- Email: [email protected]
Thiazolidinedione use is not associated with worse cardiovascular outcomes: A study in 28,332 high risk patients with diabetes in routine clinical practice
International Journal of Cardiology 167 (2013) 1380–1384
Contents lists available at SciVerse ScienceDirect
International Journal of Cardiology journal homepage: www.elsevier.com/locate/ijcard
Thiazolidinedione use is not associated with worse cardiovascular outcomes: A study in 28,332 high risk patients with diabetes in routine clinical practice Brief title: Thiazolidinedione use and mortality Ronan Roussel a, b, c,⁎, 1, Samy Hadjadj d, e, 1, Blandine Pasquet f, g, 1, Peter W.F. Wilson h, 1, Sidney C. Smith Jr. i, 1, Shinya Goto j, 1, Florence Tubach f, g, 1, Michel Marre a, b, c, 1, Avi Porath k, 1, Michel Krempf l, 1, Deepak L. Bhatt m, 1, P. Gabriel Steg b, c, n, 1 a
INSERM, U-695, 75006, Paris, France Univ Paris Diderot, Sorbonne Paris Cité, UMR-738, F-75018, Paris, France c AP-HP, Hôpital Bichat, F-75018, Paris, France d CHU, Diabetology, F-86000, Poitiers, France e INSERM, CIC, F-86000, Poitiers, France f INSERM, CIE-801, F-75018, Paris, France g APHP, Hôpital Bichat, Département d'Epidémiologie et de Recherche Clinique, F-75018, Paris, France h Atlanta VA Medical Center and Emory University School of Medicine, Atlanta, GA, USA i Center for Cardiovascular Science and Medicine, UNC School of Medicine, Chapel Hill, NC, USA j Tokai University School of Medicine, Isehara, Japan k Medical Division, Maccabi Healthcare Services and Faculty of Health Sciences, Ben-Gurion University of Negev, Beer Sheva, Israel l INSERM, UMR915, Institut du Thorax, Université de Nantes, CHU, Nantes, France m VA Boston Healthcare System, Brigham and Women's Hospital, and Harvard Medical School, Boston, MA, USA n INSERM, U-698, Paris, France b
a r t i c l e
i n f o
Article history: Received 26 October 2011 Received in revised form 4 April 2012 Accepted 6 April 2012 Available online 4 May 2012 Keywords: Thiazolidinedione Diabetes Cardiovascular risk Mortality Registry
a b s t r a c t Objective: Assess the cardiovascular safety of Thiazolidinediones (TZD) in routine clinical practice. Background: TZD are insulin-sensitizing antidiabetic drugs commonly used in type 2 diabetes, but their cardiovascular safety has been questioned. We examined the association between TZD use and major cardiovascular outcomes. Methods: We examined 2-year mortality, non-fatal myocardial infarction (MI), and congestive heart failure (CHF) rates among outpatients with high cardiovascular risk and diabetes according to TZD use in the REACH Registry. Multivariable adjustment and propensity scores were used in the analyses. Results: A total of 4997 out of 28,332 patients took TZDs at baseline. During follow-up, 1532 patients died. The mortality rates (95% confidence interval [CI]) were 6.5% (5.5–7.6) with TZD and 7.2% (6.33–8.06) without; adjusted hazard ratio (HR) was 1.06 (0.89–1.26, P=0.54). The lack of association with mortality was consistent across subgroups regardless of history of atherothrombosis or CHF. Rates of non-fatal MI (HR 1.10, 95% CI 0.83–1.45, P=0.50) and non-fatal CHF (HR 0.90, CI 0.75–1.09, P=0.27) were similar in users and non-users. TZD use was associated with an increased risk of CHF in patients aged >80 years (HR 1.59, CI 1.06–2.40, P=0.03). Conclusions: Use of TZD was not associated with increased incidence of major cardiovascular events in patients with diabetes from this large registry. Older patients experienced an increased risk of CHF over the study interval. Limitations of this study include its observational design, and thus unmeasured confounders cannot be excluded. © 2012 Elsevier Ireland Ltd. All rights reserved.
1. Introduction Abbreviations: BMI, Body Mass Index; CHF, Congestive Heart Failure; CI, Confidence Interval; eGFR, estimated Glomerular Filtration Rate; HR, Hazard Ratio; MI, Myocardial infarction; RCT, Randomized Controlled Trial; ROC, Receiver Operating Characteristic; TZD, Thiazolidinedione. ⁎ Corresponding author at: Department of Diabetology, Endocrinology and Nutrition, Bichat Hospital, 46 rue Henri Huchard, 75018 Paris, France. Tel.: + 33 140257301; fax: + 33 140258842. E-mail address: [email protected] (R. Roussel). 1 For the REACH Investigators. 0167-5273/$ – see front matter © 2012 Elsevier Ireland Ltd. All rights reserved. doi:10.1016/j.ijcard.2012.04.019
Type 2 diabetes is a chronic condition associated with a high cardiovascular burden. Recent clinical trials suggest that an intensive glucose-control strategy may decrease the risk of myocardial infarction (MI), but it does not reduce mortality [1]. Conventional anti-diabetic therapy includes lifestyle modification, metformin, sulfonylurea, and insulin. The thiazolidinediones (TZD) are ligands of Peroxisome Proliferator Activated Receptor-gamma, a transcription factor expressed
R. Roussel et al. / International Journal of Cardiology 167 (2013) 1380–1384
in the adipose tissue, but also in endothelial cells. The net effect on metabolism is sensitization to insulin and glucose lowering. The cardiovascular safety and efficacy of these drugs were tested in several trials; pioglitazone was tested as secondary prevention via the PROActive study (PROspective pioglitAzone Clinical Trial in macroVascular Events) and rosiglitazone as primary prevention via the RECORD (Rosiglitazone Evaluated for Cardiovascular outcomes in ORal combination therapy for type 2 diabetes study) trial [2,3]. These two trials did not show a clear connection between TZD and cardiovascular risk, but a meta-analysis has first raised concerns regarding the cardiovascular safety of rosiglitazone, particularly with respect to the risk of MI [4]. However, none of these studies was sufficiently powered for analyzing effects on mortality because the numbers of deaths were small; 363 of 5238 (6.9%) participants in the PROActive study and 293 of 4447 (6.6%) participants in the RECORD trial. Observational studies can complement randomized trials if they are dedicated to specific aims, are of sufficient size, and have a design allowing prospective and extensive collection of relevant characteristics of the participants in order to minimize potential confounding factors. We compared mortality, non-fatal MI, and congestive heart failure (CHF) rates in patients with diabetes according to the use of TZD in the REACH (REduction of Atherothrombosis for Continued Health) Registry, an international prospective cohort of patients with either established atherosclerotic arterial disease or at risk for atherothrombosis. 2. Methods The study design, as well as the baseline description of the REACH Registry has been published previously [5–7]. Consecutive outpatients aged ≥45 years, with established coronary artery disease, cardiovascular disease, or peripheral arterial disease, or patients with ≥3 atherothrombotic risk factors were enrolled by 5587 physician practices in 44 countries between December 2003 and December 2004. In each country, the protocol was submitted to the institutional review boards according to local requirements, and signed informed consent was obtained for all patients. A standardized international case report form, completed at each study visit, was used and data were centrally collected. Diabetes was defined by anti-diabetic drug use by a patient. Overweight was defined by a body mass index (BMI) of 25–29, and obesity at BMI of ≥ 30. The estimated glomerular filtration rate was calculated according to the Modification of Diet in Renal Disease formula; stages of kidney function were defined according to the Kidney Disease Outcomes Quality Initiative guidelines (http:// www.kidney.org/professionals/KDOQI/guideline_diabetes/guide1.htm). As prior investigations have shown race-related differences in the medical treatment and outcomes of high risk patients, race was self-reported and in instances of mixed-racial origin, patients were asked to choose the race that had the strongest personal influence [6,7]. In some countries, local rules did not allow the recording of race and data were considered missing. Data were collected from participating physicians regarding patients' clinical outcomes, vascular procedures, employment status, weight and current smoking status, as well as any medications used, at each follow-up visit until 24 months after enrolment, and dates of the outcomes were collected if appropriate; events were not adjudicated. To ensure data quality, in each country 10% of all sites (that is, physicians) that enrolled ≥1 patient were chosen randomly 6% of the time, and an additional 4% were chosen due to the number of queries and missing data from the site to undergo onsite quality control by completing a site visit. For each of the sites undergoing monitoring, 100% of case report forms for patients enrolled at that site were monitored for source documentation and accuracy. The present analysis focuses on 2-year rates for mortality, non-fatal MI, and nonfatal CHF (i.e. hospitalization for CHF) in the diabetic subset of the REACH Registry. Patients with only baseline data were not included in the analysis. To take into account confounding factors we used the propensity score method [8]. This score represents the probability of receiving TZD given the characteristics of an individual. The list of co-variables was built in a two-step process. Bivariate analyses were first conducted to determine the variables associated with TZD prescription. In this preliminary selection, the P value limit was arbitrarily set at 0.20. These variables were then introduced in order to build a multivariable logistic regression model. To ensure the robustness of the score, the variables with >5% missing data were excluded. Finally, the propensity score was calculated for every patient using this model with the individual data of the patient. The quality of the model was assessed using global evaluation (Wald test) and by calculating the area under the corresponding receiver-operating characteristic (ROC) curve. To validate the score, a minimal value of 0.7 was expected. Patient characteristics were presented as mean ± standard deviation, and compared using Student's t tests or chi-squared tests. Hazard ratios (HRs) for death, nonfatal MI, and non-fatal CHF were calculated using a Cox proportional-hazard model, involving survival time in any individual patient, with TZD use and propensity scores as co-variables, as well as other specified factors. Survival time was calculated according to the date of the outcome as collected in the registry. For patients who remained
1381
free of the outcome, data were censored at the time of the last visit with available information. Event rates are presented based on total sample sizes. The homogeneity of treatment effects across subgroups was tested by adding interaction terms to the relevant models. A P value of b 0.05 was considered significant. All analyses were performed with the use of the SAS software, version 9.1.
