Abstract
Introduction
This year, we screened more than 750 potentially eligible titles on PubMed and other common search engines for manuscripts on “exercise and diabetes” published between July 1, 2018, and June 30, 2019. This year's articles tended to focus on testing artificial pancreas prototypes and approved devices in various settings of increased physical activity. Considerable improvements have been made in detecting exercise types and intensities so that smarter algorithms and controllers can make better changes to insulin for active patients living with type 1 diabetes. We also found several papers on new ways to make exercise training more effective for diabetes‐related outcomes including glycemic control and body composition. The 10 highlighted papers below represent some of the major research publications in the field of exercise and diabetes.
Key Articles Reviewed for the Article
Turksoy K, Hajizadeh I, Hobbs N, Kilkus J, Littlejohn E, Samadi S, Feng J, Sevil M, Lazaro C, Ritthaler J, Hibner B, Devine N, Quinn L, Cinar A
Hu Y, Zhang DF, Dai L, Li Z, Li HQ, Li FF, Liu BL, Sun XJ, Ye L, He K, Ma JH
Forlenza GP, Buckingham BA, Christiansen MP, Wadwa RP, Peyser TA, Lee JB, O'Connor J, Dassau E, Huyett LM, Layne JE, Ly TT
Ortiz‐Rubio P, Oladunjoye A, Agus MSD, Steil GM
Savikj M, Gabriel BM, Alm PS, Smith J, Caidahl K, Björnholm M, Fritz T, Krook A, Zierath JR, Wallberg‐Henriksson H
Pinsker JE, Laguna Sanz AJ, Lee JB, Church MM, Andre C, Lindsey LE, Doyle FJ 3rd, Dassau E
Hangping Z, Xiaona Q, Qi Z, Qingchun L, Na Y, Lijin J, Siying L, Shuo Z, Xiaoming Z, Xiaoxia L, Qian X, Jaimovich D, Yiming L, Bin L
Petruzelkova L, Soupal J, Plasova V, Jiranova P, Neuman V, Plachy L, Pruhova S, Sumnik Z, Obermannova B
Zaharieva DP, McGaugh S, Pooni R, Vienneau T, Ly T, Riddell MC
Breton MD, Patek SD, Lv D, Schertz E, Robic J, Pinnata J, Kollar L, Barnett C, Wakeman C, Oliveri M, Fabris C, Chernavvsky D, Kovatchev BP, Anderson SM
Multivariable artificial pancreas for various exercise types and intensities
Turksoy K1, Hajizadeh I2, Hobbs N1, Kilkus J3, Littlejohn E3,4, Samadi S2, Feng J2, Sevil M1, Lazaro C5, Ritthaler J6, Hibner B6, Devine N3, Quinn L7, Cinar A1,2
Departments of 1Biomedical Engineering and 2Chemical and Biological Engineering, Illinois Institute of Technology, Chicago, IL; 3Section of Endocrinology, Department of Pediatrics and Medicine, Kovler Diabetes Center, University of Chicago, Chicago, IL; 4Sparrow Medical Group/Michigan State University, Lansing, MI; 5Department of Electrical and Computer Engineering, Illinois Institute of Technology, Chicago, IL; 6Division of Biological Sciences, University of Chicago, Chicago, IL; 7College of Nursing, University of Illinois at Chicago, Chicago, IL
Background
Blood glucose control during exercise can be challenging for individuals with type 1 diabetes (T1D). A multivariable adaptive artificial pancreas (MAAP) operates without user input and can determine insulin needs based on continuous glucose monitor (CGM) and exercise signals. Therefore, the MAAP may help reduce the burden of being physically active with T1D.
Methods
A total of 18 closed‐loop experiments with 96 exercise sessions were completed. Exercise consisted of resistance exercise and either moderate‐intensity or high‐intensity exercise each day. The primary outcome was time spent in each glycemic range during and postexercise and the secondary outcomes included average glucose and average change in glucose during exercise.
Results
At exercise onset, blood glucose levels were above target at 185.5±56.6 mg/dL, 166.9±61.9 mg/dL, and 171.7±41.4 mg/dL for moderate‐intensity, resistance, and high‐intensity exercise sessions respectively. During exercise and in recovery, glucose was in a safe range (70–250 mg/dL) 93% of the time and with target (70–180 mg/dL) 70% of the time with MAAP. Glucose levels were within the severely hypoglycemic (<55 mg/dL), moderate hypoglycemic (55–70 mg/dL), moderate hyperglycemic (180–250 mg/dL), and severe hyperglycemic (>250 mg/dL) ranges for 0.9%, 1.3%, 23.1%, and 4.8% of the time, respectively. Average glucose decline was greater during moderate‐intensity versus resistance exercise (P<0.05). MAAP recommended carbohydrates in 59% of the moderate‐intensity exercise sessions, in 50% of the high‐intensity sessions, and in 39% of the resistance exercise sessions.
