Received: 24 September 2018 Revised: 12 December 2018 Accepted: 2 January 2019 DOI: 10.1111/dom.13673 ORIGINAL ARTICLE The Diabetes Unmet Need with Basal Insulin Evaluation (DUNE) study in type 2 diabetes: Achieving HbA1c targets with basal insulin in a real-world setting Luigi F. Meneghini MD1,2 | Didac Mauricio MD3 | Emanuela Orsi MD4 | Nebojsa M. Lalic MD5 | Anna M.G. Cali MD6 | Jukka Westerbacka MD6 | Peter Stella MD6 | Christophe Candelas, DEA7 | Valerie Pilorget MD7 | Riccardo Perfetti MD8 | Kamlesh Khunti MD9 | on behalf of the DUNE investigators 1 Division of Endocrinology, University of Texas Southwestern Medical Center, Dallas, Aims: To describe in a real-world setting the achievement of physician-selected individualized Texas HbA1c targets in individuals with type 2 diabetes, newly or recently initiated with basal insulin, 2 Global Diabetes Program, Parkland Health & and the association of hypoglycaemia with target achievement. Hospital System, Dallas, Texas Materials and methods: A 12-week, prospective, single-arm, observational study of adults with 3 Department of Endocrinology and Nutrition, type 2 diabetes, either newly initiated with any basal insulin or start on basal insulin within the CIBER of Diabetes and Associated Metabolic Diseases, Hospital de la Santa Creu i Sant Pau, preceding 12 months. At enrollment, eligible participants from 28 countries were treated with Barcelona, Spain or without oral antihyperglycaemic drugs and/or GLP-1 receptor agonists. 4 Endocrine and Metabolic Diseases Unit, Results: Individualized targets for almost all of the 3139 evaluable participants (99.7%) had been Fondazione Ca' Granda IRCCS, Milan, Italy set by their physicians, with 57% of participants having HbA1c targets between 7.0% and 5 Clinic for Endocrinology, Clinical Center of <7.5% (53 and <58 mmol/mol). By week 12, 28% and 27% of newly and previously initiated par- Serbia, Faculty of Medicine, University of Belgrade, Serbia ticipants, respectively, achieved individualized HbA1c targets with modest average increases in 6 daily insulin dose of 9 and 5 U (0.10 and 0.06 U/kg), respectively, from baseline (14 and 23 U Sanofi, Paris, France 7 [0.17 and 0.29 U/kg], respectively). Overall, 16% of participants experienced at least one epi- Sanofi, Chilly-Mazarin, France 8 Sanofi, Bridgewater, New Jersey sode of hypoglycaemia. Both the incidence and frequency of hypoglycaemia, but not the sever- 9 Diabetes Research Centre, University of ity, were positively associated with a higher likelihood of achieving individualized HbA1c Leicester, Leicester, UK targets (P < 0.05). Conclusions: In this prospective real-world study, most participants using basal insulin did not Correspondence achieve the individualized HbA1c targets set by their physicians. Participants who experienced Dr Luigi Meneghini MD, Department of Internal Medicine, Division of Endocrinology, symptomatic hypoglycaemia were more likely to achieve HbA1c targets than those who did not. University of Texas Southwestern Medical Center and Parkland Health and Hospital KEYWORDS System, 5323 Harry Hines Blvd, Dallas, TX 75390, USA. basal insulin, glycaemic control, hypoglycaemia, insulin therapy, observational study, type Email: luigi.meneghini@utsouthwestern.edu 2 diabetes Funding information This study was sponsored by Sanofi. Sanofi designed and coordinated the clinical trial presented in this manuscript; Sanofi; Health Research 1 | I N T RO D UC T I O N events.4 However, approximately one half of all individuals with T2DM do not achieve the HbA1c level of less than 7.0% (<53 mmol/- Chronic hyperglycaemia is associated with long-term complications of mol)5 that is recommended in clinical practice,6,7 with lower rates of 1–3 type 2 diabetes (T2DM), including increased risk of cardiovascular target achievement often reported for those treated with basal insulin This is an open access article under the terms of the Creative Commons Attribution-NonCommercial-NoDerivs License, which permits use and distribution in any medium, provided the original work is properly cited, the use is non-commercial and no modifications or adaptations are made. © 2019 The Authors. Diabetes, Obesity and Metabolism published by John Wiley & Sons Ltd. Diabetes Obes Metab. 2019;21:1429–1436. wileyonlinelibrary.com/journal/dom 1429 1430 MENEGHINI ET AL. (BI).8–11 For example, in two international real-world studies, only enroll 4000 participants with T2DM ≥18 years of age who had newly 27% to 33% of individuals with T2DM achieved HbA1c less than initiated treatment with BI, human or analog, at the time of enroll- 7.0% (<53 mmol/mol) after initiating BI therapy.10,11 Furthermore, ment, or had been treated with BI for less than 12 months (previously failure to achieve HbA1c targets in the short term may be associated initiated), with or without oral antihyperglycaemic drugs and/or with suboptimal long-term blood glucose (BG) control, as illustrated glucagon-like peptide 1 (GLP-1) receptor agonists. Participants were by a recent retrospective analysis of medical records from Europe and required to have HbA1c between 7.5% and 11.0% (≥58-≤97 mmol/- the USA, in which those who did not achieve HbA1c less than 7.0% mol) for newly initiated BI users and between 7.5% and 10.0% (<53 mmol/mol) at 3 months were less likely to achieve this target at (≥58-≤86 mmol/mol) for previously initiated BI users, and to be willing 24 months.10 Recent international treatment guidelines advocate set- to perform self-monitoring of blood glucose (SMBG) and to complete ting