Comparative cardiovascular outcomes in the era of novel anti-diabetic agents: a comprehensive network meta-analysis of 166,371 participants from 170 randomized controlled trials

Comparative cardiovascular outcomes in the era of novel anti-diabetic agents: a comprehensive... Background: Cardiovascular (CV ) safety of one anti‑ diabetic medication over another remains partially delineated. We sought to assess the comparative effect on CV outcomes among novel anti ‑ diabetic agents. Methods: This study was registered with the International Prospective Register of Systematic Reviews (CRD 42016042063). MEDLINE, EMBASE, and Cochrane Library Central Register of Controlled Trials were searched between Jan 1, 1980, and June 30, 2016. Randomized controlled trials comparing anti‑ diabetic drugs with other comparators in adults with type 2 diabetes were included. We used network meta‑ analysis to obtain estimates for the outcomes of interests. In addition, post hoc correlation analysis of severe hypoglycemia and primary outcome as per ranking order was conducted. Outcomes were major adverse cardiovascular events (MACE) and all‑ cause mortality. Results: A total of 170 trials (166,371 participants) were included. By class and by individual, sulfonylureas (SU) ranked last. Therefore, with SU as reference, categorically sodium‑ glucose co‑ transporter 2 inhibitor (SGLT2i), insulin (INS), glucagon‑ like peptide‑ 1 receptor agonist, and dipeptidyl peptidase 4 inhibitor were significantly superior in term of MACE; as were SGLT2i and INS in term of all‑ cause mortality. Moreover, ranking orders of MACE and all‑ cause mortality were both positively correlated with that of severe hypoglycemia risk (by individual: R = 0.3178, P = 0.018; by class: R = 0.2574, P = 0.038). Conclusions: Novel anti‑ diabetic agents possess favorable CV safety profile, despite small but robust differences between individuals. In addition, increase in CV risk was again shown to be partly attributable to a concomitant increase in the risk of severe hypoglycemia, for which SU performed the worst. Keywords: Cardiovascular, Meta‑ analysis, Mortality, Diabetes, Agents *Correspondence: xiaohpeng@mail.sysu.edu.cn; liaoxinx@mail.sysu.edu.cn Xiao‑ dong Zhuang, Xin He and Da‑ ya Yang contributed equally to this work Key Laboratory on Assisted Circulation, Ministry of Health, No. 58 Zhongshan 2nd road, Guangzhou 510080, People’s Republic of China Department of Endocrinology, The First Affiliated Hospital of Sun Yat‑ sen University, Guangzhou, People’s Republic of China Full list of author information is available at the end of the article © The Author(s) 2018. This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creat iveco mmons .org/licen ses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creat iveco mmons .org/ publi cdoma in/zero/1.0/) applies to the data made available in this article, unless otherwise stated. Zhuang et al. Cardiovasc Diabetol (2018) 17:79 Page 2 of 10 of incidence of outcomes mentioned in the next section, Background and reported number of patients and events in each treat- Cardiovascular (CV) safety of anti-diabetic medications ment group; treatment durations of 24  weeks or longer. had raised notable concern, so much so that, in Decem- There is no limitation of baseline treatments as long as ber 2008, The US Food and Drug Administration (FDA) they are comparable in all of the study arms. Diabetic issued a guidance statement for industries requiring patients with concomitant diseases or CV risk factors proof of CV safety for the recently approved novel anti- were also included, but these studies would be excluded diabetic medications. In fact, the benefits and risks of in sensitivity analysis. Studies that compared the different using one anti-diabetic medication over another remain dosages or forms of the same drug were excluded. Studies largely unknown. On the one hand, high-quality head- were excluded if they were crossover trials, quasi experi- to-head comparison trials with important clinical end- ments, non-randomized trials, or enrolled patients with points, including long-term CV morbidity and mortality type 1 diabetes or patients without diabetes but only INS in particular, are still lacking. On the other, most system- resistance. atic reviews and meta-analyses to date focused predomi- nately on an individual agent or limited classes of agents Novel anti‑diabetes agents and dosages [1–17]. In order to resolve this uncertainty, we performed Novel anti-diabetic agents refer to the following three a network meta-analysis to evaluate whether differ - classes: DPP4i, GLP1ra and SGLT2i. We only included ences in CV outcomes exist between novel anti-diabetic drugs that have been approved by either US FDA or medications, including dipeptidyl peptidase 4 inhibi- European Medicines Agency. Comparators can be PLA, tors (DPP4i), glucagon-like peptide-1 receptor agonists MET, SU, TZD, INS, and another novel anti-diabetic (GLP1ra), and sodium-glucose co-transporter 2 inhibi- drug mentioned above. The treatment arm for these tors (SGLT2i), and the more traditional classes of drugs, novel drugs that used recommended dosages was ana- including insulin (INS), metformin (MET), sulfonylureas lyzed (Additional file 1: S3). (SU) and thiazolidinedione (TZD). In doing so, we aimed at providing evidence-based hierarchies of the compara- Data extraction and quality assessment tive CV safety profiles among anti-diabetic agents. Outcomes of interest were major adverse cardiovascular events (MACE), which consisted of CV death, non-fatal Methods myocardial infarction (MI), non-fatal stroke, and unsta- Study design and protocol ble angina or hospitalization for unstable angina, and all- We followed a pre-specified study protocol (Additional cause mortality. We included severe hypoglycemia as an file  1: S1) and reported our results according to the outcome during the data extraction phase for post hoc Preferred Reporting Items of Systematic Reviews and analysis. Severe hypoglycemia was defined as hypoglyce - Meta-Analyses (PRISMA) statement [18]. This study is mia episode requiring the assistance of another person or registered with the International Prospective Register of medical assistance, regardless of documentation of blood Systematic Reviews (CRD 42016042063). Network meta- glucose. analysis integrates data from direct comparisons of treat- Two reviewers independently scanned the search ments within trials and from indirect comparisons of results by reading the titles and abstracts. Data extracted interventions assessed against a common comparator in included outcomes of interest, study characteristics different trials, to compare all investigated treatments. (registry number, name the first author, whether it was The network meta-analysis was based on a frequentist international study, number of study centers, treatment model [19]. duration), participant characteristics (mean age, concom- itant high risk factor, proportion of male patients), inter- Data sources and study selection vention details (type of drug, its dosage in each arm and We searched MEDLINE, EMBASE, and the Cochrane baseline drug used across arms). Library Central Register of Controlled Trials between The methodological quality of included RCTs was Jan 1, 1980, and June 30, 2016 (search strategy in Addi- assessed using the tool described in the Cochrane col- tional file  1: S2). In order to determine whether the study laboration handbook [20]. Briefly, this tool includes reported any event of interested outcomes, data on seven components, which are random sequence genera- http://www.clini caltr ials.gov were also checked if registry tion, allocation concealment, blinding of participants and number was provided. personnel, blinding of outcome assessment, incomplete Studies meeting the following criteria were included: outcome data, selective reporting and other sources of randomized controlled trial; individuals with type 2 dia- bias. Each of these components of every included study betes; comparison of anti-diabetic drugs with other posi- received a rating of “low risk”, “unclear”, or “high risk”. tive comparator drugs or placebo (PLA); had at least one Zhuang et al. Cardiovasc Diabetol (2018) 17:79 Page 3 of 10 Statistical analysis alogliptin, linagliptin, saxagliptin, sitagliptin, vildagliptin, Stata package (version 14) was applied for statistical albiglutide, dulaglutide, exenatide, liraglutide, lixisena- analyses, using the network and mvmeta command and tide, canagliflozin, dapagliflozin, empagliflozin, MET, SU, Stata routines described elsewhere [21]. For indirect and TZD, INS and PLA. Most trials (159 [93.5%] of 170) were mixed comparisons, we used network meta-analysis to two-armed studies. Studies characteristic and outcomes obtain estimates for the outcomes, and presented these were shown in Additional file  1: S5, S6. Treatment dura- estimates as odd ratios (OR) with 95% confidence inter - tion ranged from 24 to 208 weeks. Male patients ranged vals (CI). We then estimated the relative ranking prob- from 40.9 to 75%. Mean age of patients was ranged from ability of each treatment and obtained the treatment 44.0 to 74.9 years. Ten studies enrolled subjects with high hierarchy of competing interventions using rankograms, CV risk (65,650 patients), and 9 studies were exclusively surface under the cumulative ranking (SUCRA) curve, of patients with renal impairment (1349 patients). and mean ranks. Large SUCRA scores might indicate a more effective or safer intervention [22]. We showed the Study quality assessment results using SU as reference in interval plot because it The overall quality of studies was rated as good, even consistently ranked last. In addition, we chose not to pre- though some studies did not record details about rand- sent MET in the ranking as it was used as background omization and allocation concealment and there were treatment in most of the trials. only few randomized trials at low risk of bias in every To check for the presence of inconsistency, we used question-based entry (Additional file  1: S7). Moreover, the loop-specific approach that assesses the difference no major tendency was noted for smaller studies to over- between direct and indirect estimates for a specific estimate or underestimate active treatment effects on comparison in the loop (inconsistency factor) [23]. We outcomes in the comparison-adjusted funnel plot for the assumed a common heterogeneity estimate within each network (Additional file 1: S8). loop. We used the previously described node-splitting method, which separates evidence for a particular com- Mace and all‑cause mortality parison into direct and indirect, excluding one direct Networks of eligible comparisons (among individual comparison at a time and estimating the indirect treat- agents or classes of agents) for the CV outcomes were ment effect for the excluded comparison [24]. A com - presented in Fig.  1, showing predominantly pairwise parison adjusted funnel plot of treatment estimates for comparisons of drugs with DDP4i or PLA in classed drug on CV outcomes was used to assess for evidence of groups. Except for INS and SGLT2i, SGLT2i and GLP1ra, small-study effects. as well as INS and MET, direct evidence for MACE was To investigate the generalizability of the findings, we available for all the possible pairwise treatment com- assessed the effect of characteristics of trials and partici - parisons. However, such availability was lacking between pants on the outcomes in sensitivity analyses by exclud- individual drugs. ing studies with the following design characteristics: In the network meta-analyses, MACE were reported patients with high CV risk; patients with renal impair- in 152 studies (158,786 patients with 8702 MACE). ment; and sample size less than 100 in one arm. Comparative effects of all drugs were ranked with Finally, to explore the potential impact of severe hypo- SUCRA probabilities (Additional file  1: S9). Mixed glycemia on the association between anti-diabetic drugs comparisons were in the interval plot with SU as refer- and CV outcomes, additional correlation analysis of ence (Fig. 2) and the comparisons table (Fig. 3). In term severe hypoglycemia and outcome of interest according of MACE, mixed comparisons by drug class showed to the ranking order was conducted. that SGLT2i (OR 0.70, 95% CI 0.55–0.90), INS (0.71, 95% CI 0.57–0.90), GLP1ra (OR 0.76, 95% CI 0.61– 0.94), and DPP4i (OR 0.77, 95% CI 0.62–0.9) were sig- nificantly better than SU, and SGLT2i (OR 0.72, 95% Results CI 0.54–0.97) and INS (OR 0.73, 95% CI 0.56–0.97) The PRISMA flowchart showing electronic searching were superior to TZD. Mixed comparisons by indi- processes is shown in Additional file  1: S4. There were vidual drug showed that vildagliptin (OR 0.47, 95% CI 170 randomized controlled trials including 166,371 adults 0.25–0.90), lixisenatide (OR 0.72, 95%CI 0.55–0.95) eligible for the systematic review, reporting 8702 cases of and exenatide (OR 0.74, 95% CI 0.56–0.99) were sig- MACE (33.5 per 1000 patient-year) and 4914 cases of all- nificantly better than SU. Moreover, vildagliptin (OR cause mortality (18.3 per 1000 patient-year). Seven drug 0.49, 95% CI 0.25–0.94) was significantly superior to classes were compared with PLA or each other: DPP4i, TZD and lixisenatide (OR 0.49, 95% CI 0.25–0.94) was GLP1ra, SGLT2i, MET, SU, TZD and INS. For indi- significantly superior to albiglutide. By applying the vidual comparison, 18 treatment groups were analyzed: Zhuang et al. Cardiovasc Diabetol (2018) 17:79 Page 4 of 10 Fig. 1 Network plot of treatment comparisons for major adverse cardiovascular events (MACE) and all‑ cause mortality. a Categorized drugs comparisons for MACE; b Categorized drugs comparisons for all‑ cause mortality; c Individual drugs comparisons for MACE; d Individual drugs comparisons for all‑ cause mortality. The size of the nodes represents the number of trials that study the treatments. Direct comparison of treatments is linked with a line, the thickness of which represents the number of trials that assess the comparison. SGLT2i sodium‑ glucose co‑transporter 2 inhibitor(s), GLP1ra glucagon‑like peptide ‑1 receptor agonist(s), DPP4i dipeptidyl peptidase 4 inhibitor(s), TZD thiazolidinedione, MET metformin, SU sulfonylurea, INS insulin, PLA placebo, VIL vildagliptin, EMP empagliflozin, LIX lixisenatide, ALO alogliptin, EXE exenatide, LIR liraglutide, CAN canagliflozin, DAP dapagliflozin, DUL dulaglutide, SIT sitagliptin, LIN linagliptin, ALB albiglutide, SAX saxagliptin design-by-treatment inconsistency model, we detected significantly superior to albiglutide (OR 0.84, 95% CI inconsistency in only one loop of comparisons: SU- 0.73–0.97) and saxagliptin (OR 0.77, 95% CI 0.63–0.94). TZD (P = 0.042) (Additional file 1 : S10). We did not observe any inconsistencies between evi- All-cause mortality was reported in 139 studies dence derived from direct to indirect comparisons for (159,722 patients with 4914 death). Mixed comparisons all-cause mortality using the design-by-treatment incon- were in the interval plot with SU as reference (Fig.  2) sistency model (Additional file 1: S10). and the comparisons table (Fig.  3). In term of all-cause mortality, mixed comparisons by drug class showed that Sensitivity analyses and post hoc correlation analysis SGLT2i was significantly better than GLP1ra (OR 0.72, Results for MACE were generally robust in sensitivity 95% CI 0.57–0.91), DPP4i (OR 0.72, 95% CI 0.60–0.87), analyses by excluding studies with the following design TZD (OR 0.64, 95% CI 0.42–0.97) and SU (OR 0.58, characteristics: patients with high CV risk; patients with 95% CI 0.41–0.83). Moreover, INS was significantly bet - renal impairment; and sample number less than 100 ter than DPP4i (OR 0.87, 95% CI 0.77–0.97) and SU (OR in one arm. After the sensitivity analyses, changes in 0.70, 95% CI 0.50–0.97). Mixed comparisons by individ- ORs and rankings, either categorical drugs or individ- ual drug showed that exenatide was significantly better ual drugs, did not alter the primary results appreciably than albiglutide (OR 0.68, 95% CI 0.57–0.82), saxagliptin (Additional file 1: S11). (OR 0.62, 95% CI 0.49–0.79), linagliptin (OR 0.55, 95% As is shown in the post hoc correlation analysis (Fig. 4), CI 0.35–0.87), TZD (OR 0.62, 95% CI 0.40–0.97) and for individual drugs, the ranking order of MACE or SU (OR 0.57, 95% CI 0.38–0.85), and lixisenatide was all-cause mortality was positively correlated with the Zhuang et al. Cardiovasc Diabetol (2018) 17:79 Page 5 of 10 Fig. 2 Forest plot for MACE and all‑ cause mortality of anti‑ diabetic agents compared with sulfonylurea (individual and categorized agents). Treatments are ranked by surface under the cumulative ranking (SUCRA) values. OR odds ratio, CrI credibility interval Zhuang et al. Cardiovasc Diabetol (2018) 17:79 Page 6 of 10 Fig. 3 Mixed comparison results of anti‑ diabetic agents for MACE and all‑ cause mortality, both for individual (above table) and for categorized agents (below table). Agents are reported in order of MACE ranking. Treatment at the top left corner ranks first, while the one at the bottom right corner ranks last. OR lower than 1 favor the column‑ defining treatment. Anti‑ diabetic agents in one class are painted with the same color. Significant results are in bold other individual or classes of drugs, SU are associated ranking order of severe hypoglycemia (R = 0.3178, with the highest risks of MACE and all-cause mortal- P = 0.018; R = 0.2574, P = 0.038, respectively), whereas ity. Finally, the ranking of CV risk was linearly corre- for drug classes, a similar trend was observed with TZD lated with the ranking of severe hypoglycemia risk by as an outlier. individual comparisons, with SU displaying the highest risks in both endpoints. Discussion Our study found that the newer agents in general To the best of our knowledge, our network meta-anal- showed favorable CV safety, yet there are discrepan- ysis represents the most comprehensive synthesis of cies between individual and class comparisons. In a data currently available with regard to CV outcomes recently published meta-analysis, the DPP4i vildaglip- in pharmacologically managed patients with type 2 tin was found to significantly reduce the risk of MI and diabetes. Our findings can be summarized as follows: stroke, while other agents in the same class seemed to first, among anti-diabetic agents included in the net perform less well in terms of CV outcomes [3]. In two work, SGLT2i in class comparisons, and vildagliptin recent studies, the use of DPP4i was found to be asso- in individual comparisons, respectively ranked first in ciated with improved long-term survival in diabetic terms of MACE. Furthermore, when compared with Zhuang et al. Cardiovasc Diabetol (2018) 17:79 Page 7 of 10 Fig. 4 Correlation analyses showing correlation between the ranking order of MACE and all‑ cause mortality risk and the ranking order of severe hypoglycemia risk. Color of circle represents different drugs shown above. Area of a circle reflects sample size. a, b, c, d The panels indicates the correlation relationship between severe hypoglycemia and outcome of interest according to the ranking order patients surviving a myocardial infarction [15] whereas valid, and better informed. Such a “targeted” strategy, its increase in overall risks of heart failure or exhibit nonetheless, ought not to have indiscreetly relied too within-class differences remains unresolved [7 ]. These much, if at all, on a “known class effect”, but rather should results were reiterated in our study, in that despite be individualized and outcome-specific, if (and only if ) vildagliptin displayed the best CV safety profile indi - the signals of harm were detected in pre-approval pack- vidually, and in class comparisons the ranking of DPP4i age or post-approval monitoring of these newer agents. actually dropped to the fifth place in order. Other dis - On the other hand, albeit primary results being crepancies are also identifiable and can be resolved “mixed”, our analysis had once again confirmed that SU similarly. In light of the mixed results, therefore, a