Original Paper. Pharmacology 2008;82: DOI: /

Transcription

1 Original Paper Pharmacology 2008;82: DOI: / Received: April 14, 2008 Accepted: May 2, 2008 Published online: August 1, 2008 Indirect Comparison of Once Daily Insulin Detemir and in Reducing Weight Gain and Hypoglycaemic Episodes when Administered in Addition to Conventional Oral Anti-Diabetic Therapy in Patients with Type-2 Diabetes Walid Fak hour y a Ian Lockhart a Robert W. Kotchie a Mark Aagren b Corinne LeReun c a IMS Health, London, UK; b Novo Nordisk A/S, Bagsværd, Denmark; c MTAG, Sydney, N.S.W., Australia Key Words D e te m i r Insulin Neutral protamine Hagedorn Weight gain Type-2 diabetes A b s t r a c t A i m s : Basal insulin administered to type-2 diabetic patients with poor glycaemic control when managed with oral antidiabetics () alone can lead to an increased risk of weight gain and hypoglycaemia. In the absence of head-to-head trials, an indirect comparison of the once-daily insulin detemir with insulin glargine was conducted on the following outcomes: weight gain, hypoglycaemic episodes, and HbA 1c. Methods: Parallel-group randomised controlled trials of at least 20 weeks duration that compared once-daily evening glargine or detemir with a common comparator, neutral protamine Hagedorn insulin (evening), were selected. Trials focused on insulin-naïve, type-2 diabetic patients poorly controlled with OAD. Five open-label trials were identified (n = 2,092 patients; n = 1 detemir and n = 4 glargine trials), with an indirect comparison of glargine (n = 869 patients) and detemir trials (n = 169 patients) carried out using metaregression to control for covariates. Weight gain was analysed as weighted mean differences (WMD), hypoglycaemic episodes as odds ratios (OR), and HbA 1c at the end of study as standardised mean differences (SMD). Results: Patients receiving evening detemir gained significantly less weight (unadjusted WMD 1.22 kg, 95% CI 2.15, 0.29 kg; p = 0.010) and significantly fewer of them experienced hypoglycaemic episodes versus evening glargine (unadjusted OR 0.52, 95% CI 0.28, 0.98; p = 0.044). There was no significant difference between treatments for the mean HbA 1c level at study endpoint (unadjusted SMD 0.09, 95% CI 0.16, 0.33; p = 0.480). Conclusions: Once-daily use of insulin detemir resulted in significantly less weight gain and fewer hypoglycaemic episodes than glargine, while maintaining clinically appropriate HbA 1c levels in type-2 diabetic patients currently receiving OAD. Copyright 2008 S. Karger AG, Basel Introduction Insulin therapy, while providing effective glycemic control, can result in weight gain and hypoglycaemic episodes in type-2 diabetic patients, which can lead to prob- Funding: This project was sponsored by Novo Nordisk A/S, Bagsværd, Denmark. W.F., I.L., R.W.K., and C.L. are IMS Health employees (UK), while M.A. is an employee of Novo Nordisk (Denmark). Fax karger@karger.ch S. Karger AG, Basel /08/ $24.50/0 Accessible online at: Dr. Walid Fakhoury IMS Health, Lee House 109 Hammersmith Road London W14 OQH (UK) Tel , Fax , wfakhoury@uk.imshealth.com

2 lems in adherence to long-term treatment with insulin [1]. Results from the United Kingdom Prospective Study (UKPDS), a large randomised controlled study of type-2 diabetic patients, showed that glycaemic control with insulin therapy resulted in a mean weight gain of 4 kg from baseline measures [2, 3]. There was also a greater increase in rates of hypoglycaemic episodes per year for insulin therapy compared to oral treatment with sulphonylureas or conventional diet therapy [2]. Patients who have poorly controlled type-2 diabetes with oral anti-diabetics () alone require basal insulin therapy to achieve sufficient glycaemic control to avoid long-term macrovascular and microvascular complications associated with poor control of the disease [1, 4]. Weight gain occurs most rapidly soon after initiation of insulin therapy, and may interfere with treatment adherence, since prior to insulin therapy, many type-2 diabetic patients are already overweight and anxious to avoid additional gains [1, 4]. Weight