Adv Ther (2021) 38:3857–3871 https://doi.org/10.1007/s12325-021-01773-z ORIGINAL RESEARCH Real-World Use of Insulin Glargine U100 and U300 in Insulin-Naı̈ve Patients with Type 2 Diabetes Mellitus: DosInGlar Study Natalia Duque . Esther Artime . Irene Romera . Jeremie Lebrec . Silvia Dı́az . Miriam Rubio . Antoni Sicras-Mainar . Enrique Carretero-Anibarro . Xavier Mundet . Juan J. Gorgojo-Martı́nez . Jesús Reviriego Received: February 5, 2021 / Accepted: May 5, 2021 / Published online: May 29, 2021  The Author(s) 2021 ABSTRACT Methods: In this observational, retrospective cohort study, electronic medical records in the Introduction: In the EDITION clinical trial Big-Pac database (Real Life Data) relating to programme, patients with type 2 diabetes mel- adult insulin-naı̈ve patients with T2DM who litus (T2DM) receiving insulin glargine (IGlar) initiated IGlar U100 or U300 treatment in Spain U300 required 10–15% more insulin than those in 2016–2017 and remained on treatment for receiving IGlar U100. This study sought to 18 months were selected. IGlar U100- and determine whether this difference was apparent U300-treated patients were matched 1:1 in real-world practice. (propensity score matching). The primary analysis compared changes from baseline in Supplementary Information The online version mean daily IGlar dose (U and U/kg) at 6 (± 2), contains supplementary material available at https:// 12 (± 2) and 18 (± 2) months between cohorts doi.org/10.1007/s12325-021-01773-z. (paired t tests). Changes in glycated haemoglo- bin (HbA1c) and weight were analysed N. Duque (&)  E. Artime  I. Romera  S. Dı́az  descriptively. M. Rubio  J. Reviriego Departamento Médico, Lilly Alcobendas, Avda. de la Results: The IGlar U100 and U300 cohorts Industria 30, Alcobendas, 28108 Madrid, Spain included 556 matched pairs (46.9% female) e-mail: duque_natalia@lilly.com with the following mean (standard deviation) J. Lebrec values at baseline, respectively: age 63.6 (12.8) HaaPACS GmbH, Schriesheim, Germany versus 63.7 (11.9) years; years since diagnosis 9.5 (1.4) versus 9.5 (1.3); HbA1c 8.8 (1.3) versus A. Sicras-Mainar 8.7 (1.5) %; weight 84.6 (16.9) versus 84.7 (17.1) Atrys Health SA, Health Economics and Outcomes Research, Barcelona, Spain kg. Mean IGlar dose at baseline was 0.19 U/kg/day (both cohorts). Patients receiving E. Carretero-Anibarro IGlar U300 showed a greater increase from Medicina Familiar y Comunitaria, Unidad de Gestión Clı́nica Estepa, Sevilla, Spain baseline in IGlar dose at 6, 12 and 18 months [mean dose (U/kg/day) 5.1%, 10.3% and 12.8% X. Mundet greater, respectively, in IGlar U300-treated Universitat Autonoma de Barcelona, Bellaterra, Spain patients]. Mean HbA1c was 8.1% in both cohorts at 18 months. Mean (SD) weight at J. J. Gorgojo-Martı́nez 18 months with IGlar U100 and IGlar300 was Department of Endocrinology and Nutrition, 86.8 (17.0) kg and 85.0 (17.1) kg, respectively. Hospital Universitario Fundación Alcorcón, Madrid, Spain 3858 Adv Ther (2021) 38:3857–3871 Conclusion: In real-world practice, insulin dose DIGITAL FEATURES was significantly higher in IGlar U300-treated than U100-treated patients at 6, 12 and This article is published with digital features, 18 months, with similar reductions in HbA1c. including a summary slide, to facilitate under- At equal IGlar price/unit in Spain, the increased standing of the article. To view digital features dose requirements of IGlar U300 would result in for this article go to https://doi.org/10.6084/ higher costs. m9.figshare.14535843. Keywords: Cost; Dose; Glycaemic control; Insulin glargine; Real-world evidence; Type 2 INTRODUCTION diabetes mellitus; U100; U300 Type 2 diabetes mellitus (T2DM) is a progressive disease that requires treatment intensification Key Summary Points to maintain glycaemic control over time [1]. As a result, many patients eventually require Why carry out this study? insulin [2]. When insulin therapy is first intro- In clinical trials, patients with type 2 duced into the antidiabetic regimen of a patient diabetes mellitus (T2DM) receiving insulin with T2DM, basal insulin is usually the first glargine (IGlar) U300 required 10–15% more option [3]. Insulin glargine (IGlar) was the first insulin than those receiving IGlar U100 once-daily, long-acting insulin analogue to be introduced into clinical practice [4]. At this There is limited real-world evidence about time, it was formulated as a 100 unit/ml the doses of these two IGlar preparations preparation [IGlar U100 (Lantus)] [4]. IGlar The real-world DosInGlar study aimed to U100 has demonstrated consistent efficacy and determine the doses of insulin for IGlar safety in numerous randomised controlled trials U300 versus IGlar U100 in insulin-naı̈ve [4] and real-world evidence studies [5] and rep- patients with T2DM in the 18 months after resents the ‘gold standard’ against which new treatment initiation basal insulin analogues are assessed [6]. Since the introduction of IGlar U100, a number of What was learned from this study? 100 unit/ml insulin glargine biosimilars with The results showed that insulin doses over similar efficacy and safety profiles have become time are higher in patients receiving IGlar available [7]. U300 versus IGlar U100 in the real-world IGlar is also available as a 300 unit/ml for- setting, although glycaemic control is mulation [IGlar U300 (Toujeo)] [8]. IGlar similar for both preparations U100 and IGlar U300 have different pharma- cokinetic/pharmacodynamic profiles. The mean When initiating insulin treatment, in duration of action of IGlar U100 is 25.5 h under addition to an antidiabetic effect, risk of euglycaemic clamp conditions, following a sin- hypoglycaemia and effect on body weight, gle-dose of 0.3 U/kg [9]. The pharmacokinetic physicians should assess the possible cost profiles of IGlar U100 and IGlar U300 were implications associated with total dose, compared following seven 0.4 U/kg doses with which will potentially be higher with IGlar estimated median terminal half-lives of 19 and U300 13.5 h, respectively [10]. IGlar U300 also has a flatter and more extended time-action profile than IGlar U100 [11], and this may result