3. Results 3.1. Patient characteristics Of the 65,441 patients enrolled in the REACH Registry and with data during follow-up, 28,332 patients had type 2 diabetes and available data on TZD use; they represent our study population. The Table 1 Baseline characteristics of the study population by TZD use. Characteristics
TZD use Yes (n = 4997) No (n = 23 335) p-value*
Age (years) Male sex Region of enrolment North America Latin America Western Europe Eastern Europe Middle East Asia (excluding Japan) Australia Japan Racial origin Caucasian Hispanic East Asian South Asian Other Asian Black Other Clinical and biological variables Waist (cm) BMI (kg/m2) Overweight (BMI 25–29) Obese (BMI ≥30) Smoking Former smoker Current smoker Serum creatinine (mg/L) Fasting blood glucose (mg/dL) Fasting total cholesterol (mg/dL) Fasting triglycerides (mg/dL) Hypertension Systolic blood pressure (mm Hg) Diastolic blood pressure (mm Hg) Prior arterial disease None One bed Two beds Three beds Prior CHF Baseline therapies ≥1 Antiplatelet agent Acetylsalicylic acid ≥1 Lipid-lowering agent Statins Angiotensin II receptor blockers Angiotensin-converting enzyme inhibitors Beta-blockers Calcium channel blockers Diuretics Nitrates Other antihypertensive agents Sulfonylurea Metformin Insulin Other anti-diabetic agent
b 0.0001 0.99 b 0.0001
67.1 ± 9.6 3034 (60.7)
68.6 ± 9.6 14 165 (60.7)
4063 (81.3) 61 (1.2) 338 (6.8) 27 (0.5) 26 (0.5) 309 (6.2) 19 (0.4) 154 (3.1)
8945 (38.3) 756 (3.2) 6300 (27.0) 1515 (6.5) 407 (1.7) 2401 (10.3) 844 (3.6) 2167 (9.3)
3255 (66.9) 398 (8.2) 403 (8.3) 60 (1.2) 189 (3.9) 500 (10.3) 60 (1.2)
13 001 (60.7) 1222 (5.7) 3480 (16.3) 352 (1.6) 1439 (6.7) 1184 (5.5) 735 (3.4)
104.9 ± 18.7 31.6 ± 6.7 1561 (31.8) 2642 (53.8)
100.1 ± 16.3 29.0 ± 5.9 8704 (37.9) 8435 (36.7)
1865 (38.5) 691 (14.3) 1.15 ± 0.73 145.1 ± 55.6 185.4 ± 48.2 180.3 ± 113.7 4448 (89.0) 135.2 ± 18.3 75.8 ± 10.9
9028 (40.0) 3072 (13.6) 1.13 ± 0.73 146.0 ± 53.1 191.9 ± 48.6 173.4 ± 105.2 20 360 (87.3) 139.9 ± 19.5 78.4 ± 11.4
2042 (40.9) 2314 (46.3) 561 (11.2) 80 (1.6) 704 (14.3)
6686 (28.7) 12 834 (55.0) 3374 (14.5) 441 (1.9) 3845 (16.8)
b 0.0001
3718 (74.4) 3367 (67.4) 4310 (86.3) 3981 (79.7) 1628 (32.9) 2575 (52.0)
17 196 (73.7) 14 723 (63.2) 17 663 (75.7) 16 047 (68.8) 6158 (26.6) 11 486 (49.5)
0.32 b 0.0001 b 0.0001 b 0.0001 b 0.0001 0.0015
2115 (42.6) 1654 (33.4) 2465 (49.6) 792 (16.3) 541 (11.0) 2426 (49.3) 2374 (48.2) 1027 (20.8) 417 (8.9)
10 363 (44.6) 8642 (37.2) 10 731 (46.1) 5537 (24.1) 2651 (11.5) 10 198 (43.8) 9399 (40.4) 6546 (28.1) 2681 (11.6)
0.0078 b 0.0001 b 0.0001 b 0.0001 0.35 b 0.0001 b 0.001 b 0.0001 b 0.0001
b 0.0001
Data are n (%) or mean ± standard deviation.
b 0.0001 b 0.0001 b 0.0001 b 0.0001 0.13
0.07 0.31 b 0.0001 0.0002 0.0006 b 0.0001 b 0.0001 b 0.0001
1382
R. Roussel et al. / International Journal of Cardiology 167 (2013) 1380–1384
characteristics of this population at baseline, according to TZD use, are presented in Table 1. Patients prescribed TZD differed strongly from TZD non-users regarding baseline characteristics, risk factors, and use of evidence-based therapies for secondary prevention; they were younger, more frequently obese, Caucasian, and living in North America. Based on their fasting glycemia, their glycemic control was similar. Regarding classical risk factors, TZD users had a slightly lower fasting total cholesterol concentration, higher triglyceride concentration, and blood pressure levels were lower. Use of TZD was noticeably lower in patients with a previous history of atherothrombotic events. Also, TZD users were more frequently prescribed evidence-based therapies for secondary prevention, such as antiplatelet agents, statins, and renin-angiotensin system blockers; of note, the use of diuretics was higher in TZD users (Table 1). 3.2. Propensity score The list of co-variables used to calculate the propensity score included age, geographic region of enrolment, height, BMI, smoking status, atrial fibrillation/flutter, use of lipid-lowering agents, statins, angiotensin II receptor blockers, angiotensin-converting enzyme inhibitors, beta-blockers, calcium channel blockers, nitrates, peripheral arterial claudication medications, insulin, and use of other antidiabetic agents. The propensity score reached the quality requirements because the likelihood associated with the model was strong (Wald test, P b 0.0001) and the area under the ROC curve (0.779) exceeded the 0.7 threshold (Supplementary Fig. 1). 3.3. TZD use and outcomes At the 2-year follow-up visit, 239 (event rate 6.5%, 95% confidence interval [CI] 5.5–7.6) and 1293 (7.2%, 95% CI 6.3–8.1) deaths, 103 (event rate 2.7%, 95% CI 2.0–3.3) and 480 (2.6%, 95% CI 2.1–3.1) non-fatal MI events, and 279 (event rate 6.3%, 95% CI 5.5–7.1) and 1505 (6.8%, 95% CI 6.3–7.3) non-fatal CHF events were recorded, for TZD users and non-users, respectively. No association was found between TZD use and any of these outcomes; adjusted HRs were 1.06 (0.89–1.26, P = 0.54), 1.10 (0.83–1.45, P = 0.50), and 0.90 (0.75–1.09, P = 0.27), for mortality, non-fatal MI, and non-fatal CHF, respectively (Fig. 1). Among patients who suffered a non-fatal MI, 75 subsequently died later during follow-up in similar proportions according to TZD use at baseline; 16 of 103 (15.5%) and 59 out of 480 (12.3%) in TZD users and non-users, respectively (P = 0.37). All adjusted analyses yielded consistent results—there was no difference
between death rates, non-fatal MI rates, and CHF rates, regardless of whether the rates were adjusted for age and sex, or for age, sex and propensity to receive TZDs, or all of these and differences in baseline characteristics. The HRs associated with TZD use for death, non-fatal MI, and non-fatal CHF according to relevant subgroups of patients are presented in Supplementary Fig. 2. There were no inconsistencies among subgroups for association of TZD with death or non-fatal MI (Supplementary Fig. 2). However, use of TZD interacted significantly (P for interaction = 0.002) with classes of age for risk of non-fatal CHF. A higher risk of CHF on TZD was observed among patients aged >80 years (HR 1.59, 95% CI 1.06–2.40). Patients with no established cardiovascular disease, but only risk factors had similar outcomes compared with those with established atherothrombosis in 1, 2, or 3 arterial beds (Supplementary Fig. 2). As the use of TZD was mostly prevalent in North America, we performed a sensitivity analysis restricted to US patients. The results were similar to those found for the overall population and showed a neutral association between TZD and death (HR 0.96, 95% CI 0.66–1.39, P = 0.82), non-fatal MI (HR 0.90, 95% CI 0.61–1.33, P = 0.59), and non-fatal CHF (HR 1.07, 95% CI 0.80–1.45, P = 0.64). 4. Discussion Diabetes very often coexists with other risk factors for atherosclerosis or even a history of atherosclerotic events, therefore clinicians have to take this background into account when prescribing antidiabetic drugs. In this large, international, prospective observational study of diabetic outpatients with high cardiovascular risk, the use of TZD was not associated with an increased risk for death, nonfatal MI, or non-fatal CHF, although some increased risk for CHF was noted in the subgroup of patients aged >80 years. Several studies have recently reported the cardiovascular outcomes of patients treated with TZD in usual care settings, although the present work focused on high cardiovascular risk profile, a unique characteristic [9–14]. Patients of the PROactive trial testing pioglitazone had to have evidence of extensive macrovascular disease before recruitment and were therefore at high risk [2]. Despite this, 6.9% of the participants died during an average time of observation of 34.5 months, slightly less than the mean mortality rate (7.1%) during the shorter 2-year follow-up of our registry. This may be explained by the mean age of participants, 68.3 ± 9.6 years in our study, and 61.7 ± 7.8 years in the PROactive trial. The RECORD open-label trial included a minority of patients with established atherothrombotic disease, and the proportion of death was 6.6%, but with a much longer mean observation time of 5.5 years [3]. Both studies were reassuring on
Fig. 1. Association of TZD use with mortality, non-fatal MI and CHF. Adjusted HRs (on all factors significantly associated in univariate, as well as propensity score) for mortality, nonfatal Myocardial Infarction (MI), and congestive heart failure (CHF) associated with TZD use as recorded at baseline, in the whole group of diabetic patients in the REACH Registry.
