• Wat is de huidige situatie?
• Beschrijf kort het knelpunt dat ten grondslag ligt aan de uitgangsvraag (bijv. praktijkvariatie, kosten, nieuwe therapie, acceptatie door patiënten)

Type 1 diabetes wordt gekenmerkt door (vrijwel) volledige afwezigheid van endogene insulineproductie door de betacellen in de eilandjes van Langerhans in de pancreas. Mensen met type 1 diabetes zijn daarom afhankelijk van insulinebehandeling, middels subcutane injecties meerdere malen per dag of middels een continue subcutane pomp. Ondanks intensieve insulinetherapie in combinatie met frequente glucosecontrole is optimale glucoseregulatie slechts voor een minderheid van de type 1 diabetespopulatie bereikbaar en zijn zowel hyper- als hypoglykemieën een dagelijkse realiteit.

Patiënten met type 1 diabetes die met multipele subcutane injecties worden behandeld gebruiken meerdere malen per dag een kortwerkende insuline-analoog, en daarnaast een langwerkende insuline-analoog die meestal 1x daags wordt geïnjecteerd. In de afgelopen jaren zijn er nieuwe langwerkende insuline-analogen beschikbaar gekomen. Deze uitgangsvraag gaat over de bijdrage van deze nieuwe insuline-analogen, die een ander werkingsprofiel hebben dan de al langer bestaande langwerkende insuline-analogen, aan de diabetesregulatie. Hierbij gaat het zowel om verbetering van de glucoseregulatie uitgedrukt in HbA1c, time in range (TIR) en hypoglykemieën, kwaliteit van leven, alsook om eventuele nadelige effecten. Aangezien de nieuwere langwerkende insuline-analogen in het algemeen duurder zijn dan de oudere, dient ook het kosten-aspect te worden beoordeeld

HbA1c levels (crucial outcome)

Severe hypoglycemia (crucial outcome)

Ketoacidosis (crucial outcome)

Health related quality of life (crucial outcome)

Hypoglycemia (important outcome)

Nocturnal hypoglycemia (important outcome)

Serious adverse events (important outcome)

Non-serious adverse events (important outcome)

Injection site reactions (important outcome)

All-cause mortality (important outcome)

Design

Hemmingsen (2021) performed a systematic review with meta-analysis of 26 randomized controlled trials with a duration of 24 weeks or more comparing one long-acting insulin to NPH insulin or another long-acting insulin in people with type 1 diabetes mellitus. The literature search was performed until 24 August 2020.

Participants

Non-pregnant people (adults and children) with type 1 diabetes were included.

Exposure definitions

In Hemmingsen (2021), the intervention was defined as long-term treatment with (ultra-)long-acting insulin analogues and compared with NPH insulin (neutral protamine Hagedorn) or another (ultra-)long-acting insulin analogue. In order to make the exposure definition of Hemmingsen (2021) compatible to the PICO, the exposure definitions were adapted as follows:

The intervention treatment was defined as 2nd generation long-acting insulin analogues: Insulin degludec or Insulin glargine U300.

The comparator treatment was defined as 1st generation long acting insulin analogues: Insulin glargine U100, Insulin detemir.

Comparisons included

2nd generation                               1st generation               

Insulin degludec              with       Insulin detemir               

Insulin degludec              with       Insulin glargine U100    

One study (Urakami, 2017) did not provide the concentration of glargine (neither in the SR of Hemmingsen (2021) nor in the original publication, nor in clinicaltrials.gov). In this trial the concentration of glargine was hand searched through other publications and it was found that glargine U100 was given (Danne 2020).

In Hemmingsen (2021), no studies on glargine U300 were reported. We have sent a mail to the authors to ask why these trials were not included. Their answer was that the trials on glargine U300 were compared with glargine U100 trials and that they did not include trials comparing same type of insulin in different concentration. Since another meta-analysis did compare glargine U300 with glargine U100 in type 1 diabetes (Danne 2020) and fits the PICO, this meta-analysis was added. Its design is described below.

Design

Danne (2020) performed a post hoc meta-analysis from three randomized, multicenter, 6-month similarly designed phase 3 trials in which insulin glargine U300 was compared to glargine U100 in type 1 diabetes.

Participants

People (adults and children) with type 1 diabetes were included.

Exposure definitions

The intervention treatment was defined as 2nd generation long-acting insulin analogue: Insulin glargine U300.

The comparator treatment was 1st generation long acting insulin analogue: Insulin glargine U100

The total follow-up in all 3 studies was 6 months.

