Inhaled Technosphere InsulinCompared With Injected PrandialInsulin in Type 1 Diabetes: ARandomized 24-Week TrialDiabetes Care 2015;38:2266–2273 | DOI: 10.2337/dc15-0075

OBJECTIVE

To compare the efficacy and safety of Technosphere insulin (TI) and insulin aspartin patients with type 1 diabetes.

RESEARCH DESIGN AND METHODS

This open-label noninferiority trial compared the change in HbA1c frombaseline toweek 24 of prandial TI (n = 174) with that of subcutaneous aspart (n = 171), bothwith basal insulin, in patients with type 1 diabetes and HbA1c 7.5–10.0% (56.8–86.0 mmol/mol).

RESULTS

Mean change in HbA1c in TI patients (–0.21% [–2.3 mmol/mol]) from baseline(7.94% [63.3 mmol/mol]) was noninferior to that in aspart patients (–0.40%[–4.4 mmol/mol]) from baseline (7.92% [63.1 mmol/mol]). The between-groupdifference was 0.19% (2.1 mmol/mol) (95% CI 0.02–0.36), satisfying the noninfe-rioritymargin of 0.4%. However, more aspart patients achieved HbA1c <7.0% (53.0mmol/mol) (30.7% vs. 18.3%). TI patients had a small weight loss (–0.4 kg) com-pared with a gain (+0.9 kg) for aspart patients (P = 0.0102). TI patients had a lowerhypoglycemia event rate than aspart patients (9.8 vs. 14.0 events/patient-month,P < 0.0001). Cough (generally mild) was the most frequent adverse event (31.6%with TI, 2.3% with aspart), leading to discontinuation in 5.7% of patients. Treat-ment group difference formean change frombaseline in forced expiratory volumein 1 s was small (40 mL) and disappeared upon TI discontinuation.

CONCLUSIONS

In patients with type 1 diabetes receiving basal insulin, HbA1c reduction with TIwas noninferior to that of aspart, with less hypoglycemia and less weight gain butincreased incidence of cough.

Intensive glycemic control often necessitates basal-bolus insulin treatment. How-ever, hypoglycemia, weight gain, and the burden of multiple injections often lead topoor adherence (1). Hypoglycemia is considered the main limiting factor for opti-mizing glycemic control (2). Mild hypoglycemia may be inconvenient or frightening(3), whereas severe hypoglycemia can be life threatening (4). Lack of compliancewith insulin therapy is a frequent problem in patients with type 1 diabetes due topain and embarrassment associated with insulin injections (5). Concerns aboutweight gain also contribute to reduced adherence (6). An inhaled prandial insulin

1Atlanta Diabetes Associates, Atlanta, GA2Division of Endocrinology,Metabolismand LipidResearch, Washington University School of Med-icine in St. Louis, St. Louis, MO3NewYorkHospital Queens, NewYork-PresbyterianHealthcare System, Flushing, NY4Centro de Pesquisas em Diabetes Ltda., PortoAlegre, Brazil5MannKind Corporation, Paramus, NJ6Sanofi, Bridgewater, NJ7Department of Pathology, Albert Einstein Col-lege of Medicine, Bronx, NY

Corresponding author: BruceW. Bode, bbode001@aol.com.

Received 13 January 2015 and accepted 15 June2015.

Clinical trial reg. no. NCT01445951, clinicaltrials.gov.

This article contains Supplementary Data onlineat http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc15-0075/-/DC1.

A slide set summarizing this article is availableonline.

*A complete list of the Affinity 1 Study Group canbe found in the Supplementary Data online.

© 2015 by the American Diabetes Association.Readers may use this article as long as the workis properly cited, the use is educational and notfor profit, and the work is not altered.

See accompanying articles, pp. 2200,2204, 2211, 2217, 2226, 2234, 2237,2241, 2250, 2258, 2274, and 2282.

Bruce W. Bode,1 Janet B. McGill,2

Daniel L. Lorber,3 Jorge L. Gross,4

P.-C. Chang,5 and David B. Bregman,6,7

for the Affinity 1 Study Group*

2266 Diabetes Care Volume 38, December 2015

SPEC

IALARTICLE

COLLECTION:INSU

LIN

with rapid kinetics may address some ofthese concerns and could provide im-portant therapeutic options for individ-ualized diabetes management.Technosphere insulin (TI) (MannKind

Corporation, Valencia, CA) is a dry pow-der formulation of regular human in-sulin adsorbed onto Technospheremicroparticles for oral inhalation. Theprimary component of Technosphereparticles is the excipient fumaryl diketo-piperazine (FDKP), which is highly solu-ble in water at neutral and basic pH.Under acidic pH, FDKP undergoes inter-molecular self-assembly and crystallizesinto microparticles (median diameter;2–2.5 mm) (7). Upon inhalation, thesemicroparticles can reach the deep lungwhere they dissolve rapidly because ofthe physiological pH, allowing absorp-tion of insulin and FDKP into the sys-temic circulation with a time tomaximum serum insulin concentrationof ;12–15 min (8,9). FDKP is excretedunchanged in the urine (10). TI is de-livered through the Gen2 inhaler de-vice. The Gen2, which replaced theprevious device MedTone, is smaller,simpler, and more efficient (i.e., usesless TI than the MedTone inhaler toprovide the same insulin exposure,and only one inhalation per cartridgeis necessary).The objective of the current study

was to demonstrate noninferiority of TIdelivered through the Gen2 inhaler incombination with basal insulin (TI-Gen2 group) to insulin aspart in combi-nation with basal insulin (insulin aspartgroup) in its effect on HbA1c in patientswith type 1 diabetes. In addition, thestudy compared the pulmonary safetyin patients receiving TI when admin-istered through theTI-Gen2or TI-MedToneinhalers.

