Reduced Hypoglycemia Risk in Type 2 Diabetes Patients Switched to/Initiating Insulin Glargine 300 vs 100 U/ml: A European Real-World Study

Approximately 60 million people in Europe have diagnosed type 2 diabetes (T2D), meaning that this disease affects around 10% of European adults aged ≥ 25 years [1]. Thus, T2D is a major burden on individuals and healthcare systems [2]. Joint American and European guidelines recommend lifestyle interventions as the first-line intervention, metformin as the first-line medication, and glucagon-like peptide 1 receptor antagonists as the first-line injectable medication [3]. As T2D is a progressive disease, patients will often eventually require insulin therapy [2]. However, there are a number of barriers to the initiation of insulin including fear of injections and hypoglycemia [2, 4].

A second-generation basal insulin analog (BIA), insulin glargine 300 U/ml (Gla-300), has been associated with significant reductions in some hypoglycemia outcomes (and comparable improvements in glycemic control) in randomized controlled trials (RCTs) in patients with T2D vs. the first-generation BIA, insulin glargine 100 U/ml (Gla-100) [5‐9], which has been confirmed in recent trial-level [10] and patient-level [11, 12] meta-analyses. This advantage is likely due to Gla-300’s improved pharmacokinetic/pharmacodynamic (PK/PD) profile, longer half-life, and more stable bioavailability [13‐15]. The reduced risk of hypoglycemia, in combination with additional advantages, such as reduced injection volumes and flexible dosing times, may allow second-generation BIAs to overcome some of the barriers to insulin acceptance and dose optimization [16‐19].

The importance of complementing RCT data with real-world data, both to fill the gaps in RCT comparative effectiveness data and to ensure that data derived from RCTs are generalizable to different real-world treatment settings and patient populations, is increasingly being recognized [20]. A number of real-world studies from the US have demonstrated comparable to improved glycemic control and comparable to reduced hypoglycemia with Gla-300 vs. (mainly) first-generation BIAs in insulin-experienced [21‐24] and insulin-naïve [23, 25] patients. However, there is a paucity of real-world comparative effectiveness research in European patient populations [26]. Non-comparative studies conducted in Europe have shown that switching to [27‐29] or initiating [30] Gla-300 improves glycemic control, but the impact on hypoglycemia is unclear.

The main objectives of the current study were to compare patient demographics, clinical characteristics, glycemic control, and hypoglycemic events of T2D patients prescribed Gla-300 or Gla-100 using data from clinical practice in France, Spain, and Germany. We also examined physicians’ reasons for initiating Gla-300 or Gla-100.

In total, 665 medical practitioners were recruited to the study during January to May 2018: 225 (33.8%) from Spain, 211 (31.7%) from France, and 229 (34.4%) from Germany. The majority were general/family practice physicians (n = 322; 48.4%), followed by diabetologists (n = 158; 23.8%), endocrinologists (n = 104; 15.6%) and internal medicine specialists (n = 81; 12.2%).

In this comparative real-world analysis of European data from insulin-experienced patients who switched BIA therapy to either Gla-300 or Gla-100, at 6 months after switching, improvements in glycemic control were similar in both treatment groups, with A1C reductions of about 1%. In multivariate analyses adjusting for baseline differences, switching to Gla-300 vs. Gla-100 was associated with a significantly greater reduction in the number of hypoglycemic events. Results for insulin-naïve patients initiating Gla-300 or Gla-100 were broadly similar, with A1C reductions of about 1.2–1.3% from baseline, and a significant difference in adjusted hypoglycemic events, with Gla-300 showing a decrease while Gla-100 showed a slight increase. Both groups may have benefited from further titration in terms of their A1C improvement.

At baseline, patients switching BIA therapy to Gla-300 vs. Gla-100 had a higher BMI, higher incidence of hypoglycemia, lower A1C, and a greater burden of comorbidities, while those initiating Gla-300 vs. Gla-100 were younger, were more often male and of European race/ethnicity, had a higher BMI, had more hypoglycemia, and had more comorbidities. In a previous study conducted in Germany and Austria, patients with T2D who were prescribed Gla-300 (switch or initiate) were also younger, had a higher BMI, and had lower A1C [31]. Our outcomes data are generally in line with those from the EDITION series of RCTs comparing Gla-300 and Gla-100. In the EDITION T2D studies, Gla-300 and Gla-100 were found to achieve similar glycemic control, but Gla-300 was associated with improvements in some hypoglycemia end points in insulin-experienced patients treated with basal-bolus therapy [5] or oral + basal therapy [6, 7] and in insulin-naïve patients [8, 9].

