Real‐world outcomes of addition of insulin glargine 300 U/mL (Gla‐300) to glucagon‐like peptide‐1 receptor agonist (GLP‐1 RA) therapy in people with type 2 diabetes: The DELIVER‐G study

Abstract

Aims

To provide real‐world data on the addition of basal insulin (BI) in people with type 2 diabetes mellitus (PWD2) suboptimally controlled with glucagon‐like peptide‐1 receptor agonist (GLP‐1RA) therapy. However, real‐world data on the addition of BI to GLP‐1RA therapy are limited.

Materials and Methods

We used a US electronic medical record data source (IBM® Explorys®) that includes approximately 4 million PWD2 to assess the real‐world impact of adding the second‐generation BI analogue insulin glargine 300 U/mL (Gla‐300) to GLP‐1RA therapy. Insulin‐naïve PWD2 receiving GLP‐1RAs who also received Gla‐300 between March 1, 2015 and September 30, 2019 were identified; participants were required to have data for ≥12 months before, and ≥6 months after, addition of Gla‐300.

Results

The mean (standard deviation [SD]) age of participants (N = 271) was 57.9 (10.8) years. Baseline glycated haemoglobin (HbA1c) was 9.16% and was significantly reduced (−0.97 [SD 1.60]%; P < 0.0001) after addition of Gla‐300; a significant increase in the proportion of PWD2 achieving HbA1c control was observed after addition of Gla‐300 (HbA1c <7.0%: 4.80% vs. 22.14%, P < 0.0001; HbA1c <8.0%: 19.56% vs. 51.29%, P < 0.0001). The incidence of overall (8.49% vs. 9.59%; P = 0.513) and inpatient/emergency department (ED)‐associated hypoglycaemia (0.37% vs. 0.74%; P = 1.000), as well as overall (0.33 vs. 0.46 per person per year [PPPY]; P = 0.170) and inpatient/ED‐associated hypoglycaemia events (0.01 vs. 0.04 PPPY; P = 0.466) were similar before and after addition of Gla‐300.

Conclusions

In US real‐world clinical practice, adding Gla‐300 to GLP‐1RA significantly improved glycaemic control without significantly increasing hypoglycaemia in PWD2. Further research into the effect of adding Gla‐300 to GLP‐1RA therapy is warranted.

1. INTRODUCTION

Type 2 diabetes mellitus (T2DM) is a chronic, progressive disease that affects approximately 422 million adults worldwide. ¹ Guidelines from the American Diabetes Association (ADA) and the European Association for the Study of Diabetes (EASD) suggest that people with T2DM (PWD2) aim to achieve a target of glycated haemoglobin (HbA1c) level of <7%. ² Despite a plethora of available treatments for T2DM, many PWD2 do not achieve this target. ³ , ⁴ More effective use of existing agents through earlier intensification (where appropriate), and targeted support to maintain patient motivation and improve treatment adherence, may offer a potential intervention to improve HbA1c target achievement. ⁵

Glucagon‐like peptide‐1 receptor agonists (GLP‐1RAs) are recommended treatment options at many stages of T2DM because they lower HbA1c, may show favourable effects on weight, and have a low risk of hypoglycaemia. ⁶ , ⁷ Despite clear evidence of the efficacy of GLP‐1RAs, some patients do not experience lasting glycaemic control and require treatment intensification. ⁸ , ⁹ For PWD2 with suboptimal glycaemic control during oral antidiabetic drug or GLP‐1RA therapy, the ADA and EASD recommend that sulphonylureas or insulin therapy should also be added, as required. ⁶ , ⁷ , ¹⁰ Clinical studies have demonstrated improved glycaemic control and low risk of hypoglycaemia and weight gain with the combination of insulin and GLP‐1RAs ¹¹ ; however, data on outcomes are lacking when these therapies are used together in real‐world clinical practice. Additionally, most previous studies combining insulin and GLP‐1RAs have added the latter to the former; data reporting outcomes for patients who received basal insulins (BIs) after experiencing suboptimal glycaemic control during GLP‐1RA therapy are lacking.

This retrospective study of electronic medical records (EMRs) aimed to determine the impact of adding BI analogue insulin glargine 300 U/mL (Gla‐300) to GLP‐1RA therapy in PWD2 receiving treatment intensification.

