Epidemiology/Health Services/Psychosocial Research ORIGINAL ARTICLE

The Impact of on Inhaled Insulin

1 2 ANDERS HIMMELMANN, MD, PHD ASTRID H. PETERSEN, MSC this route of administration, which re- 2 2 JOHAN JENDLE, MD, PHD ULF L. DAHL, MSC semble profiles of rapid-acting analogs 1 3 ANDERS MELLEN´ , MD PER WOLLMER, MD, PHD (4,8). Patients with diabetes have similar smoking habits to smokers without dia- betes, with as many as 17–26% of diabetic patients smoking (9). The long- OBJECTIVE — This study, one of the first to address issues of pulmonary insulin delivery in term adverse effects of smoking smokers, compared pharmacokinetics of inhaled insulin delivered via the AERx insulin Diabetes Management System (iDMS) in nondiabetic cigarette smokers and nonsmokers. are well established (10), and smoking has additional acute effects on the lungs, RESEARCH DESIGN AND METHODS — In this randomized two-period crossover increasing the permeability of the al- efficacy and safety trial in 27 nondiabetic smokers and 16 nonsmokers (18 men/25 women, veolar-capillary barrier in animal models mean age 28 years, mean BMI 23.0 kg/m2), subjects received single doses of inhaled insulin (33.8 (11) and humans (12). The effects of these IU) following overnight fasting on consecutive dosing days. On one dosing day, smokers smoked smoking-related pulmonary changes on three cigarettes immediately before insulin administration (“acute smoking”); on the other dos- the delivery and absorption of inhaled ing day, smokers had not smoked since midnight (“nonacute smoking”). After inhalation, 6-h serum insulin and serum glucose profiles were determined. drugs are not yet fully understood. Smok- ing dramatically increases the rate of ab- RESULTS — Pharmacokinetic results for evaluable subjects were derived from serum insulin sorption of the ␤ agonist terbutaline (13) profiles. The amount of insulin absorbed during the first 6 h after dosing (area under the and decreases subcutaneous absorption

exogenous serum insulin curve from 0 to 6 h [AUC(0–6h)]) was significantly greater in smokers of insulin, increasing dosage require- Ϫ Ϫ (63.2 vs. 40.0 mU l 1 h 1, P ϭ 0.0017); peak concentration was both higher and earlier in the ments (14). Smoking also acutely impairs Ͻ smokers (maximal serum concentration of insulin [Cmax] 42.0 vs. 13.9 mU/l, P 0.0001; time insulin action leading to insulin resistance to maximal serum concentration of insulin [t ] 31.5 vs. 53.9 min, P ϭ 0.0003). The estimated max (15). intrasubject variability of AUC(0–6h) was 13.7 and 16.5% for nonsmokers and smokers, respec- tively. No safety issues arose. Because insulin has a narrow thera- peutic index, the AERx iDMS, an inhala- CONCLUSIONS — Absorption of inhaled insulin via the AERx iDMS was significantly tion-activated system with insulin strips,