individualized HbA1c targets that take account of the individual's a patient diary. Exclusion criteria included treatment with rapid-acting characteristics, duration of diabetes, co-morbidities, life expectancy or premix insulin, or physician intent to intensify treatment with a and risk of hypoglycaemia.12–15 Furthermore, therapy intensification rapid-acting or premix insulin within the next 3 months. Participants is recommended if individualized targets are not reached after were also excluded if they were more likely to have type 1 diabetes 5 3 months. (<40 years old and had initiated insulin within 1 year of diabetes diag- In the case of insulin-treated individuals with T2DM, suboptimal nosis), or if they were, or planned to become, pregnant. glycaemic control may be due, in part, to ineffective treatment pre- To help eliminate bias, investigators were advised to include con- scription, therapeutic inertia, non-adherence, omission and/or dose secutive patients suitable for the study. Signed informed consent was reduction in the setting of hypoglycaemia, or fear of obtained from all participants. To mirror real-world clinical practice for hypoglycaemia.16–20 For example, concerning individuals with T2DM the management of diabetes, no fixed study visit was scheduled dur- who had received stable doses of metformin and sulfonylureas for at ing the follow-up period; rather, clinical visits, including the possibility least 6 months, those who experienced moderate or severe hypogly- of phone visits, and treatment choices were undertaken according to caemia reported poorer adherence to medication (46% vs 67%; local practice. At study entry, data were collected from participants P < 0.01) and were more likely to perceive side effects as a barrier to concerning demographics, medical history, especially concerning treatment (36% vs 14%; P < 0.001) than those who did not experi- diabetes complications, comorbidities and history of severe hypogly- ence hypoglycaemia or experienced only mild hypoglycaemia.21 How- caemia, and type of BI being used (human intermediate-acting or ever, it is less clear whether poor adherence to treatment as the result long-acting analog). The use and titration of concomitant antidiabetic of hypoglycaemia impacts attainment of HbA1c targets. An inverse medications other than BI during the study period was left at the dis- relationship appears to exist between rates of hypoglycaemia and cretion of the treating physicians. This study was observational, with HbA1c in T2DM, as reported in a meta-regression analysis of 11 ran- treatment carried out according to local practice and conducted in domized controlled trials involving insulin-treated participants,22 and a accordance with the Declaration of Helsinki and the International study in individuals treated with sulfonylureas.23 Walz et al reported Conference on Harmonization guidelines for Good Clinical Practice. that, despite poorer adherence, patients who experienced moderate or severe symptoms of hypoglycaemia achieved lower mean HbA1c 2.2 | HbA1c targets values than patients without or with only mild symptoms (7.0% vs 7.3% [53 vs 56 mmol/mol]; P < 0.05).21 At baseline, an individualized long-term HbA1c target was set for each Data concerning the extent to which patients are set and achieve participant by their physician. In the case of those for whom an indi- individualized HbA1c targets are lacking. Furthermore, the association vidualized target was not set, a general HbA1c target less than 7.0% between achievement of individualized glycaemic targets and the risk [<53 mmol/mol] was defined, based on current guidelines5 (Table 1). of hypoglycaemia in clinical practice with BI is currently not known. A separate 12-week objective was set by physicians, based on the The Diabetes Unmet Need with Basal Insulin Evaluation (DUNE) study HbA1c level they anticipated patients would be able to reach by week was a 12-week, prospective, observational study (February 2015-July 12; however, the results of this objective are not the focus of this 2016) involving individuals with T2DM, either newly (at enrollment) report. or recently (<12 months) initiated treatment with BI. The study aimed to describe the proportion of participants who achieved individualized 2.3 | Defining participants at low or high risk of or general HbA1c targets at 12 weeks, and to evaluate the impact of hypoglycaemia complications the frequency and severity of symptomatic hypoglycaemia on target achievement. Independent of any evaluation by their physicians, participants were retrospectively categorized as being at low risk or high risk of compli- cations from a hypoglycaemic event (Table 1). 2 | MATERIALS AND METHODS 2.4 | Objectives and endpoints 2.1 | Study design and participants The study had two primary objectives: first, to describe the proportion The DUNE study was a 12-week, single-arm, observational study with of participants who achieved their individualized (long-term) HbA1c prospective follow-up at week 12 (±2 weeks). The study aimed to target at 12 weeks; second, to evaluate the association of the MENEGHINI ET AL. 1431 TABLE 1 Definitions of HbA1c goals and participants at high risk of for the incidences of symptomatic hypoglycaemia of 20% in newly ini- complications from hypoglycaemia tiated participants and of 45% in previously initiated participants. Definitions Achievement of HbA1c targets at 12 weeks was modeled as a Individualized HbA1c target Overall, long-term goal; those function of symptomatic hypoglycaemia using stepwise