2. METHODS

2.1. Study design

This was a retrospective analysis of US EMR data sources (IBM® Explorys®) for the period March 1, 2014 and March 31, 2020. IBM Explorys includes longitudinal data from more than 4 million PWD2, representing outpatient, inpatient and post‐acute care settings. Data were standardized using common ontologies such as International Classification of Diseases, 9th/10th Revisions (ICD‐9/10‐CM; for diagnoses and some procedures), current procedural terminology (for procedures), logical observation identifier names and codes (for clinical and laboratory observations) and national drug codes (for prescriptions). First prescription of Gla‐300 during the identification period was defined as the index date (Figure S1). As this study uses deidentified patient records, no informed consent or ethical approval was required. The study was conducted in line with the Declaration of Helsinki.

2.2. Study population

We identified insulin‐naïve adults (aged ≥18 years) with ≥1 diagnosis of T2DM (identified using ICD‐9/‐10 codes listed in Table S1) who were receiving GLP‐1RAs (daily GLP‐1RAs: liraglutide and exenatide; weekly GLP‐1RAs: exenatide extended‐release, albiglutide and dulaglutide) and also Gla‐300 (index event) between March 1, 2015 and September 30, 2019. Patients were also required to have data for ≥12 months before index (“baseline”) and ≥6 months after index (“follow‐up”), and a valid HbA1c value of 3% to 15%, recorded within 6 months pre‐index and in the 3‐ to 9‐month post‐index period. Patients were excluded from study entry if they had type 1 diabetes mellitus identified by a ratio of type 1 to type 2 ICD diagnoses in individual patient records of >0.5 and either a prescription of glucagon with no record of antidiabetic prescription other than metformin, or a drug record of metformin, a diagnosis of polycystic ovary syndrome and no record of antidiabetic prescription other than metformin. Patients were also excluded if they had ≥1 prescription for BI (including fixed‐ratio insulin and GLP‐1RA combinations) on the index date, or a diagnosis of severe acute respiratory syndrome/coronavirus (COVID‐19) after January 20, 2020.

2.3. Study assessments

Study assessments included: baseline patient demographics, characteristics and comorbidities; change in HbA1c between baseline and follow‐up, using the last value recorded within the 3‐ to 9‐month post‐index period; proportion of patients who achieved target HbA1c (<7% or <8%) in the 6‐month follow‐up period; and the incidence and rates of hypoglycaemia with and without emergency department (ED)/inpatient events (identified by ICD‐9/‐10 code, or a blood‐glucose recording ≤70 mg/dL) during the 6‐month baseline and 6‐month follow‐up periods. Incidence and event rates (per person per year [PPPY]) of healthcare resource utilization (HCRU) encounters were also assessed and categorized as inpatient‐, ED‐ or outpatient‐related.

2.4. Statistical analyses

A minimum sample size of 250 patients was required to establish a significant difference in HbA1c reduction. The minimum sample size was calculated using a 95% significance level with 90% power to detect a difference of 0.4 (margin of error) and a standard deviation (SD) of 1.1 in HbA1c reduction, with an estimated dropout rate of 20%. Sensitivity analyses assessed HbA1c reduction and hypoglycaemia for patients receiving weekly (exenatide extended‐release, dulaglutide or albiglutide) versus daily (exenatide or liraglutide) GLP‐1RA therapies, and also for HbA1c outcomes, considered the first HbA1c value in the 3‐ to 9‐month post‐index period, as the follow‐up HbA1c. Study variables were analysed descriptively, with McNemar′s test used to evaluate differences in categorical variables and paired Student′s t‐tests for continuous variables (P < 0.05 was considered significant).

3. RESULTS

3.1. Patient characteristics

A total of 271 patients (mean [SD] age 57.9 [10.8] years) met the eligibility criteria and were included in the study (Table 1). The most common comorbidities recorded in the 12‐month pre‐index period were hypertension, hyperlipidaemia and obesity. Overall, 156 (57.6%) patients received daily GLP‐1RAs and 115 (42.4%) received weekly GLP‐1RAs before treatment intensification with Gla‐300; the mean (SD) number of background oral antidiabetic drugs received by patients was 1.97 (1.04). In the 6‐month pre‐index period, 8.5% of patients had a hypoglycaemic event (event rate: 0.17 PPPY). No significant changes in weight (108.0 vs. 108.4 kg) or body mass index (35.8 vs. 36.1 kg/m²) were seen with addition of Gla‐300 to GLP‐1RAs (baseline vs. follow‐up, respectively).