gain can also lead to insulin resistance in clinically obese patients, and is associated with an increased risk of adverse cardiovascular events [1, 5]. Hypoglycaemia is caused by falling plasma glucose concentrations that influence central nervous system responses and neuronal metabolism [6]. Mild to moderate clinical symptoms include palpitations, anxiety/arousal, hunger, fatigue, confusion, and behavioural changes, while severe symptoms include seizures and brain damage [1, 6]. The fear of hypoglycaemia can affect adherence to insulin therapy and may predispose clinicians to delay insulin therapy when commencement is clinically appropriate [1, 7, 8]. In randomised controlled trials (s) of type-2 diabetic patients, insulin detemir (Levemir, Novo Nordisk A/S, Bagsværd, Denmark), a new long-acting insulin analogue, is associated with significantly less weight gain and hypoglycaemia than insulin neutral protamine Hagedorn (NPH), the standard long-acting human insulin therapy [9 12]. Insulin glargine (Lantus, Sanofi- Aventis, Paris, France), the other long-acting insulin analogue available for this population, did not differ on weight gain, but significantly reduced the number of nocturnal hypoglycaemic episodes compared to insulin NPH [13 16]. There are no head-to-head s comparing insulin detemir with insulin glargine in patients receiving who adhere to a once-daily dosing regimen for both treatments until study completion. For this reason, an indirect comparison of published s using treat-to-target protocols with insulin NPH as the common comparator was conducted to evaluate oncedaily insulin detemir versus glargine. The objective was to assess the comparative ability of these treatments to control weight gain and minimise the occurrence of hypoglycaemia while maintaining glycaemic control in type-2 diabetic patients taking. Materials and Methods S t u dy S e l e c t i o n C r i t e r i a The Medline and Cochrane Library electronic databases were searched for English-language papers through to March Criteria for the selection of relevant research studies were: (i) parallel-group s of at least 20 weeks duration available as fulltext; (ii) either once-daily insulin glargine or insulin detemir with a common comparator, e.g. once-daily insulin NPH (evening); (iii) reporting weight gain as an outcome, and (iv) insulin-naïve, poorly controlled type-2 diabetic patients currently receiving. The literature search did not identify any s that directly compared once-daily insulin detemir with insulin glargine in type-2 diabetic patients. Relevant studies identified from the literature search were obtained, and were read and reviewed for methodological quality. Thereafter, data on the key characteristics of the study sample, outcomes measured, and results reported in each study were tabulated in a data extraction sheet. Data extraction was performed by one reviewer, with the completed table checked for accuracy by a second reviewer. Statistics and Outcome Measures The literature search identified five open-label s that met the study selection criteria. An indirect comparison of the insulin glargine and detemir trials was carried out using meta-regression analysis as recommended in the literature [16] (using Stata version 8.1). This involves initially performing direct comparison metaanalyses for each treatment of interest versus the common comparator (e.g. insulin NPH). In the present case, direct meta-analysis was performed only for the insulin glargine versus NPH trials, since there were too few once-daily insulin detemir trials (n = 1) to enable pooling of results. Thereafter meta-regression was used to estimate the degree to which one or more covariates account for differences in treatment results observed in the insulin glargine and detemir trials. In the case of an indirect treatment comparison, the type of treatment is one of these covariates. Accordingly, the weight gain, hypoglycaemia and HbA 1c results from the s comparing insulin glargine to insulin NPH were meta-analysed using a fixed-effects model. The 2 test was used to detect the presence of significant