case were associated with the highest risks of MACE and all- can perhaps be made against a “class effect” in the era of cause mortality. In fact, SU steadily brought up the rear novel anti-diabetic medications, namely, the fact that a in both individual and class ranking, even after sensitivity well-documented better (or worse) CV safety profile of analysis. When plotting the ORs of MACE or all-cause one individual agent does not necessarily justify extrap- mortality for all other comparator drugs against SU, indi- olation of such a benefit (or harm) to other agents in vidually, point estimates concordantly lie to the left of the same class. the “line of no effect”; collectively, by order of effect size, These results have practical implications. Several SGLT2i, INS, GLP1ra and DPP4i were significantly better appeals have recently been made for an appraisal of the than SU, indicating that SU in actuality possess the worst current paradigm to evaluate CV risks of novel anti- CV safety profile among these medications. diabetic medications via large-scale, long-term CV SU are currently the most widely used medications safety trials [25, 26]. The authors argued for alternative for type 2 diabetes second only to MET. However, the approaches that are more cost-effective, more externally undesirable effect of weight gain [27], the greatest risk of Zhuang et al. Cardiovasc Diabetol (2018) 17:79 Page 8 of 10 iatrogenic hypoglycemia [28–31], a potential increase in due to heterogeneity of study designs and reporting CV morbidity and mortality [32–36], and adding to that, styles in the included studies, comparison of A1c lev- the advent of novel anti-diabetic medications with argu- els as well as the degree of CV risk at baseline across ably equal glucose-lowering effectiveness [37], all render different trials were unavailable. In addition, only a few SU as less favorable [38]. In a recent commentary, the studies reported outcomes such as acute coronary syn- role of SU in the era of novel anti-diabetic medications drome and CV death, and most had few or zero events. was thoughtfully challenged [39]. And according to the Some subgroup analyses had small numbers of partici- latest management guideline jointly issued by the Ameri- pants, likely resulting in poor precision of estimates. can Association of Clinical Endocrinologists and the Second, many of the international multicenter trials American College of Endocrinologists, in combinational were conducted primarily in higher-income countries, regimens, the strength of recommendation for SU to be which would possibly interfere with the external valid- added on top of MET is the weakest [40]. ity of these results for lower-income settings. Third, One of the major controversies about SU is their CV follow-up duration in most studies was relatively too safety. For example, in the UK Prospective Diabetes short to draw any definitive conclusion for long-term Study (UKPDS), in patients treated with SU, there was a CV outcome. Fourth, as first-line data from these trials trend of a 16% decrease in MI at the end of the study, but being unaccessible to us, the present study was unable at 10-year follow up there was a significant 15% decrease to account for other possible factors correlated with in events in the same arm [41]. In a recent network meta- meaningful CV outcomes. Results from continuing analysis, the authors reported no significant differences trials would provide useful insight in answering these in the associations between nine classes of anti-diabetic questions. medications and the risk of CV or all-cause mortal- ity [37]. Unfortunately, however, due to low event rates in general as well as the statistical power being diluted Conclusions by multiple layers of analyses in particular, conclusions Our network meta-analyses showed that all three classes about true effects of the studied drugs on CV or all-cause of novel anti-diabetic medications, i.e. DPP4i, GLP1ra, mortality were far from precise. The contribution of our and SGLT2i, possess favorable CV safety profile in gen - study, then, is having further clarified the current confu - eral, notwithstanding that there are small but robust dif- sion by offering an evidence-based hierarchy of CV safety ferences among individual drugs. These results refute a profile among all major anti-diabetic medications, which simplistic rationale of generalizing the CV benefit of one revealed the steady truth that SU are associated with the single agent to the others in the same class. In addition, highest risks of MACE or all-cause mortality when com- we also observed that SU were associated with the high- pared with other individual or classes of agents. est risk of MACE and all-cause mortality, which could In our post hoc correlation analysis, the ranking of potentially explained by its concomitant increase in the MACE and mortality risk were both linearly correlated risk of severe hypoglycemia. Such correlation should with the ranking of severe hypoglycemia risk by individ- probably call for a reassessment of the role of SU as first- ual comparisons, with SU at the top of risks, corroborat- line additive to MET in the pharmacological management ing the already well-founded connection, if not causality, of type-2 diabetes. These findings should be considered between the elevated risk of iatrogenic hypoglycemia in policy-making and the development of clinical practice with the use of SU and the inferior CV safety of this class guidelines. of traditional anti-diabetic agents, although the effect of Additional file “drug–drug interaction” cannot be completely ruled out, because many of these trials had multi-drug regimens. In Additional file 1: S1. Protocol. S2. Search strategy. S3. Novel drugs addition, it is of note that TZD are shown to be an out- approved by FDA or European Medicines Agency. S4. Flow chart of the lier in both correlations, with higher ranking of CV risk study selection process. S5. Studies characteristic of the included studies. but lower ranking of hypoglycemia. Such a result could S6. Outcomes of interest in each study. S7. Quality assessment of the included studies. S8. Comparison‑adjusted funnel plot for the network. be explained by the previous observation that the use of S9. Ranking ordered according to surface under the cumulative ranking TZD (especially rosiglitazone) is associated with poten- values of outcomes. S10. Consistency analysis of direct verse indirect tial increase of CV risks independent of hypoglycemia comparisons for outcomes. S11. Summaries of sensitivity analysis. [42]. Our study is limited in several ways. First, although Abbreviations comprehensive systematic search strategies were CV: cardiovascular; FDA: The US Food and Drug Administration; DPP4i: employed, our analyses were limited by the modest dipeptidyl peptidase 4 inhibitor; GLP1ra: glucagon‑like peptide ‑1 receptor agonist; SGLT2i: sodium‑ glucose co‑transporter 2 inhibitor; INS: insulin; MET: amount of data in the included studies. To begin with, Zhuang et al. Cardiovasc Diabetol (2018) 17:79 Page 9 of 10 metformin; SU: sulfonylurea; TZD: thiazolidinedione; PLA: placebo; MACE: 6. Ferdinand KC, Botros FT, Atisso CM, Sager PT. Cardiovascular safety for major adverse cardiovascular event; MI: myocardial infarction; OR: odds ratio; once‑ weekly dulaglutide in type 2 diabetes: a pre‑specified meta‑analysis CI: confidence interval; SUCRA : surface under the cumulative ranking. of prospectively adjudicated cardiovascular events. Cardiovasc Diabetol. 2016;15:38. https ://doi.org/10.1186/s1293 3‑016‑0355‑z. Authors’ contributions 7. Verma S, Goldenberg RM, Bhatt DL, Farkouh ME, Quan A, Teoh H. Dipepti‑ LXX, XHP and ZXD: designed the study and wrote the paper, collected and dyl peptidase‑4 inhibitors and the risk of heart failure: a systematic review analyzed the data. ZXD, HX and YDY conducted the search, collected the data, and meta‑analysis. CMAJ Open. 2017;5:E152–77. https ://doi.org/10.9778/ performed the analysis and wrote the manuscript. All authors critically revised cmajo .20160 058. the manuscript. All authors read and approved the final manuscript. 8. Zhang Z, Chen X, Lu P, Zhang J, Xu Y, He W. Incretin‑based agents in type 2 diabetic patients at cardiovascular risk: compare the effect of GLP ‑1 Author details agonists and DPP‑4 inhibitors on cardiovascular and pancreatic out ‑ Department of Cardiology, The First Affiliated Hospital of Sun Yat ‑sen Univer ‑ comes. Cardiovasc Diabetol. 2017;16:31. https ://doi.org/10.1186/s1293 sity, Guangzhou, People’s Republic of China. Key Laboratory on Assisted Cir‑ 3‑017‑0512‑z. culation, Ministry of Health, No. 58 Zhongshan 2nd road, Guangzhou 510080, 9. Hoogwerf BJ, Lincoff AM, Rodriguez A, Chen L, Qu Y. Major adverse cardi‑ People’s Republic of China. Department of Endocrinology, The First Affiliated ovascular events with basal insulin peglispro versus comparator insulins Hospital of Sun Yat‑sen University, Guangzhou, People’s Republic of China. in patients with type 1 or type 2 diabetes: a meta‑analysis. Cardiovasc Diabetol. 2016;15:78. https ://doi.org/10.1186/s1293 3‑016‑0393‑6. Acknowledgements 10. Tang H, Fang Z, Wang T, Cui W, Zhai S, Song Y. Meta‑analysis of effects of We thank Miss. Yankun Sun for her assistance in image processing. sodium‑ glucose cotransporter 2 inhibitors on cardiovascular outcomes and all‑ cause mortality among patients with type 2 diabetes mel‑ Competing interests litus. Am J Cardiol. 2016;118:1774–80. https ://doi.org/10.1016/j.amjca The authors declare that they have no competing interests. rd.2016.08.061. 11. Fei Y, Tsoi MF, Kumana CR, Cheung TT, Cheung B. Network meta‑analysis Availability of data and materials of cardiovascular outcomes in randomized controlled trials of new anti‑ All data generated or analyzed during this study are included in this published diabetic drugs. Int J Cardiol. 2018;254:291–6. https ://doi.org/10.1016/j. article and its additional file.ijcar d.2017.12.039. 