greater in smokers, with a higher AUC(0–6h)and Cmax and a shorter tmax. Intrasubject variability has been developed to enable liquid aero- of AUC(0–6 h) was low and similar in nonsmokers and smokers. These data prompt more sols of bolus insulin to be delivered into extensive investigation of inhaled insulin in diabetic smokers. the deep lung. It emits a fine particle aero- sol (mass median aerodynamic diameter Diabetes Care 26:677–682, 2003 of 2–3 ␮m) from single-use insulin strips by extruding the prepacked solution through hundreds of precisely laser- ntensive insulin therapy is the corner- native (3,4). Advances in aerosol technol- evaporated holes in a single-use nozzle. stone of glycemic control in diabetic ogy have resulted in the efficient delivery The dose will be selected in AERx units in I patients, reducing complications in of smaller-sized particles (for example, the final version of the AERx iDMS (1 type 1 and type 2 diabetic subjects (1,2). AERx insulin Diabetes Management Sys- AERx unit ϭϳ1 IU given subcutane- Despite improvements in regimens, in- tem [iDMS], Exubera, Aerodose, and ously) and 1-unit increments will be pos- convenient multiple insulin injections Technospheres). The clinical perfor- sible, with a clear dose response as shown can pose a barrier to good glycemic con- mance, reproducibility, and patient satis- in previous trials (8,16). Trials have dem- trol. Alternative routes for insulin admin- faction with these emerging systems are onstrated the efficacy and reproducibility istration have been determined, with currently being investigated (3,5–7) of insulin administered from AERx iDMS inhaled insulin emerging as a viable alter- based on the insulin profiles seen using compared with subcutaneous delivery, ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● with no safety concerns or serious adverse events (8,16). From 1Sahlgrenska University Hospital, Gothenburg, Sweden; 2Novo Nordisk A/S, Copenhagen, Denmark; and 3Lund University, Lund, Sweden. The present study was performed Address correspondence and reprint requests to Associate Professor Anders Himmelmann, MD, Depart- with a prototype of AERx iDMS to evalu- ment of Clinical Pharmacology, Sahlgrenska University Hospital, SE-413 45 Gothenburg, Sweden. E-mail: ate the pharmacokinetics, safety, and tol- [email protected]. erability of inhaled insulin delivered to a Received for publication 14 August 2002 and accepted in revised form 11 November 2002. A.H.P. and U.L.D. hold stock in Novo Nordisk A/S. P.W. has received honoraria for speaking engage- smoking population, to assess the acute ments from and serves on an advisory panel for Novo Nordisk A/S. effects of smoking, and to compare these Abbreviations: AUC(0–6h), area under the exogenous serum insulin curve from 0 to 6 h; Cmax, maximal findings with those from healthy non- serum concentration of insulin; CV, coefficient of variation; iDMS, insulin Diabetes Management System; smoking volunteers. This is one of the MRT, mean residence time; tmax, time to maximal serum concentration of insulin. A table elsewhere in this issue shows conventional and Syste`me International (SI) units and conversion first studies to address these issues of pul- factors for many substances. monary insulin delivery in smokers.

DIABETES CARE, VOLUME 26, NUMBER 3, MARCH 2003 677 Impact of smoking on inhaled insulin

