Effect of Muraglitazar on Death and Major Adverse Cardiovascular Events in Patients with Type 2 Diabetes Mellitus

Home , Muraglitazar, Pioglitazone, PPAR agonist, Rosiglitazone

ORIGINAL CONTRIBUTION JAMA-EXPRESS

Effect of Muraglitazar on Death and Major Adverse Cardiovascular Events in Patients With Type 2 Diabetes Mellitus

Steven E. Nissen, MD Context Peroxisome proliferator–activated receptors (PPARs) are nuclear transcrip- Kathy Wolski, MPH tion factors that modulate gene expression. Therapeutic agents targeting 2 distinct ␣ ␥ Eric J. Topol, MD families of PPARs ( and ) have been introduced in the United States. The first dual- PPAR agonist, muraglitazar, was reviewed by a US Food and Drug Administration (FDA) EROXISOME PROLIFERATOR– advisory committee on September 9, 2005, resulting in a vote of 8:1 recommending activated receptors (PPARs) are approval for its use in controlling blood glucose levels in patients with type 2 diabetes. ligand-activated nuclear tran- Objective To evaluate the incidence of death, myocardial infarction (MI), stroke, scription factors that modu- congestive heart failure (CHF), and transient ischemic attack (TIA) in diabetic patients lateP expression of a large number of treated with muraglitazar compared with controls. genes.1 In the United States, therapeu- Design, Setting, and Participants The source material for this analysis consisted tic agents that target 2 distinct families of documents about phase 2 and 3 clinical trials released under public disclosure laws of PPARs (␣ and ␥) have been intro- for the FDA advisory committee meeting. All reviewed trials were prospective, ran- duced. Examples of PPAR-␣ agents in- domized, double-blind, multicenter studies enrolling patients with type 2 diabetes and clude the fibric acid derivatives fenofi- hemoglobin A1c levels between 7% and 10%. Patients (N=3725) were randomized to receive differing doses of muraglitazar, pioglitazone, or placebo as monotherapy or brate and gemfibrozil. These agents in combination with metformin or glyburide in trials ranging from 24 to 104 weeks. modulate lipid metabolism primarily by lowering serum triglyceride levels and Main Outcome Measures The primary outcome was the incidence of death, non- modestly increasing levels of high- fatal MI, or nonfatal stroke. A more comprehensive composite outcome included these events plus the incidence of CHF and TIA. density lipoprotein (HDL) cholesterol. In several clinical trials, these PPAR-␣ Results In the muraglitazar-treated patients, death, MI, or stroke occurred in 35 of agents have reduced cardiovascular 2374 (1.47%) patients compared with 9 of 1351 (0.67%) patients in the combined placebo and pioglitazone treatment groups (controls) (relative risk [RR], 2.23; 95% events or demonstrated slowing of ath- confidence interval [CI], 1.07-4.66; P=.03). For the more comprehensive outcome mea- 2-4 ␥ erosclerosis progression. The PPAR- sure that included TIA and CHF, the incidence was 50 of 2374 (2.11%) for mura- agonists increase insulin sensitivity and glitazar compared with 11 of 1351 (0.81%) for controls (RR, 2.62; 95% CI, 1.36- are widely used as antidiabetic agents.5 5.05; P=.004). Relative risks for each of the individual components of the composite Two drugs are currently available, pio- end point exceeded 2.1 but were not statistically significant. Incidence of adjudicated glitazone and rosiglitazone. Because CHF was 13 of 2374 (0.55%) muraglitazar-treated patients and 1 of 1351 controls both abnormal lipid levels and insulin (0.07%) (RR, 7.43; 95% CI, 0.97-56.8; P=.053). resistance are believed to promote ath- Conclusions Compared with placebo or pioglitazone, muraglitazar was associated erosclerosis in diabetic patients, sev- with an excess incidence of the composite end point of death, major adverse cardio- eral pharmaceutical companies have vascular events (MI, stroke, TIA), and CHF. This agent should not be approved to treat sought to develop dual-PPAR agonists diabetes based on laboratory end points until safety is documented in a dedicated car- that target both the ␣ and ␥ families.6 diovascular events trial. The first dual-PPAR to reach the US JAMA. 2005;294:(doi:10.1001/jama.294.20.joc50147). www.jama.com Food and Drug Administration (FDA) ␥ Author Affiliations: Department of Cardiovascular Medi- for consideration of approval is muraglitazar, a strong PPAR- agonist with cine, Cleveland Clinic Foundation, Cleveland, Ohio. moderate PPAR-␣ effects.7 The devel- Corresponding Author: Steven E. Nissen, MD, De- opment program for muraglitazar in- partment of Cardiovascular Medicine, Cleveland Clinic See editorial comment. Foundation, 9500 Euclid Ave, Cleveland, OH 44195 cluded a series of clinical trials exam- ([email protected]).

