Introduction
Improving treatment satisfaction and safety are important for treating type 2 diabetes (T2DM) and are required to gain better treatment adherence, thereby improving long-term outcomes [
1]. Poor adherence to insulin treatment results from many factors, including a busy lifestyle, travels, stress or psychosocial issues, skipped meals, and treatment-associated burden such as the number of injections, flexibility in injection time, regimen complexity, and hypoglycemia [
2]. Hypoglycemia is associated with reduced quality-of-life (QoL) due to its inherent stressful nature and worries regarding individual psychosocial consequences [
3]. Hypoglycemia itself also often results in poor treatment adherence [
4].
Second-generation long-acting basal insulin (BI) analogues, including insulin glargine 300 U/mL (Gla-300), provide a longer duration of action and even more stable pharmacokinetic/pharmacodynamic profiles than the first-generation BI analogue, Gla-100 [
5,
6]. The EDITION treat-to-target randomized controlled trial (RCT) program compared Gla-300 with Gla-100 in populations with type 1 and T2DM [
7‐
14]. EDITION 1, which included people with T2DM treated with a basal-bolus treatment regimen, showed that Gla-300 provided similar glycemic control after 6 months, but with significantly less nocturnal hypoglycemia compared with Gla-100, and a more sustained HbA
1c reduction for Gla-300 at 12 months [
8,
9]. Gla-300 may also reduce insulin treatment burden compared with Gla-100, as it allows more flexibility in the timing of injections, requires a reduced volume of injected insulin, and results in smaller glycemic excursions [
15,
16].
Most studies of BI analogues are not specifically designed to evaluate treatment satisfaction. Therefore, MAGE aimed to assess treatment satisfaction, efficacy, and safety of Gla-300 in people with T2DM receiving basal-bolus insulin in a real-world setting, and is distinct from other real-world studies, as treatment satisfaction was the primary endpoint.
Methods
Study Design
MAGE (a 6-month Multicenter, prospective, single-Arm, observational study with a 6-month extension period, evaluated treatment satisfaction, efficacy and safety of insulin Glargine 300 U/mL in people with T2DM receiving basal-bolus insulin in a rEal-world setting), was conducted in second-line diabetes clinics in Belgium. MAGE was performed in accordance with the principles of the 18th World Medical Assembly (Declaration of Helsinki, 1964) and all subsequent amendments, the European guidelines for Good Epidemiology Practice, and complied with local regulatory requirements [
17,
18]. All participants signed an informed consent form before taking part in the study and study extension.
Study Population
Participants were at least 18 years old, with T2DM for more than 1 year, HbA1c 7.0–10.0%, receiving basal and mealtime insulin (with 4–5 daily injections) for at least 6 months, and had not previously used Gla-300. Individuals were excluded if they were taking oral antihyperglycemic drugs (OADs) other than metformin or were receiving corticosteroids. Full inclusion and exclusion criteria are listed in the Supplementary Methods.
Study Endpoints
After the baseline visit, two further visits were scheduled by the physician during the first 6 months (months 3 and 6), with the possibility to plan two extra visits over an additional 6-month period (months 9 and 12). As this study was observational, visits were planned every 3 months according to the current clinical practice in Belgium. Additional visits were allowed if considered necessary by the investigator.
The primary endpoint was a change in the Diabetes Treatment Satisfaction Questionnaire status version (DTSQs) total score from baseline to month 6 in participants with T2DM treated with Gla-300 in routine care (see Supplementary Methods for details of DTSQs methodology). The primary objective was to demonstrate a change of at least 2 points in the DTSQs total score (baseline to month 6), which was deemed as a clinically relevant improvement.
Secondary endpoints included additional Patient-Reported Outcomes (PROs) such as the Diabetes Treatment Satisfaction Questionnaire change version (DTSQc); World Health Organization (WHO)-5 Well-Being Index; Hypoglycemia Fear Survey (HFS)-II; and Pittsburgh Sleep Quality Index (PSQI). See Supplementary Methods for the methodology regarding the secondary PRO analyses. Reasons for starting Gla-300 were also assessed. The study also examined mean changes in bodyweight, body mass index (BMI), HbA1c, and Gla-300 dose. The proportion of participants who continued Gla-300 therapy was also recorded.