R. Roussel et al. / International Journal of Cardiology 167 (2013) 1380–1384
cardiovascular outcomes and mortality associated with TZD, except for heart failure. The PROactive trial suggested a benefit of pioglitazone on a secondary composite of death, non fatal MI and stroke [2]. Taken as a whole, our observational study extends the cardiovascular safety profile of TZD regarding death and ischemic events in an older and at very high risk population. Several studies have recently reported the cardiovascular outcomes of patients treated with TZD in usual care settings, although the present work focused on high cardiovascular risk profile, a unique characteristic [9–14]. There are distinct differences between these studies and the present analysis, such as the present study was not limited to CHF patients (they were a minority, 16%, in the trial) or to elderly patients. Also, in the Aguilar et al. study the control group included only patients receiving noninsulin-sensitizing drugs, by design excluding patients treated with biguanides [11]. However, biguanides are possibly associated with favorable effects on the myocardium [15]. These studies concur that in stable outpatients with high risk, TZD use was safe regarding risk of CHF and mortality. The REACH Registry is prospective by design, and 1532 deaths and 583 non-fatal MI events (approximately 5 times more than in the RECORD trial) were recorded. However, the design of the registry required clinicians to provide information on class of drugs used, but not on the molecules. Thus, it is impossible to separate the effects of the two marketed TZD, pioglitazone and rosiglitazone, which is a clear limitation in our study. Rosiglitazone has been suspected of increasing the risk for MI and pioglitazone may be protective or neutral [16]. At the time of enrolment in the registry, rosiglitazone and pioglitazone shared equitably the market in the US and the world [17]. Thus, we cannot exclude from our results that an adverse effect of one drug may be more or less compensated by a positive effect of the other on clinical events, particularly, but not only, on MI. However, the issue of cardiovascular safety of rosiglitazone compared to pioglitazone is still debated, as no large randomized controlled face to face trial was published; results of retrospective cohort studies are controversial, suggesting either similar MI, CHF or all-cause death risk for patients exposed to rosiglitazone and pioglitazone [14,18], or an inferiority of rosiglitazone [19–22]. An important point to acknowledge in our study is the lack of CHF events associated with TZD use. Only elderly patients (aged >80 years) were at higher risk for CHF using TZD compared with non-users. This lack of deleterious impact of TZD on heart failure might seem unexpected. Increased risk for CHF, presumably through salt retention, is a known risk when prescribing a TZD [23]. This has been documented in randomized controlled trials (RCTs), comparing TZD with placebo or hypoglycemic drugs [24,25]. Such a deleterious effect was not shown in our registry, as in others [11,12]. Of note, no adverse effects of rosiglitazone on brain natriuretic peptide levels were seen in a RCT of patients with coronary artery disease but without clinical heart failure at baseline [25]. Major design differences may explain this apparent discrepancy. First, in RCTs the exposure to the drug begins with the study, which is not the case in a registry. We do not have information on the duration of exposure to TZD before the enrolment of participants, but as the registry began in 2004, about 5 years after the marketing of TZD in the US and many other countries, it is expected that many patients were already being treated with TZD for years. Thus, if patients were to develop heart failure or even retention edema due to TZD, they were likely to have already done so prior to enrolment, after clinicians had taken the appropriate measures. Finally, RCTs are double-blinded, but clinicians who prescribed TZD to participants to the REACH Registry were likely to be fully aware of any side effects of the medication and may have adjusted medical management to minimize them. Particularly with regard to CHF, they may have prescribed diuretics more frequently; we did observe such an increased rate of use of diuretics among patients receiving TZD (Table 1). Interestingly, TZD users were also more prone to receive other protective drugs
1383
such as renin-angiotensin system blockers or statins. This is a common feature of registries analyzed according to TZD use [10,11,13,14] and also of the randomized, but open-label, trial RECORD where statins and diuretics were prescribed 20% and 25%, respectively, more frequently at the last visit in the rosiglitazone arm [3]. Studies describing prescribing patterns (drugs but also lifestyle) before and after the initiation of TZD would be informative. Overall, TZD users were younger, healthier, and received more aggressive treatment with life saving therapy. This lack of balance between groups according to TZD use is a clear limitation of the study, intrinsic to its observational design. Despite this, careful adjustment on potential confounders, using a robust propensity score and multivariable analysis would tend to minimize falsely reassuring results, although unmeasured confounders remain distinctly possible. As glycemic control might be one of the confounders, we included fasting glycemia in the multivariate analysis, as a surrogate of HbA1c, not available in the study, since both parameters are known for a long time as being tightly correlated [26]. No direct information was available regarding diabetes duration. We hypothesize that mean duration was quite long, although heterogeneous. For example, the mean duration of diabetes was 10.4 ± 8.7 years in the BARI 2D trial in patients with type 2 diabetes and coronary artery disease [27]. We may consider requirement of insulin as a surrogate for a long history of diabetes; our results did not differ according to its use. In summary, in this international registry of high cardiovascular risk diabetic patients, TZD use was safe as it was not associated with increased rates of major cardiovascular events. These findings from a large international registry, reflective of routine practice, suggest that when clinicians decided to prescribe TZDs they coped successfully with its potential side effects and contraindications in order to optimize the risk–benefit balance. Funding This work was supported by sanofi-aventis, Bristol-Myers Squibb, and the Waksman Foundation (Tokyo, Japan). All analyses from the REACH Registry are prepared by independent authors who are not governed by the funding sponsors and are prioritized and reviewed by an academic publications committee before submission. The statistical analyses were conducted solely by an academic team (BP, PR). The funding sponsors have the opportunity to review manuscript submissions but do not have authority to change any aspect of a manuscript. The REACH Registry is sponsored by sanofi-aventis, Bristol-Myers Squibb, and the Waksman Foundation (Tokyo, Japan), who assisted with the design and conduct of the study and data collection. Conflict of interest Dr Roussel has received research grants, honoraria, or consulting fees from sanofi-aventis, MSD Chibret, Servier, Eli Lilly, Novo Nordisk, Medtronic, Janssen-Cilag, and Lifescan. Dr Hadjadj has received research grants, honoraria, or consulting fees from AstraZeneca, sanofi-aventis, MSD Chibret, Servier, Roche, Bristol-Myers Scribb. Prof Wilson has received research grants from sanofi-aventis within the past 3 years. Prof Goto has received honoraria and consulting fees from Eisai, sanofiaventis, Daiichi-Sankyo, GlaxoSmithKline, Bristol-Myers Squibb, Otsuka, Bayer, Schering-Plough, Takeda, Astellas, AstraZeneca, Novartis and Kowa. Prof Goto also received research grants from Pfizer, Ono, Eisai, Otsuka, Daiichi-Sankyo, sanofi-aventis, Takeda and Astellas within the last 3 years. Dr Tubach has received research grants from Astra Zeneca, Abbott, sanofi-aventis, Schering Plough, Servier, and Wyeth. Prof Marre has received honoraria as an adviser and for lectures: Novo-Nordisk, sanofi-aventis, Merck, Servier, and Eli Lilly. Prof Krempf has received honoraria for advisory board attendance and consulting fees from AstraZeneca, Bristol-Myers Squibb, Roche, Pfizer, Merck, sanofi-aventis,
1384
R. Roussel et al. / International Journal of Cardiology 167 (2013) 1380–1384
