient 70. In this trial 18, muscle glycogen storage significantly increased after training and there was a significant direct association between enhanced hSGLT3 transcript levels and increased muscle glycogen stores. Taken together, the data suggest that insulin-independent hSGLT3 but not insulin-dependent GLUT4 may have been involved in the observed, exercise-stimulated muscle glucose storage 18. However, this trial was the only one that reported that the subjects had ‘uncontrolled type 2 diabetes (characterized by poor glycemic control and sustained hyperglycemia)’ and was the only one reported that some of the subjects were treated with insulin 18 out of the seven studies for GLUT4 18, 23, 27, 30, 31, 35, 37. Five of the other six studies 23, 30, 31, 35, 37 for GLUT4 expression showed significant increases. Among these, two trials 30, 37 included patients with IGT; one trial included patients with type 2 diabetes in good health without insulin treatment 35 and the other two trials 23, 31 did not report that their patients had uncontrolled type 2 diabetes or took insulin. It is possible that the state of glucose control in these subjects could be better than that in Castaneda's study 18, and that this may explain the observed GLUT4 adaptation.
Unlike hSGLT3, GLUT4 is insulin-dependent, and it translocates from the intracellular vesicle storage to the sarcolemma in response to exercise and/or insulin action 71. It has been proposed that insulin resistant individuals have a defect in GLUT4 trafficking and targeting leading to reduced GLUT4 in the cell membrane in skeletal muscle 72–74. Thus, the more severe insulin resistance in Castaneda's study 18 may have caused the inability of exercise to stimulate GLUT4 function, and different degrees of insulin resistance may be associated with different pathways of exercise-enhanced glucose uptake. However, as there is only one trial 18 investigating hSGLT3 and exercise, further investigation is required to confirm these findings and assess the relevance to enhanced glucose uptake after exercise.
Enzymes/inflammatory factors/lipids and miscellaneous adaptations
Five studies reported in seven publications 22–24,27,34,, 37,42 investigated enzymes, and three of them 22–24, 27, 42 did not find any significant changes. Citrate synthase and HAD were markedly improved by aerobic training 34, 37 but not by resistance training 22–24. The changes in amount of citrate synthase and HAD (73% and 62%, respectively) after exercise training in subjects of type 2 diabetes mellitus were similar to those in a healthy control group (85% and 48%, respectively) 34. Other than exercise modality, there is no obvious explanation for this lack of adaptation in resistance training studies.
Of the three studies 19, 27, 36 measured inflammatory factors, two 19, 27 did not find any significant changes and one 36 reported some beneficial changes: increases in inhibitor κB α/inhibitor κB β/NFκB and reduction in tumor necrosis factor α. Adiponectin receptor I and II increases were reported in one aerobic training trial 41.
Eight studies 24, 27, 29, 34, 37, 40–42 tested some parameters of lipid metabolism. Some beneficial improvements were seen in this domain after exercise training. Two aerobic training trials 34, 37 reported decreases in intra-muscular triglyceride concentration. Two other aerobic training trials 27, 40 reported increases of uncoupling protein 3. However, no consistent pattern has been demonstrated for other lipid outcomes.
Notably, all positive adaptations above (enzymes/ inflammatory factors/lipids) are from uncontrolled trials 34, 36, 37, 40, 41, except one nonrandomized controlled trial, which reported the increases of uncoupling protein (UCP)3 27. These results can therefore only be considered a weak form of evidence at present, and they require confirmation in randomized controlled trials.
Time intervals between the last exercise session and biopsy are provided in Table 2. However, no correlation could be found between biopsy time and metabolic outcomes.
Limitations of this investigation and future directions for research
The biggest limitation of existing data is the small quantity (only three randomized controlled trials) and moderate quality of available literature. Since there were only five controlled trials, data available for ES calculations were quite limited, and therefore the magnitude and consistency of changes observed in many domains require confirmation in future trials. For all outcomes, inadequate reporting of mean and SD precluded calculation of ES in many studies.
Only 3 trials included resistance training, and the other 14 trials included aerobic or combined exercise training. As it has been shown in this investigation that resistance training did improve muscle fibre size as well as insulin signalling and glycogen storage (a capacity for glucose uptake and storage) in patients with type 2 diabetes or IGT, more studies focused on resistance training are required, as it may represent an important alternative approach to effectively manage diabetes and attenuate its long-term complications. In particular, direct comparisons of the relative benefits of aerobic versus resistance training for exercise-induced skeletal muscle adaptation are needed.
Unfortunately, most studies in this review included moderate- to high-intensity exercise (only one study included low intensity exercise), 30–60 min each session, 2–6 times/week, and 4–16 weeks in duration (only two studies included training more than 16 weeks). Therefore, limited data were available with which to examine dose–response effects across varying exercise characteristics. Thus, no obvious conclusions could be drawn from the extracted results that we had analyzed.
Exercise clearly improves insulin sensitivity and glucose homeostasis 9. The purpose of this review was to examine changes at the level of skeletal muscle, which may explain the overall benefits to metabolic control in this rapidly expanding cohort. The most consistent adaptations were in the areas of improved muscle fiber size and glucose transport, with inconsistent evidence for beneficial adaptations in the areas of enzymes, inflammatory factors, and lipid metabolism. Some of those studies 18, 20, 23, 30, 34, 35, 41 reported positive correlations between improved whole body insulin sensitivity and adaptations to exercise in the area of muscle fiber size, glucose transport, enzymes and lipid metabolism. However, preliminary results reviewed here still need to be confirmed in additional controlled, large trials in this cohort.
Future studies should investigate the insulin signalling pathway to better understand the mechanism of the improvements in insulin sensitivity and glucose homeostasis in response to exercise in this cohort. In addition, whole and regional body composition rising criterion methods, whole body insulin sensitivity and long-term glucose control measurements and their relationships to skeletal muscle morphological and biochemical changes should be included to advance understanding in this field. Elucidation of this pathway and its adaptation may allow the development of improved pharmacological, dietary, and physical activity interventions to address the most important mediators of glucose regulation. Ultimately, this understanding may lead to earlier diagnosis, greater prevention, and reduced clinical sequelae and morbidity associated with insulin resistance and type 2 diabetes.
Yi Wang was supported by University of Sydney International Postgraduate Research Scholarship.