Journal of Cardiac Failure Vol.
-
No.
-
2015
Clinical Investigation
Trends in Heart Failure Clinical Trials From 2001e2012 AYMAN SAMMAN TAHHAN, MD,1 MUTHIAH VADUGANATHAN, MD MPH,2 ANITA KELKAR, MD, MPH,1 VASILIKI V. GEORGIOPOULOU, MD,1 ANDREAS P. KALOGEROPOULOS, MD, MPH, PhD,1 STEPHEN J. GREENE, MD,3 GREGG C. FONAROW, MD,4 MIHAI GHEORGHIADE, MD,3 AND JAVED BUTLER, MD, MPH5 Atlanta, Georgia; Boston, Massachusetts; Chicago, Illinois; Los Angeles, California; and Stony Brook, New York
ABSTRACT Background: A systematic assessment of the temporal trends in heart failure (HF) clinical trials is lacking. Methods and Results: A total of 154 phase IIeIV HF trials including 162,725 patients published from 2001 to 2012 in 8 high-impact-factor journals were reviewed. The median number of participants and sites per trial were 367 (interquartile range [IQR] 133-1450) and 38 (5e101), respectively. Median enrollment duration was 2.2 (1.5e3.3) years. The majority of studies investigated treatment for chronic HF (82.5%) and investigated HF with reduced ejection fraction (EF) (71.4%), whereas 27 trials (17.5%) enrolled patients with mixed EF and 9 (5.8%) enrolled HF with preserved EF patients alone. Enrollment rates did not significantly change over time (median 0.49 patients site1 month1, IQR 0.34e0.98; P 5 .53). Trials meeting their primary end point decreased over time from 73.5% in 2001e2003 to 52.5% in 2010e2012 (P 5 .08) and were more often smaller and used nonmortality end points. Industry trials were larger with shorter enrollment duration, more concentrated in North America, and more likely to be positive. Trials conducted exclusively outside North America and Western Europe had the highest enrollment rates (median 1.95 patients site1 month1, IQR 1.34e4.11). Conclusions: Contemporary HF clinical trials display slow enrollment rates and decreased rates of positive outcomes over time. Positive trials tended to be smaller size with a higher proportion of surrogate end points. (J Cardiac Fail 2015;-:1e9) Key Words: Clinical trials, heart failure, outcomes, temporal trends.
Heart failure (HF) constitutes a tremendous health care burden and is the leading cause of hospitalizations among older adults in the United States.1e3 Despite recent national attempts at reform of HF care, mortality and readmission rates among HF patients remain suboptimal.4 There
remains an unmet need to develop new therapies for these patients.5 Recent HF clinical trials, however, have faced hurdles, particularly sluggish patient recruitment and retention. For example, in the EVEREST (Efficacy of Vasopressin Antagonism in Heart Failure: Outcome Study With Tolvaptan) trial, more than one-fifth of North American sites failed to enroll any patients, and even enrolling sites recruited only an average of 7 patients per site over the 28-month follow-up.6 In a recent study of the ClinicalTrials.gov registry, the major reason for early termination of cardiovascular clinical trials was lower than expected recruitment.7 Failing to meet enrollment targets delay clinical trials, drive up costs, and pose threats to both internal and external validity. Moreover, baseline characteristics, treatment patterns, protocol completion rates, and outcomes may differ by site enrollment rate, potentially influencing overall trial results.6 Comprehensive data characterizing recent trends and experience in enrollment patterns of HF clinical trials are limited. To this end, we sought to describe operational
From the 1Cardiology Division, Emory University, Atlanta, Georgia; Department of Medicine, Massachusetts General Hospital, Harvard Medical School, Boston, Massachusetts; 3Center for Cardiovascular Innovation, Northwestern University Feinberg School of Medicine, Chicago, Illinois; 4Division of Cardiology, University of California Los Angeles, Los Angeles, California and 5Division of Cardiology, Stony Brook University, Stony Brook, New York. Manuscript received February 24, 2015; revised manuscript received May 12, 2015; revised manuscript accepted June 15, 2015. Reprint requests: Javed Butler, MD, MPH, Division of Cardiology, Stony Brook University, Health Sciences Center, T-16, Room 080, Stony Brook, NY 11794, USA. Tel: þ1 631-444-1066; Fax: þ1 631-444-1054. E-mail: [email protected] See page 8 for disclosure information. 1071-9164/$ - see front matter Ó 2015 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.cardfail.2015.06.014 2
1
2 Journal of Cardiac Failure Vol.
-
No.
- -
2015
characteristics of HF trials published from 2001 to 2012 in the 8 highest-impact-factor medical journals in the categories of general medicine and cardiology. Methods Identification of Clinical Trials All HF trials published in 2001e2012 in the 8 highest-impactfactor journals in the ‘‘General and Internal Medicine’’ and ‘‘Cardiology’’ categories of the 2013 Journal Citation Reports, including New England Journal of Medicine (NEJM), Journal of the American Medical Association (JAMA), Lancet, Annals of Internal Medicine (AIM), Circulation, European Heart Journal (EHJ), Journal of the American College of Cardiology (JACC), and British Medical Journal (BMJ), were reviewed. We identified 154 phase IIeIV randomized controlled trials (RCT) by means of an electronic search of the Pubmed database with the use of the key words ‘‘trial*’’ and ‘‘random*’’ restricted to the aforementioned high-impact journals. To ensure that no trials were missed, a subsequent manual search of each individual journal edition from January 2001 to December 2012 was performed. The following studies were excluded: (1) pilot or phase I trials, (2) pediatric trials, (3) trials including hospitals as units of intervention, and (4) publications reporting interim, secondary, or post hoc analyses. We followed PRISMA guidelines for all procedures and reporting. Data Abstraction The following data were abstracted: (1) journal, (2) year of publication, (3) HF type, (4) recruitment setting and acuity, (5) intervention, (6) duration (estimated from starting and ending dates), (7) total patients enrolled, (8) total number of sites, (9) number of participating countries, (10) number of participating sites in each country, (11) primary outcomes, and (12) funding sources. For incomplete data fields, additional data were extracted from secondary publications identified in ClinicalTrials.gov. Trials were divided into trials including HF with reduced ejection fraction (HFrEF) and with preserved ejection fraction (HFpEF) and trials recruiting both types. Trials were also classified into acute and chronic (stable, ambulatory) subsets. EF cutoff points used for enrollment criteria also were collected. The trials were further divided into (1) acute HF from hospital units with short-term intervention, (2) chronic outpatient HF, (3) chronic HF recruited from the inpatient setting but with post-discharge long-term intervention, and (4) chronic HF recruited from both inpatient and outpatient settings or recruitment setting not clearly designated. Trials were divided into 6 categories based on the intervention: (1) medications, (2) devices (pacemakers, left ventricular assist devices, implantable cardioverter-defibrillators, biventricular pacemakers, intra-aortic balloon pumps), (3) surgical procedures (coronary artery bypass surgery, ventricular reconstruction), (4) nonsurgical procedures (intracoronary gene therapy, ultrafiltration), (5) others (exercise training, continuous positive airway pressure, multidisciplinary management, patient education, behavioral and lifestyle interventions), and (6) testing/imaging. Based on the ClinicalTrials.gov designations, funding source was assessed as (1) industry, (2) government, or (3) university or other nonprofit or nonfederal organizations. Trials were further classified according to the primary end point measured: (1) all-cause mortality, (2) cardiovascular or HF-related death, (3) nonmortality intermediate end point (subjective
measures that may be dependent on patient motivation or clinical judgment, such as symptom scores, hospitalizations, exercise tolerance tests), and (4) ‘‘surrogate’’ end points as indirect measures for clinically meaningful outcomes (eg, assessment of left ventricular function or biomarkers). The most common nonmortality intermediate and surrogate end points used in the included sample of trials are summarized in Table 1. A ‘‘positive’’ trial was defined when the null hypothesis was rejected for the primary end point (intervention was either superior or equivalent/noninferior according to the primary hypothesis). Reported outcomes were divided into mortality as primary outcome vs trials using surrogate end points (eg, dyspnea relief, wedge pressure, B-type natriuretic peptide, etc). Regions were divided into (1) exclusively in North America (NA), including the United States, Canada, and Mexico; (2) exclusively in Western Europe (WE), including Austria, Belgium, Bermuda, Denmark, Finland, France, Germany, Greece, Iceland, Ireland, Israel, Italy, Luxembourg, the Netherlands, Norway, Portugal, Spain, Sweden, Switzerland, and United Kingdom; (3) exclusively outside of NA and WE (rest of the world [ROW]); and (4) mixed/multiregional. Statistical Analysis Trials were divided into four 3-year periods on the basis of publication date (2001e2003, 2004e2006, 2007e2009, and 2010e2012). Continuous variables were described as mean and SD or as median and interquartile range (IQR), categoric variables as number and percentage. Enrollment rates were estimated based on the reported study duration (completion date minus start date). Continuous variables were compared across nominal categories with the use of the Kruskal-Wallis test and Bonferroni-adjusted post hoc pairwise comparisons to maintain the family-wise error a 5 0.05. Categoric variables were compared with the use of chi-square testing. A second investigator rereviewed trials conducted from 2001 to 2003 for the percentage of trials meeting their primary end point. Proportion of agreement was 0.94 (0.85e0.98) and Cohen unweighted k coefficient was 0.88 (0.77e0.99) between the 2 reviewers for this sample. Analyses were performed with the use of IBM SPSS 21 (IBM Corp, Armonk, New York).