Downloaded from www.jama.com by MarisaSitjar, on October 24, 2005 MURAGLITAZAR AND ADVERSE EVENTS

trialsperformedindiabeticpatients,pub- sessed by examination of detailed Table 1. Muraglitazar Clinical Trials in Diabetic Patients licly released by the sponsor and FDA listings of adverse patient events re- 9,10 Study, Duration, No. for the advisory panel meeting. ported by the study sponsor in the FDA and Daily Doses Randomized documents. The documents do not pro- CV 168006 (104 weeks) METHODS vide specific information on the pro- Muraglitazar Analyzed Studies 0.5 mg 236 cess used to verify these events. The 1.5 mg 259 We reviewed the FDA briefing docu- CHF events were described as cen- 5 mg 245 ments available via the FDA Web site trally adjudicated, but there is no men- 10 mg 249 20 mg 239 for the September 9, 2005, public hear- tion of an adjudication process for other Pioglitazone 15 mg 251 ing. There were 2 major documents, an types of events. CV 168018 (24 weeks) Muraglitazar analysis of data in the muraglitazar clini- For composite end points, to avoid 2.5 mg 111 cal development program provided by double counting of patients with more 5 mg 114 9 Placebo 115 FDA staff and a separate document pre- than 1 event, each patient was classi- CV 168021 (24 weeks) pared by the developers of the drug fied by the event with the greatest se- Muraglitazar 2.5 mg ϩ 191 10 glyburide (Bristol-Myers Squibb and Merck). verity. Thus, a patient with MI who sub- Muraglitazar 5 mg ϩ 193 These documents provide data for 5 sequently died as a consequence of this glyburide clinical trials that assessed safety and event was classified as experiencing car- Placebo ϩ glyburide 199 CV 168022 (78 weeks) efficacy in diabetic patients. Efficacy as- diovascular death but not MI. For ana- Muraglitazar 2.5 mg ϩ 233 sessments included primarily the ef- lytical purposes, the event rates for both metformin Muraglitazar 5 mg ϩ 205 fects of muraglitazar on various mea- comparators (placebo and piogl- metformin sures of glycemic control and lipids. itazone) were pooled and compared Placebo ϩ metformin 214 CV 168025 (26 weeks) Safety assessment included reporting of with event rates for muraglitazar. From Muraglitazar 5 mg ϩ 587 major adverse events including cardio- the crude event rates, relative risks metformin vascular outcomes. (RRs), 95% confidence intervals (CIs), Pioglitazone 30 mg ϩ 572 metformin Four of the reviewed studies were and P values were calculated for each Totals phase 3 trials and 1 study was identi- individual type of event. All muraglitazar dosages 2862 Muraglitazar dosages Յ5 mg 2374 fied as phase 2. Two different compara- The primary outcome measure was Pioglitazone or placebo 1351 tors were studied in these 5 trials: pla- a composite end point commonly used cebo and the approved PPAR-␥ agonist in cardiovascular trials: the combined pioglitazone. A high rate of edema and incidence of death, MI, or stroke. A ining the effects of this agent on lipid CHF with high doses of muraglitazar was more specific outcome measure was levels and glycemic control in diabetic observed early in the development pro- generated by substituting cardiovascu- patients. These trials also collected clini- gram and the sponsor ceased develop- lar death for all-cause mortality. A more cal outcomes data for major events in- ment of daily dosages higher than 5 mg, comprehensive outcome measure was cluding all-cause mortality, cardiovas- only requesting regulatory approval for generated by adding CHF and TIA cular mortality, myocardial infarction dosages of 5 mg/d or less.10 Accord- events to the composite. Statistical com- (MI), stroke, transient ischemic at- ingly, we restricted our analysis to treat- parisons between muraglitazar- tack (TIA), and congestive heart fail- ment groups using muraglitazar doses treated and control patients were cal- ure (CHF). of 5 mg/d or less. This analysis yielded culated using the Wald ␹2 test and SAS An advisory committee for the FDA 2374 patients exposed to muraglitazar version 8.0 software (SAS Institute Inc, Endocrinology and Metabolic Drugs Di- and 1351 patients exposed to compara- Cary, NC). PՅ.05 was considered sta- vision reviewed these studies at an open tor agents, of which 823 received pio- tistically significant. publichearingonSeptember9,2005,and