Outcomes

The systematic review of Hemmingsen (2021) reported the following outcomes: HbA1c levels, severe hypoglycemia, hypoglycemia, nocturnal hypoglycemia, diabetic ketoacidosis (DKA), health-related quality of life, serious adverse effects, adverse events, socioeconomic effects, all-cause mortality

The post hoc meta-analysis of Danne (2020) reported the following outcomes per trial: HbA1c levels, severe hypoglycemia. The original trials that were meta-analyzed by Danne (Home 2015, Mathuhisa 2016 and Danne 2020) also reported on the outcomes, hypoglycemia, nocturnal hypoglycemia, serious adverse effects and all cause-mortality. These outcomes from the original trials were added as well.

Conflicts of interest

In the study from Hemmingsen (2021), the first author was funded by The Leona M. and Harry B. Helmsley Charitable Trust as part of the Addressing the Challenge and Constraints of Insulin Sources and Supply (ACCISS) Study. The study from Danne (2020) was sponsored by the manufacturer of glargine (Sanofi) and writing assistance was provided by Sanofi.

Study characteristics

In the pooled study from Hemmingsen (2021) and Danne (2020), a total of 3538 participants were randomized: 2001 participants were randomized to second generation long-acting insulin analogues and 1537 were randomized to 1st generation long-acting insulin analogues. Three of the studies included children only and randomized 831 children, i.e. 23% of all participants. The remaining six studies included adults only. The proportion of participants finishing the studies varied from 81% to 100%. One study had a cross-over design (Urakami 2017). The remaining studies were parallel-group RCTs. All studies had an open-label design. All studies except one were multicenter studies (Urakami 2017). All studies included both genders. The mean age of the participants varied from 10 to 54 years. The average duration of type 1 diabetes varied from 3.9 to 23.2 years. All studies had predefined HbA1c as the primary outcome.

Results

In the systematic review of Hemmingsen (2021) and post hoc meta-anlysis of Danne (2020) an interaction analysis showed that results were similar in adults and children, after which the results were pooled. This strategy is also followed in the current systematic review.  

HbA1c levels (critical outcome)

The mean difference in HbA1c levels at time of baseline and end of follow-up were expressed as percentages and were reported in 8 studies. The included studies showed a mean difference of 0.07% (95% CI 0.01 to 0.19), indicating a higher, but clinically non-important reduction of HbA1c in the control group compared to the intervention group. Effects were similar for adults and children. Lost to follow-up for this outcome was >14% in some of these studies (i.e., BEGIN Basal-Bolus type 1, BEGIN Flex T1, EDITION 4, EDITION JP)

Figure 1. Forest plot of comparison 2^nd generation vs 1^st generation long acting insulins, outcome HbA1c*

* The minus signs in the forest plot (see ‘mean’ HbA1C level for the studies Edition 4, Edition JP1, BEGIN Young and Edition Junior) denote mean differences within each treatment arm

Severe hypoglycemia (critical outcome)

In the systematic reviews of Hemmingsen (2021) and post hoc meta-analysis of Danne (2020), severe hypoglycemia was defined as: requiring assistance from another person (was planned to be further categorized into 'assistance from other persons', assistance from medical staff, intravenous glucose administration, subcutaneous glucagon administration, hospitalization, intensive-care unit stay, coma). Overall, no differences in severe hypoglycemia were found between intervention and control group (risk ratio 0.98; 95% CI, 0.78-1.24) . Results were similar for adults and children. There was inconsistency between the two meta-analyses, where the three glargine U300 trials combined (reported by Danne, 2020) showed a lower risk for severe hypoglycemia for second generation long-acting insulin analogues (pooled risk ratio 0.67; 95% CI, 0.67-0.99), while the other trials (reported by Hemmingsen, 2021) showed no difference for severe hypoglycemia (pooled risk ratio 1.19; 95% CI, 0.91-1.56).

Figure 2. Forest plot of comparison 2^nd generation vs 1^st generation long acting insulins, outcome severe hypoglycemia

Diabetic ketoacidosis (critical outcome)

In the systematic review of Hemmingsen (2021), diabetic ketoacidosis was defined as potentially life-threatening condition with high levels of ketones in the body which when building up in the blood make the blood more acidic. This outcome was reported in 5 studies. Only few events were noted (n=8) making the results imprecise as shown by the large confidence intervals. Due to small numbers and imprecision the results were not pooled in the Forest plot.

