Efficacy and Safety of Liraglutide Added to Insulin Treatment in Type 1 Diabetes: The ADJUNCT ONE Treat-To-Target Randomized Trial
OBJECTIVE To investigate whether liraglutide added to treat-to-target insulin improves glycemic control and reduces insulin requirements and body weight in subjects with type 1 diabetes.
RESEARCH DESIGN AND METHODS A 52-week, double-blind, treat-to-target trial involving 1,398 adults randomized 3:1 to receive once-daily subcutaneous injections of liraglutide (1.8, 1.2, or 0.6 mg) or placebo added to insulin.
RESULTS HbA1c level was reduced 0.34–0.54% (3.7–5.9 mmol/mol) from a mean baseline of 8.2% (66 mmol/mol), and significantly more for liraglutide 1.8 and 1.2 mg compared with placebo (estimated treatment differences [ETDs]: 1.8 mg liraglutide −0.20% [95% CI −0.32; −0.07]; 1.2 mg liraglutide −0.15% [95% CI −0.27; −0.03]; 0.6 mg liraglutide −0.09% [95% CI −0.21; 0.03]). Insulin doses were reduced by the addition of liraglutide 1.8 and 1.2 mg versus placebo (estimated treatment ratios: 1.8 mg liraglutide 0.92 [95% CI 0.88; 0.96]; 1.2 mg liraglutide 0.95 [95% CI 0.91; 0.99]; 0.6 mg liraglutide 1.00 [95% CI 0.96; 1.04]). Mean body weight was significantly reduced in all liraglutide groups compared with placebo ETDs (1.8 mg liraglutide −4.9 kg [95% CI −5.7; −4.2]; 1.2 mg liraglutide −3.6 kg [95% CI −4.3; −2.8]; 0.6 mg liraglutide −2.2 kg [95% CI −2.9; −1.5]). The rate of symptomatic hypoglycemia increased in all liraglutide groups (estimated rate ratios: 1.8 mg liraglutide 1.31 [95% CI 1.07; 1.59]; 1.2 mg liraglutide 1.27 [95% CI 1.03; 1.55]; 0.6 mg liraglutide 1.17 [95% CI 0.97; 1.43]), and hyperglycemia with ketosis increased significantly for liraglutide 1.8 mg only (event rate ratio 2.22 [95% CI 1.13; 4.34]).
CONCLUSIONS Liraglutide added to insulin therapy reduced HbA1c levels, total insulin dose, and body weight in a population that was generally representative of subjects with type 1 diabetes, accompanied by increased rates of symptomatic hypoglycemia and hyperglycemia with ketosis, thereby limiting clinical use in this group.
Glycemic control is still frequently suboptimal in people with type 1 diabetes (1). Maintaining strict glycemic control by intensive insulin therapy, although effective, is challenging and often results in hypoglycemic events and weight gain (2). Adjunct therapies to insulin, like metformin, are being used off label to stabilize glucose control, limit insulin dose, and reduce weight gain. Novel combination therapies using agents commonly prescribed for type 2 diabetes show promise when added to insulin therapy in type 1 diabetes (3). Among these therapies is liraglutide, a glucagon-like peptide 1 receptor agonist (GLP-1RA), which enhances insulin secretion and suppresses elevated levels of glucagon in a glucose-dependent manner (4). In patients with type 2 diabetes, liraglutide, when used as an adjunct to insulin, has provided clinically meaningful reductions in glycated hemoglobin (HbA1c) (5).
Recently, small-scale studies and retrospective analyses have suggested a potential for liraglutide in the treatment of people with type 1 diabetes. The administration of liraglutide as an adjunct to insulin has consistently shown a reduction in daily insulin dose and weight loss compared with control subjects, and neutral or positive results in terms of HbA1c levels and rates of hypoglycemia (6–12). This study aims to address whether adding liraglutide to insulin therapy in a treat-to-target approach can safely improve glycemic control in a population that is generally representative of people with type 1 diabetes when studied in a randomized, double-blind fashion over a 52-week period.
Research Design and Methods
The trial was a 52-week randomized, placebo-controlled, double-blind, parallel-group, treat-to-target, phase 3 trial performed at 177 centers in 17 countries.