glitazone and 528 placebo. A literature recommended approval of the drug as search found a single publication re- RESULTS monotherapy for treatment of type 2 dia- porting results from 1 of these trials,11 The 5 clinical trials in patients with diabetes (by an 8:1 vote) and as combina- and only 2 of the studies are listed on a betes who were exposed to muraglita- tion therapy in patients with blood glu- clinical trials registration site (http: zar or a comparator are summarized in cose not adequately controlled with met- //www.clinicaltrials.gov). Therefore TABLE 1. These studies varied in dura- formin (by a 7:2 vote).8 Public disclosure nearly all information was derived from tion from 24 to 104 weeks and in- laws require the release of briefing docu- the FDA briefing documents. cluded patients who received mura- ments for advisory committee meetings, glitazar monotherapy or muraglitazar which are made available to the public Outcome Measures in combination with 2 other diabetes via the FDA Web site at the time of the Occurrence of all-cause mortality, car- treatments, either metformin or gly- public hearing. The current study rep- diovascular death, nonfatal MI, nonfa- buride. Several features were common resents an analysis of the muraglitazar tal stroke, CHF, and TIA were as- to all 5 trials. Patients were aged 18 to

Downloaded from www.jama.com by MarisaSitjar, on October 24, 2005 MURAGLITAZAR AND ADVERSE EVENTS

70 years, had a body mass index less than Table 2. Baseline Demographic and Laboratory Characteristics* 41, triglycerides lower than 600 mg/dL Combination (6.8 mmol/L), and hemoglobin A1c lev- Therapy Studies els between 7% and 10%.9,10 Patients Monotherapy Studies (CV 168021, CV 168022, with class III or IV CHF were ex- (CV 168006, CV 168018) CV 168025) cluded. Also excluded were patients with Muraglitazar Comparators Muraglitazar Comparators a history of MI, unstable angina, stroke, (n = 729) (n = 366) (n = 1409) (n = 985) TIA, angioplasty, or coronary artery by- Age, mean (SD), y 53.5 (10) 52.5 (9.6) 55.1 (8.7) 54.6 (9.0) pass graft surgery within 6 months prior Men, No. (%) 417 (57.2) 198 (54.1) 724 (51.4) 500 (50.8) to enrollment. The following narra- White race, No. (%) 595 (81.6) 290 (79.2) 1229 (87.2) 878 (89.1) tives briefly describe these studies. Body weight, mean (SD), kg 89.2 (17.9) 89.7 (18.5) 88.0 (18.3) 89.8 (17.9) CV 168006 was a phase 2 mono- Body mass index, mean (SD)† 31.2 (4.9) 31.6 (4.9) 31.3 (4.8) 31.7 (4.9) therapy dose-ranging trial designed to Hemoglobin A1c, mean (SD), % 8.1 (1.1) 8.2 (1.1) 8.1 (1.0) 8.1 (1.0) Fasting plasma glucose, mean 178 (54) 184 (51) 172 (48) 174 (50) explore the effects of a 40-fold range of (SD), mg/dL daily doses (from 0.5 to 20 mg) in pa- Blood pressure, mean (SD), mm Hg tients with type 2 diabetes. During the Systolic 129 (14.8) 129.4 (15.3) 131 (15.0) 131.4 (15.7) first 24 weeks, mean glucose concen- Diastolic 79.7 (8.7) 80.1 (8.30) 80.3 (8.9) 80.4 (8.8) trations were monitored and dosages LDL cholesterol, mean (SD), mg/dL 123.7 (33.3) 128.4 (37.6) 113.1 (34.4) 112.3 (33.0) were titrated subsequently to meet pre- HDL cholesterol, mean (SD), mg/dL 43.0 (9.9) 43.7 (10.6) 45.4 (10.5) 45.6 (10.7) specified levels of glycemic control. Statin use, No. (%) 148 (20.3) 64 (17.5) 343 (24.3) 238 (24.2) Abbreviations: HDL, high-density lipoprotein; LDL, low-density lipoprotein. There was a long-term extension phase SI conversions: To convert glucose to mmol/L multiply values by 0.0555; to convert cholesterol to mmol/L multiply of 104 weeks. A single control group values by 0.0259. *Values represent the number of patients for which data were submitted by the study sponsors.10 was included that received 15 mg/d of †Calculated as weight in kilograms divided by height in meters squared. pioglitazone, which could be titrated up to 45 mg/d to achieve glycemic control. Because 24.9% of patients receiv- glitazone in patients whose hypergly- stroke) occurred in 35 of 2374 mura- ing the 10-mg muraglitazar dose and cemia was not adequately controlled glitazar-treated patients (1.47%) vs 9 of 40.1% of those receiving 20 mg expe- with metformin alone. 