Insulin Formulations – a Review

European Review for Medical and Pharmacological Sciences 2001; 5: 73-83 Insulin formulations – a review

A.M. GUALANDI-SIGNORINI, G. GIORGI

Department of Pharmacology “G. Segre”, University of Siena - Siena (Italy)

Abstract. – Although the improvement Introduction on insulin therapy since it was first conceived, it is still far from mimicking physiological se- In the history of insulin therapy some mile- cretion of pancreatic b-cells and research to find new insulin formulations and new routes stones are present. In 1921, the young physician of administration continues. Human biosynthet- Frederick Banting (1881-1941) and the forth ic insulin (rapid-acting, intermediate-acting and year medical student Charles Best (1899-1978), long-acting), produced by recombinant DNA working in the laboratory of Prof. McLeod in technique, is currently available. The pharma- the Toronto University, found the final link in a cokinetic profile of rapid-acting insulin (regu- series of studies begun in 1916 by other re- lar) does not adequately reproduce the physio- searchers that guessed the pancreas secretetion logical post-prandial insulin response. This has led to the development of molecular analogues of substance, already named “insulin”, capable with slight modifications that prevent the spon- to decrease blood glucose concentrations. The taneous polymerisation underlying delayed ab- young researchers isolated this pancreas extract sorption. Fast-acting analogues such as that “cured” hyperglycemia in diabetic dogs1-2 Lyspro and Aspart can be injected immediately and in 1922 they successfully administered it before the meal, inducing a very fast and sub- for the first time to a 14-year-old diabetic pa- stantial peak of insulin, similar to that pro- 3 duced by b-cells, but have the disadvantage of tient, Leonard Thompson . short duration of action. For this reason, and In 1923, the company Lilly began marketing because of the difficulty of obtaining sufficient animal insulin. In 1928, the hormone was iden- basal insulin concentrations to control pre- tified to be a protein. Since the insulin of prandial blood glucose levels with current Banting and Best could not function for more long-acting insulins, analogues known as than 6 h, much research went into finding ways Glargine and Detemir have been synthesized. of prolonging action. In 1936, Hagedorn noted They have virtually no plasma peak and acts for about 24 h. These characteristics make it that addition of a basic protein, such as prota- ideal to cover basal insulin requirement. With mine, to the insulin preparation, kept the hor- insulin analogues, it also seems possible to mone in suspension at the injection site, delay- overcome the problem of intra- and inter-indi- ing absorption and prolonging its action4. In vidual variability in absorption after subcuta- 1946, the first insulin “Isophane NPH” neous injection. This variability is directly pro- (Neutral Protamine Hagedorn), obtained com- portional to the duration of insulin action. Research into new routes of administration has bining insulin and protamine in stoichiometric led to production of inhaled insulin powder, quantities (hence the term isophane from the soon to become commercially available. Insulin Greek equal and manifest) at neutral pH, was is absorbed through the lung alveoli. Trials to marketed5. In 1952, the first Lente insulin, re- evaluate efficacy and toleration have shown tarded with zinc and without protamine, was that inhaled insulin has a similar kinetic profile produced in Denmark6. In 1955, Sanger deter- to the fast-acting injected analogue and can mined the exact formula of bovine insulin7. The therefore be used for mealtime requirement, combined with a single daily injection of long- sixties saw the development of a radioim- 8 acting insulin. Oral insulin is currently being munoassay for insulin . In the seventies, pro- studied in type 1 diabetes prevention with duction of mono-component insulins, obtained promising results. by ion exchange chromatography and of mono- Key Words: peak ones, separated by Sephadex G50 col- umn, began. High performance chromatogra- Human insulin, Lyspro, Aspart, Glargine, Inhaled insulin, Oral insulin. phy made it possible to isolate an insulin 99% pure from proinsulin and other islet hormones9.

