Safety and efficacy of insulin glargine 300 u/mL compared with other basal insulin therapies in patients with type 2 diabetes mellitus: a network meta-analysis.

OBJECTIVE: To compare the efficacy and safety of a concentrated formulation of insulin glargine (Gla-300) with other basal insulin therapies in patients with type 2 diabetes mellitus (T2DM).
DESIGN: This was a network meta-analysis (NMA) of randomised clinical trials of basal insulin therapy in T2DM identified via a systematic literature review of Cochrane library databases, MEDLINE and MEDLINE In-Process, EMBASE and PsycINFO. OUTCOME MEASURES: Changes in HbA1c (%) and body weight, and rates of nocturnal and documented symptomatic hypoglycaemia were assessed.
RESULTS: 41 studies were included; 25 studies comprised the main analysis population: patients on basal insulin-supported oral therapy (BOT). Change in glycated haemoglobin (HbA1c) was comparable between Gla-300 and detemir (difference: -0.08; 95% credible interval (CrI): -0.40 to 0.24), neutral protamine Hagedorn (NPH; 0.01; -0.28 to 0.32), degludec (-0.12; -0.42 to 0.20) and premixed insulin (0.26; -0.04 to 0.58). Change in body weight was comparable between Gla-300 and detemir (0.69; -0.31 to 1.71), NPH (-0.76; -1.75 to 0.21) and degludec (-0.63; -1.63 to 0.35), but significantly lower compared with premixed insulin (-1.83; -2.85 to -0.75). Gla-300 was associated with a significantly lower nocturnal hypoglycaemia rate versus NPH (risk ratio: 0.18; 95% CrI: 0.05 to 0.55) and premixed insulin (0.36; 0.14 to 0.94); no significant differences were noted in Gla-300 versus detemir (0.52; 0.19 to 1.36) and degludec (0.66; 0.28 to 1.50). Differences in documented symptomatic hypoglycaemia rates of Gla-300 versus detemir (0.63; 0.19 to 2.00), NPH (0.66; 0.27 to 1.49) and degludec (0.55; 0.23 to 1.34) were not significant. Extensive sensitivity analyses supported the robustness of these findings.
CONCLUSIONS: NMA comparisons are useful in the absence of direct randomised controlled data. This NMA suggests that Gla-300 is also associated with a significantly lower risk of nocturnal hypoglycaemia compared with NPH and premixed insulin, with glycaemic control comparable to available basal insulin comparators. Published by the BMJ Publishing Group Limited. For permission to use (where not already granted under a licence) please go to http://www.bmj.com/company/products-services/rights-and-licensing/

This is the first comprehensive literature review and network meta-analysis (NMA) summarising the available clinical trial literature on the clinical benefits of the newly approved basal insulin, Gla-300, and potential basal insulin comparators, and enabling comparisons between these therapies.

The systematic literature review was limited to only English language literature; while this is likely to include all major randomised clinical trials conducted for basal insulin therapy in type 2 diabetes mellitus (T2DM), it may exclude smaller studies with no publication in English.

The NMA was conducted in accordance with National Institute for Health and Care Excellence guidance and extensive sensitivity analyses were utilised to assess the robustness of the findings.

While NMA enables the synthesis of available clinical information, it is not a substitute for head-to-head clinical trials to compare therapies, and such trials should be encouraged and conducted.

Introduction

Worldwide, approximately 348.3 million people are living with type 2 diabetes mellitus (T2DM).1 2 As T2DM progresses, insulin therapy may be required to achieve glycaemic control. The 2015 ADA/EASD Position Statement on Managing Hyperglycemia in T2DM recommends initiating basal insulin in combination with oral therapy among the appropriate options for patients who are unable to achieve their glycated haemoglobin (HbA1c) target after 3 months of metformin monotherapy.3

Insulin glargine 300 u/mL (Gla-300) is a new basal insulin that has recently (2015) been approved by the European Commission and the US Food and Drug Administration. Gla-300 is a concentrated formulation of insulin glargine 100 u/mL (Gla-100), developed to produce a more flat and more prolonged pharmacokinetic and pharmacodynamic profile.4–6 Several randomised controlled clinical safety and efficacy trials comparing Gla-300 to Gla-100 have shown that Gla-300 achieves reduction in HbA1c comparable to that of Gla-100, while lowering the risk of hypoglycaemia.6–8 Comparable HbA1c reduction is expected given that each treatment group utilised the same dose titration to achieve fasting plasma glucose of 4.4–5.6 mmol/L (ie, treat-to-target approach). The lower hypoglycaemia rates observed with Gla-300 may be due to properties inherent to the glargine molecule that lead to pharmacokinetic and pharmacodynamic differences at varying concentrations (ie, between Gla-300 and Gla-100).4 5

At the present time, head-to-head studies of Gla-300 with other available basal insulin options have not been conducted; however, such comparisons would help determine the place in therapy for this product. Meta-analysis enables the findings from multiple primary studies with comparable outcome measures to be combined.9 In absence of direct head-to-head clinical trials, mixed treatment meta-analysis (also known as network meta-analysis (NMA)) may be used to estimate comparative effects of multiple interventions using indirect evidence.9 The current report is an NMA conducted to indirectly compare the efficacy and safety of U300 versus available intermediate-acting to ultra-long-acting basal insulin formulations in the treatment of T2DM.

Methods

Systematic literature review

A systematic literature review was conducted to identify evidence for the clinical efficacy and safety of insulin regimens in T2DM according to National Institute for Health and Care Excellence (NICE) standards.9 The following electronic databases were searched: the Cochrane Library (eg, the Cochrane Central Register of Controlled Trials (CENTRAL) and the Database of Abstracts of Reviews of Effectiveness (DARE)), MEDLINE and MEDLINE In-Process (using Ovid platform), EMBASE (using Ovid Platform) and PsycINFO. Congresses searched were the European Association for the Study of Diabetes (EASD; 2011–2013), the American Diabetes Association (ADA; 2011–2013) and the International Diabetes Federation (IDF; 2011 and 2013). Key search terms included: ‘diabetes mellitus, type 2/’, ‘glargine’, ‘detemir’, ‘degludec’, ‘NPH’, ‘neutral protamine hagedorn’, ‘biphasic’, ‘aspart protamine’, ‘novomix’ and ‘premix’. Searches were limited to human, English-language only articles published from 1980 onwards. The NMA focused on studies published recently (ie, based on availability of basal insulin analogues). At the time of analysis, the Gla-300 vs Gla-100 studies were only available in clinical study reports; however, these studies have subsequently been published.6–8

Several quality control procedures were in place to ensure appropriate study selection and data extraction. Screening of abstracts and full-text was conducted by two independent researchers (a third independent researcher made a final determination for articles for which there was uncertainty). Data extraction was also conducted by two independent researchers (with reconciliation of discrepancies). Where available, full-text versions of the article were used for data extraction (an abstract or poster was not used unless it was the terminal source document). All processes were documented by the researchers and the data extraction file was also quality checked. The source materials (abstracts, full-text articles) and data extraction files were sorted, and saved on a secure server.