multivariate without an individualized target logistic regression, adjusted according to region, age, diabetes dura- set by their physician were given a general HbA1c target of <7.0% tion, HbA1c at baseline, use of sulfonylureas and/or meglitinides at (<53 mmol/mol) baseline, and use of GLP-1 receptor agonists at study entry. Participants at high riska of • Age ≥65 y complications from • Duration of diabetes >15 y hypoglycaemia • Coronary heart disease • Renal function impairment 3 | RE SU LT S • Professional driver • Myocardial infarction • Myocardial revascularization 3.1 | Study population and baseline characteristics • Peripheral vascular disease • Heart failure 3.1.1 | Study population • History of severe hypoglycaemia • Stroke The study, conducted from February 2015 to July 2016, evaluated • Diabetic retinopathy leading to 3139 patients from 28 countries (Supporting Information Figure S1), blindness • Transient ischaemic attack including 1716 (54.7%) newly initiated and 1423 (45.3%) previously • Lower extremity amputation initiated participants. Mean treatment duration for previously initiated • Severe dementia participants was approximately 5.7 months. Overall, 63% of partici- a High risk if patients ≥65 years old or with evidence of any of the comor- pants self-titrated their insulin, compared with 37% whose titrations bidities/characteristics agreed by the DUNE Steering Committee. were determined by physicians. These proportions were similar in both newly and previously initiated groups. frequency and/or severity of symptomatic hypoglycaemia with indi- vidualized HbA1c target achievement at 12 weeks. Primary endpoints 3.1.2 | Baseline characteristics were achievement of individual HbA1c target, or general HbA1c tar- There were no major clinical differences between newly and previ- get of less than 7.0% (<53 mmol/mol), and the frequency and severity ously initiated participants in terms of age, weight, BMI, concomitant of symptomatic hypoglycaemia. Symptomatic hypoglycaemia events diabetes medications (other than BI), presence of at least one micro- were classified as severe (requiring third party assistance, with or vascular complication, diabetic neuropathy, diabetes-related func- without a BG measurement) or non-severe (associated with typical tional impairment or estimated glomerular filtration rate (eGFR) hypoglycaemia symptoms and not requiring third party assistance (Table 2). The proportion of newly and previously initiated participants regardless of BG measurement).24 with a duration of diabetes longer than 10 years was 39% and 46%, Secondary endpoints included hypoglycaemia occurrence, general respectively. Participants at high risk of complications from a hypogly- HbA1c goal achievement (<7.0% and <8.0% [<53 and <64 mmol/mol]) caemic event comprised more than 60% of the newly and previously according to the level of risk of complications following hypoglycae- initiated groups (Supporting Information Table S1). mia, and individualized HbA1c target without symptomatic hypogly- caemia. The change from baseline to week 12 was determined for HbA1c, FPG as measured by SMBG, body weight and insulin dose. At 3.2 | Individualized HbA1c target the end of the study, a survey was administered to collect investigator Overall, individualized HbA1c targets were set by the investigators for opinions on whether failure to reach target was due to lack of adher- 99.7% of participants, with no major differences between targets set ence to titration/lifestyle recommendations, hypoglycaemic events, for newly and previously initiated participants (Table 3). Individualized intercurrent disease or other reasons. HbA1c targets of 7.0% to less than 7.5% (<53-<58 mmol/mol) were set for the majority of participants in both groups (Table 3). The major 2.5 | Data analysis and statistics reasons underlying physicians' decisions to set less stringent individu- alized HbA1c targets were participant age, patient acceptance of the The sample size was determined to ensure sufficient precision for treatment constraints required to meet the HbA1c target (such as evaluating the percentage of participants at HbA1c target (<7.0% SMBG, insulin dose adjustment and lifestyle changes) and the exis- [<53 mmol/mol]), assuming 27% of participants would achieve that tence of comorbidities. target at 12 weeks, as reported in real-world observational studies.8 The inclusion of 4000 patients allowed this percentage to be esti- mated with a precision of at least 1.5%, assuming that 15% of partici- 3.3 | Insulin treatment and weight pants would be non-evaluable. This sample size also enabled Long-acting BI analogs were used by 80% and 76% of newly and pre- detection of an odds ratio of at least 1.3 for the relationship between viously initiated participants, respectively. The remaining participants HbA1c target achievement at 12 weeks and occurrence of symptom- used human intermediate-acting insulin (NPH). Mean daily insulin atic hypoglycaemia (assumed to be positively associated [OR > 1]) the dose (standard deviation [SD]) at baseline was 14 (7) U (0.17 reference being at least one symptomatic hypoglycaemic event, with [0.09] U/kg) and 23 (15) U (0.29 [0.17] U/kg) for newly and previously a power of at least 80% and an alpha risk of 5%, and with assumptions initiated participants, respectively. The majority of newly initiated 1432 MENEGHINI ET AL. TABLE 2 Demographic and baseline characteristics TABLE 3 Individualized HbA1c target set by physicians Previously Individualized targets set initiated Newly Previously