73 A.M. Gualandi-Signorini, G. Giorgi

Human Insulin produced by another technique. The procedure exploits the yeast Saccharomyces cerevisi- Human insulin is a globular protein with a ae as a living laboratory by inserting purified molecular weight of about 5,800 kd, consist- DNA that dictates the synthesis of a single ing of 51 aminoacid residues organised in two chain precursor different from proinsulin in polypeptide chains (A and B), linked by two one of the yeast’s plasmids. The protein, pro- disulphide bonds. Chain A consists of 21 duced by yeast continuously during fermenta- residues with an extra disulphide bond be- tion, is harvested from the culture medium, tween A6 and A11; chain B consists of 30 isolated by centrifugation and crystallisation, aminoacids. Complete synthesis of the human converted to insulin ester by trypsin transpep- insulin molecule was achieved in 196610. tidisation and finally hydrolysed. This method Insulin exists as a monomer only at low con- guarantees high purity12,19-20. centrations while it shows propensity to aggre- Human insulins currently on the market in- gate into stable dimers at higher concentra- clude rapid-acting (Regular), intermediate- tions, in aqueous solution at pH 2-8 and into acting (NPH and Lente) and long-acting hexamers in the presence of zinc ions. The (Ultralente) formulations. The former is a hexamer, in which chain A constitutes much of clear, colourless, aqueous solution buffered at the polar surface, is almost spherical in struc- neutral pH (7-7.8). Meta-cresol is added as ture, with a diameter of 5 nm and a height of preservative, glycerol as tonic stabiliser, as 3.5 nm. Polymerisation of the hormone has well as zinc chloride. Hexamers, made stable major pharmacological implications. by zinc ions, are the predominant quaternary Before the eighties, insulin was extracted structure of pharmacological insulin; other from bovine or pig pancreas. Commercial structures include dimers and tetramers21. production of human insulin began in 1982. The molecule tends to aggregate in the vial This milestone in the history of diabetes is of and in the tissue where it is injected. The similar scientific relevance to the discovery of hexamers need to be split for the insulin ab- insulin itself. Once it became clear that it was sorption from the subcutaneous injection site. impossible to obtain sufficient human insulin For this reason, Regular insulin enters the for world requirements from cadavers, the general circulation after a lag phase about 30 techniques for synthesising insulin were de- min after injection. It achieves the plasmatic veloped. Semisynthetic insulin was obtained peak at 2 to 4 h and lasts for 6 h. by an enzymatic method, in which trypsin Absorption of NPH is delayed by prota- catalyses the substitution of alanine, in posi- mine, a protein extracted from the nucleus of tion B30 of pig insulin, with threonine. The fish sperm, where its role is to stabilise DNA. insulin thus produced has the same The commercial form is Isophane-NPH in- aminoacid sequence that the human one and sulin, a white suspension of orthorhombic it was completely free of other pancreatic and crystals containing 0.9 molecules of prota- gastroenteric hormones11-12. mine and two atoms of zinc per hexamer5. In Genetic engineering has enabled industrial the crystal, protamine regulates interactions synthesis of human insulin by the technique of between dimers and hexamers. The vehicle is recombinant DNA. Fragments of DNA cod- water buffered at pH 6.9-7.5. Phenol or meta- ing for chains A and B are inserted separately cresol are added as preservatives21. Insulin in plasmids and then in a special strain of crystals are insoluble in water and tend to Escherichia coli which acts as a microlaborato- precipitate to the bottom of the vial, which ry for in vitro synthesis. The insulin chains re- has to be tipped various times to resuspend leased into the medium are harvested (togeth- them before use22. NPH is longer-acting: in er with other bacterial proteins) and reassem- fact its blood absorption begins 1.5 h after bled with the appropriate disulphide bonds13- subcutaneous injection; it has a peak plasma 15. To avoid contamination during the various concentration at 4 to 12 h and disappears steps of the procedure, plasmids were pro- within 24 h. The “tail”, however, is relatively duced with genetic material coding for the ineffective. Although NPH has the most reg- proinsulin molecule, which is isolated from the ular absorption of all intermediate- and long- culture medium and deprived of its connecting acting insulins, inter- and intra-individual peptide16-18. Biosynthetic insulin can be also variability is high.

74 Insulin formulations – a review

Lente, delayed without addition of prota- molecules with minimal modifications and mine, is obtained by precipitating the hor- appropriate kinetics. In 1996, an analogue in mone in the presence of zinc salts. When its which the aminoacids in positions B28 and 29 molar ratio with respect to insulin is greater were inverted, was marketed24-26. The inver- than one, the zinc ion reduces the solubility sion conferred the property of prompt ab- of insulin in neutral solvent6. When an acid sorption and hence fast action. The analogue solution of insulin is brought to pH 7.4 with is known as Lyspro, from lysine and proline, an excess of zinc ions, the resulting precipi- the aminoacids transposed. The handling of tate is amorphous and has moderately de- the molecule prevents its natural tendency to layed absorption after subcutaneous injec- hexamers aggregation, since position B28 is tion. This kind of insulin formulation is crucial for its spatial configuration27. Because known as Semilente, and was produced until the interaction between B28 and B23, two several years ago. If insulin is allowed to crys- molecules of insulin align antiparallel28 form- tallise at pH 5.5 before zinc is added and pH ing a non polar dimer which in turn aggre- corrected, the obtained formulation is known gates into hexamers by linking to zinc in as Ultralente. This insulin consists of crystals pharmacological preparations29. Since modifi- insoluble in water which remain in suspen- cation of aminoacid residue B28 counteracts sion and are absorbed very slowly. A 3:7 mix- polymerisation of insulin, the analogue may ture of amorphous and crystalline insulin, be injected immediately before the meal, as it known as Lente, has an intermediate absorp- acts in a few minutes. Lyspro maintains a tion profile. In commercial preparations, plateau for 1 to 2 h and disappears by 3 to 4 methyl-parahydroxybenzoate is used as h. Need for availability of insulin analogue preservative and the suspension is buffered at was dictated by three pharmacokinetic de- pH 7-7.821. It appears in the circulation 2.5 h fects typical of Regular insulin: after injection, has the peak plasma concentration at 7 to 15 h and disappears by 20 h. 1. Its relatively slow absorption causes Both NPH and Lente insulins can be mixed lacking reproduction of physiological with Regular in the syringe. However, while profile of the b-pancreatic secretion. NPH and Regular conserve their pharmaco- 2. The comparison of the insulin curve in a kinetic characteristics, Lente/Regular mixture normal subject to that in a type 1 diabet- changes in time because the excess of zinc ic patient (injected with Regular insulin) ions binds part of the rapid-acting insulin, shows the following figures: (a) the ab- transforming it into a form similar to sence of the first peak, (b) the flat over- Semilente. all profile in the first two hours after the Long-acting insulin (Ultralente) is an aque- meal (leading to early post-prandial hy- ous suspension of zinc-insulin crystals at neu- perglycemia) and (c) the persisting tral pH, of milky appearance21. It has an on- raised insulin plasma levels during post- set of action after 4 h, achieves a slight but absorptive period. undesirable peak plasma concentration at 7 h 3. This pattern may lead to pre-prandial and sustains blood insulin levels approxi- hypoglycemic events and conditions the mately 8 to 20 h; it inconsistently mimics en- prescription of snacks for type 1 diabetic dogenous basal secretion. Moreover its ab- patients. sorption is completely irregular23 and it cannot be mixed with Regular insulin in the sy- A lot of Italian, European and American ringe because the action of the latter would trials carried out after experimentation and be excessively delayed. consolidated use of insulin Lyspro30-37 indicate that the post-prandial insulin profile after the analogue administration is very close to the physiological one. Comparison with Insulin Analogues the insulin profile after an injection of Regular insulin (even when the patient waits Since the nineteen-eighties, pharmacologi- the prescribed 30 min before eating) con- cal research has been concentrated on ana- firms the higher validity of Lyspro to mimic logues of human insulin, seeking to develop physiological glucose-stimulated insulin se-

