Article Text

Download PDFPDF

Clinical perspectives on the frequency of hypoglycemia in treat-to-target randomized controlled trials comparing basal insulin analogs in type 2 diabetes: a narrative review
  1. Julio Rosenstock1,
  2. Harpreet S Bajaj2,
  3. Ildiko Lingvay3,
  4. Simon R Heller4
  1. 1Velocity Clinical Research at Medical City, Dallas, Texas, USA
  2. 2LMC Diabetes & Endocrinology, Brampton, Ontario, Canada
  3. 3Endocrinology Division, Department of Internal Medicine and Peter O’Donnell School of Public Health, University of Texas Southwestern Medical Center, Dallas, Texas, USA
  4. 4Department of Oncology and Metabolism, University of Sheffield, Sheffield, UK
  1. Correspondence to Dr Julio Rosenstock; juliorosenstock{at}dallasdiabetes.com

Abstract

The objective of this review was to comprehensively present and summarize trends in reported rates of hypoglycemia with one or two times per day basal insulin analogs in individuals with type 2 diabetes to help address and contextualize the emerging theoretical concern of increased hypoglycemic risk with once-weekly basal insulins.

Hypoglycemia data were extracted from treat-to-target randomized clinical trials conducted during 2000–2022. Published articles were identified on PubMed or within the US Food and Drug Administration submission documents. Overall, 57 articles were identified: 44 assessed hypoglycemic outcomes in participants receiving basal-only therapy (33 in insulin-naive participants; 11 in insulin-experienced participants), 4 in a mixed population (insulin-naive and insulin-experienced participants) and 9 in participants receiving basal-bolus therapy. For the analysis, emphasis was placed on level 2 (blood glucose <3.0 mmol/L (<54 mg/dL)) and level 3 (or severe) hypoglycemia.

Overall, event rates for level 2 or level 3 hypoglycemia across most studies ranged from 0.06 to 7.10 events/person-year of exposure (PYE) for participants receiving a basal-only insulin regimen; the rate for basal-bolus regimens ranged from 2.4 to 13.6 events/PYE. Rates were generally lower with second-generation basal insulins (insulin degludec or insulin glargine U300) than with neutral protamine Hagedorn insulin or first-generation basal insulins (insulin detemir or insulin glargine U100). Subgroup categorization by sulfonylurea usage, end-of-treatment insulin dose or glycated hemoglobin reduction did not show consistent trends on overall hypoglycemia rates. Hypoglycemia rates reported so far for once-weekly basal insulins are consistent with or lower than those reported for daily-administered basal insulin analogs.

  • Hypoglycemia
  • Diabetes Mellitus, Type 2
  • Insulin

Data availability statement

All data relevant to the study are included in the article or uploaded as supplemental information.

http://creativecommons.org/licenses/by-nc/4.0/

This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/.

Statistics from Altmetric.com

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Introduction

As recommended by the American Diabetes Association (ADA) and European Association for the Study of Diabetes (EASD) 2022 consensus on the management of hyperglycemia,1 basal insulin has been replaced by injectable incretin-related therapy as the first-line injectable therapy for type 2 diabetes (T2D) insufficiently controlled by combinations of oral glucose-lowering agents. However, many individuals with T2D may still eventually require basal insulin as beta-cell function declines over time or if insufficiently controlled by injectable incretin-related therapy.2

Hypoglycemia is a common adverse event associated with insulin therapy, and hypoglycemia risk increases with intensive glycemic control and duration of T2D.3 Optimizing T2D management with insulin therapy is therefore a balancing act between achieving optimal glycemic control while minimising hypoglycemia risk, irrespective of the insulin formulation used in clinical practice. Overall hypoglycemia may be considered as any episode with blood glucose (BG) <3.9 mmol/L (<70 mg/dL) irrespective of time of day. ‘Level 1’ hypoglycemia consists of BG values between 3.0 and 3.9 mmol/L (54–70 mg/dL), which are regarded as ‘alerts’ to take action but do not have regulatory implications.4 5 Clinicians and regulators consider ‘level 2’ (or ‘clinically significant’) hypoglycemia, defined as a BG value <3.0 mmol/L (<54 mg/dL), and ‘level 3’ (or ‘severe’) hypoglycemia, defined as an episode causing sufficient cognitive impairment to require assistance from another person, to be most relevant when evaluating treatment.4 6 Both level 2 and level 3 hypoglycemia have substantial clinical impacts, increasing mortality and morbidity7 8 while also reducing quality of life9 and exerting indirect effects via increased healthcare resource utilization (ie, hospital admissions), increased workplace absenteeism and reduced productivity.10 Recent real-world analyses in the USA show alarmingly high rates of level 3 or severe hypoglycemia, supporting the urgent need for hypoglycemia-mitigating interventions.11

Over the past three decades, there have been major improvements in the pharmacokinetic (PK) and pharmacodynamic (PD) properties of basal insulins, starting with first-generation basal insulin analogs (ie, insulin detemir (detemir) and insulin glargine U100 (glargine U100)), generating more predictable, longer-acting soluble formulations with flatter PK and PD profiles among insulin analogs, resulting in reduced hypoglycemia risk compared with neutral protamine Hagedorn (NPH) insulin.12–14