in more stable and sustained glycaemic control over the 24-h inter-dosing interval [11]. An additional difference between these two insulin preparations is that, for the same number of units of insulin, the volume of IGlar U300 is Adv Ther (2021) 38:3857–3871 3859 approximately one-third that of IGlar U100 longer follow-up period than has been reported [11]. previously (18 months). In addition, we aimed The EDITION clinical trial programme [11] to describe the sociodemographic and clinical was designed to evaluate the efficacy and safety characteristics of IGlar U100- and IGlar U300- of IGlar U300 versus IGlar U100. The majority treated patients at the time of treatment initia- of the EDITION trials enrolled patients who had tion and the changes in HbA1c, fasting plasma previously used insulin, but EDITION 3, which glucose (FPG), and body weight up to 18 enrolled 878 insulin-naı̈ve patients with T2DM months after treatment initiation. from across North America, Europe and Japan, was an exception [11, 12]. In this trial, although the basal insulin dose increased in both groups, METHODS the mean dose was *17% greater in the IGlar U300 group at 6 months (0.62 U/kg/day vs. 0.53 Study Design and Patient Population U/kg/day), with no significant difference between groups in glycaemic control and noc- The DosInGlar study was an observational ret- turnal symptomatic hypoglycaemia [12]. These rospective cohort study based on electronic disparities in insulin dose between IGlar U300 medical records in the ENCePP-registered Big- and IGlar U100 reflect the results of the EDI- Pac database [20] in Spain (Fig. 1). Big-Pac is a TION programme as a whole, in which a source of dissociated and anonymised medical 10–15% higher dose of IGlar U300 versus IGlar records from 1.8 million patients attending U100 was typical [11], and are acknowledged in primary care and specialised centres. The the summaries of product characteristics for the records held by Big-Pac are representative of the IGlar U300 and U100 products [8, 13, 14]. Spanish population in terms of prevalence of The EDITION data were generated as part of specific diseases (including diabetes) and bio- a clinical trial programme. However, the chemical and anthropometric patient charac- importance of taking real-world data into teristics [21]. The medical records held by Big- account when making treatment decisions is Pac are collected monthly from a panel of [ gaining recognition [15–17], and it is for this 1200 office-based primary care physicians and reason that we conducted the DosInGlar study. 900 specialists in 7 of the 17 regions in Spain. Until now, data relating to the dosage profile of The database includes information relating to IGlar U300 vs. IGlar U100 in insulin-naı̈ve demographic characteristics, diagnoses, mor- patients with T2DM in routine clinical practice bidity and mortality data [International Classifi- were limited to a 6-month retrospective study cation of Diseases (ICD) codes], medication (REALITY), which involved 376 propensity details, clinical scales, biochemistry data, score-matched patients [18], and a retrospective anthropometric characteristics and use of study utilising medical record data to investi- resources (physician visits, hospital admissions, gate treatment dosing patterns and clinical tests ordered). Data relating to hypoglycaemic outcomes in 390 patients, published by Gupta episodes are included in the database, but mild et al. [19]. In the REALITY population, the mean and moderate cases may be under-reported. dose of IGlar U300 was *17% higher than the This study was approved by the local ethics mean dose of IGlar U100 at the end of follow- committee (Hospital de Tarrasa in Barcelona). up, with no difference between groups in mean The Big-Pac database was used to identify glycated haemoglobin (HbA1c) [18], whereas adult patients with a diagnosis of T2DM [ICD, Gupta et al. [19] reported similar doses of IGlar Tenth Revision, Clinical Modifications (ICD-10- U100 and U300 and a similar HbA1c between CM) code: E10–E14] who initiated treatment groups. with IGlar U100 or IGlar U300 (identified from The primary aim of the DosInGlar study was pharmacy claims file anatomical therapeutic to compare the dosage profiles of IGlar U100 chemical codes) during the period 1 January and IGlar U300 when used in insulin-naı̈ve 2016 to 31 December 2017 inclusive (the patients with T2DM in clinical practice over a inclusion period; Fig. 1). At this time, 3860 Adv Ther (2021) 38:3857–3871 Fig. 1 DosInGlar study design. Months 6, 12, and 18 = 6 ± 2, 12 ± 2, and 18 ± 2 months after the index date, respectively. IGlar insulin glargine LY2963016 (Abasaglar) was the only IGlar Study Endpoints and Data Extraction biosimilar available in Spain; the insulin for- mulations used by patients therefore comprised The primary study objective was to compare the Toujeo (IGlar U300), Lantus, or Abasaglar change from baseline in daily dose (U and U/kg) (both IGlar U100). The date on which each of IGlar at 6, 12 and 18 months after the index patient initiated IGlar U100 or IGlar U300 date between the IGlar U100- and IGlar U300- treatment (the index therapy) was designated as treated cohorts. Dispensed medication was used the index date, and patients were followed for for the analysis. The daily doses of IGlar U100 18 months from this date. In addition to index, and U300 (U/day) were recorded for the index the following time windows at each temporal date (first prescription). At months 6, 12 and 18, data point were defined: 6 (± 2), 12 (± 2) or 18 the daily dose was calculated as the mean of the (± 2) months (hereafter written as 6, 12 and doses [U/day (provided in the record) and 18 months). U/kg/day (calculated)] for all dispensed pre- Inclusion criteria included T2DM; scriptions within the relevant ± 2-month age C 18 years on the index date; insulin naı̈ve interval. The secondary study objectives were to for C 1 year prior to the index date; received describe the sociodemographic and clinical one or more prescription during the follow-up characteristics at index and the changes in period; two or more health records included