75 A.M. Gualandi-Signorini, G. Giorgi cretion. In clinical practice, this phenomenon In 2000, another analogue of human in- should lead to more physiologic glycemia 90- sulin, Aspart, became available. This ana- 120 min after the meal. However, while the logue is obtained by substitution of the pro- short duration of action allows for better line residue in position B28 with an aspartic control over post-prandial glucose levels, it acid residue38. This substitution eliminates does not satisfy, alone, the basal insulin re- the interaction with glicine B23 that triggers quirement in diabetic patients with low en- polymerisation. Monomer status is also aided dogenous insulin reserve (type 1). In order to by repulsion between the aspartic acid use the fast-acting analogue in the daily ther- residue and the nearby glutamic acid B21, apeutic protocol of the type 1 diabetic sub- both positively charged. Since the variation ject, a combination of Lyspro and intermedi- introduced in Aspart is not close to the recep- ate-acting insulin is therefore necessary. On tor site, there is no change in receptor bind- the other hand, this imposes the use of the ing affinity39 or in the rate of intracellular dis- syringe to mix insulins or the practice of a si- sociation40-41 with respect to native insulin. multaneous double injection by the “pen”, This manipulation has also solved the prob- which is the simplest and more pleasant actu- lem of affinity for IGF-1 receptors. The delay al method for self-administration of insulin. in making Aspart available was due to the in- Until Lyspro becomes available in “penfills” tense mitogenic activity (due to excessive with a premixed preparation of the analogue affinity for the IGF-1 receptor) of the origi- (25, 50 and 75%) and intermediate-acting in- nally proposed analogue with aspartic acid in sulin (neutral protamine Lyspro, NPL, 75, 50 B10 position39. and 25%) (expected soon), it cannot com- There are many advantages of therapy us- pletely replace Regular insulin, but it can on- ing fast-acting insulin analogues. The need to ly coexist for a more a versatile use. For ex- limit post-prandial hyperglycemia, both in ample, it can be administered to type 2 dia- type 1 and 2 diabetics, emerges from recent betic patients requiring insulin due to sec- studies with large cohorts, the results of ondary failure of oral hypoglycemic agents. which unequivocally correlate poor metabolic In this case it can be used at mealtimes, in as- control with the development and progres- sociation to intermediate- or long-acting in- sion of complications. Specifically, post-pran- sulin at bed-time, if necessary. The pharma- dial hyperglycemia seems a major factor for cokinetic of Lyspro suggests that type 2 dia- the cardiovascular manifestations so frequent betics, characterised by lacking first peak of in type 2 diabetics42-44. This is the rationale for insulin secretion, can mostly benefit from the another indication of insulin therapy in type 2 analogue. It may also be successfully used in patients, using analogues. Restoration of the those patients with iatrogenic diabetes due to early insulin peak seems to be particularly steroids, in whom hyperglycemia occurs se- important. In fact, correcting hyperglycemia lectively after meals, or in those patients 90-120 min after the meal, it is possible to re- with diabetes associated to liver disease hav- duce late hyperinsulinemia and, as a conse- ing the same feature. It may be useful in type quence, the insulin resistance highly related 1 diabetic subjects, educated to manage to the development of cardiovascular compli- “free” variations in diet, for spot correction cations45-46. Pre-prandial and nocturnal hypo- of occasionally glucose peaks, detected dur- glycemia can also be removed by fast-acting ing self-monitoring (well illustrated in the insulin37. Finally, the possibility of injecting “pizza, coca-cola and tiramisù” study)36. the hormone immediately before the meal, Another application is in the first period af- without having to wait 30 min, greatly im- ter diagnosis, when a residual basal insulin proves the quality of life, especially for young secretion is consistent. An interesting indica- patients who are dependent on injections tion is provided by patients who have an un- throughout their life. However, type 1 diabet- certain food intake (intercurrent pathologies, ic subjects still require simultaneous adminis- gastroparesis diabeticorum…) In these sub- tration of intermediate- or long-acting in- jects, Lyspro can be injected immediately af- sulin. ter the meal at a dose calculated on the basis The pharmaceutical formulation of fast- of the quantity of food eaten, without deteri- acting insulin is the same as that of regular in- orating post-prandial glucose levels. sulin.