The ‘Treat-To-Target’ Study published in 2003, comparing glargine U100 with NPH insulin, led to a paradigm shift in the accepted standard of care regarding insulin therapy for T2D.15 This trial established the concept of gradually increasing the insulin dose to attain a predefined fasting glucose target or until unacceptable hypoglycemia occurs. This ‘treat-to-target’ approach would thus enable glycated hemoglobin (HbA1c) parity between comparators, and in turn allow for differentiation based on other outcomes of interest (eg, hypoglycemia). This study reduced hypoglycemia risk using basal insulin analogs that showed a more predictable PK/PD; this enabled systematic insulin titrations resulting in fewer hypoglycemic episodes, particularly at night.16 Subsequent development of second-generation insulin analogs with an even longer duration of action, flatter insulin PK/PD profiles, and lower insulin peak:trough ratios (ie, more constant circulating insulin levels), such as insulin degludec (degludec) U100 and U200, and insulin glargine U300 (glargine U300), further reduced hypoglycemia risk compared with glargine U100.17–19

Despite these developments, hypoglycemia remains a fundamental challenge of basal insulin therapy. Fear of hypoglycemia, delays in treatment initiation partly due to the injection barrier, titration inertia and a lack of concordance are perhaps the main barriers to the effective use of insulin therapy in patients with T2D.20 21 To help overcome clinical inertia and to reduce the injection burden with daily basal insulin, two once-weekly basal insulins have been developed and are in late-stage clinical trials: insulin icodec (icodec) and basal insulin Fc efsitora alfa (efsitora). The efficacy of icodec and efsitora was demonstrated in phase 1 and phase 2 randomized clinical trials (RCTs),22–28 and further confirmed for icodec in the phase 3a ONWARDS program in insulin-naive individuals newly initiating insulin, individuals treated with basal insulin alone or individuals already established on basal-bolus insulin regimens.29–34

Given the protracted duration of action of once-weekly insulin, physicians may have a theoretical concern regarding the risk of hypoglycemia.35 Contextualizing the risk of hypoglycemia observed with once-weekly insulin in RCTs requires a clear and up-to-date understanding of the risk of hypoglycemia seen with currently available once-daily or two times per day basal analogs. However, owing to heterogeneity in trial designs, hypoglycemia definitions, titration interventions and study populations, it is difficult to compare hypoglycemia outcomes between RCTs in a formal meta-analysis. Therefore, the aim of this narrative review was twofold: (1) to define the hypoglycemia risk with current one or two times per day basal insulin analogs; and (2) to use these data to help contextualize emerging once-weekly basal insulin hypoglycemia data. Here, we comprehensively summarize mean hypoglycemia rates from treat-to-target trials of daily basal insulins in T2D conducted over the past 22 years (2000–2022), with reference to the differences between studies. Importantly, given the narrative nature of this review, no comparative statistical analyses were possible.

Methods: trial identification and data extraction

Published articles were identified in Ovid MEDLINE and Embase via targeted searches using the search strings shown in online supplemental table 1. The US Food and Drug Administration (FDA) submission documents for glargine U100,36 detemir,37 degludec38 and glargine U30039 were also used to source additional studies. Only English articles were included.

Supplemental material

Treat-to-target RCTs (sample size >30 participants/arm) assessing daily basal insulins (detemir, glargine U100, degludec and glargine U300) in adults (≥18 years of age) with T2D conducted between January 1, 2000 and December 30, 2022 and reporting BG-confirmed hypoglycemia were eligible for inclusion. RCTs assessing premix insulin or fixed-dose basal insulin/glucagon-like peptide-1 receptor agonist (GLP-1 RA) combinations without a basal-only insulin arm were not eligible. Data from phase 3a RCTs assessing icodec were added during development, as the publications became available.30–33 40

The primary outcomes of interest were the mean event rates/person-year of clinically relevant ‘level 2’ hypoglycemia (BG value <3.1 mmol/L (55 mg/dL) or <3.0 mmol/L (54 mg/dL)), severe ‘level 3’ hypoglycemia, overall hypoglycemia (any BG value <3.9 mmol/L (70 mg/dL) irrespective of time of day) and nocturnal hypoglycemia; these definitions were selected to provide the greatest data inclusivity, given the changes in hypoglycemia definitions over time. Prior to 2017 and the position statement by the International Hypoglycaemia Study Group,5 the glucose threshold for ‘level 2’ hypoglycemia was applied arbitrarily across studies; hence, both <3.1 mmol/L (55 mg/dL) and <3.0 mmol/L (54 mg/dL) thresholds have been accounted for in the analysis. Hypoglycemia rates are presented as events/person-year of exposure (PYE) reported for the defined study treatment period. Continuous glucose monitoring (CGM)-based time below range (TBR), defined as proportion of time spent with sensor glucose levels below target range (<3.9 mmol/L (<70 mg/dL)), was also captured; however, trials reporting CGM-based hypoglycemia without also providing BG-confirmed values were excluded. Study design and participant baseline characteristics (body mass index (BMI), HbA1c, age and diabetes duration), in addition to changes in HbA1c from baseline to end of treatment (EOT), were also captured. Sulfonylurea usage and continuation (yes/no) after trial randomization was also extracted. As access to the raw data was not possible, and to avoid any inadvertent introduction of errors, all data reported here employ the format and number of decimal places used in the original publication.

Overview of treat-to-target daily basal insulin trials in people with T2D (2000–2022)

Overview of trials

Overall, 57 treat-to-target RCTs were identified: 39 via the targeted literature search, 13 from the FDA submission documents (NPH insulin (n=4),41–44 glargine U100 (n=6),41–43 45–47 detemir (n=4),44 48–50 degludec (n=4),46 51–53 and glargine U300 (n=1)45) and 5 phase 3a icodec publications that were added as they became available.30–33 40

Of these 57 RCTs, 44 assessed outcomes in individuals receiving basal-only therapy (33 in insulin-naive participants; 11 in insulin-experienced participants), 4 in a mixed population of insulin-naive and insulin-experienced participants and 9 in individuals receiving basal-bolus therapy.

