Relationship Between Timing of Trial Randomization, Protocol Completion, and Clinical Outcomes Among Patients Hospitalized for Heart Failure: From the ASTRONAUT Trial

Multiple large phase III trials of hospitalized heart failure (HF) patients have tested the effects of pharmacologic agents on post-discharge endpoints, but none of these therapies have been definitively shown to improve patient outcomes. Multiple reasons for the lack of ‘positive’ trials have been proposed, but increasing attention is being directed towards clinical trial design and study site performance.1, 2 In this respect, nearly all prior phase III programmes have required eligible patients to be randomized and initiated on study therapy soon after hospital presentation, generally within 48 h. The Aliskiren Trial on Acute Heart Failure Outcomes (ASTRONAUT) was the only phase III trial of hospitalized HF patients with no such requirement, and thus provides a unique opportunity to study the association between timing of trial enrolment during hospitalization and patient profiles and event rates.3 The study design and primary results of the ASTRONAUT trial have been previously reported.3, 4 Briefly, ASTRONAUT was a prospective, multicentre, global, randomized placebo-controlled trial investigating the effects of aliskiren, a direct renin inhibitor, on clinical outcomes among patients hospitalized for HF. Eligible patients were aged ≥ 18 years with an ejection fraction ≤ 40%, elevated admission natriuretic peptide level [B-type natriuretic peptide (BNP) ≥ 400 pg/mL or N-terminal pro-BNP (NT-proBNP) ≥1600 pg/mL], and hospitalized with a primary diagnosis of worsening HF. Before randomization, patients were required to be haemodynamically stable, defined as systolic blood pressure ≥110 mmHg for ≥ 6 h and no use of intravenous vasodilators (except nitrates) or inotropes at any time since hospital presentation. Key exclusion criteria included estimated glomerular filtration rate < 40 mL/min/1.73 m2, haemodynamically significant valve disease, and stroke, myocardial infarction, or cardiac surgery within 3 months. The trial was conducted in accordance with the Declaration of Helsinki and with institutional review board/ethics committee approval at all sites. All patients provided written informed consent. The present analysis included all 1615 patients in the ASTRONAUT efficacy cohort. Patients were enrolled from 289 sites in 24 countries across North America, Latin America, Western Europe, Eastern Europe, and Asia/Pacific between May 2009 and December 2011. Patients were grouped by quartiles of time (in days) from admission to randomization. Baseline demographics, vital signs, laboratory studies, medical and medication histories, and clinical events were compared across time-to-randomization quartiles using chi-squared, analysis of variance, and Kruskal–Wallis tests as appropriate. Rates of protocol non-completion, including patient withdrawal of consent, loss to follow-up, and protocol deviation, were also compared across quartiles. All statistical analyses were performed using SAS version 9.3 (SAS Institute, Cary, NC, USA). Two-tailed P < 0.05 was considered statistically significant. Median (25th–75th percentile) time from admission to randomization in ASTRONAUT was 5 (3–8) days and ranged from < 1 day to 47 days. Time to randomization varied by geographic region, with shortest times in North America [median 2 (2–4) days] and longest times in Eastern Europe [median 6 (3–9) days] (P < 0.001). Within all regions except Latin America, there was wide variation in time to randomization by country (Figure 1). Of the 1615 total patients, 483 (29.9%) were randomized within 0–2 days of admission, 352 (21.8%) during days 3–4, 423 (26.2%) during days 5–7, and 357 (22.1%) during days 8–47. Median hospital length of stay ranged from 4 (3–8) days among patients in quartile 1 to 15 (11–20) days in quartile 4. Patients randomized earlier tended to have less severe New York Heart Association (NYHA) class at admission, but more severe functional class at randomization. With exception of patients randomized earlier being younger with marginally lower ejection fraction, baseline patient characteristics were similar by time-to-randomization quartile (Table 1). Notably, there were no significant differences in vital signs, renal function, NT-proBNP level, or co-morbidities. Likewise, there were no significant differences in rates of all-cause mortality, cardiovascular mortality, all-cause hospitalization, or HF hospitalization at 30-day, 6-month, or 12-month follow-up. Rates of protocol non-completion, including patient withdrawal of consent, loss to follow-up, and protocol deviation, were similarly low across time-to-randomization quartiles. In this global trial of patients hospitalized for HF, patients randomized early and late during hospitalization had generally similar baseline characteristics, rates of protocol completion, and post-discharge clinical events. Although ASTRONAUT did not specify an in-hospital enrolment window, review of trial inclusion/exclusion criteria and baseline patient characteristics suggests the large majority of ASTRONAUT patients enrolled within 2 days of admission (i.e. quartile 1) would have been eligible for other phase III hospitalized HF trials testing