RESEARCH DESIGN AND METHODS

This study was a randomized, multicen-ter, 24-week trial. Adults $18 yearsold with type 1 diabetes for at least12 months with HbA1c 7.5–10.0% (58.5–86.0 mmol/mol) participated in theU.S., Russia, Ukraine, and Brazil betweenFebruary 2011 and May 2013. Inclusioncriteria were nonsmoking for the pre-ceding 6 months, BMI #38 kg/m2,forced expiratory volume in 1 s (FEV1)and forced vital capacity (FVC) $70%predicted (11), and stable insulin dose(,2 IU/kg/day) with fasting plasma

glucose (FPG),220mg/dL for 3monthsbefore screening. Exclusion criteriawere significant pulmonary disease, sig-nificant abnormalities on chest X-ray,malignancy within 5 years, severe com-plications of diabetes, or two or moresevere hypoglycemic episodes within6 months.

This study was approved by appropri-ate independent ethics committees orinstitutional review boards and moni-tored by an independent Data SafetyMonitoring Board. All patients providedwritten informed consent.

Randomization was stratified by re-gion (U.S., Brazil, and Russia/Ukraine)and basal insulin (insulin glargine, insu-lin detemir, or NPH insulin) with blocksizes of six. During a 4-week basal insulinoptimization period, all patients contin-ued their preenrollment basal insulin,converted their prandial insulin to insu-lin aspart, and titrated their dose ofbasal insulin every 3 days, targetingan FPG of 100–120 mg/dL. Patientsreaching FPG ,180 mg/dL at the endof this 4-week period were randomized1:1:1 to TI-Gen2, TI-MedTone, or insu-lin aspart.

Patients randomized to insulin aspartcontinued the aspart dose used duringbasal optimization. Patients randomizedto TI converted to a starting TI dose ad-ministered at the beginning of a meal orup to 20 min after starting a meal ac-cording to a conversion table based on10 units TI-Gen2 per ;4 units insulinaspart (Supplementary Table 1). Forthe first 12 weeks of randomized treat-ment, prandial doses in each groupwereadjusted weekly. Patients randomizedto insulin aspart targeted average pre-meal self-monitored blood glucose(SMBG) values of 100–120 mg/dL (Sup-plementary Table 2); patients random-ized to TI targeted average 90-minpostmeal SMBG values of 110–160mg/dL (Supplementary Table 3). TI patientswere instructed to take a supplementaldose (10 units TI) at the time of the read-ing if a 90-min postmeal SMBG valuewas $180 mg/dL. Aspart patients werealso advised that additional dose adjust-ments may be made based on meal size,carbohydrate content, or SMBG results.Dosing was kept stable except for safetyreasons during the final 12 weeks of ran-domized treatment. After 24 weeks ofrandomized treatment, patients con-verted back to insulin aspart and were

followed for another 4 weeks to assesspulmonary safety.

The primary efficacy end point waschange in HbA1c from the end of basaloptimization (baseline) to week 24 inthe TI-Gen2 and insulin aspart groups.Key secondary efficacy end points werethe proportion of patients achievingHbA1c #7.0% (53.0 mmol/mol), propor-tion of patients achieving HbA1c #6.5%(47.5 mmol/mol), change in FPG frombaseline to week 24, seven-point plasmaglucose profile (before meals, 90 minafter meals, and bedtime) based onSMBG values, and change in bodyweight.

The primary safety end point waschange in FEV1 from baseline to week24 in the TI-Gen2 and TI-MedTonegroups. Spirometry (FEV1 and FVC) wasperformed according to American Tho-racic Society and European RespiratorySociety guidelines. Safety assessmentalso included incidence and event ratesof hypoglycemia. Nonsevere hypoglyce-mia was defined as SMBG,70 mg/dL orsymptoms of hypoglycemia; severe hy-poglycemia was an event in which theassistance of another individual was re-quired. Other safety assessments in-cluded adverse events (AEs), vitalsigns, electrocardiograms, laboratoryvalues, and physical examinations.Anti-insulin antibody levels, measuredusing radioimmunoassay (Kronus, Star,ID), were reported in Kronus units.

Three hundred fourteen patientswere targeted to be randomized to TI-Gen2 and insulin aspart groups, whichwould provide 90% power, using aone-sided a of 0.025 and an SD of 1.0for a noninferiority design to test thedifference in change from baseline toweek 24 in HbA1c between the twogroups. This assumes a noninferioritymargin of 0.4% and a dropout rate of15%.

