Is There a Place for Inhaled Insulin in the Era of Automated Insulin Delivery?

Individuals with type 1 diabetes (T1D) require lifelong insulin therapy. The pain and inconvenience of multiple daily insulin injections (MDI) often affect adherence and quality of life, increasing the risk of long-term diabetes complications associated with suboptimal glucose control (1,2). Repeated subcutaneous insulin injections may also lead to localized lipohypertrophy interfering with insulin absorption (3). Although rapid-acting analogs (RAAs) have improved pharmacokinetic profiles, current formulations still fall short of fully mimicking the physiological prandial insulin response and must be given before meals (4). Inhaled insulin is a promising alternative route for insulin delivery. The first inhaled insulin, Exubera, was approved by the U.S. Food and Drug Administration (FDA) in 2006 but ultimately withdrawn from the market due to commercialization challenges (5). Technosphere insulin (TI) (marketed as Afrezza; MannKind) received FDA approval in 2014 for use in both T1D and type 2 diabetes (6). In the previous Affinity 1 randomized controlled trial, 355 adults with T1D on MDI at baseline received prandial TI versus insulin aspart with basal insulin. At 24 weeks, the HbA_1c in the TI group was noninferior to that in insulin aspart patients (between-treatment difference 0.19%, 95% CI 0.02–0.36) (6).

TI is a dry powder formulation of regular human insulin adsorbed onto Technosphere microparticles for oral inhalation. On inhalation, these microparticles reach the deep lung, where they dissolve rapidly because of the physiological pH, allowing absorption of insulin and fumaryl diketopiperazine into the systemic circulation with a time to maximum serum insulin concentration of approximately 12–15 min (7). In a phase 1 study where TI was compared with insulin lispro in patients with T1D, a faster median time to maximum concentration (8 vs. 50 min) was seen with TI, with a return to baseline levels more quickly (180–240 vs. 280 min) (4). Faster absorption and clearance of TI leads to a faster onset, shorter duration of action, and rapid suppression of endogenous glucose production (8), such that it is theoretically more effective in meeting prandial insulin needs. Its rapid absorption facilitates flexibility in administration timing, allowing it to be taken at the start of the meal, with the option for repeat dosing 60–90 min later (4).

In this issue of Diabetes Care, Hirsch et al. (9) report the results of INHALE-3, a phase 4 open-label randomized controlled trial assessing the efficacy and safety of TI in combination with insulin degludec compared with usual care in adults with T1D across 19 centers in the U.S. Nearly 50% of the study participants were on automated insulin delivery (AID) prestudy. Among 123 enrolled participants, 62 received TI with insulin degludec (TI group), while 61 continued their usual care (UC), which included AID for 48%, nonautomated pumps for 10%, and MDI for 43% (Fig. 1). The primary outcome, HbA_1c after 17 weeks, was similar between the TI-degludec regimen and UC (adjusted difference 0.11%, 95% CI 0.10–0.33; P = 0.01 for 0.4% noninferiority margin). Throughout the 24-h period, the percent time spent with glucose in target range (TIR), 70–180 mg/dL, was higher for the TI group compared with that for MDI or nonautomated pump users in the UC group (Fig. 1). In comparisons with AID users in the UC group, however, findings showed similar daytime percent TIR but lesser nocturnal percent TIR 70–180 mg/dL in the TI group (Fig. 1). Frequency of continuous glucose monitoring (CGM)-measured hypoglycemia (<70 and <54 mg/dL) was low and similar among groups during both daytime and nighttime. While overall HbA_1c levels were comparable between the treatment arms, a larger proportion (24% vs. 0%) of the TI group who were previously on MDI or nonautomated pumps achieved HbA_1c <7.0%. (30% vs. 4%) and >70% TIR 70–180 mg/dL. The HbA_1c <7.0% achievement surpassed that in the Affinity 1 study (TI 18.3% vs. aspart 30.7%), where the two treatments were also compared in individuals with T1D (6).

The findings of INHALE-3 are notable in several respects, first being the high baseline use of contemporary diabetes technologies, especially AID. All participants were on personal continuous real-time CGM for the study duration. This allowed optimization of the TI dose, which was titrated based on 1-h CGM glucose in the first 4 weeks. The bioequivalent dose of TI in the current trial is closer to two times the RAA dose, which is higher than the 1.0- to 1.3-fold conversion in the originally FDA-approved label. A previous proof-of-concept dosing study also demonstrated that doubling the TI dose significantly reduced postprandial glucose levels without new safety concerns (10). The baseline meal challenge also confirmed that this higher TI dose was more effective in reducing postmeal hyperglycemia, especially in prestudy MDI users (11). Collectively, these findings suggest the need for and upward adjustment in the dose conversion algorithm, although larger studies are needed to refine the optimal conversion from RAA to TI. The short trial duration of 17 weeks limits long-term assessment, though results of the 13-week extension phase will be reported in future.

Interestingly, a higher proportion in the TI group showed improvements or worsening of HbA_1c by >0.5% in comparison with the UC group. In the TI group, 26% had HbA_1c worsened by >0.5%, compared with 3% in the UC group, while ∼21% had improved HbA_1c in the TI group compared with 5% in the UC group. Studies are warranted to investigate whether these individual differences relate to insulin sensitivity or user factors. The timing of administration and frequency of inhaled correction doses might differ between users. In this trial, the self-reported frequency of TI correction doses was two to three times per week for 51% of patients at week 17. Pulmonary tolerability however, continues to be of concern. In the current trial, the incidence of mild cough (21%) was similar to that in previous studies. Eight participants discontinued TI treatment due to treatment-related side effects. While the decline in pulmonary function is nonprogressive, the potential elevated risk of lung cancer (12) warrants further long-term surveillance, especially at higher TI doses. Approximately one-half of the trial participants expressed willingness to continue with inhaled insulin posttrial, suggesting its acceptability among a subset of patients.

Insulin remains a crucial, lifesaving medication for individuals with T1D and insulin-dependent diabetes. Noteworthy advancements are being made in development of targeted insulins and innovative delivery systems (13). Despite several promising phase 2 and phase 3 studies, there are no approved oral insulins currently available (14). The potential benefits of TI are still being explored in various clinical settings. Combining inhaled insulin with AID systems could enhance treatment flexibility. For example, in a small pilot study in nine individuals, a fixed preprandial dose of TI further improved TIR during the 5-h postprandial period by 21.6% and reduced postprandial glucose peak by 33 mg/dL without increasing hypoglycemia (15). The ongoing TI safety and pharmacokinetics study in pediatric T1D (clinical trial reg. no. NCT02527265, clinicaltrials.gov) is expected to report in 2025. Long-term pulmonary safety data will be particularly relevant if higher doses are to be used in younger people.

In conclusion, TI offers advantages—faster onset, freedom from subcutaneous insulin administration, convenience with dosing with meals—though based on current evidence may not surpass AID in terms of glucose outcomes. Identifying the right responder and research on optimal dosing algorithms, with confirmation of pulmonary safety, will be crucial for its future development.