Editorial
January 17, 2023
TRANSFORM-HF—Can We Close the Loop on Diuretics in Heart Failure?
Michelle M. Kittleson
JAMA. 2023;329(3):211-213. doi:10.1001/jama.2022.21692
Fluid overload is the cardinal manifestation of heart failure, and prior to the 20th century, treatment options were limited to purgatives, bloodletting, leeching, or lancing. It was not until 1920 that the concept of a diuretic was first appreciated, when physicians observed increased urination in patients given mercury as a remedy for syphilis. The 1950s and 1960s ushered in safe and effective diuretic therapy. Chlorothiazide, the first orally active diuretic, was approved for use in the United States in 1958, and furosemide, the first loop diuretic, in 1964.
By 2020, furosemide was the 19th most prescribed medication in the United States,1 yet the evidence base for the 26 million annual prescriptions includes only 7 placebo-controlled trials comprising less than 500 patients.2Loop diuretics may be more reliable, effective, and safe than mercury, leeches, or bloodletting, but like those erstwhile blunt instruments, their place in the heart failure armamentarium is based on necessity, not evidence. And while congested patients feel better, there is potential for harm: administration of loop diuretics results in an acute vasoconstrictor response3 and more neurohormonal activation,4 and higher diuretic doses are associated with increased mortality.5,6
Loop diuretics treat the symptoms of heart failure, not the underlying disease process—but fortunately, the past few decades have brought revolutionary advances in therapies that target the neurohormonal pathways implicated in the pathogenesis of heart failure. The combination of the angiotensin receptor-neprilysin inhibitor sacubitril/valsartan, an evidence-based β-blocker (bisoprolol, carvedilol, or sustained-release metoprolol), a mineralocorticoid antagonist (spironolactone or eplerenone), and a sodium-glucose cotransporter 2 inhibitor (dapagliflozin or empagliflozin) improves quality of life, reduces heart failure hospitalizations, and improves survival.7
Optimal evidence- and guideline-based therapies offset the adverse hemodynamic and neurohormonal effects of loop diuretics. However, these therapies may mitigate8,9 but do not obviate, the need for diuretics, which remain a cornerstone of heart failure management. Still, more than 50 years after the advent of furosemide, an unresolved question remains: what is the best diuretic for routine use in heart failure?
The Torsemide Comparison With Furosemide for Management of Heart Failure (TRANSFORM-HF) trial in this issue of JAMA addresses this key clinical question.10 While furosemide is the most commonly used diuretic in patients with heart failure,11 torsemide may be superior. Compared with furosemide, torsemide has increased bioavailability and potency. Unlike furosemide, torsemide downregulates the renin-angiotensin-aldosterone system, reduces myocardial fibrosis, and fosters reverse myocardial remodeling.12 In small, open-label, randomized trials of torsemide vs furosemide, torsemide improved quality of life13 and reduced hospitalization.14 In a larger nonrandomized analysis of the 2 loop diuretics, there was a more than 50% reduction in mortality with torsemide vs furosemide after 9 months.15
Against this backdrop of compelling preliminary data, TRANSFORM-HF was an open-label, pragmatic trial that randomized 2859 patients hospitalized with heart failure at 60 hospitals in the United States to torsemide or furosemide. There was no difference in mortality at 30 months: 26.1% in the torsemide group and 26.2% in the furosemide group (hazard ratio, 1.02 [95% CI, 0.89-1.18]). There was no difference in hospitalizations at 12 months: 940 hospitalizations in the torsemide group and 987 hospitalizations in the furosemide group (rate ratio, 0.94 [95% CI, 0.84-1.07]).
TRANSFORM-HF was a negative trial that advances the field of heart failure in 2 important ways. First, TRANSFORM-HF provides valuable lessons to clinicians on the importance of negative trials in clinical practice. Second, TRANSFORM-HF illustrates the benefits and challenges inherent in the design and execution of pragmatic trials.
The authors should be commended for demonstrating the feasibility of a pragmatic randomized trial in heart failure. The primary goal of a traditional randomized trial is to determine efficacy and safety of an intervention under ideal conditions. Creation of this idealized setting, however, can result in low and slow enrollment and high cost. In contrast, a pragmatic trial balances the benefit of randomization with less stringent enrollment criteria and follow-up, affording the ability to efficiently generate evidence with widespread applicability.
