Doi:10.1016/j.cardfail.2008.02.007

Journal of Cardiac Failure Vol. 14 No. 6 2008 Pioglitazone and Heart Failure: Results From a Controlled Study in Patients With Type 2 Diabetes Mellitus and Systolic Dysfunction THOMAS D. GILES, ALAN B. MILLER, MD,URI ELKAYAM, MONDIRA BHATTACHARYA, MD,AND ALFONSO PEREZ, New Orleans, Louisiana; Jacksonville, Florida; Los Angeles, California; Deerfield, Illinois Background: Thiazolidinediones are associated with fluid retention, often interpreted as worsening car-diac function, limiting their use in patients with heart failure (HF). We compared the effects of pioglita-zone and glyburide on cardiac function in patients with type 2 diabetes, systolic dysfunction, and NewYork Heart Association (NYHA) functional Class II/III HF.
Methods and Results: Participants received pioglitazone or glyburide (6insulin) for 6 months in thisdouble-blind, randomized, multicenter study. The primary end point was time to HF, a composite ofcardiovascular mortality and hospitalization or emergency room (ER) visit for HF. Secondary endpointsincluded echocardiographic and functional classification assessments. An earlier time to onset and higherincidence of the primary endpoint was noted with pioglitazone (13%) versus glyburide (8%) (P 5 .024).
Hospitalization or ER visit occurred in 30 pioglitazone and 15 glyburide participants, 19 and 12 of whom,respectively, continued treatment. Cardiac mortality (5 versus 6 participants, respectively) and cardiacfunction, as measured by change in ventricular mass index (P 5 .959), ejection fraction (P 5 .413), orfractional shortening (P 5 .280), were similar between treatments.
Conclusions: Pioglitazone was associated with a higher incidence of hospitalization for HF without anincrease in cardiovascular mortality or worsening cardiac function (by echocardiography). (J CardiacFail 2008;14:445e452)Key Words: Cardiovascular disease, thiazolidinediones, left ventricular dysfunction.
Type 2 diabetes mellitus is associated with a heavy car- Thiazolidinediones (TZDs) are insulin-sensitizing agents diovascular (CV) burden, particularly coronary heart dis- used to treat type 2 diabetes. Insulin resistance is associated The increased risk for heart failure in this patient with disturbances in glucose and lipid metabolism, adipose population is estimated to be 2 to 11 times greater than differentiation, and markers of endothelial function, coagu- that for an age-matched nondiabetic population.
lation, and inflammation,all of which contribute to From the 1Tulane University School of Medicine, New Orleans, Louisi- Dr. Thomas Giles has received honoraria for speaking/consulting ana; 2University of Florida Health Science Center, Jacksonville, Florida; services and research grants from Takeda Pharmaceuticals; Dr. Allen 3University of Southern California Keck School of Medicine, Los Angeles, Miller has received honoraria for speaking services from Takeda California and 4Takeda Global Research & Development Center, Inc, Pharmaceuticals and payment from GlaxoSmithKline for consulting services; Dr. Uri Elkayam has received honoraria for speaking/con- Manuscript received August 27, 2007; revised manuscript received Jan- sulting services and research grants from Takeda Pharmaceuticals; uary 22, 2008; revised manuscript accepted February 5, 2008.
Drs. Perez and Bhattacharyra are employees of Takeda Pharmaceu- Reprint requests: Thomas Giles, MD, Professor of Medicine, Tulane University School of Medicine, 109 Holly Drive, Metairie, LA 70005.
Funded by Takeda Pharmaceuticals (Takeda Global Research & Sponsored and funded by Takeda Global Research & Development Cen- ter, Inc. Deerfield, Illinois.
1071-9164/$ - see front matter Partial data from this study have been published previously as congress Ó 2008 Elsevier Inc. All rights reserved.
abstracts (American College of Cardiology Scientific Session 2007 and European Society of Cardiology Congress 2007).
Journal of Cardiac Failure Vol. 14 No. 6 August 2008 atherosclerosis. Despite improvements in many surrogate LVEF, left ventricular mass index (LVMI), cardiac index (CI), and markers of CV risk with TZD treatment, their use is limited fractional shortening (FS).
by the dose-dependent, generally mild to moderate, fluid re- NYHA functional classification was evaluated using standard tention observed with these agents,the mechanism of assessments and definitions at each visit. Exercise tolerance (stan- which is poorly understood, but may involve enhanced so- dardized 6-minute walk test) and global assessments of HF statuswere determined at baseline and final visit.
dium absorption across the collecting duct of the nephron, Adverse events (AEs) and clinical laboratory assessments were where PPARg receptors, the pharmacologic target of determined at each visit. Laboratory assessments were performed TZDs, are abundant.
at Clinical Reference Laboratory (Lenexa, KS) and included stan- Previously, rosiglitazone,and pioglishowed pre- dard serum chemistry (including kidney and liver function assess- served cardiac function and structure relative to placebo or ments), hematology, and urinalysis (including pH, protein, glyburide, respectively. Here we report the results of a clin- glucose, ketones, and specific gravity).