heterogeneity. Mean weight gain from baseline, a continuous outcome, was analysed and reported as a weighted mean difference (WMD). The number of patients presenting with at least one hypoglycaemic episode of any kind (i.e. symptomatic, asymptomatic, severe or nocturnal) at study completion in each treatment group was analysed as odds ratios using the Mantel-Haenszel method. Mean HbA 1c at study completion, an established outcome measure [11, 15], was analysed as a standardised mean difference (SMD). The weight gain, hypoglycaemia and HbA 1c outcomes in the insulin detemir trial were converted to these same statistical measures. Indirect Comparison of Detemir and Pharmacology 2008;82:

3 Thereafter meta-regression analysis by treatment arm (unadjusted model) was conducted, and additional analyses of other covariates for weight gain and HbA 1c were performed to control for potential sources of heterogeneity. These included, for both outcomes, mean age at baseline, study duration, duration of OAD therapy, and in addition, baseline HbA 1c for the latter outcome. Since only two of four studies focusing on insulin glargine [13, 16] reported data on hypoglycaemic episodes, it was not possible to conduct meta-regression analyses on additional covariates for this outcome. The regression model used relates the treatment effect to the study-level covariates, assuming a normal distribution for the residual errors with a within-study and an additive between-studies component of variance, the latter denoted 2. The estimated between-studies variance 2 is a measure of the residual heterogeneity once having adjusted for the covariates. An estimate of the difference between insulin detemir and insulin glargine is reported for each analysis, with p values and 95% two-sided confidence intervals are provided to demonstrate the statistical significance of the results. R e s u l t s Study/Patient Characteristics Five open-label s (n = 2,092 patients) compared once-daily evening insulin glargine or detemir with insulin NPH in insulin-naïve type-2 diabetic patients who had poor glycaemic control with. In all s, patients continued taking OAD medication when insulin therapy was initiated, and once-daily dosing of insulin was adhered to until study completion. One (n = 333 patients) compared once-daily evening insulin detemir versus evening insulin NPH in patients recruited from Western Europe and the USA [11]. Four s compared insulin glargine administered in the evening versus NPH insulin, with one study [14] (n = 764 patients) recruiting from the USA and Canada, and the remaining three (n = 459, n = 426, n = 157) from European countries [13, 15, 16]. There was some betweenstudy variation in reporting randomisation, with most using a remote treatment allocation system [11, 13 15] but none reporting the method for generating the randomisation schedule. In terms of baseline characteristics ( table 1 ), the mean age of patients was similar across s (range years), as were proportions of males and females (4/5 s had an average 10% more males than females), and study completion rates (range %) [11, 13 15]. Baseline HbA 1c was poorly controlled (range ) and study participants were either overweight or obese at baseline (range mean BMI ). Study duration differed (range 5 12 months) [11, 15], but was controlled for as a potential source of heterogeneity in the indirect comparison. Insulin dosing formulas varied ( table 1 ) with two s designating a dose of 10 international units (IU) [11, 14], and a third 10 or 20 IU, depending on prior OAD use [16]. The two remaining s determined an initial insulin dose for each patient according to a formula or specified target [13, 15]. Insulin titration differed slightly between studies, with fasting blood glucose targets reported as follows: ^ 100 mg/dl [13, 14], ^ 108 mg/dl [11], 120 mg/dl [15], and mg/dl [16]. Finally, insulin doses at study completion varied widely (range IU/day) [15, 16]. The absence of observable patterns between dosing data and study results across individual s indicates that nothing can be concluded about the influence of