Newly (within initiated initiated All (total) initiated 12 months) All (total) (n = 1716) (n = 1423) n = 3139 Characteristic n = 1716 n = 1423 n = 3139 Target set for the patient, % 99.9 99.5 99.7 Age, y (SD) 60 (11) 61 (10) 61 (11) (mmol/mol) Gender, female/ 50/50 53/47 51/49 <6.5 (<48) 1 1 1 male, % 6.5 to <7.0 (48-<53) 18 18 18 Weight, kg (SD) 85 (18) 83 (17) 84 (17) 7.0 to <7.5 (53-<58) 58 57 57 Body mass index, 30.6 (5.6) 30.4 (5.4) 30.5 (5.5) kg/m2 (SD) 7.5 to <8.0 (58-<64) 17 17 17 Mean HbA1c (SD) ≥8.0 (≥64) 6.5 7.2 6.8 % 9.14 (1.01) 8.56 (0.77) 8.88 (0.96) Reason for target (several possible) (%) mmol/mol 76.44 (11.03) 70.01 (8.46) 73.52 (10.45) Age 67 65 66 Duration of diabetes, 10 (7) 11 (7) 10 (7) Comorbidities 36 39 38 years (SD) History of severe 0.6 2.3 1.4 <1 y, % 6 5 6 hypoglycaemia 1 to 5 y, % 22 20 21 Patient acceptability 40 43 41 5 to 10 y, % 33 29 31 Other 9 10 9 >10 y, % 39 46 42 Individualized target achievement Hypoglycaemia within Patients achieving 6 months of study target (%) entry, n (%) Individualized target 28 27a 27 Severe 37 (2.2) 75 (5.3) 112 (3.6) <7.0 % (<53 mmol/mol) Symptomatic 68 (4.0) 171 (12.0) 239 (7.6) High risk 24 21 23 At least one diabetes 92 93 93 Low risk 27 28 28 medication, % <8.0 % (<64 mmol/mol) Metformin 80 80 80 High risk 65 64 65 Sulfonylureas 54 42 49 Low risk 62 65 63 DPP-4 inhibitors 30 26 28 Patients achieving 24 20 22 GLP-1 receptor 5 6 6 individual target without agonists hypoglycaemia (%) Glinides 4 4 4 Physician-reported reasons SGLT2 inhibitors 4 4 4 for failure to reach the 12-wk objectiveb (%) Thiazolidinediones 2 3 2 Lack of adherence to 44 44 44 Alpha-glucosides 2 1 2 titration inhibitors Lack of adherence to 62 58 60 Diabetes complications lifestyle At least one 39 42 40 recommendations complication, % Hypoglycaemic events 4 5 4 eGFR 84 (29) 85 (27) 84 (29) Intercurrent disease 4 5 5 (mL/min/1.73 m2), mean (SD) Other 22 27 24 a Abbreviations: DPP-4, dipeptidyl peptidase 4; eGFR, estimated glomerular P = 0.504, Chi-squared test. b filtration rate; GLP-1, glucagon-like peptide 1; SD, standard deviation; The 12-week objective was the HbA1c level that physicians anticipated SGLT, sodium-glucose co-transporter. their patients would be able to reach by week 12, and may have differed from the individualized target described in the main text. (95%) and previously initiated (89%) participants were using once- other adverse drug reactions (0.03%) or other reasons (0.4%). By week daily BI dosing; 5% and 11% of participants were using at least twice- 12, the daily insulin dose increased by an average of 0.08 U/kg in daily dosing, respectively. The median (SD) target for fasting SMPG both newly and previously initiated participants (Supporting Informa- set by physicians, according to local practice, was 118 (15) mg/dL tion Table S2). There was a modest increase in weight over the (7 [0.9] mmol/L). In participants for whom titration data were available 12 weeks in both groups (Supporting Information Table S2). (n = 3090), titration was predominantly performed every 1 to 3 days (41%) or weekly (34%). In the remaining participants, titration fre- 3.4 | Achievement of HbA1c target at 12 weeks quency was less than once a week. Eleven (0.6%) in the newly initi- ated group and 15 (1.1%) in the previously initiated group At week 12, HbA1c had reduced from baseline by 1.4% (15 mmol/- discontinued insulin use during the study because of insufficient con- mol) in newly initiated participants, and by 0.8% (8.7 mmol/mol) in trol (0.8% overall), hypoglycaemia (0.2%), lack of adherence (0.2%), previously initiated participants (Supporting Information Table S3). MENEGHINI ET AL. 1433 Overall, 27% of patients achieved their individualized HbA1c target, per year, respectively. The incidence of severe hypoglycaemia during which was most commonly between 7.0% and less than 7.5% the study period was very low in both newly and previously initiated (<53-<58 mmol/mol) (Table 3). Fewer participants achieved individu- participants. alized targets without experiencing hypoglycaemia (Table 3). Similar proportions in the self-titrated and physician-titrated groups achieved their individualized HbA1c target (Supporting Information Table S4). 3.6 | HbA1c target achievement at 12 weeks – Participants at high risk of complications following hypoglycaemia multivariate model were significantly more likely (P < 0.0001) to have higher HbA1c tar- Participants who did not experience a symptomatic hypoglycaemic event gets set by their physicians, compared with those at low risk were significantly less likely to achieve their HbA1c target than those (Supporting Information Table S5). The overall proportion of low-risk who had experienced an event (P < 0.001) (Table 5). Factors that were participants who achieved HbA1c less than 7.0% (<53 mmol/mol) was significantly associated with target achievement in this model included lower than the proportion of high-risk participants who achieved higher individualized HbA1c target set by physicians at study entry HbA1c less than 8.0% (<64 mmol/mol) (Table 3). The majority of par- (P < 0.001) and use of long-acting BI analog vs human intermediate- ticipants with an individualized HbA1c target less than 7.0% acting insulin (P < 0.05) (Supporting Information Table S6). By compari- (<53 mmol/mol) did not achieve the target, irrespective of risk status. son, factors negatively associated with target achievement included lon- In contrast, most participants with an individualized HbA1c target less ger duration of diabetes (P < 