ical study that evaluated the potential of pioglitazone versusglyburide to alter cardiac function in type 2 diabetic pa- tients with symptomatic HF. Compared with previous re- Oral diabetes therapies were stopped at the screening visit (ie, ports,this is the most advanced heart failure (HF) within 2 weeks before randomization). Treatment randomization diabetic patient population studied in a controlled clinical was to pioglitazone (30 mg) or glyburide (10 mg) in a 1:1 ratio, strat- trial of TZD therapy to date.
ified by baseline insulin use. Doses were increased as tolerated to 45mg pioglitazone or 15 mg glyburide if fasting plasma glucose (FPG)was O140 mg/dL (determined weekly for the first 2 weeks and monthly thereafter). Insulin was the only allowed rescue medica- tion. For hypoglycemia, study drug dose was decreased for noninsu-lin users; insulin dose was decreased before study drug for insulin This was a prospective, double-blind, multicenter, controlled users. Maximum tolerated study drug doses were used for the study study evaluating HF progression in participants with type 2 diabe- tes mellitus and systolic dysfunction with symptomatic HF after 6months of treatment with pioglitazone or glyburide (6insulin).
Determination of the Primary End Point Study visits occurred weekly for the first 2 weeks and monthlythereafter. Glyburide provided a reasonable estimate of back- The primary end point, HF progression, was a composite of CV ground CV event rates in this patient population.
mortality and hospitalization or ER visit for HF. An independent The International Conference for Harmonisation-Good Clinical treatmenteblinded clinical end point committee adjudicated all Practice guidance, the Declaration of Helsinki (2000), and local endpoint events using information provided by the study center regulations were followed. All participants provided written in- and all available hospital documentation (ie, admission/discharge formed consent. The study is registered on clinicaltrials.gov documents, physician notes, laboratory/radiography study results, death certificates). Documented increasing dyspnea on exertion,worsening orthopnea or nocturnal dyspnea, increasing fatigue/ decreasing exercise tolerance, renal hypoperfusion (ie, worseningrenal failure) not from adjusted diuretic or after load reduction Participants were $18 years of age with an hemoglobin A1c therapy, pulmonary edema, increased jugular venous pressure (rel- (HbA1c) $7.0%, body mass index #48 kg/m2, New York Heart ative to baseline), or radiologic evidence confirmed a HF diagno- Association (NYHA) functional Class II/III HF (ie, cardiac dis- sis. A data safety monitoring board conducted periodic, unblinded ease resulting in moderate limitation of physical activity; comfort- reviews of all safety data and had the sole authority to stop the able at rest, but symptoms of fatigue, palpitations, dyspnea, or study for safety reasons.
anginal pain occurring with less than what would be consideredordinary activity), left ventricular systolic dysfunction (ie, left Statistical Methods ventricular ejection fraction [LVEF] #40%) at screening, receiv-ing sulfonylurea therapy (6insulin) for $30 days before screen- Statistical analyses (SAS, v8.2, Cary, NC) were performed on ing, or discontinued metformin therapy within 30 days of the intent-to-treat cohort, defined as participants who received at least 1 dose of study drug. The time to first event was estimated Subjects naive to antidiabetic therapy or with type 1 diabetes, using the Kaplan-Meier method. Assuming a 10% dropout rate, serum creatinine O2.0 mg/dL (males) or O1.8 mg/dL (females), a sample size of 300 per group was estimated to allow 90% power systolic blood pressure (SBP) O150 mm Hg or diastolic blood to detect a 10-percentage point treatment differential at 24 weeks pressure (DBP) O100 mm Hg, or recent (#3 months) myocardial with a minimum event rate of 10% in 1 group and 20% in the other infarction, coronary angioplasty or bypass graft, unstable angina, group. Significance was assessed at 0.05 and reported as 2-sided.
transient ischemic attack or stroke, or severe/advanced peripheral Continuous variables were compared using a 2-way analysis of co- vascular disease were excluded.
variance with fixed effects terms of treatment, center, baseline in-sulin use, and baseline value (as covariate). Descriptive statistics for LVMI, CI, FS, and LVEF were derived using observed values.
Analyses of change from baseline to final visit used the last obser- Echocardiograms were recorded by standard protocol at screen- vation carried forward method. Low-density lipoprotein choles- ing and final visits, and read by a single treatmenteblinded reader terol (LDL-C) values coincident with triglyceride values greater (BioMedical Systems, St Louis, MO). Measured indices included than 400 mg/dL were excluded from the summary.
Pioglitazone, Type 2 Diabetes, and Heart Failure  Giles et al 447 Table 1. Demographic and Baseline Characteristics (ITT The study started in June 2000. In November 2003, the data safety monitoring board observed a treatment-group imbalance for the composite endpoint and withdrawal rate. An independent, unblinded analysis revealed a treat- ment-group difference in the composite endpoint (P 5 .09) favoring glyburide, prompting the data safety monitor- ing board recommendation for early study termination.