Table 1—Baseline characteristics Nordisk A/S; Aradigm Corporation) was administered to each volunteer via a rep- Smokers Nonsmokers resentative prototype AERx iDMS (Novo Nordisk A/S; Aradigm Corporation). The Sex (M/F) 11/16 7/9 insulin was administered at a revised dose Age (years) 27.7 Ϯ 5.7 27.7 Ϯ 5.2 of 33.8 IU (half a strip; previously calcu- Weight (kg) 71.7 Ϯ 10.0 71.3 Ϯ 7.4 lated to have an effect corresponding ap- BMI (kg/m2) 23.4 Ϯ 2.2 23.0 Ϯ 1.7 proximately to a single subcutaneous FVC (% predicted) 103 Ϯ 10.2 103 Ϯ 10.0 dose of 4.4 IU with this prototype) (8). Ϯ Ϯ FEV1 (% predicted) 97.9 9.2 96.0 8.3 Because of pronounced hypoglyce- FEV% (% predicted) 94.6 Ϯ 7.6 93.0 Ϯ 6.6 mia in one of the first two subjects tested, Ϯ Data are means SD. FEV1, forced expiratory volume during the first second; FEV%, forced expiratory four subjects initially received a lower in- volume %; FVC, forced vital capacity. sulin dose of 22.5 IU. The protocol was then amended when results from another study became available demonstrating a RESEARCH DESIGN AND and facilitated assessment of the effects of lower relative bioavailability of inhaled METHODS — The trial was con- acute smoking before insulin inhalation. insulin than perceived on original dose ducted in accordance with the Declara- Smokers were randomized to either not selection. As a result of this new data, all tion of Helsinki and was approved by the smoke from midnight until the end of subsequent subjects finally received a re- local ethics committee, and all subjects dosing day 1 (“nonacute smoking”), and vised dose of 33.8 IU. To ensure safety at gave their written informed consent be- then on dosing day 2 to smoke three cig- this higher dose, all subjects in whom in- fore entering the trial. arettes during the 30 min before insulin tolerable signs and symptoms of hypogly- dosing (“acute smoking,” to simulate the cemia (and/or duplicate blood glucose Volunteers extreme situation), or vice versa. Subjects readings on a HemoCue Glucometer of Healthy volunteers were current smokers were hospitalized and test conditions Յ2.0 mmol/l) were observed were treated or nonsmokers of either sex, with a mean were identical on both dosing days. with intravenous glucose infusion to raise age of 27.7 Ϯ 5.4 years and a mean BMI of Smoke inhalation did not differ between their blood glucose level to 2.5 mmol/l or 23.4 Ϯ 2.2 kg/m2 for smokers and 23.0 Ϯ smoking groups (acute and nonacute a level necessary to relieve symptoms, 1.7 kg/m2 for nonsmokers (Table 1). A smoking), and subjects used their own whereas the blood sampling continued total of 43 eligible volunteers had inhaled cigarettes and inhalation habits. as planned. This glucose infusion sty- insulin administered, and these individu- In the morning of each dosing day mied interpretation of pharmacodynamic als comprised 27 smokers and 16 non- and after an overnight fast, a single dose of parameters. smokers. Smokers were defined as inhaled human soluble insulin in dispos- After inhalation of the trial drug, sub- smoking, on average, Ն10 cigarettes a able insulin strips (67.5 IU/strip; Novo jects held their breath for 10 s and then day for at least 5 years. Nonsmokers were defined as not having smoked for at least the previous 3 years and as having an average total smoking history of Յ7 pack-years (e.g., one pack of 20 cigarettes a day for 7 years). Nonsmoker and smoker status were also confirmed by the absence or presence, respectively, of urine cotinine. All volunteers included in the trial had normal pulmonary function, as deter- mined by pulmonary function tests at screening. Subjects with active, chronic, or a history of pulmonary disease were excluded.

