ÌÇÐÄvlog

Key Points

QuestionÌý Do cardiac biomarkers provide incremental prognostic value in patients with type 2 diabetes?

FindingsÌý In this secondary analysis of a randomized clinical trial of 12 310 patients with overt cardiovascular disease or multiple risk factors, elevated levels of high-sensitivity troponin T, N-terminal pro–B-type natriuretic peptide, and high-sensitivity C-reactive protein significantly improved the discrimination and appropriate reclassification of risk, in particular for high-sensitivity troponin T and N-terminal pro–B-type natriuretic peptide.

MeaningÌý Biomarkers appropriately risk stratify patients with diabetes in terms of future cardiovascular events and challenge the traditional differentiation between primary and secondary prevention based simply on clinical history.

ImportanceÌý Cardiac biomarkers provide insights into pathophysiologic processes and offer an attractive strategy for the assessment of cardiovascular risk.

ObjectiveÌý To assess the incremental prognostic value of biomarkers that reflect different pathophysiologic processes in patients with type 2 diabetes.

Design, Setting, and ParticipantsÌý The Saxagliptin Assessment of Vascular Outcomes Recorded in Patients with Diabetes Mellitus (SAVOR)–Thrombolysis in Myocardial Infarction (TIMI) 53 is a randomized, double-blind, placebo-controlled clinical trial that evaluated the safety of saxagliptin vs placebo in 16 492 outpatients with type 2 diabetes with overt cardiovascular disease (CVD) or multiple risk factors. In this secondary analysis, widely used biomarkers were evaluated to ascertain whether they would provide incremental prognostic value in the risk stratification. Median follow-up was 2.1 years (interquartile range, 1.8-2.3 years). The study was performed from May 10, 2010, to June 15, 2013.

InterventionsÌý Randomization to saxagliptin vs placebo in addition to standard care.

Main Outcomes and MeasuresÌý Concentrations of high-sensitivity troponin T, N-terminal pro–B-type natriuretic peptide, and high-sensitivity C-reactive protein were analyzed continuously and by established cut points. Cardiovascular death, myocardial infarction, ischemic stroke, and hospitalization for heart failure (HF) were adjudicated by a blinded events committee.

ResultsÌý Of the 16 492 patients, 5455 (33.1%) were female and 11 037 (66.9%) were male. Mean (SD) age was 65.0 (8.5) years (range, 39-99 years). Baseline biomarkers were measured in 12 310 patients. Elevated levels of each biomarker were associated significantly with increased risk for all cardiovascular end points. When added to clinical variables, biomarkers significantly improved the discrimination and appropriate reclassification of risk. Elevated high-sensitivity troponin T was associated with an increased risk of cardiovascular death (adjusted hazard ratio [AHR], 3.07; 95% CI, 2.35-4.02; P < .001), myocardial infarction (AHR, 2.13; 95% CI, 1.69-2.67; P < .001), and hospitalization for HF (AHR, 3.85; 95% CI, 2.82-5.27; P < .001). Elevated N-terminal pro–B-type natriuretic peptide was also associated with an increased risk of cardiovascular death (AHR, 3.09; 95% CI, 2.46-3.89; P < .001), myocardial infarction (AHR, 1.95; 95% CI, 1.51-2.53; P < .001), and hospitalization for HF (AHR, 3.92; 95% CI, 3.11-4.92; P < .001). Elevated high-sensitivity C-reactive protein was more weakly associated with an increased risk of cardiovascular death (AHR, 1.49; 95% CI, 1.22-1.82; P < .001) and hospitalization for HF (AHR, 1.47; 95% CI, 1.20-1.81; P < .001). Consistent results were seen in patients with or without established CVD.