TABLE 1.

Baseline characteristics (N = 271)

Characteristic	Mean ± SD or n (%)
Age, years (SD)	57.9 (10.8)
Male, n (%)	138 (50.9)
Body mass index, kg/m2 (SD)	35.5 (6.3)
Baseline HbA1c, % (SD)	9.16 (1.51)
Patients with hypoglycaemia events, n (%)	23 (8.5)
Hypoglycaemia event rate, PPPY (SD)	0.17 (0.65)
Mean number of OADs, n (SD)	1.97 (1.04)
Medications received during baseline period, n (%)
Daily GLP‐1RA	156 (57.6)
Weekly GLP‐1RA	115 (42.4)
Any OAD	250 (92.3)
Metformin	171 (63.1)
Sulphonylureas	151 (55.7)
Sodium‐glucose cotransporter‐2 inhibitors	113 (41.7)
Dipeptidyl peptidase‐4 inhibitors	68 (25.1)
Thiazolidinediones	23 (8.5)
Mean (SD) Elixhauser comorbidity index score	3.24 (2.03)
Mean (SD) Charlson comorbidity index score	0.73 (1.27)
Comorbidities/complications (ICD‐9/‐10) in >5% of patients
Hypertension	226 (83.4)
Hyperlipidaemia	230 (84.9)
Obesity	111 (41.0)
Depression	56 (20.7)
Neuropathy	44 (16.2)
Nephropathy	21 (7.7)
Retinopathy	17 (6.3)

Abbreviations: GLP‐1RA, glucagon‐like peptide‐1 receptor agonist; HbA1c, glycated haemoglobin; ICD‐9/‐10, International Classification of Diseases, 9th/10th Revisions; OAD, oral antidiabetic drug; PPPY, per person per year; SD, standard deviation.

3.2. Glycated haemoglobin

Overall, HbA1c was significantly reduced after treatment intensification with addition of Gla‐300 (−0.97 ± 1.6%; P < 0.0001 [Figure 1A]). Significantly more patients also achieved HbA1c control (target of <7% or <8%) during follow‐up after addition of Gla‐300 (Figure 1B). A significant reduction in HbA1c was seen regardless of baseline HbA1c level (≤9% or >9%), but a larger reduction was noted for patients with HbA1c >9% at baseline (Figure 2A). Sensitivity analyses demonstrated that the addition of Gla‐300 to either a daily (n = 156) or a weekly (n = 115) GLP‐1RA resulted in a similar, significant decrease in HbA1c (daily: −0.89 ± 1.62, P < 0.001; weekly: −1.07 ± 1.58, P < 0.001 [Figure 2B]). In addition, HbA1c outcomes were also similar when considering the first HbA1c value in the 3‐ to 9‐month follow‐up period, instead of the last value (−0.90 ± 1.66; P < 0.0001).

FIGURE 1.

Baseline and follow‐up values for (A) glycated haemoglobin (HbA1c) and (B) proportion of patients achieving target HbA1c goals

FIGURE 2.

Change in glycated haemoglobin (HbA1c) from baseline to follow‐up, stratified by (A) baseline glycaemic level and (B) glucagon‐like peptide‐1 receptor agonist (GLP‐1RA) regimen frequency

3.3. Hypoglycaemia

No changes in overall incidence or event rate of hypoglycaemia were seen following addition of Gla‐300 (Figure 3). When analysed by GLP‐1RA regimen, increases in the incidence of hypoglycaemia (from 3.9% at baseline to 8.3% at follow‐up; P = 0.020) and hypoglycaemia event rate (from 0.21 PPPY to 0.44 PPPY; P = 0.017) were observed at follow‐up in patients using daily GLP‐1RAs. No significant differences were seen at follow‐up in patients using weekly GLP‐1RAs. Hypoglycaemia incidence (%) and event rates (PPPY) at baseline and follow‐up stratified by dosing frequency are provided in Table S2. No significant changes were observed in hypoglycaemia incidence (%) or event rates (PPPY) at baseline and follow‐up stratified by baseline HbA1c level (Table S3).