0.001) and higher HbA1c at study entry than 8.0% (<64 mmol/mol) did achieve this target (Supporting Infor- (P < 0.001) (Supporting Information Table S6). The frequency of symp- mation Table S5). tomatic hypoglycaemia was also significantly associated with target achievement. Compared with participants who did not experience any 3.5 | Self-reported hypoglycaemia events, those who experienced one, two to four or more than four Symptomatic hypoglycaemia within the last month prior to study events were significantly more likely (P < 0.05) to achieve target HbA1c entry was reported by 4% and 12% of newly and previously initiated (Table 5). participants, respectively. Severe hypoglycaemia within the last Achievement of individualized HbA1c targets without symptom- 6 months prior to study entry was reported in 2% and 5% of newly atic hypoglycaemia was significantly less likely with increasing dura- and previously initiated participants, respectively. During the 12-week tion of diabetes (>10 vs <1 year; OR [95% CI], 0.37 [0.25-0.54]; study period, symptomatic hypoglycaemia was reported in 14% and P < 0.001) and with higher HbA1c levels at study entry (≥9.6% vs 18% of newly and previously initiated participants, respectively <8.0% [≥81 vs <64 mmol/mol]; OR [95% CI], 0.42 [0.31-0.56]; (Table 4), corresponding to a rate of 1.5 and 2.2 events per participant P < 0.001). Participants for whom higher HbA1c targets were set at study entry were more likely to achieve individualized targets without TABLE 4 Self-reported hypoglycaemia hypoglycaemia compared with participants for whom lower targets Previously were set (eg, ≥7.5%-<8.0% vs <7.0% [≥58-<64 vs ≥53 mmol/mol]; OR initiated (within [95% CI], 3.09 [1.89-5.07]; P < 0.001). Newly initiated 12 mo) All (total) (n = 1716) (n = 1423) (n = 3139) TABLE 5 Multivariate logistic regression model of individual HbA1c Participants with at least 14 18 16 targeta achievement at 12 weeks one symptomatic hypoglycaemic Multivariate modelb OR (95% CI) P value event (%) Symptomatic hypoglycaemia Yes Reference <0.001c Number of symptomatic 0.37 (1.36), 0 to 0.55 (1.96), 0.45 (1.66), No 0.645 (0.513 <0.001 hypoglycaemic events 21 0 to 39 0 to 39 to 0.810) per participant, mean (SD), range Frequency of symptomatic 0 Reference 0.0023 hypoglycaemic events Frequency of 1 1.411 (1.004 0.047 symptomatic to 1.983) hypoglycaemia (%) 2 to 4 1.585 (1.160 0.004 0 86 82 84 to 2.166) 1 6 7 6 >4 1.946 (1.091 0.024 to 3.473) 2 to 4 7 8 8 Number of symptomatic n 1.088 (1.030 0.002 >4 2 3 2 hypoglycaemic events to 1.149) Severity of symptomatic a hypoglycaemia (%) Proportion of patients in target range: 1%, <6.5% (<48 mmol/mol); 18%, 6.5% to <7.0% (48-<53 mmol/mol); 57%, 7.0% to <7.5% (53-<58 mmol/- Absence of 85.8 81.7 84.0 mol); 17%, 7.5% to <8.0% (58-<64 mmol/mol); 7%, ≥8.0% symptomatic (≥64 mmol/mol). hypoglycaemia b Modeling was adjusted for region, age, duration of diabetes, baseline Non-severe 13.7 17.0 15.2 HbA1c, use of sulfonylureas and/or metiglinides at study entry, and use of symptomatic GLP-1 receptor agonists at study entry. c hypoglycaemia Reference P values reflect the global association between the hypogly- caemia factor and HbA1c target achievement; all other P values are com- Severe hypoglycaemia 0.5 1.3 0.8 pared to the reference. 1434 MENEGHINI ET AL. 3.7 | Predictive factors for treatment failure – than those who did not experience hypoglycaemia.21 Lower baseline multivariate model HbA1c has also been independently associated with HbA1c target achievement and risk of hypoglycaemia.31 Each 1% increase in base- Multivariate analysis was conducted to identify factors predictive of line HbA1c reduced the likelihood of achieving an HbA1c target by treatment failure, defined as failure to achieve individualized and gen- 46% and increased the risk of experiencing at least one episode of eral HbA1c target at week 12). While experience of severe hypogly- confirmed hypoglycaemia by 17%.31 Similarly, in the DUNE study, caemia during the last 6 months prior to study entry (OR [95% CI], there was a significant positive association between experiencing 0.77 [0.49-1.21]) was not significantly associated with treatment fail- symptomatic hypoglycaemia and HbA1c target achievement, and ure, a significant positive association was observed for duration of dia- between lower baseline HbA1c and 12-week HbA1c target achieve- betes (1–5 vs <1 year; OR [95% CI], 2.33 [1.60-3.41]) and for HbA1c ment. Hypothetically, this association might be explained by the fact at study entry (8.01%-8.70% vs <8.01%; OR [95% CI], 1.62 that participants who achieved target HbA1c were at greater risk of [1.27-2.08]) with treatment failure. hypoglycaemia, given their lower average BG levels. While more intensive insulin titration can also lead to greater risk of hypoglycae- mia, we observed very modest increases in insulin use in our study. 