76 Insulin formulations – a review

Long-Acting Analogues is produced from human DNA recombinant insulin, modified by the addition of two argi- All available intermediate- and long-acting nine residues, conferring a positive charge, to insulins (NPH, Lente, Ultralente) have two the C-terminal (B30a and B30b) of the mole- major handling problems. The first is that all cule. This addition shifts the isoelectric point the classical formulations have wide variations (pI) from 5.4 to 6.7, making the molecule in absorption between different individuals, more soluble at the acid pH of the vial and and even in different injection sites and at dif- less soluble at the neutral pH of subcuta- ferent moments in the same subject, in a man- neous site. The acidity of the medium in ner which is directly proportional to the dura- which insulin is conserved dictated another tion of their action23. This phenomenon is modification of the molecule, namely substi- small, but present and significant with Regular tution of asparagine in position A21 with insulin; it is greater with NPH, Lente and glycine. This prevents deamidation and con- greatest of all with Ultralente, which is rarely fers stability51. It has been demonstrated52 used for this reason. The second problem is that Glargine interacts with insulin receptors their time-action profile. After a delay of a few in a similar way to human regular insulin and hours, their absorption gives rise to an insulin also that its dissociation is similar; it does not peak and a lowering effect on blood glucose, have mitogenic effects due to activation of the major cause of nocturnal hypoglycemia, IGF-1 receptors53. often unawareness, which afflicts many in- Pharmacological studies54-55 have been car- sulin-treated patients47. The peak is followed ried out with the iso- and euglycemic clamp at an interval of 7-8 h by a significant drop in techniques to evaluate the action profile of plasma insulin concentrations, not sufficient to Glargine in normal and diabetic subjects. Its prevent morning hyperglycemia. An attempt absorption rate has been measured in type 2 was made to correct this by giving long-acting diabetics and compared with that of insulin insulin at bedtime, between 10 and 12 pm, in- NPH: the mean 25% disappearance times of stead of at mealtime, before dinner, together 125I-insulin from the injection site were 15 and Regular or analogue48. This strategy has im- 6.5 h for Glargine and NPH, respectively, and proved basal hyperglycemia in many subjects, residual radioactivity at 24 h were 54.4 and enabling a reduction in dose, and consequent- 27.9%56. The area under the blood insulin ly a reduced risk of nocturnal hypoglycemia, curve between 0 and 6 h after administration but is still far from successful in all cases. of Glargine was significantly less than that of The need for an ideal long-acting insulin, that NPH57. Maximum plasma concentration in gives rise to a square wave of plasma concentra- the first 4 h was 5.11 IU/l for Glargine and tion, or in other words, constant basal insulin 10.8 IU/l for NPH58. A recent study57 in type levels, has prompted research into long-acting 1 diabetic patients without endogenous in- analogues. The aim of the research is to obtain sulin reserve (C-peptide negative) confirmed an insulin formulation, which can be combined that the action of Glargine lasts 24 h and that with a fast-acting analogue to ensure a closer it gives no plasmatic insulin peak, whereas parallel to physiological b-cell secretion. NPH and Ultralente achieve significantly high Candidate analogues for daily use are: (i) plasma levels a few hours after injection. soluble analogues, created by varying the Moreover, NPH lasts only 12 h and the disad- electrical charge of the insulin molecule and vantage of Ultralente is its enormous variabil- hence its isoelectric point, which affects solu- ity in absorption. Compared to NPH, bility in vivo49 (insulin Glargine) and (ii) acyl Glargine therefore shows a constant increase, analogues in which a fatty acid molecule is followed by a stable plateau, which, in some linked to an aminoacid residue so as to bind patients, lasts almost 24 h. All these features with albumin in the blood, forming a “depot” indicate that the pharmacokinetical charac- product that gradually releases insulin into teristics of Glargine meet the criteria neces- the tissues50 (insulin Detemir). sary to ensure basal insulin requirement in type 1 diabetics with negligible plasma C- Glargine peptide concentrations. Many clinical trials In 2001, insulin Glargine has become com- show that Glargine provides good control of mercially available. This long-acting analogue basal glycemic levels59-64.

77 A.M. Gualandi-Signorini, G. Giorgi

Glargine insulin is a clear liquid (like fast- Many factors influence the absorption rate acting insulin) with a pH of 4. After injection, it of insulin. forms a microprecipitate in the tissues, ensur- ing slow and steady absorption. Unfortunately, Injection site it cannot be mixed with fast-acting formula- The abdominal region offers faster and tions and this is a problem since the number of more reproducible absorption (T/2=87 min) daily injections increases when therapy proto- than the arm (140 min), buttocks (155 min) col is based on the introduction of Glargine. and thighs (165 min)72. This difference is of clinical significance and is linked to regional differences in blood flow. Absorption of in- Stability of Insulin in sulin monomers is correlated with available Commercial Preparations capillary area, explaining why muscle con- traction (which causes vasodilation) around Agents as heat, repeated and strong shak- the site of a recent injection, significantly ing, exposure to hydrophobic surfaces (such speeds onset of the glucose lowering effect of as drip attachments) cause changes in insulin the hormone. Similarly, high room tempera- conformation leading to linear aggregation ture or sauna facilitate Regular but not long- and formation of insoluble fibrils65. This phe- acting insulin to be absorbed; the mechanism nomenon mainly affects intravenous infu- involved may not be related to changes in sions (drips required in acute situations) and blood flow but rather to accelerated diffusion continuous subcutaneous infusions by in the tissues. The fast action of analogues is minipump used for intensive insulin therapy. maintained irrespective of injection site25. Many measures have been devised to avoid it, such as addition of glycerol66, albumin67, Injection depth patient’s serum68 or blood69, but none of them If the injection is too deep and does not have proved practical, physiological and safe. distribute the insulin solution or suspension At temperatures above 25° C, chemical in the subcutaneous tissue but rather in the destabilisation due to deamidation or poly- muscle, absorption is faster due to the greater merisation may occur70. Current insulin for- vascularisation of muscle with respect to sub- mulations ensure stability of the hormone be- cutaneous fat. If the injection is too superfi- low 25° C for a month. Vials for daily use can cial (in the derma), the process is slower and therefore be kept at room temperature and incomplete. supplies must be kept at 2-8° C, where their physical and chemical characteristics are Concentration maintained for a period of about 30 months. Injection of smaller volumes of insulin solution, corresponding to higher concentrations, improves absorption. The influence of concentration has been eliminated by stan- Absorption dardising all commercial preparations to 100 IU/ml. Insulin must be administered parenterally because digestive enzymes destroy it. When Massage of injection site injected s.c., insulin must be split into Massage with cotton wool soaked with dis- monomers before it enters circulation. The infectant should be avoided because it in- lymphatic system only plays a secondary role creases the absorption rate by a mechanism in absorption of the hormone. Less than 5%, a apparently not related to blood flow73. clinically insignificant amount, is broken down at the injection site. As already mentioned, the problem of intra- and inter-individual variability in insulin absorption is very significant and Distribution and Elimination can cause a variation of 10-52% in the amount of insulin absorbed daily, responsible for ex- Intravenous insulin has a half-life of 4 min- cursions in glycemia by as much as 80%71. This utes. If administered subcutaneously, it is dis- problem has not yet been solved. tributed in the plasma phase in the free form,

