Frequency of hypoglycaemia with basal insulin treatments in adults with type 1 diabetes treated with basal‐bolus insulin regimens in treat‐to‐target trials: A narrative review

To summarise, in a narrative review, published data on hypoglycaemia occurrence with basal insulin therapy in adults with type 1 diabetes treated with basal‐bolus insulin regimens in treat‐to‐target randomised controlled trials.


| INTRODUCTION
Individuals with type 1 diabetes require exogenous insulin therapy, which can be delivered via a combination of shortacting and long-acting insulin injections in a basal-bolus regimen, otherwise known as a multiple daily injection regimen.Alternatively, a continuous insulin infusion regimen may be used, in which short-acting insulin is continuously delivered by an insulin pump with or without sensor-driven automated insulin delivery. 1Attention must be paid to the dose, formulation and timing of exogenous insulin delivery to avoid hyperinsulinaemia and subsequent hypoglycaemia. 2,3urrent guidelines from the American Diabetes Association (ADA) and the European Association for the Study of Diabetes (EASD) categorise hypoglycaemia as follows: level 1, encompassing measured glucose concentrations of 3.0-<3.9mmol/L (54-<70 mg/dL); level 2, encompassing measured glucose concentrations of less than 3.0 mmol/L; and level 3, encompassing severe events with altered mental and/or physical status requiring assistance for treatment. 4,5With increasing hypoglycaemia severity (from level 1 to level 3), the negative physiological effects worsen from symptoms of confusion to loss of consciousness and, very rarely, even death. 4,6Hypoglycaemia also negatively affects quality of life, with adverse effects on social, personal and professional life. 7,8onsideration of hypoglycaemia risk, particularly for level 2 and level 3 events, is clearly important when evaluating treatment benefits and risks for individuals with type 1 diabetes.For example, some evidence suggests that the second-generation basal insulin analogues insulin glargine U300 (glargine U300) and insulin degludec (degludec) have a lower risk of hypoglycaemia than the first-generation basal insulin analogues insulin glargine U100 (glargine U100) or insulin detemir (detemir).This lower risk may be due to the second-generation insulin analogues having longer durations of action and flatter insulin pharmacokinetic and pharmacodynamic profiles than the first-generation insulin analogues. 9,10ypoglycaemia remains a key risk associated with insulin treatment.Two once-weekly basal insulins, insulin icodec and basal insulin Fc (BIF; insulin efsitora alfa), are in development for the treatment of type 1 and type 2 diabetes.Based on Phase II data, these treatments may have a similar risk of hypoglycaemia and similar or improved glucose-lowering profiles to available basal insulins that require once-or twice-daily administration.2][13][14] Both treatments are in late-stage clinical development; a Phase III trial programme is underway to examine the efficacy and safety of insulin icodec as part of a basal-bolus regimen in individuals with type 1 diabetes. 15derstanding the degree of hypoglycaemia occurrence with available basal insulins administered as part of a basal-bolus regimen in individuals with type 1 diabetes would provide valuable context to evaluate data from trials of once-weekly basal insulins.Furthermore, summarising published data on hypoglycaemia occurrence by therapeutic formulation could inform treatment guidelines and clinical practice.This summary should evaluate the challenges in comparing hypoglycaemia rates across randomised controlled trials (RCTs) due to differences in study methodologies and hypoglycaemia definitions, reflecting changes in the understanding of hypoglycaemia and in its methods of detection over time.

| Objective of this narrative review
To help contextualise the emerging hypoglycaemic data from once-weekly basal insulin trials, a narrative review was conducted to summarise published data and describe trends on the occurrence of hypoglycaemia with different basal insulin formulations in adults with type 1 diabetes treated with basal-bolus insulin regimens in treat-to-target

What's new?
What is already known?
• Hypoglycaemia has negative physiological and quality-of-life effects on individuals with type 1 diabetes.

What this study found?
• Based on reported rates, hypoglycaemia risk remains a key clinical challenge in type 1 diabetes.• Continuous glucose monitoring-based rates were numerically, yet consistently, higher than self-measured blood or plasma glucose rates.
• Differences across studies in design and participant characteristics hindered comparison by insulin formulation.
What are the implications of the study?
• Summaries of published hypoglycaemia data provide valuable context to interpret the results of trials for once-weekly basal insulins.• Future research should examine variation in hypoglycaemia risk by insulin formulation and individuals' characteristics, and assess the clinical impact of continuous glucose monitoring usage.
RCTs in which insulin dosages were adjusted to meet a pre-specified treatment target.The heterogeneity in study designs and hypoglycaemia definitions among studies precluded meta-analysis; subsequently, available data were categorised by hypoglycaemia definition and summarised narratively.Importantly, the narrative nature of this review precluded any comparative statistical analyses.All authors contributed to the scope and methodological approach of the review, as well as to interpretation of the data and drafting of the manuscript.