Pierre Fabre; payment for speakers' bureau from Bristol-Myers Squibb, Pfizer, sanofi-aventis, Merck, Schering-Plough, Merck Lipha. Dr Bhatt receives research grants from Amarin, AstraZeneca (for a study of saxagliptin), Bristol-Myers Squibb, Eisai, Ethicon, Medtronic, sanofi-aventis, and The Medicines Company; is a research collaborator with Takeda (not pertaining to TZDs); and had previously worked on a study of rosiglitazone funded by GlaxoSmithKline (the PPAR study, reference #25 of this paper). Prof Steg has received research grant from sanofi-aventis (1999–2008); is on the speaker's bureau for Boehringer Ingelheim, Bristol-Myers Squibb, GlaxoSmithKline, Menarini, Medtronic, Nycomed, Pierre Fabre, sanofi-aventis, Servier, and The Medicines Company; is on the consulting/advisory boards for Astellas, AstraZeneca, Bayer, Boehringer Ingelheim, Bristol-Myers Squibb, Daiichi-Sankyo, Endotis, GlaxoSmithKline, Medtronic, MSD, Nycomed, sanofi-aventis, Servier, and The Medicines Company; and is a stockholder for Aterovax. Supplementary data related to this article can be found online at doi:10.1016/j.ijcard.2012.04.019. Author contributions Dr Roussel had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Roussel, Hadjadj, Krempf, Bhatt, Steg. Acquisition of data: Steg, Goto. Analysis and interpretation of data: Roussel, Hadjadj, Marre, Steg, Bhatt. Drafting of the manuscript: Roussel, Hadjadj, Steg, Bhatt. Critical revision of the manuscript for important intellectual content: Roussel, Hadjadj, Goto, Wilson, Porath, Smith Jr, Marre, Tubach, Krempf, Steg, Bhatt. Statistical analysis: Roussel, Pasquet, Tubach, Steg. Obtained funding: Steg, Goto. Administrative, technical, or material support: Steg, Goto. Study supervision: Steg, Bhatt. Acknowledgments The REACH Registry is endorsed by the World Heart Federation. A complete list of REACH investigators is published in JAMA 2006; 295:180–189. The REACH Registry enforces a no ghost-writing policy. This manuscript was written and edited by the authors, who take full responsibility for its content. The drafts were written by Ronan Roussel. The authors of this manuscript have certified that they comply with the Principles of Ethical Publishing in the International Journal of Cardiology. REACH Registry Executive Committee Deepak L. Bhatt, MD, MPH, VA Boston Healthcare System and Brigham and Women's Hospital, Boston, MA, USA (Chair); Ph. Gabriel Steg, MD, INSERM U-698, Université Paris 7, AP-HP, Paris, France (Chair); E. Magnus Ohman, MD, Duke University Medical Center, Durham, NC, USA; Joachim Röther, MD, Johannes Wesling Klinikum Minden, Minden, Germany; Peter W.F. Wilson, MD, Emory University School of Medicine, Atlanta, GA, USA. REACH Registry Global Publication Committee Mark J. Alberts, MD, Northwestern University Medical School, Chicago, IL, USA; Deepak L. Bhatt, MD, MPH, VA Boston Healthcare System and Brigham and Women's Hospital, Boston, MA, USA (Chair); Ralph D'Agostino, PhD, Boston University, Boston, MA, USA; Kim Eagle, MD, University of Michigan, Ann Arbor, MI, USA; Shinya Goto, MD, PhD, Tokai University School of Medicine, Isehara, Kanagawa, Japan; Alan T. Hirsch, MD, Minneapolis Heart Institute Foundation and University of Minnesota School of Public Health, Minneapolis, MN, USA; Chiau-Suong Liau, MD, PhD, Taiwan University Hospital and College of Medicine, Taipei, Taiwan; Jean-Louis Mas, MD, Centre Raymond Garcin, Paris, France; E. Magnus Ohman, MD, Duke University Medical Center, Durham, NC, USA; Joachim Röther, MD, Klinikum Minden, Minden, Germany; Sidney C. Smith, Jr, MD, University of North Carolina at Chapel Hill, Chapel Hill, NC, USA; Ph. Gabriel Steg, MD, INSERM
U-698, Université Paris 7, AP-HP, Paris, France (Chair); Peter W.F. Wilson, MD, Emory University School of Medicine, Atlanta, GA, USA. References [1] Control GroupTurnbull FM, Abraira C, et al. Intensive glucose control and macrovascular outcomes in type 2 diabetes. Diabetologia 2009;52:2288–98. [2] Dormandy JA, Charbonnel B, Eckland DJ, et al. Secondary prevention of macrovascular events in patients with type 2 diabetes in the PROactive Study (PROspective pioglitAzone Clinical Trial In macroVascular Events): a randomised controlled trial. Lancet 2005;366:1279–89. [3] Home PD, Pocock SJ, Beck-Nielsen H, et al. RECORD Study Team. Rosiglitazone evaluated for cardiovascular outcomes in oral agent combination therapy for type 2 diabetes (RECORD): a multicentre, randomised, open-label trial. Lancet 2009;373:2125–35. [4] Nissen SE, Wolski K. Effect of rosiglitazone on the risk of myocardial infarction and death from cardiovascular causes. N Engl J Med 2007;356:2457–71. [5] Ohman EM, Bhatt DL, Steg PG, et al. The REduction of Atherothrombosis for Continued Health (REACH) Registry: an international, prospective, observational investigation in subjects at risk for atherothrombotic events-study design. Am Heart J 2006;151(786):e1–10. [6] Bhatt DL, Steg PG, Ohman EM, et al. International prevalence, recognition, and treatment of cardiovascular risk factors in outpatients with atherothrombosis. JAMA 2006;295:180–9. [7] Steg PG, Bhatt DL, Wilson PW, et al. One-year cardiovascular event rates in outpatients with atherothrombosis. JAMA 2007;297:1197–206. [8] Rubin DB. Estimating causal effects from large data sets using propensity scores. Ann Intern Med 1997;127:757–63. [9] Delea TE, Edelsberg JS, Hagiwara M, Oster G, Phillips LS. Use of thiazolidinediones and risk of heart failure in people with type 2 diabetes: a retrospective cohort study. Diabetes Care 2003;26:2983–9. [10] Lipscombe LL, Gomes T, Lévesque LE, Hux JE, Juurlink DN, Alter DA. Thiazolidinediones and cardiovascular outcomes in older patients with diabetes. JAMA 2007;298:2634–43. [11] Aguilar D, Bozkurt B, Pritchett A, Petersen NJ, Deswal A. The impact of thiazolidinedione use on outcomes in ambulatory patients with diabetes mellitus and heart failure. J Am Coll Cardiol 2007;50:32–6. [12] Pantalone KM, Kattan MW, Yu C, et al. The risk of developing coronary artery disease or congestive heart failure, and overall mortality, in type 2 diabetic patients receiving rosiglitazone, pioglitazone, metformin, or sulfonylureas: a retrospective analysis. Acta Diabetol 2009;46:145–54. [13] Masoudi FA, Inzucchi SE, Wang Y, Havranek ER, Foody JM, Krumholz HM. Thiazolidinediones, metformin, and outcomes in older patients with diabetes and heart failure: an observational study. Circulation 2005;111:583–90. [14] Habib ZA, Tzogias L, Havstad SL, et al. Relationship between thiazolidinedione use and cardiovascular outcomes and all-cause mortality among patients with diabetes: a time-updated propensity analysis. Pharmacoepidemiol Drug Saf 2009;18:437–47. [15] Inzucchi SE, Masoudi FA, McGuire DK. Metformin in heart failure. Diabetes Care 2007;30:e129. [16] Loke YK, Kwok CS, Singh S. Comparative cardiovascular effects of thiazolidinediones: systematic review and meta-analysis of observational studies. BMJ 2011;342:d1309, http://dx.doi.org/10.1136/bmj.d1309. [17] Starner CI, Schafer JA, Heaton AH, Gleason PP. Rosiglitazone and pioglitazone utilization from January 2007 through May 2008 associated with five risk-warning events. J Manag Care Pharm 2008;14:523–31. [18] Wertz DA, Chang CL, Sarawate CA, Willey VJ, Cziraky MJ, Bohn RL. Risk of Cardiovascular Events and All-Cause Mortality in Patients Treated With Thiazolidinediones in a Managed-Care Population. Circ Cardiovasc Qual Outcomes 2010;3:538–45. [19] Gerrits C, Bhattacharya M, Manthena S, Baran R, Perez A, Kupfer S. A comparison of pioglitazone and rosiglitazone for hospitalization for acute myocardial infarction in type 2 diabetes. Pharmacoepidemiol Drug Saf 2007;16:1065–71. [20] Graham DJ, Ouellet-Hellstrom R, MaCurdy TE, et al. Risk of acute myocardial infarction, stroke, heart failure, and death in elderly Medicare patients treated with rosiglitazone or pioglitazone. JAMA Jul 28 2010;304(4):411–8. [21] Juurlink DN, Gomes T, Lipscombe LL, Austin PC, Hux JE, Mamdani MM. Adverse cardiovascular events during treatment with pioglitazone and rosiglitazone: population based cohort study. BMJ 2009;339:b2942. [22] Winkelmayer WC, Setoguchi S, Levin R, Solomon DH. Comparison of cardiovascular outcomes in elderly patients with diabetes who initiated rosiglitazone vs pioglitazone therapy. Arch Intern Med 2008;168:2368–75. [23] Lago RM, Singh PP, Nesto RW. Congestive heart failure and cardiovascular death in patients with prediabetes and type 2 diabetes given thiazolidinediones: a meta-analysis of randomised clinical trials. Lancet 2007;370:1129–36. [24] Komajda M, McMurray JJ, Beck-Nielsen H, et al. Heart failure events with rosiglitazone in type 2 diabetes: data from the RECORD clinical trial. Eur Heart J Apr 2010;31(7):824–31. [25] Bhatt DL, Chew DP, Grines C, et al. Peroxisome proliferator-activated receptor gamma agonists for the Prevention of Adverse events following percutaneous coronary Revascularization–results of the PPAR study. Am Heart J 2007;154:137–43. [26] Gonen B, Rubenstein A, Rochman H, Tanega SP, Horwitz DL. Haemoglobin A1: An indicator of the metabolic control of diabetic patients. Lancet Oct 8 1977;2(8041):734–7. [27] BARI 2D Study GroupFrye RL, August P, et al. A randomized trial of therapies for type 2 diabetes and coronary artery disease. N Engl J Med 2009;360(24):2503–15.