Both basal-only and basal-bolus regimens were administered with or without non-insulin glucose-lowering agents including GLP-1 RAs and/or sulfonylureas.

Hypoglycemia data collection and definitions

Across studies, hypoglycemia data were collected using self-measured blood glucose (SMBG) measurements (n=50), CGM devices (n=1) or a combination of both (n=1) (online supplemental tables 2 and 3). SMBG measurements were used to measure overall and nocturnal hypoglycemia rates, while CGM was used to measure TBR. Overall and nocturnal hypoglycemia definitions for the included trials are shown in online supplemental table 4. Most trials used definitions of BG-confirmed <3.1 mmol/L (<55 mg/dL), severe hypoglycemia or BG-confirmed <3.1 mmol/L (<55 mg/dL) or severe hypoglycemia or BG-confirmed values of <3.0 mmol/L (<54 mg/dL), which were grouped together for the purpose of this analysis.

Trial characteristics

A summary of concomitant sulfonylurea usage, trial duration, baseline characteristics, SMBG titration targets, EOT insulin dose and changes in HbA1c across all studies is presented in table 1. Additional details on an individual study basis are presented in online supplemental tables 2 and 3. Most trials were phase 3 (three were phase 4) and multicenter; two single-center studies were included (online supplemental table 2). Trial duration was ≤16 weeks for 4 studies, 22–28 weeks for the majority of studies (n=34), 36–41 weeks for 4 studies, 52–54 weeks for 10 studies, and ≥64 weeks for 5 studies. Ten of the trials reported a non-inferiority statistical framework (online supplemental table 2). Generally, the mean ranges for baseline age and duration of diabetes for insulin-experienced individuals in the basal-only (54.7–62.9 years and 13.0–15.6 years, respectively) and basal-bolus trials (56.5–62.1 years and 12.4–16.4 years, respectively) were higher than those for insulin-naive individuals receiving basal-only insulin therapy (52.9–64.0 years and 6.4–11.6 years, respectively). Across each set of studies per insulin type, the range in HbA1c reduction was improved with successive generations of insulin (table 1). All included studies employed treat-to-target protocols based on glucose targets, except one study that used both glucose and HbA1c (<7.0%) targets (online supplemental table 2).54

Table 1

Summary of trials and hypoglycemia rates per insulin type for treat-to-target basal-only and basal-bolus insulin T2D trials

Summary of overall and nocturnal hypoglycemia event rates

Overall and nocturnal hypoglycemia event rates (events/PYE) across studies are summarized in table 1. Itemized outcomes, including incidence rates and number of participants reporting hypoglycemia, for each individual trial are shown in online supplemental tables 5 and 6. Generally, there were wide variations in rates of overall hypoglycemia and nocturnal hypoglycemia across RCTs (table 1).

Across basal-only trials in insulin-naive individuals, overall BG-confirmed hypoglycemia rates (any BG threshold <3.9 mmol/L (<70 mg/dL) at any time of day (level 1 or two hypoglycemia)) ranged from 0.06 to 8.5 events/PYE41 45 48–50 55–58 for NPH insulin, detemir, glargine U100, glargine U300, and degludec with a general trend for lower rates (events/PYE) with successive generations of insulin (table 1). Severe hypoglycemia (level 3) rates across these basal insulins were low and ranged from 0.0 to 0.14 events/PYE. Nocturnal hypoglycemia rates (insufficient information to group by level) ranged from 0.03 to 1.3 events/PYE, with a trend for lower rates with second-generation analogs than with first-generation analogs (table 1).

For basal-only trials in insulin-experienced individuals, reported overall BG-confirmed hypoglycemia rates (any BG threshold <3.9 mmol/L (<70 mg/dL)) ranged from 0.28 to 8.11 events/PYE for glargine U100, glargine U300, and degludec (table 1). For these basal insulins, severe (or level 3) hypoglycemia rates ranged from 0.0 to 0.09 events/PYE, while nocturnal hypoglycemia rates ranged from 0.11 to 2.3 events/PYE. Studies with NPH or detemir in this group were either not conducted or did not report hypoglycemia rates (table 1).

For the basal-bolus trials, reported overall BG-confirmed hypoglycemia rates were substantially higher (any BG threshold <3.9 mmol/L (<70 mg/dL)), ranging from 2.7 to 26.8 events/PYE for glargine U100, glargine U300, and degludec. Nocturnal hypoglycemia rates ranged from 0.3 to 4.2 events/PYE for glargine U10059–63 and were 1.4 events/PYE for degludec.63 Studies with NPH or detemir in this group were either not conducted or did not report hypoglycemia rates (table 1).