Results Trial Characteristics
A total of 154 trials that collectively enrolled 162,725 patients were identified (Fig. 1). Forty (26%) were published in JACC, 35 (22.7%) in NEJM, and 29 (18.8%) in Circulation. The median number of participants was 367 (IQR 133e1,450), the median number of participating sites per trial was 38 (IQR 5e101), and the median duration of enrollment was 2.2 (IQR 1.5e3.3) years. The distributions of other characteristics are presented in Table 2. The majority of studies (127 trials, 82.5%) investigated chronic HF, and only 27 trials (17.5%) tested therapy in acute HF. Most chronic HF trials recruited patients from the outpatient setting; only 12 trials (7.8%) recruited chronic HF participants from the inpatient setting. For acute HF trials, only 5 of the 27 trials (18.5%) included natriuretic peptides as a key enrollment criteria. B-Type natriuretic peptide cutoffs ranged from 350 to 500 pg/
Trends in HF Clinical Trials
Samman Tahhan et al
3
Table 1. Common Surrogate and Nonmortality Intermediate Outcomes Common Surrogate Outcomes Left ventricular structure and function Echocardiographic measurements (eg, left ventricular ejection fraction, left ventricular systolic and diastolic volumes, stroke volume, mitral flow E/A ratio, deceleration time, isovolumic relaxation time, and ratio of systolic-diastolic pulmonary venous flow velocity) Magnetic resonance imaging Single-photon emission computerized tomography Radionuclide ventriculography Hemodynamic indices Cardiac index Pulmonary capillary wedge pressure Metrics of decongestion Renal function (serum creatinine, glomerular filtration rate) Electrolytes and urine output Weight loss Biomarkers Natriuretic peptides High-sensitivity C-reactive protein Soluble tumor necrosis factor Vascular adhesion molecule 1, intercellular adhesion molecule 1, P-selectin Noninvasive endothelial function studies Heart rate variability Common Nonmortality Intermediate Outcomes Exercise tolerance or capacity Cardiopulmonary exercise testing (peak exercise oxygen consumption) 6-minute walk testing Hospitalizations Total number of days hospitalized Hospitalizations and emergency or urgent care visits Heart failureerelated readmission rate Quality of life metrics Changes in New York Heart Association functional class
mL, and N-terminal proeB-type natriuretic peptide cutoffs ranged from 1,000 to 2,000 pg/dL. Biomarker-based criteria were generally accompanied by other criteria based on EF, renal function, and invasive or noninvasive hemodynamic measures. The majority were sponsored by industry (82 trials, 53.2%). Most trials (n 5 110; 71.4%) investigated HFrEF, 27 trials (17.5%) enrolled patients with mixed EF, and 9 trials (5.8%) enrolled HFpEF patients alone. Left ventricular ejection fraction (LVEF) cutoff used in HFrEF trials as one of the inclusion criteria ranged from !25% to 50%, with a median cutoff of 35% across trials. Over time, the median EF cutoff used did not significantly change (P 5 .19). For the definition of HFpEF, trials used an LVEF cutoff range from O40% to 50%, with a median cutoff of 45%. The median EF cutoff for HFpEF also did not significantly change over time (P 5 .85). Enrollment
The median enrollment rate was 0.49 (IQR 0.34e0.98) patients site1 month1, with no significant change over time (P 5 .53 for trend; Fig. 2). There was a trend toward higher recruitment in chronic HF trials from hospitals (0.92, IQR 0.41e4.66) compared with acute HF, chronic HF from outpatient setting, and mixed recruitment (0.56 [IQR 0.37e0.92], 0.45 [IQR 0.34e0.93], 0.41 [IQR 0.21e0.95], respectively; P 5 .14). Enrollment was higher in trials recruiting mixed EF patients compared with HFrEF
or HFpEF only (0.90 [IQR 0.59e1.78] vs 0.41 [IQR 0.27e0.7] and 0.54 [IQR 0.38e1.6], respectively; P ! .001). Device trials had the lowest enrollment at 0.36 (IQR 0.23e0.51) patients site1 month1. Trial Outcomes
Overall, 96 trials (62.3%) reported positive primary outcomes. Of these, 21 (21.9%) included all-cause mortality as the primary end point, 8 (8.3%) included cardiovascular or HF-related death, 29 (30.2%) included intermediate end points, and 38 (39.6%) included surrogate end points. The proportion of trials with positive outcomes decreased from 73.5% in 2001e2003 to 52.5% in 2010e2012 (P 5 .08 for trend; Fig. 3). Trial characteristics based on outcome are presented in Table 3. Trials recruiting chronic HF patients from the outpatient setting were more likely to be positive (71.6%) compared with acute HF (55.6%) and chronic HF recruited from the inpatient setting (33.3%). Positive trials were likely to have fewer sites (28 [IQR 2e73] vs 65 [IQR 10e168]) and participants (240 [IQR 106e610] vs 627 [IQR 277e1,983) per trial; P 5 .001. Single-center trials were more likely to be positive. Enrollment rate and trial duration were not different between positive and negative trials. Trials with mortality as primary outcome were less frequently positive than trials with surrogate end points (45.3% vs 74.4%; P 5 .0001). Inpatient studies were less likely to be positive compared with mixed or outpatientonly studies (55.6% vs 60.4% and 71.6%, respectively;
4 Journal of Cardiac Failure Vol.
-
No.
- -
2015
Fig. 1. PRISMA flow diagram.
P 5 .06). No major differences were observed in positive trial outcomes across EF subgroups and geographic region (all P O .1). Trial Characteristics by Funding Source
Industry trials were more likely to include larger numbers of participants, sites, and countries. Duration of trial enrollment was shorter in industry trials (2.0 y, IQR 1.5e2.5) compared to university/organization (2.5 y, IQR 1.5e4) and government-funded trials (3.2 y, IQR 2e3.9; P 5 .01). There were more industry trials in NA (41.3%), whereas university/organization funding was more common in WE (60.6%; P ! 001). Trials supported by industry and university/organization were more likely to be positive compared with those funded by government sources (64.6% and 62.2% vs 50.0%, respectively; P 5 .16). Enrollment rates were highest for university/nonfederal organizationefunded trials (0.95, IQR 0.56e3.41) compared with government sources (0.69, IQR 0.36e2.25) and industry-sponsored trials (0.40, IQR 0.27e0.58 patients site1 month1; P ! .001). Trial Characteristics by Region
Multiregional trials had more participants and sites but lower enrollment rates. The highest enrollment was seen in trials done in ROW, with a median of 1.95 patients site1 month1 (IQR 1.34e4.11), followed by WE 0.92 (IQR 0.41e4.83), NA 0.43 (IQR 0.27e0.88), and multiregional
trials 0.41 (IQR 0.33e0.61; P 5 .001). Multiregional trials more commonly had mortality as the primary outcome (80%) compared with fewer than one-third of regionspecific trials (P ! .001). Trials in ROW were more likely to enroll patients regardless of EF (46.2%) compared with other regions (NA 22.4%, WE 19.4%; P 5 .05). Most device (63%) and procedural trials (100%) were done in NA rather than other regions (P ! .001). The proportion of trials conducted in NA trended down from 44.1% in 2001e2006 to 33.3% in 2007e2012. Similar trends were seen in trials conducted in ROW 11.9% to 8.7%, whereas WE and multiregional trials increased over time from 20.3% to 27.5% and 23.7% to 30.4%, respectively. However, these trends did not reach statistical significance (P 5 .4). Discussion Our study highlights a number of important patterns of recent HF clinical trials over a 12-year time frame, including low enrollment rates, fewer positive trials over time, and only a minority of studies focusing on hospitalized HF (17%) and HFpEF (6%) patients. Efficient clinical trial enrollment facilitates timely completion and dissemination of results. Unfortunately, recent estimates suggest that up to 60%e80% of clinical trials in the United States do not meet their temporal targets owing to challenges in recruitment, and 30% of trial sites may fail to recruit even a single patient.8 These
Trends in HF Clinical Trials
Samman Tahhan et al
5
Table 2. Trial Characteristics and Enrollment Rates Trial Characteristic Journal Annals of Internal Medicine British Medical Journal Circulation European Heart Journal Journal of the American College of Cardiology Journal of the American Medical Association Lancet New England Journal of Medicine Trial classification Chronic HF recruited from outpatients Acute HF recruited from inpatients Chronic HF recruited from inpatients Chronic HF recruited from both in- and outpatients Year published 2001e2003 2003e2006 2007e2009 2010e2012 Intervention Device Medication Others Procedure Surgery Testing/imaging Heart failure classification HF with preserved EF HF with reduced EF HF regardless of EF Unclear Sponsor Government Industry University/organization Unclear Outcome Negative Positive Trial size Patients Sites Countries Duration (y) Enrollment rate* Single region Multiregional Single country Multinational Region North America Western Europe Rest of the world Mixed Primary outcome All-cause mortality Cardiovascular or HF-related death Nonmortality intermediate outcomes Surrogate outcomes
n (%)
Enrollment Rate*
P Value .001
3 2 29 19 40 12 14 35
(1.9) (1.3) (18.8) (12.3) (26) (7.8) (9.1) (22.7)
4.23 4.72 0.45 1.08 0.86 0.62 0.46 0.39
(3.24e6.04) (1.78e7.67) (0.25e0.96) (0.31e2.42) (0.41e1.95) (0.42e0.73) (0.35e0.62) (0.22e0.59)
67 27 12 48
(43.5) (17.5) (7.8) (31.2)
0.45 0.56 0.92 0.41
(0.34e0.93) (0.37e0.92) (0.41e4.66) (0.21e0.95)
34 34 46 40
(22.1) (22.1) (29.9) (26)
0.4 0.68 0.54 0.55
(0.3e0.85) (0.36e3.01) (0.35e0.92) (0.3e0.93)
33 81 24 5 4 7
(21.4) (52.6) (15.6) (3.2) (2.6) (4.5)
0.36 0.54 1.39 0.36 1.09 0.92
(0.23e0.51) (0.35e0.88) (0.68e4.23) (0.12e0.66) (0.2e88.22) (0.46e12.81)
9 110 27 8
(5.8) (71.4) (17.5) (5.2)
0.54 0.41 0.99 2.25
(0.38e1.6) (0.27e0.7) (0.59e1.78) (0.97e41.04)
30 82 37 5
(19.5) (53.2) (24) (3.2)
0.66 0.4 0.97 1.95
(0.36e2.41) (0.27e0.59) (0.46e3.83) (1.89e0)
.14
.53
!.001
!.001
!.001
.55 58 (37.7) 96 (62.3) 366.5 37.5 1 2.17 0.49 93 35 82 65
0.59 (0.41e0.95) 0.41 (0.26e1.2)
(133.25e1449.5) (5e101.5) (1e8) (1.52e3.3) (0.34e0.98) (72.3) (27.3) (55.8) (44.2)
0.64 0.41 0.73 0.41
(0.35e1.83) (0.34e0.62) (0.36e1.91) (0.29e0.62)
49 31 13 35
(38.3) (24.2) (10.2) (27.3)
0.43 0.88 1.96 0.41
(0.28e0.88) (0.41e4.08) (1.35e4.12) (0.34e0.62)
48 16 41 49
(31.2) (10.4) (26.6) (31.8)
0.41 0.6 0.43 0.68
(0.28e0.73) (0.28e0.78) (0.33e1.19) (0.41e1.89)
.03 .003 .001
.16
HF, heart failure; EF, ejection fraction. *Patients per site per month.