4 | DISCUSSION The DUNE study may have been limited by several factors. Firstly, as mentioned earlier, a high proportion of participants report- The DUNE study was an observational, multinational, prospective edly self-titrated, which may not be applicable beyond the clinical real-life study that assessed the achievement of HbA1c targets set by practice settings represented in this real-world study.26 Furthermore, physicians based on individual patient characteristics. The study fur- hypoglycaemia data were collected by physicians only at week ther explored the relationship between the occurrence of hypoglycae- 12, based on patient diaries, which may be subject to recall bias.32 mia and achievement of individualized HbA1c targets. Based on The short observational period may also reduce the generalizability of prespecified criteria, over 60% of DUNE participants were deemed at results, and the association between hypoglycaemia and target “high risk” of complications from hypoglycaemia because of their age achievement may not necessarily have persisted over a longer obser- (>65 years), health status (vascular complications or co-morbidities) or vational period. The 3-month study period may have been insufficient occupation (commercial/truck driver). to allow participants with higher baseline HbA1c levels to reach gly- While there were substantial reductions in HbA1c during the caemic targets. However, a recent real-world clinical study reported 12-week study in both newly and previously initiated patients, most that decline in HbA1c occurs, for the most part, during the first participants failed to achieve their individualized HbA1c target, set at 3 months of insulin titration, with limited further decline thereafter,10 <7.5% (<58.5 mmol/mol) for 76% of patients. This may be related, in suggesting that the HbA1c levels reached during the DUNE study part, to insufficient insulin dose titration during the 12 weeks of treat- would reflect longer-term levels. Additionally, the decline in HbA1c ment, as indicated by the 9 U and 5 U dose increases reported in the during the first 3 months may reflect the period when titration occurs newly or previously initiated groups, respectively; such absence of most often during routine practice.25 intensive titration (titration inertia) has been reported previously in It is of interest that the rates of symptomatic hypoglycaemia dur- real-world clinical practice.25 Notably, 63% of participants in this ing the study period in both newly initiated (14%) and previously initi- study were reportedly self-titrating their insulin, whereas, in wider ated (18%) participants were lower than predicted (45% and 20%, clinical practice settings, it is likely that titration is largely determined respectively) based on conservative estimates from previous random- by physicians.26 Overall, however, no major difference in individual- ized clinical trials.28,33 This suggests that such randomized trials have ized HbA1c target achievement was observed between participant- limited applicability to real-life clinical practice. For example, in treat- driven or physician-driven titration, in either newly or previously initi- to-target trials, the lower HbA1c values were achieved with intensive ated participants. While only 27% of participants achieved their indi- titration. By contrast, the limited increases in insulin dose in this study vidual HbA1c target by week 12, this is consistent with observations may have further contributed to the observed low incidence and rates from other real-world studies.8,25 By comparison, in treat-to-target tri- of hypoglycaemia. The low rate of hypoglycaemia in the DUNE study als with defined titration algorithms, close follow-up and careful moni- could have impacted the reported association between HbA1c target toring, the proportion of participants achieving HbA1c targets of less achievement and the occurrence, frequency and severity of than 7.0% (<53 mmol/mol) with BIs has been reported to be as much hypoglycaemia. as 40% to 50%.27–29 Importantly, mean daily insulin doses at the end Despite its limitations, the DUNE study benefitted from having a of these treat-to-target studies reached between approximately large global real-world population with a comprehensive, prospective 40 and 100 U/d,27–29 considerably higher than those reached in this collection of data concerning patient characteristics at baseline, thus study. providing novel insights into real-life practice, particularly in that indi- It has been suggested that the experience of hypoglycaemia may vidualized targets for most patients appear to be set by their physi- result in increased fear of future hypoglycaemic events and, conse- cians. Given that individualized targets were often set higher than the quently, a worsening in glycaemic control as the result of changes in general HbA1c goal of less than 7.0% (<53 mmol/mol), this definition adherence to medication.19,30 However, it has also been reported of glycaemic control, often used in clinical studies,6–10 might need to that, despite the fact that hypoglycaemia results in poorer adherence, be reconsidered for real-world studies. The overall profile of individu- participants experiencing hypoglycaemia achieved lower HbA1c levels alized HbA1c targets set or achieved in our study population was MENEGHINI ET AL. 1435 similar regardless of retrospective stratification of participants as research support from Novartis, Novo Nordisk, Sanofi, Eli Lilly, Pfizer, being at high or low risk for hypoglycaemia complications, assessed Boehringer Ingelheim and MSD. independently from physician evaluations; therefore, the results should be translatable to other real-world scenarios. As participants who did not experience symptomatic hypoglycae- Author contributions mia were less likely to achieve glucose targets, factors other than L. M., D. M., A. C., J. W. and K. K. were involved in study concept and experiencing hypoglycaemia, or the fear thereof, may have contrib- design. C. C. performed the statistical analyses of data. All authors uted to the failure to achieve targets. For example, multivariate ana- participated in the interpretation of data and in the writing, reviewing lyses showed