Conclusions and RelevanceÌý A substantial proportion of patients with stable type 2 diabetes with established CVD or multiple clinical risk factors have evidence of ongoing myocardial injury, hemodynamic stress, or systemic inflammation. Biomarker risk stratification thus challenges the traditional differentiation between primary and secondary prevention based simply on clinical history. Strategies to improve risk stratification in patients with type 2 diabetes, with or without CVD, should consider incorporation of biomarker data into standard risk algorithms.

Trial RegistrationÌý clinicaltrials.gov Identifier:

Type 2 diabetes accelerates atherogenesis, causes myocardial dysfunction, and significantly worsens outcomes in patients with and without established cardiovascular disease through a variety of mechanisms. As such, the presence of type 2 diabetes is commonly considered to be a risk factor for these adverse outcomes.^1-4 However, in patients with type 2 diabetes, assessing risk poses a particular challenge given the heterogeneity in the clinical and pathophysiologic characteristics in this condition and the overall higher rate of diabetes-related complications. This challenge is especially relevant when screening patients with type 2 diabetes without manifest cardiovascular disease, in whom the risk of ischemic complications varies widely based on age, duration of diabetes, and comorbidities. Because cardiac biomarkers provide insight into different pathophysiologic processes and improve risk stratification across the cardiovascular disease spectrum, we hypothesized that such biomarkers may aid in the risk stratification for the development of first or recurrent cardiovascular events in patients with type 2 diabetes.

The Saxagliptin Assessment of Vascular Outcomes Recorded in Patients with Diabetes Mellitus (SAVOR)–Thrombolysis in Myocardial Infarction (TIMI) 53 trial evaluated the cardiovascular efficacy and safety of saxagliptin, a selective dipeptidyl peptidase 4 inhibitor, in patients with type 2 diabetes with overt cardiovascular disease or at risk for cardiovascular events.⁵ During a median of 2.1 years of follow-up, saxagliptin reduced hemoglobin A_1c levels but did not alter the risk of the primary composite end point of cardiovascular death, myocardial infarction, or ischemic stroke or the secondary composite end point that added hospitalization for heart failure, unstable angina, and coronary revascularization to the primary end point, although there was a significant 27% increased relative risk of hospitalization for heart failure in patients randomized to receive saxagliptin.⁶

As part of a prespecified subgroup analysis, we evaluated biomarkers reflecting the pathophysiologic processes of myocardial injury with high-sensitivity troponin T (hsTnT), hemodynamic stress with N-terminal pro–B-type natriuretic peptide (NT-proBNP), and inflammation with high-sensitivity C-reactive protein (hsCRP) to ascertain whether these widely used biomarkers would provide incremental prognostic value in the risk stratification of patients with type 2 diabetes. We also wanted to determine whether there was any interaction among baseline levels of these 3 biomarkers and the effects of treatment with saxagliptin.

As previously described,⁵ SAVOR-TIMI 53 was a multicenter, randomized, double-blind, placebo-controlled clinical trial that randomized 16 492 patients with type 2 diabetes, hemoglobin A_1c levels between 6.5% and less than 12.0% within 6 months of randomization (to convert to proportion of total hemoglobin, multiply by 0.01), and a history of established cardiovascular disease or multiple clinical risk factors for vascular disease (dyslipidemia, hypertension, or smoking) to receive saxagliptin or matching placebo. The full eligibility criteria and analysis plan have been reported previously.^5,7 Written informed consent was obtained from all patients. All participating centers obtained approval from their local institutional review board. Biomarker analyses were performed with the approval of the Partners Human Research Committee.