Inclusion criteria

In order to be considered for the NMA, clinical studies identified by the systematic literature review had to meet the following criteria: randomised active comparator-controlled clinical studies, patient population of adults with T2DM treated with basal insulin (with or without bolus), patients could be newly initiating insulin (naïve) or already exposed to insulin, and a minimum follow-up of 20 weeks. In addition, studies were required to have patients from at least one of the following countries: the USA, France, Germany, the UK, Spain and/or Italy.

Outcome measures

Outcome measures analysed by NMA included change in HbA1c (%) from baseline, change in body weight (kg) from baseline and rates of hypoglycaemic events (documented symptomatic and/or nocturnal) per patient year. A documented symptomatic event was defined as an event during which typical symptoms of hypoglycaemia were accompanied by measured plasma glucose under a threshold value. In the EDITION trials, the results were reported using both a concentration of ≤3.0 mmol/L and of ≤3.9 mmol/L. No restriction on the threshold levels was imposed. A 3.9 mmol/L threshold for the EDITION trials was selected to be consistent with the majority of other trials in the network. Nocturnal hypoglycaemic events were defined as any event (confirmed and/or symptomatic) occurring during a period at night.

Statistical methods

All analyses were implemented using the statistical software R and OpenBUGS, specifically the packages using Markov Chain Monte Carlo (MCMC). Examples of coding used are provided in an online supplemental appendix. Randomised clinical trials that were identified from a systematic literature review and that met the study selection inclusion criteria were analysed using a random-effect Bayesian NMA, following the UK NICE guidance.9 Each outcome was analysed within the evidence network where it was reported. MCMC was used to estimate the posterior distribution for treatment comparison. Continuous outcomes (eg, change in HbA1c or body weight) were modelled assuming a normal likelihood and an identity link. Event rate data (eg, number of hypoglycaemic episodes per patient-year follow-up) were modelled using a Poisson mixed likelihood and log link. Non-informative priors were assumed.

Sensitivity analyses

Sensitivity analyses including meta-regression were conducted to evaluate the robustness of the findings. The base scenario included studies of patients on basal insulin-supported oral therapy (BOT; patients received basal insulin in combination with oral antihyperglycaemic drugs but with no bolus insulin; patients could be either—insulin naïve or insulin experienced). Additional scenarios were all studies (ie, patients receiving basal insulin with or without bolus), studies of patients on BOT excluding premixed studies, studies of insulin-naïve patients only, only studies with Week 24–28 results, and excluding degludec three times weekly (3TW) dosing. Meta-regression was conducted for key outcomes to account for study-level population characteristics, adjusting for the following: study-level baseline HbA1c, diabetes disease duration and basal-bolus population. In addition, broader definitions for hypoglycaemia were analysed. A comparison of NMA to classical meta-analysis in the base scenario (BOT) using an inverse variance-weighted method was also conducted.

Results

Over 4000 studies were identified for screening, of which 86 were identified for data extraction; from these, 41 studies were included in the NMA (figure 1A). A brief overview of these studies is provided in table 1.

Figure 1

(A) PRISMA flow diagram for studies comparing basal insulin therapies in type 2 diabetes mellitus (T2DM; N=41). aCochrane Library (eg, the Cochrane Central Register of Controlled Trials (CENTRAL) and the Database of Abstracts of Reviews of Effectiveness (DARE)), MEDLINE and MEDLINE In-Process (using Ovid platform), Embase (using Ovid Platform) and PsycINFO; If applicable, relevant results from clinical trial registry were included. Zinman et al34 report 2 distinct studies within 1 publication. bFor title/abstract and full-text review, articles were excluded based on inclusion/exclusion criteria as specified in the systematic literature review. cTwo articles analysed the same trial. dConferences searched included EASD and ADA 2011–2013, and IDF 2011. IDF 2013 was assessed when the CD-ROM became available—the end of February. Multiple abstracts examined the same trial and 14 trials were extracted. eStudies must include at least two treatment arms in the network, including: U300, insulin glargine, insulin detemir, insulin NPH, insulin degludec and premix insulin. (B) Evidence network diagram for BOT studies (n=25) reporting HbA1c (%) change from baseline. Each insulin treatment is a node in the network. The links between the nodes represent direct comparisons. The numbers along the lines indicate the number of trials or pairs of trial arms for that link in the network. Reference numbers indicate the trials contributing to each link. BOT, basal insulin-supported oral therapy; HbA1c, glycated haemoglobin; NPH, neutral protamine Hagedorn.

Table 1

Randomised comparative studies included in NMA of patients with T2DM on basal insulin treatment