insulin dosing. Despite dosing differences, insulin glargine s were sufficiently similar with regard to insulin titration formulas, trial methodology, and patient baseline characteristics to warrant combination in a direct metaanalysis (see below). Mean weight gain in kilograms and number of patients experiencing at least one hypoglycaemic episode at study completion are reported in table 2 for each. Table 3 shows the baseline and end-of-study HbA 1c values for each. Weight Gain Direct Comparison: With Insulin NPH the Common Comparator A direct fixed-effects meta-analysis of four s found no significant difference between evening insulin glargine and insulin NPH on weight gain from baseline in type-2 diabetic patients (WMD 0.31 kg; 95% CI 0.04, 0.66 kg; p = 0.086) [13 16]. No significant heterogeneity was detected in this analysis ( 2 p = 0.297). One found that patients gained significantly less weight when taking evening insulin detemir compared to NPH (WMD 0.91 kg; 95% CI 1.77, 0.05 kg; p = 0.038) [11]. Indirect Comparison: Insulin Detemir versus Insulin Using meta-regression, an indirect comparison found that evening insulin detemir patients experienced significantly less weight gain compared to evening glargine patients (unadjusted model: WMD 1.22 kg; 95% CI 2.15, 0.29 kg; p = 0.010). The statistical significance of the reduced weight gain effect favouring detemir was replicated in successive regression models adjusting for age, study duration, and duration of prior OAD treatment ( table 4 ). 158 Pharmacology 2008;82: Fakhoury /Lockhart /Kotchie /Aagren / LeReun

4 Ta b l e 1. Baseline characteristics and dosing information from the five s included for indirect comparison Study characteristics Philis-Tsimikas et al. [11], 2006 Fritsche et al. [13], 2006 Riddle et al. [14], 2003 et al. [15], 2000 et al. [16], 2006 Study design Population Sample size n = 333 n = 459 n = 764 n = 426 n = 110 Interventions Mean age, years ( 8 SD) n = 169 OAD + ID p.m. 1 n = 164 OAD + NPH p.m. 2 n = 227 OAD + IG p.m. 3 n = 232 OAD + NPH p.m. n = 367 OAD + IG p.m. n = 389 OAD + NPH p.m. n = 214 OAD + IG p.m. n = 208 OAD + NPH p.m. ID p.m.: IG p.m.: IG p.m.: IG p.m.: (SEM) NPH p.m.: NPH p.m.: NPH p.m.: NPH p.m.: (SEM) n = 61 OAD + IG p.m. n = 49 OAD + NPH p.m. IG p.m.: (SE) NPH p.m.: (SE) Male/female, n ID p.m.: 91/78 IG p.m.: 132/95 IG p.m.: 202/165 IG p.m.: 118/96 IG p.m.: 38/23 NPH p.m.: 94/70 NPH p.m.: 119/113 NPH p.m.: 218/171 NPH p.m.: 110/98 NPH p.m.: 32/17 Baseline mean BMI Study completers, % ID p.m.: (SD) IG p.m.: (SD) IG p.m.: (SD) NPH p.m.: 30.4 NPH p.m.: 28.9 NPH p.m.: (SD) (SD) (SD) IG p.m.: (SEM) NPH p.m.: (SEM) IG p.m.: (SE) NPH p.m.: (SE) ID p.m.: 90.6% IG p.m.: 99.5% IG p.m.: 91% IG p.m.: 89.2% IG p.m.: 98% NPH p.m.: 89.6% NPH p.m.: 100% NPH p.m.: 91.8% NPH p.m.: 83.3% NPH p.m.: 98% Study duration 20 weeks 24 weeks 24 weeks 52 weeks 36 weeks OAD dose Insulin initial dose, all treatment arms Insulin titration formula, all treatment arms unless otherwise stated Mean insulin dose at study completion At least half recommended maximum doses (for metformin or insulin secretagogue mono or combination therapy + TZD in USA study sites) Glimepiride 3 mg daily 10 IU (FBG value 50 mg/dl) & 10 mg/dl ID p.m. and NPH p.m.: Increase dose by 2, 4, 6, 8 IU, respectively, if FBG , , , >180 mg/dl to achieve pre-breakfast FBG levels of 108 mg/dl ID p.m.: (SD) IU/day at 20 weeks NPH p.m.: (SD) IU/day at 20 weeks Increase dose by 2, 4, 6, 8 IU, respectively, if FBG >100, 120, 140, 160 mg/ dl on 1 of 2 consecutive days to achieve FBG levels of 100 mg/dl IG p.m.: (SD) IU/day at 6 months (24 weeks) NPH p.m.: (SD) IU/day at 6 months (24 weeks) Stable doses of one or two (sulphonylureas, metformin, pioglitazone, or rosiglitazone) Pre-study doses of sulphonylureas alone or combined with acarbose, metformin, or metformin alone maintained 10 IU Discretion of investigator adjusted to achieve FBG target of 120 mg/dl Increase dose by 2, 4, 6, 8 IU, respectively, if FBG , , , 180 mg/dl on 1 of 2 consecutive days to achieve FBG levels of 100 mg/dl IG p.m.: (SE) IU/day at 6 months (24 weeks) NPH