Pharmacokinetics of insulin aerosols generated by a capillary aerosol generator

February 1, 2008

The first inhalation insulin product was recently approved for use in Europe and the US as an alternative to short-acting subcutaneous insulin injections.1,2 Exubera (Pfizer Inc.) is a dry powder inhaler (DPI) that delivers recombinant human insulin and is indicated for the treatment of Type I and II diabetes. Clinical studies have revealed this product has an onset of action similar to rapid-acting insulin analogs and has a duration of glucose-lowering activity comparable with subcutaneously regular human insulin injections.3

As a result of pulmonary bioavailability issues — including inhaler dose delivery and deposition efficiency, pulmonary metabolism and ciliary clearance of insulin — relatively large doses of inhaled insulin are required to achieve a similar therapeutic response compared with subcutaneous injection.4 In the case of Exubera, dosage guidance given to patients indicates that 1 mg of this product is equivalent to 0.11 mg (3 IU) of subcutaneous regular human insulin. From a biological prospective, the challenge of increasing insulin pulmonary bioavailability can be approached by increasing the rate and extent of absorption (absorption enhancers),5,6 protecting the insulin molecule from metabolism (encapsulation or PEGylation),7–9 and decreasing macrophage clearance of insulin (large porous particles).10,11

At present, the most readily available means of improving bioavailability is to increase inhaler efficiency to deliver and deposit a larger proportion of the metered dose in the lungs. Therefore, while Exubera has led the way in the development of a safe and effective inhaled insulin, numerous other products are also in development.12–14 Aqueous soft mist aerosols may offer an advantage compared with DPI formulations as they generate aerosols that exhibit high pulmonary deposition.15 The AERx iDMS (Novo Nordisk) uses an aqueous insulin formulation and is currently in Phase?III clinical trials.16 Another soft mist-type inhaler is the capillary aerosol generator (CAG), which is capable of producing a high-efficiency fine particle aerosol from a liquid formulation that is pumped through a heated capillary.17 For drug in water or ethanol formulations, when the aerosol particles exiting the capillary have high vehicle (water or ethanol) content, the dominant process taking place within the exiting fluid stream is evaporation, which produces a soft mist aerosol for inhalation.

Previous studies have characterized the in vitro aerosol and pharmaceutical properties of insulin aerosols generated using the CAG.17 These soft mist aerosols were characterized as producing high fine particle fractions, while the primary chemical structure of the insulin molecule remained intact and was biologically active.17 In this study, CAG aerosols are generated from three different insulin formulations and administered via intra-tracheal (IT) inhalation to rats. The pharmacokinetics and pharmacodynamics of these CAG insulin aerosols were compared across formulations and relative to intravenous?(IV) insulin.

Materials and methods

Insulin aerosol formulations. Insulin aerosols were generated from three solution formulations using the CAG (Chrysalis Technologies, a division of Philip Morris USA; Richmond, VA, USA). Aerosols were produced by controlled heating of the formulation flowing through the capillary at 10 μL/s. Immediately before in vivo dosing (which will be discussed later), aerosols were generated for 10 s into a 500 mL holding chamber. Formulation A was a 1.8% w/v insulin aqueous solution; Formulation B was the commercial Humulin R formulation (Eli Lily; Indianapolis, IN, USA); Formulation C was a 1.0% w/v insulin in Ethanol:0.1 N HCI (85:15 v/v). Recombinant Human Insulin (expressed in E. coli, approximately 28.7 IU/mg anhydrous) was used in Formulations A and C (#I0259; Sigma, St Louis, MO, USA).

Aerosol administration and pharmacokinetic study. The study animals were fasted for approximately 16 h before the study. The test formulation and control studies were each performed on a separate day. For each formulation, five male Crl:CD(SD)IGS BR VAF/Plus rats (0.5–0.6 kg) were anesthetized using a combination of ketamine and xylazine. Following anesthesia and immediately before dosing, the animals were cannulated with an endotracheal tube inserted to a level above the carina. The animal was then connected to the Harvard ventilator (Model 683; Holliston, MA, USA) operating at 80 strokes/min with a 2 mL inhaled tidal volume.

Within 1 min, the holding chamber containing the insulin aerosol was connected to the ventilator and the animal was force ventilated the aerosol for 1 min. Following dosing, the animal was removed from the ventilator. For the sham control, the animals were treated, as previously described, however, they were ventilated with only air for the 1 min dosing period. Similarly, for the IV control group (0.9 IU/kg), the animals were treated (as above), however, they were ventilated with only air for the 1 min dosing period and the IV dose of insulin was administered.

Serial blood samples were collected via vascular cannula before dosing, and up to 3 h post dose. Plasma insulin and glucose levels were measured using appropriate techniques (RIA for insulin, enzymatic method for glucose) in the sham and IV control groups, and following administration of the three aerosol formulations.

