One Bout of Exercise Alters Free-Living Postprandial Glycemia in Type 2 Diabetes

Douglas J. Oberlin; Catherine R. Mikus; Monica L. Kearney; Pamela S. Hinton; Camila Manrique; Heather J. Leidy; Jill A. Kanaley; R. Scott Rector; John P. Thyfault


Med Sci Sports Exerc. 2014;46(2):232-238. 

In This Article

Materials and Methods


Sedentary and low-activity individuals with T2D were recruited in and around Columbia, MO, who, on average, took fewer than 6000 steps per day (pedometer) and did not participate in any formal exercise program (>30 min of planned exercise two times a week). Pedometers were worn for a 3-d period to measure daily step count. All subjects were nonsmokers with a body mass index between 30 and 42 kg·m−2 who did not have orthopedic limitations to exercise safely on a treadmill and stationary bike. They were weight stable (±5%) and had no changes in medication for at least 3 months before entering the study. In addition, the subjects had controlled diabetes with HbA1c < 7.5% with no insulin use and no reported advanced retinopathy or neuropathy. Other exclusion criteria included breakfast skippers, pregnancy, sleep perturbations, night shift workers, people who have recently traveled across more than two time zones, or individuals with irregular daily schedules to avoid odd sleep or eating patterns that may have conflicted with the study design. All women enrolled in the study were postmenopausal. All subjects signed an informed consent, which was approved by the institutional review board of the University of Missouri.

After the consent meeting, the subjects came to the exercise physiology laboratory during the morning after a 10-h fast for measurement of height, weight, blood pressure, fasting blood glucose, glycated hemoglobin (HbA1c), and blood lipids (total cholesterol, LDL, HDL, and triglycerides). HbA1c was analyzed on a Siemens DCA Vantage analyzer using blood drawn in a heparin tube. Blood lipids (cholesterol, triglycerides, HDL, LDL, and total cholesterol) and fasting glucose were measured by a commercial laboratory as done previously.[24] Subjects were then given a diet log and a pedometer to measure dietary consumption and daily steps, respectively, for a 3-d period. On another visit, the subjects had their body composition measured via dual-energy x-ray absorptiometry (Hologic QDR 4500A; Hologic, Bedford, MA). The subjects then performed an exercise stress test to determine maximal oxygen consumption (V̇O2peak), maximal heart rate, and to screen for any potential cardiac abnormalities via EKG. The exercise stress test was performed on a treadmill using a standardized Bruce protocol as performed previously.[24] During the test, respiratory gases were measured by a metabolic cart (ParvoMedics True One 2400 Metabolic Measurement System; ParvoMedics, Sandy, UT), and cardiovascular function was monitored using a 12-lead EKG (Quinton Q-Stress v3.5 Exercise Test Monitor; Quinton Cardiology, Inc./Burdick, Deerfield, WI). The criteria for a maximal test were three of the following: volitional exhaustion, perceived exertion of 17 or greater, respiratory exchange ratio of greater than 1.0, or a leveling off or slight decrease in oxygen consumption. The EKG data from each exercise stress test were reviewed by a cardiologist to ensure that the participants could safely participate in an exercise session. There was a 5- to 15-d washout after the V̇O2peak test before the subjects began the study protocol.

Study Design

The study incorporated a randomized crossover design and consisted of measuring minute-to-minute interstitial glucose concentrations by CGMS during a 2-d sedentary condition (subjects maintained normal daily activities) and another 2-d condition in which the subjects exercised during the morning of the first day (followed by maintaining normal daily activities). Therefore, subjects served as their own controls. We have previously used CGMS to measure glycemic control in free-living subjects who are healthy as well as those with T2D.[24,25] During each condition, the subjects consumed the study diet, which consisted of three meals a day representing breakfast (8:00 a.m.), lunch (1:00 p.m.), and dinner (6:00 p.m.) for 4 d. The first 2 d of the diet were to acclimate the subject to the new diet. The following 2 d of the standard diet coincided with CGMS measurements to ensure that any changes to glycemic control were due to exercise and not due to dietary changes. Figure 1 shows the 4-d study period, which the subjects repeated twice, once performing an exercise (EX) bout in the morning before breakfast and once while remaining sedentary (SED). During the SED condition, the subjects continued their typical (sedentary) physical activity, which was verified using a Walk4Life Duo pedometer (Walk4Life Inc., Plainfield, IL) and an accelerometer (BodyMedia SenseWear armband body monitoring system; BodyMedia Inc., Pittsburgh, PA). Both pedometer and accelerometer were used primarily to assess physical activity within subjects, between conditions. Validation studies on the accuracy of SenseWear to measure energy expenditure have provided mixed results;[12,17] however, the device allowed a basis of comparison between the two conditions and showed that exercise increased energy expenditure. There was a 5- to 15-d washout period between the two conditions of the study.

Figure 1.

Study design for the exercise (EX) versus sedentary (SED) phases of the study. Subjects repeated two separate phases in which they wore CGMS and consumed provided meals three times per day. In both phases, CGMS was placed on the evening of the second wash-in day. On the morning of day 1, subjects either performed EX or remained SED. Glucose levels were measured by CGMS, and three prepared meals were consumed per day (six meals total) during the next 2 d.

