Glycaemic thresholds for counterregulatory hormone and symptom responses to hypoglycaemia in people with and without type 1 diabetes: a systematic review

Aim/hypothesis The physiological counterregulatory response to hypoglycaemia is reported to be organised hierarchically, with hormone responses usually preceding symptomatic awareness and autonomic responses preceding neuroglycopenic responses. To compare thresholds for activation of these responses more accurately between people with or without type 1 diabetes, we performed a systematic review on stepped hyperinsulinaemic–hypoglycaemic glucose clamps. Methods A literature search in PubMed and EMBASE was conducted. We included articles published between 1980 and 2018 involving hyperinsulinaemic stepped hypoglycaemic glucose clamps among people with or without type 1 diabetes. Key exclusion criteria were as follows: data were previously published; other patient population; a clamp not the primary intervention; and an inadequate clamp description. Glycaemic thresholds for counterregulatory hormone and/or symptom responses to hypoglycaemia were estimated and compared using generalised logrank test for interval-censored data, where the intervals were either extracted directly or calculated from the data provided by the study. A glycaemic threshold was defined as the glucose level at which the response exceeded the 95% CI of the mean baseline measurement or euglycaemic control clamp. Because of the use of interval-censored data, we described thresholds using median and IQR. Results A total of 63 articles were included, whereof 37 papers included participants with type 1 diabetes ( n =559; 67.4% male sex, aged 32.7±10.2 years, BMI 23.8±1.4 kg/m 2 ) and 51 papers included participants without diabetes ( n =733; 72.4% male sex, aged 31.1±9.2 years, BMI 23.6±1.1 kg/m 2 ). Compared with non-diabetic control individuals, in people with type 1 diabetes, the median (IQR) glycaemic thresholds for adrenaline (3.8 [3.2–4.2] vs 3.4 [2.8–3.9 mmol/l]), noradrenaline (3.2 [3.2–3.7] vs 3.0 [2.8–3.1] mmol/l), cortisol (3.5 [3.2–4.2]) vs 2.8 [2.8–3.4] mmol/l) and growth hormone (3.8 [3.3–3.8] vs. 3.2 [3.0–3.3] mmol/l) all occurred at lower glucose levels in people with diabetes than in those without diabetes (all p ≤0.01). Similarly, although both autonomic (median [IQR] 3.4 [3.4–3.4] vs 3.0 [2.8–3.4] mmol/l) and neuroglycopenic (median [IQR] 3.4 [2.8–N/A] vs 3.0 [3.0–3.1] mmol/l) symptom responses were elicited at lower glucose levels in people with type 1 diabetes, the thresholds for autonomic and neuroglycopenic symptoms did not differ for each individual subgroup. Conclusion