In the EDITION 1 study, which recruited patients on basal and meal-time insulin, switching to Gla-300 resulted in a similar reduction in A1C after 6 months (− 0.83%) [5] as that seen in our study (− 0.87%, adjusted). This result is notable given the assumed advantages of the RCT setting for achieving therapeutic end points, including treat-to-target titration schemes and the increased attention provided by the structure of the RCT, which may improve patient adherence to therapy [32]. However, it should be noted that mean baseline A1C was lower in EDITION 1 than in the present study (8.16% vs. 8.47%, respectively), body weight was higher (106 vs. 84 kg), and there was a requirement for a relatively high baseline use of basal insulin (≥ 42 U/day) in EDITION 1 [5], all factors that could potentially have an impact on observed A1C changes.

In the EDITION 3 study in insulin-naïve patients [8], mean baseline A1C was also lower than in the insulin-naïve population in the present study (8.54% vs. 8.68%); however, reductions in A1C were slightly more comparable (approximately − 1.5% vs. − 1.31% in the present study). In the current study, the small but significantly greater A1C reduction among Gla-100 vs. Gla-300 initiators (− 0.14%, p = 0.014) is unlikely to be of clinical significance. This was also at the cost of a slight increase in hypoglycemia, possibly the result of the different attitude toward patient management (i.e., greater focus on better glycemic control than on reduced hypoglycemia).

Our data also add to a growing body of real-world evidence demonstrating the effectiveness of switching to Gla-300. In previous real-world studies, insulin-experienced patients who switched to Gla-300 consistently showed important reductions in A1C [18, 21‐23, 27‐29, 32]. In those studies with a (mainly) first-generation BIA propensity score-matched comparator group, Gla-300 showed similar improvements in glycemic control [21‐23]. None of these comparative studies were conducted in Europe, but the results of two pragmatic studies that were mainly conducted in Europe have recently been published [26]. These were randomized real-world insulin-experienced and -naïve patients to Gla-300 or standard-of-care BIA (mainly Gla-100) [26]. Gla-300 was non-inferior for A1C reduction among insulin-naïve but not insulin-experienced patients [26].

In the current study, Gla-300 switchers had a larger unadjusted reduction in hypoglycemia events than Gla-100 switchers, but this was mainly due to a higher number of events during baseline as the number during follow-up was similar in both groups. However, after controlling for covariates, the reduction in hypoglycemia events with Gla-300 remained significant. Here, once again, our data are in line with other real-world studies, which have shown reduced hypoglycemia risk with Gla-300 vs. (mainly) first-generation BIAs [18, 21‐23, 33]. In the above-mentioned pragmatic studies that were mainly conducted in Europe, most hypoglycemia end points were similar for Gla-300 and standard-of-care BIA, but there were some significant reductions in nocturnal hypoglycemia with Gla-300 [26]. These hypoglycemia advantages are presumed to be derived from the more stable PK/PD profile of Gla-300 [13‐15].

The higher rate of baseline hypoglycemia in the Gla-300 switcher group is unsurprising, given that physicians cited a lower risk of hypoglycemia as a major reason for switching to Gla-300 and a lower rate of hypoglycemia was more often a primary expectation after switching in the patients for whom Gla-300 was chosen. These data are in line with a US study, where hypoglycemia advantages were stated as a reason to switch to Gla-300 [18]. However, a study from The Netherlands reported that reduced volume, improved flexibility, and better quality of life were more often cited as reasons for switching to Gla-300 than fear of hypoglycemia [19]. Interestingly, in a study that included patients with diabetes from 17 countries, the proportions who were concerned about hypoglycemia varied widely by country, from approximately 30% (in The Netherlands) to around 80% (in Algeria) [4]. In the current study, the higher rates of comorbidities in the Gla-300 group did not appear to be a driver of treatment choices, with only approximately 5% of physicians stating that this was a reason for switching, a similar proportion to that seen in the Gla-100 group. Familiarity with therapy was also previously seen to be a more frequent reason given by physicians when prescribing Gla-100 [18], and this was also observed in the present study. Gla-300 had relatively recently been approved (2015 in Germany and 2017 in France and Spain) when participating physicians were recruited to the study (early 2018), and so their reasoning for treatment selection might be expected to change as they gain experience and familiarity over time. Responses for the insulin-naïve group suggest that drivers of physician treatment choice were similar to those reported for the insulin-experienced group who were switching.