Impact of study factors on overall hypoglycemia rates

To investigate the impact of hypoglycemia definition, overall hypoglycemia rates were grouped by respective study into two categories: (1) overall severe or BG-confirmed <3.9 mmol/L (<70 mg/dL) and (2) overall severe or BG-confirmed <3.1 mmol/L (<55 mg/dL) or <3.0 mmol/L (<54 mg/dL) (figure 1A,B). Hypoglycemia event rates were variable, with broader rate ranges observed among trials using a threshold of <3.9 mmol/L (<70 mg/dL; figure 1A)45 54 56 58 61 64–66 than among those with BG <3.1 mmol/L (<55 mg/dL)46–53 55 57 59 60 63 67–69 or <3.0 mmol/L (<54 mg/dL17 24 45; figure 1B). Hypoglycemia rates were generally higher for basal-bolus trials60 61 63 than for basal-only trials in both insulin-naive32–34 41 45 46 48–52 54–58 64 67 68 70–84 and insulin-experienced individuals,47 53 65 66 69 79 85 whereas rates for mixed population trials tended to be in the middle (figure 1A,B). For overall severe or BG-confirmed <3.9 mmol/L (<70 mg/dL), hypoglycemia rates ranged from 0.7 to 8.5 events/PYE for basal-only (insulin-naive) trials (glargine U100, 0.7–8.5; glargine U300, 1.4–6.4),45 54 56 58 64 from 3.36 to 18.14 events/PYE for basal-only (insulin-experienced) trials (glargine U100, 3.36–18.14; degludec, 4.22–3.39; glargine U300, 6.76–14.01),65 66 and from 11.24 to 26.76 events/PYE for the basal-bolus trial61 (glargine U100, 11.24–26.76; glargine U300, 11.39–25.48; figure 1A). For overall severe or BG-confirmed <3.1 mmol/L (<55 mg/dL) or <3.0 mmol/L (<54 mg/dL), hypoglycemia rate ranges (events/PYE) were 0.06–7.14 for basal-only (insulin-naive) trials (NPH insulin 7.14; detemir 0.23–3.67; glargine U100, 0.06–2.9; glargine U300, 0.11–0.24; degludec, 0.09–1.6),46 48–52 55 57 67 68 77 80 83 84 0.28–5.66 for basal-only (insulin-experienced) trials (glargine U100, 0.28–2.75; glargine U300, 1.11–1.64; degludec, 0.28–5.66),47 53 69 79 85 and 2.43–13.63 for basal-bolus trials (glargine U100, 2.56–13.63; glargine U300, 2.43–3.78; degludec, 11.09)60 63 (figure 1B).

Figure 1

Overall hypoglycemia rates grouped according to hypoglycemia definitions (blood glucose thresholds: ≤3.9 mmol/L [≤70 mg/dL] [Panel A], <3.1 mmol/L [<55 mg/dL] or <3.0 mmol/L [<54 mg/dL] [Panel B]) used for treat-to-target basal-only (BO) and basal-bolus insulin T2D trials. Hypoglycemia rates are shown as events/PYE. Each data point represents a single trial. Six BO (insulin-naive) trials reported data for multiple basal insulins (Bolli et al45; Hermansen et al50; Meneghini et al101; Pan et al46; Rosenstock et al57; Zinman et al67); two BO (insulin-experienced) trials reported data for multiple basal insulins (Yki-Järvinen et al65 and Wysham et al85). BG, blood glucose; degludec, insulin degludec; detemir, insulin detemir; glargine U100, insulin glargine U100; glargine U300, insulin glargine U300; NPH, neutral protamine Hagedorn insulin; PYE, person-year of exposure; T2D, type 2 diabetes.

Overall severe or BG-confirmed hypoglycemia rates <3.1 mmol/L (<55 mg/dL) or <3.0 mmol/L (<54 mg/dL) were grouped according to concomitant sulfonylurea usage (figure 2A), EOT basal insulin dose (figure 2B) and change from baseline to EOT in HbA1c (figure 2C). No trends were observed across these parameters. Similarly, achieving EOT HbA1c <7.0% vs ≥7.0% did not consistently lead to higher severe or BG-confirmed hypoglycemia rates (online supplemental figure 1).

Figure 2

Overall hypoglycemia rates (overall severe or BG-confirmed/BG-confirmed value of <3.0 mmol/L (<54 mg/dL) or <3.1 mmol/L (<55 mg/dL)) grouped according to sulfonylurea usage (A), EOT basal insulin dose (B) and change from baseline to EOT in HbA1c (C). Hypoglycemia rates are shown as events/PYE. Each data point represents a single trial; some trials reported hypoglycemia rates for more than one basal insulin. Numbers of data points differ across panels because not all trials reported sulfonylurea usage or the insulin dose used. BG, blood glucose; BO, basal-only; degludec, insulin degludec; detemir, insulin detemir; EOT, end of treatment; glargine U100, insulin glargine U100; glargine U300, insulin glargine U300; HbA1c, glycated hemoglobin; NPH, neutral protamine Hagedorn; PYE, person-year of exposure.

Exploring trends in hypoglycemia rates across different study populations

Basal-only (insulin-naive) trials: BG-confirmed <3.9 mmol/L (<70 mg/dL) irrespective of symptoms

Only glargine U100 or U300 trials reported rates with BG-confirmed <3.9 mmol/L in insulin-naive individuals, with the largest range seen with glargine U100 (figure 1A). The highest rate was seen in Bolli et al with glargine U100; 8.5 events/PYE after 26 weeks with a corresponding reduction in HbA1c from baseline to EOT of −1.46% points (EOT insulin dose, 52 units (U)/day; titration target, 4.4–5.6 mmol/L (80–100 mg/dL)).45 Conversely, the lowest rate was observed in Yang et al with glargine U100; 1.1 events/PYE after 24 weeks, substantially lower than that for the 26-week trial, and a corresponding HbA1c reduction of −1.45% points (EOT insulin dose, 19.5 U; titration target, ≤6.1 mmol/L (≤110 mg/dL)).84 Moreover, in the 24-week trial, sulfonylurea usage was continued after randomization,84 whereas sulfonylureas were discontinued in the 26-week trial.45 In addition, mean T2D duration and baseline BMI in the 24-week trial (7.9 years and 25.6 kg/m2)64 84 were lower than in the 26-week trial (9.6 years and 33.2 kg/m2)45; these differences may help account for the higher rates of overall hypoglycemia in the 26-week trial (figure 1A).45