A screening visit (Week −2) was followed by a randomization visit 2 weeks later (Week 0). Eligible subjects were stratified by HbA1c level and BMI (HbA1c <8.5% and ≥8.5% [<69 and ≥69 mmol/mol], BMI ≤27 and >27 kg/m2), and were centrally randomized by Novo Nordisk (Clinical Supplies Coordination) to one of six parallel-treatment groups in a 3:1 ratio to receive either liraglutide (0.6, 1.2, or 1.8 mg) or a corresponding volume of placebo (0.1, 0.2, or 0.3 mL) once daily by subcutaneous injection. Subjects started with a dose of 0.6 mg liraglutide/0.1 mL placebo, and increased by 0.6 mg/0.1 mL every other week until reaching their target dose. Once reached, the liraglutide dose was maintained, with no dose reduction permitted. Subjects continued insulin treatment in addition to liraglutide or placebo throughout the trial. The total daily insulin dose was initially reduced by 25% on the day of randomization and by a further 10% on subsequent days of dose escalation for a minimum of 24 h. Subjects’ insulin dose was subsequently increased as necessary, which was in line with the treat-to-target design. Pretrial insulin treatment was considered to be background medication and was continued throughout the trial.
Key inclusion criteria were as follows: clinically diagnosed type 1 diabetes ≥12 months prior to the screening visit, treatment with basal bolus or continuous subcutaneous insulin infusion (CSII) ≥6 months, stable insulin treatment for the last 3 months, BMI ≥20 kg/m2, and age 18–75 years. A notably wide range of HbA1c levels (7.0–10% [53–86 mmol/mol]) was specified in order to include subjects clinically representative of people with type 1 diabetes. Importantly, and in contrast to most studies, the current trial did not exclude people with a history of severe hypoglycemia or hypoglycemia unawareness, or people with a history of ketoacidosis.
Key exclusion criteria included the following: prior use of GLP-1RA or dipeptidyl peptidase-4 inhibitors; the use of medication that interfered with glycemic control, including all hypoglycemic agents or steroids; history of acute or chronic pancreatitis; severely decreased renal function (estimated glomerular filtration rate <30 mL/min/1.73 m2, calcitonin level >50 ng/L at screening, personal/family history of medullary thyroid carcinoma or multiple endocrine neoplasia syndrome type 2, or severe neuropathy.
The primary objective of this study was to investigate the efficacy of liraglutide added to treat-to-target insulin therapy on glycemic control, and the reduction in total daily insulin dose and body weight loss compared with placebo in subjects with established type 1 diabetes with inadequate glycemic control. The secondary objective was to determine the efficacy of liraglutide added to insulin in reducing episodes of hypoglycemia compared with placebo.
Additional efficacy measures included fasting plasma glucose (FPG), self-determined nine-point self-monitored plasma glucose profiles (using a self-measured blood glucose [BG] device calibrated to report plasma glucose), 1,5-anhydroglucitol (1,5-AG), and quality of life (treatment-related impact measures-diabetes [TRIM-D] and Short Form [SF]-36 scores).
Key safety assessments included adverse events (AEs), hypoglycemic and hyperglycemic episodes, pulse, systolic and diastolic blood pressure, the laboratory safety variables amylase and lipase, and predefined medical events of special interest (deaths, acute coronary syndrome, cerebrovascular event, pancreatitis, neoplasm, thyroid disease, symptomatic hyperglycemic episodes with ketosis [plasma glucose >16.7 mmol/L (>300 mg/dL)], and plasma ketone level >1.5 mmol/L), severe hypoglycemic episodes, and medication errors concerning trial products. The majority of medical events of special interest were evaluated by an independent external adjudication committee (EAC). Symptomatic hypoglycemic episodes were defined by Novo Nordisk as severe according to the American Diabetes Association (ADA) and a plasma glucose value <3.1 mmol/L (56 mg/dL) with symptoms consistent with hypoglycemia (13).