Self-reported hypoglycemia (severe, symptomatic, confirmed [≤ 70 mg/dL and < 54 mg/dL]) was recorded according to time (nocturnal [between 00:00 and 05:59] and at any time of day [24 h]). Severe hypoglycemia was defined (per American Diabetes Association guidelines) as an event that required the assistance of another person to actively administer carbohydrates, glucagon, or to take other corrective actions [
19].
Safety outcomes including adverse events (AEs), serious adverse events (SAEs), and adverse events of special interest (AESI) were recorded.
Data Analysis and Statistics
The sample size calculation performed prior to the study (paired t test with nQuery Advisor Version 7.0) determined that the recruitment of 82 patients was required to detect a clinically relevant change of at least 2 points in the DTSQs total treatment satisfaction score at month 6 versus baseline (p value set at 0.05, standard deviation [SD] of 5.50, power of 90%). Taking into account an approximate 20% participant dropout rate, it was estimated that 100 patients should be enrolled to achieve the expected 82 completed patients at month 6.
The primary endpoint was assessed in the intent-to-treat (ITT) population (all individuals eligible for Gla-300 treatment at baseline) and in the per-protocol (PP) population (all participants who completed the study). Secondary efficacy endpoints were assessed in the ITT population. All endpoints were summarized by descriptive statistics. The primary endpoint was analyzed using a paired Student’s t test with p < 0.05 considered statistically significant. The normality of continuous secondary endpoints was assessed using Kolmogorov–Smirnov testing. For endpoints with more than two time points, mixed models were used, while changes from baseline were assessed with a one-sample t test. If normality was not achieved, Friedman’s testing was used for outcomes with more than two time points, and paired Wilcoxon’s testing was performed on outcomes with two time points. Statistical analysis was performed using IBM SPSS Statistics (Version 21.0) and StatXact (Version 6.0). Missing values were not replaced/extrapolated.
Discussion
Participants in MAGE are representative of people with T2DM typically followed in second-line diabetes clinics; however, the mean HbA
1c value at baseline was slightly higher than reported for the general Belgian population [
20], and all participants in MAGE were on basal-bolus insulin regimens, suggesting that the MAGE population in this study was at a more advanced stage of diabetes. Approximately one-third of participants had microvascular complications, as expected in a population of people who have long-term T2DM.
In this real-world population, switching to Gla-300 was associated with a significant improvement in treatment satisfaction at 6 and 12 months (2.8 points [
p < 0.0001] and 2.1 points [
p = 0.004], respectively); these improvements are in line with those reported in the EDITION 1 RCT (2.3 points from baseline to month 6) [
8]. It is worth noting that the MAGE study evaluated fewer participants over 6 months than planned on the basis of power calculations. However, because the amplitude of the observed difference in treatment satisfaction between baseline and month 6 was greater than the estimated 2 points, power for the primary endpoint remained above 90%. Different approaches have been used to determine minimally important differences (MID) in QoL measures. An empirical method based on a systematic review of 38 studies suggests using a cutoff of half of the SD [
21]. Hence, for the 7-point DTSQs scale, the MID should likely be approximately 3.5. A second group suggests a more stringent MID of 0.3 SD (i.e., a 2.1-point difference on the 7-point DTSQs scale) [
22]. Of note, the baseline DTSQs score was 27.93 points on a scale of 36 points, suggesting an already high treatment satisfaction prior to switching.
One of the main drawbacks of DTSQs is the “ceiling” effect, as individuals tend to score their current baseline treatment satisfaction at a high level, possibly precluding further improvements by the end of the study [
23]. The DTSQc was developed to overcome this bias by asking individuals to compare their new treatment versus their old treatment by the end of the trial. DTSQc results confirmed those from the DTSQs as they also showed a significant improvement in treatment satisfaction. Additionally, DTSQc identified significant improvements in both perceived frequency of hyperglycemia and hypoglycemia at both 6 and 12 months, suggesting improved perception of glucose control with the new treatment. This may be linked to a more stable glycemic profile when treated with Gla-300, compared with older BIs [
5,
6,
24].