Basal-only (insulin-naive) trials: BG-confirmed <3.1 mmol/L (<55 mg/dL) irrespective of symptoms

Five trials48–50 55 57 reported these hypoglycemia rates for detemir (figure 1B), including a 26-week trial48 48 and a 52-week trial,49 49 both of which used a titration target of 4.1–6.0 mmol/L (74–108 mg/dL). In these trials, the hypoglycemia rates for detemir were 0.28 and 0.23 events/PYE, respectively (EOT insulin dose, 39.5 U/day and 42 U/day, respectively), with corresponding HbA1c reductions of only −0.5% point (baseline HbA1c: 7.6% in both). However, in Rosenstock et al57 (52 weeks) and Hermansen et al50 (24 weeks), overall hypoglycemia rates for detemir were higher (2.9 and 3.67 events/PYE, respectively; EOT insulin dose, not reported and 36.1 U/day, respectively; titration target: <6.0 mmol/L (<108 mg/dL)) than those reported in DeVries et al and Rosenstock et al48 49; however, corresponding HbA1c reductions were greater (−1.5% points (52-week trial57) and −1.8% points (24-week trial50)), potentially explaining the higher hypoglycemia rates. The fifth trial, Meneghini et al (26 weeks), used a tighter titration target (>3.9–5.0 mmol/L); the reported rate was 1.15 events/PYE with a modest HbA1c reduction (−0.48% point).55 In Rosenstock et al and Hermansen et al, sulfonylureas were continued after randomization,50 57 but they were discontinued in DeVries et al, Rosenstock et al and Meneghini et al.48 49 55 Across all five trials, there was some variation in diabetes duration (range: 8.0–9.6 years), but there was no trend with respect to hypoglycemia rates. These data suggest that a combination of concomitant sulfonylurea usage with detemir and greater HbA1c reductions may have contributed to the higher hypoglycemia rates (figure 1B).

Basal-only trials (insulin-experienced): severe or BG-confirmed <3.1 mmol/L (<55 mg/dL) irrespective of symptoms

Four trials reported overall hypoglycemia rates for degludec (figure 1B).47 53 69 85 Sulfonylurea usage was generally discontinued after study randomization, although baseline use was reported for some studies.

Two 26-week trials, Pei et al and Mathieu et al,53 69 with titration targets of 4.0–5.0 mmol/L (72–90 mg/dL), reported rates of 0.48 events/PYE53 69 and 1.0 events/PYE, respectively, for degludec, with corresponding HbA1c reductions of −1.0% point and −0.7% point, respectively (baseline HbA1c: 9.0% and 7.7%, respectively). In Pei et al,69 EOT basal insulin dose was 37 U/day and diabetes duration was 11.5 years. In Mathieu et al,53 EOT basal insulin dose was not reported and diabetes duration was 12.9 years.

A 64-week trial85 and a 22-week trial,47 both with titration targets of 4.0–5.0 mmol/L (72–90 mg/dL), reported rates of overall hypoglycemia for degludec of 2.2 and 5.66 events/PYE, respectively, which were higher than those reported in Pei et al and Mathieu et al; there was a corresponding HbA1c reduction of −0.7% point in Bode et al (baseline: 8.2%)47 and −0.49% point in Wysham et al (baseline: 7.6%).85

Compared with Pei et al and Mathieu et al,53 69 diabetes duration and EOT insulin dose from Wysham et al85 (14.1 years and 83 U/day, respectively) and Bode et al47 (12.9 years and 76 U/day, respectively) were higher. No trends were seen regarding sulfonylurea usage.47 53 69 85 These data suggest that greater EOT insulin dose and diabetes duration may have contributed to higher hypoglycemia rates with degludec (figure 1B).

Basal-bolus trials: severe or BG-confirmed <3.9 mmol/L (<70 mg/dL) irrespective of symptoms

In a head-to-head trial of glargine U100 and glargine U300, both in combination with bolus insulin aspart,61 severe or BG-confirmed hypoglycemia rates (6.8 and 25.5 events/PYE, respectively) were higher than that observed in basal-only trials in insulin-naive and insulin-experienced individuals receiving glargine U10045 54 64–66 or glargine U30045 56 58 65 (figure 1B). In the head-to-head basal-bolus trial (titration target, 4.4–5.6 mmol/L (80–100 mg/dL)), both glargine U100 and glargine U300 reduced HbA1c by −0.83% point from 8.15% at baseline (EOT daily basal insulin dose not reported), and diabetes duration was similar for both groups (16.1 years (glargine U100) vs 15.6 years (glargine U300)). It is likely that the higher hypoglycemia rates with basal-bolus than with basal-only regimens are attributed to the additional bolus insulin (figure 1B).61

Basal-only (insulin-naive) trial compared with basal-bolus trial: severe or BG-confirmed <3.1 mmol/L (<55 mg/dL) irrespective of symptoms

BEGIN ONCE LONG was a 52-week trial (titration target, 3.9–5.0 mmol/L (70–90 mg/dL)) of basal-only glargine U100 compared with basal-only degludec, while BEGIN BB T2 compared basal-bolus glargine U100 with basal-bolus degludec (both in combination with insulin aspart).63 67 The rate of severe or BG-confirmed hypoglycemia was similar with basal-only glargine U100 and basal-only degludec (1.85 and 1.52 events/PYE, respectively), while the rate was statistically significantly lower with degludec than with glargine U100 in the basal-bolus comparison (11.1 vs 13.6 events/PYE; p=0.0359).63 67 Corresponding HbA1c reductions of −1.2% and −1.3% points (baseline: 8.2% and 8.4%), respectively, were reported.63 67 The higher hypoglycemia rates with basal-bolus than with basal-only regimens, despite similar glucose-lowering efficacy, can be attributed to the use of bolus insulin and longer diabetes duration (ie, disease progression) in the basal-bolus group (13.4 years vs 8.6 years) (figure 1B).63 67