in HbA1c, FPG and body weight at 6, 12 and the computer system; and remained on IGlar 18 months in the two cohorts. Comorbidity treatment, without discontinuation, burden was assessed at index by the Charlson for C 18 months from the index date. Discon- Comorbidity Index [22]. The HbA1c, FPG and tinuation was defined as a period of C 6 months body weight values used were those obtained during which no insulin was dispensed. Patients within 6 (± 2), 12 (± 2) and 18 (± 2) months were excluded from the study if they were per- after the index date. manently institutionalised, had previously received IGlar therapy, were receiving a gluca- Statistical Analysis gon-like peptide 1 receptor agonist (GLP-1 RA) or prandial insulin on the index date, had a To minimize selection bias, propensity score diagnosis of type 1 or gestational diabetes, or if matching (PSM) was used to match each patient their record was missing two or more temporal who initiated IGlar U100 therapy 1:1 with a data points for the same variable among any of patient who initiated IGlar U300. Baseline HbA1c, FPG, body weight or IGlar dose. covariates used in the PSM multivariate Adv Ther (2021) 38:3857–3871 3861 regression model included age, sex, time diagnosis of T2DM, 2492 initiated IGlar U100 from diagnosis, macrovascular complications, and 1297 initiated IGlar U300 during the microvascular complications, antidiabetic med- inclusion period. Approximately 2% of each ication, concomitant medication, HbA1c, body group were lost to follow-up, and 24% in each weight, and IGlar dose at index. Patients were group discontinued index therapy before matched using the greedy nearest neighbour 18 months (Fig. 2). After exclusions for missing method without replacement, with a calliper of data, loss to follow-up, discontinuation of the 0.2 standard deviations (SDs). index therapy after \ 18 months, and use of a Based on an IGlar daily dose SD of 25 units GLP-1 RA or prandial insulin on the index date, [23], a sample size of 500 per cohort would the IGlar U100 and IGlar U300 cohorts included allow detection of a difference in glargine dose 1109 and 559 patients, respectively. This was of C 5.1 U/day (i.e. 0.06 U/kg/day for an 85 kg reduced to 556 patients in each cohort after individual) with 90% power (two-sided two- subsequent PSM (Fig. 2). sample t test with a type I error rate of 0.05). The baseline sociodemographic characteris- These calculations did not account for match- tics of the two cohorts were similar prior to ing between samples and subsequent reduction [Table S1 in the electronic supplementary in variation of the outcome difference between material (ESM)] and after (Table 1) matching, cohorts and were therefore conservative. The with similar values in the IGlar U100 and IGlar 1:1 matching that was inherent in the PSM U300 matched cohorts for percentage of female procedure reduced the effective sample size to patients (46.9% for both), mean age (63.6 and that of the smaller of the two cohorts. 63.7 years), time since diagnosis of T2DM Changes from baseline in IGlar daily dose (U (9.5 years for both), index HbA1c (8.8 and and U/kg) at 6, 12 and 18 months were com- 8.7%), index body weight (84.6 and 84.7 kg), pared between the matched cohorts using index FPG (10.8 and 10.9 mmol/l) and Charlson paired t tests. A hierarchical testing strategy was Comorbidity Index score (2.1 and 2.2). The used (i.e. the difference between the two percentages of each cohort receiving each of a cohorts in dose change from baseline was tested range of concomitant antidiabetic medications sequentially in the following order: month 18, were also similar (Table 1). month 12, month 6). Subsequent time points At index, the mean daily dose of IGlar was were only tested if all previous time points had 0.19 U/kg in both cohorts (Table 2) (absolute shown a statistically significant difference. This mean daily doses: IGlar U100, 16.3 U; IGlar ensured overall control of the type I error. Sec- U300, 16.1 U). The mean change from baseline ondary objectives (changes in HbA1c, FPG and in IGlar dose (U/kg/day) was significantly body weight at 6, 12 and 18 months) were higher in the IGlar U300-treated patients than analysed descriptively. Absolute values of IGlar in those receiving IGlar U100 at 6, 12 and dose, HbA1c, FPG and body weight at each time 18 months, with differences of 0.02, 0.04 and point were also computed. Qualitative data 0.05 U/kg/day, respectively (p B 0.001 for all; were expressed as absolute and relative fre- mean doses of 0.44 vs. 0.39 U/kg/day and 21.1 quencies and quantitative data as mean, SD and vs. 17.2 U/day at 18 months). This represents 95% confidence intervals (CIs). A p value \ 0.05 5.1%, 10.3% and 12.8% higher doses of IGlar signified statistical significance. U300 vs. IGlar U100 at these time points All analyses were performed using SPSSWIN (Fig. 3). v23 (IBM SPSS Statistics: IBM, Armonk, NY, Mean HbA1c decreased by similar amounts USA). in both groups at 6, 12 and 18 months, achieving mean values of 8.1% and 8.1% in the IGlar U100 and IGlar U300 groups, respectively, RESULTS at 18 months (Table 2; Fig. 4). Mean FPG also decreased by similar amounts in the two groups From a total database patient population of (IGlar U100 vs. IGlar U300: 6 months, - 0.4 49,317 who were aged C 18 years with a vs. - 0.4 mmol/l; 12 months, - 0.4 vs. 3862 Adv Ther (2021) 38:3857–3871 Fig. 2 Patient flowchart. *Baseline covariates included age, replacement. GLP-1 RA glucagon-like peptide-1 receptor sex, time from diagnosis, HbA1c, weight, IGlar dose, agonist; HbA1c glycated haemoglobin; IGlar insulin macrovascular complications, microvascular complications, glargine; PSM propensity score matching; T2DM type 2 antidiabetic medication and concomitant medication. diabetes mellitus Method: 1:1 PSM; greedy nearest neighbour without - 0.6 mmol/l; 18 months, - 0.6 vs. DISCUSSION - 0.8 mmol/l) (Fig. 4), whereas mean body weight—which was similar between groups at The DosInGlar study suggests that, in real-world baseline—increased more in the IGlar U100 clinical practice, similar mean HbA1c and FPG group than in those receiving IGlar U300 at 