1351 control patients (0.67%) (RR, rienced edema, these 2 doses were not 2.23; 95% CI, 1.07- 4.66; P=.03). A used in subsequent trials.9,10 Demographics and more specific outcome measure, sub- CV 168018 was a phase 3 mono- Baseline Characteristics stituting cardiovascular death for all- therapy study of the 2.5- and 5-mg mu- TABLE 2 reports the baseline charac- cause mortality, occurred in 27 of 2374 raglitazar doses compared with a match- teristics of the patients enrolled in these muraglitazar-treated patients (1.14%) ing placebo. Only patients never trials. The FDA documents include pa- vs 7 of 1351 control patients (0.52%) previously treated with an antidia- tients who received muraglitazar or a (RR, 2.21; 95% CI, 0.96-5.08; P=.06). betic agent were eligible. This trial was comparator, pooling the placebo and A more comprehensive outcome mea- recently published.11 pioglitazone treatment groups. The pa- sure adding CHF and TIA events to the CV 168021 was a phase 3 placebo- tients were relatively young (mean Ͻ55 composite yielded an incidence of 50 controlled study comparing 2.5 or 5 mg years) and obese (mean body mass in- of 2374 for muraglitazar-treated pa- of muraglitazar with placebo in dia- dex Ͼ30). Both sexes were approxi- tients (2.11%) vs 11 of 1351 control pa- betic patients with hyperglycemia not mately equally represented. Diabetes tients (0.81%) (RR, 2.62; 95% CI, 1.36- adequately controlled with glyburide. control was relatively poor, with mean 5.05; P=.004). There was a blinded 102-week long- hemoglobin A1c levels higher than 8.0%. Table 4 also shows the incidence term phase, but these results were not rates and RRs for the individual com- available at the time of the FDA advi- Mortality and ponents of the primary outcome mea- sory panel meeting. Cardiovascular Events sure and several other composite end CV 168022 was a randomized, TABLE 3 illustrates the adverse events points typically used in cardiovascu- double-blind, placebo-controlled study occurring in the 5 reviewed clinical lar outcome trials. Individual compo- that compared muraglitazar 2.5 or 5 mg trials conducted in diabetic patients nents of the primary end point showed with placebo in patients with hyper- with muraglitazar dosages of 5 mg/d or consistently greater incidence in the glycemia not adequately controlled with less. TABLE 4 illustrates the event rates, muraglitazar-treated group compared metformin alone. RRs, and P values for muraglitazar com- with controls (RRs ranging from 2.14 CV 168025 was a phase 3 study of pared with pioglitazone or placebo. The to 7.43). However, the number of events the effect of addition of 5 mg of mura- primary outcome measure (all-cause was small and differences for indi- glitazar compared with 30 mg of pio- mortality, nonfatal MI, or nonfatal vidual components of the primary out-

Downloaded from www.jama.com by MarisaSitjar, on October 24, 2005 MURAGLITAZAR AND ADVERSE EVENTS

come measure were not statistically sig- Table 3. Adverse Cardiovascular Events in the Muraglitazar Program nificant, with P values ranging from No. of Patients Noncardiovascular Cardiovascular .053 to .34. The difference in the oc- Study and Therapy Exposed Death Death MI Stroke CHF TIA currence rate for adjudicated CHF was CV 168006 nearly significant, occurring in 13 of Muraglitazar 0.5 mg 236 0 0 0 0 0 0 2374 muraglitazar-treated patients Muraglitazar 1.5 mg 259 2 0 0 0 0 1 (0.55%) and 1 of 1351 controls (0.07%) Muraglitazar 5 mg 245 1 0 0 1 0 1 (RR, 7.43; 95% CI, 0.97-56.8; P=.053). Pioglitazone 15 mg 251 1 0 1 1 0 0 CV 168018 COMMENT Muraglitazar 2.5 mg 111 0 0 0 1 1 0 Muraglitazar 5 mg 114 0 0 0 0 0 0 Two general families of PPAR ago- Placebo 115 0 0 1 0 0 0 nists are currently approved for treat- CV 168021 ment of dyslipidemia and diabetes in the Muraglitazar 2.5 mg 191 0 0 3 2 1 1 United States. The PPAR-␣ agents in- ϩ glyburide clude the fibric acid derivatives gemfi- Muraglitazar 5 mg 193 1 1 4 2 3 1 ϩ glyburide brozil and fenofibrate, which have been Placebo ϩ glyburide 199 0 1 1 0 0 0 available for several decades, al- CV 168022 though all of their modes of action have Muraglitazar 2.5 mg 233 1 1 4 1 0 0 only recently been elucidated.12 This ϩ metformin class of agents is primarily used to treat Muraglitazar 5 mg 205 3 1 3 0 1 0 ϩ metformin dyslipidemia by reducing triglyceride Placebo ϩ metformin 214 0 0 0 0 0 1 levels while moderately elevating lev- 13 CV 168025 els of HDL cholesterol. Muraglitazar 5 mg 587 0 5 