barriers to effective recruitment are the major reason for early termination.7 Our study demonstrated that enrollment rates were persistently low over time, at !0.7 patients site1 month1. These rate estimates appear to be even lower than contemporaneously conducted general
cardiovascular trials and are paradoxically low considering the tremendous prevalence of HF in the general population.9 Moreover, the excess costs and resources incurred by slow trial enrollment may influence patient profiles and trial outcomes.6
6 Journal of Cardiac Failure Vol.
-
No.
- -
2015
Fig. 2. Enrollment rates in heart failure clinical trials. Bars represent median enrollment rate expressed as patients per site per month. The trend over time was not significant (P 5 .35).
A number of strategies have been suggested to improve recruitment of HF patients. Enhanced multidisciplinary support is likely critical, particularly in trials for hospitalized HF, where the logistics of timely patient enrollment must combat pressures for expediency of care. Targeted communication and referrals from health care providers10,11 may augment patient volume. Patient education may overcome cultural barriers to participation in certain ethnic groups.12e14 To test these strategies within sites, a pretrial registry preceding official trial enrollment has been proposed. Although yet to be implemented, a pretrial registry could help inclusion of high-quality centers capable of enrolling appropriate patients and meeting the needs of the subsequent trial.15 Over the past decade, HF trials with positive outcomes decreased, especially considering that the present data from high-tier medical journals likely overestimate the success of trials due to publication bias.16,17 These findings are unsettling when one also considers that positive trials were
Fig. 3. Heart failure trial outcomes over time. The trend for positive clinical outcomes over the 12-year study period was statistically significant (P 5 .04).
more likely to be smaller single-center experiences that had nonmortality outcomes, likely as part of early experience with drug development. The use of surrogate and composite end points in contemporary trials may be due to attempts to shorten the timeline with faster end point accrual. Although trial execution likely plays a large role in declining rates of positive trial outcomes, other factors including poor understanding of study drugs and failure to match the study drug with the correct patient population may shape these recent trends.18 The use of surrogate markers continues to play an important role in early-phase clinical trials. Surrogate end points facilitate limiting sample size requirements and increase feasibility of smaller-budget clinical trials over a shorter duration, or they allow assessing those interventions where clinical benefit is uncertain and further insights are needed before proceeding with pivotal large-scope and -budget trials. Certain surrogate end points may provide early therapeutic or safety signals that could inform future trial planning and direction. Finally, surrogate markers may have value independent from downstream morbidity and mortality, and they may capture patient-centered metrics, including functionality and quality of life. However, despite the theoretic rationale of surrogate end points in early-phase studies, there is an important disconnect between phase II and III clinical trials, especially in certain settings, eg, acute HF.18 These measures may be costly, may be difficult to standardize across sites, and may have important limiting characteristics. Therefore, surrogate end points need to continue to be carefully studied in conjunction with ‘‘hard’’ clinical events. The key factors that will influence the future utility of surrogate end points in phase II clinical trials include the use of validated metrics that are (1) driven by a biologic rationale of potential therapeutic response, (2) predictive of clinical outcomes, (3) studied in a well phenotyped HF subgroup, and (4) measured in a standardized fashion in a central core laboratory. Our study revealed that recent trials conducted outside NA achieved more efficient enrollment.19 Differences in rate of enrollment may be explained by systematic variation in risk profiles, background treatments, and systems of care.20e22 Despite higher enrollment with greater number of sites, multicenter/multiregion trials tended to have the worst enrollment rates. Although such broad geographic catchments may be necessary to generate the required study power for evaluating mortality and hospitalization end points, this pattern of globalization may influence trials in other ways. For example, in the TOPCAT (Treatment of Preserved Cardiac Function Heart Failure With an Aldosterone Antagonist) study, rates of the primary outcome were significantly reduced with the use of spironolactone in the Americas, a benefit that was not observed in the overall trial population and in Russia/Georgia. This differential effect generates the hypothesis that the neutral study results were a consequence of geographic variation in patient characteristics, site enrollment, or study execution.23 Thus,
Trends in HF Clinical Trials
Samman Tahhan et al
7
Table 3. Trial Characteristics Based on Trial Outcomes, n (%) Trial Characteristic Journal Annals of Internal Medicine British Medical Journal Circulation European Heart Journal Journal of the American College of Cardiology Journal of the American Medical Association Lancet New England Journal of Medicine Trial classification Chronic HF recruited from outpatient Acute HF recruited from the inpatient Chronic HF recruited from the inpatient Chronic HF recruited from both in/outpatient Year published 2001e2003 2003e2006 2007e2009 2010e2012 Intervention Device Medication Others Procedure Surgery Testing/imaging Heart failure classification HF with preserved EF HF with reduced EF HF regardless of EF Unclear Sponsor Government Industry University/Organization Unclear Trial size Patients Sites Countries Duration (y) Enrollment rate (patients per site per month) Single region Multiregional Single country Multinational Region North America Western Europe Rest of the world Mixed Primary outcome All-cause mortality Cardiovascular or HF-related death Nonmortality intermediate outcomes Surrogate outcomes
Negative Trials
Positive Trials
P Value .01
1 1 11 9 7 10 4 15
(33.3) (50) (37.9) (47.4) (17.5) (83.3) (28.6) (42.9)
2 1 18 10 33 2 10 20
(66.7) (50) (62.1) (52.6) (82.5) (16.7) (71.4) (57.1)
19 12 8 19
(28.4) (44.4) (66.7) (39.6)
48 15 4 29
(71.6) (55.6) (33.3) (60.4)
9 9 21 19
(26.5) (26.5) (45.7) (47.5)
25 25 25 21
(73.5) (73.5) (54.3) (52.5)
6 33 11 4 1 3
(18.2) (40.7) (45.8) (80) (25) (42.9)
27 48 13 1 3 4
(81.8) (59.3) (54.2) (20) (75) (57.1)
4 39 12 3
(44.4) (35.5) (44.4) (37.5)
5 71 15 5
(55.6) (64.5) (55.6) (62.5)
15 29 14 0
(50) (35.4) (37.8) (0)
15 53 23 5
(50) (64.6) (62.2) (100)
.06
.08
.06
.81
.16
627 65 2 2.33 0.55 31 17 25 31
(277e1983.5) (10e168) (1e16) (1.7e3.34) (0.38e0.83) (33.3) (48.6) (30.5) (47.7)
240 28 1 2.08 0.41 62 18 57 34
(106e610) (2e73) (1e4) (1.42e3.3) (0.26e1.19) (66.7) (51.4) (69.5) (52.3)
.001 .005 .01 .28 .55 .11 .03 .12
20 9 2 17
(40.8) (29) (15.4) (48.6)
29 22 11 18
(59.2) (71) (84.6) (51.4)
27 8 12 11
(46.6) (13.8) (20.7) (19)
21 8 29 38
(21.9) (8.3) (30.2) (39.6)
.003
HF, heart failure; EF, ejection fraction.
interpreting heterogeneous results garnered from a diverse patient population across numerous countries significantly affects generalizability.24 Compared with other funding sources, industrysponsored trials tended to be larger and multicenter to potentially offset sluggish enrollment. This approach is costly and inefficient, but whether funding mechanism influences trial outcome is unclear. In our study, industry-
and university/organization-funded studies tended to report more positive findings than government-supported trials, consistent with earlier exploratory analyses.25 Although definitive conclusions are impossible, given the complexity of trial conduct and execution, these data are intriguing, eg, selection of a comparator may bias the study results,26 and the interpretation of results may be confounded.27
8 Journal of Cardiac Failure Vol.
-
No.
