Efficacy and safety of GP40021 insulin lispro biphasic compared with Humalog Mix 25 in Type 2 diabetes mellitus patients

This was a Phase III, multicenter, open-label, randomized, active-controlled, parallel group, noninferiority 26-week study in 18 study sites Supplementary Material 1. Eligible study participants with T2D (n = 210) in accordance with the inclusion/exclusion criteria were randomized 1:1 to GP-Lis25 (n = 105) or Ly-Lis25 (n = 105) treatment twice-daily. We performed block randomization with the consideration of HbA1c level (7.6–9.0% or >9.0%) and previous insulin treatment (insulin-naive or currently treated with insulin). Figure 1 shows the flowchart of study subjects.

The study period consisted of screening procedure (up to 4 weeks), insulin dose titration (up to 4 weeks) and treatment period with stable doses (26 weeks). Study participants used prefilled insulin pens. During the period of insulin dose titration, insulin doses could be changed as much as needed to achieve optimal glycemic control in patients. Daily insulin doses during the treatment period should be stable. In special cases (e.g., to provide a subject’s safety if he has frequent hypoglycemic episodes), investigators could reduce insulin dose even during the treatment period with stable doses, however it was not recommended generally. To maintain stable blood glucose levels, participants should measure capillary blood glucose at least four-times a day, and every time they were supposed to have hypoglycemic episode.

The study was registered under ClinicalTrials.gov number NCT04023344.

Eligible participants were 18 to 65 years old, with T2D diagnosed for at least 6 months before screening, with HbA1c level 7.6–12.0% and BMI 18.5–35.0 kg/m² at screening.

Participants regularly taking oral antidiabetic drugs before screening should continue administration in the same dose for at least 3 months before the first dosing of the investigational insulin. However, study subjects could discontinue sulphonylurea treatment in titration period in case of high hypoglycemic risk. Investigators made a decision based on a patient’s glycemic profile.

Only T2D patients took part in the trial as T1D patients had the increased risk of DM decompensation associated with nonpriority indication of insulin mixture. This waiver from the regulatory guidelines [4,5] was approved by the Ministry of Health of the Russian Federation.

Key exclusion criteria were the following:

The full list of the exclusion criteria is provided at ClinicalTrials.gov.

Immunogenicity outcome measures were the main safety end points in this study.

The primary end point was frequency of immune response (AIA concentration >10 units/ml) at week 26.

Secondary immunogenicity end points included mean AIA concentration at weeks 12 and 26, and the change in mean AIA concentration from baseline (at screening) to weeks 12 and 26.

Blood samples of patients having immune response either at week 12, or at week 26, or both, and a more than fourfold increase of AIA concentration at week 12, or at week 26, or both were tested for insulin neutralizing antibodies.

AIA concentration was assessed using the validated enzyme immunoassay method in the central laboratory.

To detect insulin neutralizing antibodies, we used the method with iLiteTM Insulin Assay Ready Cells, introducing the firefly luciferase reporter gene to their genome under the control of an insulin-dependent promoter [8]. Neutralizing activity was established via binding of insulin alfa-chain and CD220 receptor in samples.

Efficacy end points were considered as secondary in accordance with the regulatory guidelines [4,5] and included:

The study investigators selected individual glycemic goal for each participant in accordance with the national guidelines on diabetes treatment [9] at the randomization visit. They did not change this goal during the study periods. HbA1c was measured at screening and at weeks 12 and 26.

FPG level was measured at the screening, in the beginning of treatment, at week 12 and 26. Glycemic level was also controlled daily by the study subjects at least four-times a day using glucometers given by Sponsor at the screening.

SPGP was performed by the participants using glucometers Accu-Chek Active with appropriate testing strips provided by the Sponsor. The study subjects performed SPGP three days a week before visits at weeks 4, 12 and 26, twice in two consecutive days. They measured capillary blood glucose in fasting condition, two hours after breakfast, before lunch, two hours after lunch, before dinner, two hours after dinner and at 3 a.m.

DTSQs and DTSQc questionnaires measure patients’ satisfaction with current diabetes treatment. Both questionnaires consist of eight questions estimating absolute and relevant satisfaction values for DTSQs and DTSQc, respectively. Patients completed DTSQs at randomization and at week 26, DTSQc – only at week 26.

In the study, incidence and severity of adverse events (AEs) including AEs of special interest, such as hypoglycemic episodes, injection site reactions and hypersensitivity reactions, were assessed. In addition to these secondary safety end points, changes in vital signs, changes in electrocardiogram and laboratory results compared with baseline, were also assessed. All investigators should report both all AEs observed in the study and AEs spontaneously reported by study subjects. AE details were collected for all study participants from signing the Informed Consent Form to their last visit.

Each AE was assessed by the study investigators according to the following criteria: seriousness, severity, relationship to the investigational product or comparator (WHO-UMC causality assessment). All AEs were also recorded in source documents and electronic case report forms. If a serious adverse event (SAEs) occurred, it was reported to the sponsor within 24 h. Every SAE was reviewed by the Sponsor’s central Pharmacovigilance and Medical teams based on their clinical judgment. All nonserious AEs were periodically analyzed by the Sponsor’s Medical team.