Disposition and Baseline Characteristics Time since diagnosis of T2DM (mo) Of the 518 participants receiving study drug (pioglita- Time since first diagnosis of CHF (mo) zone 262; glyburide 256), 264 were enrolled in US sites Etiology of heart failure, n (%) and 254 in non-US sites (Participant enrollment Ischemic heart disease at non-US sites included 249 participant enrolled in Argen- Idiopathic dilated cardiomyopathy tina, 1 participant in Colombia, and 4 participants in Mex- Valvular heart disease ico. Overall, 60% of the pioglitazone group and 72% of the glyburide group completed the study. During the first 8 weeks of the study, twice as many pioglitazone as glyburide participants discontinued treatment (18% [47/262] versus 9% [22/256], respectively). Thereafter, the discontinuation CHF medications, n (%) rates were similar between groups (21% versus 19%, re- Renin-angiotensin system spectively). Reasons for discontinuation were similar with antagonist or blocker the exception of AEs (24 pioglitazone versus 14 glyburide), lack of efficacy (10 versus 4, respectively), and withdrew consent (18 versus 7, respectively).
Treatment groups were balanced for baseline demograph- ics and clinical characteristics (). At baseline, approx- Others cardiac therapies (amiodarone, imately one-third of participants in each group were using digoxin, glyceryl trinitrate, orisosorbide mononitrate therapies) insulin. During the study, insulin treatment was initiated in LV ejection fraction (%) twice as many pioglitazone as glyburide users (53 [20%] Fractional shortening (%) Assessed for Eligibility Values are presented as mean (SD) unless otherwise indicated.
ITT, intent-to-treat; BMI, body mass index; SBP, systolic blood pres- sure; DBP, diastolic blood pressure; HR, heart rate; NYHA, New York HbA1c <7% N = 190 Heart Association; HbA1c, hemoglobin A1c; FPG, fasting plasma glucose; EF >40% N = 138TSH ≥3.5 mU/L N = 68 CHF, congestive heart failure; LV, left ventricular; SD, standard deviation.
*Diuretics used by at least 5% of participants in either treatment group Assigned to pioglitazone N = 262 Assigned to glyburide N = 256 Received intervention N = 262 Received intervention N = 256 versus 29 [10%], respectively); most new insulin use oc- Included in ITT N = 262 Included in ITT N = 256 curred early in the study. By Week 8, 14.9% and 9.8% of par- Completed the study N = 156 Completed the study N = 184 Did not complete the study N = 106 Did not complete the study N = 72 ticipants in the pioglitazone and glyburide groups,respectively, had added insulin. Participants received optimal Lost to follow up N = 2 Lost to follow up N = 2 HF management that included renin-angiotensin system an- Discontinued N = 104 Discontinued N = 70 tagonist or blocker, diuretic, antithrombotic, b-blocking, and Adverse event N = 24 Adverse event N = 14 Lack of efficacy N = 10 Lack of efficacy N = 4 amiodarone, digoxin, glyceryl trinitrate, or isosorbide mono- Patient noncompliance N = 3 Patient noncompliance N = 0 Protocol violation N = 1 Protocol violation N = 1 nitrate therapies. During the study, diuretic use increased Voluntary withdraw N = 18 Voluntary withdraw N = 7 from 80.2% to 82.4% in the pioglitazone group and from Investigator discretion N = 2 Investigator discretion N = 2 Terminated by sponsor N = 45 Terminated by sponsor N = 41 85.5% to 87.9% in the glyburide group.
Baseline LVEF values were similar between groups (29.7% pioglitazone versus 29.4% glyburide), although Fig. 1. Participant disposition. ITT, intent to treat; HbA1c, hemo-globin A1c; EF, ejection fraction; TSH, thyroid-stimulating the mean value for participants enrolled at non-US sites (27.5% versus 28.65%, respectively) was lower than for

Journal of Cardiac Failure Vol. 14 No. 6 August 2008 Baselineto Final Visit
Mean Percent Change from
Pioglitazone (15 –45mg QD) Glyburide (5 –15mg QD) in FPG (mg/dL)
Hospitalization for HF (%)
LS Mean Change From Baseline
No BL insulin BL Insulin Use Study Time (Weeks)
Fig. 2. (A) Kaplan-Meier estimates of probability of first event based on determination by clinical end point committee. P value is on thetreatment-group difference at final visit. (B) Hospitalization for heart failure by baseline insulin use and by geographic region (ITT anal-ysis). (C) Mean percent change from baseline to final visit in echocardiographic parameters (ITT analysis). Final visit analysis incorporatedthe last observation for each subject. (D) Mean change from baseline in FPG by study week and treatment group (ITT analysis). PIO, pio-glitazone; GLY, glyburide; ITT, intent to treat; LVMI, left ventricular mass index; CI, cardiac index; EF, ejection fraction; FS, fractionalshortening; FPG, fasting plasma glucose.
participants enrolled at US sites (32.2% versus 30.1%, re- Subgroup analyses of the primary composite end point, spectively). Similarly, FS values were lower among partic- although limited by smaller sample sizes and fewer events ipants at non-US sites (21.2% versus 20.5%) than US sites than the overall analysis, revealed higher event rates with (22.7% versus 21.7%, respectively). These findings indicate pioglitazone versus glyburide among participants using in- that participants enrolled outside of the US tended to have sulin at baseline (19.1% versus 8.4%; respectively, P 5 more advanced heart failure disease.
.032), whereas event rates among participants not using in-sulin at baseline (10.4% versus 8.1%, respectively; P 5.285), with concomitant insulin use during the study(16.9%vs 11.0%, respectively; P 5 .172), and without concomitant Progression of HF insulin use (9.2% and 6.1%, respectively; P 5 .185) were Pioglitazone was associated with an earlier time to onset similar between group. As with the primary analysis, (P 5 .024) and a higher incidence rate of the com-posite end point compared with glyburide ).