- -
2015
Fewer than 6% of recent clinical trials were conducted in patients with HFpEF, and !17% included patients hospitalized for HF. Unfortunately, these high-risk patient subsets have limited treatment options and have had largely unchanged outcomes over the past 2 decades.5 Accordingly, our analyses, coupled with recent studies of ClinicalTrials.gov, reveal important unmet needs and potential areas of future drug development. A number of limitations must be considered when interpreting these findings. Study enrollment rate was calculated on the basis of overall patient enrollment per site per a given unit of time; however, this approach does not account for site discontinuation or additions during the course of the trial. These data are not published and could result in underestimating the true enrollment rates. Given the inclusion of clinical trials published only in 8 high-impact journals rather than all HF trials, generalizability of these results may be limited and success of trial programs may be overestimated. In conclusion, recent trends in HF clinical trials have important implications on future research investment, funding mechanisms, and patient confidence. Moreover, disclosure of further trial conduct data in the public domain will help ascertain these trends more carefully. Disclosures Dr Fonarow reports significant research grants from the National Heart, Lung, and Blood Institute and Agency for Healthcare Research and Quality, significant consultancy support from Novartis and Takeda, and modest consultancy support from Bayer, Gambro, Janssen, The Medicines Company, and Medtronic. Dr Gheorghiade reports relationships with Abbott, Astellas, Astrazeneca, Bayer, Cardiorentis, Corthera, Cytokinetics, Cytopherx, Debiopharm, Errekappa Terapeutici, Glaxosmithkline, Ikaria, Intersection Medical, Johnson and Johnson, Medtronic, Merck, Novartis, Ono Pharma, Otsuka, Palatin Technologies, Pericor Therapeutics, Protein Design, Sanofi-Aventis, Sigma Tau, Solvay, Sticares Interact, Takeda, and Trevena. Dr Butler reports research support from the National Institutes of Health and European Union and is a consultant to Amgen, Bayer, BG Medicine, Celladon, Gambro, GE Healthcare, Harvest, Medtronic, Ono Pharma, Stemedica, and Trevena. The other authors report no relevant conflicts of interest.
Supplementary Data Supplementary data related to this article can be found at http://dx.doi.org/10.1016/j.cardfail.2015.06.014.
References 1. Butler J, Marti C, Pina I, DeFilippi C. Scope of heart failure hospitalization. Congest Heart Fail 2012;18(Suppl 1):S1e4.
2. Bui AL, Horwich TB, Fonarow GC. Epidemiology and risk profile of heart failure. Nat Rev Cardiol 2011;8:30e41. 3. Gheorghiade M, Vaduganathan M, Greene SJ, Mentz RJ, Adams KF Jr, Anker SD, et al. Site selection in global clinical trials in patients hospitalized for heart failure: perceived problems and potential solutions. Heart Fail Rev 2014;19:135e52. 4. Suter LG, Li SX, Grady JN, Lin Z, Wang Y, Bhat KR, et al. National patterns of risk-standardized mortality and readmission after hospitalization for acute myocardial infarction, heart failure, and pneumonia: update on publicly reported outcomes measures based on the 2013 release. J Gen Intern Med 2014;29:1333e40. 5. Butler J, Fonarow GC, Gheorghiade M. Strategies and opportunities for drug development in heart failure. JAMA 2013;309:1593e4. 6. Butler J, Subacius H, Vaduganathan M, Fonarow GC, Ambrosy AP, Konstam MA, et al. Relationship between clinical trial site enrollment with participant characteristics, protocol completion, and outcomes: insights from the EVEREST (Efficacy of Vasopressin Antagonism in Heart Failure: Outcome Study With Tolvaptan) trial. J Am Coll Cardiol 2013;61:571e9. 7. Bernardez-Pereira S, Lopes RD, Carrion MJ, Santucci EV, Soares RM, de Oliveira Abreu M, et al. Prevalence, characteristics, and predictors of early termination of cardiovascular clinical trials due to low recruitment: insights from the ClinicalTrials.gov registry. Am Heart J 2014; 168:213e219.e1. 8. Blanton S, Morris DM, Prettyman MG, McCulloch K, Redmond S, Light KE, et al. Lessons learned in participant recruitment and retention: the EXCITE trial. Phys Ther 2006;86:1520e33. 9. Butler J, Samman Tahhan A, Georgiopoulou V, Kelkar A, Lee M, Khan B, et al. Trends in characteristics of cardiovascular clinical trials 2001e2012. Am Heart J 2015. In press. doi:10.1016/j.ahj.2015.05.006. 10. Galbreath AD, Smith B, Wood P, Forkner E, Peters JI. Cumulative recruitment experience in two large single-center randomized, controlled clinical trials. Contemp Clin Trials 2008;29:335e42. 11. Campbell MK, Snowdon C, Francis D, Elbourne D, McDonald AM, Knight R, et al. Recruitment to randomised trials: strategies for trial enrollment and participation study. The STEPS study. Health Technol Assess 2007;11. iii, ix-105. 12. Baquet CR, Commiskey P, Daniel Mullins C, Mishra SI. Recruitment and participation in clinical trials: socio-demographic, rural/urban, and health care access predictors. Cancer Detect Prev 2006;30: 24e33. 13. Gul RB, Ali PA. Clinical trials: the challenge of recruitment and retention of participants. J Clin Nurs 2010;19:227e33. 14. Chang BH, Hendricks AM, Slawsky MT, Locastro JS. Patient recruitment to a randomized clinical trial of behavioral therapy for chronic heart failure. BMC Med Res Methodol 2004;4:8. 15. Greene SJ, Shah AN, Butler J, Ambrosy AP, Anker SD, Chioncel O, et al. Designing effective drug and device development programs for hospitalized heart failure: a proposal for pretrial registries. Am Heart J 2014;168:142e9. 16. Dwan K, Gamble C, Williamson PR, Kirkham JJ, Reporting Bias G. Systematic review of the empirical evidence of study publication bias and outcome reporting biasdan updated review. PLoS One 2013;8:e66844. 17. Hopewell S, Loudon K, Clarke MJ, Oxman AD, Dickersin K. Publication bias in clinical trials due to statistical significance or direction of trial results. Cochrane Database Syst Rev 2009:MR000006. 18. Vaduganathan M, Greene SJ, Ambrosy AP, Gheorghiade M, Butler J. The disconnect between phase II and phase III trials of drugs for heart failure. Nat Rev Cardiol 2013;10:85e97. 19. Kim ES, Carrigan TP, Menon V. International participation in cardiovascular randomized controlled trials sponsored by the National Heart, Lung, and Blood Institute. J Am Coll Cardiol 2011;58:671e6. 20. Glickman SW, McHutchison JG, Peterson ED, Cairns CB, Harrington RA, Califf RM, et al. Ethical and scientific implications of the globalization of clinical research. N Engl J Med 2009;360:816e23. 21. Eapen ZJ, Vavalle JP, Granger CB, Harrington RA, Peterson ED, Califf RM. Rescuing clinical trials in the United States and beyond: a call for action. Am Heart J 2013;165:837e47.
Trends in HF Clinical Trials 22. Lang T, Siribaddana S. Clinical trials have gone global: is this a good thing? PLoS Med 2012;9:e1001228. 23. Pfeffer MA, Claggett B, Assmann SF, Boineau R, Anand IS, Clausell N, et al. Regional variation in patients and outcomes in the Treatment of Preserved Cardiac Function Heart Failure With an Aldosterone Antagonist (TOPCAT) Trial. Circulation 2015;131:34e42. 24. Stough WG, Zannad F, Pitt B, Goldstein S. Globalization of cardiovascular clinical research: the balance between meeting medical needs and maintaining scientific standards. Am Heart J 2007;154:232e8.
Samman Tahhan et al
9
25. Bourgeois FT, Murthy S, Mandl KD. Outcome reporting among drug trials registered in. Ann Intern Med 2010;153:158e66. ClinicalTrials.gov. 26. Lexchin J, Bero LA, Djulbegovic B, Clark O. Pharmaceutical industry sponsorship and research outcome and quality: systematic review. BMJ 2003;326:1167e70. 27. Als-Nielsen B, Chen W, Gluud C, Kjaergard LL. Association of funding and conclusions in randomized drug trials: a reflection of treatment effect or adverse events? JAMA 2003;290:921e8.
