A Randomized Trial Comparing Inhaled Insulin Plus Basal Insulin Versus Usual Care in Adults With Type 1 Diabetes

Glycemic targets set by the American Diabetes Association (1) are met in only ∼20% of adults with type 1 diabetes (2,3). Even with automated insulin delivery (AID) systems, most users have glucose levels >180 mg/dL for ≥25% of the day (4). This inability to limit hyperglycemia is in part due to the delayed onset of action of boluses of rapid-acting analog (RAA) insulin given for a meal. The delay in effect of the RAA insulin bolus, even with use of insulins with faster onset of action (e.g., fast-acting aspart, insulin lispro-aabc), limits the ability to administer higher insulin doses due to the risk of postmeal hypoglycemia.

Inhaled Technosphere insulin (TI) (Afrezza; MannKind, Danbury, CT) is a dry-powder formulation of recombinant human insulin absorbed onto Technosphere microparticles for oral inhalation with a breath-powered inhaler. It has both a more rapid onset of action and more rapid dissipation than subcutaneous RAA insulin (5,6). Thus, TI can potentially reduce both postmeal hyperglycemia and the risk of later postmeal hypoglycemia (7,8).

Although TI was approved by the U.S. Food and Drug Administration (FDA) 10 years ago, it is not frequently prescribed despite its potential benefits. This in part may be due to the limited clinical trial data with use of a TI dose that is more bioequivalent to the RAA dose and substantially higher than what is recommended in the FDA label and the lack of data for comparing a regimen of TI combined with a long-acting basal insulin plus continuous glucose monitoring (CGM) for dosing determinations versus contemporary insulin delivery methods, including AID. An in-clinic standardized meal study showed that a TI dose approximately twice the dosing of the current FDA label reduced postmeal hyperglycemia more than the TI dose based on the FDA label (9). In a meal study performed at the baseline of the current study, this higher TI dose was shown to reduce the postmeal glucose excursion in comparison with RAA insulin following a standardized liquid meal (10).

To evaluate the efficacy and safety of TI, we conducted a randomized controlled trial (RCT) in adults with type 1 diabetes. The trial compared a regimen of TI plus a basal insulin (insulin degludec) versus a control group continuing their usual insulin delivery method consisting of AID, a nonautomated pump, or multiple daily insulin injections (MDI).

The randomized trial was conducted at 19 endocrinology practices in the U.S. (7 community-based and 12 academic centers). The protocol, which is available from clinicaltrials.gov (clinical trial reg. no. NCT05904743), was approved by a central institutional review board, and informed consent was obtained from each participant.

Major eligibility criteria included age ≥18 years old, type 1 diabetes for at least 6 months with the same treatment regimen for at least 3 months (AID, nonautomated pump, or MDI), HbA_1c <11.0%, and CGM use on a regular basis (at least 70% of the time in the prior 4 weeks). Key exclusion criteria included use of noninsulin glucose-lowering medications or inhaled insulin in the prior 3 months, asthma treatment in the prior 12 months, smoking in the prior 3 months, or history of lung cancer. (See Supplementary Table 1 for a complete listing of inclusion and exclusion criteria.)

After informed consent was obtained, participants were screened for eligibility. A blinded Dexcom G6 Pro sensor was worn for ∼14 days for collection of baseline CGM data. Eligible participants were randomly assigned (1:1) with use of a block design separately for each site to either the TI-degludec regimen or usual care (UC) group. The follow-up period for the RCT was 17 weeks. The first 4 weeks of the RCT were designated for insulin dose optimization in both groups. At baseline and after 17 weeks, HbA_1c was measured at a central laboratory, forced expiratory volume in 1 s (FEV₁) was assessed with the Vitalograph asma-1 respiratory monitor, a urine pregnancy test was performed for participants capable of becoming pregnant, CGM data for analysis were collected with a blinded Dexcom G6 Pro sensor, and patient-reported outcome surveys were completed. We assessed treatment adherence by querying participants at each visit with respect to frequency of missed insulin doses in the prior 7 days. The visit schedule and study procedures can be found in Supplementary Table 2.