78 Insulin formulations – a review that is, not bound to proteins, as does en- paratus. One or two inhalations (2-6 mg) are dogenous insulin, unless the organism has generally the appropriate dose for counter- produced a significant quantity of antibodies acting post-prandial hyperglycemia in a type against it. These antibodies act as a buffer 1 diabetic patient with poor endogenous in- system, binding insulin with a capriciously re- sulin reserve. Pharmacokinetical studies in versible bond. In normal subjects, insulin has normal and diabetic subjects80 show that its a distribution volume of 85 ml/kg74. It reaches absorption rate is similar to that of injected specific receptors in target tissues. It is elimi- analogue. A plasma peak is achieved 5-60 nated by break down in the liver (60-80%), min after inhalation, compared to 60-80 min kidney (10-20%), muscle and adipose tissue for Regular insulin. It is therefore adminis- (10-20%). A small amount is excreted in the tered immediately before the three meals. Its urine. Paradoxically, significant impairment duration of action is intermediate between of renal function reduces its clearance more analogue and Regular. A mean daily require- than severe liver disease75. The residence ment of 12 mg is equivalent to 350 IU of solu- time in subcutaneous tissue is brief for fast- ble insulin. The divergence between these acting analogues, making their action profile doses and those usually administered by in- short. Subcutaneous deposit of Regular and jection of soluble or suspended insulin is due long-acting insulin is significant and is to a series of factors that cause dispersion. achieved after repeated administrations. The size of deposit is related to the number of in- 1) The first loss of hormone is in the injections/die. For example, a dose of 40 IU/die halation apparatus, but 80-95% reaches in two injections, is calculated to deposit 22- the oral cavity. 60 IU; the same dose in multiple doses de- 2) The second loss is the quantity that de- posits 12-20 IU and continuous infusion de- posits on the mucosa of the mouth and posits 1-14 IU76. This explains why patients pharynx on the way to the alveoli. taking few doses/die are protected for longer 3) Lung tissue metabolises part of the in- against rises in blood glucose levels, whereas sulin, which is not absorbed. The effi- patients undergoing continuous infusion can- ciency of absorption is calculated at 20- not even tolerate brief interruptions without 40%, in other words, 6-8 out of 10 in- developing ketosis. sulin molecules do not enter circula- tion80-81.

Many trials have been conducted with type Alternative Routes of 1 and 2 diabetic patients, comparing tradi- Insulin Administration tional therapy with inhaled insulin at mealtimes associated to a subcutaneous Ultralente Since 1924 researches about alternative injection at bedtime. All unequivocally found routes of insulin administration (oral, nasal, the similarity in the hypoglycaemic effect of rectal, transdermal) were reported77-78, but the inhaled insulin in respect to subcutaneous until recently no realistically useful prepara- route82-85. In a 3-month study, inhalation was tions had been obtained. Of all these routes, so well tolerated and accepted that 80% of inhalation is the one most tested greatly. The patients chose to continue inhalation therapy rationale behind inhalation is the passage of for a year80. However there are still some protein molecules, including insulin, across doubts as to whether this route is suitable for the mucosa of the lung alveoli, which have a all patients, that it is actually more practical surface area of 100 m2. Insulin has shown a than subcutaneous injection and that it is not particular aptitude for this route79. A phar- subject to substantial interference by factors macological formulation of the hormone for such as smoking and intercurrent pathologies inhalation will soon become commercially of the airways81,84. available. It is a dry powder dosed in mg, in The oral route has been neglected for some blisters containing 1 or 3 mg of insulin in 5 time because of the difficulty to keep insulin mg of excipient (mannitol, glycine and sodi- intact in the gastric environment and to get it um citrate powder). The blisters are opened across the barrier of the intestinal mucosa. It and emptied directly into the inhalation ap- is currently used in experimental studies for

79 A.M. Gualandi-Signorini, G. Giorgi the prevention of type 1 diabetes mellitus, ex- 10) KATSOYANNIS PG. The synthesis of the insulin ploiting its property of modifying the immune chains and their combination to biologically active response underlying insulitis by contact with material. Diabetes Rev 1964; 13: 339-348. the tolerant mucosa of the digestive system86. 11) MORIHARA K, OKA T, T SUZUKI H. Semisynthesis of human insulin by trypsin-catalyzed replacement Moreover, at the annual meeting of the of Ala-B30 by Thr in porcine insulin. Nature 1979; American Society for Clinical Pharmacology 280: 412-423. and Therapy, Still recently presented a hexyl- 12) MARKUSSEN J, DAMGAARD U, PINGEL M, SNEL L, insulin monoconjugate 2 (HIM2) in which an SORENSEN AR, SORENSEN E. Human insulin (Novo): alkyl-polyethylene-glycol group attached to chemistry and characteristics. Diabetes Care lysine B29 confers stability in the gastric envi- 1983; 6 (Suppl 1): 4-6. ronment. The phase II trials of oral insulin 13) GOEDDEL DV, KLEID DG, BOLIVAR F, et al. Expression have been presented at American Diabetes in Escherichia coli of chemically synthesized Association annual meeting82. Oral adminis- genes for human insulin. Proc Natl Acad Sci USA tration of insulin would be a break-through 1979; 76, 106-110. for obvious reasons of patient convenience 14) MILLER WL, BAXTER JD. Recombinant DNA: a new and also because it restores the primary role source of insulin. Diabetologia 1980; 18. 431-436. of the liver as manipulator of insulin, a role 15) CHANCE RE, KROEFF EP, HOFFMANN JA, FRANK BH. lost with the introduction of subcutaneous Chemical, physical and biological properties of biosynthetic human insulin. Diabetes Care 1981; administration. Under physiological condi- 4: 147-154. tions, b-cell hormone is captured by the liver 16) FRANK BH, PETTEE JM, ZIMMERMANN RE, BURK PJ. which metabolises about 40% and then re- The production of human proinsulin and its trans- leases the remaining 60% into the peripheral formation to human insulin and C-peptide. In: circulation, whereas injected insulin is ab- Rich DH, Gross E, eds. Peptide synthesis-struc- sorbed directly into the general circulation, ture-function. Rockford: Pierce Chemical causing excessive impregnation of the periph- Company 1981: 729-739. eral tissues. This is a major aspect in which 17) JOHNSON IS. Authenticity purity of human insulin. insulin replacement therapy diverges from Diabetes Care 1982; 5 (Suppl 2): 4-12. what happens in nature. 18) GALLOWAY JA, HOOPER SA, SPRANDLIN CT. Biosynthetic human proinsulin: review of chemistry, in vitro and in vivo receptor binding, animal and human pharmacology studies, and clinical experience. Diabetes Care 1992; 14: 666-692.