Contents lists available at SciVerse ScienceDirect
International Journal of Cardiology journal homepage: www.elsevier.com/locate/ijcard
Thiazolidinedione use is not associated with worse cardiovascular outcomes: A study in 28,332 high risk patients with diabetes in routine clinical practice Brief title: Thiazolidinedione use and mortality Ronan Roussel a, b, c,⁎, 1, Samy Hadjadj d, e, 1, Blandine Pasquet f, g, 1, Peter W.F. Wilson h, 1, Sidney C. Smith Jr. i, 1, Shinya Goto j, 1, Florence Tubach f, g, 1, Michel Marre a, b, c, 1, Avi Porath k, 1, Michel Krempf l, 1, Deepak L. Bhatt m, 1, P. Gabriel Steg b, c, n, 1 a
INSERM, U-695, 75006, Paris, France Univ Paris Diderot, Sorbonne Paris Cité, UMR-738, F-75018, Paris, France c AP-HP, Hôpital Bichat, F-75018, Paris, France d CHU, Diabetology, F-86000, Poitiers, France e INSERM, CIC, F-86000, Poitiers, France f INSERM, CIE-801, F-75018, Paris, France g APHP, Hôpital Bichat, Département d'Epidémiologie et de Recherche Clinique, F-75018, Paris, France h Atlanta VA Medical Center and Emory University School of Medicine, Atlanta, GA, USA i Center for Cardiovascular Science and Medicine, UNC School of Medicine, Chapel Hill, NC, USA j Tokai University School of Medicine, Isehara, Japan k Medical Division, Maccabi Healthcare Services and Faculty of Health Sciences, Ben-Gurion University of Negev, Beer Sheva, Israel l INSERM, UMR915, Institut du Thorax, Université de Nantes, CHU, Nantes, France m VA Boston Healthcare System, Brigham and Women's Hospital, and Harvard Medical School, Boston, MA, USA n INSERM, U-698, Paris, France b
a r t i c l e
i n f o
Article history: Received 26 October 2011 Received in revised form 4 April 2012 Accepted 6 April 2012 Available online 4 May 2012 Keywords: Thiazolidinedione Diabetes Cardiovascular risk Mortality Registry
a b s t r a c t Objective: Assess the cardiovascular safety of Thiazolidinediones (TZD) in routine clinical practice. Background: TZD are insulin-sensitizing antidiabetic drugs commonly used in type 2 diabetes, but their cardiovascular safety has been questioned. We examined the association between TZD use and major cardiovascular outcomes. Methods: We examined 2-year mortality, non-fatal myocardial infarction (MI), and congestive heart failure (CHF) rates among outpatients with high cardiovascular risk and diabetes according to TZD use in the REACH Registry. Multivariable adjustment and propensity scores were used in the analyses. Results: A total of 4997 out of 28,332 patients took TZDs at baseline. During follow-up, 1532 patients died. The mortality rates (95% confidence interval [CI]) were 6.5% (5.5–7.6) with TZD and 7.2% (6.33–8.06) without; adjusted hazard ratio (HR) was 1.06 (0.89–1.26, P=0.54). The lack of association with mortality was consistent across subgroups regardless of history of atherothrombosis or CHF. Rates of non-fatal MI (HR 1.10, 95% CI 0.83–1.45, P=0.50) and non-fatal CHF (HR 0.90, CI 0.75–1.09, P=0.27) were similar in users and non-users. TZD use was associated with an increased risk of CHF in patients aged >80 years (HR 1.59, CI 1.06–2.40, P=0.03). Conclusions: Use of TZD was not associated with increased incidence of major cardiovascular events in patients with diabetes from this large registry. Older patients experienced an increased risk of CHF over the study interval. Limitations of this study include its observational design, and thus unmeasured confounders cannot be excluded. © 2012 Elsevier Ireland Ltd. All rights reserved.
1. Introduction Abbreviations: BMI, Body Mass Index; CHF, Congestive Heart Failure; CI, Confidence Interval; eGFR, estimated Glomerular Filtration Rate; HR, Hazard Ratio; MI, Myocardial infarction; RCT, Randomized Controlled Trial; ROC, Receiver Operating Characteristic; TZD, Thiazolidinedione. ⁎ Corresponding author at: Department of Diabetology, Endocrinology and Nutrition, Bichat Hospital, 46 rue Henri Huchard, 75018 Paris, France. Tel.: + 33 140257301; fax: + 33 140258842. E-mail address: [email protected] (R. Roussel). 1 For the REACH Investigators. 0167-5273/$ – see front matter © 2012 Elsevier Ireland Ltd. All rights reserved. doi:10.1016/j.ijcard.2012.04.019
Type 2 diabetes is a chronic condition associated with a high cardiovascular burden. Recent clinical trials suggest that an intensive glucose-control strategy may decrease the risk of myocardial infarction (MI), but it does not reduce mortality [1]. Conventional anti-diabetic therapy includes lifestyle modification, metformin, sulfonylurea, and insulin. The thiazolidinediones (TZD) are ligands of Peroxisome Proliferator Activated Receptor-gamma, a transcription factor expressed
R. Roussel et al. / International Journal of Cardiology 167 (2013) 1380–1384
in the adipose tissue, but also in endothelial cells. The net effect on metabolism is sensitization to insulin and glucose lowering. The cardiovascular safety and efficacy of these drugs were tested in several trials; pioglitazone was tested as secondary prevention via the PROActive study (PROspective pioglitAzone Clinical Trial in macroVascular Events) and rosiglitazone as primary prevention via the RECORD (Rosiglitazone Evaluated for Cardiovascular outcomes in ORal combination therapy for type 2 diabetes study) trial [2,3]. These two trials did not show a clear connection between TZD and cardiovascular risk, but a meta-analysis has first raised concerns regarding the cardiovascular safety of rosiglitazone, particularly with respect to the risk of MI [4]. However, none of these studies was sufficiently powered for analyzing effects on mortality because the numbers of deaths were small; 363 of 5238 (6.9%) participants in the PROActive study and 293 of 4447 (6.6%) participants in the RECORD trial. Observational studies can complement randomized trials if they are dedicated to specific aims, are of sufficient size, and have a design allowing prospective and extensive collection of relevant characteristics of the participants in order to minimize potential confounding factors. We compared mortality, non-fatal MI, and congestive heart failure (CHF) rates in patients with diabetes according to the use of TZD in the REACH (REduction of Atherothrombosis for Continued Health) Registry, an international prospective cohort of patients with either established atherosclerotic arterial disease or at risk for atherothrombosis. 2. Methods The study design, as well as the baseline description of the REACH Registry has been published previously [5–7]. Consecutive outpatients aged ≥45 years, with established coronary artery disease, cardiovascular disease, or peripheral arterial disease, or patients with ≥3 atherothrombotic risk factors were enrolled by 5587 physician practices in 44 countries between December 2003 and December 2004. In each country, the protocol was submitted to the institutional review boards according to local requirements, and signed informed consent was obtained for all patients. A standardized international case report form, completed at each study visit, was used and data were centrally collected. Diabetes was defined by anti-diabetic drug use by a patient. Overweight was defined by a body mass index (BMI) of 25–29, and obesity at BMI of ≥ 30. The estimated glomerular filtration rate was calculated according to the Modification of Diet in Renal Disease formula; stages of kidney function were defined according to the Kidney Disease Outcomes Quality Initiative guidelines (http:// www.kidney.org/professionals/KDOQI/guideline_diabetes/guide1.htm). As prior investigations have shown race-related differences in the medical treatment and outcomes of high risk patients, race was self-reported and in instances of mixed-racial origin, patients were asked to choose the race that had the strongest personal influence [6,7]. In some countries, local rules did not allow the recording of race and data were considered missing. Data were collected from participating physicians regarding patients' clinical outcomes, vascular procedures, employment status, weight and current smoking status, as well as any medications used, at each follow-up visit until 24 months after enrolment, and dates of the outcomes were collected if appropriate; events were not adjudicated. To ensure data quality, in each country 10% of all sites (that is, physicians) that enrolled ≥1 patient were chosen randomly 6% of the time, and an additional 4% were chosen due to the number of queries and missing data from the site to undergo onsite quality control by completing a site visit. For each of the sites undergoing monitoring, 100% of case report forms for patients enrolled at that site were monitored for source documentation and accuracy. The present analysis focuses on 2-year rates for mortality, non-fatal MI, and nonfatal CHF (i.e. hospitalization for CHF) in the diabetic subset of the REACH Registry. Patients with only baseline data were not included in the analysis. To take into account confounding factors we used the propensity score method [8]. This score represents the probability of receiving TZD given the characteristics of an individual. The list of co-variables was built in a two-step process. Bivariate analyses were first conducted to determine the variables associated with TZD prescription. In this preliminary selection, the P value limit was arbitrarily set at 0.20. These variables were then introduced in order to build a multivariable logistic regression model. To ensure the robustness of the score, the variables with >5% missing data were excluded. Finally, the propensity score was calculated for every patient using this model with the individual data of the patient. The quality of the model was assessed using global evaluation (Wald test) and by calculating the area under the corresponding receiver-operating characteristic (ROC) curve. To validate the score, a minimal value of 0.7 was expected. Patient characteristics were presented as mean ± standard deviation, and compared using Student's t tests or chi-squared tests. Hazard ratios (HRs) for death, nonfatal MI, and non-fatal CHF were calculated using a Cox proportional-hazard model, involving survival time in any individual patient, with TZD use and propensity scores as co-variables, as well as other specified factors. Survival time was calculated according to the date of the outcome as collected in the registry. For patients who remained
1381
free of the outcome, data were censored at the time of the last visit with available information. Event rates are presented based on total sample sizes. The homogeneity of treatment effects across subgroups was tested by adding interaction terms to the relevant models. A P value of b 0.05 was considered significant. All analyses were performed with the use of the SAS software, version 9.1.