Conclusion
Blood glucose declined during all three exercise modalities and in recovery, but to levels that were deemed to be safer overall because of preexercise hyperglycemia. Overall, hypoglycemic events occurred in 14.6% of exercise sessions, representing only 2% of the overall time in the exercise and recovery period combined.
For individuals with T1D to reduce the likelihood of hypoglycemia during exercise, significant preplanning is often required. This may include reducing basal insulin preexercise (1,2) or reducing the mealtime bolus insulin before exercise (3). However, in the case of unplanned exercise, options are more limited (i.e., supplemental carbohydrate feeding). With newer closed‐loop technologies, an exercise announcement can help communicate to the insulin pump so insulin levels can be adjusted before the onset of exercise. We agree that exercise announcements using automated insulin delivery are beneficial for attenuating hypoglycemia, but in most cases, they still require advanced notice. In this important study, Turksoy and colleagues showed that their multivariable adaptive artificial pancreas (MAAP) provided hypoglycemia early alarms and carbohydrate recommendation algorithms without any user input to announce exercise. Although hypoglycemia was not eliminated entirely during exercise, MAAP could maintain glucose levels safely and minimize the prevalence of hypoglycemia independent of the exercise type, despite requiring the system to detect exercise onset and predict a glucose decline before measures could be taken to prevent hypoglycemia. As automated insulin delivery systems continue to evolve, this study is important as it highlights many of the challenges associated with exercise and the need to test these algorithms under real‐life conditions.
Pre‐exercise blood glucose affects glycemic variation of aerobic exercise in patients with type 2 diabetes treated with continuous subcutaneous insulin infusion
Hu Y1, Zhang DF1, Dai L1, Li Z1, Li HQ1, Li FF1, Liu BL1, Sun XJ1, Ye L2, He K3, Ma JH1
1Department of Endocrinology, Nanjing First Hospital, Nanjing Medical University, Jiangsu, China; 2National Heart Research Institute Singapore, National Heart Centre Singapore, Singapore; 3Department of Endocrinology, Wuxi Hospital Affiliated to Nanjing University of Chinese Medicine, Jiangsu, China
Background
For individuals with type 2 diabetes (T2D), exercise may improve insulin sensitivity and overall glycemic control as measured by glycated hemoglobin. However, in individuals with T2D treated with insulin, the impact of exercise on glycemic variation needs to be studied further. This study examined the influence of aerobic exercise in individuals with T2D treated with continuous subcutaneous insulin infusion (CSII).
Methods
This study included 267 adults with T2D using CSII. All participants completed at least one aerobic exercise session (30 minutes of treadmill walking) and were separated into effective and ineffective groups based on incremental area under curve (AUC) from 0 to 60 minutes postexercise (AUC0–60min).
Results
A total of 776 aerobic exercise sessions were examined. Blood glucose levels decreased in the first 60 minutes (i.e., within the 30 minutes of exercise and first 30 minutes of recovery) and then began to increase slightly thereafter. Preexercise blood glucose (PEBG) was negatively correlated with AUC0–60min (standardized β=−0.386, P<0.001) and incremental AUC of blood glucose ≤ 4.4 mmol/L (standardized β=−0.078, P=0.034). PEBG was also found to be significantly higher in the effective (n=181 patients) versus ineffective (n=86 patient) group (P<0.001). The Δglucose AUC0–60min postdinner was significantly greater than during prelunch, postlunch, and predinner (all P < 0.05).
Conclusion
PEBG is positively correlated with the efficacy of aerobic exercise, and when PEBG>16.7mmol/L, aerobic exercise will not worsen hyperglycemia. In addition, postdinner exercise decreases blood glucose levels better than exercise during other times of the day.
This study was the first to investigate glycemic variation with continuous glucose monitoring (CGM) during and after aerobic exercise in individuals with T2D using CSII. For individuals living with diabetes, American Diabetes Association/American College of Sports Medicine guidelines recommend if blood glucose levels >16.7 mmol/L (>300 mg/dL) without ketones, they should only exercise if they feel well enough (4). Interestingly, the authors in this featured paper found that aerobic exercise did not worsen hyperglycemia when PEBG >16.7 mmol/L (>300 mg/dL). As such, we agree that the standards for exercise should be more relaxed in patients with T2D treated with CSII. In addition, this study found that exercise postdinner was more effective for glucose‐lowering than exercise during other times of the day, though postdinner PEBG was not higher than that of the other periods. It was also important to note that the risk of hypoglycemia increased when the PEBG was lower. Similarly, a study conducted in youth with T1D found that glycemic variability is partially explained by preexercise blood glucose levels and that near euglycemia at the start of aerobic exercise increases the risk of exercise‐related hypoglycemia (5). Therefore, a lower starting glucose level should be taken into consideration before exercise onset even in T2D. Moreover, the best time of day for exercise in patients with T2D, if given a choice, appears to be after dinner, as long as patients are motivated to do it.
