The investigation reported here was a 51-week (2 weeks of screening, 2 × 24 weeks treatment, 1 week follow-up), open-label, randomized, multi-center, crossover phase IV trial in insulin naïve type 2 diabetes patients with insufficient metabolic control and HbA1c values ≥7.0 % and ≤10.0 % (to decrease the likely need for prandial insulin supplementation) despite treatment with oral hypoglycemic agents (OHAs). The study protocol was conducted in accordance with good clinical practice and the Declaration of Helsinki, was approved by the Ethics Committee of the Landesärztekammer Baden-Württemberg (Stuttgart, Germany) on May 19th 2009, and registered with clinicaltrials.gov (NCT00941369). All patients provided written informed consent prior to inclusion.
Patient selection
Patients of either gender (aged 18–80 years) with type 2 diabetes mellitus according to the American Diabetes Association criteria [17], were considered eligible for the study. Further inclusion criteria were a body mass index (BMI) of >22 to <40 kg/m2, HbA1c of ≥7.0 to ≤10.0 %, and fasting blood glucose (FBG) of ≥120 mg/dL (6.7 mmol/L).
In order to allow for an accurate comparison of the two types of insulin, patients were excluded from the study if they had received treatment with any insulin within the 3 months prior to inclusion, treatment with more than two OHAs within the 4 weeks prior to inclusion, or continuous treatment with thiazolidinediones or glucagon-like peptide (GLP)-1 receptor agonists. Other factors that may significantly affect quality of life or emotional well-being were also indications for exclusion from the study. These included a history of ketoacidosis, a history of drug or alcohol abuse, diabetic retinopathy with surgical treatment (laser photocoagulation or vitrectomy) in the 3 months prior to study entry or which may require surgical treatment within 3 months, prior pancreatectomy, impaired hepatic function, impaired renal function, current treatment for a psychiatric illness (not further specified), systemic corticoid treatment for more than 2 months, prior bariatric surgery, or major dietary changes for weight management during the last 3 months resulting in weight reduction of >5 kg.
Study design and treatments
The study consisted of a two-week screening phase, followed by two 24-week treatment periods, without a washout period in between. After the second treatment period, patients were followed for an additional week (Fig. 1). In each study center, patients were block randomized on a 1:1 basis to either sequence A, starting with insulin glargine (period 1; weeks 1–24) and then switching to NPH insulin (period 2; weeks 25–48); or to sequence B, starting with NPH insulin (period 1) and then switching to insulin glargine (period 2). A crossover design was chosen to allow for patients to serve as their own controls.
Insulin glargine (Sanofi, Berlin, Germany) and NPH insulin (Sanofi, Berlin, Germany) were injected with the TactiPen® injector pen (Sanofi, Berlin, Germany), which is a re-usable insulin delivery device. Insulin glargine was administered by subcutaneous injection once daily, at any time, but each day at the same time. NPH basal insulin was administered at bedtime (21:00–23:00). If the NPH dose was exceeding 30 IE and/or nocturnal hypoglycaemia occured, the NPH dose was split into two doses. One dose was injected at bedtime as described and the second dose was given in the morning (07:00 – 09:00).
During the first week of the treatment phase (forced titration phase), insulin titration was carried out daily. The titration target value was FBG ≤ 5.6 mmol/L. Patients increased their insulin dose following a predefined titration algorithm until the target FBG value was reached. The starting doses were 10 units (U) of insulin glargine and 10 insulin units (I.U.) of NPH insulin per day. At 24 weeks, patients were switched to their second insulin treatment regimen, again following the predefined titration schedule. Four-week titration schemes were used to obtain the same glycemic treatment targets in both insulin therapy regimens during each treatment period, a procedure which was only limited by general limitations of diabetes treatment, such as increasing hypoglycemia or other safety aspects.
In addition to insulin treatment, one or a maximum of two OHAs were allowed (metformin, sulfonylurea, or dipetidyl peptidase (DPP)-IV inhibitors). The dosage of the OHAs remained stable during the study period. In case of postprandial blood glucose (PPG) values exceeding 11.1 mmol/L on two consecutive visits, treatment with prandial short-acting insulin was allowed.