The full analysis set, defined as all ran-domized patients, was used for efficacyanalyses. The safety population, definedas patients who had at least one dose ofstudy medication (based on treatmentseach subject received), was used forsafety analyses. All analyseswere prede-fined in the statistical analysis plan. Noimputation was applied unless other-wise specified.

Noninferiority was assessed usinga mixed-model repeated-measures(MMRM)analysis,withHbA1cmeasurement

care.diabetesjournals.org Bode and Associates 2267

as the dependent variable and explana-tory variables of region, basal stratum,visit (categorical time), treatment, andvisit by treatment as fixed effects; pa-tient as a random effect; and baselineHbA1c as covariate. Sensitivity analyseswere also performed to evaluate theimpact of missing data. The primarysafety comparison, change in FEV1

from baseline to week 24 in the TI-Gen2versus the TI-MedTone group, was an-alyzed using the MMRM model, withbaseline pulmonary function test(PFT) value, height, and age as covari-ates; treatment, race, sex, visit (cate-gorical time), and visit by treatment asfixed effects; and subject as a randomeffect.

RESULTS

Patient disposition is shown in Supple-mentary Fig. 1. A total of 1,401 patientswere screened for eligibility, 614 ofwhomentered the basal insulin optimization

phase and 518 randomized (TI-Gen2[n = 174], insulin aspart [n = 170], orTI-MedTone [n = 174]). Demographicsand baseline characteristics are shownin Supplementary Table 4. Patient demo-graphic and baseline characteristicswere balanced among the groups.Mean baseline HbA1c was ;7.95%(;63.4 mmol/mol). Of the randomizedpatients, 130 (74.7%) in the TI-Gen2group, 151 (88.8%) in the insulin aspartgroup, and 138 (79.3%) in the TI-MedTonegroup completed the 24-week treat-ment period. Dropout rates in theTI-Gen2, insulin aspart, and TI-MedTonegroups were 25.3%, 11.2%, and 20.7%,respectively, primarily due to with-drawal of consent for personal reasons(e.g., moving, changing jobs) in allgroups, whereas withdrawal becauseof an AE occurred only with TI (Supple-mentary Fig. 1).

The mean change in HbA1c at week 24was –0.21% (–2.3 mmol/mol) from a

baseline of 7.94% (63.3 mmol/mol) forthe TI-Gen2 group and –0.40% (–4.4mmol/mol) from a baseline of 7.92%(63.1 mmol/mol) for the insulin aspartgroup based on MMRM analyses (Table1). The between-group difference was0.19% (2.1 mmol/mol) (95% CI 0.02–0.36), which satisfied the predefinednoninferiority criterion (,0.4%). As-sessment of the per-protocol popula-tion also supported noninferiority(Supplementary Table 5). HbA1c levelsdeclined by ;0.6% in both groups dur-ing basal optimization, continued todecrease in the first 12 weeks of ran-domized treatment (slightly more inthe insulin aspart group), and then re-mained stable during the 12-week sta-ble dosing period (Fig. 1A).

During randomized treatment, themean dose of daily basal insulin in-creased by ;5 units in the TI-Gen2group (31.7 units at baseline to 36.8units at week 12) and ;2 units in the

Table 1—Key analyses and results (full analysis set)

TI-Gen2(n = 174)

Insulin aspart(n = 170) Treatment difference

HbA1c (% [mmol/mol])*Baseline 7.94 (63.3) 7.92 (63.1)Week 24 7.73 (61.0) 7.52 (58.7) 0.19 (2.1)Adjusted mean change 20.21 (–2.3) 20.40 (–4.4)95% CI –0.33 to –0.09 –0.52 to –0.28 0.02 to 0.36

HbA1c #7.0% (53.0 mmol/mol)†Incidence (%) 24 (18.3) 46 (30.7) OR = 0.44995% CI 0.23 to 0.86

P = 0.0158

HbA1c #6.5% (47.5 mmol/mol)†Incidence (%) 10 (7.6) 19 (12.7) OR = 0.57695% CI 0.24 to 1.38

P = 0.2144

FPG (mg/dL)*Baseline (SE) 153.9 (5.3) 151.6 (5.4)Adjusted mean change (SE) 225.3 (7.6) 10.2 (7.4) 235.4 (10.6)95% CI 256.3 to –14.6

P = 0.001

Body weight, kg‡Baseline (SE) 75.5 (16.1) 73.5 (15.3)Adjusted mean change (SE) 20.4 (0.4) 0.9 (0.4) 21.3 (0.5)95% CI 22.33 to –0.31

P = 0.0102

All hypoglycemia§Incidence (%) 167 (96.0) 170 (99.4) P = 0.0621Event rate (events/patient-month) 9.8 14.0 P , 0.0001

Severe hypoglycemia§Incidence (%) 32 (18.4) 50 (29.2) P = 0.0156Event rate (events/100 patient-months) 8.1 14.5 P = 0.1022

OR, odds ratio. *Assessed usingMMRManalysis. †Assessed using logistic regression analysis. ‡Assessed using ANCOVA. §Incidencewas based on thesafety population, which was 174 for the TI-Gen2 group and 171 for the insulin aspart group. Incidence was assessed using logistic regressionanalysis. Event rate was assessed using negative binomial analysis.

