Heterogeneity Across One Disease, Two Drugs, Three Trials, and Four Guidelines: Are we Further Ahead?

This article refers to ‘Combined neprilysin and renin–angiotensin system inhibition in heart failure with reduced ejection fraction: a meta-analysis’ by S.D. Solomon et al., published in this issue on pages 1238–1243. Mind the Gap Audible warning on the London Underground Guideline committees often differ in their approach as to how to integrate new information into a recommendation that ultimately helps shape practice, policy, and patient outcomes. Since there is a single scientific body of evidence from which to draw, guideline recommendations are often similar. However, they differ as to the context to which their guideline (and ultimately clinician and patient end-user) will apply this, and the values and preferences that shape a recommendation. How do these gaps between guidelines play a role in clinical practice, do they matter, and should we pay them heed? Sacubitril/valsartan, an angiotensin receptor–neprilysin inhibitor (ARNI), is a new medication targeting the known pathways that are altered in the setting of heart failure (HF). ARNIs are a part of the broader family of drugs that have combined neutral endopeptidase (NEP) and renin–angiotensin–aldosterone system (RAAS) inhibition properties, reviewed elsewhere.1 After the publication of a landmark randomized controlled trial (RCT) called the Prospective Comparison of ARNI with ACEI to Determine Impact on Global Mortality and Morbidity in Heart Failure (PARADIGM-HF) that tested the use of an ARNI, guideline committees, payers, clinicians, and patients are discussing the relative merits of switching from a therapy that has been a cornerstone for >25 years to a newer medication. All major guidelines assigned a relatively strong recommendation for the use of this drug (Table 1) which effectively means that there is general agreement that the treatment is beneficial for patients.2-5 However, despite the results of the single PARADIGM-HF trial (global multicentre trial, >2000 hard events, consistent results across different outcomes, and internally consistent across subgroups and time), the European and US guidelines assigned a level of evidence ‘B’ for sacubitril/valsartan.4, 5 Per their guideline construct, for ‘level A’ evidence, multiple RCTs or high quality meta-analyses are required. This puts their guidelines committee in a bind: does the robustness of the findings of PARADIGM-HF offset the above-mentioned requirements? Is this an outdated concept that needs refinement? To put this in context, diuretics were given the same I-B recommendation as sacubitril/valsartan in the ESC guidelines. In contrast, the Canadian guidelines apply the Grading of Recommendations Assessment, Development and Evaluation (GRADE) guideline technique which allows for consideration of the quality of data across multiple domains including the effect size, precision, and consistency, to name a few, and not just the number of RCTs. The relative merits of either system need careful consideration—no system is perfect. On this backdrop, the leading guidelines (representing Europe, Canada, the UK, and the USA) differ enough for us to consider the cause of this dissention—and how Solomon et al. may or may not help those that require a meta-analysis. In the current issue of the journal, Solomon et al. have presented the results of a meta-analysis of three RCTs in which a RAAS inhibitor was combined with a NEP inhibitor compared with RAAS inhibition alone in patients with heart failure with reduced ejection fraction (HFrEF).6 Only three studies met their criteria which span >20 years of clinical investigation.7-9 The trials differed in their structure: one of the three had a run-in period, was a year longer, and 1.5× larger than the next largest trial. This methodological heterogeneity is pertinent as to whether or not we should even combine these trials, and, if so, how we should interpret the pooled hazard ratio (HR). Solomon et al. chose to pool the three trials, and with a 12% reduction in all-cause death, the combined NEP/RAAS inhibitors were shown to have a mortality reduction compared with RAAS inhibition alone. The composite outcome of death or HF hospitalization was decreased in patients receiving NEP/RAAS inhibition compared with those who received the RAAS inhibition with ACE inhibitors [pooled HR 0.86, 95% confidence interval (CI) 0.76–0.97]. When applying a more contemporary definition for HF hospitalization used in the largest and most recent trial (which allows for oral diuretic intensification), a 15% reduction in the composite endpoint of death and HF hospitalization was seen. In the results of this meta-analysis, there is statistical heterogeneity (an I2 = 41% for mortality) between trials of different agents.6 Several issues deserve further consideration. Clinical heterogeneity is seen where there is variance in the patients enrolled, the intervention, the control, or the outcomes. In this instance, the intervention—two different combination drugs—are lumped together for comparison. Is this a drug class? Drugs are usually considered to be from the same class if they have similar chemical structures and similar mechanisms of action. Omapatrilat was a combined ACE and neprilysin inhibitor, but sacubitril/valsartan is an