Table 2. Incidence of Primary End Point and Contributing Whereas mortality contributed little to the overall rate Events (ITT Analysis) and was similar between treatments (5 with pioglitazone and 6 with glyburide), overnight hospitalization for HF con- tributed disproportionately to the composite endpoint andwas higher in the pioglitazone group. Thirty pioglitazone Composite event n (%) Death from CV cause and 15 glyburide participants were hospitalized or had an emergency room visit for HF; of these, 19 and 12 partici- for worsening CHF pants, respectively, continued in the study, indicating most cases of HF were reversible and not treatment CV, cardiovascular; CHF, congestive heart failure; ER, emergency room.
P value for treatment-group difference, determined by log-rank test.
Pioglitazone, Type 2 Diabetes, and Heart Failure  Giles et al 449 hospitalization for HF contributed disproportionately to the Metabolic Effects composite end point among baseline and concomitant insu- FPG was significantly decreased with glyburide relative lin users, whereas the rate of hospitalization for HF was low to pioglitazone during the first 4 weeks of treatment and balanced among participants who did not use insulin ). Nearly half of pioglitazone users compared with one third of glyburide users were at maximal dose by Participants enrolled at non-US sites tended to enter the Week 2. By Week 4, a steady and consistent increase in study with more severe disease as measured by mean EF mean FPG was noted with glyburide, whereas decreases and FS. Subgroup analyses of the composite endpoint by were noted with pioglitazone. By Week 16, and thereafter, region also revealed a higher incidence of events for the a significant difference in mean FPG was observed favoring non-US cohort (pioglitazone 17.2% versus glyburide pioglitazone. Consistent with changes in FPG was a reduc- 9.5%, P 5 .061) than for the US cohort (9.7% versus tion in HbA1c with pioglitazone compared with glyburide 6.9%, respectively P 5 .241). Within the composite end least squares (LS) (means point, the rate of hospitalization was similar among glybur- e0.98% versus e0.73%, respec- tively), that was first noted at Week 20 and sustained for the ide users irrespective of region (4.8% non-US versus 4.6% remainder of the study (at Week 24, P 5 .007).
US), but was twice as high at non-US (13.3%) versus US At Week 24, significant differences were seen between sites (6.7%) among pioglitazone users ).
pioglitazone and glyburide in triglycerides ( Higher rates of HF progression were reported for piogli- 7.6 mg/dL, P ! .001 [LS mean % change tazone users in the O64-years-of-age cohort (pioglitazone 7.2]), high-density lipoprotein cholesterol (HDL-C) (4.8 14.7%; glyburide 7.l%; P 5 .037) and among men (13.0% versus 7.1%, respectively; P 5 .030), whereas the e0.8 mg/dL, P ! .001 [LS mean % change 15.5 rates were similar between groups in the #64-years-of-age e0.5]), and LDL-C (6.9 versus e2.4 mg/dL, P 5 .016 [LS mean % change 7.0 versus 2.4], respectively).
cohort and among women (14.1% versus 11.9%; P 5 .477)(12.0% versus 9.0%, respectively; P 5 .270).
Overall rates of AEs and serious AEs were similar be- tween treatment groups, though more pioglitazone than gly- buride participants discontinued because of an AE (Hypoglycemia was more common with glyburide; Despite an increased reporting of HF with pioglitazone, edema was more common with pioglitazone. The most echocardiographic data indicate preserved rather than commonly reported AE leading to study discontinuation worsening cardiac function with similar changes in the was aggravated HF (2.3%, pioglitazone; 1.2%, glyburide).
LVMI (g/m2) (pioglitazone e4.1 [47.08]; glyburide e6.9 Weight gain, known to occur with both TZDs and insu- [37.52]; P 5 .959), FS (%) (0.2 [6.79] and 1.2 [6.42], lin, was more than twice as likely to be reported as an respectively; P 5 .280), and LVEF (%) (3.6 [10.20] and AE (6.1% versus 2.7%) and mean weight gain was greater 2.5 [9.86], respectively; P 5 .413) noted between treatmentgroups. Cardiac index (Lminm2), however, was signifi-cantly increased with pioglitazone compared with glybur- Table 3. Treatment-Emergent Adverse Events (ITT ide (0.14 6 0.79 versus e0.03 6 0.55, respectively, P 5.012) and significant increases in CI were also ob- served for pioglitazone users in both the US and non-US cohorts. No treatment-group differences were observed within the US and non-US cohorts for LVMI, FS, or CHF/cardiac failure aggravatedy LVEF. Notably, the absolute relative improvements in CI seen with pioglitazone were similar between regions de- spite the significantly poorer baseline readings at non-US Significant treatment effects on blood pressure favoring AEs causing discontinuation ($1.5%)z pioglitazone were observed at Weeks 8 and 12 for SBP CHF/cardiac failure aggravatedy and at Weeks 4, 8, 12, and final visit for DBP. There Serious AEs ($1.5%)z were no notable treatment effects on any other measured vi- CHF/cardiac failure aggravatedy tal sign, electrocardiogram parameter, physical examina- Myocardial infarction tion, 6-minute walk test, changes in NYHA functional Ventricular tachycardia classification, or laboratory parameter with the exception of heart rate. At all time points, mean decreases in heartrate were noted with pioglitazone; mean decreases up to CHF, congestive heart failure.