The clinical end points in this analysis include (1) the primary composite end point of the trial (cardiovascular death, myocardial infarction, or ischemic stroke), (2) each individual component, and (3) hospitalizations for heart failure, a component of an expanded, prespecified secondary end point. A clinical events committee, unaware of the study group assignments, adjudicated all components of the primary and secondary composite efficacy end points,^5,7 using definitions based on draft guidelines for the standardization of end points in cardiovascular trials proposed by the US Food and Drug Administration.⁸

Concentrations of hsTnT, NT-proBNP, and hsCRP were measured in samples obtained at randomization in patients who consented to participate in the biomarker assessment. Serum was isolated and stored frozen in aliquots at −20° to −80°C at the enrolling site until shipped to the Biomarker Research/TIMI Clinical Trials Laboratory, Boston, Massachusetts, where they were maintained at −80°C or colder. Serum NT-proBNP concentrations were measured at the first thaw using a sandwich immunoassay (proBNP II; Roche Diagnostics). The analytic range extends from 5 to 35 000 pg/mL (to convert to nanograms per liter, multiply by 1). The reported within-run coefficient of variation was 4.2% at a level of 44 pg/mL and 2.7% at a level of 33 606 pg/mL. The cut points for elevated levels were defined as 450 pg/mL or higher for those younger than 50 years, 900 pg/mL or higher for those aged 50 to 75 years, and 1800 pg/mL or higher for those older than 75 years.⁹ Concentrations of hsTnT were measured with an electrochemiluminescent immunoassay assay (Roche Diagnostics). The lower limit of detection of the assay is 3 pg/mL (to convert to nanograms per liter, multiply by 1).¹⁰ Levels greater than 15.0 pg/mL for men and greater than 10 pg/mL for women were used as the diagnostic cut points, or upper reference limit (URL), representing the 99th percentile in healthy individuals with a coefficient of variation less than 10%.¹¹ Concentrations of hsCRP were measured with an enhanced immunoturbidimetric assay (Roche Diagnostics) with a lower level of detection of 0.15 mg/L and a functional sensitivity of 0.3 mg/L (to convert to nanomoles per liter, multiply by 9.524). The cut point for an elevated level was defined as greater than 3 mg/L based on guidelines for cardiovascular risk stratification.¹²

Categorical variables were compared using the χ² test and continuous variables with a 2-tailed, paired t test or Wilcoxon rank sum test, as appropriate. Events rates are presented as 2-year Kaplan-Meier estimates. Biomarkers were analyzed as continuous variables, categorized as sex-specific quartiles, and based on the prespecified cut points mentioned above. Multivariable clinical models that evaluated the association between biomarkers and clinical outcomes were stratified by established cardiovascular disease vs multiple risk factors and adjusted for the following clinical variables: age (continuous), sex, systolic blood pressure (continuous), history of heart failure, duration of diabetes, prior myocardial infarction, hypertension, hyperlipidemia, smoking, estimated glomerular filtration rate (continuous), and treatment allocation (saxagliptin vs placebo).

Estimates of the C statistic for the clinical model created from previously listed variables were calculated based on the Harrell method¹³ and then compared to the models after the addition of the different biomarkers. The discriminative value of the biomarkers was further examined with the method described by Pencina and colleagues^14,15 to determine the continuous (categoryless) net reclassification improvement (NRI) (the probability that patients are appropriately assigned to a higher or lower risk) and integrated discrimination improvement (IDI) (a method to quantify mean predicted probabilities of events and nonevents based on the addition of the new biomarkers to the model). The NRI values above 0.6 are considered strong, those around 0.4 as intermediate, and those less than 0.2 as weak.¹⁵ Comparisons between saxagliptin and placebo were examined using an unadjusted Cox proportional hazards regression model stratified by estimated glomerular filtration category and baseline cardiovascular risk group with assigned treatment as a model term. Statistical analyses were performed with SAS statistical software, version 9.4 (SAS Institute Inc).

Serum and plasma biomarkers were collected in 12 310 of the 16 492 SAVOR-TIMI 53 patients, 74.6% of the entire trial population, from May 10, 2010, to June 15, 2013, and the biomarkers for this study were analyzed from September 1 to October 31, 2013. Baseline characteristics of the patients with (n = 12 310) and without (n = 4182) biomarkers are presented in eTable 1 in the Supplement. Of the 16 492 patients, 5455 (33.1%) were female and 11 037 (66.9%) were male. Mean (SD) age was 65.0 (8.5) years (range, 39-99 years). The proportions of patients with established cardiovascular disease (9619 [78.1%]) and risk factors alone (2691 [21.9%]) in the biomarker assessment were similar to the overall trial. Patients in the biomarker assessment group were more likely to be white or Hispanic and weigh more than patients without baseline biomarkers.