Participants in the TI group were provided with Dexcom G7 sensors for use throughout the study. Degludec was injected once daily. The initial dose of degludec was based on the average daily amount of basal insulin being used and then titrated, with a fasting glucose of 90–120 mg/dL without hypoglycemia targeted (Supplementary Table 3).

A meal challenge was performed in clinic at which the first dose of TI was inhaled. This initial dose of TI was approximately twice the number of units of RAA insulin received for a given amount of carbohydrate or size of a meal, a conversion ratio higher than in the TI package insert, as described in Supplementary Table 3. Results of the meal challenge, which have previously been published (10), were used to establish the starting TI dose. The dose was titrated at study visits based on the 1-h postprandial CGM glucose values. Participants were instructed to administer a correction dose of TI as soon as 60–90 min after a previous TI dose if the glucose level was >140 mg/dL, including at bedtime, and if the glucose level was >200 mg/dL overnight (Supplementary Table 3).

Participants in the UC group were instructed to follow their usual diabetes management with respect to insulin delivery, using their personal insulin, and personal CGM and blood glucose meter use. They had the same visit schedule as the TI group and received diabetes management recommendations.

The primary outcome was HbA_1c at 17 weeks, with testing for noninferiority between groups. Secondary outcomes, prespecified to be tested in a hierarchical fashion to preserve the overall type 1 error, included CGM metrics (calculated from the blinded CGM sensor worn by both groups prior to the 17-week visit) for hypoglycemia with testing for noninferiority followed by tests of superiority for other CGM metrics and HbA_1c as described in Supplementary Table 4. Other outcomes included additional HbA_1c and CGM-measured end points, weight, and patient-reported outcomes. Safety outcomes included severe hypoglycemia, diabetic ketoacidosis, other serious adverse events, change in FEV₁, and certain adverse events of special interest, as listed in Supplementary Table 4. All outcomes were prespecified except for worsening of HbA_1c by >0.5% and by >1.0% and worsening of CGM-measured percent time spent with glucose in range 70–180 mg/dL by 5% and by 10%, which were added post hoc.

Sample size was computed for the primary end point of noninferiority for HbA_1c. Assuming a noninferiority margin of 0.4%, one-side α 0.025, power 90%, SD of 17-week HbA_1c (adjusted for the correlation with baseline) 0.6%, and 1:1 randomization, the calculated sample size was 98. The sample size was increased to 120 to account for dropouts and treatment nonadherence.

The statistical analysis plan can be found in Supplementary Material. For the primary analysis, HbA_1c was compared between groups with use of a linear mixed-effects regression model with adjustment for baseline HbA_1c and clinical site (random factor), with variance modeled separately for each treatment group and missing data handled with the direct likelihood method. A 95% CI was constructed on the treatment group difference, and the upper end of the CI was compared with 0.4% for the assessment of whether the prespecified noninferiority margin was met. A similar approach was used for other continuous end points. For binary HbA_1c and CGM outcomes, risk-adjusted percentages by treatment group were computed at 17 weeks from a logistic regression model, with adjustment for baseline value of the end point and clinical site using generalized estimating equations. Analyses were performed separately for prestudy AID system users and MDI users combined with nonautomated pump users.

To control the type 1 error, we used an iterative hierarchical approach for the primary and key secondary end points as listed in Supplementary Table 4. For the other efficacy analyses, the false discovery rate was controlled with the adaptive Benjamini-Hochberg procedure and reflected in the CIs constructed around the treatment group difference point estimates. SAS, version 9.4, was used for the analyses.