*Events reported for $5% of participants in either group are shown.
Week 12 and minor increases thereafter were noted with yData are investigator reports, not adjudicated endpoint events.
zEvents reported for $1.5% of participants in either group are shown.
Journal of Cardiac Failure Vol. 14 No. 6 August 2008 (2.10 kg versus 1.23 kg, respectively, P 5 .012) with piogli- pioglitazone after 24 weeks of treatment relative to glybur- tazone than with glyburide. Among insulin users, mean ide. Changes in echocardiographic parameters were similar weight gain was 2.20 kg and 1.44 kg, respectively, whereas between treatment groups, with the exception of CI, which among noninsulin users it was 0.87 kg and 0.26 kg, respec- showed significant increases with pioglitazone versus gly- tively. Moreover, insulin use more than tripled the reporting buride. The increase in cardiac output is consistent with rate of weight gain among pioglitazone users (9.2% with the clinically relevant decreases in blood pressure noted insulin versus 2.5% without insulin).
with pioglitazone, changes that have been observed in otherTZD studies.Additionally, positive effects on triglycer- ides and HDL-C were noted with pioglitazone.
Our data also support previous studies of pioglitazone on Pioglitazone was associated with an excess of overnight other CV outcomes, such as MI and CV death. In PROac- hospitalizations for HF, resulting in a higher rate of the tive, pioglitazone was associated with a 16% (95% CI composite end point versus glyburide in this study popula- 0.72e0.98; P 5 .027) risk reduction for a composite end tion of type 2 diabetic patients with preexisting symptom- point of all-cause mortality, non-fatal MI (excluding silent atic HF. The separation between treatment groups for MI), and non-fatal stroke relative to placebo.Addition- time to first event occurred at approximately 6 weeks, sta- ally, among patients with a previous MI, the risk of recur- bilized thereafter, and was driven almost entirely by hospi- rent MI was significantly reducas was the risk of talization for HF. CV-related deaths contributed little to the recurrent stroke among patients with previous overall composite end point in either group and CV mortal- Among patients who reported serious HF in PROactive, ity was not increased with pioglitazone despite the higher treatment had no effect on mortality.Together, these rates of hospitalization for HF. Importantly, once hospital- findings indicate that the potential CV benefits of pioglita- ized for HF, participants in the pioglitazone group did not zone treatment were not offset by the higher rate of serious have worse outcome compared with participants in the gly- HF in the high-risk PROactive patient population.
buride group: HF was generally reversible, not treatment Although neither pioglitazone nor the TZD rosiglitazone limiting, and not associated with increased CV-related has been shown to have deleterious effects on cardiac struc- deaths or adverse events (excluding events of HF).
ture or function,there appear to be incompletely under- In our study, a greater proportion of pioglitazone than stood differences between these agents that impact their glyburide participants received insulin (54.2% and 43.8%, overall CV safety profile. In contrast to the trend of CV respectively), and most new insulin use was initiated early benefit observed in PROactive and in the recent meta- in the trial. Additionally, dose titration tended to take place analysis of pioglitazone studies published by Lincoff et earlier in the pioglitazone group, as most participants were reports suggest that rosiglitazone treatment is associ- receiving maximal dose by Week 2. The increased use of ated with increased risk of MI and CV Such differ- insulin and more rapid dose titration in the pioglitazone ences in outcome may, in part, be explained by the distinct group, both of which were driven by FPG targets stipulated lipid effects of these two TZDs. In a head-to-head stud in the protocol, are consistent with known properties of the of 802 patients with diabetic dyslipidemia, pioglitazone sig- 2 study medications. Glyburide and other sulfonylureas, nificantly improved triglyceride and HDL-C levels, and in- which are insulin secretagogues, lower glucose levels effec- creased LDL-C to a significantly lesser extent relative to tively in the short term but tend to lose effectiveness over rosiglitazone. In a separate study, 305 patients with diabetic time. TZDs, however, may take longer to reach peak effect dyslipidemia who were on rosiglitazone and stable statin but appear to provide equivalent or better long-term control therapy were switched to pioglitazone (statin therapy was than sulfonylureas, as was observed in the present study.
maintAfter 17 weeks of treatment, significant de- Rapid dose titration and disproportionate insulin use in creases in triglyceride, total cholesterol, and LDL-C levels the pioglitazone group also likely contributed to the in- were noted. These improvements were accompanied by fa- creased reports of HF among pioglitazone users. TZDs vorable shifts in LDL subfractions from smaller dense par- are associated in a dose-dependent fluid retention, whereas ticles to the less atherogenic large buoyant particles.
insulin, a sodium-retaining hormone, is associated with an Although no study evaluating the correlation between pio- estimated risk for HF of 14% to 24% when used as mono- glitazone-induced changes in lipids and other markers of therapy or in combination therapies in patients with type 2 atherosclerosis has been conducted, it is reasonable to hy- diabIn our study, hospitalization for HF was reported pothesis that these unique features of pioglitazone may con- for 14.1% of pioglitazone participants and 6.4% of glybur- tribute to the trend for beneficial CV outcomes observed in ide participants who used insulin, compared with 5.0% PROactive and may explain the absence of an apparent car- versus 3.4%, respectively, among participants for whom in- dioprotective effect with rosiglitazone.
sulin was not added.