The sample size was determined to establish noninferiority in HbA1c level change after 52 weeks of treatment between liraglutide 1.8 mg and placebo with 90% power, assuming a noninferiority margin of 0.3%, a two-sided t test at a 5% significance level, an SD of 1.1%, and a true difference of 0%. Furthermore, a dropout rate of 20% was assumed, and it was also assumed that none of the withdrawn subjects would contribute to the analysis. The required sample size was found to be 1,395 subjects, 349 per arm, allowing for the expected dropout rate.
Continuous data were analyzed using a mixed model for repeated measurements (MMRM) using an unstructured covariance matrix and with treatment, stratification, and country as fixed factors, and baseline value as a covariate, all nested within a visit. Insulin dose, 1.5-AG, amylase, and lipase were logarithmically transformed before analysis. The response consisted of all scheduled postbaseline measurements obtained while the subjects were receiving treatment.
Binary data were analyzed by a logistic regression model with treatment and stratification as fixed factors, and baseline as a covariate. Missing data were imputed from the MMRM used for the analysis of HbA1c. Event data (hypoglycemic/hyperglycemic episodes) were analyzed by a negative binomial regression model with a log-link function and the logarithm of the exposure time as offset, with treatment, stratification, and country as fixed factors, and the HbA1c value at baseline as a covariate. Placebo groups were pooled in all analyses.
Analyses were undertaken in an attempt to identify subgroups with an improved treatment effect. Prespecified analyses used the baseline variables age, sex, duration of type 1 diabetes, method of insulin administration, BMI, HbA1c level, hypoglycemia unawareness status, severe hypoglycemia within the last 12 months, renal function, and geographic region. Post hoc analyses (specified after unblinding) included baseline variables such as smoker status, country, and C-peptide level (less than or greater than or equal to the lower limit of quantification [LLOQ; 0.03 nmol/L]).
A total of 1,789 adults with type 1 diabetes were screened, of whom 1,398 were randomized and 1,393 were exposed to trial product (Supplementary Fig. 1).
Subjects were broadly homogeneous across treatment groups in terms of age, gender, HbA1c level, and type 1 diabetes duration (Table 1 and Supplementary Table 1). Notably, the study population included ∼37% of patients with HbA1c level ≥8.5%, ∼7% had a history of severe hypoglycemia, and 6% had hypoglycemia unawareness. The majority of the study population, ∼83%, were C-peptide negative. The remaining 17% had C-peptide levels below the normal range, which is consistent with a diagnosis of type 1 diabetes, but above the LLOQ (0.03 nmol/L).
Despite the treat-to-target design, a differential effect on HbA1c levels was seen for the different treatment groups (Fig. 1A). All groups had their HbA1c level decreased during the study, which is consistent with the treat-to-target protocol including insulin titration, and enhanced teaching and guidance during the trial. The drop from baseline was largest and dose dependent in the liraglutide-treated subjects, but was partially lost by Week 52 (0.54%, 0.49%, 0.43% for liraglutide 1.8, 1.2, and 0.6 mg and 0.34% for placebo [5.9, 5.4, 4.8, and 3.8 mmol/mol, respectively]). This reduction in HbA1c level was significantly larger for liraglutide 1.8 and 1.2 mg compared with placebo (1.8 mg liraglutide estimated treatment difference [ETD]: −0.20 [95% CI −0.32; −0.07], P = 0.0019; 1.2 mg liraglutide −0.15 [95% CI −0.27; −0.03], P = 0.0164; 0.6 mg liraglutide −0.09 [95% CI −0.21; 0.03], P = 0.1299).
There were no significant differences in FPG and 1,5-AG levels for any liraglutide group compared with the placebo group (Supplementary Fig. 2 and Supplementary Table 2). The observed reduction in the mean of the nine-point self-monitored plasma glucose profile from baseline to Week 52 was somewhat lower for the liraglutide groups compared with the placebo group (Supplementary Fig. 2 and Supplementary Table 2).
The total insulin dose decreased from baseline to Week 52 in the 1.8 mg liraglutide (−5%) and 1.2 mg liraglutide (−2%) groups, but increased in the 0.6 mg liraglutide (4%) and placebo groups (4%) (Fig. 1B). The reduction in total insulin dose, primarily bolus insulin, was significant for liraglutide 1.8 and 1.2 mg compared with placebo (estimated treatment ratio: 1.8 mg liraglutide 0.92 [95% CI 0.88; 0.96], P < 0.0001; 1.2 mg liraglutide 0.95 [95% CI 0.91; 0.99], P = 0.0148; 0.6 mg liraglutide 1.00 [95% CI 0.96; 1.04], P = 0.9615). Units of insulin per kilogram of body weight returned to approximately baseline levels in all treatment groups after 52 weeks.