3. Results 3.1. Patient characteristics Of the 65,441 patients enrolled in the REACH Registry and with data during follow-up, 28,332 patients had type 2 diabetes and available data on TZD use; they represent our study population. The Table 1 Baseline characteristics of the study population by TZD use. Characteristics
TZD use Yes (n = 4997) No (n = 23 335) p-value*
Age (years) Male sex Region of enrolment North America Latin America Western Europe Eastern Europe Middle East Asia (excluding Japan) Australia Japan Racial origin Caucasian Hispanic East Asian South Asian Other Asian Black Other Clinical and biological variables Waist (cm) BMI (kg/m2) Overweight (BMI 25–29) Obese (BMI ≥30) Smoking Former smoker Current smoker Serum creatinine (mg/L) Fasting blood glucose (mg/dL) Fasting total cholesterol (mg/dL) Fasting triglycerides (mg/dL) Hypertension Systolic blood pressure (mm Hg) Diastolic blood pressure (mm Hg) Prior arterial disease None One bed Two beds Three beds Prior CHF Baseline therapies ≥1 Antiplatelet agent Acetylsalicylic acid ≥1 Lipid-lowering agent Statins Angiotensin II receptor blockers Angiotensin-converting enzyme inhibitors Beta-blockers Calcium channel blockers Diuretics Nitrates Other antihypertensive agents Sulfonylurea Metformin Insulin Other anti-diabetic agent
b 0.0001 0.99 b 0.0001
67.1 ± 9.6 3034 (60.7)
68.6 ± 9.6 14 165 (60.7)
4063 (81.3) 61 (1.2) 338 (6.8) 27 (0.5) 26 (0.5) 309 (6.2) 19 (0.4) 154 (3.1)
8945 (38.3) 756 (3.2) 6300 (27.0) 1515 (6.5) 407 (1.7) 2401 (10.3) 844 (3.6) 2167 (9.3)
3255 (66.9) 398 (8.2) 403 (8.3) 60 (1.2) 189 (3.9) 500 (10.3) 60 (1.2)
13 001 (60.7) 1222 (5.7) 3480 (16.3) 352 (1.6) 1439 (6.7) 1184 (5.5) 735 (3.4)
104.9 ± 18.7 31.6 ± 6.7 1561 (31.8) 2642 (53.8)
100.1 ± 16.3 29.0 ± 5.9 8704 (37.9) 8435 (36.7)
1865 (38.5) 691 (14.3) 1.15 ± 0.73 145.1 ± 55.6 185.4 ± 48.2 180.3 ± 113.7 4448 (89.0) 135.2 ± 18.3 75.8 ± 10.9
9028 (40.0) 3072 (13.6) 1.13 ± 0.73 146.0 ± 53.1 191.9 ± 48.6 173.4 ± 105.2 20 360 (87.3) 139.9 ± 19.5 78.4 ± 11.4
2042 (40.9) 2314 (46.3) 561 (11.2) 80 (1.6) 704 (14.3)
6686 (28.7) 12 834 (55.0) 3374 (14.5) 441 (1.9) 3845 (16.8)
b 0.0001
3718 (74.4) 3367 (67.4) 4310 (86.3) 3981 (79.7) 1628 (32.9) 2575 (52.0)
17 196 (73.7) 14 723 (63.2) 17 663 (75.7) 16 047 (68.8) 6158 (26.6) 11 486 (49.5)
0.32 b 0.0001 b 0.0001 b 0.0001 b 0.0001 0.0015
2115 (42.6) 1654 (33.4) 2465 (49.6) 792 (16.3) 541 (11.0) 2426 (49.3) 2374 (48.2) 1027 (20.8) 417 (8.9)
10 363 (44.6) 8642 (37.2) 10 731 (46.1) 5537 (24.1) 2651 (11.5) 10 198 (43.8) 9399 (40.4) 6546 (28.1) 2681 (11.6)
0.0078 b 0.0001 b 0.0001 b 0.0001 0.35 b 0.0001 b 0.001 b 0.0001 b 0.0001
b 0.0001
Data are n (%) or mean ± standard deviation.
b 0.0001 b 0.0001 b 0.0001 b 0.0001 0.13
0.07 0.31 b 0.0001 0.0002 0.0006 b 0.0001 b 0.0001 b 0.0001
1382
R. Roussel et al. / International Journal of Cardiology 167 (2013) 1380–1384
characteristics of this population at baseline, according to TZD use, are presented in Table 1. Patients prescribed TZD differed strongly from TZD non-users regarding baseline characteristics, risk factors, and use of evidence-based therapies for secondary prevention; they were younger, more frequently obese, Caucasian, and living in North America. Based on their fasting glycemia, their glycemic control was similar. Regarding classical risk factors, TZD users had a slightly lower fasting total cholesterol concentration, higher triglyceride concentration, and blood pressure levels were lower. Use of TZD was noticeably lower in patients with a previous history of atherothrombotic events. Also, TZD users were more frequently prescribed evidence-based therapies for secondary prevention, such as antiplatelet agents, statins, and renin-angiotensin system blockers; of note, the use of diuretics was higher in TZD users (Table 1). 3.2. Propensity score The list of co-variables used to calculate the propensity score included age, geographic region of enrolment, height, BMI, smoking status, atrial fibrillation/flutter, use of lipid-lowering agents, statins, angiotensin II receptor blockers, angiotensin-converting enzyme inhibitors, beta-blockers, calcium channel blockers, nitrates, peripheral arterial claudication medications, insulin, and use of other antidiabetic agents. The propensity score reached the quality requirements because the likelihood associated with the model was strong (Wald test, P b 0.0001) and the area under the ROC curve (0.779) exceeded the 0.7 threshold (Supplementary Fig. 1). 3.3. TZD use and outcomes At the 2-year follow-up visit, 239 (event rate 6.5%, 95% confidence interval [CI] 5.5–7.6) and 1293 (7.2%, 95% CI 6.3–8.1) deaths, 103 (event rate 2.7%, 95% CI 2.0–3.3) and 480 (2.6%, 95% CI 2.1–3.1) non-fatal MI events, and 279 (event rate 6.3%, 95% CI 5.5–7.1) and 1505 (6.8%, 95% CI 6.3–7.3) non-fatal CHF events were recorded, for TZD users and non-users, respectively. No association was found between TZD use and any of these outcomes; adjusted HRs were 1.06 (0.89–1.26, P = 0.54), 1.10 (0.83–1.45, P = 0.50), and 0.90 (0.75–1.09, P = 0.27), for mortality, non-fatal MI, and non-fatal CHF, respectively (Fig. 1). Among patients who suffered a non-fatal MI, 75 subsequently died later during follow-up in similar proportions according to TZD use at baseline; 16 of 103 (15.5%) and 59 out of 480 (12.3%) in TZD users and non-users, respectively (P = 0.37). All adjusted analyses yielded consistent results—there was no difference
between death rates, non-fatal MI rates, and CHF rates, regardless of whether the rates were adjusted for age and sex, or for age, sex and propensity to receive TZDs, or all of these and differences in baseline characteristics. The HRs associated with TZD use for death, non-fatal MI, and non-fatal CHF according to relevant subgroups of patients are presented in Supplementary Fig. 2. There were no inconsistencies among subgroups for association of TZD with death or non-fatal MI (Supplementary Fig. 2). However, use of TZD interacted significantly (P for interaction = 0.002) with classes of age for risk of non-fatal CHF. A higher risk of CHF on TZD was observed among patients aged >80 years (HR 1.59, 95% CI 1.06–2.40). Patients with no established cardiovascular disease, but only risk factors had similar outcomes compared with those with established atherothrombosis in 1, 2, or 3 arterial beds (Supplementary Fig. 2). As the use of TZD was mostly prevalent in North America, we performed a sensitivity analysis restricted to US patients. The results were similar to those found for the overall population and showed a neutral association between TZD and death (HR 0.96, 95% CI 0.66–1.39, P = 0.82), non-fatal MI (HR 0.90, 95% CI 0.61–1.33, P = 0.59), and non-fatal CHF (HR 1.07, 95% CI 0.80–1.45, P = 0.64). 4. Discussion Diabetes very often coexists with other risk factors for atherosclerosis or even a history of atherosclerotic events, therefore clinicians have to take this background into account when prescribing antidiabetic drugs. In this large, international, prospective observational study of diabetic outpatients with high cardiovascular risk, the use of TZD was not associated with an increased risk for death, nonfatal MI, or non-fatal CHF, although some increased risk for CHF was noted in the subgroup of patients aged >80 years. Several studies have recently reported the cardiovascular outcomes of patients treated with TZD in usual care settings, although the present work focused on high cardiovascular risk profile, a unique characteristic [9–14]. Patients of the PROactive trial testing pioglitazone had to have evidence of extensive macrovascular disease before recruitment and were therefore at high risk [2]. Despite this, 6.9% of the participants died during an average time of observation of 34.5 months, slightly less than the mean mortality rate (7.1%) during the shorter 2-year follow-up of our registry. This may be explained by the mean age of participants, 68.3 ± 9.6 years in our study, and 61.7 ± 7.8 years in the PROactive trial. The RECORD open-label trial included a minority of patients with established atherothrombotic disease, and the proportion of death was 6.6%, but with a much longer mean observation time of 5.5 years [3]. Both studies were reassuring on
Fig. 1. Association of TZD use with mortality, non-fatal MI and CHF. Adjusted HRs (on all factors significantly associated in univariate, as well as propensity score) for mortality, nonfatal Myocardial Infarction (MI), and congestive heart failure (CHF) associated with TZD use as recorded at baseline, in the whole group of diabetic patients in the REACH Registry.
