De-intensification of basal-bolus therapy by replacing prandial insulin with once-weekly subcutaneous semaglutide in individuals with well-controlled type 2 diabetes: A single-centre, open-label randomised trial (TRANSITION-T2D)
Abstract
Aims
The study aims to examine the outcome of replacement of prandial insulin with once-weekly subcutaneous semaglutide in people with type 2 diabetes reasonably controlled on multiple daily insulin injections (MDI).
Materials and Methods
This single-centre, randomised, open-label trial enrolled a statistically predetermined sample of 60 adults with HbA1c ≤7.5% (58 mmol/mol) receiving MDI, with a total daily dose (TDD) ≤120 units/day. Participants were assigned 2:1 to subcutaneous semaglutide 1.0 mg plus insulin degludec, or to continue MDI. The primary outcome was percentage of subjects maintaining HbA1c ≤7.5% (58 mmol/mol) at Week 26.
Results
At Week 26, 90% of semaglutide and 75% of MDI subjects maintained HbA1c ≤7.5% (≤58 mmol/mol) (p = 0.18). Mean changes (95% CI) in HbA1c, weight and percentage body weight for semaglutide versus MDI, respectively, were −0.5% (−0.7, −0.3) versus 0.0% (−0.3, 0.3; p = 0.009); −8.9 kg (−9.9, −7.8) versus 1.5 kg (−0.1, 3.1; p < 0.001); and −8.6% (−9.6, −7.6) versus 1.4% (0.0, 2.8; p < 0.001). Insulin TDD decreased 56.0% (−62.3, −49.7) with semaglutide and increased 6.7% (−2.5, 16.0) with MDI (p < 0.001). Among semaglutide subjects, 58% reduced insulin TDD > 50%, 97.5% stopped prandial insulin and 45% lost >10% body weight. Participant treatment satisfaction scores trended higher with semaglutide. Hypoglycaemia frequency was similar between groups.
Conclusions
In people with type 2 diabetes well controlled (HbA1c ≤7.5% [≤58 mmol/mol]) on MDI ≤120 units/day, replacing multiple daily injections of prandial insulin with once-weekly subcutaneous semaglutide can maintain and even improve HbA1c, lower body weight and lessen the burden of management.
1 INTRODUCTION
Many individuals with type 2 diabetes initiate multiple daily injections of insulin (MDI) with basal and bolus insulin to achieve adequate glycaemic control. Despite initial improvements in glycated haemoglobin (HbA1c), long-term MDI is associated with poor glycaemic control1, 2 for reasons including undesired weight gain (in a population that already has a high prevalence of overweight/obesity), increased hypoglycaemia frequency/severity, missed injections, ineffective prandial insulin titration and increased management burden (frequent injections, heightened glycaemia vigilance). Ideally, this therapeutic strategy for type 2 diabetes would be restricted to those unable to achieve glycaemic control with other pharmacological interventions.
Once initiated, MDI typically becomes a permanent/life-long therapy, and failure to de-intensify treatment (a form of therapeutic inertia) in such individuals is well documented.3-5 The American Diabetes Association recognises that complex insulin regimens may require simplification for patients with type 2 diabetes and self-management challenges, citing the elderly, in particular.6 Yet, practical guidance on de-intensification or simplification of basal-bolus management is lacking, as is an acknowledgement of the range of possible patient-specific reasons to de-intensify or simplify MDI; these might include hypoglycaemia concerns, a desire to reduce the overall burden of management for quality-of-life reasons or weight-loss goals.
The efficacy of combining basal insulin with glucagon-like peptide-1 receptor agonists (GLP-1RAs) has been demonstrated in numerous studies involving individuals with type 2 diabetes taking only basal insulin at baseline who were randomised to treatment intensification with added GLP-1RA or prandial insulin (MDI). The combination of GLP-1RA plus basal insulin has been associated with superior glycaemic control and lower hypoglycaemia rates as compared to MDI.7-9 However, for individuals already on MDI, there remains an unmet clinical need for data on safe and effective strategies to de-intensify treatment by replacing prandial insulin with a non-insulin therapy (e.g., GLP-1RA). The few studies evaluating replacement of MDI with fixed-dose combination GLP-1RA/basal insulin or once-weekly albiglutide in combination with basal insulin have reported encouraging data, but were limited by moderate success in terms of prandial insulin discontinuation and/or entailed complex transition protocols.10-12 We posited the likelihood of even more robust clinical benefits than reported in previous studies, along with less treatment burden, by replacing prandial insulin with a more effective, long-acting GLP-1RA using a relatively simple strategy.