Hypoglycemic episodes, allergic reactions and injection site reactions were AEs of special interest (according to the study protocol), and they were evaluated separately from other AEs. Hypoglycemic episodes were recorded by the participants in their personal self-care glycemic diaries. Only episodes with capillary blood glucose level less than 3.9 mmol/l were considered hypoglycemic episodes. Physicians assessed hypoglycemia severity depending on a clinical situation, regardless the need in assistance of other people (in the same way they evaluated severity of any AEs). This fact was recorded separately. Minor and major episodes were also registered based on the need of external help. Local injection site reactions and hypersensitivity reactions were assessed during visits of every patient to the study site and were also recorded by the study participants in their diaries. The data were entered into electronic case report forms.

All these measures were used for provision of continuous benefit-risk assessment during the study.

All data analyses were made using R version 3.5.0 for Windows software in accordance with the approved statistical analysis plan. The results of statistical analysis in full analysis set population were presented.

Quantitative statistical analysis is described using:

We used analysis of covariance (ANCOVA) to estimate changes of continuous variables from baseline. For each of these variables, ANCOVA included the baseline variable value as a covariate and treatment group and study site as fixed factors. Student’s t-test or Mann–Whitney U-test (depending on normality of distribution) were used to compare absolute values of continuous variables between the study groups. If variance was not equal in the groups (evaluated with the Levene’s test), Welch’s unequal variances t-test was used instead of Student’s t-test.

Qualitative statistical analysis is described using:

We compared qualitative (categorical) variables using X²-test (with continuity correction in case any unit contains value of 6 to 10) or exact Fisher’s test (if any unit contained value of 0 to 5). We analyzed the change of dichotomous variables from baseline with McNemar test. p-value < 0.05 was accepted statistically significant.

Since insulin immunogenicity is relatively low, high study power is not required for estimation of frequency of immune response [4,5]. In comparative studies of insulin immunogenicity, sample size is estimated on the basis of HbA1c values in accordance with the regulatory guideline [4,5]. We used α level 0.05, β level 0.2, mean difference 0, standard deviation 1.1% and δ 0.4% in accordance with the regular practice for antihyperglycemic therapy [10]. With these inputs, the total number of required study subjects was estimated as 186 (93 per group). Taking into account a possible 10–15% early withdrawal rate, we randomized 210 patients.

The trial was conducted in accordance with the guidelines on Good Clinical Practice and with ethical principles for medical research involving human subjects laid down in the Declaration of Helsinki [11]. All study participants provided the written informed consent before participating in the study. The trial was approved by the Ministry of Health of the Russian federation (Clinical trial authorization No. 644 dated 11 December 2017) and by the Ethics Council (abstract from the meeting minutes No 159 dated 21 December 2017).

We screened 257 T2D patients and randomized 210 of them (105 to GP-Lis25, 105 to Ly-Lis25 groups). Three participants discontinued GP-Lis25 therapy (one subject decided to withdraw from the study, another one was lost to follow-up, the third one had adverse reactions), and two of the subjects discontinued Ly-Lis25 therapy (one subject decided to withdraw from the study, another one was lost to follow-up). Figure 1 provides the flowchart of the study participants. Baseline characteristics of the study subjects are presented in Table 1. All baseline characteristics were comparable between the groups except prior insulin therapy: more patients in Ly-Lis25 group were administered with only basal insulin at baseline.

Frequency of immune response at week 26 did not differ between the groups (Table 2).

Mean AIA concentration at week 12 was 6.13 ± 12.93 units/ml in GP-Lis25 group and 3.78 ± 6.02 units/ml in Ly-Lis25 group. By week 26, mean AIA concentration was 5.48 ± 11.24 units/ml in GP-Lis25 group and 3.37 ± 4.82 units/ml in Ly-Lis25 group. Changes in AIA levels by weeks 12 and 26 are shown in Table 2. These changes were not statistically significant in both groups.

Neither AIA concentration at week 26, nor change in AIA concentration by week 26 from baseline influenced treatment efficacy (p = 0.734 and p = 0.727, respectively) and insulin dose (p = 0.282 for AIA concentration at week 26 and p = 0.378 for change in AIA concentration by week 26).

Blood samples of 12 patients in GP-Lis25 group and eight patients in Ly-Lis25 group having more than fourfold increase of AIA concentration at week 12, or at week 26, or both, were tested for insulin neutralizing antibodies. Three out of 12 patients in GP-Lis25 group and three out of eight patients in Ly-Lis25 group developed neutralizing AIAs during the study period, but none of them had any clinical signs showing that efficacy of insulin treatment was reduced. GP-Lis25 and Ly-Lis25 groups did not differ significantly in incidence of neutralizing AIA formation (p = 0.642).