1 2 7 1 Two major outcome studies have ϩ metformin demonstrated reductions in adverse car- Pioglitazone 30 mg 572 1 0 1 1 1 0 ϩ metformin diovascular events following adminis- Totals tration of these agents for secondary car- Muraglitazar Յ5 mg 2374 8 8 15 9 13 5 diovascular prevention.2,3 PPAR-␥ agents Pioglitazone or placebo 1351 2 1 4 2 1 1 are approved for glycemic control in pa- Abbreviations: CHF, congestive heart failure; MI, myocardial infarction; TIA, transient ischemic attack. tients with type 2 diabetes. Three drugs have been introduced in this class: rosiglitazone, pioglitazone, and troglita- Table 4. Event Rates and Relative Risks zone. The latter agent was withdrawn from the market due to hepatotoxicity. No. (%) A recent outcomes study showed a trend Muraglitazar Control Relative Risk P toward reduction in vascular events for (n = 2374) (n = 1351) (95% CI) Value pioglitazone but increased incidence of Composite End Points 14 All-cause mortality plus nonfatal 35 (1.47) 9 (0.67) 2.23 (1.07-4.66) .03 CHF. MI or stroke Because of the favorable effects of All-cause mortality plus nonfatal MI, 50 (2.11) 11 (0.81) 2.62 (1.36-5.05) .004 PPAR-␣ and ␥ agonists, development of stroke, CHF, or TIA dual-PPAR agents has been considered Cardiovascular death plus nonfatal 27 (1.14) 7 (0.52) 2.21 (0.96-5.08) .06 MI or stroke a highly promising strategy for simul- Cardiovascular death plus nonfatal 42 (1.77) 9 (0.67) 2.69 (1.30-5.53) .007 taneous treatment of both hyperglyce- MI, stroke, CHF, or TIA mia and dyslipidemia in diabetic pa- All-cause mortality or nonfatal MI 27 (1.14) 7 (0.52) 2.21 (0.96-5.08) .06 tients.6 The first of these agents to be con- Cardiovascular death or nonfatal MI 19 (0.80) 5 (0.37) 2.17 (0.81-5.83) .12 sideredforFDAapprovalismuraglitazar, Individual End Points which can be characterized as a strong All-cause mortality 16 (0.67) 3 (0.22) 3.05 (0.89-10.5) .08 PPAR-␥agonist with moderate ␣effects. Cardiovascular death 8 (0.34) 1 (0.07) 4.57 (0.57-36.5) .15 With any new class of pharmaceuti- Fatal or nonfatal MI 15 (0.63) 4 (0.30) 2.14 (0.71-6.46) .18 cal agents, unexpected toxicity may Fatal or nonfatal stroke 9 (0.38) 2 (0.15) 2.57 (0.55-11.9) .23 emerge during the development pro- Fatal or nonfatal TIA 5 (0.21) 1 (0.07) 2.85 (0.33-24.4) .34 gram. However, in some cases, pharma- Adjudicated CHF 13 (0.55) 1 (0.07) 7.43 (0.97-56.8) .053 ceutical sponsors defer or withhold pub- Abbreviations: CHF, congestive heart failure; CI, confidence interval; MI, myocardial infarction; TIA, transient ischemic lication of phase 2 and 3 clinical trial data attack. until after drug approval. Accordingly,

Downloaded from www.jama.com by MarisaSitjar, on October 24, 2005 MURAGLITAZAR AND ADVERSE EVENTS documents submitted to the FDA for 2.1. The consistency in magnitude and drug showed favorable effects on tri- consideration of approval may consti- direction of the adverse effects across glycerides and HDL cholesterol in these tute the only publicly available source multiple cardiovascular end points re- same clinical trials. A related drug, gem- of objective information for newly ap- duce the likelihood that these find- fibrozil, a pure PPAR-␣ agent, has dem- proved pharmaceutical agents. This has ings result from chance alone. These re- onstrated impressive benefits in 2 major been the case for muraglitazar for which sults are particularly concerning clinical outcome trials. However, other the public disclosure of phase 2 and 3 because the significant excess of ad- PPAR-␣ and ␥ agonists have shown a data occurred via the FDA Web site verse events was observed after only variety of potential cardiovascular tox- shortly before the advisory panel con- limited drug exposure ranging from 24 icities in preclinical studies.16-18 vened to consider the drug for ap- to 104 weeks. Moreover, patients who The specific choice of composite end proval on September 9, 2005. are enrolled in clinical trials often con- point requires additional comment. We From these public disclosure docu- stitute the lowest-risk strata of pa- emphasized a primary outcome mea- ments, we observed a numerical excess tients, and the real world exposure sure that excluded CHF events be- of adverse cardiovascular events for pa- would likely substantially amplify the cause peripheral edema and