2268 Inhaled Versus Injected Insulin in Type 1 Diabetes Diabetes Care Volume 38, December 2015

insulin aspart group (28.7 units at base-line to 30.8 units at week 24) (Fig. 1B).The mean daily prandial insulin aspartdose started at 23.5 IU at randomizationand increased 9.1% during the first12 weeks of randomized treatment.The mean daily prandial dose in theTI-Gen2 group was 75.0 units at randomi-zation (10-unit cartridge of TI-Gen2 =;4units of rapid-acting insulin analog[RAA]) (Supplementary Table 1). Thedose increased to 107.4 units at week12 and then remained stable for thenext 12 weeks (Fig. 1C). Approximately60% of patients in each treatment grouptook at least one supplemental dose inaddition to their meal-associated doses;however, overall supplemental dosingwas infrequent.At week 24, the adjusted mean

change in FPG levels was –25.3 mg/dLwith TI-Gen2 and 10.2 mg/dL with insu-lin aspart. The treatment difference in

FPG change from baseline to week 24was statistically significant in favorof TI-Gen2 (–35.4 mg/dL, P = 0.001)(Table 1).

For the secondary end point of HbA1cgoal attainment, more insulin aspart pa-tients (46 [30.7%]) reached HbA1c tar-gets of #7.0% (53.0 mmol/mol) thanTI-Gen2 patients (24 [18.3%], P =0.0158) (Table 1). The between-groupdifference in attaining HbA1c #6.5%(47.5 mmol/mol) was not statisticallysignificant.

At baseline, seven-point glucose pro-files were higher at every time point forthe TI-Gen2 group than for the insulinaspart group; this difference was alsoseen at week 24, except for a lower pre-breakfast value in the TI-Gen2 group(Fig. 1D). A progressive increase inSMBG values throughout the day wasnoted for both groups at baseline andweek 24; this increase was greater in

the TI-Gen2 group. TI patients had asmall weight loss (–0.4 kg) comparedwith a gain (+0.9 kg) for aspart patients(P = 0.0102) (Table 1).

The proportions of patients reportingat least one AEwere higher with TI-Gen2and TI-MedTone (101 of 174 [58.0%]and 104 of 173 [60.1%], respectively)than with insulin aspart (74 of 171[43.3%]). AEs occurring in $2% of pa-tients are shown in SupplementaryTable 6. AEs were mostly mild ormoderate. Rates of serious AEs werelow in all groups (TI-Gen2, 5 of 174[2.9%]; TI-MedTone, 9 of 173 [5.2%]; in-sulin aspart, 7 of 171 [4.1%]). Discontin-uations due to AEs occurred with TI only(TI-Gen2, 16 of 174 [9.2%]; TI-MedTone,9 of 173 [5.2%]). One patient in the in-sulin aspart group died of accidentaldrowning, which was considered un-likely related to the study drug. Exclud-ing hypoglycemia, cough was the most

Figure 1—Mean change over time in HbA1c from screening to posttreatment (A), average daily dose of basal insulin in the TI-Gen2 and insulin aspartgroups (B), mean daily dose of prandial insulin in the TI-Gen2 and insulin aspart groups (C), and mean seven-point glucose profiles at baseline andweek 24 (D). Full analysis set is presented with SE bars. F/U, follow-up.

care.diabetesjournals.org Bode and Associates 2269

frequent AE and reported by a higherpercentage of the TI-Gen2 (55 of 174[31.6%]) and TI-MedTone (39 of 173[22.5%]) patients than the insulin aspartpatients (4 of 171 [2.3%]). Regardless ofthe inhaler, cough was predominantlya mild, dry, intermittent (66–69%) orsingle-defined event (23–27%) and oc-curred within 10 min after inhalation ofthe dry powder. Cough led to study dis-continuation in 15 patients (TI-Gen2, 10of 174 [5.7%]; TI-MedTone, 5 of 173[2.9%]) and was severe in 3 (TI-Gen2, 2of 174 [1.1%]; TI-MedTone, 1 of 173[0.6%]). Cough resolved when TI wasdiscontinued. In both the TI-Gen2 andTI-MedTone groups, the percentage ofpatients reporting new-onset cough washighest in the first week after treatmentinitiation (19.5% and 13.3%, respec-tively) and then decreased to 4.1% and2.3%, respectively, during week 2, and to0% and 0.5%, respectively, by week 8.Over 24 weeks, all groups had small

declines from baseline in mean FEV1(TI-Gen2, –0.07 L; TI-MedTone, –0.08 L;insulin aspart, –0.04 L). The decline wasgreater in the inhaled insulin groups,but this difference resolved by thefollow-up visit 4 weeks after discontinua-tion of treatment (Fig. 2). A similarpattern was seen in FVC results (datanot shown).The hypoglycemia event rate, regard-