Protocol and methods In this single-center open-label trial (Fig. 1), nonsmokers and smokers received a single dose of inhaled insulin on each of two consecutive dosing days. Within the nonsmoking group, subjects experienced identical conditions on both dosing days. A two-period crossover design was applied within the smoker group. This al- lowed smokers to act as their own control Figure 1—Schematic trial design.

678 DIABETES CARE, VOLUME 26, NUMBER 3, MARCH 2003 Himmelmann and Associates

Figure 2—Mean exogenous serum insulin curves and glucose profiles for smokers and nonsmokers. NB: intravenous glucose infusion was initiated if a subject’s blood glucose reached 2.0 mmol/l. The vertical line at 28 min marks the time of the first glucose infusion. exhaled slowly. No reference therapy forced expiratory volume percent (calcu- the systematic effect of acute smoking was used, but strips containing sterile lated as FEV1/forced vital capacity [FVC] versus nonacute smoking. The final CV water were used for inhalation training and expressed as a percentage), and FVC estimate therefore represents the intra- purposes. (Table 1). Lung function variables were subject variability that can be attributed Blood sampling for serum insulin and expressed as a percentage of predicted to chronic or long-term smoking. serum glucose was carried out at pre- normal values. The secondary pharmacokinetic end planned intervals before the dose and for Safety assessments included fre- points—maximal serum concentration of 6 h after insulin administration. Smoking quency of adverse events, pulmonary insulin (Cmax) and time to maximal serum was not permitted during the sampling function tests, physical examination, elec- concentration of insulin (tmax), initial rate period. Serum glucose was determined trocardiogram, vital signs, and standard of increase of serum insulin concentra- using a glucose oxidase method. Serum laboratory safety analyses. tion, mean residence time (MRT), and insulin and C-peptide were both mea- terminal elimination rate constant— sured using commercially available ELISA Statistical analysis were analyzed in a similar way to the methods (Dako, Cambridgeshire, U.K.). The pharmacokinetic analyses were based AUC(0–6h). Inter- and intrasubject CVs of Exogenous insulin concentration was on 23 smokers and 13 nonsmokers with Cmax and tmax were estimated for smokers derived as evaluable insulin data. and nonsmokers. The safety evaluation The primary pharmacokinetic end was based on the safety analysis group Insulinmeasured point, area under the exogenous serum (all subjects exposed to trial product [n ϭ 43]). Ϫ (insulin /C-peptide ) insulin curve from 0 to 6 h [AUC(0–6h)], initial initial was log-transformed and analyzed in an ϫ C-peptidemeasured ANOVA model with smoker/nonsmoker RESULTS groups and acute smoking/no smoking as where the initial insulin and C-peptide fixed effects and subject as random effect. Demographic characteristics concentrations were computed as the av- The areas under the curve were calculated The two groups were well matched for erage of measurements up to the time of using the trapezoidal method. The con- age, sex, and BMI at baseline (Table 1). Of administration (17). trasts within this model (smoking vs. no the 43 subjects exposed to the trial drug, All laboratory analyses were per- smoking; acute vs. nonacute smoking) 18 were men and 25 were women, with a formed in a central laboratory (Medi-Lab were tested for significance, and the dif- mean age of 28 years and a mean BMI of A/S, Copenhagen, Denmark). In the ference with 95% confidence limits was 23 kg/m2. Within the smoking group, the case of hypoglycemic symptoms, blood calculated. The intra- and intersubject co- mean duration of smoking was 13 years. glucose for the safety analysis was efficients of variation (CVs) of AUC(0–6h) Three smokers were withdrawn from the measured in duplicate at the clinic us- were estimated separately for smokers trial after dosing (Fig. 1). Of these, two ing a HemoCue B-Glucose Analyser and nonsmokers, and a 95% CI for the were withdrawn because of hypoglycemia (HemoCue AB, Angelholm, Sweden). intrasubject CV was constructed. Non- (before the protocol was amended), and In addition, pulmonary function tests smokers experienced two dosing days un- one was withdrawn because of problems were performed at baseline, on the pre- der the same conditions. For the smoking with insertion of the cannula. All with- dosing day, and before and 6 h after the group, data from the nonacute and acute drawers were excluded from the concen- insulin administration on both dosing dosing days were used. Assuming a com- tration analyses. Additionally, four other days. These included forced expiratory mon effect, the estimation of intrasubject subjects were excluded from the serum volume during the first second (FEV1), CV in smokers included adjustment for insulin analyses. Of these, one smoker

DIABETES CARE, VOLUME 26, NUMBER 3, MARCH 2003 679 Impact of smoking on inhaled insulin

Table 2—Pharmacokinetic parameters in nonsmokers and smokers and in acute and nonacute smoking

AUC(0–6h) Ϫ1 Ϫ1 n (mU l h ) Cmax (mU/l) tmax (min) Nonsmoking mean 13 40.0 13.9 53.9 Smoking mean 23 63.2 42.0 31.5 Mean ratio — 1.58 3.02 Ϫ22.4* 95% CI — 1.20 to 2.08 1.94 to 4.70 Ϫ34.2 to Ϫ10.6 P — 0.0017 Ͻ0.0001 Ϫ0.0003 Acute smoking mean 23 55.8 36.8 30.0 Nonacute smoking mean 23 71.5 47.8 33.3 Mean ratio — 0.78 0.77 Ϫ3† 95% CI — 0.70 to 0.86 0.67 to 0.89 Ϫ14 to 7.16 P — Ͻ0.0001 0.0013 0.5340

Shown are estimated means, ratios, CIs, and P values based on ANOVA with log-transformed response, adjusted for sex, period, and acute smoking. *For tmax based on mean difference (smokers minus nonsmokers); †tmax based on mean difference.

