Comparable efficacy with similarly low risk of hypoglycaemia in patient‐ vs physician‐managed basal insulin initiation and titration in insulin‐naïve type 2 diabetic subjects: The Italian Titration Approach Study

Abstract Aims People with uncontrolled type 2 diabetes (T2DM) often delay initiating and titrating basal insulin. Patient‐managed titration may reduce such deferral. The Italian Titration Approach Study (ITAS) compared the efficacy and safety of insulin glargine 300 U/mL (Gla‐300) initiation and titration using patient‐ (nurse‐supported) or physician‐management in insulin‐naïve patients with uncontrolled T2DM. Materials and methods ITAS was a multicentre, phase IV, 24‐week, open‐label, randomized (1:1), parallel‐group study. Insulin‐naïve adults with T2DM for ≥1 year with poor metabolic control initiated Gla‐300 after discontinuation of SU/glinides, and were randomized to self‐titrate insulin dose (nurse‐assisted) or have it done by the physician. The primary endpoint was change in HbA1c. Secondary outcomes included hypoglycaemia incidence and rate, change in fasting self‐monitored plasma glucose, patient‐reported outcomes (PROs), and adverse events. Results Three hundred and fifty five participants were included in the intention‐to‐treat population. At Week 24, HbA1c reduction from baseline was non‐inferior in patient‐ vs physician‐managed arms [least squares mean (LSM) change (SE): −1.60% (0.06) vs −1.49% (0.06), respectively; LSM difference: −0.11% (95% CI: −0.26 to 0.04)]. The incidence and rates of hypoglycaemia were similarly low in both arms: relative risk of confirmed and/or severe nocturnal (00:00‐05:59 hours) hypoglycaemia was 0.77 (95% CI: 0.27 to 2.18). No differences were observed for improvement in PROs. No safety concerns were reported. Conclusions In the T2DM insulin‐naïve, SU/glinides discontinued population, patient‐managed (nurse‐assisted) titration of Gla‐300 may be a suitable option as it provides improved glycaemic control with low risk of hypoglycaemia, similar to physician‐managed titration.


| INTRODUCTION
There is often a delay in initiation and titration of basal insulin (BI) in people with type 2 diabetes mellitus (T2DM) inadequately controlled by non-insulin treatment. 1,2 Lack of time and resources for healthcare providers (HCPs), complexity of the titration process, fear of hypoglycaemia, [3][4][5] along with psychological barriers to the idea itself of insulin initiation, are the most common reasons of such a delay.
To overcome these barriers, physicians recommend educational tools and support from medical staff. 4 Indeed, nurse-assisted insulin treatment has been shown to improve initiation rates and provide greater HbA 1c reductions compared with usual care. 6,7 Simple titration algorithms have also been shown to help patients to self-manage their titration. 8 Results from the AT.LANTUS and ATLAS studies, which compared patient-and physician-managed titration of glargine 100 U/mL (Gla-100), showed that patient-managed titration provided slightly, albeit significantly, better HbA 1c reductions and was well tolerated. 8,9 Insulin glargine 300 U/mL (Gla-300) has more stable pharmacokinetic and pharmacodynamic profiles compared with Gla-100 at fixed 10 and, most importantly, at clinical doses used by patients with type 1 diabetes (T1DM), 11,12 and reduces the risk of hypoglycaemia vs Gla-100 in T2DM. 13 Thus, self-titration with Gla-300 might offer advantages to patients to effectively and safely control hyperglycaemia. A recent study (TAKE CONTROL 14 ) has compared patientvs physician-managed titration of Gla-300. However, this comparison was evaluated in a T2DM population heterogeneous for antecedent insulin use and continued use of sulphonylureas (SU). No study has so far examined the question specifically in a homogeneous group of insulin-naïve T2DM people in the absence of SU or glinides.
The Italian Titration Approach Study (ITAS) aimed to test the efficacy and safety of initiation and titration of Gla-300 with the same algorithm used either by patients (nurse-assisted) or by physicians in people with uncontrolled T2DM naïve to insulin.