less of severity, as well as the incidenceof severe hypoglycemia was lower inthe TI-Gen2 group than in the insulinaspart group (P , 0.0001 and 0.0156,

respectively) (Table 1). Over the courseof the study, the event rate for severehypoglycemia was 44% lower in theTI-Gen2 group (8.1 vs. 14.5 events/100patient-months for TI vs. insulin aspart),but the difference was not statisticallysignificant (P = 0.1022). Both nonsevereand severe hypoglycemia event rates inthe TI-Gen2 group were consistentlylower regardless of the achieved HbA1cvalue at week 24 (Fig. 3A and B) andafter adjustment for HbA1c changefrom baseline to week 24 (Fig. 3C). Sim-ilarly, the hypoglycemia rate was lowerin the TI group after HbA1c was added tothe statistical model used in the dataanalysis (Supplementary Table 7). Fur-thermore, the timing of the hypoglyce-mia events with TI paralleled its rapidkinetics; hypoglycemia event rateswithin 2 h postmeal were comparablebetween TI and insulin aspart but weretwo- to threefold higher in the insulinaspart group .2–5 h after meals (Fig.3D).

At week 24, median insulin antibodylevels increased in the TI-Gen2 group(9.3 to 30.9 Kronus units/mL from base-line to week 24) and the TI-MedTonegroup (8.6 to 41.1 Kronus units/mLfrom baseline to week 24) and remainedunchanged in the insulin aspart group(8.7 and 8.9 Kronus units/mL from base-line to week 24). No differences wereseen in the type or incidence of AEs orchanges in HbA1c in patients at the top10th percentile of insulin antibody in-crease and the overall trial population.

There were no clinically meaningfulchanges in other laboratory values, vitalsigns, physical examinations, or electro-cardiogram results.

CONCLUSIONS

The results of this trial demonstratethat a regimen of prandial TI in combi-nation with basal insulin is noninferiorto a standard regimen of basal insulinplus prandial RAA for reducing HbA1c

in patients with type 1 diabetes. Themean reduction in HbA1c was greaterin the insulin aspart group than in theTI group, and more patients in the insu-lin aspart group achieved HbA1c goals of#7.0% (53.0 mmol/mol) and #6.5%(47.5 mmol/mol). Certain factors mayhave contributed to these differences.The trial compared a basal-bolus regi-men using a novel inhaled agent (TI) tothat of a standard-of-care injectablecomparator (insulin aspart) under con-ditions that would optimize the com-parator treatment regimen. Afterrandomization, the insulin aspart groupcontinued the basal-bolus regimen thatall patients received during basal opti-mization. Approximately 40% of pa-tients used insulin aspart in clinicalpractice (before trial entry). In compari-son, the inhaled insulin group switchedto an unfamiliar regimen and titrationscheme (based on 90-min postprandialrather than premeal SMBG values).Study design elements, including a newtitration scheme along with less insulinavailable 2–5 h after the meal with in-haled insulin, could account for thisHbA1c difference. Optimization of thebasal dose could potentially decreasethis difference between inhaled insulinand RAA.

Consistently less hypoglycemia wasobserved in TI patients than in insulinaspart patients. The greatest differencewas observed in the .2–5-h period af-ter the meal. This analysis of hypoglyce-mia rate as a function of time after ameal indicates that the lower hypogly-cemia rate associated with TI resultedfrom its rapid kinetics. Maximumplasma drug concentration occurssooner with TI (12–15 min) than withRAA (45–60 min) (8,12–14). Similarly,the time to return to baseline drug con-centration (180 min vs. .300 min) andtime to baseline glucose lowering effect(240 min vs. 300–360 min) is also lesswith TI than with RAA. HbA1c reduction

Figure 2—FEV1 test results from baseline toweek 28 follow-up (MMRM). Safety population. F/U,follow-up.

2270 Inhaled Versus Injected Insulin in Type 1 Diabetes Diabetes Care Volume 38, December 2015

was 0.19% less in the TI group than inthe insulin aspart group. However, thehypoglycemia advantage of TI over in-sulin aspart was maintained, even afteradjusting for the different levels ofachieved HbA1c at week 24 and whenincluding change from baseline inHbA1c in the statistical model. Thus,the small between-group difference inHbA1c reduction favoring the insulin as-part group could not entirely explain thebeneficial effect on hypoglycemia seenwith TI. Rather, at least a part of thehypoglycemia advantage is likely to re-sult from the more rapid kinetics of TI.Although a small HbA1c reduction wasseen in both groups, a small weight re-duction was seen in the TI-Gen2 group,whereas a statistically significant weightgain occurred for the insulin aspartgroup. Although the mechanism is notknown with certainty, reduced concernabout hypoglycemia may lead to less“defensive” eating (15).There was a decrease in FPG for the

TI-Gen2 group compared with an in-crease in the insulin aspart group. Again,the mechanism is not known, but it hasbeen observed previously with TI andwith another inhaled insulin (16,17). Acanine model suggested that it mightbe due to better peripheral glucose

utilization after delivery of inhaled insu-lin (18). Except for the prebreakfast(FPG) value, seven-point glucose valueswere greater for the TI-Gen2 group thanfor the insulin aspart group at all otherpoints. SMBG values increased through-out the day in both treatment groups.