Improvement in treatment satisfaction scores may potentially be biased, as participants may be tempted to answer PRO questionnaires positively. However, as no improvements were observed in the other non–diabetes-specific QoL measures (such as the WHO-5 Well-Being Index measure), one may conclude that this bias was avoided in MAGE. This is in accordance with another real-world observational study of Gla-300 in adults with T2DM, which demonstrated a non-significant improvement in the non–diabetes-specific WHO-5 measure, alongside a significant improvement in the Diabetes Medication System Rating Questionnaire measure [
15].
Gla-300 was associated with significant improvements in hypoglycemia fear scores (HFS-II total scores were in the range of MID: 2.0–5.8 or 3.6–3.9, depending on the methodology) [
25]. Reductions in the hypoglycemia behavior and worry subscales also suggested participants were less likely to adopt damaging behaviors (e.g
., insulin dose reduction or deliberately maintaining hyperglycemia during social events/important tasks), or have hypoglycemia-related worries (e.g., failing to recognize low blood glucose, losing control, or public embarrassment) [
26]. Sleep quality improved between baseline and month 6, which may also be associated with improved QoL [
27].
The overall annualized hypoglycemia rate in MAGE was comparable to that seen in EDITION 1 for Gla-300 (21 versus 26 events per participant-year, respectively), as was the rate of severe events (0.2 versus 0.3 events per participant-year) [
8]. Comparing other categories of hypoglycemia between these two studies is confounded by different definitions, but the rates of hypoglycemia confirmed at either the ≤ 70 mg/dL or < 54 mg/dL thresholds in MAGE (21 and 6.6 events per participant-year, respectively) also appear similar to the rates seen in EDITION 1 for hypoglycemia, defined as “confirmed” (≤ 70 mg/dL or < 54 mg/dL) or “severe” (25 and 3.8 events per participant-year, respectively). Although this latter definition in EDITION 1 includes severe events, these were rare and so would not have a substantial impact on the rates reported; any differences may be due to differences in study design (real-world study versus RCT with treat-to-target approach, in EDITION 1). The frequency of overall AEs was low in this study, and the most commonly experienced non-hypoglycemic AEs of diarrhea, bronchitis, and myalgia are in line with those reported in the 6-month EDITION 1 study, where the most common AEs in both treatment groups were infections, gastrointestinal events, or musculoskeletal complaints [
8].
Improvements in treatment satisfaction and reduction in hypoglycemia fear occurred without reduction in HbA
1c between baseline and months 6 or 12. This is in agreement with other studies using DTSQs, which showed that, while improvements in treatment satisfaction were not associated with improved glycemic control, they were likely to improve treatment adherence and the perception individuals have about control of the disease [
1]. Thus, improvements in treatment satisfaction may contribute to long-term stable glucose control.
There was an increase in Gla-300 dose between baseline and months 6 and 12 of up to 14%, which is in line with the EDITION trial program [
8]. However, an insulin dose increase has not been seen in all studies [
28,
29]. In this study, the dose increase occurred in parallel with clinically relevant benefits in PROs, such as increased treatment satisfaction and reduced hypoglycemia fear.
A strength of the MAGE study is the length of the study; the 6-month extension period confirmed the positive diabetes-specific PRO results from the 6-month study period. When interpreting the results of MAGE, consideration should be given to the open-label, non-interventional design of the study. Although the results are in accordance with the EDITION 1 RCT, they should be considered as complementary to the RCT data and deserve to be confirmed by additional trials. It is possible that the participants of MAGE may have decided to switch to Gla-300 as a result of dissatisfaction with their prior insulin therapy, which could have biased their perception of treatment satisfaction in favor of Gla-300. As individuals were unblinded and aware that they were receiving a “newer” and more advanced insulin, they may have been influenced to be more positive in treatment satisfaction and QoL scoring. However, the MAGE cohort appeared typical of those who attend second-line outpatient clinics, and the absence of improvement in non–diabetes-specific measures (e.g., WHO-5 Well-Being Index) makes a bias in favor of Gla-300 unlikely. Other limitations include the limited participant numbers, and the absence of a control group; however, this is reflective of the nature of this real-world study.
Acknowledgements
The authors thank the study participants, the trial staff, particularly Elke Meyssen and Frank Stessens, and the investigators for their participation.