Categorizing subjects by reduction in HbA1c level highlighted the fact that approximately twice as many subjects receiving liraglutide 1.8 mg had reductions in HbA1c level of >1% without severe hypoglycemia after 52 weeks compared with those receiving placebo. In addition, nearly twice as many subjects receiving liraglutide 1.8 and 1.2 mg had significant HbA1c reductions below the ADA target of 7.0% or HbA1c reductions <7% without severe hypoglycemia compared with those receiving placebo (Fig. 2).
Treatment with liraglutide in addition to insulin significantly reduced body weight in a dose-dependent manner, with losses of 4.0, 2.7, and 1.3 kg for liraglutide 1.8, 1.2, and 0.6 mg, respectively. Subjects randomized to receive placebo gained 0.9 kg (ETD: 1.8 mg liraglutide −4.90 [95% CI −5.65; −4.16], P < 0.0001; 1.2 mg liraglutide −3.55 [95% CI −4.29; −2.81], P < 0.0001; 0.6 mg liraglutide −2.19 [95% CI −2.91; −1.47], P < 0.0001).
Overall, the rates of all AEs increased dose dependently in the liraglutide groups compared with the placebo group (1.8 mg, 7.7 events per patient-year of exposure [PYE]; 1.2 mg, 6.0 events/PYE; 0.6 mg, 5.3 events/PYE; and placebo, 4.8 events/PYE) (Table 2).
Gastrointestinal disorder AEs were reported more frequently in the liraglutide groups compared with the placebo group and increased dose dependently (1.8 mg, 2.7 events/PYE; 1.2 mg, 1.9 events/PYE; 0.6 mg, 1.3 events/PYE; placebo, 0.76 events/PYE) with the most frequently reported complaint being nausea (1.8 mg, 0.97 events/PYE; 1.2 mg, 0.69 events/PYE; 0.6 mg, 0.47 events/PYE; placebo, 0.18 events/PYE).
The proportion of subjects with AEs leading to premature discontinuation of trial product increased dose dependently (14.7%, 12.6%, and 4.9%, respectively, for liraglutide 1.8, 1.2, and 0.6 mg) compared with placebo (3.4%), and the majority were gastrointestinal AEs. Most of the discontinuations occurred within the initial 8–12 weeks of the trial. The proportion of subjects reporting serious AEs (SAEs) was comparable among treatment groups (Table 2).
Mean (geometric) lipase levels were approximately 24 units/L at baseline and increased in all liraglutide groups compared with placebo groups, but with no clear dose dependency (35%, 36%, and 32%, respectively, for liraglutide 1.8, 1.2, and 0.6 mg, and 1% for placebo). Similarly, mean (geometric) amylase levels were initially ∼50 units/L and increased in all liraglutide groups compared with placebo, but with no clear dose dependency (10%, 10%, and 10%, respectively, for liraglutide 1.8, 1.2, and 0.6 mg, respectively, and 3% for placebo). One case of pancreatitis in the liraglutide 0.6 mg group was confirmed by the EAC (Supplementary Table 3).
More than 90% of subjects experienced a hypoglycemic episode (BG <3.9 mmol/L [70 mg/dL], by ADA definition), and ∼82% of subjects experienced symptomatic hypoglycemic episodes (Novo Nordisk definition [Table 2], which is used in this study unless otherwise stated).
