Effects of Adding Linagliptin to Basal Insulin Regimen for Inadequately Controlled Type 2 Diabetes: A ≥52-week randomized, double-blind study

This randomized, double-blind, placebo-controlled, phase III study was conducted in 167 centers in 19 countries (Argentina, Belgium, Brazil, Canada, Czech Republic, Finland, Germany, Greece, Italy, Korea, Mexico, the Netherlands, Norway, Peru, Russia, Slovakia, Spain, Taiwan, and the U.S.).

Patients were eligible if they were ≥18 years of age with a diagnosis of type 2 diabetes, had inadequate glycemic control (HbA_1c ≥7.0% [53 mmol/mol] to ≤10.0% [86 mmol/mol]), had a BMI of ≤45 kg/m², and were receiving treatment with basal insulin, alone or in combination with metformin and/or pioglitazone, for ≥12 weeks. Acceptable basal insulins were insulin glargine, insulin detemir, and neutral protamine Hagedorn insulin. The total prescribed insulin dose must not have changed by >10% of the baseline value during the 12 weeks before randomization.

Patients were ineligible if they had uncontrolled fasting hyperglycemia (glucose >13.3 mmol/L during placebo run-in); a myocardial infarction, stroke, or transient ischemic attack within 6 months before informed consent; impaired hepatic function (either alanine transaminase, aspartate transaminase, or alkaline phosphate >3 times the upper limit of normal); previous gastric bypass surgery; or any medical history of cancer (except basal cell carcinoma) in the 5 years before screening. Further exclusion criteria included hypersensitivity or allergy to the investigational products; contraindications to metformin or pioglitazone; treatment with rosiglitazone, sulfonylureas, glucagon-like peptide 1 analogs, DPP-4 inhibitors, or antiobesity drugs within the 3 months before informed consent; a history of alcohol or drug abuse in the previous 3 months; and current treatment with systemic steroids or change in dosage of thyroid hormones within 6 weeks before informed consent. Premenopausal women who were nursing, pregnant, or not practicing an acceptable method of birth control were also ineligible.

The trial protocol was approved by the independent ethics committees or institutional review boards of all participating centers. The study was carried out according to the principles of the Declaration of Helsinki and the International Conference on Harmonization Guideline for Good Clinical Practice. All patients gave written, informed consent before participation.

Patients underwent a 2-week, open-label placebo run-in period to confirm their eligibility after the initial screening and to exclude those who were nonadherent. After this placebo run-in period, eligible patients were randomly assigned (1:1 ratio) to receive double-blind linagliptin 5 mg once-daily or placebo in addition to continued basal insulin for at least 52 weeks. Treatment assignment was determined by computer-generated random sequence with an interactive voice response system. Randomization was stratified by HbA_1c (<8.5% [69 mmol/mol] vs. ≥8.5% [69 mmol/mol]), renal function (estimated glomerular filtration rate [eGFR]), and concomitant use of OADs (metformin only, pioglitazone only, metformin and pioglitazone, or none). Because the patients who were randomized early in the trial were treated until the study’s end, the maximal possible treatment duration was 110 weeks. During the first 24 weeks of treatment, the doses of basal insulin (within 10% of baseline dose) and OADs were kept unchanged. After week 24, adjustments to the dose of basal insulin (but not OADs) were allowed according to the medical judgment of the investigator, with a treatment target for FPG of 6.1 mmol/L.

Rescue therapy could be initiated during randomized treatment if a patient met the following criteria: confirmed FPG (after overnight fast) >13.3 mmol/L during the first 12 weeks, FPG >11.1 mmol/L from weeks 12 to 24, or FPG >10.0 mmol/L or HbA_1c >8.0% (64 mmol/mol) after week 24. For initiation of rescue medication, these criteria had to be confirmed by two measurements on separate days. Patients were withdrawn from the trial if the FPG remained above this threshold despite rescue therapy.

