Our study showed a 30% reduction in total insulin dose within 2 months of tirzepatide treatment in adults with T1D. Both basal and bolus insulin requirements were reduced, however, bolus insulin requirement was reduced relatively more than basal insulin leading to an increased basal/total daily insulin ratio. Besides dose and weight reductions, total daily insulin units/kg also decreased by 25% at 6 months, which may indicate an increase in insulin sensitivity.

Our study provides guidance on insulin dose adjustment when tirzepatide therapy is initiated in adults with T1D. We recommend reducing basal and bolus doses by 20–30% in line with weight loss within the first month of tirzepatide initiation (with 2.5 mg/week starting dose). In the following months, an additional 5–10% reduction compared to baseline may be needed. These reductions can be achieved through different approaches among people using AID, MDI, or CSII. The Control-IQ AID system is a basal-dependent system, which dynamically modulates the basal rates set by the users based on predicted sensor glucose levels, which means direct basal adjustments can help with the insulin reduction needed with tirzepatide. Higher correction factor (which will reduce the insulin delivery for automated boluses) and higher carbohydrate–insulin ratio (which will reduce user-initiated meal insulin boluses) adjustments would be needed to prevent hypoglycemia, especially after meals. Finally, the patient’s weight and total daily insulin settings can be manually adjusted and should be updated to achieve better glycemic management with tirzepatide treatment. Other AID systems such as Omnipod 5, Medtronic 780 G, iLet, and CamAPS FX are basal-independent systems and these systems do not use user-set basal rates for automated insulin delivery [9,10,11]. We suggest adjusting the modifiable settings (carbohydrate–insulin ratio, correction factor, active insulin time or target glucose, and so on) depending on the functionality of the AID systems with intention to reduce insulin delivery by 20–30% during tirzepatide initiation. Multiple daily injection users may be more vulnerable for hypoglycemia than AID users due to lack of basal automation and may require more frequent evaluations and updates in their insulin doses. In these patients, 15% basal reduction and 20–25% bolus reduction in the first months can be implemented with monitoring nighttime and postprandial glucose for fine-tuning the basal and bolus doses, respectively. For non-AID pumps, basal rates can be reduced for all hours, while considering diurnal rhythm and changing basal insulin needs throughout the day [10].

Moreover, insulin reduction may depend on baseline A1c. For example, people with T1D with higher A1c (A1c > 8%) may not require 20–30% reduction in insulin dose while someone with A1c close to 7% indicating more optimal insulin use may require close to 30% reduction in insulin dose. The baseline A1c of our cohort was 7.0% and therefore, we assume that for people with higher A1c (> 8%), up to 15% reduction in total insulin dose should be reasonable starting strategy. Patients with higher A1c (> 9%) probably may not need any dose reduction, especially during the first month of adjunct treatment with GLP-1RA. In this regard, further studies with larger cohorts are needed to provide evidence for insulin titration in patients with higher A1c and suboptimal glycemic management.

This study is the first to report on tirzepatide use in people with T1D using an AID system, Tandem Control-IQ. This study provides early insight into the insulin- and weight-specific patterns of people with T1D initiating tirzepatide while using an AID system and provides some clinical principles to consider. Limitations include small sample size, retrospective study design without controls, data collection from EHR, and variable tirzepatide dose titrations. Actual Control-IQ technology settings were not analyzed in this analysis, so it is possible that the insulin reduction seen in this study were due to both the user/healthcare provider changing user-modifiable insulin parameters, but also due to the Control-IQ technology automation which modulates insulin delivery as needed to maintain glucose levels in range. Moreover, our findings are only applicable with tirzepatide therapy. The various GLP-1RA such as semaglutide and dulaglutide are different in many ways, such as dose initiation, dose titration, and weight loss efficacy over time and therefore, insulin titration strategies may be different with different GLP-1RA therapies. Future studies with larger sample sizes can provide more precise insulin titration for tirzepatide use in T1D, with additional insight into AID contributions for safe insulin titration. Changes in insulin sensitivity should also be analyzed with larger cohorts.