12. Tsioufis C, Andrikou E, Thomopoulos C, Papanas N, Tousoulis D. Oral Consent for publication glucose‑lowering drugs and cardiovascular outcomes: from the negative Not applicable. RECORD and ACCORD to neutral TECOS and promising EMPA‑REG. Curr Vasc Pharmacol. 2017;15:457–68. https ://doi.org/10.2174/15701 61114 Ethics approval and consent to participate66616 12081 50642 . Not applicable. 13. de Jong M, van der Worp HB, van der Graaf Y, Visseren F, Westerink J. Pioglitazone and the secondary prevention of cardiovascular disease. Funding A meta‑analysis of randomized‑ controlled trials. Cardiovasc Diabetol. This work was funded by China national nature science founda‑ 2017;16:134. https ://doi.org/10.1186/s1293 3‑017‑0617‑4. tion (81600206) and Guangdong national nature science foundation 14. Lam KS, Chow CC, Tan KC, Ma RC, Kong AP, Tong PC. Practical consid‑ (2016A030310140/20160903). erations for the use of sodium‑ glucose co‑transporter type 2 inhibi‑ tors in treating hyperglycemia in type 2 diabetes. Curr Med Res Opin. 2016;32:1097–108. https ://doi.org/10.1185/03007 995.2016.11616 08. Publisher’s Note 15. Wang MT, Lin SC, Tang PL, Hung WT, Cheng CC, Yang JS. The impact of Springer Nature remains neutral with regard to jurisdictional claims in pub‑ DPP‑4 inhibitors on long‑term survival among diabetic patients after first lished maps and institutional affiliations. acute myocardial infarction. Cardiovasc Diabetol. 2017;16:89. https ://doi. org/10.1186/s1293 3‑017‑0572‑0. Received: 21 February 2018 Accepted: 23 May 2018 16. Schernthaner G, Lehmann R, Prazny M, Czupryniak L, Ducena K, Fasching P. Translating recent results from the Cardiovascular Outcomes Trials into clinical practice: recommendations from the Central and Eastern Euro‑ pean Diabetes Expert Group (CEEDEG). Cardiovasc Diabetol. 2017;16:137. https ://doi.org/10.1186/s1293 3‑017‑0622‑7. References 17. Schnell O, Ryden L, Standl E, Ceriello A. Updates on cardiovascular 1. Sonesson C, Johansson PA, Johnsson E, Gause‑Nilsson I. Cardiovascular outcome trials in diabetes. Cardiovasc Diabetol. 2017;16:128. https ://doi. effects of dapagliflozin in patients with type 2 diabetes and different risk org/10.1186/s1293 3‑017‑0610‑y. categories: a meta‑analysis. Cardiovasc Diabetol. 2016;15:37. https ://doi. 18. Moher D, Liberati A, Tetzlaff J, Altman DG. Preferred reporting items for org/10.1186/s1293 3‑016‑0356‑y. systematic reviews and meta‑analyses: the PRISMA statement. J Clin Epi‑ 2. Fisher M, Petrie MC, Ambery PD, Donaldson J, Ye J, McMurray JJ. Cardio‑ demiol. 2009;62:1006–12. https ://doi.org/10.1016/j.jclin epi.2009.06.005. vascular safety of albiglutide in the Harmony programme: a meta‑analy‑ 19. Caldwell DM, Ades AE, Higgins JP. Simultaneous comparison of multiple sis. Lancet Diabetes Endocrinol. 2015;3:697–703. https ://doi.org/10.1016/ treatments: combining direct and indirect evidence. BMJ. 2005;331:897– S2213 ‑8587(15)00233 ‑8. 900. https ://doi.org/10.1136/bmj.331.7521.897. 3. Savarese G, D’Amore C, Federici M, De Martino F, Dellegrottaglie S, Mar‑ 20. Green S, Higgins JPT. Cochrane handbook for systematic reviews of ciano C. Eec ff ts of dipeptidyl peptidase 4 inhibitors and sodium‑ glucose interventions. Chichester: Wiley‑Blackwell; 2008. linked cotransporter‑2 Inhibitors on cardiovascular events in patients 21. White IR. Multivariate random‑ effects meta‑regression: updates to with type 2 diabetes mellitus: a meta‑analysis. Int J Cardiol. 2016;220:595– mvmeta. Stata J. 2011;11:255–70. 601. https ://doi.org/10.1016/j.ijcar d.2016.06.208. 22. Salanti G, Ades AE, Ioannidis JP. Graphical methods and numerical sum‑ 4. Monami M, Dicembrini I, Nardini C, Fiordelli I, Mannucci E. Eec ff ts of maries for presenting results from multiple‑treatment meta‑analysis: glucagon‑like peptide ‑1 receptor agonists on cardiovascular risk: a meta‑ an overview and tutorial. J Clin Epidemiol. 2011;64:163–71. https ://doi. analysis of randomized clinical trials. Diabetes Obes Metab. 2014;16:38– org/10.1016/j.jclin epi.2010.03.016. 47. https ://doi.org/10.1111/dom.12175 . 23. White IR, Barrett JK, Jackson D, Higgins JP. Consistency and inconsistency 5. Wu JH, Foote C, Blomster J, Toyama T, Perkovic V, Sundstrom J. Eec ff ts in network meta‑analysis: model estimation using multivariate meta‑ of sodium‑ glucose cotransporter‑2 inhibitors on cardiovascular events, regression. Res Synth Methods. 2012;3:111–25. https ://doi.org/10.1002/ death, and major safety outcomes in adults with type 2 diabetes: a jrsm.1045. systematic review and meta‑analysis. Lancet Diabetes Endocrinol. 2016;4:411–9. https ://doi.org/10.1016/S2213 ‑8587(16)00052 ‑8. Zhuang et al. Cardiovasc Diabetol (2018) 17:79 Page 10 of 10 24. Veroniki AA, Vasiliadis HS, Higgins JP, Salanti G. Evaluation of inconsist‑ 34. Fisman EZ, Tenenbaum A, Boyko V, Benderly M, Adler Y, Friedensohn A. ency in networks of interventions. Int J Epidemiol. 2013;42:332–45. https Oral antidiabetic treatment in patients with coronary disease: time‑ ://doi.org/10.1093/ije/dys22 2. related increased mortality on combined glyburide/metformin therapy 25. Smith RJ, Goldfine AB, Hiatt WR. Evaluating the cardiovascular safety of over a 7.7‑ year follow‑up. Clin Cardiol. 2001;24:151–8. new medications for type 2 diabetes: time to reassess? Diabetes Care. 35. Rao AD, Kuhadiya N, Reynolds K, Fonseca VA. Is the combination of sulfo‑ 2016;39:738–42. https ://doi.org/10.2337/dc15‑2237. nylureas and metformin associated with an increased risk of cardiovascu‑ 26. Steg PG, Roussel R. Randomized trials to evaluate cardiovascular safety lar disease or all‑ cause mortality?: a meta‑analysis of observational stud‑ of antihyperglycemic medications: a worthwhile effort? Circulation. ies. Diabetes Care. 2008;31:1672–8. https ://doi.org/10.2337/dc08‑0167. 2016;134:571–3. https ://doi.org/10.1161/CIRCU LATIO NAHA.116.02191 4. 36. Currie CJ, Poole CD, Evans M, Peters JR, Morgan CL. Mortality and other 27. Bolen S, Feldman L, Vassy J, Wilson L, Yeh HC, Marinopoulos S. Systematic important diabetes‑related outcomes with insulin vs other antihypergly‑ review: comparative effectiveness and safety of oral medications for type cemic therapies in type 2 diabetes. J Clin Endocrinol Metab. 2013;98:668– 2 diabetes mellitus. Ann Intern Med. 2007;147:386–99. 77. https ://doi.org/10.1210/jc.2012‑3042. 28. Bennett WL, Maruthur NM, Singh S, Segal JB, Wilson LM, Chatterjee R. 37. Palmer SC, Mavridis D, Nicolucci A, Johnson DW, Tonelli M, Craig JC. Comparative effectiveness and safety of medications for type 2 diabetes: Comparison of clinical outcomes and adverse events associated with an update including new drugs and 2‑ drug combinations. Ann Intern glucose‑lowering drugs in patients with type 2 diabetes: a meta‑analysis. Med. 2011;154:602–13. https ://doi.org/10.7326/0003‑4819‑154‑9‑20110 JAMA. 2016;316:313–24. https ://doi.org/10.1001/jama.2016.9400. 5030‑00336 . 38. Xu J, Rajaratnam R. Cardiovascular safety of non‑insulin pharmaco ‑ 29. Tschope D, Bramlage P, Binz C, Krekler M, Plate T, Deeg E. Antidiabetic therapy for type 2 diabetes. Cardiovasc Diabetol. 2017;16:18. https ://doi. pharmacotherapy and anamnestic hypoglycemia in a large cohort of org/10.1186/s1293 3‑017‑0499‑5. type 2 diabetic patients—an analysis of the DiaRegis registry. Cardiovasc 39. Genuth S. Should sulfonylureas remain an acceptable first ‑line add‑ on to Diabetol. 2011;10:66. https ://doi.org/10.1186/1475‑2840‑10‑66. metformin therapy in patients with type 2 diabetes? No, it’s time to move 30. Seaquist ER, Anderson J, Childs B, Cryer P, Dagogo‑ Jack S, Fish L. Hypogly‑ on! Diabetes Care. 2015;38:170–5. https ://doi.org/10.2337/dc14‑0565. cemia and diabetes: a report of a workgroup of the American Diabetes 40. Garber AJ, Abrahamson MJ, Barzilay JI, Blonde L, Bloomgarden ZT, Bush Association and the Endocrine Society. Diabetes Care. 2013;36:1384–95. MA. Consensus statement by the American Association of Clinical Endo‑ https ://doi.org/10.2337/dc12‑2480. crinologists and American College of Endocrinology on the comprehen‑ 31. Zimmerman BR. Sulfonylureas. Endocrinol Metab Clin N Am. sive type 2 diabetes management algorithm—2016 executive summary. 1997;26:511–22. Endocr Pract. 2016;22:84–113. https ://doi.org/10.4158/EP151 126.CS. 32. Roumie CL, Hung AM, Greevy RA, Grijalva CG, Liu X, Murff HJ. Compara‑ 41. Holman RR, Paul SK, Bethel MA, Matthews DR, Neil HA. 10‑ year follow‑ tive effectiveness of sulfonylurea and metformin monotherapy on car ‑ up of intensive glucose control in type 2 diabetes. N Engl J Med. diovascular events in type 2 diabetes mellitus: a cohort study. Ann Intern 2008;359:1577–89. https ://doi.org/10.1056/NEJMo a0806 470. Med. 2012;157:601–10. https ://doi.org/10.7326/0003‑4819‑157‑9‑20121 42. Cobitz A, Zambanini A, Sowell M, Heise M, Louridas B, McMorn S. A retro‑ 1060‑00003 . spective evaluation of congestive heart failure and myocardial ischemia 33. Tzoulaki I, Molokhia M, Curcin V, Little MP, Millett CJ, Ng A. Risk of car‑ events in 14,237 patients with type 2 diabetes mellitus enrolled in 42 diovascular disease and all cause mortality among patients with type 2 short‑term, double ‑blind, randomized clinical studies with rosiglitazone. diabetes prescribed oral antidiabetes drugs: retrospective cohort study Pharmacoepidemiol Drug Saf. 2008;17:769–81. https ://doi.org/10.1002/ using UK general practice research database. BMJ. 2009;339:b4731. pds.1615. Ready to submit your research ? Choose BMC and benefit from: fast, convenient online submission thorough peer review by experienced researchers in your field rapid publication on acceptance support for research data, including large and complex data types • gold Open Access which fosters wider collaboration and increased citations maximum visibility for your research: over 100M website views per year At BMC, research is always in progress. Learn more biomedcentral.com/submissions http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Cardiovascular Diabetology Springer Journals

Comparative cardiovascular outcomes in the era of novel anti-diabetic agents: a comprehensive network meta-analysis of 166,371 participants from 170 randomized controlled trials

Free
10 pages

Loading next page...