Between 27 July and 20 November 2023, 123 participants were randomly assigned to the TI group (N = 62) or the UC group (N = 61). Mean ± SD age was 45 ± 15 years (range 18–77); 54% were female, 89% of White race, and 11% of Hispanic ethnicity. Mean HbA_1c was 7.6% ± 0.9%. An AID system was being used by 48%, MDI by 44%, a nonautomated pump by 6%, and a pump with predictive low glucose suspend feature by 2%. Participant characteristics appeared balanced between groups (Table 1).

The 17-week primary outcome visit was completed by 58 of the 62 participants (94%) in the TI group and by 57 of the 61 (93%) in the UC group (Supplementary Fig. 1) (last completed visit on 26 March 2024). Among the trial completers, the visit completion rate for the eight protocol follow-up visits was 99% in each group.

In addition to the four participants in the TI group who dropped from the trial, seven other participants completed the trial but discontinued the TI regimen and resumed their prestudy insulin delivery method prior to 17 weeks (Supplementary Table 5).

Mean ± SD HbA_1c was 7.57% ± 0.97% at baseline and 7.62 ± 1.06% at 17 weeks in the TI group and 7.59% ± 0.80% and 7.54% ± 0.77%, respectively, in the UC group (adjusted difference 0.11%, 95% CI −0.10 to 0.33, P value for noninferiority = 0.01 for 0.4% margin) (Table 2). The results were similar to those of the primary analysis in the per-protocol cohort (treatment group difference 0.04%, 95% CI −0.16 to 0.25) (Supplementary Table 6) and in a cohort restricted to participants with baseline HbA_1c >7.0% (treatment group difference 0.05%, 95% CI −0.34 to 0.45) (Supplementary Table 7). Various sensitivity analyses performed for assessment of the robustness of the results with different assumptions for handling missing data produced similar results (data not shown).

HbA_1c improved from baseline to 17 weeks by >0.5% (5.5 mmol/mol) in 12 (21%) in the TI group vs. 3 (5%) in the UC group and worsened by >0.5% (5.5 mmol/mol) in 15 (26%) in the TI group vs. 2 (3%) in the UC group (Table 3). At 17 weeks, 17 (30%) in the TI group vs. 10 (17%) in the UC group had HbA_1c <7.0% (53 mmol/mol). Among participants with baseline HbA_1c >7.0% (53 mmol/mol), 8 (21%) in the TI group compared with 0 in the UC group had a 17-week HbA_1c level <7.0% (53 mmol/mol). The wider spread of HbA_1c outcomes with greater numbers improving and worsening in the TI group than the UC group is evident in Supplementary Fig. 2. HbA_1c results according to AID use in the UC group are shown in Tables 2 and 3. Analyses in other subgroups did not suggest a differing treatment effect according to baseline characteristics (Supplementary Fig. 3).

CGM-measured hypoglycemia was low at baseline and showed little change at 17 weeks in both groups, meeting the prespecified criterion for noninferiority for both time spent with glucose <54 mg/dL (treatment group difference 0.0%, 95% CI −0.3 to 0.3, P value for noninferiority 0.002) and time <70 mg/dL (treatment group difference 0.1%, 95% CI −0.7 to 0.8, P value for noninferiority <0.001) (Table 2).

Mean ± SD percent time spent with glucose in range 70–180 mg/dL during the daytime was 52% ± 17% at baseline and 53% ± 21% at 17 weeks in the TI group and 51% ± 18% and 50% ± 17%, respectively, in the UC group (adjusted difference 2%, 95% CI −3 to 8). Other CGM metrics reflective of hyperglycemia also showed no meaningful differences between groups (Table 2 and Supplementary Tables 8 and 9). Time in range 70–180 mg/dL was >70% at 17 weeks in 24% of the TI group and in 13% of the UC group; 25% vs. 14%, respectively, had improvement in time in range of ≥10% from baseline, and 31% vs. 20% had worsening ≥10% (Table 3).