Exercise Testing and Exercise Session

Graded exercise tests to measure V̇O2peak were performed on a treadmill using a Bruce Protocol as done previously.[23] The EX bout consisted of 60 min of aerobic exercise broken into three 20-min sections starting at approximately 6:30 a.m. This included 20 min on a treadmill, 20 min on a stationary cycle, and another 20 min on a treadmill. The exercise session was broken into these segments because the subjects could not do 60 min on the treadmill at this intensity without needing to sit down. The exercise intensity was within 5 beats per minute of 60% of HRR for the duration of the exercise bout (as determined from a previous graded exercise stress test). Intensity was controlled during the exercise session by adjusting speed or grade on the treadmill or adjusting resistance on the stationary cycle, to maintain the target heart rate throughout the exercise bout. The exercise prescription falls in line with the dual recommendation of 150 min·wk−1 at an intensity of 40%–60% V̇O2peak for patients with T2D by the American Diabetes Association and the American College of Sports Medicine.[1] In addition, the same exercise prescription was used in a previous study from our laboratory, which showed a decrease in PPG during 5–7 d of exercise training in previously SED subjects with T2D.[24] Subjects were instructed to take their medications as prescribed throughout both conditions and were monitored postexercise to avoid hypoglycemia. Thus, in this study, we wanted to determine whether and for how many meals one bout of exercise prescribed at the same intensity and duration would improve postprandial glycemic (PPG) responses to subsequent meals.

Study Diet

As stated previously, the subjects consumed the same study diet in both phases (Table 1). The diet was prepared and packed out by study staff. The subjects were instructed to eat the meals at the same times each day and to allow 5 h between meals. Every meal had the same nutrient composition and caloric content and contained the exact same food items prepared as either breakfast, lunch, or dinner. Breakfast was a potato hash with seasoned ground beef topped with salsa and cheese served with buttered toast, applesauce, and a juice drink. Lunch was mini cheeseburgers with salsa mixed into the patties and baked french fries served with a side of applesauce and a juice drink. Dinner was a mini-meatloaf with salsa and cheese baked in and mashed potatoes served with garlic toast, applesauce, and a juice drink. Subject were instructed to leave uneaten foods in containers when returning them to the laboratory, allowing us to asses food compliance to a certain degree. True compliance to the diet was not directly assessed as we did not monitor them eating the meals directly.

The macronutrient distribution based on energy content was 51.4% carbohydrate, 30.9% fat, and 17.8% protein for the total energy content for each meal (Table 1). The glycemic load of each meal was approximately 55. The daily energy requirement for each subject was estimated using the Harris–Benedict equation and also quantified using a 3-d dietary record completed by the subject. The 3-d mean was then averaged with the Harris–Benedict estimate to determine individual energy requirements and to avoid deficits due to underreporting in the dietary records. From this information, the subjects were provided a diet containing 1600, 1800, 2000, 2200, or 2400 kcal·d−1, whichever kilocalorie level was within 100 kcal of their predicted requirements. For example, if a person was estimated at 2063 kcal, they would receive the 2000-kcal diet while an individual estimated at 2115 would receive a 2200-kcal diet. The diet was designed to simulate a typical American diet and provide consistent diet composition between meals and across subjects; it was not meant to serve as an intervention, although it did differ from their normal routine.

The subjects were given a log sheet to track when they ate their meals. They were instructed to eat all of the portions within a 15- to 20-min time frame. Subjects also noted when they went to bed at night and when they got up in the morning. No additional calories were given to replace those lost in the exercise session, which created a deficit. Although this is a known confounder, replacing the calories would have lead to a mismatch of macronutrients between conditions, which would have also been a confounding variable. Therefore, the meals were kept constant through both conditions.

PPG Control

A Medtronic iPro CGMS (Minneapolis, MN) monitor was attached to the subject's abdomen with a probe inserted beneath the skin, and the monitor was attached and taped down with Smith & Nephew IV3000 adhesive pads the night before the first measurement day. The CGMS was then worn for two consecutive days. While the CGMS was worn, the subjects recorded four daily blood glucose levels with an Accu-Chek Compact Plus glucometer, which were then used to calibrate the CGMS. We used the CGMS glucose concentration data to quantify the 24-h average glucose for each day in each condition. We also used the glucose concentrations along with subject-recorded meal times to calculate the PPG-AUC for each of the six meals (quantified as the 4-h glucose AUC response after each meal). Blood glucose values recorded every 15 min during the 4-h period were used to calculate total AUC. We also quantified 2-h postprandial glucose concentration for each of the six meals (PPG-2 h) because it is a strong predictor of cardiovascular events.[6]

Physical Activity Monitoring

Subjects wore pedometers and BodyMedia SenseWear Pro II arm band accelerometers during both phases of the study. Data from the accelerometers (estimated energy expenditure) and pedometers (daily steps) were collected.

Statistical Analysis

Statistical analysis was analyzed using the Statistical Package for the Social Sciences (version 15; IBM, Armonk, NY) on PPG-AUC, PPG-2 h, average 24-h glucose concentration, estimated caloric expenditure from accelerometer, and daily steps using a two-way repeated-measures ANOVA. The two levels were condition (EX or SED) and meal[1–6] for all except the average 24-h glucose, steps, and energy expenditure, for which the two levels were condition and day (1 or 2). Post hoc analyses were conducted using Fisher least significant difference. The level of statistical significance was set at a P value of 0.05, and the data are reported as means ± SE.