The rate of symptomatic hypoglycemic episodes was observed to be higher in liraglutide-treated subjects (1.8 mg, 16.5 events/PYE; 1.2 mg, 16.1 events/PYE; 0.6 mg, 15.7 events/PYE; placebo, 12.3 events/PYE). The rate of symptomatic hypoglycemic episodes was significantly higher for liraglutide 1.8 and 1.2 mg compared with placebo (estimated rate ratio [ERR]: 1.8 mg liraglutide 1.31 [95% CI 1.07; 1.59], P = 0.0081; 1.2 mg liraglutide 1.27 [95% CI 1.03; 1.55], P = 0.0219; 0.6 mg liraglutide 1.17 [95% CI 0.97; 1.43], P = 0.1079). In contrast, the number of hypoglycemic events assessed by the EAC and confirmed as severe was observed to be lower in all liraglutide-treated groups compared with the placebo group (45, 31, and 40 events, respectively, in the liraglutide 1.8, 1.2, and 0.6 mg groups, compared with 57 events in the placebo group) (Table 2). However, there was no statistically significant difference in the number of EAC-confirmed severe hypoglycemic episodes for any treatment comparison (ERR: 1.8 mg liraglutide 0.85 [95% CI 0.48; 1.49], P = 0.5693; 1.2 mg liraglutide 0.61 [95% CI 0.33; 1.12], P = 0.1088; 0.6 mg, 0.64 [95% CI 0.36; 1.13], P = 0.1250).
Novo Nordisk–defined asymptomatic hypoglycemic episodes are those episodes with plasma glucose levels <3.1 mmol/L (56 mg/dL) and no symptoms of hypoglycemia (Table 2).
Considering the total number of severe or BG-confirmed hypoglycemic episodes, ∼21.7%, 19.6%, and 18.2%, respectively, in the liraglutide 1.8, 1.2, and 0.6 mg groups, and 27.1% in the placebo group were asymptomatic. The distribution of asymptomatic hypoglycemic episodes did not stem from unbalanced randomization of trial subjects with a history of hypoglycemia unawareness or severe hypoglycemia (Table 1).
Nocturnal hypoglycemia (onset between 0:01 and 5:59 a.m., both inclusive) showed broadly similar patterns to diurnal hypoglycemia. The observed rates were higher for liraglutide compared with placebo, but without any apparent dose dependency.
Overall, ∼90% of subjects reported hyperglycemic episodes (plasma glucose >16.7 mmol/L [>300 mg/dL]) (Table 2). The observed rates were 33.5, 30.9, and 29.5 events/PYE for 1.8, 1.2, and 0.6 mg liraglutide, respectively, and 34.7 events/PYE for placebo. The proportion of subjects with hyperglycemic episodes with ketosis (plasma ketone level >1.5 mmol/L) was 11.2%, 7.5%, and 6.3%, respectively, for liraglutide 1.8, 1.2, and 0.6 mg, and 6.9% in the placebo group. Higher rates were observed for the liraglutide groups compared with the placebo group (1.8 mg liraglutide, 0.28 events/PYE; 1.2 mg liraglutide, 0.15 events/PYE; 0.6 mg liraglutide, 0.17 events/PYE; and placebo, 0.12 events/PYE). The treatment difference was statistically significant for liraglutide 1.8 mg compared with placebo (ERR 2.22 [95% CI 1.13; 4.34], P = 0.0205). All symptomatic hyperglycemic episodes with ketosis (plasma ketone level >1.5 mmol/L) were evaluated by the EAC to confirm that they were ketoacidosis events. A total of eight events were adjudicated as being diabetic ketoacidosis (DKA); 3, 1, and 4 events, respectively, in the liraglutide 1.8, 1.2, and 0.6 mg groups, and 0 in the placebo group (Table 2). In all but one case (observed in a subject who had been susceptible to nausea and vomiting during the treatment escalation period), there was a clinically relevant condition, such as concomitant infection, pump malfunction, and postoperative insufficient insulin treatment, that triggered the episode or a medical history of multiple DKA events. In addition, as shown in Supplementary Table 4, DKA was seen across a broad range of HbA1c values at the time of the DKA event, and all DKA cases occurred in C-peptide–negative subjects (i.e., C-peptide level less than LLOQ).
