Long-acting insulin analogues
Just as the short-acting insulin analogues were developed to overcome the perceived flaws of subcutaneously injected regular insulin, so were the long-acting insulin analogues developed to overcome perceived flaws in the basal NPH insulin. While the absorption of NPH insulin was already less variable than that of the prolonged acting Zinc-insulins, the day-to-day variability in absorption of NPH insulin was still considered too high. Also, the duration of action was considered too short and the peak-action too high to meet the qualifications of a 'true basal insulin'.
The feeling that NPH insulin might not be adequate as a basal insulin was further enforced when the short-acting insulin analogues proved unable to significantly improve HbA1c levels. This lack of effect of the short-acting insulin analogues was attributed to a perceived basal hypo-insulinaemia between meals, more specifically in the late afternoon when the effect of the NPH insulin injected the previous evening was thought to have ended. As stated, the goal became to develop insulins that had at least a 24-hr duration of action, little or no insulin peak and a low intra-individual variability in absorption. Additionally, it was reasoned that such a basal analogue should ideally also be hepatospecific in its action, since basal insulinaemia mainly functions to suppress hepatic glucose output.
Methods to prolong insulin action
While the goal of hepatospecificity proved to be elusive[a], the companies succeeded in prolonging insulin action by modifications to the insulin molecule. Thusfar, two approaches have proven to be useful.
- Changing the iso-electric point of the insulin molecule. Since insulin is an acidic protein, at normal pH it is in the ionized (charged) state and hence water soluble. By changing or adding some amino-acids, the iso-electric point (i.e. the pH at which the molecule is least ionized) of the insulin can be shifted from pH= 4 to pH =7. The resulting insulin is less water soluble at pH = 7, and hence easily self-aggregates and precipitates to form microcrystals, which act as an insulin depot preparation. This principle was used for insulin glargine (Figure 1).
- Attaching a fatty-acid tail to the insulin molecule. By attaching a fatty-acid tail of varying length to the
Figure 2. Molecular structure for insulin analogues detemir and degludec compared to human insulin. In both analogues, the insulin is truncated at position B29 and a fatty-acid side chain is attached.C-terminal end of the beta-chain of the insulin molecule, the insulin can be bound by macromolecules, particularly albumin. Since this is a relatively high affinity bond, most of the insulin molecules in plasma and tissue fluids will be bound and only a few will be present as free, active, insulin. This results in a depot of albumin-bound insulin. This principle was used for insulin detemir and insulin degludec (Figure 2).
Efficacy of long-acting analogues
While the long-acting (basal) insulin analogues, particularly glargine and degludec, have a longer duration of action than NPH insulin following s.c. injection, this has not translated into major clinical improvements in glycaemic control. Relatively few studies have been performed regarding their efficacy compared to NPH insulin in type 1 diabetes (most with detemir) and the overall picture shows a reduction in HbA1c of about 0.15% (1.5mmol/mol) and a 15% reduction in hypoglycaemia at best. In type 2 diabetes considerably more data are available, but again a Cochrane meta-analysis found only a modest benefit compared to NPH insulin with a 16-18% reduction in hypoglycaemia and no change in HbA1c (and possibly even a deterioration in HbA1c with insulin detemir). Moreover, even while company-sponsored meta-analyses found benefits of up to 35% reduction in hypoglycaemia in type 2 diabetes, it should be borne in mind that the typical hypoglycaemia event-rate for patients starting once-daily basal insulin is about 1-2 episodes per patientyear, so the relevance of differences found should not be overestimated anyway.
^ Palmer et al. Curr Med Res Opin 2004;1729
^ This should not have come as a surprise since physiological "hepatospecificity" of insulin is mainly the result of the high 'first-pass' effect: insulin produced by the pancreas will first pass the liver were 50% is extracted. The liver does not have an insulin receptor that is intrinsically different from the insulin receptors in other tissues.