Performance of Omnipod personalized model predictive control algorithm with moderate intensity exercise in adults with type 1 diabetes
Forlenza GP1, Buckingham BA2, Christiansen MP3, Wadwa RP1, Peyser TA4, Lee JB5, O'Connor J5, Dassau E6, Huyett LM5, Layne JE5, Ly TT5
1Barbara Davis Center for Diabetes, University of Colorado School of Medicine, Aurora, CO; 2Division of Pediatric Endocrinology, Department of Pediatrics, Stanford University, Stanford, CA; 3Diablo Clinical Research, Walnut Creek, CA; 4ModeAGC LLC, Palo Alto, CA; 5Insulet Corporation, Acton, MA; 6Harvard John A. Paulson School of Engineering and Applied Sciences, Harvard University, Cambridge, MA
This manuscript is also discussed in the article on Decision Support Systems and Closed Loop, page S‐47.
Background
This study assessed the performance and safety of the Omnipod® personalized model predictive control (MPC) algorithm with variable glucose set points and moderate‐intensity exercise using an investigational device in adults with type 1 diabetes (T1D).
Methods
Following a 7‐day outpatient standard treatment phase, 12 adults with T1D were supervised for 54 hours with a hybrid closed‐loop (HCL) system in a hotel setting. Participants completed two moderate‐intensity exercise sessions (duration >30 minutes) on consecutive days. For the first, the glucose set point increased from 130 to 150 mg/dL and the second, a temporary basal rate of 50%, started 90 minutes preexercise. The primary endpoints were percent time in hypoglycemia <70 mg/dL and hyperglycemia ≥250 mg/dL.
Results
Participants were aged 36.5±14.4 years (mean±standard deviation), diabetes duration 21.7±15.7 years, glycated hemoglobin (HbA1c) 7.6%±1.1%, and total daily dose 0.60±0.22 U/kg. Outcomes for the 54‐hour HCL period were mean glucose 136±14 mg/dL; percent time <70 mg/dL 1.4%±1.3%, 70–180 mg/dL 85.1%±9.3%, and ≥250 mg/dL 1.8%±2.4%. In the 12‐hour period after exercise onset, percent time <70 mg/dL was 1.4%±2.7%, with the raised glucose set point and 1.6%±3.0% with reduced basal rate. On both study nights, percent time <70 mg/dL overnight was 0%±0%.
Conclusion
The Omnipod® personalized MPC algorithm was safe and performed well in response to variable glucose set points and with temporarily raised glucose set points or a reduced basal rate set 90 min before moderate‐intensity exercise in adults with T1D.
For individuals with T1D, maintaining blood glucose levels in a safe glycemic range during and postexercise can be challenging and depending on the strategy used, sometimes ineffective. As such, automated insulin delivery systems may have the potential to reduce the burden associated with diabetes management; however, these systems also need to be able to differentiate between the type, intensity, and duration of exercise and adjust insulin needs accordingly. Automated insulin delivery systems are often able to reduce the likelihood of hypoglycemia during prolonged aerobic type activity; however, additional carbohydrate consumption is often still required before, during, or after exercise to prevent or even treat hypoglycemia. In this study, Forlenza et al. demonstrated the Omnipod personalized MPC algorithm performed well during the 54‐hour period where adults with T1D completed moderate‐intensity exercise with a temporary glucose set point increase or basal rate reduction 90 minutes preexercise. In this study, one caveat is that for both temporary glucose set point increase and basal rate reduction strategy, 5/12 participants still had supplemental carbohydrates (without an insulin bolus) before exercise that ranged from 25 to 40 grams. Future studies may consider implementing a more aggressive exercise announcement strategy, such as an 80% basal rate reduction set 90 minutes preexercise, as has been shown to work reasonably well in open‐loop studies (1). Finally, this study showed that the Omnipod personalized model predictive control algorithm with variable glucose set points is also effective in eliminating nocturnal hypoglycemia with the combination of a preexercise announcement and the algorithm attenuating insulin delivery.
Adjusting Insulin Delivery to Activity (AIDA) clinical trial: effects of activity‐based insulin profiles on glucose control in children with type 1 diabetes
Ortiz‐Rubio P1, Oladunjoye A2, Agus MSD2, Steil GM2
1Division of Endocrinology, Children's Hospital Colorado, Aurora, CO; 2Medicine Critical Care, Department of Medicine, Boston Children's Hospital, Boston, MA
Background
Increased afternoon physical activity levels in children with type I diabetes mellitus (T1D) causes hypoglycemia. This study determined if nighttime basal insulin adjustments could be titrated based on weekly review of accelerometer, CGM, and insulin pump data.