The primary efficacy end point was the change from baseline in HbA_1c after 24 weeks of treatment. Secondary end points included changes from baseline in HbA_1c and FPG with time, change from baseline in FPG after 52 weeks of treatment, the proportion of patients achieving HbA_1c <7% (53 mmol/mol), the proportion of patients achieving ≥0.5% (5.5 mmol/mol) reduction in HbA_1c, and the change from baseline in mean basal insulin dose after 52 weeks of treatment. Other end points included the use of rescue medication and mean change in body weight to the end of treatment.

Safety end points included the frequency and intensity of adverse events (AEs), including hypoglycemia and clinically relevant new or worsening findings in physical examination, 12-lead electrocardiogram, vital signs, lipid parameters, and clinical laboratory assessments. An independent external adjudication committee reviewed treatment-emergent fatal events and suspected events of stroke or cardiac ischemia (including myocardial infarction), hospitalization for heart failure, stent thrombosis, and revascularization procedures. Follow-up for all AEs, including those persisting after a patient had completed (or withdrawn prematurely from) the trial, continued until the event had resolved or been sufficiently characterized.

Allowing for SD of change in HbA_1c from baseline of 1.2% (13.1 mmol/mol), 284 patients per treatment group were sufficient to achieve 93% power to detect a 0.35% (3.8 mmol/mol) difference between groups in change in HbA_1c from baseline to week 24. The larger sample size of 600 patients in each group allowed adequate exposure data to be collected for the regulatory requirement to detect rare cardiovascular events across the entire linagliptin program.

The primary end point was evaluated with ANCOVA at the level of α = 0.025 (one-sided). The statistical model included “treatment,” “concomitant OADs,” and “baseline renal function impairment category” as fixed classification effects and “baseline HbA_1c” as linear covariate. This analysis was performed on the full analysis set (FAS), comprising all randomized patients treated with at least one dose of study medication, with a baseline HbA_1c measurement and at least one on-treatment HbA_1c measurement within the first 24 weeks of double-blind treatment. An approach of last observation carried forward (LOCF) was used to replace missing data.

Secondary end points were evaluated in the FAS with an ANCOVA model with LOCF. Changes in FPG with time were analyzed for the FAS (observed case set [OC]; i.e., patients with available data) by means of descriptive statistics. The impact of treatment on the use of rescue medication was assessed by means of logistic regression, and the time to first use of rescue therapy was evaluated by Kaplan-Meier analysis. The impact of treatment on the occurrence of hypoglycemia was investigated by means of logistic regression and Kaplan-Meier analysis. Safety end points were evaluated for the treated set (all patients who were treated with at least one dose of study medication) by means of descriptive statistics. AEs were described according to the Medical Dictionary for Drug Regulatory Affairs (version 14.0). Hypoglycemia was analyzed by three levels of intensity: plasma glucose ≤4 mmol/L accompanied by typical symptoms of hypoglycemia, plasma glucose <3 mmol/L accompanied by typical symptoms of hypoglycemia but without need for external assistance, and severe hypoglycemia requiring the assistance of another person to actively administer carbohydrate, glucagon, or other resuscitative actions.

This study was conducted between August 2009 and September 2011. A total of 1,261 patients were randomized to receive linagliptin (n = 631) or placebo (n = 630) once daily (Fig. 1). Of these, 1,063 patients (84.3%) completed the trial (543 [86.1%] receiving linagliptin vs. 520 [82.5%] receiving placebo). The main reasons for discontinuation were AEs (linagliptin 4.0%, placebo 5.2%) and refusal to continue study medication (linagliptin 3.3%, placebo 4.1%). Mean exposures to study medication were 435.5 days in the linagliptin group and 422.4 days in the placebo group (medians 444 and 448 days, respectively).

Demographic and clinical characteristics were similar between treatment groups at baseline and are presented in Table 1.