During the 24-week randomizedtreatment period, basal insulin doses in-creased in both the TI-Gen2 and the in-sulin aspart groups but increased by3 units more in the TI-Gen2 group. Thisfinding is not unexpected because dura-tion of action of TI is shorter than that ofinsulin aspart. The need to increase thedose of the basal insulin in a basal-bolusregimen was similarly observed whenRAAs with a relatively shorter time ac-tion profile compared with regular hu-man insulin were introduced in thebasal-bolus treatment regimen (19–21). When basal insulin dose was in-cluded in the statistical model used forassessing the change in HbA1c, noninferi-ority of the TI group relative to the in-sulin aspart group was maintained(Supplementary Table 8). Patients inboth the TI-Gen2 and the insulin aspartgroups continually titrated their pran-dial insulin doses during the first12 weeks of the randomized treatmentperiod. Therewas a greater increase in the

TI dose partly because the protocol-recommended conversion ratio (one10-unit cartridge of TI per 4 units insulinaspart) was intentionally conservativeto maximize safety during the switchfrom injected to inhaled insulin.

This study included a direct compari-son of pulmonary function in patientsreceiving TI through the Gen2 orMedTone inhalers. The changes in bothFEV1 and FVC observed during treatmentwith either inhaler were small. No asso-ciation was noted with either FEV1 orFVC changes and mean TI dose, age,sex, race, or cough status; mean changesin FEV1 and FVC were driven not by asmall number of subjects with largechanges (outliers) but by slight shifts indistribution because of a large numberof subjects with small changes (data notshown). Furthermore, the changes re-versed after discontinuing TI treatment.The changes to FEV1 and FVC were un-likely to be clinically significant. Thehigher incidence of cough with theGen2 inhaler was most likely a result ofthe greater amount of powder being in-haled in a single inhalation with Gen2,whereas with MedTone, the amount ofpowder inhaled per dose is distributedover two inhalations. Overall, the pul-monary findings in this study are

Figure 3—Analyses of hypoglycemia. Event rates of nonsevere and severe hypoglycemia by HbA1c category at week 24 (A and B), total hypoglycemiaas a function of change in HbA1c from baseline to week 24 (C), and total hypoglycemia as a function of time after a meal throughout the trial (D). Fullanalysis set. Based on negative binomial analysis to account for patient effect and exposure time.

care.diabetesjournals.org Bode and Associates 2271

consistent with data collected through-out the TI development program de-scribing the pulmonary tolerability andsafety of TI (12). All adults, whetherthey have diabetes, demonstrate a de-cline in pulmonary function with age.There is some evidence that the declinein pulmonary function in patients withdiabetes is greater than that of the gen-eral population (22). In clinical trials of TIof up to 2 years duration, small (40 mL or;1–1.5%) nonprogressive treatmentgroup differences in PFT decline are ob-served, which disappear after TI is dis-continued. These results suggest thatthe observed changes in PFTs are of lim-ited clinical significance. Longer studies(.5 years) are being conducted to eval-uate PFT changes associated with long-term use of TI.The potential impact of dropouts on

the noninferiority analysis was assessedusing multiple statistical models. First,the analysis method (MMRM) is a pre-ferred model for addressing missingdata (23,24). Analyses performed basedon the completers population or theconventional last observation carriedforward imputation method confirmedthe conclusion from the primaryMMRManalysis based on the full analysis set.Furthermore, sensitivity analyses, suchas pattern mixture and multiple imputa-tion analyses, concluded that the im-pact of missing data is insignificant tothe conclusion of noninferiority. Im-portantly, in all analyses, the averageHbA1c level was maintained with the TIregimen over the 24-week randomizedtreatment period, with no clinically sig-nificant loss of glycemic control. Duringthe 4-week follow-up phase after pa-tients converted back to prandial insulinaspart, mean HbA1c levels were main-tained with no clinically significantchange in the rate of hypoglycemia, al-though the difference in hypoglycemiarate between TI and aspart started todiminish (Supplementary Fig. 2).Patients receiving TI experienced an

increase in insulin antibodies, but higherantibody levels were not associatedwith adverse clinical outcomes. The in-crease in antibody levels has been re-ported with inhaled insulin and isprobably associated with the lungs’ im-munological properties (25).In summary, this 24-week study

showed that a regimen of prandialTI-Gen2 in combination with basal insulin

led to noninferior reductions in glycemicparameters, with weight neutrality andless hypoglycemia compared with sub-cutaneous aspart. However, TI-Gen2 en-abled fewer patients to achieve HbA1c,7.0% (53.0 mmol/mol) (18.3% withTI-Gen2 vs. 30.7% with aspart), mostlikely due to higher glucose levels inthe late postprandial period 2–5 h post-inhalation. The change in FPG frombaseline to week 24 was –25.3 mg/dLwith TI-Gen2 vs. 10.2 mg/dL with aspart.TI was associated with a higher inci-dence of (generally mild) cough (31.6%with TI-Gen2 vs. 2.3% with aspart),which also contributed to discontinua-tions due to AEs occurring with TI only(TI-Gen2, 16 of 174 [9.2%]; TI-MedTone,9 of 173 [5.2%]). Thus, TI is an option forprandial insulin in patients with type1 diabetes who have concerns abouthypoglycemia or injection burden. A trialto evaluate long-term pulmonary safetywill be conducted.

