Severe hypoglycaemia still represents a significant risk for many patients with type 1 diabetes. The condition is feared by patients [1
] and relatives [3
] and remains a major barrier in achieving recommended glycaemic targets [4
]. Moreover, severe hypoglycaemia may, although rarely, lead to permanent brain damage or death [5
]. The frequency of severe hypoglycaemia in type 1 diabetic populations is 1–1.7 episodes per patient-year [6
]. The distribution of episodes is, however, very skewed. Thus, 20% of patients with type 1 diabetes experience two or more episodes of severe hypoglycaemia per year [15
]. Once prone to recurrent severe hypoglycaemia, patients often remain so throughout their life.
Insulin analogues were developed to improve treatment of insulin-treated diabetes with respect to glycaemic control and avoidance of hypoglycaemic episodes. Short-acting insulin analogues (aspart, lispro and glulisine) were designed to mimic the fast physiological postprandial insulin release and long-acting insulin analogues (detemir and glargine) were designed to mimic the basal continuous insulin release with minimal peak action, thereby leading to a presumed decreased risk of hypoglycaemia. The effect of insulin analogues on glycaemic control has been documented in large trials [16
]. However, for several reasons the impact of insulin analogues on the frequency of severe hypoglycaemia is less clear. Firstly, most studies comparing the effects of human insulin with those of insulin analogues specifically exclude patients at risk of severe hypoglycaemia, such as patients with previous severe events or impaired hypoglycaemia awareness, from participation [18
]. This renders the trials insufficiently statistically powered to detect differences in the frequency of severe hypoglycaemia and the conclusions from these studies can not readily be extrapolated to patient groups prone to severe hypoglycaemia. Secondly, hypoglycaemia is often less well defined and detected, at best as a secondary endpoint, with reports including a mixture of mild, severe, and asymptomatic hypoglycaemia. Thirdly, severe hypoglycaemia should be analysed according to a statistic model that takes into account the much skewed distribution of episodes e.g. negative binomial or zero-inflated models [22
A Cochrane review comparing short-acting insulin analogues with human insulin included 49 studies, of which 10 studies were eligible in relation to hypoglycaemia [16
]. The analysis showed that the overall mean occurrence of hypoglycaemic episodes was lower for short-acting analogues in comparison to regular insulin in patients with type 1 diabetes (−0.2 per patient per month (95% confidence interval: -1.1 to 0.7)). No statistical comparative calculations were possible specifically addressing severe hypoglycaemia due to inconsistent and bias-prone definitions in the different studies. However, in the analogue group the incidence of severe hypoglycaemia ranged from 0–247 (median 22) episodes per 100 patient-year compared to the incidence in the human insulin group ranging from 0–544 (median 46).
After the initiation of the present study, a meta-analysis based on 20 studies (duration
weeks) was published comparing long-acting insulin analogues with long-acting human insulin in type 1 diabetes, including 12, 15 and 13 studies analysing any hypoglycaemia, severe hypoglycaemia, and nocturnal hypoglycaemia, respectively [17
]. The analysis showed that long-acting analogues are not associated with a significant reduction of overall risk of any hypoglycaemia in comparison with NPH insulin. When episodes of severe hypoglycaemia and nocturnal hypoglycaemia were separately analysed, insulin detemir, but not insulin glargine, was associated with a reduced risk compared to human NPH insulin (OR 0.73 (0.60 - 0.89), p
0.002 and OR 0.69 (0.55 - 0.86), p
0.001; respectively). There are important limitations with the meta-analysis. Firstly, the trials use different criteria for hypoglycaemia, different methods in recording severe hypoglycaemia, and many trials are not randomised. Secondly, the participants in the studies are relatively young (mean: 34
years) with a short duration of diabetes (mean: 13
years) suggesting that the participants’ state of hypoglycaemia awareness is relatively unaffected. Thirdly, in some trials the number of insulin injections per day are different in the human insulin group and the insulin analogue group. Finally, the vast majority of the trials in the analysis were sponsored by manufacturers of long-acting insulin analogues. To our knowledge a direct comparison of long acting insulin analogues and long-acting human insulin has not been made in patients at high risk of severe hypoglycaemia.
When it comes to patients at high risk of severe hypoglycaemia, who may have the greatest potential benefit from insulin analogue therapy, there is only one small, randomised,12-month cross-over study comparing the effect of the short-acting insulin analogue (lispro) and regular human insulin on the frequency of severe hypoglycaemia [23
]. The study population comprised 33 patients with type 1 diabetes and hypoglycaemia unawareness. There was a trend towards a lower incidence of severe hypoglycaemia during treatment with insulin lispro in comparison with human insulin (p
0.087). Another study in 322 women with type 1 diabetes, who were pregnant or planning pregnancy and therefore being at relatively high risk of severe hypoglycaemia due to very tight glycaemic control, compared insulin aspart with human insulin as meal-time insulin (with human NPH insulin as basal insulin) using an open-label, randomised, parallel-group, multicentre design. Although the risk of severe hypoglycaemia in women treated with insulin aspart was 28% lower, the difference did not reach statistical significance [24
]. In a randomised 24-week pilot study of 15 patients with type 1 diabetes complicated by severe hypoglycaemia, comparing rigorous hypoglycaemia avoidance with insulin analogue therapy (lispro and glargine), continuous subcutaneous insulin therapy (CSII) or education alone (but not including any control group), hypoglycaemia awareness was restored and further severe hypoglycaemia was prevented with concomitant improvement in glycaemic control in the analogue and CSII groups [25
The HypoAna Trial is designed to elucidate whether short-acting and long-acting insulin analogues (insulin aspart and insulin detemir) in comparison with human regular and NPH insulin are superior with respect to reducing the occurrence of severe hypoglycaemic episodes in patients with recurrent hypoglycaemia. There is an urgent need for this evidence in clinical practice and a need to elucidate if the higher costs of insulin analogues are justified in this respect. However, since the structures of different insulin analogues are not similar, the results of this study cannot be directly extrapolated to other insulin analogues (lispro, glulisine and glargine).