Linagliptin was superior to placebo in reducing HbA_1c after 24 weeks (Supplementary Table 1). The adjusted mean (SE) change in HbA_1c from baseline at week 24 for linagliptin was −0.58% (0.08) (−6.3 [0.9] mmol/mol), compared with 0.07% (0.08) (0.8 [0.9] mmol/mol) for placebo, resulting in a placebo-adjusted mean change in HbA_1c from baseline of −0.65% (95% CI −0.74 to −0.55) (−7.1 mmol/mol; P < 0.0001). After 52 weeks, the adjusted mean changes in HbA_1c from baseline were −0.48% (0.08) (−5.2 [0.9] mmol/mol) for linagliptin and 0.05% (0.08) (0.5 [0.9] mmol/mol) for placebo (placebo-adjusted difference −0.53% [95% CI −0.64 to −0.43], −5.8 mmol/mol; P < 0.0001). The treatment difference between linagliptin and placebo was maintained for 76 weeks (Fig. 2A).

At week 24, the placebo-adjusted decrease in FPG was −0.6 mmol/L (95% CI −0.9 to −0.4; P < 0.001). During the period when insulin titration was allowed and recommended, improvement in FPG was sustained in the linagliptin group, whereas FPG decreased in the placebo group in parallel with a small increase in basal insulin dose (Fig. 2B). By week 52, the mean changes in FPG from baseline were similar between the groups (linagliptin baseline 7.8 mmol/L, 52-week 7.6 mmol/L, change from baseline −0.2 [0.2] mmol/L; placebo baseline 7.8 mmol/L, 52-week 7.6 mmol/L, change from baseline −0.3 [0.2] mmol/L). The FPG target of <6.1 mmol/L was achieved by 24.6% of patients in the linagliptin group and 19.1% of the placebo group. Of these patients, 29.6% and 38.1% experienced hypoglycemia in the linagliptin and placebo groups, respectively.

Among patients with baseline HbA_1c ≥7.0% (53 mmol/mol), after 52 weeks an HbA_1c <7.0% (53 mmol/mol) was achieved by 16% and 7% of the linagliptin and placebo groups, respectively (P < 0.001). Compared with the placebo group, patients in the linagliptin group were more likely to have a decrease in HbA_1c of ≥0.5% (5.5 mmol/mol) after 52 weeks of treatment (37% linagliptin vs. 17% placebo; P < 0.0001).

Analysis of change in HbA_1c by prespecified subgroups demonstrated no significant interaction with treatment according to renal function category (P = 0.5784), type of basal insulin (P = 0.9511), age group (P = 0.1000), concomitant use of OADs (P = 0.64), sex (P = 0.98), or BMI (P = 0.99), indicating that none of these factors altered the efficacy of linagliptin. Subgroup variables with significant (P < 0.10) interactions with treatment were baseline HbA_1c (P = 0.0725), geographical region (P = 0.0548), race (P = 0.0603), and time since diabetes diagnosis (P = 0.0017). Placebo-adjusted mean (SD) changes in HbA_1c were greatest for patients with a baseline HbA_1c of ≥9.0% (75 mmol/mol) (−0.83% [0.10]), patients who were from Asia (−1.00 [0.15]) or of Asian race (−0.93% [0.14]), and patients who had been diagnosed with diabetes for >5 years (−0.72% [0.05]). The placebo-adjusted mean change from baseline in HbA_1c after 24 weeks was −0.52% (−5.7 mmol/mol) for patients taking no OADs at baseline (n = 96 [15.50%]; P < 0.0001), −0.67% (−7.3 mmol/mol) for those taking metformin only (n = 470 [76.1%]; P < 0.0001), −0.76% (−8.3 mmol/mol) for those taking pioglitazone only (n = 12 [1.0%]; P = 0.129), and −0.71% (−7.8 mmol/mol) for patients taking both metformin and pioglitazone (n = 91 [7.4%]; P < 0.0001).