6, values were achieved using 13% higher doses of 12 and 18 months, culminating in mean IGlar U300 versus IGlar U100 at 18 months after increases of 2.1 kg (IGlar U100) and 0.3 kg treatment initiation in insulin-naı̈ve patients (IGlar U300) at 18 months [between-group dif- with T2DM. Our results therefore support the ference, IGlar U100 vs. IGlar U300: 1.87 kg current American Diabetes Association (ADA)/ (95% CI 1.42–2.31)]. Mean (SD) weight at European Association for the Study of Diabetes 18 months with IGlar U100 and IGlar300 was (EASD) consensus report for the management of 86.8 (17.0) kg and 85.0 (17.1) kg, respectively. hyperglycaemia in T2DM and the European summary of product characteristics for Toujeo Adv Ther (2021) 38:3857–3871 3863 Table 1 Sociodemographic characteristics and concomitant treatments at baseline in matched cohorts Patient characteristic/treatment IGlar U100 IGlar U300 SMDa N = 556 N = 556 Age, years 63.6 (12.8) 63.7 (11.9) 0.008 Female sex (%) 46.9 46.9 0.001 Time from diagnosis, years 9.5 (1.4) 9.5 (1.3) 0.002 HbA1c, % 8.8 (1.3) 8.7 (1.5) 0.069 FPG, mmol/l 10.8 (2.2) 10.9 (2.5) 0.043 Body weight, kg 84.6 (16.9) 84.7 (17.1) 0.006 Charlson Comorbidity Index 2.2 (1.5) 2.1 (1.5) 0.067 b Antidiabetic treatment (%) Oral combination therapy 42.8 39.4 0.069 Metformin 32.0 36.2 0.089 Sulphonylurea 29.1 29.1 0.000 DPP-4 inhibitor 25.7 24.3 0.032 SGLT-2 inhibitor 9.7 11.2 0.049 Glinide 8.8 9.9 0.038 a-Glucosidase inhibitor 5.0 4.0 0.048 Glitazone 0.2 0.7 0.075 All values are mean (standard deviation) unless specified otherwise DPP dipeptidyl peptidase; FPG fasting plasma glucose; HbA1c glycated haemoglobin; IGlar insulin glargine; SGLT sodium- glucose cotransporter; SMD standardised mean difference a SMD \ 0.10 is commonly considered to signify no residual imbalance b The treatments that comprise ‘oral combination therapy’ were not recorded; therefore, the percentage of patients shown as receiving metformin represents those receiving metformin as monotherapy and is likely to be lower than the overall percentage receiving metformin (IGlar U300), which state that a 10–14% [3] or U300 and IGlar U100 is currently unknown, it 10–18% [8] higher dose of IGlar U300 versus may be due to IGlar U300 spending more time IGlar U100 may be required to achieve similar in the subcutaneous depot than IGlar U100, glycaemic targets. Our results are also aligned thereby allowing greater enzymatic inactivation with the results of randomised clinical trials and of the glargine molecule [25]. other real-world studies that have suggested The difference in dosing profile between that higher doses of IGlar U300 than IGlar U100 IGlar U100 and IGlar U300 has a number of are needed to achieve a particular level of gly- implications for both patients and healthcare caemic control [11, 18, 24–26]. In many of these systems. Based on the findings of the DosInGlar studies, as in the DosInGlar study, these study, in countries where IGlar U100 and IGlar patients with T2DM were insulin naı̈ve when U300 have the same (or similar) cost per unit IGlar therapy was initiated [18, 24, 25]. While (e.g. Spain [27]), the higher dose requirements the reason for the dose difference between IGlar of IGlar U300 to achieve a similar glycaemic 3864 Adv Ther (2021) 38:3857–3871 Table 2 Mean IGlar dose and HbA1c at baseline and 6, 12 and 18 months after initiation of treatment in matched cohorts IGlar/HbA1c Value Change from baseline Difference in change from baseline U100 U300 U100 U300 U100–U300 IGlar mean (SD) dose (U/kg/day) Baseline 0.19 (0.14) 0.19 (0.17) – – – 6 months 0.39 (0.17) 0.41 (0.22) 0.20 (0.09) 0.22 (0.12) - 0.02; p = 0.001a 12 months 0.39 (0.17) 0.43 (0.23) 0.20 (0.13) 0.24 (0.14) - 0.04; p \ 0.001a 18 months 0.39 (0.17) 0.44 (0.23) 0.20 (0.15) 0.25 (0.16) - 0.05; p \ 0.001a IGlar mean (SD) dose (U/day) Baseline 16.3 (11.9) 16.1 (14.1) – – 6 months 33.1 (16.2) 34.3 (19.4) 16.8 (6.2) 18.2 (10.2) - 1.4; p = 0.034a 12 months 33.4 (16.5) 35.9 (19.5) 17.1 (7.6) 20.8 (10.8) - 3.7; p \ 0.001a 18 months 33.5 (16.4) 37.2 (19.9) 17.2 (8.1) 21.1 (11.0) - 3.9; p \ 0.001a Mean (SD) HbA1c, % Baseline 8.78 (1.35) 8.72 (1.53) – – – 6 months 8.27 (1.32) 8.33 (1.19) - 0.51 (1.1) - 0.39 (1.4) - 0.12 (1.7) 12 months 8.26 (1.33) 8.19 (1.22) - 0.52 (1.2) - 0.53 (1.5) 0.00 (1.9) 18 months 8.15 (1.24) 8.09 (1.17) - 0.63 (1.2) - 0.63 (1.5) - 0.01 (1.9) N = 551–556 for all values HbA1c glycated haemoglobin; IGlar insulin glargine; SD standard deviation a Change from baseline in IGlar dose compared between groups using paired t tests (statistical significance, p \ 0.05) control would result in higher treatment costs Although hypoglycaemia data are not avail- (13% more expensive, on a unit/kg basis, if the able for the DosInGlar study, this topic merits price per unit is equal). Differences among discussion because it is an important consider- countries in healthcare systems, insulin pre- ation when determining which insulin prepa- scribing and utilisation patterns, etc., may need ration should be prescribed for a patient. to be considered before these results can be Although the current ADA/EASD consensus extrapolated to other countries. A further report states that IGlar U300 is associated with a implication is that it may take longer to reach lower risk of nocturnal hypoglycaemia than the optimal dose of insulin with IGlar U300 IGlar U100 [3], this is not conclusively borne than with IGlar U100. This is supported by the out by clinical trial results or real-world data results of the EDITION clinical trial programme involving insulin-naı̈ve patients with T2DM. [11] and by our data, which show that the mean For example, the annualised rate of hypogly- daily dose of IGlar U100 remained unchanged caemia was not significantly lower for IGlar between 6 and 18 months, but the dose of IGlar U300 versus IGlar U100 in any of the eight U300 still had not plateaued 18 months after categories of hypoglycaemia examined in EDI- treatment initiation (Fig. 3). TION AP [24] and in only one of the eight cat- egories (documented symptomatic Adv Ther (2021) 38:3857–3871 3865 Fig. 3 Mean dose of IGlar U300 and IGlar U100 at baseline and 