| Trial identification and data extraction
Relevant articles were identified by searching the MEDLINE and Embase databases on 15 November 2022 using search terms relating to type 1 diabetes, basal-bolus regimens (i.e., basal, bolus and specific insulin treatments used in basalbolus regimens) and hypoglycaemia.Included articles were those reporting the rate or proportion of individuals experiencing level 2 or level 3 hypoglycaemia (including nocturnal hypoglycaemia) based on treat-to-target RCTs of basal insulin treatments in adults with type 1 diabetes treated with basal-bolus insulin regimens.Insulin formulations of interest were the second-generation insulin analogues degludec and glargine U300, the first-generation insulin analogues glargine U100 and detemir and neutral protamine Hagedorn (NPH) insulin.Data from trials of NPH were included to ensure a comprehensive overview of hypoglycaemia.Multiple articles on the same study, including articles relating to trial extensions, were included if each article included unique information on hypoglycaemia occurrence.Excluded publications were scientific abstracts, articles published before 2012 or reporting on studies in fewer than 20 individuals and publications with a focus on type 2 diabetes, bolus insulins, insulin pumps, mixed insulins or insulin biosimilars.
After excluding ineligible articles by single screening of the abstracts and full texts, relevant data were extracted from eligible articles (n = 21).Extracted data were study methods (including study design, basal and bolus insulin interventions and titration targets), demographic and baseline characteristics, on-trial glycated haemoglobin (HbA 1c ) reduction, insulin dosages and level 2 and level 3 hypoglycaemia data (including definitions, proportion affected and rates).If a single publication reported hypoglycaemia data from multiple time periods, the narrative review synthesis prioritised data from the longest time period.
7][18][19] Data from these articles are discussed separately in this review because the participant characteristics and hypoglycaemia risk may not represent the general population with type 1 diabetes.High hypoglycaemia risk was defined in the studies as: having one or more hypoglycaemia risk factors (at least one level 3 hypoglycaemic episode in the previous year, moderate chronic renal failure, hypoglycaemia unawareness, diabetes duration of over 15 years or an episode of hypoglycaemia of any severity in the previous 12 weeks) (SWITCH 1 study) 18 ; having had at least two episodes of level 3 hypoglycaemia in the preceding year (HypoAna study) 16,19 ; or having had at least one episode of nocturnal level 3 hypoglycaemia in the preceding 2 years (HypoDeg study). 10,172.2 | Overview of articles

| Study characteristics
Table S1 provides key details on the study designs for the 21 included articles.Across all articles, the most frequently reported data were those on glargine U100 (n = 16 articles), followed by degludec (n = 10), glargine U300 (n = 7), detemir (n = 3), an insulin analogue regimen of detemir and insulin aspart (n = 2) and a human insulin regimen of NPH insulin and regular human insulin (n = 2).1][22] Of the articles on studies limited to high-risk individuals, three articles reported on glargine U100 and degludec, and two articles reported on an insulin analogue regimen and a human insulin regimen.
All insulins were administered once daily, except for detemir in three studies, which was administered either once-or twice-daily. 16,19,23Various titration targets, based on timing and glucose thresholds, were used across the studies.For fasting or preprandial measures, target plasma glucose ranges varied from 3.9-4.9mmol/L to 4.4-7.2mmol/L (Table S1).