There appeared to be some country-specific differences in rates of hypoglycemia following insulin switching in our study, with a difference in severe hypoglycemia in favor of Gla-300 evident in Spain alone. However, it should be noted that the country-specific subgroup analyses involved relatively small sample sizes. Additionally, interpretation of results is complicated by differences in the approach to insulin titration that may be evident in different countries. Other studies have reported differences in hypoglycemia between different European countries, although these were more evident prior to insulin initiation than afterwards [34]. Such differences may, in part, be due to country-specific differences in factors such as the way that hypoglycemia is defined or recorded, patient characteristics, or the way that BIAs are titrated. Large global observational studies such as DISCOVER (n = 15,000), which is investigating real-world treatment patterns across 38 countries, may help shed light on the reasons behind such international differences in outcomes [35]. The REALI project, a pooled analysis of European studies, may also provide further information on Gla-300 use in Europe [36].

Our study has a number of limitations. As a retrospective analysis of patient medical records sourced from self-selected physicians, the data here represent a convenience sample that may not be generalizable to all patients using Gla-300 or Gla-100 in the countries included. Given the time frame of the study and the introduction of Gla-300 to the market, more patients were perhaps being switched to Gla-300 than other BIAs. Therefore, it was relatively difficult to find eligible patients who had recently switched to Gla-100. Although the analysis plan required similar numbers of patients in both the switcher and naïve groups, despite extending the ‘recruitment’ period, fewer of these patients were included in the database relative to the other groups. The analysis was nevertheless undertaken as sufficient numbers were available.

The data submitted by physicians may be subject to channeling bias and unmeasured confounders, such as socioeconomic status, or disease severity, and country-specific differences in how hypoglycemia is reported and how insulin titration is managed. Indeed, patients in the two groups do show some differences, and physicians were known to initiate therapies for different reasons. Analyses adjusted for baseline differences should minimize their influence, but some element of confounding may remain. Propensity score matching of included patients was considered but rejected in favor of the regression method because of the reductive nature of the method. Bolus/meal-time insulin use among BIA switchers and non-insulin diabetes medication use among BIA initiators fell dramatically after switching/initiation of Gla-300 or Gla-100, which could have impacted A1C and hypoglycemia results. However, as the changes were similar in both BIA groups, this should not have overly affected the between-group comparisons. Lastly, hypoglycemia levels were likely to be underestimated because of a reliance upon recorded events that were significant enough to show up in medical charts.

Despite the limitations, there are significant strengths of this study. Medical record abstraction provides an opportunity to assess real-world data outside the highly controlled setting of clinical trials. Data were collected from relatively large samples from three European countries, healthcare professional profiles were diverse, and it was possible to assess clinical, biologic, and BIA dose data, even if based on chart reviews. The random decision rule design may have helped control for some confounding aspects of the convenience sample, such as a tendency for physicians to select the most recent charts available.

This comparative European real-world study—which examined patients who were switched to or initiated Gla-300 or Gla-100—supports the results from RCTs and US real-world comparative studies that Gla-300 provides comparable glycemic control to Gla-100, but with a reduced risk of hypoglycemia. This study provides important additional evidence in European patients, as there were previously little comparative data in this population. Regarding choice of therapy, physicians were more likely to select Gla-300 to achieve a lower risk of hypoglycemia, whereas familiarity was more important when selecting Gla-100. It would be interesting to repeat this analysis when physicians are equally familiar with the use of Gla-300 and Gla-100.