and two nonsmokers had received the evident when comparing values of MRTs Intrasubject variability of Cmax (95% lower dose and thus were excluded, and and terminal elimination rate constants. CI) was similar for both smokers and one nonsmoker was excluded because of The MRT in the smoker group was less nonsmokers, being 23.8% (18.3–34.1) analytical problems with the samples. than half of that in the nonsmoking group and 26% (18.6– 42.9), respectively, Ͻ (P 0.0001), and the apparent elimina- whereas tmax values were 59 and 43%, Pharmacokinetic results tion rate constant of exogenous serum in- respectively. Pharmacokinetic data were derived from sulin was almost twice as high in smokers exogenous serum insulin profiles ob- as in nonsmokers (P ϭ 0.0019). Because Pharmacodynamic results tained on each dosing day by correcting absorption is the rate-limiting factor for This was primarily a pharmacokinetic total insulin based on C-peptide measure- disappearance of insulin from serum, the study. However, serum glucose was mea- ments (Fig. 2), comparing area under the faster apparent elimination in smokers is sured in the trial (Fig. 2), but because of insulin-concentration time curve and more a reflection of the increased absorp- intervention with glucose infusion in 13 other pharmacokinetic parameters in the tion rate than an indication of a difference subjects (12 smokers, 1 nonsmoker), as different study groups. in the elimination process between smok- per protocol, no formal analyses could be Smoking versus nonsmoking. More in- ers and nonsmokers. made and no conclusions were drawn. sulin was absorbed in the smoking group Acute versus nonacute smoking. With- than in the nonsmoking group, as dem- in the smoking group, mean AUC(0–6h) Safety onstrated by the mean AUC(0–6h)for se- was significantly reduced (by 22%) after A total of 13 subjects (12 smokers, 1 non- rum insulin, which was ϳ60% greater for acute smoking compared with nonacute smoker) underwent glucose infusion be- smokers than nonsmokers (63.2 vs. 40.0 smoking (P Ͻ 0.0001). The mean exoge- cause of minor hypoglycemia, as per Ϫ Ϫ mU l 1 h 1, P ϭ 0.0017) (Table 2). The nous insulin profiles (Fig. 2) showed dif- protocol. There were 16 smokers who ex- mean exogenous insulin profiles (Fig. 2) ferences between acute and nonacute perienced hypoglycemia (11 infused) on demonstrated clear differences between smoking, but again only during the first the nonacute dosing day and 13 smokers smokers and nonsmokers during the first 100 min after the dose. on the acute dosing day (again, 11 were 100 min after the dose. After this, the dif- Cmax was 23% lower after acute infused). A total of 12 smokers received ferences were minimal. smoking compared with nonacute smok- glucose infusion on one or both of the ϭ Cmax was approximately three times ing (P 0.0013). No effect of acute days; one nonsmoker received glucose higher among smokers than nonsmokers smoking was seen for tmax (Table 2). infusion on both days. Note that Cmax (P Ͻ 0.0001) (Table 2). Absorption of in- The initial rate of increase of serum was reached before glucose infusion and sulin appeared to be significantly faster in insulin concentration was significantly was unaffected by subsequent insulin smokers than in nonsmokers and tmax was (33%) lower after acute smoking com- secretion. 22 min shorter (P ϭ 0.0003) (Table 2). pared with nonacute smoking (P ϭ None of the subjects withdrew be- The initial rate of increase of serum insu- 0.0008). The terminal elimination rate cause of study-related adverse events. A lin concentration (the estimated slope of constant of exogenous serum insulin ap- total of 17 (40%) subjects experienced ad- the serum insulin profile during the pe- peared to be similar on both occasions. verse events; all were mild or moderate riod from 0 to 15 min after dosing) was Intrasubject variability. The intrasub- and evenly distributed between groups. significantly higher (ϳ3.7 times) in ject variability, as measured by CV (95% Headache was the most frequently re- Ͻ ϭ smokers than in nonsmokers (P CI) of AUC(0–6h)was similar for smokers ported adverse event (n 9) but was 0.0001). and nonsmokers, being 16.5% (12.7– likely to be due to fasting or, in the smok- The more rapid absorption of insulin 23.6) and 13.7% (9.8–22.6), respec- ers, abstinence from smoking. Headache in smokers than in nonsmokers was also tively. and hypoglycemia coincided on one oc-