The median (interquartile ranges) of biomarkers were 12.0 pg/mL (8.1-18.4 pg/mL) for hsTnT, 141 pg/mL (64-332 pg/mL) for NT-proBNP, and 2.4 mg/L (1.1-5.0 mg/L) for hsCRP. In the overall population, there was a linear increase in the rates of the primary composite end point and the individual end points of cardiovascular death, myocardial infarction, ischemic stroke, and hospitalization for heart failure (Figure 1 and Table 1). The association was most striking for hsTnT and NT-proBNP, with a weaker association with elevated levels of hsCRP. The association between biomarkers and outcomes was consistent in the 9619 patients with established cardiovascular disease and in the 2691 patients with multiple risk factors alone (eTable 2 in the Supplement).

Nearly all patients (99.7%) had a detectable baseline level of hsTnT (≥3 pg/mL). Overall, 5239 patients (43.0%) had a level above the 99th percentile URL (>15 pg/mL for men and >10 pg/mL for women); this proportion was higher in patients with established cardiovascular disease (4394 [46.2%]) than in patients with cardiovascular risk factors only (845 [31.6%]) (P < .001). Overall, a level of hsTnT greater than 99th percentile URL was associated with an increased risk of the primary end point (12.3% vs 4.1%; adjusted hazard ratio [AHR], 2.19; 95% CI, 1.87-2.56; P < .001), cardiovascular death (5.9% vs 1.2%; AHR, 3.07; 95% CI, 2.35-4.02; P < .001), myocardial infarction (5.7% vs 1.9%; AHR, 2.13; 95% CI, 1.69-2.67; P < .001), ischemic stroke (2.6% vs 1.2%; AHR, 1.64; 95% CI, 1.21-2.22; P = .001), or hospitalization for heart failure (6.2% vs 0.8%; AHR, 3.85; 95% CI, 2.82-5.27; P < .001) (Table 1). Of note, in patients with clinical risk factors only, an elevated level of hsTnT identified individuals with a higher risk of the primary end point, cardiovascular death, myocardial infarction, and hospitalization for heart failure, but not ischemic stroke (Figure 2 and eTable 2 in the Supplement). Conversely, in patients with clinical risk factors only and hsTnT levels in the less than the 99th percentile URL, the rate of each of these clinical end points was individually less than 0.5% per year. Patients with established cardiovascular disease but an hsTnT level in the less than 99th percentile URL had a risk lower than those patients with risk factors alone and elevated levels of hsTnT. A similar pattern of risk was seen when categorizing patients according to the presence or absence of a history of myocardial infarction (eFigure 1 in the Supplement).

A large proportion of patients also had biomarker levels above a priori cut points for NT-proBNP (≥450 pg/mL for those younger than 50 years, ≥900 pg/mL for those 50-75 years old, and ≥1800 pg/mL for those older than 75 years; 964 patients [7.8%]) and hsCRP (>3 mg/L; 5086 [41.3%]). A level above these established cut points for each biomarker was significantly associated with cardiovascular death, myocardial infarction, ischemic stroke, and hospitalization for heart failure (Table 1). The pattern of risk was similar when biomarkers were analyzed as quartiles (eFigure 2 and eTable 3 in the Supplement).

The addition of the biomarkers, as a continuous variable or by cut points, significantly improved metrics of discrimination (C statistic and IDI) and net reclassification (NRI) when compared with the multivariable clinical model alone (see the Methods section) (Table 2 and eTable 4 in the Supplement). The improvement with NT-proBNP and hsTnT was most striking with a continuous NRI and relative IDI of greater than 0.60 for cardiovascular death and hospitalization for heart failure compared with less than 0.20 with hsCRP. A similar pattern was seen in patients with established cardiovascular disease and patients with clinical risk factors only, although hsCRP did not improve the discrimination or reclassification in patients with risk factors only (eTable 5 in the Supplement).