6, 12 and 18 months after treatment initiation. IGlar insulin glargine. Values represent mean ± 95% confidence intervals hypoglycaemia: B 3.9 mmol/l over 24 h) in reporting lower weight gain in the IGlar U100 EDITION 3 [12]. Moreover, in EDITION 3, the group than the IGlar U300 group [18, 24]. percentage of participants experiencing the However, in the DosInGlar study, mean body main secondary efficacy endpoint (one or more weight increased to a greater extent in the IGlar nocturnal confirmed or severe hypoglycaemic U100-treated group than in those receiving event) was 16% with IGlar U300 and 17% with IGlar U300. An influence of concomitant med- IGlar U100 [relative risk 0.89 (95% CI ication is unlikely to underlie this finding, as no 0.66–1.20)] [12], and in the real-world REALITY apparent differences in background antidiabetic study, similar on-study percentage increases therapy between groups were found. Weight were seen in the proportions of IGlar U100- and gain and hypoglycaemia often co-present as IGlar U300-treated patients who reported at unwanted effects of insulin treatment in least one incidence of hypoglycaemia per week patients with T2DM [28], and there is some (IGlar U100: 5.5% at baseline, 11.0% at follow- evidence for an association between weight gain up; IGlar U300: 3.3% at baseline, 8.7% at fol- and hypoglycaemia from the Diabetes Control low-up) [18]. and Complications Trial (DCCT) [29]. There- Given the potential for weight gain in insu- fore, it could be hypothesised that the increased lin-treated patients [3], body weight is another weight gain in the IGlar U100 group reflects a important consideration, and we should again relative increase in hypoglycaemic events com- look to clinical trial and real-world data pared with the IGlar U300 group. Hypogly- involving insulin-naı̈ve patients with T2DM for caemia data were not available for the insight. Other studies have consistently shown DosInGlar study, although it has been evaluated similar mean on-study changes in body weight in other studies of insulin-naı̈ve T2DM patients in IGlar U100- and IGlar U300-treated patients that compared IGlar U100 and IGlar U300. The with T2DM, with no significant differences 6-month REALITY study (18) showed that between treatments in either clinical trials or insulin-naı̈ve patients previously treated with real-world studies [18, 24, 25] and some studies IGlar U100 had higher rates of hypoglycaemia 3866 Adv Ther (2021) 38:3857–3871 Fig. 4 Mean change from baseline in HbA1c (a), mean Values in graphs represent mean ± 95% confidence FPG (b) and HbA1c (c) at 6, 12 and 18 months in intervals patients treated with IGlar U300 or IGlar U100. FPG, fasting plasma glucose; HbA1c, glycated haemoglobin; IGlar insulin glargine; m months; SD standard deviation. relative to those treated with IGlar U300, differences between cohorts in this outcome. although this was not accompanied by addi- These data should therefore be interpreted with tional weight gain. Moreover, data from the caution. EDITION-3 trial show that annualised rates of It is generally accepted that similar levels of nocturnal confirmed (B 3.9 mmol/l) or severe glycaemic control are achieved with IGlar U100 hypoglycaemic events were similar for IGlar and IGlar U300 [11], and this is supported by U100 and IGlar U300 at 6 and 12 months the results of the DosInGlar study. However, it [12, 25], with a slight (not significant) increase is notable that the mean HbA1c at 18 months in weight gain in the IGlar U100 group at 6 and after treatment initiation (8.1% in both treat- 12 months. It seems difficult to consolidate ment groups) was higher than the goal of 7.0% these previous data with the possibility that that is typically recommended for most patients increased weight gain with IGlar U100 in the [3]. As already noted, the IGlar U300 dose had DosInGlar study was related to hypoglycaemia not plateaued 18 months after treatment initi- (although it cannot be excluded). Alternatively, ation, and we therefore cannot exclude the other factors such as lifestyle interventions and possibility of a further reduction in HbA1c for socio-economic status, for which data were also IGlar U300 if follow-up had been longer. How- not collected, could have influenced this ever, the incline of the slope between 12 and parameter. In addition, it is important to note 18 months is slight (Fig. 3), which suggests that that body weight was a descriptive endpoint any further dose increase would have been and that the study was not designed to test minimal (potential for additional reduction in Adv Ther (2021) 38:3857–3871 3867 HbA1c is therefore predicted to be negligible; of the study and the fact that it was based on moreover, robust analysis of any resulting dif- electronic medical records mean that some data ference would not be possible because of the may not have been recorded. Moreover, the lack of statistical power to detect such small data available were limited (e.g. there were no dose differences). The suboptimal HbA1c values data relating to time of IGlar administration or at 18 months indicate that whichever IGlar adherence, and the hypoglycaemic event data formulation patients with T2DM receive in real- were not assessed as they were only partially world clinical practice, additional lifestyle and available because of under-reporting). The pharmacological interventions and/or intensi- analysis did not account for individual antidia- fication of insulin therapy are frequently betic therapies used as part of ‘oral combination needed. Lifestyle interventions form the therapy’. As a result, the percentages of patients cornerstone of effective diabetes management, recorded as using metformin may be artificially and clinical guidelines place an emphasis on low. The implication of increased cost for IGlar both lifestyle management and patient educa- U300 (because of the higher dose needed to tion [3]. Therapeutic inertia represents an achieve a similar antidiabetic effect to U100) is additional problem in patients with T2DM [30]. applicable where the price per unit of insulin is Nevertheless, adherence to the detailed guid- similar or