Several lessons were learned from this study. First, HF Increased left ventricular size and reduced ejection frac- associated with pioglitazone use does not diminish cardiac tion are typical with systolic dysfunction in HF. Despite the function or structure. Second, use of metformin during the earlier time to and higher incidence rate of HF, there was no run-in period may have provided sufficient glycemic con- evidence of worsening cardiac function or structure with trol early in the study, thereby eliminating the need to Pioglitazone, Type 2 Diabetes, and Heart Failure  Giles et al 451 initiate early insulin therapy. Third, starting patients at the patients with type 2 diabetes and HF (NYHA Class I and lowest dose of pioglitazone and slowly titrating based on II), whereas monitoring for edema or weight gain.
HbA1c rather than FPG would have allowed sufficienttime for pioglitazone to exert its glucose-lowering effects without a premature dose increase. Slow-dose titration isparticularly important in patients with HF given that TZD We acknowledge the members of the Clinical Endpoint treatment is associated dose-dependent fluid retention, Committee, Dr. Timothy Fagan and Dr. Jean Barbey, and which may have exacerbated symptoms of HF in this study.
the data safety monitoring board, Dr. Lloyd Fisher (Statis- Metformin was contraindicated for use in diabetic pa- tician), Dr. Sidney Goldstein (Cardiologist), and Dr. Eugene tients with heart failure at the time the study was designed.
Consequently, glycemic control was maintained via dose ti- We acknowledge the participation of the study investi- tration and addition of insulin, circumstances that increased gators: United States: Dr. Usman Ahmad, Dr. Jerome the risk for HF among pioglitazone users. The lack of an Anderson, Dr. Zaeem Ansari, Dr. Mohammad Ashraf, Dr.
oral alternative to insulin likely had a second important Kenneth Barnwell, Dr. Scott Baron, Dr. Malik Baz, Dr.
impact on the studydit contributed to the relatively high Raye Bellinger, Dr. Martin Berk, Dr. Rajinder Bhalla, Dr.
dropout rate noted with pioglitazone, most of which was at- Geetha Bhat, Dr. Seth Bilazarian, Dr. Martin Bilsker, Dr.
tributed to treatment failure and withdrawal of consent. The Victor Bradley, Dr. John Buse, Dr. Dennis Buth, Dr. Robert differential dropout rate was greatest during the first 8 Carhart, Dr. Albert Carr, Dr. Vernon Carrigan, Dr. Premin- weeks of the study (18% versus 8.6%, respectively), dra Chandraratna, Dr. Ajit Chauhan, Dr. Edward Chesne, when there was a significant treatment-group difference in Dr. Saurabh Chokshi, Dr. Harry Colfer, Dr. Clancy Cone, HbA1c favoring glyburide; thereafter, the dropout rates Dr. Michale Cooperman, Dr. Kerry Cranmer, Dr. Nizar Da- were similar (21% vs 19.1%, respectively). Since the study boul, Dr. Kola Danisa, Dr. Scott Davis, Dr. Michale Dawes, completed, the restrictions for use of metformin in patients Dr. Joshua DeLeon, Dr. Peter DeStefano, Dr. Prakash Deed- with HF have been relaxed and recent studies suggest met- wania, Dr. George Dennish, Dr. Robert DiBianco, Dr. Jerry formin may have beneficial effects in this patient popula- Drucker, Dr. Steven Duckor, Dr. Phillip Duncan, Dr. Victor Echenique, Dr. Leo Egbujiobi, Dr. Isaac Eisenstein, Dr.
Mark Eisner, Dr. M El Shahawy, Dr. Amin Elashker, Dr.
Limitations of the Study Kent Farnsworth, Dr. Marc Feldman, Dr. Richard Fleming,Dr. Ronald Freudenberger, Dr. Jasper Fullard Jr, Dr. Cyn- This study, the first to evaluate a TZD in a patient popu- thia L. Gaboury, Dr. Raul Gaona, Dr. Paul Garcia, Dr. Chris lation with symptomatic systolic dysfunction, had several Geannopoulos, Dr. Mark Goldberg, Dr. Ivan Goldsmith, Dr.
limitations, including a 2-week screening period in which Francis Goldstein, Dr. Lloyd Goodman, Dr. Mark A. Good- oral antidiabetic medications were withheld, a rapid dose man, Dr. Ronald C. Gove, Dr. Irmina Gradus-Pizlo, Dr.
titration schedule contrary to dosing recommendations for Frank J. Green, Dr. Bruce Greenspahn, Dr. Arun Grover, clinical use of pioglitazone, and dose escalation based on Dr. Richard Guthrie, Dr. Ernest A. Haeusslein, Dr. Peter FPG rather than HbA1c. These limitations likely contrib- Hanley, Dr. Robert Henry, Dr. Paul Hermany, Dr. Richard uted to the differential completion rate in each arm of the R. Heuser, Dr. General Hilliard, Dr. David Huffman, Dr.
study, which, in turn, may have affected the comparability Paul Hughes, Dr. Joseph Iemma, Dr. Henry Ingersoll, Dr.
of the treatment groups.