R. Roussel et al. / International Journal of Cardiology 167 (2013) 1380–1384
cardiovascular outcomes and mortality associated with TZD, except for heart failure. The PROactive trial suggested a benefit of pioglitazone on a secondary composite of death, non fatal MI and stroke [2]. Taken as a whole, our observational study extends the cardiovascular safety profile of TZD regarding death and ischemic events in an older and at very high risk population. Several studies have recently reported the cardiovascular outcomes of patients treated with TZD in usual care settings, although the present work focused on high cardiovascular risk profile, a unique characteristic [9–14]. There are distinct differences between these studies and the present analysis, such as the present study was not limited to CHF patients (they were a minority, 16%, in the trial) or to elderly patients. Also, in the Aguilar et al. study the control group included only patients receiving noninsulin-sensitizing drugs, by design excluding patients treated with biguanides [11]. However, biguanides are possibly associated with favorable effects on the myocardium [15]. These studies concur that in stable outpatients with high risk, TZD use was safe regarding risk of CHF and mortality. The REACH Registry is prospective by design, and 1532 deaths and 583 non-fatal MI events (approximately 5 times more than in the RECORD trial) were recorded. However, the design of the registry required clinicians to provide information on class of drugs used, but not on the molecules. Thus, it is impossible to separate the effects of the two marketed TZD, pioglitazone and rosiglitazone, which is a clear limitation in our study. Rosiglitazone has been suspected of increasing the risk for MI and pioglitazone may be protective or neutral [16]. At the time of enrolment in the registry, rosiglitazone and pioglitazone shared equitably the market in the US and the world [17]. Thus, we cannot exclude from our results that an adverse effect of one drug may be more or less compensated by a positive effect of the other on clinical events, particularly, but not only, on MI. However, the issue of cardiovascular safety of rosiglitazone compared to pioglitazone is still debated, as no large randomized controlled face to face trial was published; results of retrospective cohort studies are controversial, suggesting either similar MI, CHF or all-cause death risk for patients exposed to rosiglitazone and pioglitazone [14,18], or an inferiority of rosiglitazone [19–22]. An important point to acknowledge in our study is the lack of CHF events associated with TZD use. Only elderly patients (aged >80 years) were at higher risk for CHF using TZD compared with non-users. This lack of deleterious impact of TZD on heart failure might seem unexpected. Increased risk for CHF, presumably through salt retention, is a known risk when prescribing a TZD [23]. This has been documented in randomized controlled trials (RCTs), comparing TZD with placebo or hypoglycemic drugs [24,25]. Such a deleterious effect was not shown in our registry, as in others [11,12]. Of note, no adverse effects of rosiglitazone on brain natriuretic peptide levels were seen in a RCT of patients with coronary artery disease but without clinical heart failure at baseline [25]. Major design differences may explain this apparent discrepancy. First, in RCTs the exposure to the drug begins with the study, which is not the case in a registry. We do not have information on the duration of exposure to TZD before the enrolment of participants, but as the registry began in 2004, about 5 years after the marketing of TZD in the US and many other countries, it is expected that many patients were already being treated with TZD for years. Thus, if patients were to develop heart failure or even retention edema due to TZD, they were likely to have already done so prior to enrolment, after clinicians had taken the appropriate measures. Finally, RCTs are double-blinded, but clinicians who prescribed TZD to participants to the REACH Registry were likely to be fully aware of any side effects of the medication and may have adjusted medical management to minimize them. Particularly with regard to CHF, they may have prescribed diuretics more frequently; we did observe such an increased rate of use of diuretics among patients receiving TZD (Table 1). Interestingly, TZD users were also more prone to receive other protective drugs
1383
such as renin-angiotensin system blockers or statins. This is a common feature of registries analyzed according to TZD use [10,11,13,14] and also of the randomized, but open-label, trial RECORD where statins and diuretics were prescribed 20% and 25%, respectively, more frequently at the last visit in the rosiglitazone arm [3]. Studies describing prescribing patterns (drugs but also lifestyle) before and after the initiation of TZD would be informative. Overall, TZD users were younger, healthier, and received more aggressive treatment with life saving therapy. This lack of balance between groups according to TZD use is a clear limitation of the study, intrinsic to its observational design. Despite this, careful adjustment on potential confounders, using a robust propensity score and multivariable analysis would tend to minimize falsely reassuring results, although unmeasured confounders remain distinctly possible. As glycemic control might be one of the confounders, we included fasting glycemia in the multivariate analysis, as a surrogate of HbA1c, not available in the study, since both parameters are known for a long time as being tightly correlated [26]. No direct information was available regarding diabetes duration. We hypothesize that mean duration was quite long, although heterogeneous. For example, the mean duration of diabetes was 10.4 ± 8.7 years in the BARI 2D trial in patients with type 2 diabetes and coronary artery disease [27]. We may consider requirement of insulin as a surrogate for a long history of diabetes; our results did not differ according to its use. In summary, in this international registry of high cardiovascular risk diabetic patients, TZD use was safe as it was not associated with increased rates of major cardiovascular events. These findings from a large international registry, reflective of routine practice, suggest that when clinicians decided to prescribe TZDs they coped successfully with its potential side effects and contraindications in order to optimize the risk–benefit balance. Funding This work was supported by sanofi-aventis, Bristol-Myers Squibb, and the Waksman Foundation (Tokyo, Japan). All analyses from the REACH Registry are prepared by independent authors who are not governed by the funding sponsors and are prioritized and reviewed by an academic publications committee before submission. The statistical analyses were conducted solely by an academic team (BP, PR). The funding sponsors have the opportunity to review manuscript submissions but do not have authority to change any aspect of a manuscript. The REACH Registry is sponsored by sanofi-aventis, Bristol-Myers Squibb, and the Waksman Foundation (Tokyo, Japan), who assisted with the design and conduct of the study and data collection. Conflict of interest Dr Roussel has received research grants, honoraria, or consulting fees from sanofi-aventis, MSD Chibret, Servier, Eli Lilly, Novo Nordisk, Medtronic, Janssen-Cilag, and Lifescan. Dr Hadjadj has received research grants, honoraria, or consulting fees from AstraZeneca, sanofi-aventis, MSD Chibret, Servier, Roche, Bristol-Myers Scribb. Prof Wilson has received research grants from sanofi-aventis within the past 3 years. Prof Goto has received honoraria and consulting fees from Eisai, sanofiaventis, Daiichi-Sankyo, GlaxoSmithKline, Bristol-Myers Squibb, Otsuka, Bayer, Schering-Plough, Takeda, Astellas, AstraZeneca, Novartis and Kowa. Prof Goto also received research grants from Pfizer, Ono, Eisai, Otsuka, Daiichi-Sankyo, sanofi-aventis, Takeda and Astellas within the last 3 years. Dr Tubach has received research grants from Astra Zeneca, Abbott, sanofi-aventis, Schering Plough, Servier, and Wyeth. Prof Marre has received honoraria as an adviser and for lectures: Novo-Nordisk, sanofi-aventis, Merck, Servier, and Eli Lilly. Prof Krempf has received honoraria for advisory board attendance and consulting fees from AstraZeneca, Bristol-Myers Squibb, Roche, Pfizer, Merck, sanofi-aventis,
1384
R. Roussel et al. / International Journal of Cardiology 167 (2013) 1380–1384