More patients required rescue medication in the placebo group (50.4%) than in the linagliptin group (38.2%); the odds ratio (linagliptin vs. placebo) for rescue medication was 0.575 (95% CI 0.454–0.728; P < 0.0001). The mean (SD) change in basal insulin dose to week 24 (when dose was to have remained within 10% of baseline) was 0.1 (0.2) IU for patients treated with linagliptin and 0.4 (0.2) IU for those treated with placebo. From week 24, mean basal insulin dose increased in the linagliptin group to a lesser extent than in the placebo group (Fig. 2B). The adjusted mean (SD) changes from baseline in insulin dose at week 52 were 2.6 (0.8) IU for linagliptin and 4.2 (0.8) IU for placebo (P < 0.003).

The overall incidence of patients with ≥1 reported AE was comparable between treatment groups (linagliptin 78.4%, placebo 81.4%) (Supplementary Table 2). AEs were primarily of mild or moderate intensity; AEs of severe intensity occurred in 8.2% and 8.3% of patients in the linagliptin and placebo groups, respectively. The most commonly reported severe AEs in both treatment groups were hypoglycemia, coronary artery disease, osteoarthritis, cardiac failure, pneumonia, diarrhea, arthralgia, gastroenteritis, acute renal failure, and subdural hematoma. AEs considered to be drug-related occurred in 18.7% and 22.2% of linagliptin and placebo patients, respectively. Drug-related hypoglycemia occurred in 83 (15.1%) and 95 (13.2%) patients in the linagliptin and placebo groups, respectively, and was the only drug-related AE with an incidence greater than 2%. There were no imbalances between treatment groups for other drug-related AEs. AEs leading to discontinuation of trial medication occurred in 21 patients (3.3%) in the linagliptin group and 28 patients (4.4%) in the placebo group. There were no clinically relevant changes in vital signs or laboratory parameters in either group, including no between-group imbalance in shifts in stage of renal impairment.

The percentage of patients with investigator-defined hypoglycemia was not different between groups either at week 24 (linagliptin 22.0%, placebo 23.2%) or at the end of treatment (linagliptin 31.4%, placebo 32.9%). Incidence of severe hypoglycemia was also similar between groups (week 24 linagliptin 0.3%, placebo 0.6%; end of treatment linagliptin 1.7%, placebo 1.1%) (Supplementary Table 2).

There was no significant change from baseline in mean body weight at either week 24 (linagliptin −0.16 [0.12] kg, placebo 0.12 [0.11] kg) or at week 52 (linagliptin −0.3 [0.19] kg, placebo −0.04 [0.18] kg). Adjudicated cardiovascular events occurred in 18 linagliptin patients (2.9%) and 11 placebo patients (1.7%). Cardiovascular deaths occurred in 5 patients (0.8%) in the linagliptin group and 1 (0.2%) in the placebo group. Total mortality, however, was comparable between the arms, with 5 deaths in each group.

In patients with inadequate glycemic control despite treatment with OADs, basal insulin therapy is recommended with or without additional OADs (14). This phase III clinical trial demonstrated that, in basal-insulin treated patients with type 2 diabetes and inadequate glycemic control (15,16), the addition of linagliptin 5 mg once-daily improved glycemic control without increasing the risk of hypoglycemia or body weight gain. These improvements were not affected by concomitant use of OADs, type of basal insulin, age, or degree of renal impairment.

Previous studies have also shown that DPP-4 inhibitors can significantly decrease HbA_1c concentration when added to various insulin regimens, with the exception of sitagliptin, without an additional risk of hypoglycemia (8–13). Our study differs from the previous trials in specifically testing a more homogenous population of basal insulin–treated patients and including both a period of stable insulin dosing, permitting robust assessment of the efficacy and safety of add-on linagliptin, and an extension period, during which patients were allowed to adjust the insulin according to the investigator criteria.