This study was designed to investigate whether once-weekly subcutaneous semaglutide (in combination with once-daily basal insulin) could fully replace the need for bolus (prandial) insulin, while maintaining (or improving) glycaemic control, in persons with type 2 diabetes reasonably well controlled on MDI. A parallel goal was to achieve the transition from prandial insulin to semaglutide with a high degree of implementation simplicity for real-world practitioners.
2 MATERIALS AND METHODS
2.1 Study design and participants
This investigator-sponsored, 26-week, single-centre, open-label, randomised controlled trial was conducted at two Cleveland Clinic sites (Cleveland, Ohio, United States). The protocol was approved by the local institutional review board and the study was performed in accordance with the Declaration of Helsinki. All participants provided written informed consent. The trial was registered at ClinicalTrials.gov (identifier: NCT04538352).
Eligible subjects were adults (age 18–75 years) with type 2 diabetes and HbA1c ≤7.5% (≤58 mmol/mol) on MDI (defined as ≥3 daily injections of insulin analogues, including ≥2 injections of prandial insulin) up to a total daily dose (TDD) of 120 units and ≤0.5 units/kg per day of prandial insulin. The rationale for these cut-offs was based on clinical experience, previous studies using similar interventions and typical insulin requirements in individuals with type 2 diabetes.6 Only metformin and/or SGLT2-inhibitors were allowed as background diabetes therapy. Key exclusion criteria (see online Protocol) were positive GAD-65 antibodies, end-stage kidney disease, history of bariatric surgery and use of steroids >5 mg prednisone or equivalent.
Cleveland Clinic electronic health records were used to identify potential participants from primary care/endocrinology practices throughout Ohio, who were then referred to one of two Cleveland Clinic endocrinology practice sites for further eligibility assessment.
2.2 Randomisation
Participants were randomly assigned 2:1 to insulin degludec once daily plus subcutaneous semaglutide 1.0 mg once weekly (semaglutide group) or to continue MDI with daily insulin degludec plus prandial insulin aspart (MDI group). Block, stratified randomisation via central allocation was used to ensure equal distribution of participants receiving TDD of insulin <80 or 80–120 units/day between study arms.
2.3 Procedures
For all participants, basal insulin doses were maintained at pretrial levels if ≤0.5 units/kg per day or otherwise adjusted to a maximum of 0.5 units/kg daily.
In the semaglutide group, subcutaneous semaglutide was initiated at a dose of 0.25 mg once weekly (increased as tolerated to 0.5 and 1 mg at Weeks 4 and 8, respectively) and prandial insulin was completely stopped on the day of the first semaglutide injection. During the first 12 weeks (titration period), participants were asked to check blood glucose before meals and at bedtime and use a bolus correction dose (insulin aspart per sliding scale) before meals if glucose values were >150 mg/dL. At Week 12, correction insulin was fully discontinued, the frequency of blood glucose monitoring was decreased to once daily and insulin degludec could be self-titrated according to a basal insulin dose adjustment protocol (Table S1A). All participants were provided with the required drugs and glucose-testing materials.
Participants randomised to the MDI group were transitioned from their existing basal-bolus regimen to insulin degludec and insulin aspart. They were asked to check their glucose values before meals and at bedtime throughout the study and were given an insulin adjustment protocol to optimise management, as well as a sliding scale to correct pre-meal glucose values >150 mg/dL (Table S1).
Baseline assessments included HbA1c, GAD-65 antibodies, serum creatinine, AST/ALT, weight and BMI, and basal and prandial insulin doses. Study visits at Weeks 4, 8, 12 and 26 included a physical examination, evaluation of the need to titrate study medication dosage, recording of daily insulin doses and documentation of any hypoglycaemia events since the last visit, including a review of the participant's glucose monitor. At Weeks 12 and 26, blood samples were collected for HbA1c, serum creatinine and AST/ALT. Participants completed the Diabetes Treatment Satisfaction Questionnaire Status version (DTSQs) at baseline and Week 26 and completed the Diabetes Treatment Satisfaction Questionnaire Change version (DTSQc) at Week 26.