CHF are tients treated with muraglitazar com- risk. Taken as a whole, these data dem- known hazards of PPAR-␥ agents and pared with controls (patients treated onstrate that it is likely that muraglita- warnings are included in FDA- with either placebo or pioglitazone). zar, if approved by the FDA, would con- approved package inserts for pio- Therefore, to determine the incidence of stitute an unacceptable patient hazard. glitazone and rosiglitazone.19 Nonethe- death and major adverse cardiovascu- We believe it is always important to less, edema was very prominent in lar events, we carefully reviewed the weigh efficacy and safety together in de- studies of muraglitazar, particularly at clinical trial data available within these ciding the clinical utility of any drug. higher doses, occurring in 24.9% and documents and included all trials per- The efficacy for muraglitazar con- 40.1% of patients exposed to the 10- formed in diabetic patients submitted to sisted of a lowering of blood glucose, and 20-mg doses, respectively.9,10 the FDA. We excluded from analysis pa- reduction in triglycerides, and in- Whether muraglitazar constitutes a tients treated with the higher 10- and crease in HDL cholesterol.9,10 These are greater or lesser risk for CHF in com- 20-mg doses of muraglitazar because fur- laboratory end points and must be parison to existing PPAR-␥ drugs such ther development of these dosages was weighed in the context of the more im- as rosiglitazone and pioglitazone re- terminated after a phase 2 trial demon- portant clinical outcomes. Drugs that mains to be determined. The cur- strated a high incidence of peripheral lower blood glucose (sulfonylureas) or rently available data suggest that the in- edema. We also did not include a small, low-density lipoprotein cholesterol cidence of CHF is at least as high with short-term phase 2 trial that adminis- (ezetimibe) have been approved based muraglitazar as with approved PPAR-␥ tered the drug to persons without dia- on laboratory measures of efficacy. agonists.9,10 We also excluded TIA from betes to examine its effects on lipid lev- However, these approvals occurred af- the primary outcome measure be- els. The remaining patients consisted of ter demonstration of excellent safety in cause this is a more subjective end point 2374 participants exposed to 5 mg/d or fairly large patient populations. In con- than stroke or MI. less of muraglitazar and 1351 exposed trast, muraglitazar does not lower low- The precise mechanism underlying to placebo or pioglitazone. density lipoprotein levels and the ben- the increased cardiovascular toxicity ob- The results of this analysis are con- efits of lowering blood glucose for other served with muraglitazar is uncertain. cerning. For the most widely accepted drugs have not always shown a reduc- In different species, PPAR agonists ex- composite end point of death, MI, and tion in serious vascular complica- hibit a variety of biological effects that, stroke, the RR for muraglitazar was tions.15 Thus, it is particularly impor- if they occur in humans, might ex- 2.23. Other end points using nar- tant to weigh the efficacy results against plain the results of this analysis.16-18,20 rower definitions (including only car- the safety concerns. It must be emphasized that each of the diovascular death) or broader compos- It must be emphasized that athero- PPARs activate or suppress different ites (including CHF and TIA events) sclerotic cardiovascular disease is par- genes with only partial overlap in ac- showed similar risks. The most inclu- ticularly common in patients with type tivity. Accordingly, each agent must be sive composite end point that in- 2 diabetes, representing the cause of considered separately from the effi- cluded all-cause mortality, nonfatal MI, death in approximately 80% of dia- cacy and safety perspective. stroke, TIA, and CHF showed a highly betic patients. Thus, any drug used to The possibility of an interaction be- significant increase in RR for muragl- treat diabetes requires careful scrutiny tween muraglitazar and other antidia- itazar-treated patients (2.62; P=.004). for its effects on atherosclerosis- betic therapies must also be consid- Furthermore, there was a highly con- related outcomes, such as MI and stroke. ered. Most adverse cardiovascular sistent pattern of excess morbidity for The apparent increase in adverse car- events occurred in studies in which mu- all of the components of the major out- diovascular events in muraglitazar- raglitazar was combined with gly- come measure with all RRs exceeding treated patients is surprising because the buride or metformin. However, the

