Insulin dose adjustments with add-on glucagon-like peptide-1 receptor (GLP-1R) agonists in clinical practice

Abstract

Glucagon-like peptide-1 receptor agonists (GLP-1RAs) are gaining ground as therapeutic modalities in combination with insulin in patients with type 2 diabetes mellitus. Exploiting the multiple benefits of incretin-based therapies in certain patient populations, especially in those who would benefit most from potential weight loss or prevention of body weight gain, has provided a valuable add-on option in diabetes management. However, caution needs to be exercised when initiating such a double injectable therapy, as evidence indicates that, in most instances, the insulin dose needs to be re-adjusted. The majority of published studies suggest reduction of insulin dose, especially related to the ‘bolus’ component; however, some have also recommended that insulin dose should actually be increased, but we found no credible evidence to support the latter. An important determinant of the titration process is the insulin formulation already in use at baseline. As more potent and long-acting GLP-1RAs are introduced, optimal insulin dose scaling is a major challenge, especially in a primary setting. We provide an overview of the current knowledge in this rapidly changing field. Based on currently reported evidence, a reduction of basal insulin by 10% and a decrease of prandial insulin by 30 – 40% is recommended on addition of GLP-1RAs.

Keywords:

• add-on
• agonists
• diabetes
• incretin
• insulin
• receptor

1. Introduction

Glucagon-like peptide-1 receptor agonists (GLP-1RAs) are among the most potent and rapidly expanding therapeutic approaches recently introduced in type 2 diabetes mellitus (T2DM). In general, reduction in glycosylated haemoglobin (HbA_1c), (1.2%, on average) is accompanied by body weight loss of up to 7.3 kg [1-5]. However, individual responses are highly variable. When patients concomitantly receive insulin and a GLP-1RA, it has been reportedly shown that – on average – 67% experience reduction in both HbA_1c and body weight, 14% show reduction in body weight alone, 12% show reduction in HbA_1c alone, whereas in 7% of this patient population, both body weight and HbA_1c were increased [1].

GLP-1RAs are licensed for use in combination with oral anti-diabetic agents or with basal insulin therapy. A number of ‘real world’ studies have described off-licence uses of GLP-1RAs in combination with insulin formulations other than basal insulin. When starting a GLP-1RA, a concomitant decrease in insulin dose has been documented in almost every trial. However, the reduction in insulin dose varies widely and ranges from 0 to 65% from baseline. There is little consensus in current literature regarding the degree to which the insulin dose should be reduced – if at all – when initiating GLP-1RA therapy.

We sought to review the published evidence on insulin dose re-adjustment following initiation of add-on GLP-1RA therapies. Source articles were identified in PubMed Central (including MEDLINE); EMBASE; Cochrane Central Register of Controlled Trials (CENTRAL) via Wiley Interscience; and Cumulative Index to Nursing and Allied Health Literature (CINAHL), up to 1 May 2014.

Interestingly, the reduction in insulin dose is dependent on the insulin regime used, for example, basal insulin only, basal-bolus, premixed insulin and/or co-administration of oral antidiabetic agents (Tables 1 and 2).

Table 1. Changes in insulin dose depending on baseline insulin regimen.

	Reference	Dose (U) at baseline		Maintenance dose (U)		Δ (%) to baseline (U)
Insulin scheme		Basal	Prandial	Basal	Prandial	Basal	Prandial
Basal	Arnolds et al. [23]	40.3		41.1		0.8 (+2%)
Basal/ bolus	N/A
OAD + Basal	Buse et al. [7]	49.5		62.5		13 (+26%)
	Levin et al. [6]	0.37 U/Kg	0.32 U/kg	0.47 U/kg	0.6 U/kg	0.1 U/kg (+27%)	0.6 U/kg (+88%)
	Phillips et al. [25]	105		97		8 (-8%)
	Seino et al. [17]	24.9		23.5		1.4 (-6%)
	Riddle et al. (Get Goal L) [15]	54		50		4 (-7%)
	Riddle et al. (Get Goal Duo 1) [16]	43.4		46.7		3.3 (+8%)
OAD + Basal/ bolus	Viswanathan et al. [24]	58.4	50.4	53.1	36.6	5.3 (-9%)	13.8 (-27%)
Pre-mixed	Viswanathan et al. [24]	72.9		28.3		44.6 (-61%)
Various insulin schemes ± OAD	Yoon et al. [2]	62.9	29.4	58.4	13	4.5 (-7%)	16.4 (-56%)
	Sheffield et al. [4]	48	26	49	17	1 (+2%)	9 (-35%)
	Lind et al. [13]	109.5		62.5		47 (-42%)
	Thong et al. [1]	120		78		42 (-35%)
	Nayak et al. [3]	144		55		89 (-62%)
	Rachabattula et al. [26]	135		114		21 (-16%)
	Anholm et al. [11]	69		41		28 (-40%)
AVERAGE		75.8	35.3	57.3	22.2	-20.8 (17%)	-13.1 (39.3%)

OAD: Oral antidiabetic drugs.

Table 2. Glycaemic control with different GLP-1-RAs added-on insulin.