2.4 Outcomes
The primary end-point was the percentage of participants maintaining HbA1c ≤7.5% (≤58 mmol/mol) at Week 26. Key secondary end-points were changes from baseline to Week 26 in HbA1c, body weight (percent and absolute changes) and percent changes in TDD and daily basal insulin dose. Additional secondary end-points included absolute changes in insulin doses (basal, prandial, total), percent change in prandial insulin from baseline to Week 26, percentage of participants who had to resume prandial insulin in the semaglutide arm before Week 26 and changes in treatment satisfaction from baseline to Week 26 (assessed using the DTSQs and DTSQc).12, 13 Percentages of participants achieving >10% categorical weight loss and >50% reduction in insulin TDD were determined post hoc. Safety assessments included rate of hypoglycaemic events from participant glucose logs or glucose meter data (overall and by severity level), and numbers of adverse events, including serious adverse events and events leading to drug discontinuation or dose decrease.
2.5 Statistical analysis
Sample size calculations were performed based on the precision of control rate differences within and between groups. Based on data from a prior albiglutide study11 and utilising a two-sided 95% CI, with 60 subjects randomised 2:1, the control rate with semaglutide plus basal insulin could be estimated to within ±11% to ±15% and the difference between groups could be estimated within a range of ±19% (with pooled control rate of 85%) to ±26% (with pooled control rate of 60%) (Table S2).
All analyses were performed in the intention-to-treat population. Continuous variables were tested using the 2-sample t-test or Wilcoxon rank-sum test, depending on data distribution; categorical variables were tested using Pearson's Χ2 test or Fisher's exact test. Analysis of the primary end-point was conducted using logistic regression with Firth's adjustment, including baseline HbA1c as a covariate, with an overall significance level of 0.05. Secondary end-points were analysed using linear mixed-effect models with baseline variables as covariates. All commonly observed study times were used, and interaction between time and randomisation group was included. To control the family-wise error rate in multiplicity of hypothesis testing and the inflation of type I error probability, key secondary end-points were corrected by Bonferroni adjustment at a significance threshold of 0.01. Other secondary outcomes are presented as point estimates with 95% CIs; these were not adjusted for multiplicity, nor used for hypothesis testing-based conclusion. Generalised linear mixed-effect models with Poisson regression, or negative binomial if overdispersion was detected, were conducted to estimate the frequency of hypoglycaemic events, overall and by level of severity. Safety outcomes were exempt from multiplicity correction and are presented at a significance threshold of 0.05. Data were managed and analysed using R software (version 4.3.1, Vienna, Austria).
2.6 Role of the funding source
Novo Nordisk, the funder of the study, had no role in study design, data collection, data analysis, data interpretation or writing of the report. KMP had full access to all study data and takes responsibility for the integrity and accuracy of the final submitted article.
3 RESULTS
Between 18 January 2021 and 16 September 2022, 60 individuals were enrolled and randomised to the semaglutide group (n = 40) or the MDI group (n = 20) (Figure 1). Baseline demographic and clinical characteristics were similar between groups (Table 1). The median self-reported duration of type 2 diabetes was 18.5 years, median HbA1c was 6.8% (51 mmol/mol) and median body weight and BMI were 104.6 kg and 35.3 kg/m2, respectively.