680 DIABETES CARE, VOLUME 26, NUMBER 3, MARCH 2003 Himmelmann and Associates casion in only one subject. There were no its action and may necessitate a dosage tion, of inhaled insulin on acute smoking. clinically relevant changes in mean or in- adjustment. may also cause cutaneous vaso- dividual pulmonary function test results It is important to note that, despite an constriction, and this has been shown to in the two groups from screening to the apparent increase in absorption, there delay the absorption of subcutaneously end of the trial. No clinically relevant may not necessarily be a concomitant in- injected insulin (27). As intense exposure changes in other safety parameters were crease in insulin action. A study using the to tobacco smoke gives rise to the activa- reported. Spiros system (Eli Lilly) has shown that tion of leukocytes (18), it is also possible relative insulin resistance in smokers may that this could cause insulin to be de- CONCLUSIONS — This study, to in- lessen the response to a greater insulin graded in the alveoli. Even if this blunting vestigate pulmonary insulin absorption in peak (15). Cigarette smoke has been effect turns out to be constant for any nondiabetic smoking volunteers, high- found to acutely impair insulin action and given patient, the dose of the AERx iDMS lights the issue in smokers of an apparent to lead to insulin resistance (22,23) and given would have to be adjusted for this increased absorption compared with glucose intolerance (24) in healthy smok- smoking pattern. nonsmokers when dosed with pulmonary ers and exacerbate insulin resistance in Subjects adopted the same inhalation insulin via the AERx iDMS, an effect that patients with type 2 diabetes (25). Be- habits for both acute and nonacute smok- is somewhat blunted by acute smoking. cause pharmacodynamic parameters ing situations and were not instructed to Smokers comprise a substantial pro- were not specifically addressed in this do otherwise; therefore, it is unlikely that portion of the diabetic population; there- trial, we are unable to comment at this variable inhaling habits explain the differ- fore, it is essential to obtain an insight into time on the likely glucose-lowering effect ence in insulin absorption. pharmacokinetic patterns for inhaled in- produced by inhaled insulin in smokers. Differences observed in serum insulin sulin in these individuals. The present Several issues must be addressed if levels between acute and nonacute smok- study showed a near physiological insulin smokers are to be prescribed inhaled in- ing did not manifest in obvious changes in profile gained with the AERx iDMS in cig- sulin via the AERx iDMS. The increased serum glucose profiles, possibly because arette smokers. absorption of inhaled insulin in smokers of increased insulin resistance in the Interestingly, although administra- may have a considerable influence on smokers (22–25). tion of inhaled insulin is highly reproduc- dose and timing of delivery in relation to ible with the AERx iDMS system in meals and cigarette smoking. The rapidly Intrasubject variability smokers in this trial, and there is no rea- emerging high peaks in insulin concen- Intrasubject variability of extent of insulin son to suggest that this would not also be tration observed among smokers would absorption was low and similar in smok- the case in patients with diabetes, the ma- need to be considered when formulating ers and nonsmokers (16.5 and 13.7%, re- jor challenge would be selection of an ap- dosing guidelines in the smoking popula- spectively), which is encouraging and propriate dose depending on smoking tion. There is evidence to suggest that implicates an inherent reliability within pattern. there is improvement in alveolar-capillary the AERx iDMS system. This variability barrier function within a week after stop- compares favorably with that seen after Smoking versus nonsmoking ping smoking (26), which could help to subcutaneous injection of insulin in a re- The overall amount of insulin absorbed normalize inhaled insulin absorption af- cent trial with the AERx iDMS showing and peak insulin concentration was sig- ter . Although the cho- similar low variability of insulin and glu- nificantly greater and insulin absorption sen dose could possibly be reduced to cose after inhaled and subcutaneous insu- was significantly faster in the smoking accommodate the effect of smoking, the lin administration to type 1 diabetic population than in the nonsmoking pop- unique 1 AERx unit increment would no patients (7). This might be further im- ulation. A significantly higher elimination longer be useful, because adjusting the proved by patient education. rate constant and shorter MRT for insulin dose by 1 AERx unit would result in a The introduction of acute smoking was observed in smokers. serum insulin increment equivalent to be- did not appear to alter the reliability of It is well established in both animal tween 2 and 3 IU. insulin dosing in subjects. models (11) and humans (12) that ciga- rette smoke increases the permeability of Acute versus nonacute smoking Clinical implications the alveolar-capillary barrier. Postulated Within smoking subjects, acute smoking The data presented here support the con- mechanisms for these physiological just before insulin inhalation blunted the sistency of AERx iDMS, with its unique changes include immunological modifi- enhanced absorption of insulin signifi- “breath control” activation mechanism cations (18), increased blood perfusion cantly compared with nonacute smoking (extrusion of insulin only when inspira- (19), surfactant antioxidant depletion in the same group. However, no differ- tory flow rate is optimal for deep lung (20), or disruption in surfactant function ence in time to peak concentration was delivery, to minimize any influence of pa- (21). These mechanisms may also be oc- observed. tient breathing technique). The data also curring in smokers inhaling insulin and The reasons why acute smoking confirm that the system remains repro- would explain the insulin profiles seen. should limit physiological increases in in- ducible in a smoking population. Increased metabolism of drugs in haled insulin absorption compared with The study highlighted that different smokers, possibly due to the induction of nonacute smoking are unclear. Smoking smoking patterns (e.g., nonacute vs. acute drug-metabolizing enzymes, has also is known to reversibly constrict the smoking), between and within individu- been demonstrated (14). Any augment- smooth muscle of the airways, which may als, could theoretically influence dose re- ed metabolism of insulin would reduce influence the distribution, and absorp- quirements of inhaled insulin. Because