Subgroup analyses identified no notably different findings, with the exception of C-peptide positivity (Supplementary Tables 5–10 and Supplementary Fig. 3). In the current trial, 239 of 1,373 subjects with baseline C-peptide measurements were C-peptide positive (value above or equal to the LLOQ [0.03 nmol/L]) (Supplementary Table 1). Baseline characteristics of the C-peptide–positive subjects showed a shorter duration of type 1 diabetes and higher baseline HbA1c values (Supplementary Table 1). For both the 1.8 and 1.2 mg liraglutide dose groups, reductions in HbA1c level from baseline were greater for C-peptide–positive subjects (0.83% and 0.71%, respectively) than for subjects who were C-peptide negative (0.47% and 0.44%, respectively). For the lowest liraglutide dose group (0.6 mg), HbA1c reductions were similar for C-peptide–positive and C-peptide–negative patients (0.42% and 0.44%, respectively). C-peptide–positive subjects also experienced a lower rate of symptomatic hypoglycemia and very few episodes (two, one, zero, and one events, respectively, in the liraglutide 1.8, 1.2, and 0.6 mg, and placebo groups) of hyperglycemia with ketosis (plasma ketone level >1.5 mmol/L) compared with C-peptide–negative participants. The interaction between treatment effect and C-peptide positivity was not statistically significant for either HbA1c level or symptomatic hypoglycemia for any dose group (Supplementary Tables 6 and 8, respectively).
For all liraglutide dose groups, estimated mean reductions in HbA1c level were greater for patients with a baseline HbA1c level <8.5% of those with a baseline HbA1c level of ≥8.5% (1.8 mg 0.57% vs. 0.46%; 1.2 mg 0.54% vs. 0.40%; 0.6 mg 0.44% vs. 0.41%). However, the interaction between treatment effect and baseline HbA1c level was not statistically significant for any dose group (Supplementary Table 10).
Quality of Life
Patient-reported outcomes were generally positive; the TRIM-D total score was significantly higher for all liraglutide groups, particularly for diabetes management, compared with the placebo group (1.8 mg liraglutide: ETD 4.58 [95% CI 2.82; 6.34], P > 0.0001; 1.2 mg liraglutide: 4.85, ETD 3.09 [95% CI 1.35; 4.83], P = 0.0005; 0.6 mg liraglutide 4.71, ETD 2.95 [95% CI 1.27; 4.63], P = 0.0006). There were no significant treatment differences for any of the liraglutide groups compared with the placebo group in the SF-36 overall physical and mental scores.
ADJUNCT ONE is the first 52-week, double-blind, randomized controlled clinical trial to investigate the safety and efficacy of adding liraglutide to a continuously adjusted treat-to-target insulin dose in people with type 1 diabetes with inadequate glycemic control at baseline. The current trial demonstrated that in a population that is generally representative of people with type 1 diabetes, liraglutide added to insulin treatment results in a modest dose-dependent reduction in HbA1c level despite a reduced total daily insulin requirement for the two highest doses of liraglutide. All doses of liraglutide reduced body weight, which is consistent with observations in people with type 2 diabetes (14). Though some of the efficacy markers (HbA1c and 1,5-AG levels) drifted more toward baseline in the liraglutide groups compared with the placebo group, body weight remained more constant, and insulin dose did not increase despite the deterioration in HbA1c and 1,5-AG levels. The exact mechanisms underlying these observations remain unclear.
Previous investigations of the efficacy of liraglutide added to insulin in people with type 1 diabetes have been inconclusive; some trials (8,15) found no treatment difference compared with the placebo group, whereas others reported reductions in BG and HbA1c, which to some extent could be due to liraglutide reducing postprandial glucagon secretion and glucose excursions. In the current trial, the placebo-corrected reduction in HbA1c level reported with the highest dose of liraglutide (1.8 mg) was modest (0.2% [2.4 mmol/mol]).