For degludec, severe or BG-confirmed hypoglycemia rates were 1.5 and 11.1 events/PYE for insulin-naive (basal-only insulin) and insulin-experienced (basal-bolus insulin) individuals, respectively, with corresponding HbA1c reductions of −1.06% and −1.10% points (baseline: 8.2% and 8.3%).63 67 Similarly, despite a small difference in glycemic efficacy, this suggests that the lower hypoglycemia rates may have been driven by the bolus insulin and longer diabetes duration (13.6 years (basal-bolus) vs 9.4 years (basal-only)).63 67

CGM-measured hypoglycemia: TBR

CGM-measured TBR data were only reported for two glargine U100 basal-only trials (figure 3)71 86: a head-to head trial of glargine U100 versus degludec in insulin-experienced individuals,86 and a glargine plus sulfonylurea (gliclazide) evaluation in insulin-naïve individuals.71 Blinded CGM was used in both trials.

Figure 3

CGM-measured TBR (over a 24-hour period) for glargine U100 basal-only trials in insulin-naive individuals with T2D. Observed data are reported for Bajaj et al.71 Estimated data are reported for Goldenberg et al.86 In Bajaj et al, individuals wore blinded CGM devices during week 13. In Goldenberg et al (a 41-week crossover trial of glargine U100 vs degludec), after a 2-week screening period (week −4 to week −2) and a 2-week run-in period (week −2 to week 0), there were two consecutive 16-week treatment periods (weeks 0–16 and weeks 18–34), each followed by a 2-week maintenance period (weeks 16–18 and weeks 34–36); at the end of the second treatment period, there was a 1-week follow-up (weeks 36–37). The data shown in the figure are that collected during the two 2-week maintenance periods and were recorded by participants using a blinded CGM device. CGM, continuous glucose monitoring; degludec, insulin degludec; EOT, end of treatment; glargine U100, insulin glargine U100; GLC, gliclazide; HbA1c, glycated hemoglobin; T2D, type 2 diabetes; TBR, time below range; U, units.

In the head-to-head trial,86 the proportion of TBR was numerically lower with degludec than with glargine U100, despite an identical titration target of 3.9–5.0 mmol/L (70–90 mg/dL); both degludec and glargine U100 demonstrated similar HbA1c reductions from baseline (−0.5% vs −0.4% points, respectively; baseline: 7.6%) and comparable EOT insulin doses (56 vs 59 U/day). Overall, for glargine U100 and degludec, the proportion of TBR for level 1 hypoglycemia (sensor glucose 3.0–3.8 mmol/L (54–68 mg/dL); recommended target <4.0%) was 6.3% and 5.8%, while for level 2 hypoglycemia (sensor glucose <3.0 mmol/L (<54 mg/dL); recommended target <1.0%), it was 2.5% and 2.2%, respectively. The estimated mean treatment difference (degludec−glargine U100) in TBR for level 2 hypoglycemia was −0.24 (95% CI −0.79 to 0.31), which was not statistically significantly different.86

Overall hypoglycemia rates with once-weekly insulin

The efficacy and safety of once-weekly insulin in T2D were investigated in three phase 2 icodec studies,22 23 five phase 3 icodec studies (ONWARDS 1–5)29–33 40 and two phase 2 efsitora alfa studies.27 28 In the icodec phase 3a studies, sulfonylureas were either discontinued or reduced by 50% at the investigator’s discretion to mirror real-world conditions.29–33 40 In ONWARDS 1 and ONWARDS 3, overall level 2 and 3 hypoglycemia rates (events/PYE) in basal-only insulin-naive individuals ranged from 0.19 to 0.31 for icodec, compared with 0.16 for glargine U100 and 0.15 for degludec (figures 1B and 2), with overall HbA1c reductions ranging from −1.3% to −1.6% points (baseline range: 8.4–8.6%), confirming statistical non-inferiority and superiority of icodec versus respective once-daily comparators.32 33 In insulin-naive individuals, icodec plus a dosing guidance app was also compared against a pooled group of once-daily basal insulin analogs (degludec, glargine U100 or glargine U300) administered following real-world practice; overall combined level 2 or level 3 hypoglycemia rates (events/PYE) were 0.19 and 0.14, respectively, with corresponding HbA1c reductions of −1.68% and −1.31% points (baseline: 9.0% and 8.9%, respectively), confirming non-inferiority and superiority of icodec.40 In basal-only insulin-experienced individuals, overall combined level 2 or level 3 hypoglycemia rates (events/PYE) were 0.7 for icodec and 0.3 for degludec (figures 1B and 2), but were not statistically significant, with corresponding HbA1c reductions of −0.93% and −0.71% points, respectively (baseline: 8.2% and 8.1%, respectively), confirming non-inferiority and superiority of icodec.30 In individuals receiving basal-bolus therapy, the overall combined level 2 or level 3 hypoglycemia rates were similar for icodec+insulin aspart versus glargine U100+insulin aspart (5.6 events/PYE (both)) (figures 1B and 2), as were the corresponding HbA1c reductions (−1.16% and −1.18% points, respectively, both from 8.3% at baseline), confirming non-inferiority of icodec.31