| Participant characteristics
When reported, baseline individual characteristics, along with end-of-treatment HbA 1c change from baseline and insulin dosage, are provided in Table S2.Excluding articles on studies in high-risk individuals, mean age was 25.9-48.2years, 20,24 mean diabetes duration was 24 years or less and mean baseline body mass index (BMI) was 23-28 kg/m. 2 When reported, mean baseline HbA 1c was 57-68 mmol/mol (7.4%-8.4%).HbA 1c change from baseline to the end of treatment ranged from approximately 1 mmol/mol (0.1%-points) to approximately 10 mmol/mol (0.9%-points).Basal insulin dosage generally increased from baseline (0.29-0.38 U kg −1 day −1 ) to the end of each study (0.29-0.48 U kg −1 day −1 ).Bolus insulin dosages also generally increased from baseline to the end of each study, although less consistently than basal insulin dosages.Several studies did not report insulin dosages at baseline, at the end of the study or both.
All articles reporting CGM usage were published in or after 2018, whereas articles reporting SMBG/SMPG usage were usually published in or before 2018.For level 2 hypoglycaemia, articles reporting based on a glucose threshold of 3.1 mmol/L tended to be published before those using a glucose threshold of 3.0 mmol/L.Level 3 hypoglycaemia was the most frequently reported definition (18 articles).
Trends indicated that higher end-of-treatment HbA 1c values (Figure 2a), broader titration targets (Figure 2b) and higher end-of-treatment glucose levels (Figure 2c) may be linked with lower level 2 hypoglycaemia rates; however, the almost-complete separation of these factors by insulin formulation, the relatively low number of studies with relevant data on end-of-treatment HbA 1c or glucose levels and the lack of statistical analyses hindered these comparisons.Therefore, it was unclear whether the lower hypoglycaemia rates observed with glargine (U100 or U300) versus degludec and detemir were truly related to HbA 1c level, titration target or glucose level.
Based on observed trends, when grouped by insulin formulation, hypoglycaemia rates reported with glargine U100 showed greater variation than those with other insulins.Additionally, relatively high rates were reported for two studies examining glargine U100 and one study examining degludec, all relating to rates of level 2 hypoglycaemia that included level 3 events.The high rates were reported in two separate studies examining glargine U100 and degludec.Both treatment arms had high observed hypoglycaemia rates in the BEGIN: Flex T1 extension, a 52week, open-label Phase III study. 31High hypoglycaemia rates were observed with glargine U100, but not degludec, in the Home 2012 study, a sub-study of a 16-week, openlabel Phase II study. 29There was no clear reason for the high observed rates in these two articles in terms of study methods or participant characteristics, including SMBG measurement frequency.
In high-risk individuals receiving either glargine U100 or degludec, reported rates were 15.01-16.31events/PYE in the HypoDeg study and 20.446-21.684events/PYE in the SWITCH 1 study. 10,18These rates were similar to or lower than those observed in non-high-risk individuals (Table S4).
One study reported no level 3 hypoglycaemic events with either glargine U100 or detemir, which may be linked to the short study duration (approximately 5 weeks). 20s per the level 2 events, reported hypoglycaemia rates with glargine U100 had the widest range of all insulin formulations.No other trends were observed by insulin formulation.
3.3 | All-day hypoglycaemia measured in studies using CGM Level 2 Data on CGM-derived level 2 hypoglycaemia in two articles were based on high-risk individuals. 16,17The overall rate range was 5-182 events/PYE (events/PYE: study total, 5-73; glargine U100, 16-182; degludec, 10-120; insulin analogue regimen [detemir and aspart], 10-62; human insulin regimen [NPH and regular insulin], 5-83).Symptomatic event rates were 6-17 times lower than rates of events with or without symptoms and were similar to symptomatic rates observed with SMBG or SMPG.Rates of events with or without symptoms were higher in studies using CGM than those using SMBG or SMPG.Proportion of participants experiencing hypoglycaemia was not reported (Figure 3; Table S4).These findings aligned with a separate study, the InRange trial, which was not limited to high-risk individuals; in this study, CGM all-day level 2 rates were 2-6 times higher than SMPG rates. 35ates with glargine U100 were higher than rates with other insulins; this could not easily be explained by within-study or between-study differences in the different treatment groups, or by study methods.

| Level 3
Rates of all-day level 3 hypoglycaemia were not reported in any article that utilised CGM.
One potential outlier was identified from a sub-study of an open-label, Phase II clinical trial, in which the reported rate with glargine U100 was 12.3 events/PYE when including level 3 (severe) events. 29Rates in the degludec arm (5.1 events/PYE) were similar to those of other studies of degludec.There was no clear reason for the high glargine U100 rates observed in this study based on study methods or participant characteristics.
Glargine U100 had the widest rate range, which may reflect the higher number of studies using this insulin than other insulins.No other differences by insulin formulation were apparent.
In high-risk individuals from the SWITCH 1 and HypoDeg, studies reported hypoglycaemia rates were similar to those observed in non-high-risk individuals, with an overall range of 1.68-6.03events/PYE for glargine U100 and degludec 10,18 (Table S5).In the HypoDeg study, rates were similar for the intention-to-treat (ITT) and per-protocol (PP) analyses 10 ; corresponding information F I G U R E 3 All-day (24-h) level 2 (relating to thresholds of 3.0 mmol/L or 3.1 mmol/L) hypoglycaemia rates measured using CGM.All studies were limited to individuals at high risk of hypoglycaemia.CGM, continuous glucose monitoring; IAsp, insulin aspart; IDeg, insulin degludec; IDet, insulin detemir; IGlar U100, insulin glargine U100; NPH, neutral protamine Hagedorn; PYE, patient-year of exposure.for both ITT and PP analyses was not provided in other articles.