-
No.
-
2015
Clinical Investigation
Trends in Heart Failure Clinical Trials From 2001e2012 AYMAN SAMMAN TAHHAN, MD,1 MUTHIAH VADUGANATHAN, MD MPH,2 ANITA KELKAR, MD, MPH,1 VASILIKI V. GEORGIOPOULOU, MD,1 ANDREAS P. KALOGEROPOULOS, MD, MPH, PhD,1 STEPHEN J. GREENE, MD,3 GREGG C. FONAROW, MD,4 MIHAI GHEORGHIADE, MD,3 AND JAVED BUTLER, MD, MPH5 Atlanta, Georgia; Boston, Massachusetts; Chicago, Illinois; Los Angeles, California; and Stony Brook, New York
ABSTRACT Background: A systematic assessment of the temporal trends in heart failure (HF) clinical trials is lacking. Methods and Results: A total of 154 phase IIeIV HF trials including 162,725 patients published from 2001 to 2012 in 8 high-impact-factor journals were reviewed. The median number of participants and sites per trial were 367 (interquartile range [IQR] 133-1450) and 38 (5e101), respectively. Median enrollment duration was 2.2 (1.5e3.3) years. The majority of studies investigated treatment for chronic HF (82.5%) and investigated HF with reduced ejection fraction (EF) (71.4%), whereas 27 trials (17.5%) enrolled patients with mixed EF and 9 (5.8%) enrolled HF with preserved EF patients alone. Enrollment rates did not significantly change over time (median 0.49 patients site1 month1, IQR 0.34e0.98; P 5 .53). Trials meeting their primary end point decreased over time from 73.5% in 2001e2003 to 52.5% in 2010e2012 (P 5 .08) and were more often smaller and used nonmortality end points. Industry trials were larger with shorter enrollment duration, more concentrated in North America, and more likely to be positive. Trials conducted exclusively outside North America and Western Europe had the highest enrollment rates (median 1.95 patients site1 month1, IQR 1.34e4.11). Conclusions: Contemporary HF clinical trials display slow enrollment rates and decreased rates of positive outcomes over time. Positive trials tended to be smaller size with a higher proportion of surrogate end points. (J Cardiac Fail 2015;-:1e9) Key Words: Clinical trials, heart failure, outcomes, temporal trends.
Heart failure (HF) constitutes a tremendous health care burden and is the leading cause of hospitalizations among older adults in the United States.1e3 Despite recent national attempts at reform of HF care, mortality and readmission rates among HF patients remain suboptimal.4 There
remains an unmet need to develop new therapies for these patients.5 Recent HF clinical trials, however, have faced hurdles, particularly sluggish patient recruitment and retention. For example, in the EVEREST (Efficacy of Vasopressin Antagonism in Heart Failure: Outcome Study With Tolvaptan) trial, more than one-fifth of North American sites failed to enroll any patients, and even enrolling sites recruited only an average of 7 patients per site over the 28-month follow-up.6 In a recent study of the ClinicalTrials.gov registry, the major reason for early termination of cardiovascular clinical trials was lower than expected recruitment.7 Failing to meet enrollment targets delay clinical trials, drive up costs, and pose threats to both internal and external validity. Moreover, baseline characteristics, treatment patterns, protocol completion rates, and outcomes may differ by site enrollment rate, potentially influencing overall trial results.6 Comprehensive data characterizing recent trends and experience in enrollment patterns of HF clinical trials are limited. To this end, we sought to describe operational
From the 1Cardiology Division, Emory University, Atlanta, Georgia; Department of Medicine, Massachusetts General Hospital, Harvard Medical School, Boston, Massachusetts; 3Center for Cardiovascular Innovation, Northwestern University Feinberg School of Medicine, Chicago, Illinois; 4Division of Cardiology, University of California Los Angeles, Los Angeles, California and 5Division of Cardiology, Stony Brook University, Stony Brook, New York. Manuscript received February 24, 2015; revised manuscript received May 12, 2015; revised manuscript accepted June 15, 2015. Reprint requests: Javed Butler, MD, MPH, Division of Cardiology, Stony Brook University, Health Sciences Center, T-16, Room 080, Stony Brook, NY 11794, USA. Tel: þ1 631-444-1066; Fax: þ1 631-444-1054. E-mail: [email protected] See page 8 for disclosure information. 1071-9164/$ - see front matter Ó 2015 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.cardfail.2015.06.014 2
1
2 Journal of Cardiac Failure Vol.
-
No.
- -
2015
characteristics of HF trials published from 2001 to 2012 in the 8 highest-impact-factor medical journals in the categories of general medicine and cardiology. Methods Identification of Clinical Trials All HF trials published in 2001e2012 in the 8 highest-impactfactor journals in the ‘‘General and Internal Medicine’’ and ‘‘Cardiology’’ categories of the 2013 Journal Citation Reports, including New England Journal of Medicine (NEJM), Journal of the American Medical Association (JAMA), Lancet, Annals of Internal Medicine (AIM), Circulation, European Heart Journal (EHJ), Journal of the American College of Cardiology (JACC), and British Medical Journal (BMJ), were reviewed. We identified 154 phase IIeIV randomized controlled trials (RCT) by means of an electronic search of the Pubmed database with the use of the key words ‘‘trial*’’ and ‘‘random*’’ restricted to the aforementioned high-impact journals. To ensure that no trials were missed, a subsequent manual search of each individual journal edition from January 2001 to December 2012 was performed. The following studies were excluded: (1) pilot or phase I trials, (2) pediatric trials, (3) trials including hospitals as units of intervention, and (4) publications reporting interim, secondary, or post hoc analyses. We followed PRISMA guidelines for all procedures and reporting. Data Abstraction The following data were abstracted: (1) journal, (2) year of publication, (3) HF type, (4) recruitment setting and acuity, (5) intervention, (6) duration (estimated from starting and ending dates), (7) total patients enrolled, (8) total number of sites, (9) number of participating countries, (10) number of participating sites in each country, (11) primary outcomes, and (12) funding sources. For incomplete data fields, additional data were extracted from secondary publications identified in ClinicalTrials.gov. Trials were divided into trials including HF with reduced ejection fraction (HFrEF) and with preserved ejection fraction (HFpEF) and trials recruiting both types. Trials were also classified into acute and chronic (stable, ambulatory) subsets. EF cutoff points used for enrollment criteria also were collected. The trials were further divided into (1) acute HF from hospital units with short-term intervention, (2) chronic outpatient HF, (3) chronic HF recruited from the inpatient setting but with post-discharge long-term intervention, and (4) chronic HF recruited from both inpatient and outpatient settings or recruitment setting not clearly designated. Trials were divided into 6 categories based on the intervention: (1) medications, (2) devices (pacemakers, left ventricular assist devices, implantable cardioverter-defibrillators, biventricular pacemakers, intra-aortic balloon pumps), (3) surgical procedures (coronary artery bypass surgery, ventricular reconstruction), (4) nonsurgical procedures (intracoronary gene therapy, ultrafiltration), (5) others (exercise training, continuous positive airway pressure, multidisciplinary management, patient education, behavioral and lifestyle interventions), and (6) testing/imaging. Based on the ClinicalTrials.gov designations, funding source was assessed as (1) industry, (2) government, or (3) university or other nonprofit or nonfederal organizations. Trials were further classified according to the primary end point measured: (1) all-cause mortality, (2) cardiovascular or HF-related death, (3) nonmortality intermediate end point (subjective
measures that may be dependent on patient motivation or clinical judgment, such as symptom scores, hospitalizations, exercise tolerance tests), and (4) ‘‘surrogate’’ end points as indirect measures for clinically meaningful outcomes (eg, assessment of left ventricular function or biomarkers). The most common nonmortality intermediate and surrogate end points used in the included sample of trials are summarized in Table 1. A ‘‘positive’’ trial was defined when the null hypothesis was rejected for the primary end point (intervention was either superior or equivalent/noninferior according to the primary hypothesis). Reported outcomes were divided into mortality as primary outcome vs trials using surrogate end points (eg, dyspnea relief, wedge pressure, B-type natriuretic peptide, etc). Regions were divided into (1) exclusively in North America (NA), including the United States, Canada, and Mexico; (2) exclusively in Western Europe (WE), including Austria, Belgium, Bermuda, Denmark, Finland, France, Germany, Greece, Iceland, Ireland, Israel, Italy, Luxembourg, the Netherlands, Norway, Portugal, Spain, Sweden, Switzerland, and United Kingdom; (3) exclusively outside of NA and WE (rest of the world [ROW]); and (4) mixed/multiregional. Statistical Analysis Trials were divided into four 3-year periods on the basis of publication date (2001e2003, 2004e2006, 2007e2009, and 2010e2012). Continuous variables were described as mean and SD or as median and interquartile range (IQR), categoric variables as number and percentage. Enrollment rates were estimated based on the reported study duration (completion date minus start date). Continuous variables were compared across nominal categories with the use of the Kruskal-Wallis test and Bonferroni-adjusted post hoc pairwise comparisons to maintain the family-wise error a 5 0.05. Categoric variables were compared with the use of chi-square testing. A second investigator rereviewed trials conducted from 2001 to 2003 for the percentage of trials meeting their primary end point. Proportion of agreement was 0.94 (0.85e0.98) and Cohen unweighted k coefficient was 0.88 (0.77e0.99) between the 2 reviewers for this sample. Analyses were performed with the use of IBM SPSS 21 (IBM Corp, Armonk, New York).