	First author, year published (Regimen type)	Countries/Continents	Key inclusion criteria	N*	Randomised comparator arms	Allocation method	Study duration	Discontinuation rate†	Outcomes in current NMA‡
									A	B	C	D
Gla-300 vs Gla-100	Bolli, 20158	North America, Europe, Japan	Insulin naïveOADHbA1c 7–11%	873	Gla-300Gla-100	IVRS	6 months	Gla-300: 62/439 (14%)Gla-100: 75/439 (17%)	✓	✓	✓	✓
	Riddle, 20146	North America, Europe, South Africa	On basal bolus insulin regimenHbA1c 7–10%	806	Gla-300 + bolusGla-100 + bolus	IVRS	6 months	Gla-300: 30/404 (7.4%)Gla-100: 31/402 (7.7%)	✓	✓	✓	✓
	Yki-Järvinen, 20147	North America, Europe, Russia, South America, South Africa	On basal insulinOADHbA1c 7–10%	809	Gla-300Gla-100	IVRS	6 months	Gla-300: 36/404 (8.9%)Gla-100: 38/407 (9.3%)	✓	✓	✓	✓
Gla-100 vs premixed insulin	Aschner, 201310	NR	Insulin naïveOAD	923	PremixedGla-100±glulisine	NR	24 weeks	NR (meeting abstract)	✓	✓		✓
	Buse, 200911	Australia, Europe, India, North America, South America	Insulin naïveOADHbA1c >7%	2091	Lispro protamine/lispro 75/25Gla-100	IVRS	24 weeks	Premixed insulin:145/1045 (13.9%)Gla-100: 128/1046 (12.2%)	✓	✓
	Fritsche, 201012	Europe and Australia	Premixed insulin+/- MetforminHbA1c 7.5–11.0%	310	70/30 NPH + bolus (regular or aspart)Gla-100 + glulisine	Electronic case record system	52 weeks	Premixed insulin: 28/157 (17.8%)Gla-100:25/153 (16.3%)	✓	✓	✓
	Jain, 201013	Asia, Australia, Europe, North America, Russian Federation	Insulin naïveOADHbA1c ≥7.5–12%	484	Insulin lispro 50/50Gla-100 + lispro	TS	36 weeks	Premixed insulin: 31/242 (12.8%)Gla-100: 27/242 (11.2%)	✓	✓
	Kann, 200614	Europe	Insulin naïveOADHbA1c >7–12%	255	Insulin aspart 70/30+ metforminGla-100 + glimepiride	Sealed codes	28 weeks	Premixed insulin: 13/130 (10.0%)Gla-100: 12/128 (9.4%)	✓	✓
	Kazda, 200615	Germany	Insulin naïveHbA1c 6–10.5%	159	Protaminatedlispro/lispro 50/50LisproGla-100	NR	24 weeks	Premixed insulin: 14.8%§Bolus insulin: 7.7%§Gla-100: 15.1%§	✓	✓
	Ligthelm, 201116	USA and Puerto Rico	On basal insulinOADHbA1c ≥8%	279	Biphasic aspart 70/30Gla-100	IVRS	24 weeks	Premixed insulin: 19/137 (13.9%)Gla-100: 32/143 (22.4%)	✓	✓	✓
	Raskin, 200517	USA	Insulin naïveOADHbA1c ≥8%	222	Biphasic aspart 70/30Gla-100	Sequential numbers/codes	28 weeks	Premixed insulin:17/117 (14.5%)Gla-100: 7/116 (6.0%)	✓
	Riddle, 201118	NR	OAD	572	Protamine-aspart/aspart 70/30Glargine + 1 prandial GlulisineGla-100 + glulisine (stepwise addition)	NR	60 weeks	NR (meeting abstract)	✓			✓
	Robbins, 200719	Australia, Europe, India, North America (USA and Puerto Rico)	OADHbA1c 6.5–11%	315	Lispro 50/50 + metforminGla-100+metformin	TS	24 weeks	Premixed insulin: 15/158 (9.5%)Gla-100: 22/159 (13.8%)	✓	✓
	Rosenstock, 200820	USA and Puerto Rico	On basal insulinOADHbA1c 7.5–12%	374	Insulin lispro protamine/lisproGla-100 + lispro	TS	24 weeks	Premixed insulin: 29/187 (15.5%)Gla-100: 29/187 (15.5%)	✓	✓		✓
	Strojek, 200921	Asia, Europe, North America, South America, South Africa	Insulin naïveOADHbA1c >7–11%	469	Biphasic aspart 70/30 + metformin/glimepirideGla-100 +metformin/glimepiride	IVRS	26 weeks	Premixed insulin: 26/239 (10.9%)Gla-100: 21/241 (8.7%)	✓		✓
	Tinahones, 201322	11 countries (not specified)	On basal insulinOADHbA1c 7.5–10.5%	478	Lispro mix 25/75Gla-100 + lispro	NR	24 weeks	NR (meeting abstract)	✓	✓		✓
	Vora, 201323	NR	On basal insulin	335	Biphasic insulin aspart/aspart protamine 30/70Gla-100 + glulisine	NR	24 weeks	Premixed insulin: 23/165 (13.9%)Gla-100: 14/170 (8.2%)	✓			✓
Gla-100 vs NPH	Fritsche, 200324	Europe	Insulin naïveOADHbA1c 7.5–10.5%	695	NPHGla-100 (morning)Gla-100 (bedtime)	Sequential numbers/codes	28 weeks	NPH: 27/234 (11.5%)Gla-100 (morning): 12/237 (5.1%)Gla-100 (bedtime):18/229 (7.9%)	✓	✓		✓
	Massi Benedetti, 200325	Europe, South Africa	OAD	570	NPHGla-100	Sequential numbers/codes	52 weeks	NPH: 33/285(11.6%)Gla-100: 16/293 (5.5%)	✓			✓
	Riddle, 200326	North America	Insulin naïveOADHbA1c 7.5–10%	756	NPHGla-100	IVRS	24 weeks	NPH: 32/392 (8.2%)Gla-100: 33/372 (8.9%)	✓	✓		✓
	Rosenstock, 200127	NR	On insulinHbA1c 7–12%	518	NPHGla-100	NR	28 weeks	NPH: 21/259 (8.1%)Gla-100: 28/259 (10.8%)	✓	✓		✓
	Rosenstock 200928	North America	OADHbA1c 6–12%	1017	NPHGla-100	IVRS	5 years	NPH: 145/509 (28.5%)§Gla-100: 141/515 (27.4%)§	✓	✓
	Yki-Järvinen, 200629	Europe	Insulin naïveOADHbA1c ≥8%	110	NPHGla-100	NR	36 weeks	NPH: 1/49 (2.0%)Gla-100: 1/61 (1.6%)	✓	✓		✓
Degludec vs Gla-100	Garber, 201230	Asia (Hong Kong), Europe, Middle East (Turkey), North America, Russia, South Africa	On insulin±OADHbA1c 7–10%	1004	Degludec + aspartGla-100 + aspart	IVRS	52 weeks	Degludec: 137/755 (18.1%)Glargine:40/251 (15.9%)	✓	✓	✓	✓
	Gough, 201331	Europe, North America, Russia, South Africa	Insulin naïveOADHbA1c 7–10%	456	DegludecGla-100	IVRS	26 weeks	NR§	✓	✓	✓	✓
	Meneghini, 201332	Asia, Europe, Israel, North America, Russia, South America, South Africa	OADHbA1c 7–11%	685	Degludec (flexible)Degludec (once daily)Gla-100	IVRS	26 weeks	Degludec(flexible): 26/229 (11.4%)Degludec(once daily): 24/228 (10.5%)Gla-100: 27/230 (11.7%)	✓	✓	✓	✓
	Zinman, 201233	Europe, North America	Insulin naïveOADHbA1c 7–10%	1023	DegludecGla-100	IVRS	52 weeks	Degludec: 166/773 (21.5%)Glargine:60/257 (23.3%)	✓	✓	✓	✓
	Zinman (AM), 201334	Europe, Israel, North America, South Africa	Insulin naïveOADHbA1c 7–10%	456	DegludecGla-100	IVRS	26 weeks	Degludec: 38/230 (16.5%)Gla-100: 24/230 (10.4%)	✓	✓	✓
	Zinman (PM), 201334	Europe, North America	Insulin naïveOADHbA1c 7–10%	467	DegludecGla-100	IVRS	26 weeks	Degludec: 25/233 (10.7%)Gla-100: 25/234 (10.7%)	✓	✓	✓
Detemir vs Gla-100	Hollander, 200835	Europe and the USA	OAD and/or insulinHbA1c 7–11%	319	Detemir + aspartGla-100 + aspart	TS	52 weeks	Detemir: 43/216 (19.9%)Gla-100: 23/107 (21.5%)	✓	✓	✓
	Meneghini, 201336	Asia, South America, USA	Insulin naïveOADHbA1c 7–9%	453	DetemirGla-100	NR	26 weeks	Detemir: 38/228 (16.7%)Gla-100: 41/229 (17.9%)	✓	✓	✓
	Raskin, 200937	NR	OAD and/or insulinHbA1c 7–11%	387	Detemir + aspartGla-100 + aspart	NR	26 weeks	Detemir: 46/256 (18.0%)Gla-100: 18/131 (13.7%)	✓	✓	✓
	Rosenstock, 200838	Europe and the USA	Insulin naïveOADHbA1c 7.5–10%	582	DetemirGla-100	TS	52 weeks	Detemir: 60/291 (20.6%)Gla-100: 39/291 (13.4%)	✓	✓	✓
	Swinnen, 201039	Asia, Australia, Europe, Middle East (Turkey), North America, Russia, South America	Insulin naïveOADHbA1c 7–10.5%	964	DetemirGla-100	NR	24 weeks	Detemir: 10.1%§Gla-100:4.6%§	✓	✓		✓
Detemir vs premixed	Holman, 200740	Europe	Insulin naïveOADHbA1c 7–10%	708	Prandial insulin aspartDetemirBiphasic aspart 30	IVRS	52 weeks	Bolus: 17/239 (7.1%)Detemir: 10/234 (4.3%)Premixed insulin:13/235 (5.5%)	✓	✓
	Liebl, 200941	Europe	OADHbA1c 7–12%	715	Detemir + aspartSoluble aspart/protamine-crystallised aspart 30/70	Codes	26 weeks	Detemir: 44/541 (8.1%)Premixed insulin: 17/178 (9.6%)	✓	✓
Detemir vs NPH	Haak, 200542	Europe	HbA1c ≤12%	505	Detemir + aspartNPH + aspart	NR	26 weeks	Detemir: 26/341 (7.6%)§NPH: 8/164 (4.9%)§	✓	✓
	Hermansen, 200643	Europe	Insulin naïveOADHbA1c 7.5–10%	475	DetemirNPH	TS	24 weeks	Detemir: 4%§NPH: 5%§	✓	✓	✓
	Montañana, 200844	Spain	On insulin± metforminHbA1c 7.5–11%	271	Detemir + aspartNPH + aspart	Codes	26 weeks	Detemir:7/126 (5.6%)NPH: 12/151 (7.9%)	✓	✓	✓
	Philis-Tsimakas, 200645	North America and Europe	Insulin naïveOADHbA1c 7.5–11%	498	Detemir morningDetemir eveningNPH	IVRS	20 weeks	Detemir (morning): 19/168 (11.3%)Detemir (evening): 16/170 (9.4%)NPH: 17/166 (10.2%)	✓		✓
	Raslová, 200446	8 Countries (not specified)	On insulin±OADsHbA1c <12%	394	Detemir + insulin aspartNPH + human soluble insulin	NR	22 weeks	Detemir: 10/195 (5.1%)§NPH: 6/199 (3.0%)§	✓	✓