During daytime, there were no meaningful differences observed between groups in CGM metrics (Supplementary Table 10). During nighttime, there was a suggestion of more hyperglycemia and lower percent time in range 70–180 mg/dL in the TI group compared with the UC group, particularly in TI participants who were using an AID system prior to the study and switched to the TI regimen during the trial. As seen in Fig. 1, during the day from approximately 9:00 a.m. to 9:00 p.m., mean glucose levels appear similar with TI-degludec versus AID, but starting at 9:00 p.m. and overnight, mean glucose was lower with AID. Mean glucose appeared slightly lower throughout the 24 h of the day in TI-degludec users compared with that in users of MDI or a nonautomated pump. In all three of these groups, mean glucose appears higher than what might be expected based on the 17-week HbA_1c levels.

No significant differences were observed between treatment groups on any of the participant-reported outcome surveys (Supplementary Table 11).

Overall, 124 adverse events were reported in 48 (77%) participants in the TI group and 83 events in 39 (64%) participants in the UC group (Supplementary Tables 12 and 13). A severe hypoglycemia event occurred in one participant in the TI group (related to excessive alcohol ingestion many hours after the last TI dose) and in no participants in the UC group. There were no cases of DKA. Two hospitalizations (appendectomy, postsurgery hyperglycemia) occurred in the UC group and none in the TI group.

In the TI group, there were 31 adverse events in 23 (37%) participants considered possibly, probably, or definitely related to the TI-degludec regimen (Supplementary Table 13). TI-associated cough, which was usually transient and mild, was reported by 14 (23%) of the participants, 2 of whom permanently discontinued TI. Shortness of breath reported by five participants, wheezing by two, and bronchospasm by one were considered possibly related to TI; two dropped from the study while the other six continued to use TI. Overall, eight participants discontinued TI related to side effects.

Mean ± SD FEV₁ was 2.92 ± 0.75 L at baseline and 2.84 ± 0.77 L at 17 weeks in the TI group and 2.93 ± 0.72 and 2.92 ± 0.66 L, respectively, in the UC group (difference −0.08, 95% CI −0.19 to 0.03, P = 0.16). No participants had a decrease in FEV₁ exceeding 20%, which was the prespecified level for classification as an adverse event.

Mean ± SD ratio of the investigator-determined initial dose of TI to the participant’s prerandomization RAA dose for a standardized meal was 1.8 ± 0.3 at baseline, and the ratio increased after dose titration to 2.8 ± 1.9 at 17 weeks. Supplementary Tables 15 and 16 provide the total daily dose of basal and bolus insulin in each treatment group.

At one or more of the 8-, 13-, or 17-week visits, nine (15%) participants in the TI group and eight (13%) participants in the UC group reported missing three or more bolus doses or two or more basal insulin doses in the prior week. At the 17-week visit, participants in the TI group indicated that their daily average number of correction doses over the prior week was as follows: zero to one, 20% of participants; two to three, 51%; four to five, 22%; and more than five, 8%.

Mean ± SD weight change from baseline was 0.1 ± 2.8 kg in the TI group and 1.4 ± 3.1 kg in the UC group (difference −1.3 kg, 95% CI −2.4 to −0.2) (Supplementary Table 17).

In this 17-week RCT of adults with type 1 diabetes, change in HbA_1c was noninferior with a regimen of TI for boluses plus once daily degludec as basal insulin in comparisons with a UC control group where participants continued with their prestudy insulin delivery method, consisting primarily of AID or MDI. Both the TI group and the UC group showed little mean change from baseline in HbA_1c. This lack of change in mean HbA_1c has different connotations in the two groups. Whereas in the UC group there were few participants with substantial improvement or worsening of HbA_1c from baseline, in the TI group there were approximately similar numbers of participants who had a substantial HbA_1c improvement (>0.5% [5.5 mmol/mol]) or worsening (>0.5% [5.5 mmol/mol]).