Results Trial Characteristics
A total of 154 trials that collectively enrolled 162,725 patients were identified (Fig. 1). Forty (26%) were published in JACC, 35 (22.7%) in NEJM, and 29 (18.8%) in Circulation. The median number of participants was 367 (IQR 133e1,450), the median number of participating sites per trial was 38 (IQR 5e101), and the median duration of enrollment was 2.2 (IQR 1.5e3.3) years. The distributions of other characteristics are presented in Table 2. The majority of studies (127 trials, 82.5%) investigated chronic HF, and only 27 trials (17.5%) tested therapy in acute HF. Most chronic HF trials recruited patients from the outpatient setting; only 12 trials (7.8%) recruited chronic HF participants from the inpatient setting. For acute HF trials, only 5 of the 27 trials (18.5%) included natriuretic peptides as a key enrollment criteria. B-Type natriuretic peptide cutoffs ranged from 350 to 500 pg/
Trends in HF Clinical Trials
Samman Tahhan et al
3
Table 1. Common Surrogate and Nonmortality Intermediate Outcomes Common Surrogate Outcomes Left ventricular structure and function Echocardiographic measurements (eg, left ventricular ejection fraction, left ventricular systolic and diastolic volumes, stroke volume, mitral flow E/A ratio, deceleration time, isovolumic relaxation time, and ratio of systolic-diastolic pulmonary venous flow velocity) Magnetic resonance imaging Single-photon emission computerized tomography Radionuclide ventriculography Hemodynamic indices Cardiac index Pulmonary capillary wedge pressure Metrics of decongestion Renal function (serum creatinine, glomerular filtration rate) Electrolytes and urine output Weight loss Biomarkers Natriuretic peptides High-sensitivity C-reactive protein Soluble tumor necrosis factor Vascular adhesion molecule 1, intercellular adhesion molecule 1, P-selectin Noninvasive endothelial function studies Heart rate variability Common Nonmortality Intermediate Outcomes Exercise tolerance or capacity Cardiopulmonary exercise testing (peak exercise oxygen consumption) 6-minute walk testing Hospitalizations Total number of days hospitalized Hospitalizations and emergency or urgent care visits Heart failureerelated readmission rate Quality of life metrics Changes in New York Heart Association functional class
mL, and N-terminal proeB-type natriuretic peptide cutoffs ranged from 1,000 to 2,000 pg/dL. Biomarker-based criteria were generally accompanied by other criteria based on EF, renal function, and invasive or noninvasive hemodynamic measures. The majority were sponsored by industry (82 trials, 53.2%). Most trials (n 5 110; 71.4%) investigated HFrEF, 27 trials (17.5%) enrolled patients with mixed EF, and 9 trials (5.8%) enrolled HFpEF patients alone. Left ventricular ejection fraction (LVEF) cutoff used in HFrEF trials as one of the inclusion criteria ranged from !25% to 50%, with a median cutoff of 35% across trials. Over time, the median EF cutoff used did not significantly change (P 5 .19). For the definition of HFpEF, trials used an LVEF cutoff range from O40% to 50%, with a median cutoff of 45%. The median EF cutoff for HFpEF also did not significantly change over time (P 5 .85). Enrollment
The median enrollment rate was 0.49 (IQR 0.34e0.98) patients site1 month1, with no significant change over time (P 5 .53 for trend; Fig. 2). There was a trend toward higher recruitment in chronic HF trials from hospitals (0.92, IQR 0.41e4.66) compared with acute HF, chronic HF from outpatient setting, and mixed recruitment (0.56 [IQR 0.37e0.92], 0.45 [IQR 0.34e0.93], 0.41 [IQR 0.21e0.95], respectively; P 5 .14). Enrollment was higher in trials recruiting mixed EF patients compared with HFrEF
or HFpEF only (0.90 [IQR 0.59e1.78] vs 0.41 [IQR 0.27e0.7] and 0.54 [IQR 0.38e1.6], respectively; P ! .001). Device trials had the lowest enrollment at 0.36 (IQR 0.23e0.51) patients site1 month1. Trial Outcomes
Overall, 96 trials (62.3%) reported positive primary outcomes. Of these, 21 (21.9%) included all-cause mortality as the primary end point, 8 (8.3%) included cardiovascular or HF-related death, 29 (30.2%) included intermediate end points, and 38 (39.6%) included surrogate end points. The proportion of trials with positive outcomes decreased from 73.5% in 2001e2003 to 52.5% in 2010e2012 (P 5 .08 for trend; Fig. 3). Trial characteristics based on outcome are presented in Table 3. Trials recruiting chronic HF patients from the outpatient setting were more likely to be positive (71.6%) compared with acute HF (55.6%) and chronic HF recruited from the inpatient setting (33.3%). Positive trials were likely to have fewer sites (28 [IQR 2e73] vs 65 [IQR 10e168]) and participants (240 [IQR 106e610] vs 627 [IQR 277e1,983) per trial; P 5 .001. Single-center trials were more likely to be positive. Enrollment rate and trial duration were not different between positive and negative trials. Trials with mortality as primary outcome were less frequently positive than trials with surrogate end points (45.3% vs 74.4%; P 5 .0001). Inpatient studies were less likely to be positive compared with mixed or outpatientonly studies (55.6% vs 60.4% and 71.6%, respectively;
4 Journal of Cardiac Failure Vol.
-
No.
- -
2015
Fig. 1. PRISMA flow diagram.
P 5 .06). No major differences were observed in positive trial outcomes across EF subgroups and geographic region (all P O .1). Trial Characteristics by Funding Source
Industry trials were more likely to include larger numbers of participants, sites, and countries. Duration of trial enrollment was shorter in industry trials (2.0 y, IQR 1.5e2.5) compared to university/organization (2.5 y, IQR 1.5e4) and government-funded trials (3.2 y, IQR 2e3.9; P 5 .01). There were more industry trials in NA (41.3%), whereas university/organization funding was more common in WE (60.6%; P ! 001). Trials supported by industry and university/organization were more likely to be positive compared with those funded by government sources (64.6% and 62.2% vs 50.0%, respectively; P 5 .16). Enrollment rates were highest for university/nonfederal organizationefunded trials (0.95, IQR 0.56e3.41) compared with government sources (0.69, IQR 0.36e2.25) and industry-sponsored trials (0.40, IQR 0.27e0.58 patients site1 month1; P ! .001). Trial Characteristics by Region
Multiregional trials had more participants and sites but lower enrollment rates. The highest enrollment was seen in trials done in ROW, with a median of 1.95 patients site1 month1 (IQR 1.34e4.11), followed by WE 0.92 (IQR 0.41e4.83), NA 0.43 (IQR 0.27e0.88), and multiregional
trials 0.41 (IQR 0.33e0.61; P 5 .001). Multiregional trials more commonly had mortality as the primary outcome (80%) compared with fewer than one-third of regionspecific trials (P ! .001). Trials in ROW were more likely to enroll patients regardless of EF (46.2%) compared with other regions (NA 22.4%, WE 19.4%; P 5 .05). Most device (63%) and procedural trials (100%) were done in NA rather than other regions (P ! .001). The proportion of trials conducted in NA trended down from 44.1% in 2001e2006 to 33.3% in 2007e2012. Similar trends were seen in trials conducted in ROW 11.9% to 8.7%, whereas WE and multiregional trials increased over time from 20.3% to 27.5% and 23.7% to 30.4%, respectively. However, these trends did not reach statistical significance (P 5 .4). Discussion Our study highlights a number of important patterns of recent HF clinical trials over a 12-year time frame, including low enrollment rates, fewer positive trials over time, and only a minority of studies focusing on hospitalized HF (17%) and HFpEF (6%) patients. Efficient clinical trial enrollment facilitates timely completion and dissemination of results. Unfortunately, recent estimates suggest that up to 60%e80% of clinical trials in the United States do not meet their temporal targets owing to challenges in recruitment, and 30% of trial sites may fail to recruit even a single patient.8 These
Trends in HF Clinical Trials
Samman Tahhan et al
5
Table 2. Trial Characteristics and Enrollment Rates Trial Characteristic Journal Annals of Internal Medicine British Medical Journal Circulation European Heart Journal Journal of the American College of Cardiology Journal of the American Medical Association Lancet New England Journal of Medicine Trial classification Chronic HF recruited from outpatients Acute HF recruited from inpatients Chronic HF recruited from inpatients Chronic HF recruited from both in- and outpatients Year published 2001e2003 2003e2006 2007e2009 2010e2012 Intervention Device Medication Others Procedure Surgery Testing/imaging Heart failure classification HF with preserved EF HF with reduced EF HF regardless of EF Unclear Sponsor Government Industry University/organization Unclear Outcome Negative Positive Trial size Patients Sites Countries Duration (y) Enrollment rate* Single region Multiregional Single country Multinational Region North America Western Europe Rest of the world Mixed Primary outcome All-cause mortality Cardiovascular or HF-related death Nonmortality intermediate outcomes Surrogate outcomes
n (%)
Enrollment Rate*
P Value .001
3 2 29 19 40 12 14 35
(1.9) (1.3) (18.8) (12.3) (26) (7.8) (9.1) (22.7)
4.23 4.72 0.45 1.08 0.86 0.62 0.46 0.39
(3.24e6.04) (1.78e7.67) (0.25e0.96) (0.31e2.42) (0.41e1.95) (0.42e0.73) (0.35e0.62) (0.22e0.59)
67 27 12 48
(43.5) (17.5) (7.8) (31.2)
0.45 0.56 0.92 0.41
(0.34e0.93) (0.37e0.92) (0.41e4.66) (0.21e0.95)
34 34 46 40
(22.1) (22.1) (29.9) (26)
0.4 0.68 0.54 0.55
(0.3e0.85) (0.36e3.01) (0.35e0.92) (0.3e0.93)
33 81 24 5 4 7
(21.4) (52.6) (15.6) (3.2) (2.6) (4.5)
0.36 0.54 1.39 0.36 1.09 0.92
(0.23e0.51) (0.35e0.88) (0.68e4.23) (0.12e0.66) (0.2e88.22) (0.46e12.81)
9 110 27 8
(5.8) (71.4) (17.5) (5.2)
0.54 0.41 0.99 2.25
(0.38e1.6) (0.27e0.7) (0.59e1.78) (0.97e41.04)
30 82 37 5
(19.5) (53.2) (24) (3.2)
0.66 0.4 0.97 1.95
(0.36e2.41) (0.27e0.59) (0.46e3.83) (1.89e0)
.14
.53
!.001
!.001
!.001
.55 58 (37.7) 96 (62.3) 366.5 37.5 1 2.17 0.49 93 35 82 65
0.59 (0.41e0.95) 0.41 (0.26e1.2)
(133.25e1449.5) (5e101.5) (1e8) (1.52e3.3) (0.34e0.98) (72.3) (27.3) (55.8) (44.2)
0.64 0.41 0.73 0.41
(0.35e1.83) (0.34e0.62) (0.36e1.91) (0.29e0.62)
49 31 13 35
(38.3) (24.2) (10.2) (27.3)
0.43 0.88 1.96 0.41
(0.28e0.88) (0.41e4.08) (1.35e4.12) (0.34e0.62)
48 16 41 49
(31.2) (10.4) (26.6) (31.8)
0.41 0.6 0.43 0.68
(0.28e0.73) (0.28e0.78) (0.33e1.19) (0.41e1.89)
.03 .003 .001
.16
HF, heart failure; EF, ejection fraction. *Patients per site per month.