When compared with patients without any elevated biomarkers, there was a stepwise increase in the risk of cardiovascular events in patients who had 1, 2, or 3 elevated biomarkers by cut point. In the 481 patients (3.9%) with 3 elevated biomarkers, the 2-year rates were 17.3% for cardiovascular death, 11.7% for myocardial infarction, 4.7% for ischemic stroke, and 21.7% for hospitalization for heart failure, whereas the corresponding rates in the patients without any elevated biomarkers (4266 patients [34.7%]) were 1.0%, 1.9%, 1.0%, and 0.7%, respectively (Table 3). This stepwise increase in risk was apparent in patients with established cardiovascular disease or multiple risk factors alone (eFigure 3 in the Supplement).

Compared with placebo, patients randomized to receive saxagliptin had a similar risk of the primary and secondary end points of the trial, including the individual components of cardiovascular death and myocardial infarction, regardless of baseline biomarker level, whether analyzed as continuous variables or by cut points. The previously reported overall increased risk of hospitalization for heart failure observed in patients randomized to receive saxagliptin was also consistent across different baseline biomarkers levels (eTable 6 in the Supplement).

In patients with type 2 diabetes and established cardiovascular disease or multiple cardiovascular risk factors, 3 cardiovascular biomarkers were associated with adverse cardiovascular outcomes, independent of baseline clinical features. They significantly improved the discrimination and reclassification of the risk for the primary end point (cardiovascular death, myocardial infarction, or ischemic stroke) and the individual end points, in addition to hospitalization for heart failure. Moreover, even small elevations of hsTnT and NT-proBNP carried significant prognostic implications in patients with or without manifest cardiovascular disease. Biomarker risk stratification thus challenges the traditional differentiation between primary and secondary prevention that is based simply on a history of a clinically recognized cardiovascular event. For example, the risk of cardiovascular death in a patient with clinical risk factors alone but with an elevated hsTnT level is higher (3.7%) than in a patient with established cardiovascular disease and a low hsTnT level (1.4%). Integration of these biomarkers into routine screening and risk stratification algorithms of patients with type 2 diabetes can more accurately assess risk and thereby target patients in whom modification or intensification of diagnostic and treatment strategies, as well as the frequency of follow-up, might reduce future complications.

Despite enrollment criteria that excluded patients with recent cardiovascular events, a large proportion of patients in this cohort had levels of hsTnT, NT-proBNP, and hsCRP that would be considered elevated. Nearly half of patients with established cardiovascular disease and one-quarter of patients with risk factors alone had hsTNT levels above the 99th percentile of a healthy population, a level that if dynamic, and in the appropriate clinical context, would be consistent with a myocardial infarction according to current consensus guidelines.¹⁶

Compared with other studies^17-19 in patients with stable type 2 diabetes, the proportion of patients in SAVOR-TIMI 53 with detectable levels of hsTnT above the 99th percentile URL was in general higher, likely because of a combination of factors, including the older age of our patients, their longer duration of diabetes, and the requirement that patients have established cardiovascular disease or documented cardiovascular risk factors for inclusion. The proportion of patients with established cardiovascular disease in SAVOR-TIMI 53 who had an elevated hsTnT level was similar to the Bypass Angioplasty Revascularization Investigation in Type 2 Diabetes (BARI 2D) cohort.²⁰