Methods
Children with T1D at risk of nighttime hypoglycemia were identified from regression analysis of daytime step count versus nighttime nadir glucose. If the regression slope was significantly different from zero, subjects were managed with different nighttime basal insulin profiles, using an algorithm, following high and low activity days.
Results
Of the 20 children with T1D enrolled in the study (ages 7–17 years), regression slope analyses identified 10 children at risk for nocturnal hypoglycemia. In these 10 children, baseline nighttime nadir glucose level was lower following high activity days (120 [110–139] vs low activity days (152 [130–162] mg/dL; P=0.004). Use of activity‐based nocturnal basal profiles produced similar nighttime nadir glucose levels following high (136 [123–175] mg/dL) and low (140 [108–180] mg/dL) activity days (P=0.73), with fewer nighttime interventions to correct hypoglycemia.
Conclusions
Children with lower nighttime glucose levels following high daytime activity can be identified using step count data obtained from readily available accelerometers. Using this approach, modification to basal insulin profiles can increase nadir glucose levels and markedly lower hypoglycemia risk.
Basal insulin delivery should be reduced by about 20% overnight to protect against postexercise, late‐onset hypoglycemia (6). Being able to identify children with T1D at risk for nocturnal hypoglycemia based on daily activity levels would help make activity‐based changes to evening insulin levels more intuitive for patients and practitioners. This group of researchers demonstrated that using daily step counts from a Fitbit activity monitor, as a proxy for physical activity levels, may be a useful way to identify children at risk for nocturnal hypoglycemia. In just over half of the 20 children enrolled in the study, daily step count was a significant predictor of nighttime nadir glucose. In the other half, no relationship was observed. Based on the work of these authors, a threshold step count of about 6,000 steps by 8 p.m. appeared to be associated with increased hypoglycemia risk in more than half the children, while others had step counts well above 10,000 with little to no hypoglycemia risk. Interestingly, the children with an activity effect already had basal rates configured to prevent hypoglycemia after high activity days and these same basal rate settings tended to result in hyperglycemia overnight on less active days. This finding allowed the research team to better configure a more aggressive nocturnal basal delivery following sedentary days. These results support the notion that these families have a fear of nocturnal hypoglycemia often caused by increased physical activity levels (7). This study is important as we think about other metrics that can help us guide activity‐based changes to insulin delivery in active children with T1D.
Afternoon exercise is more efficacious than morning exercise at improving blood glucose levels in individuals with type 2 diabetes: a randomized crossover trial
Savikj M1,2, Gabriel BM1, Alm PS1, Smith J1, Caidahl K3,4, Björnholm M3, Fritz T3,4, Krook A1, Zierath JR1,3,5, Wallberg‐Henriksson H6
1Department of Physiology and Pharmacology, Section of Integrative Physiology, Karolinska Institutet, Stockholm, Sweden; 2Faculty of Medicine, University of Oslo, Oslo, Norway; 3Department of Molecular Medicine and Surgery, Section of Integrative Physiology, Karolinska Institutet, Stockholm, Sweden; 4Department of Clinical Physiology, Karolinska University Hospital, Stockholm, Sweden; 5The Novo Nordisk Foundation Center for Basic Metabolic Research, Faculty of Health and Medical Sciences, University of Copenhagen, Copenhagen, Denmark; 6Department of Physiology and Pharmacology, Section of Integrative Physiology, Karolinska Institutet, Stockholm, Sweden
Background
Regular physical activity, including high intensity interval training (HIIT), is recommended for the prevention or treatment of type 2 diabetes (8). The time of day for exercise that best promotes glycemic control in type 2 diabetes remains unclear, however. This study aimed to determine whether HIIT exercise training at two distinct times of day would have different effects on 24‐hour blood glucose levels, as measured by CGM, in men with type 2 diabetes.
Methods
A total of 11 men with type 2 diabetes (aged 45–68 years; body mass index 23–33 kg/m2), not on insulin therapy, underwent supervised HIIT (cycling with a 7‐minute warm‐up, followed by six 1‐minute bouts at a high intensity [ranging between 180 and 350 Watts], interspersed with light cycling for 1 minute), either in the morning after a light breakfast or 3 hours after lunch. The HIIT occurred three times per week for 2 weeks for each time of two strategies (i.e., morning vs afternoon), with a 2‐week washout period between strategies.