Adherence to the insulin titration algorithm was confirmed by self-report of the physician. Insulin treatment adherence and adherence to OHA treatment of patients were measured by self-report and by evaluating the amounts of prescribed insulins and OHAs, respectively.
Efficacy and safety endpoints
The primary efficacy endpoint of this study was a comparison of insulin glargine and NPH insulin used in BOT in terms of a composite DRQoL score, which was assessed at the end of each of the two treatment periods. The DRQoL consisted of a standardized and unweighted ITEQ score (Cronbach’s α = 0.93) [18], a PAID questionnaire score (Cronbach’s α = 0.86) [19, 20], and the mental health score of the SF-12® Health Survey [21]. The ITEQ was used to assess a range of factors, including leisure activities, sleep, weight control, and diabetes control, as well as general treatment satisfaction. The PAID questionnaire was designed to evaluate diabetes-specific emotional stress. The mental health score of the SF-12® included questions to indicate overall mental health as perceived by the patient. After converting the three sub-scores from these questionnaires to values in a 0–100 range, the composite score was calculated using the following formula: DRQoL = 1/3 * (ITEQ + (100 − PAID) + SF-12®). The range of values for DRQoL was 0–100, with 100 being the optimal value.
In addition to the composite DRQoL score, the individual questionnaire scores, and those from the EuroQol (EQ-5D) questionnaire [22, 23], and the Diabetes Treatment Satisfaction Questionnaire status version (DTSQs), were assessed. Each of the questionnaires was completed at baseline, crossover visit, and end of study, except for ITEQ which was not completed at baseline because participants included had no prior insulin treatment. Further secondary efficacy variables assessed the level of glycemic control by evaluating HbA1c values, FBG, and 7-point blood glucose profiles (determined by self-measured blood glucose readings). Additional parameters assessed as secondary variables were body weight, waist circumference, blood pressure, and lipids. Further secondary objectives and assessments included hypoglycemic events (symptomatic and/or severe), total daily insulin doses, and the patients’ treatment preference for insulin glargine vs. NPH insulin reported at the end of the study.
Safety endpoints were total number of serious adverse events (SAE) and adverse events (AE), including all forms of hypoglycemia, in particular, severe hypoglycemia (secondary efficacy endpoint), and localized pain, redness, or inflammation at the injection site. A 7-day follow-up period was used to ensure that all events that may have been related to treatment were included.
Statistical considerations and analysis
In a previous cross-sectional study [16], different effect sizes of insulin glargine compared to NPH insulin in terms of SF-12®, PAID, and ITEQ scores were observed (d = 0.10, 0.11, and 0.29, respectively). The average effect size of all three scales was d = 0.166. Since the present study had a crossover design, in which each participant served as his/her own control, an effect size on the primary endpoint DRQoL of d = 0.20 was expected. Such an effect can be detected with 90 % power using a paired t-test with a significance level of 5 % and with 265 patient pairs. Considering a non-evaluable rate of 20 %, a total of 332 patients were to be enrolled in order to have 265 patients (completing both treatments) evaluable for the efficacy analysis.
The primary efficacy endpoint was evaluated by analysis of covariance (ANCOVA). The model included fixed effects for treatment, sequence, and period (treatment by sequence interaction), as well as a random effect to account for subjects within sequence. The Shapiro-Wilk test was applied to test the model assumption of normality of residuals at a critical level of 0.1. Statistical tests were performed at a significance level of α = 0.05.
Wherever possible, secondary endpoints were evaluated by analyzing changes from the start of the respective treatment period to its end. If applicable, treatment comparisons for secondary efficacy variables were made by the variance analytical approach described for the primary efficacy endpoint. The number of patients with at least one hypoglycemic event and the number of hypoglycemic events per patient year of insulin treatment was analyzed. Treatment differences in hypoglycemia rates were analyzed using the McNemar test. The same test was used to analyze response rates. For the questionnaires, single domain scores were summarized descriptively by treatment period. The total scores were also analyzed by ANCOVA. For the primary and secondary efficacy variables, two subgroups (by treatment sequence) were analyzed (within patient comparison): A) starting with insulin glargine and then switching to NPH insulin, and B) starting with NPH insulin and then switching to insulin glargine. In addition, treatment comparisons were performed focusing on each period separately (between patient comparisons).
Data entry, verification, and validation were carried out using SAS version 9.2.