Hyperglycemia blunts the therapeutic effect of exercise on glycemic control in patients with type 2 diabetes

TO THE EDITOR

Randomized controlled trials show that aerobic exercise training improves glycemic control in type 2 diabetics ¹. However, inter-individual variability is large². This may be explained by genetic variability³, but ambient hyperglycemia⁴ and pancreatic beta-cell function⁵ may also contribute. We examined whether changes in glycemic control following a 12-16-week aerobic exercise training intervention were influenced by the pre-training glycemic state in 105 individuals with impaired glucose tolerance or type 2 diabetes.

Methods

Before and following a 12-16 week period of aerobic exercise training, body composition, aerobic fitness (VO₂max), and glycemic control (hemoglobin [Hb]A_1c, fasting glucose, and oral glucose tolerance test [OGTT]) were determined in a total of 105 older (61±1 years; mean±S.E.M.), overweight/obese (33±1 kg/m²) subjects with impaired glucose tolerance (N=56) or type 2 diabetes (N=49; diagnosed 4.8±0.9 years prior and not insulin-treated). Relationships between pre-intervention variables and intervention-induced changes in variables were assessed by linear and non-linear regression. The study was approved by our institutional review board and subjects provided informed-consent. See eMethods (http://archinte.jamanetwork.com) for full details of the study design.

Results

Body weight (−4.6±0.5 kg), whole-body adiposity (−1.9±0.3%), VO₂max (+0.23±0.03 L/min), fasting plasma glucose (−0.35±0.08 mmol/L), and 2-hour OGTT glucose (−0.8±0.2 mmol/L) were significantly improved following exercise training (full data is shown in eTable 1). Pre-training fasting plasma glucose did not influence exercise-induced changes in glycemic control. However, there was a non-linear quadratic relationship between pre-training 2-hour OGTT glucose and exercise-induced changes in the 2-hour glucose response (r²=0.26, P=0.06; Fig. A). Subjects with a pre-training 2-hour OGTT glucose of less than 13.1 mM showed greater exercise-induced decreases in 2-hour glucose (r=-0.44, P<0.001); while subjects with a pre-training 2-hour OGTT glucose of greater than 13.1 mM had smaller improvements in 2-hour glucose (r=0.29, P=0.07). The same non-linear quadratic relationship existed between pre-training HbA_1c and exercise-induced changes in HbA_1c (r²=0.33, P<0.05; Fig. B); where subjects with pre-training HbA_1c less than 6.2% had an exercise-induced decrease in HbA_1c (r=−0.55, P<0.01), while subjects with pre-training HbA_1c greater than 6.2% had smaller exercise-induced improvements in HbA_1c (r=0.38, P<0.05). Furthermore, pre-training HbA_1c was linearly and inversely related to the exercise-induced change in VO₂max, such that high pre-training HbA_1c predicted smaller exercise-induced increases in VO₂max (r=−0.38, P<0.01; Fig. C).

Fig 1.

Subjects with impaired glucose tolerance or type 2 diabetes underwent 12-16 weeks of moderate-intensity exercise training, 5 d/wk, 60 min/d. Individual subject data points are plotted on both panels; the X-axis represents the pre-training variable and the Y-axis indicates the exercise-induced change such that data points above the axis indicate an exercise-induced increase, and vice versa. Open circles (○) represent impaired glucose tolerant subjects; open triangles (Δ) represent subjects with type 2 diabetes. The solid line represents the regression curve; the dotted line represents the 95% confidence interval. [A] There was a non-linear quadratic relationship between pre-training 2-hour OGTT glucose and the training-induced change in 2-hour OGTT glucose (y=0.06x²−1.5x+7.6; r²=0.26, P=0.06; N=105). For every 1 mM increase in pre-training 2-hour glucose above 13.1 mM (the inflection point of the curve), there was a 0.2 mM loss of improvement in 2-hour glucose following exercise. [B] There was also a non-linear quadratic relationship between pre-training HbA_1c and the training-induced change in HbA_1c (y=0.31x²−3.8x+11.7; r²=0.33, P<0.05; N=52). For every 1% point increase in pre-training HbA_1c above 6.2% (the inflection point of the curve), there was a 0.2% point loss of improvement in HbA_1c following exercise. [C] An inverse linear relationship between pre-training HbA_1c and the training-induced change in aerobic fitness was found (y=−0.11x+0.91; r=−0.38, P=0.006; N=52). For every 1% increase in pre-training HbA_1c, there was 0.11 L/min loss of improvement in VO₂max following exercise training.

COMMENT

These findings emphasize that exercise-induced improvements in glycemic control are dependent on the pre-training glycemic level. We demonstrate that although moderate-intensity aerobic exercise can improve glycemic control, individuals with ambient hyperglycemia are the most likely to be non-responders. Our key observation is that pre-training hyperglycemia predicts exercise-induced improvements in glycemic control: for every 1 mM rise in pre-training 2-hour OGTT glucose above 13.1 mM (the curve inflection point in Fig. A) we predict a 0.2 mM loss of improvement in 2-hour OGTT glucose following exercise. Accordingly, for every 1% point increase in pre-training HbA_1c above 6.2% (the curve inflection point in Fig. B), we predict a 0.2% point loss of improvement in HbA_1c following exercise. Pre-training hyperglycemia also predicted the exercise-induced increment in aerobic fitness: for a 1% increase in pre-training HbA_1c we predict a 0.11 L/min loss of improvement in VO₂max following exercise.

Prior work shows that diabetes remission following exercise and diet intervention is more likely in individuals with a shorter disease history and lower HbA_1c⁶. Here we show that aerobic exercise-induced improvements in glycemic control are blunted by ambient hyperglycemia, particularly in subjects with type 2 diabetes. Mechanistic studies are required to help us understand this phenomenon, but underlying impairments in beta-cell function are likely to be very important⁵. That hyperglycemia blunted the cardiovascular adaptations to exercise (VO₂max) is in agreement with some prior reports⁷ and may be explained by the causal association between chronic hyperglycemia and micro- and macrovascular dysfunction⁸.

The clinical relevance of these new findings is paramount as they highlight the need to understand the metabolic “non-responder”. Because chronic hyperglycemia (>6.2% HbA_1c; >13.1 mM glucose) potentially predicts a poor therapeutic effect of aerobic exercise upon glycemic control and fitness, using exercise to treat patients with poorly-controlled type 2 diabetes may have limited chances of a successful outcome.