 
/lp/springer_journal/comparative-cardiovascular-outcomes-in-the-era-of-novel-anti-diabetic-reeayt1fib
Publisher
Springer Journals
Copyright
Copyright © 2018 by The Author(s)
Subject
Medicine & Public Health; Diabetes; Angiology; Cardiology
eISSN
1475-2840
D.O.I.
10.1186/s12933-018-0722-z
Publisher site
See Article on Publisher Site

Abstract

Background: Cardiovascular (CV ) safety of one anti‑ diabetic medication over another remains partially delineated. We sought to assess the comparative effect on CV outcomes among novel anti ‑ diabetic agents. Methods: This study was registered with the International Prospective Register of Systematic Reviews (CRD 42016042063). MEDLINE, EMBASE, and Cochrane Library Central Register of Controlled Trials were searched between Jan 1, 1980, and June 30, 2016. Randomized controlled trials comparing anti‑ diabetic drugs with other comparators in adults with type 2 diabetes were included. We used network meta‑ analysis to obtain estimates for the outcomes of interests. In addition, post hoc correlation analysis of severe hypoglycemia and primary outcome as per ranking order was conducted. Outcomes were major adverse cardiovascular events (MACE) and all‑ cause mortality. Results: A total of 170 trials (166,371 participants) were included. By class and by individual, sulfonylureas (SU) ranked last. Therefore, with SU as reference, categorically sodium‑ glucose co‑ transporter 2 inhibitor (SGLT2i), insulin (INS), glucagon‑ like peptide‑ 1 receptor agonist, and dipeptidyl peptidase 4 inhibitor were significantly superior in term of MACE; as were SGLT2i and INS in term of all‑ cause mortality. Moreover, ranking orders of MACE and all‑ cause mortality were both positively correlated with that of severe hypoglycemia risk (by individual: R = 0.3178, P = 0.018; by class: R = 0.2574, P = 0.038). Conclusions: Novel anti‑ diabetic agents possess favorable CV safety profile, despite small but robust differences between individuals. In addition, increase in CV risk was again shown to be partly attributable to a concomitant increase in the risk of severe hypoglycemia, for which SU performed the worst. Keywords: Cardiovascular, Meta‑ analysis, Mortality, Diabetes, Agents *Correspondence: xiaohpeng@mail.sysu.edu.cn; liaoxinx@mail.sysu.edu.cn Xiao‑ dong Zhuang, Xin He and Da‑ ya Yang contributed equally to this work Key Laboratory on Assisted Circulation, Ministry of Health, No. 58 Zhongshan 2nd road, Guangzhou 510080, People’s Republic of China Department of Endocrinology, The First Affiliated Hospital of Sun Yat‑ sen University, Guangzhou, People’s Republic of China Full list of author information is available at the end of the article © The Author(s) 2018. This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creat iveco mmons .org/licen ses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creat iveco mmons .org/ publi cdoma in/zero/1.0/) applies to the data made available in this article, unless otherwise stated. Zhuang et al. Cardiovasc Diabetol (2018) 17:79 Page 2 of 10 of incidence of outcomes mentioned in the next section, Background and reported number of patients and events in each treat- Cardiovascular (CV) safety of anti-diabetic medications ment group; treatment durations of 24  weeks or longer. had raised notable concern, so much so that, in Decem- There is no limitation of baseline treatments as long as ber 2008, The US Food and Drug Administration (FDA) they are comparable in all of the study arms. Diabetic issued a guidance statement for industries requiring patients with concomitant diseases or CV risk factors proof of CV safety for the recently approved novel anti- were also included, but these studies would be excluded diabetic medications. In fact, the benefits and risks of in sensitivity analysis. Studies that compared the different using one anti-diabetic medication over another remain dosages or forms of the same drug were excluded. Studies largely unknown. On the one hand, high-quality head- were excluded if they were crossover trials, quasi experi- to-head comparison trials with important clinical end- ments, non-randomized trials, or enrolled patients with points, including long-term CV morbidity and mortality type 1 diabetes or patients without diabetes but only INS in particular, are still lacking. On the other, most system- resistance. atic reviews and meta-analyses to date focused predomi- nately on an individual agent or limited classes of agents Novel anti‑diabetes agents and dosages [1–17]. In order to resolve this uncertainty, we performed Novel anti-diabetic agents refer to the following three a network meta-analysis to evaluate whether differ - classes: DPP4i, GLP1ra and SGLT2i. We only included ences in CV outcomes exist between novel anti-diabetic drugs that have been approved by either US FDA or medications, including dipeptidyl peptidase 4 inhibi- European Medicines Agency. Comparators can be PLA, tors (DPP4i), glucagon-like peptide-1 receptor agonists MET, SU, TZD, INS, and another novel anti-diabetic (GLP1ra), and sodium-glucose co-transporter 2 inhibi- drug mentioned above. The treatment arm for these tors (SGLT2i), and the more traditional classes of drugs, novel drugs that used recommended dosages was ana- including insulin (INS), metformin (MET), sulfonylureas lyzed (Additional file 1: S3). (SU) and thiazolidinedione (TZD). In doing so, we aimed at providing evidence-based hierarchies of the compara- Data extraction and quality assessment tive CV safety profiles among anti-diabetic agents. Outcomes of interest were major adverse cardiovascular events (MACE), which consisted of CV death, non-fatal Methods myocardial infarction (MI), non-fatal stroke, and unsta- Study design and protocol ble angina or hospitalization for unstable angina, and all- We followed a pre-specified study protocol (Additional cause mortality. We included severe hypoglycemia as an file  1: S1) and reported our results according to the outcome during the data extraction phase for post hoc Preferred Reporting Items of Systematic Reviews and analysis. Severe hypoglycemia was defined as hypoglyce - Meta-Analyses (PRISMA) statement [18]. This study is mia episode requiring the assistance of another person or registered with the International Prospective Register of medical assistance, regardless of documentation of blood Systematic Reviews (CRD 42016042063). Network meta- glucose. analysis integrates data from direct comparisons of treat- Two reviewers independently scanned the search ments within trials and from indirect comparisons of results by reading the titles and abstracts. Data extracted interventions assessed against a common comparator in included outcomes of interest, study characteristics different trials, to compare all investigated treatments. (registry number, name the first author, whether it was The network meta-analysis was based on a frequentist international study, number of study centers, treatment model [19]. duration), participant characteristics (mean age, concom- itant high risk factor, proportion of male patients), inter- Data sources and study selection vention details (type of drug, its dosage in each arm and We searched MEDLINE, EMBASE, and the Cochrane baseline drug used across arms). Library Central Register of Controlled Trials between The methodological quality of included RCTs was Jan 1, 1980, and June 30, 2016 (search strategy in Addi- assessed using the tool described in the Cochrane col- tional file  1: S2). In order to determine whether the study laboration handbook [20]. Briefly, this tool includes reported any event of interested outcomes, data on seven components, which are random sequence genera- http://www.clini caltr ials.gov were also checked if registry tion, allocation concealment, blinding of participants and number was provided. personnel, blinding of outcome assessment, incomplete Studies meeting the following criteria were included: outcome data, selective reporting and other sources of randomized controlled trial; individuals with type 2 dia- bias. Each of these components of every included study betes; comparison of anti-diabetic drugs with other posi- received a rating of “low risk”, “unclear”, or “high risk”. tive comparator drugs or placebo (PLA); had at least one Zhuang et al. Cardiovasc Diabetol (2018) 17:79 Page 3 of 10 Statistical analysis alogliptin, linagliptin, saxagliptin, sitagliptin, vildagliptin, Stata package (version 14) was applied for statistical albiglutide, dulaglutide, exenatide, liraglutide, lixisena- analyses, using the network and mvmeta command and tide, canagliflozin, dapagliflozin, empagliflozin, MET, SU, Stata routines described elsewhere [21]. For indirect and TZD, INS and PLA. Most trials (159 [93.5%] of 170) were mixed comparisons, we used network meta-analysis to two-armed studies. Studies characteristic and outcomes obtain estimates for the outcomes, and presented these were shown in Additional file  1: S5, S6. Treatment dura- estimates as odd ratios (OR) with 95% confidence inter - tion ranged from 24 to 208 weeks. Male patients ranged vals (CI). We then estimated the relative ranking prob- from 40.9 to 75%. Mean age of patients was ranged from ability of each treatment and obtained the treatment 44.0 to 74.9 years. Ten studies enrolled subjects with high hierarchy of competing interventions using rankograms, CV risk (65,650 patients), and 9 studies were exclusively surface under the cumulative ranking (SUCRA) curve, of patients with renal impairment (1349 patients). and mean ranks. Large SUCRA scores might indicate a more effective or safer intervention [22]. We showed the Study quality assessment results using SU as reference in interval plot because it The overall quality of studies was rated as good, even consistently ranked last. In addition, we chose not to pre- though some studies did not record details about rand- sent MET in the ranking as it was used as background omization and allocation concealment and there were treatment in most of the trials. only few randomized trials at low risk of bias in every To