| Characteristic | Overall (N = 60) | MDI (n = 20) | Semaglutide (n = 40) |
|---|---|---|---|
| Sex | |||
| Male | 35 (58%) | 12 (60%) | 23 (57%) |
| Female | 25 (42%) | 8 (40%) | 17 (42%) |
| Age | 68.6 (58.7, 71.5) | 64.7 (54.5, 69.3) | 70.1 (62.7, 72.0) |
| Ethnicity | |||
| Asian/Pacific Islander | 1 (1.7%) | 1 (5.0%) | 0 (0%) |
| Black/African American | 20 (33%) | 8 (40%) | 12 (30%) |
| Hispanic/Latino | 1 (1.7%) | 0 (0%) | 1 (2.5%) |
| Other | 1 (1.7%) | 1 (5.0%) | 0 (0%) |
| White | 37 (62%) | 10 (50%) | 27 (68%) |
| Metformin therapy | 29 (48%) | 10 (50%) | 19 (48%) |
| SGLT2i therapy | 9 (15%) | 3 (15%) | 6 (15%) |
| Weight (kg) | 104.6 (89.1, 121.8) | 111.6 (86.6, 122.8) | 103.1 (89.6, 120.6) |
| BMI, kg/m2 | 35.3 (30.0, 39.9) | 35.0 (29.0, 41.0) | 35.3 (30.5, 39.0) |
| Self-reported duration of type 2 diabetes | 18.5 (9.2, 25.0) | 16.5 (6.8, 25.2) | 19.0 (12.0, 25.0) |
| Insulin dose (U/day) | |||
| Daily basal insulin | 34.5 (23.5, 45.8) | 32.0 (20.0, 46.2) | 35.5 (24.8, 45.8) |
| Daily prandial insulin | 30.0 (20.0, 39.2) | 30.0 (23.8, 36.0) | 29.0 (20.0, 40.5) |
| Total daily dose | 64.5 (50.8, 80.5) | 62.5 (50.0, 80.5) | 70.0 (52.8, 80.5) |
| Insulin dose (U/kg per day) | |||
| Total basal dose | 0.3 (0.2, 0.4) | 0.3 (0.2, 0.4) | 0.3 (0.2, 0.4) |
| Total prandial dose | 0.3 (0.2, 0.4) | 0.3 (0.2, 0.4) | 0.3 (0.2, 0.4) |
| Total insulin dose | 0.6 (0.5, 0.8) | 0.6 (0.5, 0.8) | 0.6 (0.5, 0.9) |
| Blood lab sample | |||
| HbA1c (%) | 6.8 (6.4, 7.2) | 6.9 (6.5, 7.3) | 6.8 (6.2, 7.2) |
| HbA1c (mmol/mol) | 51 (46, 55) | 52 (48, 56) | 51 (44, 55) |
| Creatinine (mg/dL) | 1.1 (0.9, 1.5) | 1.1 (0.9, 1.4) | 1.1 (0.9, 1.5) |
| eGFR (mL/min/1.73 m2)a | 66.0 (47.5, 81.6) | 63.5 (49.8, 84.0) | 66.0 (42.8, 81.6) |
- Note: All values are shown as n (%) or median (IQR).
- Abbreviations: eGFR, estimated glomerular filtration rate; MDI, multiple daily insulin injections; MDRD, Modification of Diet in Renal Disease; SGLT2i, sodium glucose cotransporter 2 inhibitor.
- a Calculated using MDRD equation.
At Week 26, 34/40 (85%) participants in the semaglutide group were receiving semaglutide 1.0 mg, 4/40 (10%) were receiving 0.5 mg and 2/40 (5%) had discontinued semaglutide due to adverse events (one acute renal failure; one severe headaches and palpitations).
At Week 26, 36/40 participants (90%) in the semaglutide group and 15/20 (75%) in the MDI group maintained HbA1c ≤7.5% (≤58 mmol/mol) (odds ratio, 2.6; 95% CI 0.7, 11.0; p = 0.18). The model-adjusted estimated mean (95% CI) changes in HbA1c were −0.5% (−0.7, −0.3) (−5 mmol/mol [−7, −3]) in the semaglutide group versus 0% (−0.3, 0.3) (0 mmol/mol [−3, 3]) in the MDI group (between-group difference, −0.5% [−0.9, −0.1] or − 5 mmol/mol [−10, −1]; p = 0.009) (Figure 2A; Table S3).
At Week 26, mean (95% CI) body weight change from baseline was −8.9 kg (−9.9, −7.8) in the semaglutide group and +1.5 kg (−0.1, 3.1) in the MDI group (difference, 10.4 kg [8.5, 12.3]; p < 0.001) (Figure 2B). Mean percentage change in body weight at Week 26 was −8.6% (−9.6, −7.6) in the semaglutide group and 1.4% (0, 2.8) in the MDI group (difference, 10.0% [8.3, 11.8]; p < 0.001) (Figure 2C; Table S3). Weight loss >10% from baseline was achieved in 45% (18/40) of semaglutide group participants (post hoc outcome).
