Background: Exercise challenges closed loop (CL) therapy for type 1 diabetes (T1D).

Aim: To compare CL glucose homeostasis in adults with T1D undertaking high-intensity interval anaerobic exercise (AnE) versus moderate-intensity aerobic exercise (AeE).

Method: Two-stage randomised crossover study of 12 adults (9 women, 3 men) with T1D. AnE and AeE stages involved 45-min bicycle exercise of equal caloric requirement, with CL (proportional-integral-derivative with insulin feedback algorithm, Medtronic) activated 180-min pre-exercise. The CL glucose target was increased 120-min pre-exercise until 15-min post-exercise. Venous blood was collected for glucose, ketones, lactate, and insulin hourly for the first 120-min post-CL activation, then at 15-min intervals until 120-min post-exercise.

Results: Plasma glucose (mean±SE) was similar pre-exercise in AnE and AeE (9.9±0.6 vs. 9.9±0.7mmol/L; p=0.39). During exercise glucose increased progressively with AnE, and reduced with AeE (p10mmol/L AUC was greater for AnE than AeE (247.8±333.5 vs. 55.9±128.2mmol.min/L; p=0.03; figure). There was no sensor glucose <4mmol/L. During AnE, 3 participants had glucose >14mmol/L with ketones >0.6mmol/L. In AnE lactate increased from 15-min of exercise (4.6 ± 0.6; p<0.001) until 60-min post-exercise (1.1 ± 0.5; p=0.030). There were no statistically significant differences in insulin or ketone levels between AnE and AeE stages. In AnE, ketones rose post-exercise from baseline pre-exercise levels (p<0.01).

Conclusions: AnE and AeE impact plasma glucose differently during CL. CL minimised hypoglycaemia during both AnE and AeE, and limited AnE hyperglycaemic excursions. Nonetheless during AnE mean glucose increased, and ketosis occurred in 25% of participants. These observations may relate to counter-regulatory hormone action. Maintaining euglycaemia during exercise remains challenging, even with CL. Tailored AnE and AeE algorithms may improve CL performance.