Results
Postbreakfast HIIT resulted in a rise in CGM‐based glycemia from 6.4±0.3 mmol/L to 6.9±0.4 mmol/L (mean±SEM), while afternoon HIIT resulted in a decline in glucose level (to 6.2±0.3 mmol/L). Morning HIIT was also associated with elevations in glycemia at other points of the day and elevations in morning glycemia even on nontraining days. Afternoon HIIT was associated with elevated parathyroid hormone and thyroid‐stimulating hormone concentrations and reduced T4 concentration compared with pretraining concentrations. Thyroid stimulating hormone was also elevated after morning HIIT, whereas parathyroid hormone and thyroxine concentrations were unaltered with morning HIIT.
Conclusions
Afternoon HIIT is more efficacious than morning HIIT in improving blood glucose levels in type 2 diabetes. Unexpectedly, morning HIIT, but not afternoon HIIT, elevates glycemia at several time points throughout the day as compared with the pretraining state.
Based on this small but carefully controlled study of 11 men with type 2 diabetes who were not using insulin therapy, afternoon HIIT is more efficacious than morning HIIT at improving blood glucose levels in type 2 diabetes. Strikingly, morning HIIT had a deleterious effect on glycemia, with levels elevated throughout much of the day as compared with either the pretraining state or afternoon HIIT. These interesting observations mimic another study published this year indicating that morning (fasted) HIIT elevates glycemia throughout the day in well trained individuals with type 1 diabetes (9,10). Collectively, these studies highlight the importance of optimizing the timing of exercise when prescribing it as for diabetes.
Evaluation of an artificial pancreas with enhanced model predictive control and a glucose prediction trust index with unannounced exercise
Pinsker JE1, Laguna Sanz AJ1,2, Lee JB1,2, Church MM1, Andre C1, Lindsey LE1, Doyle FJ 3rd1,2, Dassau E1,2,3
1Department of Clinical Research, Sansum Diabetes Research Institute, Santa Barbara, CA; 2Harvard John A. Paulson School of Engineering and Applied Sciences, Harvard University, Cambridge, MA; 3Department of Research, Joslin Diabetes Center, Boston, MA
Background
The goal for newer artificial pancreas (AP) systems is to maximize the time in a targeted glucose range of 70–180 mg/dL, even in the face of large meal boluses and exercise. These authors recently demonstrated that their control to target model predictive control (MPC) algorithm is capable of achieving a high percentage of time in range when challenged with a 65‐g carbohydrate unannounced meal (11). The set point–based controller used in the prior study was then enhanced to include exponential weighting on hypoglycemic excursions (eMPC) and a trust index module that dynamically adapts future insulin delivery based on past glucose predictions (12,13). This study aimed to test this new AP system in settings of unannounced exercise and unannounced meals.
Methods
Fifteen adults with T1D were admitted for a 48‐hour supervised setting of exercise and meals using the AP system with 1) an eMPC controller, 2) a hypoglycemia Health Monitoring System, and 3) a trust index module that weights future insulin delivery based on past glucose predictions. Subjects started in closed‐loop AP mode on the first day just 2 hours before dinner and then were maintained on the AP system overnight and the next day at which point brisk walking occurred post breakfast. Following a lunch, the subjects were maintained on the AP system for an additional 29 hours with three additional meals provided (dinner, breakfast and lunch). The primary objective was time within the target glucose range of 70–180 mg/dL, as assessed by CGM, determining if this combination of eMPC and the trust index could provide safe and effective glucose control. Secondary measures included time in various degrees of hypo‐ and hyperglycemia, safety events, and failure analysis of the devices used.
Results
Percent time in target range (i.e., 70–180 mg/dL) was higher with AP use as compared with sensor augmented pump use (88.0%±8.0% vs 74.6%±9.4%). Moreover, time <70 mg/dL (1.5%±1.9% vs 7.8%±6.0%) and number of hypoglycemic events (0.6±0.6 vs 6.3±3.4), was less with AP versus sensor augmented pump (both P<0.001). On average, the trust index enhanced controller responsiveness to predicted hyper‐ and hypoglycemia by 26% (P<0.005). There were no persistent connectivity issues between device components. Percent time in closed loop was 98.62%. Although there were frequent “warning” alarms for developing hypoglycemia in the study, there were few treatments for hypoglycemia during AP use with a mean of 1.7±1.7 treatments over the 48‐hour study period.
Conclusions
In this population of well‐controlled patients with T1D, eMPC with trust index AP achieved nearly 90% time in the target glucose range. This new system was deemed effective and safe using the following information provided to the MPC controller: CGM readings, prior insulin delivery, basal profile, total daily insulin, and body weight.