Once-weekly efsitora alfa was compared with once-daily degludec as basal-only regimens in insulin-naive and insulin-experienced individuals using various titration targets in two phase 2 trials.27 28 In insulin-naive individuals, the titration target for both treatments was 4.4–<5.6 mmol/L (80–100 mg/dL), while in insulin-experienced individuals, the two titration targets for efsitora were ≤7.8 mmol/L (≤140 mg/dL) (algorithm 1) and ≤6.7 mmol/L (≤120 mg/dL) (algorithm 2), and the titration target for degludec was ≤5.6 mmol/L (≤100 mg/dL).27 28 In insulin-experienced individuals, baseline HbA1c was reduced after 32 weeks by −0.6% point for efsitora (both titration algorithms) and −0.7% point for degludec (baseline: 8.1% for both). The rate of level 1 hypoglycemia (<3.9 mmol/L (<70 mg/dL)) was statistically significantly lower with efsitora than with degludec (23.0 vs 30.5 events/PYE; p value not reported). For level 2 hypoglycemia (<3.0 mmol/L (<54 mg/dL)), there was no statistically significant difference between arms (2.2 vs 3.0 events/PYE; p value not reported).28 The difference in level 1 hypoglycemia may be attributed to the more stringent titration target used for degludec than for efsitora, resulting in statistically significantly lower fasting blood glucose with degludec than with efsitora (change from baseline: –1.8 mmol/L (–32.7 mg/dL) vs –0.9 mmol/L (–17.0 mg/dL), respectively; p value not reported). In insulin-naive individuals, however, there was no statistically significant difference between efsitora and degludec using the same titration target in terms of level 1 (3.29 vs 2.77 events/PYE, respectively) or level 2 hypoglycemia (0.22 vs 0.15 events/PYE, respectively); reductions in HbA1c were −1.20% points and −1.26% points, respectively (baseline: ~8.0%).27

Discussion

This narrative review comprehensively summarizes rates of hypoglycemia from T2D treat-to-target trials of one or two times per day administered basal insulins and provides relevant context for the event rates in the ongoing or completed once-weekly insulin trials, while considering the differences in trial design, hypoglycemia definitions, basal insulin type/dose, titration targets, data collection methods and HbA1c-lowering efficacy. Our narrative approach precludes definitive conclusions to be drawn regarding the relative efficacy and safety of basal insulins as this can only be achieved using participant-level analyses of head-to-head RCT data.

Nevertheless, data indicate that, in treat-to-target RCTs, rates of hypoglycemia were generally lower with second-generation basal insulins (degludec and glargine U300) than with NPH insulin or first-generation basal insulin analogs (detemir and glargine U100) and, generally, most RCTs with basal insulin analogs reported an overall hypoglycemia rate (BG <70 mg/dL (<3.9 mmol/L) at any time of the day) of more than one event/PYE. Notably, those trials reporting lower hypoglycemic event rates had lower HbA1c reductions, well above the recommended target of <7.0%. As expected, owing to the more intensive insulin regimen, the use of mealtime bolus insulin and longer diabetes duration, hypoglycemia rates were higher in basal-bolus trials than in basal-only trials, particularly for previously insulin-naive individuals. However, the reported hypoglycemia rates were highly variable across trials, largely driven by heterogeneity in trial design, trial parameters (including eligible population), hypoglycemia definitions, glycemic control and hypoglycemia data collection methods, which can all directly impact the collection, reporting and analysis of hypoglycemia data. As these elements varied greatly across trials, it renders impossible any statistical comparison of hypoglycemia data across studies, yet holistically, these data can provide valuable clinical context when gauging hypoglycemic risk with weekly insulins relative to the ranges reported for widely used first-generation and second-generation daily basal insulin analogs.

Hypoglycemia risk is driven by a complex interplay of factors such as the use of concomitant sulfonylureas,87 strictness of insulin titration regimens88 and glucose-lowering efficacy, plus the strength of counter-regulatory responses which are impaired in many individuals with diabetes.89 However, in this review, sulfonylurea usage, EOT insulin dose and the glucose-lowering efficacy of the insulins did not show consistent effects on overall hypoglycemia. Given the association between sulfonylurea usage and hypoglycemia, it was surprising that a clearer trend between the two was not seen; however, differences in other variables in the trial designs/methodologies may have had a confounding effect.

Concerns regarding greater frequency of hypoglycemia with once-weekly insulin use (icodec and efsitora) may be assuaged by contextualizing emerging once-weekly insulin data from SMBG and CGM analyses with observed hypoglycemia rates with first-generation and second-generation basal insulin analogs. Findings from this narrative review suggest that level 2 and level 3 hypoglycemia rates reported for icodec were below (in ONWARDS 1, 3, 2 and 5) or within (in ONWARDS 4) the rate ranges from basal-only (in insulin-naive and basal insulin-experienced) trials and basal-bolus insulin trials for daily-administered basal insulin analogs, respectively. Based on phase 2 data, level 2 hypoglycemia rates with efsitora were also lower than that observed with daily basal insulin analogs; however, as the phase 3 program will investigate a different titration algorithm and formulation, it remains to be seen whether these results will be replicated.