p.m.: (SE) IU/day at 6 months (24 weeks) Discretion of investigator adjusted to achieve FBG target of 120 mg/dl IG p.m.: (SEM) IU/day at 1 year (52 weeks) NPH p.m.: (SEM) IU/day at 1 year (52 weeks) Any stable dose of metformin ( 1.5 g) 10 IU (prior metformin only); 20 IU (prior metformin + sulphonylurea) Increase dose by 2 IU if FBG >100 mg/dl or 4 IU if FBG >180 mg/dl on 3 consecutive mornings to achieve FBG of mg/dl IG p.m.: (SE) IU/day at 9 months NPH: (SE) IU/ day at 9 months T2D = Type-2 diabetes; ID = insulin detemir; IG = insulin glargine; NPH = neutral protamine Hagedorn; p.m. = evening dose; TZD = thiazolidinedione; FBG = fasting blood glucose. 1 Insulin detemir evening dose; 2 insulin NPH evening dose; 3 insulin glargine evening dose. Indirect Comparison of Detemir and Pharmacology 2008;82:

5 Ta b l e 2. Individual study results for patients experiencing hypoglycaemic episodes and mean weight gain (kg) at study completion: evening insulin detemir, insulin NPH, and insulin glargine Study reference Patients experiencing at least one episode of hypoglycaemia, n OR (95% CI) Mean weight gain at study completion, kg (8 SE) WMD (95% CI) Philis-Tsimikas et al. [11], 2006 Fritsche et al. [13], 2003 Riddle et al. [14], 2003 et al. [15], 2000 et al. [16], 2006 Evening detemir: 27/169 (16%) Detemir vs. NPH: Evening detemir: Evening NPH: 53/164 (32%) 0.40 (0.24, 0.67) Evening NPH: Evening glargine: 155/227 (68%) vs. NPH Evening glargine: Evening NPH: 173/232 (75%) 0.88 (0.41, 1.89) Evening NPH: Detemir vs. NPH: 0.91 ( 1.77, 0.05) vs. NPH: 0.80 (0.08, 1.52) Evening glargine: not reported N/A Evening glargine: vs. NPH: Evening NPH: not reported Evening NPH: ( 0.35, 0.75) Evening glargine: not reported N/A Evening glargine: vs. NPH: Evening NPH: not reported Evening NPH: ( 0.41, 0.87) Evening glargine: 33/61 (54%) vs. NPH Evening glargine: Evening NPH: 28/49 (57%) 0.73 (0.41, 1.89) Evening NPH: vs. NPH: 0.90 ( 2.71, 0.91) Ta b l e 3. Baseline and study endpoint HbA 1c values for evening insulin detemir, insulin NPH, and insulin glargine Study reference Baseline HbA 1c HbA1c at study endpoint Philis-Tsimikas et al. [11], 2006 Fritsche et al. [13], 2003 Riddle et al. [14], 2003 et al. [15], 2000 et al. [16], 2006 Evening detemir (SD) Evening detemir (SD) Evening NPH (SD) Evening NPH (SD) Evening glargine (SD) Evening glargine (SD) Evening NPH (SD) Evening NPH (SD) Evening glargine (SD) Evening glargine 6.96 Evening NPH (SD) Evening NPH 6.97 Evening glargine (SEM) Evening glargine (SEM) Evening NPH (SEM) Evening NPH (SEM) Evening glargine (SE) Evening glargine (SE) Evening NPH (SE) Evening NPH (SE) Ta b l e 4. Weight gain indirect comparison of evening insulin detemir versus evening insulin glargine unadjusted and adjusted models Weight gain: Evening detemir vs. evening glargine WMD 95% CI p value Unadjusted model , Model adjusted on age (age NS, p = 0.209) , Model adjusted on duration (duration NS, p = 0.601) , Model adjusted on duration group 1 (group 1 NS, p = 0.419) , Model adjusted on duration group 2 (group 2 NS, p = 0.797) , Model adjusted on duration of OAD treatment (OAD NS, p = 0.303) , Duration group 1: duration treated as a categorical variable (20 24, 36 and 52 weeks). Duration group 2: flag for duration = 52 weeks compared to other durations (20, 24 and 36 weeks). 160 Pharmacology 2008;82: Fakhoury /Lockhart /Kotchie /Aagren / LeReun

6 Ta b l e 5. HbA 1c at study completion indirect comparison of evening insulin detemir versus evening insulin glargine unadjusted and adjusted models SMD bedtime detemir vs. bedtime glargine SMD 95% CI p value Unadjusted model , Model adjusted on age (age NS, p = 0.474) , Model adjusted on duration (duration NS, p = 0.209) , Model adjusted on duration group 1 (group 1 NS, p = 0.592) , Model adjusted on duration group 2 (group 2 NS, p = 0.215) , Model adjusted on duration of OAD treatment (OAD NS, p = 0.223) , Model adjusted on baseline HbA 1c (HbA 1c NS, p = 0.749) , Duration group 1: duration treated as a categorical variable (20 24, 36 and 52 weeks). Duration group 2: flag for duration = 52 weeks compared to other