All the studies were open-label, with the exception of Liebl et al1 (not reported).

*Safety population; exceptions: efficacy population for Buse et al,11 Raslová et al,46 Riddle et al,18 Tinahones et al22 and Vora et al.23

†Numerator for discontinuation rate=randomised patients−patients completing the study; denominator for discontinuation rate=randomised patients. Exceptions noted in footnote (§).

‡A=change in HbA1c, B=change in body weight, C=nocturnal hypoglycaemia rate, D=documented symptomatic hypoglycaemia rate.

§Exceptions to definition of discontinuation rate/or discontinuation rate not calculable with information available: Gough 2013 reported that 460 were randomised 1:1 (3 were randomised in error and were withdrawn, 1 withdrew consent (all prior to treatment)) and 228 and 229 received detemir and Gla-100, respectively, however, completion/withdrawal not described; Kazda et al15 reported ‘drop-out’ rates (however, numbers randomised to each group not provided and denominator may have been exposed rather than randomised patients); Swinnen et al's39 brief report does not make clear what the denominator was for completion rate provided (did not report number randomised to each group, only total randomised; did not report numbers of patients completing the study—only the percentages); Hermansen et al:43 denominator may be ITT population—475 were randomised but the breakdown between treatment arms is not clear; Haak et al,42 reported rates based on patients receiving treatment rather than randomised patients; Raslová et al46 reported rates reported based on ITT rather than randomised patients (ITT only 1 less than randomised, but number randomised to each treatment arm not provided in publication). In addition, Rosenstock et al28 reported data over 5 years; however, only the first year data were included in this NMA.

HbA1c, glycated haemoglobin; ITT, intention to treat; IVRS, interactive voice (or web) response system; NMA, network meta-analysis; NPH, neutral protamine Hagedorn;

NR, not reported; OAD,oral antidiabetic medication; T2DM, type 2 diabetes mellitus; TS, telephone system.

Included trials

All studies were randomised based on entry criteria, with interactive voice (or web) response system or telephone system as the main method of randomisation (n=22), followed by use of sequential numbers/codes (n=6) and electronic case record system (n=1); the method of randomisation was either not reported or not clear in the remaining studies (n=12). The majority (40/41) of studies specified an open-label in design (1 study did not specify). Loss to follow-up (ie, rates of discontinuation among randomised patients) among the studies ranged from 1.6% to 28.5%, with 10 studies reporting discontinuation rates <10%, 22 reporting 10–20% and 5 reporting >20% in at least one treatment arm (loss to follow-up was not reported in 4 studies). The baseline patient characteristics of patients in each of the 41 studies are provided in table 2.

Table 2

Patient baseline characteristics for trials included in the NMA (N=41)