check for the presence of inconsistency, we used question-based entry (Additional file  1: S7). Moreover, the loop-specific approach that assesses the difference no major tendency was noted for smaller studies to over- between direct and indirect estimates for a specific estimate or underestimate active treatment effects on comparison in the loop (inconsistency factor) [23]. We outcomes in the comparison-adjusted funnel plot for the assumed a common heterogeneity estimate within each network (Additional file 1: S8). loop. We used the previously described node-splitting method, which separates evidence for a particular com- Mace and all‑cause mortality parison into direct and indirect, excluding one direct Networks of eligible comparisons (among individual comparison at a time and estimating the indirect treat- agents or classes of agents) for the CV outcomes were ment effect for the excluded comparison [24]. A com - presented in Fig.  1, showing predominantly pairwise parison adjusted funnel plot of treatment estimates for comparisons of drugs with DDP4i or PLA in classed drug on CV outcomes was used to assess for evidence of groups. Except for INS and SGLT2i, SGLT2i and GLP1ra, small-study effects. as well as INS and MET, direct evidence for MACE was To investigate the generalizability of the findings, we available for all the possible pairwise treatment com- assessed the effect of characteristics of trials and partici - parisons. However, such availability was lacking between pants on the outcomes in sensitivity analyses by exclud- individual drugs. ing studies with the following design characteristics: In the network meta-analyses, MACE were reported patients with high CV risk; patients with renal impair- in 152 studies (158,786 patients with 8702 MACE). ment; and sample size less than 100 in one arm. Comparative effects of all drugs were ranked with Finally, to explore the potential impact of severe hypo- SUCRA probabilities (Additional file  1: S9). Mixed glycemia on the association between anti-diabetic drugs comparisons were in the interval plot with SU as refer- and CV outcomes, additional correlation analysis of ence (Fig. 2) and the comparisons table (Fig. 3). In term severe hypoglycemia and outcome of interest according of MACE, mixed comparisons by drug class showed to the ranking order was conducted. that SGLT2i (OR 0.70, 95% CI 0.55–0.90), INS (0.71, 95% CI 0.57–0.90), GLP1ra (OR 0.76, 95% CI 0.61– 0.94), and DPP4i (OR 0.77, 95% CI 0.62–0.9) were sig- nificantly better than SU, and SGLT2i (OR 0.72, 95% Results CI 0.54–0.97) and INS (OR 0.73, 95% CI 0.56–0.97) The PRISMA flowchart showing electronic searching were superior to TZD. Mixed comparisons by indi- processes is shown in Additional file  1: S4. There were vidual drug showed that vildagliptin (OR 0.47, 95% CI 170 randomized controlled trials including 166,371 adults 0.25–0.90), lixisenatide (OR 0.72, 95%CI 0.55–0.95) eligible for the systematic review, reporting 8702 cases of and exenatide (OR 0.74, 95% CI 0.56–0.99) were sig- MACE (33.5 per 1000 patient-year) and 4914 cases of all- nificantly better than SU. Moreover, vildagliptin (OR cause mortality (18.3 per 1000 patient-year). Seven drug 0.49, 95% CI 0.25–0.94) was significantly superior to classes were compared with PLA or each other: DPP4i, TZD and lixisenatide (OR 0.49, 95% CI 0.25–0.94) was GLP1ra, SGLT2i, MET, SU, TZD and INS. For indi- significantly superior to albiglutide. By applying the vidual comparison, 18 treatment groups were analyzed: Zhuang et al. Cardiovasc Diabetol (2018) 17:79 Page 4 of 10 Fig. 1 Network plot of treatment comparisons for major adverse cardiovascular events (MACE) and all‑ cause mortality. a Categorized drugs comparisons for MACE; b Categorized drugs comparisons for all‑ cause mortality; c Individual drugs comparisons for MACE; d Individual drugs comparisons for all‑ cause mortality. The size of the nodes represents the number of trials that study the treatments. Direct comparison of treatments is linked with a line, the thickness of which represents the number of trials that assess the comparison. SGLT2i sodium‑ glucose co‑transporter 2 inhibitor(s), GLP1ra glucagon‑like peptide ‑1 receptor agonist(s), DPP4i dipeptidyl peptidase 4 inhibitor(s), TZD thiazolidinedione, MET metformin, SU sulfonylurea, INS insulin, PLA placebo, VIL vildagliptin, EMP empagliflozin, LIX lixisenatide, ALO alogliptin, EXE exenatide, LIR liraglutide, CAN canagliflozin, DAP dapagliflozin, DUL dulaglutide, SIT sitagliptin, LIN linagliptin, ALB albiglutide, SAX saxagliptin design-by-treatment inconsistency model, we detected significantly superior to albiglutide (OR 0.84, 95% CI inconsistency in only one loop of comparisons: SU- 0.73–0.97) and saxagliptin (OR 0.77, 95% CI 0.63–0.94). TZD (P = 0.042) (Additional file 1 : S10). We did not observe any inconsistencies between evi- All-cause mortality was reported in 139 studies dence derived from direct to indirect comparisons for (159,722 patients with 4914 death). Mixed comparisons all-cause mortality using the design-by-treatment incon- were in the interval plot with SU as reference (Fig.  2) sistency model (Additional file 1: S10). and the comparisons table (Fig.  3). In term of all-cause mortality, mixed comparisons by drug class showed that Sensitivity analyses and post hoc correlation analysis SGLT2i was significantly better than GLP1ra (OR 0.72, Results for MACE were generally robust in sensitivity 95% CI 0.57–0.91), DPP4i (OR 0.72, 95% CI 0.60–0.87), analyses by excluding studies with the following design TZD (OR 0.64, 95% CI 0.42–0.97) and SU (OR 0.58, characteristics: patients with high CV risk; patients with 95% CI 0.41–0.83). Moreover, INS was significantly bet - renal impairment; and sample number less than 100 ter than DPP4i (OR 0.87, 95% CI 0.77–0.97) and SU (OR in one arm. After the sensitivity analyses, changes in 0.70, 95% CI 0.50–0.97). Mixed comparisons by individ- ORs and rankings, either categorical drugs or individ- ual drug showed that exenatide was significantly better ual drugs, did not alter the primary results appreciably than albiglutide (OR 0.68, 95% CI 0.57–0.82), saxagliptin (Additional file 1: S11). (OR 0.62, 95% CI 0.49–0.79), linagliptin (OR 0.55, 95% As is shown in the post hoc correlation analysis (Fig. 4), CI 0.35–0.87), TZD (OR 0.62, 95% CI 0.40–0.97) and for individual drugs, the ranking order of MACE or SU (OR 0.57, 95% CI 0.38–0.85), and lixisenatide was all-cause mortality was positively correlated with the Zhuang et al. Cardiovasc Diabetol (2018) 17:79 Page 5 of 10 Fig. 2 Forest plot for MACE and all‑ cause mortality of anti‑ diabetic agents compared with sulfonylurea (individual and categorized agents). Treatments are ranked by surface under the cumulative ranking (SUCRA) values. OR odds ratio, CrI credibility interval Zhuang et al. Cardiovasc Diabetol (2018) 17:79 Page 6 of 10 Fig. 3 Mixed comparison results of anti‑ diabetic agents for MACE and all‑ cause mortality, both for individual (above table) and for categorized agents (below table). Agents are reported in order of MACE ranking. Treatment at the top left corner ranks first, while the one at the bottom right corner ranks last. OR lower than 1 favor the column‑ defining treatment. Anti‑ diabetic agents in one class are painted with the same color. Significant results are in bold other individual or classes of drugs, SU are associated ranking order of severe hypoglycemia (R = 0.3178, with the highest risks of MACE and all-cause mortal- P = 0.018; R = 0.2574, P = 0.038, respectively), whereas ity. Finally, the ranking of CV risk was linearly corre- for drug classes, a similar trend was observed with TZD lated with the ranking of severe hypoglycemia risk by as an outlier. individual comparisons, with SU displaying the highest risks in both endpoints. Discussion Our study found that the newer agents in general To the best of our knowledge, our network meta-anal- showed favorable CV safety, yet there are discrepan- ysis represents the most comprehensive synthesis of cies between individual and class comparisons. In a data currently available with regard to CV outcomes recently published meta-analysis, the DPP4i vildaglip- in pharmacologically managed patients with type 2 tin was found to significantly reduce the risk of MI and diabetes. Our findings can be summarized as follows: stroke, while other agents in the same class seemed to first, among anti-diabetic agents included in the net perform less well in terms of CV outcomes [3]. In two work, SGLT2i in class comparisons, and vildagliptin recent studies, the use of DPP4i was found to be asso- in individual comparisons, respectively ranked first in ciated with improved long-term survival in diabetic terms of MACE. Furthermore, when compared with Zhuang et al. Cardiovasc Diabetol (2018) 17:79 Page 7 of 10 Fig. 4 Correlation analyses showing correlation between the ranking order of MACE and all‑ cause mortality risk and the ranking order of severe hypoglycemia risk. Color of circle represents different drugs shown above. Area of a circle reflects sample size. a, b, c, d The panels indicates the correlation relationship between severe hypoglycemia and outcome of interest according to the ranking order patients surviving a myocardial infarction [15] whereas valid, and better informed. Such a “targeted” strategy, its increase in overall risks of heart failure or exhibit nonetheless, ought not to have indiscreetly relied too within-class differences remains unresolved [7 ]. These much, if at all, on a “known class effect”, but rather should results were reiterated in our study, in that despite be individualized and outcome-specific, if (and only if ) vildagliptin displayed the best CV safety profile indi - the signals of harm were detected in pre-approval pack- vidually, and in class comparisons the ranking of DPP4i age or post-approval monitoring of these newer agents. actually dropped to the fifth place in order. Other dis - On the other hand, albeit primary results being crepancies are also identifiable and can be