Pharmacokinetic dose calculation. From the pharmacokinetic study, the exhaled air was captured on an aerosol filter and subsequently assayed to determine the amount of insulin in the exhaled air ('exhaled dose'). In vitro aerosol dose capture studies were performed for each of the formulations to determine the emitted dose of insulin from the exit of the endotracheal tube, following the 1 min aerosol administration procedure from the holding chamber — this was defined as the 'emitted dose'. The 'administered dose' of insulin was calculated as 'emitted dose'-'exhaled dose'.

Aerosol characterization. In vitro particle size analysis of the aerosols was performed using the Andersen and Moudi cascade impactors?(MSP Corporation, MN, USA). Scanning electron microscopy (Jeol JSM-35; Tokyo, Japan) was performed on aerosol samples collected from the cascade impactor stages. For these in vitro studies, a stability indicating HPLC method was used for quantitative insulin analysis.18

Results and Discussion

Aerosol characterization. Figures 1(a)–(c) show scanning electron micrographs of the insulin particles generated from the three formulations. The insulin aqueous solution formulation (Formulation A) generated an aerosol that was characterized with a mass median aerodynamic diameter (MMAD) of 1.3 μm (SD=0.15 μm) and a mean (SD) fine particle fraction (%FPF <5.6 μm) of 82.8% (2.7%). The geometric SD was 1.95. Aerosolization of the commercial Humulin R injection formulation (Formulation B) produced spherical particles as shown in Figure 1(b). This solution contained a number of additional excipients including glycerin, m-cresol and zinc oxide, which may have contributed to the altered morphology compared with Formulation A. The mean MMAD (SD) of Formulation B was 1.6 μm (0.2 μm), with a mean (SD) %FPF <5.6 μm of 85.5% (2.4%). The geometric standard deviation was 2.1. The ethanolic insulin solution formulation was characterized by a mean (SD) MMAD of 0.58 μm (0.07 μm) and a mean (SD) %FPF <5.6 μm of 85.6% (3.8%). The geometric standard deviation was 3.2, indicating the bimodal particle size distribution of this aerosol.

Figure 1

Pulmonary insulin pharmacokinetics.

The control IV administration of insulin (0.9 IU/kg) revealed a mean (SD) AUC180 of 13868 (2657) μIU/mL×min and a calculated total clearance of 72.6 mL/min/kg, which suggests extensive metabolism/degradation in blood. Table 1 shows the mean (SD) emitted aerosol doses and the measured exhaled dose during aerosol administration. The calculated administered dose in microgrammes is also shown in Table 1, and was calculated as 1.6 IU/kg, 3.8 IU/kg and 3.6 IU/kg for Formulations A, B and C, respectively.

Table 1

Figure 2 shows the mean plasma insulin concentration versus time profile for the three formulations. The profiles for Formulations B and C were typical of some of those observed in the literature following intra-tracheal administration of insulin with rapid systemic absorption and subsequent clearance.9,19 Peak plasma insulin levels were observed at 30 min following administration of Formulation B (88.7 μIU/mL). These levels were similar to those observed in other studies.19,20

The insulin plasma profile following administration of the aqueous insulin solution formulation aerosol (Formulation A) revealed a different profile compared with the other formulations, with a mean Cmax of 20.4 (7.4) μIU/mL occurring at 60 min. A similar profile was reported by Morimoto et al., (2001) following intra-tracheal administration of 5 IU/kg aqueous insulin solution.20 Figure 3 (a) and (b) shows the mean insulin AUC180 and Cmax for the three test formulations compared with the sham and IV controls. Despite similar doses, the systemic bioavailability of Formulation B was double that of Formulation C. This was possibly a result of the presence of excipients in Formulation B altering the rate and extent of insulin absorption. An alternative hypothesis is that differences in the initial particle size distribution of the generated aerosols changed the site of deposition within the lung and, therefore, altered the absorption kinetics of the insulin aerosols. Formulation C produced a submicron aerosol that was rapidly absorbed, perhaps as a result of peripheral lung deposition; however, this may have exposed it to more extensive metabolism within the airways resulting in a lower overall pulmonary bioavailability. The mean absolute pulmonary bioavailability calculated using the IV control for Formulations A, B and C was 7%, 10% and 4%, respectively. A similar bioavailability value (9.1%) was reported by Kobayashi et al. (1996) following the IT solution installation of insulin (3 IU).19 Liu et al. (1993) reported pulmonary bioavailability of a sodium insulin solution (6IU/kg) instilled IT as 14.7%.7 Readers should note that the pulmonary bioavailability calculated in this study is relative to the IV dose the literature contains a large number of studies reporting insulin bioavailability following aerosol administration; however, these values are often reported relative to subcutaneous injection.9

Figure 2

Figure 4 shows the pharmacodynamic response following insulin inhalation. A significant reduction in blood glucose levels was observed for each of the three formulations. The comparative time course profile of the glucose level reduction appeared to follow the same trend as the insulin plasma profiles. The glucose level reduction was slower for Formulation A and appeared to be correlated with the delayed systemic absorption observed in Figure 1.