	First author	Year	AgeMean±SD	Male (%)	Diabetes duration (years)Mean±SD	HbA1c (%), Mean±SD	Body weight (kg)Mean±SD
Gla-100 vs Gla-300	Bolli8	2015	57.7±10.1	57.8	9.8±6.4	8.5±1.1	95.4±23.0
	Riddle6	2014	60.0±8.6	52.9	15.9±7.5	8.1±0.8	106.3±20.8
	Yki-Järvinen7	2014	58.2±9.2	45.9	12.6±7.1	8.3±0.8	98.4±21.6
Gla-100 vs premixed	Aschner10	2013	NA	NA	NA	8.7±0	NA
	Buse11	2009	57.0±10	52.80	9.5±6.1	9.1±1.3	88.50±21.0
	Fritsche12	2010	60.6±7.7	50.91	12.7±6.3	8.6±0.9	85.61±15.1
	Jain13	2010	59.4±9.2	48.78	11.7±6.5	9.4±1.2	78.5±15.3
	Kann14	2006	61.3±9.1	51.4	10.25±7.1	9.1±1.4	85.4±15.5
	Kazda15	2006	59.4±9.5	54.7	5.6±2.9	8.1±1.2	NA
	Ligthelm16	2011	52.7±10.4	56.66	11.15±6.4	9.0±1.1	97.9±20.5
	Raskin17	2005	52.5±10.2	54.5	9.2±5.3	9.8±1.5	90.2±18.9
	Riddle18	2011	NA	NA	NA	NA	NA
	Robbins19	2007	57.8±9.1	49.9	11.9±6.3	7.8±1.0	88.6±19.7
	Rosenstock20	2008	54.7±9.5	52.5	11.1±6.3	8.9±1.1	99.5±20.6
	Strojek21	2009	56.0±9.9	43.96	9.3±6.0	8.5±1.1	NA
	Tinahones22	2013	NA	NA	NA	8.6±0.8	NA
	Vora23	2013	NA	NA	NA	NA	NA
Gla-100 vs NPH	Fritsche24	2003	61.0±9.0	53.7	NA	9.1±1.0	81.3±14.8
	MassiBenedetti25	2003	59.5±9.2	53.7	10.35±6.1	9.0±1.2	NA
	Riddle26	2003	55.5±9.2	55.5	8.71±5.56	8.6±0.9	NA
	Rosenstock27	2001	59.4±9.8	60.1	13.75±8.65	8.6±1.2	90.2±17.6
	Rosenstock28	2009	55.1±8.7	53.9	10.75±6.8	8.4±1.4	99.5±22.5
	Yki-Järvinen29	2006	56.5±1	63.3	9±1	9.5±0.1	93.1±2.5
Degludec vs Gla-100	Garber30	2012	58.9±9.3	54.0	13.6±7.3	8.3±0.8	92.5±17.7
	Gough31	2013	57.6±9.2	53.2	8.2±6.2	8.3±1.0	92.5±18.5
	Meneghini32	2013	56.5±9.6	53.7	10.6±6.7	8.4±0.9	81.7±16.7
	Zinman33	2012	59.2±9.8	61.9	9.2±6.2	8.2±0.8	90.0±17.3
	Zinman (PM)34	2013	57.4±10.2	57.2	8.8±3.4	8.3±0.8	91.9±18.5
	Zinman (AM)34	2013	58.2±9.8	56.9	8.9±6.1	8.3±0.9	93.3±18.8
Detemir vs Gla-100	Hollander35	2008	58.7±11	58.0	13.5±8.0	8.7±1.0	92.7±17.6
	Meneghini36	2013	57.3±10.3	56.5	8.2±6.1	7.91±0.6	82.3±16.7
	Raskin37	2009	55.8±10.3	54.6	12.3±7.0	8.4±1	95.6±18.2
	Rosenstock38	2008	58.9±9.9	57.9	9.1±6.3	8.6±0.8	87.4±17.0
	Swinnen39	2010	58.4±8.3	54.7	9.9±5.8	8.7±0.9	83.9±17.1
Detemir vs premixed	Holman40	2007	61.7±9.8	64.1	NA	8.5±0.8	85.8±15.9
	Liebl41	2009	60.7±9.2	58.5	9.3±6.4	8.5±1.1	NA
Detemir vs NPH	Haak42	2005	60.4±8.6	51.1	13.2±7.6	7.9±1.3	86.9±15.8
	Hermansen43	2006	60.9±9.2	53.1	9.7±6.4	8.6±0.8	82.6±13.8
	Montañana44	2008	61.9±8.8	40.6	16.3±8.0	8.85±1.0	81.0±12.1
	Philis-Tsimakas45	2006	58.5±10.5	56.8	10.3±7.2	9.0±1.0	NA
	Raslová46	2004	58.3±9.3	42.1	14.1±7.8	8.1±1.3	80.8±12.7

HbA1c, glycated haemoglobin; NA, not applicable; NMA, network meta-analysis; NPH, neutral protamine Hagedorn.

Twenty-five of the 41 studies (61%) were of patients on BOT (main population for this analysis; n=15 746 patients). The evidence network for the BOT studies is depicted in figure 1B. Patients in the BOT studies had a mean age ranging from 52.4 to 61.7 years, duration of diabetes 8.2–13.8 years, baseline body weight 81.3–99.5 kg and HbA1c 7.8–9.8%.

Glycaemic control

In patients with T2DM on BOT (n=25 studies), the change in HbA1c was comparable between Gla-300 and insulin detemir (−0.08; −0.40 to 0.24), neutral protamine Hagedorn (NPH; 0.01; −0.28 to 0.32), degludec (−0.12; −0.42 to 0.20) and premixed insulin (0.26; −0.04 to 0.58) (figure 2A). These changes were similar to those in the overall NMA (n=41 studies) and across the various sensitivity analyses shown in table 3A.

Figure 2

NMA findings for Gla-300 versus other basal insulins in the BOT population: (A) change in HbA1c (%); (B) change in body weight (kg); (C) risk of nocturnal hypoglycaemia; (D) risk of documented symptomatic hypoglycaemia. BOT, basal insulin-supported oral therapy; CrI, credible interval; DET, =insulin detemir; DEG, insulin degludec; HbA1c, glycated haemoglobin; NMA, network meta-analysis; NPH, neutral protamine Hagedorn; PREMIX, premixed insulin; RR, risk ratio.