These findings have several implications. Even in asymptomatic patients with diabetes without any recent acute ischemic event, we observed evidence of myocardial injury as identified by elevated levels of hsTnT. In fact, patients without established cardiovascular disease or a history of myocardial infarction but with elevated levels of hsTnT are at equal or higher subsequent risk than patients with documented atherothrombotic disease and lower levels of hsTnT. This finding may be related to persistent myocardial injury from glycemic dysregulation, leading to endothelial and microvascular dysfunction, myocardial cell death, and subsequent fibrosis.²¹ Moreover, because heart failure is now recognized to be one of the first manifestations of cardiovascular disease in patients with type 2 diabetes,²² better screening tools, such as hsTnT or natriuretic peptides, may improve earlier identification of patients at risk. Current treatment recommendations do not incorporate these biomarkers.²³

The strong association between increased concentrations of hsTnT and the risk of a first or recurrent myocardial infarction in this large and diverse population of patients with diabetes and various levels of atherothrombotic risk is noteworthy and consistent with what has been described in patients with coronary artery disease in other cohorts of patients with type 2 diabetes.²⁰ Moreover, because a large proportion of asymptomatic patients with type 2 diabetes have circulating levels of hsTnT above the myocardial infarction cut point, the evaluation of patients with type 2 diabetes presenting with a suspected acute coronary syndrome must integrate the rise and fall of hsTnT levels rather than the absolute concentration into the diagnostic algorithm to avoid inappropriate diagnoses of acute coronary syndromes.¹⁶ Consistent with prior studies^24-27 of natriuretic peptides in patients with type 2 diabetes, elevated levels of NT-proBNP were strongly associated with the risk of cardiovascular death and hospitalization with heart failure.

Simultaneous assessment of all 3 biomarkers identified particularly low- and high-risk populations. In patients with 3 elevated biomarkers, the 2-year rates of cardiovascular death approached 20%, more than 10-fold the risk of a similarly sized cohort of patients who had no elevated levels. A multimarker approach, reported to be useful in acute coronary syndromes,^28,29 may also provide an approach to discriminate risk in patients with stable type 2 diabetes more fully.

To become fully integrated into clinical care, a biomarker must fulfill several criteria.^30,31 The assay should ideally be high throughput and demonstrate adequate precision and reproducibility. Clinically, the biomarker must provide incremental prognostic information to standard risk tools, in particular through improved discrimination and reclassification as tested with contemporary statistical methods.¹⁴ On the basis of the continuous NRI and relative IDI, NT-proBNP and hsTnT provide at least moderate improvement in risk stratification (relative IDI and NRI >0.4) for all cardiovascular end points, with a particularly strong improvement in reclassification with NT-proBNP and hsTnT for cardiovascular death (NRI and relative IDI >0.60 for hsTnT and >1.9 for NT-proBNP) and hospitalization for heart failure (NRI and relative IDI >0.60 for hsTnT and >1.0 for NTproBNP).¹⁵ The incremental improvement with hsCRP was less marked (relative IDI and NRI ≤0.15 for cardiovascular death), highlighting that a general marker of inflammation, such as hsCRP, may not provide similar discrimination of risk in patients with the ongoing low-level chronic inflammatory processes associated with diabetes.

Ideally, a biomarker should also provide actionable diagnostic and treatment implications. Of the cardiovascular biomarkers, only hsTnT in the setting of acute coronary syndromes has clearly met this last hurdle. In other scenarios, hsTnT and NT-proBNP are excellent discriminators of risk but to date have not been found to definitively change treatment decisions.³² The proof-of-principle NT-proBNP Guided Primary Prevention of CV Events in Diabetic Patients (PONTIAC) study³³ found a significant reduction in cardiovascular death and hospitalization in patients with elevated levels of natriuretic peptides randomized to more aggressive up-titration of inhibitors of the renin-angiotensin-aldosterone system and β-blockers, but additional studies on larger populations with more extended observation will be required to validate these findings.

Although biomarkers clearly identified patients at increased cardiovascular risk, no biomarker identified a particular patient population that benefited or was harmed by treatment with saxagliptin. Now that several different classes of diabetes drugs have been reported to improve cardiovascular outcomes,^34,35 biomarkers may provide additional insight into the mechanism of action and identification of patients most likely to benefit from therapy.