that individuals with a longer duration of diabetes and a and editing of the manuscript, and all had final responsibility for higher HbA1c level at study entry were significantly less likely to approving the published version. L. M. is the guarantor of this work achieve target HbA1c levels. Although unlikely, it may also be possible and, as such, had full access to all the data in the study and takes that the experience of hypoglycaemia itself contributed to lower responsibility for the integrity of the data and the accuracy of the data HbA1c levels, with more frequent events leading to a disproportion- analysis. ately lower overall average HbA1c level. Alternatively, patients who achieved lower HbA1c levels may have been at higher risk of ORCID experiencing hypoglycaemia. Further studies are required to better understand the association between hypoglycaemia and HbA1c target Luigi F. Meneghini https://orcid.org/0000-0003-4539-2725 attainment, and to determine why so many individuals with T2DM are Didac Mauricio https://orcid.org/0000-0002-2868-0250 failing to achieve glycaemic targets in real-life settings. The observed limited increase in insulin dose suggests that there may be an oppor- tunity for individuals with T2DM and their physicians to titrate insulin RE FE RE NC ES more effectively to achieve desired glycaemic goals. Alternatively, the 1. Stratton IM, Adler AI, Neil HA, et al. Association of glycaemia with macrovascular and microvascular complications of type 2 diabetes use of newer glucose-lowering drugs with reduced risks of hypogly- (UKPDS 35): prospective observational study. BMJ. 2000;321: caemia as compared to BI may help individuals achieve glucose tar- 405-412. gets. To this end, further research and updated treatment guidelines 2. Turner RC, Millns H, Neil HA, et al. Risk factors for coronary artery dis- ease in non-insulin dependent diabetes mellitus: United Kingdom Pro- are required to define the most appropriate options for the intensifi- spective Diabetes Study (UKPDS: 23). BMJ. 1998;316:823-828. cation of glucose-lowering drugs and the optimal timing of insulin ini- 3. Lehto S, Ronnemaa T, Pyorala K, Laakso M. Predictors of stroke in tiation and titration. middle-aged patients with non-insulin-dependent diabetes. Stroke. 1996;27:63-68. 4. Paul SK, Klein K, Thorsted BL, Wolden ML, Khunti K. Delay in treat- ment intensification increases the risks of cardiovascular events in ACKNOWLEDGMENTS patients with type 2 diabetes. Cardiovasc Diabetol. 2015;14:100. The authors thank the study participants, trial staff and investigators 5. American Diabetes Association. Standards of medical care in diabetes- 2017: summary of revisions. Diabetes Care. 2017;40:S4-S5. for their participation (participating physicians are listed in Appendix 6. Stone MA, Charpentier G, Doggen K, et al. Quality of care of people S1). K. K. acknowledges support from The National Institute for with type 2 diabetes in eight European countries: findings from the Health Research (NIHR) Collaboration for Leadership in Applied Guideline Adherence to Enhance Care (GUIDANCE) study. Diabetes Care. 2013;36:2628-2638. Health Research and Care—East Midlands (CLAHRC-EM) and the 7. Stark Casagrande S, Fradkin JE, Saydah SH, Rust KF, Cowie CC. The NIHR Leicester Biomedical Research Centre. prevalence of meeting A1C, blood pressure, and LDL goals among Editorial and writing assistance was provided by Arthur Holland, people with diabetes, 1988-2010. Diabetes Care. 2013;36:2271-2279. 8. Dalal MR, Grabner M, Bonine N, Stephenson JJ, DiGenio A, Bieszk N. PhD, of Fishawack Communications Ltd and was funded by Sanofi. Are patients on basal insulin attaining glycemic targets? Characteristics and goal achievement of patients with type 2 diabetes mellitus treated with basal insulin and physician-perceived barriers to achieving glyce- Conflict of interest mic targets. Diabetes Res Clin Pract. 2016;121:17-26. 9. Kostev K, Dippel FW, Rathmann W. Glycemic control after initiating L. M. has served on advisory panels for Novo Nordisk, Sanofi, and basal insulin therapy in patients with type 2 diabetes: a primary care AstraZeneca, and has received consulting fees from Sanofi. D. M. has database analysis. Diabetes Metab Syndr Obes. 2015;8:45-48. served on advisory panels for Sanofi, Praxis Pharmaceutical, AstraZe- 10. Mauricio D, Meneghini L, Seufert J, et al. Glycaemic control and hypo- neca, Novo Nordisk and MSD, and has received speakers' bureau fees glycaemia burden in patients with type 2 diabetes initiating basal insu- lin in Europe and the USA. Diabetes Obes Metab. 2017;19:1155-1164. from Menarini, GlaxoSmithKline, Eli Lilly, Sanofi, Novartis, Novo Nor- 11. Khunti K, Caputo S, Damci T, et al. The safety and efficacy of adding disk and MSD. E. O. has served on advisory panels for Boehringer once-daily insulin detemir to oral hypoglycaemic agents in patients Ingelheim and Eli Lilly, and has received speakers' bureau fees from with type 2 diabetes in a clinical practice setting in 10 countries. Dia- betes Obes Metab. 2012;14:1129-1136. Sanofi, Takeda, Johnson & Johnson, Novo Nordisk and AstraZeneca. 12. American Diabetes Association. Standards of medical care in diabetes- N. L. has no conflicts of interest to declare. A. C., J. W., P. S., C. C., 2014. Diabetes Care. 2014;37(suppl 1):S14-S80. V. P. and R. P. are all employees of and stock/shareholders in Sanofi. 