This was a 24-week, multicenter, open-label, randomized, parallel-group, phase III study in patients with T1D comparing immunogenicity, safety, and efficacy of MYL-1601D with Ref-InsAsp-US. It was conducted in 149 centers in the United States. The study comprised a 4-week run-in period, a 24-week treatment period, and a 4-week follow-up period. Patients were randomized in a 1:1 ratio to receive either MYL-1601D (100 U/mL) or reference Ref-InsAsp-US (100 U/mL) once daily. A follow-up visit, via a telephone call, was scheduled 4 weeks after the last dose of MYL-1601D or Ref-InsAsp-US.

For the study, patients aged 18–65 years, with a body mass index of 18.5–35.0 kg/m2, and established diagnosis of T1D were included. All patients were on insulin treatment (stable dose of once-daily basal Lantus or Toujeo injection and multiple daily bolus Ref-InsAsp-US or Humalog injections) for a minimum of 3 months before screening. At screening, patients had a glycated hemoglobin (HbA1c) concentration of 6.5–10.0% and a hemoglobin of ≥ 10.0 g/dL. Key exclusion criteria included patients with a history of clinically significant infections and medical conditions, autoimmune disorders, history of hematological disorders, insulin pump usage in the last 3 months before screening, any non-insulin antidiabetic therapies, clinically significant abnormal laboratory data, secondary diabetic complications of moderate insulin, moderate insulin resistance, and patients who planned to receive elective surgery during the study period.

The clinical study protocol and other essential clinical documents were reviewed and approved by an institutional review board/ethics committee at each clinical site. This study was conducted in accordance with legal and regulatory requirements, International Ethical Guidelines for Biomedical Research Involving Human Patients [11], International Council for Harmonisation Good Clinical Practice (ICH-GCP) [12], and the Declaration of Helsinki [13]. Written informed consent was obtained from all patients prior to study inclusion. The study was registered at ClinicalTrials.gov (NCT03760068).

After a 3-week screening period, during the run-in period, all patients received subcutaneous injection of FlexPen® Ref-InsAsp-US at a concentration of 100 U/mL (Batch numbers: HZF7372; HZFA196; JZFC499; JZFC879; manufactured by Novo-Nordisk) at mealtime until randomization. In addition, all patients were shifted from their current basal insulin to Lantus SoloSTAR® (insulin glargine injection, 100 U/mL, manufactured by Sanofi-Aventis) once daily at the start of the run-in period and continued till study completion. The doses of Ref-InsAsp-US and Lantus were titrated during the run-in period to ensure diabetes control. During the treatment period, all patients received one of the following treatments: MYL-1601D (Batch number: BM18002196) manufactured by Biocon from a manufacturing process or Ref-InsAsp-US taken at mealtime in a prefilled disposable pen with a 3-mL cartridge. Frequent adjustments of insulin dose were discouraged. Study treatment dose and titration instructions were given to the patient at the time of medication dispense, and all subsequent study visits were as per the recommended ADA 2019 standard-of-care specifics (the targets were prandial or post prandial). The 7-point self-monitoring of blood glucose (SMBG) diary was assessed, reviewed, and discussed at each visit to ensure the effectiveness and safety of glycemic control.

The primary endpoint was treatment-emergent antibody response (TEAR) rate during the 24-week treatment period. In this study, TEAR was defined as patients who were anti-insulin antibody (AIA) negative at baseline and became positive at any timepoint post-baseline (treatment-induced AIA) or patients who were AIA positive at baseline and had 4-fold increase in titer value at any time post-baseline (treatment-boosted AIA). The criteria of a 4-fold increase was considered a scientifically reasonable margin to define treatment-boosted AIA [14, 15]. The TEAR rate was not assessed in isolation but was part of the totality of evidence including changes in HbA1c, fasting plasma glucose (FPG), insulin dose, neutralizing antibodies (NAb), and injection-site reaction (ISR) to determine if any observed changes in TEAR rate were clinically meaningful.

Efficacy endpoints were assessed as secondary endpoints and included change from baseline to week 24 in HbA1c, FPG, prandial, basal, and total daily insulin dose per unit body weight (U/kg), and 7-point SMBG profile.