DIABETES CARE, VOLUME 26, NUMBER 3, MARCH 2003 681 Impact of smoking on inhaled insulin inhaled insulin is likely to be popular native routes of administration as an 16. Heise T, Scharling B, Bellaire S, Tusek C, among smokers and nonsmokers alike, it approach to improve insulin therapy: Bott S: Dose-response of pulmonary in- is clear that greater attention needs to be update on dermal, oral, nasal and pulmo- sulin with the AERx insulin Diabetes paid to this subsection of the population, nary insulin delivery. Curr Pharm Des 7: Management System in healthy subjects. especially if the pulmonary route of drug 1327–1351, 2001 Diabetologia 44 (Suppl. 1):A212, 2001 5. Laube BL, Benedict GW, Dobs AS: The 17. Owens DR: Human Insulin: Clinical Phar- delivery is to be exploited to its full po- lung as an alternative route of delivery for tential. We await further investigation of macological Studies in Normal Man. Lan- insulin in controlling postprandial glu- caster, UK, MTP Press, 1986 inhaled insulin absorption in smokers cose levels in patients with diabetes. Chest 18. Lehr HA: Microcirculatory dysfunction and how this compares to subcutaneous 114:1734–1739, 1998 induced by cigarette smoking. Microcircu- insulin absorption in the same patients, to 6. Gerber RA, Cappelleri JC, Kourides IA, lation 7:367–384, 2000 clarify the efficacy and safety of the AERx Gelfand RA: Treatment satisfaction with 19. Nieman GF, Clark WR Jr, Paskanik AM, inhaled insulin in patients with type 1 di- iDMS in the daily life of a diabetic patient Bredenberg CE, Hakim TS: Unilateral abetes: a randomized controlled trial. Di- who smokes. Furthermore, the issue of smoke-inhalation increases pulmonary abetes Care 24:1556–1559, 2001 potential long-term effects of inhaled in- blood flow to the injured lung. J Trauma sulin in smokers and nonsmokers alike 7. Hompesch M, Kapitza C, Scharling B, Thomsen LK, Heise T: Intra-subject vari- 36:617–623, 1994 warrants further attention. ability of pulmonary insulin via the AERx 20. Li XY, Rahman I, Donaldson K, MacNee In summary, the present study has insulin Diabetes Management System ver- W: Mechanisms of cigarette smoke in- shown that dosing with AERx iDMS is sus subcutaneous insulin. Diabetologia 44 duced increased airspace permeability. highly reproducible in both the smoking (Suppl. 1):A212, 2001 Thorax 51:465–471, 1996 and nonsmoking population. It must be 8. Brunner GA, Balent B, Ellmerer M, Schaupp 21. Schmekel B, Bos JAH, Kahn R, Lachmann emphasized, however, that this pharma- L, Siebenhofer A, Jendle JH, Okikawa J, B, Venge P, Wollmer P: Integrity of the cokinetic study was performed in nondi- Pieber TR: Dose-response relation of liquid alveolar-capillary barrier and alveolar sur- abetic subjects using relatively low doses aerosol inhaled insulin in type 1 diabetic factant system in smokers. Thorax 47: of insulin. The use of inhaled insulin in a patients. Diabetologia 44:305–308, 