durations (20, 24 and 36 weeks). Hy p og lyc a e mi a Direct Comparison: With Insulin NPH the Common Comparator A fixed-effects meta-analysis of two s found no significant difference between insulin glargine and insulin NPH in the number of patients experiencing at least one episode of hypoglycaemia at study completion (OR 0.77; 95% CI 0.53, 1.10; p = 0.143) [13, 16]. No significant heterogeneity was detected in this analysis ( 2 p = 0.672). Significantly fewer evening insulin detemir patients reported at least one hypoglycaemic episode at study completion compared with insulin NPH patients (OR 0.40; 95% CI 0.24, 0.67; p = 0.001) [11]. Indirect Comparison: Insulin Detemir versus Insulin In the indirect comparison, significantly fewer evening detemir patients experienced at least one episode of hypoglycaemia at study completion compared with insulin glargine patients (unadjusted OR 0.52; 95% CI 0.28, 0.98; p = 0.044). Due to the limited number of insulin glargine s reporting hypoglycaemia, it was not possible to perform additional covariate analyses for this outcome. HbA 1c Direct Comparison: With Insulin NPH the Common Comparator A direct fixed-effects meta-analysis of four s found no significant difference between insulin glargine and NPH treatments in mean HbA 1c levels at study endpoint (SMD 0.03; 95% CI 0.12, 0.07; p = 0.548) [13 16]. No significant heterogeneity was detected in this analysis ( 2 p = 0.391). One found no significant difference between insulin detemir and NPH treatments in mean HbA 1c levels at study endpoint (SMD 0.06; 95% CI 0.16, 0.27; p = 0.593) [11]. Indirect Comparison: Insulin Detemir versus Insulin The indirect comparison of insulin detemir versus insulin glargine found no significant difference between treatment arms in mean HbA 1c levels at study endpoint (unadjusted SMD 0.09; 95% CI 0.16, 0.33; p = 0.480). This lack of statistical significance was replicated in subsequent regression models that adjusted for age, study duration, duration of prior OAD therapy, and baseline HbA 1c ( table 5 ). Discussion Once-daily basal insulin administered in conjunction with OAD is the most convenient form of insulin delivery in type-2 diabetic patients as it enables less frequent administration than shorter-acting insulin therapies. The indirect comparison demonstrates that compared to insulin glargine, use of once-daily evening insulin detemir is associated with significantly less weight gain and fewer episodes of hyperglycaemia. In addition, there was no significant difference in HbA 1c at study completion between the two insulin treatments. s of once-daily administration of long-acting insulins only were selected for two reasons. Firstly, from a patient perspective, once-daily long-acting insulin when taken with OAD (as opposed to a multiple daily basalbolus insulin regimen) represents a more convenient and flexible form of therapy, as insulin administration in- Indirect Comparison of Detemir and Pharmacology 2008;82:

7 volves fewer injections and need not be co-ordinated with meals. This insulin therapy is the most convenient option, provided control of weight, hypoglycaemia, and HbA 1c is not compromised. It was therefore important to select s with compatible dosing regimens, e.g. OAD plus once-daily insulin detemir or glargine, in order to indirectly compare performance on these clinical outcomes. Secondly, a previous study which compared once or twice daily insulin detemir plus insulin aspart versus once or twice daily NPH found no significant difference between the two insulin detemir regimens on control of HbA 1c level (p = 0.50) [9]. It was therefore of interest to assess whether a once-daily regimen of insulin determir could maintain adequate control of weight gain, hypoglycaemic episodes, and HbA 1c in type-2 diabetic patients taking OAD, compared with insulin glargine. The s included in the indirect comparison are open-label, which is a limitation. Both insulin detemir and glargine are clear solutions, whereas insulin NPH is a cloudy solution, so blinding would have been difficult to accomplish. Although the lack of patient