| Trial design and participants
ITAS was a national, multicentre, phase IV, 24-week, open-label, randomized (1:1), parallel-group study, conducted in Italy (EudraCT Number: 2015-001167-39). The primary aim of this study was to assess non-inferiority of change in HbA 1c over 24 weeks, when the same algorithm for Gla-300 dose titration was managed by the patient (nurseassisted) or the physician. Participants were insulin-naïve adults (≥18 years of age) with T2DM for ≥1 year with poor glycaemic control (HbA 1c ≥7.5 to ≤10%) on oral antihyperglycaemic drugs (OADs) and/or non-insulin injectables. After withdrawal of SU/glinides, if any, to evaluate the hypoglycaemia risk solely due to Gla-300, all participants were treated with Gla-300 and were randomized to either self-adjust their BI dose, or have their dose adjusted by a physician during visits or by telephone. The study consisted of a 2-week screening period and a 24-week treatment period (with interim analysis at 12 weeks). In each arm, visits/contacts were weekly until week 12, and then every 2 weeks until week 24. Additional unscheduled contacts (phone, on-site visit) were made available if clinically required. Full details of the methodology have recently been reported. 15 All participants provided informed, written consent. The clinical trial protocol was approved by the appropriate local Ethical Committees and IRB/IEC. The study was conducted in accordance with the Declaration of Helsinki and the ICH guidelines for good clinical practice. 16

| Basal insulin titration
All patients were instructed to self-administer a daily subcutaneous injection of Gla-300 (Toujeo SoloStar ® , Sanofi) in the evening, anytime from dinner to bedtime. Gla-300 was administered at a starting dose of 0.2 U/kg and then adjusted at each visit in the physicianmanaged arm, and weekly or even more frequently (but no more often than every 3-4 days) in the patient (nurse-assisted) managed arm to achieve a fasting self-monitored plasma glucose (SMPG) of 80-110 mg/dL. In both groups, changes in the insulin dose were based on the median of the fasting SMPG values measured on three consecutive days, of which the last was the day when titration was scheduled. Patients randomized to self-managing insulin titration received a specific educational session regarding self-adjustment of insulin dose from the study nurse, who monitored algorithm application without however exerting any influence on titration. The patient-and physician-managed titration algorithm is shown in Table S1.
Rescue therapy, if needed, either by adding a new OAD and/or by increasing the dose of an existing antihyperglycaemic non-study drug, was based on the Investigator's judgement, considering primarily the patient's individual clinical needs, but also local guidelines and Gla-300 labelling. 17 Further details of the titration protocol are also provided in the Appendix.
At randomization, investigators provided patients with a blood glucometer (MyStar Extra ® ; Sanofi) and diary to assess and record daily fasting SMPG until it was stable at the target. A seven-point SMPG profile was measured at Week 12 and Week 24. The investigator explained the need to measure glucose at the times requested by the study and to record the values correctly in the diary that the patients brought along with the glucometer at each office visit.
Patients were instructed to measure capillary plasma glucose whenever they experienced symptoms of hypoglycaemia. All hypoglycaemia episodes were recorded in the patient's diary or documented in the "hypoglycaemia screen/page" of the electronic case report form (e-CRF). Nocturnal hypoglycaemia was any event that occurred between 00:00 and 05:59. Severe hypoglycaemia was defined as an event requiring assistance of another person to actively administer carbohydrate or glucagon or to perform other resuscitative actions.

| Endpoints and other assessments
The primary endpoint of the study was change in HbA 1c from baseline to Week 24 (assessed using 0.3% as non-inferiority margin). The main secondary endpoint was the percentage of participants with ≥1 con-