19) MARKUSSEN J, DAMGAARD U, JORGENSEN KH. References Production of human monocomponent insulin. In: Guerigian JL, Bransome E, Outschoorn AS, eds. 1) ANTING EST The internal secretion of the B FG, B CH. Hormone drugs. Rockville: US Pharm Conv 1982: pancreas. J Lab Clin Med 1922; 7: 256-271. 116-126. 2) BANTING FG, BEST CH. Pancreatic extracts. J Lab 20) MARKUSSEN J, DAMGAARD U, DIERS I, et al. Biosynthesis Clin Med 1922; 7: 464-472. of human insulin in yeast via single chain precur- 3) BEST CH. The first clinical use of insulin. Diabetes sors. Diabetologia 1986; 29: 568A-569A. 1956; 5: 65-67. 21) REPERTORIO FARMACEUTICO ITALIANO. Milano: OVP 4) HAGEDORN HC, JENSEN BN, KRARUP NB, WODSTRUP I. Italia, 1999. Protamine insulinate. JAMA 1936; 106: 177-180. 22) JEHLE PM, MICHELER C, JEHLE DR, BREITIG D, BOEHM 5) KRAYENBUHL C, ROSENBERG T. Crystalline protamine BO. Inadequate suspension of neutral protamine insulin. Rep Steno Mem Hosp 1946; 1: 60-73. Hagedorn (NPH) insulin in pens. Lancet 1999; 6) HALLAS-MOLLOR K. The lente insulins. Diabetes 354: 1604-1607. 1956; 5: 7-14. 23) BINDER C, LAURITZEN T, FABER O. Insulin pharmacoki- 7) SANGER F. Chemistry of insulin. Science 1959; 129: netics. Diabetes Care 1984; 7: 188-199. 1340-1343. 24) BRANGE J, OWENS DR, KANG S, VOLUND A. 8) HALES CN, RANDLE PJ. Immunoassay of insulin with Monomeric insulins and their experimental and insulin-antibody precipitate. Biochem J 1963; 88: clinical implications. Diabetes Care 1990; 13: 137-141. 923-954. 9) MARKUSSEN J, DAMGAARD U, JORGENSEN KH. Human 25) HOWEY DC, BROWSHER RR, BRUNELLE RL, monocomponent insulin. Chemistry and charac- WOODWORTH JR. [Lys(B28), Pro(B29)] human in- teristics. Acta Med Scand 1983; 671 (Suppl 1): sulin. A rapidly absorbed analogue of human in- 99-105. sulin. Diabetes 1994; 43: 396-401.