Daytime glucose levels, measured with CGM, were similar with the TI-degludec regimen and with AID use in the UC group. However, overnight hyperglycemia was greater with the TI-degludec than AID. This pattern started at ∼9:00 p.m. suggesting that participants may have underdosed TI at dinner and not redosed with a correction dose during the evening or at bedtime, perhaps due to concern about overnight hypoglycemia, although little overnight hypoglycemia was observed. The greater beneficial effect of AID overnight compared with during the daytime has been well established (11).

Several aspects of TI dosing differed from RAA dosing. First, with TI, participants were instructed to dose at the start of the meal, whereas with RAA insulin, dosing 5–15 min prior to the meal is the standard recommendation. Second, in view of the shorter duration of effect of TI compared with RAA, participants were instructed to redose TI 60–90 min after a meal dose if the glucose level was >140 mg/dL, which is not advisable with RAA, since the meal bolus will still be active for several hours. Such a redosing schedule was shown to be safe in the current study as well as a prior study by Akturk et al. (7). Third, the number of units in an optimized dose of TI was on average two to three times the dose of RAA. The current U.S. labeling of TI is to base initial dosing on an approximate 1:1 conversion of the usual RAA dose for a patient. However, this is based on the amount of powdered insulin in a cartridge and not on the amount that is actually absorbed in the lungs. Thus, the appropriate, bioequivalent dose of TI should be at least twofold higher than the RAA dose for most patients.

With respect to safety, the frequency of CGM-measured hypoglycemia was low and similar between groups during both the daytime and nighttime, meeting prespecified noninferiority criteria. Side effects were more frequently reported in the TI group than UC group. The most common side effect was a brief cough associated with TI inhalation, with a frequency similar to that reported in prior studies (12). Pulmonary effects were infrequent, and none were serious; in most such cases the participant continued to use TI. There was a slight numerical reduction in FEV₁ that was not statistically significant. There was one severe hypoglycemia event in the TI group related to excessive alcohol intake.

To put these trial results in perspective, it is important to consider the key differences in study design between this trial and the trial reported by Bode et al. (8) that formed the basis for the FDA approval. First, the initial TI conversion dose in the current trial was much higher than in the earlier trial. The FDA label for TI, which was based on the study dosing from the work of of Bode et al., indicates that the starting initial TI dose should be 1.0–1.3 times the RAA dose, for most of the dose range, whereas clinical experience is that this dose is too low and a higher dose is needed. Second, CGM was a key component of diabetes management and TI dosing in the current trial but not the previous trial. Third, in the current trial we measured CGM metrics for hyperglycemia and hypoglycemia during the daytime and nighttime, which were not available in the previous trial. Fourth, the cohorts in the trials were quite different, with just MDI users in the prior trial and both MDI and AID users in the current trial. Fifth, the inclusion of AID users provided the ability in the current study to compare the TI-degludec regimen with AID.

Strengths of the study include a multicenter randomized trial design and inclusion of both MDI and AID system users. Limitations include the cohort being predominantly White, although this is consistent with the profile of people with type 1 diabetes, and length of the follow-up period for the RCT being only 17 weeks, although additional information from longer duration of TI use will be available from a 13-week extension phase that followed the RCT, reporting for which will be in a separate manuscript.

In conclusion, the study demonstrated that change in HbA_1c in adults with type 1 diabetes who switched to a regimen of TI and degludec was noninferior to HbA_1c change in a UC control group using predominately either AID or MDI for insulin delivery. The study results are striking in showing that approximately similar numbers of participants had substantial improvement or worsening of HbA_1c in switching to TI. Successful use of TI requires considerable engagement in glycemic self-management, with frequent TI inhalation during the day and evening. When this is done, substantial glycemic improvement is possible even in switching from an AID system. However, when this is not the case, there can be glycemic worsening as observed in a subset of study participants. Thus, while the study results support consideration of TI as a viable option for patients with type 1 diabetes, patient selection with respect to prescribing TI is important. Future work is needed, for better identification of the profile of patients who may be ideal candidates for TI as well as those for whom a switch to TI is unlikely to be successful, and for establishment of best clinical practices for optimal titration of TI dosing.