William Isley, Dr. Bruce Iteld, Dr. Bruce Jackson, Dr.
Our study confirms the need for slow dose titration, mon- Syed Jafri, Dr. Cherian John, Dr. Leonard Kanterman, Dr.
itoring for weight gain and edema, and careful use of insu- Michael Kesselbrenner, Dr. Mark Kipnes, Dr. Robert Kip- lin in this patient population, recommendations consistent perman, Dr. Michael Koren, Dr. Marc Kozinn, Dr. Jeffrey with the current ACTOS package and the American Kramer, Dr. Arvid Lade, Dr. Rafael Levites, Dr. Cameron Diabetes Association/American Heart Association guide- Lindsey, Dr. Irving Loh, Dr. Kelly Mandagere, Dr. Joseph lines for TZD use in type 2 diabetics with systolic or dia- Martin, Dr. Frank McBarron, Dr. Gregory McCarroll, Dr.
stolic dysfunction.
Frank McGrew, Dr. Darren McGuire, Dr. Keith McLean,Dr. J. Frederick McNeer, Dr. Nina Miles-Everett, Dr. Sam Miller, Dr. Mario Motta, Dr. Ashis Mukherjee, Dr. JamesNeiman, Dr. Joel Neutel, Dr. Asim Nisar, Dr. Diane Nor- In this study in patients with type 2 diabetes and systolic mandin, Dr. Thomas W. Oates, Dr. Robert Orchard, Dr. Ay- dysfunction with NYHA functional Class II/III HF, there man Osman, Dr. Gregory Pennock, Dr. Robert Pinner, Dr.
was an excess of overnight hospitalization for HF with pio- Andreas Prinz, Dr. Ramon Ramirez, Dr. Alan Rauba, Dr.
glitazone without an excess of CV mortality or worsening Marc Rendell, Dr. Theodore Ritzer, Dr. Jacob Rubin, Dr.
cardiac function. HF that occurred with pioglitazone was Tariq Saleem, Dr. Lakhjit S. Sandhu, Dr. Jaime Sandoval, reversible and not treatment limiting. These results support Dr. Ricky Schneider, Dr. Jeffrey Shanes, Dr. Bramah Singh, current guidelines that recommend initiating thiazolidine- Dr. Emil Skobeloff, Dr. Jon Slotoroff, Dr. Joseph Smiddy, dione therapy at the lowest dose and titrating gradually in Dr. William Smith, Dr. Allan Stahl, Dr. William Stavrou, Journal of Cardiac Failure Vol. 14 No. 6 August 2008 Dr. Richard Steingart, Dr. Mark Strauss, Dr. Kelley Sulli- Association Functional Class I or II Heart Failure. J Am Coll Cardiol van, Dr. John Sutherland, Dr. Sharad Swami, Dr. Melvin 9. Giles TD, Miller AB, Elkayam U, Bhattacharya M, Perez A. Preserv- Tonkon, Dr. Ricardo Ubillus, Dr. Stephen Ulrich, Dr. Kevin ing systolic function: two studies demonstrate safety of pioglitazone in Vaska, Dr. Nampalli Vijay, Dr. Krishnaswami Vijayaragha- patients with type 2 diabetes mellitus and heart failure. J Am Coll Car- van, Dr. John Walker, Dr. Louis Warrick, Dr. C. Kevin Watt, diol 2007;49(Suppl A):66A.
Dr. Jeffrey Wayne, Dr. Howard Weinberg, Dr. Edwin Whit- 10. Maru S, Koch GG, Stender M, Clark D, Gibowski L, Petri H, et al.
ney, Dr. James Wilson. Argentina: Dr. Jorge Altamirano, Antidiabetic drugs and heart failure risk in patients with type 2 diabe-tes in the U.K. primary care setting. Diabetes Care 2005;28:20 Dr. Marcos Amuchastegui, Dr. Victor Balestrini, Dr. Eduar- 11. Derosa G, Cicero AF, Dangelo A, Gaddi A, Ragonesi PD, do Blumberg, Dr. Guillermo Bortman, Dr. Aaron Bronstein, Piccinni MN, et al. Thiazolidinedione effects on blood pressure in Dr. Gustavo Caruso, Dr. Hugo Colombo, Dr. Jesu´s Cuadra- diabetic patients with metabolic syndrome treated with glimepiride.
do, Dr. Eduardo Escudero, Dr. Graciela Gimeno, Dr. Luiz Hypertens Res 2005;28:917e24.
Guzman, Dr. Emilio Kuschnir, Dr. Hugo Lastiri, Dr. Jorge 12. Basu A, Jensen MD, McCann F, Mukhopadhyay D, Joyner MJ, Rizza RA. Effects of pioglitazone versus glipizide on body fat distri- Lerman, Dr. Alberto Lorenzatti, Dr. Claudio Majul, Dr. Fe- bution, body water content, and hemodynamics in type 2 diabetes.
lipe Martinez, Dr. Olga Moises, Dr. Antonio Norkus, Dr.
Diabetes Care 2006;29:510e4.