Changes in insulin doses from baseline to Week 26 are presented in Figure 3 and Table S3. The estimated mean (95% CI) change in TDD was −37.3 (−41.1, −33.6) units/day in the semaglutide group (mean decrease of 56%), versus an increase of 5.2 (−0.3, 10.6) units/day in the MDI group (6.7% increase) (p < 0.001). Mean basal insulin dose decreased by 20.2% in the semaglutide group (−30.7, −9.7) and increased by 16.2% (0.7, 31.7) in the MDI group (p < 0.001). Mean prandial insulin dose decreased by 96.2% (−100.0, −89.5) in the semaglutide group and increased by 7.5% (−2.5, 17.4) in the MDI group. In 58% (23/40) of the semaglutide group, insulin TDD was reduced by >50% (post hoc outcome). In the semaglutide group, 39/40 (97.5%) participants replaced all prandial insulin throughout the study; one participant resumed MDI because of a severe adverse event (acute kidney injury). Cessation of all insulin, both basal and prandial, occurred in 4 out of 40 participants (10%) randomised to semaglutide.
At Week 26, mean (IQR) DTSQs scores were numerically higher in the semaglutide group (35.0 [32.0, 36.0]) versus the MDI group (32.5 [27.8, 35.5]). DTSQc scores trended towards better overall treatment satisfaction and more favourable hyperglycaemia and hypoglycaemia scores in the semaglutide group versus the MDI group (Table S4).
Mean estimated rates of total, Level 1 and Level 2 hypoglycaemic events were statistically similar (p = 0.24) between groups (Tables 2 and S5). There were only two Level 3 hypoglycaemic events, both of which occurred in the MDI group at the Week 12 visit.
| MDI (n = 20) | Semaglutide (n = 40) | |
|---|---|---|
| Hypoglycaemic episodes, estimated mean (95% CI) | ||
| Any hypoglycaemic episode | 1.4 (0.5, 3.9) | 0.7 (0.3, 1.5) |
| Rate ratio | 2.1 (0.6, 7.5) | |
| Hypoglycaemic episodes by severitya | ||
| Level 1 | 1.0 (0.4, 2.8) | 0.6 (0.3, 1.5) |
| Rate ratio | 1.6 (0.4, 5.6) | |
| Level 2 | 0.1 (0, 0.5) | 0.0 (0, 0) |
| Rate ratio | NE | |
| Level 3 | 0.0 (0, 0) | 0.0 (0, 0) |
| Rate ratio | NE | |
| Adverse events, by subject, n (%) | ||
| Any adverse event | 8 (40) | 16 (40) |
| Serious adverse event | 4 (20) | 9 (22) |
| Adverse event leading to drug discontinuation | 1 (5) | 2 (5) |
| Adverse event leading to a dose decrease | 1 (5) | 4 (10) |
| Adverse events, by event, n (%) | ||
| All adverse events | 10 | 29 |
| Serious adverse event | 5 | 12 |
| Adverse event leading to drug discontinuation | 1 | 2 |
| Adverse event leading to a dose decrease | 2 | 4 |
- Note: Statistical testing methods: Any hypoglycaemic episode and L1 hypoglycaemic episodes (GLMM with negative binomial regression); L2 hypoglycaemic episodes (GLMM with Poisson regression); L3 hypoglycaemic episodes (Exact Poisson test); adverse event rates by subject and by event (Pearson's Chi-squared test; Fisher's exact test). All commonly observed study times were used and interaction between time and randomisation group was included. None of the statistical comparisons were significant at the level of p < 0.05. See Table S5 for hypoglycaemic event data detailed by each follow-up time point.
- Abbreviations: GLMM, generalised linear mixed-effect model; L, level; NE, not estimable.
- a Regardless of symptoms, L1 was classed as glucose value <70 mg/dL (3.9 mmol/L) and ≥54 mg/dL (3 mmol/L); L2 as glucose value <54 mg/dL (3.0 mmol/L). A severe event (L3) was defined by altered mental status/and or physical status requiring assistance for treatment of hypoglycaemia.
Adverse events (excluding hypoglycaemia) were reported by 40% of participants in each group (Table 2). There were no significant between-group differences for adverse events overall, serious adverse events or adverse events leading to drug discontinuation or dose decrease.