CGM is a relatively recent technological development, so for most trials included in the review, SMBG was the primary method for hypoglycemia data collection. Consequently, the detection of non-severe hypoglycemia relied on the identification of symptoms and/or self-reported BG readings, which may have underestimated actual values. The European Medicines Agency, EASD and ADA have recently recommended that CGM data be used in conjunction with participant-reported hypoglycemia outcomes as part of regulatory trials for the approval of drugs.90 91 In RCTs, compared with SMBG, CGM systems have demonstrated greater sensitivity in detecting hypoglycemia,91–93 and when used to inform titration decisions, have led to reduced proportions of TBR (<3.9 mmol/L (<70 mg/dL) and <3.0 mmol/L (<54 mg/dL)) and number of hypoglycemic events.94 95 To date, CGM has been used in three once-weekly basal insulin studies (ONWARDS 1, 2 and 4)30 31 33 but it was not used for titration or insulin dose adjustments as the CGM data were blinded; future use of open CGM with titration algorithms may help to further reduce the occurrence of hypoglycemia with once-weekly basal insulin. The further use of CGM in conjunction with SMBG-based data in future trials, such as those for once-weekly basal icodec and efsitora, will provide greater clinical context and may help to identify populations at risk of hypoglycemia, facilitating the movement of clinical practice into personalized diabetes management.

This narrative review thoroughly collected data from late-phase RCTs with diverse and large sample sizes over a 22-year period. Most studies were open label with an evaluation period less than 6 months, and only 10 studies implemented a non-inferiority statistical framework. To enable more robust evaluations, based on recent FDA guidance, future studies should include an efficacy evaluation period of 6–12 months, be blinded (if possible), and include non-inferiority or superiority statistical frameworks with defined estimands.96–98 In addition, hypoglycemia definitions need to be standardized as much as possible. Nevertheless, the data collected here can be used to provide clinical perspective on the frequency of hypoglycemia event rates in ongoing and future clinical trials assessing the safety of basal insulin analogs. A meta-analysis was not conducted due to the considerable heterogeneity in trial protocols and hypoglycemia definitions. Accordingly, only observed trends were reported without the support of statistical rigor. Additionally, the hypoglycemia rates reported here may not directly apply to routine clinical practice as some populations with severe comorbidities may have been excluded, as is typical for RCTs. Trials may have also specifically excluded individuals prone to hypoglycemia, with prior severe hypoglycemia a common exclusion criterion. It was not possible to assess hypoglycemia rates by phase (ie, initial titration vs maintenance titration) as these were not consistently reported across studies. Similarly, not all outcomes of interest were consistently reported across studies. Finally, given the emphasis on RCT data, the generalizability of our results to real-world clinical practice may be limited.

Summary

Second-generation basal insulin analogs may reduce, but not eliminate, hypoglycemia risk. Hypoglycemia rates varied widely depending on the study population, study design, insulin titration targets, insulin dosages and diabetes duration. Overall, the frequency of event rates for level 2 and level 3 hypoglycemia in most treat-to-target RCTs testing basal insulin analogs as a basal-only insulin regimen ranged from 0.06 to 7.10 events/PYE; the hypoglycemia rate for basal-bolus regimens ranged from 2.4 to 13.6 events/PYE. These rate ranges may provide a useful reference when gaging hypoglycemia risk associated with once-weekly basal insulin analogs. Most notably, although more robust comparative data are needed, this review suggests that the hypoglycemia risk data reported so far on once-weekly basal insulins are consistent with or lower than those previously reported for daily insulin basal insulin analogs despite their longer duration of action.

Supplemental material

Data availability statement

All data relevant to the study are included in the article or uploaded as supplemental information.

Ethics statements

Patient consent for publication

Ethics approval

Not applicable.

Acknowledgments

Medical writing support was provided by Samuel Bestall of Oxford PharmaGenesis, Oxford, UK, funded by Novo Nordisk.

References

Footnotes

  • Contributors JR contributed to the design, literature search, analysis of the extracted data, and to the writing, reviewing and editing of the manuscript. HSB contributed to the design, literature search, analysis of the extracted data, and to the writing, reviewing and editing of the manuscript. IL contributed to the design, literature search, analysis of the extracted data, and to the writing, reviewing and editing of the manuscript. SRH contributed to the design, literature search, analysis of the extracted data, and to the writing, reviewing and editing of the manuscript.

  • Funding This study was funded by Novo Nordisk (not applicable).

  • Competing interests JR has served on advisory panels for Applied Therapeutics, Biomea Fusion, Boehringer Ingelheim, Eli Lilly, Endogenex, Hanmi, Novo Nordisk, Oramed, Regor, Sanofi, Scholar Rock, Structure Therapeutics, Terns Pharmaceuticals and Zealand Pharma; and has received research support, consulting fees or honoraria from Applied Therapeutics, AstraZeneca, Biomea Fusion, Boehringer Ingelheim, Eli Lilly, Endogenex, Hanmi, Novo Nordisk, Oramed, Regor, Sanofi, Scholar Rock, Structure Therapeutics, Terns Pharmaceuticals and Zealand Pharma. HSB reports trial fees paid to his institution by Amgen, AstraZeneca, Boehringer Ingelheim, Ceapro, Eli Lilly, Gilead, Janssen, Kowa Pharmaceuticals, Madrigal Pharmaceuticals, Merck, Novartis, Novo Nordisk, Pfizer, Sanofi, Ionis and Tricida. IL reports receiving grants, personal fees, or non-financial support from Sanofi, Lilly, Boehringer Ingelheim, Merck/Pfizer, Mylan, AstraZeneca, Johnson & Johnson, Intercept, Target Pharma, Zealand, Shionogi, Carmot, Structure, Bayer, Mediflix, WebMD, GI Dynamics, Intarcia Therapeutics, Mannkind, Novartis, Novo Nordisk, Structure Therapeutics and Valeritas. SRH reports consultancy with Eli Lilly, Zealand Pharma and Zucara Pharma; has participated in speaker panels for AstraZeneca, Medtronic and Novo Nordisk; and receives grant support from Dexcom.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.