Meta-Analysis: Efficacy and Safety of Inhaled Insulin Therapy in Adults With Diabetes Mellitus

Ceglia L, Lau J, Pittas AG. Ann Intern Med. 2006;145:665-675.

Despite the availability of effective diabetes therapies, less than half of US adults suffering from the disease attain a hemoglobin A_1C (A1C) level of <7%. Although insulin is an effective treatment for lowering hyperglycemia, there is considerable resistance to its use among both patients and healthcare providers, often related to the need for subcutaneous (sc) injections. As a result, individuals with diabetes may be hesitant to initiate insulin therapy, or maintain it for prolonged periods of time. The US Food and Drug Administration (FDA) approved the first inhaled form of insulin on January 27, 2006, for use among nonsmoking adults with no pulmonary disease who have either type 1 diabetes mellitus (T1DM) or type 2 (T2DM). This approval provided the first new insulin delivery option since the discovery of insulin treatment in the 1920s. Compliance with inhaled insulin, along with its efficacy and safety as a premeal, noninvasive alternative to subcutaneous (sc) regular or rapid-acting insulin, has been studied in depth.

A systematic search of MEDLINE from 1966 through June 22, 2006, and the Cochrane Controlled Clinical Trials Register from the second quarter of 2006 was conducted for English-language, randomized, controlled, and parallel or crossover trials of inhaled insulin in nonpregnant adults with T1DM or T2DM. The search terms employed were "diabetes", "blood sugar", "hyperglycemia", "glucose", "glycohemoglobin", "hemoglobin", "A1C", "inhaled", "inhalation", "aerosol", "pulmonary", "insulin", "human", and "clinical trial". Further searches were conducted for citation sections of the articles recovered and additional publications in personal reference lists. Unpublished studies of inhaled insulin were reviewed after accessing the briefing document for Exubera^® powder, which was released under public disclosure laws to the FDA advisory committee by Pfizer Inc on September 8, 2005. Two independent reviewers and/or a group conference resolved any discrepancies. Data included in the analyses were baseline characteristics of study participants, types of interventions, efficacy of glycemic control as measured by both change in A1C from baseline and the proportion of patients achieving A1C <7%, and nonglycemic outcomes, such as changes in body weight and overall patient satisfaction. Safety was evaluated by extracting data reflecting severe hypoglycemia, pulmonary symptoms, levels of circulating insulin antibodies, changes in pulmonary function variables (1-second forced expiratory volume [FEV₁]), and changes in diffusing capacity of carbon monoxide (DL_co). Because various methods were employed for reporting hypoglycemia in individual trials, the FDA definition was used to determine the incidence of hypoglycemia, ie, blood glucose levels ≤2 mmol/L (≤36 mg/dL) or those that created a need for assistance. All studies included in this analysis were open label, and the authors of this paper used allocation generation (proper randomization), intention-to-treat analysis, dropout rate, and whether the studies were designed for glycemic efficacy (eg, "noninferiority" or "superiority") or safety to assess their quality.

A total of 16 trials involving 4023 participants was included in this analysis, and A1C levels at baseline were 7% to 9.8% while body mass index (BMI) ranged from 24 kg/m² to 32 kg/m². The trials selected compared inhaled with sc regular insulin, insulin lispro, or insulin aspart among 1534 patients with T1DM, along with inhaled insulin versus regular sc insulin or oral hypoglycemic agents in 2489 patients with T2DM. The sc insulin regimens in the comparisons between patients with T1DM and T2DM included a combination of basal insulin (neutral protamine Hagedorn insulin, ultralente insulin, or insulin glargine given once or twice per day) and bolus insulin (regular insulin, insulin lispro, or insulin aspart given 2 or 3 times per day). Oral hypoglycemic treatment regimens included sulfonylureas, metformin, and thiazolidinediones taken alone or in combination with 1 or 2 other oral medications.

The analysis revealed that among patients with T1DM or T2DM, there was a small but statistically significant difference in the decrease in A1C from baseline of 0.08% (CI, 0.03% to 0.14%) that favored sc insulin. However, that difference was only reported in 2 trials of 104 weeks“ duration, and none was found between sc and inhaled insulin in studies that were <24 weeks. There were no statistically significant differences reported in the 5 trials designed to assess the number of patients with T1DM or T2DM who achieved A1C <7% when taking inhaled versus sc insulin (27.1% vs 24.6%, respectively). Comparisons of studies that measured inhaled insulin versus oral hypoglycemic agents among patients with T2DM revealed that inhaled insulin lowered A1C more effectively (weighted mean difference, –1.04% [CI, –1.59% to –0.49%]), and resulted in these patients being more likely to achieve A1C <7% (risk ratio, 1.87 [CI, 1.07 to 3.25]; 30.9% vs 16.9%, respectively).