TRANSFORM-HF lived up to its potential as a pragmatic randomized trial with broad eligibility to any patient with heart failure excluding only those with limited life expectancy or receiving dialysis; centralized and remote study follow-up, with no requirement for site visits; and streamlined data collection with no required interventions beyond standard of care. The investigators reaped the benefits of this approach with average recruitment of more than 2 patients per site per month prior to the COVID-19 pandemic, a striking improvement over the average enrollment rate of 0.68 patients per site per month in traditional randomized heart failure trials.16 TRANSFORM-HF also had greater diversity: 36.9% of trial participants were women and 33.9% were Black individuals; prior heart failure trials have typically included less than 30% women and less than 20% Black individuals.17
Publication of high-quality, randomized clinical trials with negative findings is essential to mitigate publication bias. TRANSFORM-HF joins a catalog of cautionary tales in cardiology, whereby carefully executed negative trials have refuted the misleading promise of plausible surrogate end points and preliminary data. Flecainide suppressed premature ventricular contractions after myocardial infarction but increased the risk of death,18 thiazolidinediones reduced hemoglobin A1c levels in patients with diabetes but increased the risk of heart failure,19 and nesiritide reduced pulmonary capillary wedge pressure but had no impact on heart failure hospitalizations or mortality.20 These trials are no less valuable for their negative conclusions, and neither is TRANSFORM-HF. Despite putative mechanistic advantages and optimism from small or nonrandomized studies, torsemide did not reduce heart failure hospitalizations or mortality compared with furosemide. The lesson: clinicians should have a healthy suspicion for plausible pathophysiology, surrogate end points, and nonrandomized data as the sole basis of defining superiority of an intervention.
Nonetheless, with any negative trial, one must convince oneself that the findings are evidence of absence rather than absence of evidence. TRANSFORM-HF is a large, well-designed, randomized trial, yet the possibility of absence of evidence still exists, a by-product of the pragmatic trial design. The broad entry criteria promoted rapid enrollment and widespread generalizability at the cost of a heterogeneous trial population. The limited patient monitoring streamlined trial participation for patients and study sites while sacrificing the ability to monitor patient adherence and perform detailed outcome assessment of end points relevant to heart failure other than mortality.
For example, TRANSFORM-HF included patients with both heart failure with reduced ejection fraction and preserved ejection fraction. Because the pathophysiology and natural history of these conditions differ, could there have been a differential treatment effect masked by the heterogenous study group? In addition, there was a 5.4% crossover rate between torsemide and furosemide groups at hospital discharge and information on diuretic dosage was available for only about 70% of trial participants at 1 month and about two-thirds of study participants at 1 year. Could lack of study drug adherence have impacted the assessment of efficacy? Further, while torsemide had no impact on rehospitalizations at 12 months or death at 30 months, could there have been a difference in unscheduled clinic visits, presentation to the emergency department, need for thiazide diuretics, or changes in kidney function? These questions will remain unanswered given the constraints of a pragmatic trial design.
The TRANSFORM-HF Trial should be celebrated for dispelling a longstanding myth, based on surrogate markers and small trials, of the superiority of torsemide over furosemide. Now, when faced with a patient with heart failure and congestive symptoms, clinicians can focus their energy on what really matters: not the relative merits of different loop diuretics, but rather the initiation and optimization of evidence- and guideline-based therapies to help their patients feel better and live longer.7
Nevertheless, celebration of TRANSFORM-HF should be tempered with caution. Pragmatic trials are more flexible and nimbler in design and execution, but this agility comes at a cost. An overly heterogeneous patient population can impact the trial’s ability to assess efficacy of therapies while minimally intensive follow-up precludes comprehensive outcome assessment.
In less than a century, the management of heart failure has been transformed from toxic methods of fluid removal to therapies that improve the patients’ quality of life and survival. The future of heart failure lies in pragmatic trials that merge the rigor of randomized trials with the capacity for efficient generation of generalizable evidence. The key to the success of these trials, however, will be in balancing the competing goals of rigor and efficiency. TRANSFORM-HF has closed the loop on diuretic therapy in heart failure—for now—and offers valuable lessons to inform the design of future studies that will further optimize the care of patients with heart failure.