This study has some limitations. We did not assess left ventricular function. Baseline lipid levels were not measured, although most patients were undergoing lipid-lowering therapy in this trial.

Elevated levels of hsTnT, NT-proBNP, and hsCRP improved risk stratification in more than 12 000 patients with type 2 diabetes and established cardiovascular disease or multiple cardiovascular risk factors. Strategies to improve risk stratification in type 2 diabetes, with or without manifest cardiovascular disease, should consider incorporation of biomarker data into standard risk algorithms because they provide more accurate risk stratification than clinical variables alone.

Correction: This article was corrected on October 26, 2016, to fix a typographical error in the Funding/Support section.

Corresponding Author: Benjamin M. Scirica, MD, MPH, TIMI Study Group, Cardiovascular Division, Brigham and Women’s Hospital, 75 Francis St, Boston, MA 02115 (bscirica@partners.org).

Accepted for Publication: July 14, 2016.

Published Online: September 28, 2016. doi:10.1001/jamacardio.2016.3030

Author Contributions: Dr Scirica had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Concept and design: Scirica, Bhatt, Braunwald, Raz, Cavender, Hirshberg, Steg, Morrow.

Acquisition, analysis, or interpretation of data: Scirica, Bhatt, Braunwald, Im, Mosenzon, Udell, Hirshberg, Pollack, Jarolim, Morrow.

Drafting of the manuscript: Scirica, Im.

Critical revision of the manuscript for important intellectual content: Bhatt, Braunwald, Raz, Cavender, Im, Mosenzon, Udell, Hirshberg, Pollack, Steg, Jarolim, Morrow.

Statistical analysis: Scirica, Im.

Administrative, technical, or material support: Scirica, Cavender, Udell, Jarolim.

Study supervision: Scirica, Bhatt, Braunwald, Raz, Cavender, Hirshberg, Steg, Jarolim, Morrow.