or study investigator blinding would not affect the objective measurement of weight gain and HbA 1c levels, it is unclear what impact it had on the reporting of hypoglycaemic episodes. In patients knowingly taking insulin, fear of hypoglycaemic episodes could have influenced experience and resulted in more self-reports of these events. However, provided randomization was effective, the risk of an inflated number of hypoglycaemic events would be expected to be evenly distributed between treatment arms, thereby avoiding a treatment-specific bias. Since no head-to-head s were found that maintained all insulin detemir recipients on a once-daily regimen versus insulin glargine, an indirect comparison of these treatments was conducted, with insulin NPH the common comparator. Meta-regression, a well-established method for conducting indirect comparisons, was used as it allows for the generation of covariate-adjusted models, which control for potential sources of heterogeneity. This is not possible with simpler approaches. A potential limitation of this method was the small number of published trials (n = 5) focusing on once-daily detemir or glargine, which could have limited the statistical power of the indirect comparison. However, an alternative Bayesian approach, or a simple comparison of the pooled insulin glargine outcomes with those of the single detemir trial, would have been equally limited by the number of studies available. The present analysis therefore represents an appropriate synthesis of the current published evidence on this topic. The direct meta-analysis of insulin glargine trials included an assessment of whether there was significant heterogeneity between studies. For the weight gain, hypoglycaemia, and HbA 1c analyses, no significant between-study heterogeneity was found. For the indirect comparison, analyses of the weight gain and HbA 1c outcomes controlled for heterogeneity by adjusting for key study covariates (age, study duration, prior OAD use), with none of the latter reaching statistical significance. A potential limitation of the indirect comparison were the differences in study duration (range 5 12 months), with the single insulin detemir being the shortest [11]. However, the indirect comparison assessed study duration as a covariate for the weight gain and HbA 1c outcomes, and it was found to have no significant impact on either (weight gain p = 0.601; HbA 1c p = 0.209). Since only two insulin glargine studies reported hypoglycaemic episodes [13, 16], running adjustment models for key study covariates was not possible for this outcome. However, longer study duration did not appear to be associated with a cumulatively greater number of hypoglycaemic episodes reported by patients. A study of 36 weeks duration [16] reported that 54% of patients on insulin glargine experienced at least one episode of hypoglycaemia compared with 68% in a study that lasted just 24 weeks [13]. Therefore, study duration differences do not appear to directly influence the numbers of patients reporting hypoglycaemic episodes. Finally, there were between-study differences in the clinical definition of hypoglycaemia with the following blood glucose cut-off values reported:! 3.1 mmol/l (56 mg/dl) [11], ^ 4 mmol/l [16], and! 4.2 mmol/l (! 75 mg/ dl) [13]. Higher percentages of patients with at least one hypoglycaemic episode were observed in the glargine treatment arms (74% in 24 weeks, and 54% in 36 weeks) [13, 16] than in the detemir arm (16% in 20 weeks) [11]. The latter study had a more restrictive definition of hypoglycaemia. However, it is unlikely that the wide differences in the percentage of patients reporting at least one hypoglycaemic episode in the 20-week detemir study (16%) versus the 24-week glargine study (74%) can be attributed solely to different definitions of hypoglycaemia. Furthermore, there is no linear relationship between longer study duration and the percentage of patients who experienced hypoglycaemia, since the longer glargine study had fewer patients reporting episodes than the shorter study, despite both having similar clinical definitions of hypoglycaemia. The differences in clinical definitions and in study duration do not appear to account for the wide variation in hypoglycaemic event rates ob- 162 Pharmacology 2008;82: Fakhoury /Lockhart /Kotchie /Aagren / LeReun