GLP-1R agonist		Baseline values		Outcome values		Δ (% to baseline)
		HbA1c(%)	FPG (mmol/l)	HbA1c (%)	FPG (mmol/l)	HbA1c (%)	FPG (mmol/l)
Exenatide b.i.d.	Arnolds et al. [23]	8.1	5.6	6.3	4.3	1.8 (-22%)	1.3 (-23%)
	Riddle et al. [16,31]	7.8		6.5		1.3 (-17%)
	Buse et al. [7]	8.3	8.1	6.6	6.5	1.7 (-20%)	1.6 (-20%)
	Thong et al. [1]	9.5		9		0.5 (-5%)
	Pawasakar et al. [27]	8.5		8.1		0.4 (-5%)
	Levin et al. [6]	8.8		7.6		1.2 (-14%)
	Sheffield et al. [4]	8.4		7.5		0.9 (-11%)
	Yoon et al. [2]	8.1		7.5		0.6 (-7%)
	Nayak et al. [3]	8.8		8.6		0.2 (-2%)
	Viswanathan et al. [24]	7.7		7.1		0.6 (-8%)
	Christofides et al. [28]	8.1		7.3		0.8 (-10%)
	Phillips et al. [25]	8.2		6.8		1.4 (-17%)
	Houser et al. [29]	8.1		6.9		1.2 (-15%)
	Vithian et al. [30]	8.9		8.2		0.7 (-8%)
	Rachabattula et al. [26]	9.4		8.1		1.3 (-14%)
Exenatide LAR	N/A
Liraglutide	Anholm et al. [11]	8.6		7.8		0.8 (-9%)
Lixisenatide	Riddle et al. [14](Get Goal-L) [10]	8.4	8.1	7.8	8	0.6 (-7%)	0.1 (-1%)

b.i.d.: Twice daily; FPG: Fasting plasma glucose; HbA1c: Glycosylated haemoglobin; LAR: Long-acting release.

2. Insulin dose adjustments per GLP-1RA therapy

2.1 Exenatide and insulin

Exenatide was the first GLP-1RA to be licensed for use in DM in Europe and the US and is licensed as an add-on therapy for patients already taking insulin. Hence, we mostly gain insight into the subject of insulin dose re-adjustments by gauging the wide clinical experience of its use in this setting; however, all published trials to date are based on twice daily (b.i.d.) exenatide, with no trials to date combining the new variant of long-acting exenatide (LAR) with insulin.

The largest observational study of exenatide–insulin combination was conducted by the Association of British Clinical Diabetologists and started in December 2008, as part of the national UK exenatide audit [1]. Data from 4857 patients started on exenatide were evaluated at baseline and follow-up (median 26 weeks). Of these, 1257 added exenatide to existing insulin therapy, 664 patients added insulin to existing exenatide therapy and 2936 patients used exenatide without insulin. Adding exenatide to patients receiving insulin resulted in reductions in HbA_1c and body weight (0.51% and 5.8 kg, respectively). One in six patients discontinued insulin while on exenatide treatment, and average insulin dose reduced from 1.0 ± 0.8 to 0.7 ± 0.2 U/kg/day (p < 0.001). The observational study, by Yoon et al. [2], examined patients on oral antidiabetic agents (metformin, sulfonylureas or thiazolidinediones) and a variety of insulin regimes. Notably, ∼ 70% of patients were using a basal-bolus insulin regime, ∼ 30% basal insulin and 0.1% prandial insulin alone. The introduction of exenatide resulted in a reduction in basal insulin by 7% (∼ 4.5 U), (not statistically significant) and a concomitant significant reduction in prandial insulin by 56% (∼ 16.4 U; p < 0.001).

Nayak et al. [3] evaluated 160 patients on insulin therapy, 76% injecting twice daily and 23% on > 2 injections/day at baseline. After commencing exenatide, they observed a reduction in insulin dose of 62% (∼ 89 U) after 12 months. Another benefit was weight loss of 10.2 kg at 6 months and 12.7 kg at 12 months of follow-up; however, there was non-significant change in HbA_1c. Similarly, Sheffield et al. [4] reported significant mean HbA_1c reduction by 0.87%, weight loss of 5.2 kg and a non-significant 35% decrease in prandial insulin.

Interestingly, insulin dose reduction has been associated with a higher baseline Body Mass Index (BMI). Rosenstock et al. [5] observed an insulin dose reduction compared with placebo only in those patients with a BMI 30 – 36 kg/m² (average reduction: LS mean difference, −9.2 and −12.2 units, respectively; p < 0.05), but not in those with BMI < 30 kg/m². However, HbA_1c reduction in the exenatide group was also influenced by initial BMI (< 30 kg/m²: -0.62%, 30 – 36 kg/m²: -0.85, > 36 kg/m²: -0.39%), albeit in an inverse relationship compared to that seen with insulin therapy between initial BMI and HbA_1c. Regarding weight loss, it was reported to be approximately 3 kg versus no change or even increase in weight in the placebo group (depending on initial HbA_1c).