4 DISCUSSION
This open-label, randomised, controlled trial demonstrated that in persons with reasonably well-controlled type 2 diabetes (HbA1c ≤7.5% [≤58 mmol/mol]) receiving MDI ≤120 units/day, replacing prandial insulin with once-weekly subcutaneous semaglutide was not only safe and effective, but achievable using a simple one-step approach: completely stopping prandial insulin upon semaglutide initiation, rather than gradually decreasing then stopping prandial insulin. Importantly, the transition from MDI to semaglutide plus basal insulin in TRANSITION-T2D was achieved with a high degree of safety. Unexpectedly, the estimated rate of hypoglycaemia, though numerically lower in the semaglutide group, was not statistically significantly different between groups. This finding might be attributed, at least in part, to semaglutide-induced appetite suppression and reduced food intake resulting in some unexpected hypoglycaemia events in semaglutide-treated participants, as well as a lack of proactive modulation of basal insulin dosing in the semaglutide group. Per protocol, self-adjustment of basal insulin dosing in the semaglutide group was permitted starting at Week 12 of the study; it is possible that a statistically significant difference in hypoglycaemia event rates between treatment groups may have been observed had semaglutide participants been allowed to adjust basal insulin doses earlier in the study.
Overall, the effect of semaglutide on insulin requirements was noticeable, enabling full replacement of prandial insulin in all participants who remained on semaglutide throughout the study, as well as significant reductions in basal insulin dosage. Perhaps the most remarkable observation was the gradual and marked decrease in body weight in the semaglutide group; this reached a mean of −8.9 kg (8.6% decrease) at 26 weeks, with no apparent plateau. Additionally, 45% of the semaglutide group achieved >10% weight loss from baseline to Week 26. Measures of participant satisfaction with treatment trended more favourably in the semaglutide group, which was not unexpected given the greatly reduced injection and glucose vigilance burdens when switching from multiple daily prandial insulin doses to once-weekly semaglutide.
Our data suggest that the transition could be easily implemented in real-world practice, given the lack of deterioration in HbA1c levels during the titration phase in the semaglutide group, and may entail less intensive healthcare follow-up than an MDI regimen. The range of demographic and comorbidity distributions in the TRANSITION-T2D study sample seemed generally representative of the typical real-world population receiving MDI, including participants of various ages and those with long-standing type 2 diabetes, thus suggesting broad clinical applicability of the strategy. Besides the inherent limitations of an open-label study design, another study limitation is that individuals with baseline insulin TDD of 80–120 units/day were underrepresented (25% of total population); further studies are required to confirm the generalisability of our results to such persons. Given the study's 26-week duration, the long-term durability of response and potential need for reintroduction of prandial insulin could not be determined and requires further study. Though the study population was not large, the sample size of 60 allowed for estimation with adequate precision of control rates within and between groups; increasing the sample size would not have provided clinically significant improvements in precision. Due to funding constraints, the study could not provide personal continuous glucose monitors (CGM) for all participants; although pre-existing CGM use was allowed, fingerstick glucose readings were required of all participants as the basis for guiding insulin dose adjustments during the study. Incorporating CGM use could have provided more comprehensive blood glucose data, including time in range and below range. Yet, at the time the study was conducted, HbA1c was the accepted glycaemic target for type 2 diabetes and was chosen for the primary study end-point. Random treatment assignment and demographic diversity of the sample are strengths of the study.
To our knowledge, there have been only two prior randomised, controlled studies that have investigated replacement of prandial insulin with a GLP-1RA or SGLT2 inhibitor,10, 11 using different medications and/or more complex transition protocols. In a 26-week trial evaluating the impact of replacing prandial insulin with once-weekly subcutaneous albiglutide,11 complete cessation of prandial insulin occurred in only 54% of participants (mean baseline HbA1c, 7.7% [61 mmol/mol]). In contrast, 97.5% of TRANSITION-T2D participants discontinued prandial insulin therapy, a difference possibly related to a greater effect of semaglutide on HbA1c lowering and weight loss. Indeed, the albiglutide study investigators speculated that weekly GLP-1RAs other than albiglutide, including semaglutide, might demonstrate greater efficacy as part of a strategy to replace prandial insulin. Furthermore, in the albiglutide trial, discontinuation of prandial insulin was done through a more complex protocol involving a glycaemic optimisation phase, followed by a prandial insulin down-titration phase before discontinuation, all of which would be challenging in real-world clinical practice. In addition, TRANSITION-T2D participants were already reasonably controlled at study entry, which allowed for a one-step transition in therapy, rather than a run-in period of insulin optimisation as in the albiglutide study.