A reduction in placebo-adjusted body weight of up to 4.9 kg was achieved for subjects randomized to receive liraglutide, which was comparable to that recently reported (12) with liraglutide in overweight patients with type 1 diabetes (−6.8 kg). This is noteworthy for an adjunctive therapy to insulin. The improved glycemic control obtained with intensive insulin therapy must always be balanced against associated weight gain, as overweight and obesity are in turn associated with insulin resistance and cardiovascular risk factors in a population with type 1 diabetes (16). Therefore, the weight loss associated with liraglutide is considered to be a benefit for patients with type 1 diabetes, but may be an issue in those where weight is subnormal. The current trial shows that these beneficial findings come at the expense of a higher rate of symptomatic hypoglycemia and more episodes of hyperglycemia with ketosis compared with insulin treatment alone. This finding has not been consistently reported before in people with type 1 diabetes when liraglutide was added to therapy, possibly reflecting the glycemic control and lower insulin dose in liraglutide-treated subjects, as well as difficulties in accurately adjusting bolus insulin and food intake (6,9). A more flexible insulin titration may have mitigated the risk of hypoglycemia, though this would have to be balanced with the constraints of the treat-to-target design required for wider risk-benefit analysis. Notably, rates of severe hypoglycemia were not higher in liraglutide-treated subjects compared with placebo, which in part could be due to the fact that liraglutide did not compromise the glucagon counter-regulatory response to hypoglycemia (45 mg/dL) in type 1 diabetes patients (8). Post hoc analyses, though not conclusive, also indicated a trend toward a lower rate of symptomatic hypoglycemia, almost no episodes with hyperglycemia with ketosis, and a greater reduction in HbA1c level in C-peptide–positive subjects compared with C-peptide–negative subjects. The current study, importantly, did not exclude people with a history of severe hypoglycemia and hypoglycemia unawareness, which distinguishes it from typical trials in type 1 diabetes. This subgroup did not respond differently from the overall population, pointing to a hypoglycemia risk when combining liraglutide with insulin in a broad population of people with long-standing type 1 diabetes.
The increases in mean amylase and lipase reported in the liraglutide treatment groups did not have any clinical implications and are similar to observations with this drug class in type 2 diabetes (17–19). Because pancreatic volume and enzyme secretion are often decreased in subjects with type 1 diabetes, elevations from baseline for an individual subject within the defined normal range need to be interpreted in this context (20–22).
Although hyperglycemic events occurred in an approximately equal proportion of subjects across all treatment groups and the rate of events was lower in the liraglutide-treated groups than for the placebo group, the higher rate of hyperglycemia with ketosis may have two explanations. First, it could be linked, in part, to the nausea associated with liraglutide treatment; 41% of subjects treated with liraglutide and 12.1% randomized to receive placebo reported nausea, the most frequent AE. It is possible that the dose-dependent increase in nausea may alert subjects to check ketones more frequently, resulting in a parallel dose-driven reporting of hyperglycemia with ketosis. Second, and more probably, the increased incidence of hyperglycemia with ketosis reflects the effect of lowering the insulin dose in the liraglutide groups, which in some patients may then be below that required to suppress peripheral lipolysis. Lowering the insulin dose below a critical level in patients with type 1 diabetes will result in ketone production, especially when carbohydrate levels are reduced as a result of decreased carbohydrate intake. Similar observations have been made for sodium/glucose cotransporter 2 inhibitors added to insulin therapy, where carbohydrate levels may be reduced as a result of urinary excretion (23,24).
Very few cases of DKA were observed in this trial, all of which were in the liraglutide-treated groups, though were not dose dependent. Information on a history of DKA was not systematically collected from subjects before study enrollment, and no attempt to exclude subjects prone to DKA was made (these steps were taken in agreement with the regulatory authorities in order to recruit a clinically informative trial population). The incidence of DKA in the current trial should also be considered in the context of wider reported baseline rates of DKA in the type 1 diabetes population; almost 5% of patients with type 1 diabetes in the T1D Exchange Clinical Registry reported a DKA event in the 12 months before their enrollment, compared with <1% of subjects treated with liraglutide in the current trial for the same period (25).
Paradoxically, a major strength of the ADJUNCT ONE trial, namely, the broad subject recruitment, may also be a limitation. Widening the subject population may better reflect the clinical spectrum of diabetes but may also mask the subgroups in which liraglutide adjunct to insulin would produce the most benefit. This is most clearly illustrated by the post hoc analyses, which suggest better treatment effects in subjects with residual C-peptide levels at the start of the trial. This is both consistent with the mechanism of action of liraglutide and in line with a recent report (26) in people with type 2 diabetes highlighting a better clinical response to GLP-1RAs in those subjects with residual C-peptide levels.