intravenous or oral medications. Thus, current findings may inform the utility of in-hospital enrolment windows in other programmes. Recent large trials in hospitalized HF populations have been designed to test the effects of investigational medical therapies initiated soon after hospital presentation (e.g. within 24–48 h) on post-discharge mortality and rehospitalization endpoints.5-7 This design is consistent with hospitalization representing a distinct biologic event characterized by end-organ injury, and the hypothesis that rapid introduction of study drug may attenuate such injury and improve post-discharge outcomes.8 With acknowledgement that all ASTRONAUT patients were required to be haemodynamically stable for ≥ 6 h prior to randomization, the observed consistency across vital signs, laboratory data, co-morbidities, and mortality rates among haemodynamically stable patients randomized early vs. later during HF hospitalization does not support overt biologic differences. Rather, these data suggest a randomization window anchored to time of hospital admission may not meaningfully influence longer-term outcomes of the enrolled subset. These findings are consistent with reports highlighting the subacute nature of many worsening HF episodes and potential influences of non-biologic factors (e.g. country-specific and local healthcare infrastructure/resources, financial incentives, patient preferences) in dictating exact timing or occurrence of HF hospitalization.9 Nonetheless, noting that NYHA class at randomization was more severe among patients randomized earlier in ASTRONAUT, early in-hospital enrolment may be preferable for trials focused on short-term symptomatic improvement with in-hospital endpoints (e.g. dyspnoea relief). Aside from potential inability to define post-discharge risk, mandatory randomization within a narrow in-hospital window may pose a significant burden for study sites and prevent efficient enrolment. Moreover, although there was no signal of excess protocol non-completion among patients randomized early in ASTRONAUT, it is possible that pressure to complete enrolment and randomization within strict time limits may increase rates of errors when determining patient eligibility, or worsen the quality of informed consent discussions between study staff and patients. These effects could conceivably increase chances of subsequent protocol deviations, patient withdrawal of consent, or adverse events. The feasibility of early randomization may also vary by country of enrolment and study site, as suggested in Figure 1. In the current era of generally difficult and slow enrolment in HF clinical trials, these collective data from ASTRONAUT suggest that trials of hospitalized HF patients focused on post-discharge endpoints may benefit from allowing randomization throughout the in-hospital period to improve enrolment efficiency without any meaningful impact on the clinical profile and event rates of the accrued study population. Financial and material support for the ASTRONAUT trial was provided by Novartis Pharma AG (Basel, Switzerland). Haris P. Subacius conducted all final analyses for this report with funding from the Center for Cardiovascular Innovation, Northwestern University Feinberg School of Medicine, Chicago, IL, USA, and takes responsibility for the integrity of the data. Conflict of interest: S.J.G. is supported by the National Institutes of Health (NIH) T32 post-doctoral training grant (5T32HL069749-14), a Heart Failure Society of America/Emergency Medicine Foundation Acute Heart Failure Young Investigator Award funded by Novartis, and has received research support from Amgen and Novartis. G.C.F. reports research funding from the NIH and serving as a consultant for Amgen, Bayer, Medtronic, and Novartis. S.D.S. has received grant funding, consultant fees, and travel support from Novartis. M.V. is supported by the NIH T32 postdoctoral training grant (T32HL007604). A.P.A. is supported by the NHLBI T32 postdoctoral training grant (5T32HL069749). M.F. receives research support from an American Heart Association grant (17MCPRP33460225) and the NIH T32 grant (5T32HL007101). M.B. has served as a consultant for AstraZeneca, Bayer, Boehringer-Ingelheim, Daiichi-Sankyo, AWD Dresden, Berlin-Chemie, MSD, Novartis, Pfizer, Sanofi-Aventis, and Servier. A.P.M. has served on committees of clinical studies sponsored by Amgen, Bayer, Abbott Vascular, Cardiorentis, Johnson & Johnson, and Novartis Pharma AG. F.Z. reports personal fees from Janssen, personal fees from Bayer, personal fees from Novartis, personal fees from Boston Scientific, personal fees from Resmed, personal fees from Amgen, personal fees from CVRx, personal fees from Quantum Genomics, personal fees from General Electric, personal fees from Boehringer, other from cardiorenal, personal fees from AstraZeneca, personal fees from Vifor Fresenius, personal fees from Cardior, outside the submitted work. J.B. has received research support from the NIH, PCORI and the European Union; and serves as a consultant for Amgen, Array, AstraZeneca, Bayer, Boehringer-Ingelheim, Bristol Myers Squib, CVRx, G3 Pharmaceutical, Innolife, Janssen, Luitpold, Medtronic, Merck, Novartis, Relypsa, StealthPeptide, SC Pharma, Vifor, and ZS Pharma. All other authors declare no relevant disclosures.