barriers to effective recruitment are the major reason for early termination.7 Our study demonstrated that enrollment rates were persistently low over time, at !0.7 patients site1 month1. These rate estimates appear to be even lower than contemporaneously conducted general
cardiovascular trials and are paradoxically low considering the tremendous prevalence of HF in the general population.9 Moreover, the excess costs and resources incurred by slow trial enrollment may influence patient profiles and trial outcomes.6
6 Journal of Cardiac Failure Vol.
-
No.
- -
2015
Fig. 2. Enrollment rates in heart failure clinical trials. Bars represent median enrollment rate expressed as patients per site per month. The trend over time was not significant (P 5 .35).
A number of strategies have been suggested to improve recruitment of HF patients. Enhanced multidisciplinary support is likely critical, particularly in trials for hospitalized HF, where the logistics of timely patient enrollment must combat pressures for expediency of care. Targeted communication and referrals from health care providers10,11 may augment patient volume. Patient education may overcome cultural barriers to participation in certain ethnic groups.12e14 To test these strategies within sites, a pretrial registry preceding official trial enrollment has been proposed. Although yet to be implemented, a pretrial registry could help inclusion of high-quality centers capable of enrolling appropriate patients and meeting the needs of the subsequent trial.15 Over the past decade, HF trials with positive outcomes decreased, especially considering that the present data from high-tier medical journals likely overestimate the success of trials due to publication bias.16,17 These findings are unsettling when one also considers that positive trials were
Fig. 3. Heart failure trial outcomes over time. The trend for positive clinical outcomes over the 12-year study period was statistically significant (P 5 .04).
more likely to be smaller single-center experiences that had nonmortality outcomes, likely as part of early experience with drug development. The use of surrogate and composite end points in contemporary trials may be due to attempts to shorten the timeline with faster end point accrual. Although trial execution likely plays a large role in declining rates of positive trial outcomes, other factors including poor understanding of study drugs and failure to match the study drug with the correct patient population may shape these recent trends.18 The use of surrogate markers continues to play an important role in early-phase clinical trials. Surrogate end points facilitate limiting sample size requirements and increase feasibility of smaller-budget clinical trials over a shorter duration, or they allow assessing those interventions where clinical benefit is uncertain and further insights are needed before proceeding with pivotal large-scope and -budget trials. Certain surrogate end points may provide early therapeutic or safety signals that could inform future trial planning and direction. Finally, surrogate markers may have value independent from downstream morbidity and mortality, and they may capture patient-centered metrics, including functionality and quality of life. However, despite the theoretic rationale of surrogate end points in early-phase studies, there is an important disconnect between phase II and III clinical trials, especially in certain settings, eg, acute HF.18 These measures may be costly, may be difficult to standardize across sites, and may have important limiting characteristics. Therefore, surrogate end points need to continue to be carefully studied in conjunction with ‘‘hard’’ clinical events. The key factors that will influence the future utility of surrogate end points in phase II clinical trials include the use of validated metrics that are (1) driven by a biologic rationale of potential therapeutic response, (2) predictive of clinical outcomes, (3) studied in a well phenotyped HF subgroup, and (4) measured in a standardized fashion in a central core laboratory. Our study revealed that recent trials conducted outside NA achieved more efficient enrollment.19 Differences in rate of enrollment may be explained by systematic variation in risk profiles, background treatments, and systems of care.20e22 Despite higher enrollment with greater number of sites, multicenter/multiregion trials tended to have the worst enrollment rates. Although such broad geographic catchments may be necessary to generate the required study power for evaluating mortality and hospitalization end points, this pattern of globalization may influence trials in other ways. For example, in the TOPCAT (Treatment of Preserved Cardiac Function Heart Failure With an Aldosterone Antagonist) study, rates of the primary outcome were significantly reduced with the use of spironolactone in the Americas, a benefit that was not observed in the overall trial population and in Russia/Georgia. This differential effect generates the hypothesis that the neutral study results were a consequence of geographic variation in patient characteristics, site enrollment, or study execution.23 Thus,
Trends in HF Clinical Trials
Samman Tahhan et al
7
Table 3. Trial Characteristics Based on Trial Outcomes, n (%) Trial Characteristic Journal Annals of Internal Medicine British Medical Journal Circulation European Heart Journal Journal of the American College of Cardiology Journal of the American Medical Association Lancet New England Journal of Medicine Trial classification Chronic HF recruited from outpatient Acute HF recruited from the inpatient Chronic HF recruited from the inpatient Chronic HF recruited from both in/outpatient Year published 2001e2003 2003e2006 2007e2009 2010e2012 Intervention Device Medication Others Procedure Surgery Testing/imaging Heart failure classification HF with preserved EF HF with reduced EF HF regardless of EF Unclear Sponsor Government Industry University/Organization Unclear Trial size Patients Sites Countries Duration (y) Enrollment rate (patients per site per month) Single region Multiregional Single country Multinational Region North America Western Europe Rest of the world Mixed Primary outcome All-cause mortality Cardiovascular or HF-related death Nonmortality intermediate outcomes Surrogate outcomes
Negative Trials
Positive Trials
P Value .01
1 1 11 9 7 10 4 15
(33.3) (50) (37.9) (47.4) (17.5) (83.3) (28.6) (42.9)
2 1 18 10 33 2 10 20
(66.7) (50) (62.1) (52.6) (82.5) (16.7) (71.4) (57.1)
19 12 8 19
(28.4) (44.4) (66.7) (39.6)
48 15 4 29
(71.6) (55.6) (33.3) (60.4)
9 9 21 19
(26.5) (26.5) (45.7) (47.5)
25 25 25 21
(73.5) (73.5) (54.3) (52.5)
6 33 11 4 1 3
(18.2) (40.7) (45.8) (80) (25) (42.9)
27 48 13 1 3 4
(81.8) (59.3) (54.2) (20) (75) (57.1)
4 39 12 3
(44.4) (35.5) (44.4) (37.5)
5 71 15 5
(55.6) (64.5) (55.6) (62.5)
15 29 14 0
(50) (35.4) (37.8) (0)
15 53 23 5
(50) (64.6) (62.2) (100)
.06
.08
.06
.81
.16
627 65 2 2.33 0.55 31 17 25 31
(277e1983.5) (10e168) (1e16) (1.7e3.34) (0.38e0.83) (33.3) (48.6) (30.5) (47.7)
240 28 1 2.08 0.41 62 18 57 34
(106e610) (2e73) (1e4) (1.42e3.3) (0.26e1.19) (66.7) (51.4) (69.5) (52.3)
.001 .005 .01 .28 .55 .11 .03 .12
20 9 2 17
(40.8) (29) (15.4) (48.6)
29 22 11 18
(59.2) (71) (84.6) (51.4)
27 8 12 11
(46.6) (13.8) (20.7) (19)
21 8 29 38
(21.9) (8.3) (30.2) (39.6)
.003
HF, heart failure; EF, ejection fraction.
interpreting heterogeneous results garnered from a diverse patient population across numerous countries significantly affects generalizability.24 Compared with other funding sources, industrysponsored trials tended to be larger and multicenter to potentially offset sluggish enrollment. This approach is costly and inefficient, but whether funding mechanism influences trial outcome is unclear. In our study, industry-
and university/organization-funded studies tended to report more positive findings than government-supported trials, consistent with earlier exploratory analyses.25 Although definitive conclusions are impossible, given the complexity of trial conduct and execution, these data are intriguing, eg, selection of a comparator may bias the study results,26 and the interpretation of results may be confounded.27
8 Journal of Cardiac Failure Vol.
-
No.