| Level 3
32][33]35 The overall range of rates was 0.01-0.14events/PYE (events/PYE [n estimates]: glargine U100, 0.01-0.14[4]; glargine U300, 0.03-0.1 [4]; degludec, 0.05-0.1 [3]; detemir, 0.08 [1]).Predictably, rates of level 3 events were lower than rates of level 2 events.][32][33][34][35] In the BEGIN: Flex T1 study, the hypoglycaemia rate was high in the glargine U100 arm compared with other studies of the same insulin.Meanwhile, the degludec arm had lower rates than those observed in other studies. 31Study methodology and participant characteristics could not account for the difference in rates between this study and other studies.Given the sparseness of the data, no conclusions could be drawn about the patterns of observed rates by insulin formulation.
In the Pedersen-Bjergaard 2022 article (HypoDeg study), 10 nocturnal level 3 hypoglycaemia rates were higher in high-risk individuals than non-high-risk individuals (0.26, 0.4 and 0.13 events/PYE for study total, glargine U100 and degludec, respectively) (Table S5).In this article, rates were similar for the ITT and PP analyses 10 ; corresponding information for both ITT and PP analyses was not provided in other articles.
Symptomatic event rates were 6-14 times lower than rates of events with or without symptoms.Rates were similar across insulin formulations, except that those for glargine U100 tended to be higher than those for other insulins.As per the all-day rates, the InRange trial provided CGM rates that were not limited to high-risk individuals; in this study, CGM nocturnal level 2 rates were approximately six times higher than SMPG rates. 35 3.5.2 | Level 3 Rates of nocturnal level 3 hypoglycaemia were not reported in any articles that utilised CGM.