Table 3

Additional analyses

(A) Sensitivity analyses
Outcome	Comparator
	Gla-100	Detemir	NPH	Degludec	Premix
Change in HbA1c*
BOT, insulin naïve	0.01 (−0.27 to 0.29)	−0.14 (−0.47 to 0.19)	−0.09 (−0.43 to 0.25)	−0.12 (−0.45 to 0.21)	0.08 (−0.23 to 0.39)
Adjusting for Bolus Insulin Trials	−0.01 (−0.44 to 0.42)	−0.10 (−0.55 to 0.36)	−0.05 (−0.51 to 0.41)	−0.14 (−0.60 to 0.33)	0.07 (−0.37 to 0.51)
Insulin naïve	0.04 (−0.41 to 0.48)	−0.09 (−0.59 to 0.40)	−0.06 (−0.55 to 0.43)	−0.12 (−0.62 to 0.37)	0.24 (−0.22 to 0.72)
T2DM overall	0.01 (−0.23 to 0.25)	−0.08 (−0.37 to 0.21)	−0.03 (−0.32 to 0.26)	−0.12 (−0.42 to 0.18)	0.09 (−0.18 to 0.35)
Studies reporting hypoglycaemia data	0.01 (−0.23 to 0.25)	−0.18 (−0.51 to 0.14)	−0.09 (−0.57 to 0.38)	−0.12 (−0.42 to 0.18)	0.18 (−0.12 to 0.51)
Studies with 24–28-week results	0.01 (−0.24 to 0.26)	−0.04 (−0.36 to 0.27)	−0.03 (−0.35 to 0.30)	−0.14 (−0.47 to 0.19)	0.17 (−0.10 to 0.45)
Excluding Degludec 3TW	0.02 (−0.22 to 0.28)	−0.08 (−0.37 to 0.22)	0.01 (−0.26 to 0.30)	−0.01 (−0.32 to 0.31)	0.26 (−0.02 to 0.55)
Adjusting for baseline HbA1c	0.05 (−0.49 to 0.63)	−0.03 (−0.60 to 0.56)	0.02 (−0.56 to 0.61)	−0.07 (−0.65 to 0.53)	0.13 (−0.42 to 0.72)
Adjusting for disease duration	0.03 (−0.29 to 0.34)	−0.06 (−0.41 to 0.29)	−0.01 (−0.37 to 0.35)	−0.10 (−0.46 to 0.26)	0.11 (−0.23 to 0.44)
Change in body weight
BOT, insulin naïve	−0.44 (−1.67 to 0.81)	0.58 (−0.85 to 2.03)	−0.22 (−1.68 to 1.25)	−0.52 (−1.93 to 0.92)	−1.09 (−2.44 to 0.29)
Adjusting for Bolus Insulin Trials	−0.58 (−2.54 to 1.37)	0.11 (−1.98 to 2.20)	−0.63 (−2.75 to 1.45)	−0.66 (−2.78 to 1.45)	−1.13 (−3.18 to 0.91)
Insulin naïve	−0.30 (−1.44 to 0.82)	1.18 (−0.12 to 2.47)	−0.12 (−1.39 to 1.10)	−0.46 (−1.71 to 0.80)	−1.12 (−2.39 to 0.15)
T2DM overall	−0.27 (−1.28 to 0.73)	0.42 (−0.78 to 1.62)	−0.32 (−1.54 to 0.89)	−0.35 (−1.58 to 0.88)	−0.81 (−1.96 to 0.32)
Studies reporting hypoglycaemia data	−0.28 (−1.28 to 0.71)	1.01 (−0.29 to 2.31)	0.89 (−0.90 to 2.70)	−0.36 (−1.58 to 0.86)	−1.24 (−2.59 to 0.09)
Studies with 24–28-week results	−0.28 (−1.28 to 0.74)	0.26 (−1.05 to 1.57)	−0.15 (−1.45 to 1.16)	−0.42 (−1.76 to 0.92)	−1.01 (−2.19 to 0.18)
Excluding Degludec 3TW	−0.46 (−1.34 to 0.43)	0.68 (−0.38 to 1.76)	−0.76 (−1.82 to 0.27)	−0.79 (−1.90 to 0.33)	−1.83 (−2.89 to −0.68)
Adjusting for baseline HbA1c	−0.27 (−2.03 to 1.25)	0.43 (−1.46 to 2.12)	−0.32 (−2.23 to 1.39)	−0.34 (−2.26 to 1.38)	−0.81 (−2.68 to 0.82)
Adjusting for disease duration	−0.44 (−1.91 to 1.00)	0.25 (−1.38 to 1.87)	−0.49 (−2.15 to 1.13)	−0.52 (−2.20 to 1.13)	−0.99 (−2.58 to 0.58)
Nocturnal hypoglycaemia event rate
BOT, insulin naïve	0.57 (0.33 to 0.98)	0.53 (0.28 to 1.01)	0.21 (0.10 to 0.44)	0.68 (0.36 to 1.25)	0.42 (0.21 to 0.81)
BOT, premixed excluded	0.62 (0.37 to 1.17)	0.56 (0.30 to 1.21)	0.16 (0.08 to 0.41)	0.79 (0.42 to 1.64)	N/A
Adjusting for Bolus Insulin Trials	0.56 (0.24 to 1.29)	0.52 (0.21 to 1.32)	0.20 (0.07 to 0.57)	0.66 (0.26 to 1.61)	0.50 (0.19 to 1.26)
Insulin naïve patients only	0.58 (0.12 to 2.77)	0.51 (0.07 to 3.38)	0.17 (0.02 to 1.37)	0.61 (0.10 to 3.48)	0.26 (0.03 to 2.35)
T2DM overall	0.64 (0.39 to 1.03)	0.60 (0.32 to 1.11)	0.23 (0.11 to 0.50)	0.75 (0.41 to 1.34)	0.57 (0.31 to 1.05)
Studies with 24–28-week results	0.64 (0.37 to 1.10)	0.51 (0.22 to 1.18)	0.24 (0.08 to 0.70)	0.67 (0.32 to 1.37)	0.55 (0.26 to 1.17)
Excluding Degludec 3TW	0.57 (0.33 to 0.98)	0.51 (0.24 to 1.07)	0.19 (0.07 to 0.45)	0.83 (0.42 to 1.69)	0.36 (0.17 to 0.74)
2.8–4.2 mmol/L	0.64 (0.37 to 1.11)	0.68 (0.35 to 1.34)	0.31 (0.15 to 0.63)	0.75 (0.38 to 1.46)	0.68 (0.35 to 1.29)
Adjusting for baseline HbA1c	0.37 (0.18 to 0.90)	0.35 (0.15 to 0.91)	0.13 (0.05 to 0.39)	0.43 (0.19 to 1.12)	0.33 (0.14 to 0.86)
Adjusting for disease duration	0.60 (0.31 to 1.13)	0.56 (0.26 to 1.19)	0.22 (0.09 to 0.53)	0.71 (0.34 to 1.46)	0.54 (0.25 to 1.14)
Documented symptomatic hypoglycaemia event rate
BOT, insulin naïve	0.72 (0.40 to 1.30)	0.63 (0.22 to 1.73)	0.58 (0.26 to 1.24)	0.59 (0.29 to 1.20)	0.50 (0.24 to 1.01)
BOT, premixed excluded	0.75 (0.55 to 1.05)	0.69 (0.42 to 1.23)	0.55 (0.36 to 0.91)	0.66 (0.46 to 1.01)	N/A
Adjusting for Bolus Insulin Trials	0.83 (0.35 to 1.83)	0.72 (0.22 to 2.31)	0.76 (0.28 to 1.86)	0.68 (0.26 to 1.67)	0.57 (0.22 to 1.41)
Insulin naïve patients only	0.62 (0.21 to 1.77)	0.54 (0.12 to 2.36)	0.50 (0.14 to 1.63)	0.61 (0.17 to 2.25)	0.24 (0.05 to 1.09)
T2DM overall	0.78 (0.50 to 1.23)	0.68 (0.27 to 1.70)	0.71 (0.38 to 1.30)	0.64 (0.36 to 1.16)	0.54 (0.30 to 0.98)
Studies with 24–28-week results	0.78 (0.45 to 1.34)	0.68 (0.23 to 2.01)	0.75 (0.36 to 1.60)	0.53 (0.23 to 1.20)	0.58 (0.27 to 1.25)
Adjusting for baseline HbA1c	0.71 (0.44 to 1.13)	0.61 (0.25 to 1.51)	0.64 (0.34 to 1.19)	0.58 (0.32 to 1.05)	0.49 (0.27 to 0.89)
Adjusting for disease duration	0.57 (0.32 to 0.99)	0.50 (0.18 to 1.33)	0.52 (0.25 to 1.04)	0.47 (0.23 to 0.92)	0.40 (0.19 to 0.78)

*Four additional studies were included in sensitivity analyses for HbA1c and/or body weight, but were not in the main NMA.47–50

†No direct evidence for specific comparison.

