The development of glucagon-like peptide-1 receptor agonists (GLP-1RAs), such as exenatide, liraglutide, dulaglutide, lixisenatide, and semaglutide, led to major significant advances in type 2 diabetes (T2DM) management. GLP-1RAs improve insulin secretion in a glucose-dependent manner, slow gastric emptying, increase satiety, increase glucose uptake by the muscles, decrease gluconeogenesis in the liver, and reduce postprandial glucagon secretion.¹ These agents offer many advantages given their significant reduction of hemoglobin A1c (A1C)², promotion of weight loss³, reduced risk of hypoglycemia, and reduced major adverse cardiovascular outcomes^4–6 and reduced progression of nephropathy.⁷

The American Diabetes Association recommends initiating GLP-1RAs in patients with T2DM after implementing lifestyle modifications and metformin therapy if promoting weight loss, minimizing weight gain, or decreasing hypoglycemia is desired.⁸ GLP-1RAs with proven cardiovascular disease benefits (dulaglutide, liraglutide, or semaglutide) are recommended in patients with atherosclerotic cardiovascular disease (ASCVD) or chronic kidney disease without albuminuria. Additionally, GLP-1RAs are the preferred injectable agent over insulin for patients on oral therapy needing intensification.⁸ Insulin should be prioritized over a GLP-1RA in patients with A1C >10% or blood glucose levels >300 mg/dL or weight loss AND symptoms of hyperglycemia.⁸

For patients on basal insulin, GLP-1RAs are the preferred addition over prandial insulin due to clearer dose titration and patient education and as well as a reduced risk of hypoglycemia and fluctuations in blood glucose.^8,9 Combination products like Xultophy™ (insulin degludec/liraglutide) and Soliqua™ (insulin glargine/lixisenatide) offer the advantage of reduced administrations by combining basal insulin and a GLP-1RA in a single injection.¹⁰

Due to their extended time on the receptor, longer-acting GLP-1RAs have a greater effect on FBG and PPG compared to shorter-acting agents, which have targeted effects on PPG (Table 1).⁹ Some clinical trials have described anticipatory reductions in basal insulin dose by 10-20% when initiating a GLP-1RA in more adequately controlled patients (A1C <8%). These adjustments are based on clinical knowledge and are not consistent or well-studied.⁹ Table 2 outlines clinical trials and recommendations for insulin adjustments with GLP-1RAs. However, best practices are needed for insulin adjustments when initiating a GLP-1RA.

In 2015, researchers at the Oxford Centre for Diabetes Endocrinology and Metabolism released recommendations for a 10% reduction in basal insulin and 30-40% reduction in prandial insulin upon addition of a GLP-1RA.11 Few studies specifically evaluating insulin dose adjustments in GLP-1RA initiation have been published since this recommendation. A randomized, double-blind, placebo-controlled trial in Sweden evaluated the addition of once-daily liraglutide to patients inadequately controlled on basal-bolus insulin therapy.12 Because patients had suboptimal glycemic control at baseline (A1C ≥7.5%), insulin doses were not preemptively reduced. If FBG or pre-prandial glucose levels were normal or close to normal for two consecutive days, then dose reductions were considered. After 24 weeks, reductions were found between liraglutide versus placebo for A1C (-1.13 [95% CI -1.45 to -0.81]), body weight (-3.81 kg [95% CI -4.87 to -2.76]), and total daily insulin dose (-15.8 units [95% CI -23.1 to -8.5]). There was no difference in hypoglycemia between the groups.¹²

A 26-week, randomized, open-label, active-control, multicenter, treat-to-target study compared patients with T2DM with an A1C 7.0 to 9.5% with or without metformin on a basal-bolus insulin regimen (<140 units/day and at least 3 three injections/day) versus the addition of albiglutide and reduction of prandial insulin. All patients were standardized on once-daily insulin glargine and three times-daily insulin lispro before randomization. In the albiglutide and insulin glargine (AIG) group, the insulin lispro dose was reduced by 50% and subsequently discontinued at week 4 for the remainder of the treatment period. After week 8, insulin lispro was reintroduced in patients with average PPG >180 mg/dL. The insulin glargine-lispro (IGL) arm served as the active control, and insulin lispro dose adjustments were made following an algorithm. The AIG group was non-inferior to the IGL group for change in A1C from baseline (0.06% [95% CI -0.05 to 0.17]). The proportion of patients achieving A1C <7% was similar between groups (OR 1.0 [95% CI 0.7-1.3]). In the AIG group, 72% of patients either did not require insulin lispro reintroduction or decreased the insulin lispro without increasing A1C, and 54% completely replaced insulin lispro with albiglutide at the study conclusion. There were no differences in baseline characteristics between the patients in the AIG group who did or did not require reintroduction of insulin lispro. 

Reductions were found between AIG versus IGL for average weekly injections (-16.0 ± 7.9), body weight (-4.4 kg [95% CI -4.9 to -3.8]), total daily insulin dose (-61.8 units [95% CI -65.9 to -57.8]), and hypoglycemic rate (OR 0.43 [95% CI .031-0.60]).  Of note, albiglutide was removed from the U.S. market in 2017 for economic reasons unrelated to safety or efficacy.^13,14 The study showed a similar mean insulin glargine dose between groups with a lower FBG in the AIG group, emphasizing the impact of long-acting GLP-1RAs on both FBG and PPG.¹⁵ Therefore, basal insulin dose reductions may be more appropriate with long-acting agents versus short-acting agents. Table 1 outlines recommendations for short- and long-acting GLP-1RAs initiation with various insulin regimens.

In diabetes management, a patient-centered approach and strong clinical judgment are imperative. For example, in patients at higher risk of hypoglycemia, (longer duration of diabetes, concomitant secretagogues, erratic eating patterns, and kidney disease), more substantial insulin dose reductions may be appropriate.16 Consider the patient case outlined in Figure 1. GLP-1RA initiation is appropriate in this patient, given their elevated A1C, obesity, and ASCVD. Because the A1C is relatively close to the patient’s goal, a 10-20% reduction in basal insulin dose and a 30-50% reduction in prandial insulin dose is appropriate upon starting a long-acting GLP-1RA. If the patient were less adequately controlled, less aggressive insulin dose adjustments could be considered, if at all. Close monitoring of FBG, PPG, and signs/symptoms of hypo- and hyperglycemia is warranted in patients on insulin starting any GLP-1RA.