The first insulin preparation capable of consistently lowering blood glucose was developed in 1921. But 100 years later, blood glucose control with insulin in people with diabetes is nearly universally suboptimal, with essentially the same molecule still delivered by the same inappropriate subcutaneous injection route. Bypassing this route with oral administration appears to have become technologically feasible, accelerating over the past 50 years, either with packaged insulin peptides or by chemical insulin mimetics. Some of the problems of prospective unregulated absorption of insulin into the circulation from subcutaneous depots might be overcome with glucose-responsive insulins. Approaches to these problems could be modification of the peptide by adducts, or the use of nanoparticles or insulin patches, which deliver insulin according to glucose concentration. Some attention has been paid to targeting insulin preferentially to different organs, either by molecular engineering of insulin, or with adducts. But all these approaches still have problems in even beginning to match the responsiveness of physiological insulin delivery to metabolic requirements, both prandially and basally. As would be expected, for all these technically complex approaches, many examples of abandoned development can be found. Meanwhile, it is becoming possible to change the duration of action of subcutaneous injected insulin analogues to act even more rapidly for meals, and towards weekly insulin for basal administration. The state of the art of all these approaches, and the barriers to success, are reviewed here.