This enhanced MPC with trust index achieved impressive glycemic control in this 48‐hour observational period that included several meals (two dinners, two breakfasts and two lunches, snacks, and unannounced exercise (brisk walking). What is particularly impressive is the low rate of hypoglycemia (i.e., ∼1.5% of time spent <70 mg/ dL) with the AP system even though exercise was initiated at a time when bolus insulin was in the circulation. Some limitations of this study included the rigid nature of the meals provided (all were timed accordingly and of a modest carbohydrate content) and that only one mild exercise session was evaluated. Future studies are planned to test the efficacy of this promising new AP system.
The impact on glycemic control through progressive resistance training with bioDensityTM in Chinese elderly patients with type 2 diabetes: the PReTTy2 (Progressive Resistance Training in Type 2 Diabetes) trial
Hangping Z1, Xiaona Q1, Qi Z1, Qingchun L1, Na Y1, Lijin J1, Siying L1, Shuo Z1, Xiaoming Z1, Xiaoxia L1, Qian X1, Jaimovich D2, Yiming L3, Bin L4
1Department of Endocrinology and Metabolism, Huashan Hospital, Fudan University, Shanghai, People's Republic of China; 2Performance Health Systems, Northbrook, IL; 3Department of Endocrinology and Metabolism, Huashan Hospital, Fudan University, Shanghai, People's Republic of China; 4Department of Endocrinology and Metabolism, Huashan Hospital, Fudan University, Shanghai, People's Republic of China
Background
The American Diabetes Association Standards of Medical Care recommend that patients should have an active lifestyle to improve health and longevity (14). Resistance exercise (e.g., weight training) is recommended as an important component of lifestyle intervention as it builds muscle mass and strength and improves or maintains insulin sensitivity and potentially glycemic control. However, many patients with type 2 diabetes may be unfamiliar with the proper weight training techniques or may be hesitant to adopt weight training with weights or weight machines for fear of injury. BioDensity (bD) is a new mode of progressive resistance training that may be used in some rehabilitation settings. This study aimed to evaluate the efficacy of a progressive resistance training program using bD on glycemic control and cardiovascular risk factors in older Chinese adults with type 2 diabetes.
Methods
This study enrolled 300 older adults, aged 50 to 75 years, with type 2 diabetes in a randomized resistance training program with the bD technique. Of these, 100 were control patients with no training intervention, while 200 had resistance training with bD. Anthropometry, biochemical parameters, HbA1c, and fasting plasma glucose were measured at baseline, 3‐month, and at 6‐month of treatment.
Results
A total of 265 patients completed the study with no adverse events. There were no baseline differences in HbA1c between intervention and control groups. bD‐treated patients had significant improvements at 6 months, as compared with baseline, for HDL (1.25+0.32 vs 1.17±0.26 mmol/L; P<0.001), LDL (3.23±0.89 vs 2.93±0.80 mmol/L, P<0.001), and total cholesterol (4.97±1.22 vs 4.58±1.03 mmol/L, P<0.001) levels. In a subgroup analysis, in those subjects with elevated HbA1c levels at baseline (≥ 7.5%), HbA1c levels (P=0.004) and fasting plasma glucose levels (P=0.018) were lower in the bD group compared with the control groups at 6 months of treatment.
Conclusions
Progressive resistance training for 6 months using bD improves lipid levels and reduces fasting plasma glycemia and HbA1c levels in older patients with type 2 diabetes that initially had suboptimal glycemic control.
Resistance exercise is feasible for older patients living with T2D (15). Particularly if combined with regular aerobic exercise training, resistance training can lower A1c levels (16). This large study of older Chinese patients with T2D shows that adding resistance exercise can help patients achieve their lipid and glucose control targets. Whether bD is more, or less, effective than other forms of less expensive resistance training (body weight exercises, machine weights, or free weights) is unclear.
Excellent glycemic control maintained by open‐source hybrid closed loop AndroidAPS during and after sustained physical activity
Petruzelkova L1, Soupal J2, Plasova V1, Jiranova P1, Neuman V1, Plachy L1, Pruhova S1, Sumnik Z1, Obermannova B1
1Department of Pediatrics, University Hospital Motol and 2nd Faculty of Medicine, Charles University in Prague, Prague, Czech Republic; 23rd Department of Internal Medicine, 1st Faculty of Medicine, Charles University in Prague, Prague, Czech Republic
This manuscript is also discussed in the article on Decision Support Systems and Closed Loop, page S‐47.
Background
Glycemic variability usually increases in children with T1D when physical activity is introduced. With limited availability to closed‐loop insulin delivery systems, do it yourself (DIY) artificial pancreas systems (APS) are becoming increasingly popular as a management strategy. This study compared the commercially available Medtronic SmartGuard technology system with predictive low‐glucose management (PLGM) to an open‐source hybrid closed‐loop system (AndroidAPS) in children in an extreme sports setting.