Figure 3. Forest plot of comparison 2^nd generation vs 1^st generation long acting insulins, outcome diabetic ketoacidosis

Health-related quality of life (critical outcome)

In the systematic review of Hemmingsen (2021), health-related quality of life was defined as mental and physical health-related quality of life and evaluated by a validated instrument: Short-Form-36 (higher values mean better health related quality of life)). The 3 included studies showed a similar physical health score in those randomized to second generation versus first generation long-acting insulin analogues (mean difference -0.27; 95% CI, -1.08 to 0.53). The 3 studies also showed a similar mental health score in those randomized to second generation versus first generation long-acting insulin analogues (mean difference -0.98; 95% CI, -2.93 to 0.97). The results on mental health score were inconsistent, i.e., Davies (2014) found a statistically significant effect where mental health was better in first generation long-acting insulin analogues, but the effect was small (-3.00; 95% CI, -4.44 to -1.56) and did not reach the threshold of clinical significance (0.5*SD). Results from SWITCH 1 concerned unpublished data.

Figure 4. Forest plot of comparison 2^nd generation vs 1^st generation long acting insulins, outcome health related quality of life

Hypoglycemia (important outcome)

In the systematic review of Hemmingsen (2021) hypoglycemia was defined as: hypoglycemic episodes not requiring assistance from another person. In the 3 trials that were reported by Danne (2020), hypoglycemia was reported as symptomatic hypoglycemic episode plus a glucose measurement of < 3.9 mmol/L (< 70 mg/dL ). All 8 included studies showed a consistent similar risk of hypoglycemia in those randomized to second generation versus first generation long-acting insulin analogues (pooled risk ratio 1.01; 95% CI, 1.00-1.03).

Figure 5. Forest plot of comparison 2^nd generation vs 1^st generation long acting insulins, outcome hypoglycemia

Nocturnal hypoglycemia (important outcome)

In the systematic review of Hemmingsen (2021) and the 3 trials that were reported by Danne (2020), nocturnal hypoglycemia was defined as hypoglycemia during night-time and defined as reported in studies. All 8 included studies showed no increased risk of nocturnal hypoglycemia in those randomized to second generation versus first generation long-acting insulin analogues (risk ratio 1.01; 95% CI, 0.95-1.17). There was some inconsistency as the results of Edition JP1 showed a protective effect of nocturnal hypoglycemia in those randomized to second generation long-acting insulin analogues (risk ratio 0.77; 95% CI, 0.62-0.95).

Figure 6. Forest plot of comparison 2^nd generation vs 1^st generation long acting insulins, outcome nocturnal hypoglycemia

Socioeconomic effects (important outcome)

In the systematic review of Hemmingsen (2021), socioeconomic effects were defined as direct costs defined as a consequence of admission or readmission rates; average length of stay; visits to general practitioner; accident or emergency visits; medication consumption; indirect costs defined as resources lost due to illness by the participant or their family member.

No studies reported the costs of the intervention during the study period.

Serious adverse events (important outcome)

In the systematic review of Hemmingsen (2021) and the 3 trials that were reported by Danne (2020), serious adverse events (SAE) were defined according to the International Conference on Harmonization (ICH) guidelines as: any event that leads to death, that is life-threatening, required in-patient hospitalization or prolongation of existing hospitalization, resulted in persistent or significant disability, and any important medical event which may have had jeopardized the patient or required intervention to prevent it (ICH 1997) or as reported in studies. Overall no difference was seen in SAEs between intervention and control group (risk ratio 1.04; 95% CI, 0.83-1.31), but there was imprecision as shown by the large pooled confidence interval.

Figure 7. Forest plot of comparison 2^nd generation vs 1^st generation long acting insulins, outcome serious adverse events

Non-serious adverse events (important outcome)

In the systematic review of Hemmingsen (2021), non-serious adverse events were defined as all adverse events, not classified as SAEs. In the 3 trials that were reported by Danne (2020), non-serious adverse events were defined as all adverse events that were experienced in > 5.0% of participants. Overall no difference was seen in non-serious adverse events between intervention and control group (risk ratio 1.02; 95% CI, 0.98-1.06).

Figure 8. Forest plot of comparison 2^nd generation vs 1^st generation long acting insulins, outcome non-serious adverse events

Injection site reactions (important outcome)

The outcome ‘injection site reaction’ was not reported in the systematic reviews of Hemmingsen (2021) and Danne (2020). Therefore, we searched in the original 8 studies that were reported by Hemmingsen (2021) and Danne (2020) to see if they reported injection site reactions. This outcome was reported by all 5 studies in adults (BEGIN Basal-Bolus type 1, Begin Flex T1, Davies (2014), Edition 4, Edition JP1). The outcome ‘injection site reaction’ was not reported in the 3 studies in children (BEGIN Young, Edition Junior, Urakami (2017)). Overall no difference was seen in injection site reactions between intervention and control group (risk ratio 0,95; 95% CI, 0.49-1.84), but the result was imprecise as shown by the pooled large 95% confidence interval.