equal (e.g. Spain [27], at time of ance provided by current guidelines on active publication), but this may not necessarily be the management [3] should increase the likelihood case in other regions or may be subject to future of treatment goals being achieved. change. Additional limitations include the There were some differences in the baseline requirement for patients to be on insulin for sociodemographic characteristics of this study 18 months. This means that our findings can- population versus the randomised EDITION-3 not be generalised to all IGlar-treated patients [12] multicentre trial that also compared IGlar with T2DM. It is also possible that our findings U100 and U300 in an insulin-naı̈ve population may be affected by confounding factors (e.g. with type 2 diabetes; patients in this real-world level of education and socio-economic status), study were older and had a lower body weight which may affect compliance with lifestyle and than those in the EDITION-3 study (time from medication recommendations [33–36] and for diagnosis and HbA1c were similar between which data were not available, and by selection these studies). The strict inclusion and exclu- bias. The potential for the latter was minimised, sion criteria of randomised controlled trials may however, by the use of PSM. The use of lead to discrepancies in terms of age, sex, descriptive analyses only for the secondary comorbidities, disease severity and comedica- objectives represents a further limitation. tion [31, 32]. Patients in controlled trials may The potential for real-world evidence to also exhibit different behaviours, including complement clinical trials data is increasingly higher medication adherence and increased being recognised [37], and the confirmation of tolerance of adverse events [31]. Compared to clinical trial data provided by the DosInGlar the real-world REALITY study, in a Canadian study is important. We have shown that, in population, patients in our study were older, insulin-naı̈ve patients with T2DM who initiate had shorter times from diagnosis and lower insulin therapy, use of IGlar U300 is associated HbA1c—this supports an influence of region on with higher doses of insulin than IGlar U100 in sociodemographic and disease characteristics of real-world clinical practice. Moreover, the populations with T2DM in real-world settings degree of glycaemic control achieved was simi- (i.e. a Spanish population versus a more diverse lar with the two preparations. The findings of population distributed across several countries). this real-world study are supported by those of The potential for regional differences in popu- the EDITION 3 and EDITION AP clinical trials, lation characteristics should therefore be con- which have also shown that, in previously sidered when comparing different studies. insulin-naı̈ve patients with T2DM, the dose of This study was subject to a number of limi- IGlar U300 associated with a particular level of tations. The retrospective, observational nature glycaemic control is higher than the dose of 3868 Adv Ther (2021) 38:3857–3871 IGlar U100. Taking into account the similarities and critical revision of the manuscript for between IGlar U300 and IGlar U100 in terms of important intellectual content. Silvia Dı́az was the baseline sociodemographic characteristics involved in the conception and design of the and disease burden (Charlson Comorbidity study, interpretation of the data and critical Index) of patients with T2DM who initiate revision of the manuscript for important intel- therapy, an antidiabetic effect (reduction in lectual content. Jesús Reviriego was involved in HbA1c and FPG), risk of hypoglycaemia and the conception and design of the study, acqui- effect on body weight, physicians should assess sition and analysis of the data, interpretation of the possible cost implications associated with the data and critical revision of the manuscript total dose—which will potentially be higher for important intellectual content. Xavier with IGlar U300—when initiating insulin Mundet was involved in the interpretation of treatment. the data for the study and critical revision of the manuscript for important intellectual content. Juan J. Gorgojo-Martinez was involved in the ACKNOWLEDGEMENTS interpretation of the data for the study, drafting of the manuscript and critical revision of the manuscript for important intellectual content. Funding. This study was sponsored by Eli Jeremie Lebrec was involved in the analysis of Lilly and Company, Indianapolis, IN, USA, who the data, interpretation of the data and critical also funded the rapid publication of this article. revision of the manuscript for important intel- lectual content. Enrique Carretero-Anibarro was Medical Writing Assistance. The authors involved in the analysis of the data, interpreta- would like to acknowledge Janet Douglas and tion of the data and critical revision of the Sheridan Henness (Rx Communications, Mold, manuscript for important intellectual content. UK) for medical writing assistance with the Antoni Sicras-Mainar was involved in the con- preparation of this article. Support for this ception and design of the study, acquisition and assistance was funded by Eli Lilly and Com- analysis of the data, interpretation of the data pany, Indianapolis, IN, USA. and critical revision of the manuscript for important intellectual content. Authorship. All named authors meet the International Committee of Medical Journal Prior Presentation. The data presented in Editors (ICMJE) criteria for authorship for this this paper have been previously presented (Ar- article, take responsibility for the integrity of time E et al. Dosage profile of insulin glargine the work as a whole, and have given their U100 and U300 in patients with type 2 diabetes approval for this version to be published. mellitus in real clinical practice: DosInGlar Study. XXXI Congress of the Spanish Society of Authorship Contributions. Miriam Rubio Diabetes, 21–23 June 2020; Duque N et al. was involved in the analysis of the data, inter- Dosage profile of insulin glargine U100 and pretation of the data, drafting of the manuscript U300 in