Not all studies have reported a reduction in insulin dose after commencing exenatide. Levin et al. [6] reported a 27% increase in insulin dose in patients with basal insulin and oral antidiabetic drugs (OADs) that were started on exenatide. However, it is worth noting that this finding was in the context of stricter follow-up, with a tight insulin titration. They reported a concomitant HbA_1c decrease of 1.2%. Buse et al. [7] also reported a 27% increase in total insulin dose in patients treated with insulin glargine and OADs after commencing exenatide. Nevertheless, the insulin dose was augmented much more in the control group (∼ 13U; 42%). HbA_1c decreased by 1.74% with exenatide versus 1.04% with placebo, and weight decreased by 1.8 kg versus 1.0 kg increase, respectively.

2.2 Liraglutide and insulin

As mentioned above, most studies based on insulin combined with GLP-1 RA are based on the use of exenatide. There is some data on the addition of insulin to liraglutide, but little has been published about starting liraglutide in patients already taking insulin [8-10]. This is reflected in the licensing of liraglutide which is not licensed as an add-on therapy to insulin, but insulin detemir is licensed as an add-on therapy to liraglutide.

At the 2011 American Diabetes Association Congress, Anholm et al. [11] reported a 40% decrease in basal insulin dose in patients started on liraglutide, accompanied by a significant reduction in HbA_1c of 0.8% and mean weight loss of 5.1 kg. The European Association for the Study of Diabetes (EASD) liraglutide audit found that in clinical practice [12], many patients already receiving insulin were started on liraglutide as an off-licence treatment. When compared with the EASD exenatide audit, these patients achieved greater HbA_1c reduction, but less weight reduction, probably because there was less insulin and thiazolidinedione discontinuation in the liraglutide group. The liraglutide audit took place after the exenatide audit, and it is thought that this difference reflects lessons learned with exenatide regarding over-zealous reduction of DM treatment prior to GLP-1RA initiation.

Only one other study has evaluated the effect of liraglutide on insulin dose, but these data were reported in combination with data from patients who commenced exenatide; therefore, results cannot be assumed to be representative of the effect of liraglutide only. The observational study, by Lind et al. [13] assessed the addition of GLP-1RAs to insulin; 66% of patients were treated with liraglutide and the remaining 34% started exenatide at baseline, 52% of patients were on multiple daily injections, 34% on basal insulin, 12% on pre-mixed insulin and almost 70% of patients were on metformin. By the end of the study, total insulin dose had decreased by 42%, HbA_1c was significantly reduced by 1% and mean weight loss was 7.1 kg.

2.3 Lixisenatide and insulin

Lixisenatide is one of the newest GLP-1RAs in clinical practice [14]. Phase III program clinical trials have included several studies with insulin, among which GetGoal-L and GetGoal-Duo-1 evaluated changes in insulin dosage [15-17].

GetGoal-L [15] was a double-blind, parallel-group, placebo-controlled trial; 495 patients with established basal insulin therapy were randomized to add-on lixisenatide 20 mg once daily or placebo, for 24 weeks. Basal insulin dosage was unchanged except to limit hypoglycaemia. Even so, a decrease in insulin dose at study end was seen with lixisenatide (-10%) compared with placebo (-3%). HbA_1c reduction was also significantly greater in the lixisenatide group (0.6 vs 0.3%), as well as 2-h post-prandial glucose (5.5 vs 1.7 mmol/l) and weight reduction (-1.8 vs -0.5 kg).

The results of GetGoal-Duo1 [16] were slightly different. This double-blind, parallel-group trial enrolled patients with an HbA_1c between 7 and 10%, despite oral therapy. Insulin glargine was added and systematically titrated during a 12-week run-in, after which candidates were randomized to lixisenatide 20 mg once daily or placebo for 24 weeks while insulin titration continued. HbA_1c improved by 0.71% in the lixisenatide group and 0.40% in the placebo group. Insulin dose actually increased by 7% in the lixisenatide group, however, doses increased by a greater amount in the placebo group (12%; p < 0.01).

3. Expert opinion

Results from published studies on insulin dose adjustment after initiation of GLP-1RAs show significant heterogeneity [18-23]. However, the most striking trend appears to be a reduction of up to 75 units (17% average) in basal insulin doses, and of up to 35 units (∼ 39% average) in prandial insulin. This finding is not surprising given the mechanism of action of GLP-1RAs on modulation of the incretin effect during the postprandial state, not just on the residual capacity of the failing beta cell to respond to insulinotropic stimuli, but also via suppression of glucagon action and hepatic glucose output. Based on currently reported evidence, a reduction of basal insulin by 10% and a decrease of prandial insulin by 30 – 40% is recommended on addition of GLP-1RAs.

Further well-controlled studies are needed in order to establish clinical guidance, based on solid, practice-driven evidence, especially with the long-acting forms of incretin-based therapies. Accumulating evidence from use of the newest GLP-1RAs (e.g., liraglutide or lixisenatide) in clinical practice, along with emerging data from large outcome trials with exenatide should help shape our knowledge regarding optimal insulin dosage in clinical practice.

Declaration of interest

The authors have no relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript. This includes employment, consultancies, honoraria, stock ownership or options, expert testimony, grants or patents received or pending, or royalties.