8 served between the detemir and glargine studies, suggesting that insulin treatment may be the main factor accounting for this difference. Co n c l u s i o n The current study presents the best available evidence for the comparison of once-daily insulin detemir versus insulin glargine in patients with type-2 diabetes taking. It confirms that detemir use in this population significantly minimises weight gain and reduces the number of patients experiencing hypoglycaemic episodes compared to insulin glargine, with no significant difference in control of HbA 1c between the two treatments. Insulin detemir is therefore the most appropriate once-daily long-acting insulin for type-2 diabetic patients with poor glycaemic control on, who are overweight and/or anxious about increased weight gain and incidence of hypoglycaemia. References 1 Carver C: Insulin treatment and the problem of weight gain in type 2 diabetes. Diabetes Educ 2006; 32: UK Prospective Diabetes Study (UKPDS) Group: Intensive blood-glucose control with sulphonylureas or insulin compared with conventional treatment and risk of complications in patients with type 2 diabetes (UK- PDS 33). Lancet 1998; 352: Jacob AN, Salinas K, Adams-Huet B, Raskin P: Weight gain in type 2 diabetes mellitus. Diabetes Obes Metab 2007; 9: Hermansen K, Davies M: Does insulin detemir have a role in reducing risk of insulinassociated weight gain? Diabetes Obes Metab 2007; 9: Blonde L: State of diabetes care in the United States. Am J Manag Care 2007; 13(suppl 2): S36 S40. 6 Cryer PE, Davis SN, Shamoon H: Hypoglycemia in diabetes. Diabetes Care 2003; 26: Korytkowski M: When oral agents fail: practical barriers to starting insulin. Int J Obes Relat Metab Disord 2002; 26(suppl 3):S18 S24. 8 Meneghini L: Why and how to use insulin therapy earlier in the management of type 2 diabetes. South Med J 2007; 100: Haak T, Tiengo A, Draeger E, Suntum M, Waldhausl W: Lower within-subject variability of fasting blood glucose and reduced weight gain with insulin detemir compared to NPH insulin in patients with type 2 diabetes. Diabetes Obes Metab 2005; 7: Hermansen K, Davies M, Derezinski T, Martinez Ravn G, Clauson P, Home P: A 26- week, randomized, parallel, treat-to-target trial comparing insulin detemir with NPH insulin as add-on therapy to oral glucoselowering drugs in insulin-naive people with type 2 diabetes. Diabetes Care 2006; 29: Philis-Tsimikas A, Charpentier G, Clauson P, Ravn GM, Roberts VL, Thorsteinsson B: Comparison of once-daily insulin detemir with NPH insulin added to a regimen of oral antidiabetic drugs in poorly controlled type 2 diabetes. Clin Ther 2006; 28: Raslova K, Tamer SC, Clauson P, Karl D: Insulin detemir results in less weight gain than NPH insulin when used in basal-bolus therapy for type 2 diabetes mellitus, and this advantage increases with baseline body mass index. Clin Drug Investig 2007; 27: Fritsche A, Schweitzer MA, Haring HU; 4001 Study Group: Glimepiride combined with morning insulin glargine, bedtime neutral protamine Hagedorn insulin, or bedtime insulin glargine in patients with type 2 diabetes. A randomized, controlled trial. Ann Intern Med 2003; 138: Riddle MC, Rosenstock J, Gerich J; Insulin 4002 Study Investigators: The treatto-target trial: randomized addition of glargine or human NPH insulin to oral therapy of type 2 diabetic patients. Diabetes Care 2003; 26: H, Dressler A, Ziemen M; HOE 901/300s Study Group: Less nocturnal hypoglycemia and better post-dinner glucose control with bedtime insulin glargine compared with bedtime NPH insulin during insulin combination therapy in type 2 diabetes. HOE 901/3002 Study Group. Diabetes Care 2000; 23: H, Kauppinen-Makelin R, Tiikkainen M, et al: Insulin glargine or NPH combined with metformin in type 2 diabetes: the LANMET study. Diabetologia 2006; 49: Indirect Comparison of Detemir and Pharmacology 2008;82:

9