FIGURE 3.

(A) Incidence of hypoglycaemia, and (B) hypoglycaemia event rate (per person per year [PPPY]), at 6 months pre‐ and 6 months post‐index period

3.4. Healthcare resource utilization

Significant reduction in the incidence of all‐cause (98.9% vs. 95.2%; P = 0.008) and diabetes‐associated outpatient visits (93.7% vs. 82.3%; P < 0.0001) were recorded between baseline and 6‐month follow‐up periods (Table 2). Reductions in all‐cause (10.2 vs. 9.0 PPPY; P = 0.005) and diabetes‐associated outpatient visits (5.7 vs. 4.2 PPPY; P < 0.0001) were also recorded between baseline and 6‐month follow‐up periods. No other significant changes in the incidence or rates of HCRU were recorded, including hypoglycaemia‐associated HCRU.

TABLE 2.

Healthcare resource utilization at baseline and follow‐up

	All‐cause		Diabetes‐associated		Hypoglycaemia‐associated
	BL	FU	BL	FU	BL	FU
Incidence, n (%)
Inpatient visits	21 (7.75)	25 (9.23)	13 (4.80)	15 (5.54)	1 (0.37)	0 (0.00)
ED visits	31 (11.44)	27 (9.96)	14 (5.17)	11 (4.06)	0 (0.00)	1 (0.37)
Outpatient visits	268 (98.89)	258 (95.20)*	254 (93.73)	223 (82.29)*	17 (6.27)	19 (7.01)
Endocrinology outpatient visits	35 (12.91)	39 (14.39)	33 (12.18)	29 (10.70)	2 (0.74)	3 (1.11)
Event rate, events (PPPY)
Inpatient visits	35 (0.26)	27 (0.20)	17 (0.13)	16 (0.12)	1 (0.0074)	0 (0.00)
ED visits	38 (0.28)	32 (0.24)	17 (0.13)	11 (0.08)	0 (0.00)	1 (0.0074)
Outpatient visits	1382 (10.20)	1224 (9.03)*	778 (5.74)	563 (4.15)*	28 (0.21)	35 (0.26)
Endocrinology outpatient visits	71 (0.52)	76 (0.56)	59 (0.44)	53 (0.39)	3 (0.02)	10 (0.07)
Inpatient days (PPPY)	72 (0.53)	82 (0.61)	41 (0.30)	45 (0.33)	8 (0.06)	0 (0.00)

Abbreviations: BL, baseline; ED, emergency department; FU, follow‐up; PPPY, per person per year.

^*Significant (P < 0.01) vs. baseline.

4. DISCUSSION

In US real‐world clinical practice, addition of Gla‐300 to GLP‐1RA therapy significantly improved glycaemic control in PWD2 without significantly increasing hypoglycaemia. This is highly clinically relevant in the context of PWD2 initiating GLP‐1RAs and adding BI, as most previous studies have assessed the addition of GLP‐1RAs to BI therapy. Improvements in glycaemic control were seen irrespective of the use of daily or weekly GLP‐1RAs, or baseline HbA1c level. Observations support ADA/EASD guidelines, noting that insulin should be added to GLP‐1RAs for PWD2 with inadequate glycaemic control requiring treatment intensification. ⁶ , ⁷ , ¹⁰ Addition of Gla‐300 led to a further reduction of approximately 1% in HbA1c and between two‐ and three‐times more patients reaching their target HbA1c levels. The post‐index improvements in HbA1c seen here are in line with observations from the intensification arm of the LixiLan and DUAL clinical studies, suggesting that adding Gla‐300 to GLP‐1RA therapy improves HbA1c. In these studies, intensification with Gla‐300 provided additional benefit compared with continuation of GLP‐1RAs alone. ¹² , ¹³ In patients on GLP‐1RAs requiring intensification with Gla‐300, the addition of Gla‐300 did not increase the risk of hypoglycaemia overall. A small increase was observed in the subset of patients receiving GLP‐1RA therapy daily versus weekly; however, hypoglycaemic events were rare, with event rates indicating approximately one event every 2 years (0.44 PPPY). Addition of Gla‐300 was associated with a reduction in the incidence and rate of outpatient HCRU, both overall and diabetes‐related. The reduction in diabetes‐related outpatient visits most likely reflects improved glycaemic control with Gla‐300, and in turn drives a consequent reduction in all‐cause outpatient visits. There were no differences in other measures of HCRU after intensification with Gla‐300.