Acknowledgments. The authors thank all thepatients who participated in the trial as well asthe following current and former MannKindCorporation employees for expert contributionsto trial conduct and analysis: I. Battles, D. Gash,Y. Ma, L. Ming, S. Suh, and G. Tonelli. Medicalwriting assistance was provided by F. Martino ofFPM Communications, LLC (North Brunswick,NJ).Funding. B.W.B.’s employer (Atlanta DiabetesAssociates) received grant and research supportfrom the National Institutes of Health.Duality of Interest. This study was funded byMannKind Corporation. Sanofi has a worldwidelicense for TI from MannKind Corporation.Technosphere and MedTone are registeredtrademarks of MannKind Corporation. B.W.B.is a stock/shareholder of Aseko (formerly Gly-tec); received consulting honoraria fromHalozyme, Janssen, Medtronic, Novo Nordisk,Sanofi, Tandem Diabetes Care, and Valeritas;andbelongs to speakers’bureaus for AstraZeneca,Dexcom, GlaxoSmithKline, Insulet, Janssen, EliLilly, Medtronic, Merck, Novo Nordisk, Sanofi,and Valeritas, and grant and research supportwas received by his employer (Atlanta DiabetesAssociates) from Abbott, Biodel, Bristol-MyersSquibb/Amylin, Dexcom, GlaxoSmithKline,Halozyme, Janssen, Eli Lilly/Boehringer Ingelheim,MacroGenics, MannKind Corporation, Medtronic,NovoNordisk, Sanofi, andValeritas. J.B.M. has par-ticipated in advisory panels for AstraZeneca,Merck, Janssen, Abbott, Novo Nordisk, McNeil,Sanofi, and Eli Lilly/Boehringer Ingelheim; hasreceived consulting honoraria from MannKindCorporation; received research support fromMannKind Corporation, Andromeda, Novartis,Sanofi, Novo Nordisk, and Takeda; and belongsto the speakers’ bureau for Janssen. D.L.L. hasparticipated in advisory panels for Merck, Eli Lilly/Boehringer Ingelheim, Novo Nordisk, and Roche;

is a board member and stock/shareholder of Bio-del; received consulting honoraria from Merck,Novo Nordisk, and Roche; has received researchsupport from MannKind Corporation, Janssen,Novo Nordisk, Boehringer Ingelheim, andMerck; and belongs to speakers’ bureaus forNovo Nordisk and Janssen. J.L.G. received consul-ting honoraria from Boehringer Ingelheim, Bristol-Myers Squibb, Eli Lilly, and Novo Nordisk andresearch support from Boehringer Ingelheim,Bristol-Myers Squibb, Eli Lilly, GlaxoSmithKline,and Novo Nordisk. P.-C.C. is an employee ofMannKind Corporation. D.B.B. was an employeeofMannKind Corporation at the time the trial wasperformed. Medical writing assistance wasfunded by MannKind Corporation. No otherpotential conflicts of interest relevant to thisarticle were reported.Author Contributions. B.W.B., J.B.M., D.L.L.,and J.L.G. contributed to the conduct of thestudy; data acquisition, analysis, and interpre-tation; and review and approval of the manu-script. P.-C.C. contributed to the study design;data acquisition, analysis, and interpretation;and review and approval of the manuscript.D.B.B. contributed to the data analysis and in-terpretation and reviewand approval of theman-uscript. P.-C.C. and D.B.B are the guarantors ofthis work and, as such, had full access to all thedata in the study and take responsibility for theintegrity of the data and the accuracy of the dataanalysis.Prior Presentation. Parts of this study werepresented in abstract form at the 74th ScientificSessions of the American Diabetes Association,San Francisco, CA, 13–17 June 2014.

References1. Morris AD, Boyle DI, McMahon AD, GreeneSA, MacDonald TM, Newton RW. Adherenceto insulin treatment, glycaemic control, andketoacidosis in insulin-dependent diabetesmellitus. The DARTS/MEMO Collaboration. Di-abetes Audit and Research in Tayside Scot-land. Medicines Monitoring Unit. Lancet1997;350:1505–15102. Cryer PE. Hypoglycemia in type 1 diabetesmellitus. Endocrinol Metab Clin North Am2010;39:641–6543. Jacqueminet S, Masseboeuf N, Rolland M,Grimaldi A, Sachon C. Limitations of the so-called“intensified” insulin therapy in type 1 diabetesmellitus. Diabetes Metab 2005;31:4S45–4S504. American Diabetes Association. Standards ofmedical care in diabetesd2014. Diabetes Care2014;37(Suppl. 1):S14–S805. Peyrot M, Rubin RR, Kruger DF, Travis LB.Correlates of insulin injection omission. Diabe-tes Care 2010;33:240–2456. Peyrot M, Skovlund SE, Landgraf R. Epidemi-ology and correlates of weight worry in the mul-tinational Diabetes Attitudes, Wishes andNeeds study. Curr Med Res Opin 2009;25:1985–19937. Leone-Bay A, Grant M. Technosphere/insulin:mimicking endogenous insulin release. InModified-Release Drug Delivery Technology.Vol. 2, 2nd ed. Rathbone M, Hadgraft J,Roberts M, Lane M, Eds. New York, InformaHealthcare USA, Inc., 2008, p. 673–6798. Boss AH, Petrucci R, Lorber D. Coverage ofprandial insulin requirements by means of an