Figure 9. Forest plot of comparison 2^nd generation vs 1^st generation long acting insulins, outcome injection site reactions

All-cause mortality (important outcome)

All-cause mortality was defined as death from any cause. Only few events were noted (n=6) and the follow-up of the studies to report on this outcome was relatively short (6 months to 2 years). Due to small numbers and imprecision the results were not pooled in the Forest plot.

Figure 10. Forest plot of comparison 2^nd generation vs 1^st generation long acting insulins, outcome all-cause mortality

Level of evidence of the literature

The level of evidence (GRADE method) is determined per comparison and outcome measure and is based on results from systematic review of randomized trials and therefore starts at level “high”. Subsequently, the level of evidence was downgraded if there were relevant shortcomings in one of the several GRADE domains: risk of bias, inconsistency, indirectness, imprecision, and publication bias.

The level of evidence regarding the outcome measure HbA1c levels was downgraded by 1 level because of risk of bias (1 level, incomplete outcome data).

The level of evidence regarding the outcome measure severe hypoglycemia was downgraded by 1 level because of inconsistency).

The level of evidence regarding the outcome measure diabetic ketoacidosis was downgraded by 2 levels because of imprecision (2 levels, < 10 events in all 5 included studies, in total only 8 events)

The level of evidence regarding the outcome measure health related quality of life was downgraded by 1 level because of inconsistency.

The level of evidence regarding the outcome measure hypoglycemia was downgraded by 1 level because of risk of bias (1 level outcome event somewhat subjective).

The level of evidence regarding the outcome measure nocturnal hypoglycemia was downgraded by 1 level because of risk of bias (1 level outcome event somewhat subjective). Although one trial result was inconsistent to the pooled effect, this was not taken into account in further down grading the evidence, as it was only one trial, and results were adjusted for random effects.

The level of evidence regarding the outcome measure serious adverse event was downgraded by 1 level because of imprecision (1 level, large pooled confidence interval could not rule out a harmful effect).

The level of evidence regarding the outcome measure non-serious adverse event was downgraded by 1 level because of bias (1 level outcome event somewhat subjective).

The level of evidence regarding the outcome measure injection site reaction was downgraded by 2 levels because of imprecision (1 level, large pooled confidence interval could not rule out a protective [or harmful] effect) and indirectness (1 level, outcome not studied in children).

The level of evidence regarding the outcome measure all-cause mortality was ungraded due to lack of data.

A systematic review of the literature was performed to answer the following question:

What are the (un)favorable effects of 2nd generation long-acting insulin analogues compared to 1st generation long-acting insulin analogues in patients with Diabetes mellitus type 1?

PICO

P (Patients)                                      : Children and adults with (incident) type 1 diabetes mellitus

I (Intervention)                              : 2nd generation long-acting insulin analogues: Insulin degludec, Insulin glargine U300

C (Comparison)                              : 1st generation long acting insulin analogues: Insulin, Insulin detemir, Insulin glargine U100

O (Outcomes):                               : HbA1c levels, severe hypoglycemia, hypoglycemia, nocturnal hypoglycemia, diabetic ketoacidosis (DKA), time within the target range of blood glucose, time above the target range of blood glucose, time below the target range of blood glucose, health-related quality of life, patient satisfaction (with treatment), serious adverse effects, adverse events/injection site reactions, socioeconomic effects, all-cause mortality.

Relevant outcome measures

The guideline development group a-priori considered HbA1c, severe hypoglycemia, DKA, time within normal range of blood glucose, time above normal range, time below normal range and health related quality of life as critical outcomes for decision making.

Important outcome measures included hypoglycemia, nocturnal hypoglycemia, patient satisfaction (with treatment), serious adverse effects, adverse events/injection site reactions, socioeconomic effects and all-cause mortality

The working group did not define the outcome measures listed above but used the definitions used in the studies.

The working group did not define a minimal clinically (patient) important difference for all outcomes. Therefore, the default thresholds proposed by the international GRADE working group were used: a 25% difference in relative risk (RR< 0.8 or RR>1.25) for dichotomous outcomes, and 0.5 standard deviations (SD) for continuous outcomes.

Search and select (Methods)

The databases Medline (via OVID) and Embase (via Embase.com) were searched with relevant search terms until 14 July 2021. This search strategy can be found in the Methods section. The detailed search strategy is depicted under the tab Methods. The systematic literature search resulted in 607 hits, including 247 systematic reviews and 360 randomized controlled trials (RCTs)

A total of 31 articles were initially selected based on title and abstract screening. After reading the full text, 29 studies were excluded (see the table with reason for exclusion under the tab Methods), and 2 studies were included.

Results

Two meta-analyses were included in the analysis of the literature. Important study characteristics and results are summarized in the evidence tables. The assessment of the risk of bias is summarized in the risk of bias tables.