Pierre Fabre; payment for speakers' bureau from Bristol-Myers Squibb, Pfizer, sanofi-aventis, Merck, Schering-Plough, Merck Lipha. Dr Bhatt receives research grants from Amarin, AstraZeneca (for a study of saxagliptin), Bristol-Myers Squibb, Eisai, Ethicon, Medtronic, sanofi-aventis, and The Medicines Company; is a research collaborator with Takeda (not pertaining to TZDs); and had previously worked on a study of rosiglitazone funded by GlaxoSmithKline (the PPAR study, reference #25 of this paper). Prof Steg has received research grant from sanofi-aventis (1999–2008); is on the speaker's bureau for Boehringer Ingelheim, Bristol-Myers Squibb, GlaxoSmithKline, Menarini, Medtronic, Nycomed, Pierre Fabre, sanofi-aventis, Servier, and The Medicines Company; is on the consulting/advisory boards for Astellas, AstraZeneca, Bayer, Boehringer Ingelheim, Bristol-Myers Squibb, Daiichi-Sankyo, Endotis, GlaxoSmithKline, Medtronic, MSD, Nycomed, sanofi-aventis, Servier, and The Medicines Company; and is a stockholder for Aterovax. Supplementary data related to this article can be found online at doi:10.1016/j.ijcard.2012.04.019. Author contributions Dr Roussel had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Roussel, Hadjadj, Krempf, Bhatt, Steg. Acquisition of data: Steg, Goto. Analysis and interpretation of data: Roussel, Hadjadj, Marre, Steg, Bhatt. Drafting of the manuscript: Roussel, Hadjadj, Steg, Bhatt. Critical revision of the manuscript for important intellectual content: Roussel, Hadjadj, Goto, Wilson, Porath, Smith Jr, Marre, Tubach, Krempf, Steg, Bhatt. Statistical analysis: Roussel, Pasquet, Tubach, Steg. Obtained funding: Steg, Goto. Administrative, technical, or material support: Steg, Goto. Study supervision: Steg, Bhatt. Acknowledgments The REACH Registry is endorsed by the World Heart Federation. A complete list of REACH investigators is published in JAMA 2006; 295:180–189. The REACH Registry enforces a no ghost-writing policy. This manuscript was written and edited by the authors, who take full responsibility for its content. The drafts were written by Ronan Roussel. The authors of this manuscript have certified that they comply with the Principles of Ethical Publishing in the International Journal of Cardiology. REACH Registry Executive Committee Deepak L. Bhatt, MD, MPH, VA Boston Healthcare System and Brigham and Women's Hospital, Boston, MA, USA (Chair); Ph. Gabriel Steg, MD, INSERM U-698, Université Paris 7, AP-HP, Paris, France (Chair); E. Magnus Ohman, MD, Duke University Medical Center, Durham, NC, USA; Joachim Röther, MD, Johannes Wesling Klinikum Minden, Minden, Germany; Peter W.F. Wilson, MD, Emory University School of Medicine, Atlanta, GA, USA. REACH Registry Global Publication Committee Mark J. Alberts, MD, Northwestern University Medical School, Chicago, IL, USA; Deepak L. Bhatt, MD, MPH, VA Boston Healthcare System and Brigham and Women's Hospital, Boston, MA, USA (Chair); Ralph D'Agostino, PhD, Boston University, Boston, MA, USA; Kim Eagle, MD, University of Michigan, Ann Arbor, MI, USA; Shinya Goto, MD, PhD, Tokai University School of Medicine, Isehara, Kanagawa, Japan; Alan T. Hirsch, MD, Minneapolis Heart Institute Foundation and University of Minnesota School of Public Health, Minneapolis, MN, USA; Chiau-Suong Liau, MD, PhD, Taiwan University Hospital and College of Medicine, Taipei, Taiwan; Jean-Louis Mas, MD, Centre Raymond Garcin, Paris, France; E. Magnus Ohman, MD, Duke University Medical Center, Durham, NC, USA; Joachim Röther, MD, Klinikum Minden, Minden, Germany; Sidney C. Smith, Jr, MD, University of North Carolina at Chapel Hill, Chapel Hill, NC, USA; Ph. Gabriel Steg, MD, INSERM
U-698, Université Paris 7, AP-HP, Paris, France (Chair); Peter W.F. Wilson, MD, Emory University School of Medicine, Atlanta, GA, USA. References [1] Control GroupTurnbull FM, Abraira C, et al. Intensive glucose control and macrovascular outcomes in type 2 diabetes. Diabetologia 2009;52:2288–98. [2] Dormandy JA, Charbonnel B, Eckland DJ, et al. Secondary prevention of macrovascular events in patients with type 2 diabetes in the PROactive Study (PROspective pioglitAzone Clinical Trial In macroVascular Events): a randomised controlled trial. Lancet 2005;366:1279–89. [3] Home PD, Pocock SJ, Beck-Nielsen H, et al. RECORD Study Team. Rosiglitazone evaluated for cardiovascular outcomes in oral agent combination therapy for type 2 diabetes (RECORD): a multicentre, randomised, open-label trial. Lancet 2009;373:2125–35. [4] Nissen SE, Wolski K. Effect of rosiglitazone on the risk of myocardial infarction and death from cardiovascular causes. N Engl J Med 2007;356:2457–71. [5] Ohman EM, Bhatt DL, Steg PG, et al. The REduction of Atherothrombosis for Continued Health (REACH) Registry: an international, prospective, observational investigation in subjects at risk for atherothrombotic events-study design. Am Heart J 2006;151(786):e1–10. [6] Bhatt DL, Steg PG, Ohman EM, et al. International prevalence, recognition, and treatment of cardiovascular risk factors in outpatients with atherothrombosis. JAMA 2006;295:180–9. [7] Steg PG, Bhatt DL, Wilson PW, et al. One-year cardiovascular event rates in outpatients with atherothrombosis. JAMA 2007;297:1197–206. [8] Rubin DB. Estimating causal effects from large data sets using propensity scores. Ann Intern Med 1997;127:757–63. [9] Delea TE, Edelsberg JS, Hagiwara M, Oster G, Phillips LS. Use of thiazolidinediones and risk of heart failure in people with type 2 diabetes: a retrospective cohort study. Diabetes Care 2003;26:2983–9. [10] Lipscombe LL, Gomes T, Lévesque LE, Hux JE, Juurlink DN, Alter DA. Thiazolidinediones and cardiovascular outcomes in older patients with diabetes. JAMA 2007;298:2634–43. [11] Aguilar D, Bozkurt B, Pritchett A, Petersen NJ, Deswal A. The impact of thiazolidinedione use on outcomes in ambulatory patients with diabetes mellitus and heart failure. J Am Coll Cardiol 2007;50:32–6. [12] Pantalone KM, Kattan MW, Yu C, et al. The risk of developing coronary artery disease or congestive heart failure, and overall mortality, in type 2 diabetic patients receiving rosiglitazone, pioglitazone, metformin, or sulfonylureas: a retrospective analysis. Acta Diabetol 2009;46:145–54. [13] Masoudi FA, Inzucchi SE, Wang Y, Havranek ER, Foody JM, Krumholz HM. Thiazolidinediones, metformin, and outcomes in older patients with diabetes and heart failure: an observational study. Circulation 2005;111:583–90. [14] Habib ZA, Tzogias L, Havstad SL, et al. Relationship between thiazolidinedione use and cardiovascular outcomes and all-cause mortality among patients with diabetes: a time-updated propensity analysis. Pharmacoepidemiol Drug Saf 2009;18:437–47. [15] Inzucchi SE, Masoudi FA, McGuire DK. Metformin in heart failure. Diabetes Care 2007;30:e129. [16] Loke YK, Kwok CS, Singh S. Comparative cardiovascular effects of thiazolidinediones: systematic review and meta-analysis of observational studies. BMJ 2011;342:d1309, http://dx.doi.org/10.1136/bmj.d1309. [17] Starner CI, Schafer JA, Heaton AH, Gleason PP. Rosiglitazone and pioglitazone utilization from January 2007 through May 2008 associated with five risk-warning events. J Manag Care Pharm 2008;14:523–31. [18] Wertz DA, Chang CL, Sarawate CA, Willey VJ, Cziraky MJ, Bohn RL. Risk of Cardiovascular Events and All-Cause Mortality in Patients Treated With Thiazolidinediones in a Managed-Care Population. Circ Cardiovasc Qual Outcomes 2010;3:538–45. [19] Gerrits C, Bhattacharya M, Manthena S, Baran R, Perez A, Kupfer S. A comparison of pioglitazone and rosiglitazone for hospitalization for acute myocardial infarction in type 2 diabetes. Pharmacoepidemiol Drug Saf 2007;16:1065–71. [20] Graham DJ, Ouellet-Hellstrom R, MaCurdy TE, et al. Risk of acute myocardial infarction, stroke, heart failure, and death in elderly Medicare patients treated with rosiglitazone or pioglitazone. JAMA Jul 28 2010;304(4):411–8. [21] Juurlink DN, Gomes T, Lipscombe LL, Austin PC, Hux JE, Mamdani MM. Adverse cardiovascular events during treatment with pioglitazone and rosiglitazone: population based cohort study. BMJ 2009;339:b2942. [22] Winkelmayer WC, Setoguchi S, Levin R, Solomon DH. Comparison of cardiovascular outcomes in elderly patients with diabetes who initiated rosiglitazone vs pioglitazone therapy. Arch Intern Med 2008;168:2368–75. [23] Lago RM, Singh PP, Nesto RW. Congestive heart failure and cardiovascular death in patients with prediabetes and type 2 diabetes given thiazolidinediones: a meta-analysis of randomised clinical trials. Lancet 2007;370:1129–36. [24] Komajda M, McMurray JJ, Beck-Nielsen H, et al. Heart failure events with rosiglitazone in type 2 diabetes: data from the RECORD clinical trial. Eur Heart J Apr 2010;31(7):824–31. [25] Bhatt DL, Chew DP, Grines C, et al. Peroxisome proliferator-activated receptor gamma agonists for the Prevention of Adverse events following percutaneous coronary Revascularization–results of the PPAR study. Am Heart J 2007;154:137–43. [26] Gonen B, Rubenstein A, Rochman H, Tanega SP, Horwitz DL. Haemoglobin A1: An indicator of the metabolic control of diabetic patients. Lancet Oct 8 1977;2(8041):734–7. [27] BARI 2D Study GroupFrye RL, August P, et al. A randomized trial of therapies for type 2 diabetes and coronary artery disease. N Engl J Med 2009;360(24):2503–15.