2272 Inhaled Versus Injected Insulin in Type 1 Diabetes Diabetes Care Volume 38, December 2015

ultra-rapid-acting inhaled insulin. J Diabetes SciTechnol 2012;6:773–7799. Richardson PC, Boss AH. Technosphere insu-lin technology. Diabetes Technol Ther 2007;9(Suppl. 1):S65–S7210. Potocka E, Cassidy JP, Haworth P,Heuman D, van Marle S, Baughman RA Jr.Pharmacokinetic characterization of the novelpulmonary delivery excipient fumaryl diketo-piperazine. J Diabetes Sci Technol 2010;4:1164–117311. Hankinson JL, Odencrantz JR, Fedan KB. Spi-rometric reference values from a sample of thegeneral U.S. population. Am J Respir Crit CareMed 1999;159:179–18712. Neumiller JJ, Campbell RK, Wood LD. A re-view of inhaled Technosphere insulin. Ann Phar-macother 2010;44:1231–123913. Rave K, Heise T, Heinemann L, Boss AH. In-haled Technosphere insulin in comparison tosubcutaneous regular human insulin: time ac-tion profile and variability in subjects withtype 2 diabetes. J Diabetes Sci Technol 2008;2:205–21214. Bosi E, Scavini M, Ceriello A, et al.; PRISMAStudy Group. Intensive structured self-monitoringof blood glucose and glycemic control in non-insulin-treated type 2 diabetes: the PRISMA

randomized trial. Diabetes Care 2013;36:2887–289415. Hollander PA. Insulin detemir for the treat-ment of obese patients with type 2 diabetes.Diabetes Metab Syndr Obes 2012;5:11–1916. Rosenstock J, Lorber DL, Gnudi L, et al. Pran-dial inhaled insulin plus basal insulin glargineversus twice daily biaspart insulin for type 2 di-abetes: a multicentre randomised trial. Lancet2010;375:2244–225317. Skyler JS, Weinstock RS, Raskin P, et al.; In-haled Insulin Phase III Type 1 Diabetes StudyGroup. Use of inhaled insulin in a basal/bolusinsulin regimen in type 1 diabetic subjects:a 6-month, randomized, comparative trial. Di-abetes Care 2005;28:1630–163518. Edgerton DS, Cherrington AD, Neal DW,et al. Inhaled insulin is associated with pro-longed enhancement of glucose disposal inmuscle and liver in the canine. J PharmacolExp Ther 2009;328:970–97519. Ciofetta M, Lalli C, Del Sindaco P, et al. Con-tribution of postprandial versus interprandialblood glucose to HbA1c in type 1 diabetes onphysiologic intensive therapy with lispro insulinat mealtime. Diabetes Care 1999;22:795–80020. Del Sindaco P, Ciofetta M, Lalli C, et al. Useof the short-acting insulin analogue lispro in

intensive treatment of type 1 diabetes mellitus:importance of appropriate replacement of basalinsulin and time-interval injection-meal. DiabetMed 1998;15:592–60021. Tamas G, Marre M, Astorga R, Dedov I,Jacobsen J, Lindholm A; Insulin Aspart StudyGroup. Glycaemic control in type 1 diabetic pa-tients using optimised insulin aspart or humaninsulin in a randomised multinational study. Di-abetes Res Clin Pract 2001;54:105–11422. Davis WA, KnuimanM, Kendall P, Grange V,Davis TME; Fremantle Diabetes Study. Glycemicexposure is associated with reduced pulmonaryfunction in type 2 diabetes: the Fremantle Di-abetes Study. Diabetes Care 2004;27:752–75723. Panel on Handling Missing Datain ClinicalTrials; National Research Council. The Preven-tion and Treatment of Missing Data in ClinicalTrials. Washington, DC, National AcademiesPress, 201024. Siddiqui O, Hung HM, O’Neill R. MMRM vs.LOCF: a comprehensive comparison based onsimulation study and 25 NDA datasets. J Bio-pharm Stat 2009;19:227–24625. Fineberg SE, Kawabata TT, Krasner AS,Fineberg NS. Insulin antibodies with pulmonarydelivery of insulin. Diabetes Technol Ther 2007;9(Suppl. 1):S102–S110

care.diabetesjournals.org Bode and Associates 2273