| DISCUSSION
Based on the 21 articles identified in this review, managing hypoglycaemia risk remains a key challenge associated with exogenous insulin treatment.Up to 98% of individuals with type 1 diabetes experience level 2 hypoglycaemia when receiving a basal-bolus insulin regimen, with rates generally exceeding 10 events/PYE.Up to 13.4% of individuals experience level 3 hypoglycaemia (requiring external assistance for recovery), with rates of 0.0-0.4events/PYE; nocturnal hypoglycaemia rates were generally 10-30% of the reported all-day hypoglycaemia rates.
Between-study differences in study designs (e.g., titration targets) and participant characteristics hindered comparisons of hypoglycaemia rates by insulin tion.Few consistent patterns were observed in relation to these factors and hypoglycaemia rates, except that individuals with a history of level 3 hypoglycaemia (all-day or nocturnal) had higher rates of level 3 hypoglycaemia than individuals without this history.This association could reflect hypoglycaemia unawareness in individuals with a history of hypoglycaemia. 36Alternatively, this finding may reflect the known difficulties with managing hypoglycaemia risk as age and diabetes duration increase, 4,37 given that studies in individuals with a history of level 3 hypoglycaemia (all-day or nocturnal) generally reported higher mean ages and longer diabetes durations than studies not limited to individuals with this history.The paucity of comparative effectiveness studies hampered the assessment of hypoglycaemia rates according to insulin formulation and individuals' characteristics.This lack of findings requires further exploration given the 2023 ADA guidelines, which recommend tailoring diabetes treatment to each individual's needs. 1 Further research to compare multiple treatments and titration targets using a consistent methodology could overcome the study design heterogeneity issues encountered in this review.Alongside this, real-world studies of basal insulin treatments could facilitate exploration of individuals' characteristics in more diverse populations than those generally included in RCTs.These strategies could identify techniques to individualise basal-bolus insulin regimens and titration targets.
Across the studies, the range of hypoglycaemia rates with glargine U100 was consistently larger than that with other insulin formulations.Individual studies in people with different hypoglycaemia risk background including SWITCH 1 (≥1 hypoglycaemia risk factor) and EDITION JP 1 (no additional hypoglycaemia risk factors), statistically support a relatively higher rate or incidence of level 2 or level 3 hypoglycaemia, respectively, with glargine U100 against comparator basal insulins (degludec and glargine U300, respectively). 18,33dditionally, some studies of glargine U100 reported particularly high hypoglycaemia rates.These findings may be solely due to the many studies that examined glargine U100 or may reflect a pharmacokinetic or pharmacodynamic property of glargine U100 that leads to high levels of variation in hypoglycaemia rates.A similar trend was not noted in the few studies examining detemir, another first-generation basal insulin analogue: this contrasts with previous studies suggesting that risk of hypoglycaemia is higher with first-generation than with second-generation basal insulin analogues. 9,10otably, all included glargine U100 studies involved once-daily administration, whereas three of the five studies of detemir permitted once-daily or twice-daily administration.
Except for glargine U100, there were few consistent patterns in hypoglycaemia rates by insulin formulation.All-day level 2 hypoglycaemia rates with degludec and detemir were higher than with other insulin formulations; however, this trend was not consistent across all hypoglycaemia categories.Importantly, our review only comments on observed trends and therefore cannot support any differences between basal insulin formulations with any statistical rigor.Further work to harmonise study methods and reporting could help to confirm any differences in hypoglycaemia rates across different insulin formulations.Although out of scope for the current review, it would also be informative to assess the relative contributions of basal and bolus insulin to daytime hypoglycaemia across different basal insulin formulations.Given the difficulty in comparing hypoglycaemia rates in this review owing to the evolution in hypoglycaemia definitions over time, clear and consistent reporting of hypoglycaemia is necessary to enable comparisons.9][40] A similar issue could arise when considering events with signs or symptoms of hypoglycaemia without confirmation of blood glucose level.Ideally, studies should report data on all clinically relevant hypoglycaemic events, with consideration of additional data reporting to contextualise these rates and to facilitate comparisons with earlier studies.
Along with the glucose threshold used to define hypoglycaemia, the assessment method is important.Reported hypoglycaemia rates were based on SMBG/ SMPG for nearly all articles in this review, with only three articles reporting CGM-based rates.Although widely used, self-reporting may be limited by several factors including human error (e.g., incorrect meter use and storage), meter design (e.g., accuracy) and the environment (e.g., temperature).The use of CGM overcomes some of the drawbacks of SMBG. 41Indeed, hypoglycaemia rates were numerically, yet consistently, higher with CGM than with SMBG/SMPG, which reflects the greater sensitivity of CGM to detect hypoglycaemic events, particularly if asymptomatic. 16,17,34However, it should be noted that the limited number of studies utilising CGM resulted in a wide range of hypoglycaemic events being reported, limiting the conclusions that can be drawn from these data.Additionally, caution is warranted when using CGM, given the risk of overdetection, particularly when relatively broad glucose thresholds are used to define hypoglycaemia.This be particularly relevant when measuring 'alert' level hypoglycaemic episodes, defined as episodes with blood glucose values between 3.0 and 3.9 mmol/L that are not considered clinically significant nor of major regulatory relevance. 42However, individuals with type 1 diabetes could potentially use CGM to adjust their treatment proactively to reduce time spent in hypoglycaemia through early detection and management of falling blood glucose levels. 4,43,44With CGM use becoming more prominent in trials and as standard of care in clinical practice, the advantages and disadvantages of CGM should be carefully explored and communicated to patients and clinicians. 5,44longside CGM use, automated insulin delivery through continuous insulin infusion is becoming more prevalent and changing the treatment landscape for people with T1D. 5 The present review provides valuable context for guiding clinical decisions around basal-bolus insulin therapy, which is currently widespread and expected to remain so for the foreseeable future given the variability in access to automated insulin delivery.Further research into the risk of hypoglycaemia across insulin delivery approaches, including how risk varies by individuals' characteristics (e.g., presence of comorbidities, BMI, age, glycaemic control and hypoglycaemia history), may provide valuable information to facilitate individualisation of treatment for T1D, particularly as the treatment landscape for T1D evolves.

| CONCLUSIONS
Despite advancements in diabetes therapy, 45,46 there are still difficulties in managing and understanding hypoglycaemia risk.Improvements in CGM may help to identify and to prevent hypoglycaemia, particularly through detection and treatment of asymptomatic level 1 or level 2 hypoglycaemic events; however, the risks of over-detection and consequent overtreatment must be considered.Few trends were identified by insulin formulation, study methodology or individuals' characteristics.Comparisons of multiple treatments and titration targets with consistent study methodology, as well as real-world studies of basal insulins in diverse populations, may provide more informative outputs to assess these factors than reported to date.Furthermore, improved consistency in the reporting of hypoglycaemia rates, both in terms of the definitions used and reporting of all clinically relevant hypoglycaemic events, may aid comparisons between insulin formulations and with earlier studies.
The findings of this review help to demonstrate the risk of hypoglycaemia with available basal insulin therapies and to provide context for the results of ongoing and future clinical trials, including those for two once-weekly basal insulins, insulin icodec and basal insulin Fc.