Safety endpoints included incidence of positive antibody response and NAb, impact of AIA on PD parameters, such as FPG, HbA1c, and insulin dose, change in hypoglycemia rate (30-day adjusted), incidence of hypoglycemic events, incidence of treatment-emergent adverse events (TEAEs) and serious adverse events (SAEs), ISRs, systemic reactions, hypersensitivity, and immune-mediated AEs, and device-related safety assessments. All subjects received a diary and blood glucose monitoring device (glucometer) at week − 4 to monitor blood glucose at home and to record the 7-point SMBGs. The incidence of hypoglycemic episodes was summarized by category; severe hypoglycemia (an event requiring the help of another person to actively administer carbohydrate, glucagon, or different resuscitative actions), documented symptomatic hypoglycemia (an event of hypoglycemia accompanied by a measured plasma glucose concentration ≤ 70 mg/dL [3.9 mmol/L]), asymptomatic hypoglycemia (an event of hypoglycemia not accompanied by a plasma glucose concentration ≤ 70 mg/dL [3.9 mmol/L]), probable symptomatic hypoglycemia (hypoglycemia with no glucose level activity that resolved with food intake, body covering endocrine, or endogenous glucose), relative hypoglycemia (an event of hypoglycemia accompanied by a plasma glucose concentration > 70 mg/dL [3.9 mmol/L]), and nocturnal hypoglycemia (hypoglycemia that happens from the time the patient goes to bed at nighttime until the time he or she wakes up) were considered as serious adverse events [16, 17].

Pre-dose serum samples for immunogenicity assessments were collected at baseline and at pre-specified time points of 2, 4, 8, 12, 16, 20, and 24 weeks after randomization. A conventional radioimmunoprecipitation assay was used to detect the presence of AIAs in a tiered approach (screening, confirmation, and characterization) [14, 18] using 125I-MYL-1601D as tracer. All confirmed AIA positive samples were evaluated in the characterization tier which included titer determination and evaluation of insulin cross-reactivity using excess human insulin in the confirmatory tier. During method validation, the screening cut point factor (1.27), confirmatory cut point (47.2%), and titer cut point factor (1.66) were determined in healthy human sera using the statistical methods consistent with robust procedures recommended by Shankar et al. [14] and Devanarayan et al. [19]. The sensitivity of the screening and confirmatory assays was 10.36 ng/mL and 16.03 ng/mL at the 95% consistency level, respectively, and both assays exhibited precision of < 20% CV at the low positive control level. The titration assay precision yielded a minimum significant ratio of 2 [20], sufficient to support the significance of a 4-fold increase in titer for treatment-boosted ADA. Confirmed AIA positive samples were also tested for the presence of NAb using a separate cell-based assay in which the inhibition of insulin receptor phosphorylation in a transfected Chinese hamster ovary cell line overexpressing insulin receptor was measured.

The sample size estimation was planned with approximately 500 patients based on the primary objective. The 80% power would be achieved with 250 patients per treatment group to demonstrate that the 90% confidence interval (CI) of treatment difference (MYL-1601D minus Ref-InsAsp-US) is within the margin.

The equivalence margin was provided by the regulatory agency with fixed sample size (n = 500) based on publications by Chow et al. [21, 22]. The margin was determined by the TEAR rate of the reference treatment group. However, due to the nature of binomial distribution, the margin is largest when the reference TEAR rate is 50%, where the variance is largest, and margins decrease as reference rates increase or decrease away from 50%. The final TEAR rate for primary analysis in the Ref-InsAsp-US group was 27.8% and therefore the final margin of ± 11.7% was derived based on the reference group rate and 500 planned sample size. The margins for sensitivity analyses were calculated according to different TEAR rates in the reference treatment group (Fig. S1, see electronic supplementary material [ESM]).

Missing data were imputed using multiple imputation methods for primary and sensitivity analysis for TEAR rate associated with AIA and secondary variables. A different imputation process was used for each of the two TEAR criteria: (i) logistic regression multiple imputations assuming missing not at random for baseline AIA-negative patients, post-baseline binary response (positive or negative); (ii) for baseline positive patients, missing titer values (continuous values) were imputed with the same treatment group non-missing patients using the pattern mixture model with the complete-case missing values method [23].

All statistical procedures were performed using Statistical Analysis Software (SAS®) 9.3 or higher (SAS Institute, Cary, NC, USA). The 90% CI of treatment difference in TEAR rate was established using the Wald confidence limit method. The TEAR rate with AIA of MYL-1601D was established equivalent to Ref-InsAsp-US if the 90% CI of the treatment difference was within the margin of − 11.7% to 11.7%. For secondary and safety continuous variables, a mixed model repeated measures (MMRM) model was used; for HbA1c and FPG, actual values and changes from baseline at each timepoint were captured and summarized by the treatment group. For HbA1c, treatment difference and 95% CIs were also displayed along with p-values. Sensitivity analyses were performed to evaluate the impact of missing data on primary analysis results.

Safety data were summarized using descriptive statistics. Adverse events (serious and non-serious) were graded in accordance with the NCI-CTCAE scale [24]. The hypoglycemic event rate per subject per 30 days was analyzed using MMRM. For antibody continuous variables, Fisher’s exact test or Chi-squared test was used for categorical data analyses.