2001 603–608, 1992 22. Facchini FS, Hollenbeck CB, Jeppesen J, more appropriate clinical setting must of 9. Haire-Joschu D, Glasgow R, Tibbs T: Smoking and diabetes. Diabetes Care 22: Chen YDI, Reaven GM: Insulin resistance course be further investigated before clear 1887–1898, 1999 and cigarette smoking. Lancet 339:1128– guidelines can be given for this patient 10. American Thoracic Society: Cigarette 1130, 1992 group. smoking and health. Am J Respir Crit Care 23. Attvall S, Fowelin J, Lager I, Von Schenk Med 153:861–865, 1996 H, Smith U: Smoking induces insulin re- 11. Witten ML, Lemen RJ, Quan SF, Sobonya sistance: a potential link with the insulin Acknowledgments— The study was per- RE, Roseberry H, Stevenson JL, Clayton J: resistance syndrome. J Intern Med 233: formed with financial support from Novo Nor- Acute cigarette smoke exposure increases 327–332, 1993 disk A/S. alveolar permeability in rabbits. Am Rev 24. Nakanishi N, Nakamura K, Matsuo Y, Su- Respir Dis 132:321–325, 1985 zuki K, Tatara K: Cigarette smoking and 12. Jones JG, Minty BD, Lawlor P, Hulands G, risk for impaired fasting glucose and type References Crawley JC, Weall N: Increased alveolar 2 diabetes in middle-aged Japanese men. 1. DCCT Research Group: The absence of a epithelial permeability in cigarette smok- Ann Intern Med 133:183–191, 2000 glycemic threshold for development of ers. Lancet 1:66–68, 1980 25. Targher G, Alberiche M, Zenere MB, long-term complications: the perspective 13. Schmekel B, Borgstro¨m L, Wollmer P: Bonadonna RC, Muggeo M, Bonora E: of the DCCT. Diabetes 45:1289–1298, Difference in pulmonary absorption of in- 1996 haled terbutaline in healthy smokers and Cigarette smoking and insulin resistance 2. American Diabetes Association: Implica- non-smokers. Thorax 46:225–228, 1991 in patients with non-insulin dependent tions of the United Kingdom Prospective 14. Miller LG: Cigarettes and drug therapy: diabetes mellitus. J Clin Endocrinol Metab Diabetes Study. Diabetes Care 21:2180– pharmacokinetic and pharmacodynamic 82:3619–3624, 1997 2184, 1998 considerations. Clin Pharm 9:125–135, 26. Minty BD, Jordan C, Jones JG: Rapid im- 3. Skyler JS, Cefalu WT, Kourides IA, Land- 1990 provement in abnormal pulmonary epi- schulz WH, Balagtas CC, Cheng SL, Gel- 15. Wise SD, Sathirakul K, Yeo KP, Chien JY, thelial permeability after stopping smoking fand RA: Efficacy of inhaled insulin in Aftring RP: Smoking increases the bio- cigarettes. BMJ 282:1183–1186, 1981 type 1 diabetes mellitus: a randomised availability of inhaled insulin but relative 27. Klemp P, Staberg B, Madsbad S, Kolen- proof-of-concept study. Lancet 357:331– insulin resistance ameliorates differences dorf K: Smoking reduces insulin absorp- 335, 2001 in action. Diabetologia 44 (Suppl. 1):A5, tion from subcutaneous tissue. BMJ 284: 4. Heinemann L, Pfu¨ tzner A, Heise T: Alter- 2001 237, 1982

682 DIABETES CARE, VOLUME 26, NUMBER 3, MARCH 2003