BOT, basal insulin-supported oral therapy (ie, no bolus insulin); CrI, Credible interval; HbA1c, glycated haemoglobin; NA, not applicable; NMA, network meta-analysis; NPH, neutral protamine Hagedorn; T2DM, type 2 diabetes mellitus.

Continued

Body weight

Change in body weight from baseline was reported in 36 trials in the NMA. Among patients with T2DM on BOT, no statistically significant difference in body weight change was observed between Gla-300 and detemir (difference: 0.69; 95% CrI −0.31 to 1.71), NPH (−0.76; −1.75 to 0.21) or degludec (−0.63; −1.63 to 0.35), whereas weight gain was significantly lower with Gla-300 compared with premixed insulin (−1.83; −2.85 to −0.75) (figure 2B). These changes were similar to those in the overall NMA (n=41 studies) and across the various sensitivity analyses (table 3A).

Hypoglycaemia events

Among the studies identified, 20 trials reported nocturnal hypoglycaemia event rate data and 16 reported documented symptomatic hypoglycaemia event rate data that met criteria for inclusion in the NMA. The hypoglycaemia event data from each of these clinical trials are summarised in table 4.

Table 4

Hypoglycaemia outcomes for trials included in the NMA

Study	Year	Arm	Total exposure*	Documented symptomatic	Nocturnal	Severe†
Gla-100 vs Gla-300
Bolli et al8	2013	Gla-100	218	821	41	4
		Gla-300	217	505	24	4
Riddle et al6	2014	Gla-100	200	2957	162	48
		Gla-300	201	2714	127	54
Yki-Järvinen et al7	2014	Gla-100	202	1641	140	12
		Gla-300	201	1357	78	6
Gla-100 vs premixed insulin
Aschner et al10	2013	Gla-100	213	249		5
		Premixed insulin	212	632		3
Fritsche et al12	2010	Gla-100	141		321	16
		Premixed insulin	149		353	33
Ligthelm et al16	2011	Gla-100	65		233	6
		Premixed insulin	63		273	0
Raskin et al17	2005	Gla-100	61			1
		Premixed insulin	58			0
Riddle et al18	2011	Gla-100 (plus step-wise glulisine)	220	1559
		Gla-100 (plus 1 prandial dose)	217	1565
		Premixed insulin	221	2694
Robbins et al19	2007	Gla-100	73			4
		Premixed insulin	72			8
Rosenstock et al20	2008	Gla-100	86	3866		3
		Premixed insulin	86	4000		9
Strojek et al21	2009	Gla-100	114		57	3
		Premixed insulin	110		120	3
Tinahones et al22	2013	Gla-100	111	859
		Premixed insulin	109	783
Vora et al23	2013	Gla-100	78		446
		Premixed insulin	76		273
Gla-100 vs NPH
Fritsche et al24	2003	Gla-100 (morning dosing)	109	710		6
		Gla-100 (evening dosing)	104	467		4
		NPH	107	583		13
Riddle et al26	2003	Gla-100	169	1553		14
		NPH	179	2308		9
Rosenstock et al27	2001	Gla-100	139	2012
		NPH	139	1577
Rosenstock et al28	2009	Gla-100	2556			102
		NPH	2511			151
Yki-Järvinen et al29	2006	Gla-100	42	5		0
		NPH	34	8		0
Degludec vs Gla-100
Garber et al30	2012	Degludec	671	13 821	932	40
		Gla-100	229	5361	421	11
Gough et al31	2013	Degludec	106	357	19	0
		Gla-100	107	389	30	0
Meneghini et al32	2013	Degludec (flexible dosing)‡	108	851	65	2
		Degludec (evening dosing)	105	776	63	2
		Gla-100	105	383	84	2
Zinman et al33	2012	Degludec	667	2675	167	2
		Gla-100	218	806	85	5
Zinman (AM) et al34	2013	Degludec	105		42	1
		Gla-100	106		21	1
Zinman (PM) et al34	2013	Degludec	109		22	1
		Gla-100	110		22	0
Detemir vs Gla-100
Hollander et al35	2008	Detemir	187		449	17
		Gla-100	92		265	6
Meneghini et al36	2013	Detemir	103		115	0
		Gla-100	104		91	2
Raskin et al37	2009	Detemir	128		540	11
		Gla-100	65		221	8
Rosenstock et al38	2008	Detemir	262		341	0
		Gla-100	269		350	0
Swinnen et al39	2010	Detemir	224	1491		18
		Gla-100	220	1273		35
Detemir vs premixed insulin
Holman et al40	2007	Detemir	233			4
		Premixed insulin	234			11
		Aspart	238			16
Detemir vs NPH
Hermansen et al43	2006	Detemir	106		160	1
		NPH	106		349	8
Montañana et al44	2008	Detemir	62		46	0
		NPH	73		107	3
Philis-Tsimikas et al45	2006	Detemir (morning dosing)	63		6	0
		Detemir (evening dosing)	65		19	2
		NPH	63		47	0

*Total exposure indicates the number of patient-years over which the rate for hypoglycaemic events is determined.

†Although severe events were not analysed in the NMA due to small numbers of events, they are included in the table if reported within the publication.

‡Rotating morning and evening dosing schedule (ie, 8–40 h intervals between doses).

NMA, network meta-analysis; NPH, neutral protamine Hagedorn.

Nocturnal hypoglycaemia

In patients with T2DM on BOT, Gla-300 was associated with a significantly lower nocturnal hypoglycaemia rate compared with NPH (0.18; 0.05 to 0.55) and premixed insulin (0.36; 0.14 to 0.94) and a numerically lower rate when compared with detemir (0.52; 0.19 to 1.36) and degludec (0.66; 0.28 to 1.50) (figure 2C). These changes were similar to those in the overall NMA (n=41 studies) and across the various sensitivity analyses (table 3A).

Documented symptomatic hypoglycaemia

In patients with T2DM on BOT, Gla-300 was associated with a numerically lower rate of documented symptomatic hypoglycaemic events compared with detemir (0.63; 0.19 to 2.00), NPH (0.66; 0.27 to 1.49) and degludec (0.55; 0.23 to 1.34) (figure 2D). These changes were similar to those in the overall NMA (n=41 studies) and across the various sensitivity analyses (table 3A). In the BOT population, comparative data for premixed insulin were not available for this particular outcome.