Addition of linagliptin to basal insulin was associated with a low frequency of hypoglycemia, implying that dose reduction of basal insulin to avoid hypoglycemia when coadministering linagliptin may not be necessary. Although the extension period was intended primarily to provide long-term safety data for linagliptin, it also provided additional information on utilization in clinical practice. In the previous trials of other DPP-4 inhibitors, the insulin dose was either kept stable (8,9,13) or titrated (10–12), but not both. During the extension period, the insulin dose was increased more in the placebo group than in the linagliptin group. The magnitude of the increase in insulin dose, however, was much less than recommended in the study protocol (i.e., to achieve an FPG target of 6.1 mmol/L). Because no forced titration was requested, this may have resulted in some “titration inertia.” Of note, the proportion of elderly patients and patients with renal impairment was high in the study, and it appears likely that investigators may have been more reluctant to uptitrate basal insulin because age and renal impairment both increase risk of hypoglycemia. The mean FPG change from baseline at week 52 was similar between groups, indicating that clinicians were targeting similar FPG goals. Nevertheless, significantly lower HbA_1c values were obtained with linagliptin.

Patients themselves are also often reluctant to uptitrate insulin because of the potentially increased risk of hypoglycemia. This represents a significant consideration driving the need for additional treatments such as linagliptin to improve glycemic control. The low risk of hypoglycemia with linagliptin could potentially reduce indirect costs, such as hospitalization, and this may counterbalance the added direct expense (17,18).

In contrast to other DPP-4 inhibitors, linagliptin has a primarily nonrenal route of elimination and therefore does not require dose adjustment in patients with impaired renal function (6,19). Because many patients taking insulin have impaired renal function, this is a potential advantage of linagliptin relative to other DPP-4 inhibitors. A previous study found linagliptin to be well tolerated and efficacious in long-term use in patients with severe renal impairment (20).

In conclusion, this study shows that addition of linagliptin to basal insulin improves glycemic control without increasing hypoglycemia or inducing weight gain, with the additional advantage that the dose does not need to be altered in the elderly or in those with impaired renal function.

This study was sponsored by Boehringer Ingelheim. H.Y.-J. received research support from Lilly-Amylin and Boehringer Ingelheim. H.Y.-J. also received honoraria for consultancy for Sanofi, Novartis, AstraZeneca, Merck Sharp & Dohme, and Bristol-Myers Squibb and for lectures at meetings sponsored by Sanofi and Merck Sharp & Dohme. J.R. served on scientific advisory boards and received honoraria or consulting fees from Roche, Sanofi, Novo Nordisk, Eli Lilly, MannKind, GlaxoSmithKline, Takeda, Daiichi Sankyo, Johnson & Johnson, Novartis, Boehringer Ingelheim, and Lexicon. J.R. also received grants and research support from Merck, Pfizer, Sanofi, Novo Nordisk, Roche, Bristol-Myers Squibb, Eli Lilly, GlaxoSmithKline, Takeda, Novartis, AstraZeneca, Amylin, Johnson & Johnson, Daiichi Sankyo, MannKind, Lexicon, and Boehringer Ingelheim. S.D.-G. received consulting fees from Boehringer Ingelheim. S.Pi., S.B., S.T., S.Pa., and H.-J.W. are employees of Boehringer Ingelheim. No other potential conflicts of interest relevant to this article were reported.

H.Y.-J., J.R., S.D.-G., S.Pi., S.B., S.Pa., and H.-J.W. participated in the design of the study; the conduct of the study; the collection, analysis, and interpretation of data; and the writing and revision of the article. S.B. planned and performed the statistical analysis of the data. S.T. participated in the collection, analysis, and interpretation of data and in the writing and revision of the article. All the authors were fully responsible for all content and editorial decisions, were involved at all stages of manuscript development, and have approved the final version. H.Y.-J. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Data from this study were presented in abstract form at the 72nd Scientific Sessions of the American Diabetes Association, Philadelphia, Pennsylvania, 8–12 June 2012, and at the 48th Annual Meeting of the European Association for the Study of Diabetes, Berlin, Germany, 1–5 October 2012.

The authors thank the patients and staff who participated in this study. (A complete list of participating investigators appears in the Supplementary Data online.) Medical writing assistance, supported financially by Boehringer Ingelheim, was provided by Claire Stevens and Paul Lane of Envision Scientific Solutions during the preparation of the manuscript.