Among patients with either T1DM or T2DM, there were no differences in weight gain between the inhaled and sc insulin groups (range, 0.2 to 1.5 kg). In contrast, inhaled insulin caused greater weight gain than did combinations of oral therapies (adjusted mean difference, 1.85 kg [CI, 0.98 kg to 2.73 kg]), although there was no difference between insulin and rosiglitazone monotherapy (adjusted mean difference, 0.95 kg [CI,–0.18 kg to 2.09 kg]). A higher proportion of patients with T1DM (75% taking inhaled insulin versus 78% taking sc insulin) versus those with T2DM (16% taking inhaled insulin versus 18% taking sc insulin) reported severe hypoglycemia; however, there were no differences between inhaled and sc insulin in the proportion of patients with T1DM or T2DM reporting ≥1 episode of severe hypoglycemia (risk ratio, 1.00 [Cl, 0.95 to 1.04]). More patients treated with inhaled insulin than with oral agents reported at least 1 episode of severe hypoglycemia (risk ratio, 3.06 [Cl, 1.03 to 9.07]: 9.4% vs 3.5%, respectively).

Nonproductive cough was the most commonly reported pulmonary symptom among patients who received inhaled insulin. Although it was reported more frequently by those treated with sc insulin or oral agents (risk ratio, 3.52 [CI, 2.23 to 5.56]; 16.9% vs 5.0%, respectively), there were no differences between patients with T1DM or T2DM. Cough occurred rapidly after administration of inhaled insulin, usually within seconds to minutes, was mild and not associated with changes in pulmonary function, was noted within the first month of treatment, and both frequency and severity diminished over time. Patients treated with inhaled insulin had a decrease in FEV₁ from baseline that was statistically significant compared with patients with T1DM who received sc insulin and those with T2DM who were given oral hypoglycemic agents. Inhaled insulin was also associated with a greater decrease in DL_CO from baseline compared with sc insulin among patients with T1DM (weighted mean difference, -0.902 mL/min per mm Hg [Cl, -1.546 to -0.258 mL/min per mm Hg]). Although this was evident in studies that were ≤24 weeks, there was no difference found in a 2-year study.

Combined data from all inhaled insulin trials revealed that 2% of patients with T1DM and 2.3% of those with T2DM discontinued treatment as a result of respiratory events, compared with 0% and 0.1% of patients taking sc insulin or oral hypoglycemic agents, respectively. Increased levels of insulin antibodies were observed following administration of inhaled versus sc insulin, especially among patients with T1DM. The levels of insulin antibodies measured were not associated with any specific clinical outcomes, including dosing requirements, glycemic control, or adverse outcomes, such as allergic events, pulmonary side effects, or hypoglycemia. Each of the 4 trials that measured patient satisfaction among T1DM or T2DM groups reported that it was significantly greater for those receiving inhaled versus sc insulin, and specific reasons included ease of administration, comfort, convenience, mealtime flexibility, and ease of administering multiple doses per day. Two of these studies reported statistically significant improvements in overall quality of life (QoL). In addition, patients that were randomly assigned to inhaled insulin were more likely to continue with this treatment than to voluntarily return to sc insulin.

This evidence-based assessment of randomized, controlled trials was designed to measure the efficacy and safety, along with patient preferences, of inhaled versus sc insulin and revealed that this new delivery option can enhance the QoL of adult patients with T1DM or T2DM. Patients were not only more satisfied with inhaled insulin, they preferred it to sc insulin, and therefore this treatment option could increase the proportion of patients who would choose insulin and adhere to such therapy. Specific advantages of inhaled insulin include reduction of glycemia comparable to sc regular insulin, and greater decreases than those obtained by patients taking oral agents. In fact, an equivalent proportion of patients receiving inhaled and sc insulin achieved A1C <7%. However, the authors of this paper noted that since there are no data demonstrating a superiority of inhaled to sc insulin, it is too early to assume that patients who use it to replace current sc insulin therapy can expect the same glycemic control. In addition, because dose adjustment schedules for inhaled-insulin delivery devices have not been perfected, small, incremental changes in dosing need to occur to avoid hypoglycemia. The most promising aspect of inhaled insulin is that it provides a noninvasive option that could increase patient acceptability of insulin treatment. Therefore, long-term efficacy and safety trials should be conducted in the future to provide further information about its benefits for the treatment of diabetes.