Conflict of Interest Disclosures: All authors have completed and submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Dr Scirica reported receiving research grants via the Thrombolysis in Myocardial Infarction (TIMI) Study and Brigham and Women’s Hospital from AstraZeneca, Eisai, Merck, and Poxel and consulting fees from AstraZeneca, Biogen Idec, Boehringer Ingelheim, Covance, Dr Reddy’s Laboratory, Elsevier Practice Update Cardiology, Forest Laboratory, GE Healthcare, GlaxoSmithKline, Health[at]Scale, Lexicon, Merck, Sanofi, St. Jude's Medical, and University of Calgary. For the work under consideration, Dr Braunwald reported receiving grant support to his institution from AstraZeneca. For outside the submitted work, Dr Braunwald reported receiving grants to his institution from Novartis, Duke University, Merck, Daiichi Sankyo, and GlaxoSmithKline; uncompensated consultancies and lectures with Merck; uncompensated consultancies and lectures with Novartis; consultancies with The Medicines Company, Sanofi, and Theravance; and personal fees for lectures from Daiichi Sankyo, Menarini International, Bayer, and Medscape. Dr Raz reported receiving grants from AstraZeneca and Bristol-Myers Squibb during the conduct of the study; scientific board membership from Novo Nordisk, MSD, Eli Lilly, Sanofi, Medscape, Andromeda, and Insuline; payment for lectures, including service on speaker’s bureaus for lectures from Eli Lilly, Novo Nordisk, Johnson & Johnson, Sanofi, MSD, and Novartis; and stock options for Insuline. Dr Cavender reported receiving consulting fees from Merck and AstraZeneca. Dr Morrow reported receiving grants from AstraZeneca during the conduct of the study; consultancy fees from Abbott Laboratories, BG Medicine, Critical Diagnostics, Daiichi Sankyo, Genetech, Gilead, Instrumentation Laboratories, Johnson & Johnson, Konica Minolta, Merck, Novartis, Provenchio, Roche Diagnostics, and Servier; and grants from Abbott Laboratories, Beckman Coulter, BG Medicine, Critical Diagnostics, Bristol-Myers Squibb, Daiichi Sankyo, GlaxoSmithKline, Merck, Novartis, Roche Diagnostics, Sanofi, and Gilead. Dr Jarolim reported receiving grants from Abbott, AstraZeneca, Daiichi Sankyo, Merck, Roche Diagnostics, and Waters Technologies and consultancy fees from Biosystems and Quanterix. Dr Mosenzon reported receiving grants from AstraZeneca and Bristol-Myers Squibb during the conduct of the study; consulting fees from AstraZeneca and Bristol-Myers Squibb; and support for travel to meetings for the study from AstraZeneca and Bristol-Myers Squibb; serving on the scientific advisory board of AstraZeneca, Bristol-Myers Squibb, Novo Nordisk, Eli Lilly, Sanofi, and Norvartis; and serving on the speaker’s bureaus for Novo Nordisk, AstraZeneca, Bristol-Myers Squibb, Eli Lilly, Sanofi, Norvartis, Merc Sharpe, and Dohme. Dr Udell reported receiving grants from AstraZeneca and Bristol-Myers Squibb during the conduct of the study and grants from Novartis and personal fees from Amgen, AstraZeneca, Janssen, Merck, Novartis, Sanofi Pasteur, and Servier outside the submitted work. Drs Hirshberg and Pollack reported employment by AstraZeneca and having stock or stock options in AstraZeneca. Dr Steg reported receiving personal fees from AstraZeneca (during the conduct of the study), Amarin, Bayer, Boehringer-Ingelheim, Bristol-Myers-Squibb, Daiichi Sankyo, GlaxoSmithKline, Lilly, Merck-Sharpe-Dohme, Novartis, Otsuka, Pfizer, Roche, and The Medicines Company; grants and personal fees from Sanofi and Servier; and personal fees (outside the submitted work) from Vivus. Dr Bhatt reported receiving grants from Amarin, AstraZeneca, Bristol-Myers Squibb, Eisai, Ethicon, Medtronic, Sanofi, The Medicines Company, FlowCo, PLx Pharma, Takeda, Roche, Pfizer, Forest Laboratories, Ischemix; personal fees from Duke Clinical Research Institute, Mayo Clinic, Population Health Research Institute, American College of Cardiology (as well as nonfinancial support), Belvoir Publications, Slack Publications, WebMD, and Elsevier; and other support from Medscape Cardiology, Regado Biosciences, Boston Veterans Affairs Research Institute, the Society of Cardiovascular Patient Care, American Heart Association, HMP Communications, Harvard Clinical Research Institute, Clinical Cardiology, Journal of the American College of Cardiology, Veterans Affairs, St. Jude Medical, Biotronik, and Cardax. No other disclosures were reported.

Funding/Support: Saxagliptin Assessment of Vascular Outcomes Recorded in Patients With Diabetes Mellitus (SAVOR)–TIMI 53 was sponsored by AstraZeneca and Bristol-Myers Squibb.

Role of the Funder/Sponsor: The TIMI Study Group, Hadassah Medical Organization, and the sponsors were responsible for the design and conduct of the study and collection and management of the data. The TIMI Study Group is responsible for the current analysis; interpretation of the data; preparation, review, and approval of the manuscript; and decision to submit the manuscript for publication.

Additional Information: The SAVOR-TIMI 53 Executive Committee is as follows: Eugene Braunwald, MD (study chair), Deepak L. Bhatt, MD, MPH (coprincipal investigator), Itamar Raz, MD (coprincipal investigator), Jaime A. Davidson, MD, Robert Frederich, MD (nonvoting), Boaz Hirshberg, MD, PhD (nonvoting), and P. Gabriel Steg, MD.