Participants

A total of 144 participants from Japan were randomized to receive efsitora (n = 71) or degludec (n = 73) (Table 1). Of these, 99 participants (68.8%) received the general dosing algorithm (efsitora, n = 50; degludec, n = 49) and 45 participants (31.3%) received the alternative dosing algorithm (efsitora, n = 21; degludec, n = 24).

Table 1 Demographic and baseline clinical characteristics of Japanese QWINT-2 participants

Demographic and baseline characteristics were generally balanced between the efsitora and degludec groups. As expected, on the basis of differences in the criteria for the dosing algorithms, participants who received the alternative dosing algorithm had lower mean body weight, HbA1c levels, BMI, and fasting serum glucose levels at baseline. Compared with the general dosing algorithm group, the average age of participants in the alternative dosing algorithm group was higher, with a lower percentage of male participants. Participants had received a range of different noninsulin glucose-lowering agents, with no apparent differences in patterns of agent between treatment groups or dosing algorithms.

Change in HbA1c from Baseline to Week 52

Efsitora showed comparable efficacy to degludec, irrespective of the dosing algorithm, as evidenced by the LSM change in HbA1c level from baseline to week 52 (Figs. 1A, B). Over this period, mean HbA1c decreased from 8.04% to 6.63% with efsitora (LSM change, −1.37%) and from 8.00% to 6.64% with degludec (LSM change, −1.36%; estimated treatment difference, −0.01%) (Fig. 1A). Participants who received efsitora according to the alternative dosing algorithm had a smaller decrease in mean HbA1c compared with those following the general dosing algorithm (LSM change −0.92% versus −1.56%), likely due to the lower baseline HbA1c in participants who received the alternative dosing algorithm (Fig. 1B). A mean HbA1c level of less than 7% was reached with efsitora at week 12 with both dosing algorithms and was maintained through the end of the treatment period (Fig. 1B).

Fig. 1The alternative text for this image may have been generated using AI.

Change in HbA1c (%) over time in Japanese QWINT-2 participants. Data are presented as least-squares mean (LSM) ± standard error. Error bars represent standard error. The delta symbol (Δ) indicates the LSM change from baseline to week 52. A Overall Japanese population. B General dosing algorithm and alternative dosing algorithm. C Change in HbA1c in participants who received the general and alternative dosing algorithms from weeks 0–52 by GLP-1 RA use. CI confidence interval, ETD estimated treatment difference, GLP-1 RA glucagon-like peptide-1 receptor agonist, HbA1c glycated hemoglobin

Similar trends in efficacy were observed when participants were analyzed by GLP-1 receptor agonist use (Fig. 1C). Among participants who used GLP-1 receptor agonists, the estimated treatment difference (95% CI) between efsitora and degludec was −0.12 (−0.52, 0.27) with the general dosing algorithm and −0.27 (−0.67, 0.13) with the alternative dosing algorithm. Among participants who did not use GLP-1 receptor agonists, the estimated treatment difference was +0.28 (−0.14, 0.70) with the general dosing algorithm and −0.06 (−0.45, 0.34) with the alternative dosing algorithm.

TIR from Baseline to Weeks 48–52

From weeks 48–52, TIR was comparable between participants in the efsitora and degludec groups. TIR increased from baseline to the final 4 weeks of treatment (weeks 48–52) (Fig. 2A). TBR was also comparable between the efsitora and degludec groups (Figs. 2A, B).

Fig. 2The alternative text for this image may have been generated using AI.

Percentage of time in different ranges measured by continuous glucose monitoring over 24 h in Japanese QWINT-2 participants. The percentage of time spent at various glucose levels, as measured by continuous glucose monitoring, are presented as least-squares means (LSMs). A Overall Japanese population at baseline and weeks 48–52. B Overall Japanese population at weeks 0–4, weeks 4–8, and weeks 8–12. Percentages may not total 100% because of rounding

Change in FBG from Baseline to Week 52

From baseline to week 52, mean FBG level decreased from 177.6 to 118.1 mg/dL with efsitora (LSM change, −60.6 mg/dL) and from 182.2 to 119.3 mg/dL with degludec (LSM change, −59.4 mg/dL; estimated treatment difference, −1.18) (Fig. 3A). Participants who received efsitora according to the alternative dosing algorithm had the lowest FBG levels among all treatment groups throughout the treatment period (Fig. 3B). Furthermore, with the alternative dosing algorithm, the LSM mean change in mean FBG at week 52 was larger in the efsitora group than in the degludec group (Fig. 3B).