Acknowledgments. J.B.B. is an editor of Diabetes Care but was not involved in any of the decisions regarding review of the manuscript or its acceptance.

Funding. The trial was funded byMannKind through a contract with the JAEB Center for Health Research, which then provided funding to the clinical sites. Representatives of MannKind participated in protocol development with the investigators and were involved with the JAEB Center for Health Research in study oversight. MannKind reviewed the manuscript and provided comments.

Duality of Interest. I.B.H. reports consulting for Abbott, Roche, Hagar, and Vertex Pharmaceuticals and research support from MannKind, Dexcom, and Tandem Diabetes Care. R.W.B. reports that his institution has received funding on his behalf as follows: grant funding, study supplies, and consulting fees from Insulet, Tandem Diabetes Care, and Beta Bionics; grant funding and study supplies from Dexcom; grant funding from Bigfoot Biomedical; study supplies from Medtronic, Ascencia, and Roche; consulting fees and study supplies from Eli Lilly and Novo Nordisk; and consulting fees from embecta, Sequel Med Tech, Vertex Pharmaceuticals, Hagar, Ypsomed, Sanofi, and Zucara Therapeutics. Y.K. reports receiving product support from Tandem Diabetes Care, Insulet, and Abbott and acting on the advisory board for Tandem Diabetes Care, Novo Nordisk, and Vertex Pharmaceuticals. H.K.A. received grant payments for research from the following companies: Boehringer Ingelheim, Abbott Diabetes Care, AbbVie, Covance, Dexcom, Eli Lilly, Madrigal Pharmaceuticals, Medtronic, Novo Nordisk, MannKind, Carmot Therapeutics, Quintiles, Akero Therapeutics, 89bio, Viking Therapeutics, PPD, Zydus, Kowa Pharmaceuticals America, Insulet, and VtV Therapeutics. G.A. has received consulting fees from Dexcom and Insulet. G.A. has received research support to her institution, Northwestern University, from Fractyl Health, Insulet, Tandem Diabetes Care, and Welldoc. T.B. reports participating in the speakers bureau for Eli Lilly, Novo Nordisk, Boehringer Ingelheim, AstraZeneca, and Dexcom; receiving research support from Eli Lilly, Novo Nordisk, Boehringer Ingelheim, Roche, Abbott, Tandem Diabetes Care, MannKind, Insulet, Medtronic, AbbVie, VtV Therapeutics, and Biomea Fusion; and acting on advisory boards for Insulet and MannKind. C.J.L. reports research support paid to her institution from Dexcom, Tandem Diabetes Care, Insulet, Novo Nordisk, and Abbott Diabetes Care and consulting fees from Dexcom and Tandem Diabetes Care. K.C. receives research support provided to her institution from Dexcom, Abbott, Medtronic, Eli Lilly, MannKind, and Insulet and receives consulting fees from Dexcom. A.M. reports acting on the scientific advisory board and participating in speaking engagements for MannKind; acting as a clinical trial principal investigator and participating in speaking engagements for Novo Nordisk, Sanofi, Eli Lilly, Boehringer Ingelheim, and AstraZeneca; participating in speaking engagements for Janssen Pharmaceuticals; acting as a chief clinical principal investigator, acting on the scientific advisory board, and participating in speaking engagements for Medtronic; acting on the scientific advisory board and acting as a principal investigator for AbbVie; and participating in speaking engagements for Intuity Medical. D.P. reports support from Novo Nordisk, Eli Lilly, MannKind, AstraZeneca, and Boehringer Ingelheim. D.W.S. reports consulting for Abbott Diabetes Care and research support to his institution from the National Institute of Diabetes and Digestive and Kidney Diseases, Tandem Diabetes Care, Novo Nordisk, and Abbott Diabetes Care. R.S.W. reports participation