80 Insulin formulations – a review

26) BREMS DN, ALTER NA, BECKAGE MJ. Altering the as- 39) HANSEN BF, DANIELSEN GM, DREJER K. Sustained sig- sociation properties of insulin by aminoacid re- nalling from the insulin receptor after stimulation placement. Protein Eng 1992; 5: 527-533. with insulin analogues exhibiting increased mitogenic potency. Biochem J 1996; 315: 271-279. 27) DI MARCHI RD, MAYER P, F AN L. Synthesis of fast acting insulin based on structural homology with in- 40) PLUM A, HANSEN KT, ANDERSEN L, LARSEN UD. sulin-like growth factor I. In: Smith JA, Rivier JE, Pharmacokinetics of insulin aspart, a short acting eds. Peptides: chemistry and biology. Proceedings insulin analogue in rats and dogs. Diabetes 1999; of the Twelfth American Peptide Symposium. 48 (Suppl 1): A451. Leiden: Leiden ESCOM, 1992: 26-28. 41) HOME PD, BARRIOCANAL L, LINDHOLM A. Comparative 28) BAKER EN, BLUNDELL TL, CUTFIELD JF. The structure pharmacokinetics and pharmacodynamics of the of 2Zn pig insulin crystals at 1.5 A resolution. novel rapid-acting insulin analogue, insulin as- Philos Trans R Soc Lond B Biol Sci 1988; 319: part, in healthy volunteers. Eur J Clin Pharmacol 369-456. 1999; 55: 199, 203. 29) BRANGE J. Insulin preparations. In: Brange J, 42) HANEFELD M, FOSCHER S, JULIUS U, et al. Risk factors Skelbabek-Pedersen B, Langkjaer L, eds. for myocardial infarction and death in newly de- Galenics of insulin: the physicochemical and tected NIDDM: The Diabetes Intervention Study, pharmaceutical aspects of insulin preparations. 11-year follow-up. Diabetologia 1996; 39: 1577- Berlin: Springer-Verlag, 1987: 17-31. 1583. 30) PAMPANELLI S, TORLONE E, LALLI C, et al. Improved 43) TUOMILEHTO J, QIAO Q, BORCH-JOHNSEN K, BALKAU B. postprandial metabolic control after subcuta- Glucose tolerance and mortality: Comparison of neous injection of a short-acting insulin analogue WHO and American Diabetes Association diag- in IDDM of short duration with residual pancreatic nostic criteria. Lancet 1999; 354: 617-621. b-cell function. Diabetes Care 1995; 18: 1452- 44) BONORA E, KIECHL S, OBERHOLLENZER E, EGGER G, 1459. BONADONNA RC, MUGGEO M. Impaired glucose tol- 31) GARG SK, CARMAIN JA, BRADDY KC et al. Pre-meal in- erance, type 2 diabetes mellitus and carotid ath- sulin analogue insulin lispro vs Humulin R insulin erosclerosis. Prospective results from the treatment in young subjects with type 1 diabetes. Bruneck Study. Diabetologia 2000; 43: 156-161. Diabetic Med 1996; 13: 47-50. 45) PERRIELLO G, BOLLI GB. Uso dell’analogo dell’insuli- 32) ROWE R, ANDERSON JH, GALE E. A double-blind na lispro nella terapia intensiva del diabete melli- comparison of insulin lispro and regular insulin in to di tipo 1. In: Terapia insulinica: stato dell’arte e patients on a multiple injection regimen. Diabetes prospettive. Atti Simposio Satellite XVI Con- 1996; 45 (Suppl 2): 71A-72A. gresso Nazionale Società Italiana Diabetologia, 1996: 25-36. 33) SCHERNTHANER G, WEIN W, S ANDHOLZER K, EQUILUZ- BRUCK S, BIRKETT S. Postprandial use of insulin 46) BRUTTOMESSO D, DEL PRATO S. Razionale della ter- lispro: a new therapeutic option in the treatment apia insulinica nel diabete di tipo II. In: Terapia in- of type 1 diabetic patients? Diabetologia 1996; 39 sulinica: stato dell’arte e prospettive. Atti (Suppl 1): A24-A25. Simposio Satellite XVI Congresso Nazionale Società Italiana Diabetologia, 1996: 37-44. 34) EBELING P, J ANSSON PA, SMITH U, et al. Optimal combination of insulin lispro and basal insulin im- 47) THE DIABETES CONTROL AND COMPLICATIONS TRIAL proves glycemic control in IDDM. Diabetologia RESEARCH GROUP. The effect of intensive treatment 1997; 40 (Suppl 1): A351-A352. of diabetes on the development and the progres- sion of long-term complications in insulin-depen- 35) ANDERSON JH, BRUNELLE LR, KOIVISTO VA, et al. AND dent diabetes mellitus. N Engl J Med 1993; 329: THE MULTICENTER INSULIN LISPRO STUDY GROUP. 977-986. Reduction of postprandial hyperglycemia and frequency hypoglycemia in IDDM patients on in- 48) RIDDLE MC. Evening insulin strategy. Diabetes sulin-analogue treatment. Diabetes 1997; 46: Care 1990; 13: 676-686. 265-270. 49) SEIPKE G, GEISEN K, NEUBAUER HP, PITTIUS C, ROSSKAMP 36) HEINEMANN L, HEISE T, WAHL LC, et al. Pizza, coke R, SCHWABE D. New insulin preparations with pro- and tiramisù: can type 1 diabetic patients cover longed action profiles: A21-modified arginine in- such a meal with the rapid acting insulin analogue sulin. Diabetologia 1992; 35 (Suppl 1): A4-A5. Lyspro. Diabetic Medicine 1996; 13: 625-629. 50) JENSEN-HOLM HB, RIBEL U, JONASEN I. Absorption 37) HOLLEMAN F, S CHMITT H, ROTTEIRS R REES A, and effect profile after s.c. administration of the SYMANOWSKI S, ANDERSON J, THE BENELUX-UK INSULIN long-acting insulin NN-304 to pigs. Diabetologia LISPRO STUDY GROUP. Reduced frequency of severe 1995; 35 (Suppl 1): A192. hypoglycemia and coma in well-controlled IDDM 51) ROSSKAMP RH, PARK G. Long-acting insulin analogs. patients treated with insulin lispro. Diabetes Care Diabetes Care 1999; 22 (Suppl 2): 109-113. 1997; 20:1827-1831. 52) BERTI L, SEFFER E, SEIPKE G. Human insulin analog 38) DALL V. Preclinical safety pharmacology studies on HOE 901: characteristics of receptor binding and the rapid-acting insulin analogue, insulin aspart. tyrosine kinase activation. Diabetes 1995; 44 Artzneim Forsch Drug Res 1999; 49: 463-470. (Suppl 1): 243.