Daniel Nul, Dr. Luis Orozco-Linares, Dr. Sergio Perrone, 13. Dormandy JA, Charbonnel B, Eckland DJA, Erdmandn E, Massi- Dr. Jorge Pinzetta-Martinez, Dr. Mario Russo, Dr. Fernando Benedetti M, Moules I, et al. Secondary prevention of macrovascular Sokn, Dr. Alvaro Sosa-Liprandi, Dr. Hugo Torres, Dr. Ser- events in patients with type 2 diabetes in the PROactive Study (PRO-spective pioglitAzone Clinical Trial In macroVascular Events). Lancet gio Varini, Dr. Cesar Zaidman. Colombia: Dr. Rodrigo Botero-Lopez, Carlos Cure-Cure, Dr. Efraı´n Gomez-Lopez, 14. Erdmann E, Dormandy JA, Charbonnel B, Massi-Benedetti M, Dr. Monica Jaramillo-Jaramillo, Dr. Alexis Llamas- Moules IK, Skene AM. on behalf of the PROactive investigators. The Jimenez, Dr. Hugo Luquez, Frederico Silva-Sieger, Dr.
effect of pioglitazone on recurrent myocardial infarction in 2445 pa- Jorge Vela´squez-Norena. Puerto Rico: Dr. Elizabeth Bar- tients with type 2 diabetes and previous myocardial infarctiondResultsfrom PROactive (PROactive 05). J Amer Coll Cardiol 2007;49: ranco. Mexico: Dr. Abel Pavia-Lopez.
15. Wilcox R, Bousser M-G, Pirag V, Betteridge J, Schernthaner G, Kupfer S, et al. PROactive 04: Effects of pioglitazone in patientswith type 2 diabetes with or without previous strokedresults fromPROactive (PROspective pioglitAzone Clinical Trial In macroVascu-lar Events). Stroke 2007;38:865e73.
16. Ryde´n L, Thra´insdo´ttir I, Swedberg K. Adjudication of serious heart failure in patients from PROactive:[letter]. Lancet 2007;368:189e90.
1. Haffner SM, Lehto S, Ro¨nnemaa T, Pyo¨ra¨la¨ K, Laakso M. Mortality 17. Erdmann E, Standl E, Charbonnel B, Wilcox R, Spanheimer R, Massi- from coronary heart disease in subjects with type 2 diabetes and in Benedetti M, et al. Pioglitazone use and heart failure in patients with nondiabetic subjects with and without prior myocardial infarction. N type 2 diabetes and preexisting cardiovascular disease: data from the Engl J Med 1998;339:229 PROactive study (PROactive 08). Diabetes Care 2007;30:2773 2. Nichols GA, Gullion CE, Ephross SA, Brown JB. The incidence of 18. Lincoff MA, Wolski K, Nicholls SJ, Nissen S. Pioglitazone and risk of congestive heart failure in type 2 diabetes. Diabetes Care 2004;27: cardiovascular events in patients with type 2 diabetes mellitus. JAMA 3. Reavan G. Metabolic syndrome: pathophysiology and implications for 19. Nissen SE, Wolski K. Effect of rosiglitazone on the risk of myocardial management of cardiovascular disease. Circulation 2002;106:286 infarction and death from cardiovascular causes. N Engl J Med 2007; 4. ACTOS prescribing information. Takeda Pharmaceuticals America, 20. Goldberg RB, Kendall DM, Deeg MA, Buse JB, Zagar AJ, Pinaire JA, 5. Avandia prescribing information. GlaxoSmithKline; 2007.
et al. A comparison of lipid and glycemic effects of pioglitazone and 6. Zhang H, Zhang A, Kohan DE, Nelson RD, Gonzalez FJ, Yang T. Col- rosiglitazone in patients with type 2 diabetes and dyslipidemia. Diabe- lecting duct-specific deletion of peroxisome proliferator-activated tes Care 2005;28:1547e54.
receptor g blocks thiazolidinedione-induced fluid retention. PNAS 21. Berhanu P, Kipnes MS, Khan MA, Perez AT, Kupfer SF, Spanheimer RC, et al. Effects of pioglitazone on lipid and lipoprotein 7. Guan YF, Hao C, Cha DR, Rao R, Lu W, Kohan DE, et al. Thiazoli- profiles in patients with type 2 diabetes and dyslipidaemia after treat- dinediones expand body fluid volume through PPARgstimulation of ment conversion from rosiglitazone while continuing stable statin ther- ENaC-mediated renal salt absorption. Nature Medicine 2005;11: apy. Diab Vasc Dis Res 2006;3:39e44.
22. Eurich DT, Majumdar SR, McAlister FA, Tsuyuki RT, Johnson JA.
Improved clinical outcomes associated with metformin in patients McMorn SO, Roberts JN, et al. A randomized, placebo-controlled trial with diabetes and heart failure. Diabetes Care 2005;28:2345e51.
assessing the effects of rosiglitazone on echocardiographic function 23. American Diabetes Association. Standards of medical care in diabetes.
and cardiac status in type 2 diabetic patients with New York Heart Diabetes Care 2008;31:S12e54.

Source: http://www-hsc.usc.edu/~elkayam/articles/piogiltazone%20and%20heart%20failure.pdf