Comparison of NMA to classic meta-analysis findings

The comparison of NMA results that integrate all available evidence versus those from classical meta-analysis solely based on direct evidence in the base scenario (BOT) found generally consistent effect size across all four outcomes and tighter 95% CIs with the classical meta-analysis (table 3B).

Discussion

In this NMA of randomised clinical studies comparing various basal insulin therapies in patients with T2DM, the new concentrated formulation, Gla-300, demonstrated change in HbA1c that was comparable to the change reported in studies of insulin detemir, degludec, NPH and premixed insulin. Change in body weight with Gla-300 was significantly less than that with premixed insulin and comparable to the other basal insulin. Hypoglycaemia rates appeared lower with Gla-300 and the comparator basal insulin. The rate of documented symptomatic hypoglycaemia associated with Gla-300 was numerically but not significantly different from that of other basal insulin therapies. A notable difference was that Gla-300 was associated with a significantly lower risk of nocturnal hypoglycaemia (ranging from approximately 64% to 82% lower) compared with premixed insulin and NPH.

These NMA data extend our current knowledge regarding Gla-300. Based on direct comparisons in the EDITION studies, Gla-300 was associated with comparable glycaemic control, but had a significantly lower rate of nocturnal hypoglycaemia compared with Gla-100.6–8 The more flat and more prolonged pharmacokinetic profile associated with Gla-300 compared with Gla-100 may contribute to the reduced rate of nocturnal hypoglycaemia that is observed clinically. Reasons for the difference in pharmacokinetic profile between Gla-100 and Gla-300 are not known, but may be due to factors inherent to the retarding principle of the insulin glargine molecule and a phenomenon of surface-dependent release.4 5 Gla-300 has a pH of approximately 4, at which it is completely soluble; however, once injected subcutaneously, the solution is neutralised and forms a precipitate allowing for the slow release of small amounts of insulin glargine. It has been suggested that the size (ie, surface area) of the subcutaneous deposit may determine the redissolution rate.51

The finding of a significantly lower rate of nocturnal hypoglycaemia associated with a basal insulin analogue compared with NPH is consistent with previous meta-analyses. For example, a meta-analysis of randomised clinical trials comparing long-acting basal insulin analogues (Gla-100 or detemir) with NPH showed that, among 10 studies reporting data for nocturnal hypoglycaemia, both analogues were associated with a reduced risk of nocturnal events, with an OR of 0.46 (95% CI 0.38 to 0.55) compared with NPH.52 Similarly, in the pivotal Treat-to-Target study comparing Gla-100 to NPH, the risk reduction with Gla-100 ranged from 42% to 48% for different categories of nocturnal hypoglycaemic events.26 A subsequent meta-analysis of individual patient data from 5 randomised clinical trials comparing Gla-100 to NPH, reported reductions of approximately 50% in nocturnal hypoglycaemia with Gla-100.53 Given these data, along with patient-level data from the EDITION trials,6–8 which when pooled54 demonstrated a 31% lower relative difference in the annualised rate of nocturnal events over the 6-month study period for Gla-300 compared with Gla-100, the even more pronounced difference in the rate of nocturnal events between Gla-300 and NPH in this NMA is expected.

The finding of fewer nocturnal hypoglycaemic events with Gla-300 compared with premixed insulin in this NMA is in line with ‘real-world’ data from the Cardiovascular Risk Evaluation in people with type 2 Diabetes on Insulin Therapy (CREDIT) study, an international observational study that provided insights on outcomes following insulin initiation in clinical practice.55 In CREDIT study, propensity-matched groups were evaluated 1 year after initiating insulin treatment and showed that basal insulin was associated with significantly lower rates of nocturnal hypoglycaemia compared with premixed insulin. This also held true for propensity-matched analysis of basal plus mealtime insulin versus premixed insulin groups.

The substantially lower risk of nocturnal hypoglycaemia associated with Gla-300 is an important finding given the clinical burden associated with such events.56 In a multination survey of 2108 patients with diabetes (types 1 and 2) who had recently experienced nocturnal hypoglycaemia, patients reported a negative impact on their sleep quality as well as their functioning, the day after a nocturnal hypoglycaemic event.57 Nocturnal events were associated with increased self-monitoring of blood glucose, and approximately 15% of patients reported temporary reductions in insulin dose. An economic evaluation of these data found that nocturnal hypoglycaemic events were associated with lost work productivity and increased healthcare utilisation.58 Utilisation costs were estimated to be higher among patients who injured themselves due to a trip or fall associated with their nocturnal hypoglycaemia episode (approximately $2000 per person annually).

While the findings of this NMA are promising for Gla-300, several limitations are evident. The studies included in this NMA were of open-label design, which is inherently subject to bias; however, this type of methodology is typically used in trials comparing insulin therapies due to visible differences between insulin products and/or differences in injection devices. A potential issue is that there was no multiplicity adjustment, and given that there were multiple comparisons, it is possible that positive findings were due to chance. In addition, trial-level summary data may not have been adequately powered to detect differences between products—for example, while randomised controlled studies of Gla-100 versus Gla-300 and pooled patient level data from these studies have shown that Gla-300 is associated with a significantly lower rate of nocturnal hypoglycaemia, the trial-level data comparisons in this NMA did not achieve significance for this end point. Finally, a well-recognised limitation of any NMA is that, by design, these are not randomised comparisons; however, these data can aid the decision-making process until prospective randomised comparative clinical trial data become available.

Strengths of the current NMA include that it was conducted in accordance with established NICE guidelines and that the estimates reported are in line with those in previous meta-analyses of comparative basal insulin studies.52 53 59 60 NMA provides the capability of considering different pathways simultaneously rather than simple indirect pairwise comparison through multiple pathways. Another strength is the quality of studies included in the NMA (ie, the majority had discontinuation rates <20%). The studies included were similar in design and, from a clinical standpoint, heterogeneity of the patient population was not considered an issue. Results of the NMA were internally consistent with what was reported in individual RCTs. Finally, extensive sensitivity analyses considering subsets of studies, different hypoglycaemia definitions and adjusting for trial-level characteristics, supported the robustness of the findings.

In conclusion, clinical trial findings and the results from this NMA suggest that Gla-300 in the treatment of T2DM is associated with a lower rate of nocturnal hypoglycaemia than treatment with premixed insulin and NPH, while demonstrating comparable glycaemic control versus all comparators. Change in body weight was significantly lower for Gla-300 versus premixed insulin, and comparable with other basal insulin. These NMA data, along with randomised clinical trial findings of reduced nocturnal hypoglycaemia and comparable clinical benefits for Gla-300 versus Gla-100, suggest that this new basal insulin represents an important advance in insulin treatment for patients with T2DM.