Figure 3

However, at the final time point (180 min), there appeared to be similar changes in the blood glucose levels for each of the formulations, indicating a good pharmacodynamic response to the inhaled insulin, despite the measured differences in bioavailability for each of the three formulations. Note, however, that the 3 h sampling interval did not allow assessment of the entire pharmacodynamic time course, most notably duration of action. These plasma glucose profiles are similar to those appearing in similar inhaled insulin pharmacokinetic studies.7,9,20 However, direct comparison of pharmacokinetic and pharmacodynamic data is difficult because of issues relating to accurately estimating and comparing the delivered insulin dose, and the variable techniques used to instill insulin into the rat lungs, which will have significant effects on the pharmacokinetics and pharmacodynamics.

Figure 4


Pharmacokinetic and pharmacodynamic responses were observed following aerosol exposure with the three CAG insulin formulations. High levels of in vivo variability in both the pharmacokinetic and pharmacodynamic responses were also observed. This study has provided proof of concept data regarding the use of the CAG to generate and deliver insulin aerosols, but further studies are required to elucidate the differences between the formulations.

Key points


This work was supported by Chrysalis Technologies.

Animal studies were performed at Charles River Laboratories, Argus Division and their expertise and assistance is acknowledged. All animal procedures were approved by the Institutional Animal Care and Use Committee at Charles River Laboratories, Argus Division.

Michael Hindle is associate professor at Virginia Commonwealth University (Richmond, VA, USA). He received his B.Pharm., (pharmacy) and PhD (pharmaceutical technology) from the University of Bradford (UK). He has more than 15 years of experience in the broad area of aerosol drug development, including aerosol pharmacokinetics and in vitro test method development. For the past 10 years he has collaborated with Chrysalis Technologies in the development of a novel capillary aerosol inhaler.

Jürgen Venitz is associate professor, Virginia Commonwealth University. He received his MD and PhD in physiology from the Universitaet des Saarlandes in Saarbruecken (Germany). His work experience includes directing Phase I clinical pharmacology research units in Germany and as a clinical pharmacologist in early-stage drug development. He is professionally involved with AAPS , ACCP and ASCPT, and serves on various FDA advisory committees.

Patrick D. Lilly is principle scientist, RD&E, Philip Morris?USA. After receiving MSPH and PhD from the University of North Carolina at Chapel Hill (NC, USA), Dr Lilly pursued opportunities at CIIT working on biologically-based predictive models for formaldehyde and chloroform. He joined Boehringer Ingelheim Pharmaceuticals ((Ridgefield, CT, USA) where he assumed increasing levels of responsibility for management of toxicology programmes for discovery and development compounds in the areas of antivirals, anti-inflammatory diseases and metabolic disorders. he moved to Chrysalis Technologies in 2003, where he assumed responsibility for both preclinical and clinical studies.


1. First inhaled insulin product approved. FDA Consum., 40, 28–29 (2006).

2. J. Lenzer, BMJ, 332:321 (2006).

3. W.T. Cefalu, J.S. Skyler and I.A. Kourides, Ann. Intern. Med., 134, 203–207 (2001).

4. J.S. Patton, J.G. Bukar and M.A. Eldon, Clin. Pharmacokinet., 43, 781–801 (2004).

5. F. Komada et al., J. Pharm. Sci., 83, 863–867 (1994).

6. A. Hussain et al., J. Contr. Rel., 94(1),15–24 (2004).

7. F.Y. Liu et al., Pharm. Res., 10, 228–232 (1993).

8. C.L. Leach et al., "Modifying the pulmonary absorption and retention of proteins through PEGylation, in R.N. Dalby et al., Eds, Respiratory Drug Delivery IX (Davis Healthcare International Publishing River Grove, IL, USA, 2004) pp 69–77.

9. L. Garcia-Contreras et al., AAPS PharmSci., 5, E9 (2003).

10. R, Vanbever et al., Pharm. Res., 16, 1735–1742 (1999).

11. C. Lombry et al., Am. J. Physiol. Lung Cell Mol. Physiol., 286, L1002–L1008 (2004).

12. A. Pfutzner, A.E. Mann and S.S. Steiner, Diabetes Technol. Ther., 4, 589–594 (2002).

13. S. Garg et al., Diabetologia., 49, 891–899 (2006).

14. K. Hermansen et al., Diabetes Care, 27, 162–167 (2004).

15. M. Hindle, "Soft mist inhalers: A review of current technology," in The Drug Delivery Companies Report (PharmaVentures, Oxford, Autumn/Winter, 2004) pp 31–34.

16. R.R. Henry, S.R. Mudaliar and W.C. Howland, Diabetes Care, 26, 764–769 (2003).

17. M. Hindle et al., "Adding pharmaceutical flexibility to the capillary aerosol generator", in R.N. Dalby et al., Eds. Respiratory Drug Delivery IX (Davis Healthcare International Publishing, River Grove, IL, USA, (2004) pp 247–254.

18. A. Oliva, J. Fariña and M. Llabrés, Int. J. Pharm., 143(2), 163–170 (1996).

19. S. Kobayashi, S. Kondo and K. Juni, Eur. J. Pharmaceut. Sci., 11(4), 367–372 (1996).

20. K. Morimoto et al., Drug Dev. Ind. Pharm., 27(4), 365–371 (2001).