Methods
Children 6–15 years of age (n=22, 16 girls) with T1D who were intermediate or advanced skiers participated in a three whole‐day alpine skiing camp. The participants were assigned to either the SmartGuard technology system (n=12) or AndroidAPS. The systems were compared based on mean glucose level, time in hypoglycemia (<3.9 mmol/L), and time within a targeted glucose range (3.9–10.0 mmol/L) during daytime activity and overnight/recovery.
Results
Those using AndroidAPS had a lower mean glucose level (129.6±49 vs 138.6±50 mg/dL; P<0.042) compared with those using SmartGuard technology. There was no difference between the two groups when comparing time below the target (median 5.0%±2.5% vs 3.0%±2.3%; P=0.6) and in the glucose target range (63%±9.5% vs 63%±18%; P=0.5), with no severe hypoglycemia recorded in either group. Children using AndroidAPS had a nonsignificant reduction in glycemic variability (median AndroidAPS 2.3 [2.1–2.8] vs PLGM 2.6 [2.5–2.9]; P=0.09). There was no difference between amount of carbohydrates consumed to prevent or treat hypoglycemia.
Conclusions
The DIY AndroidAPS is a safe alternative to SmartGuard technology, with a similar performance in children participating in a multiday extreme sport camp.
Prolonged multiday exercise is possible in individuals with T1D but often requires additional planning for the increased activity. Quite often, in order to avoid hypoglycemic episodes, individuals consume too many supplemental carbohydrates or are too aggressive with insulin reductions, resulting in hyperglycemia (17,18) or increased glycemic variability (19). Automatic functions such as the SmartGuard technology using PLGM or APS may reduce the burden of planning for such types of activity and aid in preventing hypoglycemic episodes during exercise and in recovery periods. A downfall of the PLGM technology is the development of rebound hyperglycemia following insulin suspension (20). APS can prevent this with a subsequent rise in insulin delivery based on the gradual rise in glucose levels. Due to the limited commercially available APS, DIY systems have become increasingly popular and resulted in improved HbA1c, time in range and improved quality of life (21). This study showed the safety and efficacy of children using the DIY AndroidAPS in a prolonged exercise setting, with lower mean glucose compared with the SmartGuard technology. The bolus and basal insulin rates were reduced by 30% to 50% during activity, which further protected against hypoglycemia, and hyperglycemia was immediately corrected with a bolus of insulin. These preset strategies likely contributed to the high time in target range throughout the study for both systems. The researchers speculate that AndroidAPS may have had additional benefits if the users had not experienced frequent cannula malfunctions, as was observed in this ski camp setting. Overall, the AndroidAPS proves to be a feasible alternative but requires further research to increase the awareness and safety of these commonly used DIY systems. How to study these nonregulated or approved artificial pancreas devices remains an ethical challenge in academia.
Improved open‐loop glucose control with basal insulin reduction 90 minutes before aerobic exercise in patients with type 1 diabetes on continuous subcutaneous insulin infusion
Zaharieva DP1, McGaugh S1, Pooni R1, Vienneau T2, Ly T3, Riddell MC1,4
1Muscle Health Research Centre, School of Kinesiology and Health Science, Faculty of Health, York University, Toronto, Ontario, Canada; 2Insulet Canada Corporation, Oakville, Ontario, Canada; 3Insulet Corporation, Billerica, MA; 4LMC Diabetes and Endocrinology, Toronto, Ontario, Canada
Background
A considerable challenge for a person with diabetes is completing moderate intensity exercise with optimal glycemic control. Use of CSII offers the advantage of being able to manipulate basal insulin levels in anticipation of a planned activity or exercise. Although it is increasingly recognized that sufficient time must be allowed for reduction in basal rates to have clinical impact, there is a gap in the literature about how much and how early basal rate reduction (BRR) should occur.
Methods
To address this gap in knowledge, 17 young adults on CSII were recruited to an in‐clinic randomized crossover trial designed to compare the effect of three different BRRs on glucose levels during afternoon exercise: cessation at onset of exercise, 80% reduction 90 minutes prior to onset of exercise, and 50% reduction 90 minutes prior to onset of exercise. Sixty minutes of moderate intensity exercise was completed on each occasion. Monitoring continued during a postexercise meal and overnight. Outcome measures were CGM glucose levels, insulin levels, and episodes of hypoglycemia requiring treatment.
Results
All 17 participants completed each of the three study scenarios. Eighty percent BRR resulted in the smallest drop in blood glucose level by end of exercise (−31mg/dL vs −47mg/dL vs −67mg/dL, 80% vs 50% vs pump suspension, all P<0.05). Pump suspension at the onset of exercise resulted in more episodes of hypoglycemia during exercise than did either 50% or 80% BRR 90 minutes prior to exercise (7 vs 1 vs 1; P<0.05). Insulin levels were higher with pump suspension than with 50% or 80% reduction (P<0.05) but were similar by the postexercise period.