Fig. 3The alternative text for this image may have been generated using AI.

Change in fasting blood glucose measured by participant-monitored glucose testing over time in Japanese QWINT-2 participants. Data are presented as least-squares mean (LSM) ± standard error. Error bars represent standard error. The adjusted fasting blood glucose values were based on participant-assigned fasting time point and the first reading between 5:00 AM and 10:00 AM. The delta symbol (Δ) indicates the LSM change from baseline to week 52. A Overall Japanese population. B General dosing algorithm and alternative dosing algorithm. CI confidence interval, ETD estimated treatment difference, SMBG self-monitored blood glucose

Change in Insulin Dose from Baseline to Week 52

At week 52, the total weekly insulin dose was 194.3 U/week (27.8 U/day) for efsitora and 226.5 U/week (32.4 U/day) for degludec, with an estimated treatment difference of –32.3 U/week (Fig. 4A). At week 52, weekly efsitora doses were also lower than weekly degludec doses for both dosing algorithms (Fig. 4B). The difference in insulin dose between the efsitora and degludec groups was greater with the alternative dosing algorithm than with the general dosing algorithm (Fig. 4B).

Fig. 4The alternative text for this image may have been generated using AI.

Change in insulin dose over time in Japanese QWINT-2 participants. Data are presented as the least-squares mean (LSM) weekly insulin dose over time. Error bars represent standard error. A Overall Japanese population. B General dosing algorithm and alternative dosing algorithm. CI confidence interval, ETD estimated treatment difference, U units

Rate and Incidence of Hypoglycemia from Baseline to Week 52

The rate of combined level 2 or 3 hypoglycemic events was generally low and similar between the efsitora and degludec groups throughout the treatment period (Table 2). No participants treated with efsitora reported a level 3 hypoglycemic event with either dosing algorithm. Level 3 hypoglycemia was reported in one participant treated with degludec with each dosing algorithm.

Table 2 Event rate per year of hypoglycemia in Japanese QWINT-2 participantsSafety of the One-Time Starting Dose of Efsitora

TIR increased from weeks 0–12 during the initial dosing period and was comparable between the efsitora and degludec groups. TBR was also comparable between efsitora and degludec during the initial dosing period, with values for efsitora versus degludec of 0.5% versus 0.3% at weeks 0–4, 0.5% versus 0.5% at weeks 4–8, and 0.7% versus 0.6% at weeks 8–12. CGM data collection during weeks 0–12 showed no evidence of increased TBR due to the one-time starting dose of efsitora.

The rates of combined level 2 or 3 hypoglycemia during the initial dosing period (weeks 0–12) were low and comparable between the efsitora and degludec groups. During this period, seven episodes of combined level 2 or 3 hypoglycemic events were reported with efsitora in 71 participants and six events were reported with degludec in 73 participants (Supplementary Table 1).

Adverse Events

The overall incidence of TEAEs was similar between the efsitora and degludec groups. TEAEs were reported in 50 participants (70.4%) who received efsitora and 57 participants (78.1%) who received degludec. A similar number of TEAEs were reported between the efsitora and degludec groups for each dosing algorithm. Serious adverse events were reported by three participants with efsitora (angina unstable, glaucoma, and cellulitis) and six participants with degludec (chronic tonsillitis, cholecystitis acute, meniscus injury, tibia fracture, osteoarthritis, benign spleen tumor, and hypoglycemia [n = 2]). No participants using either dosing algorithm discontinued the study or study treatment due to an adverse event, and no deaths were reported during the study (Supplementary Table 2).

Serious adverse events (SAEs) were comparable between the efsitora and degludec groups. Seven participants (7.1%) who received the general dosing algorithm experienced at least one SAE during the study, with similar SAE incidence rates between the efsitora and degludec groups. Among participants who received the alternative dosing algorithm, no SAEs were reported in the efsitora group, and two participants (8.3%) experienced at least one SAE in the degludec group.