81 A.M. Gualandi-Signorini, G. Giorgi

53) KURTZHALS P, S CHAEFFER L, SORENSEN AR. Correlations peptide drugs. New York: Plenum Press, 1993: of receptor binding and metabolic and mitogenic 315-350. potencies of insulin analogs designed for clinical 66) BLACKSHEAR PJ, ROHDE TD, PALMER JL. Glycerol pre- use. Diabetes 2000; 49: 999-1005. vents insulin precipitation and interruption of flow 54) SOON PC, MATTHEWS DR, ROSSKAMP R. 24h profile of in an implantable insulin infusion pump. Diabetes action of biosynthetic long-acting insulin (HOE Care 1983; 6: 387-392. 901) tested in normal volunteers by glucose 67) LOUGHEED WD, WOULFE-FLANAGAN H, CLEMENT JR. clamp methodology. Diabetes 1997; 46 (Suppl 1): Insulin aggregation in artificial delivery systems. 161. Diabetologia 1980; 19: 1-9. 55) LINKESCHOWA R, HEISE T, R AVE K. Time-action profile 68) ALBISSER AM, LOUGHEED WD, PERLMAN K. of long-acting insulin analogue HOE 901. Nonaggregating insulin solutions for long-term Diabetes 1999; 48 (Suppl 1): 97. glucose control in experimental and human dia- 56) SCHOLTZ HE, VAN NIEKERK N, MEYER BH. An assess- betes. Diabetes Rev 1980; 29: 241-243. ment of the variability in the pharmacodynamics 69) HARRIS MD, DAVIDSON MB, ROSENBERG CS. Simple (glucose lowering effect) of HOE 901 compared solution to problem of biostator-induced insulin with NPH and ultralente human insulins using the aggregation. Diabetes Care 1986; 9: 356-358. euglycaemic clamp techinique. Diabetologia 1999; 42 (Suppl 1): 235. 70) BRANGE J, HAVELUND S, HOUGHAARD P. Chemical stability of insulin: 2. Formation of higher molecu- 57) LEPORE M, PAMPANELLI S, FANELLI C, et al. lar weight transformation products during storage Pharmacokinetics and pharmacodynamics of of pharmaceutical insulin preparations. Pharm subcutaneous injection of long-acting human in- Res 1992; 9: 727-734. sulin analog Glargine, NPH insulin, ultralente human insulin and continuous subcutaneous infu- 71) LAURITZEN T, FABER OK, BINDER C.Variation in 125I- sion of insulin lispro. Diabetes 2000; 49: 2142- insulin absorption and blood glucose concentra- 2148. tion. Diabetologia 1979; 17: 291-295. 58) HEINEMAN L, LINKESCOWA R, RAVE K, HOMPESH B, 72) BERGER M, CUPPERS HJ, HEGNER H. Absorption kinet- SEDLAK M, HEISE T. Time-action profile of the long- ics and biologic effects of subcutaneously inject- acting insulin analog glargine (HOE 901) in com- ed insulin preparations. Diabetes Care 1982; 5: parison with those of NPH insulin and placebo. 77-91. Diabetes Care 2000; 23: 644-649. 73) LINDE B., PHILIP A. Massage-enhanced insulin ab- 59) COATES PA, MUKHERJEE S, LUZIO SD. sorption - increased distribution or dissociation of Pharmacokinetics of a long-acting human insulin insulin. Diabetes Res 1989; 11: 191-194. analog (HOE 901) in healthy subjects. Diabetes 74) HOFFMAN A, ZIV E. Pharmacokinetic considerations 1995; 44 (Suppl 1): 130. of new insulin formulations and routes of adminis- 60) LUZIO SD, OWENS D, EVANS M. Comparison of the tration. Clin Pharmacokinet 1997; 33: 285-301. sc absorption of HOE 901 and NPH human in- 75) RALL TW, NIES AS, TAYLOR P. The insulin. LS sulin in type 2 diabetic subjects. Diabetes 1999; Goodman, A Gilman, eds. The pharmacological 48 (Suppl 1): 111. basis of therapeutic. New York: McMillan 61) PIEBER TR, EUGENE-JOLCHINE I, DEROBERT E. The Publishing Co, 1988: 1368-1369. European Study Group of HOE 901 in type 1 dia- 76) BINDER C, BRANGE J. Chemical and pharmaceutics betes. Efficacy and safety of HOE 901 versus of insulin. In: D Porte, RS Sherwin, eds. NPH insulin in patients with type 1 diabetes. Ellenberg & Rifkin’s Ð Diabetes Mellitus. Diabetes Care 2000; 23: 157-162. Stamford: Appleton & Lange, 1997: 689-708. 62) YKI-JARVINEN H, DRESSLER A, ZIEMEN M. The HOE 77) HEUBNER W, D E JONGH SE, LAQUER E. Uber 901/3002 Study Group. Less nocturnal hypo- Inhalation von Insulin. Klin Wochenschr 1924; 51: glycemia and better post-dinner glucose control 2342-2343. with bedtime insulin glargine compared with bedtime NPH insulin during insulin combination ther- 78) JENSEN HF. Insulin. Its chemistry and physiology. apy in type 2 diabetes. Diabetes Care 2000; 23: New York: the Commonwealth Fund, 1938. 1130-1136. 79) PATTON JS. Mechanisms of macromolecule absorption by the lungs. Advanced Drug Deliv Rev 63) ROBENSTOCK J, PARK G, ZIMMERMAN J FOR THE US 1996; 19: 3-36. INSULIN GLARGINE (HOE 901) TYPE 1 DIABETES INVESTIGATOR GROUP. Basal insulin glargine (HOE 80) PATTON JS, BUKAR J, NAGAJARAN S. Inhaled insulin. 901) versus NPH insulin in patients with type 1 di- Adv Drug Delivery Rev 1999; 35: 235-247. abetes on multiple daily insulin regimens. 81) GALE E. Two cheers for inhaled insulin. Lancet Diabetes Care 2000; 23: 1137-1142. 2001; 357: 324-325. 64) BERGER M. Safety of insulin glargine. The Lancet 82) BLOOMGARDEN ZT. American Diabetes Association 2000; 356: 2013. Annual Meeting, 1999: New approaches to insulin 65) BRANGE J, LANGKJAER L. IN WANG YJ, PEARLMAN R, treatment and glucose monitoring. Diabetes Care eds. Stability and characterization of protein and 1999; 22: 2078-2082.

82 Insulin formulations – a review

83) GELFAND RA, SCHWARTZ SL, HORTON M, LAW CG, PUN 85) SKYLER JS, CEFALU WT, KOURIDES IA, et al. The EF. Pharmacological reproducibility of inhaled hu- Inhaled Insulin Phase II Study Group. Efficacy of man insulin pre-meal dosing in patients with type inhaled human insulin in type 1 diabetes mellitus: 2 diabetes mellitus (NIDDM). Diabetes 1998; 47 a randomised proof-of-concept study. Lancet (Suppl 1): A61. 2001; 357: 331-335. 84) McCassey K, Perlman K, Daneman D. Insulin 86) MARON R, MELICAN NS, WEINER HL. Regulatory Th2- therapy in children and adolescents with type type T cell lines against insulin and GAD peptides 1 diabetes. Diab Nutr Metab 1999; 12: 86- derived from orally- and nasally-treated NOD mice 95. suppress diabetes. J Autoimmun 1999; 12: 251-258.