resolved “mixed”, our analysis had once again confirmed that SU similarly. In light of the mixed results, therefore, a case were associated with the highest risks of MACE and all- can perhaps be made against a “class effect” in the era of cause mortality. In fact, SU steadily brought up the rear novel anti-diabetic medications, namely, the fact that a in both individual and class ranking, even after sensitivity well-documented better (or worse) CV safety profile of analysis. When plotting the ORs of MACE or all-cause one individual agent does not necessarily justify extrap- mortality for all other comparator drugs against SU, indi- olation of such a benefit (or harm) to other agents in vidually, point estimates concordantly lie to the left of the same class. the “line of no effect”; collectively, by order of effect size, These results have practical implications. Several SGLT2i, INS, GLP1ra and DPP4i were significantly better appeals have recently been made for an appraisal of the than SU, indicating that SU in actuality possess the worst current paradigm to evaluate CV risks of novel anti- CV safety profile among these medications. diabetic medications via large-scale, long-term CV SU are currently the most widely used medications safety trials [25, 26]. The authors argued for alternative for type 2 diabetes second only to MET. However, the approaches that are more cost-effective, more externally undesirable effect of weight gain [27], the greatest risk of Zhuang et al. Cardiovasc Diabetol (2018) 17:79 Page 8 of 10 iatrogenic hypoglycemia [28–31], a potential increase in due to heterogeneity of study designs and reporting CV morbidity and mortality [32–36], and adding to that, styles in the included studies, comparison of A1c lev- the advent of novel anti-diabetic medications with argu- els as well as the degree of CV risk at baseline across ably equal glucose-lowering effectiveness [37], all render different trials were unavailable. In addition, only a few SU as less favorable [38]. In a recent commentary, the studies reported outcomes such as acute coronary syn- role of SU in the era of novel anti-diabetic medications drome and CV death, and most had few or zero events. was thoughtfully challenged [39]. And according to the Some subgroup analyses had small numbers of partici- latest management guideline jointly issued by the Ameri- pants, likely resulting in poor precision of estimates. can Association of Clinical Endocrinologists and the Second, many of the international multicenter trials American College of Endocrinologists, in combinational were conducted primarily in higher-income countries, regimens, the strength of recommendation for SU to be which would possibly interfere with the external valid- added on top of MET is the weakest [40]. ity of these results for lower-income settings. Third, One of the major controversies about SU is their CV follow-up duration in most studies was relatively too safety. For example, in the UK Prospective Diabetes short to draw any definitive conclusion for long-term Study (UKPDS), in patients treated with SU, there was a CV outcome. Fourth, as first-line data from these trials trend of a 16% decrease in MI at the end of the study, but being unaccessible to us, the present study was unable at 10-year follow up there was a significant 15% decrease to account for other possible factors correlated with in events in the same arm [41]. In a recent network meta- meaningful CV outcomes. Results from continuing analysis, the authors reported no significant differences trials would provide useful insight in answering these in the associations between nine classes of anti-diabetic questions. medications and the risk of CV or all-cause mortal- ity [37]. Unfortunately, however, due to low event rates in general as well as the statistical power being diluted Conclusions by multiple layers of analyses in particular, conclusions Our network meta-analyses showed that all three classes about true effects of the studied drugs on CV or all-cause of novel anti-diabetic medications, i.e. DPP4i, GLP1ra, mortality were far from precise. The contribution of our and SGLT2i, possess favorable CV safety profile in gen - study, then, is having further clarified the current confu - eral, notwithstanding that there are small but robust dif- sion by offering an evidence-based hierarchy of CV safety ferences among individual drugs. These results refute a profile among all major anti-diabetic medications, which simplistic rationale of generalizing the CV benefit of one revealed the steady truth that SU are associated with the single agent to the others in the same class. In addition, highest risks of MACE or all-cause mortality when com- we also observed that SU were associated with the high- pared with other individual or classes of agents. est risk of MACE and all-cause mortality, which could In our post hoc correlation analysis, the ranking of potentially explained by its concomitant increase in the MACE and mortality risk were both linearly correlated risk of severe hypoglycemia. Such correlation should with the ranking of severe hypoglycemia risk by individ- probably call for a reassessment of the role of SU as first- ual comparisons, with SU at the top of risks, corroborat- line additive to MET in the pharmacological management ing the already well-founded connection, if not causality, of type-2 diabetes. These findings should be considered between the elevated risk of iatrogenic hypoglycemia in policy-making and the development of clinical practice with the use of SU and the inferior CV safety of this class guidelines. of traditional anti-diabetic agents, although the effect of Additional file “drug–drug interaction” cannot be completely ruled out, because many of these trials had multi-drug regimens. In Additional file 1: S1. Protocol. S2. Search strategy. S3. Novel drugs addition, it is of note that TZD are shown to be an out- approved by FDA or European Medicines Agency. S4. Flow chart of the lier in both correlations, with higher ranking of CV risk study selection process. S5. Studies characteristic of the included studies. but lower ranking of hypoglycemia. Such a result could S6. Outcomes of interest in each study. S7. Quality assessment of the included studies. S8. Comparison‑adjusted funnel plot for the network. be explained by the previous observation that the use of S9. Ranking ordered according to surface under the cumulative ranking TZD (especially rosiglitazone) is associated with poten- values of outcomes. S10. Consistency analysis of direct verse indirect tial increase of CV risks independent of hypoglycemia comparisons for outcomes. S11. Summaries of sensitivity analysis. [42]. Our study is limited in several ways. First, although Abbreviations comprehensive systematic search strategies were CV: cardiovascular; FDA: The US Food and Drug Administration; DPP4i: employed, our analyses were limited by the modest dipeptidyl peptidase 4 inhibitor; GLP1ra: glucagon‑like peptide ‑1 receptor agonist; SGLT2i: sodium‑ glucose co‑transporter 2 inhibitor; INS: insulin; MET: amount of data in the included studies. To begin with, Zhuang et al. Cardiovasc Diabetol (2018) 17:79 Page 9 of 10 metformin; SU: sulfonylurea; TZD: thiazolidinedione; PLA: placebo; MACE: 6. Ferdinand KC, Botros FT, Atisso CM, Sager PT. Cardiovascular safety for major adverse cardiovascular event; MI: myocardial infarction; OR: odds ratio; once‑ weekly dulaglutide in type 2 diabetes: a pre‑specified meta‑analysis CI: confidence interval; SUCRA : surface under the cumulative ranking. of prospectively adjudicated cardiovascular events. Cardiovasc Diabetol. 2016;15:38. https ://doi.org/10.1186/s1293 3‑016‑0355‑z. Authors’ contributions 7. Verma S, Goldenberg RM, Bhatt DL, Farkouh ME, Quan A, Teoh H. Dipepti‑ LXX, XHP and ZXD: designed the study and wrote the paper, collected and dyl peptidase‑4 inhibitors and the risk of heart failure: a systematic review analyzed the data. ZXD, HX and YDY conducted the search, collected the data, and meta‑analysis. CMAJ Open. 2017;5:E152–77. https ://doi.org/10.9778/ performed the analysis and wrote the manuscript. All authors critically revised cmajo .20160 058. the manuscript. All authors read and approved the final manuscript. 8. Zhang Z, Chen X, Lu P, Zhang J, Xu Y, He W. Incretin‑based agents in type 2 diabetic patients at cardiovascular risk: compare the effect of GLP ‑1 Author details agonists and DPP‑4 inhibitors on cardiovascular and pancreatic out ‑ Department of Cardiology, The First Affiliated Hospital of Sun Yat ‑sen Univer ‑ comes. Cardiovasc Diabetol. 2017;16:31. https ://doi.org/10.1186/s1293 sity, Guangzhou, People’s Republic of China. Key Laboratory on Assisted Cir‑ 3‑017‑0512‑z. culation, Ministry of Health, No. 58 Zhongshan 2nd road, Guangzhou 510080, 9. Hoogwerf BJ, Lincoff AM, Rodriguez A, Chen L, Qu Y. Major adverse cardi‑ People’s Republic of China. Department of Endocrinology, The First Affiliated ovascular events with basal insulin peglispro versus comparator insulins Hospital of Sun Yat‑sen University, Guangzhou, People’s Republic of China. in patients with type 1 or type 2 diabetes: a meta‑analysis. Cardiovasc Diabetol. 2016;15:78. https ://doi.org/10.1186/s1293 3‑016‑0393‑6. Acknowledgements 10. Tang H, Fang Z, Wang T, Cui W, Zhai S, Song Y. Meta‑analysis of effects of We thank Miss. Yankun Sun for her assistance in image processing. sodium‑ glucose cotransporter 2 inhibitors on cardiovascular outcomes and all‑ cause mortality among patients with type 2 diabetes mel‑ Competing interests litus. Am J Cardiol. 2016;118:1774–80. https ://doi.org/10.1016/j.amjca The authors declare that they have no competing interests. rd.2016.08.061. 11. Fei Y, Tsoi MF, Kumana CR, Cheung TT, Cheung B. Network meta‑analysis Availability of data and materials of cardiovascular outcomes in randomized controlled trials of new anti‑ All data generated or analyzed during this study are included in this published diabetic drugs. Int J Cardiol. 2018;254:291–6. https ://doi.org/10.1016/j. article and its additional file.ijcar d.2017.12.039. 