in multicenter clinical trials, through her institution, sponsored by Tandem Diabetes Care, Insulet, Diasome, Eli Lilly, and Amgen, and use of DexCom devices obtained at reduced cost for clinical research. B.W.B. reports stock ownership in Aseko (Glytec); consulting for Beta Bionics, Biomea Fusion, Eli Lilly, Medtronic, MannKind, Novo Nordisk, and Sensorics; participating in the speakers bureau of Abbott, Eli Lilly, Medtronic, MannKind, Novo Nordisk, Sanofi, and Xeris; and receiving grant and research support through his employer from Abvance Therapeutics, Abbott, Dexcom, Diasome, Dompé farmaceutici, Insulet, Eli Lilly, MannKind, Medtronic, Novo Nordisk, Sanofi, Senseonics, REMD Biotherapeutics, Xeris, and vTv Therapeutics. O.H. reports research support from Novo Nordisk, Eli Lilly, and MannKind and acting on an advisory board for Abbott Nutrition. M.K. reports participating in clinical trials for the following sponsor companies: 89bio, AbbVie, Akero Therapeutics, Amgen, AstraZeneca, Biolinq, Biomea Fusion, Boehringer Ingelheim, Corcept Therapeutics, Diamyd Medical, Dexcom, Eli Lilly, Endogenex, Fractyl Health, Gilead, Insulet, Ionis Pharmaceuticals, Kowa, MannKind, Medtronic, Novo Nordisk, Pfizer, Reata Pharmaceuticals, Sinocare, Tandem Diabetes Care, vTv Therapeutics, Zucara Therapeutics, and Zydus. M.K. participates in an advisory board for Corcept Therapeutics and a clinical events committee for Capillary Biomedical. C.J. reports consulting, acting on an advisory board for, and speaking on insulin pumps for Medtronic. J.B.B. reports research support from Bayer, Boehringer Ingelheim, Carmot Therapeutics, Corcept Therapeutics, Dexcom, Eli Lilly, Insulet, MannKind, Novo Nordisk, and vTv Therapeutics; consulting fees from Alkahest, Altimmune, Anji Pharmaceuticals, Aqua Medical, AstraZeneca, Boehringer Ingelheim, CeQur, Corcept Therapeutics, Eli Lilly, embecta, GentiBio, Glyscend, Insulet, Mediflix, Medscape, Medtronic, Mellitus Health, Metsera, Moderna, Novo Nordisk, Pendulum Therapeutics, Praetego, ReachMD, Stability Health, Tandem Diabetes Care, Terns Pharmaceuticals, and Vertex Pharmaceuticals; and stock options from Glyscend, Mellitus Health, Pendulum Therapeutics, Praetego, and Stability Health. K.R.K. has received consulting fees from Novo Nordisk. K.R.K. has received research-related contracts (paid to the institution) from Bayer, Boehringer Ingelheim, Carmot Therapeutics, Diasome, Eli Lilly, Novo Nordisk, Rhythm Pharmaceuticals, and vTv Therapeutics. G.O. has received research support paid to her institution from Dexcom, Abbott Diabetes Care, Tandem Diabetes Care, Insulet, and Novo Nordisk. No other potential conflicts of interest relevant to this article were reported.

Regarding the role of the funder, the manuscript was written by the authors without a medical writer and the content is solely the responsibility of the authors. MannKind had no rights of approval of the content of the manuscript or the decision on journal submission.

Author Contributions. I.B.H. and R.W.B. wrote the first draft of the manuscript. M.C.M. and P.C. wrote the statistical analysis plan and conducted the analyses. All other authors had a key role in the study, provided review and critical input into the writing of the manuscript, and approved the manuscript. I.B.H. and R.W.B. are the guarantors of this work and, as such, had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Prior Presentation. Parts of this study were presented in abstract form at the 84th Scientific Sessions of the American Diabetes Association, Orlando, FL, 21–24 June 2024.

Handling Editors. The journal editor responsible for overseeing the review of the manuscript was Mark A. Atkinson.