patients with type 2 diabetes mellitus and critical revision of the manuscript for in real clinical practice: DosInGlar Study. Euro- important intellectual content. Irene Romera pean Association for the Study of Diabetes 56th was involved in the conception and design of annual meeting, 21–25 September 2020). the study, interpretation of the data and critical revision of the manuscript for important intel- Disclosures. Irene Romera, Natalia Duque, lectual content. Natalia Duque was involved in Silvia Dı́az and Miriam Rubio are full-time the interpretation of the data for the study and employees of Lilly Spain. Jeremie Lebrec is a critical revision of the manuscript for important consultant for Eli Lilly and Company. Miriam intellectual content. Esther Artime was involved Rubio reports being a minor shareholder in Eli in the conception and design of the study, Lilly. Irene Romera, Natalia Duque, Esther analysis of the data, interpretation of the data Artime and Silvia Dı́az report ownerships of Eli Adv Ther (2021) 38:3857–3871 3869 Lilly shares. Jesús Reviriego reports being an Eli REFERENCES Lilly and Company employee until January 17 2020, and a minor shareholder in Eli Lilly and 1. Fonseca VA. Defining and characterizing the pro- Company. Xavier Mundet reports personal fees gression of type 2 diabetes. Diabetes Care. from Boehringer Ingelheim Pharmaceuticals 2009;32(Suppl 2):S151–6. https://doi.org/10.2337/ Inc., Eli Lilly Company and Merck Sharp & dc09-S301. Dohme. Juan J. Gorgojo-Martinez reports per- 2. Inzucchi SE, Bergenstah RM, Buse JB, et al. Man- sonal fees from Astra-Zeneca, Abbott, AbbVie agement of hyperglycemia in type 2 diabetes, 2015: Inc, Boehringer Ingelheim Pharmaceuticals a patient-centered approach: update to a position Inc., Esteve, Janssen Pharmaceuticals, Eli Lilly statement of the American Diabetes Association and the European Association for the Study of and Company, Merck Sharp & Dohme, Novo- Diabetes. Diabetes Care. 2015;38(1):140–9. https:// Nordisk, Pfizer, Roche Pharma, Sanofi-Aventis doi.org/10.2337/dc14-2441. and Mundipharma, outside the submitted work. Jeremie Lebrec reports personal fees from Eli 3. Davies MJ, D’Alessio DA, Fradkin J, et al. Manage- ment of hyperglycemia in type 2 diabetes, 2018. A Lilly, during the conduct of the study. Enrique consensus report by the American Diabetes Associ- Carretero-Anibarro and Antoni Sicras-Mainar ation (ADA) and the European Association for the have no conflicts of interest to declare. Study of Diabetes (EASD). Diabetes Care. 2018;41(12):2669–701. https://doi.org/10.2337/ Compliance with Ethics Guidelines. This dci18-0033. study was approved by the ethics committee of 4. Hilgenfeld R, Seipke G, Berchtold H, Owens DR. The the Hospital de Tarrasa in Barcelona (study ref- evolution of insulin glargine and its continuing erence number: 02-19-399-039). contribution to diabetes care. Drugs. 2014;74(8): 911–27. https://doi.org/10.1007/s40265-014-0226- 4. Data Availability. The datasets (aggregated and anonymised), analysed for this specific 5. Hirose T, Chen CC, Ahn KJ, Kiljański J. Use of study, are available (on reasonable request) at insulin glargine 100 U/mL for the treatment of type the following email address: 2 diabetes mellitus in East Asians: a review. Diabetes Ther. 2019;10(3):805–33. https://doi.org/10.1007/ ansicras@atryshealth.com s13300-019-0613-7. Open Access. This article is licensed under a 6. Standl E, Owen DR. New long-acting basal insulins: Creative Commons Attribution-NonCommer- does benefit outweigh cost? Diabetes Care. cial 4.0 International License, which permits 2016;39(Suppl 2):S172–9. any non-commercial use, sharing, adaptation, 7. Davies M, Dahl D, Heise T, Kiljanski J, Mathieu C. distribution and reproduction in any medium Introduction of biosimilar insulins in Europe. Dia- or format, as long as you give appropriate credit bet Med. 2017;34(10):1340–53. https://doi.org/10. to the original author(s) and the source, provide 1111/dme.13400. a link to the Creative Commons licence, and 8. Sanofi-Aventis Deutschland GmbH. Toujeo Sum- indicate if changes were made. The images or mary of Product Characteristics. 2020. https://www. other third party material in this article are ema.europa.eu/documents/product-information/ included in the article’s Creative Commons toujeo-epar-product-information_en.pdf. Accessed 5 Oct 2020. licence, unless indicated otherwise in a credit line to the material. If material is not included 9. Linnebjerg H, Lam ECQ, Zhang X, et al. Duration of in the article’s Creative Commons licence and action of two insulin glargine products, LY2963016 your intended use is not permitted by statutory insulin glargine and Lantus insulin glargine, in subjects with type 1 diabetes mellitus. Diabetes regulation or exceeds the permitted use, you Obes Metab. 2017;19(1):33–9. https://doi.org/10. will need to obtain permission directly from the 1111/dom.12759. copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by- 10. Becker RHA, Dahmen R, Bergmann K, et al. New insulin glargine 300 units•mL-1 provides a more nc/4.0/. even activity profile and prolonged glycemic con- trol at steady state compared with insulin glargine 3870 Adv Ther (2021) 38:3857–3871 100 units•mL-1. Diabetes Care. 2015;38(4):637–43. 20. European Network of Centres for Pharmacoepi- https://doi.org/10.2337/dc14-0006. demiology and Pharmacovigilance. 2019. http:// www.encepp.eu/encepp/viewResource.htm?id= 11. Vargas-Uricoechea H. Efficacy and safety of insulin 29236. Accessed 8 Jun 2020. glargine 300 U/mL versus 100 U/mL in diabetes mellitus: a comprehensive review of the literature. 21. Sicras-Mainar A, Enriquez JL, Hernández I, Sicras- J Diabetes Res. 2018;2018:2052101. https://doi.org/ Navarro A. Validation and representativeness of the 10.1155/2018/2052101. Spanish BIG-PAC database: Integrated computer- ized medical records for research into epidemiol- 12. Bolli GB, Riddle MC, Bergenstal RM, et al. New ogy, medicines and health resource use (real word insulin glargine 300 U/ml compared with glargine evidence). Value Health. 2019;22(3):S734. 