In the 6-month BEYOND study10 (which included participants with higher HbA1c and lower BMI and TDD of insulin compared with TRANSITION-T2D), prandial insulin was stopped upon initiation of IDegLira or IGlarLixi. However, despite a high mean HbA1c level at baseline (8.5% [69 mmol/mol]), the change at 3 months was modest (−0.6% [−6.6 mmol/mol]) and clinically insufficient, with no further decrease at 6 months. Further, 12 (12%) participants switched to IDegLira or IGlarLixi discontinued before 6 months due to lack of efficacy (8%) or adverse events (4%).
The magnitude of mean weight loss in TRANSITION-T2D was considerably higher than previously reported with semaglutide 1.0 mg at comparable time points in the SUSTAIN program (range, −4.5 to −6.4 kg14-20) and the STEP-2 study (~6% decrease),21 and also higher than noted in non-semaglutide trials using similar strategies to TRANSITION-T2D,10-12 including BEYOND (basal insulin plus GLP-1RA group, −1.9 kg10) and the albiglutide study (albiglutide plus basal insulin group, −2.0 kg11). The combination of semaglutide's weight-reduction properties, along with significantly lower insulin exposure (an agent with known adverse effects on body weight), likely placed study participants in a ‘virtuous cycle’: the more they lost weight, the less insulin they needed, so they experienced further weight loss.
These findings have highly practical implications for real-world management of individuals with type 2 diabetes who have achieved good glycaemic control on MDI regimens. Replacing prandial insulin with a long-acting weekly GLP1-RA like semaglutide can safely optimise blood glucose control with less intensive follow-up than would be required for ongoing titration of complex insulin regimens. Full cessation of prandial insulin at the time of semaglutide initiation not only simplifies therapy, but also maximises semaglutide's effect, as exogenous insulin leads to a feedback inhibition of endogenous insulin secretion.13 Most individuals with type 2 diabetes retain some degree of insulin-secretory capacity; by withdrawing exogenous insulin and adding an agent that can restore beta-cell function,22, 23 along with other key benefits (such as reduced glucagon secretion, delayed gastric emptying, decreased food intake and consequently reduced prandial insulin requirements), exogenous prandial insulin was fully replaced without major hyperglycaemic events. The few events of mild hyperglycaemia in the semaglutide group during the titration phase were controlled with low-dose correction doses of rapid-acting insulin.
The clinical characteristics of the TRANSITION-T2D participants could be presumed to be representative of potentially good candidates for transitioning from prandial insulin to a GLP-1RA as described in this study, including HbA1c ≤7.5% (≤58 mmol/mol) on MDI (≥3 daily injections of insulin analogues, including ≥2 injections of prandial insulin) up to a TDD of 120 units and ≤0.5 units/kg of prandial insulin, no GAD-65 antibodies and no end-stage kidney disease. It should be noted that the study intentionally screened out individuals suspected of having autoimmune diabetes and extra caution is advisable if attempting this approach in persons with a strong family history of autoimmune diabetes, prior autoimmune antibody positivity, poorly controlled type 2 diabetes despite intensive insulin therapy, or leaner BMI. Determination of GAD-65 antibodies would be prudent before considering this approach if there is any suspicion of possible type 1 diabetes. Lastly, in specific cases, a random assessment of C-peptide and glucose to assess residual insulin-producing capacity may be helpful. It is also important to highlight that implementing this intervention in real-world practice may increase the risk of euglycemic diabetic ketoacidosis, particularly in patients using background therapy with SGLT2 inhibitors during metabolically stressful events. Therefore, it is crucial to take the necessary precautions by educating patients on recognising symptoms, identifying warning signs and measuring ketones when such events occur.