Downloaded from www.jama.com by MarisaSitjar, on October 24, 2005 MURAGLITAZAR AND ADVERSE EVENTS number of events in any single study small, primarily because these trials en- propriate dedicated trial to assess car- is too few to draw definitive conclu- rolled only stable patients without a re- diovascular outcomes is performed. sions about the RRs of muraglitazar cent history of cardiovascular events Published Online: October 20, 2005 (doi:10.1001/ with or without concomitant therapy and exposed patients to muraglitazar for jama.294.20.joc50147). Author Contributions: Dr Nissen had full access to all with other agents. Other dual-PPARs in a limited duration (24-104 weeks). Be- of the data in the study and takes responsibility for development may or may not exhibit cause we did not have access to trial da- the integrity of the data and the accuracy of the data ␹2 analysis. similar hazards. Major differences in the tabases, we used simple analysis Study concept and design: Nissen, Topol. effects of pure PPAR-␥ agonists have rather than the more powerful time- Acquisition of data: Nissen, Wolski. Analysis and interpretation of data: Nissen, Wolski, been observed. Pioglitazone and rosi- to-event methods. Accordingly, risk es- Topol. glitazone have differing effects on lip- timates, CIs, and P values could not be Drafting of the manuscript: Nissen, Topol. 21 Critical revision of the manuscript for important in- ids and neither exhibit the hepato- adjusted for other covariates, such as tellectual content: Nissen, Wolski, Topol. toxicity that resulted in market treatment center and duration of ex- Statistical analysis: Nissen, Wolski, Topol. withdrawal of troglitazone.22 Our find- posure to the drug. The number of Study supervision: Nissen. Financial Disclosures: Dr Nissen has consulted for As- ings emphasize the need for robust events was too few to perform formal traZeneca, Abbott, Atherogenics, Bayer, Lipid Sci- safety data for this class of drugs prior testing for heterogeneity. ences, Wyeth, Novartis, Pfizer, Sankyo, Haptogard, Hoffman-LaRoche, Kemia, Takeda, Kowa, Sanofi- to regulatory approval. Nonetheless, some important con- Aventis, Protevia, Novo-Nordisk, Eli Lilly, Kos Phar- There are important limitations to clusions are warranted. Muraglitazar ap- maceuticals, GlaxoSmithKline, Forbes Medi-tech, Vas- genix, Vascular Biogenics, Isis Pharmaceuticals, Viron our analysis. The data used to deter- pears to increase the risk for morbidity Therapeutics, Roche, and Merck–Schering Plough. Dr mine incidence rates and RRs were de- and mortality in diabetic patients dur- Nissen has lectured on behalf of AstraZeneca and Pfizer. Companies are directed to pay honoraria or consult- rived from review of publicly dis- ing relatively short-term treatment. The ing fees directly to charity; no reimbursement is paid closed regulatory documents. We did estimated magnitude of this risk is sub- to Dr Nissen and there is no tax deduction involved. not have access to the original trial da- stantial with RRs indicating a doubling Dr Nissen has been involved in clinical trials receiving research support from AstraZeneca, Eli Lilly, Takeda, tabases. The exact definitions for MI, for irrevocable, major end points and Sankyo, Sanofi-Aventis, Pfizer, Atherogenics, and Lipid stroke, and other events were not avail- composite outcomes. The consistency of Sciences. All reimbursement is directed to the Cardio- vascular Coordinating Center at the Cleveland Clinic able. Other than CHF, adverse events these RRs suggests that this result is not and Dr Nissen receives no personal remuneration. Ms were investigator- and sponsor- due to chance. Accordingly, muraglita- Wolski and Dr Topol reported no financial disclosures. reported, not centrally adjudicated. The zar should not be used or approved to Funding/Support: The authors did not receive any actual number of analyzed events was treat patients with diabetes until an ap- financial support for this analysis.

