Benefits and Harms of Once-Weekly Glucagon-Like Peptide-1
Total Page:16
File Type:pdf, Size:1020Kb
This online-first version will be replaced with a final version when it is included in the issue. The final version may differ in small ways. Annals of Internal Medicine REVIEW Benefits and Harms of Once-Weekly Glucagon-like Peptide-1 Receptor Agonist Treatments A Systematic Review and Network Meta-analysis Francesco Zaccardi, MD; Zin Zin Htike, MD; David R. Webb, PhD; Kamlesh Khunti, PhD; and Melanie J. Davies, MD Background: Once-weekly glucagon-like peptide-1 receptor exenatide and albiglutide for fasting plasma glucose (–0.7 agonists (GLP-1RAs) are new drugs for the treatment of type 2 mmol/L [CI, –1.1 to –0.2 mmol/L]; –12.6 mg/dL [CI, –19.8 to –3.6 diabetes. mg/dL]), and taspoglutide, 20 mg, and dulaglutide, 0.75 mg, for body weight (–1.5 kg [CI, –2.2 to –0.8]). Clinically marginal or no Purpose: To summarize evidence for the cardiometabolic effi- differences were found for blood pressure, blood lipid levels, cacy and adverse effects of once-weekly GLP-1RAs in adults with and C-reactive protein levels. Once-weekly exenatide increased type 2 diabetes. heart rate compared with albiglutide and dulaglutide (1.4 to 3.2 Data Sources: Electronic databases (PubMed, Web of Science, beats/min). Among once-weekly GLP-1RAs, the risk for hypogly- Cochrane Central Register of Controlled Trials, U.S. Food cemia was similar, whereas taspoglutide, 20 mg, had the great- and Drug Administration, European Medicines Agency, est risk for nausea (odds ratios, 1.9 to 5.9). ClinicalTrials.gov) and congress abstracts from inception Limitation: Data were unavailable for semaglutide, definitions through 26 September 2015. of outcomes were heterogeneous, the last-observation-carried- Study Selection: Randomized, controlled trials (≥24 weeks of forward imputation method was used in 73% of trials, and pub- follow-up) studying albiglutide, dulaglutide, once-weekly ex- lication bias is possible. enatide, semaglutide, and taspoglutide and reporting a cardio- Conclusion: Compared with other once-weekly GLP-1RAs, du- metabolic (primary outcome, hemoglobin A1c [HbA1c]) or safety laglutide 1.5 mg, once-weekly exenatide, and taspoglutide, 20 outcome. mg, showed a greater reduction of HbA1c, fasting plasma glu- Data Extraction: Extraction was done in duplicate, and risk of cose, and body weight. Taspoglutide, 20 mg, had the highest bias was assessed. No language restriction was applied. risk for nausea; risk for hypoglycemia among once-weekly GLP- 1RAs was similar. Data Synthesis: 34 trials (21 126 participants) were included. Compared with placebo, all once-weekly GLP-1RAs reduced Primary Funding Source: Sanofi Aventis (grant to the Univer- sity of Leicester). HbA1c and fasting plasma glucose; taspoglutide, 20 mg, once- weekly exenatide, and dulaglutide, 1.5 mg, reduced body Ann Intern Med. doi:10.7326/M15-1432 www.annals.org weight. Among once-weekly GLP-1RAs, the greatest differences For author affiliations, see end of text. were found between dulaglutide, 1.5 mg, and taspoglutide, 10 This article was published online first at www.annals.org on 8 December mg, for HbA1c (–0.4% [95% CI, –0.7% to –0.2%]), once-weekly 2015. ype 2 diabetes mellitus is a progressive metabolic cently, GLP-1RAs that are administered once weekly Tdisorder associated with an increased risk for vascu- have emerged, reducing the number of injections lar diseases and whose main and diagnostic character- and side effects and potentially improving patient istic is hyperglycemia (1–3). In the past decade, the compliance (6). number of available treatments for hyperglycemia has Several randomized, controlled trials (RCTs) have increased significantly (4). Glucagon-like peptide-1 re- assessed the efficacy and safety of once-weekly GLP- ceptor agonists (GLP-1RAs) are a relatively new class of 1RAs compared with daily GLP-1RAs or other glucose- drug that stimulate insulin and inhibit glucagon secre- lowering therapies. However, to date, no direct com- tion, slow gastric emptying, and reduce food intake (5). parisons between once-weekly GLP-1RAs have been In clinical studies, these drugs improve glucose control available. In the absence of direct evidence, network meta- and reduce body weight, without an increased risk for analysis is an increasingly used statistical methodology hypoglycemia (6). The American Diabetes Association that allows the estimation of the comparative effective- (ADA) and European Association for the Study of Dia- ness of multiple treatments (9, 10). In this context, we betes (EASD) recommend GLP-1RAs as a therapeutic conducted a systematic review and network meta- option in patients receiving metformin with or without another glucose-lowering treatment if individualized metabolic targets are not achieved (7). See also: The first 2 approved GLP-1RAs are administered as subcutaneous daily injections (twice-daily exenatide Web-Only and once-daily liraglutide). A direct comparison of Supplement these “daily” drugs has shown differences in terms of CME quiz tolerability, efficacy, and pharmacokinetics (8). More re- www.annals.org Annals of Internal Medicine 1 Downloaded From: http://annals.org/ by a Bibl Central Clinica Bologna User on 12/09/2015 This online-first version will be replaced with a final version when it is included in the issue. The final version may differ in small ways. REVIEW Comparative Clinical Profiles of Once-Weekly GLP-1RAs analysis to assess the comparative efficacy and safety of not possible to extract relevant information for the pri- the once-weekly GLP-1RAs albiglutide, dulaglutide, ex- mary outcome from published reports, we searched for enatide, semaglutide, and taspoglutide. trial results in ClinicalTrials.gov and contacted study authors. Study-level quality was assessed by using the items METHODS reported in the Cochrane risk-of-bias tool (random- Data Sources and Searches sequence generation, allocation concealment, blinding This study was done according to a prespecified of participants and personnel and outcome assess- protocol (Supplement 1, available at www.annals.org) ment, and incomplete data and selective reporting) and followed the standard guidelines for the conduct (14). If the 2 independent reviewers disagreed on the and reporting of systematic reviews and network meta- eligibility of an article, extracted information, or quality analysis (11–13). We searched the following databases assessment, consensus was reached by reevaluation of from inception to 26 September 2015: PubMed, ISI the article and consultation with a third reviewer. Web of Science, Cochrane Central Register of Con- trolled Trials, ClinicalTrials.gov, U.S. Food and Drug Data Synthesis and Analysis Administration (www.fda.gov), European Medicines Stata, version 14 (Stata Corp, College Station, Agency (www.ema.europa.eu/ema), and the abstract Texas), was used for all analyses, and results are re- databases of major diabetes conferences (ADA and ported with 95% CIs. We performed pairwise random- EASD from 2012 onward). Reference lists of eligible effects meta-analyses by using the Knapp–Hartung studies, as well as systematic reviews and meta- method with the metareg command (15). We did net- analyses of GLP-1RAs, were manually scanned for addi- work meta-analyses within a frequentist framework us- tional relevant studies. No language restriction was ap- ing the network suite (16). By using the mvmeta com- plied. Detailed information on the search strategy is mand (17), network performs network meta-analysis as provided in Supplement 1. multivariate random-effects meta-analysis and meta- Study Selection regression, as recently proposed (18); the analysis as- sumes that all treatment contrasts have the same heter- Phase 3 RCTs in adults with type 2 diabetes that ogeneity variance (18). We used the network rank lasted 24 weeks or more were included. We included option to estimate the ranking probabilities and the RCTs with at least 1 once-weekly GLP-1RA study group netleague command to report relative treatment effects (albiglutide, dulaglutide, exenatide, semaglutide, and for all pairwise comparisons estimated with the network taspoglutide), regardless of the comparator (placebo meta-analysis (19). or another glucose-lowering drug). We required that Because our goal was to assess differences by RCTs report data on cardiometabolic outcomes (pri- comparing once-weekly GLP-1RAs with each other, we mary outcome, HbA ; secondary outcomes, fasting 1c combined twice-daily exenatide and once-daily liraglu- plasma glucose; body weight; systolic and diastolic tide treatments into one group of daily GLP-1RA thera- blood pressure; heart rate; C-reactive protein; and pies. We also combined insulin glargine and detemir blood lipid values [total, low-density lipoprotein, and (basal insulin) treatments. We defined a single group high-density lipoprotein cholesterol and triglycerides]) for albiglutide because the majority of RCTs were de- or safety outcomes (documented or symptomatic hypo- signed to titrate the drug to 50 mg if necessary, and glycemic events; severe hypoglycemic events; nausea dose-specific data were available only for 2 studies (20, and injection-site reaction events; fatal and nonfatal 21); for these studies, we used data on the 50-mg dose. cardiovascular and cancer events; all-cause mortality). We report the characteristics and summary data of We excluded RCTs involving patients with chronic kid- included studies. For both pairwise and network meta- ney disease. Although the clinical trial program of ta- analyses, we used arm-specific mean differences from spoglutide was stopped in