Further investigation of this de-intensification strategy would be of value to explore feasibility in patient types beyond those included in our study. TRANSITION-T2D participants had a baseline HbA1c ≤7.5% (≤58 mmol/mol), whereas the albiglutide and BEYOND studies enrolled participants with higher baseline HbA1c; perhaps transition from MDI to basal insulin plus weekly semaglutide is more successful after reasonable glycaemic control is obtained, since removing the effects of glucotoxicity may optimise the effectiveness of GLP-1RAs. Nevertheless, given the outstanding results obtained from the present trial in terms of the percentage of participants fully replacing prandial insulin and achieving further improvement in their HbA1c, testing this approach in persons on higher insulin doses (>120 units/day), and/or with HbA1c levels >7.5% (>58 mmol/mol), is reasonable not only to establish efficacy and safety in such individuals, but also to determine whether baseline glycaemic status or TDD influences transition outcomes or protocols. This is an especially appealing research area now that a higher dose of subcutaneous semaglutide (2.0 mg) has been demonstrated to provide additional reductions in HbA1c and weight loss in persons with type 2 diabetes.24 Lastly, in the analysis performed, we approached it as a superiority trial. We compared the groups using standard tests, without any prespecified ranges for non-inferiority. Future research might be better able to conclusively demonstrate non-inferiority of control rates, using these results as a guide to develop an appropriate sample size to study this question.
In summary, TRANSITION-T2D is the first randomised trial, to our knowledge, to demonstrate safe and effective de-intensification of MDI in persons with well-controlled type 2 diabetes, where nearly 100% of study participants initiating semaglutide were able to stop prandial insulin. Of equal importance, the approach we used is a simple one, making it clinically accessible for both clinicians and patients in real-world settings. MDI no longer needs to be accepted as the final therapy destination for people with type 2 diabetes. An ostensibly simple therapy transition (simultaneously stopping prandial insulin and starting subcutaneous semaglutide) not only reduces treatment and management burdens (fewer weekly injections, plus an expected reduction in glucose monitoring frequency/intensity), but also facilitates weight loss and, presumably, other weight loss–related benefits. Moreover, in persons with type 2 diabetes and established cardiovascular disease, there are further benefits of replacing prandial insulin with subcutaneous semaglutide,17 which is FDA approved to reduce major cardiovascular event risk. This approach has the potential to revolutionise management of persons with reasonably well-controlled type 2 diabetes who, until now, have been assumed to require lifelong MDI, especially those with obesity.
AUTHOR CONTRIBUTIONS
PR, BB and KMP conceived and designed the TRANSITION-T2D study. PR, NB, JB and HX collected, analysed and interpreted the data. KJ provided administrative support, and assisted with organising and conducting the trial; DI, MLG and KZ helped manage study participants. All authors had full access to the data, contributed to the drafting and review of the manuscript and approved the final article for publication.
ACKNOWLEDGEMENTS
The authors express their sincere appreciation to the individuals who participated in this study and made the trial possible. Medical writing and editorial support were provided by Sandra Westra, PharmD, and Alison Terry, both funded by Cleveland Clinic.
FUNDING INFORMATION
This study was supported by a research grant from Novo Nordisk Inc. Investigator Sponsored Studies Program; TRANSITION-T2D ClinicalTrials.gov number, NCT04538352. Novo Nordisk had no influence on or involvement in trial design or conduct, data collection, analysis or manuscript development. The opinions in this article are those of the authors and do not necessarily represent the opinion of Novo Nordisk.
CONFLICT OF INTEREST STATEMENT
DI reported receiving speakers' honoraria from Novo Nordisk and Lilly. MLG reported receiving speakers' honoraria from Asofarma. BB reported receiving research support from Novo Nordisk. KMP reported receiving consulting fees from AstraZeneca, Bayer, Corcept Therapeutics, Diasome, Eli Lilly, Merck, Novo Nordisk, Sanofi and Twin Health; speakers' honoraria from AstraZeneca, Corcept Therapeutics, Merck and Novo Nordisk; research support (to Cleveland Clinic) from Bayer, Merck, Novo Nordisk and Twin Health; and a patent application, Identifying Patients for Intensive Hyperglycemia Management (US Provisional Patent Application No. 62/982195). No other disclosures were reported.
Open Research
PEER REVIEW
The peer review history for this article is available at https://www.webofscience.com/api/gateway/wos/peer-review/10.1111/dom.16057.
DATA AVAILABILITY STATEMENT
Individual participant data are not available for public access. Requests for deidentified patient-level data may be considered pending DUA and IRB approval.