12. Tsioufis C, Andrikou E, Thomopoulos C, Papanas N, Tousoulis D. Oral Consent for publication glucose‑lowering drugs and cardiovascular outcomes: from the negative Not applicable. RECORD and ACCORD to neutral TECOS and promising EMPA‑REG. Curr Vasc Pharmacol. 2017;15:457–68. https ://doi.org/10.2174/15701 61114 Ethics approval and consent to participate66616 12081 50642 . Not applicable. 13. de Jong M, van der Worp HB, van der Graaf Y, Visseren F, Westerink J. Pioglitazone and the secondary prevention of cardiovascular disease. Funding A meta‑analysis of randomized‑ controlled trials. Cardiovasc Diabetol. This work was funded by China national nature science founda‑ 2017;16:134. https ://doi.org/10.1186/s1293 3‑017‑0617‑4. tion (81600206) and Guangdong national nature science foundation 14. Lam KS, Chow CC, Tan KC, Ma RC, Kong AP, Tong PC. Practical consid‑ (2016A030310140/20160903). erations for the use of sodium‑ glucose co‑transporter type 2 inhibi‑ tors in treating hyperglycemia in type 2 diabetes. Curr Med Res Opin. 2016;32:1097–108. https ://doi.org/10.1185/03007 995.2016.11616 08. Publisher’s Note 15. Wang MT, Lin SC, Tang PL, Hung WT, Cheng CC, Yang JS. The impact of Springer Nature remains neutral with regard to jurisdictional claims in pub‑ DPP‑4 inhibitors on long‑term survival among diabetic patients after first lished maps and institutional affiliations. acute myocardial infarction. Cardiovasc Diabetol. 2017;16:89. https ://doi. org/10.1186/s1293 3‑017‑0572‑0. Received: 21 February 2018 Accepted: 23 May 2018 16. Schernthaner G, Lehmann R, Prazny M, Czupryniak L, Ducena K, Fasching P. Translating recent results from the Cardiovascular Outcomes Trials into clinical practice: recommendations from the Central and Eastern Euro‑ pean Diabetes Expert Group (CEEDEG). Cardiovasc Diabetol. 2017;16:137. https ://doi.org/10.1186/s1293 3‑017‑0622‑7. References 17. Schnell O, Ryden L, Standl E, Ceriello A. Updates on cardiovascular 1. Sonesson C, Johansson PA, Johnsson E, Gause‑Nilsson I. Cardiovascular outcome trials in diabetes. Cardiovasc Diabetol. 2017;16:128. https ://doi. effects of dapagliflozin in patients with type 2 diabetes and different risk org/10.1186/s1293 3‑017‑0610‑y. categories: a meta‑analysis. Cardiovasc Diabetol. 2016;15:37. https ://doi. 18. Moher D, Liberati A, Tetzlaff J, Altman DG. Preferred reporting items for org/10.1186/s1293 3‑016‑0356‑y. systematic reviews and meta‑analyses: the PRISMA statement. J Clin Epi‑ 2. Fisher M, Petrie MC, Ambery PD, Donaldson J, Ye J, McMurray JJ. Cardio‑ demiol. 2009;62:1006–12. https ://doi.org/10.1016/j.jclin epi.2009.06.005. vascular safety of albiglutide in the Harmony programme: a meta‑analy‑ 19. Caldwell DM, Ades AE, Higgins JP. Simultaneous comparison of multiple sis. Lancet Diabetes Endocrinol. 2015;3:697–703. https ://doi.org/10.1016/ treatments: combining direct and indirect evidence. BMJ. 2005;331:897– S2213 ‑8587(15)00233 ‑8. 900. https ://doi.org/10.1136/bmj.331.7521.897. 3. Savarese G, D’Amore C, Federici M, De Martino F, Dellegrottaglie S, Mar‑ 20. Green S, Higgins JPT. Cochrane handbook for systematic reviews of ciano C. Eec ff ts of dipeptidyl peptidase 4 inhibitors and sodium‑ glucose interventions. Chichester: Wiley‑Blackwell; 2008. linked cotransporter‑2 Inhibitors on cardiovascular events in patients 21. White IR. Multivariate random‑ effects meta‑regression: updates to with type 2 diabetes mellitus: a meta‑analysis. Int J Cardiol. 2016;220:595– mvmeta. Stata J. 2011;11:255–70. 601. https ://doi.org/10.1016/j.ijcar d.2016.06.208. 22. Salanti G, Ades AE, Ioannidis JP. Graphical methods and numerical sum‑ 4. Monami M, Dicembrini I, Nardini C, Fiordelli I, Mannucci E. Eec ff ts of maries for presenting results from multiple‑treatment meta‑analysis: glucagon‑like peptide ‑1 receptor agonists on cardiovascular risk: a meta‑ an overview and tutorial. J Clin Epidemiol. 2011;64:163–71. https ://doi. analysis of randomized clinical trials. Diabetes Obes Metab. 2014;16:38– org/10.1016/j.jclin epi.2010.03.016. 47. https ://doi.org/10.1111/dom.12175 . 23. White IR, Barrett JK, Jackson D, Higgins JP. Consistency and inconsistency 5. Wu JH, Foote C, Blomster J, Toyama T, Perkovic V, Sundstrom J. Eec ff ts in network meta‑analysis: model estimation using multivariate meta‑ of sodium‑ glucose cotransporter‑2 inhibitors on cardiovascular events, regression. Res Synth Methods. 2012;3:111–25. https ://doi.org/10.1002/ death, and major safety outcomes in adults with type 2 diabetes: a jrsm.1045. systematic review and meta‑analysis. Lancet Diabetes Endocrinol. 2016;4:411–9. https ://doi.org/10.1016/S2213 ‑8587(16)00052 ‑8. Zhuang et al. Cardiovasc Diabetol (2018) 17:79 Page 10 of 10 24. Veroniki AA, Vasiliadis HS, Higgins JP, Salanti G. Evaluation of inconsist‑ 34. Fisman EZ, Tenenbaum A, Boyko V, Benderly M, Adler Y, Friedensohn A. ency in networks of interventions. Int J Epidemiol. 2013;42:332–45. https Oral antidiabetic treatment in patients with coronary disease: time‑ ://doi.org/10.1093/ije/dys22 2. related increased mortality on combined glyburide/metformin therapy 25. Smith RJ, Goldfine AB, Hiatt WR. Evaluating the cardiovascular safety of over a 7.7‑ year follow‑up. Clin Cardiol. 2001;24:151–8. new medications for type 2 diabetes: time to reassess? Diabetes Care. 35. Rao AD, Kuhadiya N, Reynolds K, Fonseca VA. Is the combination of sulfo‑ 2016;39:738–42. https ://doi.org/10.2337/dc15‑2237. nylureas and metformin associated with an increased risk of cardiovascu‑ 26. Steg PG, Roussel R. Randomized trials to evaluate cardiovascular safety lar disease or all‑ cause mortality?: a meta‑analysis of observational stud‑ of antihyperglycemic medications: a worthwhile effort? Circulation. ies. Diabetes Care. 2008;31:1672–8. https ://doi.org/10.2337/dc08‑0167. 2016;134:571–3. https ://doi.org/10.1161/CIRCU LATIO NAHA.116.02191 4. 36. Currie CJ, Poole CD, Evans M, Peters JR, Morgan CL. Mortality and other 27. Bolen S, Feldman L, Vassy J, Wilson L, Yeh HC, Marinopoulos S. Systematic important diabetes‑related outcomes with insulin vs other antihypergly‑ review: comparative effectiveness and safety of oral medications for type cemic therapies in type 2 diabetes. J Clin Endocrinol Metab. 2013;98:668– 2 diabetes mellitus. Ann Intern Med. 2007;147:386–99. 77. https ://doi.org/10.1210/jc.2012‑3042. 28. Bennett WL, Maruthur NM, Singh S, Segal JB, Wilson LM, Chatterjee R. 37. Palmer SC, Mavridis D, Nicolucci A, Johnson DW, Tonelli M, Craig JC. Comparative effectiveness and safety of medications for type 2 diabetes: Comparison of clinical outcomes and adverse events associated with an update including new drugs and 2‑ drug combinations. Ann Intern glucose‑lowering drugs in patients with type 2 diabetes: a meta‑analysis. Med. 2011;154:602–13. https ://doi.org/10.7326/0003‑4819‑154‑9‑20110 JAMA. 2016;316:313–24. https ://doi.org/10.1001/jama.2016.9400. 5030‑00336 . 38. Xu J, Rajaratnam R. Cardiovascular safety of non‑insulin pharmaco ‑ 29. Tschope D, Bramlage P, Binz C, Krekler M, Plate T, Deeg E. Antidiabetic therapy for type 2 diabetes. Cardiovasc Diabetol. 2017;16:18. https ://doi. pharmacotherapy and anamnestic hypoglycemia in a large cohort of org/10.1186/s1293 3‑017‑0499‑5. type 2 diabetic patients—an analysis of the DiaRegis registry. Cardiovasc 39. Genuth S. Should sulfonylureas remain an acceptable first ‑line add‑ on to Diabetol. 2011;10:66. https ://doi.org/10.1186/1475‑2840‑10‑66. metformin therapy in patients with type 2 diabetes? No, it’s time to move 30. Seaquist ER, Anderson J, Childs B, Cryer P, Dagogo‑ Jack S, Fish L. Hypogly‑ on! Diabetes Care. 2015;38:170–5. https ://doi.org/10.2337/dc14‑0565. cemia and diabetes: a report of a workgroup of the American Diabetes 40. Garber AJ, Abrahamson MJ, Barzilay JI, Blonde L, Bloomgarden ZT, Bush Association and the Endocrine Society. Diabetes Care. 2013;36:1384–95. MA. Consensus statement by the American Association of Clinical Endo‑ https ://doi.org/10.2337/dc12‑2480. crinologists and American College of Endocrinology on the comprehen‑ 31. Zimmerman BR. Sulfonylureas. Endocrinol Metab Clin N Am. sive type 2 diabetes management algorithm—2016 executive summary. 1997;26:511–22. Endocr Pract. 2016;22:84–113. https ://doi.org/10.4158/EP151 126.CS. 32. Roumie CL, Hung AM, Greevy RA, Grijalva CG, Liu X, Murff HJ. Compara‑ 41. Holman RR, Paul SK, Bethel MA, Matthews DR, Neil HA. 10‑ year follow‑ tive effectiveness of sulfonylurea and metformin monotherapy on car ‑ up of intensive glucose control in type 2 diabetes. N Engl J Med. diovascular events in type 2 diabetes mellitus: a cohort study. Ann Intern 2008;359:1577–89. https ://doi.org/10.1056/NEJMo a0806 470. Med. 2012;157:601–10. https ://doi.org/10.7326/0003‑4819‑157‑9‑20121 42. Cobitz A, Zambanini A, Sowell M, Heise M, Louridas B, McMorn S. A retro‑ 1060‑00003 . spective evaluation of congestive heart failure and myocardial ischemia 33. Tzoulaki I, Molokhia M, Curcin V, Little MP, Millett CJ, Ng A. Risk of car‑ events in 14,237 patients with type 2 diabetes mellitus enrolled in 42 diovascular disease and all cause mortality among patients with type 2 short‑term, double ‑blind, randomized clinical studies with rosiglitazone. diabetes prescribed oral antidiabetes drugs: retrospective cohort study Pharmacoepidemiol Drug Saf. 2008;17:769–81. https ://doi.org/10.1002/ using UK general practice research database. BMJ. 2009;339:b4731. pds.1615. Ready to submit your research ? Choose BMC and benefit from: fast, convenient online submission thorough peer review by experienced researchers in your field rapid publication on acceptance support for research data, including large and complex data types • gold Open Access which fosters wider collaboration and increased citations maximum visibility for your research: over 100M website views per year At BMC, research is always in progress. Learn more biomedcentral.com/submissions

Journal

Cardiovascular DiabetologySpringer Journals

Published: Jun 5, 2018

References

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off