- -
2015
Fewer than 6% of recent clinical trials were conducted in patients with HFpEF, and !17% included patients hospitalized for HF. Unfortunately, these high-risk patient subsets have limited treatment options and have had largely unchanged outcomes over the past 2 decades.5 Accordingly, our analyses, coupled with recent studies of ClinicalTrials.gov, reveal important unmet needs and potential areas of future drug development. A number of limitations must be considered when interpreting these findings. Study enrollment rate was calculated on the basis of overall patient enrollment per site per a given unit of time; however, this approach does not account for site discontinuation or additions during the course of the trial. These data are not published and could result in underestimating the true enrollment rates. Given the inclusion of clinical trials published only in 8 high-impact journals rather than all HF trials, generalizability of these results may be limited and success of trial programs may be overestimated. In conclusion, recent trends in HF clinical trials have important implications on future research investment, funding mechanisms, and patient confidence. Moreover, disclosure of further trial conduct data in the public domain will help ascertain these trends more carefully. Disclosures Dr Fonarow reports significant research grants from the National Heart, Lung, and Blood Institute and Agency for Healthcare Research and Quality, significant consultancy support from Novartis and Takeda, and modest consultancy support from Bayer, Gambro, Janssen, The Medicines Company, and Medtronic. Dr Gheorghiade reports relationships with Abbott, Astellas, Astrazeneca, Bayer, Cardiorentis, Corthera, Cytokinetics, Cytopherx, Debiopharm, Errekappa Terapeutici, Glaxosmithkline, Ikaria, Intersection Medical, Johnson and Johnson, Medtronic, Merck, Novartis, Ono Pharma, Otsuka, Palatin Technologies, Pericor Therapeutics, Protein Design, Sanofi-Aventis, Sigma Tau, Solvay, Sticares Interact, Takeda, and Trevena. Dr Butler reports research support from the National Institutes of Health and European Union and is a consultant to Amgen, Bayer, BG Medicine, Celladon, Gambro, GE Healthcare, Harvest, Medtronic, Ono Pharma, Stemedica, and Trevena. The other authors report no relevant conflicts of interest.
Supplementary Data Supplementary data related to this article can be found at http://dx.doi.org/10.1016/j.cardfail.2015.06.014.
References 1. Butler J, Marti C, Pina I, DeFilippi C. Scope of heart failure hospitalization. Congest Heart Fail 2012;18(Suppl 1):S1e4.
2. Bui AL, Horwich TB, Fonarow GC. Epidemiology and risk profile of heart failure. Nat Rev Cardiol 2011;8:30e41. 3. Gheorghiade M, Vaduganathan M, Greene SJ, Mentz RJ, Adams KF Jr, Anker SD, et al. Site selection in global clinical trials in patients hospitalized for heart failure: perceived problems and potential solutions. Heart Fail Rev 2014;19:135e52. 4. Suter LG, Li SX, Grady JN, Lin Z, Wang Y, Bhat KR, et al. National patterns of risk-standardized mortality and readmission after hospitalization for acute myocardial infarction, heart failure, and pneumonia: update on publicly reported outcomes measures based on the 2013 release. J Gen Intern Med 2014;29:1333e40. 5. Butler J, Fonarow GC, Gheorghiade M. Strategies and opportunities for drug development in heart failure. JAMA 2013;309:1593e4. 6. Butler J, Subacius H, Vaduganathan M, Fonarow GC, Ambrosy AP, Konstam MA, et al. Relationship between clinical trial site enrollment with participant characteristics, protocol completion, and outcomes: insights from the EVEREST (Efficacy of Vasopressin Antagonism in Heart Failure: Outcome Study With Tolvaptan) trial. J Am Coll Cardiol 2013;61:571e9. 7. Bernardez-Pereira S, Lopes RD, Carrion MJ, Santucci EV, Soares RM, de Oliveira Abreu M, et al. Prevalence, characteristics, and predictors of early termination of cardiovascular clinical trials due to low recruitment: insights from the ClinicalTrials.gov registry. Am Heart J 2014; 168:213e219.e1. 8. Blanton S, Morris DM, Prettyman MG, McCulloch K, Redmond S, Light KE, et al. Lessons learned in participant recruitment and retention: the EXCITE trial. Phys Ther 2006;86:1520e33. 9. Butler J, Samman Tahhan A, Georgiopoulou V, Kelkar A, Lee M, Khan B, et al. Trends in characteristics of cardiovascular clinical trials 2001e2012. Am Heart J 2015. In press. doi:10.1016/j.ahj.2015.05.006. 10. Galbreath AD, Smith B, Wood P, Forkner E, Peters JI. Cumulative recruitment experience in two large single-center randomized, controlled clinical trials. Contemp Clin Trials 2008;29:335e42. 11. Campbell MK, Snowdon C, Francis D, Elbourne D, McDonald AM, Knight R, et al. Recruitment to randomised trials: strategies for trial enrollment and participation study. The STEPS study. Health Technol Assess 2007;11. iii, ix-105. 12. Baquet CR, Commiskey P, Daniel Mullins C, Mishra SI. Recruitment and participation in clinical trials: socio-demographic, rural/urban, and health care access predictors. Cancer Detect Prev 2006;30: 24e33. 13. Gul RB, Ali PA. Clinical trials: the challenge of recruitment and retention of participants. J Clin Nurs 2010;19:227e33. 14. Chang BH, Hendricks AM, Slawsky MT, Locastro JS. Patient recruitment to a randomized clinical trial of behavioral therapy for chronic heart failure. BMC Med Res Methodol 2004;4:8. 15. Greene SJ, Shah AN, Butler J, Ambrosy AP, Anker SD, Chioncel O, et al. Designing effective drug and device development programs for hospitalized heart failure: a proposal for pretrial registries. Am Heart J 2014;168:142e9. 16. Dwan K, Gamble C, Williamson PR, Kirkham JJ, Reporting Bias G. Systematic review of the empirical evidence of study publication bias and outcome reporting biasdan updated review. PLoS One 2013;8:e66844. 17. Hopewell S, Loudon K, Clarke MJ, Oxman AD, Dickersin K. Publication bias in clinical trials due to statistical significance or direction of trial results. Cochrane Database Syst Rev 2009:MR000006. 18. Vaduganathan M, Greene SJ, Ambrosy AP, Gheorghiade M, Butler J. The disconnect between phase II and phase III trials of drugs for heart failure. Nat Rev Cardiol 2013;10:85e97. 19. Kim ES, Carrigan TP, Menon V. International participation in cardiovascular randomized controlled trials sponsored by the National Heart, Lung, and Blood Institute. J Am Coll Cardiol 2011;58:671e6. 20. Glickman SW, McHutchison JG, Peterson ED, Cairns CB, Harrington RA, Califf RM, et al. Ethical and scientific implications of the globalization of clinical research. N Engl J Med 2009;360:816e23. 21. Eapen ZJ, Vavalle JP, Granger CB, Harrington RA, Peterson ED, Califf RM. Rescuing clinical trials in the United States and beyond: a call for action. Am Heart J 2013;165:837e47.
Trends in HF Clinical Trials 22. Lang T, Siribaddana S. Clinical trials have gone global: is this a good thing? PLoS Med 2012;9:e1001228. 23. Pfeffer MA, Claggett B, Assmann SF, Boineau R, Anand IS, Clausell N, et al. Regional variation in patients and outcomes in the Treatment of Preserved Cardiac Function Heart Failure With an Aldosterone Antagonist (TOPCAT) Trial. Circulation 2015;131:34e42. 24. Stough WG, Zannad F, Pitt B, Goldstein S. Globalization of cardiovascular clinical research: the balance between meeting medical needs and maintaining scientific standards. Am Heart J 2007;154:232e8.
Samman Tahhan et al
9
25. Bourgeois FT, Murthy S, Mandl KD. Outcome reporting among drug trials registered in. Ann Intern Med 2010;153:158e66. ClinicalTrials.gov. 26. Lexchin J, Bero LA, Djulbegovic B, Clark O. Pharmaceutical industry sponsorship and research outcome and quality: systematic review. BMJ 2003;326:1167e70. 27. Als-Nielsen B, Chen W, Gluud C, Kjaergard LL. Association of funding and conclusions in randomized drug trials: a reflection of treatment effect or adverse events? JAMA 2003;290:921e8.