13. Inzucchi SE, Bergenstal RM, Buse JB, et al. Management of hypergly- cemia in type 2 diabetes, 2015: a patient-centered approach: update K. K. has served on advisory panels, is a board member of, and has to a position statement of the American Diabetes Association and the received consulting and speakers' bureau fees from Bayer, Novartis, European Association for the Study of Diabetes. Diabetes Care. 2015; Novo Nordisk, Sanofi, Eli Lilly, Servier and MSD, and has received 38:140-149. 1436 MENEGHINI ET AL. 14. Inzucchi SE, Bergenstal RM, Buse JB, et al. Management of hypergly- 26. Khunti K, Davies MJ, Kalra S. Self-titration of insulin in the manage- cemia in type 2 diabetes: a patient-centered approach: position state- ment of people with type 2 diabetes: a practical solution to improve ment of the American Diabetes Association (ADA) and the European management in primary care. Diabetes Obes Metab. 2013;15:690-700. Association for the Study of Diabetes (EASD). Diabetes Care. 2012;35: 27. Riddle MC, Bolli GB, Ziemen M, et al. New insulin glargine 300 units/- 1364-1379. mL versus glargine 100 units/mL in people with type 2 diabetes using 15. Garber AJ, Abrahamson MJ, Barzilay JI, et al. AACE comprehensive dia- basal and mealtime insulin: glucose control and hypoglycemia in a betes management algorithm 2013. Endocr Pract. 2013;19:327-336. 6-month randomized controlled trial (EDITION 1). Diabetes Care. 16. Peyrot M, Barnett AH, Meneghini LF, Schumm-Draeger PM. Insulin 2014;37:2755-2762. adherence behaviours and barriers in the multinational Global Atti- 28. Bolli GB, Riddle MC, Bergenstal RM, et al. New insulin glargine tudes of Patients and Physicians in Insulin Therapy study. Diabet Med. 300 U/ml compared with glargine 100 U/ml in insulin-naive people 2012;29:682-689. with type 2 diabetes on oral glucose-lowering drugs: a randomized 17. Garber AJ. The importance of titrating starting insulin regimens in controlled trial (EDITION 3). Diabetes Obes Metab. 2015;17:386-394. patients with type 2 diabetes. Diabetes Obes Metab. 2009;11(suppl 5): 29. Pan C, Gross JL, Yang W, et al. A multinational, randomized, open- 10-13. label, treat-to-target trial comparing insulin degludec and insulin glar- 18. Brod M, Rana A, Barnett AH. Adherence patterns in patients with type gine in insulin-naive patients with type 2 diabetes mellitus. Drugs R D. 2 diabetes on basal insulin analogues: missed, mistimed and reduced 2016;16:239-249. doses. Curr Med Res Opin. 2012;28:1933-1946. 30. Mitchell BD, Vietri J, Zagar A, Curtis B, Reaney M. Hypoglycaemic 19. Ahren B. Avoiding hypoglycemia: a key to success for glucose- events in patients with type 2 diabetes in the United Kingdom: associ- lowering therapy in type 2 diabetes. Vasc Health Risk Manag. 2013;9: ations with patient-reported outcomes and self-reported HbA1c. BMC 155-163. Endocr Disord. 2013;13:59. 20. Russell-Jones D, Pouwer F, Khunti K. Identification of barriers to insu- 31. Riddle MC, Vlajnic A, Zhou R, Rosenstock J. Baseline HbA1c predicts lin therapy and approaches to overcoming them. Diabetes Obes Metab. attainment of 7.0% HbA1c target with structured titration of insulin 2018;20:488-496. glargine in type 2 diabetes: a patient-level analysis of 12 studies. Dia- 21. Walz L, Pettersson B, Rosenqvist U, Deleskog A, Journath G, betes Obes Metab. 2013;15:819–825. Wandell P. Impact of symptomatic hypoglycemia on medication 32. Khunti K, Alsifri S, Aronson R, et al. Rates and predictors of hypogly- adherence, patient satisfaction with treatment, and glycemic control caemia in 27 585 people from 24 countries with insulin-treated type in patients with type 2 diabetes. Patient Prefer Adherence. 2014;8: 1 and type 2 diabetes: the global HAT study. Diabetes Obes Metab. 593-601. 2016;18:907-915. 22. Mullins P, Sharplin P, Yki-Jarvinen H, Riddle MC, Haring HU. Nega- 33. Yki-Jarvinen H, Bergenstal R, Ziemen M, et al. New insulin glargine tive binomial meta-regression analysis of combined glycosylated 300 units/mL versus glargine 100 units/mL in people with type 2 dia- hemoglobin and hypoglycemia outcomes across eleven Phase III and betes using oral agents and basal insulin: glucose control and hypogly- IV studies of insulin glargine compared with neutral protamine Hage- cemia in a 6-month randomized controlled trial (EDITION 2). Diabetes dorn insulin in type 1 and type 2 diabetes mellitus. Clin Ther. 2007; Care. 2014;37:3235-3243. 29:1607-1619. 23. Kalra S, Deepak MC, Narang P, Singh V, Maheshwari A. Correlation SUPPOR TI NG I NFORMATION between measures of hypoglycemia and glycemic improvement in sul- fonylurea treated patients with type 2 diabetes in India: results from Additional supporting information may be found online in the Sup- the OBSTACLE hypoglycemia study. J Postgrad Med. 2014;60: porting Information section at the end of this article. 151-155. 24. International Hypoglycaemia Study Group. Glucose concentrations of less than 3.0 mmol/L (54 mg/dL) should be reported in clinical trials: a How to cite this article: Meneghini LF, Mauricio D, Orsi E, joint position statement of the American Diabetes Association and the European Association for the Study of Diabetes. Diabetes Care. 2017; et al. The Diabetes Unmet Need with Basal Insulin Evaluation 40:155-157. (DUNE) study in type 2 diabetes: Achieving HbA1c targets 25. Khunti K, Damci T, Meneghini L, Pan CY, Yale JF, SOLVE Study Group. with basal insulin in a real-world setting. Diabetes Obes Metab. Study of Once Daily Levemir (SOLVE): insights into the timing of insu- lin initiation in people with poorly controlled type 2 diabetes in rou- 2019;21:1429–1436. https://doi.org/10.1111/dom.13673 tine clinical practice. Diabetes Obes Metab. 2012;14:654-661.