In conclusion, the current trial demonstrated that adding liraglutide to insulin therapy for a population generally representative of people with type 1 diabetes resulted in better glycemic control, less insulin, greater body weight loss, and a greater proportion of subjects achieving the ADA target of HbA1c <7% (53 mmol/mol). This dose-dependent effect was accompanied by a higher rate of symptomatic hypoglycemia and hyperglycemic episodes with ketosis, limiting the clinical use of GLP1-RAs in people with type 1 diabetes.
Acknowledgments. The authors thank all subjects and ADJUNCT ONE investigators involved in the study. The authors also thank Salvatore Calanna and Thomas Jon Jensen of Novo Nordisk A/S for their review and input to the manuscript and Daniel Hayward of Novo Nordisk A/S for providing medical writing assistance during the preparation of this manuscript. Submission support was provided by Watermeadow Medical, an Ashfield Company, part of UDG Healthcare plc, funded by Novo Nordisk.
Duality of Interest. This study was sponsored by Novo Nordisk A/S (NN9211-3919). The sponsor (Novo Nordisk) was involved in the study design and protocol development, reviewed the manuscript for scientific accuracy, and provided statistical support. C.M. has served on advisory panels for Novo Nordisk, Sanofi Aventis, Merck Sharp & Dohme Ltd., Eli Lilly, Novartis, Bristol-Myers Squibb, AstraZeneca LP, Pfizer, Jansen Pharmaceuticals, and Hanmi; has received research support from Novo Nordisk, Sanofi Aventis, Merck Sharp & Dohme Ltd., Eli Lilly, and Novartis; and has served on speakers' bureaus for Novo Nordisk, Sanofi Aventis, Merck Sharp & Dohme, Eli Lilly, Novartis, and AstraZeneca. B.Z. has served as a consultant for Abbott, AstraZeneca, Boehringer Ingelheim, Eli Lilly, Merck, Novo Nordisk, and Sanofi and has received grant support from AstraZeneca, Boehringer Ingelheim, and Novo Nordisk. J.U.H. has served on advisory panels for Novo Nordisk and Sanofi Diabetes; has received lecture fees from Novo Nordisk, Sanofi Diabetes, and Eli Lilly; has served as a consultant for Novo Nordisk; and has received grant support from InfuCare-Dexcom. V.W. has received honoraria for speaking and for participation on advisory boards and in clinical trials from Novo Nordisk, Eli Lilly, Sanofi, Boehringer Ingelheim, and AstraZeneca. P.C. has served on advisory panels for Novo Nordisk, Eli Lilly, and Sanofi Diabetes and has received lecture fees from Novo Nordisk, Eli Lilly, and Sanofi Diabetes. E.C. is an employee and shareholder of Novo Nordisk A/S. M.L. is an employee of Novo Nordisk A/S. B.B. has served as consultant for and owns stocks in Aseko; has served on a speakers' bureau for Merck; and has served on a speakers' bureau and as a consultant for Valeritas. His employer/institution has received research support from Abbott, AstraZeneca, Dexcom, Lexicon, Roche, and Senseonics; his institution has received speakers' bureau and consultant fees and research support from Janssen, Medtronic, Novo Nordisk, and Sanofi; and his institution has received speakers' bureau and research support from Boehringer Ingelheim/Lilly and GlaxoSmithKline. No other potential conflicts of interest relevant to this article were reported.
Author Contributions. C.M., E.C., and B.B. contributed to the study design and contact and data collection and analysis. B.Z., J.U.H., V.W., and P.C. contributed to study contact and data collection and analysis. M.L. contributed to the data analysis. All authors were involved in the writing of the manuscript and approved the final version of the manuscript. C.M. 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.
Prior Presentation. Parts of this study were presented in abstract form at the American Association of Clinical Endocrinologists 25th Annual Meeting and Clinical Congress, Orlando, FL, 25–29 June 2016.
↵* A complete list of the ADJUNCT ONE Investigators can be found in the Supplementary Data online.
This article contains Supplementary Data online at http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc16-0691/-/DC1.
See accompanying article, p. 1693.
- Received March 30, 2016.
- Accepted July 16, 2016.
- © 2016 by the American Diabetes Association.
Readers may use this article as long as the work is properly cited, the use is educational and not for profit, and the work is not altered. More information is available at http://www.diabetesjournals.org/content/license.