To our knowledge, this is the first study to provide insights into the characteristics of insulin‐naïve PWD2 on GLP‐1RA treatment intensified with Gla‐300 in a real‐world setting, whilst also assessing their clinical response to intensification with Gla‐300 over a 6‐month follow‐up period. Such patients are typically not well represented in randomized controlled trials; therefore, these results provide valuable information regarding this patient population for clinicians, healthcare‐delivery systems, patients and payers. Furthermore, although the addition of GLP‐1RA therapy to oral medication with and without insulin has been well studied, GLP‐1RA therapy is now typically used early and before insulin in the treatment of T2DM. Therefore, the findings from this study, showing the effect of adding insulin to GLP‐1RA therapy, are of high clinical relevance.

The study did have some potential limitations. First, it had a retrospective design with a relatively short follow‐up period (6 months). Second, hypoglycaemia may have been underreported in the study, as only the clinically significant events were likely to have been captured (i.e., there were no data from self‐monitoring of blood glucose or continuous blood‐glucose monitoring). Third, data extracted from the database were mainly from patients from the Northwestern and Southern states of the US and thus may not be representative of the US national or global T2DM landscape. Lastly, there are several considerations when interpreting analyses of EMR data. For example, diagnoses were based on ICD‐9/‐10 codes, and as EMR data may not link the actual diagnosis name, this could have resulted in some misclassification. Furthermore, as EMRs only capture medication prescription, not dispensing or consumption, prescription information may not reflect actual drug usage in real life. In addition, dosage data were missing in a high percentage of the EMRs and therefore dose information and titration intensity/timing could not be addressed in this study. It would of course be interesting to see the impact of dose titration and this could be an avenue for further study.

In summary, results from this US real‐world study showed that intensification of GLP‐1RA therapy with Gla‐300 was an effective option for improving glycaemic control without any apparent increase in risk of hypoglycaemia. In addition, intensification with Gla‐300 was associated with a reduction in outpatient resource use. The findings from this real‐world study warrant further assessment of the effect of adding Gla‐300 to GLP‐1RA therapy in PWD2.

AUTHOR CONTRIBUTIONS

All authors contributed to the interpretation of the data, drafting and critical review of the manuscript, and approved the final version for submission.

CONFLICTS OF INTEREST

T.S.B has received research support from Abbott Diabetes, Abbott Rapid Diagnostics, Biolinq, Capillary Biomedical, Dexcom, Eli Lilly, Kowa, Livongo, Mannkind, Medtronic, Novo Nordisk, REMD, Sanofi, Sanvita, Senseonics, Viacyte, vTv Therapeutics and Zealand Pharma; consulting honoraria from Abbott, CeQur, Lifescan, Mannkind, Medtronic, Novo Nordisk and Sanofi; and speaking honoraria from BD, Medtronic and Sanofi. C.N., J.G. and J.W. are employees and stockholders of Sanofi. M.J.S. and L.S. are employees of Accenture, which has received research funding for this analysis.

FUNDING INFORMATION

This study was funded by Sanofi.

PEER REVIEW

The peer review history for this article is available at https://publons.com/publon/10.1111/dom.14739.

Supporting information

Appendix S1 Supporting Information

Bailey TS, Gill J, Jones S. M, Shenoy L, Nicholls C, Westerbacka J. Real‐world outcomes of addition of insulin glargine 300 U/mL (Gla‐300) to glucagon‐like peptide‐1 receptor agonist (GLP‐1 RA) therapy in people with type 2 diabetes: The DELIVER‐G study. Diabetes Obes Metab. 2022;24(8):1617‐1622. doi: 10.1111/dom.14739

DATA AVAILABILITY STATEMENT

Aggregate data are provided in the manuscript. Scripts used for pulling data from IBM Explorys platform are available on request from Merwyn Jones S. Restrictions apply to the availability of the source patient‐level data, which were used under license for this study.