| Statistical analyses
Full details of the statistical analyses have previously been reported. 15 The primary efficacy analysis was performed in the intention-to-treat (ITT) population, comprising all randomized patients who received at least one dose of Gla-300 and had a baseline assessment of primary efficacy variables, irrespective of compliance with the study protocol and procedures. The per-protocol (PP) population consisted of all patients in the ITT population without major protocol deviations and was used for the supportive analysis of the primary efficacy endpoint.
The safety population was defined as all randomized patients who received at least one dose of Gla-300 and had at least one safety variable collected.
The primary endpoint was analysed using a linear mixed-effect model (LMEM) with titration approach and centre as the fixed effect and the HbA 1c baseline value as the covariate. To assess the noninferiority of patient-managed vs physician-managed titration, the upper bound of the 95% confidence interval (CI) for the estimated difference in the mean change of HbA 1c from baseline to endpoint at Week 24 between the two titration approaches was compared with the predefined non-inferiority margin of 0.3% HbA 1c . Non-inferiority was shown if the upper bound of the two-sided 95% CI of the estimated difference for the ITT population was <0.3%.
The risk of hypoglycaemic events in the two titration groups was compared in terms of relative risk, whereas a rate ratio was computed to compare the annual incidence rates of same type of events in the two-titration arms. All statistical comparisons between the two titration groups were based on 95% CI. The cumulative number of confirmed and/or severe hypoglycaemic events per patient was provided by titration approach and described through descriptive statistics for continuous variables. Moreover, the cumulative mean functions of anytime (24 hours) or nocturnal confirmed [≤70 mg/dL (≤3.9 mmol/L)] and/or severe hypoglycaemia were estimated and graphically presented.
FPG, fasting SMPG and body weight were analysed using the same LMEM described above using the baseline value as a covariate.
The frequency and percentage of patients at different HbA 1c targets were summarized descriptively by titration group, and chi-square tests were applied for comparison.
Changes in PAID-5 and DES-SF total scores were computed and analysed with a LMEM using the effect of the titration approach and the baseline total score as covariates. DTSQ at the end of the treatment is reported through descriptive statistics in each study arm.
Safety endpoints were analysed descriptively.

| Participant flow
Of the 458 patients enrolled, 359 were randomized after exclusion of 72 patients who did not meet the HbA 1c criteria ( Figure S1). Three hundred and thirty nine (94.4%) patients completed the trial at 46 Italian sites. During the 24-week study period, 8.6% and 9.4% of patients in the patient and physician groups, respectively, initiated at least one rescue therapy (mainly dipeptidyl peptidase-4 inhibitors or sodiumglucose co-transporter-2 inhibitors) for poor glycaemic control or post-prandial hyperglycaemia.

| Baseline demographics
Baseline demographics are presented in Table 1 and were not different between the two groups. Over 90% of participants were receiving metformin as monotherapy or combination treatment (Table S2)

| Change in HbA 1c
At baseline, the mean HbA 1c was 8.77%  between the two treatment groups (Figure 2A,B) Figure 3A and  Figure 3B and Table S3). Incidence and rates of confirmed [<54 mg/dL (<3.0 mmol/L)] and/or severe hypoglycaemic events occurring at night (00:00-05:59 hours) were not different between treatment groups ( Figure 3A,B and Table S3).

| Incidence and rates of hypoglycaemia at any time of day (24 hours)
No differences were seen in the incidence or annualized rates of confirmed [≤70 mg/dL (≤3.9 mmol/L)] hypoglycaemic events at any time of day (24 hours) between the patient-and the physician-managed groups ( Figure 3A,B and Table S3). Incidence and rates of confirmed [<54 mg/dL (<3.0 mmol/L)] and/or severe hypoglycaemic events occurring at any time were not different between treatment groups ( Figure 3A,B and Table S3).

| Cumulative number of confirmed and/or severe hypoglycaemic events
The estimated cumulative mean function of anytime (24 hours) or nocturnal confirmed [≤70 mg/dL (≤3.9 mmol/L)] and/or severe hypoglycaemia was not different between the two treatment groups ( Figure 4A,B). and/or severe hypoglycaemic events occurring between 00:00 hour and pre-breakfast were not different between treatment groups ( Figure 3A,B and Table S3).

| Severe hypoglycaemia
There were four episodes of severe hypoglycaemia in three patients: one in the patient-managed group and three in the physician-managed groups.

| Composite endpoints
The proportion of patients achieving HbA 1c <7.0% without severe and/or confirmed hypoglycaemia were also not different between the groups (29.3% and 26.1% in the patient-and physician-managed groups, respectively) (P = .50). There was also no significant difference between the patient-and physician-managed groups among patients who achieved HbA 1c <7.0% without severe and/or confirmed hypoglycaemic events and no weight increase (19.8% vs 14.6%, respectively) (P = .21).