angiotensin receptor and neprilysin inhibitor. Few would group ACE inhibitors and ARBs in the same class, and guidelines differentiate and do not equate these based on the outcomes of patients enrolled in the large trials of either agent. Nevertheless, they do have a similar mechanism of action and, as outlined below, similar clinical outcomes but with potentially differing safety profiles. The clinical investigative community, having been burned before, should exercise caution in equating these two agents. The trials also differed in their background therapies consistent with the era in which they were conducted. For example, the rate of beta-blocker use was higher in the PARADIGM-HF trial (92%) as compared with the Omapatrilat Versus Enalapril Randomized Trial of Utility in Reducing Events (OVERTURE) study (51%) or the Inhibition of MetalloProtease by omapatrilat in a Randomized Exercise and Symptom Study of heart failure (IMPRESS) trial (30%), and there were few implantable defibrillators in any of the trials. Remarkably, the trials enrolled patients of similar age (∼63 years old), four-fifths were men with similar heart rates, systolic blood pressures, and (by design) mean EFs all under 30%. The study of Solomon et al. is an aggregate data meta-analysis and the investigators had no access to the patient-level data of the IMPRESS and OVERTURE trials, and thus analysis of information on subpopulations or other variation is not possible. Although access to patient-level data may have resolved some of the issues mentioned above, it may not fully address all the issues of heterogeneity, e.g. different drugs and doses. Another factor is the use of a pre-randomization run-in phase in PARADIGM-HF but not in the IMPRESS and OVERTURE trials. In PARADIGM-HF, 10521 patients entered sequential, single-blind run-in periods (first enalapril then sacubitril/valsartan) to ensure that randomized patients will tolerate the target doses of the study drugs. Among those who entered the run-in phase, 10.5% failed to complete the enalapril step and a further 9.3% (10.3% of the remaining population) failed to complete the sacubitril/valsartan run-in phase.10 This makes cross-trial comparisons of efficacy and safety difficult, especially on rare events such as angioedema. Although there was a clear rationale for having an active treatment run-in phase, it has caused controversy as to the generalizability of the findings of this trial to the broader population of patients with HF, and, thus, does a meta-analysis shine light on this issue? A secondary analysis of patients with baseline characteristics resembling those who failed to complete the run-in phase did not alter the apparent benefit of sacubitril/valsartan over enalapril.10 From this, we could conclude that the run-in phase will have limited effect on the pooled estimates of efficacy in this meta-analysis, but does not solve an issue regarding the underestimation of the adverse events. Across all three trials, combined NEP/RAAS inhibition was shown to be associated with more hypotension, but less renal dysfunction and hyperkalaemia compared with ACE inhibition alone. Since these trials have used different definitions for adverse events, should they be combined? For example, the IMPRESS trial provided the vague definition of ‘significantly elevated potassium’ for hyperkalaemia, OVERTURE has failed to report it (and was not found by Solomon et al.), and PARADIGM-HF defined it as potassium >6.0 mg/dL. Hypotension is reported as ‘symptomatic hypotension’ in PARADIGM-HF, whilst it is reported merely as ‘hypotension’ in both of the other trials. The safety assessment could be more reliable if the authors were able to perform an individual participant data meta-analysis rather than aggregating the adverse events with disparate definitions. Thoughtful analysis of well-curated data via open access to data in the future will offer a new window, and most probably provide additional comfort to those who see patients—especially those patients who are on the margins of what was tested in the RCTs. In general, although there are remaining uncertainties associated with NEP/RAAS inhibition, this meta-analysis may not provide rigid guideline constructs the information needed to modify their level of evidence. Given the significant clinical and methodological heterogeneity evident in the meta-analysis of these three trials, it is doubtful that most will look beyond the single recent pivotal trial and most clinicians won't ‘mind the gap’ they see between guidelines when caring for their patients. The largest gap of efficiently translating RCT evidence to clinical practice still needs to be traversed and is a far more worthy objective for guidelines and their respective medical societies. Conflict of interest: J.A.E. is a Population Health Investigator supported by Alberta Innovates–Health Solutions (Alberta, Canada), is the Chair of the Canadian Cardiovascular Society Guidelines on Heart Failure, and has received research funding or honoraria from Amgen, Bayer, Merck, Novartis, Servier, and Trevena. N.S. has no conflicts to declare.

Récupéré en direct depuis OpenAlex et désinversé. Les résumés ne sont pas conservés dans cette base de données : les index inversés représentent 8,6 Go des 9,3 Go de texte de la base, et le serveur dispose de 13 Go libres.