Conclusions
Overall, a 50%–80% BRR applied 90 minutes preexercise did not cause preexercise hyperglycemia and attenuated the glucose drop during moderate intensity exercise. In comparison with pump suspension at onset of exercise, there were fewer episodes of hypoglycemia without compromising the postexercise meal or overnight glucose control.
This study adds to the growing evidence to assist persons with diabetes using CSII with exercise. Reducing basal rate insulin 90 minutes prior to an hour of moderate‐intensity exercise by either 50% or 80% results in less hypoglycemia than pump suspension at the onset of exercise. It is reassuring that in all three scenarios the postexercise meal and overnight glycemic control following afternoon exercise was similar. The difference between 80% and 50% BRR was small and not enough to systematically recommend one approach over the other. The authors report an observation common to many exercise studies; that is, considerable interindividual variation and individualization may be needed when adopting BRR for planned exercise.
Continuous glucose monitoring and insulin informed advisory system with automated titration and dosing of insulin reduces glucose variability in type 1 diabetes mellitus
Breton MD, Patek SD, Lv D, Schertz E, Robic J, Pinnata J, Kollar L, Barnett C, Wakeman C, Oliveri M, Fabris C, Chernavvsky D, Kovatchev BP, Anderson SM
Center for Diabetes Technology, University of Virginia, Charlottesville, VA
Background
Newer advisory systems for artificial pancreas control of diabetes should attempt to maintain overall glucose control, as measured by time in range using continuous glucose monitoring (CGM), while reducing glucose variability (GV). Increased GV may be linked with increased diabetes‐related complications including hypoglycemia and cardiovascular disease. This research tested the efficacy of the University of Virginia Decision Support System (UVA‐DSS) that consists of engineering logic strategies for automated insulin titration, bolus insulin calculation, and carbohydrate treatment advice.
Methods
Twenty‐four participants with T1D who were either on insulin pump therapy (n=16) or multiple daily insulin injections (n=8) completed two randomized, crossover 48‐hour in‐clinic visits wearing a blinded Dexcom G4 CGM and using either usual care or the UVA‐DSS. During each visit, subjects were exposed to a variety of meal sizes with varying macronutrient content and two 45‐minute bouts of moderate‐intensity exercise performed 3 hours and 2 hours after breakfast on days 1 and 2, respectively. GV and glucose control were assessed using the blinded CGM.
Results
Over the course of the protocol, participants collected an average of 37.9±14.0 days of CGM data, associated with 358±242 blood glucose measurements, while consuming on average 211±182 meals or snacks and injecting 258±197 units of bolus insulin. With UVA‐DSS, GV was reduced from 36±8 % (with standard care) to 33±6% (P=0.045), with much of the observed benefit seen during the daytime because of reduced low blood glucose excursions. With respect to the exercise periods, pre‐ and postexercise glucose concentrations were not improved with UVA‐DSS, but the number of hypoglycemia events was significantly reduced (from 4 events to 1 event below 70 mg/dL) compared with usual care. Moreover, the amount of carbohydrate intake was less with the UVA‐DSS when compared with usual care on both days (6.3±9.5 g vs 4.8±10.2 g on day 1 and 9.4±13.4 g vs 3.3±7.7 g on day 2; P=0.011), likely because insulin delivery during exercise was significantly reduced with the UVA‐DSS (1.11±0.67 U vs 0.85±0.63 U on day 1 and 0.96±0.66 U vs 0.63±0.57 U on day 2; P=0.026) and preexercise carbohydrate intake was higher (1.2±5.6 g vs 4.1±13.0 g on day 1 and 0.6±2.9 g vs 11.6–17.9 g on day 2; P=0.003).
Conclusions
Overall the CGM‐based UVA‐DSS was shown to be safe and feasible in a limited, controlled, crossover clinical trial of adults with type 1 diabetes undergoing various challenges of exercise and meals. Moreover, the system significantly reduced GV, likely through the reduction of exposure to hypoglycemia, without increasing average glycemia or exposure to hyperglycemia.
Decision support around exercise and meal sizes remains a major challenge for patients with T1D on pump or insulin injections. This novel work showed that the UVA‐DSS can reduce glycemic variation and reduce the time spend below and above a targeted glucose range. While the system still required that participants consume carbohydrate snacks for exercise, which can be frustrating for those wanting to lose weight, the use of the UVA‐DSS did reduce the amount of carbohydrates required because it also reduced the insulin delivered around the time of exercise. Having the ability to use the UVA‐DSS in the home rather than just in the clinic setting in future studies should help further reduce patient burden around exercise and meal challenges.
Footnotes
Author Disclosure Statement
No competing financial interests exist.