| Patient-reported outcomes
There was no difference between study groups for improvement in DTSQ treatment satisfaction between baseline and Week 24 ( Figure S4)

| Adverse events
A summary of treatment-emergent adverse events (TEAEs) can be found in Table S4. There were no drug-related serious adverse events.
Three TEAEs [bilateral peripheral oedema (n = 1) in the patientmanaged group and coronary artery disease (n = 1) and pregnancy F I G U R E 4 Cumulative mean number of (A) anytime (24 hours) or (B) nocturnal (00:00-05:59 hours) confirmed and/or severe hypoglycaemic events (ITT population). ITT, intention-to-treat (n = 1) in the physician-managed group] led to treatment discontinuation. Seven TEAEs were suspected of being treatment-related (five in the patient-and two in the physician-managed group) and were as follows: mild oedema, severe hypoglycaemia, moderate weight increase, injection site mild pain, and injection site mild pruritus in the patient-managed group and injection site mild pain and injection site mild pruritus in the physician-managed group. LANTUS (insulin-naïve and pre-treated participants) 8 and ATLAS (insulin-naïve participants) 9 ]. These two studies showed even greater reductions in HbA 1c with patient-vs physician-managed titration.
However, overall incidence of hypoglycaemia was higher in the patient-vs physician-managed titration arms in AT.LANTUS (P < .01) and ATLAS (P = .02 for symptomatic hypoglycaemia; P = .002 for nocturnal hypoglycaemia). In ITAS, there is a lower risk of hypoglycaemia (any definition) in both patient-and physician-managed BI titration groups, likely because the insulin-naïve population studied had lower hypoglycaemia risk, and also because Gla-300 carries a lower risk of hypoglycaemia vs Gla-100. 13 Recently, the TAKE CONTROL study 14  Regardless of the mechanism, reducing hypoglycaemia on BI treatment in T2DM is an important goal to improve long-term adherence of patients to insulin treatment and titration to target, as demonstrated by the high retention of patients in ITAS in both groups for the entire duration of the study. Minimizing the risk of even nonsevere hypoglycaemia, as in ITAS, is important to prevent long-term hypoglycaemia unawareness, 20 and the subsequent risk for severe hypoglycaemia, 21 which is associated with higher risk of cardiovascular morbidity and mortality. 22,23 Interestingly, there was no increase in body weight with Gla-300 neither in patient-nor in physician-managed BI titration, in line with previous observations. 13 As in TAKE CONTROL, ITAS reported no difference in total PRO scores in the patient-and physician-managed titration arms, demonstrating that patients felt comfortable with self-titration. In addition, incidence of TEAEs was not different in the patient-and physicianmanaged groups, as shown in TAKE CONTROL, AT.LANTUS, and ATLAS. 8,9,14 ITAS tried to mimic certain features of pragmatism (ie, not too selected study population; easy and safe titration algorithm) comparing physician managed titration with patient (nurse-assisted) self-titration. Both approaches are similarly successful and safe in ITAS, thus proving the feasibility and convenience of the simpler patient (nurseassisted) approach.

The limitations of this study include the open-label design. It
should also be noted that the good results of ITAS have been obtained with a number of visits/contacts, which do not necessarily reflect the lower level of interaction that patients would often receive in routine clinical practice in diabetes clinics. Finally, hypoglycaemia events were pragmatically reported and analysed by patient diaries or e-CRF and not by independent glucose monitoring tools.
In conclusion, ITAS provides additional evidence that initiation and titration of BI Gla-300 are not only efficacious as several previous studies have shown but also safe with low risk for any hypoglycaemia, especially when SU/glinides, a likely confounder, are suspended. Interestingly, these positive results, which include neutrality of BI Gla-300 on body weight, are achieved similarly when either patients (nurse assisted) or physicians manage BI titration. This encourages development of programmes for wider use of the patient-managed (nurseassisted) approach for BI titration, shifting the locus of control to fight clinical inertia.