REFERENCES 1. Vamecq J, Latruffe N. Medical significance of per- /2005-4169M2.pdf. September 9, 2005. Accessibility data to clinical use: available data in humans and clini- oxisome proliferator-activated receptors. Lancet. 1999; verified October 13, 2005. cal applications. Diabetes Metab. 2003;29:207-222. 354:141-148. 9. Table of contents. Endocrinologic and Metabolic 16. Sambandam N, Morabito D, Wagg C, et al. 2. Frick MH, Elo O, Haapa K, et al. Helsinki Heart Drugs Advisory Committee. FDA Web site. Available Chronic activation of PPAR ␣ is detrimental to car- Study: primary-prevention trial with gemfibrozil in at: http://www.fda.gov/ohrms/dockets/ac/05/briefing diac recovery following ischemia. Am J Physiol Heart middle-aged men with dyslipidemia: safety of treat- /2005-4169B2_02_00-FDA-TOC.htm. Accessibility Circ Physiol. September 2005. Epub ahead of print. ment, changes in risk factors, and incidence of verified October 13, 2005. 17. Lygate CA, Hulbert K, Monfared M, et al. The coronary heart disease. N Engl J Med. 1987;317: 10. Advisory Committee briefing document: Par- PPAR-␣ activator rosiglitazone does not alter remod- 1237-1245. gluva (muraglitazar, BMS-298585). FDA Web site. eling but increases mortality in rats post-myocardial 3. Rubins HB, Robins SJ, Collins D, et al. Gemfibrozil Available at: http://www.fda.gov/ohrms/dockets/ac infarction. Cardiovasc Res. 2003;58:632-637. for the secondary prevention of coronary heart dis- /05/briefing/2005-4169B2_01_01-BMS-Pargluva 18. Xu Y, Gen M, Lu L, et al. PPAR-gamma activa- ease in men with low levels of high-density lipopro- .pdf. Accessibility verified October 13, 2005. tion fails to provide myocardial protection in ische- tein cholesterol. N Engl J Med. 1999;341:410-418. 11. Buse JB, Rubin CJ, Frederich R, et al. Muraglita- mia and reperfusion in pigs. Am J Physiol Heart Circ 4. Effect of fenofibrate on progression of coronary- zar, a dual (alpha/gamma) PPAR activator: a random- Physiol. 2005;288:H1314-H1323. artery disease in type 2 diabetes: the Diabetes Ath- ized, double-blind, placebo-controlled, 24-week mono- 19. Nesto RW, Bell D, Bonow RO, et al. Thiazolidine- erosclerosis Intervention Study, a randomised study. therapy trial in adult patients with type 2 diabetes. Clin dione use, fluid retention, and congestive heart fail- Lancet. 2001;357:905-910. Ther. 2005;27:1181-1195. ure: a consensus statement from the American Heart 5. Campbell IW. The clinical significance of PPAR 12. Vu-Dac N, Schoonjans K, Kosykh V, et al. Fi- Association and American Diabetes Association. gamma agonism. Curr Mol Med. 2005;5:349-363. brates increase human apolipoprotein A-II expres- Circulation. 2003;108:2941-2948. 6. Staels B, Fruchart JC. Therapeutic roles of peroxi- sion through activation of the peroxisome proliferator- 20. Molavi B, Rasouli N, Mehta JL. Peroxisome pro- some proliferator-activated receptor agonists. Diabetes. activated receptor. J Clin Invest. 1995;96:741-750. liferator-activated receptor ligands as antiathero- 2005;54:2460-2470. 13. Despres JP, Lemieux I, Robins SJ. Role of fibric acid genic agents: panacea or another Pandora’s box? 7. Devasthale PV, Chen S, Jeon Y, et al. Design and derivatives in the management of risk factors for coro- J Cardiovasc Pharmacol Ther. 2002;7:1-8. synthesis of N-[(4-methoxyphenoxy)carbonyl]-N-[[4- nary heart disease. Drugs. 2004;64:2177-2198. 21. Goldberg RB, Kendall DM, Deeg MA, et al; GLAI [2-(5-methyl-2-phenyl-4-oxazolyl)ethoxy]phenyl] 14. Dormandy JA, Charbonnel B, Eckland DJ, et al. Study Investigators. A comparison of lipid and glyce- methyl]glycine [Muraglitazar/BMS-298585], a novel Secondary prevention of macrovascular events in pa- mic effects of pioglitazone and rosiglitazone in pa- peroxisome proliferator-activated receptor alpha/ tients with type 2 diabetes in the PROactive Study tients with type 2 diabetes and dyslipidemia. Diabe- gamma dual agonist with efficacious glucose and lipid- (PROspective pioglitAzone Clinical Trial In macroVas- tes Care. 2005;28:1547-1554. lowering activities. J Med Chem. 2005;48:2248-2250. cular Events): a randomised controlled trial. Lancet. 22. Rezulin to be withdrawn from the market [press 8. Endocrinologic and Metabolic Drugs Advisory Com- 2005;366:1279-1289. release]. Available at: http://www.fda.gov/ohrms mittee Meeting. FDA Web site. Available at: http: 15. Riveline JP, Danchin N, Ledru F, et al. Sulfonyl- /dockets/ac/00/backgrd/3634b1a_tab6c.htm. Ac- //www.fda.gov/ohrms/dockets/ac/05/minutes ureas and cardiovascular effects: from experimental cessibility verified October 14, 2005.

Downloaded from www.jama.com by MarisaSitjar, on October 24, 2005