HbA1c level was significantly decreased from baseline at week 12 for 1.36 ± 1.22% in GP-Lis25 group (p < 0.001) and for 1.34 ± 1.16% in Ly-Lis25 group (p < 0.001). HbA1c level was significantly decreased from baseline at week 26 for 1.59 ± 1.23% in GP-Lis25 group (p < 0.001) and for 1.60 ± 1.21% in Ly-Lis25 group (p < 0.001). These changes did not differ between the groups either at week 12 (p = 0.192), or at week 26 (p = 0.381). HbA1c changes are shown in Figure 2.

Intergroup difference of HbA1c level change at week 26 from baseline was (95% CI) 0.01 (-0.27–0.28)%. The CI does not cross the established 0.4% noninferiority margin demonstrating that GP-Lis25 is noninferior to Ly-Lis25 in terms of hypoglycemic efficacy.

Twenty-six patients (25%) in GP-Lis25 group and 23 (22%) patients in Ly-Lis25 group could achieve the pre-established individual glycemic goal at week 26 (p = 0.714).

FPG level was significantly decreased from baseline at week 12 for 2.29 ± 3.82 mmol/l in GP-Lis25 group (p < 0.001) and for 1.78 ± 4.35 mmol/l in Ly-Lis25 group (p = 0.001). FPG level was also decreased significantly from baseline at week 26 for 2.55 ± 4.24 mmol/l in GP-Lis25 group (p = 0.001) and for 1.98 ± 4.22 mmol/l in Ly-Lis25 group (p = 0.001). These changes did not differ between the groups either at week 12 (p = 0.296), or at week 26 (p = 0.092). FPG changes are shown in Figure 3.

SPGP changes did not differ between the groups either at week 12 (p = 0.896, p = 0.811, p = 0.208, p = 0.432, p = 0.569, p = 0.214, p = 0.423 in fasting condition, two hours after breakfast, before lunch, two hours after lunch, before dinner, two hours after dinner and 3 a.m., respectively), or at week 26 (p = 0.407, p = 0.492, p = 0.832, p = 0.639, p = 0.801, p = 0.748, p = 0.762, for the abovementioned points, respectively). SPGP changes are shown in Figure 4.

Total insulin dose was significantly increased in both groups at week 26 compared with week 4 for 1.70 ± 5.97 units/day in GP-Lis25 group (p = 0.004) and 1.36 ± 4.79 units/day in Ly-Lis25 group (p = 0.005), but these changes did not differ between the groups (p = 0.878). The changes in basal insulin dose and bolus insulin dose were also not statistically significant and did not differ between the groups (p = 0.878 and p = 0.892, respectively).

Body weight of the patients was increased at week 26 compared with baseline values for 0.69 ± 2.33 kg in GP-Lis25 group (p = 0.004) and for 0.63 ± 2.29 kg in Ly-Lis25 group (p = 0.004). These changes did not differ between the groups (p = 0.832).

Treatment satisfaction was increased in both groups at week 26. The increase in DTSQs total score was 7.72 ± 8.44 points in GP-Lis25 group (p < 0.001) and 7.89 ± 8.65 points in Ly-Lis25 group (p < 0.001) and was similar between the groups (p = 0.258). DTSQc scores were 11.02 ± 6.68 points in GP-Lis25 group and 12.05 ± 6.28 points in Ly-Lis25 group and did not differ between the groups (p = 0.308).

The incidence of AEs, SAEs, discontinuations due to AEs, and adverse drug reactions reported in GP-Lis25 group were similar to those in Ly-Lis25 group (Table 3). One patient in GP-Lis25 group discontinued the clinical trial because of an AE (stroke, not related to the study treatment, the patient met the exclusion criteria). The relationship of adverse drug reactions to the investigational drug was assessed by the investigators as unlikely in all 6 cases (5 in GP-Lis25 group and 1 in Ly-Lis25 group).

Most of these AEs were mild. There were five SAEs: four in GP-Lis25 group and one in Ly-Lis25 group, none of them were related to the study drugs. Three out five study participants with SAEs continued the study therapy, whereas two of them stopped using the investigational insulin drug.

One patient had one injection site reaction in GP-Lis25 group, and patients in Ly-Lis25 group did not develop injection site reactions. No hypersensitivity reactions were observed in either of the groups. There were no differences in any laboratory safety values or vital signs between the groups (data not shown).

Table 4 contains the summary of hypoglycemic episodes. The number of patients experiencing hypoglycemic episodes did not differ between the groups (p = 0.782). Ly-Lis25 group had less hypoglycemic episodes for 2.32 events/patient year. There were no severe hypoglycemic episodes (loss of consciousness, need in other people assistance) during the study. The investigators also evaluated severity of the episodes depending on a clinical situation, regardless the need in assistance of other people (in the same way they evaluated severity of any AEs). Most patients in both groups experienced mild hypoglycemia. Most hypoglycemic episodes were symptomatic, and there were less asymptomatic episodes in GP-Lis25 group (p < 0.001), but the number of patients having symptomatic or asymptomatic episodes did not differ between the groups (p = 0.344). There were no major (required external help) hypoglycemic episodes in this trial.