100 U/ml in insulin-naı̈ve people with type 2 dia- betes on oral glucose-lowering drugs: a randomized 22. Austin SR, Wong YN, Uzzo RG, et al. Why summary controlled trial (EDITION 3). Diabetes Obes Metab. comorbidity measures such as the Charlson 2015;17(4):386–94. https://doi.org/10.1111/dom. Comorbidity Index and Elixhauser score work. Med 12438. Care. 2015;53(9):e65–72. https://doi.org/10.1097/ MLR.0b013e318297429c. 13. Eli Lilly Nederland B.V. Abasaglar summary of product characteristics. 2020. https://www.ema. 23. Giorgino F, Benroubi M, Sun JH, Zimmermann AG, europa.eu/en/documents/product-information/ Pechtner V. Efficacy and safety of once-weekly abasaglar-previously-abasria-epar-product- dulaglutide versus insulin glargine in patients with information_en.pdf. Accessed 5 Oct 2020. type 2 diabetes on metformin and glimepiride (AWARD-2). Diabetes Care. 2015;38(12):2241–9. 14. Sanofi-Aventis Deutschland GmbH. Lantus sum- https://doi.org/10.2337/dc14-1625. mary of product characteristics. 2015. https://www. ema.europa.eu/en/documents/product- 24. Ji L, Kang ES, Dong X, et al. Efficacy and safety of information/lantus-epar-product-information_en. insulin glargine 300 U/mL versus insulin glargine pdf. Accessed 27 Oct 2020. 100 U/mL in Asia Pacific insulin-naı̈ve people with type 2 diabetes: The EDITION AP randomized con- 15. United States Food and Drug Administration. Real- trolled trial. Diabetes Obes Metab. 2020;22(4): world evidence. 2020. https://www.fda.gov/ 612–21. https://doi.org/10.1111/dom.13936. science-research/science-and-research-special- topics/real-world-evidence. Accessed 13 July 2020. 25. Bolli GB, Riddle MC, Bergenstal RM, et al. Gly- caemic control and hypoglycaemia with insulin 16. Katkade VB, Sanders KN, Zou KH. Real world data: glargine 300U/mL versus insulin glargine 100U/mL an opportunity to supplement existing evidence for in insulin-naı̈ve people with type 2 diabetes: the use of long-established medicines in health care 12-month results from the EDITION 3 trial. Dia- decision making. J Multidiscip Healthc. 2018;11: betes Metab. 2017;43(4):351–8. https://doi.org/10. 295–304. https://doi.org/10.2147/JMDH.S160029. 1016/j.diabet.2017.04.007. 17. Zaccardi F, Davies MJ, Khunti K. The present and 26. Ritzel R, Roussel R, Giaccari A, Vora J, Brulle- future scope of real-world evidence research in Wohlhueter C, Yki-Järvinen H. Better glycaemic diabetes: What questions can and cannot be control and less hypoglycaemia with insulin glar- answered and what might be possible in the future? gine 300 U/mL vs glargine 100 U/mL: 1-year Diabetes Obes Metab. 2020;22(Suppl 3):21–34. patient-level meta-analysis of the EDITION clinical https://doi.org/10.1111/dom.13929. studies in people with type 2 diabetes. Diabetes Obes Metab. 2018;20(3):541–8. https://doi.org/10. 18. Abitbol A, Brown RE, Jiandani D, Sauriol L, Aronson 1111/dom.13105. R. Real-world health outcomes of insulin glargine 300 U/ml vs insulin glargine 100 U/ml in adults 27. Base de Datos de medicamentos del Consejo Gen- with type 1 and type 2 diabetes in the Canadian eral de Farmacéuticos (Bot PLUS 2.0). 2020. https:// LMC Diabetes Patient Registry: The REALITY study. botplusweb.portalfarma.com. Accessed 10 June Can J Diabetes. 2019;43(7):504-509.e1. https://doi. 2020. org/10.1016/j.jcjd.2019.04.012. 28. DeVries JH, Desouza C, Bellary S, et al. Achieving 19. Gupta S, Wang H, Skolnik N, et al. Treatment dos- glycaemic control without weight gain, hypogly- ing patterns and clinical outcomes for patients with caemia, or gastrointestinal adverse events in type 2 type 2 diabetes starting or switching to treatment diabetes in the SUSTAIN clinical trial programme. with insulin glargine (300 units per milliliter) in a Diabetes Obes Metab. 2018;20(10):2426–34. real-world setting: a retrospective observational https://doi.org/10.1111/dom.13396. study. Adv Ther. 2018;35(1):43–55. https://doi.org/ 10.1007/s12325-017-0651-3. Adv Ther (2021) 38:3857–3871 3871 29. Purnell JQ, Dev RK, Steffes MW, et al. Relationship patients with type 2 diabetes: a mediation analysis. of family history of type 2 diabetes, hypoglycemia, BMJ Open Diabetes Res Care. 2016;4(1):e000184. and autoantibodies to weight gain and lipids with https://doi.org/10.1136/bmjdrc-2015-000184. intensive and conventional therapy in the Diabetes Control and Complications Trial. Diabetes. 34. Rahman M, Nakamura K, Hasan SMM, Seino K, 2003;52(10):2623–9. https://doi.org/10.2337/ Mostofa G. Mediators of the association between diabetes.52.10.2623. low socioeconomic status and poor glycemic con- trol among type 2 diabetics in Bangladesh. Sci Rep. 30. Khunti K, Gomes MB, Pocock S, et al. Therapeutic 2020;10(1):6690. https://doi.org/10.1038/s41598- inertia in the treatment of hyperglycaemia in 020-63253-8. patients with type 2 diabetes: a systematic review. Diabetes Obes Metab. 2018;20(2):427–37. https:// 35. Parajuli J, Saleh F, Thapa N, Ali L. Factors associated doi.org/10.1111/dom.13088. with nonadherence to diet and physical activity among Nepalese type 2 diabetes patients; a cross 31. Munk NE, Knudsen JS, Pottegård A, et al. Differ- sectional study. BMC Res Notes. 2014;7:758. ences between randomized clinical trial partici- https://doi.org/10.1186/1756-0500-7-758. pants and real-world empagliflozin users and the changes in their glycated hemoglobin levels. JAMA 36. Kirkman MS, Rowan-Martin MT, Levin R, et al. Netw Open. 2020;3(2):e1920949. https://doi.org/ Determinants of adherence to diabetes medica- 10.1001/jamanetworkopen.2019.20949. tions: findings from a large pharmacy claims data- base. Diabetes Care. 2015;38(4):604–9. https://doi. 32. Monti S, Grosso B, Todoerti M, Caporali R. Ran- org/10.2337/dc14-2098. domized controlled trials and real-world data: dif- ferences and similarities to untangle literature data. 37. Bartlett VL, Dhruva SS, Shah ND, Ryan P, Ross JS. Rheumatology (Oxford). 2018;57(57 Suppl 7):54–8. Feasibility of using real-world data to replicate https://doi.org/10.1093/rheumatology/key109. clinical trial evidence. JAMA Netw Open. 2019;2(10):e1912869. https://doi.org/10.1001/ 33. Houle J, Lauzier-Jobin F, Beaulieu MD, et al. jamanetworkopen.2019.1286. Socioeconomic status and glycemic control in adult